(Methods in Molecular Biology 1781) Qing Yan-Psychoneuroimmunology-Springer New York - Humana Press (2018)

Methods in
Molecular Biology 1781
Qing Yan Editor
Psychoneuro-
immunology
Methods and Protocols
Second Edition
METHODS IN MOLECULAR BIOLOGY
Series Editor
John M. Walker
School of Life and Medical Sciences
University of Hertfordshire
Hatfield, Hertfordshire, AL10 9AB, UK
For further volumes:

http://www.springer.com/series/7651
Psychoneuroimmunology
Methods and Protocols
Second Edition
Edited by
Qing Yan
PharmTao, Santa Clara, CA, USA
University of Maryland University College, Adelphi, MD, USA
Editor
Qing Yan
PharmTao
Santa Clara, CA, USA
University of Maryland University College
Adelphi, MD, USA
ISSN 1064-3745 ISSN 1940-6029 (electronic)

Methods in Molecular Biology
ISBN 978-1-4939-7827-4 ISBN 978-1-4939-7828-1 (eBook)
https://doi.org/10.1007/978-1-4939-7828-1
Library of Congress Control Number: 2018939428
© Springer Science+Business Media, LLC, part of Springer Nature 2018

This work is subject to copyright. All rights are reserved by the Publisher, whether the whole or part of the material is
concerned, specifically the rights of translation, reprinting, reuse of illustrations, recitation, broadcasting, reproduction
on microfilms or in any other physical way, and transmission or information storage and retrieval, electronic adaptation,
computer software, or by similar or dissimilar methodology now known or hereafter developed.
The use of general descriptive names, registered names, trademarks, service marks, etc. in this publication does not imply,
even in the absence of a specific statement, that such names are exempt from the relevant protective laws and regulations
and therefore free for general use.
The publisher, the authors and the editors are safe to assume that the advice and information in this book are believed to
be true and accurate at the date of publication. Neither the publisher nor the authors or the editors give a warranty,
express or implied, with respect to the material contained herein or for any errors or omissions that may have been made.
The publisher remains neutral with regard to jurisdictional claims in published maps and institutional affiliations.
Printed on acid-free paper
This Humana Press imprint is published by the registered company Springer Science+Business Media, LLC part of
Springer Nature.
The registered company address is: 233 Spring Street, New York, NY 10013, U.S.A.
Preface
As a multidisciplinary area, the biopsychosocial models in psychoneuroimmunology (PNI)

are becoming the central theme for understanding health and diseases. Such comprehension
would enable more accurate diagnosis and better therapeutics for the development of
personalized and systems medicine.
This book provides systems-based methodologies and innovative technologies that can
be used for solving complicated problems of complex systems. A wide range of theoretical
and experimental approaches are introduced for practical applications. In addition, this book
focuses on translational medicine by applying PNI methods in clinical practice. One of the
major challenges in current bioscience is the translation of basic scientific discoveries into
better therapeutic outcomes. This book is written in response to this challenge by high-
lighting the clinical implications of PNI.
Part I of the book introduces the basic and novel concepts in PNI, especially the
relationships among stress, inflammation, and psychophysiological disorders. Systems-
based models are presented with the emphasis on homeostasis, systemic diseases, and the
dynamical balance of health.
This part explains how PNI may provide the scientific basis for understanding the
“whole mind-body system” for the practice of dynamical systems medicine (see
Chapter 1). Systems biology models of the complex adaptive systems (CASs), such as a
conceptual framework of “Yin-Yang dynamics,” may be helpful for interpreting the complex
mechanisms in PNI. The disturbance in the Yin-Yang dynamical balance may result in stress,
inflammation, and various disorders including insomnia, Alzheimer’s disease, obesity, dia-
betes, cardiovascular diseases, and cancer. Such perception may contribute to the discovery
of systems-based biomarkers and therapeutic targets.
Specifically, studies in microbial endocrinology and PNI support the models that
emphasize the intersections between the neuroimmune and microbiota systems with their
effects on stress responses (see Chapter 2). In addition, life experience has fundamental
biological relevance with impacts on all adaptive systems such as the immune and nerve
systems with profound meanings on health and diseases (see Chapter 3). Although often
ignored, experience deserves more attention for understanding the dynamic interplay
among mind, body, and environment. The systems-based models should also consider the
ecological context and human variations by applying the concepts of biological anthropol-
ogy to achieve a more complete understanding of stress, inflammation, and well-being (see
Chapter 4).
The neuroimmune imbalances and disturbances in the Yin-Yang dynamics have been
found as the important features and potential biomarkers for psychiatric disorders including
anxiety and depression (see Chapter 5). As the brain maintains homeostasis via complex
signaling networks, the bio-behavioral feedback and feedforward patterns can be discovered
as the systems-based mechanisms of stress, anxiety, and depression (see Chapter 6).
Such studies may contribute to the promotion of resilience and well-being. As an
example, complex disorders such as Gulf War Illness (GWI) can be distinguishable by
measuring the co-expression of multiple markers of endocrine and immune functions (see
Chapter 7). Systems-based models can be constructed for neurobehavioral regulations and
v
vi Preface
signaling networks to analyze the information flow in the bio-behavioral circuitry and to
support behavioral stress management therapy (see Chapter 8).
In addition, studies in PNI such as those focusing on systemic inflammation and gut
microbiota may help elucidate the neuroimmune mechanisms of comorbid disorders includ-
ing depression in patients with heart failure (see Chapter 9). In systemic disorders such as
cancer and autoimmune diseases, the imbalances in cytokine networks, the hypothalamic-
pituitary-adrenal axis, and circadian rhythms can be critical markers for prognosis and
disease control (see Chapter 10).
Part II of this book introduces various cutting-edge technologies and methods for PNI
studies. These technologies include the utilizations of mouse models, the chromium release
whole blood assay, imaging techniques, as well as vaccine models.
For example, novel tools including optogenetics and chemogenetics may empower the
studies of the brain-immune interactions in PNI (see Chapter 11). Natural killer (NK) cells
are sensitive barometers of the effects of stressors on the immune system. A chromium
(51Cr) release whole blood bioassay can be used to examine the target cell killing capacity of
NK cells (see Chapter 12).
In addition, mouse models have extensive applications in PNI studies. Immunobeha-
vioral phenotyping is a first-line method for exploring the neuroimmune system. Behavioral
tests are frequently used to examine neuroimmune activation in mice (see Chapter 13). The
murine MRL model with high validity in analyzing principal pathogenic circuits has been
considered indispensable in understanding the brain-immune links and autoimmune dis-
eases (see Chapter 14).
Positron emission tomography (PET) imaging is a tool for measuring brain metabolism
and target molecules. By detecting brain variables, PET imaging can be combined with
other experimental and clinical model systems for PNI studies (see Chapter 15).
Furthermore, vaccination models are very useful for analyzing the effects of psychosocial
factors on immunity (see Chapter 16). Such protocols can help elucidate the association
between stress and vaccination responses. The modern multiplex techniques would
empower PNI research for promoting vaccine responses among at-risk populations (see
Chapter 17).
Moreover, the approaches for analyzing the fetal cholinergic signaling on systems and
cellular scales are very useful for the studies of PNI phenotypes (see Chapter 18). The
protocols for the research in prenatal stress and postnatal brain development would contrib-
ute to the development of perinatal PNI (see Chapter 19).
By covering topics from systems-based models to advanced technologies, this book can
be used by biomedical students and professionals at all levels who are interested in integra-
tive studies in psychology, psychiatry, neuroscience, immunology, molecular biology, genet-
ics, bioengineering, physiology, pathology, microbiology, systems biology, and clinical
medicine. Written by leading experts in the field, this book intends to provide a practical,
state-of-the-art, and holistic view for the translation of PNI into better preventive and
personalized medical practice.
I would like to thank all of the authors for sharing their profound thoughts and
experiences, and for making valuable contributions to this exciting new field. I also thank
the series editor, Dr. John Walker, for his help with the editing.
Santa Clara, CA, USA Qing Yan

Contents
Preface . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . v
Contributors. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . ix
PART I STRESS AND IMMUNITY: BIOPSYCHOSOCIAL MODELS

AND CLINICAL IMPLICATIONS
1 Stress and Systemic Inflammation: Yin-Yang Dynamics in Health

and Diseases. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 3
Qing Yan
2 Intersections Between Neuroimmune and Microbiota . . . . . . . . . . . . . . . . . . . . . . . 21
Colette G. Ngo Ndjom, Xavier F. Gonzalez, and Harlan P. Jones
3 Psychoneuroimmunology: The Experiential Dimension. . . . . . . . . . . . . . . . . . . . . . 37
Elling Ulvestad
4 Ecological Context and Human Variation: Applying the Principles
of Biological Anthropology to Psychoneuroimmunology . . . . . . . . . . . . . . . . . . . . 55
Eric C. Shattuck
5 Neuroimmune Imbalances and Yin-Yang Dynamics in Stress, Anxiety,
and Depression . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 77
Qing Yan
6 Increasing Resilience to Traumatic Stress: Understanding the Protective
Role of Well-Being . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 87
J. Tory Toole, Mark A. Rice Jr, Jordan Cargill, Travis J. A. Craddock,
Barry Nierenberg, Nancy G. Klimas, Mary Ann Fletcher,
Mariana Morris, Joel Zysman, and Gordon Broderick
7 Exploring the Diagnostic Potential of Immune Biomarker Co-expression
in Gulf War Illness . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 101
Gordon Broderick, Mary Ann Fletcher, Michael Gallagher,
Zachary Barnes, Suzanne D. Vernon, and Nancy G. Klimas
8 Breaking Away: The Role of Homeostatic Drive in Perpetuating
Depression . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 121
J. Tory Toole, Mark A. Rice Jr, Travis J. A. Craddock,
Barry Nierenberg, Nancy G. Klimas, Mary Ann Fletcher,
Joel Zysman, Mariana Morris, and Gordon Broderick
9 Neuroimmune Mechanisms of Depression in Adults with Heart Failure . . . . . . . 145
Jessica A. Jiménez, Christine Tara Peterson, and Paul J. Mills
10 How to Monitor the Neuroimmune Biological Response
in Patients Affected by Immune Alteration-Related Systemic Diseases . . . . . . . . . 171
Paolo Lissoni, Franco Rovelli, Luigi Vigorè, Giusy Messina,
Arianna Lissoni, Giorgio Porro, and Giuseppe Di Fede
vii
viii Contents
PART II TECHNOLOGIES AND METHODS IN PSYCHONEUROIMMUNOLOGY STUDIES
11 Application of Chemogenetics and Optogenetics to Dissect

Brain-Immune Interactions . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 195
Ben Korin and Asya Rolls
12 Psychoneuroimmunology and Natural Killer Cells:
The Chromium-Release Whole-Blood Assay . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 209
Mary Ann Fletcher, Zachary Barnes, Gordon Broderick,
and Nancy G. Klimas
13 Mouse Testing Methods in Psychoneuroimmunology 2.0:
Measuring Behavioral Responses. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 221
Albert E. Towers, Jason M. York, Tracy Baynard,
Stephen J. Gainey, and Gregory G. Freund
14 The MRL Model: A Valuable Tool in Studies of Autoimmunity-Brain
Interactions . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 259
Boris Šakić
15 PET Imaging in Psychoneuroimmunology Research . . . . . . . . . . . . . . . . . . . . . . . . 287
Jonas Hannestad
16 The Vaccination Model in Psychoneuroimmunology Research: A Review . . . . . . 309
Anna C. Whittaker
17 Measuring Vaccine Responses in the Multiplex Era. . . . . . . . . . . . . . . . . . . . . . . . . . 327
Kieran Ayling, Kavita Vedhara, and Lucy Fairclough
18 Sculpting the Sculptors: Methods for Studying the Fetal
Cholinergic Signaling on Systems and Cellular Scales. . . . . . . . . . . . . . . . . . . . . . . . 341
Martin G. Frasch, Patrick Burns, Javier Benito, Marina Cortes,
Mingju Cao, Gilles Fecteau, and André Desrochers
19 Perinatal Psychoneuroimmunology: Protocols for the Study
of Prenatal Stress and Its Effects on Fetal and Postnatal Brain
Development. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 353
Martin G. Frasch, Carlos J. Baier, Marta C. Antonelli,
and Gerlinde A. S. Metz
Index . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 377
Contributors
MARTA C. ANTONELLI Facultad de Medicina, Instituto de Biologı́a Celular y Neurociencia

“Prof. Eduardo De Robertis”, Universidad de Buenos Aires, Buenos Aires, Argentina
KIERAN AYLING Division of Primary Care, School of Medicine, University of Nottingham,
Nottingham, UK
CARLOS J. BAIER Departamento de Biologı́a, Instituto de Investigaciones Bioquı́micas de
Bahı́a Blanca (INIBIBB), Consejo Nacional de Investigaciones Cientı́ficas y Técnicas
(CONICET), Bioquı́mica y Farmacia (DBByF), Universidad Nacional del Sur (UNS),
Bahı́a Blanca, Argentina
ZACHARY BARNES Department of Medicine, University of Miami, Miami, FL, USA;
Diabetes Research Institute, University of Miami, Miami, FL, USA
TRACY BAYNARD Department of Kinesiology and Nutrition, University of Illinois, Chicago,
IL, USA
JAVIER BENITO Faculty of Veterinary Medicine, Department of Clinical Sciences, University
of Montreal, St-Hyacinthe, QC, Canada
GORDON BRODERICK Center for Clinical Systems Biology, Rochester General Hospital
Research Institute, Rochester, NY, USA; Department of Biomedical Engineering, Rochester
Institute of Technology, Rochester, NY, USA; College of Psychology, Nova Southeastern
University, Ft. Lauderdale, FL, USA; Department of Medicine, University of Alberta,
Edmonton, AB, Canada
PATRICK BURNS Bristol Veterinary School, University of Bristol, Bristol, UK
MINGJU CAO Department of Obstetrics and Gynaecology, CHU Ste-Justine Research
Centre, University of Montreal, Montreal, QC, Canada; Department of Neurosciences,
CHU Ste-Justine Research Centre, University of Montreal, Montreal, QC, Canada
JORDAN CARGILL College of Psychology, Nova Southeastern University, Ft. Lauderdale,
FL, USA
MARINA CORTES Centre de Recherche en Reproduction Animale (CRRA), University of
Montreal, St-Hyacinthe, QC, Canada
TRAVIS J. A. CRADDOCK College of Psychology, Nova Southeastern University,
Ft. Lauderdale, FL, USA; Institute for Neuro-Immune Medicine, Nova Southeastern
University, Ft. Lauderdale, FL, USA
ANDRÉ DESROCHERS Faculty of Veterinary Medicine, Department of Clinical Sciences,
University of Montreal, St-Hyacinthe, QC, Canada
LUCY FAIRCLOUGH School of Life Science, University of Nottingham, Nottingham, UK
GILLES FECTEAU Faculty of Veterinary Medicine, Department of Clinical Sciences,
University of Montreal, St-Hyacinthe, QC, Canada
GIUSEPPE DI FEDE Institute of Biological Medicine, Milan, Italy
MARY ANN FLETCHER Department of Medicine, University of Miami, Miami, FL, USA;
Institute for Neuro-Immune Medicine, Nova Southeastern University, Ft. Lauderdale, FL,
USA; Miami Veterans Affairs Medical Center, Miami, FL, USA
MARTIN G. FRASCH Department of Obstetrics and Gynaecology, University of Washington,
Seattle, WA, USA; Department of Obstetrics and Gynaecology, CHU Ste-Justine Research
Centre, University of Montreal, Montreal, QC, Canada; Department of Neurosciences,
CHU Ste-Justine Research Centre, University of Montreal, Montreal, QC, Canada; Centre
ix
x Contributors
de Recherche en Reproduction Animale (CRRA), University of Montreal, St-Hyacinthe,

QC, Canada
GREGORY G. FREUND Division of Nutritional Sciences, University of Illinois, Urbana, IL,
USA; Department of Animal Sciences, University of Illinois, Urbana, IL, USA;
Department of Pathology, Program in Integrative Immunology and Behavior, College of
Medicine, University of Illinois, Urbana, IL, USA
STEPHEN J. GAINEY Department of Animal Sciences, University of Illinois, Urbana,
IL, USA
MICHAEL GALLAGHER Department of Family Medicine, University of Calgary, Calgary,
AB, Canada
XAVIER F. GONZALEZ Department of Biomedical Sciences, Texas A&M University Corpus
Christi, Corpus Christi, TX, USA
JONAS HANNESTAD Department of Psychiatry, Yale University School of Medicine, New
Haven, CT, USA
JESSICA A. JIMÉNEZ Department of Psychology, College of Letters and Sciences, National
University, La Jolla, CA, USA
HARLAN P. JONES Department of Microbiology, Immunology and Genetics, University of
North Texas Health Science Center, Fort Wort, TX, USA
NANCY G. KLIMAS Institute for Neuro-Immune Medicine, Nova Southeastern University,
Ft. Lauderdale, FL, USA; Miami Veterans Affairs Medical Center, Miami, FL, USA
BEN KORIN Department of Immunology, Rappaport Faculty of Medicine, Technion—Israel
Institute of Technology, Haifa, Israel; Department of Neuroscience, Rappaport Faculty of
Medicine, Technion—Israel Institute of Technology, Haifa, Israel
ARIANNA LISSONI Institute of Biological Medicine, Milan, Italy
PAOLO LISSONI Institute of Biological Medicine, Milan, Italy
GIUSY MESSINA Institute of Biological Medicine, Milan, Italy
GERLINDE A. S. METZ Department of Neuroscience, Canadian Centre for Behavioural
Neuroscience, University of Lethbridge, Lethbridge, AB, Canada
PAUL J. MILLS Department of Family Medicine and Public Health, University of
California, San Diego, La Jolla, CA, USA
MARIANA MORRIS Institute for Neuro-Immune Medicine, Nova Southeastern University,
Ft. Lauderdale, FL, USA; Miami Veterans Affairs Medical Center, Miami, FL, USA
COLETTE G. NGO NDJOM Department of Microbiology, Immunology and Genetics,
University of North Texas Health Science Center, Fort Wort, TX, USA
BARRY NIERENBERG College of Psychology, Nova Southeastern University, Ft. Lauderdale,
FL, USA
CHRISTINE TARA PETERSON Department of Family Medicine and Public Health, University
of California, San Diego, La Jolla, CA, USA
GIORGIO PORRO Institute of Biological Medicine, Milan, Italy
MARK A. RICE JR College of Psychology, Nova Southeastern University, Ft. Lauderdale, FL,
USA; Center for Clinical Systems Biology, Rochester General Hospital Research Institute,
Rochester, NY, USA
ASYA ROLLS Department of Immunology, Rappaport Faculty of Medicine, Technion—Israel
Institute of Technology, Haifa, Israel; Department of Neuroscience, Rappaport Faculty of
Medicine, Technion—Israel Institute of Technology, Haifa, Israel
FRANCO ROVELLI Institute of Biological Medicine, Milan, Italy
BORIS ŠAKIĆ Department of Psychiatry and Behavioral Neurosciences, McMaster University,
Hamilton, ON, Canada
Contributors xi
ERIC C. SHATTUCK Department of Anthropology, University of Texas at San Antonio,

San Antonio, TX, USA
J. TORY TOOLE College of Psychology, Nova Southeastern University, Ft. Lauderdale,
FL, USA
ALBERT E. TOWERS Division of Nutritional Sciences, University of Illinois, Urbana,
IL, USA
ELLING ULVESTAD Department of Microbiology, Haukeland University Hospital,
Department of Clinical Science, University of Bergen, Bergen, Norway
KAVITA VEDHARA Division of Primary Care, School of Medicine, University of Nottingham,
Nottingham, UK
SUZANNE D. VERNON The Bateman Horne Center, Salt Lake City, UT, USA
LUIGI VIGORÈ Institute of Biological Medicine, Milan, Italy
ANNA C. WHITTAKER School of Sport, Exercise and Rehabilitation Sciences, University of
Birmingham, Birmingham, UK
QING YAN PharmTao, Santa Clara, CA, USA; University of Maryland University College,
Adelphi, MD, USA
JASON M. YORK School of Molecular and Cellular Biology, University of Illinois, Chicago,
IL, USA
JOEL ZYSMAN Center for Computational Science, University of Miami, Miami, FL, USA
Part I
Stress and Immunity: Biopsychosocial Models and Clinical

Implications
Chapter 1
Stress and Systemic Inflammation: Yin-Yang Dynamics

in Health and Diseases
Qing Yan
Abstract
Studies in psychoneuroimmunology (PNI) would provide better insights into the “whole mind-body
system.” Systems biology models of the complex adaptive systems (CASs), such as a conceptual framework
of “Yin-Yang dynamics,” may be helpful for identifying systems-based biomarkers and targets for more
effective prevention and treatment. The disturbances in the Yin-Yang dynamical balance may result in stress,
inflammation, and various disorders including insomnia, Alzheimer’s disease, obesity, diabetes, cardiovas-
cular diseases, skin disorders, and cancer. At the molecular and cellular levels, the imbalances in the cytokine
pathways, mitochondria networks, redox systems, and various signaling pathways may contribute to
systemic inflammation. In the nervous system, Yin and Yang may represent the dynamical associations
between the progressive and regressive processes in aging and neurodegenerative diseases. In response to
the damages to the heart, the Yin-Yang dynamical balance between proinflammatory and anti-inflammatory
cytokine networks is crucial. The studies of cancer have revealed the importance of the Yin-Yang dynamics
in the tumoricidal and tumorigenic activities of the immune system. Stress-induced neuroimmune imbal-
ances are also essential in chronic skin disorders including atopic dermatitis and psoriasis. With the
integrative framework, the restoration of the Yin-Yang dynamics can become the objective of dynamical
systems medicine.
Key words Complex adaptive systems, Dynamical medicine, Immune, Inflammation, Mind-body,
Psychoneuroimmunology, Stress, Systems biology, Systems medicine, Yin-Yang
1 Yin-Yang Dynamics in the Complex Adaptive Systems: A Conceptual Framework
Studies in psychoneuroimmunology (PNI) would provide better

insights into the “whole mind-body system” for the construction of
dynamical systems medicine [1, 2]. The revolutionary approaches
in systems biology and PNI would enable the recognition of the
disease “roots,” i.e., the dysfunctions and imbalances in the holistic
system. Such approaches may replace the reductionism-based
methods of combating diseases by killing pathogens or tumor
cells, which often lead to unfavorable results.
Systems biology models such as the conceptual framework of
“Yin-Yang dynamics” proposed in this chapter may be helpful for
Qing Yan (ed.), Psychoneuroimmunology: Methods and Protocols, Methods in Molecular Biology, vol. 1781,
https://doi.org/10.1007/978-1-4939-7828-1_1, © Springer Science+Business Media, LLC, part of Springer Nature 2018
3
4 Qing Yan
identifying systems-based biomarkers and targets for more effective

prevention and treatment. Based on such an integrative biopsycho-
social framework, the objective of personalized and systems medi-
cine is to support the normal functions of the self-regulatory
networks [3], fine-tune the dynamical balances, and restore
homeostasis [2].
To achieve the dynamical “balance of health,” understanding
the principles of complex adaptive systems (CASs) is useful for
interpreting the spatiotemporal mechanisms of health and diseases
[1, 2]. Specifically, the principle of “emergence” in CASs states that
the behaviors or pathological symptoms may be the outcomes of
the dynamical interactions among various components of the whole
system.
For example, the neuroimmune functions are influenced by the
human microbiota, especially the multidirectional communications
in the microbiota-gut-brain (MGB) axis, rather than just one or
two neurotransmitters or cytokines [2, 4]. The human microbiota
is an ecosystem that has the key role in visceral perception, drug and
carcinogen neutralization, as well as systemic inflammation
[5]. The dynamical interactions and balances in the MGB axis are
critical for the prevention and treatments of various inflammatory
diseases from depression to diabetes [4].
Such interrelationships may also illustrate the principle of
“nonlinearity” in CASs because most of the connections wouldn’t
lead to linear outcomes. A typical example is that continuous
increasing of the dosages in chemotherapy often leads to adverse
effects rather than better outcomes [1]. Chemotherapy may be
effective in controlling the tumor sizes at the early stages of treat-
ments but can induce secondary tumors with overdosages. Such
nonlinearity also indicates the transformation of the opposite fea-
tures that are often contained in one thing or process, just as the
relationships between “Yin” and “Yang.”
In addition, the characteristics of “adaptation,” “self-organiza-
tion,” and “robustness” are pivotal for living CASs in stress
responses to keep psychophysiological coherence and homeostasis,
as well as for disease recovery [1, 2]. For instance, common or
similar feedback loops of oxidation and inflammation have been
identified in a broad range of diseases from anxiety to premature
aging [6]. Preventive and therapeutic goals may concentrate on the
promotion of the self-regulation ability against the oxidative-
inflammatory activities to achieve healthy neuroimmune
capabilities.
Furthermore, the “dynamical” features of CASs may be identi-
fied as systems-based biomarkers to construct the across-scale spa-
tiotemporal profiles from cells to humans, from seconds to years
[1]. For example, the rhythmic variabilities including the heart rate
dynamics can be elucidated at different systems levels from mito-
chondrial bioenergetics to neuroimmune activities [7]. Circadian
Stress and Systemic Inflammation: Yin-Yang Dynamics in Health and Diseases 5
rhythms may affect the complex networks associated with sleep

patterns, xenobiotic metabolism, and drug detoxification
[2, 8]. Integrative preventive or therapeutic targets can be discov-
ered based on the dynamical biomarkers or profiles including the
rhythmic patterns or disturbances [9].
While these different concepts may be used to illustrate the
features of CASs, they may also make the understanding more
complicated when simplicity and unity would be needed to depict
the whole picture. These complex features of dynamical balancing,
counteracting, interconnecting, interdependent, complementing,
and transforming components in CASs can best be integrated,
represented, and interpreted by the concepts of “Yin” and “Yang.”
Specifically, the bidirectional, dynamical, and nonlinear inter-
actions between Yin and Yang components may result in the “emer-
gence” of health and diseases. The communications and
interdependence between Yin and Yang factors would enable the
“adaptation” to internal and external stressors. The counteracting
and balancing, self-adjusting, and complementing Yin-Yang ele-
ments would enable the “self-organization” and feedback loops
to restore homeostasis and balance of health. The transforming
processes would allow for the “robustness” of repair and recovery.
In addition to these spatial characteristics and activities, Yin and
Yang may also represent the dynamic and rhythmic flow for the
temporal changes such as the dark and light cycles, providing an
integrative and holistic vision across various spatiotemporal scales
(see Subheading 4). For instance, the Yin and Yang features have
been identified in sleep, wakefulness, attention, and insomnia (see
Subheading 2).
Many scientific literatures have used the concepts of “Yin” and
“Yang” to illustrate the two opposite/complementary aspects or
effects of the same/similar thing or condition. A good example is
inflammation. As a basic protective response to internal and exter-
nal stressors, inflammation has been found as the “key element” in
many pathogenic processes that may result in tissue damage, aging,
cardiovascular disease, diabetes, autoimmune diseases, and cancer
[10]. The understanding of the two roles of inflammation in health
and disease, i.e., the protective and damaging effects, is critical in
the discovery of more effective therapeutic methods for different
diseases.
At the system and psychological levels, the Yin and Yang ele-
ments have been recognized in personality. The active, easy-going,
and dynamic personalities resemble the features of Yang, while the
passive, meticulous, and static personalities have the Yin aspects
([11]; also see Table 1). In the neuroimmune system, the interde-
pendent and simultaneously counteracting actions between
dendrimer-activated monocytes and autologous natural killer
(NK) cells have been depicted with “Yin” and “Yang” features,
representing a complicated cross talk between the two cell
groups [12].
6 Qing Yan
Table 1
Yin-Yang dynamics and neuroimmune imbalances in health and diseases
Health conditions Yin-Yang factors/interactions References

Personality Active and dynamic vs. passive [11]
and static
Sleep and attention HCRT vs. MCH; excitatory/ [13]
inhibitory neuropeptides
Wake vs. sleep; sleep vs. selective [14]
attention
Sleep deprivation Pro- vs. anti-inflammatory [15]
cytokines
Insomnia, weight gain Ghrelin vs. leptin [16]
Aging, age-related disorders, neurodegenerative diseases Vascular damage vs. repair [17]
(neuroprotection vs. neurodegeneration) Progressive vs. regressive events [18]
EP2 receptor in [19]
neuroprotection
vs. neurodegeneration
Calpain-1 vs. calpain-2 [20]
The p19Arf-p53 pathway in [21]
cellular senescence/repair
Pro- vs. anti-inflammatory [22]
subsets in microglia
Trem1 vs. Trem2 regulated by [23]
LPS
Alzheimer’s disease The complement cascade in [24]
toxicity/clearance
PD Dopamine in appetitive drive [25]
vs. inhibitory control
Ocular diseases; PD VEGF vs. PEDF [26]
Frailty syndrome (FS) Redox imbalance; GSSG/GSH [27]
ratio
MetS, depression Autonomic nervous and [28]
immune imbalance
Fatty liver disease Lipogenesis vs. gluconeogenesis [29]
Obesity, MetS Energy imbalance via PPAR [30]
Type 2 diabetes ProNGF/NGF ratios [31]
Cardiac autonomic imbalance; [32]
adiponectin/leptin ratio
Insulin vs. leptin on WAT [33]
Atherosclerosis Effects of AIF-1 vs. IRT-1 on [34]
VSMC
Cardiovascular disease Mitophagy vs. cell death in stress [35]
Myocardial infarction Ischemic vs. bleeding conditions [36]
(continued)
Table 1
(continued)

Acute PPCM Pro- vs. anti-inflammatory [37]
cytokines
Stroke GABA and glutamate signaling [38]
in brain excitability
Inflammation, wound healing, tumorigenesis, cancer Growth arresting vs. growth [39, 40]
promoting
Autophagy vs. cell growth in [41]
stress
The p53 pathway in low [42]
vs. intense stress
Autoimmunity, cancer IL-21 in B cell apoptosis [43]
vs. proliferation
Colorectal neoplasia Oxidative and immune [44]
imbalances
GBM Redox imbalances in astrocytes, [45]
microglia
Stress, skin diseases Th1 vs. Th2 immune responses [47]
Stress-induced immune [46, 48]
imbalances
NGF in stress and immune [49]
imbalances
Psoriasis miR-31/miR-203 vs. hsa-miR- [50]
99a/miR-125b
In the following sections, the Yin and Yang characteristics will

be analyzed in various disorders to illustrate the dynamical mechan-
isms of the “balance of health” and the imbalances in diseases,
especially their roles in stress responses and inflammation. These
examples are summarized in Table 1. More examples in psychiatric
disorders will be discussed in Chapter 5. Many of these systems-
based features have been considered as the potential biomarkers for
more precise diagnosis, prevention, and treatment.
2 The Yin and Yang in Sleep and Insomnia
Sleep is essential for health. Sleep problems such as short sleep and
sleep disturbances have been closely associated with cellular altera-
tions and aging-associated diseases [4]. Inflammation has been
identified as the biological link between sleep problems and the
elevated risks of depression, pain, and infectious diseases. In addi-
tion, the connections between chronic stress and obesity may
8 Qing Yan
disturb the temporal patterns of the elements including leptin and

triglycerides, leading to the comorbidity of various diseases from
diabetes to cardiovascular diseases [4].
The Yin and Yang patterns have been used to describe wakeful-
ness and sleep (see Table 1). For instance, the hypocretin (HCRT)
and melanin-concentrating hormone (MCH) have been considered
a pair with Yin-Yang features [13]. The excitatory versus inhibitory
neuropeptides have been detected at the lateral hypothalamic level
and in the activating and hypnogenic neuronal regions. High fre-
quency in the neural activity is a feature of the waken condition
[14]. Sleep has been characterized with slow-wave oscillations
including prolonged up states and down states of neural activities.
In addition, sleep and selective attention have been found with
the Yin-Yang effects on synaptic plasticity, with the features of
opposing, complementary, regulating, and supporting each other
[14]. Studies have suggested that sleep is critical in cognitive func-
tions including learning and attention. Even though sleep and
attention are very different physiological conditions, they both
have the feature of inhibiting external stimuli with similar mechan-
isms. Sleep has the pivotal role in reaching the optimal levels of
attention, while attentional requests are important for the regula-
tion of sleep [14].
Sleep deprivation and circadian misalignment may affect the
levels of cortisol and the Yin-Yang balance between proinflamma-
tory and anti-inflammatory cytokine biomarkers [15]. Studies have
found that circadian misalignment may lead to elevated levels of
plasma tumor necrosis factor-alpha (TNF-α), interleukin
10 (IL-10), and C-reactive protein (CRP). Such imbalances may
be meaningful as potential biomarkers for stress responses and
associated diseases.
For instance, chronic insomnia has been closely associated with
weight gain. Among insomnia patients, lower levels of ghrelin
across the night have been observed [16]. The levels of the two
essential hormones ghrelin and leptin are critical in the mainte-
nance of the energy balance. They are involved in the regulation
of appetite and body weight. The alteration of these hormones may
lead to the energy imbalance and rhythmic disturbance among
insomnia patients and result in weight gain (see Subheading 4).
3 Yin-Yang Dynamics in Aging and Neurodegenerative

Diseases at Various Systems Levels
The neuroendocrine-immune plasticity enables stress responses and

adaptation [4]. Chronic stress has been correlated with immune
imbalances, premature aging, and aging-associated disorders. Ele-
vated levels of proinflammatory cytokines in the brain may impair
the neuronal plasticity, leading to neurobehavioral problems and

cognitive deficits including depression and neurodegenerative dis-
eases. The inflammatory, synaptic, and neurotrophic networks are
pivotal in aging and associated diseases [4].
The framework of Yin-Yang dynamics based on systems biology
and CAS principles may be helpful for understanding these com-
plex mechanisms in aging. For example, the cellular and molecular
alterations are critical in the endogenous regeneration of the vessel
wall, such as the changes in the inflammatory cells and circulating
precursors. The Yin-Yang imbalances between vascular damage and
repair have been considered as the biomarkers and therapeutic
targets for age-related comorbidities including diabetes, atheroscle-
rosis, neurodegenerative diseases, and cancer [17].
Activities in the nervous system have shown the Yin-Yang
features at molecular, cellular, and circuit levels. Yin and Yang
may well represent the dynamical associations between the progres-
sive and regressive processes in neural development and diseases
[18]. Many signaling networks in the neural development and
maintenance have been associated with the balance between neural
degeneration and regeneration, including the activities in neural
wiring/neurodegeneration and axon attraction/repulsion.
For example, at the molecular level, the prostaglandin receptor
EP2 in the brain has been related to the functions of anti-
inflammation, anticancer, and neuroprotection. The ultimate func-
tions of EP2 receptor have been suggested to be decided by the
Yin-Yang balance between neurons and glia ([19]; also see Table 1).
EP2 receptors in neurons may enhance the cAMP/PKA-
associated neuroprotection [19]. However, the glial EP2 activation
has been related to neurotoxicity and neurodegeneration in associ-
ation with cAMP/Epac signaling-mediated increase of proinflam-
matory cytokines. Therefore, the net effect of EP2 activation may
be decided by the spatiotemporal regulation from the correlated
molecular networks in the Yin-Yang manner [19].
Various signaling networks are involved in both synaptic plas-
ticity and neuronal degeneration. Examples include the two calpain
isoforms in the brain, calpain-1 and calpain-2. Calpain-1 and
calpain-2 have been found with the opposite roles in synaptic
plasticity and neurodegeneration in association with different
downstream signaling networks ([20]; also see Table 1). Specifically,
calpain-1 activation is involved in the long-term potentiation (LTP)
and considered neuroprotective. However, calpain-2 activation
may inhibit the extent of potentiation and considered
neurodegenerative.
The Yin-Yang dynamics has been identified in calpain-1 and
calpain-2 with their counteractive roles in synaptic plasticity,
learning and memory, as well as balance between neuroprotection
and neurodegeneration. The understanding of the Yin-Yang rela-
tionships between calpain-1 and -2 may enable the design of novel
10 Qing Yan
treatment strategies for a broad range of diseases related to learning

impairment and neurodegeneration [20]. For example, calpain-1
activators may be applied to suppress calpain-2 to promote learning
and neuroprotection.
In addition, the p19Arf-p53 pathway has been found with the
Yin-Yang roles in aging ([21]; also see Table 1). The p53-associated
networks are critical in stress responses as acute stress may trigger
the network toward apoptosis or cellular senescence for early aging.
However, chronic stress may stimulate the p53 network for a
temporary cell cycle arrest to repair damages and promote cell
functions.
Studies using mice models have shown that the Cdkn2a locus
of p16Ink4a and p19Arf may also have the Yin-Yang effects on each
other as p19Arf may counterbalance the pro-aging activities of
p16Ink4a [21]. When p19Arf is not available to counteract
p16Ink4a, the higher levels of p16Ink4a may lead to the
pro-aging microenvironment. However, when p16Ink4a is not
available, p19Arf may not accumulate, referring to their Yin-Yang
interdependent relationship.
At the cellular level, the Yin-Yang features have been recog-
nized in microglia as their beneficial or damaging effects are
decided by the factors such as stress signals, timing of the stimulus,
microenvironment, as well as communications with other cells
[22]. For instance, two subsets have been identified in macro-
phages/microglia by their different molecular phenotypes, effector
functions, and activation pathways ([22]; also see Table 1). The
“classically activated” M1 macrophages are proinflammatory.
They can generate oxidative metabolites and proinflammatory
cytokines to defend against pathogens and tumor cells. The “alter-
natively activated” M2 macrophages are anti-inflammatory and may
induce tissue repair and angiogenesis.
In the regulation of myeloid cell inflammation, the triggering
receptor expressed on myeloid cells (TREM) family of proteins has
important roles [23]. Different TREM receptors have different
functions in the myeloid immune processes. Lipopolysaccharide
(LPS) may regulate microglial Trem1 and Trem2 gene expressions
in an opposite way by promoting Trem1 and inhibiting Trem2
expressions. Neuroinflammatory activities have been found to
affect the Yin-Yang balance in TREM expressions and homeostasis,
and contribute to neurodegenerative disorders ([23]; also see
Table 1).
Therapeutic strategies may be designed for neurodegenerative
disorders based on the mechanisms of such Yin-Yang dynamics. For
example, in Alzheimer’s disease, the functions of the complement
cascade have been labeled with Yin and Yang characteristics ([24];
also see Table 1). Complement activations may lead to not just
inflammation and tissue damage, but also the removal of cell debris
and toxic proteins. The dynamical balances in these events may
influence the neuronal functions, while the imbalances between the

toxicity and clearance may lead to the neurodegenerative pathogen-
esis. Such features may be important for the discovery of more
effective treatment strategies by restoring the dynamical balances.
In Parkinson’s disease (PD), the Yin-Yang model has been
proposed in the appetitive drive and inhibitory regulation in the
impulse control disorders ([25]; also see Table 1). Among the PD
patients, dopaminergic depletion may reduce the tonic
D2-receptor stimulation in the regions of ventral striatum/NAcc.
However, among susceptible patients, a constitutionally higher
level of tonic dopamine may result in the relatively normal levels
of tonic D2-receptor stimulation in the same region. Such mechan-
isms may be meaningful for treatment purposes via the regulation
of the appetitive drive and inhibitory regions [25].
In addition, vascular endothelial growth factor (VEGF) and
pigment epithelium-derived factor (PEDF) have been found with
multiple roles in the pathogenesis of many disorders, including
Parkinson’s disease and cancer [26]. Recent studies have identified
the Yin-Yang dynamics in their complex cross talk in various net-
works such as the PI3K/Akt signaling pathway. For example, the
balance between the pro-angiogenic factor VEGF-A and anti-
angiogenic factor PEDF has been associated with the survival of
retinal neurons in ocular disorders such as macular edema and
macular degeneration ([26]; also see Table 1).
However, although PEDF and VEGF-A may have the counter-
active functions in vascularization in the peripheral tissues, they may
share neuroprotective effects in a reciprocal way in the brain. The
understanding of such Yin-Yang associations between VEGF and
PEDF may contribute to the treatments of different diseases espe-
cially neurodegenerative disorders such as Parkinson’s disease [26].
Furthermore, frailty syndrome (FS) is an aging-associated
problem featured with stress-vulnerability physiological weakness.
Studies have indicated the important roles of oxidative stress and
glutathione imbalance in FS ([27]; also see Table 1). For example, a
study of 62 elderly outpatients observed the altered glutathione
(GSH) with elevated levels of the oxidized glutathione (GSSG),
TNF-α, and 4-hydroxy-2,3-nonenal (HNE) protein plasma
adducts. The oxidative imbalances, especially the altered GSSG/
GSH ratio and plasma protein adducts that may predict the frailty
status, have been suggested as the reliable and useful biomarkers for
FS among elderly patients [27].
4 Yin-Yang Dynamics in Obesity, Diabetes, and Metabolic Syndrome
PNI evidences have suggested inflammation as the transitional

connection between psychosocial stress and obesity [4]. Psychoso-
cial stress may interact with genes, diets, and lifestyle to affect
12 Qing Yan
weight gain. The maladaptation to chronic stress may influence

energy intake and expenditure with the stimulation of appetite
and the reduction of physical activity. The imbalances and dysfunc-
tions in the hypothalamic-pituitary-adrenal (HPA) axis and asso-
ciated cellular networks have been correlated with upper body
obesity and sleep disturbance in metabolic disorders [4].
Specifically, psychological distress including perceived stress
and depression has been closely associated with metabolic syn-
drome (MetS) showing high blood pressure, high blood glucose,
dyslipidemia, and abdominal obesity [28]. An important mecha-
nism underlying such connections is the autonomic nervous imbal-
ance featured with lower heart rate variability (HRV). In addition,
higher levels of inflammatory biomarkers such as CRP indicate the
immune imbalances.
The rhythmic flow of metabolic intermediates is essential in
maintaining the Yin-Yang dynamical balance and interconnection
between lipid and glucose generations in the liver ([29]; also see
Table 1). During the light cycle, carbohydrate catabolism may
assimilate the intermediates into lipids. During the dark cycle,
lipid oxidation may turn the intermediates to glucose generation.
Excessive lipogenesis may suppress gluconeogenesis, while the
inhibition of lipid oxidation may also block gluconeogenesis
[29]. Normally, the complex interactions between hepatic lipid
and carbohydrate metabolism flow in the natural cycle toward the
balance in a rhythmic pattern. The disturbance of the Yin-Yang
dynamical balance may result in the conditions including fatty liver
disease [29].
An important feature of obesity is the energy imbalance in
relation to chronic inflammation and fat tissues. The profiles of
obese patients may address the vascular complications as the fat
tissues have the critical roles in energy imbalance associated with
the peroxisome proliferator-activated receptor (PPAR) and inflam-
matory pathways ([30]; also see Table 1). The molecules involved in
the white adipose tissue (WAT) activities include interleukin-6
(IL-6), TNF-α, and adiponectin.
Studies have shown that glycine may possess anti-inflammatory
activities to decrease the levels of TNF-α and IL-6 [30]. Glycine
may have the regulatory roles in energy balance and affect the
inflammatory cytokines via PPAR-γ. As a potential anti-
inflammatory agent, the glycine signaling networks have been sug-
gested useful for the adjustment of energy and inflammatory imbal-
ances related to obesity and vascular disorders [30].
In diabetes, the disturbance of the homeostasis of retinal nerve
growth factor (NGF) may be caused by excessive oxidative stress in
the disease. This condition may lead to the accumulation of the
precursor proNGF and lower levels of NGF in correlation with
neuronal and retinal dysfunctions. Such imbalance of the
ProNGF/NGF ratios has been suggested as a potential biomarker
for the early diagnosis of diabetic complications especially diabetic

retinopathy ([31]; also see Table 1).
In the early stages of type 2 diabetes, cardiac autonomic imbal-
ance has been related to the biomarkers of adipose tissue-derived
inflammation ([32]; also see Table 1). The imbalance-associated
biomarkers may include lower heart rate variability, elevated levels
of the proinflammatory cytokine IL-6, as well as alterations in the
adiponectin-to-leptin ratio. IL-6 has the negative associations with
the measures of autonomic balance.
In addition, insulin and leptin are two key hormonal adiposity
signals in the regulation of energy homeostasis by the central
nervous system (CNS). The two hormones have been found essen-
tial with the opposite Yin-Yang roles in the regulation of WAT
lipolysis and de novo lipogenesis by interacting with the mediobasal
hypothalamus (MBH) [33]. MBH leptin may reduce de novo
lipogenic protein expression, while MBH insulin may promote de
novo lipogenesis and suppress WAT lipolysis. The promotion or
suppression of the sympathetic outflow to WAT may be critical in
these processes. The imbalances in the hypothalamic regulation of
WAT metabolism have been related to inflammation, insulin resis-
tance, and type 2 diabetes ([33]; also see Table 1).
5 Stress, Yin-Yang Imbalances, and Cardiovascular Diseases
PNI studies have revealed the important impacts of psychophysio-

logical stress on the immune system and coronary artery disease
(CAD) such as the endothelial dysfunctions and chemotaxis
[4]. The neurohormonal and cytokine activities are critical in the
comorbid disorders of cardiovascular and psychiatric diseases. The
cytokine imbalances have been related to coronary atherosclerosis.
PNI studies would help identify systems-based biomarkers for car-
diovascular diseases and relevant problems.
For example, in atherosclerosis, the Yin-Yang roles have been
identified between allograft inflammatory factor-1 (AIF-1) and
interferon-responsive transcript-1 (IRT-1) on migration and pro-
liferation of human vascular smooth muscle cell (VSMC) ([34];
also see Table 1). The higher levels of AIF-1 may trigger VSMC
migration and proliferation, while IRT-1 has the opposite func-
tions. The expression of AIF-1 mRNA in human carotid plaques
may result in a more proinflammatory plasma profile for plaque
rupture. However, the expression of IRT-1 mRNA has been corre-
lated with a less aggressive phenotype and less VSMCs at the
plaque [34].
In addition, mitochondria and mitophagy have the pivotal roles
in the adaptation to stress, cardiovascular health, and diseases ([35];
also see Table 1). Mitochondria are essential in maintaining the
balance between life and death under stress conditions. Mild stress
14 Qing Yan
may lead to the damages in some mitochondria that can be seques-

tered by autophagosomes. However, severe stress can result in
serious mitochondrial damages that cannot be removed by autop-
hagosomes. Such damaged mitochondria may generate excessive
reactive oxygen species (ROS) and pro-death proteins including
cytochrome c that are associated with cell death pathways [35].
Stress responses to conditions including ischemia and reperfu-
sion may lead to pro-survival or pro-death pathways with a delicate
Yin-Yang balance [35]. The final results are decided by the cross
talk among these networks. Mitophagy may provide early cardio-
protection to keep cellular homeostasis for the adaptation to stress
conditions with the elimination of dysfunctional mitochondria. On
the other hand, higher levels of oxidative stress and apoptotic
proteases may inhibit mitophagy and move toward cell death.
These responses may cause the loss of cardiac myocytes and the
occurrence of heart failure [35].
In the treatment of myocardial infarction, a balance is needed
between the prevention of ischemic and bleeding problems ([36];
also see Table 1). Although the two events seem opposite, they are
also interdependent with the features of Yin and Yang.
In acute peripartum cardiomyopathy (PPCM), the Yin-Yang
dynamical balance between proinflammatory and anti-
inflammatory cytokines is crucial in response to the damages to
the heart ([37]; also see Table 1). Higher levels of the inflammatory
biomarker CRP have been observed together with the proinflam-
matory cytokine TNF. Studies have suggested that the cytokine
imbalances in PPCM may be important therapeutic targets. The
restoration of the cytokine balances and the timing of interventions
may be potential solutions to the disorder [37].
The features of Yin and Yang have also been observed in the
brain excitability in stroke ([38]; also see Table 1). During the early
stages of stroke, the high levels of excitability are deleterious.
However, during the later stages of recovery, the similar signaling
systems may turn to be beneficial. These opposite and contradic-
tory stages are connected via the gamma-aminobutyric acid
(GABA) and glutamate signaling networks [38]. The accurate
understanding of the Yin-Yang dynamics in the opposite and com-
plementary interactions in the signaling networks during different
stages of stroke progression may be critical for personalized preven-
tion and treatment.
6 Inflammation and Yin-Yang Imbalances in Cancer
More and more evidences have emphasized the important roles of

psycho-oncology such as cognitive deficits in cancer [4]. Stress and
depression are often experienced among cancer patients. Better
understanding of the interactions among perceived stress, HPA
axis, and inflammatory networks would contribute to the discovery
of more effective prevention and therapies in dynamical systems

medicine for cancer.
The Yin-Yang features have been used to describe the growth-
arresting versus growth-promoting processes and the tumoricidal
and tumorigenic activities in inflammation and cancer ([39, 40];
also see Table 1). The Yin-Yang dynamics may have pivotal roles in
vasculature responses and factors of apoptosis and wound healing.
Representing the Yin characteristics, the apoptosis and growth-
arrest responses from the immune cells may be activated to destroy
external stimuli and damaged tissues. The Yang features have been
suggested to characterize the “growth factors” in wound healing
and recovery [39, 40].
Oxidative stress may alter the Yin-Yang balance of immune
functions toward tumorigenesis and angiogenesis [40]. The Yin
and Yang features in the inflammatory activities have also been
related to the cross talk between immune and nonimmune systems
such as the neuroendocrine and metabolic pathways in the pro-
cesses of removing foreign stimuli, ending inflammation, and
repairing the wounded tissues.
The understanding of the Yin-Yang dynamical balances and
interactions between cell growth and autophagy may be necessary
for the treatment of cancer with their roles in tumorigenesis and
suppression effects on each other [41]. In the conditions of cell
stresses such as nutrient deprivation, autophagy may be triggered
with lower cell growth. The mammalian target of rapamycin
(mTOR) pathway has been associated with the processes in both
autophagy and cell growth ([41]; also see Table 1).
Another example acting the Yin-Yang roles is the p53 pathway.
Stress stimulation may activate p53 and its interactions with other
genes, leading to cell protection or cell death decided by the
intensity of the stress ([42]; also see Table 1). Such responses are
necessary for the protection against carcinogenesis. Intense stress
may activate p53-associated networks with pro-apoptotic and
prooxidant genes toward cell death or senescence. However, low
or moderate levels of stress may activate p53-associated pro-survival
networks to protect cells from damages, such as those related to
ROS suppression, mitochondrial function, and autophagy for cell
viability.
In the studies of cancer, allergy, and autoimmunity, IL-21
signaling may have the Yin-Yang effects on the naive B cells
depending on different conditions ([43]; also see Table 1). IL-21
may cause the apoptosis of naive B cells with the lack of the signal to
the B cell antigen receptor but with pathogen-derived signal to toll-
like receptors (TLR). When triggered by the B cell antigen receptor
signal and IL-4, IL-21 may promote the proliferation of naive B
cells and plasma cell differentiation.
In colorectal neoplasia, the combined measures of pro- and
antioxidant exposures using an oxidative balance score (OBS)
16 Qing Yan
have been found meaningful as the potential biomarkers for oxida-

tive stress and inflammation [44]. In glioblastoma multiforme
(GBM), chronic inflammation has been observed in the brain tis-
sues, in association with the higher levels of oxidative stress in
astrocytes and microglia, as well as the dysfunctions of DNA repair
enzymes ([45]; also see Table 1).
The imbalances in the immune functions, redox status, meta-
bolic activities, as well as mitochondrial DNA may contribute to the
prooxidant and proinflammatory environment for tumor cell pro-
liferation and immune escape [44, 45]. Systems-based therapeutics
may target such mechanisms by restoring the Yin-Yang dynamical
balances in these networks.
7 Stress, Inflammation, and Yin-Yang Imbalances in Skin Diseases
Stressful life events have been closely correlated with the onset of
inflammatory skin diseases [4]. Depression is common among
patients with dermatological problems. PNI evidences have
revealed the stress-induced neuroimmune imbalances in chronic
skin disorders including atopic dermatitis, psoriasis, malignant mel-
anoma, scleroderma, lichen sclerosus, and eosinophilic fasciitis
[4, 46].
The dynamical balances in the psycho-neuro-immuno-endo-
crine-cutaneous networks rely on the interactions among various
signaling pathways of neuropeptides, hormones, and immune mes-
sengers. For example, stress-induced alterations in the CNS may
disturb the balance between cell-mediated (Th1) and humoral
(Th2) immune functions, leading to the development of various
skin disorders ([47]; also see Table 1).
In itches and other allergic reactions, environmental factors
including allergens and psychosocial stress may stimulate the gen-
eration of neuropeptides and activate mast cells, leading to the
stress responses from peripheral organs including the skin
[48]. The Yin-Yang dynamical balances in neuropeptides and neu-
rotrophins may affect the promotion or inhibition of tissue regen-
eration and inflammation.
For instance, among patients with allergic dermatitis, perceived
stress may lead to higher levels of the neurotrophin nerve growth
factor (NGF), leading to elevated levels of proallergic cytokines
([49]; also see Table 1). Such immune imbalance may result in
cutaneous inflammation.
In the heterogeneity disease of psoriasis, the imbalanced
miRNA axis has been identified between miR-31/miR-203 and
hsa-miR-99a/miR-125b with the Yin-Yang features of opposite
yet complementary, interconnected, and interdependent correla-
tions [50]. The higher levels of the pair of miR-31/miR-203 and
lower levels of the pair of hsa-miR-99a/miR-125b may be involved
in the regulation of proliferation and differentiation in psoriatic

keratinocytes. The imbalanced miRNA axis has also been related
to the inflammatory activities in the psoriatic lesions. Such
Yin-Yang axis can be potential systems-based biomarkers and treat-
ment targets [50].
8 Conclusion
In conclusion, PNI and systems biology evidences as discussed

above support a conceptual framework of “Yin-Yang dynamics” in
health, diseases, and wellness. The imbalanced conditions in the
Yin-Yang relationships may result in stress, inflammation, and vari-
ous disorders including insomnia, Alzheimer’s disease, obesity,
diabetes, cardiovascular diseases, skin disorders, and cancer.
As summarized in Fig. 1, the Yin-Yang dynamical balances are
essential at various systems levels. At the environment level, the
Yin-Yang factors in the light/dark and seasonal cycles may interact
with those at other levels including genes, neuroimmune functions,
and sleep patterns. The disturbances of the Yin-Yang dynamics or
Environmental Factors Light/Dark Cycles

Seasonal Cycles
Life Style and
Psychological Personality
Status Emotions Anxiety
Sleep Patterns Imbalances Depression
Yin-Yang
Stress Responses Insomnia
Dynamics
Nutritional Status
Interactions Neuroimmune Functions Obesity

at System Levels HPA Axis Diabetes
MGB Axis Imbalances Heart Disease
Metabolism Cancer
Yin-Yang
Dynamics
Skin Diseases
Cytokine Pathways
Neurotransmitters
Interactions Hormones
at Molecular/ Metabolic Pathways
Cellular Levels Redox Systems Imbalances Systemic
Yin-Yang
Epigenetics Inflammation
Dynamics
Mitochondrial Functions
Circadian Networks
Fig. 1 The conceptual framework of “Yin-Yang dynamics” in health and diseases at various systems levels
18 Qing Yan
the imbalances in lifestyle and psychological conditions, such as

altered stress responses and malnutrition, may lead to problems
including anxiety, depression, and insomnia.
At the system levels, the imbalances and dysfunctions in the
neuroimmune systems, HPA axis, MGB axis, and metabolic systems
may result in obesity, diabetes, heart disease, skin diseases, and
cancer (see Fig. 1). At the molecular and cellular levels, the imbal-
ances in the cytokine pathways, mitochondrial functions, redox
systems, and various signaling networks may contribute to systemic
inflammation.
In addition, the imbalances and alterations at each spatial level
or temporal scale may interact with and affect the conditions at
other levels, forming various feedback and feedforward loops. For
example, systemic inflammation and relevant pathways are the piv-
otal mechanisms in a broad range of disorders from depression to
diabetes (see Fig. 1).
These imbalances and alterations are also the candidates for
systems-based biomarkers and therapeutic targets. With such an
integrative framework, the restoration of the Yin-Yang dynamical
balances can become the primary objective of dynamical systems
medicine [1, 2].
References
1. Yan Q (2014) From pharmacogenomics and 8. Selfridge JM, Gotoh T, Schiffhauer S et al
systems biology to personalized care: a frame- (2016) Chronotherapy: intuitive, sound, foun-
work of systems and dynamical medicine. ded. . .but not broadly applied. Drugs
Methods Mol Biol 1175:3–17 76:1507–1521
2. Yan Q (2017) Toward dynamical systems med- 9. Yan Q (2012) The role of psychoneuroimmu-
icine: personalized and preventive strategies. nology in personalized and systems medicine.
Pers Med 14:377–380 Methods Mol Biol 934:3–19
3. Goldman AW, Burmeister Y, Cesnulevicius K 10. Blackman MA, Yates JL, Spencer CM et al
et al (2015) Bioregulatory systems medicine: (2014) The Yin and Yang of inflammation.
an innovative approach to integrating the sci- Curr Mol Med 14:1238–1243
ence of molecular networks, inflammation, and 11. Lee SJ, Park SH, Chae H (2016) Biopsycho-
systems biology with the patient’s autoregula- logical structure of Yin-Yang using Cloninger’s
tory capacity? Front Physiol 6:225 temperament model and Carver and White’s
4. Yan Q (2016) Psychoneuroimmunology: sys- BIS/BAS scale. PeerJ 4:e2021
tems biology approaches to mind-body medi- 12. Poupot M, Turrin C-O, Caminade A-M et al
cine. Springer, New York (2016) Poly(phosphorhydrazone) dendrimers:
5. Elisei C, de Castro AP (2017) Insight into role Yin and Yang of monocyte activation for
of microbiota-gut-brain peptides as a target for human NK cell amplification applied to immu-
biotechnology innovations. Front Biosci (Elite notherapy against multiple myeloma. Nanome-
Ed) 9:76–88 dicine 12:2321–2330
6. Vida C, González EM, De la Fuente M (2014) 13. Torterolo P, Chase MH (2014) The hypocre-
Increase of oxidation and inflammation in ner- tins (orexins) mediate the “phasic” compo-
vous and immune systems with aging and anxi- nents of REM sleep: a new hypothesis. Sleep
ety. Curr Pharm Des 20:4656–4678 Sci 7:19–29
7. Sturmberg JP, Bennett JM, Picard M, Seely 14. Kirszenblat L, van Swinderen B (2015) The Yin
AJE (2015) The trajectory of life. Decreasing and Yang of sleep and attention. Trends Neu-
physiological network complexity through rosci 38:776–786
changing fractal patterns. Front Physiol 6:169
15. Wright KP, Drake AL, Frey DJ et al (2015) 29. Sun Z, Lazar MA (2013) Dissociating fatty
Influence of sleep deprivation and circadian liver and diabetes. Trends Endocrinol Metab
misalignment on cortisol, inflammatory mar- 24:4–12
kers, and cytokine balance. Brain Behav 30. Almanza-Perez JC, Alarcon-Aguilar FJ,
Immun 47:24–34 Blancas-Flores G et al (2010) Glycine regulates
16. Motivala SJ, Tomiyama AJ, Ziegler M et al inflammatory markers modifying the energetic
(2009) Nocturnal levels of ghrelin and leptin balance through PPAR and UCP-2. Biomed
and sleep in chronic insomnia. Psychoneuroen- Pharmacother 64:534–540
docrinology 34:540–545 31. Mysona BA, Matragoon S, Stephens M et al
17. Madonna R, Novo G, Balistreri CR (2016) (2015) Imbalance of the nerve growth factor
Cellular and molecular basis of the imbalance and its precursor as a potential biomarker for
between vascular damage and repair in ageing diabetic retinopathy. Biomed Res Int
and age-related diseases: as biomarkers and tar- 2015:571456
gets for new treatments. Mech Ageing Dev 32. Lieb DC, Parson HK, Mamikunian G, Vinik AI
159:22–30 (2012) Cardiac autonomic imbalance in newly
18. Bai G, Pfaff SL (2011) Protease regulation: the diagnosed and established diabetes is associated
Yin and Yang of neural development and dis- with markers of adipose tissue inflammation.
ease. Neuron 72:9–21 Exp Diabetes Res 2012:878760
19. Jiang J, Dingledine R (2013) Prostaglandin 33. Scherer T, Buettner C (2011) Yin and Yang of
receptor EP2 in the crosshairs of anti- hypothalamic insulin and leptin signaling in
inflammation, anti-cancer, and neuroprotec- regulating white adipose tissue metabolism.
tion. Trends Pharmacol Sci 34:413–423 Rev Endocr Metab Disord 12:235–243
20. Baudry M, Bi X (2016) Calpain-1 and calpain- 34. Berglund LM, Kotova O, Osmark P et al
2: the Yin and Yang of synaptic plasticity and (2012) NFAT regulates the expression of
neurodegeneration. Trends Neurosci AIF-1 and IRT-1: Yin and Yang splice variants
39:235–245 of neointima formation and atherosclerosis.
21. Baker DJ, Jin F, van Deursen JM (2008) The Cardiovasc Res 93:414–423
Yin and Yang of the Cdkn2a locus in senes- 35. Kubli DA, Gustafsson ÅB (2012) Mitochon-
cence and aging. Cell Cycle 7:2795–2802 dria and Mitophagy: the Yin and Yang of cell
22. Czeh M, Gressens P, Kaindl AM (2011) The death control. Circ Res 111:1208–1221
Yin and Yang of microglia. Dev Neurosci 36. Brar SS (2016) The Yin and Yang of long-term
33:199–209 dual antiplatelet therapy. J Am Coll Cardiol
23. Owens R, Grabert K, Davies CL et al (2017) 67:1155–1157
Divergent neuroinflammatory regulation of 37. Fett JD, Ansari AA (2010) Inflammatory mar-
microglial TREM expression and involvement kers and cytokines in peripartum cardiomyopa-
of NF-κB. Front Cell Neurosci 11:56 thy: a delicate balance. Expert Opin Ther
24. Shen Y, Meri S (2003) Yin and Yang: comple- Targets 14:895–898
ment activation and regulation in Alzheimer’s 38. Carmichael ST (2012) Brain excitability in
disease. Prog Neurobiol 70:463–472 stroke. Arch Neurol 69:161–167
25. Cilia R, van Eimeren T (2011) Impulse control 39. Thompson PA, Khatami M, Baglole CJ et al
disorders in Parkinson’s disease: seeking a (2015) Environmental immune disruptors,
roadmap toward a better understanding. inflammation and cancer risk. Carcinogenesis
Brain Struct Funct 216:289–299 36:S232–S253
26. Falk T, Gonzalez RT, Sherman SJ (2010) The 40. Khatami M (2014) Chronic inflammation: syn-
Yin and Yang of VEGF and PEDF: multiface- ergistic interactions of recruiting macrophages
ted neurotrophic factors and their potential in (TAMs) and Eosinophils (Eos) with host mast
the treatment of Parkinson’s disease. Int J Mol cells (MCs) and tumorigenesis in CALTs.
Sci 11:2875–2900 M-CSF, suitable biomarker for cancer diagno-
27. Serviddio G, Romano AD, Greco A et al sis! Cancers (Basel) 6:297–322
(2009) Frailty syndrome is associated with 41. Neufeld TP (2012) Autophagy and cell
altered circulating redox balance and increased growth—the Yin and Yang of nutrient
markers of oxidative stress. Int J Immuno- responses. J Cell Sci 125:2359–2368
pathol Pharmacol 22:819–827 42. Budanov AV (2014) The role of tumor sup-
28. Chen Y-J, Lin C-L, Li C-R et al (2016) Asso- pressor p53 in the antioxidant defense and
ciations among integrated psychoneuroimmu- metabolism. Subcell Biochem 85:337–358
nological factors and metabolic syndrome. 43. Spolski R, Leonard WJ (2008) The Yin and
Psychoneuroendocrinology 74:342–349 Yang of interleukin-21 in allergy,
20 Qing Yan
autoimmunity and cancer. Curr Opin Immunol 47. Tausk F, Elenkov I, Moynihan J (2008)
20:295–301 Psychoneuroimmunology. Dermatol Ther
44. Kong SYJ, Bostick RM, Flanders WD et al 21:22–31
(2014) Oxidative balance score, colorectal ade- 48. Peters EMJ (2012) The neuroendocrine-
noma, and markers of oxidative stress and immune connection regulates chronic inflam-
inflammation. Cancer Epidemiol Biomark matory disease in allergy. Chem Immunol
Prev 23:545–554 Allergy 98:240–252
45. Salazar-Ramiro A, Ramı́rez-Ortega D, Pérez 49. Peters EMJ, Liezmann C, Spatz K et al (2011)
de la Cruz V et al (2016) Role of redox status Nerve growth factor partially recovers inflamed
in development of Glioblastoma. Front Immu- skin from stress-induced worsening in allergic
nol 7:156 inflammation. J Invest Dermatol 131:735–743
46. Peters EMJ (2016) Stressed skin?—a molecular 50. Wang M-J, Xu Y-Y, Huang R-Y et al (2016)
psychosomatic update on stress-causes and Role of an imbalanced miRNAs axis in patho-
effects in dermatologic diseases. J Dtsch Der- genesis of psoriasis: novel perspectives based on
matol Ges 14:233–252; quiz 253 review of the literature. Oncotarget
8:5498–5507
Chapter 2
Intersections Between Neuroimmune and Microbiota

Colette G. Ngo Ndjom, Xavier F. Gonzalez, and Harlan P. Jones
Abstract
Multidiscipline-based research holds promise toward revealing complex mechanisms that determine health
and disease. For decades, scientists have conducted studies defining the relationships between neuroendo-
crine and immune function culminating into the discipline of psychoneuroimmunology (PNI). In addition,
the discipline of microbial endocrinology has similarly enhanced our understanding of disease processes.
With an increase in genetic-based sequencing technologies, the convergence of neuroendocrine-
immunological-microbial research is expected to significantly further such knowledge needed for medical
discoveries. In this chapter, we provide a review of the current findings that support the conceptual
framework linking microbiota, immunity, and neuroendocrine disciplines.
Key words Microbiota, Neuroendocrine, Stress, Immunity, Health
1 Introduction
The integration of biologic systems while complex is believed to

enhance our understanding of health and disease. Psychoneuroim-
munology (PNI) is one example in which defining the mechanistic
intersections between the neuroendocrine and immune systems has
led to increased knowledge of the etiology of numerous diseases
[1–3]. Over the past three decades, the PNI literature has grown,
elevating its impact within the biomedical research community as
an established discipline (Fig. 1).
In our previous commentary “Immune Cells Listen to What
Stress Is Saying,” we provided an overview defining the bidirec-
tional pathways between immune and central nervous systems.
Since then, a third intersection of disciplines that promises to
bridge new territories for medical discovery has been proposed,
and includes the neuroendocrine-immune and microbial systems
[4, 5]. In this chapter, we present an integrated perspective on
neuroendocrine, immune, and microbiologic systems as a basis for
fostering interdisciplinary approaches to combat disease.
21
22 Colette G. Ngo Ndjom et al.
Psychoneuroimmunology (596 Total Citations)

40
30
Citations per year
20
10
17
19 5
19 6
87
19 8
89
19 0
19 1
19 2
19 3
94
19 5
19 6
19 7
19 8
20 9
20 0
01
20 2
03
20 4
20 5
06
20 7
08
20 9
20 0
20 1
20 2
13
20 4
20 5
20 6
8
8
9
9
9
9
9
9
9
9
9
0
0
0
0
1
1
1
1
1
1
19
19
19
19
20
20
20
20
20
Year
Fig. 1 Number of citations between 1985 and 2017. Bar graph represents the number of citations found in
Pubmed database (https://www.ncbi.nlm.nih.gov/pubmed/advanced) using key words, “Psychoneuroimmu-
nology” and “Disease”
2 Microbial Endocrinology
The term “stress” has been broadly defined, but its meaning is
largely accepted as the neuronal activation and release of neurohor-
monal intermediaries from the sympathetic (SNS), parasympathetic
(PNS), and autonomic nervous (ANS) systems that mediate host
behavioral and physiological functions.
Exposure to various forms of stress (physical, psychological,
social, and infectious) activates the central nervous system, leading
to a change in neuronal tone of SNS, PNS, and ANS networks.
Dr. Lyte speaks of his experience in addressing the question of why
should one consider neuroendocrine hormones as part of a micro-
bial in vitro culture. His response, “because we do not have tryptic
soy broth and brain heart infusion media flowing through our veins
and arteries and until we use media that reflect the same environ-
ment that bacteria must survive in, then we will never fully under-
stand the mechanisms underlying the ability of infectious agents to
cause disease.” The implications of his response mirror that of PNI
research and reinforce the value of a multidiscipline framework
toward understanding mechanisms related to the pathogenesis of
human disease [6].
Initial studies determining the effects of stress on bacterial
resistance were instrumental in raising the notion of the potential
for neuroendocrine factors to affect bacterial physiology [7]. Such
studies emerged primarily from PNI research which defined stress-
induced influences on immunity as a determinant of susceptibility
Intersections Between Neuroimmune and Microbiota 23
to infectious and inflammatory disease [1, 3, 8, 9]. Subsequently,

studies by Freestone and Lyte, whose intent was to create an
environmental condition to best reflect the anatomical site from
which microbial species invade or reside, initiated studies including
neuroendocrine “stress” hormones in bacterial cultures [6, 10]. In
1992, Lyte and Ernst originally defined microbial endocrinology
through their sentinel studies demonstrating the influence of neu-
roendocrine hormones on bacterial growth [11]. Additional
reports discovered that numerous hormonal factors and their
receptors could be expressed by microorganisms similar to those
of humans and other vertebrates [12, 13]. Over the past two
decades of study, seminal findings have facilitated significant inter-
ests in the role of neuroendocrine hormones in the pathogenesis of
infectious disease [5, 14–19]. Most notable were reports of neuro-
endocrine factors’ role in mediating enteric pathogens
[20–23]. This was of no surprise based on the existence of a
neuro-immune network within the gastric mucosa referred to as
the “brain-gut-axis.” The consequence of these studies was an
emergence in defining direct influences of stress-associated hor-
monal factors on bacterial physiology within the gut mucosa and
broadly across the human microbiota ecosystem.
3 Neuroendocrine Effects on Microbiota and Pathogenic Species
The significance of microbial endocrinology has benefited from

recent genetic technological advances that have permitted the dis-
covery of new microbiota niches along the gut, skin, vaginal, and
upper and lower respiratory tissues [24–27]. Culture-independent
methodologies such as metagenomics and whole-genome microbial
sequencing have been significantly useful for microbial community
characterizations [26]. In 2008, the National Institutes of Health
(NIH) initiated a multisite project with the goal of identifying and
characterizing the microorganisms found on the human host under
healthy and disease conditions known as the Human Microbiome
Project (HMP). Through identification of human microbial niches,
the goal of this 5-year project was to test how alterations in the
human microbiome could impact human health. The knowledge
gained from the HMP has undoubtedly advanced human health,
considering that microbes which exist within and on humans have a
tenfold greater genome. While most are bacterial in origin, archaea,
protists, fungi, and viruses are also noteworthy constituents of the
human microbiota. The HMP genome reference database identifies
3055 hits of microbial organisms across skin, intestinal, respiratory,
and vaginal tissues to illustrate the existence and complexity of the
human microbial ecosystem (http://hmpdacc.org/catalog/grid.
php?dataset¼genomic).
Keyword Terms (Microbiota and Neurohormones)

60
Citations per year

40
20
0
82
04
05
07
08
09
10
11
12
13
14
15
16
19
20
20
20
20
20
20
20
20
20
20
20
20
Year
Fig. 2 Number of citations between 1982 and 2016. Bar graph represents the
number of citations found in Pubmed database (https://www.ncbi.nlm.nih.gov/
pubmed/advanced) using key words, “Microbiota and Neurohormones”
Since 1982, research defining the interplay between neuroen-

docrine responses and microbial niches in humans has significantly
grown (Fig. 2). The subsections that follow provide a brief over-
view of what is currently known of neuroendocrine-mediated
effects on the microbiota and pathogenic species.
3.1 Catecholamine As a constituent of the sympathetic nervous system, catecholamine

Effects activity plays a significant role in maintaining homeostasis in the
human body. Two major neurotransmitters, epinephrine (E) and
norepinephrine (NE), which are synthesized from tyrosine com-
prise of the catecholamines. Studies on these neurotransmitters
mainly revolved around their role in the “fight or flight” response,
transmitting signals across a chemical synapse and modulating
blood flow throughout the body. Their mode of action is facilitated
by the synaptic junctions within highly innervated tissues including
blood vessels, liver kidney, intestines, lung, heart, brain, and
immune organs (e.g., spleen, lymph nodes, and lymphocyte popu-
lations). Catecholamine signals are transmitted through adrenergic
receptors that comprise the G protein-coupled receptor family
(GPCR). These adrenergic receptors are subclassified into nine
isotypes: three alpha-1 types (α1A, α1B, and α1C), three alpha-
2 types (α2A, α2B, and α2C), and three beta-types (β1, β2, and β3).
Most of what we know regarding microbial responses to cate-
cholamines originates from studies of the brain-gut microbiota
interactions [11, 28–31]. Catecholamines have been shown to
augment the growth of species including Escherichia coli, Salmo-
nella typhi, [32, 33], Campylobacter jejuni, and Bordetella
bronchiseptica and others [34–36]. Catecholamines have also been

shown to directly modulate bacterial virulence. For instance, NE
was shown to stimulate inflammatory and secretory responses
caused by E. coli O157:H7, and augmented the microbe’s attach-
ment to intestinal mucosa [37, 38]. NE was also reported to
increase both the cellular cytotoxicity and enterotoxicity of Vibrio
parahaemolyticus infection through the induction of type III secre-
tion system-1 genes [39, 40].
In recent studies, catecholamines have been found to augment
the microflora of other tissues. Due to high mortality risks asso-
ciated with respiratory disease, one might expect as a critical need to
understand the role of neuroendocrine responses that mediate
commensal and pathogenic organisms along the respiratory tract.
Specifically, Sandrini et al., 2014, demonstrated that NE could
promote pneumococcal growth and biofilm formation [41]. Sec-
ondly, Gonzales et al., 2014, showed that NE reduced lung adher-
ence through iron binding as a potential mechanism [42]. Skin also
comprises a complex ecosystem of microbial populations on the
human body that are indispensable as a protective barrier
[43, 44]. Previous studies have shown that neuropeptides diffuse
in significant amounts along epidermal skin layers [45]. In fact,
numerous skin commensals such as S. aureus, Pseudomonas aerugi-
nosa, and Staphylococcus epidermidis are responsive to catechola-
mines, resulting in an increase in their ability to adhere, grow, and
invade skin by enhancement of virulence factors [46, 47]. Further
investigation of catecholamines’ effect on skin microbiota may
provide important insight into the mechanisms involved in skin
diseases and wound-healing responses, which could lead to novel
therapies. In addition to gut, lung, and skin, oral and vaginal
mucosal tissues harbor distinctive microbial communities impacted
by catecholamines [48–52].
How microbial species sense and utilize catecholamines stems
from the principle knowledge of a microorganisms’ need to acquire
iron (Fe) in order to support its growth, adherence, and other
metabolic functions [53]. First identified from the plant extract
catechin, catechols are a type of siderophore which are defined as
small, high-affinity iron-chelating compounds secreted by micro-
organisms. Catechols are also present as catecholamines in neuro-
transmitters, including dopamine, epinephrine, and
norepinephrine, as well as in polymerized forms of dopamine,
such as the skin pigment melanin [54, 55]. The most common
catechol produced by bacteria to promote sequestration and uptake
of Fe is 2,3-dihydroxybenzoic acid, enterobactin [56]. Enterobactin
and other siderophores including petrobactin are produced by
various enteric bacteria, including E. coli, Marinobacter hydrocarbo-
noclasticus, and Bacillus anthracis [57–59]. In 2003, studies by
Freestone et al. demonstrated that the ability to stimulate microbial
growth in serum could be enhanced by NE, E, and dopamine as
Host Transferrin, lactoferrin and Ferritin

(non-heme iron proteins)
Catecholamine
Release Scavanges Fe From
Bacterial Species
Microorganisms CA-Tf/Lf complex

Iron (Fe)
Binding to bacterial
Siderophores =Utilization of Fe
Fig. 3 Catecholamine’s role in facilitating Fe uptake by bacterial species. Commensals depend on nutrients for
survival. Acquiring iron (Fe) in the blood through high-affinity ferric-binding proteins such as lactoferrin and
transferrin. However, when certain endocrine factors such as norepinephrine are introduced, the iron can
become available to the bacteria. Norepinephrine and other catecholamines can bind to these ferric-binding
proteins resulting in the coordinated reduction of Fe (III) to Fe (II), an iron valency for which the proteins have
low affinity
well as by their metabolites through sequestration of Fe-containing

medium (e.g., serum or blood), typically unavailable due to
Fe-transferrin/lactoferrin complexes [60]. Thus, growth-
stimulating effects of catecholamines have been largely related to
the catechol-containing moiety forming a complex with the Fe
within transferrin (Tf) or lactoferrin (Lf). This complex weakens
Fe binding and so enables bacteria to acquire the normally inacces-
sible complexed Fe (Fig. 3).
Catecholamine inotropes used in the treatment of acutely ill
patients also have been shown to increase staphylococcal and pseu-
domonad biofilm formation and promote recovery from antibiotic
damage [61, 62]. Researchers have shown that host NE hormone
sensing can take place directly via bacterial adrenergic receptors
(BARs) [63]. Freestone et al. showed that administration of cate-
cholamine α-receptor antagonists, in contrast to antagonists for
β-adrenergic receptors, could block catecholamine-induced growth
in these bacteria [64]. To date, similar ligands/proteins responsible
for stress hormone recognition in Gram-positive bacteria have been
identified.
NE is also able to produce key signaling molecules known as
autoinducers [65]. Lyte et al. demonstrated that catecholamines
can induce growth of Gram-negative bacteria, particularly enteric
species involving a growth stimulator, termed noradrenaline-
induced autoinducer (NA-AI). The NA-AI induces its own
synthesis and is heat stable, highly cross-species acting activity
that stimulates increases in growth of magnitude similar to that
achievable with the catecholamines [65]. It was also shown that

NA-AI induces bacterial growth independently of transferrin
(Tf) or lactoferrin (Lf) [60], while also being able to rapidly stimu-
late the recovery to active growth of viable, but non-culturable
E. coli O157:H7 or Salmonella ssp., as well as increase the rate of
germination of Bacillus anthrax spores [66, 67]. Furthermore, the
NA-AI is recognized by periodontal pathogens [68]. Interestingly,
its synthesis and induction of the E. coli O157:H7 NA-AI were
found to require exposure to catecholamines [65, 69]. This indi-
cates that the effects of catecholamine released during acute stress
could have lasting and wide-acting effects after catecholamine levels
in the host animal have returned to normal.
3.2 Glucocorticoid Glucocorticoids are a class of corticosteroids that bind to the

Effects glucocorticoid receptor that are produced by the adrenal cortex.
Cortisol, a prominent glucocorticoid, is involved in carbohydrate,
protein, and fat metabolism and as an anti-inflammatory agent. To
date, research has proven its potential effects on bacterial physio-
logical responses. For example, Pseudomonas aeruginosa was found
to produce a protease factor that inhibits cortisol-binding globulin
as a mechanism of blocking plasma cortisol transport [70, 71]. In
addition, Verbrugghe, E. et al. have shown that glucocorticoids
such as cortisol can directly impact bacterial functioning, showing
that Salmonella typhimurium proliferation was increased within
macrophages exposed to cortisol in pigs, but not in the presence
of catecholamines [72]. Most recently, Morris, D.J. et al., 2017,
showed that gut microbial metabolic products of endogenous
adrenocorticosteroids could be factors for hypertension
[73]. Using established experimental murine models of aversive
stress and respiratory disease, our published and unpublished stud-
ies demonstrate how stress-induced neuroendocrine activation
affects immune and inflammatory responses in the lung. We also
provide the first evidence that corticotropin-releasing hormone
(CRH), a 41-amino-acid peptide expressed in central nervous and
peripheral tissues, can directly influence the virulence of a common
respiratory commensal and opportunistic pathogen, Streptococcus
pneumoniae. Specifically, our studies demonstrated that CRH
induced a higher number of colony-forming units at lower bacterial
concentrations in vitro compared to unexposed Streptococcus pneu-
moniae. We also demonstrated that preexposing Streptococcus pneu-
moniae with CRH increases bacterial burden in lung. Recently we
have shown that CRH exposure can increase serotype-specific cap-
sule formation and promote antibiotic resistance [74].
3.3 Cholinergic The autonomic nervous system (ANS) controls visceral activity
Effects within the body through three main divisions, all of which have
preganglionic neurons in the central nervous system (CNS) that
synapse with ganglionic neurons and that release acetylcholine
(Ach), which in turn activates nicotinic Ach receptors. Interestingly,

certain microbes can also produce ACh [75]. The first evidence of
ACh production by a microbe was shown a century ago, when ACh
was isolated from the ergot fungus Claviceps purpurea, and three
decades later this neurotransmitter was isolated from the bacterium
Lactobacillus plantarum [76]. More recently, ACh was measured
using radioimmunoassay from Saccharomyces cerevisiae, Bacillus
subtilis, E. coli, and Staphylococcus aureus [77]. The function of
ACh in these organisms and the conditions that lead to its synthesis
remain unknown. ACh activity in bacteria has been examined
in vitro in 60 strains of the following bacteria: Escherichia, Enter-
obacter, Erwinia, Serratia, Proteus, Alcaligenes, Flavobacterium,
Bacillus, Agrobacterium, Micrococcus, Staphylococcus, Sarcina, Cory-
nebacterium, Arthrobacter, Brevibacterium, Aeromonas, Protamino-
bacter, Xanthomonas, and Pseudomonas. From these strains,
strong specific hydrolysis was found only in P. fluorescens, while
weak hydrolysis was detected in the Aeromonas and Arthrobacter
families [78, 79]. Examples of microbes for which there is pharma-
cological evidence of the presence of nAChRs include Trypanosoma
evansi and Trypanosoma cruzi [80–82].
4 Perspective on the Neuroimmune-Microbial-Axis
Reciprocal influences between immune and microbial responses are

believed to have a major impact on delineating complex mechan-
isms of disease. Commensal microbes are exponentially greater
than the total number of cells on the human anatomy. The pressure
for symbiotic interactions undoubtedly holds many mysteries
underlining disease pathogenesis. Thus, intentional studies to
uncover imbalances or “dysbiosis” of microbial ecosystems hold
promise for the development of novel therapies. To date, altera-
tions in microbiota niches have been considered to be related to
numerous disease etiologies including obesity, autoimmunity,
asthma, metabolic syndrome, colitis, and cancer [83–87]. The
immune system is thus believed to be a mechanistic link, mediating
changes in microbiota that in turn part dictates disease [5, 88–93].
4.1 Maturation As a first line of defense, the immune system is required to arm the
of the Immune System: host through a balanced mechanism of defining “self” versus “non-
Role of the Microbiota self” antigenic peptides. This complex mechanism requires mutual
mediation with the trillions that inhabit host tissues. This rational is
best demonstrated in experimental studies, utilizing germ-free mice
(e.g., devoid of microbes at birth) as a model. Without a functional
microbiome, the immune system of germ-free mice is severely
immature compared to conventional mice [94, 95]. In the gut,
for example, lymphoid Peyer’s patches are significantly underdevel-

oped, and aberrations of the gut-associated mucosal tissues that are
altered through reduced mucus production, and declined goblet
cells, can also be noted. This mechanism was confirmed from
studies demonstrating that introducing various microbes within
germ-free mice could generate maturation of gut immunity
[96, 97]. Maturation of immune responses was found to depend
on the presence of host microbial pattern recognition receptors
(PPRs) expressed on innate immune system cells [98, 99]. PPRs
are able to recognize conserved microbial components and sense
microbial colonization through the detection of conserved micro-
bial components (e.g., nucleic acids, lipopolysaccharide) present on
all bacteria termed pathogen-associated molecular patterns
(PAMPs) and recently described on viral pathogens (e.g., viral
and bacterial DNA) [100]. Through this mechanism of recognition
by the immune system, innate and tailored host immune responses
become activated which are believed to be dependent on the com-
mensal species and their location. Furthermore, there is increasing
evidence that specific microbiota may have specific influences that
categorize them into certain groups such as those microbiota that
stimulate an inflammatory environment and those that are immu-
noregulatory. These specific “roles” are feasible due to what is
known with respect to the tailored functional immune responses
researchers have described by microbial commensal from in vitro
and in vivo studies of microbial commensals [101, 102]. Based on
this knowledge and ongoing research in the field, it is conceivable
to understand how functional aspects of microbiota can influence
disease.
5 Concluding Remarks: Implications of Neuroendocrine-Microbial-Immune

Interactions
For decades, microbiota composition has been thought synony-

mous with human health. Numerous examples supporting this
claim include the etiology of intestinal disease, nonalcoholic liver
disease, diabetes, experimental autoimmune encephalomyelitis, and
other metabolic disorders. Similarly, skin, respiratory, and vaginal
disease have been linked to altered microbiota. In order to advance
medical breakthroughs, the screening and mapping of genetic,
metabolic, and proteomic pathways are necessary. Based on the
ever-increasing knowledge defining host-microbial interactions, it
is noteworthy to recognize the interdependency between microbial
endocrinology and neuroendocrine immune-based pathways as an
opportunity to forge new ideas related to human health. In this
regard Lyte and others provide evidence for this concept [4, 12, 53,
103]. The human host containing its microbial niches produces
A.
Neuroendocrine B.
Response
C.
Bacterial Immune
Response Response
D.
Health
Outcomes
Fig. 4 Neuroendocrine-immune-microbiota axis. Conceptual framework. This

model highlights bidirectional interactions between three major disciplines
(microbiology, neurobiology, and immunology) that form a complex integrative
network defining neuroendocrine-bacterial (a); neuroendocrine-immune (b); and
bacterial-immune (c) proposed to determine health outcomes (d)
and responds to neuroendocrine stress factors and their metabo-

lites. Acquisition and utilization of neuro-compounds facilitate
specialized functions such as the augmentation of colonization,
adherence, and virulence potential as well as species competition
as part of the microbiota. For example, researchers have shown
experimentally through genetic sequence analysis of microbiota
that its composition is sensitive to stress conditions, allowing out-
growth of pathogenic organisms [104]. Likewise, human immune
and neuronal systems acquire and deliver signals that sustain sym-
biosis with its microbial counterparts. This conceptual interface of
host neuroendocrine-immunology and microbiology disciplines
has great impact whereby multidirectional interactions are at play
and that the sum of these interactions dictates health outcomes
(Fig. 4). Given this perspective, the utility of prebiotics and pro-
biotics, along with neuroendocrine and immune-targeted treat-
ments, may give rise to a goal of precision-based therapies.
Further studies that dissect the interactions between microbiota
and neuroimmune systems may aid in reaching this goal.
Acknowledgments
The authors would like to thank the Department of Microbiology,

Immunology and Genetics for financial support and resources nec-
essary for the completion of this work. The authors would also like
to thank Ms. Mira Bakine for her contribution in figure develop-
ment for this review.
References
1. Glaser R, Kiecolt-Glaser J (2005) How stress immunization targeting beta-hemolysin and
damages immune system and health. Discov CAMP factor. J Invest Dermatol 131:401–409
Med 5:165–169 16. Dashper SG, Seers CA, Tan KH, Reynolds EC
2. Kiecolt-Glaser JK (2009) Psychoneuroimmu- (2011) Virulence factors of the oral spirochete
nology: psychology’s gateway to the biomed- Treponema denticola. J Dent Res
ical future. Perspect Psychol Sci 4:367–369 90:691–703
3. Straub RH, Cutolo M (2017) 17. Zhou Y, Lin P, Li Q, Han L, Zheng H, Wei Y,
Psychoneuroimmunology-developments in Cui Z, Ni Y, Guo X (2010) Analysis of the
stress research. Wien Med Wochenschr microbiota of sputum samples from patients
4. Dinan TG, Cryan JF (2017) Microbes, immu- with lower respiratory tract infections. Acta
nity, and behavior: psychoneuroimmunology Biochim Biophys Sin Shanghai 42:754–761
meets the microbiome. Neuropsychopharma- 18. Walk ST, Blum AM, Ewing SA, Weinstock JV,
cology 42:178–192 Young VB (2010) Alteration of the murine
5. Lyte M (2016) Microbial endocrinology: an gut microbiota during infection with the par-
ongoing personal journey. Adv Exp Med Biol asitic helminth Heligmosomoides polygyrus.
874:1–24 Inflamm Bowel Dis 16:1841–1849
6. Freestone PP, Lyte M (2008) Microbial endo- 19. Stecher B, Chaffron S, Kappeli R,
crinology: experimental design issues in the Hapfelmeier S, Freedrich S, Weber TC,
study of interkingdom signalling in infectious Kirundi J, Suar M, McCoy KD, von
disease. Adv Appl Microbiol 64:75–105 Mering C, Macpherson AJ, Hardt WD
7. Lyte M (1993) The role of microbial endocri- (2010) Like will to like: abundances of closely
nology in infectious disease. J Endocrinol related species can predict susceptibility to
137:343–345 intestinal colonization by pathogenic and
commensal bacteria. PLoS Pathog 6:
8. Ashley NT, Demas GE (2017) e1000711
Neuroendocrine-immune circuits, pheno-
types, and interactions. Horm Behav 20. Sokol H, Vasquez N, Hoyeau-Idrissi N,
87:25–34 Seksik P, Beaugerie L, Lavergne-Slove A,
Pochart P, Marteau P (2010) Crypt abscess-
9. Schubert C (2014) Psychoneuroimmunology associated microbiota in inflammatory bowel
of the life span: impact of childhood stress on disease and acute self-limited colitis. World J
immune dysregulation and inflammatory dis- Gastroenterol 16:583–587
ease in later life. Psychother Psychosom Med
Psychol 64:171–180 21. O’Keefe SJ (2010) Tube feeding, the micro-
biota, and Clostridium difficile infection.
10. Freestone P, Lyte M (2010) Stress and micro- World J Gastroenterol 16:139–142
bial endocrinology: prospects for ruminant
nutrition. Animal 4:1248–1257 22. Mallozzi M, Viswanathan VK, Vedantam G
(2010) Spore-forming Bacilli and Clostridia
11. Lyte M (1992) The role of catecholamines in in human disease. Future Microbiol
gram-negative sepsis. Med Hypotheses 5:1109–1123
37:255–258
23. Endt K, Stecher B, Chaffron S, Slack E,
12. Lyte M (2004) Microbial endocrinology and Tchitchek N, Benecke A, Van Maele L, Sirard
infectious disease in the 21st century. Trends JC, Mueller AJ, Heikenwalder M, Macpher-
Microbiol 12:14–20 son AJ, Strugnell R, von Mering C, Hardt
13. Lyte M (2014) The effect of stress on micro- WD (2010) The microbiota mediates patho-
bial growth. Anim Health Res Rev gen clearance from the gut lumen after
15:172–174 non-typhoidal Salmonella diarrhea. PLoS
14. Lofgren JL, Whary MT, Ge Z, Pathog 6:e1001097
Muthupalani S, Taylor NS, Mobley M, 24. Koyanagi T, Sakamoto M, Takeuchi Y,
Potter A, Varro A, Eibach D, Suerbaum S, Ohkuma M, Izumi Y (2010) Analysis of
Wang TC, Fox JG (2011) Lack of commensal microbiota associated with peri-implantitis
flora in Helicobacter pylori-infected INS-GAS using 16S rRNA gene clone library. J Oral
mice reduces gastritis and delays intraepithe- Microbiol 2
lial neoplasia. Gastroenterology 140:210–220 25. Montagner F, Gomes BP, Kumar PS (2010)
15. Lo CW, Lai YK, Liu YT, Gallo RL, Huang CM Molecular fingerprinting reveals the presence
(2011) Staphylococcus aureus hijacks a skin of unique communities associated with paired
commensal to intensify its virulence:
samples of root canals and acute apical 36. Everest P (2007) Stress and bacteria: micro-
abscesses. J Endod 36:1475–1479 bial endocrinology. Gut 56:1037–1038
26. Bokulich NA, Rideout JR, Mercurio WG, 37. Moreira CG, Sperandio V (2016) The
Shiffer A, Wolfe B, Maurice CF, Dutton RJ, epinephrine/norepinephrine/autoinducer-3
Turnbaugh PJ, Knight R, Caporaso JG Interkingdom signaling system in Escherichia
(2016) mockrobiota: a public resource for coli O157:H7. Adv Exp Med Biol
microbiome bioinformatics benchmarking. 874:247–261
mSystems 1(5) 38. Dowd SE (2007) Escherichia coli O157:H7
27. Han M, Yang P, Zhou H, Li H, Ning K gene expression in the presence of catechol-
(2016) Metagenomics and single-cell omics amine norepinephrine. FEMS Microbiol Lett
data analysis for human microbiome research. 273:214–223
Adv Exp Med Biol 939:117–137 39. Nakano M, Takahashi A, Sakai Y, Kawano M,
28. Lyte M, Ernst S (1992) Catecholamine Harada N, Mawatari K, Nakaya Y (2007)
induced growth of gram negative bacteria. Catecholamine-induced stimulation of
Life Sci 50:203–212 growth in Vibrio species. Lett Appl Microbiol
29. Asano Y, Hiramoto T, Nishino R, Aiba Y, 44:649–653
Kimura T, Yoshihara K, Koga Y, Sudo N 40. Waldor MK, Sperandio V (2007) Adrenergic
(2012) Critical role of gut microbiota in the regulation of bacterial virulence. J Infect Dis
production of biologically active, free cate- 195:1248–1249
cholamines in the gut lumen of mice. Am J 41. Sandrini S, Alghofaili F, Freestone P, Yesilkaya
Physiol Gastrointest Liver Physiol 303: H (2014) Host stress hormone norepineph-
G1288–G1295 rine stimulates pneumococcal growth, biofilm
30. Mittal R, Debs LH, Patel AP, Nguyen D, formation and virulence gene expression.
Patel K, O’Connor G, Grati M, Mittal J, BMC Microbiol 14:180
Yan D, Eshraghi AA, Deo SK, Daunert S, 42. Gonzales XF, Castillo-Rojas G, Castillo-Rodal
Liu XZ (2017) Neurotransmitters: the critical AI, Tuomanen E, Lopez-Vidal Y (2013) Cat-
modulators regulating gut-brain Axis. J Cell echolamine norepinephrine diminishes lung
Physiol 232:2359–2372 epithelial cell adhesion of Streptococcus
31. Ney DM, Murali SG, Stroup BM, Nair N, Pneumoniae by binding iron. Microbiology
Sawin EA, Rohr F, Levy HL (2017) Metabo- 159:2333–2341
lomic changes demonstrate reduced bioavail- 43. Baldwin HE, Bhatia ND, Friedman A, Eng
ability of tyrosine and altered metabolism of RM, Seite S (2017) The role of cutaneous
tryptophan via the kynurenine pathway with microbiota harmony in maintaining a func-
ingestion of medical foods in phenylketon- tional skin barrier. J Drugs Dermatol
uria. Mol Genet Metab 121:96–103 16:12–18
32. Hegde M, Wood TK, Jayaraman A (2009) 44. Maguire M, Maguire G (2017) The role of
The neuroendocrine hormone norepineph- microbiota, and probiotics and prebiotics in
rine increases Pseudomonas aeruginosa PA14 skin health. Arch Dermatol Res 309:411–421
virulence through the las quorum-sensing 45. Zouboulis CC, Picardo M, Ju Q, Kurokawa I,
pathway. Appl Microbiol Biotechnol Torocsik D, Biro T, Schneider MR (2016)
84:763–776 Beyond acne: current aspects of sebaceous
33. Nietfeld JC, Yeary TJ, Basaraba RJ, Schauen- gland biology and function. Rev Endocr
stein K (1999) Norepinephrine stimulates Metab Disord 17:319–334
in vitro growth but does not increase patho- 46. Ali MF, Soto A, Knoop FC, Conlon JM
genicity of salmonella choleraesuis in an (2001) Antimicrobial peptides isolated from
in vivo model. Adv Exp Med Biol skin secretions of the diploid frog, Xenopus
473:249–260 tropicalis (Pipidae). Biochim Biophys Acta
34. Anderson MT, Armstrong SK (2006) The 1550:81–89
Bordetella bfe system: growth and transcrip- 47. Beasley FC, Marolda CL, Cheung J, Buac S,
tional response to siderophores, catechols, Heinrichs DE (2011) Staphylococcus aureus
and neuroendocrine catecholamines. J Bacter- transporters Hts, Sir, and Sst capture iron
iol 188:5731–5740 liberated from human transferrin by
35. Cogan TA, Thomas AO, Rees LE, Taylor AH, Staphyloferrin A, Staphyloferrin B, and cate-
Jepson MA, Williams PH, Ketley J, Hum- cholamine stress hormones, respectively, and
phrey TJ (2007) Norepinephrine increases contribute to virulence. Infect Immun
the pathogenic potential of campylobacter 79:2345–2355
jejuni. Gut 56:1060–1065
48. Brotman RM, Ravel J, Cone RA, Zenilman 60. Freestone PP, Haigh RD, Williams PH, Lyte
JM (2010) Rapid fluctuation of the vaginal M (2003) Involvement of enterobactin in
microbiota measured by Gram stain analysis. norepinephrine-mediated iron supply from
Sex Transm Infect 86:297–302 transferrin to enterohaemorrhagic Escherichia
49. Cocco JF, Antonetti JW, Burns JL, Heggers coli. FEMS Microbiol Lett 222:39–43
JP, Blackwell SJ (2010) Characterization of 61. Choi JM, Jo JY, Baik JW, Kim S, Kim CS,
the nasal, sublingual, and oropharyngeal Jeong SM (2017) Risk factors and outcomes
mucosa microbiota in cleft lip and palate indi- associated with a higher use of inotropes in
viduals before and after surgical repair. Cleft kidney transplant recipients. Medicine (Balti-
Palate Craniofac J 47:151–155 more) 96:e5820
50. Fardini Y, Chung P, Dumm R, Joshi N, Han 62. Freestone PP, Al-Dayan N, Lyte M (2016)
YW (2010) Transmission of diverse oral bac- Staphylococci, catecholamine inotropes and
teria to murine placenta: evidence for the oral hospital-acquired infections. Adv Exp Med
microbiome as a potential source of intrauter- Biol 874:183–199
ine infection. Infect Immun 78:1789–1796 63. Lyte M, Ernst S (1993) Alpha and beta adren-
51. Spear GT, Sikaroodi M, Zariffard MR, Landay ergic receptor involvement in catecholamine-
AL, French AL, Gillevet PM (2008) Compar- induced growth of gram-negative bacteria.
ison of the diversity of the vaginal microbiota Biochem Biophys Res Commun
in HIV-infected and HIV-uninfected women 190:447–452
with or without bacterial vaginosis. J Infect 64. Freestone PP, Haigh RD, Lyte M (2007)
Dis 198:1131–1140 Blockade of catecholamine-induced growth
52. Tada A, Hanada N (2010) Opportunistic by adrenergic and dopaminergic receptor
respiratory pathogens in the oral cavity of antagonists in Escherichia coli O157:H7, Sal-
the elderly. FEMS Immunol Med Microbiol monella enterica and Yersinia enterocolitica.
60:1–17 BMC Microbiol 7:8
53. Freestone PP, Sandrini SM, Haigh RD, Lyte 65. Freestone PP, Haigh RD, Williams PH, Lyte
M (2008) Microbial endocrinology: how M (1999) Stimulation of bacterial growth by
stress influences susceptibility to infection. heat-stable, norepinephrine-induced autoin-
Trends Microbiol 16:55–64 ducers. FEMS Microbiol Lett 172:53–60
54. Salton MR (1964) Requirement of dihydrox- 66. Voigt B, Schweder T, Sibbald MJ, Albrecht D,
yphenols for the growth of micrococcus lyso- Ehrenreich A, Bernhardt J, Feesche J, Maurer
deikticus in synthetic media. Biochim Biophys KH, Gottschalk G, van Dijl JM, Hecker M
Acta 86:421–422 (2006) The extracellular proteome of Bacillus
55. Shearer N, Walton NJ (2016) Dietary cate- licheniformis grown in different media and
chols and their relationship to microbial under different nutrient starvation condi-
endocrinology. Adv Exp Med Biol tions. Proteomics 6:268–281
874:101–119 67. Reissbrodt R, Rienaecker I, Romanova JM,
56. Neilands JB (1984) Siderophores of bacteria Freestone PP, Haigh RD, Lyte M,
and fungi. Microbiol Sci 1:9–14 Tschape H, Williams PH (2002) Resuscita-
57. Lee JY, Janes BK, Passalacqua KD, Pfleger BF, tion of salmonella enterica serovar typhimur-
Bergman NH, Liu H, Hakansson K, Somu ium and enterohemorrhagic Escherichia coli
RV, Aldrich CC, Cendrowski S, Hanna PC, from the viable but nonculturable state by
Sherman DH (2007) Biosynthetic analysis of heat-stable enterobacterial autoinducer. Appl
the petrobactin siderophore pathway from Environ Microbiol 68:4788–4794
Bacillus anthracis. J Bacteriol 189:1698–1710 68. Roberts A (2005) Bacteria in the mouth.
58. Koppisch AT, Browder CC, Moe AL, Shelley Dent Update 32:134–136, 139–140, 142
JT, Kinkel BA, Hersman LE, Iyer S, Ruggiero 69. Lyte M, Frank CD, Green BT (1996) Produc-
CE (2005) Petrobactin is the primary side- tion of an autoinducer of growth by norepi-
rophore synthesized by Bacillus anthracis str. nephrine cultured Escherichia coli O157:H7.
Sterne under conditions of iron starvation. FEMS Microbiol Lett 139:155–159
Biometals 18:577–585 70. Simard M, Hill LA, Lewis JG, Hammond GL
59. Hickford SJ, Kupper FC, Zhang G, Carrano (2015) Naturally occurring mutations of
CJ, Blunt JW, Butler A (2004) Petrobactin human corticosteroid-binding globulin. J
sulfonate, a new siderophore produced by Clin Endocrinol Metab 100:E129–E139
the marine bacterium Marinobacter hydrocar- 71. Simard M, Hill LA, Underhill CM, Keller BO,
bonoclasticus. J Nat Prod 67:1897–1899 Villanueva I, Hancock RE, Hammond GL
(2014) Pseudomonas aeruginosa elastase dis- activity and lipid peroxidation in the brain
rupts the cortisol-binding activity of and spinal cord of rats infected with Trypano-
corticosteroid-binding globulin. Endocrinol- soma evansi. Vet Parasitol 175:237–244
ogy 155:2900–2908 82. Wolkmer P, da Silva CB, Paim FC, Da Silva
72. Verbrugghe E, Boyen F, Van Parys A, Van AS, Tavares KC, Lazzarotto CR, Palma HE,
Deun K, Croubels S, Thompson A, Thome GR, Miletti LC, Schetinger MR,
Shearer N, Leyman B, Haesebrouck F, Pas- Lopes ST, Mazzanti CM (2012) Biochemistry
mans F (2011) Stress induced Salmonella detection of acetylcholinesterase activity in
Typhimurium recrudescence in pigs coincides Trypanosoma evansi and possible functional
with cortisol induced increased intracellular correlations. Exp Parasitol 132:546–549
proliferation in macrophages. Vet Res 42:118 83. Bolino CM, Bercik P (2010) Pathogenic fac-
73. Morris DJ, Ridlon JM (2017) Glucocorti- tors involved in the development of irritable
coids and gut bacteria: “the GALF hypothe- bowel syndrome: focus on a microbial role.
sis” in the metagenomic era. Steroids Infect Dis Clin N Am 24:961–975, ix
125:1–13 84. Craig OF, Quigley EM (2010) Bacteria,
74. Ngo Ndjom CG, Kantor LV, Jones HP genetics and irritable bowel syndrome. Expert
(2017) CRH affects the phenotypic expres- Rev Gastroenterol Hepatol 4:271–276
sion of sepsis-associated virulence factors by 85. Fox JG, Feng Y, Theve EJ, Raczynski AR,
streptococcus pneumoniae serotype 1 in vitro. Fiala JL, Doernte AL, Williams M, McFaline
Front Cell Infect Microbiol 7:263 JL, Essigmann JM, Schauer DB, Tannenbaum
75. Weinstein LI, Revuelta A, Pando RH (2015) SR, Dedon PC, Weinman SA, Lemon SM, Fry
Catecholamines and acetylcholine are key reg- RC, Rogers AB (2010) Gut microbes define
ulators of the interaction between microbes liver cancer risk in mice exposed to chemical
and the immune system. Ann N Y Acad Sci and viral transgenic hepatocarcinogens. Gut
1351:39–51 59:88–97
76. Stanaszek PM, Snell JF, O’Neill JJ (1977) 86. Fujimura KE, Slusher NA, Cabana MD,
Isolation, extraction, and measurement of Lynch SV (2010) Role of the gut microbiota
acetylcholine from Lactobacillus plantarum. in defining human health. Expert Rev Anti-
Appl Environ Microbiol 34:237–239 Infect Ther 8:435–454
77. Pandey S, Sree A, Sethi DP, Kumar CG, 87. Ganal-Vonarburg SC, Fuhrer T, Gomez de
Kakollu S, Chowdhury L, Dash SS (2014) A Aguero M (2017) Maternal microbiota and
marine sponge associated strain of Bacillus antibodies as advocates of neonatal health.
subtilis and other marine bacteria can produce Gut Microbes 8(5):479–485
anticholinesterase compounds. Microb Cell 88. Smith PD, Smythies LE, Shen R, Greenwell-
Factories 13:24 Wild T, Gliozzi M, Wahl SM (2011) Intestinal
78. Horiuchi Y, Kimura R, Kato N, Fujii T, macrophages and response to microbial
Seki M, Endo T, Kato T, Kawashima K encroachment. Mucosal Immunol 4:31–42
(2003) Evolutional study on acetylcholine 89. Okada H, Kuhn C, Feillet H, Bach JF (2010)
expression. Life Sci 72:1745–1756 The ‘hygiene hypothesis’ for autoimmune and
79. Yamada T, Fujii T, Kanai T, Amo T, allergic diseases: an update. Clin Exp Immu-
Imanaka T, Nishimasu H, Wakagi T, nol 160:1–9
Shoun H, Kamekura M, Kamagata Y, 90. Martin R, Nauta AJ, Ben Amor K, Knippels
Kato T, Kawashima K (2005) Expression of LM, Knol J, Garssen J (2010) Early life: gut
acetylcholine (ACh) and ACh-synthesizing microbiota and immune development in
activity in Archaea. Life Sci 77:1935–1944 infancy. Benef Microbes 1:367–382
80. Costa MM, Silva AS, Paim FC, Franca R, Dor- 91. Belkaid Y, Harrison OJ (2017) Homeostatic
nelles GL, Thome GR, Serres JD, Schmatz R, immunity and the microbiota. Immunity
Spanevello RM, Goncalves JF, Schetinger 46:562–576
MR, Mazzanti CM, Lopes ST, Monteiro SG 92. Luo A, Leach ST, Barres R, Hesson LB,
(2012) Cholinesterase as inflammatory mar- Grimm MC, Simar D (2017) The microbiota
kers in a experimental infection by Trypano- and epigenetic regulation of T helper 17/reg-
soma evansi in rabbits. An Acad Bras Cienc ulatory T cells: in search of a balanced
84:1105–1113 immune system. Front Immunol 8:417
81. da Silva AS, Monteiro SG, Goncalves JF, 93. Lloyd CM, Marsland BJ (2017) Lung homeo-
Spanevello R, Oliveira CB, Costa MM, Jaques stasis: influence of age, microbes, and the
JA, Morsch VM, Schetinger MR, Mazzanti immune system. Immunity 46:549–561
CM, Lopes ST (2011) Acetylcholinesterase
94. Costa MC, Santos JR, Ribeiro MJ, Freitas GJ, 98. Chu H, Mazmanian SK (2013) Innate
Bastos RW, Ferreira GF, Miranda AS, Arifa immune recognition of the microbiota pro-
RD, Santos PC, Martins Fdos S, Paixao TA, motes host-microbial symbiosis. Nat Immu-
Teixeira AL, Souza DG, Santos DA (2016) nol 14:668–675
The absence of microbiota delays the inflam- 99. McDermott AJ, Huffnagle GB (2014) The
matory response to Cryptococcus gattii. Int J microbiome and regulation of mucosal immu-
Med Microbiol 306:187–195 nity. Immunology 142:24–31
95. Chung H, Pamp SJ, Hill JA, Surana NK, 100. Hansen JD, Vojtech LN, Laing KJ (2011)
Edelman SM, Troy EB, Reading NC, Villa- Sensing disease and danger: a survey of verte-
blanca EJ, Wang S, Mora JR, Umesaki Y, brate PRRs and their origins. Dev Comp
Mathis D, Benoist C, Relman DA, Kasper Immunol 35:886–897
DL (2012) Gut immune maturation depends 101. Salzman NH (2011) Microbiota-immune sys-
on colonization with a host-specific micro- tem interaction: an uneasy alliance. Curr Opin
biota. Cell 149:1578–1593 Microbiol 14:99–105
96. Fransen F, van Beek AA, Borghuis T, 102. Brown RL, Clarke TB (2017) The regulation
Meijer B, Hugenholtz F, van der Gaast-de of host defences to infection by the micro-
Jongh C, Savelkoul HF, de Jonge MI, Faas biota. Immunology 150:1–6
MM, Boekschoten MV, Smidt H, El Aidy S,
de Vos P (2017) The impact of gut microbiota 103. Bailey MT (2016) Psychological stress,
on gender-specific differences in immunity. immunity, and the effects on indigenous
Front Immunol 8:754 microflora. Adv Exp Med Biol 874:225–246
97. Geva-Zatorsky N, Sefik E, Kua L, Pasman L, 104. Bailey MT, Dowd SE, Parry NM, Galley JD,
Tan TG, Ortiz-Lopez A, Yanortsang TB, Schauer DB, Lyte M (2010) Stressor expo-
Yang L, Jupp R, Mathis D, Benoist C, Kasper sure disrupts commensal microbial popula-
DL (2017) Mining the human gut microbiota tions in the intestines and leads to increased
for immunomodulatory organisms. Cell 168 colonization by Citrobacter rodentium.
(928–943):e911 Infect Immun 78:1509–1519
Chapter 3
Psychoneuroimmunology: The Experiential Dimension

Elling Ulvestad
Abstract
Accumulating evidence has made clear that experience—the knowledge an individual acquires during a
lifetime of sensing and acting—is of fundamental biological relevance. Experience makes an impact on all
adaptive systems, including the endocrine, immune, and nerve systems, and is of the essence, not only for
the unfolding of an organisms’ healthy status, but also for the development of malfunctional traits.
Nevertheless, experience is often excluded from empirical approaches. A variety of complex interactions
that influence life histories are thereby neglected. Such ignorance is especially detrimental for psychoneu-
roimmunology, the science that seeks to understand how the exquisite and dynamic interplay between
mind, body, and environment relates to behavioral characteristics. This chapter reviews claims for incorpor-
ating experience as a member of good explanatory standing in biology and medicine, and more specifically
claims that experiential knowledge is required to enable meaningful and relevant explanations and predic-
tions in the psychoneuroimmunological realm.
Key words Experience, Microbiome, Umwelt, Development, Evolution, Function
1 Introduction
Ideas that the nervous, immune, and endocrine systems work in

close concert with each other as well as with external inputs were
not in high vogue prior to the 1980s. Scientists were still preoccu-
pied with elaborations of the internal workings of the three adaptive
systems, and rarely made crossover connections. That such an
integrative effort would be rewarding was, however, highlighted
in a 1981 landmark publication entitled Psychoneuroimmunology
[1]. The book, edited by Robert Ader, consisted of a collection of
reviews on emerging work, and made a fascinating but also a
challenging reading.
In the book’s foreword, Robert A. Good [2] outlined a
research agenda for the integrative efforts that should follow:
The question that remains is how these three major networks—the nervous
system, the endocrine system, and the immunologic system—interact and,
how, by understanding these interactions in precise quantitative terms, we
can learn to predict and control them. ([2], p. xix).
37
38 Elling Ulvestad
In the immediate follow-up article, Psychosocial factors in infec-

tious disease, S. Michael Plaut and Stanford B. Friedman [3] addi-
tionally stated that psychoneuroimmunology needs to understand
how various psychosocial factors of the experiencing subject can
modify external challenges. They thus elaborated further on an
often-observed phenomenon—that not all individuals infected
with a certain infectious agent come down with disease—and
emphasized that there must be “something more” involved in
pathogenesis than just a battle between the infectious agent and
the immune system. They made references to results from human
studies which demonstrated that the meaning a person attaches to a
phenomenon makes an impact on disease outcome, and so claimed
that:
The relevant question is not whether a given disease is caused by a patho-
genic agent or by psychological factors, but rather to what extent the disease
can be related to each of a number of factors in the history, makeup, and
environment of the organism. ([3], p. 7)
By this claim they highlighted a shift in research focus—not

only should scientists investigate why individuals are susceptible to
illness, but they should as well investigate why individuals are
resistant to disease. The “received view,” that infectious agents
and psychological stressors are sufficient causes of disease, should
therefore be replaced with a more comprehensive and interdisci-
plinary understanding of causation.
The challenge posed by Plaut and Friedman—that the human
organism’s life history needs to be included in the explanatory
framework—is demanding. For not only did the two researchers
claim that the life history should be approached from a scientific
viewpoint, but it should in addition be approached from the indi-
vidual’s perspective—from the meaning the susceptible person and
his adaptive systems attach to precipitating situations. These are
hard tasks indeed—for they ask of science more than science is
allowed to deliver, the reason being that science in its “craving for
generality” actually takes a “contemptuous attitude towards the
particular case” ([4], p. 18).
This idea can be explicated by way of a dilemma from scientific
publishing. On the one hand, editors are reluctant to communicate
case reports because cases are subject to a variety of uncontrolled
and uncontrollable influences, and so generalization of the individ-
ual outcome is a precarious undertaking. On the other hand,
editors endorse group-based investigations, because such studies
are amenable to strict control and thus generalization. However, as
publishers well know—results from group-based investigations lack
an important virtue of the case. Individual characteristics and con-
textual parameters are seldom irrelevant for the outcome, and by
treating these as confounders group-based studies thereby neglect
The Experiential Dimension 39
the experiential dimension and thus lose touch with the very indi-
viduals they represent.
Even a science that acknowledges the experiential dimension
and thus attempts to incorporate experiential parameters in the
form of major life events—e.g., divorce, death of a spouse, or
serious disease—often fails the task. As highlighted in a critical
review, individuals rarely apply the same meaning to a life event
[5]. Different subjects interpret events differently, and this differ-
ential interpretation is a determinant of how individuals respond
psychoendocrinologically. The “objective” characteristics of an
event, which are foundational in group-based investigations, are
thus not objective in a strict sense—they are rather interpreted in an
idiosyncratic manner by each different participator.
Knowledge of group characteristics is highly relevant but nev-
ertheless insufficient for understanding the individual, and it thus
appears that science needs a theory of the organism that also allows
for the emergence of “private” responses. This does not, however,
imply that science should become subjective and so comply with
the slogan “anything goes.” Science’s ultimate task is to give objec-
tive accounts of nature, also of individual experience. To stay true to
its ideal, science should therefore give well-grounded accounts of
subjective experience. Subjective accounts of experience, although
important for individual behavior, are nevertheless, at least for the
time being, outside the scientific realm.
In the following I provide an exploration of the challenge posed
by Plaut and Friedman. In so doing I will invoke the age-old
dilemma between the one and the many, and investigate how this
applies to the objective perspective taken by an external observer
and to the subjective perspective of a participant. The importance
of the subject’s environment, exemplified by the microbial com-
munities of our intestines, will also be highlighted, as will the role of
subjective interpretations of environmental stimuli over the life
cycle. These deliberations will hopefully reveal the complexity of
the experiential challenges facing psychoneuroimmunology.
2 The Meaning of Perception
A good starter for objective appropriation of subjective experiences

can be found in the Estonian zoologist Jakob von Uexküll’s
(1864–1944) elaborations on the individual organism’s dealing
with nature. When confronted with contemporary views of the
organism, Uexküll noted a discrepancy between what he believed
was the animal’s world—an active organism that interacts with its
surroundings in a meaningful manner—and the scientific concep-
tualization of it—an animal that mechanically adapts its behavior to
a given environment. To rectify this incongruity, he set out to build
a new biology in which the animal’s perspective was retained
40 Elling Ulvestad
[6]. And in so doing, he came to emphasize the importance of

perception. As he saw it, the animal’s perceptual perspective is not
something gained by passively receiving inputs in the shape of
information; rather, it involves an active interpretation of signals
rendered meaningful by the animal’s previous experiences.
Upon portraying animals as developmental structures with
communicative capabilities, Uexküll also came to notify that envi-
ronmental stimuli are of unequal importance for different kinds of
animals. He thus elaborated a distinction between the animal’s
environment, i.e., its physical surroundings, and the animal’s
umwelt—the meaning-carrying structure that contains a “sign or
symbol that members of the same species can understand, but that
those of another species cannot comprehend” ([7], p. 77). Hence,
even though animals of different species may share the same envi-
ronment, their differing umwelts make them experience the same
environment differently.
By emphasizing the animal’s perspective, Uexküll reached a
surprising insight—environmental signals are already meaningful
as they reach the animal. To see how this may come about, one
has to think of the couplings between organisms and their sur-
roundings as emerging from activity played out during two distinct
temporalities—one during the evolutionary history of the species,
when perceptual abilities are being shaped, and the other during the
developmental history of the individual, when the same perceptual
capabilities are being structured in relation to external inputs.
Perception of the umwelt thus consists of a phylogenetic compo-
nent, which allows for a fairly stereotyped pattern of behavior, and
an ontogenetic component that serves to diversify and tailor each
organism’s behavior to the actual environment (Fig. 1) [8].
Although Uexküll initially received many followers, including
the ethologist Konrad Lorenz and the philosopher on human
experience Martin Heidegger, his work had vitalistic undertones
and was therefore regarded unscientific. His emphasis on the quali-
tative aspects of animal perception and behavior was therefore soon
replaced by more quantitative and mathematically oriented the-
ories, and investigations related to the organism’s perceptual cou-
plings to the environment were thereby relayed to the background.
However, much of this changed in the 1970s when Uexküll’s ideas
were revitalized by the emerging field of biosemiotics [9], and not
the least by neurophysiologists and immunologists who began to
reorient their investigations along similar integrative lines. To
achieve the most from universalizing investigations, while at the
same time avoiding loss of the individual, it would thus be fruitful
to integrate psychoneuroimmunology with the biosemiotic view of
the world. But as humans differ from other animals in several
respects, the Uexküllian path needs to be adapted to the specific
human trajectory and situation.
Fig. 1 Each individual is the result of the lineage’s adaptive processes during evolutionary time and adaptive
processes during developmental time. These two adaptive processes are integrated in every organism. The
concave lens depicts the evolutionary resources of the organism, and includes its genetic makeup. The convex
lens, in contrast, depicts the developmentally shaped resources of the organism, including the epigenetic
makeup and the wirings of the central nervous system, the immune system, and the endocrine system. The
umwelt makes up the context in which the experiences make their impact. Dependent upon how these
systems integrate within the organism, the response may turn out as functional or dysfunctional. And, since
the organism’s responses are in many ways instructive of latter responses, the responses are depicted to feed
back to alter both the umwelt and latter responses to the same or dissimilar challenges
3 The Human Umwelt
Compared to other mammals and primates, humans differ in sev-

eral respects. Human life stories are distinguished by having an
exceptionally long life span, an extended period of juvenile depen-
dence, and support of reproduction by older postreproductive
females as well as by males [10]. In contrast to other animals,
with the exception of primates, humans also possess specific brain
structures that enable them to experience and interpret internal
physiological processes [11]. When such processes, which may be
pain, motion, nausea, thirst, or hunger, reach consciousness, they
create a subjective experience of own body. Variations in this struc-
ture may thus be part of the explanation for the great variation of
experience of a given bodily state by different humans [12].
Human beings are also distinguished by having unique psycho-
logical characteristics—a capacity for self-reflection, for designing
sophisticated symbolic structures, for attaching metaphorical con-
cepts to experiences, and for building models and categories with
42 Elling Ulvestad
the aid of the imagination. There is thus a creative element in man’s

dealing with his world—our brains create “a fantasy that coincides
with reality” ([13], p. 111). And exceptionally—when this reality
seems unfit, humans additionally have the capacity to alter the same
reality in radical ways during a process termed niche construction
[14]. Man’s umwelt is thus not only ecological, but also cultural—
all the way [15]. And as culture is an important part of man’s
umwelt, it should not come unexpected that artifacts of culture
may act as a selective force, thus feeding back on human beings in a
form strong enough to even alter the genome [16]. In the same
vein, alterations in perception and thought processes feed back on
the brain, thus altering neurophysiological and neurochemical
activities involved in perception, action, and emotional control
[17]—thus mind matters in a literal sense.
Man’s experiences are not entirely determined by the way the
world is—man is himself active in acquisition, selection, interpreta-
tion, and organization of the information. And as culture is shaped
as a cooperative effort along the generations, the human organism
is always and without exception a lived body in which history and
biography are woven together with interpersonal meaning as well as
individual purpose. While humans create and convey meaning in
coexistence with other humans, every person nevertheless inter-
prets experience within his own horizon which only partly coincides
with «all the others», even within the same cultural circle or society.
Such interaction does not disappear in reality, even when scientific
methodology excludes these elements from the study—and thereby
from science [18].
By disregarding experience, science also disregards the effects
of culture on human action. Since the meaning of a situation has
strong cultural bindings—something regarded as an upsetting
event in one culture may go quite unnoticed in another—there
are reasons to anticipate a major contribution of culture to variation
in psychoneuroimmunological development and function. The
habit of including a relatively homogenous group of participants
from Western, Educated, Industrialized, Rich, and Democratic
(WEIRD) societies in psychoneuroimmunological studies thus
effectively precludes the chance of understanding variation across
human populations [19]. There are even reasons to believe that
these WEIRD individuals are among the least representative mem-
bers one could use to generalize across human populations.
4 Restructuring the Explanatory Gap
Human beings come into the world with naturally selected coping
mechanisms. And since these mechanisms have evolutionary pre-
conditions, it is a task for science to ask whether or not such
preconditions interfere with man’s perception of the umwelt in a
true manner. There are two reasons why this question is impor-
tant—first, a science that aims to understand humans has to know
how humans experience the world, and second, a science that aims
to understand the world must have an idea of science’s own foun-
dations for knowledge acquisition. Exactly such preconditions and
preconceptions have been given critical attention by researchers in
the phenomenological tradition [20] and their investigations are
thus in many ways supplementary to the Uexküllian tradition.
The phenomenological tradition has made huge efforts to
understand the human experiential dimension and thus make it
accessible to investigation. As the phenomenologist sees it, any
biological individual accesses the umwelt through a first-person
perspective. For humans, this is the world as they know it, imbued
with meaning and emotions. Any human being has access to a wide
range of historically situated knowledge that helps him to respond
to external challenges in a meaningful way. The knowledge of each
generation is different, and knowledge also differs in different parts
of the world. Such knowledge is therefore spatiotemporally
restricted.
But human beings can also access knowledge that is true irre-
spective of time and place. To obtain such valued knowledge, man
has to “bracket” his first-person perspective on the world. He has to
take a God-like perspective, be the spectator who takes a view from
nowhere. The third-person perspective, which is the foundational
view of science, is not easily achieved. To reach the goal of true
knowledge, scientists have to act as disinterested, emotionless, and
neutral observers, and so have to undergo a long and arduous
training to achieve control over their inborn perceptual capabilities.
This is a complicated task, as they have to erase some of their
developmentally learned presuppositions.
The degree to which the third-person perspective can be
achieved varies widely between the sciences. While mathematicians
and logicians can be trained to master their subject in a true
“disinterested” manner, it is more questionable whether biologists,
social scientists, and humanists can achieve the same degree of
perfection. And for a simple reason—biological entities, including
human beings, are historically situated; they have a history that
matters as to what and who they are. That biological entities,
including the nervous, immune, and endocrine systems, have a
history does not, however, imply that they elude investigation by
a science with universalizing ambitions. But it does imply that
science should make more precise which aspects of the historical
entities it can reach firm conclusions about and which it cannot.
Although an arduous task, especially since science constantly
develops new concepts and exploratory technologies that push the
line of demarcation between knowledge and ignorance, it does
appear evident, at least for the time being, that science cannot
reach the innermost experiences of an individual. Science can for
44 Elling Ulvestad
example explore the general effects of major life events, but how
each individual experiences a divorce or the loss of a beloved one is
a private matter. It thus appears necessary to make an analytic
distinction between a public first-person perspective, which is ame-
nable for scientific investigations, and a private first-person perspec-
tive which is not. The private perspective is a specific characteristic
of each individual, be it a human being or a perceptual system. And
as such, it has no characteristics that can be generalized—it is thus
located beyond the realms of science. The public first-person per-
spective is, on the other hand, accessible from the outside. It
includes perceptual traits that are specific for a given species, and
is as such co-extensional with the animal’s umwelt.
While the public first-person aspect can be made explicit by
means of genetic, environmental, and developmental investiga-
tions, the private first-person aspect is an experiential dimension
and as such not accessible for scientific investigation. Such experi-
ence does not lend itself easily to standardized interpretation; it is
always an experience of something for someone, in a unique con-
text. There is thus a gap between what science can explain and what
it cannot—and the gap goes straight through the individual,
between the private and public aspects of the first person. None-
theless, exactly where the line of accessibility should be drawn is a
matter on which science should have a saying. The explanatory gap
should be made more precise, but prospects for its closing are for
the time being dim.
5 Relations: All the Way
Some of the most surprising knowledge coming out of the geno-

mics research programs has been novel insights into man’s relations
with the environment as well as to himself. Not only have the
sequencing efforts yielded rich empirical crops, but they have in
addition highlighted the importance of different analytic perspec-
tives. The latter is perhaps made most apparent by recent efforts to
understand relations between hosts and their microbes.
Ever since the microbiological revolution in the late 1800s, led
by Louis Pasteur and Robert Koch, microbes have been conceptua-
lized as external enemies against which man had to fight a war.
Although the perspective has been extremely rewarding in terms of
lives saved by vaccines and antibiotics, the war metaphor has been
seriously misleading as an aid to understand man’s microbial
umwelt. Only about 1500 microbial species regularly infect
human beings and cause disease, while millions of others either
ignore us or cooperate with us in an evolutionarily selected manner,
thus making their absence—not their presence—the real problem.
Several new observations derived from investigations of the
human genome as well as from the microbial communities that
colonize our mucous membranes and skin have made evident that it
is no longer possible to conceptualize microbes as simply “exter-
nal.” They are internal and cooperative as well. The sequencing of
the human genome made clear that our chromosomes are teemed
with microbially derived elements. The genome consists of 45% of
transposons—DNA sequences that are able to copy and move
within chromosomes—of which approximately 8% are retrovirus-
like [21]. Some of these retroviral integrations have been of great
importance for vertebrate physiological development. Although
most transposons that accumulate in the genome have no known
function, they contribute a large potential substrate for the evolu-
tion and development of regulatory networks [22, 23].
The genome also contains bacterially derived DNA, some of
which regulates the interaction between the eukaryotic cells and
their bacterially derived mitochondrial symbionts. The mitochon-
dria, which have evolved to become an integral part of the host’s
cells, have transferred some of their genes to the cell’s nucleus. And
in so doing, they lost the ability to reproduce freely. This loss has,
however, been matched by a comparative gain in survival capacity—
mitochondria, by their very location, have become shielded off
from immune destruction. The importance of keeping the interac-
tion between the eukaryotic cell and its mitochondria tightly regu-
lated is dramatically spelled out during debilitating physical trauma
in which mitochondria relocate or become destroyed. This leads to
a breakdown of the conditions for cooperation between the host
cell and the symbiont, and the host may thus develop a dangerous
systemic inflammatory response. The response includes fever, low
blood pressure, and increased heart rate [24], and is thus analogous
to the inflammatory process observed as a result of contaminating
bacteria during sepsis.
Another surprising observation that came out of the sequenc-
ing of the human genome was the relative paucity of genes. Based
on complexity estimates, man was thought to have about 100,000
genes prior to the sequencing. But only about 20,000 genes were
detected. Man was as complex as before, so how could the com-
plexity be accounted for by so few genes? One answer has to do
with the way the DNA is used for making proteins and regulatory
factors, and it has turned out that this process is far more efficient
than first thought [25]. But this is not the whole story; additional
data have since revealed that humans also have access to a plethora
of genes not coded for in the genome. And these genes, which are
located within bacteria and viruses on the skin and the mucous
membranes, by far outnumber the genes in the cellular nucleus.
Estimates have indicated that an adult human being can be
described as a superorganism consisting of 50% prokaryotic and
50% eukaryotic cells. Since every bacterium may be infected with as
many as 100 bacteriophages, thus giving an estimate of ten billion
46 Elling Ulvestad
viruses in each gram of human feces, there is definitely a plethora of

genes available within the human niche [26, 27].
The community of intestinal microbes, termed the microbiota,
which establishes itself shortly after birth, reaches adult levels in
early childhood. Although influenced by changes in diet and life
events, the microbiota appears to be relatively resilient to alterations
caused by stressful life events and antibiotic treatments. Its nonran-
dom organization depends on both host genetics and environmen-
tal exposure of microbes [28–30], but relatively little is known
about the rules of its assembly or how the human body controls
microbiota composition [31]. Neither is much known about what
constitutes a healthy microbiome—the collection of genes in these
organisms—nor on how this in turn influences human health.
Nevertheless, evidence increasingly converges on the hypothesis
that gut microbes may shape the host metabolic and immune
systems and thus influence the development of obesity, diabetes,
and other inflammatory diseases [32, 33].
It is by now well established that the microbiota regulates the
developing immune system [34], and that it likely played a critical
role in the evolution of the adaptive immune system [35]. There is
also accumulating evidence demonstrating that the gut microbiota
can modulate brain development and thus behavior. For example, a
recent study revealed that mice raised in germ-free conditions have
significantly increased motor activity and decreased anxiety as com-
pared to mice with normally colonized intestines [36]. Further-
more, when recolonized with microbes, the developmental deficits
in the germ-free pups normalized while recolonization of adult
germ-free mice did not, thus suggesting that there is a develop-
mental window during which the microbiota is critical to brain
development.
The mechanisms by which the gut microbiota effectuate
changes in synaptic connections, which provide the essential sub-
strate for functional brain networks that underlie perception, cog-
nition, and action, are still not known. But since the microbiota has
an effect on immune cells, it seems likely that some of the effects are
mediated by signals from these cells. This interpretation is sup-
ported by evidence showing that the immune system is capable of
modulating brain function during both development and adult-
hood [37]. In addition, the vagus nerve, which plays an important
role in the transmission of immune information from gut to brain
as well as from brain to gut [38, 39], apparently also plays an
important role during development of the microbiota-brain
communication.
Given the bidirectional flow of regulatory signals between the
microbiota and the brain, it should come as little surprise that
psychological stress leads to altered intestinal barrier function
[40] and host-microbiota interactions [41]. Increasing values of
psychological stress also negatively affect the immune system [42],
as demonstrated by reduced antibody responses to vaccines

[43]. There thus appears to be a close connection between the
hypothalamic-pituitary-adrenal (HPA) axis, autonomic nervous
system, gut, kidneys, and immune system, and this connection is
mediated via cortisol, neuronal transmitters, cytokines, and
hormones [44].
The long-time observation—that infectious disease is too com-
plex to be analyzed exclusively in terms of mechanistic interactions
between the immune system and the pathogen—has thus received
rich empirical support. Adaptive systems are relational all the way,
and to understand a given interaction we have to understand a
whole lot more than the target system. To define immunocompe-
tence singularly in terms of internal molecular and cellular proper-
ties of the immune system is, accordingly, misconceived. This way
of understanding immunocompetence provides a one-sided and
thus insufficient understanding. Immunocompetence should rather
be understood as a relational property that transcends the bound-
aries of the organism. To understand immunocompetence is thus to
understand how the individual’s immune system relates to the
other adaptive systems as well as to the organism’s umwelt. Accord-
ingly, organisms may be immunocompetent despite harboring defi-
cient immune resources. And conversely, immunocompetence may
be reduced despite the presence of a well-functioning immune
system. It all comes down to how the relations develop.
6 The “Early Origins” Scheme
The relevance of external inputs, in the form of infection or stress, is

in many ways dependent upon the internal wirings of the experien-
cing organism’s adaptive systems. And since these wirings are laid
down during the developmental process, it follows that develop-
ment is of tremendous importance for the organism’s adaptability.
The now obvious idea that development is an integral part of
evolutionary biology was, nevertheless, largely ignored by evolu-
tionary biologists from about 1900 to about 1980 [45]. The sepa-
ration of the fields was so extensive that when Ernst Mayr [46] in an
influential paper discussed cause and effect in biology, he still
distinguished between evolutionary and developmental biology as
of two separate explanatory fields that differed in methods, explan-
atory projects, and concepts.
During the making of the “modern synthesis” of evolution in
the 1930s, in which Darwin’s theory of natural selection was
blended with the rediscovered Mendelian genetics, evolution was
portrayed as an interplay between mutation and selection, with the
former providing a supply of variation and the latter acting as a
fitness-based sieve [45]. In the case of unicellular organisms, this
representation is fairly accurate. But in the case of multicellular
48 Elling Ulvestad
organisms, where genes serve as modulators of biochemical and

physiological parameters that in turn influence the growth of
embryonic tissues, the effects of mutation on fitness are not directly
accessible for selection. Since selection works on phenotypes and
their functional characteristics, since development is a major deter-
minant on the multicellular organism’s phenotype, and since some
ontogenetic trajectories are better for reproducing and survival
than their competitors, development is important for the pathway
taken by natural selection.
Development impinges on evolution because it ties the organ-
ism up in a system of references to other living and nonliving
entities in between fertilization and death. Hence, organismal life
is not simply conforming to a predetermined trajectory but follow-
ing a variable path upon which developmental decisions are influ-
ential. Genes, the “master modulators” of the modern synthesis,
are thus acting more as context-sensitive difference makers than as
determining factors; genes make regulatory factors, signaling mole-
cules, enzymes, and receptors that interact with each other in highly
regulated networks, and these are all strongly modulated by epige-
netic processes, including histone modification and DNA methyla-
tion [47]. Thus, identical twins with the same genetic makeup may
turn out quite different owing to epigenetic processes and develop-
mental plasticity [48].
Epigenetic and developmental processes have been evolution-
arily selected because they adapt organisms to the environment.
But, as has been increasingly recognized, they have maladaptive
potential as well. This may occur if environmental signals, for
instance such that were required for the establishment of proper
DNA transcription or stable patterns of interaction between cells of
the adaptive systems, change in salient ways. The so-called hygiene
hypothesis, the best-reasoned theory for the epidemic-like recent
increase in allergy and autoimmunity, utilizes this explanatory
framework. According to the hypothesis, humans of today experi-
ence an absence of stimuli from microbes which are important for
the functional development of the immune system. This creates an
input-deficiency syndrome, thus leading to malfunctional develop-
ment of the regulatory cells of the immune system
[49, 50]. Although little is known about why one kind of inflam-
matory disease develops instead of another, or why it develops in
one individual but not in another, compelling evidence indicates
that the malfunctioning develops as a consequence of perturbations
to the long coevolutionary relationships between intestinal
microbes and their vertebrate hosts [51].
Not only does the hygiene hypothesis tell a story of how the
microbial umwelt affects the maturing immune system, but it addi-
tionally tells the story of how human beings affect their microbial
umwelt. Man, being an expert niche constructor [14], is capable of
changing his environment at an astonishing rate—for better and
worse. On the better side, epidemiological data from European

countries have taught us that human life expectancy was about
25 years until the mid-eighteenth century [52]. Up to that time
the leading cause of death was infectious diseases in childhood, and
so the increasing life expectancy primarily reflected progress in the
control of infectious disease: in the mid-nineteenth century by
means of hygiene, in the late nineteenth century by vaccines, and
in the mid-twentieth century by antibiotics. The adaptations were
thus of cultural type rather than adjustments of immunity by natu-
ral selection.
The downside is that the constructed niche gives rise to a
mismatch between man’s biologically derived response patterns
and environmental challenges. The westernization of society has,
for example, made food available in large quantities. And along
with better housing and health conditions, the struggle for daily
survival has almost vanished. But this change has by no means
ended life’s struggles—man has instead become increasingly sus-
ceptible to developmental aberrations and precipitation of various
diseases, including coronary heart disease, diabetes, hypertension,
as well as cognitive and psychological impairment [53, 54].
7 The Paradox of Deterioration
As of today, individuals in low- and middle-income countries in

Africa have a life expectancy of 49 years, while people in high-
income European countries may expect to live until they reach
the age of 80 years [55]. It is still the young that die in Africa—
46% of all deaths in Africa are children aged under 15 years, whereas
only 20% are 60 years or older. In contrast, only 1% of deaths in
high-income countries are in children less than 15 years, whereas
84% are aged 60 years and over. This uneven distribution of death is
matched by a similar uneven distribution of causes—while infec-
tious disease is still the major cause of death in Africa, people in
Western societies die from cardiovascular disease and cancer.
Owing to the remarkable postponement of death that has
occurred during the last 100 years, folk increasingly develop degen-
erative diseases of the adaptive systems, including diabetes, Alzhei-
mer’s disease, and immunodeficiency—death rates for people over
65 years of age compared to people aged 25–44 are, for example,
43-fold for cancer and 89-fold for pneumonia and influenza
[56]. Ageing people thus struggle with a loss of integrity, in many
ways a truly astonishing phenomenon since it suggests that the
adaptive systems, which produce and maintain themselves during
development, are unable to perform the seemingly much simpler
task of maintaining what is already formed. This paradox of deteri-
oration is a real challenge for scientists that aim to predict and
control the psychoneuroimmunological systems. Unfortunately,
50 Elling Ulvestad
the paradox’ solution provides little theoretical support for the

achievement of therapeutic control.
As summarized by Ernst Mayr [46], there are two principal
kinds of cause in biology—proximate causes that give explanations
in terms of developmental and physiological mechanisms, and ulti-
mate causes which provide explanations in terms of evolutionary
mechanisms. The two are connected by evolutionary time—the
ultimate causes shape the proximate causes. And since natural
selection is a progressive force, one would expect evolution to
shape developmental systems to near-optimal functioning. But, as
elaborated by George Williams [57], natural selection works on
genes that enhance reproduction, not longevity. And the genes
responsible for ageing may thus be kept in the gene pool by
selection on their beneficial effects to the young that possess them
and not owing to their detrimental effects in senescence. This
phenomenon, termed antagonistic pleiotropy, explains why the
selective pressure on machinery responsible for maintenance of
genomic and cellular integrity in aging tissues has been
insignificant [58].
Since infectious disease has been a major threat to the survival
of young children and thus to their reproductive potential, natural
selection should be expected to shape the immune system so as to
increase its efficiency during the early years of life. And as evidenced
by observational data, production of inflammatory mediators by
the innate immune system complies well with the antagonistic
pleiotropic framework. The importance of a highly active innate
immune system has been corroborated by comparative data
between African and European populations. The data strongly
suggest that individuals of African ancestry have a more active
inflammatory response, perhaps owing to a greater burden of infec-
tious disease [59]. Furthermore, emerging evidence indicates that
pro-inflammatory genotypes are associated with a higher incidence
of inflammatory disease in later life, including atherosclerosis, dia-
betes, and cancer [60]. The selection for a strong pro-inflammatory
immune response, which is necessary to resist otherwise fatal infec-
tions in early life, is thus—as predicted by Williams’ hypothesis—a
double-edge sword; the overproduction of inflammatory molecules
may cause inflammatory diseases and even death later in life. Natu-
ral selection thus gives rise to mechanisms that both create and
destroy the organism.
Surprising data from the last couple of years have even shown
that this overproduction may be enhanced by various cultural
“practices.” Early experiences, which can affect adult health either
by cumulative damage over time or by adversities that take place
during sensitive periods [61], can take dramatic and often unex-
pected courses. Experience of maltreatment in childhood is, for
example, a strong predictor of adult inflammation [62], and,
more specifically, increases the risk for autoimmune disease
[63]. To control such malfunctions psychoneuroimmunologists

thus have to treat culture no less than biology.
Also the adaptive immune system follows the logic laid down
by Williams, but in a modified form. Newborn children come with
immature adaptive immune systems and thus have to rely on mater-
nally derived IgG and IgA for their first 6 months of life. However,
this immunodeficiency of the young does not contradict Williams’
prediction. Adaptive systems are designed by natural selection to
mature over the life course and so their seeming failure in early life is
part of their developmental program. The same goes for their
deteriorating function with age, as evidenced by increasing tenden-
cies to autoimmunity and immunodeficiencies, and as predicted by
the antagonistic pleiotropy framework. For the adaptive immune
system this malfunctioning is partly owing to a reconfiguration of
T-cell immunity, manifesting as the accumulation of senescent and
dysfunctional cells [64], and a shift in subpopulation frequency as
well as expressed repertoire of antibodies and T-cell receptors [65].
8 Summing Up
Compelling evidence has demonstrated that early environments are

important determinants of nervous, endocrine, and immune func-
tions over the life course. As adaptive systems seem inherently
disposed to degeneration, and since the prospects of controlling
such evolutionarily selected disintegration seem dim, a major aim of
psychoneuroimmunological investigations should be to lay out
early conditions that serve to increase the integrative processes
and, of no less importance, to delay the disintegrative processes.
Such investigations should acknowledge the importance of the
experiential dimension, and should take a life-cycle perspective in
which the organism’s timely unfolding is correlated to salient envi-
ronmental contingencies. To develop, the organism needs to
extract resources from the environment, and variation in the organ-
ism’s local ecology will thus in large part determine the levels of
available resources and thus the developmental course.
References
1. Ader R (1981) Psychoneuroimmunology. Aca- 4. Wittgenstein L (1960) The blue and brown
demic Press, New York books. Harper & Row, New York
2. Good RA (1981) Foreword: interactions of the 5. Biondi M, Picardi A (1999) Psychological
body’s major networks. In: Ader R stress and neuroendocrine function in humans:
(ed) Psychoneuroimmunology. Academic the last two decades of research. Psychother
Press, New York, pp xvii–xxix Psychosom 68:114–150
3. Plaut SM, Friedman SB (1981) Psychosocial 6. Kull K (2001) Jakob von Uexküll: an introduc-
factors in infectious disease. In: Ader R tion. Semiotica 134:1–59
(ed) Psychoneuroimmunology. Academic 7. Uexküll J (1982) The theory of meaning
Press, New York, pp 3–30 (Original 1940). Semiotica 42:25–82
52 Elling Ulvestad
8. Ulvestad E (2007) Defending life. The nature 25. Sholtis SJ, Noonan JP (2010) Gene regulation
of host-parasite relations. Springer, Dordrecht and the origins of human biological unique-
9. Hoffmeyer J (2008) Biosemiotics. An exami- ness. Trends Genet 26:110–118
nation into the signs of life and the life of signs. 26. Dethlefsen L, McFall-Ngai M, Relman DA
University of Scranton Press, Scranton (2007) An ecological and evolutionary per-
10. Hill K, Kaplan H (1999) Life history traits in spective on human-microbe mutualism and
humans: theory and empiricial studies. Annu disease. Nature 449:811–818
Rev Anthropol 28:397–430 27. Pennisi E (2011) Going viral: exploring the
11. Craig AD (2009) How do you feel—now? The role of viruses in our bodies. Science 331:1513
anterior insula and human awareness. Nat Rev 28. Spor A, Koren O, Ley R (2011) Unravelling
Neurosci 10:59–70 the effects of the environment and host geno-
12. Craig AD (2004) Human feelings: why are type on the gut microbiome. Nat Rev Micro-
some more aware than others? Trends Cogn biol 9:279–290
Sci 8:239–241 29. Turnbaugh PJ, Hamady M, Yatsunenko T et al
13. Frith C (2008) Making up the mind. How the (2009) A core gut microbiome in obese and
brain creates our mental world. Blackwell Pub- lean twins. Nature 457:480–484
lishing, Oxford 30. Reyes A, Haynes M, Hanson N et al (2010)
14. Odling-Smee FJ, Laland KN, Feldman MW Viruses in the faecal microbiota of monozy-
(2003) Niche construction. The neglected gotic twins and their mothers. Nature
process in evolution. Princeton University 466:334–338
Press, Princeton 31. Gonzalez A, Clemente JC, Shade A et al
15. Sterelny K (2007) Social intelligence, human (2011) Our microbial selves: what ecology
intelligence and niche construction. Philos can teach us. EMBO Rep 12:775–784
Trans R Soc Lond Ser B Biol Sci 362:719–730 32. Nicholson JK, Holmes E, Wilson ID (2005)
16. Laland KN, Odling-Smee J, Myles S (2010) Gut microorganisms, mammalian metabolism
How culture shaped the human genome: and personalized health care. Nat Rev Micro-
bringing genetics and the human sciences biol 3:431–438
together. Nat Rev Genet 11:137–148 33. Musso G, Gambino R, Cassader M (2011)
17. Beauregard M (2007) Mind does really matter: Interactions between gut microbiota and host
evidence from neuroimaging studies of emo- metabolism predisposing to obesity and diabe-
tional self-regulation, psychotherapy, and plates. Annu Rev Med 62:361–380
cebo effect. Prog Neurobiol 81:218–236 34. Ivanov II, Littman DR (2011) Modulation of
18. Getz L, Kirkengen AL, Ulvestad E (2011) The immune homeostasis by commensal bacteria.
human biology—saturated with experience. J Curr Opin Microbiol 14:106–114
Nor Med Assoc 131:683–687 35. Lee YK, Mazmanian SK (2010) Has the micro-
19. Henrich J, Heine SJ, Norenzayan A (2010) biota played a critical role in the evolution of
The weirdest people in the world? Behav the adaptive immune system? Science
Brain Sci 33:61–83 330:1768–1773
20. Gallagher S, Zahavi D (2008) The phenome- 36. Heijtz RD, Wang S, Anuar F et al (2011) Nor-
nological mind. An introduction to philosophy mal gut microbiota modulates brain develop-
of mind and cognitive science. Routledge, ment and behavior. Proc Natl Acad Sci U S A
London 108:3047–3052
21. International human genome consortium 37. Boulanger LM (2009) Immune proteins in
(2001) Initial sequencing and analysis of the brain development and synaptic plasticity. Neu-
human genome. Nature 409:860–921 ron 64:93–109
22. Kurth R, Bannert N (2010) Beneficial and det- 38. Van Der Zanden EP, Boeckxstaens GE, de
rimental effects of human endogenous retro- Jonge WJ (2009) The vagus nerve as a modu-
viruses. Int J Cancer 126:306–314 lator of intestinal inflammation. Neurogas-
23. Venner S, Feschotte C, Biémont C (2009) troenterol Motil 21:6–17
Dynamics of transposable elements: towards a 39. Tracey KJ (2010) Understanding immunity
community ecology of the genome. Trends requires more than immunology. Nat Immunol
Genet 25:317–323 11:561–564
24. Zhang Q, Raoof M, Chen Y et al (2010) Cir- 40. Gareau MG, Silva MA, Perdue MH (2008)
culating mitochondrial DAMPs cause inflam- Pathophysiological mechanisms of stress-
matory responses to injury. Nature induced intestinal damage. Curr Mol Med
464:104–107 8:274–281
41. Söderholm JD, Yang PC, Ceponis P et al cancer and the prospects for the human race.
(2002) Chronic stress induces mast cell- Princeton University Press, Princeton
dependent bacterial adherence and initiates 53. Gluckman P, Hanson M (2005) The fetal
mucosal inflammation in rat intestine. Gastro- matrix: evolution, development and disease.
enterology 123:1099–1108 Cambridge University Press, Cambridge
42. Padgett DA, Glaser R (2003) How stress influ- 54. Osmond C, Barker DJ (2000) Fetal, infant,
ences the immune response. Trends Immunol and childhood growth are predictors of coro-
24:444–448 nary heart disease, diabetes, and hypertension
43. Li J, Cowden LG, King JD et al (2007) Effects in adult men and women. Environ Health Per-
of chronic stress and interleukin-10 gene poly- spect 108(Suppl 3):545–553
morphisms on antibody response to tetanus 55. World Health Organization (2008) The global
vaccine in family caregivers of patients with burden of disease 2004 update. WHO Press,
Alzheimer’s disease. Psychosom Med Geneva
69:551–559 56. Troen BR (2003) The biology of aging. Mt
44. McEwen BS, Gianaros PJ (2011) Stress- and Sinai J Med 70:3–22
allostasis-induced brain plasticity. Annu Rev 57. Williams GC (1957) Pleiotropy, natural selec-
Med 62:431–445 tion, and the evolution of senescence. Evolu-
45. Hamburger V (1980) Embryology and the tion 11:398–411
modern synthesis in evolutionary theory. In: 58. De S (2011) Somatic mosaicism in healthy
Mayr E, Provine WB (eds) The evolutionary human tissues. Trends Genet 27:217–223
synthesis. Harvard University Press, Cam-
bridge, pp 96–112 59. Pennington R, Gatenbee C, Kennedy B et al
(2009) Group differences in proneness to
46. Mayr E (1961) Cause and effect in biology. inflammation. Infect Genet Evol 9:1371–1380
Science 134:1501–1506
60. Licastro F, Candore G, Lio D et al (2005)
47. Reik W (2007) Stability and flexibility of epige- Innate immunity and inflammation in ageing:
netic gene regulation in mammalian develop- a key for understanding age-related diseases.
ment. Nature 447:425–432 Immun Ageing 2:8
48. Fraga MF, Ballestar E, Paz MF et al (2005) 61. Shonkoff JP, Boyce WT, McEwen BS (2009)
Epigenetic differences arise during the lifetime Neuroscience, molecular biology, and the
of monozygotic twins. Proc Natl Acad Sci U S childhood roots of health disparities: building
A 102:10604–10609 a new framework for health promotion and
49. Rook GA (2010) 99th Dahlem conference on disease prevention. JAMA 301:2252–2259
infection, inflammation and chronic inflamma- 62. Danese A, Pariante CM, Caspi A et al (2007)
tory disorders: darwinian medicine and the Childhood maltreatment predicts adult inflam-
‘hygiene’ or ‘old friends’ hypothesis. Clin Exp mation in a life-course study. Proc Natl Acad
Immunol 160:70–79 Sci U S A 104:1319–1324
50. Okada H, Kuhn C, Feillet H et al (2010) The 63. Dube SR, Fairweather D, Pearson WS et al
‘hygiene hypothesis’ for autoimmune and aller- (2009) Cumulative childhood stress and auto-
gic diseases: an update. Clin Exp Immunol immune diseases in adults. Psychosom Med
160:1–9 71:243–250
51. Fitzsimmons CM, Dunne DW (2009) Survival 64. Pawelec G, Akbar A, Caruso C et al (2004) Is
of the fittest: allergology or parasitology? immunosenescence infectious? Trends Immu-
Trends Parasitol 25:447–451 nol 25:406–410
52. Cairns J (1997) Matters of life and death. Per- 65. Miller RA (1996) The aging immune system:
spectives on public health, molecular biology, primer and prospectus. Science 273:70–74
Chapter 4
Ecological Context and Human Variation: Applying

the Principles of Biological Anthropology
to Psychoneuroimmunology
Eric C. Shattuck
Abstract
There is considerable research interest overlap between biological anthropology and psychoneuroimmu-
nology (PNI), particularly given recent anthropological interest in endocrine and immune system func-
tioning over the life span and in different environmental contexts. In this chapter, I argue that conducting
research on non-WEIRD populations and applying an anthropological, evolutionary approach to PNI can
greatly strengthen our understanding of immune-endocrine-behavior connections. This chapter reviews
population-level variation in the human immune and endocrine systems, as well as genetic and environ-
mental contributions to this variation. The effects of culture on shaping health outcomes and stress
responses are also considered. Finally, this chapter discusses some noninvasive sampling methodologies
appropriate to field research and alternatives to laboratory-based research designs. By confronting variable
social and environmental contexts, PNI can greatly expand on its existing contributions to the treatment
and understanding of depression, mood disorders, stress, and other aspects of health and well-being.
Key words Psychoneuroimmunology, Life history theory, Human biological variation, Ecoimmunol-
ogy, Hormones, Stress, Culture, Methods
1 Introduction
Over the past three decades or so, researchers working under the
umbrella of psychoneuroimmunology have made many fundamen-
tal contributions to our understanding of basic human physiology
and psychology. Foremost of these is the realization that the
immune and endocrine systems regularly cross-communicate, and
the discovery of the bidirectional effects of mood on immune
function. Although these advances have reshaped how we think
about immunity and stress, it must be acknowledged that this
research has almost uniformly been conducted in Western countries
utilizing Western study participants. This is, of course, a matter of
convenience and all discoveries must begin somewhere before find-
ings can be generalized across a range of conditions. Yet what can
55
56 Eric C. Shattuck
we truly conclude about the human brain-behavior-immunity

nexus without a better understanding of how it operates outside
of Western laboratories?
During the same decades, biological anthropology has made its
own discoveries about human physiology (including the immune
and endocrine systems), growth, and development as well as the
causes and consequences of variation in these same domains. Addi-
tionally, anthropology has a traditional focus on the ways that
culture can influence biology. Anthropologists have recently
embarked on their own research into the immune and endocrine
systems with research questions frequently based on evolutionary
theory. As such, these two fields are entirely complementary. Com-
bining the two approaches would significantly strengthen both.
In this chapter, I highlight some sources of human biological
variation of possible interest to PNI researchers and review anthro-
pological contributions to our understanding of the same. I also
discuss the possible ways in which culture could affect the function-
ing of psychoneuroimmunological systems. Finally, I also touch on
noninvasive sampling methodologies appropriate to field research
and alternatives to laboratory-based research designs.
It must be said that other authors have called for an apprecia-
tion of anthropological contributions to PNI in the past
[1, 2]. Others have recognized the necessity of considering ecolog-
ical and evolutionary contexts [3, 4]. My goal is to build on this
previous work. Additionally, because psychoneuroimmunology is
comprised of researchers from many disciplines with highly diverse
interests, it may be that some readers are already familiar with the
data, questions, or approaches presented here. For those readers, I
hope that this chapter reinforces the importance of considering
human biological variation in their research. For unfamiliar readers,
I hope that it inspires new research questions and serves as a starting
point for productive collaborations.
2 Human Biological Variation, Biological Anthropology, and Life History Theory
Since its beginnings, anthropology has concerned itself with

making sense of human variation, both cultural and biological.
Developments through the twentieth century saw a growing inter-
est in the contributions of environment to human biology (e.g.,
[5]), often within an adaptationist perspective [6]. This approach
recognizes the significant duration of human residence in various
environments and posits that these groups have developed various
biological and cultural responses to local challenges. Any biological
adaptations can be plastic (i.e., arise during development and mod-
ified by the current environment) or encoded in the genome. High-
altitude adaptations are a particularly good example of both. At the
developmental level, individuals raised at high altitude grow larger
Ecological Context and Human Variation: Applying the Principles of. . . 57
lungs [7], while populations that have lived at high altitude for
generations, such as native Tibetans, show evidence of natural
selection on genes, including EPAS1 and PPARA, that affect eryth-
ropoiesis and vasculogenesis, among other effects [8].
Similar evolutionary logic has been applied to PNI as well.
Chronic inflammatory diseases have been attributed to a “mis-
match” between ancestral human states and current lifestyles [9];
depression has been put forth as an evolved mechanism to limit
contact with, and improve recovery from, infectious disease [10];
stress reactions, via glucocorticoids, drive leukocyte migration to
peripheral tissues in anticipation of wounding and subsequent
pathogen contact [11]. What is often missing from this adaptation-
ist approach is a consideration of short-term, variable ecological
contexts.
Life history theory, a currently booming topic in biological
anthropology and ecoimmunology, provides just such a theoretical
framework. A major focus of life history theory is an individual’s
partitioning of limited resources (i.e., time and energy) into
growth, somatic maintenance (including immune function), stor-
age, or reproduction. Resources allocated to one category cannot
be used in another which gives rise to a number of life history trade-
offs, such as stunted growth following infection due to temporary
overinvestment in immune responses [12]. Optimal life history
strategies to maximize reproductive fitness are numerous and
highly context dependent, varying based on life stage and local
ecologies. Anthropology adds another constraint to this model,
namely culture [13], which sets boundaries for appropriate beha-
viors. With this perspective, human behavior is seen as tremen-
dously variable across both time and space, which correctly
implies that findings from one population (or even at different
time points in the same population) are not likely to be relevant
to others. This simple truth encourages researchers to not overgen-
eralize, though this is unfortunately sometimes forgotten. It is
certainly true that many biological traits have evolved to meet a
universal need, but the assumption that these traits operate the
same everywhere should be questioned, particularly in the case of
those traits that have significant effects on health or are intimately
involved with internalizing external environmental cues, as is the
case for both the endocrine and immune systems.
3 Contrasting Populations: WEIRD and Anthropological
Henrich and colleagues’ important review introduced the concept

of “WEIRD” (Western, Educated, Industrialized, Rich, and Dem-
ocratic) populations to psychologists and allied researchers
[14]. The authors outlined some of the ways that this population
differs from others across the world. For instance, results from
58 Eric C. Shattuck
visual perception tasks are markedly different when comparing

WEIRD individuals to San foragers from the Kalahari; spatial cog-
nition could be affected by fundamental differences in language
systems; concepts of personal independence differ between Western
and non-Western subjects, and so on. WEIRD research participants
are, of course, convenient to sample and the necessary laboratory
equipment is often difficult or impossible to transport to the field.
However, anthropologists have effectively met these challenges and
contributed enormously to our understanding of human biology
and evolution over the years.
Major long-term projects include, among others, the Tsimane
Health and Life History Project [15], demographic and ecological
research among the Hadza of Tanzania [16, 17] and the Aché in
Paraguay [18], and a longitudinal study of stress and health in
Dominican children [19]. These projects, representing decades of
research and the effort of dozens of anthropologists and local
collaborators, have provided tremendous insight into the effects
of different environments, diets, and social structures on the
demography and socio-ecology of contemporary small-scale popu-
lations as well as human life histories and evolution [15]. These
groups can allow interested researchers to probe both the adaptive
and contextual nature of PNI theories and physiological systems,
such as an exploration of depression as a host defense against
pathogens in the Tsimane [20]. Other major findings of interest
include consistently increased levels of immune biomarkers, relative
to Americans, through the life span [21] and earlier development of
anti-helminthic immune defenses in childhood in the Tsimane
[22], as well as the critically important relationships between family
environment, stress, and health outcomes in children
[19, 23]. More generally, anthropological research has consistently
found differences in endocrine and immune function relative to
Western populations. Whether these differences affect the func-
tioning of the immune-endocrine-behavior axis remains unknown,
but it is a fertile area for research.
4 Biological Differences Between Populations
4.1 Population With the advent of cheaper and less technologically burdensome
Differences endocrinological assays such as enzyme immunoassays (EIAs), and
in Hormone Levels particularly with the development and validation of salivary EIAs,
biological anthropologists have cemented their interests in hor-
mones as a mechanistic link between health outcomes and several
related domains, particularly energetics and growth/development.
A critically important anthropological contribution to the wider
field of endocrinology has been the discovery of population-
dependent variation in hormone levels and the exploration of pos-
sible determinants of this variation.
Ellison and co-authors compared male salivary testosterone

levels between the Congolese Lese horticulturalists, Nepalese
Tamang agropastoralists, Ache foragers from Paraguay, and resi-
dents of Massachusetts, four populations with large genetic, life
history, and ecological differences [24]. While members of each
group exhibited age-related declines in testosterone levels (i.e.,
andropause), levels varied significantly based on age. Levels in
men over 45 converged to nonsignificance, while there was marked
variation in younger men. In this latter group, American men had
the highest mean values while Ache men had the lowest [24]. It
should be noted that similar age-related differences have been
found within the USA as well. While Mexican-American males in
the NHANES III dataset had higher testosterone than Caucasian
or African-American men, younger (20–44 years old) Caucasians
had lower total testosterone levels than the other two ethnicities;
older (70+ years) African-Americans had the lowest levels by a
significant margin [25]. These differences may arise early in life.
In a sample of 12–15-year-old males from the NHANES III data-
set, Mexican-American adolescents had the highest levels of testos-
terone, while African-Americans had the lowest [26].
While sex steroids have been widely studied within an anthro-
pological framework, other hormones show similar differences
between ethnicities/populations. Leptin, a hormone central to
signaling energetic states and one that may play a role in shaping
fever responses [27], was shown to be highest in Asian-Indian men
and women, relative to their European and Creole (mixed Mala-
gasy/African descent with Indian and European admixture) coun-
terparts [28]. The Ache of Paraguay are characterized by extremely
low leptin levels, generally equivalent to those of anorectic Western
individuals, yet have higher adiposity than healthy Western
subjects [29].
Whether these differences in hormones have any evolutionary,
or indeed functional, meaning is an open question in many cases.
Given the importance of hormones across many physiological
domains, it is likely that population variation in testosterone, leptin,
and other hormones does not affect normal physiological
(to include linkages between hormones and immune function)
functioning. Basal hormone levels are likely influenced more by
local resource availability and life history strategies than genetic
factors.
Nevertheless, more research is needed. For instance,
corticotropin-releasing hormone (CRH) stimulates adrenocortico-
tropic hormone (ACTH) and subsequent cortisol release. Given this
important role, it is perhaps not surprising that the CRH gene
promoter region is highly similar across disparate species, suggesting
strong conservation pressure [30]. Yet there are significant allelic
differences between major population groups (Caucasian, black
60 Eric C. Shattuck
South African, East Asian, Cameroonian, etc.) and concomitant sig-

nals of strong disruptive selection, perhaps due to infectious disease
pressure [30].
4.2 Population Inflammation varies based on ethnicity within and between

Differences countries as well. C-reactive protein (CRP) levels in American
in Inflammation women were highest in African-American women while Chinese
and Immunity or Japanese ethnicity was negatively associated with CRP
[31]. CRP levels in the Tsimane during early life are higher than
several other populations [32] and although the Tsimane live with
high pathogen and parasite prevalences, elevated CRP does not
appear to be wholly related to acute infections [21].
In comparative studies using the Bacillus Calmette-Guerin
(BCG) vaccine, several population differences in adaptive immune
responses have been observed. In Gambian and Malawian infants, T
cells show strong Th1 responses dominated by IFN-γ, while Indo-
nesian infants show additional IL-5 and IL-13 responses which are
not seen in Malawian, Gambian, South African, or British children
[33, 34]. While timing of vaccine administration may explain some
of these differences in immune response, associations with immu-
nogenicity are still highly variable which suggests the presence of
other explanatory variables [35]. Although genetics may play some
role, environmental and life history variables, including exposure to
environmental mycobacteria, birth seasonality, and maternal BMI,
appear to account for a large portion of this variation [35]. Other
population differences have been seen with one type of Haemophi-
lus influenzae b (Hib) vaccine that showed strong efficacy in Finnish
children, but no protective effect whatsoever in indigenous Alaskan
children, while a Hib-tetanus conjugate vaccine resulted in higher
antibody formation among South American infants than in Israeli,
European, or American children [35]. As in the case of the BCG
vaccine, environmental conditions explained this discrepancy; in
South American children, higher antibody titers were found in
children with more household members, suggesting increased
exposure to bacteria that cross-react with the Hib vaccine
[35]. On the other hand, high heritability (>90%) of differences
in immune responses to the measles vaccine suggests a genetic
basis, as does significant interindividual variability in responses to
the influenza vaccine for individuals living in the same environment
[36]. Population differences in antibody titers or duration have
been found with hepatitis B (Hep B), tetanus, diphtheria-tetanus-
pertussis-Hib (DTaP-Hib), polio, cholera, and other vaccines as
well [35].
4.3 Sources One possible explanation for population differences is genetic dif-
of Variation: ferences. Single-nucleotide polymorphisms in cytokine promoter
Population Genetics regions such as IL6-174G and TNF-308A have been shown to
increase cytokine concentrations (for extensive review of cytokine
polymorphisms, see [37]), and while research has generally focused

on associated immunological and health outcomes (e.g., [38]),
these SNPs could also affect PNI-related outcomes as well. For
instance, carriers of the high-production alleles might show more
severe inflammation-related depression or lethargy following
immune activation. These alleles may be selected for by higher
rates of pathogen pressure, as in the case of the Tsimane of
Bolivia [15].
The IL6-174 and TNF-308 SNPs have also linked with differ-
ences in affect. IL6-174C/C individuals, who produce less IL-6
than the G/C or G/G phenotypes, reported fewer symptoms of
depression during ribavirin and interferon-α treatment for hepatitis
C [39] while the G allele is associated with greater mood distur-
bance during virtual infection [40]. Other cytokine polymorphisms
have been shown to contribute to mood disturbances (IL10-592A)
and fatigue (IFNγ+874T) during infection as well [40, 41].
Studies in several populations across the globe have found that,
in general, non-Caucasian populations have much higher frequen-
cies (80–100%) of the IL6-174G allele than Caucasians [42], while
the opposite appears to be true for TNF-308A which is more
prevalent in European and Asian populations (Tables 2 and
3 from [43]). Whether this distribution represents the effects of
any selective pressures is yet to be determined.
Genetic variation is certainly not limited to immune system
genes. A number of polymorphisms in human glucocorticoid
receptor (GR) genes are garnering interest for their roles in stress
vulnerability and resilience. A SNP in the second GR intron has
been shown to increase the expression of receptors in vitro,
although they were determined to be less transcriptionally active
forms [44, 45]. Variants in the Tth111 GR gene have been asso-
ciated with higher basal cortisol levels while carriers of the N363S
variant of Bcl1 showed elevated cortisol responses to the Trier social
stress test [46]. Genetic variation related to CRH, (nor)epineph-
rine, and serotonergic pathways can also have downstream effects
on HPA axis functionality [47]. Interestingly, some of the function-
ally active GR SNPs appear to have sex-specific effects, with
increased HPA axis reactivity in males, but reduced reactivity in
females [47]. As with cytokine polymorphisms, whether there is
any selective benefit to carrying these GR variants is an open
question.
4.4 Sources Both nutritional status and pathogen exposure during childhood
of Variation: can have significant effects on the development and action of several
Nutritional Status important physiological systems, including the endocrine and
and Pathogen immune systems. In addition to negative effects on growth, high
Exposure pathogenic environments during childhood may “prime” the
immune system differently. Temporary fasting results in a dysfunc-
tional HPA stress response that can be counteracted with glucose
replacement [48], for instance, and protein-energy malnutrition
62 Eric C. Shattuck
(but not energy restriction alone) is known to be immunosuppres-

sive [49]. Because pathogen exposure throughout life in the West is
minimal, and food is (over)abundant, these domains may represent
the areas of largest differences between Western and some
non-Western populations.
As noted above, immunological profiles among the Tsimane
are quite different from Western populations and are likely attrib-
utable to their high pathogen environment. In addition to elevated
inflammation markers, eosinophils, and immunoglobulins, among
others, Tsimane show higher NK cell counts beginning in child-
hood and increasing thereafter [21]. This is in contrast to
age-related declines in Western populations. Whether these altered
immunological profiles and elevated inflammatory states have any
effects on mood or behavior remains to be seen. Considering the
tight links between inflammation and depression, a reasonable
hypothesis might be that Tsimane experience more or worse
inflammation-mediated depression symptoms than Western coun-
terparts. However, it is notable that other conditions associated
with chronic inflammation, namely atherosclerosis and diabetes, are
largely absent in the Tsimane [50] suggesting that they may be
more resilient to the negative effects of inflammation, likely a
critical developmental strategy in such an environment.
Interestingly, data from the Philippines show that adult CRP
levels are negatively associated with markers of childhood pathogen
exposure, but not morbidity measures [51]. Cord blood-derived
antigen-presenting cells (APCs) derived from Australian infants
have also been shown to be more functionally active following
stimulation than APCs from Papuan infants [52]. This lends sup-
port to the “old friends” hypothesis of evolutionary medicine,
which states that regular exposure to relatively benign pathogens
in childhood shapes the development of immune responses [51]. In
this regard, our Western immune responses may actually reflect
some degree of dysregulation. Indeed, elevated CRP was only
associated with perceived stress in Filipino adults who had fewer
microbial exposures in early life, a situation akin to Western chil-
dren rather than to our putative ancestral state [53], and depression
was not related to either CRP or IL-6 in the same sample [54]. This
evidence, although not conclusive, highlights the need to deter-
mine the nature of immune-endocrine interactions in diverse popu-
lations and environments.
Considering the high metabolic costs of immune responses
(e.g., 7–13% of daily metabolism per degree Celsius increase in
temperature, 15–30% increase in metabolic rate due to cost of
antibody synthesis after vaccination [55]), adequate nutrition is
critically important for mounting effective defenses against patho-
gens. In addition to calorie and macronutrient content, some
micronutrients affect inflammation. Ratios of omega-6 and
omega-3 fatty acids, which have increased markedly from
approximately 2:1 or 3:1 in putative ancestral hunter-gatherer diets

to as much as 17:1 in modern Western diets [56], have been linked
with increased production of TNF-α during stressful periods
[57]. Higher ratios also interacted with higher depressive symptom
scores to predict increased IL-6 and TNF-α production [58]. Zinc
and iron are two more micronutrients critical for effective immune
function [59, 60].
It is important to note that while undernutrition has clear
effects on immune function, so too does overnutrition. Excessive
zinc consumption, for instance, impairs lymphocyte stimulation
responses and neutrophil function, and obesity is associated with
impaired cell-mediated immune responses, NK cell activity, and
decreased lymphocyte responses [60]. This could have important
implications for health among Western populations, as well as those
transitioning into Western-based diets.
5 Culture and Psychoneuroimmunology
“Culture” has been anthropology’s bailiwick since the field’s incep-

tion. Over time, definitions and understandings of culture have
shifted away from broad archetypes following a linear advancement
and universal laws to a more nuanced understanding incorporating
historical processes, environment and economies, and structures of
power. This has permitted much finer grained analysis and an
appreciation of the ways that individuals or some groups engage
with, contest, and find meaning within a given cultural context.
Such an approach is important for health research as well.
Recently, Singer and colleagues pointed out the problems with
current uses of “culture” in the biomedical and health literature
[61]. Chief among these problems is the tendency to use crude
measures, such as ethnicity or stereotypical beliefs (e.g., machismo,
fatalism), in place of true measures of cultural difference [61]. Such
measures cannot capture the dynamic, personal nature of culture.
Furthermore, it should be remembered that Western notions of
relevant domains (e.g., well-being) are not necessarily universal.
Researchers interested in the cultural forces that shape health
should therefore take care to understand the context of their
research populations and select the most relevant domains based
on their research questions [61]. This last point is particularly
salient, as a single study can never account for all of a culture’s
complexity, nor the multiple cultures that we all inhabit (e.g.,
cultures of work, home, or other social groups). Finally, question-
naires regarding beliefs of interest should be tested for content and
construct validity to ensure that respondents understand what is
being asked and measured. In spite of these caveats, researchers are
beginning to appreciate the role that shared, internalized beliefs
play in shaping health and associated behaviors.
64 Eric C. Shattuck
5.1 Cultural There is a growing literature on ethnic/“cultural” differences in

Differences experiences of depression and stress. Of particular interest are
in Concepts of Stress differences in somatization (i.e., the propensity to experience psy-
and Depression chological symptoms as physical complaints) between groups. For
instance, Deisenhammer and colleagues report differences in soma-
tization rates in clinically depressed Turkish and Austrian women
[62]. Higher somatization in Asian or Asian-descended individuals
has also been reported in the literature and may be an explanation
for low rates of depression in some Asian countries, particularly if
there is a cultural bias toward reporting physical, rather than affec-
tive, symptoms of depression [63]. If the survey(s) used to assess
depression, either in a clinical or research setting, do not account
for this variance, this raises obvious questions of validity. Addition-
ally, group differences in stigma related to mental illness could
easily affect willingness to self-identify as depressed [64] or express
behaviors and attitudes characteristic of that condition.
Some exemplary work illustrating the power of perception on
stress comes from work on skin coloration in Puerto Rico. Ethnic
differences in blood pressure and hypertension are well known, and
many researchers have proposed a genetic component to explain
the variance. Gravlee and Dressler found that the discrepancy
between self-perceived color (a cultural construct encompassing
several factors related to social classification, including skin color,
hair texture, facial features, wealth, and family background) and
objective skin pigmentation was associated with higher systolic
blood pressure in Puerto Ricans [65]. Notably, skin pigmentation
alone was not associated with blood pressure. Taking this model a
step further, Gravlee et al. report that the effects of color on blood
pressure are also independent of genetic markers of ancestry
[66]. The effects of social classification systems are clearly powerful
enough to shape health outcomes.
There is also evidence that culture can shape hormonal
responses to stress, beginning at a very early age. A comparison of
both behavioral responses and cortisol levels following vaccination
found that although Caucasian-American infants displayed greater
affect and a longer latency to quiet after inoculation, Japanese
infants had a higher increase in cortisol [67]. The authors suggest
that these differences could be due to socialization practices, such
as the tendency for Japanese parents to keep their children closer.
Increased physical distance between Caucasian-American infants
and their parents could necessitate a greater physical and vocal
signaling of discomfort, which could reduce cortisol reactivity in
turn. Differences in parenting styles, namely the degree of psycho-
logical control imposed on the child by the parent, appear to
influence post-stressor cortisol secretion in children across cultures,
though this parenting style is more accepted in East Asia
[68]. These biological differences in stress reactivity may persist
into adulthood. Older adult Brazilians showed a greater increase in
cortisol levels after the Trier social stress test than their Canadian
counterparts, although a number of different factors, including
socioeconomic status, could contribute to this finding [69]. Fur-
thermore, perceived racial discrimination during adolescence pre-
dicted flatter diurnal cortisol slopes in both African-American and
Caucasian-American adults, as well as lower waking and average
cortisol levels in African-American adults only [70].
5.2 Cultural “Health,” “disease,” and allied concepts are known to vary between
Differences groups as well. Although they were speaking specifically about
in Conceptions mental illness, Rao and colleagues note that diagnoses are made
of Health based on deviations from sociocultural and behavioral norms
[71]. The same reasoning could apply to other health states as
well. Physical illness or disability is defined in relation to normative
states that certainly differ between individuals; cultural norms and
practices can shape what is defined as “normal.”
Different conceptions of sickness and health may translate into
different subjective illness experiences and, subsequently, variation
in “objective” clinical measures or other patient assessment out-
comes. For instance, there is a considerable literature on ethnic
differences in pain responses. A recent meta-analysis found that
African-American individuals reported lower cold pain thresholds
and tolerances, as well as higher cold pain intensity than Caucasians
[72]. African-Americans also exhibited lower thresholds and toler-
ances for pain elicited through pressure and electrical stimuli.
Biological mechanisms that may account for these differences
include genetic [73, 74] and hormonal [75] contributions to pain
tolerance and regulation. There is also a considerable role for
sociocultural and psychological factors in shaping pain responses.
These include spiritual or religious beliefs, socioeconomic status,
socialization/learned responses, mood, coping strategies, and the
like [72]. Some groups, including Mexicans, Mexican-Americans,
and Quichuas, traditionally value stoicism in the face of pain
[76, 77]. Other factors, such as the belief that illness or pain is a
test from God and must therefore be endured, have been shown to
affect treatment seeking [78].
6 Field Methods
6.1 Sample As mentioned above, biological assay technologies have been devel-
Collection Outside oping rapidly, making research on hormones and immunological
the Lab factors far more feasible than in the past. Although blood (i.e.,
plasma or serum) is still the gold standard for biological samples,
it is now possible to quantify many biomarkers in less invasive
matrices, such as saliva and urine. This has several practical benefits,
including ease of collection, reduced supplies cost, increased par-
ticipant compliance, and reduced exposure to potentially
66 Eric C. Shattuck
pathogenic material. In terms of field research, such noninvasive

collection could make it easier to collect samples in groups wary of
giving blood for fear of exploitation and in circumstances where
centrifugation or even electricity is not readily available.
For example, the Havasupai of Arizona believe that blood has
great spiritual worth, and that samples kept in laboratory biobanks
can prevent their donor’s spirit from transitioning to the afterlife.
Havasupai samples originally collected for diabetes research were
later used, without permission, for population genetics research.
The conclusions of this research were religiously offensive to the
tribe and this incident (and subsequent legal battles) has engen-
dered mistrust toward scientists, doctors, and medical procedures.
It is critically important that researchers work closely with their
participants to address concerns of negligence or abuse and respect
local culture and history [79], even if collecting noninvasive
samples.
A critical caveat to the use of noninvasive samples is that, unlike
blood, they may not represent circulating, central levels of the
biomarker of interest. For instance, while salivary enzyme immu-
noassays for IL-6 are commercially available, levels of the cytokine
in saliva may be reflective of oral health, inflammation, or smoking
habits [80, 81] rather than systemic conditions. Another example is
the measurement of peripheral oxytocin in saliva or urine, which
McCullough et al. critique extensively in a recent review [82].
A possible “middle ground” of sample collection is the use of
dried blood spots (DBS). No venipuncture is needed, and only
small volumes are collected. Following a finger stick with a sterile,
disposable lancet, up to five drops of blood (with a volume of
approximately 50 μL per drop) are transferred to filter paper (What-
man #903) [83]. Neither researchers nor field assistants need med-
ical training for this method, and participants may be able to collect
their own samples in their own homes. Samples do not need
separation, centrifugation, or further processing and immediate
freezing is not necessary. Samples can be sealed in
gas-impermeable plastic bags and shipped at ambient temperature,
with far fewer regulations than those associated with frozen whole
blood, blood products, saliva, etc. [83]. Despite these advantages,
blood spots are by no means the perfect sampling strategy. Small
total collection volumes may limit the possibility of multiple assays,
and the process of drying can affect the integrity of some cell types
and analytes [83]. Furthermore, because many commercial assays
require plasma or serum, adjustments to protocols may need to be
made and validated, or new in-house assays developed [84]. Never-
theless, DBS have been used to measure several markers of interest
to PNI researchers, including cortisol, EBV antibodies, and
CRP [84].
The use of saliva samples is likely well known to PNI research-
ers, given the widespread use of methodologies such as cortisol-
awakening responses (CAR). Several hormones (particularly steroid

hormones) can be easily measured in saliva. These hormones repre-
sent the free hormone fraction in the body [85, 86]. Saliva is easy to
collect; participants can use a small plastic straw to gently push
saliva into a cryovial or a commercially available product such as a
Salivette® swabs. Citric acid can stimulate saliva production, but
may interfere with some assays, including testosterone [85]. The
passive drool method is preferred to spitting saliva into a tube, as
this dramatically increases the number of bacteria in the sample
[85]. Specimens should be frozen as soon as possible to prevent
analyte degradation and bacterial growth and protease activity.
Rates of degradation will depend on the analyte of interest. Kirsch-
baum and Hellhammer report no significant reduction in salivary
cortisol levels after samples were kept at 20 C for a month [86]. -
Freeze-thaw cycles should be minimized, again to prevent
degradation.
Commercial assays exist for several immune markers such as
salivary immunoglobulin-A (sIgA), CRP, IL-6, and IL-1β, but
because saliva also contains several immune factors, such as com-
plement and lysozyme, it is possible to use whole saliva as a non-
specific measure of innate immunity. To that end, Muehlenbein and
colleagues have recently developed and validated a salivary bacterial
killing assay [87]. In contrast with measuring nonspecific measures
of inflammation such as CRP or IL-6, such measures give a better
sense of actual immune function. Readers interested in similar
functional assays are referred to Demas and co-authors’ excellent
review [88]. As noted above, it is important to remember that
salivary measures of inflammation and immunity may reflect oral,
rather than systemic, health. Nevertheless, some specific diseases,
including HIV, hepatitis, and dengue, can be diagnosed through
saliva [89].
Other noninvasive methods of sampling include collecting
urine and, recently, hair. If using urine samples, it is necessary to
assay creatinine levels to account for variation in urine volume and
concentration. Analyte values are then corrected by dividing by
creatinine values. As with saliva, urine should be frozen as soon as
possible to ensure stability. Collection methods are simple, though
participants should clean around their urethra, and urine can be
collected in large quantities. An additional benefit of urine is the use
of dipsticks to quickly measure specific gravity, pH, hematuria, and
other measures of possible interest, including those related to
metabolism or body condition [90]. Unlike blood or saliva, urinary
hormone concentrations will reflect secretion over a given sampling
period or since last urination.
Within the last few years, researchers have developed the use of
hair for hormone analysis, the benefit of which is the long-term
index of hormonal activity [91]. Although there is still some ques-
tion about the precise physiology of how hormones move into hair
68 Eric C. Shattuck
follicles, plasma cortisol has been shown to correlate with concen-

trations in hair [91]. Cleaning products and hair treatments, water,
and UV radiation exposure may contribute to cortisol degradation
in distal areas of the hair strand and so should be controlled for
[92]. Additionally, while hair color and texture do not appear to
affect hormone levels, ethnicity (which can determine hair growth
rate) is understudied as a possible covariate [92].
Sample collection is simple. Hairs are cut or shaved close to the
skin, generally from the posterior vertex of the scalp. Cutting or
shaving instead of pulling ensures that no follicles are included in
the sample and prevents any contamination by blood [91]. Hairs
are minced or ground and hormones are extracted with methanol
and reconstituted after evaporation. Depending on which section
of the strand is measured (i.e., closer or further from the scalp), one
can obtain monthly hormone measures [91, 93]. Unlike liquid
matrices, hair samples can be stored at room temperature for
extended periods, a distinct benefit in fieldwork, and because of
the long time frame of hair hormone deposition it is unlikely that
levels will be affected by immediate conditions [94]. In addition to
cortisol, other hormones such as DHEA, testosterone, and estra-
diol have been recovered from hair samples via LC-MS and immu-
noassay methods [94–96]. As with saliva and urine, hormones
found in hair represent the free, unbound fraction.
6.2 Alternatives While entirely possible, stimulating an immune response outside of

to LPS: In Vivo Immune a controlled laboratory or clinical settings carries with it a number
Stimulation Outside of complications. Sheldon Cohen and colleagues have famously and
the Lab successfully used rhinovirus and influenza inoculation to address a
number of questions, including the effects of social ties [97],
sociability [98], stress [99], and childhood socioeconomic status
[100] on disease resistance. Although a live, replicating pathogen
combined with a known time of exposure is an exceptional means of
following the course of an infection, participants were kept in
isolation which can unfortunately limit the research questions that
can be answered.
A related methodology, and one that does not require partici-
pant isolation, is to use commercial vaccines to elicit inflammation
and immune responses (see [101] for a review of vaccine research in
PNI). Combined with noninvasive sampling, vaccination could
theoretically substitute for natural infection and model the effects
of inflammation on mood and behavior. Caution with using sali-
vary, rather than serum, measures of some cytokines is warranted.
Only 3 (IL-6, IFN-γ, and MIP-1β) of 27 cytokines analyzed in one
study showed any significant correlation between salivary and
blood samples [102].
As an additional benefit, vaccination is often part of normal
health care, and some groups (e.g., international travelers, veter-
inarians, and military personnel) are urged or even required to
receive vaccines, making them convenient for sampling. However,

vaccines have a short window of immunological activation; depend-
ing on the biomarker, levels generally peak within a few hours
following inoculation and tail off within 24 h [103]. This window
unfortunately limits the timing of sample collection. Furthermore,
if research questions are focused on behavioral or affect outcomes,
care should be taken when recruiting participants from organiza-
tions such as veterinary schools or the military, since institutional
requirements (e.g., seminars, physical training) may minimize
variability in these outcome measures [104]. Finally, data from
the Philippines suggests that inflammatory responses to vaccines
are modulated by baseline inflammatory markers and a recent
history of infectious disease symptoms [105]. These two factors
may be important confounding variables in areas of regular, fre-
quent exposure to pathogens. In spite of these caveats, vaccine
studies could be an excellent way to explore PNI in the field.
Another option is to recruit participants with active infections.
This strategy would ostensibly have the same benefits of controlled
inoculation with a live pathogen, although it is impossible to con-
trol the precise time of infection. Furthermore, obtaining the nec-
essary sample size may be difficult. In one study of the effects of
malaria infection on human testosterone levels in Honduras, the
researcher spent 6 months at a local clinic in order to recruit the
17 cases included in the study [106]. Disease prevalence, study
personnel, surveillance methods, access to clinics/hospitals, and
even willingness to seek care can all shape recruitment strategies
and success.
7 Assessing Culture: Qualitative Data Methodologies
Collecting qualitative data has been central to anthropological

research from the beginning, as this information is critical to
anthropology’s goal of emic (that is, from the perspective of the
respondent) understandings of culture and behavior. While quali-
tative data are clearly unstandardized and subjective, they may be
useful in some situations, such as when examining highly contex-
tual and nuanced topics. Interviews, ranging from formal to infor-
mal, or written text can easily provide more information than
surveys, and participants may be more willing to speak with the
investigator or write about their emotions than to complete a
lengthy survey. These initial interviews could assist in generating
new hypotheses or in constructing new survey instruments, or even
be units of analysis themselves. An example of the utility of qualita-
tive data in PNI research is Gravlee’s use of cultural consensus
methods to generate a novel measure of the color construct [107],
later used to find that perceptions of skin color and discrimination
better predict health outcomes than objective measures of color
70 Eric C. Shattuck
[65]. As research accumulates indicating that perceptions (i.e., emic

understandings) are often more important than objective reality in
shaping stress and health (e.g., [108–110]), qualitative data should
be considered as a powerful companion to quantitative measures.
Several methods or methodological categories exist for system-
atically analyzing text, including discourse analysis, content analy-
sis, sentiment analysis, and natural language processing. This latter
category encompasses a number of techniques to model language,
including lemmatization, stemming, identifying nouns, and verbs.
These processes provide a foundation for further exploration,
including sentiment analysis, or quantification of attitudinal or
emotional language (e.g., neutral, positive, or negative) in a corpus.
Sentiment analysis also distinguishes between rational sentiments
devoid of emotional content and sentiment that reflects an under-
lying psychological state of mind, and the intensity thereof
[111]. In contrast to methods that require the researcher to sub-
jectively code their interviews, researchers have compiled lists of
positive or negatively valenced words [111], lending some degree
of objectivity to coding and analysis. Sentiment analysis could be a
more precise tool than standardized questionnaires in depression
research if applied to, say, participants’ journals or similar self-
reflective writing. Pennebaker and co-authors provide a masterful
review of natural language analysis and word use in psychological
research [112].
Discourse analysis is more interpretive than sentiment analysis,
and is largely concerned with how concepts are given meaning
within varying social contexts and how this meaning permits actors
to make sense of the world and operate within it [113]. Analysis
often focuses on contested definitions or understandings of these
concepts, and as such might be best applied to considerations of the
effects of social forces on health within a PNI framework. For
instance, there is a great interest in how masculinity is constructed
and conceived, particularly as this discourse relates to gender dis-
parities in health and healthcare utilization [114] and shapes indi-
vidual illness experiences [115]. There is some evidence linking
masculine sex roles to testosterone levels [116, 117]. How robust
these connections are, and whether they contribute to health out-
comes directly (e.g., the connections between elevated testosterone
and cardiovascular disease or prostate cancer) or through immuno-
logical pathways, remains unexplored. A study utilizing a discourse
analysis of participants’ ideas about, and adherence to, culturally
dominant notions of masculinity and combined with measurements
of testosterone, cortisol, and other biomarkers could be extremely
illuminating in this regard.
8 Changing PNI in a Changing World
The goal of this chapter has been to encourage the cross-pollination

of PNI with some of the methods and theories of biological anthro-
pology. A consideration of ecological and cultural contexts in shap-
ing both stress and immune responses becomes critical in a world
confronted with numerous, rapid environmental and social
changes.
Evidence from Western Samoa indicates that increased psycho-
social stress with concomitant declines in cell-mediated immune
function in children and adolescents is likely associated with West-
ernization and urban expansion [2]. The island of Utila, Honduras,
also demonstrates some of the potential health effects of immigra-
tion, development, and inequalities often inherent therein. An
influx of immigrant mainland Hondurans to the island has con-
tributed to substantial socioeconomic disparities and marginaliza-
tion [118]. Measures of access to valued material goods (e.g., car,
television, home ownership), which reflect a discrepancy between
what individuals possess and what they believe necessary for a
sufficient lifestyle, predicted cortisol levels, such that individuals
with perceived discrepancies demonstrated blunted diurnal cortisol
slopes and AUCs [118]. In turn, immigration status and poor
access to sanitation both predicted higher perceived lifestyle dis-
crepancies [118]. Similarly, Epstein-Barr virus antibodies were pos-
itively associated with lifestyle incongruity markers among a sample
of the indigenous Yakut of Siberia, who were undergoing a shift
toward a market economy [119].
In addition to socioeconomic status, a number of other factors
can affect immigrant health. As discussed above, increased malnu-
trition can have strong effects on immune function, and changing
body composition following immigration to Western countries
appears to contribute to altered hormone profiles [120]. The act
of migration itself is, of course, often highly stressful as is the
lengthy process of acculturation. Mexican-American immigrant
women, for instance, seem to suffer from competing demands of
Mexican and American cultural traditions, which value mother-
hood and a dual role as both mother and worker, respectively
[121]. The stress of being caught between these cultures may affect
both maternal health and inflammation [122], but also infant birth
weight [123].
The “natural experiments” [124] provided by the profound
social and environmental changes happening across the world pres-
ent major challenges to human health and well-being, as well as
fertile opportunities for collaboration between anthropology and
psychoneuroimmunology. By confronting variable social and envi-
ronmental contexts, PNI can greatly advance its contributions to
the understanding and treatment of depression, mood disorders,
stress, and other health domains.
72 Eric C. Shattuck
References
1. McDade TW (2001) Lifestyle incongruity, Applied evolutionary anthropology: Darwin-
social integration, and immune function in ian approaches to contemporary world issues.
Samoan adolescents. Soc Sci Med Springer, New York, pp 1–11
53:1351–1362 14. Henrich J, Heine SJ, Norenzayan A (2010)
2. McDade TW, Stallings JF, Worthman CM The weirdest people in the world? Behav
(2000) Culture change and stress in Western Brain Sci 33:61–83
Samoan youth: methodological issues in the 15. Gurven M, Stieglitz J, Trumble B et al (2017)
cross-cultural study of stress and immune The Tsimane health and life history project:
function. Am J Hum Biol 12:792–802 integrating anthropology and biomedicine.
3. Demas GE, Carlton ED (2015) Ecoimmu- Evol Anthropol 26:54–73
nology for psychoneuroimmunologists: con- 16. Marlowe FW (2010) The Hadza: hunter-
sidering context in neuroendocrine-immune- gatherers of Tanzania. University of California
behavior interactions. Brain Behav Immun Press, Berkeley
44:9–16 17. Blurton Jones N (2016) Demography and
4. Segerstrom SC (2010) Resources, stress, and evolutionary ecology of Hadza hunter-
immunity: an ecological perspective on gatherers. Cambridge University Press,
human psychoneuroimmunology. Ann Behav Cambridge
Med 40:114–125 18. Hill K, Hurtado AM (1996) Ache life history:
5. Boas F (1912) Changes in the bodily form of the ecology and demography of a foraging
the descendants of immigrants. Am Anthro- people. Aldine de Gruyter, New York
pol 14:530–562 19. Flinn MV (2008) Why words can hurt us:
6. Little M (2010) History of the study of social relationships, stress, and health. In:
human biology. In: Muehlenbein M Trevathan W, Smith EO, McKenna J (eds)
(ed) Human evolutionary biology. Cam- Evolutionary medicine and health. Oxford
bridge University Press, New York, pp 29–48 University Press, Oxford, pp 247–258
7. Frisancho AR (1977) Developmental adapta- 20. Stieglitz J, Trumble BC, Thompson ME et al
tion to high altitude hypoxia. Int J Biome- (2015) Depression as sickness behavior? A test
teorol 21:135–146 of the host defense hypothesis in a high path-
8. Simonson TS, McClain DA, Jorde LB et al ogen population. Brain Behav Immun
(2012) Genetic determinants of Tibetan 49:130–139
high-altitude adaptation. Hum Genet 21. Blackwell AD, Trumble BC, Maldonado
131:527–533 Suarez I et al (2016) Immune function in
9. Straub RH, Del Rey A, Besedovsky HO Amazonian horticulturalists. Ann Hum Biol
(2007) Emerging concepts for the pathogen- 43:382–396
esis of chronic disabling inflammatory dis- 22. Blackwell AD, Gurven MD, Sugiyama LS et al
eases: neuroendocrine-immune interactions (2011) Evidence for a peak shift in a humoral
and evolutionary biology. In: Ader R response to helminths: age profiles of IgE in
(ed) Psychoneuroimmunology. Elsevier, San the Shuar of Ecuador, the Tsimane of Bolivia,
Diego, CA, pp 217–232 and the US NHANES. PLoS Negl Trop Dis
10. Anders S, Tanaka M, Kinney DK (2013) 5:e1218
Depression as an evolutionary strategy for 23. Flinn MV, England BG (2003) Childhood
defense against infection. Brain Behav stress: endocrine and immune responses to
Immun 31:9–22 psychosocial events. In: Wilce JM (ed) Social
11. Dhabhar FS (2002) Stress-induced augmen- and cultural lives of immune systems. Routle-
tation of immune function—the role of stress dge Press, London, pp 107–147
hormones, leukocyte trafficking, and cyto- 24. Ellison PT, Bribiescas RG, Bentley GR et al
kines. Brain Behav Immun 16:785–798 (2002) Population variation in age-related
12. McDade TW, Reyes Garcı́a V, Tanner S et al decline in male salivary testosterone. Hum
(2008) Maintenance versus growth: investi- Reprod 17:3251–3253
gating the costs of immune activation among 25. Rohrmann S, Nelson WG, Rifai N et al
children in lowland Bolivia. Am J Phys (2007) Serum estrogen, but not testosterone,
Anthropol 136:478–484 levels differ between black and white men in a
13. Gibson MA, Lawson DW (2014) Applying nationally representative sample of Americans.
evolutionary anthropology to a changing J Clin Endocrinol Metab 92:2519–2525
world. In: Gibson MA, Lawson DW (eds)
26. Lopez DS, Peskoe SB, Joshu CE et al (2013) 38. Doyle WJ, Casselbrant ML, Li-Korotky H-S
Racial/ethnic differences in serum sex steroid et al (2010) The interleukin 6 -174 C/C
hormone concentrations in US adolescent genotype predicts greater rhinovirus illness. J
males. Cancer Causes Control 24:817–826 Infect Dis 201:199–206
27. Carlton ED, Demas GE (2014) Leptin med- 39. Bull SJ, Huezo-Diaz P, Binder EB et al (2008)
iates seasonal variation in some but not all Functional polymorphisms in the interleukin-
symptoms of sickness in Siberian hamsters. 6 and serotonin transporter genes, and
Horm Behav 66:802–811 depression and fatigue induced by inter-
28. Lilja M, Rolandsson O, Shaw JE et al (2010) feron-α and ribavirin treatment. Mol Psychia-
Higher leptin levels in Asian Indians than try 14:1095–1104
Creoles and Europids: a potential explanation 40. Piraino B, Vollmer-Conna U, Lloyd AR
for increased metabolic risk. Int J Obes Relat (2012) Genetic associations of fatigue and
Metab Disord 34:878–885 other symptom domains of the acute sickness
29. Bribiescas RG (2005) Serum leptin levels in response to infection. Brain Behav Immun
Ache Amerindian females with normal adipos- 26:552–558
ity are not significantly different from Ameri- 41. Vollmer-Conna U, Piraino BF, Cameron B
can anorexia nervosa patients. Am J Hum Biol et al (2008) Cytokine polymorphisms have a
17:207–210 synergistic effect on severity of the acute sick-
30. Baerwald CG, Mok CC, Fife MS et al (1999) ness response to infection. Clin Infect Dis
Distribution of corticotropin-releasing hor- 47:1418–1425
mone promoter polymorphism in different 42. Berger FG (2004) The interleukin-6 gene: a
ethnic groups: evidence for natural selection susceptibility factor that may contribute to
in human populations. Immunogenetics racial and ethnic disparities in breast cancer
49:894–899 mortality. Breast Cancer Res Treat
31. Kelley-Hedgepeth A, Lloyd-Jones DM, Col- 88:281–285
vin A et al (2008) Ethnic differences in 43. Shattuck EC, Muehlenbein MP (2015)
C-reactive protein concentrations. Clin Human sickness behavior: ultimate and prox-
Chem 54:1027–1037 imate explanations. Am J Phys Anthropol
32. Gurven M, Kaplan H, Crimmins E et al 157:1–18
(2008) Lifetime inflammation in two epide- 44. Russcher H, Smit P, van den Akker ELT et al
miological worlds: the Tsimane of Bolivia and (2005) Two polymorphisms in the glucocor-
the United States. J Gerontol A Biol Sci Med ticoid receptor gene directly affect
Sci 63A:196–199 glucocorticoid-regulated gene expression. J
33. Dujuardi Y, Sartono E, Wibowo H et al Clin Endocrinol Metab 90:5804–5810
(2010) A longitudinal study of BCG vaccina- 45. Russcher H, van Rossum EFC, de Jong FH
tion in early childhood: the development of et al (2005) Increased expression of the glu-
innate and adaptive immune responses. PLoS cocorticoid receptor-a translational isoform as
One 5:e14066 a result of the ER22/23EK polymorphism.
34. Lalor MK, Floyd S, Gorak-Stolinska P et al Mol Endocrinol 19:1687–1696
(2011) BCG vaccination induces different 46. Derijk RH, De Kloet ER (2008) Corticoste-
cytokine profiles following infant BCG vacci- roid receptor polymorphisms: determinants
nation in the UK and Malawi. J Infect Dis of vulnerability and resilience. Eur J Pharma-
204:1075–1085 col 583:303–311
35. Kollmann TR (2013) Variation between 47. Derijk RH (2009) Single nucleotide poly-
populations in the innate immune response morphisms related to HPA axis reactivity.
to vaccine adjuvants. Front Immunol 4:81 Neuroimmunomodulation 16:340–352
36. Bucasas KL, Franco LM, Shaw CA et al 48. Rohleder N, Kirschbaum C (2007) Effects of
(2011) Early patterns of gene expression cor- nutrition on neuro-endocrine stress
relate with the humoral immune response to responses. Curr Opin Clin Nutr Metab Care
influenza vaccination in humans. J Infect Dis 10:504–510
203:921–929 49. Ritz BW, Gardner EM (2006) Malnutrition
37. Gallagher G, Eskdale J, Bidwell JL (2003) and energy restriction differentially affect viral
Cytokine genetics: polymorphisms, functional immunity. J Nutr 136:1141–1144
variations, and disease associations. In: 50. Gurven M, Kaplan H, Winking J et al (2009)
Thomas AW, Lotze MT (eds) The cytokine Inflammation and infection do not promote
handbook. Academic Press, London, pp arterial aging and cardiovascular disease risk
19–57
74 Eric C. Shattuck
factors among lean horticulturalists. PLoS Americans: is there evidence for differential
One 4:e6590 item functioning of depressive symptoms?
51. McDade TW, Rutherford J, Adair L et al Psychol Med 44:2567–2578
(2010) Early origins of inflammation: micro- 64. Abdullah T, Brown TL (2011) Mental illness
bial exposures in infancy predict lower levels stigma and ethnocultural beliefs, values, and
of C-reactive protein in adulthood. Proc Biol norms: an integrative review. Clin Psychol Rev
Sci 277:1129–1137 31:934–948
52. Lisciandro JG, Prescott SL, Nadal-Sims MG 65. Gravlee CC, Dressler WW (2005) Skin pig-
et al (2012) Neonatal antigen-presenting cells mentation, self-perceived color, and arterial
are functionally more quiescent in children blood pressure in Puerto Rico. Am J Hum
born under traditional compared with mod- Biol 17:195–206
ern environmental conditions. J Allergy Clin 66. Gravlee CC, Non AL, Mulligan CJ (2009)
Immunol 130:1167–1174 Genetic ancestry, social classification, and
53. McDade TW, Hoke M, Borja JB et al (2013) racial inequalities in blood pressure in South-
Do environments in infancy moderate the eastern Puerto Rico. PLoS One 4:e6821
association between stress and inflammation 67. Lewis M, Ramsay DS, Kawakami K (1993)
in adulthood? Initial evidence from a birth Differences between Japanese infants and
cohort in the Philippines. Brain Behav Caucasian American infants in behavioral and
Immun 31:23–30 cortisol response to inoculation. Child Dev
54. McDade TW, Borja JB, Adair LS et al (2012) 64:1722–1731
Depressive symptoms are not associated with 68. Doan SN, Tardif T, Miller A et al (2017)
inflammation in younger and older adults in Consequences of ‘tiger’ parenting: a cross-
the Philippines. Evol Med Public Health cultural study of maternal psychological con-
2013:18–23 trol and children’s cortisol stress response.
55. Lochmiller RL, Deerenberg C (2000) Trade- Dev Sci 20:e12404
offs in evolutionary immunology: just what is 69. Souza-Talarico JN, Plusquellec P, Lupien SJ
the cost of immunity? Oikos 88:87–98 et al (2014) Cross-country differences in basal
56. Cordain L, Eaton SB, Sebastian A et al (2005) and stress-induced cortisol secretion in older
Origins and evolution of the Western diet: adults. PLoS One 9:e105968
health implications for the 21st century. Am 70. Adam EK, Heissel JA, Zeiders KH et al
J Clin Nutr 81:341–354 (2015) Developmental histories of perceived
57. Maes M, Christophe A, Bosmans E et al racial discrimination and diurnal cortisol pro-
(2000) In humans, serum polyunsaturated files in adulthood: a 20-year prospective study.
fatty acid levels predict the response of proin- Psychoneuroendocrinology 62:279–291
flammatory cytokines to psychologic stress. 71. Rao D, Feinglass J, Corrigan P (2007) Racial
Biol Psychiatry 47:910–920 and ethnic disparities in mental illness stigma.
58. Kiecolt-Glaser JK, Belury MA, Porter K et al J Nerv Ment Dis 195:734–744
(2007) Depressive symptoms, omega-6: 72. Rahim-Williams B, Riley JL 3rd, Williams
omega-3 fatty acids, and inflammation in AKK et al (2012) A quantitative review of
older adults. Psychosom Med 69:217–224 ethnic group differences in experimental pain
59. Fabris N, Mocchegiani E, Muzzioli M et al response: do biology, psychology, and culture
(1991) The role of zinc in neuroendocrine- matter? Pain Med 13:522–540
immune interactions during aging. Ann N Y 73. Fillingim RB, Kaplan L, Staud R et al (2005)
Acad Sci 621:314–326 The A118G single nucleotide polymorphism
60. Samartın S, Chandra RK (2001) Obesity, of the mu-opioid receptor gene (OPRM1) is
overnutrition and the immune system. Nutr associated with pressure pain sensitivity in
Res 21:243–262 humans. J Pain 6:159–167
61. Singer MK, Dressler W, George S et al (2016) 74. Nielsen CS, Stubhaug A, Price DD et al
Culture: the missing link in health research. (2008) Individual differences in pain sensitiv-
Soc Sci Med 170:237–246 ity: genetic and environmental contributions.
62. Deisenhammer EA, Çoban-Başaran M, Man- Pain 136:21–29
tar A et al (2011) Ethnic and migrational 75. Mechlin B, Morrow AL, Maixner W et al
impact on the clinical manifestation of depres- (2005) African Americans show alterations in
sion. Soc Psychiatry Psychiatr Epidemiol endogenous pain regulatory mechanisms and
47:1121–1129 reduced pain tolerance to experimental pain
63. Kalibatseva Z, Leong FTL, Ham EH (2014) procedures. Psychosom Med 62:517–523
A symptom profile of depression among Asian
76. Calvillo ER (2013) Insights on the pain expe- 89. Lima DP, Diniz DG, Moimaz SAS et al
rience in Mexican Americans. In: (2010) Saliva: reflection of the body. Int J
Incayawar M, Todd K (eds) Culture, brain, Infect Dis 14:e184–e188
and analgesia. Oxford University Press, 90. Ryan D, Robards K, Prenzler PD et al (2011)
New York, pp 49–61 Recent and potential developments in the
77. Incayawar M, Maldonado-Bouchard S (2013) analysis of urine: a review. Anal Chim Acta
We feel pain too: asserting the pain experience 684:17–29
of the Quichua people. In: Incayawar M, 91. Meyer JS, Novak MA (2012) Hair cortisol: a
Todd K (eds) Culture, brain, and analgesia. novel biomarker of hypothalamic-pituitary-
Oxford University Press, New York, pp 61–75 adrenocortical activity. Endocrinology
78. Gonzalez-Swafford MJ, Gutierrez MG 153:4120–4127
(1983) Ethno-medical beliefs and practices 92. Wosu AC, Valdimarsdóttir U, Shields AE et al
of Mexican-Americans. Nurse Pract 8:29–34 (2013) Correlates of cortisol in human hair:
79. Pacheco CM, Daley SM, Brown T et al (2013) implications for epidemiologic studies on
Moving forward: breaking the cycle of mis- health effects of chronic stress. Ann Epide-
trust between American Indians and research- miol 23:797–811
ers. Am J Public Health 103:2152–2159 93. Gao W, Stalder T, Foley P et al (2013) Quan-
80. Rathnayake N, Åkerman S, Klinge B et al titative analysis of steroid hormones in human
(2013) Salivary biomarkers of oral health: a hair using a column-switching LC–AP-
cross-sectional study. J Clin Periodontol CI–MS/MS assay. J Chromatogr B 928:1–8
40:140–147 94. Stalder T, Kirschbaum C (2012) Analysis of
81. Teles RP, Likhari V, Socransky SS et al (2009) cortisol in hair: state of the art and future
Salivary cytokine levels in chronic periodonti- directions. Brain Behav Immun
tis and periodontally health subjects. A cross- 26:1019–1029
sectional study. J Periodontal Res 95. Gao W, Kirschbaum C, Grass J et al (2016)
44:411–417 LCMS based analysis of endogenous steroid
82. McCullough ME, Churchland PS, Mendez hormones in human hair. J Steroid Biochem
AJ (2013) Problems with measuring periph- Mol Biol 162:92–99
eral oxytocin: can the data on oxytocin and 96. Yang HZ, Lan J, Meng YJ et al (1998) A
human behavior be trusted? Neurosci Biobe- preliminary study of steroid reproductive hor-
hav Rev 37:1485–1492 mones in human hair. J Steroid Biochem Mol
83. McDade TW (2013) Development and vali- Biol 67:447–450
dation of assay protocols for use with dried 97. Cohen S, Doyle WJ, Skoner DP et al (1997)
blood spot samples. Am J Hum Biol 26:1–9 Social ties and susceptibility to the common
84. McDade TW, Williams S, Snodgrass JJ (2007) cold. JAMA 277:1940–1944
What a drop can do: dried blood spots as a 98. Cohen S, Doyle WJ, Turner R et al (2003)
minimally invasive method for integrating Sociability and susceptibility to the common
biomarkers into population-based research. cold. Psychol Sci 14:389–395
Demography 44:899–925 99. Cohen S, Doyle WJ, Skoner DP (1999) Psy-
85. Chiappin S, Antonelli G, Gatti R et al (2007) chological stress, cytokine production, and
Saliva specimen: a new laboratory tool for severity of upper respiratory illness. Psycho-
diagnostic and basic investigation. Clin Chim som Med 61:175–180
Acta 383:30–40 100. Cohen S (2004) Childhood socioeconomic
86. Kirschbaum C, Hellhammer DH (1994) Sali- status and host resistance to infectious illness
vary cortisol in psychoneuroendocrine in adulthood. Psychosom Med 66:553–558
research: recent developments and applica- 101. Phillips AC (2012) The vaccination model in
tions. Psychoneuroendocrinology 19:313–333 psychoneuroimmunology research: a review.
87. Muehlenbein MP, Prall SP, Chester E (2011) In: Yan Q (ed) Psychoneuroimmunology:
Development of a noninvasive salivary mea- methods and protocols. Humana Press,
sure of functional immunity in humans. Am J Totowa, NJ, pp 355–370
Hum Biol 23:267–268 102. Williamson S, Munro C, Pickler R et al
88. Demas GE, Zysling DA, Beechler BR et al (2012) Comparison of biomarkers in blood
(2011) Beyond phytohaemagglutinin: asses- and saliva in health adults. Nurs Res Pract
sing vertebrate immune function across eco- 2012:246178. https://doi.org/10.1155/
logical contexts. J Anim Ecol 80:710–730 2012/246178
76 Eric C. Shattuck
103. Lim PW, Garssen J, Sandalova E (2016) 115. Charteris-Black J, Seale C (2009) Men and
Potential use of salivary markers for longitu- emotion talk: evidence from the experience of
dinal monitoring of inflammatory immune illness. Gender Lang 3:81–113
responses to vaccination. Mediat Inflamm 116. Pletzer B, Petasis O, Ortner TM et al (2015)
2016:6958293 Interactive effects of culture and sex hor-
104. Shattuck EC, Muehlenbein MP (2017) mones on the sex role self-concept. Front
Human sickness behavior not expressed in Neurosci 9:240
response to the rabies vaccine. Am J Phys 117. Willer R, Rogalin CL, Conlon B et al (2013)
Anthropol 162(S64):294 Overdoing gender: a test of the masculine
105. McDade TW, Borja JB, Kuzawa CW et al overcompensation thesis. Am J Sociol
(2015) C-reactive protein response to influ- 118:980–1022
enza vaccination as a model of mild inflamma- 118. Garcı́a AR, Gurven M, Blackwell AD (2017)
tory stimulation in the Philippines. Vaccine A matter of perception: perceived socio-
33:2004–2008 economic status and cortisol on the island of
106. Muehlenbein MP, Alger J, Cogswell F et al Utila, Honduras. Am J Hum Biol 34:e23031
(2005) The reproductive endocrine response 119. Sorensen MV, Snodgrass JJ, Leonard WR et al
to Plasmodium vivax infection in Hondurans. (2009) Lifestyle incongruity, stress and
Am J Trop Med Hyg 73:178–187 immune function in indigenous Siberians:
107. Gravlee CC (2005) Ethnic classification in the health impacts of rapid social and eco-
Southeastern Puerto Rico: the cultural nomic change. Am J Phys Anthropol
model of “color”. Soc Forces 83:949–970 138:62–69
108. Alcántara C, Chen C-N, Alegrı́a M (2014) Do 120. Heald AH, Patel J, Anderson SG et al (2007)
post-migration perceptions of social mobility Migration is associated with lower total, but
matter for Latino immigrant health? Soc Sci not free testosterone levels in South Asian
Med 101:94–106 men. Clin Endocrinol 67:651–655
109. Cohen S, Alper CM, Doyle WJ et al (2008) 121. Lagana K (2003) Come bien, camina y no se
Objective and subjective socioeconomic sta- preocupe—eat right, walk, and do not worry:
tus and susceptibility to the common cold. selective biculturalism during pregnancy in a
Health Psychol 27:268–274 Mexican American community. J Transcult
110. Sturgeon JA, Arewasikporn A, Okun MA et al Nurs 14:117–124
(2016) The psychosocial context of financial 122. Rosenberg N, Daviglus ML, DeVon HA et al
stress. Psychosom Med 78:134–143 (2017) The association between parity and
111. Liu B (2017) Many facets of sentiment analy- inflammation among Mexican-American
sis. In: Cambria E, Das D, Bandyopadhyay S, women of reproductive age varies by accultur-
Feraco A (eds) A practical guide to sentiment ation level: results of the National Health and
analysis. Springer, New York, pp 11–41 Nutrition Examination Survey (1999–2006).
112. Pennebaker JW, Mehl MR, Niederhoffer KG Womens Health Issues 27:485–492
(2003) Psychological aspects of natural lan- 123. Fleuriet KJ, Sunil TS (2014) Perceived social
guage use: our words, our selves. Annu Rev stress, pregnancy-related anxiety, depression
Psychol 54:547–577 and subjective social status among pregnant
113. Dunn KC, Neumann IB (2016) Undertaking Mexican and Mexican American women in
discourse analysis for social research. Univer- South Texas. J Health Care Poor Under-
sity of Michigan Press, Ann Arbor, MI served 25:546–561
114. Courtenay WH (2000) Constructions of 124. Garruto RM, Little MA, James GD et al
masculinity and their influence on men’s (1999) Natural experimental models: the
well-being: a theory of gender and health. global search for biomedical paradigms
Soc Sci Med 50:1385–1401 among traditional, modernizing, and modern
populations. Proc Natl Acad Sci U S A
96:10536–10543
Chapter 5
Neuroimmune Imbalances and Yin-Yang Dynamics

in Stress, Anxiety, and Depression
Qing Yan
Abstract
Evidences from psychoneuroimmunology (PNI) and systems biology studies support a conceptual frame-
work of “Yin-Yang dynamics” for understanding the “whole mind-body system.” The Yin-Yang dynamical
balances in the stress response networks may be critical for health and diseases, especially mental health and
psychiatric disorders. Specifically, the neuroimmune imbalances have been found as the important features
and potential biomarkers of stress, anxiety, depression, and systemic inflammation. At the system levels,
factors such as psychosocial stress and obesity, especially a leaky gut, may result in the imbalance between
regulatory and proinflammatory T cells. At the molecular and cellular levels, the imbalances in multiple
networks including the cytokine and redox pathways, immune-kynurenine networks, HPA axis, and
synaptic plasticity in the hypothalamus are the key factors in depression. The recognition of the neuroim-
mune imbalances and the restoration of the Yin-Yang dynamical balances need to become a high priority
toward the development of dynamical systems medicine for psychiatric diseases including depression and
schizophrenia.
Key words Anxiety, Depression, Dynamical, Inflammation, Immune, Psychoneuroimmunology,

Schizophrenia, Stress, Systems medicine, Yin-Yang
1 Introduction: Neuroimmune Imbalances in Stress and Anxiety
Evidences from psychoneuroimmunology (PNI) and systems biol-

ogy studies support a conceptual framework of “Yin-Yang dynam-
ics” for understanding the “whole mind-body system” (see
Chap. 1). Complex interactions among various spatiotemporal
scales such as the dynamical balances in the hypothalamic-pitui-
tary-adrenal (HPA) axis are essential in psychophysiological coher-
ence and homeostasis [1–3]. Individual variations in personality,
emotionality, and cognition may have significant impacts on stress
responses, neuroimmune functions, and systemic inflammation [3].
The integrative framework of Yin-Yang dynamics can be used
to represent the dynamical balancing, counteracting, interconnect-
ing, interdependent, complementing, and transforming factors in
77
78 Qing Yan
Table 1
Yin-Yang dynamics and neuroimmune imbalances in psychiatric disorders

Hair loss in stress TH1/TH2 cytokines [4]
Stress, anxiety Neurotransmitter, immune imbalances [5]
disorders Pro- vs. anti-inflammatory cytokines [6]
Oxidative and cytokine imbalances [7]
Internal growth vs. external stress signaling [8]
Oxidative imbalances [9, 10]
Cardiac autonomic imbalance [11]
Approach vs. avoidance on neural substrates [12]
Psychiatric disorders Imbalances in the kynurenine pathway [13]
Depression Regulatory vs. proinflammatory T cells [14]
Left vs. right DLPFC imbalance [15]
Pro- vs. anti-inflammatory cytokines [16]
Imbalance in the leptin networks [17]
NK-1R vs. NK-2R in monocytes [18]
Brain oxytocin vs. vasopressin [19]
Zinc deficiency on phospholipid/protein [20]
Depression in Estradiol/total testosterone (E2/T) [21]
obesity
Depression Imbalances in kynurenine networks [22, 23]
Imbalances in kynurenine networks; KynA/QA [24]
Neurotransmitter imbalance [25]
Monoaminergic imbalance [26]
Depression, Type 1 vs. type 2 immunity; neurotoxic vs. neuroprotective [27, 28]
schizophrenia kynurenine metabolites
MDD Glutamatergic imbalance [29]
Dopamine and immune imbalances [30]
Adaptive vs. maladaptive T cells; neurodegenerative vs. -regenerative [31]
T-helper 1/T-helper 2 cytokines; IL-1β/IL-10 [32]
Cytokine and redox imbalances [33]
Hippocampal redox imbalance [34, 35]
HIV-1, depression Cytokine and redox imbalances [36]
the complex adaptive systems (CASs) of health and diseases (see

Chap. 1). As shown in the examples in Table 1, the neuroimmune
imbalances have been identified as the important features and
potential biomarkers for psychiatric disorders including anxiety,
depression, and schizophrenia.
For example, a study of hair loss showed that high-stress con-
ditions may lead to the imbalances in the cytokine ratio of T-helper
cells (TH1/TH2) and slow hair growth ([4]; also see Table 1). The
study examined the effects of strong stressors such as exams among
medical students. The elevated levels of type 1 T-helper cell
Neuroimmune Imbalances and Yin-Yang Dynamics in Stress, Anxiety, and Depression 79
cytokines and apoptosis of epithelial cells were observed, with the

results of adaptive immunity cytokine imbalance and hair loss. In
addition, psychosocial stress such as job insecurity has been asso-
ciated with neurotransmitter imbalances, adrenal fatigue, anxiety,
and sleep disturbance [5]. These alterations may lead to various
health problems including immune deficiencies, digestive disor-
ders, headaches, and heart diseases.
At the molecular and cellular levels, the immune imbalances
between pro- and anti-inflammatory cytokines, relatively higher
levels of proinflammatory responses, and lower levels of anti-
inflammatory responses have been characterized in generalized
anxiety disorder (GAD). Studies have identified the altered ratios
of TNF-α/IL10, TNF-α/IL4, IFN-γ/IL10, and IFN-γ/IL4
among those with GAD ([6]; also see Table 1). Cytokine signaling
pathways have the essential functions in the brain and are involved
in neurotransmitter metabolism, synaptic plasticity, as well as neural
circuitry of mood. The imbalances in these neuroimmune networks
may lead to neuroinflammation, anxiety, depression, and cognitive
dysfunctions [7].
These examples demonstrate that the Yin-Yang dynamical bal-
ances in the stress response networks may be critical for health and
diseases, especially mental health and psychiatric disorders (see
Chap. 1). Healthy living cells need to maintain the balance between
such “internal growth” and “external stress” signaling in the cellu-
lar transcriptome [8]. The disturbance in the complex balances may
lead to the loss of growth or dysfunctional stress responses in living
cells.
Specifically, redox imbalances have been identified with the key
roles in the development of anxiety. Such imbalances may be fea-
tured with the higher levels of malondialdehyde, alterations of
antioxidant enzymes in erythrocytes, as well as dysfunctions of
mitochondrial proteins, inflammatory cytokines, and neurotrophic
factors [9]. These mechanisms implicate that the oxidative imbal-
ance can be the potential targets for the prevention and treatment
of anxiety disorders. For example, the elevated anxiety level has
been related to sleep bruxism (SB). The oxidant/antioxidant
imbalance has been associated with SB and suggested as a potential
biomarker ([10]; also see Table 1).
At the system level, social stress may result in cardiac autonomic
imbalance such as the alterations in heart rate variability (HRV).
The lower levels of HRV are commonly seen among the patients
with various anxiety problems including panic and post-traumatic
stress disorder (PTSD) [11]. Potential biomarkers such as the
corticotropin-releasing factor (CRF), neuropeptide Y (NPY), and
proinflammatory cytokines including interleukin-6 (IL-6) may
indicate the comorbidity between stress-associated mental disor-
ders and autonomic imbalance [11].
80 Qing Yan
In the anxiety disorders including PTSD, the features of

approach-avoidance have been suggested as resembling Yin and
Yang that are dynamic rather than static. The imbalances in the
Yin-Yang dynamical systems may lead to the alterations of the
neural substrates in the regions of amygdala, insula, and medial
prefrontal cortex [12]. Such imbalances may result in the same
symptoms even under different conditions including hyper- or
hypodopaminergic status. These mechanisms suggest that integra-
tive interventions may be necessary to restore the dynamical bal-
ance in the approach-avoidance processes ([12]; also see Table 1).
Moreover, cytokines including interferon- α (IFN-α) and
interleukin-2 (IL-2) have been applied in the therapies of cancers,
viral infections, and multiple sclerosis. However, these cytokine
therapies often cause neuropsychiatric side effects including the
symptoms of anxiety, depression, psychosis, hypomanic mood,
and cognitive dysfunctions [13]. These psychiatric problems may
come from the imbalances in the catabolic kynurenine and neuro-
transmitter networks with the alterations in the neuroprotective
functions.
Studies have shown that some proinflammatory cytokines may
increase the activities of indoleamine-2,3-dioxygenase (IDO) with
the elevation of tryptophan degradation into kynurenine
[13]. Such changes may reduce the availability of tryptophan for
the generation of serotonin, the key neurotransmitter for maintain-
ing the normal mood condition. More discussions about these
networks are provided in Subheading 2.
2 Neuroimmune Imbalances and Yin-Yang Dynamics in Depression
2.1 Identifying Understanding in the neuroimmune imbalances and Yin-Yang

the Imbalances at dynamics may enable the identification of potential systems-based
Various System Levels biomarkers and strategies for the better treatment of major depres-
sive disorder (MDD). At the system levels, factors such as psycho-
social stress and obesity, especially a leaky gut, may result in the
imbalance between regulatory and proinflammatory T cells [14].
In severe MDD and negative emotional judgment, the imbal-
ance between left and right dorsolateral prefrontal cortex (DLPFC)
has been observed ([15]; also see Table 1). Left DLPFC hypoactiv-
ity has been related to negative emotional judgment. Right DLPFC
hyperactivity has been associated with attentional modulation and
depression severity.
At the molecular and cellular levels, stress may affect the mono-
amine neurotransmitter levels and the balance between proinflam-
matory and anti-inflammatory cytokines [16]. Depression has been
characterized with altered inflammatory cytokine activities in
peripheral and hippocampal areas. In addition, the imbalances in
the leptin network and hypothalamus synaptic plasticity may be
important factors in stress-induced depression [17]. Chronic

unpredictable mild stress may lead to depression-like behavior
with lower levels of serum leptin and hypothalamic expression of
leptin receptor (LEPR), altered synaptic plasticity, and hyperactivity
of HPA axis. The imbalance in the expression of neurokinin (NK)-1
and NK-2 receptors (NK-1R and NK-2R) in monocytes has also
been associated with recurrent MDD ([18]; also see Table 1).
The Yin-Yang dynamics in the brain oxytocin and vasopressin
may be critical in the regulation of anxiety, stress coping, and
sociality. Central oxytocin may have anxiolytic and antidepressive
functions ([19]; also see Table 1). On the other hand, vasopressin
may have anxiogenic and depressive functions. The restoration of
the dynamical balances in these neuropeptide systems using social
interventions and pharmacotherapy may promote emotional health
to relieve anxiety and depression [19].
Furthermore, nutritional status is also important. For instance,
zinc deficiency has been found to cause the alterations in the blood
phospholipid-protein balance and result in depressive disorders
([20]; also see Table 1). Zinc deficiency may lead to the reduced
levels of phospholipids and structural alterations in proteins. More-
over, the imbalance in the sex hormones has the pivotal role in the
depressive symptoms among obese men [21].
Studies have observed that obese male patients with depression
had the elevated levels of estradiol and estradiol/total testosterone
ratio (E2/T) with reduced testosterone levels ([21]; also see
Table 1). Such imbalance between the two hormones may be
essential as the depressive symptoms have been linked to the sever-
ity of the alterations. These factors may become the preventive or
therapeutic targets for MDD.
2.2 The Imbalances Multiple pathways have been associated with the development of
in the Immune- depression, especially those of monoamine metabolism and neuro-
Kynurenine Networks endocrine functions. For example, the immune-kynurenine net-
works have the essential roles in the communications between
stress and neuroimmune systems in MDD ([3, 22]; also see
Table 1).
Chronic stress may elevate proinflammatory cytokine levels and
activities of IDO. The imbalances in the downstream kynurenine
pathway may have neurotoxic results in the brain with damages in
the glial-neuronal network, leading to higher risks of depression
[22]. As the important connections among stress, inflammation,
and depression, the kynurenine networks have been proposed as
the potential antidepressant therapeutic targets [3, 23].
In addition to stressful events, sleep disturbance has been
closely related to depression. Studies have connected sleep distur-
bance with higher levels of the inflammatory marker C-reactive
protein (CRP) and lower ratios of kynurenic acid (KynA)/quino-
linic acid (QA) among patients with depression [24]. With the
82 Qing Yan
elevations in the neurotoxic metabolites, the levels of the neuro-

protective compound KynA may be reduced. Such imbalances in
the networks of kynurenine metabolites may be critical in sleep
disturbance-associated depression (see Table 1).
In depression, neurotransmitter imbalances may be caused by
the higher levels of type A monoamine oxidase (MAOA) and the
lower levels of serotonin and norepinephrine in the brain
[25]. MAOA has been closely correlated with the onset and devel-
opment of neuropsychiatric disorders. In depression and social
anhedonia, monoaminergic imbalance and lower levels of cAMP
response element-binding protein (CREB) and β-arrestin-1,2 in
the nucleus accumbens have also been characterized ([26]; also see
Table 1).
Furthermore, psychiatric disorders including depression and
schizophrenia have been correlated with the imbalance between
type 1 and type 2 immune activations. These immune functions
may affect the enzyme IDO in the central nervous system (CNS) in
opposite Yin-Yang manners ([27]; also see Table 1). When the type
1 immune responses are partially inhibited, the higher type 2 reac-
tions may be recorded [3, 28].
Various studies have confirmed the importance of the glutama-
tergic imbalance and microglial activation in MDD, especially their
associations with neuroinflammatory markers and less resilience in
the neuroendocrine factors [29]. The imbalance may also have a
critical role in schizophrenia. Although different processes may be
involved in the development of depression and schizophrenia, they
may share the similar mechanisms in the inflammatory networks.
Moreover, the cytokine IFN-α has been found to cause depres-
sion, slow psychomotor, and fatigue ([30]; also see Table 1). IFN-α
treatment may result in altered cerebrospinal fluid levels of the
dopamine metabolite and dysfunctions of neural circuits related
to anxiety and alarm. It may influence IDO and cytokine signaling
pathways involving p38 mitogen-activated protein kinase. Such
changes may affect the production and reuptake of serotonin. The
imbalances in these networks may contribute to the behavioral
changes and are recognized as potential therapeutic targets.
2.3 The Imbalances T cells have the key roles in maintaining normal behaviors and
in the T-Cell Functions immune homeostasis. The imbalance between adaptive and mal-
adaptive functions of T cells, together with the imbalance between
neurodegenerative and regenerative repair activities, has been iden-
tified as the features of the neuroimmune pathways in depression
([31]; also see Table 1).
Specifically, peripheral naı̈ve T cells are involved in the regula-
tion of neural plasticity. T lymphocytes have the protective roles
against maladaptive behavioral responses linked to depression
[31]. However, psychogenic stress may affect the T cells’ activities,
and alter the productions of neurotrophic factors and depression-
associated cytokines in the brain. Therefore, T-cell functions are

critical in the dynamical balance between susceptibility and resil-
ience to the development of MDD [3].
The imbalance in the ratio of T-helper 1/T-helper 2 cytokines
is a significant feature of acute-phase MDD ([32]; also see Table 1).
Patients with MDD and melancholic characteristics may have ele-
vated levels of TNF-α, IL-1β, and IL-1β/IL-10 ratio. These altera-
tions can be potential biomarkers and therapeutic targets.
2.4 Neuro- Chronic neuroinflammation has a critical role in MDD by altering

inflammation the hippocampal functions and the activities of glucocorticoid
and Imbalances receptors (GRs) with the higher levels of proinflammatory cyto-
in the Redox System kines [33]. The hippocampal redox imbalance has been suggested
critical in the stress-induced depressive-like behavior [34, 35]. The
loss of the pro/antioxidative balance in the hippocampus may be
characterized with elevated levels of lipid peroxidation, superoxide
dismutase (SOD), and glutathione peroxidase. Such alterations
may be accompanied with the lower levels of catalase (CAT) activity
and higher SOD/CAT ratio, the marker of pro-oxidative status (see
Table 1).
For example, among HIV-1 patients with depression, altera-
tions in the oxidant/antioxidant balance have been observed ([36];
also see Table 1). These patients may have the higher levels of
proinflammatory cytokines including interleukin-15 (IL-15), inter-
feron γ-induced protein 10 (IP-10), IL-12 p40/p70, as well as
granulocyte colony-stimulating factor (G-CSF).
The hippocampal redox imbalances have been confirmed in
various studies ([34, 35]; also see Table 1). On the other hand,
folic acid therapy has been found to restore the levels of the antiox-
idant enzymes with the lower levels of lipid peroxidation in the
hippocampus. Folic acid has been suggested as the potential anti-
depressant with the possible functions in the restoration of the
redox balances.
3 Conclusion
As demonstrated by the above examples, the imbalances in the

Yin-Yang dynamics in the neuroimmune networks such as the
cytokine and neurotransmitter pathways may contribute to psychi-
atric disorders including anxiety, depression, and schizophrenia
[3]. Based on such understanding, systemic PNI profiles can be
developed by addressing the principles of complex adaptive systems
(CASs) including adaptation, self-organization, robustness, and
nonlinearity ([2], also see Chap. 1).
These principles can be integrated in the framework of
Yin-Yang dynamics for the identification of more comprehensive
and precise biomarkers for more effective prevention and
84 Qing Yan
therapeutic strategies (see Chap. 1). The recognition of the neu-

roimmune imbalances and the restoration of the Yin-Yang dynami-
cal balances need to become a high priority toward the
development of dynamical systems medicine for psychiatric
diseases.
References
1. Yan Q (2014) From pharmacogenomics and related to interferon and interleukins treat-
systems biology to personalized care: a frame- ment. Metab Brain Dis 24:55–68
work of systems and dynamical medicine. 14. Haroon E, Raison CL, Miller AH (2012)
Methods Mol Biol 1175:3–17 Psychoneuroimmunology meets neuropsycho-
2. Yan Q (2017) Toward dynamical systems med- pharmacology: translational implications of the
icine: personalized and preventive strategies. impact of inflammation on behavior. Neurop-
Pers Med 14:377–380 sychopharmacology 37:137–162
3. Yan Q (2016) Psychoneuroimmunology: sys- 15. Grimm S, Beck J, Schuepbach D et al (2008)
tems biology approaches to mind-body medi- Imbalance between left and right dorsolateral
cine. Springer, New York prefrontal cortex in major depression is linked
4. Peters EMJ, Müller Y, Snaga W et al (2017) to negative emotional judgment: an fMRI
Hair and stress: a pilot study of hair and cyto- study in severe major depressive disorder. Biol
kine balance alteration in healthy young Psychiatry 63:369–376
women under major exam stress. PLoS One 16. Szelényi J, Vizi ES (2007) The catecholamine
12:e0175904 cytokine balance: interaction between the brain
5. Head KA, Kelly GS (2009) Nutrients and bota- and the immune system. Ann N Y Acad Sci
nicals for treatment of stress: adrenal fatigue, 1113:311–324
neurotransmitter imbalance, anxiety, and rest- 17. Ge J-F, Qi C-C, Zhou J-N (2013) Imbalance
less sleep. Altern Med Rev 14:114–140 of leptin pathway and hypothalamus synaptic
6. Hou R, Garner M, Holmes C et al (2017) plasticity markers are associated with stress-
Peripheral inflammatory cytokines and induced depression in rats. Behav Brain Res
immune balance in generalised anxiety disor- 249:38–43
der: case-controlled study. Brain Behav Immun 18. Bardelli C, Amoruso A, Manzetti E et al (2013)
62:212–218 Recurrent major depressive disorder: imbal-
7. Salim S, Chugh G, Asghar M (2012) Inflam- ance of neurokinin (NK)-1 and NK-2 receptor
mation in anxiety. Adv Protein Chem Struct expression in monocytes. Pharmacol Res
Biol 88:1–25 68:24–30
8. Han HW, Ohn JH, Moon J, Kim JH (2013) 19. Neumann ID, Landgraf R (2012) Balance of
Yin and Yang of disease genes and death genes brain oxytocin and vasopressin: implications for
between reciprocally scale-free biological net- anxiety, depression, and social behaviors.
works. Nucleic Acids Res 41:9209–9217 Trends Neurosci 35:649–659
9. R K, D M A, C N, S N W, C D (2014) Oxida- 20. Depciuch J, Sowa-Kućma M, Nowak G et al
tive imbalance and anxiety disorders. Curr (2017) The role of zinc deficiency-induced
Neuropharmacol 12:193–204 changes in the phospholipid-protein balance
10. Kara MI, Yanık S, Keskinruzgar A et al (2012) of blood serum in animal depression model by
Oxidative imbalance and anxiety in patients Raman, FTIR and UV-Vis spectroscopy.
with sleep bruxism. Oral Surg Oral Med Oral Biomed Pharmacother 89:549–558
Pathol Oral Radiol 114:604–609 21. Monteagudo PT, Falcão AA, Verreschi ITN,
11. Wood SK (2014) Cardiac autonomic imbal- Zanella M-T (2016) The imbalance of
ance by social stress in rodents: understanding sex-hormones related to depressive symptoms
putative biomarkers. Front Psychol 5:950 in obese men. Aging Male 19:20–26
12. Stein MB, Paulus MP (2009) Imbalance of 22. Won E, Kim Y-K (2016) Stress, the autonomic
approach and avoidance: the Yin and Yang of nervous system, and the immune-kynurenine
anxiety disorders. Biol Psychiatry pathway in the etiology of depression. Curr
66:1072–1074 Neuropharmacol 14:665–673
13. Myint AM, Schwarz MJ, Steinbusch HWM, 23. Christmas DM, Potokar J, Davies SJ (2011) A
Leonard BE (2009) Neuropsychiatric disorders biological pathway linking inflammation and
depression: activation of indoleamine 30. Miller AH (2009) Norman cousins lecture.

2,3-dioxygenase. Neuropsychiatr Dis Treat Mechanisms of cytokine-induced behavioral
7:431–439 changes: psychoneuroimmunology at the
24. Cho HJ, Savitz J, Dantzer R et al (2017) Sleep translational interface. Brain Behav Immun
disturbance and kynurenine metabolism in 23:149–158
depression. J Psychosom Res 99:1–7 31. Toben C, Baune BT (2015) An act of balance
25. Naoi M, Maruyama W, Shamoto-Nagai M between adaptive and maladaptive immunity in
(2018) Type a monoamine oxidase and seroto- depression: a role for T lymphocytes. J Neu-
nin are coordinately involved in depressive dis- roimmune Pharmacol 10:595–609
orders: from neurotransmitter imbalance to 32. Huang T-L, Lee C-T (2007) T-helper 1/T-
impaired neurogenesis. J Neural Transm helper 2 cytokine imbalance and clinical phe-
(Vienna) 125(1):53–66 notypes of acute-phase major depression. Psy-
26. Lipina TV, Fletcher PJ, Lee FH, Wong AHC, chiatry Clin Neurosci 61:415–420
Roder JC (2013) Disrupted-in-schizophrenia- 33. Kim Y-K, Na K-S, Myint A-M, Leonard BE
1 Gln31Leu polymorphism results in social (2016) The role of pro-inflammatory cytokines
anhedonia associated with monoaminergic in neuroinflammation, neurogenesis and the
imbalance and reduction of CREB and neuroendocrine system in major depression.
β-arrestin-1,2 in the nucleus accumbens in a Prog Neuro-Psychopharmacol Biol Psychiatry
mouse model of depression. Neuropsycho- 64:277–284
pharmacology 38:423–436 34. Freitas AE, Bettio LEB, Neis VB et al (2014)
27. Müller N, Schwarz MJ (2008) A psychoneur- Agmatine abolishes restraint stress-induced
oimmunological perspective to Emil Kraepelins depressive-like behavior and hippocampal anti-
dichotomy: schizophrenia and major depres- oxidant imbalance in mice. Prog Neuro-
sion as inflammatory CNS disorders. Eur Arch Psychopharmacol Biol Psychiatry 50:143–150
Psychiatry Clin Neurosci 258(Suppl 2):97–106 35. Budni J, Zomkowski AD, Engel D et al (2013)
28. Miüller N, Schwarz MJ (2007) The immuno- Folic acid prevents depressive-like behavior and
logical basis of glutamatergic disturbance in hippocampal antioxidant imbalance induced by
schizophrenia: towards an integrated view. J restraint stress in mice. Exp Neurol
Neural Transm Suppl:269–280 240:112–121
29. Réus GZ, de Moura AB, Silva RH et al (2017) 36. Rivera-Rivera Y, Garcı́a Y, Toro V et al (2014)
Resilience dysregulation in major depressive Depression correlates with increased plasma
disorder: focus on glutamatergic imbalance levels of inflammatory cytokines and a dysregu-
and microglial activation. Curr lated oxidant/antioxidant balance in HIV-1-
Neuropharmacol infected subjects undergoing antiretroviral
therapy. J Clin Cell Immunol 5
Chapter 6
Increasing Resilience to Traumatic Stress: Understanding

the Protective Role of Well-Being
J. Tory Toole, Mark A. Rice Jr, Jordan Cargill, Travis J. A. Craddock,
Barry Nierenberg, Nancy G. Klimas, Mary Ann Fletcher, Mariana Morris,
Joel Zysman, and Gordon Broderick
Abstract
The brain maintains homeostasis in part through a network of feedback and feed-forward mechanisms,
where neurochemicals and immune markers act as mediators. Using a previously constructed model of
biobehavioral feedback, we found that in addition to healthy equilibrium another stable regulatory program
supported chronic depression and anxiety. Exploring mechanisms that might underlie the contributions of
subjective well-being to improved therapeutic outcomes in depression, we iteratively screened 288 candi-
date feedback patterns linking well-being to molecular signaling networks for those that maintained the
original homeostatic regimes. Simulating stressful trigger events on each candidate network while main-
taining high levels of subjective well-being isolated a specific feedback network where well-being was
promoted by dopamine and acetylcholine, and itself promoted norepinephrine while inhibiting cortisol
expression. This biobehavioral feedback mechanism was especially effective in reproducing well-being’s
clinically documented ability to promote resilience and protect against onset of depression and anxiety.
Key words Computational modeling, Reverse engineering, Homeostatic regulation, Depression,

Well-being, Positive psychology
1 Introduction
Psychological well-being has been a topic of debate for centuries,

but only within the last few decades and with the rise of positive
psychology has it been objectively defined and analyzed. Ryff and
colleagues [1] applied common factor analysis to deconstruct and
define subjective well-being in terms of six dimensions of wellness:
autonomy, environmental mastery, personal growth, positive rela-
tions with others, purpose in life, and self-acceptance. These
dimensions have been the accepted norm for the objective under-
standing of psychological well-being for over 20 years. Increased
well-being has been associated with significant reductions in both
psychopathological conditions and physical illness, and correlates
87
88 J. Tory Toole et al.
positively with improved immune function [2, 3]. Its protective and
therapeutic benefits are numerous, and an empirically validated
treatment manual for well-being therapy was produced within the
last year [4]. As can often be the case, interventions based on
subjective well-being have found their way into clinical practice
yet little is known of the underlying physiological mechanisms
involved in its regulation.
In this work, we construct a causal model from literature
describing regulatory associations linking key neurotransmitters,
hormones, and behavioral constructs. We use this model as a basis
for reverse engineering the possible biobehavioral mechanisms
involved in regulating subjective well-being by working back from
known outcomes such as the documented reduction in recidivism
rate in depression [2]. In the current study, we consider subjective
well-being as a single entity for the sake of simplicity, and iteratively
propose stimulatory and inhibitory control circuitry linking subjec-
tive well-being to a model neurotransmission network. Each set of
proposed mechanisms is then tested for its ability to instill increased
resilience to the onset of depression through repeated simulations.
This computer-based exploration of mechanisms supporting the
role of subjective well-being in depression suggests that the latter
may interact directly with norepinephrine, cortisol, acetylcholine,
and dopamine as part of a biobehavioral regulatory loop.
2 Materials
The foundation of this computer-based exploration of the

biological underpinnings of subjective well-being is a control net-
work model capturing known biobehavioral signaling mechanisms
(Subheading 2.1). This core set of signaling mechanisms is
described in detail in the following section. The model then serves
as the basis for iteratively testing proposed physiological connec-
tions to well-being such that the overall behavior of the augmented
regulatory network adheres to the known qualitative effects of
positive psychological intervention.
2.1 A Biobehavioral An important obstacle in building models across regulatory systems

Regulatory Model remains the scarcity of detailed human in vivo kinetic data as its
collection can present significant health risks to subjects. Fortunately,
the physiology and biochemistry describing the architecture of these
signaling networks are much better documented. Moreover, analysis
of the connection patterns that make up these networks is sufficient
to describe the range and type of regulatory behaviors available to
the system. Here, we capitalize on this broader body of knowledge to
construct a discrete logic representation of documented mechanisms
linking a core set of neurotransmitters, hormones, and behavioral
constructs. Specific signaling molecules include cortisol,
Computational Discovery of Resilience Mechanisms in Depression 89
epinephrine, dopamine, neuropeptide Y, norepinephrine, GABA,

serotonin, glutamate, and acetylcholine.
Cortisol and epinephrine are key players in regulating stress
response. Cortisol is a steroid hormone that plays a significant role
in the psychology and onset of maladaptive behavioral states like
depression. Literature reports that cortisol is necessary in activating
the synthesizing enzyme responsible for epinephrine [5], which has
also been associated with increases in anxiety [6]. Furthermore,
cortisol can increase levels of dopamine and glutamate in the
brain. Specifically, glucocorticoids contribute to an increase in
dopaminergic activity [7]. They also increase glutamatergic trans-
mission in the prefrontal cortex through modifying postsynaptic
NMDA and AMPA receptors [8]. Conversely, increased cortisol
inhibits serotonin levels by increasing serotonin uptake [9].
Dopamine and GABA are broad-acting neurotransmitters that
work in concert. Dopamine, applied iontophoretically, consistently
attenuates the inhibitory actions of GABA [10]. Similarly, dopa-
mine interacts antagonistically with serotonin [11], neuropeptide Y
(NPY) [12], norepinephrine [13], and acetylcholine [14]. Accord-
ing to current literature, neuropeptide Y, or NPY, can block the
phase shift of glutamate, reducing its cortical levels [15]. Increases
in NPY can also downregulate acetylcholine and norepinephrine
[16], as well as serotonin levels [17]. According to Herman et al.
(2003) [18], norepinephrine pathways modulate the synthesis of
GABA in central limbic stress circuits. Elevated levels of norepi-
nephrine also promote NPY [19]. This neurotransmitter is also a
behavioral modifier and can increase anxiety [20], hyperarousal
[21], attention [22], and working memory [23] while decreasing
the severity of depressive mood [24]. Conversely, GABA, or
gamma-aminobutyric acid, is a primary inhibitory neurotransmit-
ter. It is known to downregulate dopamine [25], norepinephrine
[26], and serotonin [27]. Elevated GABA can also lead to
decreased levels of attention [28]. Contrary to GABA, glutamate
is a primary excitatory neurotransmitter. It acts by increasing levels
of norepinephrine [29], cortisol [30], and GABA [31], as well as
improving working memory [32]. It can, however, inhibit atten-
tion [33] and exacerbate depressive symptoms [34].
Serotonin is perhaps one of the better known neurotransmit-
ters involved in depression [35] and is an established target of
pharmaceutical intervention in this illness. Because of this key
role, its place in our neurotransmission model is of particular
importance. Serotonin can promote glutamate release [36]
through a subset of receptors. Similarly, it can depolarize certain
neuronal cell types to increase the release of acetylcholine [37], a
neurotransmitter that acts on both the peripheral nervous system
(PNS) and the central nervous system (CNS). Acetylcholine is
known to promote increased levels of epinephrine [38], NPY
[39], and cortisol [40] while downregulating levels of
norepinephrine [41] and serotonin [42]. Conversely, serotonin has

been found to exercise a reciprocal relationship with norepineph-
rine [43], dopamine [44], cortisol [45], and GABA [27]. As a
behavioral modifier, serotonin can increase hyperarousal [46]
while inhibiting anxiety [47] and reducing depressive mood
[48]. Chronically high levels of serotonin, however, have been
shown to promote increased physical fatigue [49]. Likewise,
increased acetylcholine activity is also a known behavioral modifier,
increasing hyperarousal [50], attention [51], and improved work-
ing memory [52].
Collectively these regulatory associations serve to create a min-
imal biobehavioral signaling network that captures the actions of a
principal set of neurotransmission mechanisms (Fig. 1). It is impor-
tant to note that this circuitry is not illness specific but instead is
assembled from what is believed to be regulatory signaling mechan-
isms through which “normal” homeostatic regulation is ensured in
an average person.
3 Methods
The analysis that serves as the focus of this chapter is divided into
three discrete steps. The first step (Subheading 3.1) involves apply-
ing a discrete ternary logic analysis to the biobehavioral regulatory
model. Using this logic to dictate allowable departures from a given
state, a Monte Carlo simulation framework is applied to determine
the homeostatic steady states of the system (Subheading 3.2). In an
attempt at reverse engineering documented clinical outcomes of
improved subject well-being, augmented models are created which
include a subjective well-being node. In each of these models,
candidate mechanisms are proposed whereby subjective well-
being directly modulates two molecular targets and in turn is
directly promoted or inhibited by two members of this same molec-
ular network as part of a biobehavioral feedback loop. The behavior
supported by each of these candidate regulatory feedback loops is
simulated to determine (1) which possible connection patterns
continue to support established steady states, and (2) which of
these also reproduce the observed protective effects of increasing
subjective well-being (Subheading 3.3).
3.1 Discrete Ternary The discrete ternary logical network analysis used in the present
Logical Signal work is an extension of a methodology proposed by Mendoza and
Processing Xenarios (2006) [53] and Thomas (1991) [54], and has been
reported previously by our group [55, 56]. We encode documented
feedback mechanisms within the neuropsychological system using
only the direction (source and target) and type (activator or inhibi-
tor) of interaction. As data describing the relative magnitude of
cellular responses of these various signals remain limited, we con-
sider model components to be equally sensitive to all signals.
Fig. 1 A basic logic model of behavior and neurotransmission. A simple causal regulatory network model
linking 15 soluble mediators in the brain and associated behavioral constructs informed by published literature
where green connectors with arrowhead terminators indicate stimulatory actions whereas red connectors with
dot terminators indicate inhibitory actions. Also illustrated is a potential biobehavioral feedback circuit derived
from simulations of documented therapeutic outcomes and describing the regulatory interactions of subjective
well-being and neuro-signaling
Similarly, we also consider the response to these signals to be

equivalent in magnitude regardless of source. Using this formalism,
we determine the number and type of stable resting states sup-
ported by the regulatory circuitry as well as the specific qualitative
neuropsychological signatures at each of these stable points without
requiring detailed kinetic information. That is, we determine where
the system would eventually come to rest even though we may not
know how quickly this equilibrium will be reached. In this model,
signaling molecules and behavioral constructs are represented as
individual variables, each capable of adopting three discrete states:
1 (down-expressed), 0 (nominal), and 1 (up-expressed). At any
point in time t, the state of a system with N variables can be
represented by the vector ~ x ðt Þ, such that
~
x ðt Þ ¼ ðx 1 ðt Þ; x 2 ðt Þ; . . . ; x N ðt ÞÞ ð1Þ
where xi(t) represents the state of the ith variable of the

N variable system at time t. The image vector ~ x ðt þ 1Þ describes
the preferred state toward which the system evolves in the next time
increment. The state value of the image vector for the ith variable is
determined from its current state and a set of balanced ternary logic
statements based on the current value of variable and the mode of
action (i.e., activate or inhibit) of the neighboring input variables.
These logic statements are expressed as follows (Eq. (2)):
8 A A A
I I I

< x i1 ðt Þ∨x i2 ðt Þ∨ . . . ∨x iX ðt Þ ∇ x i1 ðt Þ∨x i2 ðt Þ∨ . . . ∨x iY ðt Þ
A A A
x i ðt þ 1Þ ¼ x i1 ðt Þ∨x i2 ðt Þ∨ . . . ∨x iX ðt Þ
: I I I
˜ x i1 ðt Þ∨x i2 ðt Þ∨ . . . ∨x iY ðt Þ
ð2Þ
where the ∇, ∨, and ¬ symbols are ternary HIGH/LOW PASS,
OR, and NOT operators; x ijA is the state of the ith variable’s jth
activator; and x ikI is the state of the ith variable’s kth inhibitor. The
ternary operators given in Eq. (2) are described in further detail in
Craddock et al. (2014) [55] and Fritsch et al. (2014) [56]. The first
entry in Eq. (2) is used when the variable possesses X activators and
Y inhibitors, the middle when the variable has only X activators,
and last when the activator has only Y inhibitors.
The number of nodes provides an upper bound for the total
number of possible states such that a set of N independent nodes
support 3N combinations of states. The states available to a model
will occupy a smaller subset of these since variables are no longer
independent but are constrained by the control circuitry to
co-express in specific combinations only. Steady states are defined
as those state nodes from which there is no allowable escape or
which possesses an out degree of 0. Since the current regulatory
network model (Fig. 1) contains 15 soluble mediators and behav-
ioral constructs, the number of possible combinatorial states for a
variable set of this size is 315 ¼ 14,348,907. Though quite large this
remains a computationally tractable problem and an exhaustive

search of all possible state combinations that are allowed by the
model circuitry and which constitute a steady homeostatic state as
in [56], Fritsch et al. (2014), is possible.
3.2 Using a Monte Evolution of the system through the sequence of state transitions
Carlo Environment supported by the model is simulated by applying a Monte Carlo
to Simulate Response algorithm. From any initial starting state, allowable state transitions
are determined based on Eq. (2). Applying the latter to each
variable in the model for the mth state of the system, ~ x m ðt Þ, defines
m th
the image vector ~ x ðt þ 1Þ for the m state. With ~ x m ðt þ 1Þ
defined, the system is updated asynchronously following the
generalized logical analysis of [54] Thomas (1991). According to
this method the ith variable of the mth state vector x im ðt Þ is moved
one step toward its preferred image x im ðt þ 1Þ (e.g., If x im ðt Þ ¼ 1
and x im ðt þ 1Þ ¼ 1, then x im ðt þ 1Þ is set to 0). Thus, for each
current state of the system there are potentially several subsequent
states toward which it may asynchronously evolve. From the allow-
able transitions a target state is chosen at random using a uniform
equal distribution then used to generate the next set of allowable
target states. Executing the simulation multiple times gives a distri-
bution of paths that is used to determine the behavior of the system
from any given start state. For additional details the reader is
referred to our previous work [55, 56].
3.3 Identifying As described above, homeostatic states are defined as those states to
Stable Regulatory which the system naturally returns following a perturbation. Even
Behaviors simple biological circuits often support more than one stable equilib-
rium state. These were determined here by enumerating all possible
states that may be occupied by the model shown in Fig. 1. In the
current example, this exhaustive search pointed to two steady homeo-
static states (SS) where neurotransmitter and psychobehavioral con-
structs converged to the levels described in Fig. 2. In addition to a
typical healthy resting state (SS0) we found that this biobehavioral
model could also accommodate an alternate equilibrium state where a
persistent depressive mood and increased anxiety are accompanied by
reduced physical fatigue and impaired attention. Serotonin is regu-
lated to chronically low levels while levels of glutamate, cortisol,
GABA, and epinephrine are maintained high. This finding would
suggest that a persistent depressive mood might be perpetuated at
least in part by normal regulatory drive (see Note 1).
3.4 Discovering Given the current scarcity of data linking subjective well-being to
Biobehavioral physiological markers, we propose a reverse engineering analysis
Regulatory where we iteratively construct possible biobehavioral feedback
Mechanisms mechanisms and simulate their ability to mimic observed clinical
outcomes. Specifically, we propose circuits where well-being
3.4.1 A Circuit Model directly modulates two molecular targets and is itself modulated
for Subjective Well-Being by two feedback signals. These candidate circuits are refined
Fig. 2 Multiple naturally occurring stable states. This basic model of

neurotransmission supports an alternate homeostatic state where physiologic
regulation may also support a persistent depressive state with increased anxiety
accompanied by reduced physical fatigue, impaired attention, and where
serotonin is regulated to chronically low levels while glutamate, cortisol,
GABA, and epinephrine are consistently upregulated
iteratively using a numerical optimization scheme to identify which

candidate feedback mechanisms best minimize the simulated
migration from normal homeostasis toward a persistent state of
depression and anxiety, qualitatively mimicking clinically observed
outcomes. First a set of candidate biobehavioral circuits are created
de novo from all possible circuits linking subjective well-being to
two molecular targets and two sources. Using the same exhaustive
search scenario, these augmented candidate models are tested for
the continued support of the steady-state attractors identified pre-
viously from the basic model (Subheading 3.3). Of the 288 possible
feedback circuits involving well-being, 97 supported these two
basic equilibrium states suggesting that they might be biologically
plausible (see Note 2).
3.4.2 Identifying These biologically plausible feedback circuits involving subjective

Plausible Subjective Well- well-being were then assessed in terms of their ability to impart
Being Feedback resilience to depression and mimic the improved outcomes
Mechanisms observed in clinic using a positive psychology approach. Accord-
ingly, the behavior of these 97 networks was then simulated 1000
times at 1000 random start points with well-being maintained high
for a total of one million simulations per network. Performance was
measured as the fraction of these simulations where the biobehav-
ioral network settled into the pathologic resting state despite high
levels of subjective well-being instead of recovering normal homeo-
static balance. Candidate feedback circuits were ranked on the basis
of this performance metric and Cook’s outlier distance used as the
statistical basis for identifying circuits that performed significantly
better than the others. Only 3 of these 97 candidate circuits per-
formed significantly better than average at a p < 0.05 confidence
level. With a Cook’s distance of 0.45, compared to <0.1 for the
next best candidate, one specific feedback configuration stood out
as an especially interesting candidate for future experimental
validation.
According to this candidate feedback network, subjective
well-being acts directly to upregulate norepinephrine and down-
regulate cortisol levels. Conversely increases in acetylcholine and
dopamine both directly lead to upregulation of subjective well-
being (see Note 3).
This candidate biobehavioral feedback circuit effectively mini-
mized susceptibility to onset of chronic anxiety or depression given
an elevated sense of well-being. Though not by any means defini-
tive, this simulation-based reverse engineering approach nonethe-
less provides a first plausible physiological mechanism of action for
the protective role of well-being, one that can be explored further
through experimentation (Fig. 3) (see Note 4 and Note 5).
4 Notes
1. Much like a computer may execute multiple programs, even

simple biological circuits are capable of supporting more than
one stable regulatory mode. As presented in greater detail in an
accompanying chapter by our group, we assemble from litera-
ture a basic circuit model of biobehavioral regulation and dem-
onstrate that this biological computer can in theory support a
stable regulatory program where serotonin levels are persis-
tently downregulated while GABA, glutamate, epinephrine,
and cortisol levels are upregulated resulting in stable depressive
mood accompanied by increased anxiety. This is consistent with
recent work by Müller and Schwarz (2007) [57] who propose
that dysregulation of feedback linking hypothalamic-pituitary-
adrenal (HPA) axis function with control of serotonin and
0.5
0.45
0.4
0.35
Cook's distance
0.3
0.25
0.2
0.15
0.1
0.05
0
0 20 40 60 80 100
Network sorted rank
Fig. 3 Ranking biobehavioral feedback mechanisms for subjective well-being. Constraining the circuit
architecture for subjective well-being to two incoming and two outgoing control actions, simulations show
that three candidate feedback circuits perform significantly better than average ( p < 0.05, dotted line) in
reproducing observed clinical outcomes, as measured by Cook’s outlier distance. One circuit in particular that
outperformed the others in reproducing the protective effect of subjective well-being consisted of direct
downregulation of cortisol and upregulation of norepinephrine, with positive feedback from acetylcholine and
dopamine
glutamate production might provide a comprehensive frame-

work for the study of depression. This regulatory model now
serves in this chapter as a basis for exploring the role of subjec-
tive well-being in increasing resilience to this condition.
2. The current work illustrates how a model-based numerical
protocol may be devised and used to reverse engineer candidate
causal mechanisms describing biobehavioral feedback when
these are largely unknown but when convincing characteristic
clinical outcomes have been observed. Though not definitive,
the resulting candidate mechanisms constitute a valuable basis
for the design of focused experimental investigations. The use
of computational models is slowly gaining acceptance as a tool
for exploring the complexity of illnesses such as depression
[58, 59]. By requiring only connectivity and enforcing the
correct sequence of events, the discrete logic paradigm pro-
posed here offers the possibility of modeling broader more
complex regulatory systems and delivering qualitatively
insightful predictions of their dynamic behavior with little or
no data. This minimal data requirement makes this an especially
attractive strategy for discovery and hypothesis testing.
3. Applying this model-based protocol to the real-world issues of
treatment resistance and relapse in depression has suggested
the potential involvement of norepinephrine, cortisol, acetyl-

choline, and dopamine as key players in the physiological man-
ifestation of subjective well-being. An association of changes in
subjective well-being with altered levels of cortisol, dopamine,
and norepinephrine is echoed in Ryff, Singer, and Love’s 2004
study on connecting well-being with biology [3].
4. As this framework enforces causality, this analysis extends
beyond correlates of well-being to reveal candidate biobehav-
ioral feedback mechanisms through which this psychological
construct may mediate neurotransmission in order to achieve
its protective role. In this example, computer simulations mim-
icking observed clinical outcomes highlighted four possible
control actions bridging from the psychological construct of
subjective well-being to various elements of physiological brain
chemistry. Specifically, subjective well-being may act directly to
downregulate cortisol and increase norepinephrine, and be
promoted directly in a positive feedback by increases in acetyl-
choline and dopamine. Such a role in enhancing well-being
would be consistent with the reported synergistic actions of
acetylcholine and dopamine in consolidating attention and
promoting reward-based learning [60].
5. The assembly of literature-based regulatory models of psycho-
biology may serve as a useful starting point in the exploration of
neuromechanisms that underlie the actions of relatively unex-
plored constructs such as subjective well-being. Indeed, this
protocol and those like it may find a broad range of applications
in the study of complex “mental phenomena,” especially those
with an established base of clinical and interventional data [61].
By providing a formal quantitative framework for improving
our theoretical understanding of the neurochemistry and its
bridge to the psychological realm, it provides an environment
for the development of clearly defined and testable hypotheses
for further, more detailed, and more specific research into these
constructs and how they may be affected therapeutically.
Acknowledgments
Funding was provided by US Department of Defense Congressio-

nally Directed Medical Research Program (CDMRP) awards
(http://cdmrp.army.mil/) GW093042, GW140142 (Broderick—
PI), and GW120045 (Morris—PI). This research was conducted in
collaboration with the high-performance computing team at the
University of Miami Center for Computational Science (CCS)
(http://ccs.miami.edu).
Disclaimer: The opinions and assertions contained herein are the

private views of the authors and are not to be construed as official or
as reflecting the views of the Department of Defense.
References
1. Ryff CD, Keyes CLM (1995) The structure of 13. El Mansari M, Guiard BP, Chernoloz O,
psychological well-being revisited. J Pers Soc Ghanbari R, Katz N, Blier P (2010) Relevance
Psychol 69(4):719 of norepinephrine–dopamine interactions in
2. Wood AM, Joseph S (2010) The absence of the treatment of major depressive disorder.
positive psychological (eudemonic) well-being CNS Neurosci Ther 16(3):e1–e17
as a risk factor for depression: a ten year cohort 14. Calabresi P, Picconi B, Parnetti L, Di Filippo M
study. J Affect Disord 122(3):213–217 (2006) A convergent model for cognitive dys-
3. Ryff CD, Singer BH, Love GD (2004) Positive functions in Parkinson’s disease: the critical
health: connecting well-being with biology. dopamine–acetylcholine synaptic balance. Lan-
Philos Trans R Soc Lond B Biol Sci 359 cet Neurol 5(11):974–983
(1449):1383 15. Biello SM, Golombek DA, Harrington ME
4. Fava GA (2016) Well-being therapy. In: The (1997) Neuropeptide Y and glutamate block
Wiley handbook of positive clinical psychology. each other’s phase shifts in the suprachiasmatic
Wiley, Hoboken, p 395 nucleus in vitro. Neuroscience 77
5. Wurtman RJ, Pohorecky LA, Baliga BS (1972) (4):1049–1057
Adrenocortical control of the biosynthesis of 16. Serfozo P, Bartfai T, Vizi ES (1986) Presynap-
epinephrine and proteins in the adrenal tic effects of neuropeptide Y on [3 H] nor-
medulla. Pharmacol Rev 24(2):411–426 adrenaline and [3 H] acetylcholine release.
6. Janssen SA, Arntz A, Bouts S (1998) Anxiety Regul Pept 16(2):117–123
and pain: epinephrine-induced hyperalgesia 17. Dryden S, McCarthy HD, Malabu UH,
and attentional influences. Pain 76 Ware M, Williams G (1993) Increased neuro-
(3):309–316 peptide Y concentrations in specific hypotha-
7. Piazza PV, Rougé-Pont F, Deroche V, lamic nuclei of the rat following treatment with
Maccari S, Simon H, Le Moal M (1996) Glu- methysergide: evidence that NPY may mediate
cocorticoids have state-dependent stimulant serotonin’s effects on food intake. Peptides 14
effects on the mesencephalic dopaminergic (4):791–796
transmission. Proc Natl Acad Sci 93 18. Herman JP, Renda A, Bodie B (2003) Norepi-
(16):8716–8720 nephrine–gamma-aminobutyric acid (GABA)
8. Yuen EY, Liu W, Karatsoreos IN, Feng J, McE- interaction in limbic stress circuits: effects of
wen BS, Yan Z (2009) Acute stress enhances reboxetine on GABAergic neurons. Biol Psy-
glutamatergic transmission in prefrontal cortex chiatry 53(2):166–174
and facilitates working memory. Proc Natl 19. Wahlestedt CLAES, Hakanson ROLF, Vaz CA,
Acad Sci 106(33):14075–14079 Zukowska-Grojec ZOFIA (1990) Norepi-
9. Tafet GE, Toister-Achituv M, Shinitzky M nephrine and neuropeptide Y: vasoconstrictor
(2001) Enhancement of serotonin uptake by cooperation in vivo and in vitro. Am J Phys
cortisol: a possible link between stress and Regul Integr Comp Phys 258(3):R736–R742
depression. Cogn Affect Behav Neurosci 1 20. Bremner JD, Krystal JH, Southwick SM, Char-
(1):96–104 ney DS (1996) Noradrenergic mechanisms in
10. Bergstrom DA, Walters JR (1984) Dopamine stress and anxiety: II. Clinical studies. Synapse
attenuates the effects of GABA on single unit 23(1):39–51
activity in the globus pallidus. Brain Res 310 21. Southwick SM, Bremner JD, Rasmusson A,
(1):23–33 Morgan CA, Arnsten A, Charney DS (1999)
11. Daw ND, Kakade S, Dayan P (2002) Oppo- Role of norepinephrine in the pathophysiology
nent interactions between serotonin and dopa- and treatment of posttraumatic stress disorder.
mine. Neural Netw 15(4):603–616 Biol Psychiatry 46(9):1192–1204
12. Gillard ER, Dang DQ, Stanley BG (1993) Evi- 22. Bymaster FP, Katner JS, Nelson DL, Hemrick-
dence that neuropeptide Y and dopamine in the Luecke SK, Threlkeld PG, Heiligenstein JH,
perifornical hypothalamus interact antagonisti- Morin SM, Gehlert DR, Perry KW (2002)
cally in the control of food intake. Brain Res Atomoxetine increases extracellular levels of
628(1):128–136 norepinephrine and dopamine in prefrontal
cortex of rat: a potential mechanism for efficacy
in attention deficit/hyperactivity disorder. Denosologization of biological psychiatry or

Neuropsychopharmacology 27(5):699–711 the specificity of 5-HT disturbances in psychi-
23. Zhang K, Grady CJ, Tsapakis EM, Andersen atric disorders. J Affect Disord 13(1):1–8
SL, Tarazi FI, Baldessarini RJ (2004) Regula- 36. Aghajanian GK, Marek GJ (1999) Serotonin–-
tion of working memory by dopamine D4 glutamate interactions: a new target for anti-
receptor in rats. Neuropsychopharmacology psychotic drugs. Neuropsychopharmacology
29(9):1648 21:S122–S133
24. Delgado PL, Moreno FA (1999) Role of nor- 37. MacDermot J, Higashida H, Wilson SP,
epinephrine in depression. J Clin Psychiatry Matsuzawa H, Minna J, Nirenberg M (1979)
61:5–12 Adenylate cyclase and acetylcholine release
25. Casey DE, Gerlach J, Christensson E (1980) regulated by separate serotonin receptors of
Behavioral aspects of GABA-dopamine interre- somatic cell hybrids. Proc Natl Acad Sci 76
lationships in the monkey. Brain Res Bull (3):1135–1139
5:269–273 38. Keely SL, Lincoln TM, Corbin JD (1978)
26. Moises HC, Woodward DJ (1980) Potentia- Interaction of acetylcholine and epinephrine
tion of GABA inhibitory action in cerebellum on heart cyclic AMP-dependent protein kinase.
by locus coeruleus stimulation. Brain Res 182 Am J Phys Heart Circ Phys 234(4):
(2):327–344 H432–H438
27. Bankson MG, Yamamoto BK (2004) Seroto- 39. Dixon JS, Jen PY, Gosling JA (2000) The dis-
nin–GABA interactions modulate MDMA- tribution of vesicular acetylcholine transporter
induced mesolimbic dopamine release. J Neu- in the human male genitourinary organs and its
rochem 91(4):852–859 co-localization with neuropeptide Y and nitric
28. Edden RA, Crocetti D, Zhu H, Gilbert DL, oxide synthase. Neurourol Urodyn 19
Mostofsky SH (2012) Reduced GABA concen- (2):185–194
tration in attention-deficit/hyperactivity disor- 40. Walker SW, Strachan MWJ, Lightly ERT, Wil-
der. Arch Gen Psychiatry 69(7):750–753 liams BC, Bird IM (1990) Acetylcholine stimu-
29. Russell VA, Wiggins TM (2000) Increased lates cortisol secretion through the M3
glutamate-stimulated norepinephrine release muscarinic receptor linked to a
from prefrontal cortex slices of spontaneously polyphosphoinositide-specific phospholipase
hypertensive rats. Metab Brain Dis 15 C in bovine adrenal fasciculata/reticularis
(4):297–304 cells. Mol Cell Endocrinol 72(3):227–238
30. Mathew SJ, Coplan JD, Smith EL, Schoepp 41. Kalsner S, Quillan M (1988) Presynaptic inter-
DD, Rosenblum LA, Gorman JM (2001) Glu- actions between acetylcholine and adrenergic
tamate—hypothalamic-pituitary-adrenal axis antagonists on norepinephrine release. J Phar-
interactions: implications for mood and anxiety macol Exp Ther 244(3):879–891
disorders. CNS Spectr 6(07):555–564 42. Leboulenger F, Benyamina M, Delarue C,
31. Carlsson A, Waters N, Holm-Waters S, Netchitailo P, Saint-Pierre S, Vaudry H
Tedroff J, Nilsson M, Carlsson ML (2001) (1988) Neuronal and paracrine regulation of
Interactions between monoamines, glutamate, adrenal steroidogenesis: interactions between
and GABA in schizophrenia: new evidence. acetylcholine, serotonin and vasoactive intesti-
Annu Rev Pharmacol Toxicol 41(1):237–260 nal peptide (VIP) on corticosteroid production
by frog interrenal tissue. Brain Res 453
32. Aultman JM, Moghaddam B (2001) Distinct (1):103–109
contributions of glutamate and dopamine
receptors to temporal aspects of rodent work- 43. Guiard BP, El Mansari M, Merali Z, Blier P
ing memory using a clinically relevant task. (2008) Functional interactions between dopa-
Psychopharmacology 153(3):353–364 mine, serotonin and norepinephrine neurons:
an in-vivo electrophysiological study in rats
33. Carrey NJ, MacMaster FP, Gaudet L, Schmidt with monoaminergic lesions. Int J Neuropsy-
MH (2007) Striatal creatine and glutamate/ chopharmacol 11(5):625–639
glutamine in attention-deficit/hyperactivity
disorder. J Child Adolesc Psychopharmacol 17 44. Korsgaard S, Gerlach J, Christensson E (1985)
(1):11–17 Behavioral aspects of serotonin-dopamine
interaction in the monkey. Eur J Pharmacol
34. Paul IA, Skolnick P (2003) Glutamate and 118(3):245–252
depression. Ann N Y Acad Sci 1003
(1):250–272 45. O’hara R, Schröder CM, Mahadevan R,
Schatzberg AF, Lindley S, Fox S et al (2007)
35. Van Praag HM, Kahn RS, Asnis GM, Wetzler S, Serotonin transporter polymorphism, memory
Brown SL, Bleich A, Korn ML (1987) and hippocampal volume in the elderly:
association and interaction with cortisol. Mol 54. Thomas R (1991) Regulatory networks seen as
Psychiatry 12(6):544–555 asynchronous automata: a logical description. J
46. Connor KM, Davidson JR (1998) The role of Theor Biol 153(1):1–23
serotonin in posttraumatic stress disorder: neu- 55. Craddock TJ, Fritsch P, Rice MA Jr, del
robiology and pharmacotherapy. CNS Spectr 3 Rosario RM, Miller DB, Fletcher MA, Klimas
(S2):42–51 NG, Broderick G (2014) A role for homeo-
47. Charney DS, Woods SW, Goodman WK, static drive in the perpetuation of complex
Heninger GR (1987) Serotonin function in chronic illness: gulf war illness and chronic
anxiety. Psychopharmacology 92(1):14–24 fatigue syndrome. PLoS One 9(1):e84839
48. Owens MJ, Nemeroff CB (1994) Role of sero- 56. Fritsch P, Craddock TJ, del Rosario RM, Rice
tonin in the pathophysiology of depression: MA, Smylie A, Folcik VA, de Vries G, Fletcher
focus on the serotonin transporter. Clin MA, Klimas NG, Broderick G (2013) Suc-
Chem 40(2):288–295 cumbing to the laws of attraction: exploring
49. Roelands B, Meeusen R (2010) Alterations in the sometimes pathogenic versatility of discrete
central fatigue by pharmacological manipula- immune logic. Systems Biomedicine 1
tions of neurotransmitters in normal and high (3):179–194
ambient temperature. Sports Med 40 57. Müller N, Schwarz MJ (2007) The immune-
(3):229–246 mediated alteration of serotonin and gluta-
50. Czermak C, Staley JK, Kasserman S, Bois F, mate: towards an integrated view of depression.
Young T, Henry S, Tamagnan GD, Seibyl JP, Mol Psychiatry 12:988–1000
Krystal JH, Neumeister A (2008) β2 Nicotinic 58. Soeteman DI, Miller M, Kim JJ (2012) Mod-
acetylcholine receptor availability in post- eling the risks and benefits of depression treat-
traumatic stress disorder. Int J Neuropsycho- ment for children and young adults. Value
pharmacol 11(3):419–424 Health 15(5):724–729
51. Botly LC, De Rosa E (2008) A cross-species 59. Gruwez B, Poirier MF, Dauphin A, Olié JP,
investigation of acetylcholine, attention, and Tod M (2007) A kinetic-pharmacodynamic
feature binding. Psychol Sci 19 model for clinical trial simulation of antidepres-
(11):1185–1193 sant action: application to clomipramine–-
52. Daniel JM, Dohanich GP (2001) Acetylcholine lithium interaction. Contemp Clin Trials 28
mediates the estrogen-induced increase in (3):276–287
NMDA receptor binding in CA1 of the hippo- 60. Holca-Lamarre R, Lücke J, Obermayer K
campus and the associated improvement in (2017) Models of Acetylcholine and Dopa-
working memory. J Neurosci 21 mine Signals Differentially Improve Neural
(17):6949–6956 Representations. Front. Comput Neurosci
53. Mendoza L, Xenarios I (2006) Theoretical 11:54
biology and medical Modelling. Theor Biol 61. Cipresso P, Immekus JC (2017) Back to the
Med Model 3:13 Future of Quantitative Psychology and Mea-
surement: Psychometrics in the Twenty-First
Century. Front Psychol 8:2099
Chapter 7
Exploring the Diagnostic Potential of Immune Biomarker

Co-expression in Gulf War Illness
Gordon Broderick, Mary Ann Fletcher, Michael Gallagher, Zachary Barnes,
Suzanne D. Vernon, and Nancy G. Klimas
Abstract
Complex disorders like Gulf War illness (GWI) often defy diagnosis on the basis of a single biomarker and
may only be distinguishable by considering the co-expression of multiple markers measured in response to a
challenge. We demonstrate the practical application of such an approach using an example where blood was
collected from 26 GWI, 13 healthy control subjects, and 9 unhealthy controls with chronic fatigue at three
points during a graded exercise challenge. A 3-way multivariate projection model based on 12 markers of
endocrine and immune function was constructed using a training set of n ¼ 10 GWI and n ¼ 11 healthy
controls. These groups were separated almost completely on the basis of two co-expression patterns. In a
separate test set these same features allowed for discrimination of new GWI subjects (n ¼ 16) from
unhealthy (n ¼ 9) and healthy control subjects with a sensitivity of 70% and a specificity of 90%.
Key words Cytokine profile, Co-expression patterns, Exercise response, Gulf War illness, Regression
model, Diagnostic classification, Partial least squares, Batch PLS
1 Introduction
Gulf War illness (GWI) is a complex multi-symptom illness of

unknown etiology associated with deployment to the Persian Gulf
between 1990 and 1991. Symptom presentation shows a strong
overlap with the case definition for chronic fatigue syndrome (CFS)
with 4.9% of the 693,826 Gulf War troops deployed satisfying the
latter [1]. Our work and the work of others strongly suggest that
the pathogenesis of GWI and CFS includes both a neuroendocrine
[2–5] and an immunologic component [6–11]. Earlier work in this
area has led to debates over the pro-inflammatory [12, 13] or anti-
inflammatory [14, 15] nature of these disorders and more recent
results suggest that they may not fit easily into either of these
conventional descriptive categories.
Common to the majority of existing studies is a focus on the
difference in expression of individual cytokines, hormones, and
101
102 Gordon Broderick et al.
neurotransmitters. Even though the nervous, endocrine, and

immune systems are highly integrated their biomarkers continue
to be analyzed individually leaving patterns of coordinated response
largely unexplored. In addition, investigations have focused on
immune status in patients at rest and have not sought to actively
probe the immune response. Used recently to investigate CFS [16],
exercise can be used to elicit a coordinated “fight or flight”
response in the nervous, endocrine, and immune systems. We
chose a similar standardized exercise challenge to amplify immune
and neuroendocrine response in GWI patients [9–11]. In an initial
analysis of the data we found that the patterns of association linking
immune and endocrine biomarkers differed very significantly in
GWI patients. Under effort a characteristic remodeling of the
association network occurred around nodes for CD19+ B-cell
abundance, IL-5 and IL-6 responsiveness in culture, as well as
soluble CD26 [11]. Further analysis indicated that initial IL-1a
concentration and CD2+/CD26+ cell abundance strongly influ-
enced the course of this remodeling.
Though key in gaining new insight into the basic disease
mechanisms of GWI, this type of analysis requires at least 15–30
samples to support the mathematical assembly of association net-
works in each illness group. Therefore with current technology
such an approach is impractical for use as a clinical assay for the
diagnosis of individual patients. Even though the association pat-
terns linking immune biomarkers differ from one illness group to
the next, we ask the inverse question in this work in order to reduce
the sample burden. Namely, is there a subset of associations shared
somewhat consistently by all subjects that might be preserved to a
different extent in GWI, versus CFS and healthy controls? This
question requires much less data to answer making this yardstick a
more plausible tool to deploy in clinic. Constructing a multivariate
statistical model to capture patterns of change shared by various
cytokines, cortisol, and neuropeptide Y (NPY) across the entire
time course we show that such patterns can indeed serve in distin-
guishing GWI subjects from controls. More importantly, we for-
mally test the predictive performance of this approach using a
separate validation set where we show that cytokine patterns iden-
tified in the training set also distinguish new GWI subjects from
patients with CFS, an illness with significant overlap in clinical
presentation.
2 Materials
The use of multivariate statistical models in defining molecular

phenotypes for complex illness is the principal focus of this chapter.
The foundation for conducting such an analysis is a dataset consist-
ing of a large number of candidate biomarkers. In general these
Immune Response Patterns in Gulf War Illness 103
biomarkers are substantially cross-correlated or partially redundant.

They will have been measured in samples collected from an appro-
priately recruited cohort of subjects (Subheading 2.1) assigned to
clinically relevant diagnostic groups using standard assessment
tools (Subheading 2.2). Typically samples of blood, saliva, or
other fluids will have been collected and analyzed using industry
standard assays (Subheading 2.3).
2.1 A Standardized In this example, subject inclusion criteria were derived from Fukuda
Case Definition et al. (1998) [17], and consisted in identifying veterans deployed to
and Psychometric the theater of operations between August 8, 1990, and July
Tools 31, 1991, with one or more symptoms present after 6 months
from at least two of the following: fatigue; mood and cognitive
complaints; and musculoskeletal complaints (joint pain, stiffness, or
muscle pain). The suitability of the Fukuda definition for identify-
ing GWI patients is supported by the work of Collins et al.
(2002) [18].
In support of the case definition psychometric questionnaires
used to assess illness severity included the Multidimensional
Fatigue Inventory (MFI) [19], a 20-item self-report instrument
designed to measure fatigue, and the Medical Outcomes Study
36-item short-form survey (SF-36) [20] assessing health-related
quality of life. Typically all subjects also receive a physical examina-
tion and provided a medical history including the GWI symptom
checklist as per the case definition.
2.2 Instruments Considering the response to a physiological challenge can enhance

for Assessment the expression of candidate biomarkers. In this case a Graded
of Exercise Response eXercise Test (GXT) was administered using a Vmax Spectra 29c
Cardiopulmonary Exercise Testing Instrument, Sensor-Medics
Ergoline 800 fully automated cycle ergometer, and SensorMedics
Marquette MAX 1 Stress ECG (SensorMedics Corp., Yorba Linda,
CA).
2.3 Standard Assays The data used here to illustrate the identification of biomarkers
for Measuring Immune co-expression patterns is based on measurements of immune cell
Markers in Blood signaling. Peripheral blood mononuclear cells (PBMCs) were
recovered from heparinized samples of whole blood and cultured
for 48 h with phytohemagglutinin at 37 C, 5% CO2. Following
incubation, culture supernatants were collected and frozen at
70 C until analyzed for concentrations of IL-1α, IL-5, IL-6,
IL-10, TNFα, and IFNγ. Levels of IL-6, IL-10, and TNFα were
also determined in plasma samples that were separated with 2 h,
stored at 70 C, and not thawed prior to analysis. In all cases
measurement was performed using commercial ELISA kits (Immu-
notech, Miami, FL), quality-controlled using National Institute for
Biological Standards and Control/World Health Organization
cytokine reference standards. Plasma sCD26 was also measured by
ELISA (Bender Med Systems, Burlingame, CA) and NPY was

measured using a RIA from Alpco Diagnostics (Salem, NH). Sali-
vary cortisol was determined by immunoassay using the Salimetrics
high sensitivity kit (State College, PA). This measure has been
shown to correlate well with cortisol in blood during exercise [21].
2.4 Computational Differences in the expression of individual markers were assessed

Tools using the nonparametric Wilcoxon rank-sum test and the signifi-
cance of first-order trends was based on the Friedman test for
2.4.1 Standard repeated measures. A parametric two-way ANOVA was also used
Hypothesis Tests to test for the significance of group and time effects as well as the
time-group interaction effect. Statistical significance was set at
p < 0.05. All hypothesis testing was performed using the MatLab
Statistics Toolbox (MathWorks, Natick, MA).
2.4.2 Multivariate In this example we used a projection to latent structure technique

Regression Analysis (PLS) [22, 23], also called partial least squares, to identify patterns
of cytokine co-expression that could classify GWI subjects. This
method enables the construction of A composite features, each
consisting of different weighted sums of the original variables.
Like the basic model for principal component analysis (PCA)
these A features are computed for the input space X (Eq. (1))
described by k markers measured in n subjects. However in PLS
composite features are also computed simultaneously for an output
space Y (Eq. (2)) described by m response markers measured in the
same n subjects. These composite features are defined by the load-
ing parameters P and C (Eqs. (1) and (2)) that describe the contri-
bution of each variable to each feature. These loading parameters
are then adjusted iteratively by exchanging information between
the new feature spaces T and U with residual error H using the
inner relation in Eq. (3). This is done to maximize correspondence
of the input features with the features summarizing the responses.
The objective is to identify a set of features that best capture the
variability in the input marker space X by controlling for residual
error E while also maximizing their relevance to the response vari-
ables Y by minimizing residual error F. The NIPALS algorithm
(nonlinear partial least squares) [24] was used to solve for the
optimal set of parameter values in Eqs. (1)–(3) presented here in
standard matrix form, i.e., subscript being row dimension, and
superscript being column dimension. Importantly these features
are computed with the constraint that they be mutually orthogonal
or statistically uncorrelated. This basic property supports the unbi-
ased parameter estimates with minimal error and the decomposi-
tion of output variability into fractions directly assignable to each
model component:
n ½X
n
k ¼ n ½T n A A ½P 0 A k þ n ½E n k ð1Þ
n ½Y
n
m ¼ n ½U n A A ½C 0 A m þ n ½F n m ð2Þ
n ½U A ¼ n ½T n A þ n ½H n A ð3Þ
n
Here the objective of the analysis is to assign a subject to either

the GWI or the healthy control group on the basis of their profile in
12 markers of neuroendocrine and immune function. This requires
a variant of PLS called PLS discriminant analysis (PLS-DA) [25]
where a dummy outcome variable Y is created that describes mem-
bership to diagnostic class. In this case two such binary variables
were created assigning membership yik of each subject i to the
control group k ¼ 1 and the GWI group k ¼ 2, respectively, with
yik ∈ [0,1]. Each feature a A is expressed as a linear combination
of the original cytokine, hormone, and neurotransmitter concen-
trations xk, each weighted by a loading coefficient pka.
Uncertainty in the estimation of the feature weights expressed
in terms of their standard error was computed using leave-one-out
cross-validation [26, 27] where the model is repeatedly estimated
with each subject omitted in turn. These multiple cross-validation
models provide populations of parameter estimates from which
significance may be estimated. For example the significance of the
difference between a feature score for two individual subjects was
based on the dispersion of score values obtained during cross-
validation for each subject.
All multivariate methods such as principal component analysis
(PCA), partial least squares (PLS), and partial least squares discrim-
inant analysis (PLS-DA) calculations were conducted with the
Simca software (Umetrics, Kinnelon, NJ).
2.4.3 Statistical Tools Classification performance statistics and receiver operating charac-
for Assessing teristic (ROC) curves were calculated using functions found in the
the Diagnostic SPSS Statistics software package (IBM, Armonk, NY) and the
Performance MatLab Statistics Toolbox software (MathWorks, Natick, MA).
3 Methods
The analysis that is the focus of this chapter is best described by

working through the following example involving the recruitment
and assessment of a suitable subject group (Subheading 3.1), the
measurement of a relevant physiological response (Subheading
3.2), the collection and analysis of biological samples (Subheading
3.3), and finally the identification and assessment of diagnostic
markers (Subheading 3.4).
Application of the abovementioned tools, protocols, and assays
resulted in the creation of a dataset consisting of 58 human subjects
distributed across 3 diagnostic groups and both genders. In the
data subset of interest each subject was described by 12 markers of

neuroendocrine and immune function assessed in blood collected
at three points in time during a standard exercise challenge.
3.1 Recruitment of a In this example, a first subset of 10 GWI and 11 control subjects
Study Cohort. described previously [9] were recruited from the Miami Veterans
Administration Medical Center and studied. Subjects were male
and ranged in age between 30 and 55. Subjects were in good health
prior to 1990, and had no current exclusionary diagnoses that
could reasonably explain the symptoms and their severity. Subjects
taking medications that may impact immune function were
excluded (e.g., steroids, immunosuppressive agents). Control sub-
jects consisted of Gulf War-era sedentary veterans and were
matched to GWI subjects by age, gender, body mass index
(BMI), and ethnicity. Controls were also selected and matched to
patients on the basis of their general fitness level, estimated a priori
from their history of activity, and confirmed during exercise chal-
lenge. Consistent with these criteria and specific to this analysis a
second group was subsequently recruited consisting of an addi-
tional 16 male and 10 female GWI subjects as well as 9 male CFS
subjects and 2 new male control subjects. This new group of sub-
jects was used as a separate validation set in this example.
3.2 Applying The McArdle protocol [28] for graded exercise was used to elicit an
Standard Exercise immune-endocrine response. Subjects pedal at an initial output of
Challenge 60 W for 2 min, followed by an increase of 30 W every 2 min until
the subject (1) reaches a plateau in maximal oxygen consumption
(VO2); (2) reaches a respiratory exchange ratio above 1.15; or
(3) stops the test.
3.3 Sample Prior to the exercise challenge a first blood draw was conducted
Collection after subjects sat quietly for 30 min. Second and third blood draws
and Analysis were conducted upon reaching peak effort (VO2 max) and at 4 h
post-exercise, respectively. At each blood draw, five 8 mL tubes of
blood were collected in CPT vacutainers (BD Biosciences, San Jose,
CA). Importantly, diurnal variations in this and other indicators
were controlled by conducting assessments at the same time of day
for all subjects. Blood samples were processed and markers
recorded as described in Subheading 2.3.
3.4 Numerical 1. Calculate the change in median expression of each cytokine

Analysis occurring between the GWI and the healthy control subjects
and test for significance. A standard Mann-Whitney nonpara-
3.4.1 Assessing Changes metric test can be used for this purpose and applied at each time
in the Expression point separately.
of Individual Biomarkers Although results presented in Table 1 suggest a general over-
expression of cytokines in the GWI group only a subset of these
Table 1
Changes in the expression of signaling molecules in GWI
Signal molecule GWI(t0) Ctrl(t0) GWI(t1) Ctrl(t1) GWI(t2) Ctrl(t2)

Saliva
Cortisol 0.13 (0.15) 0.04 (0.49) 0.07 (0.06)
PHA-stimulated blood culture
IL-1α 3.40 (0.66) 9.00 (0.95) 3.00 (0.68)
IL-5 81.00 (0.00) 30.00 (0.00) 41.50 (0.00)
IL-6 111.00 (0.38) 1185.50 (0.10) 2398.50 (0.19)
IL-10 19.00 (0.96) 177.50 (0.42) 269.50 (1.00)
TNFα 206.50 (0.05) 53.00 (0.68) 167.50 (0.10)
INFγ 100.00 (0.07) 170.70 (0.02) 158.20 (0.02)
Plasma
IL-6 10.75 (0.01) 11.85 (0.03) 9.50 (0.02)
IL-10 1.80 (0.27) 3.10 (0.50) 1.65 (0.43)
TNFα 2.00 (0.49) 1.30 (0.95) 1.75 (0.21)
sCD26 85.50 (0.11) 12.00 (0.81) 14.50 (0.86)
NPY 5.03 (0.65} 3.95 (0.55) 11.69 (0.97)
Composite features
Level 1 Feature 1 0.39 (0.01) 0.62 (0.01) 0.46 (0.00)
Level 1 Feature 2 0.01 (0.81) 0.06 (0.86) 0.31 (0.38)
Difference in median expression values (significance p-value, Mann-Whitney) between GWI and healthy controls (Ctrl)
prior to exercise (t0), at peak effort (t1), and 4 h post-exercise (t2). PHA-stimulated ex vivo culture cytokines are reported
as pg/105 lymphocyte cells. All signals measured in plasma are expressed in pg/mL. Salivary cortisol is reported in μg/dL
achieved statistical significance. In particular we observed sig-

nificantly higher response to PHA stimulation in GWI subjects
for IL-5 at all three sample times ( p 0.001) and in IFNγ with
challenge ( p ¼ 0.02, 0.02). Marginally higher TNF-α respon-
siveness was also observed ( p ¼ 0.05) in PHA-stimulated
culture but only at rest (t0). This was accompanied by signifi-
cantly higher concentrations of IL-6 in plasma ( p 0.03) both
prior to and throughout the challenge. Interestingly, no signif-
icant differences were observed in salivary cortisol, NPY, and
ex vivo-stimulated culture for IL-1, IL-6, and sCD26. The
same applies to IL-10 measured both in culture and in plasma.
2. Evaluate significance of trends in the expression of each marker
across time. The nonparametric Friedman test for repeated
measures and the standard two-way ANOVA were used for
this purpose.
In this example the majority of significant trends were found in

the response of healthy control subjects. A systematic progres-
sion across time was found for IL-1α ( p ¼ 0.01) and IL-5
( p ¼ 0.05) in PHA-stimulated culture for healthy controls as
well as in the expression of IL-10 in culture ( p ¼ 0.04) and in
plasma ( p ¼ 0.05). Similar exercise-induced changes in NPY
( p 0.001) were also observed. In comparison GWI subjects
as a group displayed a lower responsiveness to exercise. Only
IL-5 concentrations in PHA-stimulated culture showed any
significant trend ( p ¼ 0.03) in GWI and produced a significant
group x time interaction effect based on a parametric two-way
ANOVA. Though trends in both salivary cortisol ( p ¼ 0.06
controls; 0.15 GWI) and TNF-α ( p ¼ 0.10 controls; 0.08
GWI) failed significance under the Friedman test, results from
the two-way ANOVA showed a significant interaction of diag-
nostic group with time for these markers ( p ¼ 0.04, 0.01).
3.4.2 Isolating Patterns The nervous, immune, and endocrine systems are generally recog-
of Biomarker nized as being highly integrated regulatory networks. As a result,
Co-expression even though there were relatively few differences in the absolute
expression of individual markers, there may still be co-regulatory
imbalances. Because the basic PLS algorithm described in Subhead-
ing 2.4.2 can be applied to all data points independently of sample
time the dynamics of the immune response are not captured. To
address this we applied a two-step analysis commonly referred to as
three-way PLS or batch PLS [29, 30]. This model is described in
Fig. 1.
1. First PLS was applied to all measurements across all individual
markers using sample time t as a response Y in level 1 of the
model (steps 1–2). This was done to enforce the sequence of
values when identifying the features that best described the
patterns of co-expression linking the 12 markers (X in level
1). In a pre-processing step all regressor variables in the X block
were log transformed. Each log-transformed variable was then
scaled by dividing with the corresponding standard deviation
computed about zero instead of the mean. The response vari-
able Y describing the progression of each batch or exercise
challenge was not transformed nor was it centered. To capture
the unequal spacing in time of the three observations made
during each trial we used an approximate time vector of [t0, t1,
t2] ¼ [0, 30, 270] in minutes. In addition, the training set used
here contains roughly 8% missing array elements. Missing data
was handled implicitly by the NIPALS algorithm using an
extension to the original computation proposed by Christof-
fersson (1970) [31] that consists essentially in iteratively sub-
stituting missing values with their model predictions. This
approximate method performs well with up to 10–20% missing
Fig. 1 PLS-DA classification model structure. Diagram of the input and output relationships defined at each
level of the PLS model for classification of subjects to the GWI or healthy control class
data and has been implemented in the Simca software package

mainly because of its rapid and robust convergence properties.
A comparison with alternate approaches more suitable for
higher fractions of missing data may be found in Grung and
Manne (1998) [32].
In the current example each sample is described by 12 biomar-
kers, 6 of which were measured in culture, 5 in plasma, and 1 in
saliva. This level 1 analysis produced a base model supported by
two significant features that captured approximately 89% of the
total variability (Rx2 ¼ 0.89) in the data (Table 2). The contri-
bution of each biomarker to these features is shown in Fig. 2.
Accounting for 82% of the variability, the first feature described
a broad-spectrum response to exercise. Expression of NPY and
cytokines measured in plasma as well as in ex vivo culture all
correlated positively with one another and negatively with
salivary cortisol. Accounting for 7% of the overall variability, a
second pattern was defined by the co-expression of salivary
cortisol, IL-6 and IL-10 in plasma, and an opposite IL-1α
response in culture.
2. Features identified in step 1 are then used to define a new input
array for level 2 of the analysis. At level 2 the input variables to
the classification model were the scores Ta(t,i) in each feature
Table 2
PLS model fit and structure
Component
Model a R2X R2X(cum) Eigenvalues R2Y R2Y(cum) Q
2
Q2(cum)
Level 1 Model
Observations (N ) ¼ 69 1 0.83 0.83 9.91 0.40 0.40 0.40 0.40
K ¼ 12 Biomarkers 2 0.07 0.89 0.81 0.06 0.46 0.04 0.42
vs. Time
Level 2 Model
Observations (N ) ¼ 21 1 0.53 0.53 3.19 0.38 0.38 0.33 0.33
2 Level-1 features at 2 0.32 0.85 1.93 0.09 0.46 0.04 0.36
3 times (K ¼ 6)
R2x and R2y are Pearson correlation coefficients or total sum of squared regression (SSR) captured by the partial least
squares (PLS) model for the regressor X (molecular signals) and predicted variable Y (diagnostic class) spaces, respec-
tively. The value Q2 ¼ 1-PRESS/SST expresses overall model performance in terms of reduction in the predicted residual
sum of squared errors (PRESS) in leave-one-out cross-validation
a evaluated for each subject i at each time point t in step 1 of

the analysis, namely T1(t0,i), T1(t1,i), T1(t2,i), T2(t0,i), T2(t1,
i), and T2(t2,i). These feature scores were not transformed and
were scaled to unit variance before estimation of the level
2 model. PLS-DA was then applied (see Subheading 2.4.2) to
manage the temporal relationships linking feature scores at
each sample point in this new level 2 feature space. Because
PLS-DA carries information about the diagnostic classes, the
time course features extracted in level 2 are those that provide
optimal discrimination of GWI patients from the control
subjects.
This second step produces a set of level 2 features, each con-
sisting of weighted sums in T1(t0,i), T1(t1,i), T1(t2,i), T2(t0,i),
T2(t1,i), and T2(t2,i). In this way the full time course in each
co-expression pattern (T1 and T2) identified in level 1 is cap-
tured for each subject. Moreover by weighting individual ele-
ments of the time course it is possible to emphasize or diminish
the contribution of a level 1 score at specific phases of the
challenge. Scores produced at level 2 of the analysis now sum-
marize for each subject the complete progression across all time
points for all 12 markers. The position of a subject’s profile in
this new level 2 feature space is then used by the PLS-DA
model to predict a degree of membership to the GWI illness
class.
In this example two level 2 features were identified, each sum-
marizing almost exclusively the corresponding time course
from the first level of the model. The progression in time of
all 12 markers for each subject may now be expressed in terms
Fig. 2 Structure of composite features. The relative contribution pka of each biomarker k to each of the two
level 1 model features (a ¼ 1, 2) that optimally separate the subjects into their respective diagnostic groups.
The first feature captures a broad-spectrum disturbance across all biomarkers. The second feature captures a
pattern of co-expression whereby increases in IL-1a are offset by a corresponding increase in cortisol, IL-6,
and IL-10. Whiskers represent the standard error in estimation for each weight pka. All level 1 feature
1 weights are significant at p 0.001. Only the feature 2 weights for cortisol ( p 0.001), IL-1a ( p ¼ 0.024),
IL-6 ( p ¼ 0.008), and IL-10 ( p 0.001) were significant at p < 0.05
of only two level 2 scores (Fig. 3). Together they summarized

roughly 85% of the variation across time of both level 1 features
(Table 2). They also captured close to half of the total varia-
bility in diagnostic class membership (Ry2 ¼ 0.46) even though
a relatively narrow assessment of immune function was used.
Leave-one-out cross-validation indicated that this model was
also reasonably robust in predicting diagnostic class for data
excluded from model fitting (Q2 ¼ 0.36). Close to 40% of
variability in class membership was accounted for by broad-
spectrum changes across time (feature 1). Another 9% was
captured by the progression in feature 2. Interestingly this
sequential set of scores contributed positively to the classifica-
tion model’s performance (Q2 ¼ 0.04) even though no statis-
tically significant separation of individual median values was
found for feature 2. This underscores the fact that feature
2 as computed in the PLS regression model adds important
detail to feature 1 and should be used in conjunction with the
latter.
Fig. 3 Separation of subjects in feature space. Distribution of GWI and control subjects in a feature space
defined by the co-expression of 12 biomarkers of neuroendocrine-immune status. Each axis is defined by a
specific linear combination of biomarkers and the evolution in these patterns over the course of a graded
exercise challenge. Each point captures the entire time course recorded for that subject. In this feature space
4 of 11 control subjects appears to group with the 10 GWI subjects
3. Using model cross-validation as an estimate of variability in

individual time-course profiles we can now compare one sub-
ject to another on the basis of changes in level 1 patterns
across time.
To illustrate this we compared the progression of level 1 feature
scores in a case-matched (age, BMI, and % of predicted maxi-
mum VO2) pair of GWI patients and healthy controls: CS201
and PC 105 (Fig. 4a, b). While these subjects separate on the
basis of feature 1 alone, controlling for other sources of varia-
bility by comparing a matched pair now reveals significant
differences in feature 2 especially at peak effort
(t1) ( p ¼ 0.05). Examination of the level 2 classification
model revealed that the contribution of feature 2 scores at
time t1 was weighted an order of magnitude higher
(0.180 t1) than those at time t0 and t2 (0.016 t0, 0.017 t2).
In fact only at t1 were feature 2 scores weighted similarly to
those for feature 1 (0.278 t0, 0.227 t1, 0.299 t2).
Fig. 4 Evolution of feature scores in time for matched pairs. Median expression of level 1 feature 1 (panel A)
and feature 2 (panel B) prior to exercise (t0), at peak effort (t1), and 4 h post-exercise (t2) is shown in panels A
and B for GWI subject CS 201 (blue diamonds) and case-matched control subject PC 105 (red squares).
Panels C and D show data for GWI subject VR 203 (blue diamonds) and control subject CC 104 (red squares).
Error bars display the median absolute deviation from median (MADM) as computed from repeated cross-
validation for these observations. In panels A and B time and illness effects can be seen in both level 1 feature
1 and feature 2. Significant deviations at peak effort in feature 2 illustrate why this time point is highly
weighted in the level 2 model. In panels C and D we observe significant time effects in level 1 feature 1 and
while illness results in a persistent offset in the expression of feature 2
The important detail added by feature 2 can also be visualized

further by comparing two individuals who share very similar
feature 1 scores: GWI subject VR 203 and control subject CC
104. Though these individuals share a very similar broad-
spectrum baseline immune activation (feature 1), they differ
dramatically in their expression of feature 2 scores across time
(Fig. 4c, d). Both these examples and the structure of the
classification model suggest that differences in feature 2 score,
in particular at peak effort, may be of specific diagnostic value
and that this contribution may be obscured by high within-
group dispersion. Interestingly feature 2 scores summarized at
level 2 of the model correlated significantly with differences in

physical fatigue ( p < 0.01) and to a lesser extent general fatigue
( p ¼ 0.06) among GWI subjects but not in control subjects
( p ¼ 0.38, 0.11, respectively). No significant correlation with
either physical or general fatigue was found in GWI for feature
1 ( p ¼ 0.27, 0.85). In contrast these were borderline signifi-
cant in healthy controls ( p ¼ 0.05, 0.05). This would suggest
that feature 2 might serve as an indicator of relative illness
severity in GWI.
3.4.3 Exercise Response To assess classifier performance the continuous scale for class mem-
as a Basis for Classification bership can be transformed to a discrete class assignment by apply-
ing a membership threshold. Detection sensitivity and assignment
specificity values can then be computed using these binary assign-
ments. This is repeated over a range of threshold values to produce
a receiver-operator characteristic (ROC) curve [33].
1. In this example separate receiver-operator characteristic (ROC)
curves were computed for classification using the broad-
spectrum disturbance alone (feature 1) as well as classification
including the inflammatory imbalance pattern (feature 2)
(Fig. 5, Table 3). Both classifiers performed similarly on the
training set of 10 male GWI and 11 male healthy control
subjects, each producing an area under the curve (AUC) sig-
nificantly better than random assignment ( p ¼ 0.001 Table 3).
Nonetheless, at 85% specificity the use of both features delivers
a sensitivity of 90% rather than 70% possible with one feature
alone.
2. Performance differences became very noticeable when these
models were applied to a test set. Our first validation set con-
sisted of 16 new male GWI subjects, 9 male CFS controls, and
2 new male healthy controls. The corresponding ROC curves
for the 1 and 2 feature classifiers are shown in Fig. 5. When
tested on these new male subjects, classification using only the
first feature does not differ significantly from a random assign-
ment ( p ¼ 0.23 AUC). Inclusion of the second feature deliv-
ered a significant increase in AUC ( p ¼ 0.04 Table 3) with 90%
specificity available at 70% sensitivity. Recall that the training set
contained only male GWI and control subjects. To test the
impact of gender we constructed a second validation set by
adding ten female GWI cases to the first set. Classification
performance decreased drastically. Neither model performed
better than random assignment although two features were still
preferable ( p ¼ 0.11) to one ( p ¼ 0.31) (Table 3).
3. We intentionally included CFS subjects in the validation set as a
disease control group since CFS is clinically similar to GWI.
Using the ROC analysis performed on the training set we
Fig. 5 Linear classifier performance. Receiver-operator characteristic (ROC) curves for a linear discriminant
model constructed from cytokine, NPY, and cortisol co-expression in 10 GWI and 11 male control subjects.
Sensitivity and specificity of classification obtained at various membership threshold values are presented for
the original training set as well as for a validation test set consisting of 9 male CFS patients as well as 16 new
GWI and 2 new male control cases
choose a predicted membership threshold of 0.35 above which

we assigned a given case to the GWI group. This corresponded
in the training set to a sensitivity of 100% and a specificity of
70% or the correct assignment of 7 of 11 healthy controls and
all 10 GWI subjects. Among the GWI test cases, we found
correct assignment of 11 out of 16 male cases but only of
3 of 10 female cases. With regard to the CFS control subjects,
only two of nine subjects were classified as GWI with the other
seven classified correctly as non-GWI. Recall that no CFS con-
trols were used to train this classifier. Therefore these patterns
of immune marker co-expression appear reasonably specific to
GWI in male patients even when presented with clinically simi-
lar CFS subjects.
Table 3
Receiver-operator characteristics (ROC) for PLS-DA classifiers
Asymptotic 95%
Area Std.
under Error Asymptotic Upper Lower
Model Data set Positive:Negative curve (a) Sig. (b) bound bound
Feature Training set 10:11 Ctrl 0.900 0.064 0.001 0.774 1.026
1 Only
Feature Training set 10:11 Ctrl 0.900 0.065 0.001 0.772 1.028
1 and 2
Feature Test set (male 16:9 CFS, 2 Ctrl 0.646 0.122 0.234 0.407 0.885
1 Only only)
Feature Test set (male 16:9 CFS, 2 Ctrl 0.750 0.110 0.042 0.534 0.966
1 and 2 only)
Feature Test set (male 26 (incl. 10 female): 0.615 0.117 0.308 0.387 0.844
1 Only and female) 9 CFS, 2 Ctrl
Feature Test set (male 26 (incl. 10 female): 0.679 0.113 0.113 0.458 0.901
1 and 2 and female) 9 CFS, 2 Ctrl
Detailed statistics for receiver-operator curves (ROC) describing the sensitivity as a function of 1 specificity obtained in
classification models based on partial least squares discriminant analysis (PLS-DA). These include the area under the
curve (AUC) with the associated standard error of estimation, the corresponding significance p-value, as well as the upper
and lower confidence limits at the 0.05 significance level. Statistics are reported for classification in the training set, a
male-only test set, and a mixed-gender set
4 Notes
1. In recent work [11] we demonstrated that while the expression

of individual immune and endocrine markers provides some
diagnostic information in the study of GWI, the patterns
through which these markers coordinate their expression
offer significant new insight into this illness. Indeed while
relatively few individual markers changed significantly in
expression, the co-expression network linking these markers
underwent widespread changes in organization and structure
in GWI.
2. While structural changes in the patterns of biomarker
co-expression offer a highly discriminatory fingerprint the sam-
ple burden required for identifying these changes in individual
patients is substantial and an alternative method must be con-
sidered if molecular phenotyping is to be deployed clinically
[48]. Here we have attempted to identify a subset of immune
marker associations that are preserved across all subjects but
might be expressed to a different extent and perhaps in a
different direction in GWI patients. We have also focused on
linear interactions to make the problem even more tractable.
3. The analysis presented here identified two such shared patterns

linking 12 immune mediators measured in blood and saliva.
The majority of GWI subjects separated from healthy controls
on the basis of a common trend whereby a higher-than-
expected expression of most immune markers coincided with
a lower-than-expected concentration of salivary cortisol (fea-
ture 1). As reported previously [11] this broad-spectrum
response included the joint overexpression of individual inflam-
matory markers, Th1 and Th2 cytokines, namely IL-6 in
plasma as well as IL-5, TNF-α, and INF-ɣ in PHA-stimulated
culture. Recent evidence suggests that sensitization with a
foreign antigen mimicking self can induce a similar mixed
Th2/Th1 cytokine profile [34]. Human proteins with struc-
tural similarity to exogenous allergens can cross-react with
allergen-specific IgE antibodies, inducing sustained inflamma-
tion and lymphocyte proliferation [35–38]. Among these Hom
s 2 [37] and Hom s 4 [36] are known to induce IgE reactivity
as well as a strong IFN-γ response.
4. While an offset in baseline immune status (feature 1) was
sufficient to identify most GWI subjects, a subset of these
could not be distinguished on this basis alone. This subset
supported the identification of a second co-expression pattern
(feature 2). This pattern suggests that GWI subjects might
differ from controls in their acute-phase regulation of cortisol,
IL-6, IL-10, and possibly TNFα. This was accompanied by an
altered lymphocyte IL-1α responsiveness to PHA stimulation, a
cytokine associated with “sickness behavior” [39]. Because of
the high degree of heterogeneity in this preliminary cohort,
differences in this second co-expression pattern achieved sig-
nificance only in case-matched GWI and control subjects. This
was especially true at peak effort where the contribution of
feature 2 was significantly weighted.
5. Though subtle, small changes in feature 2 were nonetheless
instrumental in maintaining high specificity and sensitivity in
the classification of new test subjects with GWI as well as
subjects suffering from CFS (90% specificity and 70% sensitiv-
ity). Interestingly this signature involves important myokines
and metabolically active cytokines, such as IL-6 and TNF-α
with key roles in energy metabolism. IL-6 has been called a true
“exercise factor” as levels of this cytokine increase exponentially
up to 100-fold during exercise [40]. This is generally followed
by a compensatory expression of the anti-inflammatory cyto-
kine IL-10 and cortisol. The release of cortisol is also mediated
in part by IL-6 via feedback to the HPA axis [41]. Independent
of this, IL-6 is known to activate one of the principal regulators
of cellular energy status: the AMP-activated protein kinase
(AMPK) pathway. IL-6 has also been recently reported to
have an insulin-sensitizing effect [42–44]. Conversely increases

in TNFα promote insulin resistance [45] and expression of this
cytokine are inhibited by IL-6 [46]. Availability of metabolic
energy stores therefore relies at least in part on a balance of
these hormones and cytokines.
In this respect it was interesting to note that subjects with CFS,
another fatiguing illness, did not manifest this pattern of cyto-
kine co-expression to the same extent as GWI subjects. Indeed
only two of the nine CFS controls were assigned to the GWI
group by the classification model. Consequently even though
metabolic repercussions may be similar, the basic mechanisms
driving neuroendocrine-immune dysfunction might be quite
different in GWI and these illnesses may constitute very distinct
regulatory regimes [47].
6. Though our results appear to extrapolate well to new subjects
in a separate validation cohort the overall number of subjects
examined remains relatively small. Therefore we suggest this
analysis be considered preliminary even though the group size
is consistent with that used recently in related studies [4, 5]. In
addition the panel of cytokines used was relatively narrow and
by no means constituted a complete survey of the immune
response. In recognition of this a simple linear classification
model was chosen as a more conservative approach to guard
against over-fitting the data.
Acknowledgments
This work was supported by grants from the NIAAA:

R21AA016635 (PI MA Fletcher); NIAID: R01AI065723
(PI MA Fletcher); CFIDS Assoc. of America [2]: (PI N Klimas,
PI G Broderick); NIAID: UO1 AI459940 (PI N Klimas);
NIAMSD R01 AR057853-01A1 (PI N Klimas); VA merit awards
(PI N Klimas); Dynamic Modeling in GWI,CFS/ME; GW080152
(PI N Klimas); and the University of Alberta (PI G Broderick).
References
1. Kang HK, Natelson BH, Mahan CM et al cortisol secretion in patients with fibromyalgia
(2003) Post-traumatic stress disorder and and/or chronic fatigue syndrome. Brain Behav
chronic fatigue syndrome-like illness among Immun 18:314–325
Gulf War Veterans: a population-based survey 4. Golier JA, Schmeidler J, Legge J et al (2006)
of 30,000 veterans. Am J Epidemiol Enhanced cortisol suppression to dexametha-
157:141–148 sone associated with Gulf War deployment.
2. Unwin C, Blatchley N, Coker W et al (1999) Psychoneuroendocrinology 31
Health of UK servicemen who served in Per- (10):1181–1189
sian Gulf War. Lancet 353:169–178 5. Golier JA, Schmeidler J, Legge J et al (2007)
3. Crofford LJ, Young EA, Engleberg NC et al Twenty-four hour plasma cortisol and adreno-
(2004) Basal circadian and pulsatile ACTH and corticotropic hormone in gulf war veterans:
relationships to posttraumatic stress disorder and baseline characteristics. Control Clin Trials
and health symptoms. Biol Psychiatry 62 23(3):333–353
(10):1175–1178 19. Smets EM, Garssen B, Bonke B et al (1995)
6. Maher K, Klimas NG, Fletcher MA (2003) The multidimensional fatigue inventory (MFI)
Immunology. In: Jason LA, Fennell PA, Taylor psychometric qualities of an instrument to
RR (eds) Handbook of chronic fatigue syn- assess fatigue. J Psychosom Res 39(3):315–325
drome. Wiley, Hoboken, pp 124–151 20. Ware JE Jr, Sherbourne CD (1992) The MOS
7. Vojdani A, Thrasher JD (2004) Cellular and 36-item short-form health survey (SF-36). I
humoral immune abnormalities in Gulf War Conceptual framework and item selection.
veterans. Environ Health Perspect 112 Med Care 30(6):473–483
(8):840–846 21. Cadore E, Lhullier F, Brentano M et al (2008)
8. Siegel SD, Antoni MH, Fletcher MA et al Correlations between serum and salivary hor-
(2006) Impaired natural immunity, cognitive monal concentrations in response to resistance
dysfunction, and physical symptoms in patients exercise. J Sports Sci 26(10):1067–1072
with chronic fatigue syndrome: preliminary 22. Wold S, Trygg J, Berglund A et al (2001) Some
evidence for a subgroup? J Psychosom Res 60 recent developments in PLS modeling. Che-
(6):559–566 mom Intell Lab Syst 58:131–149
9. Whistler T, Fletcher MA, Lonergan W et al 23. Eriksson L, Antti H, Gottfries J et al (2004)
(2009) Impaired immune function in Gulf Using chemometrics for navigating in the large
War Illness. BMC Med Genet 2:12 data sets of genomics, proteomics, and meta-
10. Broderick G, Fuite J, Kreitz A et al (2010) A bonomics (gpm). Anal Bioanal Chem 380
formal analysis of cytokine networks in chronic (3):419–429
fatigue syndrome. Brain Behav Immun 24 24. Wold S, Esbensen K, Geladi P (1987) Principal
(7):1209–1217 component analysis. Chemom Intell Lab Syst
11. Broderick G, Kreitz A, Fuite J et al (2011) A 2:37–52
pilot study of immune network remodeling 25. Barker M, Rayens W (2003) Partial least
under challenge in Gulf War Illness. Brain squares for discrimination. J Chemom 17
Behav Immun 25(2):302–313 (3):166–173
12. Moss RB, Mercandetti A, Vojdani A (1999) 26. Efron B, Gong G (1983) A leisurely look at the
TNF-alpha and chronic fatigue syndrome. J bootstrap, the jackknife, and cross-validation.
Clin Immunol 19(5):314–316 Am Statist 37:36–48
13. Gaab J, Rohleder N, Heitz V et al (2005) 27. Krzanowski WJ (1987) Cross-validation in
Stress-induced changes in LPS-induced proin- principal component analysis. Biometrics
flammatory cytokine production in chronic 43:575–584
fatigue syndrome. Psychoneuroendocrinology 28. McArdle WD, Katch FI, Katch VL (2007)
30:188–198 Exercise physiology: energy, nutrition, and
14. Skowera A, Cleare A, Blair D et al (2004) High human performance. Lippincott Williams &
levels of type 2 cytokine-producing cells in Wilkins, London, p 954
chronic fatigue syndrome. Clin Exp Immunol 29. Wold S, Kettaneh N, Friden H et al (1998)
135(2):294–302 Modelling and diagnostics of batch processes
15. ter Wolbeek M, van Doornen LJ, Kavelaars A and analogous kinetic experiments. Chemom
et al (2007) Longitudinal analysis of pro- and Intell Lab Syst 44:331–340
anti-inflammatory cytokine production in 30. Antti H, Bollard ME, Ebbels T et al (2002)
severely fatigued adolescents. Brain Behav Batch statistical processing of H-1 NMR-der-
Immun 21(8):1063–1074 ived urinary spectral data. J Chemom
16. Light AR, White AT, Hughen RW et al (2009) 16:461–468
Moderate exercise increases expression for sen- 31. Christoffersson A (1970) The one component
sory, adrenergic, and immune genes in chronic model with incomplete data. Ph.D. thesis,
fatigue syndrome patients but not in normal Uppsala University
subjects. J Pain 10(10):1099–1112
32. Grung B, Manne R (1998) Missing values in
17. Fukuda K, Nisenbaum R, Stewart G et al principal component analysis. Chemom Intell
(1998) Chronic multisymptom illness affecting Lab Syst 42:125–139
Air Force veterans of the Gulf War. JAMA
280:981–988 33. Greiner M, Pfeiffer D, Smith RD (2000) Prin-
ciples and practical application of the receiver-
18. Collins JF, Donta ST, Engel CC et al (2002) operating characteristic analysis for diagnostic
The antibiotic treatment trial of Gulf War tests. Prev Vet Med 45(1–2):23–41
Veterans’ illnesses: issues, design, screening,
34. Bünder R, Mittermann I, Herz U et al (2004) cortisol in humans. Am J Physiol Endocrinol

Induction of autoallergy with an environmen- Metab 285(2):E433–E437
tal allergen mimicking a self protein in a murine 42. Carey AL, Steinberg GR, Macaulay SL et al
model of experimental allergic asthma. J (2006) Interleukin-6 increases insulin-
Allergy Clin Immunol 114(2):422–428 stimulated glucose disposal in humans and glu-
35. Fedorov AA, Ball T, Mahoney NM et al (1997) cose uptake and fatty acid oxidation in vitro via
The molecular basis for allergen cross- AMP-activated protein kinase. Diabetes 55
reactivity: crystal structure and IgE-epitope (10):2688–2697
mapping of birch pollen profilin. Structure 5 43. Choy EH, Isenberg DA, Garrood T et al
(1):33–45 (2002) Therapeutic benefit of blocking
36. Aichberger KJ, Mittermann I, Reininger R et al interleukin-6 activity with an anti-interleukin-
(2005) Hom s 4, an IgE-reactive autoantigen 6 receptor monoclonal antibody in rheumatoid
belonging to a new subfamily of calcium- arthritis: a randomized, double-blind, placebo-
binding proteins, can induce Th cell type controlled, dose-escalation trial. Arthritis
1-mediated autoreactivity. J Immunol 175 Rheum 46(12):3143–3150
(2):1286–1294 44. Wallenius V, Wallenius K, Ahrén B et al (2002)
37. Mittermann I, Reininger R, Zimmermann M Interleukin-6-deficient mice develop mature-
et al (2008) The IgE-reactive autoantigen onset obesity. Nat Med 8(1):75–79
Hom s 2 induces damage of respiratory epithe- 45. Plomgaard P, Bouzakri K, Krogh-Madsen R
lial cells and keratinocytes via induction of et al (2005) Tumor necrosis factor-alpha
IFN-gamma. J Invest Dermatol 128 induces skeletal muscle insulin resistance in
(6):1451–1459 healthy human subjects via inhibition of Akt
38. Valenta R, Mittermann I, Werfel T et al (2009) substrate 160 phosphorylation. Diabetes 54
Linking allergy to autoimmune disease. Trends (10):2939–2945
Immunol 30(3):109–116 46. Starkie R, Ostrowski SR, Jauffred S et al (2003)
39. Dantzer R, O’Connor JC, Freund GG et al Exercise and IL-6 infusion inhibit endotoxin-
(2008) From inflammation to sickness and induced TNF-alpha production in humans.
depression: when the immune system subju- FASEB J 17(8):884–886
gates the brain. Nat Rev Neurosci 9 47. Ben-Zvi A, Vernon SD, Broderick G (2009)
(1):46–56. Review Model-based therapeutic correction of
40. Pedersen BK, Akerström TC, Nielsen AR et al hypothalamic-pituitary-adrenal axis dysfunc-
(2007) Role of myokines in exercise and tion. PLoS Comput Biol 5(1):e1000273
metabolism. J Appl Physiol 103 48. Broderick G, Craddock RC, Whistler T et al
(3):1093–1098. Review (2006) Identifying illness parameters in fatigu-
41. Steensberg A, Fischer CP, Keller C et al (2003) ing syndromes using classical projection meth-
IL-6 enhances plasma IL-1ra, IL-10, and ods. Pharmacogenomics 7(3):407–419.
Chapter 8
Breaking Away: The Role of Homeostatic Drive

in Perpetuating Depression
J. Tory Toole, Mark A. Rice Jr, Travis J. A. Craddock, Barry Nierenberg,
Nancy G. Klimas, Mary Ann Fletcher, Joel Zysman, Mariana Morris,
and Gordon Broderick
Abstract
We propose that the complexity of regulatory interactions modulating brain neurochemistry and behavior is
such that multiple stable responses may be supported, and that some of these alternate regulatory programs
may play a role in perpetuating persistent psychological dysfunction. To explore this, we constructed a
model network representing major neurotransmission and behavioral mechanisms reported in literature as
discrete logic circuits. Connectivity and information flow through this biobehavioral circuitry supported
two distinct and stable regulatory programs. One such program perpetuated a depressive state with a
characteristic neurochemical signature including low serotonin. Further analysis suggested that small
irregularities in glutamate levels may render this pathology more directly accessible. Computer simulations
mimicking selective serotonin reuptake inhibitor (SSRI) therapy in the presence of everyday stressors
predicted recidivism rates similar to those reported clinically and highlighted the potentially significant
benefit of concurrent behavioral stress management therapy.
Key words Computational modeling, Homeostatic regulation, Depression, SSRI, Stress, Neuro-
transmitters, Glutamate, Serotonin, Network complexity, Regulatory logic, Multi-stability
1 Introduction
In 2012, an estimated 16 million adults in the USA had at least one

major depressive episode [1]. A closer look reveals a broad and
complex condition where over the course of a year 6.9% of adults
in the USA experience during a period of 2 weeks or longer either
depressed mood or loss of interest or pleasure, and at least four
other symptoms reflecting a change in function, such as problems
with sleep, eating, energy, concentration, and self-image [2]. This
also means that each year, 1 of every 14 adults is at increased risk of
suicide, the tenth leading cause of death in the USA. Apart from the
obvious personal ramifications of suffering from depression, this
disease also has a serious impact on loved ones, caregivers, and the
121
economy. In 2010, the annual costs associated with major depres-

sive disorder were $210.5 billion. This amount of money is not
only staggering, but it has also risen significantly from the $173.3
billion annual cost estimated in 2005 [3]. These costs come in
many forms, from treatment with medications and psychotherapy
to lost productivity in the workplace from increased absenteeism,
annual sick days, and higher rates of short-term disability. With
such a high cost, one might ask what is being done to alleviate
this crippling issue. There are several common forms of treatment
available for people suffering from depression [4]. One of the most
common forms involves administration of psychotropic medica-
tions. Selective serotonin reuptake inhibitors (SSRIs) and norepi-
nephrine reuptake inhibitors (SNRIs) are just two types of drugs
commonly prescribed to reduce depressive symptoms. More
recently, inhibition of gamma-aminobutyric acid (GABA) transport
has been investigated in animal models of depression [5]. Cognitive
behavioral therapy and interpersonal therapy are among the various
types of psychotherapies used to treat depression [6]. Where med-
ications act directly on the brain chemistry to alleviate symptoms,
psychotherapy acts on mental constructs like thought patterns and
interpersonal relationships and teaches coping skills to reduce
depression. Regardless of the form of treatment, depression has
an unusually high recurrence rate. Half of all patients who recover
from a first episode of depression have one or more additional
episodes in their lifetime, and nearly 80% of people with a history
of two episodes will have at least another recurrence. Individuals
with a history of depression will have five to nine separate episodes
in their lifetime [7]. The numbers alone reveal the gravity of this
situation. The reason why depression even when treated has a
significant recurrence rate is as shrouded in mystery as the very
etiology of the disease. Despite decades of intensive research, the
etiology of depression remains elusive. What is clear is that there are
biological, genetic, environmental, and interpersonal aspects that
combine to cause major depression. Until we better understand
where depression comes from, and what exactly is happening on a
system level, we can never hope to directly target and redirect illness
mechanisms.
In this work, we depart from the more conventional reduction-
ist focus on individual defects in the biology to consider major
depression as an imbalance in the regulatory actions coordinating
the interactions between these elements. Feedback and feed-
forward control of brain chemicals such as neurotransmitters plays
a significant role in maintaining physiological and behavioral stabil-
ity or homeostasis. We propose that these components form an
overarching regulatory system that is capable of supporting multi-
ple stable regulatory programs by virtue of its complexity. Indeed,
many psychological issues such as depression, anxiety, personality
disorders, and post-traumatic stress present as persistent or quasi-
Computing an Escape From Depression 123
stable conditions, resembling a regulatory trap where the person

suffering is “stuck” in a vicious cycle of distress. To explore this
concept, we model interactivity in a simple neurotransmission net-
work to explore the role of naturally occurring brain regulatory
programs and the perpetuation of depression and anxiety as regu-
latory traps. We create an integrated map of known neurotransmis-
sion feedback and feed-forward mechanisms and their association
with basic behavioral constructs and identify the stable regulatory
programs supported by this neurobehavioral circuitry. Applying a
simple ternary logic to simulate the flow of information through
this neurotransmission network indicated that a depressive state
with characteristics of major depression did coincide with a natu-
rally occurring stable regulatory mode and as such would be resis-
tant to intervention by definition. Indeed, simulating SSRI
treatment with this model produced recidivism rates similar to
those observed in clinic. Simulations showed that GABA inhibition
alone was only slightly more effective than SSRI treatment. In all
cases we found a significant contribution of behavioral stress man-
agement to improving predicted rates of remission beyond that
achieved by pharmacologic intervention alone.
2 Materials
The use of causal mechanistic models in identifying complex illness

conditions that may constitute regulatory traps and become resis-
tant to standard therapeutic intervention is the principal focus of
this chapter. The foundation for conducting such an analysis is the
assembly of a connectivity map describing the known and docu-
mented regulatory interactions linking the biological elements of
interest. Kinetic information describing the rate of regulatory
actions in biological systems is typically very sparse except in
lower organisms that can be interrogated in vitro. In contrast
information describing physiological and biochemical connectivity
in human subjects is considerably more accessible. Here we
describe how this information may be harnessed to assemble a
regulatory circuit from literature in Subheading 2.1 and how a
basic regulatory logic defined in Subheading 2.2 may be applied
to the simulation of biological circuit behavior (Subheading 2.3). A
global optimization framework where such simulations may be
used to iteratively refine the design of intervention courses is
described in Subheadings 2.4 and 2.5. This same simulation proto-
col is used in Subheading 2.6 to explore how small excursions in the
levels of individual neurotransmitters and hormones may increase
vulnerability of being drawn into a pathologic regulatory state, in
this case depression.
2.1 A An important obstacle in building models across regulatory systems

Neurobehavioral remains the scarcity of detailed human in vivo kinetic data as its
Regulatory Model collection can present significant health risks to subjects. We cir-
cumvent this by using a discrete logic representation based solely
on literature of physiological and biochemical connectivity to pro-
vide a qualitative description of system behavior. In this exploratory
effort, model content was focused on a core set of key neurochem-
icals that play central roles in psychological function. The neuro-
transmission network nodes in this canonical network were cortisol,
epinephrine, dopamine, neuropeptide Y (NPY), norepinephrine,
GABA, serotonin, glutamate, and acetylcholine. The regulatory
interactions linking these neurochemicals were then derived
through a focused literature review of their known and reported
modes of action.
Cortisol is a steroid hormone released in response to stress, and
therefore is a key element in any psychological homeostatic regime,
as stress plays a significant role in psychology and induction of
maladaptive psychological states like depression. A review of the
literature suggests that cortisol plays a necessary role in activating
enzymes responsible for the synthesis of epinephrine [8]. Further-
more, cortisol can increase levels of dopamine and glutamate in the
brain. Specifically, glucocorticoids contribute to an increase in
dopaminergic activity [9]. They also increase glutamatergic trans-
mission in the prefrontal cortex through modification of postsyn-
aptic NMDA and AMPA receptors [10]. Increased cortisol also
inhibits serotonin levels by increasing serotonin uptake [11].
Elevated levels of epinephrine are known to increase anxiety
[12]. Dopamine, applied iontophoretically, consistently attenuates
the inhibitory actions of GABA [13]. Similarly, dopamine acts
antagonistically with serotonin [14], NPY [15], and norepineph-
rine [16]. Dopamine also has an inverse regulatory relationship
with acetylcholine [17]. According to current literature, NPY can
block the phase shift of glutamate, reducing its cortical levels
[18]. Increases in NPY can also inhibit levels of acetylcholine and
norepinephrine [19], and serotonin [20]. According to Herman
et al. (2003) [21], norepinephrine pathways modulate the synthesis
of GABA in central limbic stress circuits and elevated levels also
promote NPY [22]. Behaviorally, this neurotransmitter can increase
anxiety [23], hyperarousal [24], attention [25], and working mem-
ory [26], and reduce depressive symptom severity [27]. GABA, or
gamma-aminobutyric acid, is a primary inhibitory neurotransmit-
ter of dopamine [28], norepinephrine [29], and serotonin
[30]. Behaviorally active, elevated GABA can also lead to decreased
levels of attention [31]. Contrary to GABA, glutamate is a primary
excitatory neurotransmitter in the brain. It acts by increasing levels
of norepinephrine [32], cortisol [33], and GABA [34], and also
increasing working memory [35] and depression [36]. It can,
however, also inhibit attention [37].
Serotonin is one of the primary neurotransmitters involved in

depression [38] and is often the target of pharmaceutical interven-
tion. Because of the key role serotonin plays in this disorder, it is a
particular important component of this regulatory circuit model.
Elevated serotonin levels are known to have broad-reaching effects.
Certain serotonin receptors can facilitate glutamate release
[39]. Serotonin also depolarizes certain cell types to increase ace-
tylcholine release [40]. Conversely it has been found to have a
reciprocal relationship with norepinephrine [41], dopamine [42],
cortisol [43], and GABA [30]. With regard to behavior, elevated
serotonin can promote hyperarousal [44], but inhibit both anxiety
[45] and depression [46]. One study also found that keeping
serotonin levels high for prolonged periods can lead to increased
physical fatigue [47].
Acetylcholine is a neurotransmitter that acts on both the
peripheral nervous system (PNS) and the central nervous system
(CNS). For the purposes of this model, we consider its effects
within the CNS only. Acetylcholine can promote increased levels
of epinephrine [48], NPY [49], and cortisol [50] while downregu-
lating norepinephrine [51] and serotonin [52]. As a behavioral
modifier, increased acetylcholine synthesis can promote hyper-
arousal [53] as well as increases in attention [54], and working
memory [55].
As mentioned above, each of these synergistic and competing
effects exerted by one neurotransmitter on the other as well as their
collective effects on behavior were represented abstractly as wires in
a logic circuit, with each wire transmitting an agonistic or antago-
nistic control action from a source neurochemical to a target neu-
rochemical. The final circuit was meant to capture a simple basic
regulatory network of the core neurotransmission processes as they
would function under normal circumstances in the absence of
injury- or illness-incurred damage. Although there are likely to be
person-to-person deviations due to a given individual’s genetic
makeup and acquired epigenetic modifications, we focus here on
the use of a single common model to illustrate how a biobehavioral
network might be simulated (Subheading 2.2) and how predicted
responses might be used to design a treatment plan (Subheading
2.3) for escaping persistent illness.
2.2 A Discrete To direct the flow of information through this model circuit we use
Signaling Logic a simple discrete ternary logic where each model behavioral con-
Framework struct and signaling molecule could be expressed at a low, nominal,
or high level. The specific formalism used in this work is an exten-
sion of a paradigm proposed by Mendoza and Xenarios (2006) [56]
and Thomas (1991) [57], and has been reported previously by our
group [58, 59]. We encode documented regulatory mechanisms
within the neuropsychological system using only the direction
(source and target) and type (activator or inhibitor) of interaction.
As data describing the relative magnitude of response to these
control actions remains limited, we consider all neurotransmitter

components and behavioral constructs to be equally responsive to
the control actions of their respective incoming mediators. Accord-
ingly, we also consider the resulting neurochemical and behavioral
response to be equivalent regardless of source stimuli. Each signal-
ing molecule and behavioral construct xi(t) may adopt three dis-
crete states at time t, namely 1 (downregulated), 0 (nominal), and
1 (upregulated). At any point in time t, the state of a system with
N variables can be represented by the vector~ x ðt Þ, such that
~
x ðt Þ ¼ ðx 1 ðt Þ; x 2 ðt Þ; . . . ; x N ðt ÞÞ ð1Þ
where xi(t) is the state of the ith variable in the N variable system
at time t. The image vector ~ x ðt þ 1Þ describes the preferred state
toward which the system evolves in the next time increment. The
state value of the image vector for the ith variable is determined
from its current state and a set of balanced ternary logic statements
based on the current value of each variable and the mode of action
(i.e., promote or inhibit) of the neighboring input variables. These
logic statements are expressed as follows (Eq. (2)):
8 A I
>
> x i1 ðt Þ∨x i2
A
ðt Þ∨ . . . ∨x iXA
ðt Þ ∇ x i1 ðt Þ∨x i2
I
ðt Þ∨ . . . ∨x iY
I
ðt Þ
< A
x i ðt þ 1Þ ¼ x i1 ðt Þ∨x i2A
ðt Þ∨ . . . ∨x iX
A
ðt Þ
>
> I
:
˜ x i1 ðt Þ∨x I
i2 ðt Þ∨ . . . ∨x I
iY ð t Þ
ð2Þ
where the ∇, ∨, and ¬ symbols represent ternary HIGH/LOW

PASS, OR, and NOT operators; x ijA is the state of the ith variable’s jth
activator; and x ikI is the state of the ith variable’s kth inhibitor. The
ternary operators given in Eq. (2) are described in further detail in
Craddock et al. (2014) [58] and Fritsch et al. (2014) [59]. The first
entry in Eq. (2) is used when the variable possesses X activators and
Y inhibitors, the middle when the variable has only X activators,
and last when the activator has only Y inhibitors.
Using this logic formalism and the specific connectivity archi-
tecture of the network, we determine the number and type of stable
resting states supported by the regulatory circuitry as well as the
specific qualitative neuropsychological signatures at each of these
stable points without requiring detailed kinetic information. That
is, we determine where the system would eventually come to rest
even though we may not know how quickly this equilibrium will be
reached. The number of nodes determines the total number of
allowable states available to a model, such that a model of
N nodes possesses 3N states. Steady states are defined as those
state nodes from which there is no allowable escape or which
possesses an out degree of 0. Since the current regulatory network
model (Fig. 1) contains 15 soluble mediators and behavioral con-
structs, the number of possible system states is 315 ¼ 14,348,907.
Computing an Escape From Depression
Fig. 1 Basic logic model of behavior and neurotransmission. A basic causal regulatory network model representing the principal signaling mechanisms that
127
coordinate the expression of 15 soluble mediators in the brain and associated behavioral constructs. Green connectors with an arrowhead terminator represent
stimulatory actions whereas red connectors with a dot terminator represent inhibitory actions
Though quite large this remains a computationally tractable prob-

lem and an exhaustive search of all possible state nodes was con-
ducted to determine which of the latter constituted a steady
homeostatic state as in Fritsch et al. (2014) [59].
2.3 A Monte Carlo The evolution of state transitions supported by the model was
Discrete Event simulated by applying the decision logic described in Subheading
Simulation 2.2 iteratively using a Monte Carlo algorithm. From any initial
Environment starting state, allowable state transitions are determined based on
Eq. (2). Applying the latter to each variable in the model for the mth
state of the system, ~ x m ðt Þ, defines the image vector ~
x m ðt þ 1Þ for
the m state. With ~
th
x ðt þ 1Þ defined, the system is updated
m
asynchronously allowing only one variable to change at a time in

order to maintain the stochasticity inherent to biological systems.
This is consistent with generalized logical analysis of [57] Thomas
(1991). According to this method the ith variable of the mth state
vector x im ðt Þ is moved one step toward its preferred image
x im ðt þ 1Þ (e.g., If x im ðt Þ ¼ 1 and x im ðt þ 1Þ ¼ 1, then
x im ðt þ 1Þ is set to 0). Thus, for each current state of the system
there are potentially several subsequent states toward which it may
asynchronously evolve. From these allowable transitions a target
state is chosen at random using a uniform equal distribution then
used to generate the next set of allowable target states, and so on.
Executing the simulation multiple times gives a distribution of
paths that is used to determine the behavior of the system from
any given start state. For additional details the reader is referred to
our previous work [58, 59].
2.4 Defining The above simulation paradigm can serve as a basis for the in silico
an Idealized Virtual testing of strategies for rescuing the neurobehavioral system from a
Treatment stable but pathological regulatory program. Indeed, by extending
the paradigm in Subheading 2.3 to include externally applied state
changes it becomes possible to predict the outcome of therapeutic
strategies consisting of robust sequences of discrete treatments with
a select combination of drug actions (agonist or antagonist),
applied to specific targets at carefully timed intervals. In this case
simulated treatment is directed toward supporting the recovery of
normal homeostasis and delivering lasting remission in the absence
of any further therapy. Each treatment event applied to the system
at time t in the treatment course is represented by the vector T ~ ðt Þ
specifying the change in state applied to each of the N model state
variables, in this case behavioral constructs and signaling molecules
such that
~ ðt Þ ¼ ðT 1 ðt Þ; T 2 ðt Þ; . . . ; T N ðt ÞÞ
T ð3Þ
where Ti is a ternary value describing the action of the treat-

ment on the ith state variable of the system: 1 (suppressing),
0 (untreated), and 1 (elevating). At those time points where a
treatment is applied, the image vector ~ x ðt þ 1Þ describing the

preferred state toward which the system should evolve is now
redefined by extending Eq. (2) as follows:
~
x ðt þ 1Þ ¼ ~ ~ ðt Þ
x ðt Þ þ T ð4Þ
Consistent with the ternary nature of this logic no value can

extend beyond the range of 1 to 1 and values beyond this range
are constrained accordingly (i.e., if xi(t) ¼ 1 and Ti(t) ¼ 1, then
xi(t + 1) ¼ xi(t) + Ti(t) ¼ 2 is rescaled to 1). At times t when no
treatment is applied (i.e., all Ti ¼ 0), state transition continues
unaltered according to Eq. (2).
2.5 A Global Search Using the framework described in Subheading 2.4, treatment
Iterative Simulation courses can be simulated and the desirability of their outcomes
Engine assessed with respect to a specific treatment goal. Repeating this
trial-and-error process iteratively makes it possible to gradually
improve the candidate treatment but only if (1) we learn efficiently
from past attempts while (2) continuing to maintain a broad view of
potential intervention choices. These two features are often and
odds, with conventional methods typically favoring a very efficient
search which all too quickly becomes very narrow in scope leading
to the equivalent of numerical tunnel vision. In an attempt to avoid
settling prematurely on a mediocre design a family of so-called
global search algorithms prescribe and conduct broad sets of con-
current simulations, retaining not only the best design but also the
runner-up solutions. The best features of each of these competing
treatment solutions are combined to create a new generation of
candidate designs. With a basic form that naturally accommodates
discrete state models, we apply in this work a simple genetic algo-
rithm (GA) [60] to balance breadth with efficiency in this trial-and-
error process.
A treatment course vector C ~ with M treatment events is
therefore defined as

~¼ T
C ~ ðt 1 Þ; T
~ ðt 2 Þ; . . . ; T
~ ðt M Þ ð5Þ
where T ~ ðt i Þ is the treatment vector at the ith time point. In

keeping with the asynchronous updating of the model (Subheading
2.3) each treatment vector T ~ contains only one exogenous treat-
ment variable Ti acting on system state variable i at a given time step
t. This treatment variable is selected from the set of allowable
candidate targets by the optimization algorithm as part of the
next best treatment course to be assessed.
A typical iterative optimization problem is deployed and
resolved by repeating the following stepwise protocol.
Step 1. Initiation. The GA optimization starts by generating a
population of 1000 candidate treatment courses, each
composed of a specific number of randomly selected interven-

tions consisting of candidate treatment actions, e.g., serotonin
reuptake inhibition, GABA inhibition, or both. These are
applied at random points along the time course yielding a
proposed treatment course.
Step 2. Treatment simulation. The response of the system to each
candidate treatment course in this initial generation of candi-
dates is simulated for 1000 time steps. Over the course of these
time steps the state of the system evolves according to Eq. (2),
except at those times when interventions are applied. At these
intervention events the state transition follows Eq. (5).
Step 3. Treatment assessment. These 1000 iterations conducted in
Step 2 provide a distribution of paths that are then ranked
according to a fitness function. In this work solution fitness
or desirability is based on the number of times a candidate
treatment successfully reaches the stable healthy state or
healthy homeostatic mode (% HHM).
Step 4. Treatment retention and refinement. After all treatments in
the current generation have been ranked, the top 10% are
retained without change and validated further in the next
round of simulations. New candidate solutions for the next
generation of simulations are created by choosing random
pairs from the total set of previous candidates (including the
top 10th percentile) and combining different
N elements of these
(crossover recombination). Combination, , of two treatment
courses C~1 and C~2 is performed at a single random splice
point s to create a new treatment course C~1 0 for the next
generation, such that
N
C~1 C~2 ) C~1 0
N
T~1 ðt 1 Þ;. .. ; T~1 ðt s Þj T~1 ðt sþ1 Þ;. ..; T~1 ðt M Þ T~2 ðt 1 Þ;. ..; T~2 ðt s Þj T~2 ðt sþ1 Þ;.. .; T~2 ðt M Þ

) T~1 ðt 1 Þ; .. .; T~1 ðt s Þj T~2 ðt sþ1 Þ; ... ; T~2 ðt M Þ
ð6Þ
The system response to these new treatment courses, including

members of the previous top 10th percentile, is then simulated
once again and the results are ranked to form a basis for design of
still another generation of candidate treatments.
This process is continued iteratively, typically for 1000 genera-
tions in this work. The final treatment course with the highest %
HHM is stored as the best treatment solution for this set of
attempts given the candidate set of targets selected initially as
therapeutically modifiable. In this work, candidate target sets were
evaluated repeatedly over 100 optimization runs. Optimal inter-
ventions identified in this way were then assessed again with the
addition of environmental stress simulated as increases in cortisol
level. This external stressor was added to the treatment vector

randomly across the time course, such that the stressful events
occurred at a prescribed frequency, e.g., 30% or one out of every
three time points.
2.6 Measures To explore factors that may increase vulnerability to falling into a
of Vulnerability persistent depressed state, we define and conduct sensitivity
and Impact analysis-based simulations of the regulatory logic supported by
the biobehavioral circuit model. Using a normal healthy regulatory
balance as a start point, an exogenous agonist or antagonist was
applied to each signaling molecule in turn such that corresponding
levels were maintained high or low while the response of the
regulatory network to a stressful triggering event was simulated.
This was repeated 1000 times for both abnormally high and low
levels of each signaling molecule and the results used to determine
the frequency with which response exceeded the tipping point and
degenerated into a persistent depressed state. This simulated fre-
quency of onset in the wake of an idealized triggering event was
used as a measure for describing the robustness of the biobehavioral
network to spurious excursions in the levels of individual signaling
neurotransmitters and hormones.
3 Methods
The relevance and value of these tools and the framework described
in Subheading 2 in creating new insight into complex illnesses are
best described using an example problem, in this case a model-
based analysis of regulatory dynamics in depression. First an analysis
of regulatory stability is applied (Subheading 3.1) to identify natu-
rally occurring regulatory programs that might promote persis-
tence of depression and resistance to conventional treatment.
Fluctuations in neurotransmitter and hormone levels that increase
vulnerability and facilitate access to pathologic programs, in this
case depression, are then identified (Subheading 3.2) using com-
puter simulations and the most direct course of onset associated
with such fluctuations is predicted based on the signaling logic
(Subheading 3.3). Extending this analysis of depression as a persis-
tent regulatory trap, the potential for delivering lasting remission
using conventional SSRI therapy (Subheading 3.4.1) is simulated
and results are compared to clinical observations of treatment
efficacy and recidivism rates. The same numerical protocol is
applied to assess the predicted efficacy of GABA inhibition (Sub-
heading 3.4.2) as well as behavioral intervention (Subheading
3.4.3) used in combination with conventional SSRI treatment.
The most synergistic way in which these individual intervention
components might be combined and scheduled was analyzed for-
mally using the model-based framework described in Subheading 2.
Rather than applying trial and error, simulations were iteratively

prescribed by a global search algorithm (Subheading 2.5), with the
objective of identifying an externally applied sequence and spacing of
interventions involving serotonin reuptake inhibition, GABA sup-
pression, or a combination of both that deliver the highest predicted
rate of remission.
3.1 Identifying As described above, steady homeostatic states are defined as resting
Stable Regulatory states to which the system naturally returns after perturbation.
Programs Robust and adaptable, even relatively simple biological systems
are capable of supporting more than one such resting state. These
were determined computationally in this work by enumerating all
possible states that may be occupied by the biobehavioral circuit
model shown in Fig. 1 and the corresponding control logic. An
exhaustive search using the simulation paradigm described in Sub-
heading 2.2 identified two homeostatic states (SS) where neuro-
transmitter and psychobehavioral constructs converged to
equilibrium levels described in Fig. 2. The reference steady state
(SS0) described typical healthy equilibrium with all markers
expressed at their nominal levels encoded as zero. In addition to
this baseline equilibrium state (SS0) we found that this regulatory
circuitry can also accommodate a persistent depressive state also
characterized by increased anxiety accompanied by reduced physical
fatigue and impaired attention. This same regulatory program main-
tained chronically low levels of serotonin while promoting elevated
cortisol glutamate, GABA, and epinephrine levels. This result would
suggest that a persistent depressive mood might be perpetuated
at least in part by homeostatic regulatory action (see Note 1).
3.2 Sensitivity With the possibility that persistent depression may at least partially
Analysis coincide with a naturally occurring regulatory trap it is relevant to
and Predisposing ask what factors may promote the initiation of this pathologic
Factors of Onset biobehavioral program. Repeated simulations were conducted
using the protocol and measures described in Subheading 2.6 to
assess the sensitivity of normal regulatory programming to anoma-
lous excursions in the levels of individual signaling molecules. In
the current example, the results of this sensitivity analysis (Fig. 3)
indicated that under normal equilibrium conditions, individual
deviations in almost all of the neurotransmitters and hormones
provided no increased risk for falling into the depressed steady
state beyond random chance. There were, however, three signifi-
cant exceptions. Increased levels of NPY were protective, prevent-
ing the biobehavioral network from escaping normal homeostasis
and falling into a depressive regulatory mode in 776 out of 1000
simulations. Interestingly abnormally low levels of NPY did not
significantly affect vulnerability. Changes in glutamate, however,
produced significant effects when either abnormally high or low.
Spurious increases in glutamate levels facilitated migration from
Fig. 2 Multiple naturally occurring stable states. This basic model of

neurotransmission supports an alternate homeostatic state where physiologic
regulation now ensures a sustained depressive state with increased anxiety
accompanied by reduced physical fatigue, impaired attention, and chronically
low levels of serotonin in the context of elevated cortisol, GABA, and epinephrine
normal homeostasis into persistent depression in 894 out of 1000

simulations. Conversely, low levels of glutamate prevented a lapse
from normal homeostasis into depression in 775 out of 1000
simulations (see Note 2).
3.3 Trajectory The sensitivity analysis in Subheading 3.2 suggested that abnor-
Analysis and Mapping mally elevated glutamate provides a high-risk environment in which
the Course of Onset a depressive illness state is most easily reached when the system is
put under stress. However, this analysis does not inform on how
gradual or precipitous the onset might be, nor does it highlight
telltale signs of how far this migration into depression has pro-
gressed. Knowledge of the expected course of illness has important
clinical implications. In order to explore this aspect a trajectory
analysis was conducted to determine the sequence of states that
To Health
To Illness
894
775 776
668
615
550 576 549
516 526 522 507 507
484 505
495 474 478 493 493
450 451
424
385
332
225 224
106
Fig. 3 Sensitivity analysis of key factors affecting vulnerability to depression. Each node was artificially held
high, and then low, one at a time. During each orientation, the system was allowed to naturally evolve under
stress 1000 times. Of those 1000 iterations for each orientation, the number of times the system settled in the
illness state versus the health state was recorded to determine which individual nodes represented the largest
risk or protective factor
lie along the most precipitous course of onset given a stressful

trigger event introduced in the context of elevated glutamate.
Numerically this equates to the shortest state transition path sup-
ported by the signal transmission circuitry and regulatory logic. In
this example, this shortest path analysis indicates that persistent
depression may be reached under abnormally high glutamate levels
in as few as five intermediate steps (Fig. 4). At baseline, glutamate
and cortisol are set to elevated levels to simulate a stressful event
during an excursion in glutamate concentration. In a first incre-
mental response, attention levels fall. This is followed sequentially
by an increase in depressive mood, and then an increase in GABA
which in turn triggers a decrease in serotonin. Finally, decreased
serotonin levels promote increased anxiety and reduced physical
fatigue scores. At this point, regulatory control has been fully
reset and the system has now settled into a stable depressed resting
state. Interestingly, examination of this sequence of events would
suggest that depressed serotonin levels occur late in the course of
onset and present downstream of a prior increase in GABA. This
may explain at least in part why modulating serotonin, though
significant in reducing symptom burden, may only be partially
effective as a disease-modifying treatment (see Note 3).
1 1 1 1 1 1 1 1 1 1
0 0 0 0 0 0 0 0 0 0
1 2 3 4 5 6 7 8 9 10
-1 -1 -1 -1 -1 -1 -1 -1 -1
'Depression' 'Acetylcholine' 'Norepinephrine'
'Glutamate' 'Neuro_Peptide_Y' 'Serotonin'
'Dopamine' 'GABA' 'Physical_Fatigue'
'Attention' 'Working_Memory' 'Hyperarousal'
'Cortisol' 'PASAT'
Fig. 4 Sequential migration path of acute illness onset. Simulated sequence of state changes (1 low,
0 nominal, +1 elevated) along the shortest average transition path from normal homeostatic regulation to
persistent depression. This transition path is triggered by a simulated stressful event under initial conditions of
a spurious elevation in glutamate. Only five state transition steps separate normal from pathological regulation
under such conditions and the sequence of events adheres strictly to and is derived directly from the
regulatory signaling circuitry
3.4 Prediction In addition to exploring vulnerability to depression and possible

of Treatment course of onset, the simulation protocol described in Subheading
Outcomes 2.4 may be applied repeatedly toward the high-throughput screen-
ing of combinatorial treatment courses using the efficient global
search framework outlined in Subheading 2.5. Before applying this
approach to the design of novel treatment courses, we first test the
model further by attempting to recover clinically reported rates of
relapse into depression following conventional SSRI treatment
(Subheading 3.4.1). Means of further improving treatment out-
come are then explored by iteratively refining the delivery of GABA
antagonists (Subheading 3.4.2) and behavioral therapy (Subhead-
ing 3.4.3) such that these optimally synergize with, and potentiate,
conventional SSRI treatment.
3.4.1 Reproducing A common pharmaceutical treatment for depression is directed at

Conventional Treatment modulating the neurotransmitter serotonin. Specifically, selective
Outcomes with SSRI serotonin reuptake inhibitors (SSRIs) increase the amount of
serotonin in the brain by blocking its reuptake into presynaptic

neurons. This increase in serotonin affects mood, and can decrease
depression, consistent with the wiring of our circuit model. While
reported efficacy rates of SSRIs vary greatly, current statistics put
the efficacy rate of SSRIs in clinic at around 60–70%. As a partial
validation of our simple neurotransmission model we attempted to
simulate SSRI treatment computationally. First a numerical optimi-
zation was applied to identify the interval for administering exoge-
nous serotonin that delivered the highest predicted rate of lasting
remission. Because the state of the model biobehavioral signaling
network evolves in a probabilistic manner, we simulated in a second
step the artificial introduction of exogenous serotonin according to
this optimal schedule in 100 repeated treatment courses, each in
100 simulated subjects. This was performed based on a schedule of
predicted optimal treatment intervals separating between two and
eight treatment cycles (see Note 4).
As real-world subjects must deal with environmental stressors,
we included these in our model and superimposed random stressful
events in the simulations at a frequency of occurrence of 30%
according to the protocol described in Subheading 2.5. Under
these simulated real-world conditions the proportion of predicted
treatment responses potentially leading to lasting remission
increased with repeated treatment, eventually reaching a saturation
level of roughly 75% after eight simulated cycles of SSRI treatment
(Fig. 5).
3.4.2 Designing Optimal Treatment with GABA inhibitors is also often used as a second line
Combination Therapies of therapy. Once again, we applied the numerical optimization
with GABA Inhibition scheme of Subheading 2.5, this time allowing for the optimal
scheduling of both SSRI administration and GABA inhibition.
The optimal treatment schedules for this combination therapy
were identified for two up to eight cycles of treatment. As before
each optimal treatment course was simulated 100 times in each of
100 virtual subjects. Overall, we found that these optimal treat-
ment courses included GABA inhibition only very sparingly and
that the burden of treatment was carried mainly by SSRI adminis-
tration. Nonetheless, the inclusion of GABA blockade improved
the predicted remission rates slightly under simulated real-world
conditions of 30% stressful events. However, this improvement did
not reach statistical significance.
3.4.3 Behavioral Since pharmacological intervention in depression is typically

Intervention and Stress accompanied by behavioral therapy, we repeated the experiments
Management described above, assuming this time that the virtual subjects
received behavioral therapy that provided them with effective stra-
tegies for managing stressful events. Accordingly, optimization of
pharmacological treatment schedules and their subsequent
100
90
% Mean remission (SD)
80
70
60
Mean SSRI wo stress

50 Mean SSRI cw stress (30% frequency)
Mean GABA cw stress (30% frequency)
40
30
2 3 4 5 6 7 8
Rounds of treatment"
Fig. 5 Simulated intervention time steps. Summary of simulated intervention time courses with the maximum
predicted frequency of remission as a function of repeated intervention events simulated in (A) SSRI with
behavioral stress management (blue line), and treatment in the presence of randomly occurring stressful
events (30% frequency) without behavioral intervention using (B) SSRI only (red line) and (C) SSRI with GABA
blockade (green line)
simulation was conducted in the absence of stress altogether, or in

the context of optimally effective stress management therapy. We
found that this simulated behavioral intervention increased the
efficacy of SSRI treatment significantly. The predicted rate of remis-
sion rose almost uniformly by up to 10% of treatment attempts
across the range of treatment cycles (Fig. 5; Table 1). This is far
more significant than the simulated use of combination drug ther-
apy in the absence of counseling and lifestyle changes (see Notes 5
and 6).
4 Notes
1. Regardless of the form of treatment, depression has an unusu-

ally high recurrence rate. Indeed, nearly half of all individuals
who suffer from one episode of major depression will suffer
from at least one more later in life, and nearly 80% of those who
suffer from two episodes will suffer from at least a third [7]. In
this work, we propose that the persistent recurrence and resis-
tance to treatment observed in depression and many other
Table 1
Remission rates of simulated SSRI treatment under various conditions
Number of treatment pulses 1 2 3 4 5 6 7 8

SSRI with stress management or no Mean 46.02 61.50 65.37 76.51 77.90 84.35 86.57
superimposed stress SD 5.04 4.31 4.64 4.10 3.91 3.41 3.46
SSRI treatment with 30% stressful events Mean 40.05 48.13 58.04 70.81 72.85 78.63 73.08
SD 4.82 4.41 5.12 4.27 4.34 4.34 4.79
Combined SSRI/GABA with 30% stressful Mean 44.52 51.07 63.47 71.47 76.53 79.45 83.97
events SD 5.11 4.78 4.92 4.00 4.73 3.83 3.34
Mean and standard deviation (SD) of remission rates obtained in simulations of multiple interventions using SSRI alone
with an idealized behavioral stress management (BSM) therapy (0% stressful events), SSRI treatment without BSM in the
presence of external stressors (30% frequency of stressful events) both alone and supplemented with a GABA inhibitor
multifactorial illnesses may be promoted at least in part by the

complex control dynamics regulating neurotransmitter and
hormone levels. Much like a computer may execute multiple
programs, even simple biological circuits are capable of sup-
porting more than one stable regulatory mode. In this work,
we assemble from literature a basic circuit model of biobehav-
ioral regulation and demonstrate that this biological computer
can in theory support a stable regulatory program where sero-
tonin levels are persistently downregulated while GABA, gluta-
mate, epinephrine, and cortisol levels are upregulated resulting
in stable depressive mood accompanied by increased anxiety.
This is consistent with recent work by Müller and Schwarz
(2007) [61] who propose that an integrated feedback system
linking dysregulation of the hypothalamic-pituitary-adrenal
(HPA) axis, with an underproduction of serotonin, and an
overproduction of glutamate might provide a more compre-
hensive framework for the study of depression.
2. A systematic perturbation of neurotransmitters and hormones
included in this signal processing model suggested that high
levels of glutamate could serve as a risk factor for developing
depression. This result is supported in the literature by reports
of glutamate antagonist showing promise in the treatment of
depression [62]. This same analysis also suggested that
increased levels of NPY may serve a protective role against the
onset of depression. This is also supported by empirical litera-
ture, as Morgan et al. (2002) [63] found that greater levels of
NPY release during stress are associated with reduced psycho-
logical distress, suggesting that NPY supports anxiolytic activ-
ity. Similarly, Thorsell (2010) [64] examined the therapeutic
implications of NPY as a stress mediator further supporting this
finding and pointing to the potential use of NPY in preventa-
tive treatment in individuals at higher risk of depression.
3. Understanding the most likely course for the system to take

toward depression is also of vital importance [65]. In this work,
we show that simulations of neurotransmitter and hormone
regulation can be used to predict the sequence of events lead-
ing to a recurrence of depression. For example, these simula-
tions indicated that a triggering stressful event occurring
potentiated by elevated glutamate levels may reduce separation
from a depressive state to as few as six intermediate states.
Analysis of these paths of illness progression has important
implications for treatment, providing insight into the most
appropriate targets and effective treatment strategies at each
phase. It is important to note, however, that the logic model
presented here captures the order of events but does not yet
inform on rate of progression.
4. Though SSRIs are commonly used to treat depression their
efficacy is estimated to be in the order of 60%, compared to 40%
response with placebos [66]. By modulating serotonin to
mimic the introduction of an SSRI treatment, the basic
model presented here was able to predict similar results. This
offered a partial validation of this minimal signal transmission
model while also suggesting that additional granularity could
further improve accuracy. For example, impaired immune func-
tion is a common and significant component of depression [67]
with recent work by Yirmiya et al. (2015) [68] pointing to
involvement of microglia and discovery of lymphatic vascula-
ture in the central nervous system providing a bridge to periph-
eral immune function [69].
5. Stress is a common trigger of depressive episodes and a major
cause of recurrence [70]. In this work, stressful events were
simulated as transient increases in cortisol levels. Introducing
these simulated stressful events during and after the SSRI
treatments interfered with normal homeostatic regulation lead-
ing to a predicted efficacy of 75%, approximating clinical obser-
vations. Even without pharmacological intervention,
psychotherapies have been shown to effectively reduce depres-
sive symptoms. Cognitive behavioral therapy (CBT) was found
effective in reducing depression scores (Beck Depression
Inventory) by changing automatic thoughts and dysfunctional
attitudes, adding alternate cognitive trajectories for reducing
experienced stress [71, 72]. Nonetheless in keeping with the
simulation results presented here, literature reports that psy-
chotherapy used in conjunction with SSRIs typically produces
better outcomes than either treatment used alone [73]. Indeed,
March et al. (2004) [74] found that a combination of CBT
with the SSRI, fluoxetine, significantly reduced depression
scores (Chidren’s Depression Rating Scale-Revised) in 71.0%
of subjects, compared to 60.6% of subjects on fluoxetine alone,

43.2% of subjects on CBT alone, and 34.8% on placebo.
6. The use of computational models is slowly gaining acceptance
as a tool for exploring the complexity of illnesses such as
depression and combinatorial therapies capable of addressing
this complexity. For example, Soeteman et al. (2012) [75] used
a variant of an empirical decision model to predict the impact
on suicide rate in major depressive disorder of SSRI treatment,
cognitive behavioral therapy (CBT), and a combination of the
two. They found that combination treatment worked signifi-
cantly better in the short term but that this advantage
dissipated if suicidal behavior persisted. Using a more conven-
tional engineering model Gruwez et al. (2007) [76] studied
the pharmacodynamics of antidepressants clomipramine and
lithium to predict how these should be administered in a
clinical trial. Though engineering models such as these offer a
high degree of temporal fidelity, they also require detailed
quantitative knowledge of the rate of change and for this
reason are typically applied to systems where response dynamics
are well characterized and supported by a significant body of
experimental data. With the exception of some isolated systems
much more is known about anatomical and biochemical con-
nectivity than about the rate at which signals travel through
these networks. By requiring only connectivity and enforcing
the correct sequence of events, the discrete logic paradigm
proposed here offers the possibility of modeling broader
more complex regulatory systems and delivering qualitatively
insightful predictions of their dynamic behavior with little or
no data. This minimal data requirement makes this an attractive
strategy for discovery and hypothesis testing. For example, the
basic model presented here predicts elevated levels of both
cortisol and epinephrine in depression, two neurochemicals
that haven’t typically implicated in the study of this illness and
have only very recently become of interest [77]. Most impor-
tantly perhaps, our use of a control circuit model emphasizes
the possible role of regulatory entrapment in the persistence of
depression, offering a new perspective on resistance to treat-
ment and the design of effective strategies for overcoming this.
Acknowledgments
Funding was provided by US Department of Defense Congressio-

nally Directed Medical Research Program (CDMRP) awards
(http://cdmrp.army.mil/) GW093042, GW140142 (Broderick—
PI) and GW120045 (Morris—PI). This research was conducted in
collaboration with the high-performance computing team at the
University of Miami Center for Computational Science (CCS)

(http://ccs.miami.edu).
Disclaimer: The opinions and assertions contained herein are the
private views of the authors and are not to be construed as official or
as reflecting the views of the Department of Defense.
References
1. Substance Abuse and Mental Health Services glutamatergic transmission in prefrontal cortex
Administration (2013) Results from the 2012 and facilitates working memory. Proc Natl
National Survey on Drug Use and Health: Acad Sci 106(33):14075–14079
Mental Health Findings, NSDUH Series 11. Tafet GE, Toister-Achituv M, Shinitzky M
H-47, HHS Publication No. (SMA) (2001) Enhancement of serotonin uptake by
13–4805. Substance Abuse and Mental Health cortisol: a possible link between stress and
Services Administration, Rockville, MD depression. Cogn Affect Behav Neurosci 1
2. American Psychiatric Association (2013) Diag- (1):96–104
nostic and statistical manual of mental disor- 12. Janssen SA, Arntz A, Bouts S (1998) Anxiety
ders: DSM-V. 5. American Psychiatric and pain: epinephrine-induced hyperalgesia
Association, Washington, DC and attentional influences. Pain 76
3. Greenberg PE, Fournier A, Sisitsky T, Pike CT, (3):309–316
Kessler RC (2014) The economic burden of 13. Bergstrom DA, Walters JR (1984) Dopamine
adults with major depressive disorder in the attenuates the effects of GABA on single unit
United States (2005 and 2010). J Clin Psychi- activity in the globus pallidus. Brain Res 310
atry 76(2):155–162 (1):23–33
4. Antonuccio DO, Danton WG, DeNelsky GY 14. Daw ND, Kakade S, Dayan P (2002) Oppo-
(1995) Psychotherapy versus medication for nent interactions between serotonin and dopa-
depression: challenging the conventional wis- mine. Neural Netw 15(4):603–616
dom with data. J Psychother Pract Res 26 15. Gillard ER, Dang DQ, Stanley BG (1993) Evi-
(6):574–585 dence that neuropeptide Y and dopamine in the
5. Sałat K, Podkowa A, Kowalczyk P, Kulig K, perifornical hypothalamus interact antagonisti-
Dziubina A, Filipek B, Librowski T (2015) cally in the control of food intake. Brain Res
Anticonvulsant active inhibitor of GABA trans- 628(1):128–136
porter subtype 1, tiagabine, with activity in 16. El Mansari M, Guiard BP, Chernoloz O,
mouse models of anxiety, pain and depression. Ghanbari R, Katz N, Blier P (2010) Relevance
Pharmacol Rep 67(3):465–472 of norepinephrine–dopamine interactions in
6. Cuijpers P, van Straten A, Andersson G, van the treatment of major depressive disorder.
Oppen P (2008) Psychotherapy for depression CNS Neurosci Ther 16(3):e1–e17
in adults: a meta-analysis of comparative out- 17. Calabresi P, Picconi B, Parnetti L, Di Filippo M
come studies. J Consult Clin Psychol 76 (2006) A convergent model for cognitive dys-
(6):909 functions in Parkinson’s disease: the critical
7. Burcusa SL, Iacono WG (2007) Risk for recur- dopamine–acetylcholine synaptic balance. Lan-
rence in depression. Clin Psychol Rev 27 cet Neurol 5(11):974–983
(8):959–985 18. Biello SM, Golombek DA, Harrington ME
8. Wurtman RJ, Pohorecky LA, Baliga BS (1972) (1997) Neuropeptide Y and glutamate block
Adrenocortical control of the biosynthesis of each other’s phase shifts in the suprachiasmatic
epinephrine and proteins in the adrenal nucleus in vitro. Neuroscience 77
medulla. Pharmacol Rev 24(2):411–426 (4):1049–1057
9. Piazza PV, Rougé-Pont F, Deroche V, 19. Serfozo P, Bartfai T, Vizi ES (1986) Presynap-
Maccari S, Simon H, Le Moal M (1996) Glu- tic effects of neuropeptide Y on [3 H] nor-
cocorticoids have state-dependent stimulant adrenaline and [3 H] acetylcholine release.
effects on the mesencephalic dopaminergic Regul Pept 16(2):117–123
transmission. Proc Natl Acad Sci 93 20. Dryden S, McCarthy HD, Malabu UH,
(16):8716–8720 Ware M, Williams G (1993) Increased neuro-
10. Yuen EY, Liu W, Karatsoreos IN, Feng J, McE- peptide Y concentrations in specific hypotha-
wen BS, Yan Z (2009) Acute stress enhances lamic nuclei of the rat following treatment with
methysergide: evidence that NPY may mediate hypertensive rats. Metab Brain Dis 15
serotonin’s effects on food intake. Peptides 14 (4):297–304
(4):791–796 33. Mathew SJ, Coplan JD, Smith EL, Schoepp
21. Herman JP, Renda A, Bodie B (2003) Norepi- DD, Rosenblum LA, Gorman JM (2001) Glu-
nephrine–gamma-aminobutyric acid (GABA) tamate—hypothalamic-pituitary-adrenal Axis
interaction in limbic stress circuits: effects of interactions: implications for mood and anxiety
reboxetine on GABAergic neurons. Biol Psy- disorders. CNS Spectr 6(07):555–564
chiatry 53(2):166–174 34. Carlsson A, Waters N, Holm-Waters S,
22. Wahlestedt CLAES, Hakanson ROLF, Vaz CA, Tedroff J, Nilsson M, Carlsson ML (2001)
Zukowska-Grojec ZOFIA (1990) Norepi- Interactions between monoamines, glutamate,
nephrine and neuropeptide Y: vasoconstrictor and GABA in schizophrenia: new evidence.
cooperation in vivo and in vitro. Am J Phys Annu Rev Pharmacol Toxicol 41(1):237–260
Regul Integr Comp Phys 258(3):R736–R742 35. Aultman JM, Moghaddam B (2001) Distinct
23. Bremner JD, Krystal JH, Southwick SM, Char- contributions of glutamate and dopamine
ney DS (1996) Noradrenergic mechanisms in receptors to temporal aspects of rodent work-
stress and anxiety: II. Clinical studies. Synapse ing memory using a clinically relevant task.
23(1):39–51 Psychopharmacology 153(3):353–364
24. Southwick SM, Bremner JD, Rasmusson A, 36. Paul IA, Skolnick P (2003) Glutamate and
Morgan CA, Arnsten A, Charney DS (1999) depression. Ann N Y Acad Sci 1003
Role of norepinephrine in the pathophysiology (1):250–272
and treatment of posttraumatic stress disorder. 37. Carrey NJ, MacMaster FP, Gaudet L, Schmidt
Biol Psychiatry 46(9):1192–1204 MH (2007) Striatal creatine and glutamate/
25. Bymaster FP, Katner JS, Nelson DL, Hemrick- glutamine in attention-deficit/hyperactivity
Luecke SK, Threlkeld PG, Heiligenstein JH, disorder. J Child Adolesc Psychopharmacol 17
Morin SM, Gehlert DR, Perry KW (2002) (1):11–17
Atomoxetine increases extracellular levels of 38. Van Praag HM, Kahn RS, Asnis GM, Wetzler S,
norepinephrine and dopamine in prefrontal Brown SL, Bleich A, Korn ML (1987) Deno-
cortex of rat: a potential mechanism for efficacy sologization of biological psychiatry or the
in attention deficit/hyperactivity disorder. specificity of 5-HT disturbances in psychiatric
Neuropsychopharmacology 27(5):699–711 disorders. J Affect Disord 13(1):1–8
26. Zhang K, Grady CJ, Tsapakis EM, Andersen 39. Aghajanian GK, Marek GJ (1999) Serotonin–-
SL, Tarazi FI, Baldessarini RJ (2004) Regula- glutamate interactions: a new target for anti-
tion of working memory by dopamine D4 psychotic drugs. Neuropsychopharmacology
receptor in rats. Neuropsychopharmacology 21:S122–S133
29(9):1648 40. MacDermot J, Higashida H, Wilson SP,
27. Delgado PL, Moreno FA (1999) Role of nor- Matsuzawa H, Minna J, Nirenberg M (1979)
epinephrine in depression. J Clin Psychiatry Adenylate cyclase and acetylcholine release
61:5–12 regulated by separate serotonin receptors of
28. Casey DE, Gerlach J, Christensson E (1980) somatic cell hybrids. Proc Natl Acad Sci 76
Behavioral aspects of GABA-dopamine interre- (3):1135–1139
lationships in the monkey. Brain Res Bull 41. Guiard BP, El Mansari M, Merali Z, Blier P
5:269–273 (2008) Functional interactions between dopa-
29. Moises HC, Woodward DJ (1980) Potentia- mine, serotonin and norepinephrine neurons:
tion of GABA inhibitory action in cerebellum an in-vivo electrophysiological study in rats
by locus coeruleus stimulation. Brain Res 182 with monoaminergic lesions. Int J Neuropsy-
(2):327–344 chopharmacol 11(5):625–639
30. Bankson MG, Yamamoto BK (2004) Seroto- 42. Korsgaard S, Gerlach J, Christensson E (1985)
nin–GABA interactions modulate MDMA- Behavioral aspects of serotonin-dopamine
induced mesolimbic dopamine release. J Neu- interaction in the monkey. Eur J Pharmacol
rochem 91(4):852–859 118(3):245–252
31. Edden RA, Crocetti D, Zhu H, Gilbert DL, 43. O’hara R, Schröder CM, Mahadevan R,
Mostofsky SH (2012) Reduced GABA concen- Schatzberg AF, Lindley S, Fox S, Weiner M,
tration in attention-deficit/hyperactivity disor- Kraemer HC, Noda A, Lin X, Gray HL, Hall-
der. Arch Gen Psychiatry 69(7):750–753 mayer JF (2007) Serotonin transporter poly-
32. Russell VA, Wiggins TM (2000) Increased morphism, memory and hippocampal volume
glutamate-stimulated norepinephrine release in the elderly: association and interaction with
from prefrontal cortex slices of spontaneously cortisol. Mol Psychiatry 12(6):544–555
44. Connor KM, Davidson JR (1998) The role of 55. Daniel JM, Dohanich GP (2001) Acetylcholine
serotonin in posttraumatic stress disorder: neu- mediates the estrogen-induced increase in
robiology and pharmacotherapy. CNS Spectr 3 NMDA receptor binding in CA1 of the hippo-
(S2):42–51 campus and the associated improvement in
45. Charney DS, Woods SW, Goodman WK, working memory. J Neurosci 21
Heninger GR (1987) Serotonin function in (17):6949–6956
anxiety. Psychopharmacology 92(1):14–24 56. Mendoza L, Xenarios I (2006) A method for
46. Owens MJ, Nemeroff CB (1994) Role of sero- the generation of standardized qualitative
tonin in the pathophysiology of depression: dynamical systems of regulatory networks.
focus on the serotonin transporter. Clin Theor Biol Med Model 3:13
Chem 40(2):288–295 57. Thomas R (1991) Regulatory networks seen as
47. Roelands B, Meeusen R (2010) Alterations in asynchronous automata: a logical description. J
central fatigue by pharmacological manipula- Theor Biol 153(1):1–23
tions of neurotransmitters in normal and high 58. Craddock TJ, Fritsch P, Rice MA Jr, del
ambient temperature. Sports Med 40 Rosario RM, Miller DB, Fletcher MA, Klimas
(3):229–246 NG, Broderick G (2014) A role for homeo-
48. Keely SL, Lincoln TM, Corbin JD (1978) static drive in the perpetuation of complex
Interaction of acetylcholine and epinephrine chronic illness: gulf war illness and chronic
on heart cyclic AMP-dependent protein kinase. fatigue syndrome. PLoS One 9(1):e84839
Am J Phys Heart Circ Phys 234(4): 59. Fritsch P, Craddock TJ, del Rosario RM, Rice
H432–H438 MA, Smylie A, Folcik VA, de Vries G, Fletcher
49. Dixon JS, Jen PY, Gosling JA (2000) The dis- MA, Klimas NG, Broderick G (2013) Suc-
tribution of vesicular acetylcholine transporter cumbing to the laws of attraction: exploring
in the human male genitourinary organs and its the sometimes pathogenic versatility of discrete
co-localization with neuropeptide Y and nitric immune logic. Sys Biomed 1(3):179–194
oxide synthase. Neurourol Urodyn 19 60. Whitley D, Garrett D, Watson JP (2003,
(2):185–194 January) Quad search and hybrid genetic algo-
50. Walker SW, Strachan MWJ, Lightly ERT, Wil- rithms. In: Genetic and evolutionary computa-
liams BC, Bird IM (1990) Acetylcholine stimu- tion—GECCO 2003. Springer, Berlin, pp
lates cortisol secretion through the M3 1469–1480
muscarinic receptor linked to a 61. Müller N, Schwarz MJ (2007) The immune-
polyphosphoinositide-specific phospholipase mediated alteration of serotonin and gluta-
C in bovine adrenal fasciculata/reticularis mate: towards an integrated view of depression.
cells. Mol Cell Endocrinol 72(3):227–238 Mol Psychiatry 12:988–1000
51. Kalsner S, Quillan M (1988) Presynaptic inter- 62. Niciu MJ, Henter ID, Luckenbaugh DA,
actions between acetylcholine and adrenergic Zarate CA Jr, Charney DS (2014) Glutamate
antagonists on norepinephrine release. J Phar- receptor antagonists as fast-acting therapeutic
macol Exp Ther 244(3):879–891 alternatives for the treatment of depression:
52. Leboulenger F, Benyamina M, Delarue C, ketamine and other compounds. Annu Rev
Netchitailo P, Saint-Pierre S, Vaudry H Pharmacol Toxicol 54:119–139
(1988) Neuronal and paracrine regulation of 63. Morgan CA, Rasmusson AM, Wang S, Hoyt G,
adrenal steroidogenesis: interactions between Hauger RL, Hazlett G (2002) Neuropeptide-
acetylcholine, serotonin and vasoactive intesti- Y, cortisol, and subjective distress in humans
nal peptide (VIP) on corticosteroid production exposed to acute stress: replication and exten-
by frog interrenal tissue. Brain Res 453 sion of previous report. Biol Psychiatry 52
(1):103–109 (2):136–142
53. Czermak C, Staley JK, Kasserman S, Bois F, 64. Thorsell A (2010) Brain neuropeptide Y and
Young T, Henry S, Tamagnan GD, Seibyl JP, corticotropin-releasing hormone in mediating
Krystal JH, Neumeister A (2008) β2 Nicotinic stress and anxiety. Exp Biol Med 235
acetylcholine receptor availability in post- (10):1163–1167
traumatic stress disorder. Int J Neuropsycho- 65. Schubert KO, Clark SR, Van LK et al (2017)
pharmacol 11(3):419–424 Depressive symptom trajectories in late adoles-
54. Botly LC, De Rosa E (2008) A cross-species cence and early adulthood: A systematic review.
investigation of acetylcholine, attention, and Aust N Z J Psychiatry 51(5):477–499
feature binding. Psychol Sci 19 66. Anderson IM (1998) SSRIs versus tricyclic
(11):1185–1193 antidepressants in depressed inpatients: a
meta-analysis of efficacy and tolerability. 73. Chisholm D, Sanderson K, Ayuso-Mateos JL,

Depress Anxiety 7(S1):11–17 Saxena S (2004) Reducing the global burden
67. Reiche EMV, Nunes SOV, Morimoto HK of depression. Br J Psychiatry 184(5):393–403
(2004) Stress, depression, the immune system, 74. March J, Silva S, Petrycki S, Curry J, Wells K,
and cancer. Lancet Oncol 5(10):617–625 Fairbank J, Burns B, Domino M, McNulty S,
68. Yirmiya R, Rimmerman N, Reshef R (2015) Vitiello B, Severe J (2004) Fluoxetine,
Depression as a microglial disease. Trends Neu- cognitive-behavioral therapy, and their combi-
rosci 38(10):637–658 nation for adolescents with depression: treat-
69. Louveau A, Smirnov I, Keyes TJ, Eccles JD, ment for adolescents with depression study
Rouhani SJ, Peske JD, Derecki NC, Castle D, (TADS) randomized controlled trial. JAMA
Mandell JW, Lee KS, Harris TH, Kipnis J 292(7):807–820
(2015) Structural and functional features of 75. Soeteman DI, Miller M, Kim JJ (2012) Mod-
central nervous system lymphatic vessels. eling the risks and benefits of depression treat-
Nature 523(7560):337–341 ment for children and young adults. Value
70. Hammen C (2005) Stress and depression. Health 15(5):724–729
Annu Rev Clin Psychol 1:293–319 76. Gruwez B, Poirier MF, Dauphin A, Olié JP,
71. Furlong M, Oei TP (2002) Changes to auto- Tod M (2007) A kinetic-pharmacodynamic
matic thoughts and dysfunctional attitudes in model for clinical trial simulation of antidepres-
group CBT for depression. Behav Cogn Psy- sant action: application to clomipramine–-
chother 30(03):351–360 lithium interaction. Contemp Clin Trials 28
(3):276–287
72. Farrer L, Christensen H, Griffiths KM, Mack-
innon A (2011) Internet-based CBT for 77. Peacock BN, Scheiderer DJ, Kellermann GH
depression with and without telephone track- (2017) Biomolecular aspects of depression: a
ing in a national helpline: randomised con- retrospective analysis. Compr Psychiatry
trolled trial. PLoS One 6(11):e28099 73:168–180
Chapter 9
Neuroimmune Mechanisms of Depression in Adults

with Heart Failure
Jessica A. Jiménez, Christine Tara Peterson, and Paul J. Mills
Abstract
Heart failure (HF) is a major and costly public health concern, and its prognosis is grim—with high
hospitalization and mortality rates. HF affects millions of individuals across the world, and this condition
is expected to become “the epidemic” of the twenty-first century (Jessup et al., 2016). It is well docu-
mented that individuals with HF experience disproportionately high rates of depression and that those who
are depressed have worse clinical outcomes than their nondepressed counterparts. The purpose of this
chapter is to introduce the reader to the study of depression in HF, and how psychoneuroimmunologic
principles have been applied to further elucidate mechanisms (i.e., neurohormonal and cytokine activation)
linking these comorbid disorders.
Key words Heart failure, Depression, Inflammation, Renin-angiotensin-aldosterone system, Sympa-

thetic nervous system, Gut microbiota and metabolites
1 Introduction
Heart failure (HF) is a major public health concern, especially in

societies where a sizable proportion of the population is over
65 years of age. HF is often the last stage of cardiovascular disease,
and its prognosis is grim—with high hospitalization and mortality
rates. Interestingly, heart disease, including HF, is often accompa-
nied by a psychological symptom complex, including low mood,
hostility, anger, and poor quality of life [1].
In recent years, the study of depression in HF has garnered
scientific interest due to its high prevalence in individuals with HF
and its strong tendency to worsen medical prognosis
[1–6]. Although the etiology of depression in HF remains unclear,
the disorders appear to share a similar pathogenesis involving dis-
turbance of the balance between sympathetic and parasympathetic
tone and increased inflammation, as evidenced by elevated circulat-
ing levels of proinflammatory cytokines [2, 4]. Considering that
depression has also been associated with incident HF [7, 8], most
145
146 Jessica A. Jiménez et al.
scholars favor a bidirectional pathophysiology, in which depression

may precede or follow the development of HF.
The purpose of this chapter is to introduce the reader to the
study of depression in HF, and how psychoneuroimmunologic
principles have been applied to further elucidate the mechanisms
linking these comorbid disorders. We begin the chapter with a brief
discussion of the epidemiology and pathophysiology of HF, and
then of the characteristics and consequences of depression in HF,
and conclude with discussion and presentation of relevant psycho-
neuroimmunological findings concerning the shared pathophysiol-
ogy of depression and HF.
2 Heart Failure
2.1 Epidemiology In the United States, among those aged 20 years and older, the
of Heart Failure prevalence of HF is 2.2% [9, 10]. At 40 years of age, the lifetime risk
of developing HF is 1 in 5. HF incidence rates in men approxi-
mately double with each 10-year age increase from 65 to 85 years;
however, the HF incidence rate triples for women between ages
65 and 84 years [11]. Researchers from the cohort study Multi-
Ethnic Study of Atherosclerosis found that African-Americans had
the highest risk of developing HF, followed by Hispanic, White,
and Chinese-Americans (4.6, 3.5, 2.4, and 1.0 per 1000 person-
years, respectively). This higher risk may reflect racial and ethnic
disparities as evidenced by the prevalence of hypertension, DM, and
low socioeconomic status among various groups [12].
Despite significant advances in treatment, the prognosis for
patients remains grim: 29.6% of HF patients die within 1 year of
diagnosis and 50% die within 5 years [13]. Approximately three
million patients get hospitalized each year with a primary or sec-
ondary diagnosis of HF making it one of the most common causes
of hospitalization in the elderly population in the United States
[14]. HF costs approximately $31 billion, and is projected to cost a
total of $70 billion by 2030 [11].
2.2 Definition The American College of Cardiology (ACC) and American Heart
and Classification Association (AHA) define HF as a complex clinical syndrome that
of Heart Failure can result from any structural or functional cardiac disorder that
impairs the ability of the ventricles to fill with or eject blood
[15]. The upper chambers of the heart are composed of the right
and left atria, and the lower chambers include the right and left
ventricles. The ventricles are muscular chambers that contract to
pump blood (systole). After systole, the ventricle muscles normally
relax during diastole, allowing blood from the atria to fill the
ventricles. The heart’s ability to pump can be compromised via
two mechanisms: (1) reduction in the volume of oxygenated
blood ejected from the left ventricle as a result of diminished
Neuroimmune Mechanisms of Depression in Adults with Heart Failure 147
myocardial contractility and (2) inadequate venous return to heart,

resulting from impaired ventricle filling and relaxation.
Although HF varies in its etiologies and clinical features, it can
be broadly classified into two categories: HF with systolic dysfunc-
tion (also known as “HF with reduced ejection fraction” (HFrEF)),
characterized by a reduced left ventricle ejection fraction (LVEF),
which is a measure of the percentage of blood that is ejected from
the heart into the aorta with each systole, and HF with preserved
ejection fraction (HFpEF) which is a complex disorder, where
LVEF is normal or mildly abnormal. As far as treatment and out-
come are concerned, patients with HFrEF respond favorably to the
standard pharmacological treatment regimen and demonstrate bet-
ter prognosis. In contrast, patients with HFpEF have not been
shown to respond to standard pharmacological treatments, except
for nitrates, and therefore have a poor prognosis, especially during
the decompensated phase of HF [16]. Other left ventricle
(LV) abnormalities include abnormal relaxation and filling, con-
centric remodeling, hypertrophy, increased extracellular matrix,
abnormal relaxation and filling, decreased diastolic distensibility,
and abnormal calcium handling.
There are two primary scales that are used to classify HF. The
New York Heart Association (NYHA) functional scale, which is
based on severity of symptoms and exercise intolerance, classifies
HF in categories from I to IV (Table 1). However, symptom
severity correlates poorly with many measures of LV functions.
Although there is a clear relationship between severity of symptoms
and survival, even patients with mild symptoms may still have an
increased risk of hospitalization and death [17]. The other scale is
the American College of Cardiology/American Heart Association
(ACC/AHA) scale, a newer classification that stages patients as
either A, B, C, or D based on structural changes and symptoms
(Table 1) [15]. The ACC/AHA staging system classifies HF as a
progressive disease, and once a particular stage is reached there is no
opportunity to transition to a lower stage (e.g., a stage C HF
patient cannot return to stage B). This system is often complemen-
ted by the NYHA functional classification system. In contrast,
ACC/AHA Stage C patients can shift between functional classes
I–IV at any given time. Movement up and down NYHA classes is
common, depending on the clinical status of the patient during the
time of assessment.
3 Pathophysiology of Heart Failure
HF is characterized by activation of the renin-angiotensin-aldoste-

rone system (RAAS) and the sympathetic nervous system (SNS), as
well as inflammatory pathways. Regardless of its etiology and clas-
sification, HF begins with injury to the myocardium (e.g., years of
Table 1
Functional classifications and disease progression stages of heart failure
New York Heart Association Functional (NYHA) classes
Definition Examples
NYHA No limitation of physical activity Ordinary physical activity does not cause
Class I undue fatigue, palpitation, or dyspnea
(shortness of breath)
NYHA Slight limitation of physical activity Comfortable at rest, but ordinary physical
Class activity results in fatigue, palpitation, or
II dyspnea
NYHA Marked limitation of physical activity Comfortable at rest, but less than ordinary
Class activity causes fatigue, palpitation, or
III dyspnea
NYHA Unable to carry out any physical activity Symptoms of cardiac insufficiency at rest. If
Class without discomfort any physical activity is undertaken,
IV discomfort is increased
American College of Cardiology/American Heart Association stages of heart failure
Definition Examples
Stage A High risk for developing HF, but without Hypertension, diabetes mellitus, CAD, family
structural heart disease or symptoms of HF history of cardiomyopathy
Stage B Structural heart disease, but asymptomatic Previous myocardial infarction, left ventricular
dysfunction, valvular heart disease
Stage C Structural heart disease with previous or Structural heart disease, dyspnea and fatigue,
current symptoms, but managed with impaired exercise tolerance
medical treatment
Stage D Marked symptoms at rest despite maximal Advanced disease requiring hospital-based
medical therapy support, a heart transplant, or palliative care
sustained hypertension and/or myocardial infarction), which

reduces cardiac output. In response, the body engages in a series
of compensatory mechanisms, including (1) maintaining perfusion
pressure by increasing the circulation of blood volume; (2) activat-
ing immune and inflammatory pathways; and (3) restructuring
cardiac muscle cells and reshaping the ventricle chamber (called
remodeling). This systematic response involves complex interac-
tions between the RAAS and SNS, which are collectively referred
to as neurohormonal responses.
Neurohormonal activation and cytokine activation are
designed for acute responses to injury, but prolonged activation
of these compensatory mechanisms eventually leads to further
declines in cardiac functioning. Currently, the most successful phar-
macological therapies for HF block aspects of the body’s
compensatory responses to myocardial injury [15]; thus, there is

increasing scientific interest in understanding neurohormonal and
cytokine activation in the context of HF.
3.1 Renin- The RAAS system maintains renal blood flow after the myocardium
Angiotensin- has sustained injury via its effects on remodeling the vasculature
Aldosterone System and increasing plasma volume. Decreased renal perfusion pressure
results in secretion of renin by juxtaglometer cells lining the affer-
ent renal arterioles. Specifically, renin cleaves angiotensinogen to
form decapeptide angiotensin I. The angiotensin-converting
enzyme (ACE) cleaves two C terminal amino acids to form angio-
tensin II, the primary effector of the system. Receptors for angio-
tensin II are divided into subtypes, AT-1 and AT2. AT-1 is the
predominate subtype in the vascular endothelium and a primary
target for pharmacologic blockade. Binding of angiotensin II to
AT-1 receptors results in increases in the release of intracellular
calcium from the sarcoplasma reticulum via activation of protein
kinase C. The binding of angiotensin II in the vasculature results in
an increase in systematic vasculature resistance and restoration of
blood pressure.
Prolonged compensatory actions of the RAAS in HF bring
adverse consequences, however, including increased vascular resis-
tance. Increases in resistance create undue myocardial burden and
decreased cardiac output, resulting in left ventricular hypertrophy.
Angiotensin II initiates apoptosis and interstitial fibrosis, which
contribute to the remodeling of the extracellular matrix in the
myocardium (e.g., myocyte hypertrophy). The effects of angioten-
sin II on the myocardium and peripheral vasculature result in
decreased cardiac output and renal perfusion. Angiotensin II is
also involved in the increase of plasma volume by initiating produc-
tion of mineralocorticoid aldosterone by the adrenal cortex. Aldo-
sterone acts on the distal tubules of the renal nephron and activates
a sodium-potassium exchange, which results in the retention of
sodium and water. The increased plasma volume exacerbates fluid
overload and peripheral edema. Chronic excess of aldosterone leads
to increased fibrosis in the atria, ventricles, kidneys, and
perivasculature.
3.2 Sympathetic Vascular baroreceptors respond to declines in cardiac output and

Nervous System stroke volume by increasing sympathetic nerve activity and conse-
quent release of the catecholamine norepinephrine. Sympathetic
activation improves cardiac output by increasing heart rate, myo-
cardial contractility, and stroke volume. Sympathetic activation also
increases systematic resistance and blood pressure in the peripheral
vasculature, and via catecholamines increases renin release and
angiotensin II production, further increasing vascular resistance
and afterload.
The direct effects of sympathetic activation on the myocardium

itself are primarily mediated via two classes of β-adrenergic recep-
tors, namely the β-1 and β-2 receptors. In a normal heart, β-1s
comprise approximately 80% of the total β-adrenergic receptor pool
[18]; however, chronic sympathetic activation significantly down-
regulates β-1 receptors leaving a greater proportional presence of
β-2 receptors of approximately 40% [18]. While β-2 receptors are
less downregulated, they are susceptible to inactivation from repet-
itive agonist stimulation, becoming less responsive to adrenergic
agonists. Although the release of norepinephrine acutely increases
myocardial contractility, chronic stimulation in HF worsens cardiac
function (direct cytotoxic effects), resulting in progressive dysfunc-
tion of the left ventricle, worsening pulmonary edema, and poten-
tially death. Higher levels of circulating norepinephrine have been
associated with poorer survival, and greater functional decline
in HF.
3.3 Systemic The “cytokine hypothesis” proposes that HF progression is an

Inflammation inflammatory process and that elevated levels of proinflammatory
cytokines worsen LV dysfunction. There is a significant body of
evidence to suggest that elevated levels of cytokines are associated
with cardiac decline in HF. Particularly, the inflammatory cytokines
tumor necrosis factor-alpha (TNF-α) and interluekin-6 (IL-6) are
among the most widely studied cytokines in HF.
TNF-α is a polypeptide that activates endothelial cells, recruits
inflammatory cells, and enhances the production of other proin-
flammatory cytokines. TNF-α is secreted from immune cells during
the early stages of HF; however in the final stages, the cardiac
myocytes secrete TNF-α in high quantities. TNF-α appears to be
particularly important in the transition from compensated to acute
decompensated HF, the latter being a state of exacerbated HF
requiring hospitalization. TNF-α has been extensively studied in
animal models, and overexpression of TNF-α by the cardiac myo-
cytes leads to inflammatory myocarditis and subsequent myocyte
hypertrophy, LV dilatation, and progressive LV dysfunction. For
example, exogenous administration of TNF-α at concentrations
comparable to those observed in HF produces significant declines
in myocardial contractility, worsening LV dysfunction and increas-
ing pulmonary edema.
IL-6 is also elevated in HF, particularly in the end stages of the
disease. Although IL-6 was initially thought to have only proin-
flammatory effects like those of TNF-α, research in murine models
suggests that IL-6 also plays an immune-modulatory role in
response to secretion of TNF-α. IL-6 has direct effects on the
myocardium, including decreasing contractility, activating matrix
metalloproteinases, and contributing to LV remodeling. Like
TNF-α, IL-6 is secreted from the myocardial cells during the end
stages of HF, but not during mild or moderate stages of the disease.
4 Depression in Heart Failure
4.1 Epidemiology Depression in HF has been extensively studied because of its high
of Depression in Heart prevalence in HF patients and its tendency to worsen medical
Failure Patients prognosis [1–6]. Studies show that depression is one of the most
important modifiable risk factors because it is responsible, in part,
for the high readmission rates among HF patients. The estimated
prevalence of depression in HF is 24–42%, which is 2–3 times
higher than the general population. The prevalence rate of minor
depression, as defined by the Beck Depression Inventory 10, is as
high as 35.5%. Although the majority of studies demonstrate that
depression worsens cardiac prognosis, the magnitude of the effects
has varied greatly depending on how depression is measured [2].
Depression has also been associated with incident HF. In a
community sample of 2501 patients (mean follow-up 14 years),
Williams and colleagues found that depression was an independent
predictor of developing HF in women (HR ¼ 1.96, 95% CI ¼ 1.11,
3.46, p ¼ 0.02), but not in men [7]. In a study of 4538 older adult
patients enrolled in the Systolic Hypertension in the Elderly Pro-
gram (SHEP), depressed patients were 2.82 times (95% CI ¼ 1.71,
4.67; p < 0.001) more likely to develop HF over a 4.5-year follow-
up period [8]. The association between depression and HF did not
significantly vary by sex. Although not all studies have found gen-
der effects, Williams et al.’s findings follow trends seen in
depressed, non-HF populations: the National Comorbidity Survey,
for example, reported a 1.7 greater odds of women developing
depression at some point in their lifetime compared to men [19].
4.2 Patient The prevalence of depression in HF appears to significantly differ by

Characteristics health status, demographics, and social factors. In their meta-
and Depression analysis of depression in HF, Rutledge et al. found that 11–25%
of outpatients and 35–70% of inpatients were depressed
[20]. Rutledge and colleagues also found that depressive symptom-
atology increased with severity of HF diagnosis, ranging from 11%
of patients with New York Heart Association (NYHA) functional
class I to 42% of NYHA class IV patients [20]. Several studies have
found higher depression in younger as well as female HF patients
[2, 21]. Social factors may contribute to incident depression in
HF. In a study of 245 HF patients without depression at baseline,
living alone, alcohol abuse, perception of medical care as being a
substantial economic burden, and poor health status were indepen-
dent predictors of developing depressive symptoms [22]. The
effects appear to be multiplicative in nature: 15.5% of patients
developed depression when only one of the factors were present,
36.2% developed depression when two factors were present, and
69.2% developed depression when three or more were present.
4.3 Clinical Of great clinical significance, studies find that depression has
Outcomes in Heart adverse effects on the course and prognosis of HF. Increased psy-
Failure Patients chological surveillance of HF patients over the past 15 years has
with Depression highlighted the pivotal role of depression in HF [23]. Development
of depressive symptoms after the diagnosis of HF is associated with
a 1.5- to 2.2-fold increase in all-cause and cardiovascular mortality
[24]. Depression is an independent risk factor for hospital readmis-
sion in patients with HF [25–27]. In the Telemonitoring to
Improve Heart Failure Outcomes Trial, the 30-day readmission
rate was 17.1% [28]. Kato et al. found that depression was asso-
ciated with more cardiac mortality or HF hospitalization in both
HFrEF (55% vs. 12%, p < 0.01) and HFpEF (35% vs. 11%,
p < 0.05; [29]). Sherwood et al. [30] reported that depressive
symptomatology was associated with a 1.56 (95% CI; 1.07, 2.29;
p < 0.001) increased risk of death or hospitalization during a
median 3-year follow-up period. In a sample of 374 patients hospi-
talized for HF, Jiang and colleagues [31] found that HF patients
with major depression had 2.23 greater odds (95% CI 0.04, 4.77;
p ¼ 0.04) of mortality and 3.07 (95% CI 1.41, 6.66; p ¼ 0.005)
greater odds of readmission at 1 year compared to HF patients with
no depression. In a sample of 1006 hospitalized HF patients, Jiang
et al. [32] found that patients whose Beck Depression Inventory
(BDI) scores were 5 to 9, 10 to 18, and 19 were 21%, 53%, and
83%, respectively, more likely to die than patients whose BDI score
was 5 ( p < 0.001). Vaccarino et al. [33] also found that there was
a grade associated between the number of depressive symptoms and
increased risk of death or decline of daily living at 6 months. In this
prospective study of 391 patients with decompensated HF on
admission to the hospital, patients with 11 depressive symptoms,
compared with those with <6 depressive symptoms, had an 82%
higher risk of either functional decline or death. In a study of
longitudinal outcomes (mean follow-up 39 months) in HF patients
with comorbid atrial fibrillation, Frasure-Smith and colleagues [34]
found that elevated depressive symptoms significantly predicted
cardiovascular mortality (HR: 1.57; 95% CI 1.20, 2.07;
p < 0.001). The authors also commented that the increased risk
of death was similar to risks associated with not taking standard
medications to manage HF, such as anticoagulants and aldosterone
antagonists. Worsening depressive symptomatology over time is
also associated with increased risk of adverse outcomes. A study of
147 HF outpatients found that a 1 point change in BDI scores was
associated with 1.07 increased risk of death or cardiovascular hos-
pitalization (95% CI 1.02, 1.12, p ¼ 0.007) [24]. The results from
these studies indicate that depression is an independent predictor of
worse clinical outcomes in HF patients [23].
4.4 Quality of Life In addition to its cardiotoxic effects, depressed HF patients suffer
in Heart Failure from reduced physical functioning and of course worse quality of
Patients life: depressed patients report poorer quality of life and greater
with Depression functional impairment than nondepressed patients, even when
compared with patients of a higher (i.e., worse) NYHA functional
class, which may suggest that patient perceptions of physical func-
tioning, rather than the clinical status itself, predict quality-of-life
outcomes [35, 36]. In a small study (n ¼ 58) of associations
between disease severity, functional status, depression, and daily
quality of life, greater depression severity was positively associated
with worse self-reported physical and emotional quality of life in
HF patients. A recent study by Hallas et al. [37] conducted in
146 HF patients found that patients with more negative beliefs
about the consequences of HF, and less perceived control, were
more anxious and depressed compared to patients with more posi-
tive beliefs. Greater depression ratings also predicted poorer quality
of life. Patients with more negative beliefs also had more maladap-
tive behaviors and less coping resources, which may also have
downstream effects on quality of life. In a study of 155 HF patients,
Gottlieb et al. [21] found that depressed patients scored signifi-
cantly worse than nondepressed patients on all components of the
quality-of-life questionnaires. In a more recent study, Gottlieb and
colleagues [38] demonstrated that depression is minimally related
to objective assessments of HF severity, such as peak O2 consump-
tion, B-type natriuretic peptide (BNP) levels, or ejection fraction.
However, depression significantly affects subjective measurements
of HF severity, such as NYHA classification or 6-min walk test
[10]. Undoubtedly, depression negatively affects quality of life,
but there is building evidence to also suggest that depression alters
patient perceptions of physical functioning and disease severity,
which may result in poorer ratings on subjective measures.
4.5 Screening Given that depressive symptoms are a strong predictor of HF out-
for Depression in Heart comes, screening has become an important part of patient assess-
Failure Populations ment [20]. The American Heart Association (AHA) advocates a
two-step screening process using two versions of the Patient Health
Questionnaire (PHQ): the PHQ-2, which includes two items from
the PHQ-9 is administered as a first-line screening for depression.
Positive results on the PHQ-2, the PHQ-9 is subsequently admi-
nistered for further evaluation [39]. The usefulness of the PHQ-2
or the PHQ-9 results as a predictor of prognosis in patients with
HF has been examined in many studies [40–42]. However, it is
unclear if there is any advantage in using a two-step screening
process as opposed to using the PHQ-2 alone or the PHQ-9
alone [43].
4.6 Treatment It is difficult to diagnosis depression in HF due to the overlapping

of Depression on Heart symptoms of the two conditions. Considering the prevalence of
Failure Outcomes depression in HF, as well as the magnitude of its negative effects on
the clinical outcomes, it is surprising that as few as 7% of patients
receive antidepressant therapies [16]. Depressive symptoms con-
tinue to be persistent in HF patients even in the face of antidepres-
sant therapy. In the Heart Failure and a Controlled Trial
Investigating Outcomes of Exercise Training (HF-ACTION)
study, a randomized trial of aerobic exercise in patients with systolic
HF, Gottlieb et al. reported that 29% of patients taking antidepres-
santshad at least mild depression (BDI 10; [38]). However, even
under the most ideal conditions, such as the Sertraline against
Depression and Heart Disease in Chronic Heart Failure
(SADHART-CHF) where patients in the intervention group
receive 12 weeks of antidepressant therapy as well as a nurse facili-
tated support intervention, 46% of the treatment group did not
remit during the trial [44]. Although there is no strong evidence to
indicate that treating depression improves cardiovascular outcomes
in HF [6], the SADHART-CHF study found that remission from
depression, regardless of treatment assignment, was associated with
fewer cardiovascular events [45].
Serotonergic agents are the first line of antidepressant treat-
ment even though placebo-controlled, randomized trials of sertra-
line and escitalopram have not demonstrated improved HF
outcomes [1]. There is evidence to suggest that more complex
interventions that incorporate cognitive-behavioral therapy may
reduce depressive symptomatology, anxiety, and fatigue, but there
is no evidence to suggest that these interventions significantly
improve cardiovascular outcomes or self-care practices related to
HF management, such as low-sodium diet, exercising, taking pre-
scribed medications, and monitoring edema [46]. Given that HF
patients are disproportionately affected by depression, and those
suffering with it experience worsened health outcomes, under-
standing relationships between HF and depression and its treat-
ment is of growing importance among many clinicians.
5 Psychoneuroimmunology: Understanding Pathophysiological Links Between

Heart Failure and Depression
Given the adverse effects of depression on cardiac prognosis, under-

standing the biological pathways that link HF and depression may
provide routes for pharmacological and behavioral interventions.
The comorbid nature of HF and depression suggests that they share
a similar pathophysiology: inflammation.
After more than a decade of research on this topic, although the
mechanistic relationships between depression and inflammation are
not fully understood, much progress has been made. Depressed

patients without HF have significantly higher levels of IL-1β, IL-6,
TNF-α, and interferon-gamma (IFN-γ) [47]. Potential sources of
inflammatory activation in HF include SNS activation and hyperac-
tivity of the hypothalamic-pituitary-adrenal axis [48, 49].
5.1 Lessons Learned One of the main lines of investigation in our laboratory has been
on Inflammatory the application of the cytokine model and other theoretical models
and Other Immune of broader immune activation in the context of depression in
Responses HF. We have conducted several studies that lend support to the
in Depression in HF theory that systematic inflammation, as well as broader dysregula-
tion of immune system, may underlie the relationship between
depressive symptoms and progression of HF. Our approach to the
study of these topics is the use of both cross-sectional and prospec-
tive studies. For the former, we have assessed a broad range of
inflammatory and cell adhesion biomarkers in patients with estab-
lished HF and with a range of depressive symptomatology. For the
latter, we are studying ACC/AHA Stage B patients, individuals
who were at risk for developing symptomatic HF but who were
not symptomatic. In these individuals, we are assessing a broad
range of inflammatory and cell adhesion biomarkers as well as
depression, but this time repeatedly over the course of several
years. The intention is to temporally model inflammation as it
relates to the onset and offset of depression and to the onset and
progression from non-symptomatic to symptomatic HF.
5.2 Assessing A potential explanation for the link between depression and HF
Depressive Symptoms may be inherent due to the diagnosis of depression, and measure of
its severity. The diagnostic criteria for major depressive disorder as
well as screening for depressive symptoms include both cognitive
and somatic symptoms. Thus, depressive symptomatology, such as
fatigue, loss of energy, problems concentrating, weight loss or gain,
and sleep disturbance, may be the result of underlying cardiac
dysfunction [5]. More recent studies have recognized this overlap,
and now it is favored to report somatic and cognitive symptomatol-
ogy separately as we have done in our studies.
In our studies we have primarily measured depression via the
Beck Depression Inventory (version–IA; BDI-IA), which is a
21-item self-administered assessment of extent to which patients
experience depressive symptoms [50]. Scores of 0–9 indicate mini-
mal or no depression, 10–18 mild-moderate depression, 19–29
moderate-severe depression, and 30–63 severe depression. The
reliability of this measure in our samples has been α > 0.90. A
BDI score of less than 10 is usually not considered to be clinically
significant depression, but yet studies have shown scores of 4–9 to
be associated with increased mortality in post-myocardial infarction
patients [2]. Since there has been some evidence to suggest that
cognitive/affective and somatic aspects of depression differentially
relate to clinical course in HF [51], we also independently examine

these BDI subscales of depression in our studies: (1) the cognitive/
affective subscale assesses symptoms such as sadness and dissatisfac-
tion (13 items, score range 0–39) and (2) the somatic subscale
assesses features such as changes in appetite and feelings of fatigue
(7 items, score range 0–21).
5.3 Assessing Circulating TNF-α, IL-6, and IL-1β levels are determined in plasma
Cytokines and Cellular by commercial ELISA. For IL-6, the intra-assay CV (%) is 2.2, the
Adhesion Molecules inter-assay CV (%) is 3.9, and the assay sensitivity is <0.71 pg/mL;
for TNF-α, the intra-assay CV (%) is 8.0, the inter-assay CV (%) is
16.3, and the assay sensitivity is <0.18 pg/mL; for IL-1β, the intra-
assay CV (%) is 6.8, the inter-assay CV (%) is 8.3, and the assay
sensitivity is <0.1 pg/mL.
Circulating CRP levels are determined in plasma using com-
mercial ELISA. Intra- and inter-assay coefficients of variation are
<5%. Precision and sensitivity performance values are excellent:
intra-assay CV (%) < 1.0, inter-assay CV (%) ¼ 1.6, and sensitivity
<0.05 mg/L.
Soluble intercellular adhesion molecule-1 (sICAM-1, sCD54)
and sP-selectin (sCD62P) are too determined by commercial
ELISA. The precision and sensitivity performance values are as
follows: sICAM-1 (intra-assay CV (%) ¼ 4.6, inter-assay CV
(%) ¼ 6.6, sensitivity <0.35 ng/mL) and sCD62P (intra-assay
CV (%) ¼ 5.1, inter-assay CV (%) ¼ 8.8, sensitivity <0.5 ng/mL).
Proinflammatory cytokines, such as TNF, IL-6, and CRP, are
associated with cardiac dysfunction in both human and animal
models. However, in 2009 Wirtz et al. [52] in our group provided
the first study investigating whether depressive symptoms were
associated with exercise-induced increases in circulating levels of
adhesion molecules expressed on endothelial cells (sP-selectin and
soluble sICAM-1), leukocytes (sICAM-1), and platelets
(sP-selectin). Using data from 39 middle-aged male HF patients
and 19 male control subjects, the authors found that higher depres-
sion symptomatology moderated greater increases in sP-selectin
levels in response to an acute exercise challenge over time in HF
patients as compared with control subjects (F ¼ 3.25, p ¼ 0.05).
Post hoc testing revealed that in HF patients, higher depression
scores (BDI) were significantly associated with greater increases in
sP-selectin levels over time in response to the exercise (Fig. 1). Also,
in the HF patients, higher BDI scores were associated with higher
sP-selectin levels at pre-exercise and post-exercise time points (main
effect of BDI: F ¼ 4.86, p ¼ 0.035). These effects were not found
for the control subjects. These findings suggested that levels of
sP-selectin are higher before and after exercise and have greater
increases in response to exercise in male HF patients with increasing
depressive symptom severity. These findings could have
90
Low BDI score (n=23)
High BDI score (n=15)
P-selectin (sCD62p, pg / ml)

80
70
60
50
40
30 Exercise
-5 0 5 10 15 20 25 30 35
Time in minutes
Fig. 1 Changes in sP-selectin in response to acute exercise in HF patients with

high (>10, n ¼ 15) and low (10, n ¼ 23) BDI scores
implications for acute coronary syndromes associated with exercise

and thereby may impact mortality.
While it is widely acknowledged that indictors of inflammation
are cross-sectionally associated with both depression and HF sever-
ity, our laboratory was among the first to explore whether different
types of inflammatory markers prospectively predict depressive
symptom in HF patients. Wirtz et al. [53] assessed the relationship
of proinflammatory cytokines and cellular adhesion molecules on
depressive symptoms at 12 months following initial study of 30 HF
patients. The authors found that sICAM-1—but not IL-6 or
C-reactive protein (CRP)—was associated with depression scores
12 months later (r ¼ 0.38, p ¼ 0.045). Hierarchical linear regres-
sion models revealed that sICAM-1 significantly predicted depres-
sion scores at the 12-month follow-up, with sICAM-1
independently explaining between 7% (β ¼ 0.26, p ¼ 0.040) and
10% (β ¼ 0.35, p ¼ 0.045) of the total variance in depression scores.
These findings suggest that the adhesion molecule sICAM-1 is an
independent, prospective predictor of depressive symptoms in
HF. The prospective nature of these findings supports the sug-
gested role for inflammation in increasing the severity of future
depressive symptomatology.
5.4 Assessing Until recently, immune cell migration, particularly chemotaxis, has
Chemotaxis been largely ignored in respect to depression symptoms and
and Cellular Immunity HF. Chemokines are essential for providing signaling to leukocytes
for extravasation from the blood and directing locomotion
[54, 55]. When overexpressed, recruitment and migration factors
are injurious to the cardiovascular system [56] and can generate
angiogenesis and fibrous tissue deposition, which can lead to myo-
cardial dysfunction in HF [57, 58]. Particularly, studying acute
physiologic responses to controlled challenges serves as a window
into the complex physiologic processes involved in cardiac

diseases [59].
One of the members of our group therefore developed an
in vitro chemotaxis assay to assess functional capacity of peripheral
blood mononuclear cells (PBMCs). PBMCs are separated from
whole blood using Ficoll-Hypaque, washed, and then resuspended
in RPMI 1640 with 20 mmol/L HEPES (serum-free media). In a
modified Boyden chamber, the patient’s PBMCs are then incu-
bated with either the bacterial peptide f-met leu phe (fMLP), the
physiologic chemokine stromal cell-derived factor-1 (SDF-1), or
the adrenergic agonist isoproterenol, or chemotaxis buffer. Che-
motaxis responsiveness of PBMCs to the bacterial peptide fMLP is
commonly used to measure nonspecific natural immune activity.
The chemokine SDF-1 binds to its specific receptor CXCR4, and
subsequently stimulates lymphocyte adhesion and transendothelial
migration, playing a role in adaptive cellular immunity. Levels of
SDF-1 and CXCR4 are elevated in patients with HF and have been
found to attenuate cardiac myocyte contractility [37]. PBMCs are
incubated with these agents for 2 h at 37 C, then the top of the
membrane is then gently rinsed with phosphate-buffered saline,
and the non-migrated cells are removed. The membrane is then
removed from the plate and briefly submerged in phosphate-
buffered saline. Once dry, the membrane is read by a fluorescence
plate reader (CytoFluor) at an excitation of 485 nm and emission of
530 nm.
Redwine and colleagues [60] studied the relationship between
depressive symptoms and PBMC chemotaxis both at rest and in
response to a moderate acute exercise challenge in 65 middle-aged
HF patients and 45 non-HF control subjects. Chemotaxis of
PBMCs was examined in vitro to either fMLP or SDF-1 immedi-
ately before and after the exercise. The author found that HF
patients had reduced chemotaxis to SDF-1 compared with
non-CHF subjects ( p < 05). The authors also found that higher
BDI scores were significantly associated with reduced baseline che-
motaxis to SDF-1 in both CHF and non-CHF subjects
( p ¼ 0.025). In contrast, higher BDI scores were associated with
increased chemotaxis to fMLP and SDF-1 in response to exercise in
the HF patients ( p ¼ 0.027; Fig. 2). The authors also found that
cognitive depressive symptoms, but not somatic depressive symp-
toms, were inversely associated with baseline chemotaxis to fMLP
and SDF-1 in HF and controls. When stratified by HF diagnosis,
these associations persisted when controlling for covariates. How-
ever, neither cognitive nor somatic symptoms were associated with
changes in chemotaxis from pre- to post-exercise task.
The results from the study suggested a shift in immune cell
mobility in HF patients with greater depressive symptom severity,
with reduced chemotaxis to a physiologically specific chemokine at
rest but increased chemotaxis to both nonspecific and specific
0.20
0.15
0.10
Chemotaxis
Stimulation 0.05
Index (log
transformed) 0.00
-0.05
-0.10
-0.15
Pre- exercise Post- exercise
CHF- hi BDI
CHF- lo BDI
non CHF- hi BDI
non CHF- lo BDI
0.15
0.10
0.05
Chemotaxis
Stimulation 0.00
Index (log
transformed) -0.05
-0.10
-0.15
-0.20
-0.25
Pre- exercise Post- exercise
CHF- hi BDI
CHF- lo BDI
non CHF- hi BDI
non CHF- lo BDI
Fig. 2 Changes in a logarithmic transformed stimulation index (SI) of chemotaxis to fMLP (TOP PANEL) and the
chemokine SDF-1 (BOTTOM PANEL) in HF patients and non-HF controls with high (hi) and low (lo) Beck
Depression Inventory (BDI) scores in response to exercise. Data are expressed as means SEM
chemical attractants in response to physical activity. Findings could

have implications for cardiac repair and remodeling in HF patients
and therefore disease progression.
A second chemotaxis study conducted by Redwine and collea-
gues [61] investigated if depressive symptoms were related to
alterations in the sensitivity of PBMCs to the β-adrenergic agonist
isoproterenol in patients at rest and after acute exercise in
77 patients with HF and 44 controls. As mentioned previously,
sympathetically modulated immune dysregulation is a part of the
pathophysiology of HF; however, this process may be exacerbated
in the presence of depression. The study results indicated that
depressive symptom severity ( p ¼ 0.001) and higher resting levels
of plasma norepinephrine ( p ¼ 0.003) were associated with greater
chemotaxis after exercise in patients with HF (Fig. 3). The authors
concluded that patients with HF with higher depressive symptoms
and plasma norepinephrine exhibit increased circulating immune
cell chemotaxis to isoproterenol, suggesting greater adrenergic
sensitivity. Increased immune cell migration in patients with HF
0.20
0.15
0.10
Chemotaxis 0.05
Stimulation
Index (log
0.00
transformed)
-0.05
-0.10
-0.15
1 nM 10 nM 100 nM
Isoproteronol
non-HF lo BDI
non-HF hi BDI
HF lo BDI
HF hi BDI
Fig. 3 Change scores (pre- minus post-exercise) and chemotaxis to three concentrations of isoproterenol
(1 nM, 10 nM, and 100 nM/l) in HF patients and non-HF controls. High (hi) versus low (lo) depression are
determined by scores 10 and <10 on the Beck Depression Inventory (BDI). Data expressed as means
SEM
who have elevated depressive symptoms could be associated with

cardiac remodeling and HF disease progression.
Following the results of the previous study, which suggested
increased β-AR sensitivity, Redwine and colleagues explored the
sensitivity of the leukocyte beta-adrenergic receptor and depression
sensitivity [62]. Depression was measured using the self-report
(BDI) and the Structured Clinical Interview for the DSM-IV
(SCID). Patients with major depression determined by SCID had
significantly higher β-AR sensitivity than nondepressed. However,
the BDI revealed a more complex relationship. Minimal, mild, and
moderate-to-severe depression symptom groups had significant
differences in β-AR sensitivity (F(7, 73) ¼ 7.03, p ¼ 0.002,
η2 ¼ 0.18), with mild symptoms corresponded with reduced
β-AR sensitivity and moderate-to-severe symptoms with higher
β-AR sensitivity.
5.5 Th1/Th2 Ratio Although, as we have discussed, inflammatory cytokines have been
implicated as a possible mediator of psychological symptoms of
depression and HF, it has been unclear if systematic inflammation
represents a broader dysregulation of the immune system. Particu-
larly, cellular immunity is important for protection against infec-
tion. Th1 cells promote cellular immunity by rapidly producing a
range of cytokines such as IFN-γ that activate other Th1 cells to
fight infectious agents. Th1 cells also exert a negative regulatory
role on Th2 cells that produce cytokines such as IL-4 and IL-10.
Th2 cytokines, on the other hand, attenuate immune defenses if
they are locally over-expressed, by decreasing activities of major
effectors such as Th1 cells [63, 64]. A Th2 shift may have a
profound effect on the susceptibility of the organism to infection
[65], increase inflammation, and lead to dilated cardiomyopathy
and HF [66]. Maintaining Th1/Th2 homeostasis is important for
preserving health. Thus, examining Th1/Th2 ratios can provide
information on the balance of cellular immune activation versus
negative regulation of cellular immunity.
Redwine et al. [67] examined the relationship of depressive
symptoms with cellular immune activity measured by the
Th1/Th2 ratio and cardiac rehospitalization and/or death in
18 HF patients (mean age ¼ 62, NYHA classes II–IV). The authors
reported that higher baseline depression scores were associated
with a prospective increase in incidence of cardiac-related hospita-
lizations and/or death ( p ¼ 0.037). Lesser IFN-γ/IL-10-expres-
sing CD4+ T cell ratios were related to higher depressive symptom
scores at baseline ( p ¼ 0.005, Fig. 4) and a prospective increased
incidence of cardiac-related hospitalization or death over a 2-year
period ( p ¼ 0.05). The results suggest that a shift in the Th1/Th2
ratio may play a role in the association between depressive symp-
toms and morbidity and mortality in HF patients, suggesting
broader immune dysregulation.
3.5
3.0
IFN gamma/IL-10 ratio
2.5
2.0
1.5
1.0
.5
0.0
0 10 20 30 40
HAM-D Score
Fig. 4 Relationship between IFN-γ/IL-10 ratios and Hamilton Depression Scores in HF patients
5.6 Alterations A burgeoning new area of interest in our laboratory is the role of
of the Gut, Gut the “heart-gut axis” and the “gut hypothesis” of HF, which are
Microbiota, recently emerging concepts that attempt to address the complex
and Metabolites pathophysiology of HF. HF is associated with altered gastrointesti-
in Heart Failure nal function due at least in part to ischemic conditions and conges-
tion within the gut. The host may then be affected by impaired gut
barrier function, resulting in systemic inflammatory responses and
oxidative stress. In the context of the microbiota, altered composi-
tion and/or function may influence metabolic processes in HF. For
example, the gut microbiota metabolizes dietary choline and
L-carnitine into trimethylamine which is then oxidized to trimethy-
lamine N-oxide (TMAO) in the liver [68]. Circulating TMAO is an
important predictive risk factor for cardiovascular disease and has
been observed as elevated in HF patients compared to age- and
sex-matched controls [69, 70]. Increased TMAO levels are also
associated with a greater-than-threefold increase in mortality risk
[70]. In addition, plasma concentrations of TMAO, choline, and
betaine have been observed as elevated in patients with chronic HF
compared to control subjects, and patients with New York Heart
Association classes III and IV have the highest levels [71]. More-
over, elevated TMAO has been observed as anticorrelated to long-
term survival in HF patients even after controlling for cardiorenal
parameters. As TMAO is detoxified through the kidneys, renal
functional status may affect TMAO levels. In HF patients, TMAO
concentration has been positively correlated with renal impairment

severity [72]. Gut dysbiosis can lead to increased TMAO and
uremic toxin levels and subsequently the progression of chronic
HF and kidney disease. The biological mechanisms by which the
gut microbiota or TMAO directly influences the development of
HF have not been fully elucidated. Further studies are needed to
clarify the mechanisms of impact of TMAO on cardiorenal function
and disease.
HF is associated with increases in various gut pathogens, intes-
tinal permeability, and inflammation [73]. Increased populations of
adherent gut bacteria have been identified in the mucosa of HF
patients compared to controls [74]. Altered gut microbiota com-
position can lead to increased systemic inflammation and uremic
toxins [75], originating from microbial metabolism, which have
both been associated with HF and cardiorenal compromise. In
addition, patients with chronic HF have increases in gut bacterial
pathogens such as Campylobacter, Salmonella, Shigella, and Yersi-
nia compared to controls. Candida yeast are also elevated in HF
compared to controls [76]. The intestinal microbial overgrowth
and increased intestinal permeability are associated with disease
severity, venous congestion, and inflammation. Numerous studies
suggest that decreased gut mucosal barrier function allows
gut-derived lipopolysaccharide (LPS) to enter the systemic circula-
tion. This systemic LPS exposure leads to generation of cytokines
such as IL-6 and TNF-α and subsequent inflammation and may
contribute to the development of cardiometabolic disease
[77–79]. Moreover, circulating LPS has been observed in patients
with diabetes and HF and may increase systemic inflammation in
these populations [80, 81]. In addition to endotoxin translocation,
hypoxia may also lead to increased intestinal permeability. The
reduced cardiac output, edema, and systemic congestion in HF
further increase the risk of intestinal ischemia [82]. In HF, intesti-
nal ischemia and congestion may lead to altered intestinal morphol-
ogy, function, and permeability [74, 83]. Since the gut maintains a
high oxygen demand, the small intestinal villi are susceptible to
ischemia from conditions that cause reduced blood flow. Intestinal
ischemia in turn leads to dysfunction such as reduced nutrient
absorption and pH of the gut mucosa [82].
Intestinal function and permeability are significantly impaired
in chronic HF [74]. In these patients, the wall of the colon is often
significantly thickened [84]. Edema localized to the intestinal wall
in HF patients due to systemic congestion and reduced abdominal
blood flow may increase LPS leakage into the circulation and
cytokine production. Cytokine production not only promotes
inflammation but also fibrosis and microvascular and myocardial
disease. HF patients with edema have higher plasma concentrations
of LPS compared to patients presenting without edema [85]. In
addition, patients with HF and low rate of intestinal blood flow
have higher serum levels of LPS antibody [86]. Numerous proin-

flammatory cytokines, such as IL-1, IL-2, IL-6, TNF-α, and
C-reactive protein, are observed as elevated in HF patients due in
part to circulating endotoxin [87, 88]. However, clinical trials
targeting the removal of these cytokines have largely failed; thus
endotoxins have been suggested as the trigger for increased cyto-
kine production in HF. The intestinal hypoperfusion and conges-
tion due to reduced cardiac output may further increase intestinal
permeability and bacterial translocation to create a vicious cycle of
additional inflammation, HF exacerbation, and potential renal
damage. Additional mechanistic studies as well as novel therapies
targeting the gut microbiota and metabolic dysfunction in HF are
warranted.
5.7 Altered Plasma HF is associated with depression, and key differences have been
Metabolites in Patients identified in the metabolome of depressed compared to nonde-
with Heart Failure pressed HF patients [89]. The amino acids aspartate and serine
and Depression are significantly elevated in depressed HF patients compared to
controls. Aspartate, the main excitatory amino acid in the central
nervous system (CNS), can activate, while serine functions as
co-agonist, N-methyl D-aspartate (NMDA) receptors, which are
cation channels that mediate fast synaptic transmission in the brain
and are important in memory and behavioral functioning. Elevated
concentrations of these amino acids in the synaptic space may
promote selective neuronal loss and various chronic neurological
disorders [90]. The amino acids leucine and isoleucine are also
elevated in depressed compared to nondepressed HF patients.
These branch-chain amino acids have stress response and protein
regulatory functions. Leucine is also relevant to key brain metabolic
functions.
In addition, numerous dicarboxylic acid (DCA) species, which
are produced in the context of fatty acid oxidation dysfunction or
saturation of mitochondrial β-oxidation, are elevated in depressed
compared to control HF patients. Decreased concentrations of the
ketone body 3-hydroxybutyrate, which are used as energy for the
CNS and generated through β-oxidation of fatty acids, are found in
the depressed subjects. During β-oxidation of lipids, the acetyl-
CoA generated may then enter the tricarboxylic acid cycle (TCA);
however excess concentrations of acetyl-CoA will saturate the TCA
cycle and contribute to the formation of ketone bodies such as
3-hydroxybutyrate. Thus, concentrations of 3-hydroxybutyrate
are decreased while DCA species are increased in depressed com-
pared to nondepressed HF patients, which suggests a metabolic
shift away from the primary route of fatty acid metabolism via
β-oxidation toward ω-oxidation of fatty acids in which the
ω-carbon is oxidized to carboxylic acid to form DCA. These results
are consistent with previous associations between altered amino
acid neurotransmitters, fatty acid metabolism, and depression;

however, additional studies are needed to identify the role of HF
pathophysiology on the alterations of these plasma concentrations.
5.8 Behavioral The management of HF relies on a complex regimen of self-care

Pathways practices, and patients who do not adhere may suffer from worse
cardiac prognosis [52]. Self-care maintenance refers to the
decision-making process underlying the performance of healthy
practices and self-care management refers to the behaviors used to
manage signs and symptoms of illness [91]. Self-care maintenance
practices in HF may include adherence to medications and low-salt
diets as prescribed, as well as limiting alcohol consumption and
avoiding tobacco use [91]. The most widely discussed behavioral
pathway in the literature is medication adherence [52]. Cardiac
patients, including those with HF, who have comorbid depression
have three times higher risk of cardiac medication nonadherence
compared to nondepressed patients [53]. Self-care management
behaviors include the recognition of early signs and symptoms of
worsening physical or mental condition, and a subsequent response
to address them [91]. A large multisite trial found that patients with
HF and depression were 45% more likely to delay hospital admis-
sion for more than 72 h as opposed to those HF patients without
depression [92]. Patients with depression tend to wait longer to
consult a health-care provider in response to worsening HF symp-
toms as opposed to HF patients who are not depressed.
6 Conclusions
HF is a major and costly public health concern, and its prognosis is

grim—with high hospitalization and mortality rates. It is well
documented that HF patients experience disproportionately high
rates of depression and that depressed HF patients have worse
clinical outcomes than their nondepressed counterparts. Thus,
understanding mechanisms that link HF and depression has
become a major area of scientific interest. A psychoneuroimmuno-
logical approach to examining these relationships is proving fruitful
and merits increasing attention. The work conducted thus far in our
laboratory group, as well as other groups, suggests that HF and
depression may be linked by increased neuroimmune activation and
possibly gut dysbiosis.
References
1. Shapiro PA (2017) Psychiatric aspects of heart 2. Ghosh RK, Ball S, Prasad V et al (2016)
disease and cardiac aspects of psychiatric disease Depression in heart failure: intricate relation-
in Critical Care. Crit Care Clin 33:619–634 ship, pathophysiology and most updated
evidence of interventions from recent clinical for Heart and Lung Transplantation. J Am
studies. Int J Cardiol 224:170–177 Coll Cardiol 53:e1–e90
3. Kent LK, Shapiro PA (2009) Depression and 16. Inamdar AC (2016) Heart failure: diagnosis,
related psychological factors in heart disease. management and utilization. J Clin Med 5(7)
Harv Rev Psychiatry 17:377–388 17. Ponikowski P et al (2016) 2016 ESC guide-
4. Norra C, Skobel EC, Arndt M et al (2008) lines for the diagnosis and treatment of acute
High impact of depression in heart failure: and chronic heart failure. Eur Heart J
early diagnosis and treatment options. Int J 18:891–975
Cardiol 125:220–231 18. Bristow MR (2000) β-adrenergic receptor
5. Watson K, Summer KM (2009) Depression in blockade in chronic heart failure. Circulation
patients with heart failure: clinical implications 101:558–569
and management. Pharmacotherapy 29:49–63 19. Kessler RC, Berglund P, Demler O et al (2003)
6. Shapiro PA (2009) Treatment of depression in The epidemiology of major depressive disor-
patients with congestive heart failure. Heart der: results from the National Comorbidity
Fail Rev 14:7–12 Survey Replication (NCS-R). JAMA
7. Williams SA, Kasl SV, Heiat A et al (2002) 289:3095–3105
Depression and risk of heart failure among the 20. Rutledge T, Reis VA, Linke SE et al (2006)
elderly: a prospective community-based study. Depression in heart failure: a meta-analytic
Psychosom Med 64:6–12 review of prevalence, intervention effects, and
8. Abramson J, Berger A, Krumholz HM et al associations with clinical outcomes. J Am Coll
(2001) Depression and risk of heart failure Cardiol 48:1527–1537
among older persons with isolated systolic 21. Gottlieb SS, Khatta M, Friedmann E et al
hypertension. Arch Intern Med (2004) The influence of age, gender, and race
161:1725–1730 on the prevalence of depression in heart failure
9. Bui AL, Horwich TB, Fonarow GC (2010) patients. J Am Coll Cardiol 43:1542–1549
Epidemiology and risk profile of heart failure. 22. Havranek EP, Spertus JA, Masoudi FA et al
Nat Rev Cardiol 8:30–41 (2004) Predictors of the onset of depressive
10. Roger VL, Go AS, Lloyd-Jones DM et al symptoms in patients with heart failure. J Am
(2010) Heart disease and stroke statistics— Coll Cardiol 44:2333–2338
2011 update: a report from the American 23. Tousoulis D, Antonopoulos AS, Antoniades C
Heart Association. Circulation 123:18–209 et al (2010) Role of depression in heart fail-
11. Mozaffarian D, Benjamin EJ, Go AS et al ure—choosing the right antidepressive treat-
(2016) Executive summary: heart disease and ment. Int J Cardiol 140:12–18
stroke statistics—2016 update: a report from 24. Fan H, Yu W, Zhang Q et al (2014) Depression
the American Heart Association. Circulation after heart failure and risk of cardiovascular and
133:447–454 all-cause mortality: a meta-analysis. Prev Med
12. Bahrami H, Kronmal R, Bluemke DA et al 63:36–42
(2008) Differences in the incidence of conges- 25. Chaudhry S, McAvay G, Chen S et al (2013)
tive heart failure by ethnicity: the Multi-Ethnic Risk factors for hospital admission among older
Study of Atherosclerosis. Arch Intern Med persons with newly diagnosed heart failure:
168:2138–2145 findings from the cardiovascular health study.
13. Levy D, Kenchaiah S, Larson MG et al (2002) J Am Coll Cardiol 61:635–642
Long-term trends in the incidence of and sur- 26. Vader JM, LaRue SJ, Stevens SR (2016)
vival with heart failure. N Engl J Med Timing and causes of readmission after acute
347:1397–1402 heart failure hospitalization—insights from the
14. Go AS, Mozaffarian D, Roger VL et al (2013) heart failure network trials. J Card Fail
Heart disease and stroke statistics—2013 16:30059–30058
update: a report from the American Heart 27. Freedland KE, Carney RM, Rich MW et al
Association. Circulation 127:e6–e245 (2016) Depression and multiple
15. Hunt SA, Abraham WT, Chin MH et al (2009) re-hospitalizations in patients with heart fail-
2009 Focused update incorporated into the ure. Clin Cardiol 39:257–262
ACC/AHA 2005 guidelines for the diagnosis 28. Krumholz HM, Chaudhry SI, Spertus JA et al
and management of heart failure in adults: a (2015) Do nonclinical factors improve predic-
report of the American College of Cardiology tion of readmission risk? Results from the tele-
Foundation/American Heart Association Task HF study. JACC Heart Fail 15:653–658
Force on Practice Guidelines Developed in 29. Kato N, Kinugawa K, Shiga T et al (2012)
Collaboration with the International Society Depressive symptoms are common and
associated with adverse clinical outcomes in 40. Rollman BL, Herbeck Belnap B, Mazumdar S
heart failure with reduced and preserved ejec- et al (2012) A positive 2-item Patient Health
tion fraction. J Cardiol 60:2330 Questionnaire depression screen among hospi-
30. Sherwood A, Blumenthal JA, Trivedi R et al talized heart failure patients is associated with
(2007) Relationship of depression to death or elevated 12-month mortality. J Card Fail
hospitalization in patients with heart failure. 18:238–245
Arch Intern Med 167:367–373 41. Lee KS, Lennie TA, Heo S, Moser DK (2012)
31. Jiang W, Alexander J, Christopher E et al Association of physical versus affective depres-
(2001) Relationship of depression to increased sive symptoms with cardiac event-free survival
risk of mortality and rehospitalization in in patients with heart failure. Psychosom Med
patients with congestive heart failure. Arch 74:452–458
Intern Med 161:1849–1856 42. Moraska AR, Chamberlain AM, Shah ND et al
32. Jiang W, Kuchibhatla M, Clary GL et al (2007) (2013) Depression, healthcare utilization, and
Relationship between depressive symptoms death in heart failure: a community study. Circ
and long-term mortality in patients with heart Heart Fail 6:387–394
failure. Am Heart J 154:102–108 43. Lee KS, Moser DK, Pelter M et al (2017)
33. Vaccarino V, Kasl SV, Abramson J et al (2001) Two-step screening for depressive symptoms
Depressive symptoms and risk of functional and prediction of mortality in patients with
decline and death in patients with heart failure. heart failure. Am J Crit Care 26:240–247
J Am Coll Cardiol 38:199–205 44. O’Connor CM, Jiang W, Kuchibhatla M et al
34. Frasure-Smith N, Lesperance F, Habra M et al (2010) Safety and efficacy of sertraline for
(2009) Elevated depression symptoms predict depression in patients with heart failure: results
long-term cardiovascular mortality in patients of the SADHART-CHF (sertraline against
with atrial fibrillation and heart failure. Circu- depression and heart disease in chronic heart
lation 120:134–140 failure) trial. J Am Coll Cardiol 56:692–699
35. Sullivan M, Levy WC, Russo JE et al (2004) 45. Jiang W, Krishnan R, Kuchibhatla M et al
Depression and health status in patients with (2011) Characteristics of depression remission
advanced heart failure: a prospective study in and its relation with cardiovascular outcome
tertiary care. J Card Fail 10:390–396 among patients with chronic heart failure
36. Jiang W, Kuchibhatla M, Cuffe MS et al (2004) (from the SADHART-CHF study). Am J Car-
Prognostic value of anxiety and depression in diol 107:545–551
patients with chronic heart failure. Circulation 46. Freedland KE, Carney RM, Rich MW et al
110:3452–3456 (2015) Cognitive behavior therapy for depres-
37. Hallas CN, Wray J, Andreou P, Banner NR sion and self-care in heart failure patients: a
(2011) Depression and perceptions about randomized clinical trial. JAMA Intern Med
heart failure predict quality of life in patients 175:1773–1782
with advanced heart failure. Heart Lung 47. Maes M (2001) Psychological stress and the
40:111–121 inflammatory response system. Clin Sci
38. Gottlieb SS, Kop WJ, Ellis SJ et al (2009) 101:193–194
Relation of depression to severity of illness in 48. Penninx BW, Kritchevsky SB, Yaffe K et al
heart failure (from Heart Failure and a Con- (2003) Inflammatory markers and depressed
trolled Trial Investigating Outcomes of Exer- mood in older persons: results from the health,
cise Training [HF-ACTION]). J Am Coll aging and body composition study. Biol Psy-
Cardiol 103:1285–1289 chiatry 54:566–572
39. Lichtman JH, Bigger JT, Blumenthal JA et al 49. Bremmer MA, Beekman AT, Deeg DJ et al
(2008) Depression and coronary heart disease: (2008) Inflammatory markers in late-life
recommendations for screening, referral, and depression: results from a population-based
treatment: a science advisory from the Ameri- study. J Affect Disord 106:249–255
can Heart Association Prevention Committee 50. Beck A, Steer RA (1987) Manual for the
of the Council on Cardiovascular Nursing, revised Beck Depression Inventory. Psycholog-
Council on Clinical Cardiology, Council on ical Corportation, San Antonio
Epidemiology and Prevention, and Interdisci- 51. Linke SE, Rutledge T, Johnson BD et al
plinary Council on Quality of Care and Out- (2009) Depressive symptom dimensions and
comes Research: endorsed by the American cardiovascular prognosis among women with
Psychiatric Association. Circulation suspected myocardial ischemia: a report from
118:1768–1775 the National Heart, Lung, and Blood Institute-
Sponsored Women’s Ischemia Syndrome Eval- 65. Fairweather D, Frisancho-Kiss S, Yusung SA

uation. Arch Gen Psychiatry 66:499–507 et al (2004) Interferon-γ protects against
52. Wirtz PH, Hong S, Redwine LS et al (2009) chronic viral myocarditis by reducing mast cell
Depressive symptoms are associated with solu- degranulation, fibrosis, and the profibrotic
ble P-selectin reactivity to acute exercise in cytokines transforming growth factor-β1,
heart failure. Biol Psychiatry 65:801–807 interleukin-1β, and Interleukin-4 in the heart.
53. Wirtz PH, Redwine LS, Mills PJ (2010) Am J Pathol 165:1883–1894
Response regarding inflammation as a predic- 66. Afanasyeva M, Georgakopoulos D, Belardi DF
tor of depression in heart failure. Brain Behav et al (2005) Impaired up-regulation of CD25
Immun 24:174–175 on CD4+ T cells in IFN-γ knockout mice is
54. Szabo MC, Butcher EC, McIntyre BW et al associated with progression of myocarditis to
(1997) RANTES stimulation of T lymphocyte heart failure. Proc Natl Acad Sci U S A
adhesion and activation: role for LFA-1 and 102:180–185
ICAM-3. Eur J Immunol 27:1061–1068 67. Redwine LS, Mills PJ, Hong S et al (2007)
55. Chen J, Fujimoto C, Vistica BP et al (2006) Cardiac-related hospitalization and/or death
Active participation of antigen-nonspecific associated with immune dysregulation and
lymphoid cells in immune-mediated inflamma- symptoms of depression in heart failure
tion. J Immunol 177:3362–3368 patients. Psychosom Med 69:23–29
56. Adamopoulos S, Parissis J, Kroupis C et al 68. Wang Z et al (2011) Gut flora metabolism of
(2001) Physical training reduces peripheral phosphatidylcholine promotes cardiovascular
markers of inflammation in patients with disease. Nature 472:57–63
chronic heart failure. Eur Heart J 22:791–797 69. Tang WH et al (2013) Intestinal microbial
57. Pyo RT, Sui J, Dhume A et al (2006) CXCR4 metabolism of phosphatidylcholine and cardio-
modulates contractility in adult cardiac myo- vascular risk. N Engl J Med 368:1575–1158
cytes. J Mol Cell Cardiol 41:834–844 70. Tang WH et al (2014) Prognostic value of
58. Frangogiannis NG, Dewald O, Xia Y et al elevated levels of intestinal microbe-generated
(2007) Critical role of monocyte chemoattrac- metabolite trimethylamine-N-oxide in patients
tant protein-1/CC chemokine ligand 2 in the with heart failure: refining the gut hypothesis. J
pathogenesis of ischemic cardiomyopathy. Cir- Am Coll Cardiol 64:1908–1191
culation 115:584–592 71. Troseid M et al (2015) Microbiota-dependent
59. Linden W, Gerin W, Davidson K (2003) Car- metabolite trimethylamine-N-oxide is asso-
diovascular reactivity: status quo and a research ciated with disease severity and survival of
agenda for the new millennium. Psychosom patients with chronic heart failure. J Intern
Med 65:5–8 Med 277:717–726
60. Redwine LS, Wirtz PH, Hong S et al (2009) 72. Tang WH et al (2015a) Intestinal microbiota-
Potential shift from adaptive immune activity dependent phosphatidylcholine metabolites,
to nonspecific inflammatory activation asso- diastolic dysfunction, and adverse clinical out-
ciated with higher depression symptoms in comes in chronic systolic heart failure. J Card
chronic heart failure patients. J Card Fail Fail 21:91–96
15:607–615 73. Lim GB (2016) Heart failure: gut flora—path-
61. Redwine LS, Wirtz PH, Hong S et al (2010) ogenic role in chronic heart failure. Nat Rev
Depression as a potential modulator of beta- Cardiol 13:61
adrenergic-associated leukocyte mobilization 74. Sandek A et al (2007) Altered intestinal func-
in heart failure patients. J Am Coll Cardiol tion in patients with chronic heart failure. J Am
56:1720–1727 Coll Cardiol 50:1561–1569
62. Redwine LS, Hong S, Rutledge T et al (2014) 75. Tang WH et al (2015b) Gut microbiota-
Leukocyte ß-adrenergic receptor sensitivity and dependent trimethylamine N-oxide (TMAO)
depression severity in patients with heart fail- pathway contributes to both development of
ure. Psychosom Med 76:726–731 renal insufficiency and mortality risk in chronic
63. Bot A, Smith KA, von Herrath M (2004) kidney disease. Circ Res 116:448–455
Molecular and cellular control of T1/T2 76. Pasini E et al (2016) Pathogenic gut flora in
immunity at the interface between antimicro- patients with chronic heart failure. JACC Heart
bial defense and immune pathology. DNA Cell Fail 4:220–227
Biol 23:341–350 77. Cani PD et al (2007) Metabolic endotoxemia
64. Elenkov IJ (2004) Glucocorticoids and the initiates obesity and insulin resistance. Diabetes
Th1/Th2 balance. Ann N Y Acad Sci 56:1761–1772
1024:138–146
78. Qin J et al (2012) A metagenome-wide associ- 86. Sandek A et al (2014) Intestinal blood flow in
ation study of gut microbiota in type 2 diabetes. patients with chronic heart failure: a link with
Nature 490:55–60 bacterial growth, gastrointestinal symptoms,
79. Cani PD et al (2009) Changes in gut micro- and cachexia. J Am Coll Cardiol
biota control inflammation in obese mice 64:1092–1102
through a mechanism involving GLP-2-driven 87. Levine B, Kalman J, Mayer L et al (1990)
improvement of gut permeability. Gut Elevated circulating levels of tumor necrosis
58:1091–1103 factor in severe chronic heart failure. N Engl J
80. Charalambous BM, Stephens RC, Feavers IM Med 323:236–241
et al (2007) Role of bacterial endotoxin in 88. Maeda K et al (2000) High levels of plasma
chronic heart failure: the gut of the matter. brain natriuretic peptide and interleukin-6
Shock 28L:15–23 after optimized treatment for heart failure are
81. Sandek A, Rauchhaus M, Anker SD et al independent risk factors for morbidity and
(2008) The emerging role of the gut in chronic mortality in patients with congestive heart fail-
heart failure. Curr Opin Clin Nutr Metab Care ure. J Am Coll Cardiol 36:1587–1593
11:632–639 89. Steffens DC et al (2010) Metabolomic differ-
82. Sandek A et al (2012) Studies on bacterial ences in heart failure patients with and without
endotoxin and intestinal absorption function major depression. J Geriatr Psychiatry Neurol
in patients with chronic heart failure. Int J 23:138–146
Cardiol 157:80–85 90. Meldrum BS (1986) Drugs acting on amino
83. Nagatomo Y, Tang WH (2015) Intersections acid neurotransmitters. Adv Neurol
between microbiome and heart failure: revisit- 43:687–706
ing the gut hypothesis. J Card Fail 21:973–980 91. Riegel B, Lee CS, Dickson VV (2011) Self care
84. Sandek A, Anker SD, von Haehling S (2009) in patients with chronic heart failure. Nat Rev
The gut and intestinal bacteria in chronic heart Cardiol 8:644–654
failure. Curr Drug Metab 10:22–28 92. Jaarsma T et al (2004) Design and methodol-
85. Niebauer J et al (1999) Endotoxin and ogy of the COACH study: a multicenter ran-
immune activation in chronic heart failure: a domised coordinating study evaluating
prospective cohort study. Lancet outcomes of advising and counselling in heart
353:1838–1842 failure. Eur J Heart Fail 6:227–233
Chapter 10
How to Monitor the Neuroimmune Biological Response

in Patients Affected by Immune Alteration-Related
Systemic Diseases
Paolo Lissoni, Franco Rovelli, Luigi Vigorè, Giusy Messina,
Arianna Lissoni, Giorgio Porro, and Giuseppe Di Fede
Abstract
The clinical management of patients affected by systemic diseases, including cancer and autoimmune
diseases, is generally founded on the evaluation of the only markers related to the single disease rather
than the biological immuno-inflammatory response of patients, despite the fundamental role of cytokine
network in the pathogenesis of cancer and autoimmunity is well known. Cancer progression has appeared to
be associated with a progressive decline in the blood levels of the main antitumor cytokines, including IL-2
and IL-12, in association with an increase in those of inflammatory cytokines, including IL-6, TNF-alpha,
and IL-1-beta, and immunosuppressive cytokines, namely TGF-beta and IL-10. On the other hand, the
severity of the autoimmune diseases has been proven to be greater in the presence of high blood levels of
IL-17, TNF-alpha, IL-6, IL-1-beta, IFN-gamma, and IL-18, in association with low levels of TGF-beta and
IL-10. However, because of excessive cost and complexity of analyzing the data regarding the secretion of
the single cytokines, the relation between lymphocyte-induced immune activation and monocyte-
macrophage-mediated immunosuppression has been recently proven to be expressed by the simple
lymphocyte-to-monocyte ratio (LMR). The evidence of low LMR values has appeared to correlate with a
poor prognosis in cancer and with a disease control in the autoimmune diseases. Moreover, since the in vivo
immunoinflammatory response is physiologically under a neuroendocrine modulation, for the evaluation of
patient biological response it would be necessary to investigate the function of at least the two main
neuroendocrine structures involved in the neuroendocrine modulation of the immune responses, consist-
ing of the hypothalamic-pituitary-adrenal axis and the pineal gland, since the lack of physiological circadian
rhythm of cortisol and pineal hormone melatonin has appeared to be associated with a worse prognosis in
the human systemic diseases.
Key words Autoimmunity, Biological response, Cancer, Cannabinoid system, Cytokine network,
Immunotherapy, Lymphocyte-to-monocyte ratio, Melatonin, Neuroimmunomodulation, Opioid sys-
tem, Pineal gland, Synchronization
1 Introduction to Psycho-Neuro-Endocrino-Immunology
Systemic human diseases may be defined as those pathologies,

which potentially may affect the whole living organism, being due
171
172 Paolo Lissoni et al.
to alterations of the functionless of the immune system, which

influences not only the immune responses, but also the overall
biological functions, including nervous, endocrine, and cardiovas-
cular systems [1]. The so-called human systemic disease, including
cancer and autoimmune diseases, has appeared to be due to a
profound alteration in the functionless of the immune system,
unable to destroy the malignant cells in the neoplastic diseases
and paradoxically activated against self-antigens in the autoimmune
pathologies. After several years of investigations, at present it is
possible to affirm that cancer- and autoimmunity-related immune
alterations mainly depend on an abnormally high activation in
cancer and on an abnormally low activity in the autoimmunity of
regulatory T lymphocytes (T reg), which constitute the main cells
responsible for the immunosuppression of the immune functionless
[2]. However, the recent advances in the psycho-neuro-endocrino-
immunology (PNEI) knowledge have demonstrated that the
in vivo immune responses do not depend only on the activity of
immune cells themselves, but also on the modulation of the
immune system, which is determined by the same molecules pro-
duced by the immune cells, including cytokines and chemokines,
but also by hormones, neurohormones, and neurotransmitters
[3]. Then, the cytokine network cannot be separated from its
psychoneuroendocrine regulation, which constitutes the biochem-
ical mediation of the psychospiritual life. In any case, despite the
great complexity apparently beyond the capacity of human mind
itself, the neuroimmunomodulation (NIM) of the immune system
may be synthetized into the identification of two main major fun-
damental neuroendocrine functional axes, consisting of brain opi-
oid system-hypothalamic-pituitary-adrenal axis [4] and brain
cannabinergic system-pineal-thymus axis [5, 6]. Each neurotrans-
mitter, neuromodulator, and neurohormone may potentially influ-
ence the immune responses, but from a synthetic point of view the
NIM is essentially realized by two major brain neuromodulatory
systems, consisting of brain opioid system and brain cannabinergic
system, which are both modulated by the pineal gland through the
release of its main hormone, melatonin (MLT) [7]. At the other
side, the great complexity of the immune interactions may be
synthetized into two essential immune functions, consisting of
the immuno-activation, mainly mediated by the functional axis
constituted by dendritic cells through the release of IL-12 [8];
T-helper-1 lymphocytes (TH1) through the release of IL-2,
which constitutes the main T-cell growth factor [9]; and cytotoxic
T lymphocytes, and at the other side the immunosuppression of the
immune responses, which is substantially mediated by the mono-
cyte-macrophage system and T reg cells [2], which block the
immune responses, but whose generation is induced by
macrophage-related chronic inflammatory response itself [10],
namely through the action of transforming growth factor-beta
How to Monitor the Neuroimmune Biological Response in Patients Affected by. . . 173
(TGF-beta) [11], which is produced by both macrophage precur-

sors to stimulate T reg cell generation and T reg cells to suppress the
immune response, including the anticancer reaction. The func-
tional neuroendocrine axis constituted by brain opioid system-
pituitary-adrenal gland mainly promotes the suppression of the
immune responses by either stimulating the generation of T reg
lymphocytes through a direct action on mu-opioid receptors
expressed by T cells themselves [12] or inducing lymphocytolysis
due to an enhanced cortisol production. On the other hand, the
brain cannabinergic system-pineal-thymus functional neuroendo-
crine axis activates the immune system by a direct stimulation of
IL-12 production from dendritic cells and IL-2 secretion from
TH1 lymphocytes induced by the pineal hormone MLT [6] by
acting on specific MLT receptors expressed by both immune cells
as well as by counteracting the macrophage-mediated immunosup-
pression related to the chronic inflammatory response. Brain can-
nabinergic system would promote the immune response through
its endogenous cannabinoids by either stimulating the endocrine
activity of the pineal gland [5] or inhibiting the macrophage-related
inflammatory response, which suppresses TH1 cell functions [10],
and by reducing IL-17 secretion [13]. From a molecular point of
view, the suppression of T-lymphocyte functions, namely those
realized by TH1 lymphocytes, has appeared to be mediated by the
expression of specific cell surface receptors, which are able to block
the immune responses, including the anticancer immunity, the
so-called immune checkpoints, the most active of them being
represented by the programmed death-1 (PD-1), with its ligands
PD-L1 and PD-L2 [14, 15]. In the absence of PD-1 expression, T
lymphocytes may exert their effector activities, including the TH-1-
related activation of the whole immune system, and the cytotoxic
activity against infected or transformed cells. In contrast, the
expression of PD-1 or its ligands induces the evolution of T-cell
effector functions in T reg cell ones, with a consequent suppression
of the immune responses against both tumor cells in the neoplastic
diseases and self-antigens in the autoimmune pathologies.
1.1 The Biological It is a general common opinion that the biological response is
Response consisting only of the acute and chronic immuno-inflammatory
reactions. On the contrary, since both the immune and the inflam-
matory responses are under a physiological psychoneuroendocrine
regulatory control, the investigation of host biological response in
the human systemic diseases would have to include at least the
evaluation of the immune functionless, the degree of the inflamma-
tory response, and the functional status of the main nervous and
endocrine structures involved in the neuroendocrine modulation
on the biological immuno-inflammatory responses, which have
appeared to be represented by the hypothalamic-pituitary-adrenal
(HPA) axis and by the pineal gland for the endocrine system and by
brain opioid and cannabinergic systems for the nervous control of

the immune system. Moreover, since the inflammatory response
depends on the different cytokines produced by the various acti-
vated immune cells, the same inflammatory response could be
simply considered as one of the numerous effects induced by the
activation of the immune system, which include the defense against
pathogenic agents, modulation of the angiogenic processes, and
stimulation of tissue wound repair. Each human systemic disease
may be clinically investigated and monitored by evaluating its spe-
cific markers, consisting of tumor markers for the neoplastic dis-
eases and type of involved autoantibody in the autoimmunity, or in
a new manner by concomitantly evaluating the biological response
of patients according to the recent advances in the knowledge of the
cytokine network.
1.2 Clinical By investigating the complexity of the biological response, it has to

Investigation of the be taken into consideration that the first biological response of
Synchronization living organisms, including humans, is represented by the ability
Status to regulate the different biological functions in relation to the
universal light conditions, with the consequent generation of circa-
dian rhythms in the activity of the various biological systems. Not
only this, but also the progressive decline in the ability to respond
to the circadian rhythms represents one of the most evident signs of
aging, which affect both the endocrine and the immune systems
[16]. The loss of the circadianicity in the endocrine functions may
also represent one of the first signs of severe human diseases, such as
depression and cancer. Despite the complexity of the endocrine and
immune circadian rhythms, it is commonly accepted that the simple
circadian rhythm of cortisol may be considered as synthetic of the
functional status of the overall main biological rhythms [17]. The
presence of a physiological circadian rhythm of cortisol with high
levels in the morning and decline during the afternoon is com-
monly defined as synchronization status [17]. Then, the loss of
cortisol rhythm is not only a marker of an endocrine dysfunction,
but it would also represent a sign of a more general alteration of the
neuroendocrine control of the biological rhythms, and this evi-
dence has been proven to be associated with a poor prognosis and
with a lower survival in most metastatic tumor patients [18, 19], as
well as in the psychiatric disturbances, by representing the end
result of a more general brain neurochemical disorder. The other
important biomarker of the synchronization status is consisting of
the presence of a physiological light/dark circadian rhythm of the
pineal indole MLT, normally produced mainly during the dark
period of the day [7]. As well as for cortisol rhythm, the progressive
loss of MLT light/dark rhythm may represent one of the main
biological markers of severe human diseases, including cancer and
depression [7, 20]. Moreover, the loss of MLT rhythm is generally
associated with a concomitant lack of cortisol rhythm, being the
HPA axis at least in part under a direct pineal modulation, as

suggested by the possibility to restore a normal cortisol rhythm
by an exogenous chronic circadian administration of MLT during
the dark period of the day [21].
1.3 Clinical The inflammatory response is the most ancient manner of living
Investigation organisms to biologically respond. All inflammatory response-
of the Inflammatory related subjective symptoms are the end result of several and com-
Response plex cytokine interactions. Fever itself has been proven to be due to
the action of inflammatory cytokines, including IL-1-beta, IL-6,
TNF-alpha, IL-2, and IL-12, at hypothalamic sites, while other
cytokines also provided by some inflammatory effects, such as
IL-18, do not induce fever. The inflammatory response may be
clinically identified and evaluated by determining several biomar-
kers; most of them are related to the macrophage response, includ-
ing ESR, CRP, neopterin, and soluble IL-2 receptor (sIL-2R),
which express the same biological significance, since their physio-
pathological and prognostic significance is similar, because of the
evidence that abnormally high values of one or more inflammatory
markers have been proven to predict a poor prognosis and a lack of
response to the various anticancer therapies in patients with
advanced neoplasms [22]. From a physiopathological point of
view, the inflammatory response may be generated from four essen-
tial origins, consisting of (1) the macrophage system through the
production of several inflammatory cytokines, namely IL-1-beta,
IL-6, and TNF-alpha [22, 23]; (2) the TH-1 lymphocytes through
the secretion of IL-2, which induces a general activation and prolif-
eration of lymphocytes [9]; (3) the TH-17 lymphocytes by secret-
ing IL-17, which plays a fundamental role in the induction of the
acute inflammatory response by either stimulating IL-2 production
from TH1-lymphocytes, or inhibiting the immunosuppressive
activity of T reg lymphocytes [24]; and (4) the endothelial cells,
in particular by the secretion of IL-18 [25], which also exerts
inflammatory effects by stimulating interferon (IFN)-gamma secre-
tion. Macrophage production of IL-6 is under a stimulatory role
played by the same IL-1-beta released from macrophages them-
selves. Moreover, IL-6 is the major stimulator of the hepatic pro-
duction of acute-phase inflammatory proteins. In any case, the
inflammatory response depends not only on the degree of the
same inflammatory response, but also on the antagonistic action
played by the anti-inflammatory cytokines. Then, according to their
major effect of the inflammatory response, the overall cytokines
may be subdivided into three major classes, consisting of (1) -
pro-inflammatory cytokines IL-1-beta, IL-6, IL-4, IL-5, IL-8,
IL-13, IL-17, and IL-18; (2) anti-inflammatory cytokines: IL-10
and TGF-beta; (3) modulatory cytokines with both pro- and anti-
inflammatory effects: IL-2 and IL-12. IL-18, which is mainly pro-
duced by endothelial cells, macrophages, and dendritic cells, may
also have both inflammatory and anti-inflammatory effects in some

conditions. The main anti-inflammatory effects of IL-18 are con-
sisting of inhibition of IL-17 secretion and stimulation of T reg cell
functions, with a following suppression of the inflammatory
response [25]. On the other side, IL-18 may also display inflamma-
tory effects due to stimulation of IFN-gamma production by TH1
cells, stimulation of TH2-differentiation, and stimulation of IgE
production [25]. Moreover, IL-18 could be the main cytokine
involved in the occurrence of macrophage-activating syndrome
(MAS), also called hemophagocytic syndrome, which would be
due to a deficiency in IL-18-binding protein, with a consequent
inappropriate action of IL-18 itself with its detrimental effects,
induced by a concomitant exaggerated activation of both TH lym-
phocyte and macrophage systems. One of the main controversial
cytokine interactions is that concerning the reciprocal effects
among TGF-beta, mainly produced by T reg cells, IL-17 released
from TH17 lymphocytes, and IL-12 secreted by antigen-activated
dendritic cells [8, 11]. TGF-beta would inhibit IL-17 production,
whereas in association with IL-6 and IL-23 stimulates the genera-
tion and maintenance of TH17 lymphocytes [24]. IL-12 may
inhibit T reg cell-induced TGF-beta production, and it may also
suppress IL-17 secretion in the presence of IL-21 [26]. Another
controversial cytokine in relation to the interactions between effects
on the inflammatory response and antitumor immunity is IL-10
[27]. IL-10, as well as TGB-beta, may stimulate T reg cell activa-
tion, and then it may suppress the antitumor immunity. However, it
has been recently demonstrated that IL-10 may also play some
anticancer effects, namely consisting of the induction of specific
cytotoxic T lymphocyte-cell memory. The anti-inflammatory action
of IL-10 could also contribute to its potential anticancer activity,
since the inflammatory response contributes to cancer progression
[10]. The cytokine network involved in the immuno-inflammatory
response is illustrated in Fig. 1.
1.4 Clinical The immune functionless may be evaluated by determining the

Investigation count of the single-immune cells with their various subsets and
of Immune System the status of their activation by detecting the blood concentrations
Functionless of the main cytokines. Lymphocyte subpopulations may be routi-
narely measured by monoclonal antibody against specific clusters of
differentiation (CD). At present, most lymphocyte subsets, whose
morphology is similar, may be identified through anti-CD mono-
clonal antibodies, except for TH1 and TH2 lymphocytes, which are
responsible for the activation of a preferential cytotoxic cellular or
humoral antibody-mediated immune response, respectively, and
which may be identified only on the basis of the type of secreted
cytokines, consisting of IL-2 and IFN-gamma for TH1 and IL-4,
IL-5, IL-10, and IL-13 for TH2 cells. In any case, by considering
that the single immune response is the result of two opposite
Fig. 1 The cytokine network and its relation to the main immune cells
dynamics, consisting of activation or suppression of the immune

reactivity, from a clinical point of view it would be sufficient to
detect the blood levels of the main immunosuppressive (IL-6,
IL-10, TGF-beta) and antitumor immunostimulatory cytokines
(IL-2, IL-12) in the neoplastic diseases, and the main inflammatory
(IL-6, IL-1-beta, IL-17, IL-18, TNF-alpha) and anti-inflammatory
cytokines (IL-10, TGF-beta) in the autoimmune diseases. Obvi-
ously, from a clinical point of view, to monitor the clinical course of
the systemic diseases, it is also necessary to evaluate the blood levels
not only of the main cytokines, but also of the main endogenous
inhibitors of the action of cytokines, which are represented by
soluble TNF-alpha receptor 1 and 2 (sTNF-R1, sTNF-R2) for
TNF-alpha, soluble IL-1-beta receptor 1 and 2 (sIL-1-R1, sIL-1-
R2) for IL-1-beta, and IL-18-binding protein (IL-18BP) for
IL-18 [25].
2 Clinical Investigations of PNEI Condition of Patients
The complexity of the neuroendocrinoimmune interactions would

require a great number of laboratory analyses to clinically explore
not only the immune performance of human subjects, but also the
functional links between neuroendocrine and immune systems.
Therefore, to explore the PNEI status of patients the detection of
at least the main lymphocyte subsets and the main immunostimu-
lating cytokines including IL-2 and IL-12, as well as the most
suppressive ones, such as TGF-beta, IL-6, and IL-10 for evaluating
the immune status would be required, and on the other side the
investigation of the two mean modulatory neuroendocrine func-
tional units by determining the circadian rhythms of cortisol and
MLT, which would represent the main immunosuppressive and
immunostimulatory hormones, respectively [4, 6]. In the reality,
the whole functionless of the immune system is simply consisting of
the balance between activation and suppressive events. Lymphocyte
count reflects the immune activation, whereas the immunosuppres-
sion is related to an enhanced function of the monocyte-macro-
phage system, since T reg cell generation itself is induced by the
macrophage-related chronic inflammatory response [10]. Several
biomarkers have been identified to put into evidence an enhanced
macrophage system activation, including CRP, IL-6, neopterin,
and soluble IL-2 receptor [22, 23]. However, recent experimental
and clinical observations have demonstrated that lymphocyte sub-
set detection may be simply synthetized by the TH-to-T reg lym-
phocyte ratio [28], since TH and T reg cells reflect the status of
immunoactivation or immunosuppression, respectively, as well as
more surprisingly that the simple monocyte count would reflect the
chronic inflammatory and immunosuppressive status [29], being
related to T reg cell count [10], and in the case of the neoplastic
disease it has been proven to correlate with the degree of tumor
macrophage infiltration, which stimulates tumor cell growth and
dissemination [10]. The same biomarkers of the macrophage-
mediated inflammatory response are in relation to the simple
monocyte count. In the same way, lymphocyte count has appeared
to be positively correlated with IL-2 and IL-12 concentrations
[23]. Then, at least from a routinary clinical point of view, the
PNEI status of patients may be synthetically investigated by the
simple lymphocyte-to-monocyte ratio (LMR) [30] for the immune
functionless and the evaluation of cortisol and MLT circadian

rhythms in regard to the analysis of the psychoneuroendocrine
regulation of the immune system.
2.1 Clinical The clinical evaluation of the activity of brain opioid system may be
Investigation of Opioid performed by detecting both blood and liquoral concentrations of
System Functionless the main endogenous opioids, including beta-endorphin, enkepha-
lins, and dynorphins, which act as mu-, delta-, and kappa-opioid
agonists, respectively, or in a more clinically simple manner by
analyzing the endocrine response of cortisol and LH to the oral
administration of a mu-opioid antagonist, such as naltrexone
(NTX) [31]. In normal conditions, NTX at a dose of 50 mg induces
a peak in LH and cortisol secretions, whereas the lack of LH and
cortisol response to NTX would indicate the existence of an
enhanced brain opioid system activity. On the contrary, the evi-
dence of an endocrine response to low-dose NTX would be the
expression of a diminished brain opioid system activity. The most
evident immunomodulatory effects of mu-opioid agonists, such as
beta-endorphin and morphin, consist of the induction of an
immunosuppressive status due to the inhibition of TH1 and den-
dritic cell functions, with a consequent decline in IL-2 and IL-12
production, respectively, in association with a stimulatory effect on
T reg lymphocytes, TH2 cells, and macrophages, with a following
increase in IL-10 and TGF-beta secretion [12, 34]. Then, because
of the immunosuppressive activity of the mu-opioid system, its
block by the mu-opioid antagonist NTX may improve the immune
functionless [32]. In contrast, the immunomodulating effects of
delta- and kappa-opioids are still controversial, being dependent on
dosage and experimental conditions, since both immune stimula-
tory and inhibitory effects have been described [12].
2.2 Clinical The endogenous cannabinoid system, which operates as an inter-

Investigation neuronal neuromodulatory system, produces two main cannabi-
of the Endogenous noids, consisting of the arachidonyl-ethanol-amide (AEA), also
Cannabinergic System called anandamide, and 2-arachidonyl-glycerol (2-AG) [13]. The
main immunomodulatory effects of cannabinoids are represented
by an inhibitory action on macrophage- and Th17 lymphocyte-
mediated inflammatory events, while their action on TH1 and T
reg lymphocytes is still controversial, and in particular it may be
different in vivo and in vitro, because of the in vivo existence of
functional connections between cannabinergic system and pineal
gland [5, 13], whose fundamental immunostimulatory role has
been well demonstrated [6, 7]. The functional status of brain
cannabinergic system may be clinically explored by detecting the
only blood levels of the two main endogenous cannabinoids, AEA
and 2-AG, since liquoral concentrations are positively correlated
with the blood ones [13]. Alternatively and in a more simple way,
brain cannabinoid activity may be analyzed by measuring the blood
levels of the main enzyme involved in cannabinoid degradation, the

fatty acid amide hydrolase (FAAH) [13]. The evidence of abnor-
mally high levels of FAAH is associated with low concentrations of
cannabinoids as a consequence of their enhanced degradation, and
then it is an expression of a decreased brain cannabinoid activity. On
the contrary, low levels of FAAH may be considered as a sign of an
enhanced brain cannabinoid activity.
2.3 Clinical The pineal gland may be clinically investigated in its endocrine
Investigation function by evaluating the light/dark circadian rhythm of its most
of the Pineal Gland investigated hormone, MLT, even though it is known that the
pineal may produce several other hormones; the most important
of them are represented by the 5-methoxytryptamine (5-MTT) and
the beta-carboline 6-methoxy-1,2,3,4-tetra-hydro-beta-carboline,
also called pinoline or pinealine [33], which are both provided by
anticancer and antidepressant activity. MLT circadian secretion may
be explored by measuring MLT blood levels at least at the four
main phases of the light/dark period, including morning, noon,
afternoon, and night, or in a more clinically simple way by evaluat-
ing the light and dark urinary excretion of the main MLT metabo-
lite, the 6-sulfatoxy-melatonin (6-MTS) [7]. The MLT rhythm may
be considered as normal when night blood levels of MLT or night
urinary excretion of 6-MTS are at least two times greater with
respect to those found during the light phase of the day.
2.4 Clinical Both cytokine and endocrine secretions are under a control realized
Investigation of Neuro- by several feedback mechanisms. The endocrine secretions are
Endocrino-Immune mainly regulated by negative feedback mechanisms, whereas the
Interactions secretion of cytokines is preferentially under both negative and
positive feedback circuits. Moreover, there are feedback circuits
operating between endocrine and immune systems. The existence
of cytokine-endocrine feedback mechanisms would suggest that
some endocrine secretions are not only under a control played by
other hormones, but also under a regulation exerted by cytokines,
and in the same way some cytokine secretions are controlled by
hormones and neuroactive molecules. Then, according to Descar-
tes’s method suggesting the need to reduce the complexity of a
phenomenon to only some essential evidences for its interpretation,
within the great number of simultaneous interactions between
endocrine and immunocytokine secretions, it is necessary to iden-
tify the fundamental mechanisms responsible for the regulation of
the link between endocrine and immune system, and at present it is
possible to identity four main cytokine-endocrine circuits [34]:
(1) inflammatory cytokine-HPA axis: most inflammatory cytokines,
including IL-1-beta, IL-6, TNF-alpha, IFN, as well as IL-2 and
IL-12, stimulated cortisol secretion by acting on both
hypothalamic-pituitary sites and directly at adrenal level; (2) inflam-
matory cytokine-pineal gland: most inflammatory cytokines,
namely IL-1-beta, IL-6, IL-12, TNF-alpha, and IL-2, may inhibit

MLT secretion from the pineal gland, which in contrast stimulates
the secretion of both IL-2 and IL-12, whereas it inhibits that of
IL-6, by realizing a well-defined cytokine-pineal feedback circuit
[35]. The inhibitory effect of MLT on cancer- and AIDS-related
cachexia would be mainly due to the inhibitory effect of MLT on
TNF-alpha production [7, 35]. The effects of cytokines on the
pineal, however, would depend also on their dosage, since
low-dose IL-2 or low-dose TNF-alpha have been proven in vivo
to stimulate MLT secretion; (3) IL-12–IL-10-cannabinoid system
circuit: IL-12 inhibits FAAH activity, with a following increase in
cannabinoid concentrations, whereas IL-10 stimulates FAAH activ-
ity, with a consequent decrease in cannabinoid endogenous levels
[36], by suggesting that IL-12-induced immunoactivation and
IL-10-induced immunosuppression are associated with a concomi-
tant increased or decreased activity of the endogenous cannabinoid
system, respectively; (4) leptin-cytokine network circuit [37]: leptin
(LPT) is a protein produced by adipocytes, which inhibits the
appetite by acting at hypothalamic sites through a stimulation of
NPY release and FAAH activity, with a following decline in brain
cannabinoid system activity, which in contrast amplifies the percep-
tion of the appetite [13]. However, LPT has been proven to exert
several other biological effects, including a promoting action on
cancer development through several mechanisms, consisting of
stimulation of cancer cell proliferation, angiogenic activity, and
stimulatory effect on the secretion of immunosuppressive-
inflammatory cytokines, such as IL-6. Normally, food intake stimu-
lates and starvation inhibits LPT secretion. LPT secretion is also
stimulated by the inflammatory cytokines, and this finding could
explain the reduced appetite occurring in the chronic inflammatory
diseases. In contrast, LPT levels are high in the obesity because of a
possible reduced sensitivity to LPT itself. Obesity-related enhanced
risk of cancer development would depend at least in part on the
pro-tumoral activity played by LPT itself, whose secretion is clearly
enhanced in the obese people. From a clinical investigation point of
view, however, to explore the cytokine-endocrine system interac-
tions it may be sufficient to evaluate the endocrine response to
subcutaneous low-dose IL-2 [36]. In normal condition, IL-2
induces a cortisol increase within some hours after its injection.
The evidence of an absent or a reduced response of cortisol to IL-2
injection may be considered as a sign of a reduced sensitivity of
cytokine-HPA axis, which may predispose to the onset of autoim-
mune reactions because of lack of the inhibitory action of cortisol
on T lymphocytes activated in a nonspecific manner by cytokines
released by antigen-activated T lymphocytes, including possible
autoreactive T lymphocytes [25, 35]. Figure 1 synthetizes the
regulation of the cytokine network into seven main immune cells,
including macrophages, dendritic cells, TH1 lymphocytes, cyto-

toxic T lymphocytes, T reg lymphocytes, NK-LAK cell system,
and TH17-lymphocytes.
3 Biomarkers and Physiopathology of Human Systemic Diseases
Obviously, the clinical usefulness of biological markers depends on

their importance in reflecting the physiopathological condition and
in influencing the prognosis of the various diseases. Then, the
identification of a clinical biomarker requires a well-defined knowl-
edge of the pathogenesis of each single disease. Cytokines are the
equivalent for the immune system than the hormones for the
endocrine system, and then today each pathogenesis of disease,
that excludes the investigation of the cytokine network, has to be
considered as partial and not defined. According to the data avail-
able up to now, the physiopathology of cytokine network has been
more investigated and better defined for the neoplastic diseases
than for the autoimmune ones, since most autoimmune diseases
are characterized by similar cytokine alterations, and at present it is
not possible to correlate the single autoimmune disease with a
specific cytokine alteration. In any case, before investigating the
cytokine network and the neuroimmune interactions in the sys-
temic diseases, including cancer and autoimmunity, it may be
important to analyze the neuroimmune condition of two of the
main events, which predispose to the human systemic diseases,
consisting of stress and aging itself.
3.1 Cytokine Despite the controversial results, the major evidences have demon-
Network of Aging strated that aging is characterized by a progressive decline in IL-2
and an increase in IL-17 concentrations [38]. These two events
would be connected between them, since IL-2 decline induces an
increase in IL-17 levels because of the inhibitory effect of IL-2 on
IL-17 secretion. Age-related decline in IL-2 production would be
at least in part the consequence of the concomitant age-related
decrease in the pineal function, because of the stimulatory effect
of MLT on IL-2 secretion by TH1 cells [6, 7, 35]. The decrease
with age in IL-2 levels and the increase in those of IL-17 are already
sufficient to explain aging-related enhanced frequency of both
neoplastic and autoimmune diseases. On the contrary, as far as T
reg cell behavior with age is concerned, both high and low T reg
cell percentages have been reported in aged subjects [38, 39]. Since
low or high T reg cell percentage may predispose to the autoim-
mune diseases or to the neoplastic diseases, respectively, the differ-
ent behavior of T reg cell population in aged subjects may influence
the type of disease risk for each single subject.
3.2 Cytokine The stress is one of the main causes, which predispose to the main
Network human diseases, including cancer and autoimmunity. The neuroim-
and Endocrine mune effects induced by stress and the different effects of the
Interactions of Stress different types of stress need to be further investigated and defined.
However, according to the knowledge available up to now, stress
induces three main neuroimmune effects, as follows [12, 40, 41]:
(1) activation of HPA axis, with a consequent enhanced production
of cortisol; (2) activation of brain opioid system; and (3) activation
of the sympathetic system, mainly induced by the hypothalamic
release of CRH. Both cortisol and catecholamines may induce
lymphocytolysis, inhibition of IL-2 and IL-12 production, and
stimulation of IL-10 secretion. Catecholamines inhibit lymphocyte
functions by acting on the beta-adrenergic receptor expressed by
the immune cells. Then, catecholamine-induced immunosuppres-
sion is a beta-adrenergic-mediated phenomenon, which may be
potentially abrogated by the administration of beta-adrenergic
antagonists. Mu-opioids would act by stimulating T reg cell and
by inhibiting TH-1-dependent functions, with a following increase
in IL-10 and decline in IL-2 and IL-12 levels. The stress has been
proven to predispose to both cancer and autoimmune diseases.
Then, the main question is to establish through which mechanisms
the same condition of stress may predispose to two different and
opposite immunobiological responses, as well as those involved in
cancer and in autoimmune diseases. To better understand stress-
related neuroimmune events, firstly it is necessary to evaluate the
functional status of the two major brain structures involved in the
psychoneuromodulation of the immune responses, consisting of
brain opioid and cannabinergic systems. Stress-induced enhanced
risk for cancer development has appeared to be abrogated by the
administration of mu-opioid antagonists, such as naltrexone, by
suggesting an opioid mediation of stress-related promoting effect
on tumor development, as well as the existence of an increased
brain opioid tone during cancer progression [40]. In contrast,
stress conditions predisposing to the onset of autoimmune diseases
would be characterized by a diminished brain opioid tone, with a
consequent reduced opioid-mediated immunosuppression to
counteract the possibility of an exaggerated immune response dur-
ing the activation of the immune system, such as that occurring
during the infective processes [35]. On the other hand, brain
cannabinergic system activity, which mediates both the perception
of pleasure and the spiritual sensitivity, would be reduced in both
cancer and autoimmune diseases, with a consequent potential
enhanced endogenous production of IL-17, its secretion being
physiologically under an inhibitory control played by the endoge-
nous cannabinoids [13]. IL-17-enhanced endogenous production
may predispose to both autoimmunity and cancer, because of its
fundamental role in inducing autoimmunity-related inflammatory
response and its promoting effect on cancer cell proliferation [24].
3.3 Cytokine The great variety of cancer-related immune alterations may be

Network synthetized into the presence of a diminished activity of TH1
in the Neoplastic lymphocytes and dendritic cells, with a following progressive
Disease decline in IL-2 and IL-12 production, in association with an
enhanced activity of T reg cells and macrophage system, with a
consequent increase in the production of the immunosuppressive
cytokines TGF-beta and IL-10 [35]. Then, the great number of
cancer-related lymphocyte subset alterations may be synthetized by
the evidence of a progressive decline in T helper-to-T reg cell ratio
[28]. In any case, the main negative prognostic biomarker of the
disseminated neoplastic diseases is represented by the evidence of
lymphocytopenia itself, and in a more complete manner by the
evidence of a progressive decline in LMR values [30]. Moreover,
in addition to the importance of considering cancer-related cyto-
kine secretion, it has to be also taken into consideration the fact that
several years ago the first biological sign of the malignant transfor-
mation was considered to be the loss of cell-cell contact inhibition.
Since the inhibition of cell proliferation induced by cell-cell contact
would involve the intercellular connections, it is probable that the
intercellular junctions may play an essential role in cell contact
inhibition. In fact, the alterations of intercellular junctions have
been proven to play a fundamental role in determining intercellular
matrix change-related tumor neo-angiogenesis, which allows
tumor tissue infiltration and metastatic dissemination. Then, cyto-
kines may influence cancer development also by controlling inter-
cellular junction structures.
3.4 Cytokine It is known that autoimmune diseases are due to a severe dysregula-
Network tion of the cytokine network, consisting of an exaggerated activa-
of the Autoimmunity tion of the inflammatory response. However, because of the great
number of in vivo cytokine interactions, it is difficult to establish the
role played by the single cytokine in relation to the different auto-
immune diseases [42–52]. IL-17 would play a major role in the
maintenance of autoimmunity-related inflammatory response, and
the severity of disease would be enhanced by a concomitant
increased secretion of IL-18. In any case, most systemic autoim-
mune diseases, including rheumatoid arthritis (RA), systemic lupus
erythematosus (SLE), inflammatory bowel disease, multiple sclero-
sis, psoriasis, and Sjogren’s syndrome, are characterized by a non-
specific increase in blood levels of the main inflammatory cytokines,
namely IL-1-beta, IL-6, TNF-alpha, IL-8, IL-13, IFN-gamma,
IL-17, and IL-18. The pathogenetic role of TNF-alpha would be
particularly relevant in the case of RA. On the contrary, the role of
IL-2 in the autoimmunity is still controversial, since it stimulates
both TH and T reg lymphocytes, and in any case it has been shown
that autoimmunity-related inflammatory status may be maintained
also in the absence of IL-2 [24]. IL-21 and IL-23 would be also
involved in the pathogenesis of the autoimmunity, because of the
stimulatory role of IL-21 on B lymphocyte differentiation into

plasma cells and on their antibody production, including that of
autoantibodies [53], as well as the promoting effect of IL-23 on
TH17 cell maintenance. In any case, the role of IL-21 in both
neoplastic and autoimmune diseases is still controversial, because
of its ability in association with IL-12 to inhibit both T reg and
TH17 cell differentiation [26]. The evidence of high IL-21 levels
has appeared to predict a more severe disease in patients with SLE
[53]. Moreover, the stimulatory effect of estrogens on both IL-21
production and on the activity of dendritic cells as antigen-present-
ing cells could explain at least in part the higher frequency of
autoimmune diseases in women than in men. In any case, it has to
be remarked that the clinical course of the autoimmune diseases
depends not only on the action of pro-inflammatory agents, but
also on the anti-inflammatory response, mainly mediated by IL-10
and TGF-beta within the cytokine group, as well as by the soluble
receptors or binding proteins for the different cytokines, which
reduce and neutralize their biological activity. In fact, the prognosis
of most autoimmune diseases, including SLE and RA, has been
proven to be better in the presence of high levels of IL-10 and in
particular of TGF-beta. In the same way, the evidence of high
concentrations of soluble TNF-alpha-R1 and R2 receptor has
appeared to predict a disease control in the RA, as well as that of
high blood levels of soluble IL-1-beta R1 and R2 receptors, which
would be associated with a more favorable prognosis in most auto-
immune diseases. As far as autoimmunity-related histological
alterations are concerned, an enhanced TH17 cell lesion infiltration
has been observed, while that of T reg cells may be low or
enhanced, but in any case inadequate to counteract TH-17 cell-
enhanced activity. In case, the role of TGF-beta in the autoimmune
disease is still controversial, since it may either stimulate T reg cell
functions with a following enhanced control of the inflammatory
response or promote TH-17 activity in association with IL-6 and
IL-23. Despite the controversial data, however, the prognosis of
the autoimmune diseases may be predicted on the basis of the
behavior of some cytokines and other immune molecules. In
more detail, the prognosis of the autoimmune diseases is better in
the presence of high blood levels of IL-10, TGF-beta, soluble
receptor of TNF-alpha and IL-1-beta, and IL-18BP, whereas it is
worse in the presence of high levels of IL-6, TNF-alpha, IL-17, and
IL-18, and low concentrations of TGF-beta, IL-10, and IL-18BP.
3.5 Cytokine The allergic diseases, namely those due to an enhanced IgE pro-
Network in Allergic duction, are generally characterized by low IFN-gamma and IL-12
Diseases blood levels in association with high concentrations of IL-18,
which has been proven to stimulate IgE production, even though
it may also play an inhibitory effect on IgE secretion in association
with IL-12 [25].
3.6 Cytokine Atherosclerosis itself has to be considered as an inflammatory and

Network immune-mediated disease, being induced by immune processes, in
in the Atherosclerosis particular through the infiltration of the arterial wall of TH17
lymphocytes, as well as of TH1 lymphocytes, with a consequent
enhanced production of IL-17 and IFN-gamma, respectively [54],
which stimulate arterial infiltration by other immune cells, includ-
ing macrophages. IL-18 has also been proven to be involved in the
atherosclerotic process [25], by stimulating macrophage and lym-
phocyte arterial wall infiltration. Finally, the prognosis of acute
myocardial infarction has appeared to be worse in the presence of
high blood levels of IL-18, as well as of TNF-alpha [25], because
the cytokine-induced inflammatory response may extend the
degree of tissue damage. Therefore, the monitoring of cytokine
network response may be clinically useful also in the cardiologic
pathologies.
3.7 Cytokine Neurodegenerative diseases, namely Alzheimer’s disease (AD), are

Network of the due to glial cell-induced chronic inflammatory status as the end
Neurodegenerative result of an imbalance between pro-inflammatory and anti-
Diseases inflammatory cytokines produced by glial cells themselves [55],
which originate from the monocyte-macrophage system. In AD
the inflammatory response would be mainly due to amyloid-beta
protein deposition with the consequent brain plaque formation.
The main cytokine alterations involved in AD, as well as in other
neurodegenerative diseases, are represented by an enhanced pro-
duction of IL-1-beta, TNF-alpha, and IL-17 in association with a
diminished secretion of the anti-inflammatory cytokine
TGF-beta [55].
4 Comparison in the Biological Response Between Cancer and Autoimmunity
The opposite biological response occurring in the neoplastic and in

the autoimmune diseases, generally considered as the expression of
an immunosuppressive and an immunostimulatory status, respec-
tively, is also confirmed by the fact that the recent immunotherapies
of cancer, carried out to amplify the activity of T lymphocyte system
by cytokines or checkpoint inhibitors, may have the induction of
autoimmune reactions as a side effect in the case of therapeutic
efficacy [14, 15]. By synthetizing, the main characteristics of
cancer-related biological response are consisting of the association
between low lymphocyte count, particularly TH1 cell number, and
enhanced macrophage activity, which is reflected by monocyte
count and which is positively correlated with T reg cell percentage
[10, 28, 30, 34]. At the other side, the autoimmune diseases are
characterized by an increased T lymphocyte activity, namely that of
both TH1 and TH2, depending on the preferential autoantibody
or autoreactive cytotoxic cell mediation of the autoimmune
response, in association with a concomitant enhanced macrophage

function, while T reg cell percentage may be low or normal, but in
any case inadequate to counteract the activation of the immune
response. Therefore, LMR ratio tends to decrease with cancer
progression [28, 30, 34], whereas a progressive increase in LMR
values may predict a disease control and a better prognosis. On the
other hand, LMR tends to be normal or abnormally high in the
autoimmune disease, and a progressive decline in LMR values may
be a sign of disease stabilization [34]. Moreover, because of the
immunomodulatory effects on the anticancer immunity exerted by
the different anticancer therapies, including chemotherapy itself
[56], each kind of cancer cure, irrespectively of the type of treat-
ment, to be effective has to improve the immune status of patients
by increasing lymphocyte count and decreasing monocyte and T
reg cell percentages, with a following increase in LMR values
[57]. Lymphocyte increase has been proven to represent the main
positive prognostic factor predicting the efficacy of IL-2 cancer
immunotherapy [9, 22, 23]. On the contrary, at present no prog-
nostic biomarker related to the biological response of patients has
been identified into the blood to predict the therapeutic results of
the more recent cancer immunotherapies with PD-1 inhibitors,
whose efficacy seems at present to be related to the only expression
of PD-L1 at tumor levels or at tumor microenvironment, even
though preliminary results would suggest that the evidence of an
increase in LMR, due to stimulation of lymphocyte functions and
inhibition of the monocyte-macrophage system, would seem to
correlate with the efficacy of treatment at least in terms of disease
control [14, 15]. As far as tumor characteristics are concerned,
further prognostic information could be reached by determining
in addition to the most routinary markers the expression by tumor
cells of MLT receptor (MT-R) and FAS-ligand (FAS-L), which
could constitute the most tumor-related favorable and negative
prognostic factors, respectively [6, 7, 11], since tumor cells expres-
sing MT-R have been proven to have a less biological malignancy,
whereas in contrast tumor cells expressing FAS-L show the greatest
biological aggressiveness, because of their ability to induce the
apoptosis of T cells expressing FAS receptor in the case of cell-cell
contact.
5 Conclusions
Since lymphocytosis is generally associated with a concomitant

increase in TH1 lymphocyte count [22, 23], with the only excep-
tion of the severe lymphocytopenia in cancer patients, being char-
acterized by an increase in T reg cell percentage [28], and on the
basis of the fact that monocyte count is positively correlated with
the degree of macrophage system activation and in the case of
cancer patients also with tumor macrophage infiltration [10],

which stimulates tumor growth, the simple LMR value could be
identified and proposed as the most inexpensive and general bio-
marker of the biological response for the overall human systemic
diseases either in basal conditions or under therapy, being the end
result of several interactions between immune and inflammatory
responses. In other words, since systemic disease-related immune
alterations are consisting of a loss of the physiological balance
between immunostimulatory and immunosuppressive events,
namely mediated by lymphocytes and macrophages, respectively,
LMR may be identified as the most simple biomarker to monitor
the balance between immunostimulatory and immunosuppressive
events during the clinical course of the human systemic diseases,
including cancer and autoimmune diseases, namely in relation to
the periods of disease stabilization or recurrence [10, 29, 30]. In
fact, since both cancer and autoimmune diseases would depend on
the interactions between lymphocyte and macrophage systems, and
on the basis of the fact that in the neoplastic diseases lymphocyte
count positively correlated with the blood levels of the main anti-
tumor cytokines, IL-2 and IL-12, and monocyte number reflects
macrophage- and T reg-related immunosuppressive events, as well
as on the basis of the fact that in the autoimmune diseases both
lymphocyte and monocyte-macrophage functions are abnormally
increased, the simple LMR may be considered as a nonspecific, but
universal, biomarker to monitor the biological response of patients
affected by systemic diseases during their clinical course [30]. If
lymphocyte count is low in metastatic patients, generally Il-2 and
IL-12 levels are concomitantly low, and in a same way if monocyte
count is high, generally IL-1-beta, IL-6, and TNF-alpha levels are
concomitantly high. Then, not from a scientific but from a clinical
point of view, it would not be necessary to measure the blood of
those cytokines, whose values are already related to those of lym-
phocyte and monocyte count, and the following LMR. Therefore,
in addition to the determination of LMR values, from a clinical
point of view it will be sufficient to associate the measurement of
those cytokine concentrations, whose values may allow to identify
different pathogenesis and different prognosis in both autoimmune
and neoplastic illnesses, such as TGF-beta concentrations in meta-
static cancer because of its fundamental immunosuppressive role, as
well as IL-17, IL-18, TNF-alpha, IL-2, IL-10, and TGF-beta in the
autoimmune diseases, being high levels of IL-17, IL-18,
TNF-alpha, and IL-2 associated with a more aggressive disease,
and high levels of IL-10 and TGF-beta with a more favorable
prognosis. By synthetizing, if the main clinically relevant cytokines
for the neoplastic diseases are at present represented by IL-2 and
IL-12 for their anticancer effects, and IL-6, TNF-alpha, IL-17, and
IL-18 for their involvement in the induction of the inflammatory
response, it has to be remarked that lymphocyte count is generally
positively correlated with IL-2 and IL-12 blood levels [22, 23], and
in the same way the production of inflammatory cytokines generally
correlates with macrophage activation, which is reflected by the
simple monocyte count. Then then the interaction between
immunosuppressive and immunostimulatory cytokines may be
sinthetized by the simple LMR value. The only cytokines, whose
blood detection may be useful to identify the main origin of the
inflammatory response occurring in the single patient, would be
represented by IL-10, because of its potential suppressive and
stimulatory immune effects, and IL-17, being an alternative source
of the inflammatory response with respect to the most ancient
inflammatory response mainly mediated by macrophages through
the release of IL-6 and IL-1-beta. In conclusion, today without a
clinical investigation and monitoring of cytokine network during
the clinical course of disease it is not possible to understand the real
pathogenesis of human systemic diseases and their prognosis. The
only problem is to identify which may be the fundamental bio-
marker of the biological response in relation to the different pathol-
ogies, in an attempt to avoid an excessive social medical cost, due to
the detection of several cytokine levels and lymphocyte subsets,
with the risk of having to analyze many and often controversial
immune parameters. Therefore, if it is true that the investigation of
cytokine network would require the detection of several cytokines,
on the basis of the fact that cytokine-related immune alterations of
the systemic diseases are essentially due to an imbalance between
inflammatory and anti-inflammatory cytokines, it is also true that
from a clinical point of view it could be sufficient to detect LMR
values, which may constitute an inexpensive routinary biomarker to
monitor the clinical course of human systemic diseases, since low
values of LMR would reflect a decreased T lymphocyte function in
association with an enhanced macrophage-T reg cell system activity,
as occurring in the neoplastic diseases, whereas the evidence of a
normal or enhanced LMR would be the expression of a concomi-
tant enhanced activation of both lymphocyte and macrophage
systems, as occurring in the autoimmune diseases [34]. Not only
this, but the monitoring of LMR values during the clinical course of
the neoplastic and autoimmune diseases, if its significance is
explained to each patient, could also contribute to improve the
medical and human relationship with patients, who become more
active and conscious in living their disease.
References
1. Rubinow DR (1987) Brain, behaviour and 3. Antony MH (2003) Psychoneuroimmunology

immunity: an interactive system. J Natl Cancer of cancer. Brain Behav Immun 17:84–91
Invest Monogr 10:79–82 4. Besedovsky HO, Sorkin E, Muller I (1975)
2. Zou W (2006) Regulatory T cells, tumor Hormonal changes during immune response.
immunity and immunotherapy. Nat Rev Proc Soc Exp Biol Med 150:466–471
Immunol 6:295–307
5. Lissoni P, Resentini M, Mauri R et al (1986) 22. Lissoni P (1996) Prognostic markers in

Effects of tetrahydrocannabinol on melatonin interleukin-2 therapy. Cancer Biother Radio-
secretion in man. Horm Metab Res 18:77–78 pharm 11:285–287
6. Mestroni GJM (1993) The immunoneurone- 23. Fumagalli L, Lissoni P, Di Felice G (1999)
docrine role of melatonin. J Pineal Res 14:1–10 Pretreatment serum markers and lymphocyte
7. Brzezinski A (1997) Melatonin in humans. N response to interleukin-2 therapy. Br J Cancer
Engl J Med 336:186–195 80:407–411
8. Banks RE, Patel PM, Selby PJ (1995) 24. Korn T, Bettelli E, Oukka M, Kuchroo VK
Interleukin-12: a new clinical player in cytokine (2009) IL-17 and Th17 cells. Annu Rev
therapy. Br J Cancer 71:655–659 Immunol 27:485–517
9. Grimm EA, Mazumder A, Zhang HZ et al 25. Dinarello CA (2007) Interleukin-18 in the
(1982) Lymphokine-activated killer cell phe- pathogenesis of inflammatory diseases. Semin
nomenon. J Exp Med 155:1823–1841 Nephrol 27:98–114
10. Mantovani A, Allavena P, Sica A et al (2008) 26. Tian Y, Yuan C, Ma D et al (2011) IL-21 and
Cancer-related inflammation. Nature IL-12 inhibit differentiation of T reg and
454:436–444 TH17 cells and enhance cytotoxicity of periph-
11. Kim R, Emi M, Tanabe K et al (2004) The role eral blood mononuclear cells in patients with
of Fas ligand and transfroming growth factor cervical cancer. Int J Gynecol Cancer
beta in tumor progression. Cancer 21:1672–1678
100:2281–2291 27. Dennis KL, Blatner NR, Gounari F, Khazaie K
12. Manfredi B, Sacerdote P, Bianchi M (1993) (2013) Currents status of IL-10 and regulatory
Evidence for an opioid inhibitory tone on T T-cells in cancer. Curr Opin Oncol
cell proliferation. J Neuroimmunol 44:43–46 25:637–645
13. Grotehnermen F (2004) Pharmacology of can- 28. Brivio F, Fumagalli L, Parolini D et al (2008)
nabinoids. Neuroendocrinol Lett 25:14–23 T-helper/T-regulatory lymphocyte ratio as a
new immunobiological index to quantify the
14. Aswell S, Janetka JW, Zabludoff K (2008) anticancer immune status in cancer patients.
Keeping checkpoin kinase in line: new selective In Vivo 22:647–650
inhibitors in clinical trials. Expert Opin Inves-
tig Drugs 17:1331–1340 29. Grivennikov SI, Greten FR, Karin M (2010)
Immunity, inflammation, and cancer. Cell
15. Chen J, Jiang CC, Jin L, Zhang XD (2016) 295:883–899
Regulation of PD-L1: a novel role of
pro-survival signaling in cancer. Ann Oncol 30. Eo WK, Chang HJ, Kwon SH et al (2016) The
27:409–416 lymphocyte-to-monocyte ratio predicts patient
survival and aggressiveness of ovarian cancer. J
16. Miller RA (1996) The aging immune system: Cancer 7:289–296
primer and prospectus. Science 273:70–74
31. Lewis JW, Shavit Y, Terman GV (1983) Appar-
17. Riley V (1981) Psychoneuroendocrine influ- ent involvement of opioids peptides in stress-
ence on immunocompetence and neoplasia. induced enhancement of tumor growth. Pep-
Science 212:1100–1109 tides 4:635–638
18. Mormont MC, Levi D (1997) Circadian sys- 32. Hassan ATM, Hassan ZM, Moazzeni SM
tem alterations during cancer processes: a (2009) Naloxone can improve the antitumor
review. Int J Cancer 70:241–247 immunity by reducing the CD4+CD25+Foxp3
19. Lissoni P, Messina G, Balestra A et al (2008) + regulatory T cells in BALB/c mice. Int J
Efficacy of cancer chemotherapy in relation to Immunopharmacol 9:1381–1386
synchronization of cortisol rhythm, immune 33. McIsaac WM (1961) Formation of 1-methyl-
status and psychospiritual profile in metastatic 6-methoxy-1,2,3,4-tetrahydro-2-carboline
non-small cell lung cancer. In Vivo under physiological conditions. Biochem Bio-
22:257–262 phys Acta 52:607–610
20. Bartsch C, Bartsch H (1999) Melatonin in 34. Lissoni P, Messina G, Tantarelli R et al (2017)
cancer patients and in tuor-bearing animals. The psychoneuroimmunotherapy of human
Adv Exp Med Biol 467:247–264 immune-mediated systemic diseases, including
21. Brivio F, Fumagalli L, Fumagalli G et al (2010) cancer and autoimmune. J Mol Oncol Res 1
Synchronization of cortisol circadian rhythm (1):7–13
by the pineal hormone melatonin in untreata- 35. Lissoni P (1999) The pineal gland as central
ble metastatic solid tumor patients and its pos- regulator of cytokine network. Neuroendocri-
sible prognostic signioficance on tumor nol Lett 20:343–349
progression. In Vivo 24:239–242
36. Maccarone M, Valensise H, Bari M et al (2001) systemic autoimmune disease. J Exp Med
Progesterone up-regulates anandamide hydro- 203:2785–2791
lase in human lymphocytes: role of cytokines 48. Paramalingam SS, Thumboo J, Vasoo S et al
and implication in fertility. J Immunol (2007) In vivo pro- and anti-inflammatory
166:7183–7189 cytokines in normal and patients with rheuma-
37. Candelari PV, Rampoldi A, Harbuzariv A et al toid arthritis. Ann Acad Med Singapore
(2017) Leptin signaling and cancer chemore- 36:96–99
sistance: perspectives. Word J Clin Oncol 49. Gold R, Luhder F (2008) Interleukin-17.
8:106–119 Extended features of a key player in multiple
38. Tesar BM, Shirali AC, Walker WE et al (2009) sclerosis. Am J Pathol 172:8–10
Aging augments IL-17 T -cell alloimmune 50. Katsifis GE, Rekka S, Moutsopoulos NM et al
responses. Am J Transplant 9:54–63 (2009) Systemic and local interleukin-17 and
39. Dejaco C, Duftner C, Schirmer M (2006) Are linked cytokines associated with Sjogren’s syn-
regulatory T cells linked with aging? Exp Ger- drome immunopathogenesis. Am J Pathol
ontol 41:339–345 175:1167–1177
40. Elenkov IJ, Papanicolaou DA, Wilder RL et al 51. Murugaiyan G, Saha B (2009) Protumor vs
(1996) Modulatory effects of glucocorticoids antitumor functions of IL-17. J Immunol
and cathecolamines on human interleukin-12 183:4169–4175
and interleukin-10 production: clinical implica- 52. Harrison OJ, Srinivasan N, Pott J, Schiering C
tions. Proc Assoc Am Physicians 108:374–381 et al (2015) Epithelial-derived IL-18 regulates
41. Zagon IS, Donahue RN, Bonneau RH et al Th17 cell differentiation and Foxp3+ T reg cell
(2011) T lymphocyte proliferation is sup- function in the intestine. Mucosal Immunol
pressed by the opioid growth factor met (5)- 8:1226–1236
enkephalin - opioid growth factor receptor 53. Lee J, Shin EK, Lee SY et al (2014) Oestrogen
axis: implication for the treatment of autoim- up-regulates interleukin-21 production by
mune diseases. Immunobiology 216:579–590 CD4+ T lymphocytes in patients with systemic
42. Aringer A, Smolen JS (2003) Complex cyto- lupus erythematosus. Immunology
kine effects in a complex autoimmune disease: 142:573–580
tumor necrosis factor in systemic lupus erythe- 54. Eid RE, Rao DA, Zhou J et al (1982)
matosus. Arthritis Res Ther 5:172–177 Interleukin-17 and interferon-gamma are pro-
43. Fujino S, Andoh A, Bamba S et al (2003) duced concomitantly by human coronary
Increased expression of interleukin-17 in artery-infiltrating T cells and act synergistically
inflammatory bowel disease. Gut 52:65–70 on vascular smooth muscle cells. Circulation
44. Arican O, Aral M, Sasmaz S et al (2005) Serum 119:1424–1432
levels of TNF-alpha, IFN-gamma, IL-6, IL-8, 55. Zheng C, Zhou XW, Wang JZ (2016) The dual
IL-12, IL-17, and IL-18 in patients with active role of cytokines in Alzheimer’s disease: update
psoriasis and correlation with disease severity. on interleukins, TNF-alpha, TGF-beta and
Mediators Inflamm 2005(5):273–279 IFN-gamma. Transl Neurodegener 5:7–18
45. Langrish CL, Chen Y, Blumenschein W et al 56. Ehrke MJ, Mihich E, Berd D et al (1982)
(2005) Divergent pro- and antiinflammatory Effects of anticancer drugs on the immune sys-
roles for IL-23 and IL-12 in joint autoimmune tem in humans. Semin Oncol 16:230–239
inflammation. J Exp Med 201:233–240 57. Lissoni P, Brivio F, Fumagalli L et al (2009)
46. Dinarello CA (2006) Interleukin 1 and inter- Effects of the conventional antitumor thera-
leukin 18 as mediators of inflammation and the pies: surgery, chemotherapy, radiotherapy and
aging process. Am J Clin Nutr 83:447S–455S immunotherapy on regulatory T lymphocytes
47. Lohr J, Knoechel B, Wang J et al (2006) Role in cancer patients. Anticancer Res
of IL-17 and regulatory T lymphocytes in a 29:1847–1852
Part II
Technologies and Methods in Psychoneuroimmunology

Studies
Chapter 11
Application of Chemogenetics and Optogenetics to Dissect

Brain-Immune Interactions
Abstract
For many years, the complexity and multifactorial nature of brain-immune interactions limited our ability to
dissect their underlying mechanisms. An especially challenging question was how the brain controls
immunity, since the repertoire of techniques to control the brain’s activity was extremely limited. New
tools, such as optogenetics and chemogenetics (e.g., DREADDs), developed over the last decade, opened
new frontiers in neuroscience with major implications for neuroimmunology. These tools enable mapping
the causal effects of activating/attenuating defined neurons in the brain, on the immune system. Here, we
present a detailed experimental protocol for the analysis of brain-immune interactions, based on chemo-
genetic or optogenetic manipulation of defined neuronal populations in the brain, and the subsequent
analysis of immune cells. Such detailed and systematic dissection of brain-immune interactions has the
potential to revolutionize our understanding of how mental and neurological states affect health and
disease.
Key words Immune system, Brain, Neuroimmunology, Neuroscience, Central nervous system,
Immunity, DREADDs, Chemogenetics, Optogenetics
1 Introduction
Neuroimmunology fills the gap between the organism’s two most

critical systems, the nervous and the immune systems. Surprisingly,
this field of research is mostly unidirectional, focused on the effects
of the immune system on the brain, while our understanding of
how the brain affects immunity is still limited, and is focused mainly
on the effects of stress on the immune system. Over the last decade,
novel tools in neuroscience have been developed, allowing us to
dissect, characterize, and control precise neuronal networks, and
can be used to identify their effects on the immune system
[1–3]. Here we focus on two such tools, optogenetics [4] and
chemogenetics [5, 6], which enable us to establish causal effects
of brain activity on the immune system.
195
196 Ben Korin and Asya Rolls
1.1 Harnessing Both optogenetics and chemogenetics enable the control of neuro-
Advances nal activity in vitro and in vivo [6–12]. These two key technologies
in Neuroscience share many similarities, while the differences between them expand
to Study the range of their potential applications.
Neuroimmunology In optogenetics, light-activated membrane ion channels,
known as opsins (e.g., channelrhodopsin and halorhodopsin), are
expressed in neurons using site-specific viral vector delivery. The
expressed ion channel is activated by light of a defined wavelength,
using optic fibers inserted in the area of viral expression in vivo
[4]. The stimulation of each ion channel provides a specific out-
come in the target neuron, depending on the channel type. For
example, channelrhodopsin proteins induce neuronal excitation
due to the inflow of cations following exposure to blue light
(~480 nm), and halorhodopsin channels induce neuronal activity
attenuation by the inflow of chloride ions following exposure to
yellow-green light (~570 nm) [13]. The frequency of light trans-
mission can determine the frequency of excitation of the neuron,
thus providing highly accurate temporal control of neuronal activ-
ity [14]. Expression and manipulation of more than one type of
light-sensitive ion channel in each neuron are possible, by selecting
different excitation wavelengths, allowing an even more accurate
control over the neuronal state [11, 15].
In chemogenetics, a similar viral vector delivery method is
used, but the vector encodes cell receptors activated by small mole-
cules (and not by light) [16]. For example, designer receptors
exclusively activated by designer drugs (DREADDs) are genetically
modified G protein-coupled receptors (GPCRs), activated by a
specific ligand (e.g., clozapine N oxide; CNO) [12]. Once trig-
gered by their ligand, these modified GPCRs induce a chain of
intracellular processes, consequently altering neuronal activity. For
example, the Gq DREADD (hM3Dq) initiates a signaling pathway
that induces neuronal activation, and the Gi DREADD (hM4Di)
initiates a signaling pathway that results in neuronal silencing
[17]. Once expressed in neurons in vivo, DREADDs can be acti-
vated using systemic injection of the ligand, or by oral administra-
tion. Depending on the ligand’s half-life and available
concentration in vivo, its administration results in prolonged mod-
ification of neuronal activity (residual activity may last for hours
[18, 19]).
1.2 Experimental There are several levels at which neuronal pathways in the brain may
Considerations be defined:
in Optogenetics l Spatial (e.g., ventral tegmental area; VTA).
and Chemogenetics
l Neuronal cell type (e.g., dopaminergic neurons in the VTA).
l Neuronal cell projection (e.g., dopaminergic neurons that proj-
ect from the VTA to the nucleus accumbens).
Application of Chemogenetics and Optogenetics to Dissect Brain-Immune. . . 197
l Temporal (e.g., the ability to select the initiation time and

duration of the neuronal manipulation; optogenetics also allows
the control over the specific firing pattern).
The genetic basis and properties of the channel/receptor used
in optogenetics/chemogenetics enable control over all the para-
meters listed above. Thus, the experimental design should carefully
consider these elements. For example, spatial specificity can be
achieved by stereotaxic injection of the viral vector to express the
chemogenetic receptor or the optogenetic ion channel at the target
site. To this end, accurate and validated coordinates are crucial. In
addition, the chosen type of virus for transduction, the titer, and
the injection volume should also be taken into account and cali-
brated to achieve the desired outcome.
To achieve cellular specificity, i.e., defined cell type, one should
drive the viral vector expression using a promoter expressed
uniquely in the target cell. Although it is possible to express the
DREADD or the optogenetic channel directly under the target
promoter (usually using a lentivirus), most studies use the Cre-lox
recombination system. In this system, the genetic information of
choice, conveyed by the viral vector, is expressed only in cells that
express the Cre-recombinase enzyme. This enzyme cleaves and
properly positions the inverted sequence of choice. Expressing the
Cre recombinase under a cell type-specific promoter (e.g., the
tyrosine hydroxylase (TH) promoter for dopaminergic neurons)
permits the optogenetic or chemogenetic manipulation in a defined
population of cells. Additionally, as the delivery of light in optoge-
netics provides spatial specificity by itself, the injection of a targeted
viral vector is often unnecessary when using the proper transgenic
animal model. Thus, one can use transgenic mice expressing the
optogenetic channel or DREADD in specific neuronal populations
(cell type specificity), and control, for spatial specificity by the local
application of light or the DREADD-activating agent, CNO. For
example, the Chat-ChR2-EYFP mouse line can be used to specifi-
cally activate cholinergic neurons in defined brain areas using light
(e.g., the basal forebrain), as determined by the site of optic fiber
placement.
Cell projection specificity is especially important for allowing
independent manipulations of the inputs and outputs of the exam-
ined brain region. In optogenetics, cell projection specificity can be
achieved by injecting the viral vector with the genetic information
for the opsin of choice at the projection site, and placing the optical
fiber at the target site. Using chemogenetics, one may gain projec-
tion specificity using two (or more) viral injection combinations.
For example, one could inject a virus that can carry the information
to the soma, containing the Cre-recombinase gene under the
expression of a cell type-specific promoter, to the axonal projection
of a specific site, and a second injection to the site of the cell soma,
using a different virus containing the floxed (loxP) genetic infor-

mation for a chemogenetic receptor (e.g., DREADDs). Thus, only
the neurons that project to the site of choice will have the complete
genetic information required to execute the neuronal
manipulation.
Here, we focus mainly on the chemogenetic approach, since its
extended timescale of neuronal manipulation, noninvasive activa-
tion, and relative simplicity, may be more appropriate for the analy-
sis of brain-immune interactions. Nonetheless, the chemogenetic
and optogenetic techniques have many conceptual and experimen-
tal similarities, and thus, throughout the protocol, we highlight
points that may be relevant for the use of optogenetics as well.
These tools constantly develop; hence, new variations may expand
the available tool box.
1.3 Immune System After selecting the neuronal manipulation technique, the
Characterization subsequent experimental design, and more specifically, the timing
Following Neuronal of immune analysis, should be considered. The effects of optoge-
Manipulation netics and chemogenetics on neuronal activation are immediate.
However, subsequent effects on immune cells in the periphery may
not be instantaneous, and will most likely occur on different time-
scales, depending on the evaluated component. When analyzing
mRNA levels, one may need to consider the time required for
mRNA synthesis and/or degradation. In the case of protein expres-
sion (e.g., intracellular and extracellular receptors) a longer lag
period (several hours) should be expected. Moreover, under an
experimental design dependent on accumulative neural transmis-
sion to the periphery, requiring prolonged and/or repeated neuro-
nal manipulation, several hours to several days from the beginning
of the experiment may be needed. In the case of chemogenetics, the
actual initiation time and duration of the chemogenetic receptor
activation (depending on ligand availability) may also affect the
timescale of the response.
One may also consider introducing an immunological stimulus;
this is expected to provide a more orchestrated and distinctive
response by the immune system, depending on the stimulus. For
example, when we performed reward system activation in naı̈ve
mice, we mainly saw small effects that we interpreted as priming
[1]. However, once we introduced a bacterial challenge, the effects
of reward system activation on the antibacterial response became
functionally evident [1]. Taken together, the features of the applied
neuronal manipulation, the timing of experimental analysis, and the
immunological context are especially important. These crucial vari-
ables should be chosen based on the proposed experimental
hypothesis.
Another crucial factor to consider in the experimental design is
the appropriate control group(s). The control group should
account for the effects of surgery, viral expression, and
administration of a ligand (such as CNO in DREADDs), or fiber

and light delivery (for optogenetics). Therefore, most studies use as
a control group a viral vector lacking the DREADD/light-sensitive
channel coding sequence, containing only the fluorescent reporter
gene (e.g., mCherry). These control groups also have to be simi-
larly treated with the chosen ligand (chemogenetics), or light
(optogenetics).
To characterize the effects of specific neuronal manipulations
on the peripheral immune system, we often refer to the use of flow
cytometry. Flow cytometry has been a key scientific tool in biology
and more specifically immunology. This fluorescence-based tech-
nology allows a rapid, efficient, and reliable high-throughput char-
acterization of cells in single-cell suspension. However, its
fundamental feature, the fluorescence overlap, limits this analysis
to a relatively small number of markers. Accordingly, flow cytome-
try is better suited for a predefined and hypothesis-based analysis.
In the absence of such a hypothesis, and for a broad depiction of cell
phenotypes of the immune system, high-throughput and high-
dimensional technologies, such as single-cell RNA sequencing
[20], proteomics [21–23], or CyTOF mass cytometry [24–26],
may be more appropriate. These technological advancements
allow us to characterize multiple markers, and thereby gain a
broader picture of the immunological milieu.
In this chapter, we present a systematic in vivo approach for
exploring the effects of specific brain neuronal networks on the
peripheral immune system. This experimental design is composed
of two central parts: one is the site-specific neuronal manipulation
using chemogenetics or optogenetics, and the other is the analysis
of subsequent effects on the peripheral immune system in periph-
eral immune tissues. As an example of cell-specific and site-specific
neuronal manipulation, we describe here the experimental protocol
of the activation of dopaminergic neurons (expressing TH) in the
VTA, as previously shown [1].
2 Materials
2.1 Animals 1. Cre-dependent transgenic mice: Age: 8–10 weeks. For exam-
ple, for targeting dopaminergic neurons: TH-Cre mice (e.g.,
B6.Cg-Tg(Th-Cre)1Tmd/J).
2. All experiments should be performed in accordance with the
institution’s guidelines for the care and use of laboratory
animals.
2.2 Chemogenetics 1. Anesthetics: Ketamine/xylazine mixture (ketamine 80 mg/kg;

xylazine 15–20 mg/kg) in sterile saline solution (NaCl 0.9%).
Made fresh, kept in 4 C or on ice. Optional: Inhalational

anesthesia with isoflurane.
2. Digital Lab Stereotaxic frame (Stoelting).
3. Binocular surgical microscope (Zeiss).
4. Motorized stereotaxic injector—Quintessential Stereotaxic
Injector (Stoelting).
5. 2–10 μL Micro-syringe with blunt needle (Stoelting).
6. Cordless micro drill (Stoelting).
7. Warming pad (37–38 C).
8. Povidone-iodine solution.
9. Sterilized cotton swabs.
10. Vetbond™ Tissue Adhesive (3M).
11. Sterile saline solution (NaCl 0.9%).
12. Duratears eye ointment (Alcon).
13. Sterilized surgery tools: Scalpels, scissors, tweezers.
14. Ice bucket.
15. Viral vectors:
l For neuronal activation: Gq (hM3Dq) or Gs (Gs-D)
DREADD viral vector (e.g., AAV-hM3D(Gq)-mCherry;
Addgene).
l For neuronal attenuation: Gi (hM4Di) DREADD (e.g.,
AAV-hM4D(Gi)-mCherry; Addgene).
l Control vector (e.g., AAV-mCherry; Addgene).
l All viral vectors should be stored at 80 C, unless specified
otherwise.
l Avoid freeze–thaw cycles, and divide into aliquots
(5–10 μL each).
Caution: The use of viral vectors should be done with the
appropriate approvals, and in accordance with the institu-
tion’s safety regulations.
16. Clozapine N oxide (CNO; Sigma-Aldrich): Keep in the dark
(light sensitive) at room temperature. Once dissolved, keep at
4 C in the dark, use fresh.
17. Dimethyl sulfoxide (DMSO).
18. 1 mL Syringe.
19. 10% Bleach (sodium hypochlorite) solution, for disposal of
viral waste.
20. 70% Ethanol for disinfection and cleaning of surgical tools.
2.3 Adaptations 1. Viral vectors:

for Optogenetics l For neuronal activation: For example, channelrhodopsin
viral vector (e.g., AAV-ChR2-GFP; Addgene).
l For neuronal attenuation: For example, halorhodopsin viral
vector (e.g., AAV-HaloR-GFP; Addgene).
l Control vector (e.g., AAV-GFP; Addgene).
Caution: The use of viral vectors should be done with the
appropriate approvals and in accordance with the institu-
tion’s safety regulations.
2. Optical fiber implant or cannula guide (length is determined
based on the experimental target).
3. Doric connector or injector cannula and dummy cannula
(length is determined based on the experimental target).
4. Light source: Laser, LED, or other source with the appropriate
wavelength filter.
5. Optical fiber cord and connector.
6. Metabond dental cement (e.g., C & B Metabond Quick Adhe-
sive Luting Cement; Parkell).
7. Pulse generator.
2.4 Immuno- 1. 4% Paraformaldehyde (PFA) in PBS.

fluorescence 2. Sucrose LR: Dissolve for 30% (w/vol) sucrose in PBS.
Validation of Viral
3. Surgery tools: Scissors, tweezers.
Expression
and Localization 4. Falcon 15 mL conical tubes.
5. Crushed dry ice.
6. Cryosection device.
7. Tissue slides: Superfrost Plus (Thermo Scientific).
8. Cryosection glue: Tissue-Tek® O.C.T. Compound (Sakura).
9. Anti-mCherry antibody (Novus Biologicals; Catalog # NBP1-
96752; clone 1C51).
10. Anti-tyrosine hydroxylase antibody (Merck; Catalog
#MAB318; clone LNC1).
11. Anti-c-Fos (4–17) antibody (Merck; Catalog #PC38).
12. Fluorescently labeled secondary antibodies.
3 Methods
3.1 Site-Specific 1. Prepare the surgical environment in advance; make sure that all
Delivery of Viral Vector materials and instruments are readily available and properly
for Chemogenetics sterilized.
2. Prior to viral delivery, the work environment must be approved

by the institutional safety authorities.
3. Use a disposable lab coat, gloves, shoe covers, eye protection,
and a respirator, as regulations require.
4. Fully anesthetize mice for surgery.
5. Monitor vital signs and state of the animal throughout the
procedure.
6. Validate depth of anesthesia, for example, with toe pinch to
verify the absence of a withdrawal reflex.
7. Remove hair from the top of the scull using a hair shaver.
8. Aseptically prepare the shaved skin with povidone-iodine
solution.
9. Using surgical scissors, make a midline longitudinal cut, start-
ing between the ears and finishing before the nose.
10. Gently spread and detach the skin and membranous tissue from
the skull.
11. Place mouse in the stereotaxic device.
12. Place a small amount of sterile eye ointment on each eye to
prevent drying.
13. Make sure that the skin remains moist using sterile saline
solution.
14. Make sure that the animal is kept warm throughout the surgery
using a heating pad (37–38 C) and suitable covering (e.g.,
fabric, paper, or cotton).
15. Gently remove membranous tissue from the skull until bregma
and lambda are visible.
16. Equate the coordinates at the bregma to zero.
17. Make sure that the skull is properly aligned, by measuring
dorsal-ventral and anterior-posterior coordinates at the
bregma.
18. In case of any change in stereotaxic alignment, restore the
bregma coordinates to zero.
19. Validate the aligned location of the bregma once more before
moving to the desired coordinates; for example, the VTA is
localized at (anterior–posterior 3.2 mm; medial–lateral
0.48 mm; dorsal–ventral 4.7 mm) [1].
20. After identifying the location for the insertion of the needle,
carefully drill a small hole, using a drill designed for small
animal surgery (e.g., micro drill), and confirm that the needle
can be properly inserted.
21. Insert the needle according to the desired coordinates.
22. After the needle is in the proper location, wait for 3 min for
tissue adjustment. Monitor for bleeding or any signs of stress.
23. During this time, keep skin tissue moist using a cotton swab
with sterile saline solution (NaCl 0.9%) or sterile PBS / .
24. Inject the viral vector at a rate of 0.1 μL/min (e.g., 0.7 μL of
viral vector over 7 min).
25. Following injection, keep the needle in place for an additional
5 min, to enable optimal cell transfection, and avoid backflow
of the viral suspension.
26. Slowly and carefully pull back the needle. Monitor for bleeding
or any abnormalities.
27. Disinfect the surgical area in the skull using a cotton swab with
povidone-iodine solution.
28. Using tissue glue (can also be done with stitches or staples),
reattach the two skin flaps while fully covering the skull. Make
sure to fully reattach the tissue.
29. To support postsurgical recovery, inject (subcutaneous, s.c.)
1 mL of pre-warmed (37 C) 0.9% NaCl.
30. Following the surgical procedure, let mice rest in a controlled
warm environment, and monitor until they are awaken.
31. Carefully monitor the condition of the mice (i.e., mobility, fur,
behavioral changes, weight) daily during the following week.
32. Provide analgesics (e.g., buprenorphine 0.05–0.1 mg/kg) as
required. Exclude animals with unresolved complications.
33. Allow recovery and effective viral expression for 21–30 days.
3.2 Protocol 1. Following viral injection, wait for 10 min for virus to fully
Adjustments diffuse.
for Optogenetics In transgenic mice that heritably express the opsin of choice,
(See Fig. 1) continue as follows (i.e., without the viral injection).
2. Using a stereotactic holder, slowly insert the optical fiber
implant based on the desired coordinates, i.e., in the same
location as drilled previously.
3. Secure doric/cannula implant location with Metabond dental
cement.
4. Release the optical fiber implant from the stereotaxic holder.
5. To the extent possible, reattach the two flaps of the skin using
tissue glue.
6. To support postsurgical recovery, inject (subcutaneous, s.c.)
1 mL of pre-warmed (37 C) 0.9% NaCl.
7. Following the surgical procedure, let mice rest in a controlled
warm environment, and monitor until they are awaken.
204
Chemogenetics Optogenetics
CNO Blue Yellow
Promoter DREADD Reporter
Virus expression Recovery (30 days)
Promoter ChR2 Reporter
Ready for
Promoter Reporter (control)
neuronal manipulation
hM3Dq or hM4Di Na+ or Cl-
(excitation) (attenuation) (excitation) (attenuation)
Fig. 1 Viral delivery with spatial resolution. Schematic representation of the stereotaxic injection of a viral vector that encodes the chemogenetic/optogenetic
information, and a fluorescent reporter (or control–fluorescent reporter only). Following injection, mice are allowed to recover for at least 30 days
8. Carefully monitor the state of the mice (i.e., mobility, fur,

behavioral changes, weight) on each day during the
following week.
9. Provide analgesics (e.g., buprenorphine 0.05–0.1 mg/kg) as
required. Exclude animals with unresolved complications.
10. Allow recovery and effective viral expression for 30 days.
3.3 In Vivo Activation 1. DREADD activation is performed by the intraperitoneal (i.p.)

of DREADDs injection of CNO (for example, 1 mg/kg) in sterile saline
solution (NaCl 0.9%), freshly prepared at the day of injection.
2. Calibration of CNO injection for the use of minimal concen-
tration is recommended, to avoid possible residual effects [18].
3. For example (depending on the final volume needed), dissolve
1 mg of CNO in 10 μL of DMSO, and then move to 10 mL of
sterile saline solution (NaCl 0.9%).
4. Keep at 4 C or on ice, and cover to protect from light.
5. Using this solution, for example, at 1 mg/kg of CNO, inject
10 μL of the solution per 1 g mouse weight (for example,
250 μL for a 25 g mouse).
6. CNO rapidly penetrates the brain, and in mice it remains in
body fluids for at least 60 min following i.p. injection [5].
7. The timing of the subsequent immune analysis may vary, and
should be determined experimentally.
8. Validation of effects of neuronal manipulation may be per-
formed by behavioral tests (e.g., conditioned place preference,
CPP), according to the experimental paradigm and expected
behavioral outcome.
9. Validation of neuronal activation may be performed using
c-Fos immunohistochemical analysis on mice sacrificed
90 min after CNO injection, as previously shown with dopami-
nergic neurons in the VTA [1].
3.4 In Vivo Activation 1. Following surgical recovery, habituate mice to the use of the
of Light-Sensitive optical fiber cord.
Channels (See Fig. 2) 2. On the day of the experiment, attach the optical cord extension
to the mouse implant, and secure tightly.
3. Connect the optical cord to the appropriate light source.
4. Provide the desired light pulse (manually or automatically
applied) to stimulate opsin-expressing cells.
5. Note pulse intensity, frequency, and duration of light pulses,
along with the overall number of trials.
CNO
Chemogenetic Tissue extraction
stimulation and processing Flow cytometry
Immune
system analysis Mass cytometry
Light Proteomics
Optogenetic
stimulation RNA sequencing
Fig. 2 Illustration of chemogenetic/optogenetic activation and immunological analysis. Following stereotactic

viral injection, recovery, and neuronal manipulation, chemogenetic or optogenetic receptor-expressing mice
and controls are subsequently subjected to a broad analysis of immunological organs
3.5 Brain Tissue 1. Following extraction of relevant immunological tissues

Extraction for (or following sacrifice of the animal), separate the head from
Immunofluorescence the body using scissors.
Validation of Viral 2. Detach the skin from the upper part of the skull using scissors.
Expression 3. Make a small incision on the most caudal part of the skull,
and Localization taking care not to cut into the brain.
4. Make another incision in the rostral part, between the eyes.
5. Carefully peel the top section of the skull using tweezers.
/
6. Extract the brain into a small dish with PBS .
7. Transfer to a 15 mL tube with 7 mL 4% PFA in PBS; keep at
4 C.
/
8. After 48 h (at least), move the brain to 30% sucrose in PBS ,
and keep at 4 C.
9. Once the tissue is submerged and no longer floats (usually after
48–72 h), take the brain and briefly dry it on a tissue paper,
then fast freeze using crushed dry ice.
10. Keep at 80 C until sliced in a cryosection device.
11. Use cryosection glue to properly place the tissue.
12. Tissue sections should be between 6 and 12 μm.
13. DREADDs express a fluorescent marker (e.g., m-Cherry),
which is usually visible without additional staining.
14. For validation of neuronal cell type-specific expression, stain
with the appropriate antibodies (primary and secondary, using
a suitable fluorescent marker). Compare the number of
DREADD-positive neurons which are also positive for the
cell lineage marker. Note that expression intensity of the fluo-
rescent marker may vary between the control and DREADD
groups.
15. Optogenetic viral delivery may be confirmed in a similar
manner.
4 Notes
1. Viral delivery and expression:

(a) The optimal volume of viral injection may vary; it should
typically range from 0.3 to 1 μL, and requires calibration
for optimal results.
(b) It is always recommended to validate viral expression and
function, before evaluation of subsequent immune effects.
2. Neuronal manipulation:
(a) CNO may be applied chronically by daily i.p. injections or
by addition to drinking water.
(b) CNO intake by drinking eliminates injection stress, yet
may result in greater variability between the animals, due
to the differences in the timing of intake and the volume.
For acute (one time) activation of DREADDs, the timing
is imperative, and thus injection is preferred.
(c) Lower concentrations of CNO for DREADD activation
are often possible, but the optimal concentration should
be validated.
(d) Validation of neuronal activity following chemogenetic/
optogenetic stimulation can be done by photometry
recordings [27, 28], electrical recordings [29], or immuno-
histochemistry (IHC)/immunofluorescence (IF) staining
for markers associated with neuronal activation (e.g.,
c-Fos) [1].
Acknowledgments
We would like to thank S. Schwarzbaum for editing the paper, and

T.L. Ben-Shaanan, M. Schiller, and H. Azulay-Debby for their help
and advice. Our research is supported by the Israeli Ministry of
Science, Technology & Space (MOST; 3-12070), Prince Center for
Neurodegenerative Diseases, Israeli Society for Science (1862/15),
the Colleck Research Fund and the ADELIS Foundation. A.R. is a
Howard Hughes Medical Institute-Wellcome Trust researcher.
References
1. Ben-Shaanan TL, Azulay-Debby H, Dubovik 20:700–707. https://doi.org/10.1038/nn.
T et al (2016) Activation of the reward system 4526
boosts innate and adaptive immunity. Nat Med 3. Pavlov VA, Tracey KJ (2017) Neural regulation
22:940–944. https://doi.org/10.1038/nm. of immunity: molecular mechanisms and clini-
4133 cal translation. Nat Neurosci 20:156–166.
2. Abe C, Inoue T, Inglis MA et al (2017) C1 https://doi.org/10.1038/nn.4477
neurons mediate a stress-induced anti-inflam- 4. Deisseroth K (2011) Optogenetics. Nat Meth-
matory reflex in mice. Nat Neurosci ods 8:26–29
5. Roth BL (2016) DREADDs for neuroscien- Neuropsychopharmacology 37:296–297.

tists. Neuron 89:683–694. https://doi.org/ https://doi.org/10.1038/npp.2011.179
10.1016/j.neuron.2016.01.040 18. Gomez JL, Bonaventura J, Lesniak W et al
6. Urban DJ, Roth BL (2015) DREADDs (2017) Chemogenetics revealed: DREADD
(designer receptors exclusively activated by occupancy and activation via converted cloza-
designer drugs): chemogenetic tools with ther- pine. Science 357:503–507. https://doi.org/
apeutic utility. Annu Rev Pharmacol Toxicol 10.1126/science.aan2475
55:399–417. https://doi.org/10.1146/ 19. Raper J, Morrison RD, Daniels JS et al (2017)
annurev-pharmtox-010814-124803 Metabolism and distribution of clozapine-N-
7. Cardin JA, Carlén M, Meletis K et al (2010) oxide: implications for nonhuman primate che-
Targeted optogenetic stimulation and record- mogenetics. ACS Chem Nerosci
ing of neurons in vivo using cell-type-specific 8:1570–1576. https://doi.org/10.1021/
expression of channelrhodopsin-2. Nat Protoc acschemneuro.7b00079
5:247–254. https://doi.org/10.1038/nprot. 20. Ziegenhain C, Vieth B, Parekh S et al (2017)
2009.228 Comparative analysis of single-cell RNA
8. Furman M, Zhan Q, McCafferty C et al (2015) sequencing methods. Mol Cell 65:631–643.
Optogenetic stimulation of cholinergic brain- e4. https://doi.org/10.1016/j.molcel.2017.
stem neurons during focal limbic seizures: 01.023
effects on cortical physiology. Epilepsia 56: 21. Matsumoto M, Matsuzaki F, Oshikawa K et al
e198–e202. https://doi.org/10.1111/epi. (2017) A large-scale targeted proteomics assay
13220 resource based on an in vitro human proteome.
9. Zeng H, Madisen L (2012) Mouse transgenic Nat Methods 14:251–258. https://doi.org/
approaches in optogenetics. Prog Brain Res 10.1038/nmeth.4116
196:193–213. https://doi.org/10.1016/ 22. Nyman TA, Lorey MB, Cypryk W, Matikainen
B978-0-444-59426-6.00010-0 S (2017) Mass spectrometry-based proteomic
10. Zhang F, Gradinaru V, Adamantidis AR et al exploration of the human immune system:
(2010) Optogenetic interrogation of neural focus on the inflammasome, global protein
circuits: technology for probing mammalian secretion, and T cells. Expert Rev Proteomics
brain structures. Nat Protoc 5:439–456. 14:395–407. https://doi.org/10.1080/
https://doi.org/10.1038/nprot.2009.226 14789450.2017.1319768
11. Sidor MM, Davidson TJ, Tye KM et al (2015) 23. Rieckmann JC, Geiger R, Hornburg D et al
In vivo optogenetic stimulation of the rodent (2017) Social network architecture of human
central nervous system. J Vis Exp 95:51483. immune cells unveiled by quantitative proteo-
https://doi.org/10.3791/51483 mics. Nat Immunol 18:583–593. https://doi.
12. Dong S, Rogan SC, Roth BL (2010) Directed org/10.1038/ni.3693
molecular evolution of DREADDs: a generic 24. Cheung RK, Utz PJ (2011) Screening:
approach to creating next-generation RASSLs. CyTOF—the next generation of cell detection.
Nat Protoc 5:561–573. https://doi.org/10. Nat Rev Rheumatol 7:502–503
1038/nprot.2009.239 25. Spitzer MH, Nolan GP (2016) Mass cytome-
13. Deisseroth K (2015) Optogenetics: 10 years of try: single cells, many features. Cell
microbial opsins in neuroscience. Nat Neurosci 165:780–791
18:1213–1225. https://doi.org/10.1038/nn. 26. Brodie TM, Tosevski V (2017) High-
4091 dimensional single-cell analysis with mass cyto-
14. Gunaydin LA, Yizhar O, Berndt A et al (2010) metry. Curr Protoc Immunol
Ultrafast optogenetic control. Nat Neurosci 118:5.11.1–5.11.25. https://doi.org/10.
13:387–392. https://doi.org/10.1038/nn. 1002/cpim.31
2495 27. Gunaydin LA, Grosenick L, Finkelstein JC et al
15. Prigge M, Schneider F, Tsunoda SP et al (2014) Natural neural projection dynamics
(2012) Color-tuned channelrhodopsins for underlying social behavior. Cell
multiwavelength optogenetics. J Biol Chem 157:1535–1551. https://doi.org/10.1016/j.
287:31804–31812. https://doi.org/10. cell.2014.05.017
1074/jbc.M112.391185 28. Akemann W, Mutoh H, Perron A et al (2010)
16. Sternson SM, Roth BL (2014) Chemogenetic Imaging brain electric signals with genetically
tools to interrogate brain functions. Annu Rev targeted voltage-sensitive fluorescent proteins.
Neurosci 37:387–407. https://doi.org/10. Nat Methods 7:643–649
1146/annurev-neuro-071013-014048 29. LeChasseur Y, Dufour S, Lavertu G et al
17. Ferguson SM, Neumaier JF (2012) Grateful (2011) A microprobe for parallel optical and
DREADDs: engineered receptors reveal how electrical recordings from single neurons
neural circuits regulate behavior. in vivo. Nat Methods 8:319–325
Chapter 12
Psychoneuroimmunology and Natural Killer Cells: The

Chromium-Release Whole-Blood Assay
Mary Ann Fletcher, Zachary Barnes, Gordon Broderick,
and Nancy G. Klimas
Abstract
Natural killer (NK) cells are an essential component of innate immunity. These lymphocytes are also
sensitive barometers of the effects of endogenous and exogenous stressors on the immune system. This
chapter describes a chromium (51Cr)-release bioassay designed to measure to the target cell killing capacity
of NK cells (NKCC). Key features of the cytotoxicity assay are that it is done with whole blood and that
numbers of effector cells are determined for each sample by flow cytometry and lymphocyte count. Effector
cells are defined as CD3-CD56+ lymphocytes. Target cells are the K562 erythroleukemia cell line. Killing
capacity is defined as number of target cells killed per effector cell, at an effector cell/target cell ratio of 1:1
during a 4-h in vitro assay.
Key words Natural killer cells, NK cells, Natural killer cell cytotoxicity, NKCC, Lymphocytes, K562
target cells, Chromium-release assay, Innate immunity, Flow cytometry
1 Introduction
The natural killer (NK) cell is a large, granular lymphocyte with the
ability to lyse tumor cells and virus-infected cells without prior
exposure and immunization [1]. These cells can prolong asymp-
tomatic states in HIV-infected persons with low CD4 counts [2]
and protect against malignancy [3]. The NK cell is a reliable marker
of neuroendocrine–immune interactions. Stressful life events that
trigger the fight or flight response, such as a natural disaster, can
alter lymphocyte trafficking and function, leading to elevated
( p < 0.001) natural killer cell cytotoxicity (NKCC) and number
of circulating NK cells ( p < 0.000) as were seen in post-Hurricane
Andrew samples [4]. According to a meta-analysis by Segerstrom
and Miller [5], the mobilization of NK cells during acute psycho-
logic stressors is one of the most replicated and robust findings in
human psychoneuroimmunology. For example, 45 first-time tan-
dem parachutists were examined for NK activity 2 h before,
209
210 Mary Ann Fletcher et al.
Table 1
Friedman test: effect of aerobic exercise on NK cell counts in Gulf War illness cases (T0 ¼ baseline;
T1 ¼ VO2; T2 ¼ 4 h)
Column name Count Ranked sum Average rank

T0 CD3-CD56+ cells 37 70.50 1.64
T1 CD3-CD56+ cells 37 119.00 2.77
T2 CD3-CD56+ cells 37 68.50 1.59
Column name Mean Std. dev. Median 25–75 Percentiles
T0 CD3-CD56+ cells 138.13 72.13 127.00 90.00 168.25
T1 CD3-CD56+ cells 461.66 276.09 417.00 252.00 560.00
T2 CD3-CD56+ cells 141.74 73.39 132.00 93.00 173.00
Chi square ¼ 38.03000
Probability ¼ 0.000000
immediately after, and 1 h after jumping. Functional capacity of NK

cells increased immediately after jumping followed by a decrease
significantly below starting values 1 h later [6]. In laboratory stud-
ies with Gulf War illness patients and healthy controls, serial mea-
surements showed a significant and rapid positive effect of an
aerobic exercise challenge (V02 max) on the number of NK cells
as shown in Tables 1 and 2. NK cell activity is also affected as shown
in Fig. 1 [7, 8]. Acute psychological stress may result in increased
risk of infection, for example development of colds in rhinovirus-
inoculated volunteers [9]. Sustained stress resulted in lower natural
killer cell cytotoxicity (NKCC) in Japanese critical care physicians in
individuals taking 0–3 days off per month, as compared to those
taking four or more days off [10]. Abnormalities of the stress
response are hypothesized as a trigger or mediator of chronic
fatigue syndrome (CFS) which is associated with low NKCC [11]
(Figs. 2 and 3). The NK cells from patients with CFS have dimin-
ished intracellular perforin [12].
The process of cytolysis can be dived into three stages: conju-
gate formation (binding of the effector cell to the target cell),
triggering of the lytic process (signal transduction), and lethal hit
(granule exocytosis). Once triggered, the cytolytic process may
propagate as effector cells detach from lysed targets and recycle to
initiate new lytic interactions. The binding of the NK cell to the
target cell involves receptors and ligands, but unlike cytotoxic T
cells does not result from antigen recognition. Following the tar-
get–effector cell contact, intracellular granules move to the surface
of the NK cell. Perforin from these granules facilitates the release of
serine proteases, granzymes, and their passage through the target
cell membranes. Granule exocytosis requires binding of the
secreted perforin to the target cell membrane and polymerization
Psychoneuroimmunology and Natural Killer Cells: The Chromium-Release. . . 211
Table 2
Friedman test: effect of aerobic exercise on NK cell counts in healthy controls (T0 ¼ baseline;
T1 ¼ VO2; T2 ¼ 4 h)
Column name Count Ranked sum Average rank

T0 CD3-CD56+ cells 35 59.50 1.42
T1 CD3-CD56+ cells 37 119.00 2.83
T2 CD3-CD56+ cells 37 73.50 1.75
Column name Mean Std. dev. Median 25–75 Percentiles
T0 CD3-CD56+ cells 148.75 80.94 147.00 82.00 204.00
T1 CD3-CD56+ cells 642.74 377.05 602.00 358.00 802.00
T2 CD3-CD56+ cells 176.29 106.73 172.00 109.00 203.25
Chi square ¼ 46.08300
Probability ¼ 0.000000
20
N = 40; p = .001
16
12
% NKCC
0
Baseline V02 max
Fig. 1 Effect of exercise challenge to V02 max on NKCC (% of target cells killed at a target cell ratio of 1:1) in
40 healthy controls
of this perforin to polyperforin. Intracellular perforin can be quan-

titatively measured and is a marker for NK cell function [12]. Tran-
sient expression of CD107a occurs on the NK cell surface following
perforin release and can be used as surrogate marker for NK
activity [13].
This chapter describes a chromium (51Cr)-release bioassay
designed to measure to the target cell killing capacity of NK cells.
This type of assay was first used over 40 years ago as a quantitative
tool to assess lymphocyte-associated cytolysis [14]. Radioactive
chromium binds to intracellular proteins. Upon lysis of the target
Fig. 2 Histogram of NKCC (% of target cells killed at a target cell ratio of 1:1) for 230 healthy controls
cell, the intracellular 51Cr is released in an amount proportional to

the amount of cell lysis. Key features of this NKCC assay are that it
is done with whole blood and that numbers of effector cells are
determined for each sample by flow cytometry and lymphocyte
count [15, 16]. The use of fresh, whole blood is important for
studies in psychoneuroimmunology. The NK cells are not subjected
to centrifugation and separation from their soluble and cellular
milieu. The whole-blood samples are incubated with 51Cr-labeled
target cells at four target-to-effector cell ratios, in triplicate for 4 h.
2 Materials
2.1 Equipment 1. Multiparameter flow cytometer.

2. Electronic hematology instrument.
3. Automatic gamma counter.
4. Centrifuge with swing-out holders for 96-well culture plates.
5. Electronic timer.
6. Biological hood.
7. Aerobic filter unit.
8. Water-jacketed incubator (37 C. 5% CO2, 95% humidity).
9. Repeat dispenser micropippetter.
Fig. 3 Histogram of NKCC (% of target cells killed at a target cell ratio of 1:1) for 176 chronic fatigue syndrome
cases
2.2 Supplies 1. Sterile polypropylene centrifuge tubes.

2. Sterile micropipettes tips (ranging from 1 μL to 1 mL).
3. Sterile transfer pipettes.
4. Sterile 75 mL tissue culture flasks.
5. Sterile 96-well flat-bottom tissue culture plates.
2.3 Reagents 1. Triton X.

2. Fetal bovine serum (FBS).
3. RPMI1640 medium (1), without glutamine.
4. Minimum essential medium: Nonessential amino acids (MEM,
10 mM, 100).
5. Sodium pyruvate (1 mM).
6. L-Glutamine 200 mM (100), aliquoted into 5 mL and stored
at 20 C7.
7. Penicillin (5000 U) and streptomycin (5000 μg/mL) (100)
aliquoted into 5 mL and stored at 20 C (Pen-Strep).
8. Trypan blue (0.4%).

9. Triton X-100.
51
10. Cr (5 mCi).
3 Pre-assay Preparations (in Biohazard Hood with Sterile Technique)
3.1 Tissue Culture 1. Stock media (SM): In a flask mix 500 mL of RPMI, 5 mL of
phosphate-buffered saline (PBS), 5 mL MEM, 5 mL sodium
pyruvate. Filter through 0.2 μm pore size filter.
2. Assay media (AM): Mix and filter (0.2 μm pore size filter) 90 mL
of SM and 10 mL of FBS.
3. Culture media (CM): Mix and filter (0.2 μm pore size filter)
85 mL of SM and 15 mL of FBS.
4. Storage requirements: Chromium in solution, stock media,
assay media, nonessential amino acids, FBS, MEM, and sodium
pyruvate should all be kept in the refrigerator at 2–8 C for up to
1 week. L-Glutamine and Pen-Strep should be kept in the
freezer at 20 C.
5. Target cells: The usual target cell is the K562 erythroleukemic
cell line available from the ATCC http://www.atcc.org/. The
NK-sensitive K562 cell line is maintained in a humidified 5%
CO2 atmosphere at 37 C in stationary suspension in CM. Cells
are subcultured twice weekly to produce log-phase growth (see
Note 1; 4 106 cells in 20 mL of CM will yield approximately
16 106 cells).
3.2 Patient 1. No special preparation is necessary. However, if possible the

Preparation sample should be collected in the morning so that the assay,
which requires several hours to complete, may be done in the
same day. Also, NK cell count varies in a circadian rhythm over
24 h [17].
3.3 Specimen 1. Using standard aseptic technique, collect by venipucture 3 mL

Collection whole blood into a sodium heparin tube and 1 mL of whole
and Handling blood into an ethylenediaminetetraacetic acid (EDTA) tube.
Conditions Universal precautions should be observed when handling the
blood sample and all biohazardous materials. Chromium used in
this procedure is radioactive and should be handled using appro-
priate safety standards. Except when stated otherwise, all trans-
fers of cells and biological fluids should be done in the biological
hood. However, after the final incubation is complete, samples
may be handled outside of the hood, but still with caution as to
avoid cross-contamination.
4 Assay Procedure
4.1 Target Cell 1. Pour the entire volume of cultured cells in flask into a 50 mL
Preparation conical tube. Remove 100 μL of the cell suspension and com-
(in Biohazard Hood bine them with 100 μL of trypan blue for the initial cell count.
with Sterile Technique) Spin 50 mL conical tube at low speed (400 g) for 5 min.
2. Pour off the supernatant into waste, and add 10 mL of room-
temperature SM to resuspend the cells. Centrifuge for 5 min at
400 g.
3. Remove the supernatant, and add the chromium solution equal
to 100 μCi, for each 20 106 cell mix. The amount of 51Cr
depends on weeks of “age” (time elapsed since the date of
calibration of the chromium solution, see Note 2). Incubate
for 1 h at 37 C in a humidified 5% CO2 atmosphere, with gentle
shaking every 15 min.
4. In the same 50 mL tube, add 10 mL of 37 C AM to the
chromium/cell solution, mix gently, and spin at 400 g for
5 min. Pour off the supernatant into the radioactive waste
container. Repeat this washing four times.
5. After the fourth wash add 10 mL of 37 C AM. Remove 50 μL
of cell/assay media solution to a test tube; add 50 μL trypan blue
for the final cell count in order to determine the volume of assay
media needed to produce a cell count of 2 106 K562 cells.
With remaining solution in the 50 mL tube, centrifuge for 5 min
at 400 g.
4.2 NKCC Reaction 1. Pour off supernatant from E above, add the volume of room-
Setup temperature AM proscribed by the final cell count, and mix
gently. In separate tubes make graded dilutions to produce the
four target cell concentrations as shown in Table 3. An assay
with four blood samples requires a minimum of 2 mL of each
target cell dilution.
2. Add 150 μL well-mixed blood from EDTA tube to each well of
one row in culture plate.
3. For spontaneous-release control (SR), dispense 150 μL AM to
one row of 12 wells.
4. For total-release control (TR), dispense 150 μL 1% Triton
X-100 to one row of 12 wells.
5. Add 50 μL cell suspensions to each of the three wells for each of
the four dilutions (change tips if progressing from stronger to
weaker concentration; otherwise tips can be conserved for one
row).
6. Cover plate and centrifuge for 10 min at room temperature at
400 g.
Table 3
Dilution of target cells
Final concentration Amount of initial cell suspension (2 106 cells/mL) Amount of media
2 10 cells/mL
6
2 mL 0 mL
1 10 cells/mL
6
1 mL 1 mL
0.5 106 cells/mL 0.5 mL 1.5 mL
0.25 10 cells/mL
6
0.25 mL 1.75 mL
7. Incubate for plate at 37 C in a humidified 5% CO2 atmosphere

for 4 h.
8. After 4 h add 100 mL of cold (below 9 C) assay media to all
wells to stop the NKCC reaction, and centrifuge for 5 min at
400 g.
4.3 Harvesting 1. Transfer 100 μL supernatant to counting test tubes without

and Counting disturbing cell pellet. Start with the highest concentration of
TR controls, followed by SR controls and finally patient samples.
After the last well insert at least three empty test tubes to serve as
background.
2. Load the racks into gamma counter and count released 51Cr.
4.4 Lymphocyte This information is required for calculation of NKCC using Eq. (1).
Count This should be done using an electronic hematology analyzer,
which will provide the hematocrit that is also needed for Eq. (1).
4.5 Flow Cytometry Lymphocytes that are CD45+, CD14, CD3, and CD56+ con-
stitute the bulk of the cells capable of NK cell cytotoxic activity.
Using four-color flow cytometry, determine the percent of cells in
the lymphocyte gate that meet this requirement. Multiply this by
the lymphocyte count to obtain the number of NK cells for Eq. (1).
4.6 Calculations Subtract the background counts (b) from all experimental release
(ER), SR, and TR and calculate the percent cytotoxicity for each
dilution of target cells according to Eq. (1):
2 n o 3
ðER b Þ V t ðVVb HCT Þ
ðSR b Þ
% cytotoxicity ¼ 4 5 100
t
ðTR b Þ ðSR b Þ
ð1Þ
Vt is the total volume in the well; Vb is the volume of blood in

the well; HCT is the hematocrit of blood sample.
When a constant number of NK cells are incubated with various

numbers of 51Cr-labled target cells, numbers of target cells lysed
may be expressed as percent cytotoxicity times the number of target
cells in the assay. A curve is generated by plotting the number of
target cells versus the number of cells lysed. Because the kinetics of
cytotoxicity resemble those of enzyme-substrate interactions, the
Michaelis–Menten rate equation as developed by [18] defines the
velocity of such reactions (Eq. 2):
V max ½T
v¼ ð2Þ
K m þ ½T
where v is the number of target cells lysed; T is the number of

target cells in the assay; Vmax is the number of target cells lysed
when the number of target cells is infinite; and Km is the number of
target cells required for ½ Vmax.
Callewaert and Mahle [19] showed that estimates for Vmax for
NKCC are equal to the concentration of NK cells times the mean
lytic activity per NK cells. Percent NKCC for the four concentra-
tions of target cells is transformed to the number of target cells
killed at each concentration using Eq. (3):
v ¼ %cytotoxicity ½T ð3Þ
The data are fitted to the Cleland equation. Percent cytotoxic-
ity is determined for effector cell-to-target cell ratio of 1:1, where
the number of effector cells is defined as CD3-CD56+
lymphocytes.
4.7 Quality Control There is no commercial proficiency test available for this procedure.
for Assay For quality control, always perform the assay in triplicate and have
blinded duplicates submitted to the laboratory. The SR should be
<20% of TR and the correlation coefficient for the Michaelis–Men-
ten equation should be >90%. Each laboratory must determine an
expected range for healthy individuals. In our laboratory, the
230 healthy controls shown in Fig. 1 had a mean % cytotoxicity at
an effector-to-target cell ratio of 1:1 of 28%. Table 4 gives the
parameters of NKCC for controls and a patient group, CFS.
4.8 Performance Outlying individual scores when found in triplicate data sets (i.e.,
Parameters from three wells from the same tube) should be excluded, and the
scores which are closest to one another in the set should be aver-
aged for results.
4.9 Limitations This procedure is done with fresh blood samples. Blood that is <8 h
of the Procedure old is preferred. Samples that are <8 but >24 h can be run.
However, their results must be compared to the control range
determined for samples of that age.
Table 4
The parameters of NKCC for controls and a patient group CFS
Natural killer cell cytotoxicity for chronic fatigue syndrome and healthy
controls Statistic Std. error
CFS Mean 15.94688 0.858119
95% Confidence interval for mean Lower bound 14.25328
Upper bound 17.64047
Median 12.05000
Variance 129.601
Std. deviation 11.384233
Minimum 1.200
Maximum 55.300
Range 54.100
Interquartile range 13.675
Skewness 1.392 0.183
Kurtosis 1.718 0.364
HC Mean 27.83922 0.808923
95% Confidence interval for mean Lower bound 26.24533
Upper bound 29.43310
Median 28.00000
Variance 150.502
Std. deviation 12.267920
Minimum 2.000
Maximum 69.000
Range 67.000
Interquartile range 17.025
Skewness 0.211 0.160
Kurtosis 0.033 0.320
5 Notes
1. Splitting K562 Cell Cultures

Note: In labs performing the NKCC assay on a daily or weekly
basis, the existing cell line of K562 is cultured and continually
split using the following procedure which is repeated iden-
tically every week, with the exceptions of long weekends or
periods when the test is not performed. Tissue culture flasks
should always be labeled with the date when they were
prepared, written on them.
(a) On Monday: Label four flasks with the date, “K562,” and
the volumes included (for example, if 2 mL of cells and
16 of culture media are added, label “2 + 16 media”).
(b) From the incubator, remove one of the tissue culture
flasks from the previous Friday (should be dated), and
withdraw 4 mL and put into a flask labeled for Monday’s
date. Repeat. Then withdraw 2 mL from the same flask
from Friday, and add this to a third Monday flask. Repeat

for a fourth Monday flask.
(c) Complete each of these flasks to 20 mL with culture
media.
(d) Place the newly composed flasks in the incubator, and
discard the Friday one from which transfers were made.
(e) On Tuesday: Use the remaining flask from the previous
Friday for any NK samples to be done on Tuesday
(no splitting on Tuesday).
(f) On Wednesday: Read fully: Repeat the procedure from
Monday, except using one of Monday’s flasks for transfers,
and label the new flasks with Wednesday’s date. Note,
also, that you will this time only compose one flask using
2 mL from the Monday culture, instead of two. The other
two new flasks will be 4 mL additions (complete to 20 mL
with culture media).
(g) On Thursday: Use the remaining flask from Monday for
any NK samples to be done on Tuesday (no splitting on
Thursday).
(h) On Friday: Split one of Wednesday’s flasks to four new
Friday flasks, with 1.5 mL going to two of them, and
.5 mL going to two other new Friday flasks. Complete
to 20 mL with culture media.
(i) Repeat this process weekly. If necessary to do a Saturday
NK, make an additional 2 mL flask on Wednesday.
2. Radioactive Chromium-51 Usage
For the addition of chromium in Step 1C, different volumes of
the chromium solution are added depending on the “age,” or
time elapsed since the date of calibration of the chromium
solution. The calibration date of the particular solution vial is
on the label.
Week 1 90 mL of chromium (precalibration days)

Week 2 100 mL of chromium calibration date
Week 3 120 mL of chromium post-calibration days
Acknowledgments
This work was supported by grants from the NIAAA:

R21AA016635 (PI MA Fletcher); NIAID: R01AI065723 (PI MA
Fletcher); CFIDS Assoc. of America: (PI N Klimas); NIAID: UO1
AI459940 (PI N Klimas); NIAMSD R01 AR057853-01A1 (PI N
Klimas); VA merit awards (PI N Klimas) Dynamic Modeling in GWI,
CFS/ME; and GW080152 (PI N Klimas).
References
1. Herberman RB, Holden HT (1979) Natural physicians in Japan: health assessment by
killer cells as antitumor effector cells. J Natl immune variables (CD4, CD8, CD56, and
Cancer Inst 62:441–445 NK cell activity) at the beginning of work. J
2. Solomon GF, Benton D, Harker JO, Occup Health 50:136–146
Bonavida B, Fletcher MA (1994) Prolonged 11. Fletcher MA, Zeng XR, Maher K, Levis S,
asymptomatic states in HIV-seropositive per- Hurwitz B, Antoni M, Broderick G, Klimas
sons with fewer than 50 CD4+ T cells per NG (2010) Biomarkers in chronic fatigue syn-
MM3. Psychoneuroimmunologic findings. drome: evaluation of natural killer cell function
Ann N Y Acad Sci 741:85–90 and dipeptidyl peptidase IV/CD26. PLoS One
3. Su DM, Vankayalapati R (2010) A new avenue 5(5):e10817
to cure cancer by turning adaptive immune T 12. Maher K, Klimas NG, Fletcher MA (2005)
cells to innate immune NK cells via reprogram- Chronic fatigue syndrome is associated with
ming. J Mol Cell Biol 2:237–239 diminished intracellular perforin. Clin Exp
4. Ironson G, Wyings C, Schneiderman N, Immunol 142:505–511
Baum A, Rodriquez M, Greenwood D, 13. Alter G, Malenfant JM, Altfeld M (2004)
Benight C, Antoni M, LaPerriere A, Huang CD107a as a functional marker for the identi-
H-S, Klimas NG, Fletcher MA (1997) Post- fication of natural killer cell activity. J Immunol
traumatic stress symptoms, intrusive thoughts, Methods 294:15–22
loss and immune function after Hurricane 14. Brunner K, Mauel TJ, Cerottini J-C, Chapuis B
Andrew. Psychosom Med 59:128–141 (1968) Queantitative assay of the lytic action of
5. Schedlowski M, Jacobs R, Stratmann G, immune lymphoid cells on Cr labeled alloge-
Richter S, H€adicke A, Tewes U, Wagner TO, neic target cells in vitro Inhibition by isoanti-
Schmidt RE (1993) Changes of natural killer body and by drugs. Immunology 14:181
cells during acute psychological stress. J Clin 15. Baron GA, Klimas NG, Fischl MA, Fletcher
Immunol 13:119–126 MA (1985) Decreased natural killer cell-
6. Segerstrom SC, Miller GE (2004) Psychologi- mediated cytotoxicity per effector cell. Diagn
cal stress and the human immune system: a Immunol 3:197–204
meta-analytic study of 30 years of inquiry. 16. Fletcher MA, Baron GA, Ashman MA, Klimas
Psychol Bull 130:601–630 NG (1987) Use of whole blood methods in
7. Klimas NG, Rosenthal M, Fletcher MA (2006) assessment of immune parameters in immuno-
Immune effects of an acute exercise challenge deficiency syndromes. Clin Diag Immunol
in Gulf War illness. Assoc Military Surgeons 5:69–81
US, San Antonio, TX 17. Bourin P, Mansour I, Doinel C, Roué R,
8. Harvey JM, Barnes Z, Zeng X, Sol C, Fletcher Rouger P, Levi F (1993) Circadian rhythms of
MA, Klimas NG (2011) Exercise effects on circulating NK cells in healthy and human
biomarkers of CFS/ME and GWI. Presented immunodeficiency virus-infected men. Chron-
to the International Conference on CFS/ME, obiol Int 10:298–305
Ottawa, Canada 18. Cleland WW (1967) The statistical analysis of
9. Cohen S, Tyrrell DA, Smith AP (1991) Psy- enzyme kinetic data. Adv Enzymol 29:1
chological stress and susceptibility to the com- 19. Callewaert DM, Mahle NH (1985) Kinetic
mon cold. N Engl J Med 325:606–612 models of natural killer cells and their use for
10. Okamoto H, Tsunoda T, Teruya K, Takeda N, studying activated NK cells. In: Herberman
Uemura T, Matsui T, Fukazawa S, Ichikawa K, RB, Callewaert DM (eds) Mechanisms of cyto-
Takemae R, Tsuchida K, Takashima Y (2008) toxicity by NK cells. Academic Press, Orlando,
An occupational health study of emergency FL, p 381
Chapter 13
Mouse Testing Methods in Psychoneuroimmunology 2.0:

Measuring Behavioral Responses
Albert E. Towers, Jason M. York, Tracy Baynard, Stephen J. Gainey,
and Gregory G. Freund
Abstract
The field of psychoneuroimmunology (PNI) aims to uncover the processes and consequences of nervous,
immune, and endocrine system relationships. Behavior is a consequence of such interactions and manifests
from a complex interweave of factors including immune-to-neural and neural-to-immune communication.
Often the signaling molecules involved during a particular episode of neuroimmune activation are not
known but behavioral response provides evidence that bioactives such as neurotransmitters and cytokines
are perturbed. Immunobehavioral phenotyping is a first-line approach when examining the neuroimmune
system and its reaction to immune stimulation or suppression. Behavioral response is significantly more
sensitive than direct measurement of a single specific bioactive and can quickly and efficiently rule in or out
relevance of a particular immune challenge or therapeutic to neuroimmunity. Classically, immunobehavioral
research was focused on sickness symptoms related to bacterial infection but neuroimmune activation is
now a recognized complication of diseases and disorders ranging from cancer to diabesity to Alzheimer’s.
Immunobehaviors include lethargy, loss of appetite, and disinterest in social activity/surrounding environ-
ment. In addition, neuroimmune activation can diminish physical activity, precipitate feelings of depression
and anxiety, and impair cognitive and executive function. Provided is a detailed overview of behavioral tests
frequently used to examine neuroimmune activation in mice with a special emphasis on pre-experimental
conditions that can confound or prevent successful immunobehavioral experimentation.
Key words Mouse, Maze, Exploration, Brain based, Biobehaviors, Memory, Motor activity,
Anhedonia
1 Introduction
Since its inception as an interdisciplinary field of science in the

1970s, behavior has been an integral part of psychoneuroimmunol-
ogy (PNI). Indeed, PNI is generally defined as the study of the
interactions between behavior, neural, immune, and endocrine
system functions [1, 2]. Behavior can be, and is largely, used to
assess whether a particular stimuli or experimental treatment has
the potential to activate the neuroimmune system [1, 2]. An impor-
tant concept to recognize in PNI is the bidirectionality that exists
221
222 Albert E. Towers et al.
between the nervous system and the immune system [2]. This is to
say that both neural-to-immune and immune-to-neural signaling
occur. Shared pathways exist between the nervous and immune
systems that use a repertoire of signaling molecules such as cyto-
kines and neurotransmitters [3] that are capable of interacting with
both immune and nervous system cells. These bioactives can con-
vey the state of peripheral immunity to the neuroimmune system,
communicate the status of neuroimmunity to the peripheral
immune system [3–5], and provide immunoactivating and deacti-
vating signals to immune cells throughout the body [6].
Observation of innate immune-mediated behavioral change
(immunobehaviors) is largely used as a method of measuring neu-
roimmune activation in response to pathogenic insults of infectious
[6] or non-infectious [7] etiologies. Sickness behavior, in a classical
sense, is a set of coordinated behavioral changes aimed at conserv-
ing and redirecting body energy stores toward combating illness
and promoting recovery [4, 8]. Immunobehaviors are best known
for their manifestation in association with bacterial infection [8],
but materialize in a spectrum of conditions and diseases including
cancer [8], autoimmune disorders [9], wounding [10], depression
[11], obesity [12], and neurodegenerative diseases [13]. In any
circumstance in which the innate immune system is activated,
peripheral inflammatory mediators can impact the brain, altering
normal function and causing symptoms of illness/loss of well-being
[4, 8]. Typical sickness behavior symptoms include reduction in
food intake, lethargy, malaise, loss of interest in social and/or
environmental surroundings, changes in sleep patterns, and
impaired cognition [4, 8, 11]. Furthermore, continued or dysre-
gulated activation of the neuroimmune system can progress beyond
acute sickness symptoms and transition to behaviors observed in
the anxious or depressed [11]. Fatigue is often a lingering compli-
cation of neuroimmune activation [6] and can present as purely
mental or physical but, most commonly, in a combinational form
[14]. In rodents, exercise behaviors like spontaneous wheel run-
ning (SWR) [15] are helping to unravel the complex biology of
physical fatigue. Tests examining memory formation (learning) and
memory recall, as well as those capable of measuring decision-
making, are being used to explore mental fatigue [16] and
impairment [17].
Finally, immunobehaviors are a powerful indicator of neuroim-
mune status offering insight into the pro-inflammatory milieu of
the brain. Altered behavior manifests prior to detectable changes in
brain-based bioactives and lingers past their resolution. Such con-
ditions indicate that the brain is very sensitive to small perturba-
tions and that traditional chemical bioassays are often not sensitive
enough or appropriately targeted to detect brain-based dysfunc-
tion. Hence, use of behavioral testing provides highly sensitive and
phenomenologically relevant information in regard to brain
Mouse Testing Methods in Psychoneuroimmunology 2.0: Measuring Behavioral. . . 223
function but lacks significant specificity from a mechanistic stand-

point. This is either due to an evolution-derived paucity of immu-
nobehavioral phenotypes or a current knowledge/technical
deficiency in the ability to parse such behaviors of biologically
relevant subsets. From a clinical standpoint, immunobehaviors are
critical to understanding all aspects of human condition. Thus,
determining behavioral end points ensures that parallel mechanistic
studies are grounded in phenotypic relevance. In this review, meth-
ods for measuring sickness and depressive/anxietal, cognitive, and
physical activity behaviors in mice are described. The tests were
chosen based on common usage and validated outcomes.
2 Pre-experimental Considerations
Behavior is a valuable tool for gauging the presence, severity, and

duration of innate immune activation. Prior to behavioral experi-
mentation, preparatory procedures are required so that meaningful
and repeatable results are obtained. Mice, like most animals used
for laboratory research, are responsive to the environment. Consis-
tency and reproducibility of results depend on decisions made and
precautions taken before initiation of testing. While the following
does not account for every possible pre-experimental housing and
husbandry scenario, it does seek to articulate and define significant
areas of pre-experimental bias. Every animal facility, like every
laboratory, is unique with differences obvious and subtle. Standar-
dizing and controlling for clear confounds related to mouse strain,
sex, and scientific model are key. Maintaining housing and hus-
bandry practices that support animal behavioral well-being and
physical health are essential. Finally, the following is not intended
to be an absolute guide for “correct” mouse immunobehavioral
experimentation, for it is critical that the investigator identify,
develop, and hone best practice related to their particular field. In
addition, a firm eye on federal laboratory animal regulations and
local institutional rules and guidelines is paramount.
2.1 Model and Strain A first consideration should be the animal model and strain chosen.
Choice The vast majority of PNI behavioral testing utilizes rats and mice.
Porcine models are becoming more prevalent [18–20], as have
other types of rodents, especially prairie voles [21]. In addition
zebra fish are emerging as a high-throughput model for many
behaviors, especially anxiety-like behaviors [22]. Mice are particu-
larly useful in neuroimmune and immunobehavioral research due
to their ability to reproduce and mature rapidly, and the relative
ease to which genetic modification can be applied through muta-
tional, transgenic, and knockout approaches [23]. Different behav-
ioral phenotypes exist between strains. Therefore, it is important to
be aware of and control for potential inter-strain and inter-sub-
strain variances [23]. This includes intra-strain variation between

mice raised/housed by disparate commercial venders and institu-
tional facilities. Furthermore, genetically altered/modified mice are
especially prone to immunobehavioral impairment and should be
selected with such in mind. With genetically altered/modified
mice, diligent baseline testing in comparison to wild-type animals
is essential to ensure that the behavior observed is caused by the
state and not by the trait of the animal [24]. Animal choice can also
trigger technological barriers. Certain behavioral software packages
utilize color-based tracking. Thus data analysis can fail if the animal
is indistinguishable from the background (e.g., white mice on light-
colored bedding or black mice in a dark bucket of water).
2.2 Sex Differences Depending on strain, genetic alteration/modification, and behav-

ioral test, male and female mice perform differently. For example, in
the elevated plus maze (EPM) females exhibit less general activity
[25]. When female mice are compared to male mice in a freely
explorable open-field arena, female mice are less active, less willing
to leave their home cage to enter the open field, and less likely to
explore the open field [26]. Female mice generally run for shorter
distances in the voluntary wheel running paradigm [27] and run
different distances depending on their current state in the estrous
cycle [28]. Thus, regardless of identical housing and husbandry
practices, care must be taken in mixing sexes during behavioral
testing. Such care can be especially frustrating when using geneti-
cally altered/modified animals due to in-house breeding deficien-
cies and difficulties in acquiring adequate numbers of similarly aged
animals from commercial suppliers. Importantly, current NIH
guidelines underscore the need for comparison so as to better
understand sex differences and to limit the promulgation of sex
bias [29].
2.3 Age Natural aging modulates immune function with subject variation
dependent on factors such as lifelong physiological stressor expo-
sure [30]. It comes with little surprise that immunobehaviors are
also affected by age. While sickness symptoms appears beneficial to
young mice, they may be maladaptive in older mice [31]. Age can
also negatively impact physical activity since aged mice run less
[32]. Young mice can be difficult to analyze in running tests, as
well, due to a progressive increase in distance traveled [33]. There-
fore, investigators need to consider confounds related to non-age-
matched experimental animals.
2.4 Transportation Environmental factors play a significant role in how mice behave
and respond to neuroimmune stimuli and immunobehavioral treat-
ments. Mice should be allowed a transition phase to acclimatize to a
new environment, whether a procedure/experimentation room
itself [34, 35], or the housing room following arrival from an
outside supplier [34, 36]. Transportation stress increases plasma

corticosterone levels in mice regardless of travel duration, and up to
48 h of acclimation is required for corticosterone to return to
pre-transportation values [35]. Of note, behavioral change in
response to transport has been shown to persist for 4 days [35],
while loss of body weight bounces back to pre-transport levels
within 4 days [34, 36]. Therefore, 5 days of acclimation is generally
recommended as a minimum for mice undergoing behavioral test-
ing following off-site transit [36]. A minimum of 24 h of acclima-
tization time following on-site transit (i.e., between housing and
experimental rooms) is also recommended [36]. Longer times for
acclimation may be necessary depending on the type of mice used,
duration of transport, and magnitude of difference between where
the mice were housed and where they are tested [34, 36]. As with
any experimental test, validation studies are recommended to con-
firm pre-experimental choice decisions.
2.5 Light Cycle Alteration of light cycle impacts natural murine behavioral patterns
[36], and immunobehaviors (e.g., anxiety) [37]. Mice are under
control of a genetically driven circadian clock that serves to regulate
physiological and behavioral processes in a diurnal fashion
[38]. Therefore, it is important to ensure that experimental
rooms and housing rooms have similar light cycles. In addition, it
is advisable to initiate behavioral experiments at the same time of
“day,” especially when performing repeat testing. Mice are noctur-
nal; thus most testing should occur during the dark cycle. Reverse
light cycle housing is advantageous so as not to put undue burden
on personnel performing the behavioral tests. Mice are crepuscular
[39], with heightened activity during the early (dusk) and late
(dawn), and should be tested during these peak activity times.
Methods for determining the timing of these active periods are
described in Subheading 3.2.
2.6 Temperature/ Temperature should be comparable across animal housing facilities,

Humidity as it is regulated by the Office of Laboratory Animal Welfare
(OLAW) in the United States [40]. According to the Guide for
the Care and Use of Laboratory Animals, mice should be housed in
a room/environment with temperatures ranging from 68 to 79 F
(20–26 C) [40]. Mice show neuroimmune and behavioral sensi-
tivity to both heat [41] and cold stress [42] indicating the potential
for altered behaviors with temperature fluctuations. Relative
humidity should also be maintained at similar levels across animal
housing facilities, and the Guide for the Care and Use of Labora-
tory Animals indicates a range of 30–70%. Cage style, construction,
bedding, and enrichment materials as well as housing density affect
temperature and humidity within the cage microenvironment
[40]. It is therefore important to recognize that room temperature
and humidity may not necessarily reflect intra-cage temperature and
humidity, depending on housing factors.
2.7 Noise Noise has the potential to activate neuroimmune signaling path-
ways and alter behavior, as bell ringing [43] and noise produced by
vacuuming [44], as example, stress laboratory mice by activating
the hypothalamic-pituitary-adrenal (HPA) axis [45, 46]. White
noise generators that create a constant background are useful in
reducing behavioral responses to sudden loud noises [47].
2.8 Odors Mice use odors for communication, marking territory, and in rec-
ognition signaling [36]. In addition to using patterns of urine
deposition for communication, mice produce specialized odors
via several glandular secretions [36]. Evidence indicates that neu-
roimmune activation alters odor production, which, in turn, pre-
cipitates behavioral change. As example, mice administered
lipopolysaccharide (LPS) generate olfactory cues to indicate that
they may have a transmissible pathogen. This causes healthy cage
mates to socially withdraw from “sick” mice [48]. A similar phe-
nomenon is seen in healthy mice housed with tumor-bearing cage
mates [49]. Finally, exposure to foreign/strange odors (e.g.,
human-associated odors) can result in stress responses [36]. Unfor-
tunately, no specific research has been performed examining the
duration of olfactory stress responses in mice, nor is there an
identified acclimation or exhaustion time for evocative scents. It
should also go without mention that eliminating as many olfactory
cues in/on rooms, cages, and experimental equipment (e.g.,
through use of 70% v/v ethanol) is best practice. Mouse handler
odors (i.e., perfumes and predator scents (e.g., cat odor)) should be
minimized and/or eliminated.
2.9 Handling Physical handling of mice is a well-studied modifier of mouse

physiology and behavior [50]. Therefore, it is advisable to handle
mice at least daily to acclimate them to human physical contact.
Handled mice respond more consistently to experimental treat-
ment and succumb less to handling stress [51]. Interestingly, the
handling method used appears important. Mice respond to
handling more readily when removed from their cage passively,
such as with a tube or cupped hands, as opposed to the more
traditional removal by grasping the base of the tail with the
thumb and forefinger or soft forceps [52]. Hurst and West noted
that mice develop a consistent response (measured as voluntary
interaction with handlers) by the ninth day of a single 60-s handling
session [52]. Of note, mice removed from housing by the base of
the tail had a lower level of voluntary interaction when compared to
mice removed by the passive tube and cupping methods [52]. Fur-
thermore, mice acclimated by tube or cupping methods were less
impacted by scruff restraint, when compared to mice acclimated by
base of the tail grabs [52]. Thus, a period of at least 9 days of daily
handling appears sufficient to promote a reduction in aversive
behavior in mice. If needle sticks are a component of a behavioral
experiment, a passive method of mouse cage removal is favored.
2.10 Housing Mice are social animals and should be housed, if possible, with
Method/Environmental other mice [40]. Several studies have investigated the impact that
Enrichment individual housing/social isolation has on neuroimmunity and
immunobehaviors [53]. In general, social isolation induces aggres-
sion in male mice [54]. Individually housed male mice also appear
prone to developing anxiety and depressive-like behaviors follow-
ing unpredictable chronic mild stress [55] despite their increased
propensity to explore [26, 35]. Group-housed mice order them-
selves into a social hierarchy, with 1–2 dominant mice and several
subordinate mice. Subordinates as well as dominant mice show
similar exploratory behavior in an open-field context [26] but this
seems dependent on the relatedness of the group-housed mice. As
evidence, an introduced non-sibling-dominant intruder mouse
evokes social stress and immune cell glucocorticoid resistance in
the group-housed mice [56]. Therefore, housing mice in groups at
a density of one 25–30 g mouse per 77.4–96.7 cm2 of cage floor
area by 12.7 cm of cage ceiling height is advantageous [40].
Unintuitively, clean cage environment [36, 57] and novel cage
construction materials can reduce mouse welfare and cause aberrant
behaviors [36, 40]. Introduction to a clean/novel cage increases
plasma corticosterone in mice, and increases physical activity in the
first 24 h of exposure. Metal cages, as opposed to plastic cages [40],
are colder to the touch, more conducive to noise generation, and
less permeable to light [36, 40]. Additionally, the use of solid
flooring with absorbent bedding is recommended by most institu-
tional care and use committees because wire mesh flooring can lead
to paw injury [40]. Such injury can confound behavioral experi-
ments by triggering the innate immune and pain systems [6].
Environmental enrichment is thought to enhance mouse well-
being by providing motor and sensory stimulation. Environmental
enrichment may include nesting material, structures, and/or shel-
ters within the cage [40]. Lack of environmental enrichment dam-
pens mouse reactivity and alertness in many behavioral tests
[58]. Environmental enrichment is, however, somewhat strain
dependent because the loss of reactivity and alertness mentioned
earlier observed in BALB/c mice but not in C57BL/6 mice. In
fact, Van de Weerd et al. concluded that male BALB/c mice housed
in enriched environments are anxious [58]. Olsson and Dahlborn
noted that changing a barren cage environment by placing objects
within it does not necessarily lead to “enrichment” [59]. Instead, it
is important to measure behavioral and physiological changes that
occur in enriched-housed animals, and if said enrichments result in
long-term improved health and/or well-being. Some “environ-
mental enrichers” are felt to result in stress and/or anxiety
[58, 59]. Thus, the environment within the cage may be as impor-
tant as the environment in which the cage is housed when establish-
ing appropriate pre-experimental procedures.
3 Sickness Behaviors
Sickness behavior is classically defined as the nonspecific set of

symptoms associated with the body’s response to innate immune
challenge [6]. Symptoms of sickness behavior include anorexia,
fatigue, malaise, reduced locomotor activity, loss of interest in
environmental and social surroundings, and disappearance of
body-care activity [6, 60]. These changes occur in response to
brain-based increases in the pro-inflammatory cytokines interleu-
kin-1β (IL-1β) and tumor necrosis factor-α (TNF-α), resulting in
changes in the motivational state of an organism [6, 11]. Sick
individuals also experience psychological symptoms such as anxiety
and depression, and cognitive deficits such as learning and memory
loss. Frequently used methods for quantifying sickness behavior
include social exploratory activity [61], locomotor activity
[4, 62], food disappearance [4, 62], and rotarod testing [63, 64].
3.1 Social The paradigm of social exploration appears to have evolved from
Exploration of a Novel work done by Thor and Holloway showing that adult laboratory
Juvenile rats actively investigate and form social memories of conspecifics
[65], principally via anogenital sniffing, nosing, mutual grooming,
and close following [66]. Rodents do not show behavioral habitu-
ation during social investigation, provided that the conspecific
juvenile is novel at each presentation into the home cage of the
adult [67]. Furthermore, adult male rats do not exhibit aggression
toward prepubertal male juveniles but do toward unrelated post-
pubertal males. This is a function of androgen-related odors from
the postpubertal rat, which can elicit aggressive attacks from the
adult due to unfamiliarity [65]. These aspects of social recognition
first allowed Dantzer et al. to show that social memory could be
modulated by neurological peptides [61], and were likely the first
experiments that used social exploration as a tool to measure the
effects of neuroactive compounds. Social exploratory behavior was
adapted from social memory testing by using a different juvenile at
each observation time point [68]. Due to a lack of habituation
when using a novel juvenile, social exploration is routinely used as
a sensitive test of immunobehavioral perturbation. Bluthé et al.
adapted the rat-based Dantzer procedure to mice [66].
3.1.1 Procedure All observations should be made during the dark/active cycle of the
mice. Mice should be housed individually at least 24 h prior to the
initial measurement and be allowed to acclimatize to the procedure
room (if procedure room differs from housing room) for the time
period in which they are individually housed. In some studies,
infrared light is used with the aid of infrared or night-vision capable
cameras [66, 68], but use of red-tinted lighting is also acceptable, as
mice only have limited ability to detect light from the red portion of
the visible spectra [40]. All observations of adult–juvenile interac-

tion are video recorded during social exploration testing for later
analysis. Within each observation/recording session, each adult
mouse observed (subject mouse) is only exposed to a juvenile
(challenge mouse) once. Social exploration is initially measured
immediately prior to any experimental treatments. This serves as a
baseline of social exploratory activity for each subject mouse. Social
exploration is subsequently measured at specific times following
experimental treatment, usually every 2 h for the first 12 h post-
treatment, and then every 12 h thereafter [66, 68]. A test session/
observation time point consists of introducing a novel challenge
mouse (conspecific juvenile mouse of the same sex) into the home
cage of the subject mouse for 5 min before returning the juvenile to
its own home cage [66, 68, 69]. In some instances, the juvenile
challenge mouse is housed in a clean 7.62 cm 7.62 cm 7.62 cm
wire mesh enclosure (baby cage) when introduced into the home
cage of the subject mouse [70]. This avoids undesired aggression
toward the juvenile seen with certain mouse strains like C57BL/6
mice. If a baby cage is used, it is best to acclimate the subject mice to
the wire mesh enclosure by including it in their home cage during
single housing. Once the experiment is completed, the duration
(in s) of exploration investigation of the challenge mouse by the
subject mouse is recorded from analysis of the video record [66, 68,
70]. Preferably, analysis of the video record should utilize auto-
mated video tracking software [71] such as that developed by
Noldus Information Technology (Leesburg, VA). In general, track-
ing software eliminates observer bias and provides more consistent
results. When using tracking software, care must be given as to
mouse color vs. background color so that the animal is easily
distinguished. If tracking software is not available, a trained
observer blinded to the treatment groups (blinded trained
observer) can manually quantify social exploration without signifi-
cant prejudice [66, 68]. Observer training for behavior testing is
best accomplished by having personnel new to scoring evaluate
previously analyzed videos that have been scored and validated.
When the scoring skills of the trainee observer are within 10% of a
validated reviewer in three consecutive evaluations of a novel vali-
dated video, the trainee is deemed proficient. This same training
should be followed when educating personnel to the use of auto-
mated tracking software. During manual video review the blinded
trained observer uses a stopwatch to record the time of interaction
initiated by the subject mouse with the challenge mouse through-
out the 5-min designated investigational period. Social exploration
is considered to be subject mouse to challenge mouse investigation
(not the opposite), including anogenital sniffing, nosing, following,
and grooming. With the use of a baby cage, nose-to-cage contact is
considered exploration. Social exploration is typically shown in a
graphical display using either raw seconds of exploration at each
time point [66] or as percent control or percent baseline [70].
3.2 Locomotor Since lethargy is a core symptom of sickness behavior, locomotion

Activity [60] can be used as a technically easy and high-throughput measure
of sickness. Spontaneous locomotor activity is advantageous in that
it can be assessed without moving mice from their home cage, and
automated video tracking software can be easily used for analysis. In
addition, wide-screen video capture allows for several mice to be
observed at once. Alternatively, specifically designed activity cham-
bers (Versamax from AccuScan Instruments, Columbus, OH) with
built-in infrared beam detection systems can be used [72] allowing
for real-time analysis. Such testing platforms, however, introduce
an element of novelty.
mice. Mice should be housed individually at least 24 h prior to the
initial measurement and be allowed to acclimatize to the procedure
room (if procedure room differs from the housing room) for the
time period in which they are individually housed. As with social
exploration, infrared or red lights are used to provide illumination.
At each time point of interest, mouse movement is recorded for
5 min [12], with a camera placed over the center of a single cage or
a grouping of cages. If multiple mice are being recorded during a
given observation point, they should be shielded from one
another’s view. Movement including total distance traveled, veloc-
ity, and time spent moving is best determined from the video record
using automated tracking software. However, if tracking software is
not available, videos can be hand scored by a blinded trained
observer. To score manually, a thin-line grid comprised of six
equally sized rectangles is affixed to the television or monitor screen
directly over the cage and the blinded trained observer counts the
number of times the mouse crosses a line (line crossing) through-
out the 5-min designated investigational period. A mouse is only
considered to have line crossed if both fore and hind limbs cross a
line [69]. A more powerful method for assessing mouse locomotor
activity is through long-duration (hours–days) tracking [73]. This
is required when detailing mouse crepuscular movement. While
video recording can be used for such evaluation, the data collection,
storage, and interpretation can be burdensome due to video file
sizes. Therefore, biotelemetry is a useful method for this type of
testing [74]. In this method, a biotelemetric emitter is surgically
implanted within the peritoneum of a mouse and a receiver pad
linked to a PC running data collection software is placed directly
underneath the home cage of the implanted mouse (Mini Mitter,
Bend, OR). Mouse movement is tracked and recorded automati-
cally. Specific procedures and training for this method should be
provided by the manufacturer of the device chosen.
3.3 Food Food consumption (or disappearance, as discussed below) is an

Consumption indicator of sickness, as individuals experiencing illness often
exhibit anorexia. Food consumption gives an indication of whether
anorexia is present, which could further be used to indicate if
sickness behavior is occurring [4, 62]. Food consumption can be
measured in at least two different ways, as outlined below. The first
uses “food disappearance,” which is often interpreted/estimated as
food consumption.
3.3.1 Food Mice should be individually housed as per social exploration at least
Disappearance Procedure 24 h prior to experimentation. With single housing, food should be
moved from the overhead cage food hopper and placed in an 8 cm
diameter 5 cm stainless steel bowl or glass dish in the cage. Steel
and glass are preferred over ceramic because ceramic containers can
absorb water if they are not completely glazed. In addition, steel
dishes can be magnetically secured to the cage bottom or side with
the use of a strong magnet. This prevents mice from tipping the
bowl and spilling food which can easily occur with plastic bowls.
After the acclimation period, and just prior to initiation of testing,
new food should be added to the bowl and the bowl weighed. This
process should occur at the very beginning of the dark/active cycle
of the mice in order not to disturb mice during their sleep cycle.
Food disappearance is measured by weighing the bowl plus food at
fixed intervals, such as every 24 h. For longer term experiments,
food can be re-added to the bowl and reweighed [62]. The term
food disappearance is used in place of food consumption because
not all food is ingested. Some food inadvertently falls in the cage
bedding [75]. Mice should be individually housed as per social
exploration at least 24 h prior to experimentation. 24 h prior to
testing mice should be fasted but allowed full access to water. The
empty 8 cm diameter 5 cm food bowl should be present in the
cage to reduce novelty effects. One hour prior to testing, mice
should be removed to similarly sized cage without bedding but
with full water access. Testing is initiated by cleaning the bottom of
the bedding-less cage and placing a pre-weighed bowl with food in
the cage. After 1 h, the food bowl is removed and weighed as are
any food remnants within the cage. The difference between food
bowl food disappearance and food collected from the cage floor is
considered food consumed [72]. A more powerful method for
assessing food disappearance and/or consumption is through use
of automated food and water intake measurement systems where
food and/or water intake initiated by the animal is evaluated by
computer-controlled electronics (BioDAQ, New Brunswick, NJ).
Specific procedures and training for this method should be
provided by the manufacturer of the device chosen.
3.4 Rotarod Testing Inducers of neuroimmune activation and sickness behavior impair
motor coordination and induce physical fatigue [63]. The rotarod
performance test can measure motor coordination [64] by
assessing how well mice avoid falling off a rotating rod [76]. Some
strains of mice progressively perform better on the rotarod test
during repeated trials at the same rotational speed indicating a
physical training or memory component to this procedure
[77, 78]. Rotarod apparatuses are available via commercial vendors
such as Med Associates (Fairfax, WT), with some variance in fea-
tures (number of lanes and/or rod diameter, for example). In
general, rotarod apparatuses have the same basic design featuring
a 3–9 cm diameter rod [64, 77] partitioned by plastic divider disks
spaced evenly longitudinally along the rod. The end point
measured is latency to fall from the rod [79]. Fall detection ranges
from pressure-sensitive pads located under the rod to infrared
beams that automatically stop an integrated timer when hit or
blocked. Rotarod performance can, however, be assessed manually
from a video record [79] by a blinded trained observer or by
automated tracking software.
3.4.1 Procedure Single housing prior to testing is not required but, like with social
exploration, mice need at least 24 h of acclimation to the procedure
room. Use of pre-experimentation acclimation to the rotarod is not
agreed upon. Some have exposed (trained) mice to the rotating
rod, by placing mice on the rod at a low speed (4 rpm [63] and
18 rpm [64]), while others have not [77]. Pre-experimentation
exposure to the rod has been done 1 week in advance of testing
[64] and immediately prior to testing [63]. Finally, the test itself
can be performed in one of the two ways. The rotarod performance
test measures the duration of time a mouse can remain on the
rotating rod at a single or several fixed speeds [79]. The accelerating
rotarod performance test measures the maximum speed of rotation
the mouse can tolerate before it falls from the rod in a fixed amount
of time [77, 79]. All pretest conditioning and testing should be
made during the dark/active cycle of the mice [63], although
testing has been performed during the light/inactive cycle [64].
3.4.2 Rotarod Rotarods should be calibrated such that they rotate at a constant
Performance Test speed. Rods should be kept clean and as odor free as possible
between trials, as urine and feces on the rod can affect performance
[79]. Testing is initiated by placing mice on the rotating rotarod
which rotates at a fixed speed. Mice are allowed to maintain them-
selves on the rod as long as they can and the test session continues
until they fall or a designated time point is achieved. At such time,
the latency to fall is recorded. Fixed-speed trials are best used after
significant validation testing on the strain of mouse chosen. Fixed-
speed trials are best used on mice with significant loss of coordina-
tion because small losses of coordination may not manifest at the
speed or time chosen. Some testing protocols investigate several
different speeds increasing with each trial. An example of increasing

speeds used is 5, 8, 15, 20, 24, 31, 33, and 44 rpm [79]. Mouse rest
time between increasing speed trials ranges from 10 to 60 min
[77, 79].
3.4.3 Accelerating Preconditions are similar as for non-accelerating rotarod testing.

Rotarod Performance Test This test differs in that the rod accelerates at a constant rate through
some specified range of speed (4–40 rpm) over a fixed amount of
time (5 min). Mice remain on the rod for as long as they can and
speed of the rod at the time of falling is the recorded end point
[77]. For both methods of rotarod testing, motor learning can be
assessed by performing daily trials to determine if mouse time spent
on the rod (fixed rod speed) or rod speed endured (accelerating rod
speed) improves [77].
4 Depressive/Anxiety-Like Behaviors
Depression and anxiety are well-known consequences of neuroim-

mune activation [11, 72, 80] and such behaviors can be difficult to
distinguish from sickness behaviors. Sickness-induced loss of loco-
motion is a key confound in that most tests designed to measure
depressive/anxiety-like behaviors require mouse movement
[11]. For this reason, behavioral testing for depressive/anxiety-
like behaviors following exogenous activation of the neuroimmune
system should be performed only after overt physical symptoms of
sickness have subsided and spontaneous locomotion has returned
to pretreatment levels. Importantly, the presence of depressive/
anxiety-like behaviors should be confirmed using antidepressive
and/or anxiolytic therapies that block or mitigate the purported
depressed or anxious behavior [11]. Tests for depressive/anxiety-
like behaviors include burrowing [12, 81], the elevated plus maze
(EPM) [82], the open-field test (OFT) [83, 84], the zero maze
[77], the tail suspension test (TST) [11], sucrose/saccharin prefer-
ence [11], and forced swim (aka Porsolt) [11]. The forced swim
test (FST) is the test best validated for depression due to its respon-
siveness to antidepressives [85]. Investigators should refrain from
using any single test as a definitive measure of depressive/anxiety-
like behaviors. Such behaviors are best examined using a battery of
tests. This is especially important in anxiety-like behavior tests so as
not to conflate state and trait anxiety [86]. As an example, Moon
et al. demonstrated that IL-4 knockout mice were anxious in the
EZM but not in the OFT, suggesting that these mice might be
fearful of height but not bright open spaces [87]. Unfortunately,
there is no ideal combination of tests because confounds such as
strain/model/sex can be experiment specific. As example sucrose/
saccharin preference testing should not be used on mice with serum
leptin deficiency due to the impact of leptin on sweet taste detec-
tion by the tongue [24].
4.1 Burrowing Rodents are well-known burrowers [81]. This behavior is related to
tunnel maintenance and possibly defense. Defensive burying is a
known indicator of anxiety and can, itself, be measured [81]. Bur-
rowing appears to be largely hippocampal driven but mouse strain
differences exist with C57BL/6 mice burrowing more than
129S2/Sv mice [81]. Burrowing is associated with depressive/
anxiety-like behavior where reduced burrowing reflects a
depressive/anxiety-like state [12]. As burrowing utilizes relatively
simple equipment and minimal labor, it is an easy and inexpensive
method for evaluating immunobehaviors [12, 81].
mice. Mice should be individually housed as per social exploration
at least 24 h prior to experimentation and, with single housing, a
clean empty burrowing tube should be placed in the cage. Burrow-
ing tubes can be constructed from standard white 6.8 cm diameter
PVC piping cut to 20 cm in length [81]. The open end of the
burrowing tube is elevated 3 cm by bolting two 50 mm machine
screws 1 cm from the open end, and spaced so that the tube
entrance is elevated. This elevation keeps burrowing substrate
from spilling out of the open end. The closed end is sealed with a
standard PVC end cap [12]. Testing should begin 3 h prior to the
onset of the dark/active phase of the light cycle and is initiated with
addition of burrowing substrate to the tube [12]. The burrowing
substrate used needs to be suitable to the mouse strain and experi-
mental treatment. Pelleted mouse chow, gravel, or sand is the
common material used for burrowing [81]. The burrowing tube
can be completely filled [81] or filled with a fixed amount of
substrate [12] (if ceiling effects are not a concern). Ceiling effects
arise with vigorous burrowers, because these mice will remove all
substrate from a tube in a rapid time frame obscuring any difference
in burrowing activity relative to time. After substrate is placed in the
burrow, the burrow plus substrate is weighed and returned to the
cage. If mouse chow is used as a substrate, food from the cage food
hopper should be removed for the duration of the burrowing test
[81]. Depending on anticipated mouse burrowing activity, experi-
mental observation time points can range from 1 to 24 h. Amount
burrowed (in grams) is calculated from the pre- and post-
burrowing weight of the tube plus substrate. Following a measure-
ment, the burrowing tube can be refilled, reweighed, and returned
to the cage for additional testing [81]. Alternatively, a single mea-
surement of burrowing can be utilized [12]. Occasionally, with
poorly burrowing mice, one or several training sessions may be
necessary [88]. Such practice runs improve burrowing activity and
reduce variability between animals [81], likely by mitigating the
novel object effect of the burrowing substrate.
4.2 Elevated The EPM is a straightforward method to measure anxiety-like

Plus Maze behavior in mice. Anxiety is assessed by comparing the time spent
in the open (exposed) vs. closed (walled) arms of a four-arm radial
maze [82]. An advantage of the EPM is that it eschews use of
noxious stimuli like foot shock, food/water deprivation, and/or
loud noise. Instead, it relies on the predilection of mice to favor
dark enclosed spaces over open and obviously elevated environ-
ments [82]. Accommodation and/or learning can occur with
repeated exposures to the maze. Therefore, EPM is generally admi-
nistered as a single exposure with control mice for comparison [82].
mice. Single housing prior to testing is not required but, like with
social exploration, mice need at least 24 h of acclimation to the
procedure room. The maze can be made of a variety of materials but
those that can be easily wiped clean between each mouse tested are
recommended. Maze shape is that of a plus sign where the four
arms are spaced 90 apart, radiating from an open central
5 cm 5 cm platform. Arm length and width are 25 cm 5 cm,
respectively. Maze elevation should be at least 40 cm from the floor
[82, 89]. Arm wall height is 15 cm in the “closed” arms and there
are no side walls in the “open” arms. The central 5 cm 5 cm
platform has no walls. The open arms are at 180 from each other,
likewise with the closed arms. Unlike the tests described earlier, the
EPM should be well lit by overhead white light. Significant arm
wall-generated shadows, especially those confined to a single arm,
should be avoided. Testing is initiated by placing the mouse in the
open central 5 cm 5 cm platform. Each subject mouse is intro-
duced to the maze in a similar fashion and placed on the maze in the
same spot with analogous orientation [82]. Mouse exploration is
video recorded for 5 [82] to 10 min [90]. Time spent in open and
closed arms, number of entries between arms (defined as all four
paws of the mouse crossing the threshold of an arm), frequency of
head dips (downward movement of the mouse head toward the
floor from an open arm), rears, and stretch-attend postures [82] are
best determined from the video record using automated tracking
software [69]. If tracking software is not available videos can be
hand scored by a blinded trained observer [82].
4.3 Open-Field Test The OFT can be used to measure movement [84] and anxiety-like
behavior [83, 84]. OFT apparatuses are walled arenas that vary in
shape (square, rectangle, and circle) and size (250–2500 cm2)
[84]. OFT testing should not be used as a surrogate test for
spontaneous locomotor activity because the OFT uses a novel
environment [84]. Anxiety-like behavior in the OFT is evaluated
by examining mouse movement throughout the arena with a spe-
cial focus on the amount of time the mouse spends/moves next to
walls of the OFT apparatus (thigmotaxis). The novelty, size, and
white light illumination of the OFT contribute to anxiogenesis

[84]. Procedures vary considerably but the open-field arena is
usually brightly lit in studies investigating anxiety [83].
procedure room. For testing of anxiety-like behaviors the arena
should be larger than 1600 cm2 [84]. Arena wall height should
be at least 35 cm so as to limit the ability of the mouse to see over
and above the arena. The arena can be made of a variety of materials
but those that can be easily wiped clean between each mouse tested
are recommended. Testing is initiated by placing the mouse in the
center of the arena, and each subject mouse needs to be introduced
to the arena in a similar fashion and placed in the same spot with
analogous orientation. Movement through the arena is video
recorded for 5–10 min and analysis of movement is best documen-
ted with automated tracking software because thigmotaxis is easily
appreciated with this method. Path tracing is a key aid so that
patterns of movement can be evaluated. These pattern tracings
supplement the usual measurements of time spent adjacent to the
arena walls, wall preferences, time spent not adjacent to the arena
walls, overall distance traveled, velocity, and time spent moving.
Videos can be manually examined using a line-crossing scoring
approach (similar to that described for spontaneous locomotor
activity) but this method should be carefully validated due to the
complex grid pattern needed to ascertain time spent close to the
arena walls. Due to this complexity, some have used the end point
of total distance moved plus time spent in the central 25% of the
arena when utilizing manual scoring [89]. Finally, OFT can be used
as a repeated measure to determine if therapeutics improves perfor-
mance over time [91].
4.4 Elevated Like the EPM, the zero maze measures anxiety-like behaviors in
Zero Maze mice [77]. Time spent in the open indicates a reduced level of fear/
anxiety as validated by use of anxiolytic agents that increase the time
mice spent in the open area of the zero maze [77]. The advantage of
the zero maze over the EPM is the elimination of the central
platform, which can complicate analysis of open/closed arm
comparisons [92].
mice. Single housing prior to testing is not required but, like social
room. Maze design varies but in general is comprised of a circular
track 30–45 cm in diameter that is 3–5 cm wide. Maze elevation
should be at least 40 cm from the floor [77, 93]. The track should
be divided into four quadrants with two quadrants having no side
walls and two quadrants having side walls at least 15 cm in height.

These open and closed areas should alternate. As with all maze
constructions, materials that are easily wiped clean between each
mouse tested are recommended [77, 92, 93]. Ample but dim
(40–60 lux) white lighting should be used to achieve similar illumi-
nation of both the open and closed quadrants [92]. Testing is
initiated by introducing the subject mouse to the middle of a closed
quadrant (designated as the starting quadrant). Each subject mouse
needs to be introduced to the maze in a similar fashion and placed
on the maze in the same location and orientation. Mouse explora-
tion is video recorded for 5 min [92]. Time spent in open and
walled arms, number of transitions between open and walled quad-
rants, number of rears, number of head dips (the actual dipping of
the head over the edge of the track in an open quadrant), time spent
grooming, number of stretch-attend postures, and number of fecal
boli in each type of quadrant are best determined by using a
combination of automated tracking software and direct observation
after testing (fecal boli) [77, 92]. If tracking software is not avail-
able, videos can be hand scored by a blinded trained observer. Entry
into a quadrant occurs when all four paws cross the threshold of an
open or walled area [92].
4.5 Light/Dark Box Like the EPM and EZM the light/dark box measures anxiety-like
behaviors in mice by exploiting the innate avoidance behavior of
mice to bright lights [94]. In this test, increased time spent in the
light part of the arena is a marker of reduced anxiety-like behavior.
Importantly, mice administered benzodiazepines demonstrate
increased time in the lit section, indicating that the light/dark
box paradigm is responsive to anti-anxietal therapies.
room. Test boxes can range from 44 21 cm to 91 92 cm, and
vary in the proportion of light to dark areas. Generally, the light/
dark portions are 1:1 or 2:1 [95]. Increasing the light:dark ratio
exacerbates the anxiogenic properties of the arena by creating an
open-field-like space. Light and dark areas are typically separated by
either a simple wall with a small opening (usually no more than
7 cm) or small tunnel between the two sides (5 7 10 cm).
Arenas are typically constructed of translucent and opaque plastics
to create the light and dark areas. Fully automated computer-
connected arenas are helpful (Omnitech Electronics, Columbus,
Ohio), and can easily track mouse movement automatically within
the testing period. Animals should be introduced to the testing
arena in a standardized place and orientation, usually into the light
part of the arena [96]. Testing should be video recorded for
5–10 min. Time spent in the light, time spent in the dark, latency
time to first light/dark transition, transitions, rears, and total

movement should all be determined using a combination of auto-
mated tracking software and direct observation by a blinded
observer. A transition is recorded when all four paws are within
either the light or the dark sections.
4.6 Tail The TST is a commonly used behavioral test for assessing
Suspension Test depressive-like behavior in mice. It is thought to induce an escape
response [97]. With increased depressive-like behavior mice fail to
extricate themselves from the apparatus and become immobile.
Increased immobility indicates a greater degree of depressive symp-
toms [98]. Importantly, antidepressants shorten immobility offer-
ing a degree of validation to the test’s usefulness in measuring
depressive-like behaviors [97]. The TST can be automated through
use of commercially available apparatuses that utilize computer-
linked linear load cells and load cell filters to determine mouse
movement/struggle (Med Associates, St. Albans, VT). As with
any commercially purchased device, specific procedures and train-
ing should be provided by the manufacturer. However, certain
basic procedures should be followed and considered with use of
the TST including the difficulty in examining young mice (espe-
cially C57BL/6) due to their robust tail climbing behavior and
penchant for escaping.
4.6.1 Procedure Unlike all the previous behavioral tests described the TST should be
administered during the light/inactive cycle of the mice. Single
housing prior to testing is not required but, like social exploration,
mice need at least 24 h of acclimation to the procedure room.
Testing is initiated by affixing the mouse to the apparatus “hook”
with adhesive tape wrapped around the tail at three-quarters of its
length from the tail base. Attach the mouse to the apparatus
through the tape as close to the tail as possible. The tail should
remain straight so as not to injure the mouse. Mice should be
suspended as uniformly as possible, and if multi-mouse devices
are used the mice should be shielded from each other’s view
[98]. Immobility versus movement/struggle should be measured
for 6 min. Nonautomated devices can be constructed, which are
essentially chambers with hooks. Mouse behavior can be video
recorded and immobility determined from the video record by
automated tracking software [99] or a blinded trained observer
[100]. With any of the aforementioned analyses, time of immobility
is compared between control and experimental groups of
mice [98].
4.7 Forced The FST (also called the Porsolt test for the investigator who
Swim Test developed the test in rodents) like the TST is a tool for assessing
depressive-like behavior in mice. The FST is relatively easy to
administer [101] and felt to be the best validated test for depression
by the pharmaceutical industry [102]. This test evolved from the

observation that rats will develop an immobile posture after an
initial attempt to escape from an inescapable cylinder filled with
water. FST-induced immobility is thought to represent behavioral
despair (failure of persistent escape behavior) or a development of
passive behavior that causes the animal to stop actively coping with
a stressor [101]. The FST has several disadvantages when compared
to the TST. The FST appears to be more stressful for mice and
carries a risk of hypothermia [98]. Mice of varying fatness are also
difficult to assess due to inherent buoyancy differences.
4.7.1 Procedure Like the TST, the FST should be administered during the light/
inactive cycle of the mice [102]. Single housing prior to testing is
not required but, like social exploration, mice need at least 24 h of
acclimation to the procedure room. As with most device requiring
tests, equipment design varies. A simple setup is to use clean white
or black cylindrical PVC containers 16 cm in diameter and 31 cm in
height (essentially 2 gallon pails) containing 20 cm of water main-
tained at 25 1 C [103]. The FST should be performed under
30 lux white light [104]. Testing is initiated by introducing the
subject mouse to the water-filled container. Mouse swimming is
video recorded for 6 min [103]. Immobility is determined from the
video record from the last 5 min of the FST using either automated
tracking software [105] or a blinded trained observer [103]. Immo-
bility scoring should not include movements necessary for the
mouse to maintain its head above water [98]. Like the TST, time
of immobility is compared between control and experimental
groups of mice.
4.8 Sucrose/ Anhedonia, or the inability to gain pleasure from otherwise enjoy-
Saccharin Preference able experiences, is one of the features of depression [106, 107]. In
mice, their preference for sweetened solutions has been exploited to
measure anhedonia. The decreased consumption of a sweet-tasting
solution is indicative of anhedonia and can be reversed with anti-
depressives [107]. Sucrose [107] and saccharin [12] solutions are
commonly used opposite normal tap water in a two-choice test. For
investigators concerned with mouse caloric intake, saccharin is the
recommended sweetener [12]. Advantages to the sucrose/saccha-
rin preference test are that it can be run continuously for many days
without significant concern of adaptation or learning.
4.8.1 Procedure Three days prior to testing mice should be singly housed in stan-
dard cages adapted for two-bottle water access (adaptation phase).
If the experimental design requires mice to be challenged with a
neuroimmune activator, each bottle should contain either saccharin
as a 0.4% sodium saccharin solution (1% for sucrose can be sub-
stituted for saccharin) or water. If the experimental design does not
require exogenous challenge as with a comparison of mice of
different strains or sexes, the adaptation phase should consist of

both bottles being filled with water. The adaption phase is especially
important to experiments using endogenous immune activators so
as not to elicit the behavior of conditioned aversion where the
mouse associates a newly introduced substance like saccharin/
sucrose with the cause of their loss of well-being [108]. After the
adaptation phase, mice are usually administered a challenge at the
beginning of their dark/active cycle and then returned to the cage
in which they were adapted in the presence of both water and
saccharin (testing phase). Fluid consumption is recorded every
24 h. Fluid consumption is determined by bottle weight [12]. In
order to control for the development of bottle bias, bottle position
of water vs. sweetened solution should be switched on a regular but
defined basis such as halfway through the experiment or every 24 h.
Bottle switching should also be practiced during the adaptation
phase usually at 24-h intervals. Sweetened solution preference is
generally reported as a percentage of sweetened solution consump-
tion/disappearance to total fluid consumption/disappearance
[107]. A 50/50 consumption of sweetened solution to water
equates to anhedonia but a ratio in which water consumption/
disappearance exceeds sweetened solution consumption/disap-
pearance is indicative of aversion and should trigger concerns as
to the applicability of the results to the measurement of depressive-
like behaviors [12].
5 Cognitive Behaviors
Neuroimmune activation can dramatically impact cognitive func-

tion [9] causing learning and memory deficits. Most mouse-based
behavioral tests for cognitive function make use of memory. Gen-
erally, the memory tasks test the ability of mice to form new
memories (memory formation) or recall old memories (memory
retention). Memories involving location (spatial memory) are uti-
lized [9]. A cornucopia of cognitive function tests exist. Some of
these tests have been specifically designed to identify specialized
aspects of learning and memory such as olfactory memory
[109]. Given that peripheral innate immune-driven neuroimmune
activation is relatively brain region nonspecific, cognitive tests that
cover more global aspects of brain function are favored by PNI
investigators.
5.1 Novel Object Novel object recognition is a test of working memory in mice. The
Recognition test exploits the innate tendency of mice to investigate a new entity
[110]. Novel object testing is one of the simplest of cognition tests,
but test variations are described that add significant complexity
through mixing of objects, object placement (novel location test-
ing), and testing arena conditions [12, 110–112]. The setup for
novel object recognition typically depends on what sort of memory

function a researcher desires to investigate [110]. An advantage of
novel object testing over other seemingly more powerful maze-
based memory tests is its adaptability to repeated measure testing
[110]. In essence, as long as the mouse is well adapted to the
familiar object, changing out the novel object after each exposure
allows for a new round of testing. This feature is very useful when
looking at recovery of memory, especially from short-term amnesic
challenge [113].
5.1.1 Memory Recall All observations should be made during the dark/active cycle of the
Procedure mice. Mice should be housed individually at least 24 h prior to the
initial measurement. The 24-h training phase is initiated by intro-
ducing two identical objects into the home cage (standard shoebox
cage size; 28 cm in length; 17 cm in width; 12.5 cm in height) of
the singly housed mouse. The objects are placed 10 cm apart at the
short-side wall end, 5 cm from the short-side wall, and 3.5 cm from
the long-side wall. Tall (3–5 cm in height) complex objects are
preferred because when a tall complex object is introduced during
the testing phase it provokes significantly more exploration time.
Tall complex objects can be constructed from Lego® blocks
(Enfield, CT). Magnets can be used to secure the structures to
the cage floor. All structures should be taken apart and cleaned
prior to reuse. After the 24-h training phase, the mouse is exposed
to the chosen neuroimmune activator. At relevant times after the
applied immunobehavioral challenge, the memory recall testing
phase is initiated by placing the mouse in a home-cage-like arena
(including bedding) which contains a similar object setup as in the
training phase, where one of the familiar objects has been replaced
by a novel object. The mouse should be introduced at the cage end
opposite the objects. Object exploratory behavior is video recorded
for 5 min and object investigation is determined from the video
record by either automated tracking software or a blinded trained
observer. Object exploration is considered as contact by mouth,
nose, or paw. Accidental contact such as bumping into an object
while passing should not be considered [111]. Mice with a memory
recall deficit should examine both the familiar and novel objects
equally [114]. Once recovered from neuroimmune activation, mice
should explore the novel object more frequently than the familiar
object. This test cannot be performed as a repeated measure and,
thus, requires separate groups of mice to determine at what time
after neuroimmune activation cognitive recovery occurs.
5.1.2 Memory Formation All observations should be made during the dark/active cycle of the
Procedure mice. Single housing prior to testing is not required, but, like with
procedure room. Memory formation testing differs from recall
testing in that training occurs after endogenous activation of the
neuroimmune system instead of before [110]. The procedure is

identical to the above except at relevant times after the applied
immunobehavioral challenge mice are trained with the two familiar
objects placed on either end of a large shoebox-size
(48 26 16 cm) testing arena. Then, after 30 min of training,
mice are returned to the home cage. After 2 h in the home cage,
testing is initiated by placing mice back in the testing arena where
one of the familiar objects has been replaced by a novel object.
Recording time and scoring are identical to the above. Mice with a
formation deficit should examine both the familiar and novel
objects equally. Once recovered from neuroimmune activation,
mice should explore the novel object more frequently than the
familiar object. This test can be performed as a repeated measure
as long as the novel object is always new.
5.2 Novel Location The novel location recognition test is the spatial memory analog to
Recognition the novel object recognition test. This test is performed in an
identical manner as the novel object recognition test with two
minor alterations. First, spatial cues are placed around the outside
of the testing arena; these cues can be simple tape markings on at
least three sides of the cage [115, 116]. The second alteration is
that rather than replacing one of the objects with a new, novel
object, one of the objects is simply moved to the opposite end of
the cage than where it was during training. Performed in this
manner the novel location recognition test can also be used for
repeated measure testing by moving the object to a new area within
the testing arena for each subsequent test.
5.2.1 Memory Recall The memory recall procedure for the novel location recognition
Procedure task is identical to the procedure described in Subheading 5.1.1,
combined with the following changes. The 24-h training phase is
initiated by introducing two identical objects into the home cage
(standard shoebox cage size; 28 17 12.5 cm), with spatial cues
on three of the four sides. Objects are placed 10 cm apart at the
short-side wall end, 5 cm from the short-side wall, and 3.5 cm from
the long-side wall. The memory recall testing phase is initiated by
placing the mouse in a home-cage-like arena which contains a
similar object setup as in the training phase where one of the objects
has been moved to the opposite end of the cage, 5 cm from the
short-side wall. The mouse should be introduced at the cage
between the two objects.
5.2.2 Memory Formation The memory formation procedure for the novel location recogni-
Procedure tion task is identical to the procedure described in Section 5.1.2,
combined with the following changes. The two familiar objects are
placed on one short end of a testing arena (28 17 12.5 cm)
with spatial cues on three of four sides for 30 min and then returned
to the home cage. After 2 h in the home cage, testing is initiated by
placing the mouse back in the testing arena where one of the
familiar objects has been moved to the opposite short end of
the cage.
5.3 Fear-Conditioned Fear-conditioned learning is a form of classical (Pavlovian) condi-

Learning tioning where an association between a stimuli and its aversive
consequence(s) is made [117]. Fear conditioning is a highly con-
served behavior that occurs in mice both in the laboratory and in
the wild. In PNI research, it is a useful tool for evaluating emotional
memory formation and recall [117]. Fear-conditioned learning is
generally a one-trial learning procedure and is unique from other
cognitive tests in that the investigator regulates the parameters of
the stimulus. Factors that impact mouse hearing, like age, are
important to consider prior to use of fear-conditioned learning, as
auditory sensory decline will negatively affect sound-based contex-
tual cues [117]. Like novel object testing, fear-conditioned
learning can be used to test memory formation and memory recall.
For memory formation testing, the training phase (see below) is
conducted after neuroimmune stimulation. For memory recall
testing, the training phase is conducted prior to neuroimmune
stimulation.
5.3.1 Cued Fear- All observations should be made during the dark/active cycle of the
Conditioning Procedure mice. Mice should be single housed for this behavioral procedure
and, like social exploration, mice need at least 24 h of acclimation to
the procedure room. The training/testing should take place in a
separate area of the procedure room, as animals in the same space
may be able to hear auditory cues, unless using soundproof fear-
conditioning chambers. Automated commercially available fear-
conditioning apparatuses (San Diego Instruments, San Diego,
CA) are the easiest way to adapt this testing paradigm. General
apparatus parameters are fairly uniform. There is a shock generator
and scrambler that delivers a 0.1–1.0 mA foot shock through a wire
grid floor in concert with a sound generator that produces auditory
cues, all contained in a shoebox cage-sized chamber [117]. It is
recommended that sound meters and voltmeters are used to verify
and record stimulus intensities [117]. Prior to testing mice require
training. In the initial training session, mice are placed in the fear-
conditioning apparatus for 120 s (Phase A) before the presentation
of a 30-s sound cue (Phase B). A 2-s foot shock is delivered
immediately after the sound cue (Phase C). Mice are returned to
their home cages 30 s after the shock ends. Repeat training can be
utilized to reinforce the memory. As noted earlier, training relative
to neuroimmune stimulation determines whether memory forma-
tion or recall is being tested. Testing is usually initiated 24 h post-
training and consists of reintroducing the mouse to the fear-
conditioning apparatus and representing the sound cue. The
sound cue now lasts for 180 s. Mouse behavior during this 180-s
period is recorded with a side-mounted video camera. However,

apparatuses with a beam detection grid system linked to a PC can
automate analysis. With video recording, freezing and nonfreezing
behavior is scored by a blinded trained observer at 10-s intervals.
Freezing is considered a complete lack of mouse movement
[117]. Fear conditioning is presented as the number of freezing
episodes. With an automated detection system, actual time spent
frozen can be determined.
5.3.2 Contextual Fear- Contextual fear conditioning uses the same pre-experimental and
Conditioned Learning scoring procedures as cued fear conditioning. However, in contex-
Procedure tual fear conditioning, no sound cues are delivered. The mouse is
expected to associate the apparatus with the foot shock. Testing
time is 180 s. Complexity can be added by using an alterable
microenvironment within the fear-conditioning apparatus (altered
contextual fear conditioning). A variety of cues from visual to
olfactory can theoretically be utilized [117].
5.4 Spontaneous Spontaneous alternation is the simplest spatial memory test to

Alternation perform in mice. Increases and decreases in spontaneous or perfect
alternations reflect improvements and impairments, respectively, in
spatial memory function [12, 118]. Spontaneous alternation can be
performed in a variety of maze types including radial arm, T and Y
[12, 119]. Subforms of spontaneous alternation testing are also
described including forced-trial alternation, where one arm of the
maze is closed off, forcing the mouse to enter the open arm without
choice. In subsequent testing, the closed arm is then made available
[119]. Interestingly, certain mouse strains are biased in their turn-
ing direction [119], and this should be considered and controlled
for. Like novel object testing, spontaneous alternation is useful to
test memory formation and memory recall. For memory formation
testing, spontaneous alternation is conducted after neuroimmune
stimulation. For memory recall testing, the mouse is tested in the
Y-maze (which serves as a training period), exposed to a neuroim-
mune activator and then retested in the Y-maze. In addition, spon-
taneous alternation can be performed as a repeated measure and
maze performance usually increases with repetition. Repeated mea-
sure testing is generally preferred because a “one-and-done” testing
strategy is tied more to locomotor activity and less dependent on
spatial memory [119].
procedure room. As with all mazes described previously, the
Y-maze used for spontaneous alternation should be made of a
material that can be easily wiped clean between each mouse tested.
Clear plastics are preferred so that a different black-colored design
(lines, circles, or triangles) can be affixed to the outside wall of each

arm to provide intra-maze visual cues. For Y-maze testing, the maze
base is an opaque blue. Maze shape is three equally spaced arms
120 from each other (radial Y). Arm length is 40 cm, arm width is
9 cm, and arm wall height is 16 cm [12]. Testing is initiated by
placing the mouse into the distal end of a randomly chosen arm
(as assigned by a random number generator). Each subject mouse
needs to be introduced to the maze in a similar fashion and placed
on the maze with analogous orientation. Mouse exploration is
recorded for 5 min (3 and 15 min have also been used [120]). If
the experimental design allows, mice should be tested every 24 h
for 4 consecutive days. Perfect alternations are determined from the
video record by a blinded trained observer. A perfect alternation is
defined as exploration of two different arms of the maze sequen-
tially before a return to the starting arm (e.g., beginning in arm
“C,” moving to arm “A,” then to arm “B,” before returning to arm
“C” again) [12]. Number of “regular” alternations should also be
scored. Regular alternations are defined as entering all three arms
within a sequence of four arm entries (e.g., ACAB is considered an
alternation, whereas ACAC is not) [119]. Arm entries occur when
all four paws of the mouse pass the threshold of the arm entrance
[12]. Results are represented as total alternations or perfect alter-
nations divided into the total possible alternations or perfect alter-
nations, respectively [119]. Perfect alternation scoring is
considered more rigorous.
5.5 Barnes Maze The Barnes maze, like spontaneous alternation, is a test of spatial
memory. This test combines several aspects of the previously men-
tioned mazes including elevation, open/exposed illuminated space,
and a dark enclosed area [121]. Use of the Barnes maze was
popularized as an alternative to the Morris water maze (MWM)
(described in the next section) because swimming may produce
anxiety [122]. Removal of water favors testing of mice different
fat density due to elimination of the buoyancy effect. Importantly,
the Barnes maze appears to rely on the same hippocampal-
dependent memory function as the MWM [121]. As with any
maze designed to test spatial memory, extra- and intra-maze cues
serve as location reference points, and without these cues mice
perform less well [121]. Like spontaneous alternation, the Barnes
maze can be used to test memory formation and recall depending
on when neuroimmune activation is triggered relative to the train-
ing period. However, recall is significantly simpler to measure when
using transient memory impairment paradigms.
room. As with all mazes, construction materials should be easily
cleanable. A typical Barnes maze is a 90 cm diameter white acrylic

disk with 12 equally spaced through-and-through holes arranged
5 cm from the outer edge of the disk. Hole diameter is 5 cm and the
maze is situated 56 cm off the floor [121]. A tunnel-like extension
attached to an enclosed sealable 8 8 8 cm chamber (escape
chamber) needs to be freely attachable to one of the holes from
underneath the maze. Thus, for the mouse to escape the maze
surface, it must enter a hole. Extra maze cues, such as different
geometric shapes, are placed around the maze and on the walls of
the room [122]. Prior to testing, multiday training is required.
Training occurs four times per day (during the dark/active cycle
of the mouse) for a 5-day period. In each session, the mouse is
introduced to the maze (lit at 1200 lux) via a nontransparent
holding chamber placed in the center of the maze. Time spent in
the holding chamber is 30 s. After the holding chamber is unsealed,
the mouse is allowed to explore the maze freely for 5 min. During
each training session, the escape chamber should remain under the
same assigned escape hole with all other holes blocked. During the
5-min exploration period, the mouse should find the escape cham-
ber hole. If the mouse fails to find the escape chamber hole, it is
picked up and placed near the entrance of the escape chamber hole
and allowed to enter. Once the mouse enters the escape chamber
the mouse is removed from the maze and the training session is
ended [121]. One hour after the final training session, a probe trial
is conducted in which all of the holes are blocked (preventing any
escape) and the mouse is allowed to explore the maze for 5 min.
Successfully trained mice with functional spatial memory should
actively search for the remembered escape chamber hole in the
appropriate location. Mouse introduction to the maze during the
probe trail utilizes the holding chamber technique as performed
during training. After immunobehavioral stimulation, testing is
initiated by re-performing a single 5-min training procedure.
Mouse behavior is video recorded and maze performance evaluated
from the video record using a combination of automated tracking
software and observation. The mouse should use the extra-maze
visual cues to locate the remembered escape chamber [121]. Scor-
ing the trials consists of tallying the frequency of errors committed
before entering the escape chamber (examination of the incorrect
hole), timing the latency to find/enter the escape chamber, and
determining the path length to the escape chamber. Different
variations of the Barnes maze exist and include a hidden-target
fixed-location modification in which the extra-maze cues are always
in the same location, but the maze is rotated [121].
5.6 Morris The MWM is used for assessing spatial or place learning. Advan-
Water Maze tages of the MWM include no requirement for pre-training, high
reliability across different tank designs, and proved validity in mea-
suring hippocampal-dependent spatial and reference memory.
Learning impairments in the MWM are independent of locomotor

deficits, as locomotor reductions do not seem to affect swim speed
[123]. The MWM is an open circular pool filled with water. Mice
must swim and search for a small, hidden platform just below the
surface of the water in a fixed location utilizing extra-maze cues.
The maze is constructed of a circular tank 122 cm in diameter, with
51 cm high walls and nonreflective inner surfaces. Contrasting
colors between the inside of the tank and the mouse allow easy
deployment of automated tracking software for the analysis of swim
path, duration, and location. As the purpose of the MWM test is to
force mice to use distal cues, any seams or recognizable patterns on
the inside of the pool are not recommended. The platform (usually
plastic) is typically square or circular in shape and is clear or
matched in color to the inside of the swim tank. The platform is
positioned just below the water level (0.5–1 cm). Water tempera-
ture should range between 24 and 26 C [124]. The room in which
the MWM test takes place should allow for ample distal visual cues
like for the Barnes maze. Studies show that lack of cues can nega-
tively affect MWM performance [123]. The investigator should be
aware that his/her presence in the procedure room during MWM
testing makes him/her a distal visual cue. All variations of the
MWM procedure should take place in an illuminated, albeit indi-
rectly lit, room (to avoid reflections or glare on the water surface,
which can make scoring with automated tracking software difficult
[123]). The maze is divided into four, equally sized quadrants with
the platform positioned in the center of one of these quadrants
[123]. The procedures for spatial acquisition and reversal learning
are described below but a variety of modifications exist. A key
concern in MWM spatial memory testing is that the probe trial
should be spaced temporally from the last training session to effec-
tively and reliably measure reference memory formation [123].
5.6.1 Spatial Acquisition All testing should take place during the dark/active cycle of the
Procedure mice. Single housing prior to testing is not required but, like social
room. The mouse is placed (not dropped) at the selected start
position in the maze, facing the tank wall. Video tracking should
start immediately at the placement of the mouse in the maze. The
video remains on until the mouse reaches (comes in contact with)
the platform. Standard trial limits of 1–2 min per trial are usually
used, and mice that have not reached the platform should be placed
on or guided to it by the investigator [123]. The animal should
remain on the platform for 15 s. This step helps mice orient their
position in space relative to the extra-maze cues [123]. Following
the inter-trial interval, the mouse is again placed in the maze, this
time in a different but predetermined location (most protocols start
mice from one of the four positions—south (the investigator’s
position), north (opposite the investigator’s position), east (to the
right of the investigator’s position)), and west (to the left of the
investigator’s position)). Trials are repeated four times per day for
5 days. Following training, the experimental treatment is adminis-
tered and time is allowed for the treatment to take effect, or in the
case of neuroimmune activators, for sickness behavior to resolve so
as not to confound the results [11]. The probe trial is run, during
which the platform is removed from the maze. The probe trial is
video recorded and lasts 60 s, after which the mouse is removed.
The objective of the probe trial is to determine whether or not the
mouse can recall memories of where the platform was during
training sessions based on the distal visual cues [123]. End points
measured in spatial acquisition include the number of platform site
crossovers, time and distance swam in the target quadrant relative
to the other quadrants, time in a predefined radius around the
original platform position (larger than the original platform itself),
average distance swam to target site, and latency to first target site
crossover. For investigators without automated tracking capabil-
ities, blinded trained observers should use a timer to calculate the
time spent in the aforementioned areas, as distance traveled is not
reliably measurable. Percent time spent in the target quadrant or
percent of distance swam in the target quadrant is the most com-
mon reported end points in MWM spatial acquisition
testing [123].
5.6.2 Spatial Reversal Spatial reversal testing determines the ability of mice to extinguish a
Testing Procedure particular memory in favor of a new one [123]. In this paradigm,
training procedures are the same as they were for spatial acquisition,
but the probe trial differs. During the reversal training probe trial,
the platform is moved, typically to the opposite side of the maze,
but cues remain in their same position as during training trials
[124]. Mice are placed first on the platform for 30 s to allow
them to gain some spatial cues as to where the new platform
location is. Mice are then given 1–3 trials to reach the platform,
starting from different locations if necessary [124]. The same end
points are used in spatial reversal training as with spatial acquisition
[123]. Since the platform remains in the maze, latency to reach the
platform, swim speed, and total distance swam are also useful end
points [124]. Some variations of the MWM include repeated
learning, latent learning, and cued learning [123].
5.7 5-Choice The 5-choice reaction time task (5CRT) is a useful test for assessing
Reaction Time Task a variety of cognitive functions, specifically attention, reaction time,
and executive function. The test was adapted from protocols for
testing attention in human subjects [125]. Typically, the test is
performed in a single chamber (25 25 cm) with one curved
wall containing five (2.5 2.5 cm) openings. In automated devices,
each opening is bisected by a photo beam capable of detecting
mouse movement [126], and contains a light cue that the animal
must respond to in order to receive a food reward [127]. To ensure

motivation for the reward, mice are typically fasted overnight prior
to testing or training. A variety of measurements can be evaluated
during each testing phase. Latency from time of signal onset to
response can measure neural processing speed, while latency to
reward retrieval can measure motivation. Several types of errors
can occur during the test that provides valuable information, such
as premature responses for impulsivity, and nonresponses to cues
for attention [128]. However, the most usual measurement is the
number of correct responses to cues compared to total number of
cues during the testing phase [127].
5.7.1 Procedure For the serial reaction time procedure the training phase is initiated
by placing mice into the testing chamber for a period of 1 min to
acclimate them to the chamber. Following acclimation, the first light
cue is activated in one of the openings selected at random. Beam
crosses into the opening are recorded following cue initiation and
the light is turned off following a beam cross into the correct
opening. Once mice select the correct opening a food reward is
given on the opposite end of the chamber. A new trial begins 2 s
after the mouse retrieves the food reward. A completion of this cue-
to-reward circuit is a single trial. Trials continue until a specific
amount of trials or specific time determined by the researcher is
met [129]. Computer software combined with human observers
can measure a variety of parameters such as percent of mice reaching
trial criterion, percent correct trials, percent premature responses,
percent incorrect responses, and percent omissions [129].
6 Physical Activities
An acute reduction in physical activity is a sign/symptom of sick-

ness and is associated with fatigue [11]. Chronic low-grade inflam-
mation is also linked to altered patterns of physical exertion and
fatigue [11]. Physical activity and fatigability can be measured with
techniques adopted from exercise research and include voluntary
running wheel, exhaustive/forced running, and grip strength tests.
These tests are more powerful than spontaneous and long-duration
locomotion testing described earlier in that they can tease out more
subtle activity differences. Animals that engage in more spontane-
ous physical activity generally have less fatigue, higher fitness levels,
and better performance in forced exercise testing [130]. Important
behavioral differences appear to exist with spontaneous and forced
exercise. Wheel running is spontaneous and thought to be under
central nervous system control. The concept of motivation is criti-
cal to this behavior and sickness-associated fatigue appears to be a
modulating factor [131, 132]. In contrast, exhaustive exercise,
such as forced treadmill running (FTR), is generally controlled by
muscle and/or cardiovascular limitations [133]. The rapidity with

which an animal discontinues an exhaustive exercise test may also be
governed by immunobehavioral fatigue [133]. PNI investigators
most often use spontaneous wheel running (SWR) [32] and forced
treadmill running (FTR) [131] to probe the impact of immunobe-
haviors on physical activity, and forelimb grip strength (FGS) tests
are highly applicable to studies relating to aging. SWR is preferred
for the high-throughput testing in that it can be remotely moni-
tored with little demand on personnel. FTR is much more labor
intensive but allows for considerable customization including alter-
ability of duration, frequency, and intensity. FTR is also consider-
ably more stressful to mice. Grip strength is a high-throughput
method that can easily examine muscle strength and motor
function.
6.1 Spontaneous A key advantage of SWR is that it can be assessed without moving
Wheel Running mice from their home cage and the length of examination time can
be very long. A disadvantage is that mice need to be singly housed.
Specialized caging is needed to accommodate the running wheel
and bedding must be correctly adjusted and monitored so as not to
interfere with the wheel. A basic running wheel may measure only
revolutions of the wheel and may need manual resetting at each
data collection point. Advanced running wheel systems (Mini Mit-
ter, Bend, OR) can obtain hourly, daily, or weekly run distance.
Regardless of the running wheel sophistication, wheels need to be
clean and well lubricated. Running wheel size and structure should
also accommodate the size of mouse used. In long-duration stud-
ies, cage cleaning and contact with animal facility and/or investiga-
tive personnel can result in an acute reduction in running. For the
below procedure, an automated, multichannel running wheel sys-
tem (Mini Mitter, Bend, OR) is utilized. Specific procedures and
training for any given wheel should be provided by the manufacture
of the device chosen.
6.1.1 Procedure Mice need to be individually housed for experiments with running
wheels, and need to acclimate to the procedure room for at least
24 h prior to experimentation. Groups of mice in cages with locked
running wheels and in cages with no wheels should be included for
proper experimental controls. Prior to experimentation (such as
neuroimmune activation), a baseline measurement is recorded in
case post hoc normalization of distance run is required. After
immunobehavioral stimulation, mice are immediately returned to
their cage and allowed access to the wheel (rotating or locked) or
cage environment (no wheel). A 10-day course of wheel running is
recommended. With automated running wheel systems, total dis-
tance run is reported. With manual wheels, wheel revolutions are
recorded and distance traveled calculated by multiplying the wheel
circumference by the number of revolutions. A limitation in SWR is
the absence of a running intensity marker. However, some sophis-

ticated wheel systems can record revolutions/min providing some
insight into intensity.
6.2 Forced Treadmill FTR better measures mouse fatigue [133]. Like running wheel
Running systems, mouse treadmills vary in sophistication with some allowing
both uphill and downhill running (IITC Inc. Life Science, Wood-
land Hills, CA). Treadmills coupled to oxygen consumption sys-
tems can be used to determine mouse “fitness.” Non-rodent
treadmills (Jog-A-Dog, Ottawa Lake, MI) divided into lanes with
plastic dividers allow for high-throughput studies of up to 20 mice.
Treadmills should contain a protective end (foam) to prevent mice
from being thrown from the device and to provide an impetus to
move forward should the mouse reduce its speed or stop running.
Mice will respond to the contact of their tail/hind portion to the
foam end. A ventilated cover is also recommended. Intra-
experimental prodding can be used if a mouse or mice appear to
predominantly “ride” the treadmill, but this encouragement can
lead to bias due to the difficulty of applying prodding evenly to
every subject mouse.
mice. Mice do not need to be single housed prior to this procedure
but should be allowed to acclimate for at least 24 h to the procedure
room. Prior to experimentation, mice should be trained daily for
3 days at speeds of 14–20 m/min (speed depends on mouse age
and strain). Mice that cannot learn the treadmill task should not be
included in experimental studies. Training sessions should last until
mouse exhaustion (1–2 h). Immunobehavioral stimulation is deliv-
ered 24 h after the final training session. Testing is initiated by
conducting a treadmill run to exhaustion. Exhaustion is considered
as a cease in running that is not motivated by protective end contact
with the foam end. Time to exhaustion is the measured end point.
Distance to exhaustion can be calculated from the time run and
velocity of the treadmill.
6.3 Forelimb Grip FGS is a reliable measure of muscle strength and motor function in
Strength mice. Measurement devices for FGS vary widely, from simple force
dynamometers to advanced computerized grip strength meters
such as the GPM-100 (Melquest, Toyama, Japan). During the
test, the animal is pulled by the tail while holding onto the dynam-
ometer with increasing force until it can no longer hold on. This
provides an accurate measurement of the amount of force a single
mouse can produce. In studies of young and old mice, older mice
produce less force during FGS testing, providing a validation of this
technique as a measure of muscle strength [134]. NF-kB blockade
has also been shown to increase FGS performance in the MDX
mouse model [135].
mice. Mice do not need to be individually housed prior to this
procedure, but should be allowed to acclimate to the testing room
for 24 h. The test can be performed either horizontally or vertically,
with vertical testing shown to be more consistent across multiple
tests [134]. Testing is initiated by allowing mice to grab hold of the
bars on the force gauge. Once the mouse has grabbed the rods, the
provided force gauge should be reset to 0. The investigator then
pulls the animal by the tail in a direction directly opposite the force
gauge until the mouse falls or is unable to hold itself onto the bars.
Force exerted immediately prior to the mouse releasing is recorded.
Testing should be repeated 3–6 times with approximately 1 min
between each successive test. Results are an average and reported as
force (kg) recorded on the force gauge.
7 Conclusion
Behavioral testing is a fundamental element of PNI research and

mice provide a powerful tool for exploring the origins and relevance
of sickness symptoms. Like with any experimental procedure,
uniform agreement on exact techniques between scientists has not
been achieved. Therefore, the above should be considered an over-
view of how to measure sickness and depressive/anxietal, cognitive,
and physical activity behaviors. As important as appropriate proce-
dures are to successful behavioral testing, pre-experiment consid-
erations are likely the greatest determinate to relevance and
reproducibility. It is essential that mice be housed in environments
devoid of negative stressors and be well adapted to any housing
changes. Variations in equipment and experimental design are usu-
ally irrelevant when compared to unexpected and unpredictable
housing conditions. In fact, wet cages, noise, and unfamiliar
odors are often used as elicitors of adverse biobehaviors. Thus,
consistency, concern, and care in handling mice afford the best
foundation for success. Finally, keen observation is an additional
reward, and making sure to note unanticipated or unusual beha-
viors during testing may lead to innovative discoveries and creation
of new behavioral tests and immunobehavioral paradigms.
Acknowledgments
This research was supported by the National Institutes of Health

(DK064862, NS058525, and AA019357 to G.G.F.), USDA
National Institute of Food and Agriculture, Hatch project
#ILLU971-32.
References
1. Ader R (2000) On the development of muscle damage. Am J Physiol Regul Integr
psychoneuroimmunology. Eur J Pharmacol Comp Physiol 291:R1344–R1348
405(1–3):167–176 16. Rönnb€ack L, Hansson E (2004) On the
2. Maier SF, Watkins LR, Fleshner M (1994) potential role of glutamate transport in men-
Psychoneuroimmunology—the interface tal fatigue. J Neuroinflammation 1:22
between behavior, brain and immunity. Am 17. Tanila H (2017) Testing cognitive functions
Psychol 49(12):1004–1017 in rodent disease models: present pitfalls and
3. Kerschensteiner M, Meinl E, Hohlfeld R future perspectives. Behav Brain Res. https://
(2009) Neuro-immune crosstalk in CNS dis- doi.org/10.1016/j.bbr.2017.05.040
eases. Neuroscience 158:1122–1132 18. Warren EJ et al (1997) Coincidental changes
4. Kelley KW et al (2003) Cytokine-induced in behavior and plasma cortisol in unre-
sickness behavior. Brain Behav Immun 17: strained pigs after intracerebroventricular
S112–S118 injection of tumor necrosis factor-α. Endocri-
5. Irwin MR (2008) Human psychoneuroim- nology 138(6):2365–2371
munology: 20 years of discovery. Brain 19. Antonson AM, Radlowski EC, Lawson MA
Behav Immun 22:129–139 et al (2017) Maternal viral infection during
6. Dantzer R (2004) Cytokine-induced sickness pregnancy elicits anti-social behavior in neo-
behavior: a neuroimmune response to activa- natal piglet offspring independent of postna-
tion of innate immunity. Eur J Pharmacol tal microglial cell activation. Brain Behav
500:399–411 Immun 59:300–312. https://doi.org/10.
7. Johnson DR et al (2007) Acute hypoxia acti- 1016/j.bbi.2016.09.019
vates the neuroimmune system, which diabe- 20. Rytych JL, Elmore MRP, Burton MD et al
tes exacerbates. J Neurosci 27(5):1161–1166 (2012) Early life iron deficiency impairs spatial
8. Dantzer R (2001) Cytokine-induced sickness cognition in neonatal piglets. J Nutr
behavior: mechanisms and implications. Ann 142:2050–2056. https://doi.org/10.3945/
N Y Acad Sci 933:222–234 jn.112.165522
9. Gibertini M (1998) Cytokines and cognitive 21. Grippo AJ (2009) Mechanisms underlying
behavior. Neuroimmunomodulation altered mood and cardiovascular dysfunction:
5:160–165 the value of neurobiological and behavioral
research in animal models. Neurosci Biobehav
10. Wingfield JC et al (2006) Contexts and ethol- Rev 33(2):171–180
ogy of vertebrate aggression: implications for
the evolution of hormone-behavior interac- 22. Johnson A, Hamilton TJ (2017) Modafinil
tions. In: Nelson RJ (ed) Biology of aggres- decreases anxiety-like behaviour in zebrafish.
sion. Oxford University Press, New York PeerJ 5:e2994. https://doi.org/10.7717/
peerj.2994
11. Dantzer R et al (2008) From inflammation to
sickness and depression: when the immune 23. Crawley JN (2003) Behavioral phenotyping
system subjugates the brain. Nat Rev Neu- of rodents. Comp Med 53(2):140–146
rosci 9:46–56 24. Shigemura N et al (2004) Leptin modulates
12. Lavin DN et al (2011) Fasting induces an behavioral responses to sweet substances by
anti-inflammatory effect on the neuroim- influencing peripheral taste structures. Endo-
mune system which a high-fat diet prevents. crinology 145(2):839–847
Obesity 19(8):1586–1594 25. Rodgers RJ, Cole JC (1993) Influence of
13. Yi B, Sahn JJ, Ardestani PM et al (2017) Small social isolation, gender, strain and prior nov-
molecule modulator of sigma 2 receptor is elty on plus-maze behaviour in mice. Physiol
neuroprotective and reduces cognitive deficits Behav 54(4):729–736
and neuroinflammation in experimental mod- 26. Palanza P, Gioiosa L, Parmigiani S (2001)
els of Alzheimer’s disease. J Neurochem Social stress in mice: gender differences and
140:561–575. https://doi.org/10.1111/ effects of estrous cycle and social dominance.
jnc.13917 Physiol Behav 73:411–420
14. Paul RH et al (2000) Fatigue and its impact 27. Lightfoot JT et al (2004) Genetic influence on
on patients with Myasthenia Gravis. Muscle daily wheel running activity level. Physiol
Nerve 23(9):1402–1406 Genomics 19:270–276
15. Carmichael MD et al (2006) Role of brain 28. Basterfield L, Lumley LK, Mathers JC (2009)
IL-1β on fatigue after exercise-induced Wheel running in female C57BL/6J mice:
impact of oestrus and dietary fat and effects
on sleep and body mass. Int J Obes (Lond) behavior of male mice. Asian J Biol Sci
33:212–218 4:362–368
29. Clayton JA, Collins FS (2014) Policy: NIH to 43. Goshen I et al (2003) The role of endogenous
balance sex in cell and animal studies. Nature interleukin-1 in stress-induced adrenal activa-
509:282–283. https://doi.org/10.1038/ tion and adrenalectomy-induced adrenocorti-
509282a cotropic hormone hypersecretion.
30. Hawkley LC, Cacioppo JT (2004) Stress and Endocrinology 144:4453–4458
the aging immune system. Brain Behav 44. Naff KA et al (2007) Noise produced by
Immun 18(2):114–119 vacuuming exceeds the hearing thresholds of
31. Dilger RN, Johnson RW (2008) Aging, C57BL/6 and CD1 mice. J Am Assoc Lab
microglial cell priming, and the discordant Anim Sci 46(1):52–57
central inflammatory response to signals 45. Turnbull AV, Rivier C (1995) Regulation of
from the peripheral immune system. J Leukoc the HPA axis by cytokines. Brain Behav
Biol 84(4):932–939 Immun 9(4):253–275
32. Martin SA et al (2011) Voluntary-wheel exer- 46. Beishuizen A, Thijs LG (2003) Review: endo-
cise training attenuates the visceral adipose, toxin and the hypothalamo-pituitary-adrenal
but not central, inflammatory response to (HPA) axis. Innate Immun 9(1):3–24
LPS in aged C57BL/6J mice. Brain Behav 47. Pfaff J (1974) Noise as an environmental
Immun 25(S2):S217–S218. (Abstract) problem in the animal house. Lab Anim
33. Ma H et al (2010) Effects of diet-induced 8:347–354
obesity and voluntary wheel running on 48. Arakawa H, Cruz S, Deak T (2011) From
bone properties in young male C57BL/6J models to mechanisms: odorant communica-
mice. Calcif Tissue Int 86:411–419 tion as a key determinant of social behavior in
34. Obernier JA, Baldwin RL (2006) Establishing rodents during illness-associated states. Neu-
an appropriate period of acclimatization fol- rosci Biobehav Rev 35(9):1916–1928
lowing transportation of laboratory animals. 49. Alves GJ et al (2010) Odor cues from tumor-
ILAR J 47(4):364–369 bearing mice induces neuroimmune changes.
35. Tuli JS, Smith JA, Morton DB (1995) Stress Behav Brain Res 214:357–367
measurements in mice after transportation. 50. Conour LA, Murray KA, Brown MJ (2006)
Lab Anim 29:132–138 Preparation of animals for research—issues to
36. Jennings M et al (1998) Refining rodent hus- consider for rodents and rabbits. ILAR J 47
bandry: the mouse. Report of the Rodent (4):283–293
Refinement Working Party. Lab Anim 32 51. Balcombe JP, Barnard ND, Sandusky C
(3):233–259 (2004) Laboratory routines cause animal
37. Clénet F et al (2006) Light/dark cycle manip- stress. Contemp Top Lab Anim Sci 43
ulation influences mice behavior in the ele- (6):42–51
vated plus maze. Behav Brain Res 166 52. Hurst JL, West RS (2010) Taming anxiety in
(1):140–149 laboratory mice. Nat Methods 7
38. Ciarleglio CM et al (2009) Population encod- (10):825–826
ing by circadian clock neurons organizes cir- 53. Sun L, Min L, Zhou H et al (2017) Adoles-
cadian behavior. J Neurosci 29 cent social isolation affects schizophrenia-like
(6):1670–1676 behavior and astrocyte biomarkers in the PFC
39. Goulding EH et al (2008) A robust auto- of adult rats. Behav Brain Res 333:258–266.
mated system elucidates mouse home cage https://doi.org/10.1016/j.bbr.2017.07.
behavioral structure. Proc Natl Acad Sci U S 011
A 105(52):20575–20582 54. Koike H et al (2009) Behavioral abnormality
40. National Research Council of the National and pharmacologic response in social
Academies (2011) Guide for the care and isolation-reared mice. Behav Brain Res 202
use of laboratory animals. National Academy (1):114–121
of Sciences, Washington, DC 55. Ma X et al (2011) Social isolation-induced
41. Buchanan JB et al (2008) Cognitive and neu- aggression potentiates anxiety and
roinflammatory consequences of mild depressive-like behavior in male mice sub-
repeated stress are exacerbated in aged mice. jected to unpredictable chronic mild stress.
Psychoneuroendocrinology 33(6):755–765 PLoS One 6(6):e20955
42. Ajarem JS, Safar E, Ahmad M (2011) Effect of 56. Avitsur R, Stark JL, Sheridan JF (2001) Social
ethanol and thermal stresses on the social stress induces glucocorticoid resistance in
subordinate animals. Horm Behav 39 70. Sherry CL et al (2009) Behavioral recovery

(4):247–257 from acute hypoxia is reliant on leptin. Brain
57. Pardon M et al (2004) Repeated sensory con- Behav Immun 23(2):169–175
tact with aggressive mice rapidly leads to an 71. Cao JL et al (2010) Mesolimbic dopamine
anticipatory increase in core body tempera- neurons in the brain reward circuit mediate
ture and physical activity that precedes the susceptibility to social defeat and antidepres-
onset of aversive responding. Eur J Neurosci sant action. J Neurosci 30(49):16453–16458
20(4):1033–1050 72. Basso AM et al (2009) Behavioral profile of
58. Van De Weerd HA et al (1994) Strain specific P2X7 receptor knockout mice in animal mod-
behavioural response to environmental els of depression and anxiety: relevance for
enrichment in the mouse. J Exp Anim Sci neuropsychiatric disorders. Behav Brain Res
36:117–127 198:83–90
59. Olsson AS, Dahlborn K (2001) Improving 73. York JM, Blevins NA, McNeil LK, Freund GG
housing conditions for laboratory mice: a (2013) Mouse short- and long-term locomo-
review of ‘environmental enrichment’. Lab tor activity analyzed by video tracking soft-
Anim 36:243–270 ware. J Vis Exp:e50252–e50252. https://
60. Kent S et al (1992) Sickness behavior as a new doi.org/10.3791/50252
target for drug development. Trends Pharma- 74. Buchanan JB, Sparkman NL, Johnson RW
col Sci 13:24–28 (2010) A neurotoxic regimen of metham-
61. Dantzer R et al (1987) Modulation of social phetamine exacerbates the febrile and neu-
memory in male rats by neurohypophyseal roinflammatory response to a subsequent
peptides. Psychopharmacology (Berl) peripheral immune stimulus. J Neuroinflam-
91:363–368 mation 7:82
62. Park SE et al (2011) Central administration of 75. Fanelli MT, Kaplan ML (1978) Effects of high
insulin-like growth factor-I decreases fat and high carbohydrate diets on the body
depressive-like behavior and brain cytokine composition and oxygen consumption of
expression in mice. J Neuroinflammation 8:12 ob/ob mice. J Nutr 108(9):1491–1500
63. Krzyszton CP et al (2008) Exacerbated 76. Jones BJ, Roberts DJ (1968) The quantitative
fatigue and motor deficits in interleukin-10- measurement of motor incoordination in
deficient mice after peripheral immune stimu- naı̈ve mice using and accelerating rotarod. J
lation. Am J Physiol Regul Integr Comp Pharm Pharmacol 20(4):302–304
Physiol 295(4):R1109–R1114 77. Tarantino LM, Gould TJ, Druhan JP, Bucan
64. Pecaut MJ et al (2002) Behavioral conse- M (2000) Behavior and mutagenesis screens:
quences of radiation exposure to simulated the importance of baseline analysis of inbred
space radiation in the C57BL/6 mouse: strains. Mamm Genome 11:555–564
open field, rotarod, and acoustic startle. 78. Dang MT et al (2006) Disrupted motor
Cogn Affect Behav Neurosci 2(4):329–340 learning and long-term synaptic plasticity in
65. Thor DH, Holloway WR (1982) Social mem- mice lacking NMDAR1 in the striatum. Proc
ory of the male laboratory rat. J Comp Physiol Natl Acad Sci U S A 103(41):15254–15259
Psychol 96(6):1000–1006 79. Carter RJ, Morton AJ, Dunnett SB (2001)
66. Bluthé RM, Dantzer R, Kelley KW (1991) Motor coordination and balance in rodents.
Interleukin-1 mediates behavioural but not Curr Protoc Neurosci Chapter 8:Unit 8.12
metabolic effects of tumor necrosis factor α 80. Loftis JM, Huckans M, Morasco BJ (2010)
in mice. Eur J Pharmacol 209:281–283 Neuroimmune mechanisms of cytokine-
67. Bluthé RM, Schoenen J, Dantzer R (1990) induced depression: current theories and
Androgen-dependent vasopressinergic neu- novel treatment strategies. Neurobiol Dis
rons are involved in social recognition in 37:519–533
rats. Brain Res 519:150–157 81. Deacon RMJ (2006) Burrowing in rodents: a
68. Dantzer R, Bluthé RM, Kelley KW (1991) sensitive method for detecting behavioral dys-
Androgen-dependent vasopressinergic neuro- function. Nat Protoc 1(1):118–121
transmission attenuates interleukin-1-induced 82. Walf AA, Frye CA (2007) The use of the
sickness behavior. Brain Res 557:115–120 elevated plus maze as an assay of anxiety-
69. Abraham J et al (2008) Aging sensitizes mice related behavior in rodents. Nat Protoc 2
to behavioral deficits induced by central (2):322–328
HIV-1 gp120. Neurobiol Aging 29:614–621 83. Cryan JF, Holmes A (2005) The ascent of
mouse: advances in modeling human
depression and anxiety. Nat Rev Drug Discov 96. Takao K, Miyakawa T (2006) Light/dark
4:775–790 transition test for mice. J Vis Exp:
84. Gould TD, Dao DT, Kovacsics CE (2009) e104–e104. https://doi.org/10.3791/104
The open field test. In: Gould T (ed) Mood 97. Cryan JF, Mombereau C, Vassout A (2005)
and anxiety related phenotypes in mice: char- The tail suspension test as a model for asses-
acterization using behavioral tests. Humana sing antidepressant activity: review of pharma-
Press, New York cological and genetic studies in mice.
85. Petit-Demouliere B, Chenu F, Bourin M Neurosci Biobehav Rev 29(4–5):571–625
(2004) Forced swimming test in mice: a 98. Porsolt RD et al (2001) Rodent models of
review of antidepressant activity. Psychophar- depression: forced swimming and tail suspen-
macology (Berl) 177:245–255 sion behavioral despair tests in rats and mice.
86. Avgustinovich DF, Lipina TV, Bondar NP Curr Protoc Neurosci:8.10A.1–8.10A.10
et al (2000) Features of the genetically 99. Lad HV et al (2007) Quantitative traits for the
defined anxiety in mice. Behav Genet tail suspension test: automation, optimiza-
30:101–109 tion, and BXD RI mapping. Mamm Genome
87. Moon ML, Joesting JJ, Blevins NA et al 18:482–491
(2015) IL-4 knock out mice display anxiety- 100. Steru L et al (1985) The tail suspension test: a
like behavior. Behav Genet 45:451–460. new method for screening antidepressants in
https://doi.org/10.1007/s10519-015- mice. Psychopharmacology (Berl)
9714-x 85:367–370
88. Muralidharan A, Kuo A, Jacob M et al (2016) 101. Cryan JF, Markou A, Lucki I (2002) Asses-
Comparison of burrowing and stimuli-evoked sing antidepressant activity in rodents: recent
pain behaviors as end-points in rat models of developments and future needs. Trends Phar-
inflammatory pain and peripheral neuropathic macol Sci 23(5):238–245
pain. Front Behav Neurosci 10:88. https:// 102. Cryan JF, Page ME, Lucki I (2005) Differen-
doi.org/10.3389/fnbeh.2016.00088 tial behavioral effects of the antidepressants
89. Tonelli LH et al (2009) Allergic rhinitis ¬reboxetine, fluoxetine, and moclobemide in
induces anxiety-like behavior and altered a modified forced swim test following chronic
social interaction in rodents. Brain Behav treatment. Psychopharmacology (Berl)
Immun 23:784–793 182:335–344
90. Komada M, Takao K, Miyakawa T (2008) 103. Moreau M et al (2008) Innoculation of Bacil-
Elevated plus maze for mice. J Vis Exp 22: lus Calmette-Guerin to mice induces an acute
e1088 episode of sickness behavior followed by
91. Fromm L et al (2004) Magnesium attenuates chronic depressive-like behavior. Brain Behav
post-traumatic depression/anxiety following Immun 22(7):1087–1095
diffuse traumatic brain injury in rats. J Am 104. Udo H et al (2008) Enhanced adult neuro-
Coll Nutr 23(5):529S–533S genesis and angiogenesis and altered affective
92. Shepherd JK et al (1994) Behavioural and behaviors in mice overexpressing vascular
pharmacological characterisation of the ele- endothelial growth factor 120. J Neurosci
vated “zero-maze” as an animal model of anx- 28(53):14522–14536
iety. Psychopharmacology (Berl) 116:56–64 105. Hédou G et al (2001) An automated analysis
93. Heisler LK et al (1998) Elevated anxiety and of rat behavior in the forced swim test. Phar-
antidepressant-like responses in serotonin macol Biochem Behav 70(1):65–76
5-HT1A receptor knockout mutant mice. 106. DSM-IV (1994) Diagnostic and statistical
Proc Natl Acad Sci U S A 95:15049–15054 manual of mental disorders, 4th edn. Ameri-
94. Ennaceur A (2014) Tests of unconditioned can Psychological Association, Washington,
anxiety — pitfalls and disappointments. DC
Physiol Behav 135:55–71. https://doi.org/ 107. Strekalova T, Steinbusch H (2009) Factors of
10.1016/j.physbeh.2014.05.032 reproducibility of anhedonia induction in a
95. Hascoë M, Bourin M (2009) The mouse chronic stress depression model in mice. In:
light–dark box test. In: Gould T (ed) Mood Gould T (ed) Mood and anxiety related phe-
and anxiety related phenotypes in mice, Neu- notypes in mice: characterization using
romethods, vol 42. Humana Press, Totowa, behavioral tests. Humana Press, New York
NJ. https://doi.org/10.1007/978-1- 108. Niemi MB et al (2008) Neuro-immune asso-
60761-303-9_11 ciative learning. In: Lajtha A, Galoyan A,
Besedovski HO (eds) Handbook of
neurochemistry and molecular neurobiology. 121. Harrison FE et al (2006) Spatial and nonspa-
Springer, New York tial escape strategies in the Barnes maze.
109. Brennan PA, Keverne EB (1997) Neural Learn Mem 13(6):809–819
mechanisms of mammalian olfactory learning. 122. O’Leary TP, Brown RE (2009) Visuo-spatial
Prog Neurobiol 51(4):457–481 learning and memory deficits on the Barnes
110. Bryan KJ et al (2009) Chapter 1: transgenic maze in the 16-month-old APPswe/PS1dE9
mouse models of Alzheimer’s disease: mouse model of Alzheimer’s disease. Behav
¬behavioral testing and considerations. In: Brain Res 201:120–127
Buccafusco JJ (ed) Methods of behavior anal- 123. Vorhees CV, Williams MT (2006) Morris
ysis in neuroscience, 2nd edn. CRC Press, water maze: procedures for assessing spatial
Boca Raton, FL and related forms of learning and memory.
111. Bevins RA, Besheer J (2006) Object recogni- Nat Protoc 1(2):848–858
tion in rats and mice: a one-trial non-match- 124. Rosczyk HA, Sparkman NL, Johnson RW
ing-to-sample learning task to study (2008) Neuroinflammation and cognitive
‘recognition memory’. Nat Protoc 1 function in aged mice following minor sur-
(3):1306–1311 gery. Exp Gerontol 43:840–846
112. Stefanko DP et al (2009) Modulation of long- 125. Carli M, Robbins TW, Evenden JL, Everitt BJ
term memory for object recognition via (1983) Effects of lesions to ascending norad-
HDAC inhibition. Proc Natl Acad Sci U S A renergic neurones on performance of a
106(23):9447–9452 5-choice serial reaction task in rats; implica-
113. Chiu GS, Chatterjee D, Darmody PT et al tions for theories of dorsal noradrenergic
(2012) Hypoxia/reoxygenation impairs bundle function based on selective attention
memory formation via adenosine-dependent and arousal. Behav Brain Res 9:361–380
activation of caspase 1. J Neurosci 126. Horner AE, Heath CJ, Hvoslef-Eide M et al
32:13945–13955. https://doi.org/10. (2013) The touchscreen operant platform for
1523/JNEUROSCI.0704-12.2012 testing learning and memory in rats and mice.
114. Win-Shwe TT, Fujimaki H (2012) Acute Nat Protoc 8:1961–1984. https://doi.org/
administration of toluene affects memory 10.1038/nprot.2013.122
retention in novel object recognition test 127. Bushnell PJ, Strupp BJ (2009) Assessing
and memory function-related gene expression attention in rodents. CRC Press/Taylor &
in mice. J Appl Toxicol 32(4):300–304 Francis, Boca Raton, FL
115. Gainey SJ, Kwakwa KA, Bray JK et al (2016) 128. Young JW, Light GA, Marston HM et al
Short-term high-fat diet (HFD) induced (2009) The 5-choice continuous perfor-
anxiety-like behaviors and cognitive mance test: evidence for a translational test
impairment are improved with treatment by of vigilance for mice. PLoS One 4:e4227.
glyburide. Front Behav Neurosci 10:156. https://doi.org/10.1371/journal.pone.
https://doi.org/10.3389/fnbeh.2016. 0004227
00156 129. Grissom NM, Herdt CT, Desilets J et al
116. Towers AE, Oelschlager ML, Patel J et al (2015) Dissociable deficits of executive func-
(2017) Acute fasting inhibits central caspase- tion caused by gestational adversity are linked
1 activity reducing anxiety-like behavior and to specific transcriptional changes in the pre-
increasing novel object and object location frontal cortex. Neuropsychopharmacology
recognition. Metabolism 71:70–82. https:// 40:1353–1363. https://doi.org/10.1038/
doi.org/10.1016/j.metabol.2017.03.005 npp.2014.313
117. Wehner JM, Radcliffe RA (2004) Cued and 130. Dishman RK et al (2006) Neurobiology of
contextual fear conditioning in mice. Curr exercise. Obesity 14:345–346
Protoc Neurosci Chapter 8:Unit 8.5C 131. Leasure JL, Jones M (2008) Forced a volun-
118. Deacon RMJ, Rawlins JNP (2006) T-maze tary exercise differentially affect brain and
alternation in the rodent. Nat Protoc 1 behavior. Neuroscience 156:456–465
(1):7–12 132. Garland TH Jr et al (2011) The biological
119. Lalonde R (2002) The neurological basis of control of voluntary exercise, spontaneous
spontaneous alternation. Neurosci Biobehav physical activity and daily energy expenditure
Rev 26:91–104 in relation to obesity: human and rodent per-
120. Bekker A et al (2006) Isoflurane preserves spectives. J Exp Biol 214(pt 2):206–229
spatial working memory in adult mice after 133. Noakes TD (2011) Time to move behind a
moderate hypoxia. Anesth Analg brainless exercise physiology: the evidence for
102:1134–1138
complex regulation of human exercise perfor- 135. Messina S, Bitto A, Aguennouz M et al

mance. Appl Physiol Nutr Metab 36:23–35 (2006) Nuclear factor kappa-B blockade
134. Takeshita H, Yamamoto K, Nozato S et al reduces skeletal muscle degeneration and
(2017) Modified forelimb grip strength test enhances muscle function in Mdx mice. Exp
detects aging-associated physiological decline Neurol 198:234–241. https://doi.org/10.
in skeletal muscle function in male mice. Sci 1016/j.expneurol.2005.11.021
Rep 7:42323. https://doi.org/10.1038/
srep42323
Chapter 14
The MRL Model: A Valuable Tool in Studies of Autoimmunity-

Brain Interactions
Boris Šakić
Abstract
The link between systemic autoimmunity, brain pathology, and aberrant behavior is still a largely unex-
plored field of biomedical science. Accumulating evidence points to causal relationships between immune
factors, neurodegeneration, and neuropsychiatric manifestations. By documenting autoimmunity-
associated neuronal degeneration and cytotoxicity of the cerebrospinal fluid from disease-affected subjects,
the murine MRL model had shown high validity in revealing principal pathogenic circuits. In addition,
unlike any other autoimmune strain, MRL mice produce antibodies commonly found in patients suffering
from lupus and other autoimmune disorders. This review highlights importance of the MRL model as a
useful preparation in understanding the links between immune system and brain function.
Key words Autoimmunity, Lupus, Behavioral dysfunction, Neurodegeneration, Immunopsychiatry,

Animal model
1 Introduction
1.1 Regulatory The basic principle of life is to adapt to the relentlessly changing
Metasystem environment, thus providing the basis for survival of an individual
and continuity of a species. When challenged by external and
internal stressors, functional homeostasis in mammals regulates
itself through a coordinated network of the regulatory metasystem,
which comprises diverse interactions between nervous, endocrine,
and immune systems (Fig. 1). While the nervous system is hard-
wired to endocrine glands and immune organs via autonomic
fibers, the immune system communicates with other tissues by
secreting various soluble messengers, such as cytokines, chemo-
kines, and proteins of the complement system [1]. These mediators
can affect brain function by activating the vagal and other nerves
[2], and the secondary messenger system of endothelial cells in
brain vasculature [3], or by diffusing into the brain parenchyma if
the blood–brain barrier becomes more permeable. These factors
can also alter behavior indirectly by changing hormonal activity in
259
260 Boris Šakić
Regulatory Metasystem
Behavior
• physical stressors (light, sound, heat, etc.)

• psychological stressors
Nervous System
HOMEOSTASIS
cytokines
Immune System Endocrine System
hormones (e.g., steroids)
• viruses/bacteria
external stressors
• toxins
• oncogens
internal stressors
• autoantigens
Fig. 1 The principal network of mammal regulatory metasystem. When challenged by various external and
internal stressors, functional homeostasis regulates itself through coordinated interactions between neuroen-
docrine and immune systems, which form the regulatory metasystem. While the CNS is hard-wired to
endocrine glands and immune organs via fibers of the autonomic nervous system, ANS, the immune system
communicates with other organs and tissues via release of various soluble messengers that affect the brain
either directly via neural pathways (e.g., nervous vagus) or indirectly, by affecting hormonal activity in major
endocrine pathways, such as hypothalamic-pituitary-adrenal axis. Imbalanced homeostasis (or allostasis)
ultimately results in increased production of steroids and altered behavior, which are part of the adaptive
responses to acute and chronic stressors. Acute sickness behavior and cortisol release (corticosterone in
animals) are examples of an adaptive mechanism to acute exposure to pathogens. Different forms of mental
and neurodegenerative CNS illnesses are proposed to ensue in genetically susceptible organisms when
disturbances in their regulatory metasystem become severe and chronic. Systemic autoimmune diseases
triggered by myriad of autoantigens (such as in SLE) are example of chronic activation in the regulatory
metasystem
The MRL Model 261
major endocrine pathways, such as the hypothalamic-pituitary-

adrenal axis [4]. Imbalanced homeostasis (or allostasis) in this
regulatory metasystem ultimately results in altered behavior,
which is an integral part of the adaptive responses to acute and
chronic stressors [5].
It is well established that acute and chronic stimuli from the
external environment activate neuroendocrine organs and hor-
mones in several stages [6]. Similarly, infiltrated pathogens and
toxins activate immune cells and organs, which subsequently signal
the central neuronal circuits that control diverse aspects of behav-
ior. Sickness behavior is an archetype of an acute interaction
between immune response and an adaptive behavioral response
[7]. Namely, initial activation of the immune system and behavioral
changes occur often within several hours or days. However, chronic
activation of the immune system in genetically susceptible indivi-
duals may result in various forms of brain dysfunction because of
sustained allostasis in the regulatory metasystem [8–10]. Chronic
inflammation, allergy, and autoimmune reactions are examples of
sustained immune responses that have the potential to irreversibly
harm the brain and disturb the regulatory metasystem in a pro-
longed manner. If viewing this homeostatic network more broadly,
oncogenes and autoantigens can be considered internal stressors
which trigger and maintain sustained immune reactions accompa-
nied by diverse immune cells and their products. Given that immu-
nocytes have a unique capacity to inactivate, destroy, and eliminate
other cells (including cells of own body), it is conceivable to assume
that they have potential to profoundly compromise brain morphol-
ogy and function. Autoimmune diseases are notorious for their
chronicity and severity. As such, they pose a major threat to overall
homeostasis and normal brain functioning. Unfortunately, com-
partmentalization of modern biomedical science into separate dis-
ciplines is largely responsible for our ongoing inability to integrate
knowledge from different disciplines and gain deeper insight into
etiologies of brain disorders accompanied by sustained inflamma-
tory and autoimmune reactions [11]. The chapters below summa-
rize a useful animal model that has confirmed its face and construct
validity over the past 25 years and has become an indispensable tool
in unraveling principal connections between systemic autoimmu-
nity and neuroendocrine system.
1.2 Involvement of Systemic lupus erythematosus (SLE) is a chronic autoimmune/

the Nervous System in inflammatory disorder with a broad spectrum of clinical manifesta-
Systemic Lupus tions. Despite its complexity in presentation and etiology, SLE is a
Erythematosus prime example of causal relationships between systemic autoimmu-
nity, brain pathology, and aberrant behavior. Neurologic and psy-
chiatric (NP) manifestations of unknown etiology occur in up to
75% of patients and are proposed to represent a more severe form of
SLE [12]. They range from diffuse manifestations (e.g., depression,
262 Boris Šakić
anxiety, confusion, psychosis) to focal CNS manifestations (i.e.,

seizures, cerebrovascular disease, chorea, myelopathy, transverse
myelitis, demyelinating syndrome, aseptic meningitis). Moreover,
disorders of the peripheral nervous system (e.g., polyneuropathies,
mononeuropathies, plexopathy, myasthenia gravis) are also com-
mon [13]. A significant number of patients experience NP mani-
festations well before disease onset, at the time of diagnosis, or
within the first year after diagnosis is established [14]. While a
histologically normal brain is a possible finding in NP SLE
(or CNS SLE), hypoperfusion [15–18] and regional metabolic
abnormalities [19–21] are common neuropathological findings
on brain imaging. Brain atrophy, however, is the most frequent
observation on CT scans [22–27] and is proposed to reflect wide-
spread and progressive neuronal loss [28, 29].
Apart from complications induced by kidney damage, infec-
tions, and steroid therapy, more recent studies confirm autoimmu-
nity as a primary mechanism in the etiology of CNS lupus.
Autoantibodies in the serum and cerebrospinal fluid (CSF) of
lupus patients have been proposed as a key factor in brain damage
[30]. Increased intrathecal synthesis (as revealed by an elevated IgG
index and oligoclonal banding) in patients with CNS dysfunction
[31–33] and antigen-specific autoantibodies in the CSF [34] are
shown to be associated with NP manifestations [35]. In many cases
however, the correlational nature of clinical data has led to the
necessity for animal models. In experimental studies, interactions
between autoimmune/inflammatory phenomena and brain func-
tion can be examined in a more systematic and direct way. More
importantly, due to ethical reasons, cause-effect relationships
between specific immune factors and changes in behavior cannot
be tested in clinical studies. They can be studied exclusively with
well-controlled animal models that have significant face, construct,
and predictive validity [36].
2 The MRL Model
2.1 Development Animal models in psychiatry are not replicas of human mental
disorders. Nevertheless, they represent useful preparations where
analogous disease traits, underlying mechanisms, or drug effects
can be studied in a systematic and controlled manner. When asses-
sing the validity of a specific model it is important to identify its
limitations and characteristics that distinguish it from human
pathologies. It is equally important to then consider the choice of
a proper control group and the assessment of central nervous
system damage induced specifically by disease progress.
Most commonly studied spontaneous models of SLE include
the NZB, (NZB NZW)F1 hybrid, BXSB, and MRL mouse
strains. They are characterized by a wide spectrum of autoimmune
The MRL Model 263
manifestations [37] and share common characteristics, such as

hypergammaglobulinemia and elevated antinuclear antibodies
(ANA), which are serological hallmarks of SLE. Each of these
strains has distinct features that make them less or more useful in
studying certain aspects of CNS SLE. Due to profound deficits in
behavior of MRL/MpJ-Faslpr/J (MRL/lpr) mice that appear at a
high frequency during the onset of spontaneous CNS SLE-like
disease, my research over the past 25 years had focused on the
nature of behavioral dysfunction during systemic autoimmunity
and validity of the MRL model in studies of CNS SLE pathogenesis
[38, 39].
There is no doubt that many murine models have made signifi-
cant contributions to our understanding of SLE in general, and
CNS SLE in particular [40]. However, the MRL model has several
advantages over the other strains. First, MRL mice do not have a
high incidence of inherited brain abnormalities [41] that may con-
found assessment of autoimmunity-induced changes in brain func-
tion. They also produce antibodies that are common in SLE
patients. Furthermore, MRL mice are convenient for longitudinal
studies because they develop autoimmune manifestations early in
life [42]. Since MRL mice are among the biggest in the mouse
kingdom, technically demanding procedures are feasible (e.g., CSF
collection from the cisterna magna, retro-orbital bleeding, intra-
cerebro-ventricular cannulation). Most importantly, there are two
congenic substrains that differ in <0.1% of their genetic back-
ground and disease onset. High levels of circulating cytokines,
autoantibodies, and behavioral deficits appear in the
MRL/MpJ-Faslpr/J substrain (MRL/lpr, stock 485 in The Jackson
Laboratory) within the first 2 months of life [43]. Conversely, the
congenic MRL/MpJ (MRL +/+) mice (stock 486) develop lupus-
like disease much later in life and are asymptomatic at this age, thus
representing an adequate control. Indeed, the traits above have
helped in establishing the MRL model of CNS SLE, which is largely
based on discrepancy in behavior and brain morphology between
MRL/lpr and MRL +/+ mice [44, 45].
The MRL/MpJ lymphoproliferation wild-type strain was gen-
erated at The Jackson Laboratory from a series of crosses with
strains C57BL/6J (0.3%), C3H/HeDi (12.1%), AKR/J (12.6%),
and LG/J (75%) (Fig. 2). During development of this strain, the
spontaneous mutation Faslpr was found in MRL/lpr mice at F12.
This mutation of a single autosomal recessive gene (designated
lymphoproliferative, lpr gene) on chromosome 19 results in mas-
sive lymphadenopathy, induced by the accumulation of abnormal
T-lymphocytes [46]. There is also a spontaneous loss of function
associated with the Fas mutation, which leads to a deficit in apo-
ptotic Fas receptor (FasR) expression in MRL/lpr mice
[47, 48]. Largely due to aberrant apoptosis of self-reactive clones
in the thymus [46], these mice develop an accelerated form of
264 Boris Šakić
LG/J (75%) AKR/J (12.6%) C3H/HeDi (12.1%) C57BL/6J (0.3%)
MRL/MpJ substrain
(stock 486)
MRL/MpJ-Faslpr/J substrain
(stock 485)
MRL/MpJ lpr/Fas substrain

(stock 6825)
Fig. 2 Genetic background of the MRL mouse strain. Four different strains of mice were used in mid-1970s to
produce the lupus-prone MRL (originally Murphy—Root Large) strain in The Jackson Laboratory. In compari-
son to the original stock 486, spontaneous Faslpr mutation accounted for accelerated disease development in
the stock 485 (MRL/lpr mice, shown on the picture). However, unexplained decline in autoimmune phenotype
was observed in this stock over the past decade. The subsequent stock was labeled 6825. As the stock 485, it
carries the Faslpr mutation, thus rendering this group a more adequate control in behavioral studies that test
causal links between systemic autoimmunity and brain damage
SLE-like disease, characterized by skin lesions, alopecia, arthritis,

and immune complex glomerulonephrosis. Starting around
3 months of age, levels of circulating immune complexes rise
promptly in the MRL/lpr substrain, but not in MRL +/+ controls
[42]. Female MRL/lpr mice die at an average age of 17 weeks of
age and males at 22 weeks.
In a series of early studies, we documented brain pathology,
CSF cytotoxicity, and deficits in emotional reactivity of both male
and female MRL/lpr mice [39]. However, we were also aware that
the expression of FasR plays an important role in neuronal apopto-
sis during postnatal development [49]. Although the effectiveness
of immunosuppressive treatment in preserving normal neuronal
morphology pointed to a pathogenic role of autoimmunity
[50–53], the possibility that the FasR deficiency in the CNS con-
tributes to aberrant behavior in the MRL/lpr substrain could not
be rejected until a new, long-lived stock of MRL/lpr mice was
generated.
Namely, a gradual decline in the autoimmune phenotype was
observed in the MRL/lpr colony over the past decade. This phe-
nomenon of unknown origin was also accompanied by a gradual
The MRL Model 265
decline in the number and severity of behavioral deficits [54]. In

2007, the supplier (The Jackson Laboratory, Bar Harbor, ME)
publicly announced phenotype loss in the stock 485 and reassigned
it into stock 6825 (MRL/MpJ-Faslpr/JJ), which still carried the
FasR deficiency in their genome (http://jaxmice.jax.org/strain/
006825.html). Since embryos of the original stock 485 were cryo-
preserved in 1993, the original MRL/lpr stock was reestablished in
the summer of 2008. In brief, stock 485 has been proposed as a
model of CNS SLE in reference to control stock 486 [38, 55],
while stock 6825 (“mutated” stock 485) emerged as a new,
improved control group because it shares FasR deficiency with the
stock 485.
2.2 Autoimmunity- MRL/lpr and MRL+/+ mice are comparable in many respects
Associated Behavioral (e.g., appearance, size, reproductive age), except in the onset of
Syndrome autoimmunity and neurobehavioral dysfunction. Disease progres-
sion in MRL/lpr mice parallels emergence of behavioral deficits
often seen after exposure to chronic stressors [38]. The constella-
tion of behavioral differences between age- and sex-matched
MRL/lpr and MRL +/+ mice was operationally termed “autoim-
munity-associated behavioral syndrome” (AABS), which has been
largely characterized by progressive anxiety- and depressive-like
behaviors [39]. These constructs were supported by increased
thigmotaxis of MRL/lpr mice in large arenas, impaired exploration
of novel objects and spaces, performance deficits in the plus-maze
and step-down tests, excessive floating in the forced swim test
[56–58], reduced responsiveness to palatable stimulation
[51, 59], and reduced isolation-induced inter-male fighting
[60]. Moreover, their “cognitive inflexibility” and poor spatial
learning were noted in the Morris water maze [61, 62] and sponta-
neous alternation test [53]. In addition, diseased MRL/lpr animals
show lower nocturnal and open-field activity levels, as well as
significant deficiencies in neurological [62, 63] and psychomotor
tests [64].
2.3 Breached The blood–brain barrier (BBB) is formed by endothelial cells that
Blood–Brain Barrier tightly line capillaries and blood vessels of the brain. The BBB has
an important role in maintaining a well-regulated CNS microenvi-
ronment for reliable neuronal signaling. When its normal function
is disrupted, large molecules and cells can infiltrate into the brain
and lead to CNS damage. In many CNS SLE patients, the BBB
becomes transiently or permanently breached, as evidenced by an
increased albumin quotient [65]. Increased BBB permeability is an
important permissive condition which allows diffusion of cytokines
and anti-neuronal autoantibodies into cerebrospinal fluid (CSF)
and accounts for subsequent CNS manifestations [66].
Similar to patients, a breached BBB has been observed in
diseased MRL/lpr mice. This damage occurs at a very early age
266 Boris Šakić
(7–8 weeks) and is evidenced by perivascular leakage of IgG anti-

bodies and increased albumin levels in the CSF [67, 68]. In addi-
tion to soluble factors, significant infiltration of macrophages, T
cells, B cells, and plasma cells can be demonstrated in the choroid
plexus, meninges, and brain parenchyma of MRL/lpr mice
[62, 69–71].
2.4 CNS The role of neuroinflammation is emerging as an important com-

Inflammation ponent of lupus-like disease. In addition to the previously men-
tioned infiltration of immunocytes into the brain parenchyma,
several other types of inflammation-related pathologies have been
observed in brains of MRL/lpr mice. They include upregulation of
adhesion molecules [72, 73], expression of mRNA for
pro-inflammatory cytokines [74, 75], and deposition of comple-
ment proteins [76] within brains of MRL/lpr mice. Major histo-
compatibility complex (MHC) upregulation [77] and F4/80
microglia staining provide additional evidence for microglia-
induced neuronal excitotoxicity in these animals [78]. When asses-
sing the global pattern of neuronal damage in MRL/lpr brains
[79], a pattern of degeneration emerges that is typically seen in
hydrocephalus, meningoencephalitis, and hypoglycemic encepha-
lopathy [80–84], all of which result in cerebritis. Supporting this
notion, others have reported that MRL/lpr mice have an increased
incidence of hydrocephalus [85], meningoencephalitis [69], and
altered glucose metabolism [86], with complement activation as a
likely precursor to cerebral edema [87].
Congenic MRL+/+ animals were found to have greater cell loss
and a more aggressive, sustained microglial inflammatory response
following mechanical injury and breakdown of the BBB
[88]. Unlike MRL/lpr animals, which develop spontaneous BBB
disruption, MRL+/+ mice do not show evidence of CNS damage
under normal conditions. However, the MRL strain may have an
inherent propensity toward exaggerated CNS inflammatory
responses. For example, considering that the substantia nigra of
mice has more microglia than other areas [89], one may assume
that this neuroanatomical trait in lupus mice reflects the region-
specific susceptibility to inflammatory and excitotoxic metabolites
produced by activated microglia [90].
While neuroinflammation appears to be a contributing factor
to disease in MRL/lpr animals, a distinct inflammatory response is
not commonly seen in the brains of lupus patients. Some of the
most common neuropathological findings in SLE, however, are
small vessel cerebral vasculopathy and microinfarcts. These
observed features likely reflect the end result of repeated episodes
of acute inflammation in the small brain vessels [91]. Elevated levels
of soluble adhesion molecules in serum and CSF of patients with
CNS involvement [92] also suggest that neuroinflammatory con-
ditions may play an understated role in certain NP manifestations.
The MRL Model 267
2.5 Brain Pathology Clinical studies clearly demonstrate that NP manifestations are
accompanied by cerebral atrophy [93], progressive neuronal loss
[20, 28], and parenchymal lesions [94]. Similar to CNS SLE
patients and effects of chronic stress, MRL/lpr mice show brain
atrophy and ventricular enlargement alongside behavioral deficits
detected at the onset of SLE-like disease [53, 95, 96].
The MRL strain does not show a high incidence of inherited
neuroanatomical abnormalities [41], which minimize the possibil-
ity of congenital defects confounding the study of disease-induced
neurodegeneration. At the onset of autoimmune symptoms in
MRL/lpr mice, reports of reduced complexity of pyramidal neu-
rons, reduced brain weights [96], and selectively neurotoxic CSF
[97] provided indirect evidence of neuronal damage in diseased
animals. Direct evidence of neuronal death, however, was first
confirmed in MRL/lpr brains using the Fluoro Jade B (FJB) cyto-
chemical stain (specific for dying neurons). A small percentage of
these neurons were subsequently found to contain TdT-labeled
apoptotic nuclei, and co-localized with FJB [79] and anti-
neurofilament staining [76]. Moreover, while the size of hippocam-
pal fields and neuronal density are not reduced in young
Fas-deficient lpr mice [98], cell densities are reduced within the
hippocampus, cortex [53], and midbrain [90] of aged/diseased
lupus mice.
In addition to mature neurons, recent findings suggest that
progenitor cells also degenerate in MRL/lpr brains. More specifi-
cally, the subventricular zone [99], subgranular zone [78, 79], and
substantia nigra [90], all of which are known to contain prolifera-
tive progenitor cells capable of neurogenesis [100], show signs of
damage. CSF from diseased lupus mice is also cytotoxic to neurons
and neuronal progenitor cells in vitro [101], thus supporting a link
between toxic CSF IgG and neuronal/progenitor cell damage
[68]. If in vitro findings are predictive of in vivo events, then
autoimmune-induced lesions of germinal layers may reduce the
developmental and regenerative capacity of MRL/lpr brains. An
impairment in this process would likely exacerbate subsequent
autoimmune/inflammatory-mediated neuronal death and behav-
ioral deficits. For example, an impaired capacity for hippocampal
neurogenesis could account for the cognitive impairments
observed in these animals [53]. Stress hormones, known to be
chronically elevated in lupus mice [102], have also been shown to
inhibit cell proliferation and neurogenesis [103]. Therefore, one
may assume that such mechanisms account for impaired brain
growth and regeneration along the progression of autoimmune
disease.
Despite parallels between the emergence of behavioral dysfunc-
tion and systemic autoimmunity, there is no firm evidence that
brain pathology accounts for aberrant behavior in the MRL
model. However, significant correlations suggest principal links
268 Boris Šakić
between structural brain damage and functional/behavioral impair-

ments in MRL/lpr mice. For example, deficits in spatial learning/
memory emerge concomitantly with hippocampal damage [53],
aberrant performance in the sucrose preference test coincides with
lesions of the nucleus accumbens [104], and decreased locomotor
activity accompanies degeneration in the substantia nigra
[90]. Although the mechanisms underlying these deficits are not
well understood [105], recent pharmacological evidence supports a
link between dopaminergic circuit damage and AABS.
2.6 Proposed The causative role of autoimmunity and inflammation in the path-
Neuropathogenic ogenesis of AABS has been supported by studies employing the
Factors and immunosuppressive drug cyclophosphamide (CY). Sustained treat-
Mechanisms ment with CY from an early age prevents several behavioral deficits
and brain pathology in MRL/lpr mice [51–53, 70]. More specifi-
cally, CY abolishes substrain differences in anxiety- and motivation-
related behaviors, as suggested by restored novel object explora-
tion, increased responsiveness to a palatable sucrose solution, and
normalized nocturnal activity. Although systemic autoimmunity
and inflammation have been proposed as key factors, the possibility
that subtle genetic dissimilarities, imbalanced hormonal produc-
tion, and peripheral tissue involvement contribute to certain
aspects of AABS could not be discounted. However, the use of
newly developed stock 6825 rejected the possibility that AABS is
entirely accounted for by FasR mutation in neuronal cells. Namely,
the constellation of differences between the lpr stocks 485 and
6825 confirmed the hypothesis that the lpr mutation per se does
not fully account for the brain pathology and altered behavior in
the stock 485 [106]. Together with significant correlations
between immunological and behavioral measures, this study sug-
gested that soluble immune factors play a key role in CNS patho-
genesis. In addition, combined use of immunoprecipitation with
homogenates of unaffected brains, 2-dimensional differential in-gel
electrophoresis, and mass spectrometry revealed strong binding of
CSF IgG antibodies to cytoskeletal antigens in brains of MRL/lpr
mice. This finding is consistent with the proposed pathogenic role
of brain-reactive autoantibodies (BRA) in the etiology of AABS.
CNS SLE is frequently accompanied by BRA cross-reactive
with diverse brain-specific and systemic antigens [12]. Most of
these autoantibodies have been identified on the basis of their
binding to tissues and cells, including neuroblastoma and glioblas-
toma cell lines [107]. There are also autoantibodies against lym-
phocytes, capable of being adsorbed by brain tissue
[108, 109]. Some antibodies can react specifically to CNS neurons
[110, 111], neuronal cytoplasm [112], and neuronal receptors
[113, 114]. A current literature review proposes approximately
20 pathogenic BRA [115], including anti-ribosomal, anticardioli-
pin, antiphospholipid, and more recently antibodies to an NMDA
The MRL Model 269
receptor, or anti-NR2 antibodies [30, 113, 116–121]. Consistent

with clinical findings in CNS SLE patients [122–125], our recent
study with MRL/lpr mice revealed significant reactivity of their
CSF and serum IgG molecules to cytoskeletal proteins
[106]. When the source of BRA is considered, both clinical studies
and studies with MRL/lpr mice suggest that BRA from CSF is
more pathogenic than BRA from serum [97, 101,
126–128]. While the mechanism by which circulating BRAs access
the brain is not well understood, aberrant behavioral and emotional
manifestations in human and murine forms of lupus suggest that
multiple CNS antigens and sites are targeted [113, 119, 120,
129–131]. A significant relationship between aberrant behavioral
performance and specific BRA has been reported in diseased
MRL/lpr mice [132–134]. However, certain subsets of BRA
might be more important in the induction of permanent neuronal
damage, while other subsets might merely affect neuronal function-
ing in a transient fashion [135]. Given these complex modes of
action, more direct, invasive studies are required to prove causality
between specific classes of pathogenic BRA and certain behavioral
deficit/brain pathology.
A dysregulated cytokine network is also hypothesized to play an
important role in the etiology of CNS SLE. In general,
pro-inflammatory cytokines are instrumental in expanding periph-
eral immune reactions to the CNS via endothelial activation
[65, 136]. Early increases in serum levels of IL-1 beta,
TNF-alpha, IL-6, and interferon-gamma are principal events facil-
itating the hyperproduction and maintenance of autoantibodies in
the MRL/lpr substrain [59, 137–139]. A significant correlation
between serum IL-6 levels and responsiveness to a palatable stimu-
lus was documented in MRl/lpr mice [140]. Using adeno-vector
methodology, a direct cause-effect relationship was shown for both
IL-6 and interferon-gamma [141, 142]. Similarly, administration
of TNF-alpha enhanced intracellular adhesion molecule (ICAM)-
dependent leukocyte-endothelial interactions in MRL/lpr brains
[143]. Furthermore, these interactions can be prevented by anti-
body blockade of pro-inflammatory cytokines, or ICAM
[63, 72]. It is important to emphasize that small amounts of
pro-inflammatory cytokines can also cross the BBB by specific
transport mechanisms [144–146] and activate receptors on endo-
thelial cells of the brain vasculature to release other mediators (e.g.,
cytokines, nitric oxide, prostaglandins) into the CSF and brain
parenchyma [147, 148]. However, it remains to be determined
whether physiological doses of circulating cytokines can compro-
mise viability of central neurons. On the other hand, it is well
documented that pro-inflammatory cytokines have the ability to
affect the stress hormone system and alter behavior
[149–153]. More specifically, these cytokines can regulate cortico-
steroid levels and autoimmunity through receptors in adrenal gland
270 Boris Šakić
and pituitary [152, 154–157]. Sustained activation of the pituitary-

adrenal axis in MRL/lpr mice is evidenced by increased central
expression of arginine/vasopressin mRNA [158, 159] and high
levels of corticosterone [102]. Although an imbalanced neuro-
immuno-endocrine network is confirmed to play a key role in the
etiology of brain damage, it is still not clear to which extent central
neurons are damaged by an inflammation-driven upregulation in
corticosterone production versus direct cytotoxicity of self-reactive
immunocytes and their products. Corticosterone-induced atrophy
of neurons may be reversible, but it can also be indicative of an early
stage of neurodegeneration [160, 161]. Indeed, diseased MRL/lpr
mice show profound neuronal spine loss [96] which can be further
exacerbated with chronic pretreatment with corticosterone.
Namely, sustained corticosterone administration attenuated signs
of autoimmune disease, but led to profound dendritic spine deteri-
oration, as revealed by the Golgi method [162]. Therefore, based
on chronically elevated serum corticosterone levels in MRL/lpr
mice [102], one may hypothesize that sustained endogenous
immunosuppression is a precursor and necessary factor for neuro-
degenerative events that occur when autoimmune mechanisms
prevail at later stages of the disease (Fig. 3). Indeed, changes in
the morphology of neuronal dendrites, cerebral atrophy, and
immunoreactive ubiquitin particles (denoting axon terminal
degeneration) occur by 14 weeks of age in MRL/lpr brains, but
progressive neurodegeneration and microglial activation do not
become pronounced until terminal stages of the disease
(~5 months). Few MRL/lpr mice survive beyond 6 months of
age [37], which may be attributed to profound CNS damage and
brain edema [86]. It is viable that such sustained “allostatic load”
may ultimately provide the basis for vulnerability of central neurons
[163], bona fide neurodegeneration, and behavioral dysfunction
[53, 79, 96]. Future studies examining the effects of cytokines on
the endocrine axes and the possibility that adrenalectomy prevents
(or delays) central neurodegeneration in MRL/lpr mice are
warranted.
Aberrant cytokine production also modulates systemic autoim-
munity by sustained activation of B-cells, which later differentiate
into pathogenic autoantibody-forming cells. These pathogenic
autoantibodies are a prelude to immune complex disease [164], a
common feature of lupus. Indeed, deposition of antigen-antibody
complexes (immune aggregates) in choroidal blood vessels has
been associated with NP manifestations, while vascular deposits
within the choroid plexus (CP) are accompanied by histopatholo-
gical evidence of inflammation [165]. Therefore, autoantibodies to
endothelial cells, as well as the pathogenic action of circulating-
immune complexes (CIC) on microvessels, likely contribute to, if
not cause, endothelial cell damage and breakdown of the BBB in
lupus patients and MRL/lpr mice [65, 68, 71]. Subsequent
The MRL Model 271
Inflammatory phase - functional damage
altered mood and emotional reactivity
cytokine production in the brain
leukocyte entrapment
7.
ACTH
over-expression of CAM activation of nervous vagus
4.
3.
lymphocyte hyperactivity 6. 1.
SHs
5. 1. SHs
2.
Pro-inflammatory cytokines
Autoimmune phase - structural damage
psychosis, dementia, seizures
leukocyte infiltration via CP, BRA to neurons and neural stem cells
microglial activation
9.
8.
Fig. 3 Proposed phases and pathways of the CNS damage during systemic autoimmune disease. Behavioral
dysfunction and brain damage in lupus-like disease may result from chronic stress-like response induced by
sustained autoimmunity and inflammation. In SLE patients and lupus-prone MRL/lpr mice, spontaneous onset
of systemic inflammation and autoimmunity are characterized by increased levels of pro-inflammatory
cytokines, which may activate pituitary-adrenal axis and promote sustained release of glucocorticoids. In
turn, steroid hormones suppress the immune system at multiple levels. Due to chronic nature of the disease,
glucocorticoids, cytokines, and other immune components remain elevated, thus compromising the integrity
of the blood–brain barrier and neuronal function. (A) The inflammatory phase is largely associated with early
272 Boris Šakić
infiltration of various immunocytes into the brain parenchyma of

MRL/lpr mice [70, 71] may merely be another step in a cascade of
neuropathogenic events.
There are numerous factors that can induce excitotoxic dam-
age. In an injured or immunologically challenged brain, cytokine-
producing microglia appear to play an important role. Microglia
readily activate by transforming from a ramified, resting state into
amoeboid cells that express MHC molecules [89], as seen in
MRL/lpr brains [77]. Inflammatory responses are then perpetu-
ated by both cyclooxygenase (COX) and nitric oxide (NO). In
MRL/lpr mice, however, abnormalities in prostaglandin produc-
tion were reported [166], likely rendering COX inhibition ineffec-
tive in ameliorating AABS [78]. Indirectly, these results suggest
that inducible NO synthase and glutamate system are important
mediators of neuroinflammation and neuronal apoptosis in these
animals [76]. Ultrastructural evidence obtained from electron
microscopy [78] and a study reporting significant increases in
glutamine, glutamate, and lactate concentrations in MRL/lpr
brains [86] support the notion of excitotoxic neuronal death. In
the case of glutamate toxicity, this hypothesis is supported by
evidence of anti-NMDA receptor antibodies resulting in neuronal
apoptosis in the mouse brain [113], a similar IgG-mediated mech-
anism induced by CSF from MRL/lpr mice [101], and the pres-
ence of high levels of anti-NMDA receptor antibodies in their CSF
(Betty Diamond, personal communication). Considering that anti-
CD4 treatment [167] and complement inhibitors [76, 168] ame-
liorated CNS disease in MRL/lpr mice, one may assume that
combination of cellular inflammatory and autoimmune factors
(operational at different stages of the disease) underlies brain
pathology and aberrant behavior (summarized in Fig. 4).
ä
Fig. 3 (continued) functional damage of the brain. The upregulation in circulating pro-inflammatory cytokines
(e.g., IL-1β, IL-6, TNF-α, and IFN-γ) is an initial serologic event that disturbs the activity of the immune
network. These cytokines can also activate the hypothalamic-pituitary-adrenal axis [1], which downregulates
peripheral inflammation via increased production of steroid hormones, SHs [2]. However, in addition to the
activation of major inhibitory signals from the hippocampus to hypothalamic paraventricular nucleus,
sustained binding of steroids to receptors in central neurons [3] induces stress-like manifestations (e.g.,
emotional disturbances, impaired mood) which are largely under the control of the limbic system. This effect
on brain function is further amplified by cytokine-induced activation of the nervous vagus and activation of
glial cells in the hypothalamus [4]. Moreover, activated lymphocytes [5] and cytokine-induced overexpression
of cell adhesion molecules on endothelial cells of the BBB and choroid plexus, CP [6], are conducive of
immunocyte entrapment [7]. (B) The autoimmune phase is largely characterized by structural damage, such
as neurodegeneration and brain atrophy. Chronic inflammatory responses increase the permeability of the
BBB and CP, thus leading to infiltration of immunocytes into perivascular spaces and cerebrospinal fluid
[8]. Structural brain damage can result from neurotoxic metabolites that accumulate after sustained activation
of microglia [9] and chronic binding of brain-reactive antibodies (BRA) to adult and immature neurons
[10]. Loss of periventricular and cortical mass may underlie psychosis, dementia, and seizures that frequently
accompany neuropsychiatric lupus. Note: Dashed lines represent inhibitory pathways
The MRL Model 273
neurons
B cells
antibodies
MAC
complement
T cells chemokines
cytokines MMP
ROS oligodendrocytes
macrophages
blood-brain barrier
PGE
microglia
astrocytes
Fig. 4 Summary of putative factors and cellular mechanisms underlying neuronal damage in CNS lupus. When
BBB is breached, various immune cells and mediators can compromise the viability of brain cells at different
stages of disease progress, age, and genetic deficits in affected individuals. These factors include cytotoxic T
cells, macrophages, brain-reactive antibodies to surface and intracellular receptors, the C5b-9 MAC, MMPs,
and reactive oxygen species, ROS
In comparison to other neurotransmitters, central dopamine

system activity (implicated in reward, movement, and cognitive
processes) is most profoundly altered in brains of MRL/lpr mice
[169]. A series of pharmacological studies suggest that damage to
central dopaminergic circuits accounts for at least some behavioral
deficits in this substrain. In particular, chronic injection with the
selective D2/D3 agonist quinpirole induced self-injurious behavior
[170], while acute injection with the selective D1/D2dopamine
agonist apomorphine increased rotational behavior [90]. In the
sucrose preference paradigm, acute injection with the indirect
dopamine agonist D-amphetamine failed to alter the response
rates of diseased animals to sucrose solutions [104]. Taken
together, these results link neuropathological findings of dopami-
nergic cell death in nigrostriatal, mesolimbic, and mesocortical
pathways to behavioral deficits in locomotion, motivated behavior,
and learning.
Despite the above evidence, more recent studies suggest that
genetic, endocrine, and systemic factors also contribute to the
altered behavioral profile of MRL/lpr mice (Fig. 5). For example,
274 Boris Šakić
Genetics
Behavioral phenotype
(PNS + CNS function)
Autoimmunity Endocrine System

(prenatal + postnatal)
Fig. 5 Interactions among genetic, immune, and endocrine factors in the

etiology of an altered behavioral profile in autoimmune MRL/lpr mice. The
ultimate, yet daunting, task in this model is to distinguish central from
peripheral autoimmunity-induced dysfunction that affects behavioral
performance in tasks reflecting neurological function, spontaneous locomotor
activity, emotional reactivity, and learning/memory capacity
Wen and colleagues assessed the role of B cells and autoantibodies

in the etiology of behavioral manifestations and brain pathology
using B cell-deficient (JhD/MRL/lpr) and B cell-depleted
(hCD20-DTA/MRL/lpr) mice [171]. They found that mice
from new strains, similar to lupus-prone MRL/lpr mice, exhibit
profound depression-like behavior in the forced swim test, “cogni-
tive” deficits in the object placement and object recognition tests,
increased blood–brain barrier permeability, brain cell apoptosis,
and upregulated cytokine expression in comparison to the
MRL/MpJ controls. The authors inferred that B cells and/or
autoantibodies in serum and CSF are not required for key features
of “neuropsychiatric” disease in murine CNS SLE. The results
obtained are in line with our results indicating that excessive immo-
bility is also seen in 7-week-old 6825 mice, which are asymptomatic
at this age [172]. Another recent study with bone marrow chimeras
revealed that transfer of healthy MRL+/+ bone marrow to MRL/lpr
mice led to marked attenuation of systemic disease, but did not alter
their behavioral phenotype [173]. Taken together, these results
suggest that the contribution of non-lpr-related genes accounts
for performance differences in certain behavioral paradigms
between the lpr stocks (485, 6825) and the wild-type MRL+/+
substrain (stock 486). However, the role of autoantibodies in the
etiology of AABS cannot be readily discounted because early brain
damage can be induced by maternal autoantibodies in embryogen-
esis [174, 175]. In addition, recent studies employing novel
immune markers and treatment modalities support the notion
that certain behavioral deficits and brain pathologies are induced
by systemic autoimmunity and inflammation [176–182].
When compared to age-matched 6825 controls, 7-week-old
485 mice exhibit elevated serum corticosterone, enlarged left
The MRL Model 275
adrenal gland, and enhanced hematoxylin/eosin staining in the

hypothalamic paraventricular nucleus [172]. In addition to changes
in the endocrine system, such young animals also have increased
IgG levels in their cerebrospinal fluid (CSF). This was observed well
before high serum autoantibody levels and splenomegaly are
detectable—findings that were recently confirmed by another
research group [171]. With respect to disturbed sensory function,
diseased MRL/lpr mice also exhibit reduced gustatory nerve and
behavioral responses to bitter and sweet solutions [183]. In con-
trast, their responses are comparable when salty and sour com-
pounds are used. These results suggest that type II taste receptor
cells, which are essential for bitter and sweet taste reception and
signaling, are selectively affected in MRL/lpr mice. Although the
studies were conducted at an age when lupus-like disease is florid,
they bring into question the exclusivity of our premise on a defi-
ciency in central reward circuits when younger cohorts of MRL/lpr
mice were tested [59, 172]. Along the same line, subtle deficits in
olfactory function, such as hyposmia [184], may contribute to
altered performances in tasks that depend on olfaction, yet are
traditionally considered to reflect anxiety-like behavior [56]. There-
fore, although performance deficits in certain behavioral tasks may
be induced by the development of lupus-like disease, more recent
results raise the possibility that damage to sensory inputs plays a
significant role in their etiology.
3 Summary
Compared to other models of systemic autoimmune disease, the

MRL model has several key characteristics that render it a valuable
tool in studies of autoimmunity–brain interactions [44]. First,
compared to “induced” models of CNS SLE, MRL/lpr mice spon-
taneously develop manifestations that match the human disorder in
complexity, chronicity, and severity. This includes a variety of intra-
thecal BRA and brain atrophy, which are both characteristic of more
severe forms of CNS SLE. Second, in behavioral studies, the MRL
model is well controlled with MRL+/+ (stock 486) and MRL/lpr/
JJ groups (stock 6825). The differences in genomes are <0.1%
(when compared to stock 486) and even less in case of stock
6825. Lastly, the MRL model is well defined at genetic, cellular,
and behavioral levels. In particular, the lpr lesion (encoding for
FasR deficiency) on chromosome 19, “double-negative” clones of
lymphocytes, neurodegeneration, and behavioral deficits are well-
explored phenomena, replicated by different research groups. This
abundance of knowledge allows diverse manipulations at all system
levels, thus advancing our understanding of relationships between
genes, autoimmunity, and brain dysfunction.
276 Boris Šakić
The roles of stress hormones in initial modulation of brain

morphology and behavior of MRL/lpr mice appear important
[102] and may act similar to iatrogenic effects of sustained cortico-
steroid therapy on brains of CNS SLE patients [26]. Progressive
neuronal death and microglial activation are concomitant with the
development of more severe systemic autoimmune/inflammatory
disease and more diverse AABS. Although other neurotransmitter
systems are likely involved, dopaminergic neurons seem to be a
specific target of autoimmune reactions, possibly accounting for
early deficits in emotional reactivity and motivated behavior. Fur-
ther interplay among activated microglia, neuroactive cytokines,
BRA, and cytotoxic T cells likely leads to an accumulation of
neurotoxic metabolites, edema, and brain atrophy. Although
adult neurons and neural progenitors are targeted in diseased
MRL/lpr mice [90, 97, 101, 185], the possible effect of autoim-
munity on brain cells may be extended into prenatal life and influ-
ences from the maternal immune system [175].
While the MRL/lpr substrain displays many characteristics that
resemble human CNS SLE, there are limitations which need be
considered when studying this model. First, human CNS SLE has
relapsing-remitting presentations of symptoms, while MRL/lpr
mice show a progressive and unrelenting course of disease. Second,
human SLE shows a strong gender preference (i.e., about 9–10
times more female than male patients). Although the disease starts
few weeks earlier in female mice, the MRL/lpr substrain does not
show such strong gender bias, possibly due to a different hormonal
milieu in human and murine forms of lupus. Lastly, while more
sophisticated assessments are administered in diagnosing human
CNS SLE, “psychiatric manifestations” in MRL/lpr mice are
merely constructs that can be proposed from dissimilar perfor-
mances in behavioral tests. Despite these limitations, the MRL
model remains useful and unique in understanding the complex
immuno-neuroendocrine interactions. There is no doubt that dee-
per understanding of pathogenic pathways and neurotoxic media-
tors in these animals may help in elucidating CNS SLE etiology and
provide a basis for new treatment modalities in brain disorders with
autoimmune origin.
Acknowledgments
This work was supported by the grants from the Ontario Mental
Health Foundation and Canadian Institutes of Heath Research.
The MRL Model 277
References
1. Ader R, Felten D, Cohen N (2001) Psycho- manifestations. Clin Exp Rheumatol 14
neuroimmunology, vol 1-2, 3rd edn. Aca- (Suppl. 16):S23–S29
demic Press, New York 14. van Dam AP, Wekking EM, Oomen HA
2. Goehler LE, Gaykema RP, Hansen MK, (1991) Psychiatric symptoms as features of
Anderson K, Maier SF, Watkins LR (2000) systemic lupus erythematosus. Psychother
Vagal immune-to-brain communication: a vis- Psychosom 55(2–4):132–140
ceral chemosensory pathway. Auton Neurosci 15. Colamussi P, Giganti M, Cittanti C,
85(1–3):49–59 Dovigo L, Trotta F, Tola MR, Tamarozzi R,
3. Benveniste EN, Huneycutt BS, Shrikant P, Lucignani G, Piffanelli A (1995) Brain single-
Ballestas ME (1995) Second messenger sys- photon emission tomography with
tems in the regulation of cytokines and adhe- Tc-99m-HMPAO in neuropsychiatric sys-
sion molecules in the central nervous system. temic lupus erythematosus: relations with
Brain Behav Immun 9:304–314 EEG and MRI findings and clinical manifesta-
4. Dunn AJ (2000) Cytokine activation of the tions. Eur J Nucl Med 22:17–24
HPA axis. Ann N Y Acad Sci 917:608–617 16. Handa R, Sahota P, Kumar M, Jagannathan
5. McEwen BS (2000) Allostasis and allostatic NR, Bal CS, Gulati M, Tripathi BM, Wali JP
load: implications for neuropsychopharma- (2003) In vivo proton magnetic resonance
cology. Neuropsychopharmacology 22 spectroscopy (MRS) and single photon emis-
(2):108–124 sion computerized tomography (SPECT) in
6. Johnson EO, Kamilaris TC, Chrousos GP, systemic lupus erythematosus (SLE). Magn
Gold PW (1992) Mechanisms of stress: a Reson Imaging 21(9):1033–1037
dynamic overview of hormonal and behavioral 17. Huang WS, Chiu PY, Tsai CH, Kao A, Lee
homeostasis. Neurosci Biobehav Rev CC (2002) Objective evidence of abnormal
16:115–130 regional cerebral blood flow in patients with
7. Dantzer R, O’Connor JC, Freund GG, John- systemic lupus erythematosus on Tc-99m
son RW, Kelley KW (2008) From inflamma- ECD brain SPECT. Rheumatol Int 22
tion to sickness and depression: when the (5):178–181
immune system subjugates the brain. Nat 18. Lopez-Longo FJ, Carol N, Almoguera MI,
Rev Neurosci 9(1):46–56 Olazaran J, onso-Farto JC, Ortega A,
8. McEwen BS (2003) Mood disorders and allo- Monteagudo I, Gonzalez CM, Carreno L
static load. Biol Psychiatry 54(3):200–207 (2003) Cerebral hypoperfusion detected by
SPECT in patients with systemic lupus erythe-
9. McEwen BS, Biron CA, Brunson KW, matosus is related to clinical activity and
Bulloch K, Chambers WH, Dhabhar FS, cumulative tissue damage. Lupus 12
Goldfarb RH, Kitson RP, Miller AH, Spencer (11):813–819
RL, Weiss JM (1997) The role of adrenocor-
ticoids as modulators of immune function in 19. Komatsu N, Kodama K, Yamanouchi N,
health and disease: neural, endocrine and Okada S, Noda S, Nawata Y, Takabayashi K,
immune interactions. Brain Res Rev 23 Iwamoto I, Saito Y, Uchida Y, Ito H,
(1–2):79–133 Yoshikawa K, Sato T (1999) Decreased
regional cerebral metabolic rate for glucose
10. Cotman CW, Brinton RE, Galaburda A, in systemic lupus erythematosus patients
McEwen B, Schneider DM (1987) The with psychiatric symptoms. Eur Neurol 42
neuro-immune-endocrine connection. Raven (1):41–48
Press, New York
20. Brooks WM, Sabet A, Sibbitt WL, Barker PB,
11. Kapadia M, Sakic B (2011) Autoimmune and van Zijl PC, Duyn JH, Moonen CT (1997)
inflammatory mechanisms of CNS damage. Neurochemistry of brain lesions determined
Prog Neurobiol 95(3):301–333 by spectroscopic imaging in systemic lupus
12. Hanly JG (2005) Neuropsychiatric lupus. erythematosus. J Rheumatol 24
Rheum Dis Clin North Am 31(2):273–298 (12):2323–2329
13. Tincani A, Brey R, Balestrieri G, Vitali C, 21. Volkow ND, Warner N, McIntyre R,
Doria A, Galeazzi M, Meroni PL, Valentine A, Kulkarni M, Mullani N, Gould
Migliorini P, Neri R, Tavoni A, Bombardieri L (1988) Cerebral involvement in systemic
S (1996) International survey on the manage- lupus erythematosus. Am J Physiol Imaging
ment of patients with SLE .2. The results of a 3(2):91–98
questionnaire regarding neuropsychiatric
278 Boris Šakić
22. Gonzalez-Scarano F, Lisak RP, Bilaniuk LT, 32. Hirohata S, Hirose S, Miyamoto T (1985)
Zimmerman RA, Atkins PC, Zweiman B Cerebrospinal fluid IgM, IgA, and IgG
(1979) Cranial computed tomography in the indexes in systemic lupus erythematosus.
diagnosis of systemic lupus erythematosus. Their use as estimates of central nervous sys-
Ann Neurol 5(2):158–165 tem disease activity. Arch Intern Med 145
23. Kaell AT, Shetty M, Lee BC, Lockshin MD (10):1843–1846
(1986) The diversity of neurologic events in 33. Winfield JB, Shaw M, Silverman LM, Eisen-
systemic lupus erythematosus. Prospective berg RA, Wilson HA III, Koffler D (1983)
clinical and computed tomographic classifica- Intrathecal IgG synthesis and blood-brain
tion of 82 events in 71 patients. Arch Neurol barrier impairment in patients with systemic
43(3):273–276 lupus erythematosus and central nervous sys-
24. Miguel EC, Pereira RM, Pereira CA, Baer L, tem dysfunction. Am J Med 74(5):837–844
Gomes RE, de Sa LC, Hirsch R, de Barros 34. Yoshio T, Hirata D, Onda K, Nara H, Minota
NG, de Navarro JM, Gentil V (1994) Psychi- S (2005) Antiribosomal P protein antibodies
atric manifestations of systemic lupus erythe- in cerebrospinal fluid are associated with neu-
matosus: clinical features, symptoms, and ropsychiatric systemic lupus erythematosus. J
signs of central nervous system activity in Rheumatol 32(1):34–39
43 patients. Medicine 73(4):224–232 35. Greenwood DL, Gitlits VM, Alderuccio F,
25. Omdal R, Selseth B, Klow NE, Husby G, Sentry JW, Toh BH (2002) Autoantibodies
Mellgren SI (1989) Clinical neurological, in neuropsychiatric lupus. Autoimmunity 35
electrophysiological, and cerebral CT scan (2):79–86
findings in systemic lupus erythematosus. 36. Henn FA, McKinney WT (1987) Animal
Scand J Rheumatol 18(5):283–289 models in psychiatry. In: Meltzer HY
26. Ainiala H, Dastidar P, Loukkola J, (ed) Psychopharmacology: the third genera-
Lehtimaki T, Korpela M, Peltola J, Hietaharju tion of progress. Raven Press, New York, pp
A (2005) Cerebral MRI abnormalities and 687–695
their association with neuropsychiatric mani- 37. Dixon FJ, Andrews BS, Eisenberg RA, McCo-
festations in SLE: a population-based study. nahey PJ, Theofilopoulos AN, Wilson CB
Scand J Rheumatol 34(5):376–382 (1978) Etiology and pathogenesis of a spon-
27. Waterloo K, Omdal R, Jacobsen EA, Klow taneous lupus-like syndrome in mice. Arthritis
NE, Husby G, Torbergsen T, Mellgren SI Rheum 21:S64–S67
(1999) Cerebral computed tomography and 38. Szechtman H, Sakic B, Denburg JA (1997)
electroencephalography compared with neu- Behaviour of MRL mice: an animal model of
ropsychological findings in systemic lupus disturbed behaviour in systemic autoimmune
erythematosus. J Neurol 246(8):706–711 disease. Lupus 6(3):223–229
28. Sibbitt WL, Sibbitt RR (1993) Magnetic res- 39. Sakic B, Szechtman H, Denburg JA (1997)
onance spectroscopy and positron emission Neurobehavioral alteration in autoimmune
tomography scanning in neuropsychiatric sys- mice. Neurosci Biobehav Rev 21(3):327–340
temic lupus erythematosus. Rheum Dis Clin 40. Alexander JJ, Quigg RJ (2007) Systemic
North Am 19(4):851–868 lupus erythematosus and the brain: what
29. Sibbitt WL, Haseler LJ, Griffey RH, Hart BL, mice are telling us. Neurochem Int 50
Sibbitt RR, Matwiyoff NA (1994) Analysis of (1):5–11
cerebral structural changes in systemic lupus 41. Sherman GF, Galaburda AM, Behan PO,
erythematosus by proton MR spectroscopy. Rosen GD (1987) Neuroanatomical anoma-
AJNR Am J Neuroradiol 15:923–928 lies in autoimmune mice. Acta Neuropathol
30. Jennekens FG, Kater L (2002) The central (Berl ) 74:239–242
nervous system in systemic lupus erythemato- 42. Theofilopoulos AN (1992) Murine models of
sus. Part 2. Pathogenetic mechanisms of clini- lupus. In: Lahita RG (ed) Systemic lupus
cal syndromes: a literature investigation. erythematosus, 2nd edn. Churchill Living-
Rheumatology (Oxford) 41(6):619–630 stone, New York, pp 121–194
31. McLean BN, Miller D, Thompson EJ (1995) 43. Andrews BS, Eisenberg RA, Theofilopoulos
Oligoclonal banding of IgG in CSF, blood- AN, Izui S, Wilson CB, McConahey PJ, Mur-
brain barrier function, and MRI findings in phy ED, Roths JB, Dixon FJ (1978) Sponta-
patients with sarcoidosis, systemic lupus neous murine lupus-like syndromes. Clinical
erythematosus, and Behcet’s disease involving and immunopathological manifestations in
the nervous system. J Neurol Neurosurg Psy- several strains. J Exp Med 148:1198–1215
chiatry 58(5):548–554
The MRL Model 279
44. Gulinello M, Putterman C (2011) The 57. Sakic B, Szechtman H, Denburg S,

MRL/lpr mouse strain as a model for neurop- Carbotte R, Denburg JA (1993) Emotionality
sychiatric systemic lupus erythematosus. J in lupus-prone mice: relationships to autoim-
Biomed Biotechnol 2011:207504 munity. Annual Research Day, Dept. of Bio-
45. Jeltsch-David H, Muller S (2014) Neuropsy- medical Sciences, McMaster University
chiatric systemic lupus erythematosus and 58. Sakic B, Szechtman H, Keffer M,
cognitive dysfunction: the MRL-lpr mouse Talangbayan H, Stead R, Denburg JA
strain as a model. Autoimmun Rev 13 (1992) A behavioral profile of autoimmune
(9):963–973. https://doi.org/10.1016/j. lupus-prone MRL mice. Brain Behav Immun
autrev.2014.08.015 6:265–285
46. Watanabe-Fukunaga R, Brannan CI, Cope- 59. Ballok DA, Szechtman H, Sakic B (2003)
land NG, Jenkins NA, Nagata S (1992) Lym- Taste responsiveness and diet preference in
phoproliferation disorder in mice explained by autoimmune MRL mice. Behav Brain Res
defects in Fas antigen that mediates apoptosis. 140(1–2):119–130
Nature 356:314–317 60. Sakic B, Gurunlian L, Denburg SD (1998)
47. Nagata S (1994) Mutations in the Fas antigen Reduced aggressiveness and low testosterone
gene in lpr mice. Semin Immunol 6(1):3–8 levels in autoimmune MRL-lpr males. Physiol
48. Singer GG, Carrera AC, Marshakrothstein A, Behav 63(2):305–309
Martineza C, Abbas AK (1994) Apoptosis, fas 61. Sakic B, Szechtman H, Denburg SD, Car-
and systemic autoimmunity: the MRL-Ipr/ botte RM, Denburg JA (1993) Spatial
Ipr model. Curr Opin Immunol 6:913–920 learning during the course of autoimmune
49. Park C, Sakamaki K, Tachibana O, disease in MRL mice. Behav Brain Res
Yamashima T, Yamashita J, Yonehara S 54:57–66
(1998) Expression of Fas antigen in the nor- 62. Hess DC, Taormina M, Thompson J, Sethi
mal mouse brain. Biochem Biophys Res Com- KD, Diamond B, Rao R, Feldman DS (1993)
mun 252(3):623–628 Cognitive and neurologic deficits in the
50. Sakic B, Kolb B, Whishaw IQ, Gorny G, MRL/lpr mouse: a clinicopathologic study. J
Szechtman H, Denburg JA (2000) Immuno- Rheumatol 20:610–617
suppression prevents neuronal atrophy in 63. Brey RL, Amato AA, Kagan-Hallet K, Rhine
lupus-prone mice: evidence for brain damage CB, Stallworth CL (1997) Anti-intercellular
induced by autoimmune disease? J Neuroim- adhesion molecule-1 (ICAM-1) antibody
munol 111(1–2):93–101 treatment prevents central and peripheral ner-
51. Sakic B, Denburg JA, Denburg SD, Szecht- vous system disease in autoimmune-prone
man H (1996) Blunted sensitivity to sucrose mice. Lupus 6(8):645–651
in autoimmune MRL-lpr mice: a curve-shift 64. Sakic B, Szechtman H, Stead R, Denburg JA
study. Brain Res Bull 41(5):305–311 (1996) Joint pathology and behavioral perfor-
52. Sakic B, Szechtman H, Denburg SD, Den- mance in autoimmune MRL-lpr mice. Physiol
burg JA (1995) Immunosuppressive treat- Behav 60(3):901–905
ment prevents behavioral deficit in 65. Abbott NJ, Mendonca LL, Dolman DE
autoimmune MRL-lpr mice. Physiol Behav (2003) The blood-brain barrier in systemic
58(4):797–802 lupus erythematosus. Lupus 12(12):908–915
53. Ballok DA, Woulfe J, Sur M, Cyr M, Sakic B 66. Diamond B (2010) Antibodies and the brain:
(2004) Hippocampal damage in mouse and lessons from lupus. J Immunol 185
human forms of systemic autoimmune dis- (5):2637–2640
ease. Hippocampus 14(5):649–661 67. Vogelweid CM, Johnson GC, Besch-Williford
54. Williams S, Sakic B, Hoffman SA (2010) Cir- CL, Basler J, Walker SE (1991) Inflammatory
culating brain-reactive autoantibodies and central nervous system disease in lupus-prone
behavioral deficits in the MRL model of MRL/lpr mice: comparative histologic and
CNS lupus. J Neuroimmunol 218 immunohistochemical findings. J Neuroim-
(1–2):73–82 munol 35:89–99
55. Ballok DA (2007) Neuroimmunopathology 68. Sidor MM, Sakic B, Malinowski PM, Ballok
in a murine model of neuropsychiatric lupus. DA, Oleschuk CJ, Macri J (2005) Elevated
Brain Res Rev 54(1):67–79 immunoglobulin levels in the cerebrospinal
56. Sakic B, Szechtman H, Talangbayan H, Den- fluid from lupus-prone mice. J Neuroimmu-
burg SD, Carbotte RM, Denburg JA (1994) nol 165(1–2):104–113
Disturbed emotionality in autoimmune 69. Alexander EL, Murphy ED, Roths JB, Alex-
MRL-lpr mice. Physiol Behav 56(3):609–617 ander GE (1983) Congenic autoimmune
280 Boris Šakić
murine models of central nervous system dis- 82. Alexander EL, Alexander GE (1983) Aseptic
ease in connective tissue disorders. Ann meningoencephalitis in primary Sjogren’s
Neurol 14:242–248 syndrome. Neurology 33:593–598
70. Farrell M, Sakic B, Szechtman H, Denburg JA 83. Auer RN, Wieloch T, Olsson Y, Siesjo BK
(1997) Effect of cyclophosphamide on leuco- (1984) The distribution of hypoglycemic
cytic infiltration in the brain of MRL/lpr brain damage. Acta Neuropathol 64
mice. Lupus 6(3):268–274 (3):177–191
71. Ma X, Foster J, Sakic B (2006) Distribution 84. Fujioka M, Okuchi K, Hiramatsu KI,
and prevalence of leukocyte phenotypes in Sakaki T, Sakaguchi S, Ishii Y (1997) Specific
brains of lupus-prone mice. J Neuroimmunol changes in human brain after hypoglycemic
179(1–2):26–36 injury. Stroke 28(3):584–587
72. McHale JF, Harari OA, Marshall D, Haskard 85. Denenberg VH, Sherman GF, Rosen GD,
DO (1999) TNF-alpha and IL-1 sequentially Morrison L, Behan PO, Galaburda AM
induce endothelial ICAM-1 and VCAM-1 (1992) A behavior profile of the MRL/Mp
expression in MRL/lpr lupus-prone mice. J lpr/lpr mouse and its association with hydro-
Immunol 163(7):3993–4000 cephalus. Brain Behav Immun 6:40–49
73. Zameer A, Hoffman SA (2003) Increased 86. Alexander JJ, Zwingmann C, Quigg R (2005)
ICAM-1 and VCAM-1 expression in the MRL/lpr mice have alterations in brain
brains of autoimmune mice. J Neuroimmunol metabolism as shown with [(1)H-(13)C]
142(1–2):67–74 NMR spectroscopy. Neurochem Int 47
74. Tomita M, Holman BJ, Williams LS, Pang (1–2):143–151
KC, Santoro TJ (2001) Cerebellar dysfunc- 87. Alexander JJ, Bao L, Jacob A, Kraus DM,
tion is associated with overexpression of Holers VM, Quigg RJ (2003) Administration
proinflammatory cytokine genes in lupus. J of the soluble complement inhibitor, Crry-Ig,
Neurosci Res 64(1):26–33 reduces inflammation and aquaporin 4 expres-
75. Tomita M, Holman BJ, Santoro TJ (2001) sion in lupus cerebritis. Biochim Biophys Acta
Aberrant cytokine gene expression in the hip- 1639(3):169–176
pocampus in murine systemic lupus erythe- 88. Hampton DW, Seitz A, Chen P, Heber-
matosus. Neurosci Lett 302(2–3):129–132 Katz E, Fawcett JW (2004) Altered CNS
76. Alexander JJ, Jacob A, Bao L, Macdonald RL, response to injury in the MRL/MpJ mouse.
Quigg RJ (2005) Complement-dependent Neuroscience 127(4):821–832
apoptosis and inflammatory gene changes in 89. Lawson LJ, Perry VH, Dri P, Gordon S
murine lupus cerebritis. J Immunol 175 (1990) Heterogeneity in the distribution and
(12):8312–8319 morphology of microglia in the normal adult
77. McIntyre KR, Ayer-LeLievre C, Persson H mouse brain. Neuroscience 39(1):151–170
(1990) Class II major histocompatibility com- 90. Ballok DA, Earls AM, Krasnik C, Hoffman
plex (MHC) gene expression in the mouse SA, Sakic B (2004) Autoimmune-induced
brain is elevated in the autoimmune MRL/ damage of the midbrain dopaminergic system
Mp-lpr/lpr strain. J Neuroimmunol in lupus-prone mice. J Neuroimmunol 152
28:39–52 (1–2):83–97
78. Ballok DA, Ma X, Denburg JA, Arsenault L, 91. Hess DC (1997) Cerebral lupus vasculopathy.
Sakic B (2006) Ibuprofen fails to prevent Mechanisms and clinical relevance. Ann N Y
brain pathology in a model of neuropsychia- Acad Sci 823:154–168
tric lupus. J Rheumatol 33(11):2199–2213 92. Baraczka K, Nekam K, Pozsonyi T, Jakab L,
79. Ballok DA, Millward JM, Sakic B (2003) Szongoth M, Sesztak M (2001) Concentra-
Neurodegeneration in autoimmune MRL-lpr tion of soluble adhesion molecules (sVCAM-
mice as revealed by Fluoro Jade B staining. 1, sICAM-1 and sL-selectin) in the cerebro-
Brain Res 964(2):200–210 spinal fluid and serum of patients with multi-
80. Del Bigio MR (1993) Neuropathological ple sclerosis and systemic lupus erythematosus
changes caused by hydrocephalus. Acta Neu- with central nervous involvement. Neuroim-
ropathol 85(6):573–585 munomodulation 9(1):49–54
81. Gerber J, Raivich G, Wellmer A, Noeske C, 93. Chinn RJS, Wilkinson ID, Hallcraggs MA,
Kunst T, Werner A, Bruck W, Nau R (2001) A Paley MNJ, Shorthall E, Carter S, Kendall
mouse model of Streptococcus pneumoniae BE, Isenberg DA, Newman SP, Harrison
meningitis mimicking several features of MJG (1997) Magnetic resonance imaging of
human disease. Acta Neuropathol 101 the brain and cerebral proton spectroscopy in
(5):499–508
The MRL Model 281
patients with systemic lupus erythematosus. 104. Anderson KK, Ballok DA, Prasad N,
Arthritis Rheum 40(1):36–46 Szechtman H, Sakic B (2006) Impaired
94. Rocca MA, Agosta F, Mezzapesa DM, response to amphetamine and neuronal
Ciboddo G, Falini A, Comi G, Filippi M degeneration in the nucleus accumbens of
(2006) An fMRI study of the motor system autoimmune MRL-lpr mice. Behav Brain
in patients with neuropsychiatric systemic Res 166:32–38
lupus erythematosus. Neuroimage 30 105. Brey RL, Cote S, Barohn R, Jackson C,
(2):478–484 Crawley R, Teale JM (1995) Model for the
95. Sled JG, Spring S, van Eede M, Lerch JP, neuromuscular complications of systemic
Ullal S, Sakic B (2009) Time course and lupus erythematosus. Lupus 4(3):209–212
nature of brain atrophy in the MRL mouse 106. Loheswaran G, Stanojcic M, Xu L, Sakic B
model of central nervous system lupus. (2010) Autoimmunity as a principal patho-
Arthritis Rheum 60(6):1764–1774 genic factor in the refined model of neurop-
96. Sakic B, Szechtman H, Denburg JA, sychiatric lupus. Clin Exp Neuroimmunol
Gorny G, Kolb B, Whishaw IQ (1998) Pro- 1:141–152
gressive atrophy of pyramidal neuron den- 107. How A, Dent PB, Liao SK, Denburg JA
drites in autoimmune MRL-lpr mice. J (1985) Antineuronal antibodies in neuropsy-
Neuroimmunol 87(1–2):162–170 chiatric systemic lupus erythematosus. Arthri-
97. Maric D, Millward JM, Ballok DA, tis Rheum 28:789–795
Szechtman H, Barker JL, Denburg JA, Sakic 108. Williams GW, Bluestein HG, Steinberg AD
B (2001) Neurotoxic properties of cerebro- (1981) Brain-reactive lymphocytotoxic anti-
spinal fluid from behaviorally impaired auto- body in the cerebrospinal fluid of patients
immune mice. Brain Res 920(1–2):183–193 with systemic lupus erythematosus: correla-
98. Kovac AD, Grammig J, Mahlo J, Steiner B, tion with central nervous system involvement.
Roth K, Nitsch R, Bechmann I (2002) Com- Clin Immunol Immunopathol 18
parison of neuronal density and subfield sizes (1):126–132
in the hippocampus of CD95L-deficient 109. Bluestein HG, Zvaifler NJ (1976) Brain-
(gld), CD95-deficient (lpr) and nondeficient reactive lymphocytotoxic antibodies in the
mice. Eur J Neurosci 16(1):159–163 serum of patients with systemic lupus erythe-
99. Sakic B, Maric I, Koeberle PD, Millward JM, matosus. J Clin Invest 57(2):509–516
Szechtman H, Maric D, Denburg JA (2000) 110. Bresnihan B, Hohmeister R, Cutting J, Tra-
Increased TUNEL-staining in brains of auto- vers RL, Waldburger M, Black C, Jones T,
immune Fas-deficient mice. J Neuroimmunol Hughes GR (1979) The neuropsychiatric dis-
104(2):147–154 order in systemic lupus erythematosus: evi-
100. Yamashita T, Ninomiya M, Hernandez AP, dence for both vascular and immune
Garcia-Verdugo JM, Sunabori T, mechanisms. Ann Rheum Dis 38(4):301–306
Sakaguchi M, Adachi K, Kojima T, Hirota Y, 111. Golombek SJ, Graus F, Elkon KB (1986)
Kawase T, Araki N, Abe K, Okano H, Sawa- Autoantibodies in the cerebrospinal fluid of
moto K (2006) Subventricular zone-derived patients with systemic lupus erythematosus.
neuroblasts migrate and differentiate into Arthritis Rheum 29(9):1090–1097
mature neurons in the post-stroke adult stria- 112. Quismorio FP, Friou GJ (1972) Antibodies
tum. J Neurosci 26(24):6627–6636 reactive with neurons in SLE patients with
101. Sakic B, Kirkham DL, Ballok DA, neuropsychiatric manifestations. Int Arch
Mwanjewe J, Fearon IM, Macri J, Yu G, Allergy Appl Immunol 43:740–748
Sidor MM, Denburg JA, Szechtman H, 113. DeGiorgio LA, Konstantinov KN, Lee SC,
Lau J, Ball AK, Doering LC (2005) Prolifer- Hardin JA, Volpe BT, Diamond B (2001) A
ating brain cells are a target of neurotoxic CSF subset of lupus anti-DNA antibodies cross-
in systemic autoimmune disease. J Neuroim- reacts with the NR2 glutamate receptor in
munol 169(1–2):68–85 systemic lupus erythematosus. Nat Med 7
102. Lechner O, Dietrich H, Oliveira dos SA, Wie- (11):1189–1193
gers GJ, Schwarz S, Harbutz M, Herold M, 114. Tanaka S, Matsunaga H, Kimura M,
Wick G (2000) Altered circadian rhythms of Tatsumi K, Hidaka Y, Takano T, Uema T,
the stress hormone and melatonin response in Takeda M, Amino N (2003) Autoantibodies
lupus-prone MRL/MP-fas(Ipr) mice. J Auto- against four kinds of neurotransmitter recep-
immun 14(4):325–333 tors in psychiatric disorders. J Neuroimmunol
103. Mirescu C, Gould E (2006) Stress and adult 141(1–2):155–164
neurogenesis. Hippocampus 16(3):233–238
282 Boris Šakić
115. Zandman-Goddard G, Chapman J, Shoenfeld lupus erythematosus using an immunopro-

Y (2007) Autoantibodies involved in neurop- teomic approach. Arthritis Rheum 56
sychiatric SLE and antiphospholipid syn- (10):3420–3432
drome. Semin Arthritis Rheum 36 125. Williams RC Jr, Sugiura K, Tan EM (2004)
(5):297–315 Antibodies to microtubule-associated protein
116. Isshi K, Hirohata S (1996) Association of 2 in patients with neuropsychiatric systemic
anti-ribosomal P protein antibodies with neu- lupus erythematosus. Arthritis Rheum 50
ropsychiatric systemic lupus erythematosus. (4):1239–1247
Arthritis Rheum 39(9):1483–1490 126. Yoshio T, Onda K, Nara H, Minota S (2006)
117. Denburg SD, Denburg JA (2003) Cognitive Association of IgG anti-NR2 glutamate
dysfunction and antiphospholipid antibodies receptor antibodies in cerebrospinal fluid
in systemic lupus erythematosus. Lupus 12 with neuropsychiatric systemic lupus erythe-
(12):883–890 matosus. Arthritis Rheum 54(2):675–678
118. Hanly JG, Urowitz MB, Siannis F, Farewell V, 127. Arinuma Y, Yanagida T, Hirohata S (2008)
Gordon C, Bae SC, Isenberg D, Dooley MA, Association of cerebrospinal fluid anti-NR2
Clarke A, Bernatsky S, Gladman D, Fortin glutamate receptor antibodies with diffuse
PR, Manzi S, Steinsson K, Bruce IN, neuropsychiatric systemic lupus erythemato-
Ginzler E, Aranow C, Wallace DJ, Ramsey- sus. Arthritis Rheum 58(4):1130–1135
Goldman R, van Vollenhoven R, Sturfelt G, 128. Stanojcic M, Loheswaran G, Xu L, Hoffman
Nived O, Sanchez-Guerrero J, Alarcon GS, SA, Sakic B (2010) Intrathecal antibodies and
Petri M, Khamashta M, Zoma A, Font J, brain damage in autoimmune MRL mice.
Kalunian K, Douglas J, Qi Q, Thompson K, Brain Behav Immun 24:289–297
Merrill JT (2008) Autoantibodies and neu- 129. Hoffman SA, Madsen CS (1990) Brain spe-
ropsychiatric events at the time of systemic cific autoantibodies in murine models of sys-
lupus erythematosus diagnosis: results from temic lupus erythematosus. J Neuroimmunol
an international inception cohort study. 30:229–237
Arthritis Rheum 58(3):843–853
130. Hoffman SA, Arbogast DN, Ford PM, Shu-
119. Kowal C, DeGiorgio LA, Nakaoka T, card DW, Harbeck RJ (1987) Brain-reactive
Hetherington H, Huerta PT, Diamond B, autoantibody levels in the sera of ageing auto-
Volpe BT (2004) Cognition and immunity; immune mice. Clin Exp Immunol 70:74–83
antibody impairs memory. Immunity 21
(2):179–188 131. Crimando J, Hoffman SA (1992) Detection
of brain-reactive autoantibodies in the sera of
120. Huerta PT, Kowal C, DeGiorgio LA, Volpe autoimmune mice using ELISA. J Immunol
BT, Diamond B (2006) Immunity and behav- Methods 149:87–95
ior: antibodies alter emotion. Proc Natl Acad
Sci U S A 103(3):678–683 132. Sakic B, Szechtman H, Denburg SD, Car-
botte RM, Denburg JA (1993) Brain-reactive
121. Kowal C, DeGiorgio LA, Lee JY, Edgar MA, antibodies and behavior of autoimmune
Huerta PT, Volpe BT, Diamond B (2006) MRL-lpr mice. Physiol Behav 54:1025–1029
Human lupus autoantibodies against NMDA
receptors mediate cognitive impairment. Proc 133. Gao HX, Sanders E, Tieng AT, Putterman C
Natl Acad Sci U S A 103(52):19854–19859 (2010) Sex and autoantibody titers determine
the development of neuropsychiatric manifes-
122. Ndhlovu M, Preuss BE, Dengjel J, tations in lupus-prone mice. J Neuroimmunol
Stevanovic S, Weiner SM, Klein R (2011) 229(1–2):112–122
Identification of alpha-tubulin as an autoanti-
gen recognized by sera from patients with 134. Gao HX, Campbell SR, Cui MH, Zong P,
neuropsychiatric systemic lupus erythemato- Hee-Hwang J, Gulinello M, Putterman C
sus. Brain Behav Immun 25(2):279–285 (2009) Depression is an early disease manifes-
tation in lupus-prone MRL/lpr mice. J Neu-
123. Colasanti T, Delunardo F, Margutti P, roimmunol 207(1–2):45–56
Vacirca D, Piro E, Siracusano A, Ortona E
(2009) Autoantibodies involved in neuropsy- 135. Hoffman SA, Sakic B (2008) Autoimmunity
chiatric manifestations associated with sys- and brain dysfunction. In: Siegel A, Zalcman
temic lupus erythematosus. J S (eds) The neuroimmunological basis of
Neuroimmunol 212(1–2):3–9 behavior and mental disorders. Springer-
Verlag, Boston, MA
124. Lefranc D, Launay D, Dubucquoi S, de SJ,
Dussart P, Vermersch M, Hachulla E, Hatron 136. Meroni PL, Tincani A, Sepp N, Raschi E,
PY, Vermersch P, Mouthon L, Prin L (2007) Testoni C, Corsini E, Cavazzana I,
Characterization of discriminant human brain Pellegrini S, Salmaggi A (2003) Endothelium
antigenic targets in neuropsychiatric systemic
The MRL Model 283
and the brain in CNS lupus. Lupus 12 and nitric oxide metabolites in neuropsychia-
(12):919–928 tric lupus erythematosus. Ann Rheum Dis 60
137. Tang B, Matsuda T, Akira S, Nagata N, (4):372–379
Ikehara S, Hirano T, Kishimoto T (1991) 149. Hayley S, Merali Z, Anisman H (2003) Stress
Age-associated increase in interleukin 6 in and cytokine-elicited neuroendocrine and
MRL/lpr mice. Int Immunol 3:273–278 neurotransmitter sensitization: implications
138. Singh AK, Lebedeva TV (1994) Interleukin-1 for depressive illness. Stress 6(1):19–32
contributes to high level IgG production in 150. Hu Y, Dietrich H, Herold M, Heinrich PC,
the murine MRL/lpr lupus model. Immunol Wick G (1993) Disturbed immuno-
Invest 23:281–292 endocrine communication via the hypotha-
139. Boswell JM, Yui MA, Endres S, Burt DW, lamo- pituitary-adrenal axis in autoimmune
Kelley VE (1988) Novel and enhanced IL-1 disease. Int Arch Allergy Immunol
gene expression in autoimmune mice with 102:232–241
lupus. J Immunol 141:118–124 151. Rivier C, Rivest S (1993) Mechanisms med-
140. Sakic B, Szechtman H, Braciak TA, Richards iating the effects of cytokines on neuroendo-
CD, Gauldie J, Denburg JA (1997) Reduced crine functions in the rat. In: Chadwick D,
preference for sucrose in autoimmune mice: a Marsh J, Ackrill K (eds) Corticotropin-
possible role of interleukin-6. Brain Res Bull releasing factor, vol Clayton Foundation
44(2):155–165 Laboratories for Peptide Biology, Salk Insti-
141. Kwant A, Sakic B (2004) Behavioral effects of tute, San Diego, CA 92128, Ciba Foundation
infection with interferon-gamma adenovec- Symposium, vol 172. John Wiley & Sons,
tor. Behav Brain Res 151(1–2):73–82 Chichester, pp 204–225
142. Sakic B, Szechtman H, Gauldie J, Denburg JA 152. Spangelo BL, Gorospe WC (1995) Role of
(2001) Behavioral effects of infection with the cytokines in the neuroendocrine-immune
IL-6 adenovector. Brain Behav Immun 15 system axis. Front Neuroendocrinol 16
(1):25–42 (1):1–22
143. Marshall D, Dangerfield JP, Bhatia VK, Larbi 153. Miller AH, Maletic V, Raison CL (2009)
KY, Nourshargh S, Haskard DO (2003) Inflammation and its discontents: the role of
MRL/lpr lupus-prone mice show exagger- cytokines in the pathophysiology of major
ated ICAM-1-dependent leucocyte adhesion depression. Biol Psychiatry 65(9):732–741
and transendothelial migration in response to 154. Del Rey A, Besedovsky HO (2000) The
TNF-alpha. Rheumatology (Oxford) 42 cytokine-HPA axis circuit contributes to pre-
(8):929–934 vent or moderate autoimmune processes. Z
144. Banks WA, Kastin AJ, Gutierrez EG (1994) Rheumatol 59(Suppl 2):II/31–II/35
Penetration of interleukin-6 across the 155. Lorton D, Lubahn C, Bellinger DL (2003)
murine blood-brain barrier. Neurosci Lett Potential use of drugs that target neural-
179(1–2):53–56 immune pathways in the treatment of rheu-
145. Banks WA, Kastin AJ, Gutierrez EG (1993) matoid arthritis and other autoimmune dis-
Interleukin-1 alpha in blood has direct access eases. Curr Drug Targets Inflamm Allergy 2
to cortical brain cells. Neurosci Lett 163 (1):1–30
(1):41–44 156. Spangelo BL, Judd AM, Call GB, Zumwalt J,
146. Gutierrez EG, Banks WA, Kastin AJ (1993) Gorospe WC (1995) Role of the cytokines in
Murine tumor necrosis factor alpha is trans- the hypothalamic-pituitary-adrenal and
ported from blood to brain in the mouse. J gonadal axes. Neuroimmunomodulation 2
Neuroimmunol 47(2):169–176 (5):299–312
147. Tsai CY, Wu TH, Tsai ST, Chen KH, 157. Spangelo BL, Judd AM, Isakson PC,
Thajeb P, Lin WM, Yu HS, Yu CL (1994) MacLeod RM (1989) Interleukin-6 stimu-
Cerebrospinal fluid interleukin-6, prostaglan- lates anterior pituitary hormone release
din E2 and autoantibodies in patients with in vitro. Endocrinology 125(1):575–577
neuropsychiatric systemic lupus erythemato- 158. Shanks N, Moore PM, Perks P, Lightman SL
sus and central nervous system infections. (1999) Alterations in hypothalamic-pituitary-
Scand J Rheumatol 23(2):57–63 adrenal function correlated with the onset of
148. Svenungsson E, Andersson M, Brundin L, murine SLE in MRL +/+ and lpr/lpr mice.
van Vollenhoven R, Khademi M, Brain Behav Immun 13(4):348–360
Tarkowski A, Greitz D, Dahlstrom M, 159. Sakic B, Laflamme N, Crnic LS,
Lundberg I, Klareskog L, Olsson T (2001) Szechtman H, Denburg JA, Rivest S (1999)
Increased levels of proinflammatory cytokines Reduced corticotropin-releasing factor and
284 Boris Šakić
enhanced vasopressin gene expression in Putterman C (2016) B cell and/or autoanti-

brains of mice with autoimmunity-induced body deficiency do not prevent neuropsychia-
behavioral dysfunction. J Neuroimmunol 96 tric disease in murine systemic lupus
(1):80–91 erythematosus. J Neuroinflammation 13
160. McEwen BS (1999) Stress and the aging hip- (1):73. https://doi.org/10.1186/s12974-
pocampus. Front Neuroendocrinol 20 016-0537-3
(1):49–70 172. Loheswaran G, Kapadia M, Gladman M,
161. Woolley CS, Gould E, McEwen BS (1990) Pulapaka S, Xu L, Stanojcic M, Sakic B
Exposure to excess glucocorticoids alters den- (2013) Altered neuroendocrine status at the
dritic morphology of adult hippocampal pyra- onset of CNS lupus-like disease. Brain Behav
midal neurons. Brain Res 531(1–2):225–231 Immun 32:86–93
162. Ballok DA, Sakic B (2008) Purine receptor 173. Stock AD, Wen J, Doerner J, Herlitz LC,
antagonist modulates serology and affective Gulinello M, Putterman C (2015) Neuropsy-
behaviors in lupus-prone mice: evidence of chiatric systemic lupus erythematosus persists
autoimmune-induced pain? Brain Behav despite attenuation of systemic disease in
Immun 22(8):1208–1216 MRL/lpr mice. J Neuroinflammation
163. Sapolsky RM (2000) Glucocorticoids and 12:205. https://doi.org/10.1186/s12974-
hippocampal atrophy in neuropsychiatric dis- 015-0423-4
orders. Arch Gen Psychiatry 57(10):925–935 174. Denenberg VH, Mobraaten LE, Sherman
164. Peress NS, Roxburgh VA, Gelfand MC GF, Morrison L, Schrott LM, Waters NS,
(1981) Binding sites for immune components Rosen GD, Behan PO, Galaburda AM
in human choroid plexus. Arthritis Rheum 24 (1991) Effects of the autoimmune uterine/
(3):520–526 maternal environment upon cortical ectopias,
behavior and autoimmunity. Brain Res
165. Schwartz MM, Roberts JL (1983) Membra- 563:114–122
nous and vascular choroidopathy: two pat-
terns of immune deposits in systemic lupus 175. Lee JY, Huerta PT, Zhang J, Kowal C,
erythematosus. Clin Immunol Immuno- Bertini E, Volpe BT, Diamond B (2009) Neu-
pathol 29(3):369–380 rotoxic autoantibodies mediate congenital
cortical impairment of offspring in maternal
166. Reilly CM, Oates JC, Cook JA, Morrow JD, lupus. Nat Med 15(1):91–96
Halushka PV, Gilkeson GS (2000) Inhibition
of mesangial cell nitric oxide in MRL/lpr 176. Li Y, Eskelund AR, Zhou H, Budac DP,
mice by prostaglandin J2 and proliferator acti- Sanchez C, Gulinello M (2015) Behavioral
vation receptor-gamma agonists. J Immunol deficits are accompanied by immunological
164(3):1498–1504 and neurochemical changes in a mouse
model for neuropsychiatric lupus (NP-SLE).
167. O’Sullivan FX, Vogelweid CM, Beschwilliford Int J Mol Sci 16(7):15150–15171. https://
CL, Walker SE (1995) Differential effects of doi.org/10.3390/ijms160715150
CD4(+) T cell depletion on inflammatory
central nervous system disease, arthritis and 177. Wen J, Doerner J, Weidenheim K, Xia Y,
sialadenitis in MRL/lpr mice. J Autoimmun Stock A, Michaelson JS, Baruch K,
8:163–175 Deczkowska A, Gulinello M, Schwartz M,
Burkly LC, Putterman C (2015) TNF-like
168. Jacob A, Hack B, Bai T, Brorson JR, Quigg weak inducer of apoptosis promotes blood
RJ, Alexander JJ (2010) Inhibition of C5a brain barrier disruption and increases neuro-
receptor alleviates experimental CNS lupus. J nal cell death in MRL/lpr mice. J Autoim-
Neuroimmunol 221(1–2):46–52 mun 60:40–50. https://doi.org/10.1016/j.
169. Sakic B, Lacosta S, Denburg J, Szechtman H jaut.2015.03.005
(2002) Altered neurotransmission in brains of 178. Wen J, Chen CH, Stock A, Doerner J,
autoimmune mice: pharmacological and neu- Gulinello M, Putterman C (2016) Intracer-
rochemical evidence. J Neuroimmunol 129 ebroventricular administration of TNF-like
(1–2):84–96 weak inducer of apoptosis induces
170. Chun S, McEvilly R, Foster JA, Sakic B depression-like behavior and cognitive dys-
(2008) Proclivity to self-injurious behavior function in non-autoimmune mice. Brain
in MRL-lpr mice: implications for Behav Immun 54:27–37. https://doi.org/
autoimmunity-induced damage in the dopa- 10.1016/j.bbi.2015.12.017
minergic system. Mol Psychiatry 13 179. Marcinko K, Parsons T, Lerch JP, Sled JG,
(11):1043–1053 Sakic B (2013) Effects of prolonged treat-
171. Wen J, Doerner J, Chalmers S, Stock A, ment with memantine in the MRL model of
Wang H, Gullinello M, Shlomchik MJ,
The MRL Model 285
CNS lupus. Clin Exp Neuroimmunol 3 disease in murine lupus. Clin Immun
(3):116–128 (Orlando, Fla) 185:100–108. https://doi.
180. Kapadia M, Zhao H, Ma D, Sakic B (2017) org/10.1016/j.clim.2016.08.019
Sustained immunosuppression alters olfac- 183. Kim A, Feng P, Ohkuri T, Sauers D, Cohn ZJ,
tory function in the MRL model of CNS Chai J, Nelson T, Bachmanov AA, Huang L,
lupus. J Neuroimmune Pharmacol 12 Wang H (2012) Defects in the peripheral
(3):555–564. https://doi.org/10.1007/ taste structure and function in the MRL/lpr
s11481-017-9745-6 mouse model of autoimmune disease. PLoS
181. Mahajan SD, Tutino VM, Redae Y, Meng H, One 7(4):e35588
Siddiqui A, Woodruff TM, Jarvis JN, 184. Kapadia M, Stanojcic M, Earls AM,
Hennon T, Schwartz S, Quigg RJ, Alexander Pulapaka S, Lee J, Sakic B (2012) Altered
JJ (2016) C5a induces caspase-dependent olfactory function in the MRL model of
apoptosis in brain vascular endothelial cells CNS lupus. Behav Brain Res 234(2):303–311
in experimental lupus. Immunology 148 185. Stanojcic M, Burstyn-Cohen T, Nashi N,
(4):407–419. https://doi.org/10.1111/ Lemke G, Sakic B (2009) Disturbed distribu-
imm.12619 tion of proliferative brain cells during lupus-
182. Chalmers SA, Wen J, Shum J, Doerner J, like disease. Brain Behav Immun 23
Herlitz L, Putterman C (2017) CSF-1R inhi- (7):1003–1013
bition attenuates renal and neuropsychiatric
Chapter 15
PET Imaging in Psychoneuroimmunology Research

Jonas Hannestad
Abstract
Positron-emission tomography (PET) imaging is a valuable research tool that enables in vivo quantification
of molecular targets in the brain or of a physiologic process. PET imaging can be combined with various
experimental and clinical model systems that are commonly used in psychoneuroimmunology research. As
PET imaging can be used in animals and humans, promising results can therefore often be translated from
an animal model to human disease.
Key words PET, FDG, PBR, TSPO, Inflammation, Microglia, Endotoxin
1 Introduction
1.1 The Basics Nuclear imaging refers to in vivo imaging modalities that measure
of Positron-Emission radioactivity in a tissue or an organ after injection of a radioactive
Tomography pharmaceutical (radiopharmaceutical) that binds specifically to a
molecular target in that tissue or that is incorporated in a physio-
logic process. The radioactivity measured in the tissue being stud-
ied can be used to quantify the density of a target or the activity of a
physiologic process. Radiopharmaceuticals are interchangeably
called radioligands or radiotracers; the latter refers to the very
small mass dose (usually less than 10 μg) of radiopharmaceutical
administered in PET studies. Such “trace” amounts rarely have any
measurable pharmacologic effects because the radiotracer only
binds to 1–5% of the available target molecules. Because of this,
the regulatory requirements for first-in-human studies of radio-
tracers are less strict than for first-in-human studies of new com-
pounds administered at pharmacological doses. For instance, a
single-dose toxicity study in a single species is sufficient for regu-
latory approval to administer a new radiotracer in humans. Radio-
tracers are designed to measure a certain physiologic process (e.g.,
glucose metabolism, oxygen uptake, blood flow) or to bind to a
specific extra- or intracellular molecular target (e.g., a receptor or
transporter, or a molecule that is a marker of a specific cell type); the
287
288 Jonas Hannestad
latter use gives rise to the concept of molecular imaging (which is a

subtype of nuclear imaging). Radiotracers have three main uses in
research and drug development: (1) quantification of target density,
e.g., cross-sectional comparisons of the density of a target, such as a
receptor, between an experimental condition or disease population
and a control population; (2) determination of target occupancy by
a drug in development; for example, a drug in Phase 1 development
is administered at several dose levels to block the binding of the
radiotracer to the target, and the degree of blockade of the binding
site is used to determine the relationship between the dose or
plasma levels and target occupancy; and (3) quantification of
changes in the target density over time as the result of a therapeutic
intervention or natural course, e.g., changes beta-amyloid in Alz-
heimer’s disease during a clinical trial of an anti-amyloid therapeutic
or changes of neuroinflammation during an intervention against an
immune target. Positron-emission tomography (PET) is a type of
nuclear imaging in which the radiotracer emits positrons. This
distinguishes PET imaging from another common modality of
nuclear imaging—single-photon emission computed tomography
(SPECT)—in which the radiotracer emits gamma photons. Table 1
lists common concepts and definitions in PET imaging.
1.2 Radionuclides An element is determined by the number of protons in its nucleus.

and Radiotracers An element can have several isotopes or nuclides, i.e., atoms with
the same number of protons but different number of neutrons. For
instance, the element carbon always has six protons, but there are
several nuclides or isotopes of carbon such as 11C and 12C, which
have five and six neutrons, respectively. Because the number of
neutrons has no impact on the chemical properties of an atom, in
the chemical synthesis of a radiotracer, a radioactive isotope (radio-
nuclide) can be chemically substituted for a nonradioactive isotope
of the same element. Nuclides with a high proton-to-neutron ratio
are unstable and decay by positron emission: A proton emits a
positron and neutrino and becomes a neutron. This type of decay
is called β+ decay or positron emission. This differs from β decay in
which a neutron emits an electron and an anti-neutrino and
becomes a proton. The differential in energy between the
pre-decay nuclide and the post-decay nuclide is called transition
energy or decay energy. Proton-rich nuclides with a decay energy
1.022 MeV are positron emitters, while proton-rich nuclides with
a decay energy <1.022 MeV decay by electron capture.
Radioactivity is measured in disintegrations per second. Two
systems of units are used: Becquerel (Bq) in the Système Interna-
tional d’Unités and Curie (Ci), which denotes the radioactivity of
1 g of 226Ra (1 Ci ¼ 3.7 1010 Bq). Based on long-standing
practice, doses of radiotracers administered in PET imaging are
often measured using the Ci unit. For example, doses of radiotracer
that are injected into human subjects during PET imaging are
PET in Psychoneuroimmunology 289
Table 1
Definitions of commonly used names and PET imaging
Radiotracer A pharmaceutical or similar molecule in which a radionuclide has been substituted for an
original atom
Synonyms: Radioligand; radiopharmaceutical
Radionuclide A nuclide that decays and emits radiation
Synonym: Radioisotope
Decay Emission of radiation due to an unstable configuration of the nucleus
Synonym: Transition
Activity The amount of disintegrations per minute of a specific radionuclide
Physical half- The time it takes for the activity to be halved
life
Biological half- The time it takes for the body to clear half of the radiotracer molecules
life
Effective half- The combination of physical and biological half-life
life
Attenuation The process of a gamma photon interacting with tissue and not reaching the detectors
Scatter The process of a photon interacting with tissue and giving rise to false counts
Specific Binding to the target molecule
binding
Binding Binding to other molecules
potential
Target The density of target molecules that are available to bind the radiotracer
availability
Input function The changes in radioactivity in the blood
mostly in the 5–20 mCi range. In scientific publications, however,

radioactivity is mostly referred to using Bq units. The radioactivity
of a radionuclide is the amount of radiation, in this case positrons it
emits per unit of time. Atoms decay randomly; however, the physi-
cal half-life, i.e., the time it takes for a given radioactivity to be
reduced by 50%, is constant for each radionuclide. The effective
half-life of a radiotracer takes into account both the physical half-
life and the biological half-life of the radiotracer; the latter depends
on metabolism and excretion. Therefore, the effective half-life is the
time it takes an organism to eliminate half of the radioactivity
administered. For radionuclides such as 11C, which has a half-life
of 20 min, the elimination of radioactivity is a combination of decay
(physical half-life) and elimination of the radiotracer molecule
(biological half-life). For radionuclides with very long physical
half-lives (e.g., 124I which has a half-life of 4.2 days), the effective
half-life is for practical purposes identical to the biological half-life.
In other words, a radiotracer labeled with one of these nuclides
290 Jonas Hannestad
Table 2
Two commonly used radionuclides in PET imaging
Half-life
Radionuclide (min) Advantages Disadvantages
F-18 110 Can be synthesized elsewhere and Can only be used once in a day
transported to a PET facility
C-11 ~20 Allows for multiple scans in 1 day Requires a cyclotron and a
radiochemical lab on-site for
synthesis
would be metabolized and excreted before the radionuclide loses

any significant amount of radioactivity. Conversely, for radionu-
clides with very short physical half-lives (e.g., 15O with a half-life
of ~2 min), the effective half-life is practically identical to the
physical half-life, because the radionuclide decays so fast that the
radioactivity is close to zero before any significant amount of radio-
tracer has been excreted from the body.
Radionuclides with long or intermediate half-lives (e.g., 124I or
18
F for PET and 123I for SPECT) can be purchased from a vendor
and the nuclide can be incorporated in the radiotracer precursor
molecule in the radiochemistry lab. A radiotracer (already labeled
and ready to use) can also be shipped to a facility where the PET or
SPECT camera is; this is often the approach used in clinical imaging
(e.g., oncology or cardiology), rather than on-site synthesis of each
dose. For radionuclides with short half-lives (e.g., 11C) the nuclide
must be produced on-site in a cyclotron and rapidly incorporated
into the radiotracer molecule in the radiochemistry lab. Table 2
summarizes some of the differences between two radionuclides
most commonly used in PET research.
1.3 Annihilation When a positron-emitting nuclide decays in a tissue, which is like a

and Photon sea of electrons, the positron can only travel a short distance (usu-
Interactions ally <1 mm) before interacting with an electron. This interaction
leads to the annihilation of both the positron and the electron.
Because the momentum at the time of this interaction is almost
zero (and because momentum has to be conserved), the annihila-
tion cannot give rise to a single photon moving in one direction.
Rather, the annihilation gives rise to two photons traveling in
approximately opposite direction (Fig. 1). These photons, called
annihilation photons, are in the gamma spectrum of energy and
therefore have the same properties as gamma photons that arise
directly from decay of a gamma-emitting nuclide (e.g., the radio-
nuclides used in SPECT imaging). Because of their high energy,
many of these gamma photons will escape the body, and the PET
camera can detect them as described in detail below; however, some
of these photons will interact with atoms in the tissue through
Fig. 1 Annihilation. The positron emitted during decay of a positron-emitting

nuclide collides with an electron, both are annihilated, and two annihilation
photons emerge. These travel in opposite directions and will either be detected
by the PET system’s detector crystals or interact with the tissue. Adapted from
Essential Atlas of Cardiovascular Disease (Volume 1, Chapter 14, Published May
21, 2009, Authors: Dilsizian V, Di Carli F, Narula J). Copyright 2009 by Current
Medicine, Inc. Reproduced with kind permission from Springer Science
+Business Media B.V
which they travel. Interactions with atoms in the tissue contribute

to two important phenomena:
Attenuation refers to the fact that deeper parts of an organ
(e.g., the thalamus in the brain) appear to emit less radioactivity
than superficial parts (e.g., the cortex) because the photons from
the thalamus must traverse more tissue before reaching the PET
camera detectors than photons from the cortex. Therefore, fewer
photons (per volume of tissue) from the thalamus will reach the
detectors, and the apparent radioactivity will be lower than that of
the cortex. To account for this, attenuation correction is per-
formed. Annihilation photons interact with the tissue in three
ways: the photoelectric process, Compton scattering, and pair pro-
duction. The photoelectric process occurs when a low-energy annihi-
lation photon ejects an electron from an atom, and all the energy of
the photon is transferred to that electron. The resulting X-ray
photon cannot escape the tissue to reach the PET camera and
therefore the annihilation photon is completely “absorbed”; that
is, there is no detectable event. Thus, the photoelectric process
contributes to attenuation (i.e., loss of radioactive signal intensity).
In Compton scattering, which is the prevalent form of interac-
tion between annihilation photons and a tissue, a high-energy
photon interacts with an outer shell electron. This electron is
ejected, together with a scattered photon with less energy than
the incident photon. Compton scatter can contribute to
292 Jonas Hannestad
attenuation (if the photon never reaches the PET camera) or to

background noise (if the photon is detected by the PET camera). If
the PET camera detects a scattered photon, it will misread the
localization of the annihilation event because the photon changed
directions. To reduce the likelihood of detecting scattered photons
as “true” counts, the PET camera has a specific energy window;
Compton scattering leads to a loss of energy so that scattered
photons will be more likely to have energy below the lower limit
of this window. In addition, correction for scatter is done during
image reconstruction in an attempt to reduce the background
signal from scatter.
The third type of photon-tissue interaction is pair production,
which occurs when a photon with an energy 1.022 MeV interacts
with the nucleus of an atom, producing a positron and an electron.
This new positron will undergo annihilation as described above and
give rise to a new pair of annihilation photons which can be
detected by the PET camera; however, this new annihilation does
not represent the localization of a radiotracer molecule.
These three types of interactions reduce the signal-to-noise
ratio. The aim of the PET detection system and subsequent correc-
tions is therefore to avoid (as much as possible) counting photons
that are produced by these three types of interaction, and to count
as many of the photons that are produced during annihilation of the
positron that emits from the radiotracer and that reach the detec-
tion material without interacting with the tissue (“true counts” or
“trues”).
1.4 PET Detection PET systems detect annihilation photons using a solid scintillation
Systems detector, which is made of a crystal material with which the photon
interacts to produce visible-light photons (scintillations). The most
common solid scintillation crystals in PET imaging are bismuth
germanate (BGO), lutetium oxyorthosilicate with cerium (LSO),
and yttrium oxyorthosilicate with cerium (YSO). These crystals
vary in some important properties including (1) stopping power
(the distance the annihilation photon travels in the crystal before it
interacts with an atom and deposits its energy); (2) scintillation
decay time (the time, in nanoseconds, it takes from when the
annihilation photon excites an atom in the detector crystal until
this atom emits a visible-light photon); (3) the light output (how
much light is emitted when an annihilation photon excites a crystal
atom); and (4) the energy resolution (the fraction of the 511 keV
energy that is “converted” to visible light). Also, these crystals
should have a high ratio of photoelectric to Compton interactions
(the opposite of what a tissue has) because photoelectric interac-
tions will deposit the energy in the crystal, whereas Compton
scattering results in photons entering adjacent detectors, contribut-
ing to misreads. The visible-light photons that are produced in the
detector crystals are in turn detected by a photomultiplier tube,
Fig. 2 PET detector pairs detect photon pairs that emerge from a positron
annihilation. Source: Atlas of Cardiac Imaging (Volume 1, Chapter 10,
Published January 23, 2002. Authors: Brunken R, Wong CO, Chen E, Go R,
Lee R, Braunwald E). Copyright 1998 by Current Medicine. Reproduced with kind
permission from Springer Science+Business Media B.V
which converts the energy of these photons into an electrical pulse.

This pulse is amplified and sorted by a pulse-height analyzer. The
photomultiplier tube only accepts photons within a certain energy
window (usually 350–650 keV) to avoid scattered photons (which
lose energy during Compton interactions as described above). The
narrower this window, the more precise the energy discrimination,
but the lower the detection efficiency (the ability of the detector
material to detect as many of the annihilation photons as possible).
High detection efficiency allows scan times to be shorter and radio-
tracer doses to be lower. The detector elements are connected by a
coincidence circuit with a window of 6–20 ns to ensure that a
photon pair originates from the same positron annihilation. The
pair of photons must be detected along a line, i.e., by two detectors
that are facing each other (Fig. 2). This line does not pass exactly
through the site where the decay occurred because the positron
travels ~1 mm before annihilation. For radionuclides with high
energy (e.g., 15O) the positron range is higher, and this can cause
a blurring effect of several millimeters. For more commonly used
radionuclides (e.g., 11C and 18F) the positron range is lower and
the blurring effect is less than 0.2 mm. In this case, photon range is
therefore not a limiting factor in the resolution of PET systems used
for humans.
An annihilation arising anywhere along the line connecting a
pair of detectors is detected by the PET camera as a coincident
event. The PET camera is not able to determine the exact location
where the event occurred, only that it occurred somewhere along a
294 Jonas Hannestad
certain line connecting two detectors. Although the annihilating

photons will be at close to a 180 angle, this is approximate because
any residual momentum that the positron-electron pair had before
annihilation will cause the angle to be slightly less than or more
than 180 , a phenomenon called noncollinearity. The blurring
effect of noncollinearity is amplified in large-diameter PET cameras,
and for human cameras the blurring effects due to noncollinearity
can be up to 2 mm.
All the counts from all the detector pairs over the course of a
PET scan are combined through a process called computed tomog-
raphy, an algorithm that computes cross-sectional images that
approach the actual concentration of radioactivity in the tissue.
This allows the event count for each pixel to be determined,
which in turn is used to construct a three-dimensional image of
radioactivity in the organ or tissue. The PET image can be
co-registered to a magnetic resonance (MR) image of the same
brain (because MR has higher anatomical resolution) to aid in
region-of-interest analysis.
PET measurements of radiation emitted from the radiotracer in
the brain can be static (a continuous measure of radioactivity during
a certain amount of time) or dynamic, i.e., a sequence of contigu-
ous acquisitions which can each last anywhere from 10 s to over
20 min. In dynamic PET imaging, data from each “frame” is
independently reconstructed to form a set of images. Longer acqui-
sition times have better counting statistics but poor temporal reso-
lution, while shorter acquisitions have better temporal resolution,
but are noisier. In list-mode data acquisitions one can directly use
the arrival time to estimate a dynamic range, and this approach
provides extremely high temporal resolution with full spatial
resolution.
1.5 Data Correction As mentioned above, several corrections must be applied to ensure
that each voxel value represents (as close as possible) the real tissue
radioactivity concentration. Because each detector pair may have
different detection efficiencies, a calibration process called normal-
ization is used. A phantom (a cylinder that emits positrons in a
homogeneous manner) is placed in the PET camera and any differ-
ences in measured radioactivity will be used to calibrate (normalize)
the detector pairs before the PET camera is used for research
subjects. As described above, annihilation photons undergo atten-
uation when traversing a tissue. A transmission scan, in which an
external radioactive beam goes through the organ to be imaged
(e.g., the head), is used to measure the degree to which different
parts of the organ attenuate radioactivity. This is often done before
the emission scan (the scan which measures radioactivity emitted
from the radiotracer in the organ). If no direct measurement of
attenuation is performed (i.e., a transmission scan), attenuation
correction can be done using a mathematical model. Several
methods are used to correct for scatter, and a description of these is

beyond the scope of this chapter. Additional corrections are per-
formed for random coincidences (photons that do not originate
from the same annihilation event, but that reach a detector pair
within the time window by coincidence) and for dead time (the
time it takes for the PET camera to process an annihilation event, a
time during which another event may occur but cannot be
detected).
1.6 Radiotracer After intravenous administration, a radiotracer will distribute in the

Kinetics body according to its physicochemical properties (e.g., lipophilicity,
protein binding, volume of distribution), and it will undergo
metabolism and elimination. Metabolism of most radiotracers, as
is the case with most pharmaceuticals, occurs in the liver. When a
radiotracer is developed, it is important to avoid molecules that give
rise to a radioactive metabolite that can cross the blood-brain
barrier (BBB), because these will confound the quantification of
the PET signal. Free unmetabolized (“parent”) radiotracer in
plasma is able to cross the BBB and enter the brain parenchyma.
Likewise, free brain radiotracer is able to leave the brain and enter
the circulation. At equilibrium, the concentration of free plasma
radiotracer will equal the concentration of free brain radiotracer.
The free plasma concentration can be measured, and any additional
radioactivity in the brain is due to specific and nonspecific binding
of the radiotracer. For many radiotracers, the amount of binding in
the brain has to be calculated through kinetic modeling. This
requires the measurement of changes in radioactivity in the plasma
over time. This is called the input function and requires arterial
catheterization because of the high frequency of blood sampling.
If a radiotracer does not undergo metabolism in the brain and
forms no radioactive metabolites that can cross the BBB, the con-
centration of the radiotracer in the brain will depend on (1) the
density of the target, (2) the affinity of the radiotracer for the
target, and (3) the amount of nonspecific binding of the radiotracer
(which in turn depends on its lipophilicity and protein binding).
1.7 Specific, To disentangle how much of total binding is due to specific binding
Nonspecific, and Non- to the target and how much is due to nonspecific binding, one can
displaceable Binding use several approaches. The most robust approach is to do a block-
ing study, which is the administration of a high dose of a “cold”
(nonradioactive) ligand that binds to and blocks access to the same
binding site as the radiotracer (this ligand can be the same molecule
as the radiotracer without the positron-emitting isotope or a differ-
ent molecule). Blocking reduces radioactivity because the binding
sites are occupied and the radiotracer cannot bind to as many sites.
Any residual radioactivity (non-displaceable binding) is due to non-
specific binding. The ratio of specific to nonspecific binding is
important because that is the radiotracer’s signal-to-noise ratio.
296 Jonas Hannestad
For many applications, a high signal-to-noise ratio is necessary to

give the adequate dynamic range of measurements, especially mea-
surements of subtle changes in target density over time. During the
characterization of a new radiotracer, blocking studies are con-
ducted in rats or nonhuman primates and, if possible, in human
subjects to determine the level of non-displaceable binding. Block-
ing studies are also used to confirm the existence of a reference
region, a region that is devoid of specific binding. Subsequently, a
reference region can be used for simpler approaches to quantifica-
tion of binding, such as a ratio of the region of interest over a
reference region.
It is important to keep in mind that specific binding represents
the binding to available target-binding sites. For certain target
molecules (e.g., neurotransmitter receptors or transporters),
endogenous molecules compete with the radiotracer. PET can
therefore sometimes be used to measure changes in the levels of
an endogenous ligand, e.g., dopamine. An intervention that
increases endogenous levels of dopamine will cause a reduction in
binding of a dopamine receptor radiotracer because the dopamine
prevents the radiotracer from binding to dopamine receptor-
binding sites (see for example [1]).
Several measures of radiotracer binding are used in the litera-
ture; however, increasingly the field has moved toward a consensus
nomenclature [2]. A detailed description of measures of binding is
beyond the scope of this chapter; however, one frequently used
measure is volume of distribution (VT), which refers to the ratio of
how many mL of blood contains the same amount of radiotracer as
1 cm3 of brain tissue. For instance, a VT ¼ 5 mL·cm3 means that
5 mL of plasma has the same amount of radiotracer as 1 cm3 of
brain; that is, the concentration in the brain is five times higher than
in plasma; however, VT does not indicate whether the radiotracer in
the brain is specifically bound. Rather, VT indicates the total
amount of radiotracer in the tissue, which is the sum of free
radiotracer + specifically bound (to the target) + nonspecifically
bound (e.g., to lipids or proteins). For a radiotracer that diffuses
freely across the BBB, at equilibrium the concentration of free
tissue radiotracer equals the concentration of free plasma radio-
tracer. Therefore, at equilibrium, VT ¼ specific + nonspecific bind-
ing. For a detailed discussion of the relationship between VT and
binding potential, please see [2]. The fraction of VT that corre-
sponds to specific binding (Vs) indicates the binding potential for
the radiotracer.
2 Basics of Psychoneuroimmunology
It is well known that systemic inflammation can have profound

effects on the brain. The impact on the brain, and the pathways
through which this occurs, depends in part on the magnitude of
systemic inflammation. For instance, during severe systemic inflam-

mation (e.g., sepsis), the BBB can be disrupted, and leukocytes,
inflammatory cytokines, and bacterial toxins can enter the brain
parenchyma from the blood. This may lead to significant disruption
of brain function, including neuronal death [3], which may explain
why sepsis in humans is associated with long-term cognitive decline
[4]. In milder forms of systemic inflammation, even though the
BBB remains intact, there are clear, measurable effects on the brain.
This happens because the brain has evolved to receive information
from the rest of the body, including the immune system; this is
necessary to coordinate the activities of various systems and organs.
When faced with “danger signals” (e.g., pathogen-associated mole-
cules or molecules indicative of tissue damage), the immune system
releases inflammatory cytokines and other mediators that have both
local and systemic effects. The systemic effects include fever, activa-
tion of the hypothalamic-pituitary-adrenal (HPA) axis and the
autonomic nervous system, and changes in behavior, emotions,
and cognition, all effects that are mediated by the brain [5]. The
evolutionary purpose of these responses is believed to be the opti-
mization of the ability of the immune system to fight off the
pathogen while minimizing collateral damage to tissues and organs,
thus enhancing chances of survival [6]. The notion that the brain
receives and interprets information from the immune system is
supported by several lines of evidence. For instance, inflammatory
cytokines released during viral infections are associated with
changes in mood [7]. After experimental exposure of human sub-
jects to rhinovirus or influenza virus, increased blood levels of
inflammatory cytokines were associated with a reduction in positive
mood the following day [8]. Other mild immune stimuli also are
associated with an increase in depressive symptoms in humans
[9–13]. In the rodent literature, the support for this is substantial
and it shows us how the immune system communicates with the
brain [14–17]. One commonly used experimental paradigm in
rodents is systemic administration of endotoxin, a component of
the wall of gram-negative bacteria. Endotoxin binds to the Toll-like
4 receptor, which is found on innate immune cells, initiating a
cascade that eventually leads to release of various inflammatory
cytokines, e.g., tumor necrosis factor (TNF) and interleukin-6
(IL-6). These cytokines then signal to the brain, leading to a
constellation of behaviors similar to depression in humans: anhe-
donia, decreases in novelty-induced and social behaviors, reduced
food intake, and sleep disturbance [18]. Depending on the dose of
endotoxin used, depressive-like behaviors in rodents may continue
after the acute sickness behavior ends [19]. Such delayed effects
from immune stimuli may occur because increased brain and
peripheral levels of cytokines can last for several weeks
[20, 21]. Endotoxin administration in humans also produces
depressive-like symptoms, e.g., fatigue and anhedonia [22]. The
298 Jonas Hannestad
endothelium of the BBB plays an essential role in mediating inflam-

matory signals from the bloodstream to the brain parenchyma
[23, 24]. The luminal surface of BBB endothelial cells has receptors
for inflammatory cytokines and for endotoxin; binding of any of
these triggers a pathway involving the transcription factor nuclear
factor κB [25, 26], which allows transduction of the signal across
the BBB. In addition to the endothelium, the brain can also detect
peripheral inflammatory signals through afferent vagal fibers
[27–29]. Ultimately, systemic inflammation induces expression of
inflammatory mediators in brain parenchyma [16]. This has also
been demonstrated in human and nonhuman primates [30, 31]. It
is therefore reasonable to assume that immune-to-brain pathways
are similar in rodents and humans. In addition to the effects of the
immune system on the brain, the brain can modulate immune
responses. One important anti-inflammatory pathway is the hypo-
thalamic-pituitary-adrenal (HPA) axis: activation of the HPA axis
leads to release of cortisol, which has potent anti-inflammatory
effects [32]. The efferent vagus nerve also exerts anti-inflammatory
effects through the release of acetylcholine, which may act directly
on nicotinic receptors on immune cells [33] or indirectly through
activation of the splenic nerve, which releases anti-inflammatory
norepinephrine in the spleen [34]. Therefore, the communication
between the brain and the immune system can be viewed as a loop,
in which the immune system provides information to the brain
about events in the body, and the brain uses this information to
modulate the function of the immune system. Whether dysregula-
tion in any of these pathways plays a role in the pathogenesis of
psychiatric and neurologic disorders, in particular depression, con-
tinues to be a topic of great interest and controversy [5].
3 PET Modalities
3.1 Functional PET Functional imaging refers to a brain imaging modality that is used
Imaging: FDG to obtain an estimate of cellular (neuronal) activity. If a disease state
or an experimental intervention is associated with changes in neu-
ronal activity, functional imaging can measure this to identify which
brain regions are involved. The measures are obviously proxies for
cellular activity, and the assumption is that these proxy measures
correlate strongly with changes in the activity of neurons. An
important caveat is that functional imaging cannot distinguish
between neuronal activity and that of other cells in the brain (e.g.,
astrocytes). One commonly used modality of functional imaging is
functional MR imaging (fMRI), in which MR is used to measure
changes in oxygenated blood in the brain. The assumption is that
an increase in regional blood flow (which is under tight physiologic
control) is indicative of increased metabolic demand and therefore
increased cellular activity in a brain region. Functional imaging can
also be performed with PET. In this case, a molecule that is used

metabolically by cells, e.g., oxygen or glucose, is labeled with a
radionuclide, and PET imaging is used to measure changes in the
uptake of this molecule. For instance, fluorine-18-labeled 2-deoxy-
D-glucose (FDG) is commonly used, both clinically (oncology)
and in research (neuroscience). Deoxyglucose, like glucose, is
transported across the BBB and taken up by cells in the brain.
Inside the cells, both molecules are phosphorylated by the enzyme
hexokinase and converted to deoxyglucose-6-PO4 and glucose-6-
PO4, respectively. Unlike glucose-6-PO4 which enters the glyco-
lytic pathway, deoxyglucose-6-PO4 is not a substrate of the enzyme
phosphoglucose isomerase or any other enzyme; it therefore gets
“stuck” inside the cell. The biological half-life of FDG is therefore
very long, which means that the effective half-life is equivalent to
the physical half-life of 18F (110 min). In other words, FDG will
emit positrons for several hours after it was administered to the
subject, and therefore the cells that had the highest uptake of FDG
at the time of administration will emit the most radioactivity. This
can be an advantage in study design (see below). The amount of
FDG taken up by a given cell depends on how much glucose that
cell needs in that moment, a process that is believed to be tightly
regulated. That is, the degree of FDG uptake is proportional to the
metabolic activity of the cell at the time of FDG administration.
Therefore, the radioactive signal detected in a certain brain region is
an indication of the metabolic activity in that region at the time of
FDG administration. A detailed discussion of the theoretical under-
pinnings of FDG imaging is beyond the scope of this chapter, but
several excellent review papers have been published, e.g., [35].
The fact that the signal from FDG can be detected with a PET
camera several hours after injection, while the amount of radiation
from a tissue depends on the metabolic activity at the moment of
injection, is an advantage of FDG-PET over fMRI. This enables the
use of FDG-PET imaging to measure neuronal activity while the
subject is not in the scanner. One can take advantage of this prop-
erty of FDG and design research studies in which a subject under-
goes an experimental intervention that cannot be performed in the
scanner but during which the injection of FDG can allow us to
“capture” brain metabolic activity at that time (e.g., if systemic
inflammation is induced and it is necessary to monitor the subject
during the initial phase of that process). Another factor in the
decision to use FDG-PET over fMRI for functional imaging is
whether the measurement of absolute metabolism, rather than
relative changes in blood flow, is important for the study question.
With FDG-PET, provided that one has an appropriate input func-
tion (either through an arterial line or a cardiac imaging) one can
obtain precise estimates of absolute glucose utilization in brain
regions of interest. On the other hand, fMRI compares blood
flow between two different states, but does not give an absolute
300 Jonas Hannestad
value of metabolic rate. One important advantage of fMRI over

FDG-PET is cost; the latter is several times more expensive.
FDG was approved by FDA in 1994; therefore, unlike most
other PET radiotracers, its use in human subjects does not require
an FDA investigational new drug application (IND). There are
many examples in the literature of the use of FDG-PET imaging
in psychoneuroimmunology research. For instance, Semmler et al.
used FDG-PET in rats that received high doses of endotoxin and
found that systemic inflammation was associated with a global
reduction in metabolism, especially in the cortex, and that this
correlated with necrosis in postmortem samples [3]. Capuron
et al. used FDG-PET to measure changes in glucose metabolism
in patients who received treatment with interferon-alpha, a mild,
iatrogenic, inflammatory stimulus that is associated with emergence
of depressive symptoms, and found that changes in glucose metab-
olism in the nucleus accumbens and putamen correlated with
interferon-induced fatigue [36]. Our group used FDG-PET to
measure changes in glucose metabolism in human subjects who
received endotoxin, which induces a state of transient systemic
inflammation as described above. Endotoxin administration in
humans requires a physician-sponsored IND application with
FDA. For alternatives to endotoxin in human-subject research, see
Note 1. Endotoxin for human use can be provided by the NIH
Clinical Center upon request (contact person: Dr. Anthony Suffre-
dini). For the background on Clinical Center Reference Endotoxin
(CCRE), see Note 2. Intact endotoxin vials should be stored in the
refrigerator (2–8 C). The preparation of endotoxin for human use
should be performed by a pharmacist.
3.2 Molecular PET Molecular or receptor imaging refers to the use of PET imaging to
Imaging: TSPO measure the availability of specific receptors or other molecules of
interest (e.g., neurotransmitter transporters, cell-specific markers).
Radiotracers are designed to bind to a specific molecule so that its
distribution and density can be measured in vivo. The development
of new radiotracers is a laborious process, the description of which
is beyond the scope of this chapter. Although some radiotracers are
commercially available, the majority of radiotracers that have been
developed at PET facilities around the world require on-site
synthesis.
There is considerable interest in imaging neuroinflammation
in vivo. Most PET imaging of neuroinflammation uses radiotracers
that bind to the translocator protein (TSPO; see below); however,
there have been several promising new developments in this field
that are worth summarizing.
1. PET tracers that bind to the cannabinoid receptor type
2 (CB2) have been developed as markers of neuroinflamma-
tion. Examples include 11C-A836339 [37, 38], and 11C-RS-
016 [39]. Although PET targeting CB2 appears to be a good

marker of neuroinflammation, this target has similar limitations
to TSPO; that is, it does not provide much information about
underlying molecular processes.
2. Cyclooxygenase-2 (COX-2) is an enzyme that converts arachi-
donic acid (AA) to prostaglandins, which are essential inflam-
matory mediators. Several PET tracers have been developed to
measure COX-2 activity (reviewed by Tietz, Marshall, Wuest,
Wang, and Wuest 2013); however, so far none of these tracers
has enabled in vivo imaging of COX-2 activity [40].
3. The enzyme monoacylglycerol lipase converts
2-arachidonoylglycerol, enabling prostaglandin E2 synthesis,
and a PET tracer for monoacylglycerol lipase, 11C-
SAR127303, was recently developed [41].
4. Since COX-2 does not convert AA to other inflammatory
mediators such as leukotrienes, PET tracers that, rather than
measuring COX-2 activity, measure AA turnover, e.g.,
1-[11C]-AA [42] and [18F]-FAA [43], may be useful in under-
standing the role of AA in inflammation.
5. P2X(7) is a member of the ATP-gated ion channel family and
plays an essential role in inflammatory processes [44] and in
neurological diseases in which inflammation appears to play a
role, such as Parkinson’s and Alzheimer’s disease [45].
Recently, two PET tracers for P2X(7) were developed. No
in vivo data are available yet on [11C]-GSK1482160 [46],
and [11C]-A-740003 [47] has very low brain uptake.
6. Activated microglia release β-glucuronidase, and the PET tracer
[18F]FEAnGA was developed to measure β-glucuronidase activ-
ity in vivo. Despite low brain uptake, [18F]FEAnGA was able to
detect an increase in β-glucuronidase in an animal model of
neuroinflammation [48].
7. Matrix metalloproteinases (MMP) are expressed after tissue
injury or during inflammation and these enzymes have been
implicated in the pathophysiology of Alzheimer’s disease [49]
and stroke [50]. The MMP-3 inhibitor CGS 27023A was
radiolabeled with 18F for PET and 123I for SPECT imaging
[51, 52].
If any of these radiotracers are suitable for in vivo studies they
may prove very useful in deepening our understanding of the
molecular underpinnings of various types of inflammatory pro-
cesses. Despite these promising new developments in the field of
PET imaging of neuroinflammation, most studies still use TSPO
radiotracers. Because TSPO is found at high levels in activated
microglia, a brief description of these cells is therefore provided
next as background.
302 Jonas Hannestad
3.2.1 Microglia Microglia (resident macrophages of the brain) are involved in a

variety of physiologic and pathologic processes, most importantly,
in the initiation and maintenance of neuroinflammation. When
provided with appropriate molecular signals from neurons or astro-
cytes, resting microglia become activated and release substances
that can cause neuronal dysfunction and damage, e.g., inflamma-
tory cytokines and reactive oxygen species [53], by interfering with
neurotransmission, inhibiting neuroplasticity, and causing neuronal
death [54]. Microglia can also be activated by systemic inflamma-
tion. For instance, peripheral administration of endotoxin in
rodents is associated with activation of microglia [3, 20]. Once
stimulated, microglia can stay activated for several months and
continue to express inflammatory mediators [20, 54, 55]. Activa-
tion of microglia by endotoxin or inflammatory cytokines leads to
further release of inflammatory cytokines and other potentially
neurotoxic substances [54], while exposure to anti-inflammatory
cytokines induces a neuroprotective microglial phenotype
[56–58]. Because of the pivotal role that microglia play in a variety
of neurodegenerative, neuroinflammatory, and ischemic disorders
of the brain, the ability to measure activation of microglia in vivo
with PET imaging is an area of great interest.
3.2.2 Translocator The translocator protein (TSPO) is an 18 kDa mitochondrial pro-

Protein tein that is expressed in steroid-synthesizing cells, in which its role is
to allow transport cholesterol across the mitochondrial membranes.
The expression of TSPO is very low in healthy brain tissue, but it
increases in pathologies associated with microglial activation
including stroke, trauma, infection, and autoimmune and neuro-
degenerative disorders [59, 60]; therefore PET imaging of TSPO
can be used to measure microglial activation [61]. Several PET
radiotracers that bind to TSPO are available [62].
The first radiotracer developed for TSPO was 11C-PK11195,
which has low signal-to-noise ratio. In the last decade and a half,
several second-generation TSPO tracers have been developed [63].
These new TSPO tracers have much better sensitivity, but, unlike
11
C-PK11195, their affinity for TSPO is affected by a polymor-
phism in the TSPO gene [64]. Fortunately, genotyping for this
polymorphism is straightforward and can be used to determine
whether a subject is a high-affinity, mixed-affinity, or low-affinity
binder.
Another limitation of TSPO PET imaging is that it is not clear
whether it can distinguish between various microglial phenotypes
(e.g., M1 vs. M2) and that it cannot distinguish between microglial
activation and reactive astrogliosis. In other words, TSPO PET
imaging is a gross measure of “neuroinflammation” that does not
provide detailed information about the underlying molecular pro-
cesses. New tracers that target microglial phenotypes would be
helpful to increase our understanding of immune-brain

interactions.
The PET tracer [11C]PBR28 binds to TSPO and has excellent
signal-to-noise [65, 66]. Due to the short half-life of 11C, [11C]
PBR28 must be synthesized on-site (for 18F alternatives, see Note
3). On the other hand, the 20-min half-life of 11C permits multiple
measurements of microglial activation within a short time period
which allows for a variety of research-design options. For instance,
our group used this radiotracer to measure binding to TSPO at
various times after endotoxin administration in baboons and in
humans. The use of baboons (instead of rodents) is advantageous
because of the similarity with human brain size and receptor sys-
tems, and because the body size, peripheral metabolism, and clear-
ance allow for quantification through kinetic modeling; however,
PET imaging of microglial activation can also be performed in
rodents.
The baboon endotoxin model was used because it is physiolog-
ically and immunologically very similar to human sepsis [67]. In
our experiments, we used female baboons (Papio anubis) that had
been ovariohysterectomized at least 3 months prior to the study
because estrogen can affect TSPO expression [68]. Recently, this
tracer was also used to detect a decrease in TSPO after administra-
tion of the colony-stimulating factor receptor 1 (CSF1R) inhibitor
PLX3397 (Hillmer et al. 2017). In summary, despite the limita-
tions of TSPO as a microglia target, these experiments show that
TSPO levels can be modulated in vivo and that this modulation can
be detected with PET imaging.
4 Summary
The few examples given in this chapter are only a small subset of the
types of experiments that can be designed to incorporate PET
imaging as a research tool. The two main limitations of using
PET imaging in psychoneuroimmunology research are cost and
tracer/PET center availability. If those two limitations can be
addressed, there are a variety of PET tracers that can be used to
address myriad questions in this field. For instance, PET can be
used to measure changes in striatal dopamine levels, and this
approach was recently combined with the human endotoxin admin-
istration model, showing that endotoxin increased the amount of
synaptic dopamine [69].
304 Jonas Hannestad
5 Notes
1. Endotoxin administration is one of several methods that can be

used to induce mild systemic inflammation in human subjects
for research purposes (see for review [5]). Other immune sti-
muli that are frequently used include typhoid vaccination and
interferon-alpha (which can be used as treatment or it can be
administered to healthy subjects). The latter two are approved
by FDA for human administration and do therefore not require
approval beyond the local institutional review board.
2. Clinical Center Reference Endotoxin injectable dosage forms
are manufactured by the Bureau of Biologics. CCRE is a pur-
ified lipopolysaccharide prepared from Escherichia coli O:113
(U.S. Standard Reference Endotoxin) and vialed under good
manufacturing practice guidelines. CCRE comes in 5 mL clear-
glass vials, each containing a white, lyophilized powder which
consists of 10,000 endotoxin units (approximately 1 mcg of
reference endotoxin), 10 mg of lactose, and 1 mg of polyethyl-
ene glycol 6000. The dose of endotoxin used in humans ranges
from 0.2 to 4 ng/kg body weight. If the main purpose is to
induce a state of mild systemic inflammation with subtle symp-
toms of depression, a dose of less than 1 ng/kg is
recommended [5].
3. Due to the short half-life of 11C, [11C]PBR28 must be synthe-
sized on-site. There are several 18F TSPO tracers, including
[18F]PBR06 [70] which is a close analog of [11C]PBR28 with
similar performance. Although [18F]PBR06 could be synthe-
sized by a vendor and shipped to the PET site, it is not com-
mercially available; however, it may be possible to arrange by
collaboration with a different site that has the ability to synthe-
size [18F]PBR06.
References
1. Morris ED, Constantinescu CC, Sullivan JM, 4. Iwashyna TJ, Ely EW, Smith DM, Langa KM
Normandin MD, Christopher LA (2010) (2010) Long-term cognitive impairment and
Noninvasive visualization of human dopamine functional disability among survivors of severe
dynamics from PET images. Neuroimage 51 sepsis. JAMA 304:1787–1794
(1):135–144 5. DellaGioia N, Hannestad J (2010) A critical
2. Innis RB, Cunningham VJ, Delforge J, Fujita review of human endotoxin administration as
M, Gjedde A, Gunn RN et al (2007) Consen- an experimental paradigm of depression. Neu-
sus nomenclature for in vivo imaging of rever- rosci Biobehav Rev 34:130–143
sibly binding radioligands. J Cereb Blood Flow 6. Raison CL, Capuron L, Miller AH (2006)
Metab 27:1533–1539 Cytokines sing the blues: inflammation and
3. Semmler A, Hermann S, Mormann F, the pathogenesis of depression. Trends Immu-
Weberpals M, Paxian SA, Okulla T et al nol 27:24–31
(2008) Sepsis causes neuroinflammation and 7. Bucks RS, Gidron Y, Harris P, Teeling J,
concomitant decrease of cerebral metabolism. Wesnes KA, Perry VH (2008) Selective effects
J Neuroinflammation 5:38 of upper respiratory tract infection on
cognition, mood and emotion processing: a 20. Qin L, Wu X, Block ML, Liu Y, Breese GR,
prospective study. Brain Behav Immun Hong JS et al (2007) Systemic LPS causes
22:399–407 chronic neuroinflammation and progressive
8. Janicki-Deverts D, Cohen S, Doyle WJ, Turner neurodegeneration. Glia 55:453–462
RB, Treanor JJ (2007) Infection-induced proin- 21. Moreau M, Andre C, O’Connor JC, Dumich
flammatory cytokines are associated with SA, Woods JA, Kelley KW et al (2008) Inocu-
decreases in positive affect, but not increases in lation of Bacillus Calmette-Guerin to mice
negative affect. Brain Behav Immun 21:301–307 induces an acute episode of sickness behavior
9. Hermann DM, Mullington J, Hinze-Selch D, followed by chronic depressive-like behavior.
Schreiber W, Galanos C, Pollmacher T (1998) Brain Behav Immun 22(7):1087–1095
Endotoxin-induced changes in sleep and sleep- 22. Hannestad J, DellaGioia N, Ortiz N, Pittman
iness during the day. Psychoneuroendocrinol- B, Bhagwagar Z (2011) Citalopram reduces
ogy 23:427–437 endotoxin-induced fatigue. Brain Behav
10. Reichenberg A, Yirmiya R, Schuld A, Kraus T, Immun 25:256–259
Haack M, Morag A et al (2001) Cytokine- 23. Gosselin D, Rivest S (2008) MyD88 signaling
associated emotional and cognitive distur- in brain endothelial cells is essential for the
bances in humans. Arch Gen Psychiatry neuronal activity and glucocorticoid release
58:445–452 during systemic inflammation. Mol Psychiatry
11. Wright CE, Strike PC, Brydon L, Steptoe A 13:480–497
(2005) Acute inflammation and negative 24. Ching S, Zhang H, Belevych N, He L, Lai W,
mood: mediation by cytokine activation. Brain Pu XA et al (2007) Endothelial-specific knock-
Behav Immun 19:345–350 down of interleukin-1 (IL-1) type 1 receptor
12. Eisenberger NI, Inagaki TK, Rameson LT, differentially alters CNS responses to IL-1
Mashal NM, Irwin MR (2009) An fMRI depending on its route of administration. J
study of cytokine-induced depressed mood Neurosci 27:10476–10486
and social pain: the role of sex differences. 25. Dauphinee SM, Karsan A (2006) Lipopolysac-
Neuroimage 47(3):881–890 charide signaling in endothelial cells. Lab
13. Capuron L, Fornwalt FB, Knight BT, Harvey Invest 86:9–22
PD, Ninan PT, Miller AH (2009) Does cyto- 26. Magder S, Neculcea J, Neculcea V, Sladek R
kine-induced depression differ from idiopathic (2006) Lipopolysaccharide and TNF-alpha
major depression in medically healthy indivi- produce very similar changes in gene expres-
duals? J Affect Disord 119(1–3):181–185 sion in human endothelial cells. J Vasc Res
14. De La Garza R 2nd (2005) Endotoxin- or pro- 43:447–461
inflammatory cytokine-induced sickness behav- 27. Kapas L, Hansen MK, Chang HY, Krueger JM
ior as an animal model of depression: focus on (1998) Vagotomy attenuates but does not pre-
anhedonia. Neurosci Biobehav Rev vent the somnogenic and febrile effects of lipo-
29:761–770 polysaccharide in rats. Am J Physiol 274:
15. Dunn AJ, Swiergiel AH, de Beaurepaire R R406–R411
(2005) Cytokines as mediators of depression: 28. Opp MR, Toth LA (1998) Somnogenic and
what can we learn from animal studies? Neu- pyrogenic effects of interleukin-1beta and lipo-
rosci Biobehav Rev 29:891–909 polysaccharide in intact and vagotomized rats.
16. Pecchi E, Dallaporta M, Jean A, Thirion S, Life Sci 62:923–936
Troadec JD (2009) Prostaglandins and sickness 29. Konsman JP, Luheshi GN, Bluthe RM, Dant-
behavior: old story, new insights. Physiol Behav zer R (2000) The vagus nerve mediates beha-
97(3-4):279–292 vioural depression, but not fever, in response to
17. Anisman H (2009) Cascading effects of stres- peripheral immune signals; a functional anato-
sors and inflammatory immune system activa- mical analysis. Eur J Neurosci 12:4434–4446
tion: implications for major depressive 30. Reyes TM, Coe CL (1996) Interleukin-1 beta
disorder. J Psychiatry Neurosci 34:4–20 differentially affects interleukin-6 and soluble
18. Larson SJ, Dunn AJ (2001) Behavioral effects interleukin-6 receptor in the blood and central
of cytokines. Brain Behav Immun 15:371–387 nervous system of the monkey. J Neuroimmu-
19. Frenois F, Moreau M, O’Connor J, Lawson M, nol 66:135–141
Micon C, Lestage J et al (2007) Lipopolysacchar- 31. Raison CL, Borisov AS, Majer M, Drake DF,
ide induces delayed FosB/DeltaFosB immunos- Pagnoni G, Woolwine BJ et al (2009) Activa-
taining within the mouse extended amygdala, tion of central nervous system inflammatory
hippocampus and hypothalamus, that parallel pathways by interferon-alpha: relationship to
the expression of depressive-like behavior. Psy- monoamines and depression. Biol Psychiatry
choneuroendocrinology 32:516–531 65(4):296–303
306 Jonas Hannestad
32. Bierhaus A, Wolf J, Andrassy M, Rohleder N, 43. Pichika R, Taha A, Gao F, Kotta K, Cheon Y,
Humpert PM, Petrov D et al (2003) A mecha- Chang L, Kiesewetter D, Rapoport S, Eckel-
nism converting psychosocial stress into mono- man W (2012) The synthesis and in vivo phar-
nuclear cell activation. Proc Natl Acad Sci U S macokinetics of fluorinated arachidonic acid:
A 100:1920–1925 implications for imaging neuroinflammation. J
33. Pavlov VA, Tracey KJ (2005) The cholinergic Nucl Med 53(9):1383–1391
anti-inflammatory pathway. Brain Behav 44. Di Virgilio F (2015) P2X receptors and inflam-
Immun 19:493–499 mation. Curr Med Chem. 22(7):866–877
34. Vida G, Pena G, Deitch EA, Ulloa L (2011) 45. Sáez-Orellana F, Godoy P, Silva-Grecchi T,
alpha7-cholinergic receptor mediates vagal Barra K, Fuentealba J (2015) Modulation of
induction of splenic norepinephrine. J Immu- the neuronal network activity by P2X receptors
nol 186:4340–4346 and their involvement in neurological disor-
35. Wienhard K (2002) Measurement of glucose ders. Pharmacol Res. 101:109–115
consumption using [(18)F]fluorodeoxyglu- 46. Gao M, Wang M, Green M, Hutchins G,
cose. Methods 27:218–225 Zheng Q (2015) Synthesis of [(11)C]
36. Capuron L, Pagnoni G, Demetrashvili MF, Law- GSK1482160 as a new PET agent for targeting
son DH, Fornwalt FB, Woolwine B et al (2007) P2X(7) receptor. Bioorg Med Chem Lett 25
Basal ganglia hypermetabolism and symptoms of (9):1965–1970
fatigue during Interferon-alpha therapy. Neu- 47. Janssen B, Vugts D, Funke U, Spaans A, Schuit
ropsychopharmacology 32:2384–2392 R, Kooijman E, Rongen M, Perk L, Lam-
37. Horti A, Gao Y, Ravert H, Finley P, Valentine mertsma A, Windhorst A (2014) Synthesis
H, Wong D, Endres C, Savonenko A, Dannals and initial preclinical evaluation of the P2X7
R (2010) Synthesis and Biodistribution of receptor antagonist [11C]A-740003 as a
[11C]A-836339, a New Potential Radioligand novel tracer of neuroinflammation. J Label
for PET Imaging of Cannabinoid Type Compd Radiopharm 57(8):509–516
2 Receptors (CB2). Bioorg Med Chem 18 48. Antunes I, Doorduin J, Haisma H, Elsinga P,
(14):5202–5207 van Waarde A, Willemsen A, Dierckx R, de
38. Savonenko A, Melnikova T, Wang Y, Ravert H, Vries E (2012) 18F-FEAnGA for PET of b-
Gao Y, Koppe I, Lee D, Pletnikova O, Cho E, Glucuronidase Activity in Neuro- inflamma-
Sayyida N, Hiatt A, Troncoso J, Davies P, Dan- tion. J Nucl Med. 53(3):451–458
nals R, Pomper M, Horti A (2015) Cannabi- 49. Wang X, Tan M, Yu J, Tan L (2014) Matrix
noid CB2 receptors in a mouse model of Ab metalloproteinases and their multiple roles in
amyloidosis: immunohistochemical analysis alzheimer’s disease. Biomed Res Int
and suitability as a PET biomarker of neuro- 2014:908636
inflammation. PLoS One 10(6):e0129618 50. Yang Y, Rosenberg G (2015) Matrix metallo-
39. Slavik R, Herde A, Bieri D, Weber M, Schibli R, proteinases as therapeutic targets for stroke.
Kr€amer S, Ametamey S, Mu L (2015) Synthe- Brain Res 1623:30–38
sis, radiolabeling and evaluation of novel 4- 51. Wagner S, Breyholz H, Höltke C, Faust A,
oxo-quinoline derivatives as PET tracers for Schober O, Sch€afers M, Kopka K (2009) A
imaging cannabinoid type 2 receptor. Eur J New 18F-Labelled Derivative of the MMP
Med Chem 92:554–564 Inhibitor CGS 27023A for PET: Radiosynth-
40. Tietz O, Marshall A, Wuest M, Wang M, Wuest esis and Initial Small-Animal PET Studies. Appl
F (2013) Radiotracers for molecular imaging of Radiat Isot 67(4):606–610
cyclooxygenase-2 (COX-2) enzyme. Curr Med 52. Matusiak N, van Waarde A, Bischoff R, Olten-
Chem 30(35):4350–4369 freiter R, van de Wiele C, Dierckx R, Elsinga P
41. Wang C, Placzek M, Van de Bittner G, Schroe- (2013) Probes for non-invasive matrix metal-
der F, Hooker J (2016) A novel radiotracer for loproteinase-targeted imaging with PET and
imaging monoacylglycerol lipase in the brain SPECT. Curr Pharm Des 19(25):4647–4672
using positron emission tomography. ACS 53. Tambuyzer BR, Ponsaerts P, Nouwen EJ
Chem Neurosci. 7(4):484–489 (2009) Microglia: gatekeepers of central ner-
42. Esposito G, Giovacchini G, Liow J, Bhattachar- vous system immunology. J Leukoc Biol
jee A, Greenstein D, Schapiro M, Hallett M, 85:352–370
Herscovitch P, Eckelman W, Carson R, Rapo- 54. Block ML, Zecca L, Hong JS (2007) Micro-
port S (2008) Imaging neuroinflammation in glia-mediated neurotoxicity: uncovering the
Alzheimer’s disease with radiolabeled arachido- molecular mechanisms. Nat Rev Neurosci
nic acid and PET. J Nucl Med 49 8:57–69
(9):1414–1421 55. Maeda J, Higuchi M, Inaji M, Ji B, Haneda E,
Okauchi T et al (2007) Phase-dependent roles
of reactive microglia and astrocytes in nervous 64. Owen D, Yeo A, Gunn R, Song K, Wadsworth
system injury as delineated by imaging of G, Lewis A, Rhodes C, Pulford D, Bennacef I,
peripheral benzodiazepine receptor. Brain Res Parker C, StJean P, Cardon L, Mooser V, Mat-
1157:100–111 thews P, Rabiner E, Rubio J (2012) An 18-kDa
56. Li L, Lu J, Tay SS, Moochhala SM, He BP translocator protein (TSPO) polymorphism
(2007) The function of microglia, either neu- explains differences in binding affinity of the
roprotection or neurotoxicity, is determined by PET Radioligand PBR28. J Cereb Blood
the equilibrium among factors released from Flow Metab 32(1):1–5
activated microglia in vitro. Brain Res 65. Brown AK, Fujita M, Fujimura Y, Liow JS,
1159:8–17 Stabin M, Ryu YH et al (2007) Radiation
57. Hanisch UK, Kettenmann H (2007) Micro- dosimetry and biodistribution in monkey
glia: active sensor and versatile effector cells in and man of 11C-PBR28: a PET radioligand
the normal and pathologic brain. Nat Neurosci to image inflammation. J Nucl Med
10:1387–1394 48:2072–2079
58. Pinteaux E, Rothwell NJ, Boutin H (2006) 66. Imaizumi M, Briard E, Zoghbi SS, Gourley JP,
Neuroprotective actions of endogenous inter- Hong J, Fujimura Y et al (2008) Brain and
leukin-1 receptor antagonist (IL-1ra) are whole-body imaging in nonhuman primates
mediated by glia. Glia 53:551–556 of [(11)C]PBR28, a promising PET radioli-
59. Cosenza-Nashat M, Zhao ML, Suh HS, Mor- gand for peripheral benzodiazepine receptors.
gan J, Natividad R, Morgello S et al (2009) Neuroimage 39:1289–1298
Expression of the translocator protein of 18 67. Redl H, Bahrami S (2005) Large animal mod-
kDa by microglia, macrophages and astrocytes els: baboons for trauma, shock, and sepsis stud-
based on immunohistochemical localization in ies. Shock 24(Suppl 1):88–93
abnormal human brain. Neuropathol Appl 68. Hillmer AT, Sandiego CM, Hannestad J,
Neurobiol 35:306–328 Angarita GA, Kumar A, McGovern EM,
60. Batarseh A, Papadopoulos V (2010) Regula- Huang Y, O’Connor KC, Carson RE, O’Mal-
tion of translocator protein 18 kDa (TSPO) ley SS, Cosgrove KP (2017) In vivo imaging of
expression in health and disease states. Mol translocator protein, a marker of activated
Cell Endocrinol 327:1–12 microglia, in alcohol dependence. Mol Psychi-
61. Venneti S, Lopresti BJ, Wiley CA (2006) The atry 22(12):1759–1766
peripheral benzodiazepine receptor (Translo- 69. Petrulli JR, Kalish B, Nabulsi NB, Huang Y,
cator protein 18kDa) in microglia: from Hannestad J, Morris ED (2017) Systemic
pathology to imaging. Prog Neurobiol inflammation enhances stimulant-induced
80:308–322 striatal dopamine elevation. Transl Psychiatry.
62. Rupprecht R, Papadopoulos V, Rammes G, 7(3):e1076
Baghai TC, Fan J, Akula N et al (2010) Trans- 70. Fujimura Y, Kimura Y, Simeon FG, Dickstein
locator protein (18 kDa) (TSPO) as a thera- LP, Pike VW, Innis RB et al (2010) Biodistri-
peutic target for neurological and psychiatric bution and radiation dosimetry in humans of a
disorders. Nat Rev Drug Discov 9:971–988 new PET ligand, (18)F-PBR06, to image
63. Vivash L, O’Brien T (2016) Imaging micro- translocator protein (18 kDa). J Nucl Med
glial activation with TSPO PET: lighting up 51:145–149
neurologic diseases? J Nucl Med. 57
(2):165–168
Chapter 16
The Vaccination Model in Psychoneuroimmunology

Research: A Review
Anna C. Whittaker
Abstract
This chapter explores the reasoning behind using the vaccination model to examine the influence of
psychosocial factors on immunity. It then briefly discusses the mechanics of the vaccination response and
the protocols used in psychoneuroimmunology vaccine research, before giving examples from the research
literature of the studies examining relationships such as the association between stress and vaccination
response. It also explores the ways the vaccination model can be used to answer key questions in psycho-
neuroimmunology, such as the following: “Does it matter when stressful life events occur relative to when
the vaccine is received?” “What are the effects of prior exposure to the antigen?” “Do other psychosocial
factors influence vaccine response besides stress?” Finally, it briefly considers the mechanisms underlying
psychosocial factors and vaccination response associations and the future research needed to understand
these better, and indeed to use current and future knowledge to improve and enhance vaccine responses in
key at-risk populations.
Key words Caregiving, Influenza, Interventions, Social support, Stress, Vaccine
1 Introduction: Why Study Vaccination in the Context of PNI Research?
1.1 Alternative There are many methods for examining the effects of psychological
Approaches: factors on immunity. Early work concentrated on the influence of
Enumerative Measures psychosocial stress on enumerative measures of immunity. For
example, individuals exposed to chronic stress showed reduced
numbers of certain immune cells including reduced numbers of
B-lymphocytes [1, 2], helper T-lymphocytes [1, 3, 4], cytotoxic
T-lymphocytes [1, 5], natural killer (NK) cells [1, 5], and lowered
concentrations of secretory immunoglobulin A in saliva [6–10],
compared to matched controls. However, it is difficult to determine
the clinical significance of such enumerative changes, given that
they lie within the normal range for healthy participants [11] and
may simply reflect cell migration and recirculation rather than
increased production or better function [11]. Additionally, cell
number changes could be a consequence of shifts in plasma volume
309
310 Anna C. Whittaker
and hemoconcentration; in such circumstances, changes in cell

number would reflect increased density of a lymphocyte population
rather than signaling a true increase in absolute cell numbers.
Further, even absolute changes in cell number might not necessar-
ily reflect alteration in the capacity of the immune system to mount
an effective response to antigenic challenge [11]. Consequently,
measuring changes in cell number is perhaps not the optimal means
of determining variations in the functional capacity of the immune
system, and hence the likely clinical implications of psychosocial
variables for disease resistance and susceptibility.
1.2 In Vitro Measures In vitro measures of immune function, such as cell proliferation to
stimulation with an antigen (foreign material, e.g., bacteria), or cell
cytotoxicity (killing ability), have been argued to provide a better
indication of the functional capacity of the immune system
[11]. These measures have been demonstrated to be susceptible
to impairment by chronic stress in many studies, e.g., [12–15]. For
example, recently unemployed individuals showed poorer lympho-
cyte proliferation to antigen than those in employment [16]. Fur-
ther, compared to non-bereaved controls, individuals who have
suffered bereavement showed lower neutrophil superoxide produc-
tion, one of their key cytotoxic capacities through which they
eradicate bacteria such as pneumonia [17]. Nevertheless, the
isolated testing of any particular network of immune cells provides
only limited information about the overall status of what is a highly
integrated and complex system [11], and an imperfect understand-
ing of the relationship between psychosocial factors and vulnerabil-
ity to disease [18].
2 The Vaccination Model
2.1 Benefits of the A clinically relevant model which examines the impact of psychoso-
Vaccination Model cial factors on the integrated response of the immune system to a
challenge would avoid these disadvantages. The antibody response
to vaccination provides us with such a model. Vaccines act as real
immune system challenges, although they are altered in such a way
so as not to induce disease either by being inactivated or killed, or
only a component of the actual pathogen, so are really “imitation
infections.” Therefore, by measuring the antibody levels in
response to vaccination we can assess directly how well the immune
system responds to infectious challenge. It is also clinically relevant
in that antibody levels or titers are directly related to susceptibility
and resistance to infectious disease.
2.2 The Vaccination The vaccination response involves the coordination of a wide vari-
Response ety of immune cells. Antigen is initially recognized and presented
by professional antigen-presenting cells, such as dendritic cells.
Vaccine Research in PNI 311
Thus presented, the antigen is then recognized by specific helper T

cells which process and present the antigen to B cells, the antibody
factories of the immune system; this is termed a thymus-dependent
response. There are other types of vaccination which are also
recognized by B cells without the necessity for T-cell help, thus
termed a thymus-independent response, which do not elicit as
strong or maintained a response as thymus-dependent vaccines. A
final type of vaccination, called a conjugate vaccination, is used to
improve the response to thymus-independent antigens by attaching
a protein to the antigen, which then stimulates an immune response
involving helper T-cell recognition.
When stimulated by an antigen, B-lymphocytes replicate and
mature into short-lived plasma cells which produce the earliest
antibody or immunoglobulin, IgM. In a primary response to an
antigen not previously encountered, the peak IgM response occurs
around 5 days after vaccination. Interaction between activated T
and B cells leads to the production of high-affinity or very specific
antibodies in bodily fluids: IgG (found mainly in the blood) and
IgA (found mainly at mucosal surfaces, e.g., in saliva). This more
specific response peaks around 28 days after vaccination. Other
types of antibody include IgE (part of the allergic response) and
IgD. IgG is particularly important, as being the most prolific anti-
gen in the blood and a more specific match to the particular antigen
makes it more effective at antigen elimination. Secondary antibody
responses, in which the immune system has been previously
exposed to the antigen, are more rapid and of greater magnitude;
this is because some activated T and B cells become long-lived
resting memory cells, remaining in the immune system ready to
respond quickly to challenge with previously encountered antigens.
Part of the response to vaccination is also cell mediated as well as
humoral (involving antibodies) such that helper T cells are also
activated and initiate production of key pro-inflammatory cyto-
kines, including IL-1β, IL-6, and IL-8, in response to vaccine
antigens, which can also be measured quantitatively as part of the
vaccination response, although it is less commonly done so.
Not all individuals react with a strong antibody response to
vaccination, particularly older adults who are only protected against
influenza disease in 30–50% of cases following vaccination
[19–22]. Further, the increase in vaccination availability has not
been paralleled by decreased influenza-related mortality [23]. This
variation in the vaccination response allows for the investigation of
other factors which might influence this aspect of immunity
between individuals. The relevance of the response to infectious
disease risk provides the clinically relevant imperative to do so
beyond the interest in increasing knowledge on how various factors
affect immune function. As well as age, psychosocial factors, such as
stress, may alter both the quantity and quality of antibody present
at different times after immunization, meaning that individuals
suffering from higher levels of stress are more at risk of infectious
disease.
3 Stress and Vaccination Response
3.1 Stress The most common psychosocial factor examined in the context of
Questionnaires the vaccination response is stress. This is usually assessed via life
event checklists or perceived stress measures. Life event checklists
consist of a list of major and minor life events, e.g., bereavement
and moving house, and usually require participants to indicate
which have occurred during the past month or year [24]. Some
also ask participants to indicate how stressful each event was on a
rating scale. Life events have been shown to predict a variety of
important physical health outcomes, including infectious disease
[25], and mortality, particularly in the context of little emotional
support [26]. In contrast, perceived stress scales measure indivi-
duals’ feelings about how stressful their lives are rather than the
direct occurrence of events [27]. Thus these measures are more
susceptible to subjective bias, and are better predictors of subjective
health outcomes, such as angina, rather than objective outcomes,
such as myocardial infarction [28].
3.2 Caregiver Control Another common way of assessing stress in the context of vaccina-
Models tion is to examine antibody responses among those with a key
chronic stressor versus a sociodemographically matched control
group, for example, older adults caregiving for a spouse with
dementia. The stress of caregiving has been shown to relate to
poor health and mortality [29], and can thus be considered an
important source of ongoing psychological stress. Other stress
studies compare matched controls to other groups subject to
chronic stressors such as bereavement, marital separation/divorce,
or unemployment.
3.3 Protocol for In order to fully test the impact of psychosocial factors on the
Stress and Vaccination response to vaccination, both pre-vaccination and post-vaccination
Studies blood samples are required for assessment of antibody levels. This is
due to the impact that prior vaccination or environmental exposure
to the infectious agent can have on pre-vaccination antibody levels,
and consequently post-vaccination levels. Without taking a
pre-vaccination baseline, it is difficult to state whether stress is
affecting the antibody response to a vaccination administered dur-
ing a research study or simply the maintenance of previous antibody
levels. For example, in 37 nursing-home residents, those who
reported higher levels of perceived stress had lower pre-vaccine
antibody titers to two influenza vaccine components
[30]. However, it is not clear what this means, given that

pre-vaccine titers could reflect differences in prior vaccine history
or exposure. In this same study, social support was also negatively
correlated with pre- and post-vaccination titers against the A/Pan-
ama influenza strain yet positively with pre-vaccination antibody
titers against the A/New Caledonia strain [30], making interpreta-
tion of the findings very difficult. However, some of the early
studies of stress and vaccination in students were opportunistic; in
other words they collected stress data from students who opportu-
nistically had already received a prior vaccination. Although more
complex to interpret, given lack of baseline or prior exposure
information, these studies are able to show that psychological stress
does seem to affect the maintenance of antibody titers over time.
For example, one study examined the association between life event
stress and hepatitis B antibody titer in medical students, vaccinated
either in the past 12 months or at least 13 months previously
[31]. Whereas life event exposure was not related to antibody
response in the recently vaccinated cohort, participants in the ear-
lier vaccinated cohort who reported higher life events over the past
year were over twice as likely to show an inadequate antibody titer
as those with lower life event exposure, providing some evidence
that psychosocial stress can have effects on the rate of deterioration
of antibody protection [31]. Similarly, the maintenance of higher
antibody levels after the conjugate meningococcal C vaccination
was associated with lower perceived stress [32].
3.4 Key Stress and One of the most common vaccinations studied in the context of
Vaccination Findings stress and antibody response is the influenza vaccination, particu-
larly in undergraduate student and older caregiver samples. The
influenza vaccination is a commonly utilized vaccine and consists of
three components or strains, usually two A strains and one B strain,
which change each year depending on the key circulating varieties.
A meta-analysis of 13 studies of psychological stress and influenza
vaccination concluded that there is a significant negative relation-
ship between psychological stress and antibody titer following
influenza vaccination [33]. These studies included five caregivers
and eight assessing the impact of stressful life events or perceived
stress. The meta-analysis concluded that psychological stress, how-
ever measured, had a similar negative impact on influenza vaccine
response, but that antibody responses to A/H1N1 and B-influenza
types were more sensitive to the influence of stress [33]. However,
it is difficult at this stage to explain why antibodies against influenza
strains are differentially associated with stress. One possibility is that
strain novelty influences the associations observed [11, 34], with
more novel strains being more susceptible to stress effects.
The impact on certain A-strains and on B-strains is clearly
shown in several studies of students. For example, those reporting
higher stressful life event exposures and/or higher perceived stress
prior to vaccination showed poorer responses to the A-strains of the

vaccine at 5 weeks (around the time of the peak response) and
5 months post-vaccination (indicating the decay in antibody
response over time) [35]. This was replicated for the numbers
and severity of stressful life events prior to vaccination with the
response to the B/Shandong influenza strain at both 5 weeks and
5 months post-vaccination [36]. Similarly, in a study of the effects
of daily stress and feelings of being overwhelmed during the
10 days following vaccination, higher stress ratings were associated
with lower antibody titers to the A/New Caledonia strain at both
1 and 4 months following vaccination [37]. In older adults too, we
observed that the stress of bereavement in the year prior to influ-
enza vaccination was associated with a poorer antibody response to
two of the influenza strains in a community sample of 184 adults
aged 65 and over [38]. Overall negative life event exposure was not
associated with vaccine response in this study, as has similarly been
found for self-reported stress in another study of the influenza
vaccination in older adults, although only one item was used to
assess stress [39]. However, the effect found above for bereavement
suggests that stressful events are related to pervasive immune effects
throughout the life course, although what constitutes life events
stress will vary depending on the age of the sample studied. Taken
together, these studies provide evidence that stressful life events
both preceding and in the period immediately following vaccina-
tion can influence the antibody response. They also show that both
the peak antibody response at around 4 weeks and the decay in
antibody protection over time are susceptible to influence by stress-
ful life events.
Interestingly, in the majority of these studies, where measured,
self-reported or perceived stress has not been associated with the
antibody response to vaccination, whereas the occurrence of stress-
ful life events has. This suggests that actual stressful event occur-
rence is more detrimental than individuals’ perceptions of stress,
which may more closely reflect individual differences in personality
and coping style than a quantitative count of stressful occurrences.
On the whole, the vaccination response in older adults has
mainly been considered in the context of the chronic stress of
caregiving for a spouse with dementia. Studies have shown that
caregivers have poorer antibody responses to vaccination in com-
parison to matched control participants [40–42]. Similarly, care-
givers who exhibited repetitive negative thoughts about their
situation had lower antibody titers following influenza vaccination
[43]. More recently, in a study in Hong Kong, older caregivers had
significantly lower cell-mediated but not humoral (antibody)
responses to the influenza vaccination at 12 weeks compared to
non-caregivers [44]. However, in younger populations, such as
multiple sclerosis spousal caregivers, there was no difference in
antibody response to influenza vaccination between caregivers and
controls [45]. This raises the issue of whether the poor antibody
response observed in older caregivers is, to an extent, a function of
an interaction between chronic stress exposure and
immunosenescence [46].
There is an alternative explanation for the discrepancy in out-
comes among the caregiver vaccination studies. Rather than
immune ageing, perhaps it is the intensity of the stress experienced
that determines whether caregiving becomes an issue for immunity
[45]. Dementia is a disease characterized by much more severe
cognitive and behavioral disturbances than multiple sclerosis
[47–50], and older spousal caregivers of dementia patients have
been found to report greater distress than younger multiple sclero-
sis caregivers [45]. Further, the results of two recent meta-analyses
indicate that caregivers of dementia patients generally experience
greater burden and report more symptoms of depression than those
caring for non-dementia, e.g., cancer, patients, [51, 52]. Thus, it
might be hypothesized that, irrespective of the caregiver’s age,
caring for someone with severe cognitive and behavioral problems
will compromise immunity.
We have been able to test this hypothesis recently using a
caregiving model in younger adults, young parents caring for chil-
dren with developmental disabilities. Dealing with severe cognitive
difficulties and behaviors that are problematic and distressing is the
main challenge of such caring [53–56]. In our own studies of
30 caregivers for a child with a developmental disability (mainly
Autism) versus matched controls, we have demonstrated that care-
givers report high levels of stress, anxiety, depression, child problem
behaviors, and low levels of social support. These caregivers also
exhibited a poorer antibody response to a pneumonia vaccination
than parents caring for typically developing children at both 1 and
6 months post-vaccination [57]. Of the psychological variables
considered, child problem behaviors mediated this effect. In addi-
tion, within the caregivers, parents reporting more child conduct
problems, a component of the child problem behavior measure,
mounted a poorer antibody response at 1 month than parents
reporting less conduct problems [57]. Similarly, these parents
mounted a poorer antibody response to the B/Malaysia strain of
an influenza vaccine at 1 and 6 months post-vaccination, which
again appeared to be mediated by differences in child problem
behaviors [58].
These recent findings in younger caregivers reinforce the
hypothesis that an ageing immune system is not a prerequisite for
a poor response to medical vaccination in caregivers. Nevertheless,
among our parental caregivers, older caregivers tended to have a
poorer antibody response to B/Malaysia at 1 month, suggesting
that we cannot dismiss the hypothesis that chronic stress and
immunosenescence may have synergistic effects [46].
4 Different Vaccine Factors
Vaccination studies also have the advantage of being able to incor-

porate research questions such as the following: “Does it matter
when stressful life events occur relative to when the vaccine is
received?” “What are the effects of prior exposure to the antigen?”
The next section of this chapter addresses some of these issues of
timing.
4.1 Timing of Stress This issue of the timing of stress assessment has been developed in
Measurement studies of various vaccinations including hepatitis B, which is useful
in this context, as the vaccination schedule consists of three inocu-
lations over a 6-month period. The largest of these studies exam-
ined the association between life event stress and final antibody titer
in students, vaccinated either in the past 12 months or at least
13 months previously [31]. Whereas life event exposure was not
related to antibody response in the recently vaccinated cohort,
participants in the earlier vaccinated cohort who reported higher
life events over the past year were over twice as likely to show an
inadequate antibody titer as those with lower life event exposure.
This finding suggests that the immunogenicity, the ability to induce
a strong vaccination response, of hepatitis B vaccination may initi-
ally override the influence of life event stress, although there was
also more power to detect effects in the earlier vaccinated cohort as
more participants exhibited inadequate antibody titers [31]. Never-
theless, this study provides some evidence that psychosocial stress in
the period following vaccination can have effects on the rate of
deterioration of antibody protection [18].
In a study where a low dose of hepatitis B vaccine was adminis-
tered, a higher stress index, comprising life event exposure and
psychological symptoms, measured at 2 months post-vaccination
(thus considering the period post-vaccination) was associated with
a poorer final 6-month antibody response, and the stress index at
6 months also tended to relate negatively to antibody response
[59]. However, as only the final antibody titer was measured, it is
difficult to determine whether, in this instance, stress predomi-
nantly influenced initial formation or maintenance of antibody
levels. Also, the inclusion of psychological symptoms in the com-
posite stress index makes it difficult to ascribe this finding to any
specific aspect of stress [18]. A similar study using the full-dosage
hepatitis B vaccination did not yield any significant stress effects
[60], although it is possible that this was due to the absence of a
2-month assessment of stress, which was the main predictor of
antibody response in the previous study by this group. In a study
measuring perceived stress and anxiety during the vaccination
period, i.e., post-vaccination, these were not associated with the
final antibody response to hepatitis B [40]. Further, life event stress
prior to vaccination and perceived stress at the time of the initial

vaccination were not related to antibody status 5 months following
the initial inoculation in a more recent study [61]. On the whole,
this would suggest either that stress prior to vaccination is less
detrimental to the antibody response than stress post-vaccination
or that it is difficult to observe stress effects early on with the full-
dose hepatitis B vaccination, due to its immunogenicity. Given the
findings with the influenza vaccination and stress, this latter seems
the more likely explanation.
In contrast to the studies of hepatitis B discussed thus far, one
study reported a positive association between perceived life event
stress, depression, and anxiety during the vaccination period and
hepatitis B antibody status 9 months following the initial vaccina-
tion [62]. This anomalous result has been attributed to the rela-
tively low levels of stress experienced by the participants in this
study, suggesting that moderate levels of life change stress experi-
enced during the initial stages of antibody formation may be bene-
ficial to the antibody response, although high levels may be
detrimental [62]. Such an interpretation receives support from
animal research where moderate stress at the time of vaccination
has been associated with an enhanced antibody response (see e.g.,
[63]). This will be discussed further in the section on acute stress
below.
4.2 Primary and Vaccination with an antigen to which the participant has not been
Secondary Exposure to previously exposed induces a primary antibody response whereas
Vaccine Antigens vaccination against more common pathogens such as influenza
induce a secondary immune response. By examining the effect of
stress on both primary and secondary immune responses, we can
begin to determine which aspects of the immune response are most
susceptible to stress-induced modulation.
Hepatitis B vaccination has been used in this context due to the
vaccination schedule and the low likelihood of prior naturalistic
exposure to this pathogen. In an earlier study, individuals reporting
higher mean perceived stress and anxiety over the vaccination
period were less likely to have seroconverted (produced a protective
antibody level) by the time of the second inoculation [40]. How-
ever, an emotional disclosure intervention group did not differ
from controls in antibody levels at the time of the second inocula-
tion [64]. However, psychological stress levels were not measured,
making it difficult to interpret these data. More recently, we have
used hepatitis A as a primary antigen. Students who reported a
higher number and severity of life events had a poorer antibody
response to hepatitis A at the 18-week, but not 4-week, follow-up,
suggesting that stress can impact upon the maintenance of antibody
levels [65]. Early studies using the vaccination model used novel
nonpathogenic antigens to examine the antigen-specific antibody
response. Keyhole limpet hemocyanin (KLH), a protein, has been
used in this context; the KLH-specific IgG antibody response was

lower at 8 weeks, but not 3 weeks, post-vaccination in participants
reporting fewer positive life events prior to vaccination [66].
The consensus of this evidence suggests that stress can influ-
ence the primary antibody response, particularly the maintenance
of responses to novel antigens. It also supports the idea that life
event stress effects are more likely to be evident with novel vaccine
types [36]. As discussed above, the secondary antibody response to
hepatitis B vaccination has produced mixed results, but there
appears to be stronger evidence for a negative effect of psychologi-
cal stress on the secondary response to this antigen [18, 67], in line
with the findings for the influenza vaccine.
4.3 Thymus- A further advantage to the vaccination model is that there are
Dependent Versus different types of vaccination, which can be used to help elucidate
Thymus-Independent which cells involved in the vaccination response are influenced by
Vaccines psychological factors. Most vaccinations, which consist of inacti-
vated or dead viruses like influenza, induce a thymus-dependent
antibody response, as described above. A few vaccinations, how-
ever, protect against bacterial infections or toxins, like meningo-
coccal A or tetanus, respectively, which do not require T-cell help.
There are also conjugate vaccines, in which substances that elicit a
T-cell response are conjugated to a thymus-independent pathogen,
such as a protein, in order to boost the efficiency of the antibody
response against the thymus-independent pathogen. If psychologi-
cal factors are consistently associated with the response to thymus-
dependent and conjugate vaccinations but not with thymus-inde-
pendent response, this would imply that it is T cells that are partic-
ularly liable to psychological influence.
Indeed, there is evidence to suggest that stress may exert its
effects mainly on T cells; we showed that higher frequency and
intensity of stressful life events were associated with a poorer
response to influenza and meningococcal C (following previous
conjugate meningococcal C vaccination), but not to thymus-
independent meningococcal A [36]. Similarly, no association was
found between stress and antibody response to a thymus-indepen-
dent pneumonia vaccination in preschool children [68]. However,
as older caregivers have been reported to show poorer maintenance
of antibody levels over time following pneumonia vaccination than
controls [41], it is possible that other factors such as age and
severity of stress may interact to impair antibody-mediated immu-
nity more generally than just the T-cell response.
It should be noted that in the study of caregivers and pneumo-
nia vaccination, perceived stress did not differ between the caregiver
and controls, but there was a significant difference in social support.
This might suggest that thymus-dependent vaccinations are sus-
ceptible to the effects of stressful life events, but that thymus-
independent vaccinations are more vulnerable to other psychosocial
factors such as lower social support. There is some evidence for this
suggestion. In our own laboratory, we found that social support,
but not life event stress, was positively associated with the response
to a thymus-independent pneumococcal vaccine in young healthy
students [65, 69].
The comparison between thymus-dependent and -independent
vaccination responses suggests that both types of response are suscep-
tible to psychosocial influence, but that there are key variables which
influence whether an effect on vaccination response is observed.
These include the type of psychosocial factor studied (i.e., stress
versus social support), and the age of the population sampled.
4.4 Acute Versus Following on from the discussion above in Subheading 4.1 regard-
Chronic Stress ing when stress is measured, such that moderate or less severe stress
at the time of vaccination might actually have a beneficial effect, it
has been suggested in recent years that acute (minutes or hours)
stress may be immune enhancing when experienced close to the
immune challenge. Such immune enhancement by acute stress
would be an adaptive mechanism, and might be regarded as an
integral component of the fight or flight response, and circum-
stances that elicit such a response are likely to also involve exposure
to antigens and, therefore, a robust immune response would be
adaptive for survival [63]. Our laboratory examined the effect of
acute psychological stress on antibody response to vaccination in
humans. Participants completed a 45-min time pressured, socially
evaluated mental arithmetic task, or a resting control period, imme-
diately prior to influenza vaccination. An enhancement of the anti-
body response to one of the influenza viral strains was found in
women in the psychological stress group compared to control
[70]. Similarly, in men, the antibody response to a meningococcal
A vaccination was enhanced by acute psychological stress
[71]. That these effects emerged for only one gender or the other
in these studies might be explained by examining the antibody
responses for each gender. In each case, stress was associated with
the antibody response in those with the poorest increase in anti-
bodies in response to vaccination: women for the influenza A/Pan-
ama strain and men for the meningococcal A vaccine. This latter
study [71] also provides further evidence that both thymus-
dependent and -independent vaccinations are responsive to the
impact of stress, as discussed above. Further, although not a psy-
chological stressor per se, acute eccentric exercise (arm contrac-
tions) was also shown to enhance the antibody response to
influenza vaccination in women, and the cell-mediated IFNγ pro-
duction response to stimulation with the influenza vaccine in men
[72]. This and similar studies have also lent weight to the conten-
tion that effects of behavioral factors on vaccination responses are
most likely to be observed in groups with the poorest antibody
response, or to vaccine strains which are not very immunogenic,
i.e., they engender lower antibody titers [72, 73].
4.5 Timing of As well as the timing and duration of stress measurements asso-
Vaccination ciated with the vaccine response, other behavioral factors have been
found to impact upon the antibody response. One such factor is
that of the time of day of vaccination. In our study of the effects of
psychological stress on vaccination responses in the 184 older
adults [38], we observed that the time of day of vaccine administra-
tion significantly influenced antibody titer [74]. Men responded
better in the morning than the afternoon; 41% of men showed a
twofold response when vaccinated in the morning versus 24% of
men vaccinated in the afternoon. This effect was independent of
current illnesses, medication, vaccination history, and our reported
findings of the effects of bereavement and marital quality. Women
tended to show the reverse pattern. We also observed the same
pattern in a study of younger adults’ antibody response to the
hepatitis A vaccination [74]. However, these studies were not
fully randomized, and there was little opportunity to examine the
biological mechanisms, such as cytokine and stress hormone levels.
Consequently, there was a clear and pressing need for a randomized
controlled trial of the impact of time of influenza vaccination on
antibody response and vaccine efficacy in older adults in a National
Health Service (NHS) setting. We were able to conduct such a
cluster-randomized trial within the National Health Service in
Birmingham, UK, and confirmed that a simple manipulation of
the time of vaccination can improve the immune response against
influenza in older adults. Two hundred and seventy-six participants
were randomized to have the annual influenza vaccination at one of
24 general practice surgeries in either the morning (9–11 a.m.) or
the afternoon (3–5 p.m.). This trial provided some evidence that
morning vaccination produced higher antibody levels, at least for
the H1N1 A-strain with a trend for the same effect for the B
influenza strain [75, 76]. Consequently, although this requires a
larger scale trial for confirmation, we believe that morning vaccina-
tion could be employed as an easy-to-adopt intervention within the
health services, at little or no added cost. The potential benefits
would be a decreased incidence of influenza infection and
influenza-related mortality in older adults, although a multicenter
very large trial following up on verified influenza incidence and
health outcomes would be essential to fully prove this.
5 Other Psychosocial Factors and the Vaccination Model
5.1 Social Support The support of friends and loved ones is an important determinant
of immune health, and is relatively easily measured in vaccination
studies via validated questionnaires. Studies have assessed both
functional social support, a measure of the quality and availability
of social resources a person has, and structural social support, the
number of friends a person can call on, in the context of
vaccination. First, students reporting greater social support demon-

strated a stronger combined immune response to the third inocu-
lation of the three-dose hepatitis B vaccination [40]. Second,
loneliness and smaller social network size were associated with a
poorer antibody response to the A/New Caledonian strain of the
influenza vaccination in college students [77]. Third, students with
greater functional social support showed higher titers to the
A/Panama influenza strain at both 5 weeks and 5 months following
vaccination [36]. In older nursing-home residents, social support
was also negatively correlated with pre- and post-vaccination titers
against the A/Panama influenza strain yet positively with
pre-vaccination antibody titers against the A/New Caledonia strain
[30], a finding which even the authors were unable to explain.
Along with the caregiver study discussed above, these studies gen-
erally show that a lack of social support has a strong negative impact
on antibody levels following vaccination.
Marriage is also a source of social support. In our own work,
older adults who were married, and particularly those who were
happily married, showed a better antibody response to the influenza
vaccination than those who were unmarried or less happily married
[38]. However, more general functional social support and social
network size were not associated with antibody response in this
older population [38]. Further, for children, the relationship with
their parents is the key source of support, and negative parent-child
interactions such as conflict have been associated with a less robust
antibody response to the meningococcal C vaccination over
6 months [78]. These findings perhaps lend weight to the sugges-
tion that the age of the population studied influences which psy-
chosocial factors are important for the vaccination response.
5.2 Personality Personality factors, although often examined in the context of

health outcomes, again using validated questionnaires (see, e.g.,
[79]), have scarcely been investigated relative to the vaccination
response. First, among a group of 12-year-old girls, those charac-
terized by higher internalizing scores and lower self-esteem at
baseline exhibited lower antibody titers following rubella vaccina-
tion [80]. A similar concept, neuroticism, was negatively associated
with both the peak antibody response to the A/Panama strain of an
influenza vaccination and the maintenance of antibody titers to this
strain in students [81]. Among female graduate students, trait-
negative affect/mood was negatively associated with the antibody
response to the second hepatitis B injection [61]. Further, inde-
pendently of negative affect, trait-positive affect was associated with
a better antibody response following a second hepatitis B vaccina-
tion in graduate students [82]. Thus, both negative and positive
traits appear to be able to influence this aspect of immune function
and disease protection. However, in exercising and sedentary
elderly individuals, dispositional optimism was not found to be
associated with antibody titers following influenza vaccination

[83]. Inconsistencies in these results could be attributable to the
different measures of personality studied, or the different ages of
the populations used, which will now be discussed in more detail.
6 Future Directions: Mechanisms and Interventions
The studies reviewed above outline the different methods of exam-

ining associations between psychological factors and antibody
response to vaccination. These studies show the strong associations
between psychological stress, other psychosocial factors, and
immune response to vaccination, such that stressful psychological
circumstances are associated with poorer antibody responses, while
positive factors such as social support relate to a better immune
response to vaccination. Taken together, these findings suggest two
main directions for future research. First, despite the range of
vaccinations used in such studies, as yet little is known about the
exact mechanisms by which stress and other factors can influence
antibody responses to vaccination. Research incorporating a range
of measurements, such as stress hormones, immune system mes-
sengers (cytokines), and function of key cells in the vaccination
response, such as antigen-presenting cells, would be necessary to
further our understanding regarding exactly how stress gets inside
the body to affect this clinically relevant immune outcome. Second,
the clinical implications, in terms of susceptibility to disease, arising
from a better understanding of the relationships between psycho-
logical factors and vaccination response are important, particularly
in the context of older adults who already display poor vaccination
responses. Psychological interventions to improve vaccination
response in these populations could include techniques such as
stress management, relaxation, cognitive behavioral therapy, and
emotional disclosure.
Regarding such interventions, one study showed an improve-
ment in the ability of older caregivers for a spouse with dementia to
mount a fourfold increase in antibody titer following influenza
vaccination relative to matched controls, although the mechanisms
of effect were unclear and the intervention group was not randomly
sampled [84]. Similarly, participants taking part in a written emo-
tional disclosure intervention, where they wrote about their emo-
tions about a previously undisclosed stressful event, showed
significantly higher antibody titers at 4 and 6 months following
vaccination with hepatitis B compared to a control
non-intervention group [64]. A different clinical application of
the vaccination model has arisen from the positive immune effects
demonstrated in response to acute stress and exercise, as discussed
above [70]. These preliminary findings suggest that the develop-
ment of such a behavioral challenge that could be applied in general
practitioner settings could be a way forward for improving the

vaccination response. This would be particularly important for
groups at risk of infectious disease such as older adults, the
bereaved, and caregivers. At this stage, more work is required to
establish exactly what types of intervention in which age groups and
are likely to be the most beneficial for psychological, and hence
immunological, health. Behavioral interventions, such as the time
of day of vaccination, may also be important in this context.
7 Conclusion
In conclusion, vaccination has had a substantial impact on public

health, although not everyone mounts a satisfactory and protective
antibody response to vaccination. This increasingly appears to be
the case with progressing age. Studying antibody responses to
vaccination is now contributing to the understanding of how psy-
chosocial exposures can influence immunity and, consequently,
resistance to disease. The current challenges are to build upon the
methodology that has been developed through these studies to
unravel the underlying mechanisms and to develop and apply feasi-
ble behavioral interventions to boost the response to vaccination
and, thus, optimize our resistance against infectious disease.
References
1. McKinnon W, Weisse CS, Reynolds CP et al 7. Jemmott JB 3rd, Borysenko JZ, Borysenko M

(1989) Chronic stress, leukocyte subpopula- et al (1983) Academic stress, power motiva-
tions, and humoral response to latent viruses. tion, and decrease in secretion rate of salivary
Health Psychol 8:389–402 secretory immunoglobulin A. Lancet
2. Schaeffer MA, Baum A, Reynolds CF et al 1:1400–1402
(1985) Immune status as a function of chronic 8. Jemmott JB 3rd, Magloire K (1988) Academic
stress at Three Mile Island. Psychosom Med stress, social support, and secretory immuno-
47:85 globulin A. J Pers Soc Psychol 55:803–810
3. Futterman AD, Wellisch DK, Zighelboim J 9. McClelland DC, Alexander C, Marks E (1982)
et al (1996) Psychological and immunological The need for power, stress, immune function,
reactions of family members to patients under- and illness among male prisoners. J Abnorm
going bone marrow transplantation. Psycho- Psychol 91:61–70
som Med 58:472–480 10. Gallagher S, Phillips AC, Evans P et al (2008)
4. Kiecolt-Glaser JK, Glaser R, Shuttleworth EC Caregiving is associated with low secretion
et al (1987) Chronic stress and immunity in rates of immunoglobulin A in saliva. Brain
family caregivers of Alzheimer’s disease victims. Behav Immun 22:565–572. https://doi.org/
Psychosom Med 49:523–535 10.1016/j.bbi.2007.11.007
5. De Gucht V, Fischler B, Demanet C (1999) 11. Vedhara K, Fox JD, Wang ECY (1999) The
Immune dysfunction associated with chronic measurement of stress-related immune dys-
professional stress in nurses. Psychiatry Res function in psychoneuroimmunology. Neu-
85:105–111 rosci Biobehav Rev 23:699–715
6. Deinzer R, Kleineidam C, Stiller-Winkler R 12. Bartrop RW, Luckhurst E, Lazarus L et al
et al (2000) Prolonged reduction of salivary (1977) Depressed lymphocyte function after
immunoglobulin A (sIgA) after a major aca- bereavement. Lancet 1:834–836
demic exam. Int J Psychophysiol 37:219–232 13. Esterling BA, Kiecolt-Glaser JK, Glaser R
(1996) Psychosocial modulation of cytokine-
induced natural killer cell activity in older 28. Macleod J, Davey Smith G, Heslop P et al
adults. Psychosom Med 58:264–272 (2002) Psychological stress and cardiovascular
14. Irwin M, Patterson T, Smith TL et al (1990) disease: empirical demonstration of bias in a
Reduction of immune function in life stress and prospective observational study of Scottish
depression. Biol Psychiatry 27:22–30 men. BMJ 324:1247–1251
15. Schleifer SJ, Keller SE, Camerino M et al 29. Schulz R, Beach SR (1999) Caregiving as a risk
(1983) Suppression of lymphocyte stimulation factor for mortality: the Caregiver Health
following bereavement. JAMA 250:374–377 Effects Study. JAMA 282:2215–2219
16. Arnetz BB, Wasserman J, Petrini B et al (1987) 30. Moynihan JA, Larson MR, Treanor J et al
Immune function in unemployed women. Psy- (2004) Psychosocial factors and the response
chosom Med 49:3–12 to influenza vaccination in older adults. Psy-
17. Khanfer R, Lord JM, Phillips AC (2011) Neu- chosom Med 66:950–953
trophil function and cortisol:DHEAS ratio in 31. Burns VE, Carroll D, Ring C et al (2002)
bereaved older adults. Brain Behav Immun 25 Stress, coping and hepatitis B antibody status.
(6):1182–1186. https://doi.org/10.1016/j. Psychosom Med 64:287–293
bbi.2011.03.008 32. Burns VE, Drayson M, Ring C et al (2002)
18. Burns VE, Carroll D, Ring C et al (2003) Perceived stress and psychological well-being
Antibody response to vaccination and psycho- are associated with antibody status following
social stress in humans: relationship and meningitis C conjugate vaccination. Psycho-
mechanisms. Vaccine 21:2523–2534 som Med 64:963–970
19. Allsup S, Haycox A, Regan M et al (2004) Is 33. Pedersen AF, Zachariae R, Bovbjerg DH
influenza vaccination cost effective for healthy (2009) Psychological stress and antibody
people between ages 65 and 74 years? A rando- response to influenza vaccination: a meta-
mised controlled trial. Vaccine 23:639–645 analysis. Brain Behav Immun 23:427–433.
20. Patriarca PA (1994) Editorial: A randomized https://doi.org/10.1016/j.bbi.2009.01.004
controlled trial of influenza vaccine in the 34. Gulati U, Keitel WA, Air GM (2007) Increased
elderly: Scientific scrutiny and ethical responsi- antibodies against unfolded viral antigens in
bility. JAMA 272:1700–1701 the elderly after influenza vaccination. Influ-
21. Goodwin K, Viboud C, Simonsen L (2006) enza Other Respi Viruses 1:147–156. https://
Antibody response to influenza vaccination in doi.org/10.1111/j.1750-2659.2007.00017.
the elderly: a quantitative review. Vaccine x
24:1159–1169. https://doi.org/10.1016/j. 35. Burns VE, Carroll D, Drayson M et al (2003)
vaccine.2005.08.105 Life events, perceived stress and antibody
22. Nichol KL, Margolis KL, Wuorenma J et al response to influenza vaccination in young
(1994) The efficacy and cost effectiveness of healthy adults. J Psychosom Res 55:569–572
vaccination against influenza among elderly 36. Phillips AC, Burns VE, Carroll D et al (2005)
persons living in the community. N Engl J The association between life events, social sup-
Med 331:778–784 port and antibody status following thymus-
23. Simonsen L, Reichert TA, Viboud C et al dependent and thymus-independent vaccina-
(2005) Impact of influenza vaccination on sea- tions in healthy young adults. Brain Behav
sonal mortality in the US elderly population. Immun 19:325–333
Arch Intern Med 165:265–272 37. Miller GE, Cohen S, Pressman S et al (2004)
24. Brown GW, Harris TO (1989) Social origins of Psychological stress and antibody response to
depression: a study of psychiatric disorder in influenza vaccination: when is the critical
women. Routledge, London period for stress, and how does it get inside
the body? Psychosom Med 66:215–223
25. Cohen S, Tyrell DAJ, Smith AP (1991) Psy-
chological stress and susceptibility to the com- 38. Phillips AC, Carroll D, Bums VE et al (2006)
mon cold. N Engl J Med 325:606–612 Bereavement and marriage are associated with
antibody response to influenza vaccination in
26. Rosengren A, Orth-Gomer K, Wedel H et al the elderly. Brain Behav Immun 20:279–289.
(1993) Stressful life events, social support, and https://doi.org/10.1016/j.bbi.2005.08.003
mortality in men born in 1933. BMJ
307:1102–1105 39. Talbot HK, Coleman LA, Zhu Y et al (2015)
Factors associated with maintenance of anti-
27. Cohen S, Kamarck T, Mermelstein R (1983) A body responses to influenza vaccine in older,
global measure of perceived stress. J Health Soc community-dwelling adults. BMC Infect Dis
Behav 24:385–396. https://doi.org/10. 15:195. https://doi.org/10.1186/s12879-
2307/2136404 015-0926-8
40. Glaser R, Kiecolt-Glaser JK, Bonneau RH et al 53. Floyd FJ, Gallagher EM (1997) Parental stress,
(1992) Stress-induced modulation of the care demands, and use of support services for
immune response to recombinant hepatitis B school-age children with disabilities and behav-
vaccine. Psychosom Med 54:22–29 ior problems. Fam Relat 46:359–371
41. Glaser R, Sheridan JF, Malarkey W et al (2000) 54. Hastings RP, Daley D, Burns C et al (2006)
Chronic stress modulates the immune response Maternal distress and expressed emotion:
to a pneumococcal pneumonia vaccine. Psy- cross-sectional and longitudinal relationships
chosom Med 62:804–807 with behavior problems of children with intel-
42. Vedhara K, Cox NK, Wilcock GK et al (1999) lectual disabilities. Am J Mental Retard
Chronic stress in elderly carers of dementia 111:48–61
patients and antibody response to influenza 55. Higgins DJ, Bailey SR, Pearce JC (2005) Fac-
vaccination. Lancet 353:627–631 tors associated with functioning style and cop-
43. Segerstrom SC, Schipper LJ, Greenberg RN ing strategies of families with a child with an
(2008) Caregiving, repetitive thought, and autism spectrum disorder. Autism 9:125–137
immune response to vaccination in older 56. Maes B, Broekman TG, Dosen A et al (2003)
adults. Brain Behav Immun 22:744–752. Caregiving burden of families looking after
https://doi.org/10.1016/j.bbi.2007.11.004 persons with intellectual disability and beha-
44. Wong SY, Wong CK, Chan FW et al (2013) vioural or psychiatric problems. J Intellect Dis-
Chronic psychosocial stress: does it modulate abil Res 47:447–455
immunity to the influenza vaccine in Hong 57. Gallagher S, Phillips AC, Drayson MT et al
Kong Chinese elderly caregivers? Age (Dordr) (2009) Parental caregivers of children with
35:1479–1493. https://doi.org/10.1007/ developmental disabilities mount a poor anti-
s11357-012-9449-z body response to pneumococcal vaccination.
45. Vedhara K, McDermott MP, Evans TG et al Brain Behav Immun 23:338–346. https://
(2002) Chronic stress in non-elderly care- doi.org/10.1016/j.bbi.2008.05.006
givers: psychological, endocrine and immune 58. Gallagher S, Phillips AC, Drayson MT et al
implications. J Psychosom Res 53:1153–1161 (2009) Caregiving for children with develop-
46. Graham JE, Christian LM, Kiecolt-Glaser JK mental disabilities is associated with a poor
(2006) Stress, age, and immune function: antibody response to influenza vaccination.
toward a lifespan approach. J Behav Med Psychosom Med 71:341–344. https://doi.
29:389–400 org/10.1097/PSY.0b013e31819d1910
47. Keegan BM, Noseworthy JH (2002) Multiple 59. Jabaaij L, Grosheide PM, Heijtink RA et al
sclerosis. Annu Rev Med 53:285–302. https:// (1993) Influence of perceived psychological
doi.org/10.1146/annurev.med.53.082901. stress and distress on antibody response to
10390953/1/285 low dose rDNA hepatitis B vaccine. J Psycho-
48. Poser CM, Paty DW, Scheinberg L et al (1983) som Res 37:361–369
New diagnostic criteria for multiple sclerosis: 60. Jabaaij L, Van Hattum J, Vingerhoets AJJM
guidelines for research protocols. Ann Neurol et al (1996) Modulation of immune response
13:227–231. https://doi.org/10.1002/ana. to rDNA hepatitis B vaccination by psycholog-
410130302 ical stress. J Psychosom Res 41:129–137
49. Gregory CA, Hodges JR (1996) Clinical fea- 61. Marsland AL, Cohen S, Rabin BS et al (2001)
tures of frontal lobe dementia in comparison to Associations between stress, trait negative
Alzheimer’s disease. J Neural Transm Suppl affect, acute immune reactivity, and antibody
47:103–123 response to hepatitis B injection in healthy
50. Neary D, Snowden JS, Gustafson L et al young adults. Health Psychol 20:4–11
(1998) Frontotemporal lobar degeneration: a 62. Petry J, Weems LB, Livingstone JNI (1991)
consensus on clinical diagnostic criteria. Neu- Relationship of stress, distress, and the immu-
rology 51:1546–1554 nologic response to a recombinant hepatitis B
51. Pinquart M, Sorensen S (2003) Associations of vaccine. J Fam Pract 32:481–486
stressors and uplifts of caregiving with caregiver 63. Dhabhar FS, McEwen BS (1996) Stress-
burden and depressive mood: a meta-analysis. J induced enhancement of antigen-specific cell-
Gerontol B Psychol Sci Soc Sci 58:P112–P128 mediated immunity. J Immunol
52. Pinquart M, Sorensen S (2003) Differences 156:2608–2615
between caregivers and noncaregivers in psy- 64. Petrie KJ, Booth RJ, Pennebaker JW et al
chological health and physical health: a meta- (1995) Disclosure of trauma and immune
analysis. Psychol Aging 18:250–267. https:// response to a hepatitis B vaccination program.
doi.org/10.1037/0882-7974.18.2.250 J Consult Clin Psychol 63:787–792
65. Gallagher S, Phillips AC, Ferraro AJ et al https://doi.org/10.1016/j.vaccine.2016.04.

(2008) Psychosocial factors are associated 032
with the antibody response to both thymus- 76. Long JE, Drayson MT, Taylor AE et al (2016)
dependent and thymus-independent vaccines. Corrigendum to ‘Morning vaccination
Brain Behav Immun 22:456–460 enhances antibody response over afternoon
66. Snyder BK, Roghmann KJ, Sigal LH (1990) vaccination: a cluster-randomised trial’ [Vac-
Effect of stress and other biopsychosocial fac- cine 34 (2016) 2679-2685]. Vaccine
tors on primary antibody response. J Adolesc 34:4842. https://doi.org/10.1016/j.vaccine.
Health Care 11:472–479 2016.08.031
67. Cohen S, Miller GE, Rabin BS (2001) Psycho- 77. Pressman SD, Cohen S, Miller GE et al (2005)
logical stress and antibody response to immu- Loneliness, social network size, and immune
nization: a critical review of the human response to influenza vaccination in college
literature. Psychosom Med 63:7–18 freshmen. Health Psychol 24:297–306
68. Boyce WT, Adams S, Tschann JM et al (1995) 78. O’Connor TG, Wang H, Moynihan JA et al
Adrenocortical and behavioral predictors of (2015) Observed parent-child relationship
immune response to starting school. Pediatr quality predicts antibody response to vaccina-
Res 38:1009–1017 tion in children. Brain Behav Immun
69. Gallagher S, Phillips AC, Ferraro AJ et al 48:265–273. https://doi.org/10.1016/j.bbi.
(2008) Social support is positively associated 2015.04.002
with the immunoglobulin M response to vacci- 79. Smith TW, Glaser K, Ruiz JM et al (2004)
nation with pneumococcal polysaccharides. Hostility, anger, aggressiveness, and coronary
Biol Psychol 78:211–215. https://doi.org/ heart disease: an interpersonal perspective on
10.1016/j.biopsycho.2008.01.001 personality, emotion, and health. J Pers
70. Edwards KM, Burns VE, Reynolds T et al 72:1217–1270
(2006) Acute stress exposure prior to influenza 80. Morag M, Morag A, Reichenberg MA et al
vaccination enhances antibody response in (1999) Psychological variables as predictors of
women. Brain Behav Immun 20:159–168 rubella antibody titers and fatigue - a prospec-
71. Edwards KM, Burns VE, Adkins AE et al tive double blind study. J Psychiatr Res
(2008) Meningococcal A vaccination response 33:389–395
is enhanced by acute stress in men. Psychosom 81. Phillips AC, Carroll D, Burns VE et al (2005)
Med 70:147–151 Neuroticism, cortisol reactivity, and antibody
72. Edwards KM, Burns VE, Allen LM et al (2007) response to vaccination. Psychophysiology
Eccentric exercise as an adjuvant to influenza 42:232–238
vaccination in humans. Brain Behav Immun 82. Marsland AL, Cohen S, Rabin BS et al (2006)
21:209–217 Trait positive affect and antibody response to
73. Edwards KM, Campbell JP, Ring C et al (2010) hepatitis B vaccination. Brain Behav Immun
Exercise intensity does not influence the effi- 20:261–269. https://doi.org/10.1016/j.bbi.
cacy of eccentric exercise as a behavioural adju- 2005.08.009
vant to vaccination. Brain Behav Immun 83. Kohut ML, Cooper MM, Nickolaus MS et al
24:623–630. https://doi.org/10.1016/j.bbi. (2002) Exercise and psychosocial factors mod-
2010.01.009 ulate immunity to influenza vaccine in elderly
74. Phillips AC, Gallagher S, Carroll D et al (2008) individuals. J Gerontol 57:557–562
Morning vaccine administration is associated 84. Vedhara K, Bennett PD, Clark S et al (2003)
with an enhanced response to vaccination in Enhancement of antibody responses to influ-
men. Psychophysiology 45:663–666 enza vaccination in the elderly following a
75. Long JE, Drayson MT, Taylor AE et al (2016) cognitive-behavioural stress management
Morning vaccination enhances antibody intervention. Psychother Psychosom
response over afternoon vaccination: a cluster- 72:245–252
randomised trial. Vaccine 34:2679–2685.
Chapter 17
Measuring Vaccine Responses in the Multiplex Era

Kieran Ayling, Kavita Vedhara, and Lucy Fairclough
Abstract
Vaccine studies in psychoneuroimmunology (PNI) provide an insight into biopsychosocial interactions and
their role in infectious diseases. How to measure vaccine responses is therefore of critical importance for
PNI researchers. In this chapter, traditional and modern immunoassays for the assessment of vaccine
responses are discussed, highlighting how modern multiplex techniques provide researchers with greater
capacity and opportunity for novel research relating to vaccine outcomes.
Key words Vaccination, ELISA, Immunoassay, Multiplex, Microarray
1 Introduction
Studies exploring the determinants of vaccine responses provide

psychoneuroimmunology (PNI) researchers with a window into
how psychological and behavioral factors can impact an individual’s
susceptibility and response to real-world infections [1–4]. Vaccine
administration causes an immune response, which, broadly
speaking, mirrors natural infection exposure. In both vaccination
and natural exposure, antigens enter the body and are recognized as
foreign by the immune system, ultimately resulting in the forma-
tion of antigen-specific antibodies that can protect the host from
infection. Vaccination is therefore a controlled and biologically
relevant model in which the effects of psychological and behavioral
factors on naturally acquired infection can be examined without the
associated risks [5, 6]. Beyond this, vaccination responses, in their
own right, are of significant clinical interest, being frequently sub-
optimal in certain populations such as the elderly [7]. Indeed, there
is increasing research interest concerning psychobehavioral inter-
ventions as a method of enhancing suboptimal vaccination
responses [8–10].
Given the prominence of vaccine studies in PNI, the issue of
how to best measure vaccination responses is of critical importance.
This chapter describes and evaluates several methods of measuring
327
328 Kieran Ayling et al.
vaccine responses, comparing traditional immunoassays with more

modern techniques that allow PNI researchers to examine vaccine
responses with greater efficiency and higher throughput.
2 Immunological Outcome Measures of Vaccine Response
By far the most common immune parameter used to assess vaccina-

tion responses is antigen-specific antibodies (i.e., post-vaccination
levels of antibodies which bind to the antigens present in the
vaccine). Antigen-specific antibodies are the most widely accepted
immune correlate of vaccine-induced protection and vaccines are
typically developed and evaluated on their ability to induce these
antibodies [11]. Antibodies play a critical role in the protection of a
host from infectious diseases. Their primary role in the immune
system is to combat pathogens by:
1. Neutralization—binding directing to invading pathogens, pre-
venting their ability to invade cells or replicate.
2. Opsonization—chemically tagging pathogens and thus pro-
moting phagocytosis (ingestion and destruction) of the patho-
gen by macrophages and clearance of the infection.
3. Complement activation—triggering so-called complement pro-
teins that results in cell lysis.
4. Antibody-dependent cell-mediated cytotoxicity—binding to
pathogens allowing the triggering of effector cell mechanisms
(e.g., in NK cells, eosinophils) to destroy the pathogen.
The primary antibody focus of vaccine research is a specific
antibody isotype called IgG, which represents the primary mecha-
nism for antibody-mediated protection from infectious diseases in
humans and makes up approximately 70–75% of the antibody pool
found in serum. IgG can be further split into four subclasses (IgG1-
4), of which IgG1 is the most abundant. IgA and IgM responses
also play prominent roles in vaccine-induced immunity and there-
fore are occasionally used as alternative correlates of vaccine-
induced protection (IgA is important for pathogen clearance on
mucosal surfaces and IgM is the first antibody produced after
exposure).
While antigen-specific antibody levels are by far the most com-
monly examined immune outcome in the context of vaccine
responses it is worth noting here that several cellular immune
markers have also been of interest to PNI researchers. For example,
granzyme B has been suggested as a better correlate of protection in
older adults receiving influenza vaccination [12, 13] and cytokine
profiles are increasingly frequently examined in combination with
antibody levels to better understand mechanisms by which psycho-
behavioral factors exert their effects on vaccine responses [3, 14].
Measuring Vaccine Responses in the Multiplex Era 329
A discussion of these goes beyond the scope of this chapter, which

will primarily focus on methods by which antigen-specific antibo-
dies can be measured in sera.
3 Traditional Assays for Measuring Vaccine Responses
3.1 Enzyme-Linked ELISA is an extremely common laboratory method used to detect a

Immunosorbent Assay substance (typically an antigen or antibody) in a liquid sample.
(ELISA) Many variations of ELISA exist (e.g., direct and competitive
ELISA), but to assess vaccine responses a variation known as a
sandwich ELISA (so called because the detected substance is “sand-
wiched” between two other substances) is often used to detect
antigen-specific antibodies in a given sample. In this context,
ELISA can be used to assess levels of vaccine-specific IgG antibody
levels (or many other isotypes or immune correlates) and involves:
1. Coating of wells on reactive microtiter plates with the antigens
present in the vaccine.
2. Incubation with the human serum samples (see Fig. 1): Anti-
bodies specific to the antigens then bind to the antigen on the
coated wells, with other non-binding sera contents
washed away.
3. A detection antibody is then added, for example an antibody to
detect human IgG, which binds to the antibodies from the
serum sample.
Detection
Antibody
Serum
Antibody IgG
Antigen Antibody
Antigen “A” coated well Antigen “B” coated well Antibody coated control
well
Fig. 1 Schematic overview of an ELISA. Wells are coated on a 96-well plate with antigens of interest or a serial
dilution of human IgG (to generate a standard curve). After overnight incubation, the plate is washed and then
blocked, before addition of serum samples. After further incubation, the plate is again washed and a detection
antibody is added, followed by addition of substrate. The plate is read on a plate reader to obtain optical
density values for both the standard curve and the specific antibody responses
4. Finally, an enzymatic substrate solution is added that reacts

with the detection antibody resulting in a measurable color
change that can be read by an optical scanner, with increasing
optical absorbance indicating higher quantities of antibody.
Samples with higher quantities of antibody are indicative of a
participant with greater clinical protection.
Alongside the wells containing serum samples, a calibration
curve (also known as a standard curve) is also included on the
plate made up of a serially diluted sample with a known concentra-
tion. Typically, this would be a commercially purchased human IgG
“standard.” The optical absorbance of the known sample at differ-
ent dilutions can then be plotted graphically, against which other
samples can be compared to accurately quantify the level of anti-
body in all the other samples. This process is known as
interpolation.
3.1.1 Materials 1. 96-Well or 384-well reactive ELISA plates.

2. Purified vaccine antigens (see Note 1).
3. Biotinylated anti-human IgG (or alternative for other
isotypes).
4. Human IgG standard (or alternative for other isotypes).
5. Carbonate-bicarbonate buffer tablets.
6. Phosphate-buffered saline (PBS).
7. Washing solution: 0.05% Tween-20 in PBS.
8. Blocking solution: 3% Bovine serum albumin in PBS.
9. Streptavidin-HRP.
10. Tetramethylbenzidine substrate solution.
11. Sulfuric acid, 1 N.
3.1.2 Advantages ELISAs are relatively simple to perform, low cost, and the only
and Limitations of ELISA specialist equipment required is the optical scanner that is common
to most laboratories. The assay produces a continuous measure of
antibody levels, through colorimetric change, and can be easily
adapted to measure any antibody isotype (e.g., IgA, IgM). Higher
antibody levels are indicative of greater clinical protection,
although there are no accepted thresholds at which antibody levels
as measured by ELISA are sufficient to be considered protective.
Further, in ELISA, absorbance is measured objectively by an optical
scanner and requires fewer consumables per sample than other
traditional assays (such as the HAI assay discussed below), as titra-
tion is not performed.
There are, however, multiple limitations of ELISA. To begin,
ELISA relies on colorimetric change, which has a comparatively
small dynamic range (values at which there is a linear relationship
between absorbance and antibody quantity) compared to more

modern fluorescence-based assays. This means that serum samples
often require considerable dilution for the majority to fall within
ELISA dynamic ranges. Even with such dilution, samples with
particularly high or low antibody levels may fall outside this
dynamic range and therefore require reanalysis at a different dilu-
tion, potentially exaggerating differences [15]. Further, while
ELISA requires less consumables than some assays—in traditional
ELISA assays (using a 96-well plate) reagent, sera, and labor
requirements are still relatively high as only one antibody type can
be measured per well. This limits the suitability of ELISA for where
many samples require processing. Miniaturized ELISA assays, using
a 384-well plate, reduce these requirements considerably, although
these tend to require appropriately programmed liquid-handling
robots to perform the necessary steps of the ELISA protocol
(see Note 2).
3.2 Hemagglu- The HAI assay is a quasi-functional assay primarily used in the
tination Inhibition measurement of antibody response to influenza vaccination. It
Assay (HAI) relies on the characteristic property of red blood cells to form a
lattice-like suspension structure in the presence of the influenza
virus [16]. This process, known as hemagglutination, occurs
because receptors on the surface of red blood cells bind to hemag-
glutinin glycoproteins that are present on the surface of the influ-
enza virus. Importantly, if present in sufficient quantities,
antibodies specific to the hemagglutinin on the influenza viruses
(e.g., those produced following vaccination) can inhibit or prevent
this hemagglutination, because they bind to the same glycoproteins
on the viral surface, preventing them from interacting with the red
blood cells.
Practically, these characteristics of red blood cells, influenza
virus, and hemagglutination-inhibiting antibodies are manipulated
in the HAI assay (see Fig. 2).
1. Nonspecific agglutinins are removed by incubating samples
overnight with a receptor-destroying enzyme.
2. Serum samples are serially diluted to increasing levels (e.g., 1:4,
1:8, 1:16, 1:32) and then incubated in a microtiter plate with a
standardized quantity of influenza viral antigen (see Note 3).
3. A standardized quantity of red blood cells are then added to the
plate and if antibodies are present in sufficient quantities this
will inhibit the hemagglutination process, with red blood cells
settling in a small pellet at the bottom of the well. Antibody
titers are then read as the reciprocal of the highest dilution at
which hemagglutination is inhibited.
No Virus Control
1:10240
1:1280
1:2560
1:5120
1:160
1:320
1:640
1:10
1:20
1:40
1:80
Sample A
Sample B
Sample C
Settled Red
Sample D Blood Cells
Sample E No Reaction
Sample F
(Control Well)
Sample G
Sample H
No Serum Control
Serum Antibody
bound to virus
Lattice Suspension of
Red Blood Cells and
Settled Red Virus
Blood Cells
Hemagglutination Hemagglutination
Inhibited
Fig. 2 Schematic overview of a hemagglutination inhibition assay. Viral antigens and serially diluted sera are
incubated together in a nonreactive 96-well plate, alongside controls. A standardized quantity of red blood
cells are then added. If antibodies are present in sufficient quantities this will inhibit the hemagglutination
process, with red blood cells settling in a small pellet at the bottom of the well. Antibody titers are read by eye
as the reciprocal of the highest dilution at which hemagglutination is fully inhibited
3.2.1 Materials 1. Nonreactive V-bottomed (if using chicken or turkey RBCs) or

U-bottomed (if using guinea pig or human type O RBCs)
96-well plates.
2. Purified vaccine antigens.
4. Red blood cells in PBS at 0.5% for chicken or turkey RBCs. Use
0.75% if using guinea pig or human type O RBCs (see Note 4).
5. Receptor-destroying enzyme.
3.2.2 Advantages The main advantages of the HAI assay are that it is relatively of low
and Limitations of HAI cost, is simple to perform, and does not require much specialist
equipment. Further, the HAI assay has been examined in multiple
viral challenge and epidemiological studies, meaning that estimated
thresholds for clinical protection have been established
[17–19]. This allows researchers to categorize samples according
to whether they reach “protective” levels.1
However, the HAI assay has many limitations. First, results of
the assay are read visually by the researcher, potentially adding a
1
Note that due to the variability in the real-world exposure to viral particles, it is not possible to truly say that an
individual is protected from infection based on antibody levels. Rather, protection estimates are probabilistic.
degree of interpretation and potential biases, especially when the

researcher is not blind to the research questions. Second, while
simple to perform, the HAI assay process is labor intensive, espe-
cially if multiple replicates are desired to improve reliability. A
standard 96-well microtiter plate can only hold 7 sera diluted to
12 concentrations and 1 negative control. Therefore, if 100 sera
require analysis for antibody titers for each of the three influenza
vaccine strains, each with 3 replicates, this would require at least
129 plates to be processed, as well as considerable volumes of sera,
antigens, and red blood cells. Third, the HAI assay cannot distin-
guish between different antibody isotypes, so may be unsuitable for
certain research questions. Finally, the most limiting feature of the
HAI assay is that due to its requirement for titration (the increasing
dilution of sera), samples can only be designated 1 of up to 12 pos-
sible discrete values (i.e., 10, 20, 40). This means significant varia-
tion is lost and actual antibody levels cannot be known with a high
degree of specificity.
4 Multiplex Assays
Recent technological advances in immunoassays mean that it is now

possible to analyze vaccine responses in large numbers of samples
and/or for multiple immunological outcomes comparatively
quickly compared to traditionally employed assays. Central to this
is the development of miniaturized and multiplex assays (so called
because they can measure multiple analytes of interest in a single
processing of a sample). While the use of multiplex assays is still
relatively rare for the analysis of vaccine responses in PNI, the
advent of the “multiplex era” has implications for PNI research.
PNI has historically received criticism for an overreliance on small,
often underpowered, studies [20] with the cost and time associated
with performing more traditional assays often a critical limiting
factor on sample size. Multiplex assays overcome some of these
difficulties, typically being more efficient, with higher throughput,
and lower costs per analyte (excluding any up-front costs of equip-
ment purchases). The multiplex methods discussed below should
therefore be of increasing interest and relevance for PNI researchers
moving forward.
4.1 Bead-Based Bead-based multiplex assays are characterized by their use of mini-
Assays ature polymer or glass beads (typically <10 μm in diameter) with
identifiable fluorescent signatures that can be detected using flow
cytometry techniques. Multiple sets of beads, each with different
fluorescent signatures, can be coated with specific capture antibo-
dies or antigens of interest and added together to a single sample
allowing the quantification of multiple analytes simultaneously.
Currently, there are two main types of bead-based assays:
cytometric bead assays (CBA) and Luminex. CBA employs beads

with just one fluorescent color to differentiate bead sets and has the
advantage that it can be performed on standard clinical flow cyt-
ometer equipment found in many laboratories. The use of one
fluorescent color per bead set does, however, limit the number of
analytes that can be measured in each sample, which may limit its
utility for certain research questions. Luminex, in contrast, relies on
beads with more complex fluorescence signatures allowing for
more extensive multiplex capacity. However, a significant disadvan-
tage of Luminex assays is that they require specialist dedicated
analyzers, limiting their utility and availability for some researchers.
In both CBA and Luminex assays the procedures are broadly
similar.
1. First, each set of beads (see Note 5) are activated and incubated
with a capture antigen or antibody for each analyte of interest,
which bind to the beads three-dimensional surface (see Fig. 3).
2. Next, a standardized quantity of each bead set is added to each
diluted sample, as well as to a serially diluted standard curve
made up of relevant standards (e.g., human IgG standard).
3. Beads are then washed, and incubated with a conjugated detec-
tion/reporter antibody (e.g., anti-human IgG).
Multiple
Bead Conjugated
Bead Sets
with Antigen
Detection
Antibody
Sample
Serum Antibody
Fig. 3 Schematic overview of bead-based assays. Separate bead sets are conjugated with antigens of interest,
before being added to and incubated with serum samples. Detection antibodies are then added and left to
incubate further. Beads are then washed and are read using an analytical flow chamber supporting individual
bead separation. Signals for each bead set can then be read using laser fluorescence techniques and
compared to a standard/calibration curve
4. After a final wash they can be detected via fluorescent signatures

using an analytical flow chamber supporting individual bead
separation. Fluorescence signals in each sample are then read
using laser fluorescence techniques with the signal detected for
each bead set proportional to the level of each analyte present
in the sample. These are then interpolated against the produced
standard curves to quantify the level of each analyte in the
sample. Samples with higher antibody levels give greater fluo-
rescence signals, in turn indicating a greater probability of
clinical protection.
4.1.1 Materials 1. Capture beads (either commercially prepared for analytes of

choice or with conjugation reagents as per the manufacturer’s
guidelines).
2. Cytometer setup beads.
4. PE-conjugated anti-human IgG (or alternative for other
isotypes).
6. Blocking solution: 0.1% Bovine serum albumin, 0.02% Tween-
20, 0.05% sodium azide in PBS.
8. Assay buffer (according to the manufacturer’s guidelines, often
0.1% bovine serum albumin in PBS).
9. Analytical flow chamber supporting individual bead separation.
4.1.2 Advantages As with all multiplex assays, the central advantage of bead-based
and Limitations of Bead- assays is that multiple analytes of interest (e.g., different antibody
Based Assays strains, isotypes) can be measured simultaneously in a given sample.
This is particularly valuable if sample volumes are limited and/or a
wide variety of analytes are required to be measured. As bead-based
assays rely on fluorescence signals, the assay also typically has a large
dynamic range. While these advantages are significant, it is impor-
tant to note that these assays require a greater level of expertise than
the traditional assays described above (e.g., use of flow cytometry
equipment). Further, unless already available to researchers, the
equipment costs associated with bead-based assays can be
prohibitive.
4.2 Microarray Microarray assays [21] have been widely used in genetic research,
Assays most frequently to examine the expression of multiple gene
sequences by simultaneously “printing” DNA or RNA in minute
quantities to a reactive glass slide [22, 23]. More recently, this
technology has been adapted to allow antibodies against specific
antigens to be measured in multiple, miniaturized assays (so-called
Detection
Slide Holder Antibody
Serum
Antibody
Printed Antigen
Glass Slide
Slide
Fig. 4 Schematic overview of microarray assays. Antigens of interest and serial dilutions of human IgG
(to generate a standard curve) are printed onto a reactive slide surface by an arrayer. The remainder of the
slide surface is then blocked to prevent further nonspecific binding. Diluted serum samples are then added to
the surface. After washing, a detection antibody is added, followed by a fluorescent dye that binds to the
detection antibody. Slides are read using a laser scanner to obtain fluorescence values for the standard curve
and specific antibody responses
antigen microarrays) of which thousands can theoretically be per-

formed on a single glass slide [15, 24, 25].
Antigen microarrays are a type of protein microarray in which
tiny spots of antigen (<200 μm in diameter) are “printed” onto a
reactive surface (in this case a treated glass slide) to detect antibo-
dies specific to the antigen in a given sample. Theoretically, antigen
microarray assays are, in essence, miniaturized versions of the
ELISA assay. However, unlike ELISA methods, the miniaturized
nature of antigen microarray assays means that many replicates can
be performed simultaneously, for multiple antigens, with small
quantities of sample and high throughput. The generalized process
of antigen microarray is described below and illustrated in Fig. 4.
1. First, the antigens of interest are bound in small “spots” onto a
reactive surface before the remainder of the surface is blocked
to prevent further binding.
2. Diluted serum samples are then added to the surface, at which
point antibodies in the sample with specificity to epitopes on
the antigen surface bind to the antigen.
3. After unbound serum is washed away, a secondary antibody
that binds specifically to human IgG antibodies (known as anti-
human IgG) is added. This secondary antibody is “labeled”
with biotin, a small molecule that can be readily recognized by
the protein streptavidin.
4. Finally, a fluorescent dye (cyanine5) which is conjugated to
streptavidin is added.
5. This, in turn, can be detected by laser scanning. The fluores-
cence of a given sample (measured in arbitrary fluorescence
units) is proportional to the number of antibodies in the
serum sample that bound to the printed antigen, with greater
fluorescence indicating more antibodies and in turn greater

probability of clinical protection. As with ELISA, a calibration
curve of human IgG standard is also printed and probed,
against which fluorescence values of all samples can be quanti-
fied through interpolation.
It is common practice for microarray data to be systematically
pre-processed and filtered in line with predefined criteria to remove
the influence of aberrant spots that can occur in printing or probing
stages. Criteria for removing aberrant spots can be more or less
stringent depending on the reliability desired, with potential filter
variables including spot circularity or size. This is a particular
advantage of microarray methods and can dramatically improve
the reliability of the assay [26].
4.2.1 Materials 1. 96-Well and 384-well plates.

2. Aminosilane-coated glass slides.
3. Slide holders with hydrophobic barriers/gaskets.
4. Robotic micro-arrayer (e.g., Biorobotics MicroGridII arrayer).
5. Fluorescence slide scanner (e.g., GenePix 4200AL).
7. Antibody diluent.
8. Biotinylated anti-human IgG (or alternative for other
isotypes).
11. Printing buffer: PBS Trehalose Tween.
13. Blocking solution: 5% Bovine serum albumin in PBS.
14. Streptavidin cyanin-5 dye.
4.2.2 Advantages One of the most significant advantages of antigen microarray assays
and Limitations is that they are miniaturized and multiplex, meaning that many
of Microarray Assays types of antibodies can be detected within a single processing of a
sample [24]. This vastly reduces the quantities of sera, antigen,
reagents, and time required to perform the assay [15]. Large parts
of the process are automated and resultant fluorescence signals are
read through laser scanning techniques, which produce a highly
specific continuous outcome with a large dynamic range. Further,
antigen microarrays can be made increasingly robust as a large
number of replicates can be performed simultaneously and they
can be adapted to include multiple internal quality control mea-
sures [24]. The most limiting factor of this assay is that it requires
expensive, specialist equipment and expertise. However, such

equipment is becoming increasingly common in modern labora-
tories due to their many potential applications.
5 Conclusion
Technological advances now afford PNI researchers novel oppor-

tunities to assess vaccine responses in a variety of ways. Multiplex
assays overcome the limiting features of traditional immunoassays
(such as ELISA and HAI) allowing researchers to efficiently exam-
ine multiple analytes simultaneously in small volumes of sample.
This increased capacity opens possibilities for larger sample sizes
among PNI studies examining vaccination responses, with greater
reliability and robustness in outcome measures. Further, these
techniques open a world of new research possibilities, including
the exploration of concurrent and interactive changes in hormonal,
inflammatory, and immune processes over time. This allows
researchers to further explore the mechanisms by which psycholog-
ical and behavioral factors influence immune pathways and
response to challenges, such as vaccination. While traditional assays
will still have their place in PNI, the onset of the multiplex era
presents an exciting new frontier to be explored.
6 Notes
1. The appropriate dilutions of vaccine antigens and serum sam-

ples depend on a number of factors and need to be determined
prior to running all samples for analysis. Post-vaccination sam-
ples are likely to contain high levels of antigen-specific IgG, so
samples may need considerable dilution to fit within the
dynamic range of the assay (more so for ELISA than micro-
array). One way of determining optimal dilutions is to take a
selection of pre- and post-vaccination samples and process
these simultaneously at a range of dilutions. You can then
choose the most appropriate dilution for your needs.
2. We would not advise using a 384-well plate for ELISAs without
access to a liquid-handling robot, as the miniaturized nature of
the wells makes manual performance of the assay tricky.
3. Very detailed procedures for performing HAI assays and pre-
paring required buffers/solutions are provided in the Manual
for the laboratory diagnosis and virological surveillance of
influenza published by the World Health Organization (avail-
able from http://www.who.int/influenza/gisrs_laboratory/
manual_diagnosis_surveillance_influenza/en/). Where the
description provided here is fairly brief, we strongly
recommend researchers consult these more detailed guidelines

when conducting HAI assays.
4. If you are not preparing the packed RBCs in-house and are
commercially purchasing these, do note that the shelf life of
RBCs is quite short and therefore delivery of these should be
arranged accordingly.
5. Many analytes of interest (particularly cytokines) are available
commercially pre-conjugated to allow for easier processing.
References
1. Vedhara K, Cox NK, Wilcock GK et al (1999) 10. Vedhara K, Bennett PD, Clark S et al (2003)
Chronic stress in elderly carers of dementia Enhancement of antibody responses to influ-
patients and antibody response to influenza enza vaccination in the elderly following a
vaccination. Lancet 353:627–631 cognitive-behavioural stress management
2. Glaser R, Sheridan J, Malarkey WB et al (2000) intervention. Psychother Psychosom
Chronic stress modulates the immune response 72:245–252
to a pneumococcal pneumonia vaccine. Psy- 11. Siegrist C (2012) Vaccine immunology. In:
chosom Med 62:804–807 Plotkin S, Orenstein W, Offit P (eds) Vaccines,
3. Hayney MS, Coe C, Muller D et al (2014) Age 6th edn. Elsevier Inc., Amsterdam, pp 17–36
and psychological influences on immune 12. McElhaney JE, Ewen C, Zhou X et al (2009)
responses to trivalent inactivated influenza vac- Granzyme B: correlates with protection and
cine in the meditation or exercise for prevent- enhanced CTL response to influenza vaccina-
ing acute respiratory infection (MEPARI) trial. tion in older adults. Vaccine 27:2418–2425
Hum Vaccin Immunother 10:2759–2767 13. McElhaney JE, Xie D, Hager WD et al (2006)
4. Ayling K, Fairclough L, Tighe P et al (2018) T cell responses are better correlates of vaccine
Positive mood on the day of influenza vaccina- protection in the elderly. J Immunol
tion predicts vaccine effectiveness: a prospec- 176:6333–6339
tive observational cohort study. Brain Behav 14. Lange T, Dimitrov S, Bollinger T et al (2011)
Immun 67:314–323. https://doi.org/10. Sleep after vaccination boosts immunological
1016/j.bbi.2017.09.008 memory. J Immunol 187:283–290
5. Phillips A (2012) The vaccination model in 15. Leng SX, McElhaney JE, Walston JD et al
psychoneuroimmunology research: a review. (2008) ELISA and multiplex technologies for
In: Yan Q (ed) Psychoneuroimmunology: cytokine measurement in inflammation and
methods and protocols. Springer Science & aging research. J Gerontol A Biol Sci Med Sci
Business Media, LLC, Santa Clara, CA, pp 63:879–884
355–370 16. Pedersen J (2014) Hemagglutination-
6. Burns VE, Gallagher S (2010) Antibody Inhibition Assay for Influenza Virus Subtype
response to vaccination as a marker of in vivo Identification and the Detection and Quantita-
immune function in psychophysiological tion of Serum Antibodies to Influenza Virus.
research. Neurosci Biobehav Rev 35:122–126 In: Spackman E (ed) Animal influenza virus,
7. Chen W, Kozlovsky B, Effros R (2009) Vacci- methods in molecular biology. Springer Sci-
nation in the elderly: an immunological per- ence & Business Media, LLC, New York, pp
spective. Trends Immunol 30:351–359 11–25
8. Woods JA, Keylock KT, Lowder T et al (2009) 17. Hobson D, Curry RL, Beare AS et al (1972)
Cardiovascular exercise training extends influ- The role of serum haemagglutination-
enza vaccine seroprotection in sedentary older inhibiting antibody in protection against chal-
adults: the immune function intervention trial. lenge infection with influenza A2 and B viruses.
J Am Geriatr Soc 57:2183–2191 J Hyg (Lond) 70:767–777
9. Edwards KM, Burns VE, Reynolds T et al 18. de Jong JC, Palache AM, Beyer WEP et al
(2006) Acute stress exposure prior to influenza (2003) Haemagglutination-inhibiting anti-
vaccination enhances antibody response in body to influenza virus. Dev Biol (Basel)
women. Brain Behav Immun 20:159–168 115:63–73
19. Coudeville L, Bailleux F, Riche B et al (2010) 23. Jiang H, Deng Y, Chen H et al (2004) Joint
Relationship between haemagglutination- analysis of two microarray gene-expression data
inhibiting antibody titres and clinical protec- sets to select lung adenocarcinoma marker
tion against influenza: development and appli- genes. BMC Bioinfomatics 5:1–12
cation of a bayesian random-effects model. 24. Negm OH, Hamed MR, Dilnot EM et al
BMC Med Res Methodol 10:18 (2015) Profiling humoral immune responses
20. Segerstrom S, Smith G (2012) Methods, vari- to clostridium difficile-specific antigens by pro-
ance, and error in psychoneuroimmunology tein microarray analysis. Clin Vacc Immunol
research: the good, the bad, and the ugly. In: 22:1033–1039
Segerstrom S (ed) The Oxford handbook of 25. Selvarajah S, Negm OH, Hamed MR et al
psychoneuroimmunology. Oxford University (2014) Development and validation of protein
Press, New York, pp 421–432 microarray technology for simultaneous
21. Chang T (1983) Binding of cells to matrixes of inflammatory mediator detection in human
distinct antibodies coated on solid surface. J sera. Mediators Inflamm 2014:820304
Immunol Methods 65:217–223 26. Tighe P, Ryder RR, Todd I et al (2015) ELISA
22. Shena M, Shalon D, Davis RW et al (1995) in the multiplex era: potentials and pitfalls.
Quantitative monitoring of gene expression Proteomics Clin Appl 9:406–422
patterns with a complimentary DNA microar-
ray. Science 270:467–470
Chapter 18
Sculpting the Sculptors: Methods for Studying the Fetal

Cholinergic Signaling on Systems and Cellular Scales
Martin G. Frasch, Patrick Burns, Javier Benito, Marina Cortes,
Mingju Cao, Gilles Fecteau, and André Desrochers
Abstract
The non-neuronal, immunological effects of the cholinergic signaling are exerted on the system’s scale of
observation via the vagus nerve and on the cellular scale via α7 nicotinic acetylcholine receptor (nAChR)
signaling in myeloid cells of the periphery or brain’s microglia and astrocytes. The developmental effects of
such multi-scale signaling can be conceived of as an example of psychoneuroimmunological (PNI) home-
okinesis and, while reported in the literature, are not yet systematically well studied. To be better under-
stood, the intricacy of the multi-scale interactions requires relevant preclinical animal models. Chronically
instrumented non-anesthetized fetal sheep model comes with a strong track record of bench-to-bed
translation and a large body of evidence for its strong resemblance to and relevance for human physiology
on various scales of organization. Recently, there has been growing interest in pleiotropic effects of vagus
nerve stimulation (VNS) on various organ systems such as innate immunity, metabolism, and emotion with
implications for programming of PNI phenotype. Here we describe the procedures required to record and
manipulate the vagus nerve activity in this large pregnant mammalian organism. Extending this in vivo
model to in vitro, on the cellular scale, we present the method to manipulate the cholinergic signaling in
ovine fetal microglia and astrocytes and analyze their responses on protein and RNA levels. Together these
models can provide multi-scale-level mechanistic insights into the effects of cholinergic signaling on PNI
phenotype.
Key words Vagus nerve, Vagotomy, Nerve stimulation, ENG, Animal model, Physiology, Surgery,
Chronic experimentation, Multivariate data acquisition, Neuroscience
1 Introduction
Vagal nerve stimulation(VNS) has been long used for treatment of

drug-resistant epilepsy [1]. More recently, this clinically well-
tolerated treatment approach has been explored in multiple animal
experimental models and clinical trials for treatment of a number of
conditions involving brain’s control of the innate immune system
[2]. The underlying substrate for this approach is the cholinergic
anti-inflammatory pathway (CAP), mediated by the vagus nerves.
Much remains to be studied about manipulating the vagus nerve
341
342 Martin G. Frasch et al.
activity and the physiology of its pleiotropic effects. The emerging

field of the bioelectronic medicine has aimed to tackle this
challenge.
In the field of studying the fetal development, the chronically
instrumented pregnant sheep has been used extensively for 50 years
as a highly translational animal model [3–8]. The extensive use of
the fetal sheep model is due to the unique amenability of the sheep
to the surgical placement and maintenance of catheters and elec-
trodes, allowing repetitive blood sampling, recording of bioelectri-
cal activity, application of electric stimulation, in vivo brain imaging
[9], postnatal follow-up of the lambs, and ability to derive primary
brain cell cultures, such as microglia and astrocytes, which permits
modeling multi-hit scenario of fetal and postnatal insults shaping
the glial phenotype and hence glial-neuronal interactions in a spa-
tiotemporally highly dynamic brain. This model enables studying
the programming consequences of perinatal insults on the individ-
ual developing psychoneuroimmunological (PNI) phenotype.
Here we describe the VNS methodology in near-term fetal
sheep model which allows the study of fetal vagus nerve activity in
relation to other biophysical and biochemical parameters. This can
be useful to further elucidate the function of neuroimmunological
circuits such as the CAP and in the broader context of the PNI. For
example, in an accompanying chapter we review studies and models
of prenatal stress and its effects on brain development. As discussed
in that chapter, the vagus nerve influences brain function and body
metabolism in a pleiotropic manner [10, 11]. VNS stimulates CRH
expression in hypothalamus [12], while CRH receptor 1 agonism
increases vagal modulation of the heart rate variability (HRV), an
important proxy biomarker of vagal modulatory activity [13–15].
This reciprocal CRH–vagus nerve circuitry provides an important
diagnostic and therapeutic link between stress and autonomic ner-
vous system (ANS) of which vagus nerve is a key player. Such
findings emphasize the promise models of systemic cholinergic
manipulation carry to unlock possibilities of boosting endogenous
homeokinetic mechanisms which influence PNI phenotype [16].
This does not need to remain a basic science affair. Novel
noninvasive methods of VNS are being developed which do not
require surgical cervical VNS implants, have minimal to no side
effects, and are of low cost [17–20]. It is now possible to conceive
of VNS treatment of neonates and other age groups where invasive
surgical implants cannot be a treatment of choice.
Due to the highly dynamic nature of human development from
fetus to newborn and into adulthood on the one hand and the
active role of peripheral macrophages in innate immune responses
and brain’s microglia in synaptogenesis on the other hand, the
effects of the cholinergic stimulation can be profound and long-
lasting on the programming of the immune function and the brain
wiring. These effects translate into a spectrum of the yet fully to be
Sculpting the Sculptors: Methods for Studying the Fetal Cholinergic. . . 343
Manipulation of Cholinergic Signaling

In vivo
Vagus Nerve
Control Sham Inhibition Stimulation Data Acquisition
Systems Scale
Vagus nerve in
Fetal Sheep – Electrocardiogram (ECG)
– Vagus electroneurogram (VENG)
– Heart rate variability (HRV)
Analysis
– Complex signals bioinformatics
NaCl Lipopoly- Vagotomy Vagus

saccharide Nerve
Stimulation
In vitro
α7 Nicotinic Acetylcholine Receptor
Fetal Brain Control Sham Inhibition Stimulation
Cellular Scale
Data Acquisition
Astrocytes: – ELISA
– RNAseq
α7nAChR Analysis
Microglia:
– Bioinformatics
Drugs: Vehicle Lipopoly- Antagonist Agonist

saccharide
Fig. 1 The in vivo–in vitro approach to cholinergic manipulation in a large mammalian developing organism.
α7nAChR, α7 nicotinic acetylcholine receptor. In vivo, unanesthetized fetal sheep instrumentation allows for
chronic multimodal recording and stimulation of the cervical vagus nerve unilaterally or bilaterally. This is
followed by a set of data analyses to tie together the findings. Mirroring the cholinergic manipulation on the
systems scale, the in vitro primary glia (astrocytes or microglia) cultures can be derived from any of the four
described in vivo experimental groups. This results in creation of naı̈ve (no in vivo exposure, i.e., control) and
second-hit in vitro experimental conditions. Protein- and transcriptome-level responses to manipulation of
cholinergic signaling via the α7nAChR can be tested under conditions of exposure to endotoxin (lipopolysac-
charide) with the presence of a α7nAChR antagonist (equivalent of an in vivo vagotomy) or α7nAChR agonist
(equivalent of an in vivo vagus nerve stimulation). This is again followed by bioinformatics analyses. Creating
one cohesive PNI-type physiological framework from the systems-scale and cellular-scale bioinformatics
results and models represents an exciting challenge for the immediate future
defined psychological phenotypes, influenced by processes sculpt-

ing the activity of these peripheral and central immune cells such as
nutrition, stress, or infection during pregnancy [21–23].
Together, there is strong evidence that vagus nerve activity is a
key player in control of neuroinflammation and participates in
endogenous homeokinetic mechanisms which may be leveraged
to improve developmental brain trajectories early postnatally.
We refer the interested reader to a publication by Burns et al.
reporting the detailed protocol on the elementary fetal sheep
model setup [8]. The in vivo–in vitro model protocol is described
in detail by Cortes et al. [24]. The broader picture of the approach
is summarized in Fig. 1.
In the following we build on these openly available protocols by
briefly reviewing them. We then focus on the manipulation of the
cholinergic signaling by presenting in more detail the steps required
to instrument and manipulate the fetal sheep vagus nerves in vivo as
well as to manipulate glial cholinergic signaling in vitro in primary
cultures derived from the animals manipulated in vivo.
2 In Vivo Protocol
2.1 Cholinergic Pregnant time-dated ewes are instrumented at the gestational age
Signaling of choice, usually starting at around 105 days of gestation (dGA,
Manipulation In Vivo 145 dGA term) with arterial, venous, and amniotic catheters and
Via Vagus Nerve ECG electrodes [8]. Ovine singleton fetuses of mixed breed are
Stimulation surgically instrumented with sterile technique under general anes-
thesia (both ewe and fetus). In case of twin pregnancy, the larger
2.1.1 Anesthesia and fetus is chosen based on palpating and estimating the outer inter-
Surgical Procedure orbital distance. The total duration of the surgical procedure (i.e.,
skin to skin and not counting preparation and anesthesia induction)
is about 2 h. Antibiotics are administered preoperatively to the
mother intravenously (trimethoprim sulfadoxine 5 mg/kg) as well
as to the fetus intravenously and into the amniotic cavity (ampicillin
250 mg) at the time when fetus is returned into the amniotic cavity
and the uterus has been closed. Amniotic fluid lost during surgery is
replaced with warm saline which in our experience averages
120 mL. The catheters exteriorized through the maternal flank
were secured to the back of the ewe in a plastic pouch. For the
duration of the experiment the ewe is returned to the metabolic
cage, where she can stand, lie, and eat ad libitum while the
non-anesthetized fetus is being monitored without sedating the
mother. During postoperative recovery, antibiotic administration
is continued for 3 days. Arterial blood is sampled for evaluation of
maternal and fetal condition and catheters are flushed with hepa-
rinized saline to maintain patency (1 U/mL, 2 mL maximum at a
time and no more than 10 mL daily).
Bilateral cervical vagotomy is performed in sheep fetuses as
follows. The head is extended and the neck is kept straight and
stable while approaching and exposing the vagal nerves. A bilateral
cervical vagotomy (Vx) is then performed and the VNS/VENG
electrodes are installed around the nerve. Vx step can be omitted.
Vx has the advantage that a more mechanistic dissection of efferent
and afferent VNS can be undertaken. The skin is then closed in a
continuous pattern. The animals not receiving Vx or instrumenta-
tion with VNS/VENG probe are subjected to a sham neck surgery
with all steps except Vx.
During surgery (once the first fetal arterial catheter was in place
and before returning the fetus to the uterus) a 3 mL fetal arterial
blood sample is taken for blood gas, lactate, and cytokine
measurements.
2.2 Vagus Nerve ECG, FHR, and arterial blood pressure can be monitored continu-
Electroneurogram ously (for example using the 1902 amplifier and micro3 1401 ADC
Recordings (VENG) and by CED, Cambridge, UK, and NL108A, NeuroLog, Digitimer,
Nerve Stimulation Hertfordshire, UK) and sampled at 1000 and 256 Hz, respectively.
(VNS) VNS can be applied via NeuroLog’s NL512/NL800A using pulse
sequence pre-programmed in Spike 2. The VNS settings can be as

follows: DC rectangular 5 V, 100 uA, 2 ms, 1 Hz according to
Borovikova et al. [25]. VENG should be recorded at 20,000 Hz.
Various VNS settings used so far to achieve effects on neuroim-
mune axis are reviewed by Kwan et al. [2]
2.3 Materials and Data acquisition system can be used such as a 1902 amplifier and
Methods micro3 1401 ADC by CED, Cambridge, UK. VENG should be
recorded at 20,000 Hz sampling rate. Care should be taken in
designing the VNS protocol, as this area is far from being well
defined and understood [2]. Nerve stimulation system can be
used such as NL108A, NeuroLog, Digitimer, Hertfordshire, UK.
The general framework of the pregnant sheep approach has
been published in [8]. This paper reports the precise materials
and steps needed to implement the approach.
3 In Vitro Protocol
3.1 Cholinergic Mechanisms of fetal neuroinflammation can be studied in a large

Signaling mammalian brain using the in vivo–in vitro multi-hit model of
Manipulation In Vitro: pregnant sheep [24, 26]. The proteomic and transcriptomic
Primary Microglia and changes can be studied following modulation of cholinergic signal-
Astrocyte Culture ing through α7nAChR with selective antagonistic drug
Approach α-bungarotoxin and agonistic drug AR-R17779 [27]. While the
proof-of-principle approach is presented with LPS as the stimulant,
other triggers can be used. Incubation with LPS in vitro for 6 h
results in activation of major inflammatory pathways JAK-STAT
and NFKB in naı̈ve microglia [24, 26]. The activation of these
inflammatory pathways in microglia exposed to LPS and pretreated
with α-bungarotoxin is further enhanced. Both transcriptomic acti-
vation and protein-level IL-1β secretion patterns are enhanced in
α-bungarotoxin-pretreated microglia compared to microglia
exposed to LPS alone. Our findings extend the in vitro observations
in mature rodent primary microglia cultures to a large mammalian
developing brain exposed to LPS in vivo and in vitro [28].
The present model and findings represent a direct in vitro
validation of the earlier indirect in vivo observations in the fetal
sheep brain of the same gestational age when we proposed the
existence of a fetal cerebral cholinergic anti-inflammatory pathway
with in situ evidence of the expression of α7nAChR on microglia
[29]. Due to key role of fetal microglia in sculpting the neurogen-
esis and synaptic development during the vulnerable periods of
human development, understanding the mechanisms by which
these sculptors of brain development can themselves be manipu-
lated, i.e., sculpted, is of great interest, as it both sheds light onto
the endogenous mechanisms of homeokinetic control via afferent
cholinergic neuroimmune pathways and provides avenues for
testing novel ways to restore neuroimmunological signaling pat-

terns and functional alterations in brain behavioral phenotype
which respond to cholinergic signaling.
3.1.1 In Vitro Microglia Fetal sheep brain tissues are obtained during sheep autopsy after
Culture Protocol completion of the in vivo experiment to conduct the in vitro study.
The non-instrumented, untreated twins were designated “naı̈ve”
(no LPS exposure in vivo). Instrumented animals that received LPS
in vivo were used for second-hit LPS exposure in vitro.
Fetal sheep microglia culture protocol was adapted from an
established human adult and fetal microglia culture protocol that
was modified to include a myelin removal step following the high-
speed centrifugation [30].
Briefly, fetal sheep cells were plated on poly-L-lysine (PLL)-
coated tissue culture flasks at a concentration of 2 106 cells/
mL in DMEM with 5% heat-inactivated fetal bovine serum (Gibco,
Canada Origin), 1% penicillin/streptomycin, and 1% glutamine (5%
DMEM), in which microglia grow best [30]. Cells were allowed to
incubate for 7 days at 37 C, 5% CO2, followed by a media change
by centrifugation and the addition of resuspended cells back to the
culture flask. Cells were continued to incubate for seven more days
with 5% DMEM at 37 C, 5% CO2, before the floating cells were
collected. After carefully collecting the floating microglia to avoid
contamination with astrocytes and oligodendrocytes, the cells were
incubated in 24-well plate at 1 105 cells/mL with 5% DMEM for
another 4–5 days, and then treated with or without LPS (100 ng/
mL, Sigma L5024, from E. coli O127:B8) for 6 h. Cell-conditioned
media were collected for cytokine analysis, and 0.5 mL TriZol per
well was added for RNA extraction.
To verify microglia purity, a portion of floating cells was
cultured in 24-well plate under the above conditions for flow
cytometry analysis. The cell morphology was documented with
light microscopy. Another portion of floating cells was plated
onto Lab-Tek 8-well chamber glass slide (Thermo Scientific) and
treated with or without LPS for immunocytochemistry (ICC)
analysis.
3.1.2 In Vitro Astrocyte Astrocytes are the major adherent cell population in flask. Astro-
Culture Protocol cytes are purified by passage into a new T75 flask (no PLL coated is
required) for 4–5 times before any manipulations and treatments.
After floating microglia collection, detach the adherent cells by
trypsinization (trypsin 0.25% + EDTA 0.1%, Wisent Cat. No
325-043-EL) and replate into a new flask; culture for another
7 days with 10% all-dressed DMEM. Repeat the above step 2.6.2
for 4–5 times; the cell passage procedure takes 4–5 weeks until
purified astrocytes can be used for treatment.
Plate pure astrocytes into 24-well plate at 1 105 cells/mL
with 10% DMEM for another 7 days, and then treat with or
without LPS for 6 h. We chose LPS here as an example of an in vitro

stimulus; this can be changed depending on your research question.
Similar to microglia culture, collect cell-conditioned media for
cytokine analysis; use Qiagen RNeasy Mini Kit (Cat no. 74104) for
RNA extraction. To verify astrocyte purity, plate a portion of cells
into Lab-Tek 8-well chamber glass slide (Thermo Scientific) for
ICC analysis. Use glial fibrillary acidic protein (GFAP) as an astro-
cyte marker, and counterstain cells with Hoechst.
3.1.3 Cell Culture and Prior to exposure to LPS, cells are pretreated for 1 h with either
Treatment 10 nM AR-R17779 hydrochloride (Tocris Bioscience Cat# 3964),
a selective α7nAChR agonist, or 100 nM α-bungarotoxin (Tocris
Bioscience Cat# 2133), a selective α7nAChR antagonist. Optimal
dose of AR-R17779 (A) or α-bungarotoxin (B) was chosen based
on a dose-response experiment with LPS; AR-R17779 hydrochlo-
ride was dissolved in DMSO into a stock solution. α-Bungarotoxin
was reconstituted with culture media into a stock solution and
underwent serial dilutions. AR-R17779 and α-bungarotoxin pre-
parations are added well by well; the same volume of vehicle (either
DMSO or cell culture media) was added in control wells. There-
fore, in a complete cell culture experiment, we have four experi-
mental groups: control (naı̈ve control or NC), LPS (naı̈ve LPS or
NL), LPS + B (naı̈ve LPS + B or NB), and LPS + A (naı̈ve LPS + A
or NA). If the ovine fetuses were preexposed to LPS (or another
experimental stimulus of choice) in vivo systemically, this constitu-
tes a second hit in vitro. Second-hit cell cultures are designed with
the same pattern and divided into four experimental groups: con-
trol (SHC), LPS (SHL), LPS + B (SHB), and LPS + A (SHA).
3.1.4 Measurements of The inflammatory responses are assessed on protein level with
Inflammatory Responses ELISA and on transcriptome level with RNAseq methods which
are described below in detail.
Measurement of Cytokines Cytokine concentrations in cell culture media (IL-1β) are deter-
in Plasma and Cell Culture mined by using an ovine-specific sandwich ELISA reported in detail
Media in [24, 26] which is available as in open access.
RNAseq Approach The overall experimental design can be divided into three phases:
sequencing, quantification, and discovery.
3.1.5 RNA Extraction and Total RNA is extracted from cultured microglia using TRIzol
RNA Quantification Reagent (Life Technologies). RNA quantity and quality (RNA
integrity number, RIN) are established by using a RNA Nano
Chip (Agilent RNA 6000 Nano Chips) with Agilent 2100
BioAnalyzer.
RNAseq libraries can be prepared using Illumina TruSeq RNA
Sample Preparation v2 kit (Illumina) and quality control can be
performed on the BioAnalyzer. Single-end 5 bp sequencing can be

performed at high throughput on an Illumina HiSeq2500 (which
in this example was located at the CHU Ste-Justine Core Facility
Sequencing Platform).
3.1.6 RNAseq Data Read alignment to the reference genome. To maximize the number
Analysis of genes covered, raw data are mapped to the reference genome of
the sheep Ovis aris v3.1 from Ensembl (GCA_000298735.1) as the
reference genome. Index of the reference fasta file is built with
Bowtie2 [31]. Trim the adaptor of the fastQ files with TrimGalore
and map reads to the reference genome with Tophat2 [32]. From
the aligned reads from Tophat2, the number of reads aligned is
counted with HTseq and assembled into a matrix containing the
read count of each gene per sample [33].
3.1.7 Normalization and In order to find differentially expressed genes we used DESeq2 to
Transcriptome Analysis normalize the dataset, and generate Log2-fold changes and adja-
cent p values (padj) [34]. Customarily, a gene is considered differ-
entially expressed if its adjacent p-value is strictly lower than 0.1.
Pools of up- and downregulated differentially expressed genes are
clustered and visualized into heat maps, generated in R using the
log2-normalized counts and the heatmap.2 method of the gplots
library [35].
3.1.8 Gene Selection and The sheep genome is not yet supported by most gene ontology
Gene Ontology (GO) platforms; therefore, downstream analyses are performed with
orthologs in the human genome Homo sapiens. To select the rele-
vant genes among the upregulated and downregulated genes, per-
form gene enrichment analysis for biological process and molecular
function with ToppGenes and FDR <0.05 [36, 37]. Bar diagram of
significant GO terms ( p < 10–3) is presented on a –Log (P) scale.
Protein-protein interaction networks are generated with the
STRING database [38]. Gene Ontology can also be performed in
parallel with PantherDB and only biological processes be presented
in the pie charts [39].
3.1.9 Validation of The expression profiles of differentially expressed genes related to

RNAseq Data By Real-Time inflammation and iron metabolism should be validated by real-time
Quantitative RT-PCR qRT-PCR. Total RNA (50 ng) is subjected to cDNA synthesis
using a qScript cDNA SuperMix (Quanta BioSciences) at 25 C
for 5 min, 42 C for 30 min, and 85 C for 5 min. The mRNAs of
the selected genes are quantified by qRT-PCR using the AB SYBR
Select MasterMix Kit (Applied Biosystem) with StepOne Plus Real-
Time PCR Systems (Applied Biosystems, V2.2.2). PCR is imple-
mented as per the manufacturer’s protocol. The mRNA relative
expression was calculated by the 2 ΔΔCt method over
housekeeping gene GAPDH compared to baseline or LPS
treatment depending on the experimental design [40]. Sheep-

specific primers are designed with primer3 [41], and GAPDH
primers are designed using Integrated DNA Technologies online
tool.
3.2 Materials and Conduct tissue preparation and cell culture under aseptic condi-
Methods tions; every material used needs to be sterile.
1. Tissue culture T75 or T175 flask.
3.2.1 In Vitro Culture
Protocol 2. 24-Well cell culture plate.
3. Poly-L-lysine (PLL) solution (working concentration of 10 μg/
mL): Dilute PLL stock (1 mg/mL) by 1:100 with ultrapure
water. Prepare and use fresh. Prepare all the solutions and
buffers with ultrapure water (distilled and deionized water to
attain a sensitivity of 18 MΩ-cm at 25 C).
4. Microglia culture media (5% DMEM): DMEM with 5% heat-
inactivated fetal bovine serum (Gibco, Canada Origin), 1%
penicillin/streptomycin, and 1% glutamine. Microglia grow
preferably in DMEM with 5% heat-inactivated fetal bovine
serum. In the flask, microglia are cells of the floating portion;
care should be taken to collect those cells to avoid contamina-
tion with other cell types.
5. Astrocyte culture media (10% DMEM): DMEM with 10%
heat-inactivated fetal bovine serum (Gibco, Canada Origin),
1% penicillin/streptomycin, and 1% glutamine. Astrocytes
grow better in DMEM with 10% heat-inactivated fetal bovine
serum. Astrocytes are the adherent cells that need to be
detached by trypsinization after several passages and washes.
6. Lipopolysaccharides (LPS, Sigma L5024, from E. coli O127,
B8): Reconstitute the lyophilized LPS powder with sterile
saline, keep the stock at +4 C, and prepare LPS working
solution fresh at a concentration of 100 ng/mL.
7. Cell trypsinization solution: Trypsin 0.25% + EDTA 0.1%,
Wisent Cat. No 325-043-EL.
8. Lab-Tek 8-well chamber glass slide (Thermo Scientific).
9. TriZol reagent (Life Technology) or Qiagen RNeasy Mini Kit
(Cat no. 74104).
10. Microglia marker Iba1polyclonal antibody (Wako Chemicals,
Cat# 019-19741).
11. Astrocyte maker glial fibrillary acidic protein (GFAP) antibody
(Sigma).
12. Hoechst 33342 (Sigma): For immunocytochemistry (ICC)
nuclear count staining.
13. AR-R17779 hydrochloride (Tocris Bioscience Cat# 3964), a
selective α7nAChR agonist.
14. α-Bungarotoxin (Tocris Bioscience Cat# 2133), a selective

α7nAChR antagonist.
15. Dimethyl sulfoxide (DMSO, Sigma), vehicle for dissolving
AR-R17779 hydrochloride.
3.2.2 RNAseq Approach: 1. RNA Nano Chip (Agilent RNA 6000 Nano Chips).
RNA Extraction and 2. Agilent 2100 BioAnalyzer (Agilent).
Quantification, RNAseq
3. Illumina TruSeq RNA Sample Preparation v2 kit (Illumina).
Library Preparation
4. Illumina HiSeq2500 (Illumina).
3.2.3 Validation of 1. qScript cDNA SuperMix (Quanta BioSciences).

RNAseq Data by Real-Time 2. AB SYBR Select MasterMix Kit (Applied Biosystem).
qRT-PCR
3. StepOne Plus Real-Time PCR Systems (Applied Biosystems,
V2.2.2).
4 Conclusions
The presented in vivo–in vitro approach (Fig. 1) to manipulating

cholinergic signaling on systems and cellular scales ties together
study of neuroimmunological responses to various stimuli in one
and the same subject during one of the most sensitive periods of
human development when the brain experiences a rapid neuronal
growth and small perturbations may have a profound impact on
phenotype expression during the life span. We propose that this
approach contributes to the study of developmental origins of PNI
phenotype and the mechanisms enabling its manipulation.
Acknowledgments
The authors gratefully acknowledge funding support from the

Molly Towell Perinatal Research Foundation, Canadian Institutes
of Health Research (CIHR), and Fonds de Recherche du Qué-
bec—Santé (FRQS) (to MGF).
The authors wish to thank Esther Simard, Marco Bosa, Carl Ber-
nard, Dr. Lucien Daniel Durosier, Hai Lun Liu, and Carmen
Movila for technical assistance and Jan Hamanishi for graphical
design.
References
1. Fraschini M, Demuru M, Puligheddu M, in pharmaco-resistant epileptic patients who

Floridia S, Polizzi L, Maleci A, Bortolato M, respond to VNS. Neurosci Lett 580:153–157
Hillebrand A, Marrosu F (2014) The 2. Kwan H, Garzoni L, Liu HL, Cao M,
re-organization of functional brain networks Desrochers A, Fecteau G, Burns P, Frasch
MG (2016) VNS in inflammation: systematic autonomic nervous system. Cleve Clin J Med
review of animal models and clinical studies. 76(Suppl 2):S86–S90
Bioelectron Med 3:1–6 16. Que CL, Kenyon CM, Olivenstein R, Macklem
3. Wang X, Rousset CI, Hagberg H, Mallard C PT, Maksym GN (2001) Homeokinesis and
(2006) Lipopolysaccharide-induced inflamma- short-term variability of human airway caliber.
tion and perinatal brain injury. Semin Fetal J Appl Physiol (1985) 91:1131–1141
Neonatal Med 11:343–353 17. Clancy JA, Mary DA, Witte KK, Greenwood
4. Gotsch F, Romero R, Kusanovic JP, Mazaki- JP, Deuchars SA, Deuchars J (2014)
Tovi S, Pineles BL, Erez O, Espinoza J, Hassan Non-invasive vagus nerve stimulation in
SS (2007) The fetal inflammatory response healthy humans reduces sympathetic nerve
syndrome. Clin Obstet Gynecol 50:652–683 activity. Brain Stimul 7:871–877
5. Svedin P, Kjellmer I, Welin AK, Blad S, Mallard 18. Liu RP, Fang JL, Rong PJ, Zhao Y, Meng H,
C (2005) Maturational effects of lipopolysac- Ben H, Li L, Huang ZX, Li X, Ma YG, Zhu B
charide on white-matter injury in fetal sheep. J (2013) Effects of electroacupuncture at auricu-
Child Neurol 20:960–964 lar concha region on the depressive status of
6. Nitsos I, Rees SM, Duncan J, Kramer BW, unpredictable chronic mild stress rat models.
Harding R, Newnham JP, Moss TJ (2006) Evid Based Complement Alternat Med
Chronic exposure to intra-amniotic lipopoly- 2013:789674
saccharide affects the ovine fetal brain. J Soc 19. Ylikoski J, Lehtimaki J, Pirvola U, Makitie A,
Gynecol Investig 13:239–247 Aarnisalo A, Hyvarinen P, Ylikoski M (2017)
7. Yan E, Castillo-Melendez M, Nicholls T, Non-invasive vagus nerve stimulation reduces
Hirst J, Walker D (2004) Cerebrovascular sympathetic preponderance in patients with
responses in the fetal sheep brain to low-dose tinnitus. Acta Otolaryngol 137(4):426–431
endotoxin. Pediatr Res 55:855–863 20. Frangos E, Ellrich J, Komisaruk BR (2015)
8. Burns P, Liu HL, Kuthiala S, Fecteau G, Non-invasive Access to the vagus nerve central
Desrochers A, Durosier LD, Cao M, Frasch projections via electrical stimulation of the
MG (2015) Instrumentation of near-term external ear: fMRI evidence in humans. Brain
fetal sheep for multivariate chronic Stimul 8:624–636
non-anesthetized recordings. J Vis Exp (105): 21. Diz-Chaves Y, Pernia O, Carrero P, Garcia-
e52581 Segura LM (2012) Prenatal stress causes altera-
9. Carmel E, Burns P, Durosier D, tions in the morphology of microglia and the
Duchatellier C, Cao M, Desrochers A, inflammatory response of the hippocampus of
Fecteau G, Frasch M (2012) Fetal brain MRI adult female mice. J Neuroinflammation 9:71
- experiences in the ovine model of cerebral 22. Diz-Chaves Y, Astiz M, Bellini MJ, Garcia-
inflammatory response. Reprod Sci 19 Segura LM (2013) Prenatal stress increases
(3):347A–348A the expression of proinflammatory cytokines
10. Pavlov VA, Tracey KJ (2012) The vagus nerve and exacerbates the inflammatory response to
and the inflammatory reflex—linking immunity LPS in the hippocampal formation of adult
and metabolism. Nat Rev Endocrinol male mice. Brain Behav Immun 28:196–206
8:743–754 23. Slusarczyk J, Trojan E, Glombik K,
11. Pavlov VA, Tracey KJ (2015) Neural circuitry Budziszewska B, Kubera M, Lason W,
and immunity. Immunol Res 63:38–57 Popiolek-Barczyk K, Mika J, Wedzony K,
12. Hosoi T, Okuma Y, Nomura Y (2000) Electri- Basta-Kaim A (2015) Prenatal stress is a vulner-
cal stimulation of afferent vagus nerve induces ability factor for altered morphology and
IL-1beta expression in the brain and activates biological activity of microglia cells. Front
HPA axis. Am J Physiol Regul Integr Comp Cell Neurosci 9:82
Physiol 279:R141–R147 24. Cortes M, Cao M, Liu HL, Burns P, Moore C,
13. Farrokhi CB, Tovote P, Blanchard RJ, Blan- Fecteau G, Desrochers A, Barreiro LB, Antel
chard DC, Litvin Y, Spiess J (2007) Cortagine: JP, Frasch MG (2017) RNAseq profiling of
behavioral and autonomic function of the primary microglia and astrocyte cultures in
selective CRF receptor subtype 1 agonist. near-term ovine fetus: a glial in vivo-in vitro
CNS Drug Rev 13:423–443 multi-hit paradigm in large mammalian brain.
J Neurosci Methods 276:23–32
14. Porges SW (1995) Cardiac vagal tone: a physi-
ological index of stress. Neurosci Biobehav Rev 25. Borovikova LV, Ivanova S, Zhang M, Yang H,
19:225–233 Botchkina GI, Watkins LR, Wang H,
Abumrad N, Eaton JW, Tracey KJ (2000)
15. Porges SW (2009) The polyvagal theory: new Vagus nerve stimulation attenuates the
insights into adaptive reactions of the
systemic inflammatory response to endotoxin. 33. Anders S, Pyl TP, Huber W (2015) HTSeq — a
Nature 405:458–462 python framework to work with high-
26. Cao M, Cortes M, Moore CS, Leong SY, throughput sequencing data. Bioinformatics
Durosier LD, Burns P, Fecteau G, 31(2):166–169
Desrochers A, Auer RN, Barreiro LB, Antel 34. Love MI, Huber W, Anders S (2014) Moder-
JP, Frasch MG (2015) Fetal microglial pheno- ated estimation of fold change and dispersion
type in vitro carries memory of prior in vivo for RNA-Seq data with DESeq2. Genome Biol
exposure to inflammation. Front Cell Neurosci 15(12):550
9:294 35. Warnes GR (2008) Gplots: various R program-
27. Cortes M, Cao M, Liu HL, Moore CS, Duro- ming tools for plotting data
sier LD, Burns P, Fecteau G, Desrochers A, 36. Chen J, Bardes EE, Aronow BJ, Jegga AG
Barreiro LB, Antel J (2017) α7 nicotinic ace- (2009) ToppGene Suite for gene list enrich-
tylcholine receptor signaling modulates the ment analysis and candidate gene prioritiza-
inflammatory and iron homeostasis in fetal tion. Nucleic Acids Res 37:W305–W311
brain microglia. bioRxiv:097295 37. Kaimal V, Bardes EE, Tabar SC, Jegga AG,
28. Shytle RD, Mori T, Townsend K, Vendrame M, Aronow BJ (2010) ToppCluster: a multiple
Sun N, Zeng J, Ehrhart J, Silver AA, Sanberg gene list feature analyzer for comparative
PR, Tan J (2004) Cholinergic modulation of enrichment clustering and network-based dis-
microglial activation by alpha 7 nicotinic recep- section of biological systems. Nucleic Acids Res
tors. J Neurochem 89:337–343 38:W96–W102
29. Frasch MG, Szynkaruk M, Prout AP, 38. Franceschini A, Szklarczyk D, Frankild S,
Nygard K, Cao M, Veldhuizen R, Kuhn M, Simonovic M, Roth A, Lin J,
Hammond R, Richardson BS (2016) Minguez P, Bork P, von Mering C, Jensen LJ
Decreased neuroinflammation correlates to (2013) STRING v9.1: protein-protein interac-
higher vagus nerve activity fluctuations in tion networks, with increased coverage and
near-term ovine fetuses: a case for the afferent integration. Nucleic Acids Res 41:D808–D815
cholinergic anti-inflammatory pathway? J Neu- 39. Mi H, Poudel S, Muruganujan A, Casagrande
roinflammation 13:103 JT, Thomas PD (2016) PANTHER version 10:
30. Durafourt BA, Moore CS, Blain M, Antel JP expanded protein families and functions, and
(2013) Isolating, culturing, and polarizing pri- analysis tools. Nucleic Acids Res 44:
mary human adult and fetal microglia. Meth- D336–D342
ods Mol Biol 1041:199–211 40. Livak KJ, Schmittgen TD (2001) Analysis of
31. Langmead B, Salzberg SL (2012) Fast gapped- relative gene expression data using real-time
read alignment with Bowtie 2. Nat Methods quantitative PCR and the 2( Delta Delta C
9:357–359 (T)) Method. Methods 25:402–408
32. Kim D, Pertea G, Trapnell C, Pimentel H, 41. Untergasser A, Cutcutache I, Koressaar T, Ye J,
Kelley R, Salzberg SL (2013) TopHat2: accu- Faircloth BC, Remm M, Rozen SG (2012)
rate alignment of transcriptomes in the pres- Primer3—new capabilities and interfaces.
ence of insertions, deletions and gene fusions. Nucleic Acids Res 40:e115
Genome Biol 14:R36
Chapter 19
Perinatal Psychoneuroimmunology: Protocols for the Study

of Prenatal Stress and Its Effects on Fetal and Postnatal
Brain Development
Martin G. Frasch, Carlos J. Baier, Marta C. Antonelli,
and Gerlinde A. S. Metz
Abstract
Prenatal stress (PS) impacts early behavioral, neuroimmune, and cognitive development. Pregnant rat
models have been very valuable in examining the mechanisms of such fetal programming. A newer pregnant
sheep model of maternal stress offers the unique advantages of chronic in utero monitoring and manipula-
tion. This chapter presents the techniques used to model single and multigenerational stress exposures and
their pleiotropic effects on the offspring.
Key words DOHaD, Stress, Pregnancy, Animal models, Multi-hit, Generations, Rat, Sheep
1 Introduction
Maternal and paternal prenatal stress represent some of the most

common, widely observed acute and chronically acting insults that
are known to program fetal and postnatal brain development [1]. A
wider spectrum of subtle-to-profound consequences include
behavior, especially emotional control, and neuroimmune function,
notably via programming of glial function which impacts neuro-
genesis and susceptibility to insults in later life [2–6]. Effects of
prenatal stress are highly pleiotropic regarding the timing of the
insult, the developing physiological systems affected, the individual
geno- and phenotype, and the environmental interactions. Conse-
quently, understanding the underlying mechanisms relevant to
human health requires studies in multiple animal models, each
with its unique advantages. We focus here on the maternal contri-
bution of stress experience on the offspring’s phenotype via single,
multi-, and transgenerational exposure and inheritance mechanisms
modeled in pregnant rat and sheep. These animal models are cho-
sen because of their track record in the field. Pregnant rat models
353
are well suited for single- and multigenerational experiments and

neurodevelopmental follow-up [7–9]. Pregnant sheep models
allow extensive and well-tolerated in utero instrumentation, fetal
monitoring, and manipulation which translate into clinically rele-
vant scenarios due to established physiological similarities in brain
development and patterns of injuries [10, 11].
Maternal stress during pregnancy can program physiological
responses, lifetime health trajectories outcomes, and birth out-
comes. In this highly dynamic field of research, most recent revi-
sions on the influence of maternal stress on the offspring’s
developmental outcome agree that the effects are mediated by a
set of stress-transfer mechanisms acting synergistically and affecting
several organs and systems [12, 13]. Among other systems, stress in
the mother may produce cognitive, neurological, and immunolog-
ical development disorders on the infant. Most of these disorders
share genetic risk factors but the onset of the disorders cannot be
explained solely by the inheritance. Therefore, current understand-
ing presumes that the interactions of multiple agents, including
environmental input, lead to a disorder.
The hypothesis most widely accepted is that the exposure to
different influences during in utero and/or postnatal stages of life
affects the development of target organs, disrupting the homeosta-
sis and increasing the risk of adult diseases. Several models have
been proposed stemming in Barker’s hypothesis of Fetal Basis of
Adult Diseases (FeBAD) based on their studies on adult cardiovas-
cular diseases [14], evolving later into the Predictive Adaptive
Response (PAR model) [15], the Developmental Origin of Health
and Disease (DOHaD) [16], and most recently the Developmental
Origins of Behavior, Health and Disease (DOB-HaD) proposed by
Van den Bergh (2011) [17]. Another model named LEARn
(Latent Early life Associated Regulation) considers that each envi-
ronmental exposure is a “hit” acting through induced latent epige-
netic changes [18]. According to this model, disorders develop
according to an “n” number of hits. The first hit is the early
environmental exposure that leads to epigenetic perturbations,
and after a long latency period a second trigger is necessary for
the disease to develop.
Although these models hold true for almost all diseases, partic-
ular attention has been focused on disorders related to the central
nervous system since brain sculpting is related to the conditioning
of the host defense system that depends on communication with
the developing brain [19]. The brain is particularly sensitive to
alterations of the perinatal microenvironment during early develop-
ment, although the consequences of prenatal damage may not
necessarily be apparent until a critical age when neurodevelopmen-
tal defects may be unmasked or precipitated by a subsequent expo-
sure to other insults.
Perinatal Psychoneuroimmunology: Protocols for the Study of Prenatal. . . 355
Notably, stress in early life programs health risks not only across
a single lifetime, but also across multiple generations [9, 20]. For
example, preterm birth risk was suggested to propagate through
generations [21], which is supported by our observation that stress
across generations programs higher vulnerability to preterm birth,
adverse offspring development, and epigenetic markers of human
preterm birth [22].
For example, David Olson and his team [23, 24] assessed
chronic, lifelong stressors and related these to spontaneous preterm
birth by designing a new questionnaire administered to postpartum
mothers who delivered preterm or at term. In these studies, the
combined score of childhood and adult abuse was significantly
associated with higher preterm birth risk [23, 24]. Moreover, the
proportion of women showing shortened gestation gradually
increased with a greater number of adverse childhood experiences.
These findings suggest a close association between adverse early-life
experiences and preterm birth risk, which represents a risk factor for
adverse health outcomes in later life, with psychoneuroimmunolo-
gical mechanisms thought to play a key mediating role [25]. Indi-
vidual psychoneuroimmunological traits are in turn influenced by
preterm birth which informs a transgenerational chain of events
[26, 27].
In the following, we present the pregnant rat and sheep models
of single-, trans-, and multigenerational prenatal stress.
2 Single-Generational Models of Prenatal Stress
2.1 Pregnant Rat For the last 20 years, the focus of our research has been centered in
Model of Prenatal the study of the outcomes of various neurotransmitter pathways
Stress and hormones in a rodent model of prenatal restraint stress trig-
gered by the hypothesis that stressful situations suffered prenatally
2.1.1 Overview are related to the propensity to develop psychiatric abnormalities in
the human adult life. Results obtained along these years showed
impairments at the biochemical, morphological, and behavioral
levels in the offspring of gestationally stressed dams and reversion
by a cross-fostering model.
In a brief summary of the results obtained along these years
with the single-generation rodent model of PS, we would like to
highlight the following results: (1) PS impairs dopamine metabo-
lism throughout development of the male offspring: In a model of
PS in pregnant rats, induced by restriction of movement and
subsequent cross fostering, PS increased DA D2 receptors in limbic
areas, decreased DA-stimulated release in cortical areas, and
impaired the expression of specific transcription factors during
development as well as the expression of TH and transporters. PS
altered the asymmetry in D2-type receptors in the nucleus accum-
bens, an area associated with impulsivity [28–33]. (2) PS impairs
the hypothalamic-pituitary-gonadal axis in male offspring: PS

impairments of DA metabolism were differentially affected after
puberty, suggesting that perinatal events might render the DA
circuitry more vulnerable to puberty variation of the hormonal
circulating hormones. PS induces long-term imbalance of male
sexual hormone concentrations in serum, advanced spermatogene-
sis development, and age-dependent misbalance on alpha-receptor
expression on PFC and HPC brain areas [34, 35]. (3) PS alters
behavioral and morphological parameters in offspring: A direct
relation was found between the anxiety states and levels of benzo-
diazepine (BDZ) receptors. At the morphological level, we found
reduced dendritic arborization and astroglial hypertrophy with
synaptic loss suggesting a possible alteration of glutamate metabo-
lism. Glutamate transporters were altered in frontal cortex and
hippocampus of PS offspring [36, 37].
We also found that the expression of the gene gpm6a encoding
for M6a membrane glycoprotein is increased in hippocampal rat
brain areas of prenatally stressed offspring. Also, the CpG island
promoter region of the same gene showed a lower percentage of
methylation in rats subjected to EP compared to control. Since
there is a correlation between the degree of methylation of the
CpG island and the expression of gmp6a, a gene previously identi-
fied to be modulated by stress, it is postulated that prenatal stress
may regulate the expression of gpm6a through epigenetic
mechanisms [38].
We also developed a “double hit” protocol to study whether
the stress suffered in utero (first hit) affects the vulnerability of the
nigrostriatal dopaminergic neurons to degenerate after an intras-
triatal 6-OHDA injection in adult life (second hit) [28]. Employing
this model we demonstrated that prenatally restrained stressed rats
showed an increased number of TH+ cells in the VTA (PS–sham)
and that a neurochemical insult during adulthood (PS–6-OHDA)
reduced this population to the levels of control cells (control–-
sham/6-OHDA) [39]. Additionally, the population of nNOS
(neuronal nitric oxide synthase) neurons was quantitatively differ-
ent between control- and PS–sham rats in two areas corresponding
to the mesolimbic system: nucleus accumbens core (NAcC) and
ventral tegmental area (VTA). Moreover, we found no alterations in
the number of TH+- or nNOS+-expressing cells in the nigrostriatal
system. These results were in agreement with previous results from
our laboratory showing that prenatal restraint stress affected mostly
limbic areas of the offspring rather than motor areas [40].
Taken together, the results would suggest that in an animal
model, the insults received during intrauterine life are critical in the
development of biochemical and behavioral responses in adults, as
well as crucial maternal care in the first weeks of life. In the brain,
these changes are more evident in limbic areas than motor, antici-
pating increased vulnerability in pathologies associated with
behavior and emotionality. These results suggested to us that this

model of gestational stress could provide clues to understanding
the mechanisms by which an insult in early life would be contribut-
ing to the breakdown of the balance in neurotransmission and the
formation of aberrant cortical connections which would entail the
establishment of abnormal cognitive behaviors, and, in turn, pro-
vide clues to develop predictive biomarkers of translational value.
2.1.2 Model Design Vaginal smears were collected daily for 8 days before mating to
determinate the stage of the estrus cycle and the day of conception.
On the day of proestrus, sexually experienced male Wistar rats
weighing 250–300 g were introduced for mating. Vaginal smears
were taken on the following morning. The day on which sperma-
tozoa were found in the smear was designated day 1 of pregnancy.
A constant light/dark cycle (on at 06:00 h, off at 18:00 h) and the
temperature of 21–25 C were maintained.
Prenatal Stress. Pregnant dams are randomly assigned to either
the control (C) or the prenatal stress (PS) group and individually
housed with ad libitum access to standard rat chow and water. C
rats are left undisturbed in the home cage, while PS dams are
subjected to a restraint stress procedure, which involves rats being
transferred to an experimental room where the stressor is applied.
Pregnant females are individually placed into a transparent plastic
restrainer fitted closely to body size for three 45-min periods per
day (09:00, 12:00, and 16:00 h) between days 14 and 21 of preg-
nancy. The restrainer has ventilation holes, and dimensions appro-
priate for a pregnant rat of 350 g: internal diameter 64 mm, and an
adjustable length of 149–208 mm. This type of stressor is chosen
because it has an indirect influence on the fetuses via a direct stress
on the mother. The sessions are performed in a lit environment. No
other subjects should be present in the experimental room during
the stress exposure.
Postnatal Procedures. On the day of parturition, litter charac-
teristics are recorded and litters are culled to ten pups, maintaining
similar number of males and females, when possible. Weaning is
performed at postnatal day (PND) 21. The male and female off-
spring are housed in separated cages, with no more than five rats per
cage, and in standard housing conditions. To avoid litter effects,
1–2 pups from each litter (a female and a male) should be tested for
each experiment. Behavioral testing is conducted at that period
using the elevated plus maze and light/dark box tests for anxiety-
like behaviors and sucrose anhedonia test for depressive-like behav-
ior. Brains are collected after decapitation at the end of behavioral
tests at PND 35 or another date depending on the milestones to be
assessed in the given study.
6-OHDA Lesion Procedure. Control and prenatal stressed males
at postnatal day (PND) 75 of age (average weight * 320 g) are
injected with desipramine (25 mg/kg i.p.) 1 h prior to surgery to
protect noradrenergic pathways against 6-OHDA neurotoxicity.

Fifty minutes later, animals are anesthetized with a ketamine/xyla-
zine mixture (50 mg/kg and 8 mg/kg, respectively) and placed in a
small animal stereotaxic apparatus. Following incision of the scalp, a
small drill hole is made in the skull and a pulled-glass injection
cannula is slowly inserted into the right striatum at the following
stereotaxic coordinates: 0.5 mm anterior and 3.5 mm lateral to
bregma, and 5.0 mm ventral from dura mater. Following this,
6-OHDA hydrobromide (8.75-lg of 6-OHDA in 2-lL of 0.9%
NaCl containing 0.02% ascorbic acid) is then injected [41]. The
cannula is left in place for 5 min at the end of the injection to
prevent reflux and to allow for toxin diffusion. Control and prena-
tally stressed animals which received an ipsilateral intrastriatal injec-
tion of the vehicle alone (2 uL of 0.9% NaCl containing 0.02%
ascorbic acid) serve as the sham groups. The scalp wound is then
sutured, and the animals are housed singly until sacrifice 28 days
later. As a result, four experimental groups are obtained: (1) con-
trol–sham (control rats injected with vehicle), (2) PS–sham (prena-
tally stressed rats injected with vehicle), (3) control–6-OHDA
(control rats injected with 6-OHDA), and (4) PS–6-OHDA (pre-
natally stressed rats injected with 6-OHDA) [39].
2.1.3 Materials and Materials

Methods
1. Virgin females Wistar rats weighing 250 g.
Prenatal Stress Protocol in 2. Sexually experienced male Wistar rats weighing 250–300 g.
Pregnant Rat (Single- 3. Transparent plastic restrainer: The restrainer has ventilation
Generation Stress) holes and dimensions appropriate for a pregnant rat of 350 g:
internal diameter of 64 mm with an adjustable length of
149–208 mm. This type of stressor is chosen because it has
an indirect influence on the fetuses via a direct stress on the
mother.
Methods
Mating Procedure
1. A maximum of four rats are housed per cage with ad libitum
access to standard rat chow. A constant light/dark cycle (on at
06:00 h, off at 18:00 h) and the temperature of 21–25 C are
maintained during the entire protocol. All procedures should
agree with the standards for the care of laboratory animals as
outlined in the NIH Guide for the Care and Use of Laboratory
Animals. Take care to minimize the number of rats used.
2. Collect vaginal smears daily for 8 days before mating to deter-
minate the stage of the estrus cycle and the day of conception.
3. On the day of proestrus, introduce for mating sexually experi-
enced male Wistar rats.
4. Take vaginal smears on the following morning.

5. The day on which spermatozoa is found in the smear is desig-
nated day 1 of pregnancy.
Prenatal Stress
1. Designate female pregnant rats randomly to either the control
(C) or the prenatal stress (PS) group, and individually house
them with ad libitum access to standard rat chow and water.
2. C rats are left undisturbed in the home cage.
3. PS dams are subjected to a restraint stress procedure, which
involves rats being transferred to an experimental room where
the stressor is applied.
4. Then, PS pregnant females are individually placed into a trans-
parent plastic restrainer fitted closely to body size for three
45-min periods per day (09:00, 12:00, and 16:00 h) between
days 14 and 21 of pregnancy. The sessions are performed in a lit
environment. No other subjects should be present in the exper-
imental room during the stress exposure.
Postnatal Procedures
1. On the day of parturition, litter characteristics are recorded and
litters are culled to ten pups, maintaining similar number of
males and females, if possible.
2. Weaning is performed at postnatal day (PND) 21.
3. The male and female offspring are housed in separated cages,
with no more than five rats per cage, and in standard housing
conditions.
4. To avoid litter effects, 1–2 pups from each litter (a female and a
male) should be tested for each experiment.
B. 6-OHDA Lesion Procedure
Materials
Intraperitoneal Injection (i.p.)
1. 1 mL Disposable syringes.
2. Desipramine solution: Prepare a 13.33 mg/mL desipramine
solution in sterile saline (0.9% NaCl).
3. Ketamine/xylazine mixture for injection is prepared in sterile
saline: 35.6 mg/mL and 5.45 mg/mL, respectively, from
commercial ketamine (50 mg/mL) and xylazine (20 mg/mL)
solutions.
6-Hydroxydopamine (6-OHDA) Lesion
1. Ascorbic acid solution: Prepare an ascorbic acid solution 100,
i.e., 20 mg/mL or 2% in sterile saline, and then ascorbic acid
solution 1: 10 μL from ascorbic acid solution 100 and
990 μL of saline. Solutions should be freshly prepared.
2. 6-Hydroxydopamine (6-OHDA) solution: Prepare a 6-OHDA

solution of 4.375 mg/mL in saline containing 0.02% ascorbic
acid prepared as described above. Solutions should be freshly
prepared and protected from exposure to light. 6-OHDA is
light sensitive.
3. Pulled-glass injection cannulas.
4. Animal stereotaxic apparatus.
5. Surgical drill.
6. Microsyringe pump.
Methods
Desipramine Administration
1. One hour prior to 6-OHDA lesion, weigh each rat and record
the weight.
2. Administer desipramine at 25 mg/kg by intraperitoneal injec-
tion. Maximum volume injected is 0.7 mL. Desipramine is
administered to protect noradrenergic pathways against
6-OHDA neurotoxicity.
Anesthesia and Animal Preparation
1. Fifty minutes later, anesthetize the rats with ketamine/xylazine
mixture (50 mg/kg and 8 mg/kg, respectively) by intraperito-
neal injection. Maximum volume injected is 0.7 mL.
2. Control the complete animal anesthesia analyzing that the
palpebral and corned blink reflexes, as well as the pedal reflex,
are lost.
3. Shave the top of the rat’s head and clean the head of the animal
with povidone-iodine, 10% solution.
4. Place the rat in the stereotaxic apparatus.
5. Make an incision along the midline of the skin on top of the
skull with a scalpel. Dry/clean the area with gauze.
6-OHDA Lesion Procedure
1. Make a small drill hole at the appropriated coordinates: 0.5 mm
anterior and 3.5 mm lateral to bregma.
2. Load the pulled-glass injection cannulas with the 6-OHDA
solution. Carry out the tasks protected to the light to prevent
6-OHDA degradation.
3. Slowly insert the glass cannula (100 μm/s) into the right
striatum at the following stereotaxic coordinates: 0.5 mm ante-
rior and 3.5 mm lateral to bregma, and 5.0 mm ventral from
dura mater.
4. Following this, inject 6-OHDA (8.75 μg of 6-OHDA in 2 μL
of 0.9% NaCl containing 0.02% ascorbic acid) via a
microsyringe pump at a rate of 1 μL/min. Sham rats are

injected with 2 μL of 0.9% NaCl containing 0.02%
ascorbic acid.
5. Upon completion of administration of 6-OHDA or vehicle,
leave the cannula in the place for 5 min at the end of the
injection to prevent reflux and to allow for toxin diffusion.
6. Slowly move away the glass cannula.
7. Suture the scalp.
Post-6OHDA Lesion Animal Care
1. Remove the rat from stereotaxic apparatus, and place it in a
recovery cage until the animal regains consciousness.
2. House the rats in standard cages after complete anesthesia
recovery, with no more than five rats per cage. Provide with
standard rat chow and water ad libitum.
3. Until 28 days later 6OHDA lesion the animals are used for
histological and/or biochemical analysis.
2.2 Pregnant Sheep The protocol has been established and validated by Schwab et al.
Model of Prenatal [10, 12, 42, 43]. The following protocol represents a minimalistic
Stress version for the specific experiments to be designed on its basis.
Time-dated pregnant sheep (singletons or twins) are delivered to
the research facility and allowed ~20 days for adaptation and quar-
antine (to check Q fever status as per local requirements—a condi-
tion that may vary across the jurisdictions). The acclimation time
also minimizes any maternal stress due to transportation and new
environment prior to commencing the experiments. During a ster-
ile instrumentation at ~102 dGA, the ewes are equipped with a
venous catheter while the fetus receives precordial ECG leads
sutured in the skin, two arterial and one venous catheters, and an
amniotic catheter exiting via the uterus to the maternal flank and
secured in a pouch together with telemetric transmitter, following a
protocol established in our laboratory [11]. Note that while telem-
etry is advantageous in behavioral studies and allows for longer
periods of data acquisition in social housing with animals undis-
turbed, a completely wired solution to instrumentation is also
possible. All tubing resides in a closed, sterile loop to prevent
bacterial contamination from the mother’s skin or the environ-
ment. The instrumentation procedure is followed by 3–4 days of
postoperative recovery.
2.2.1 Maternal Stress The animals are randomly assigned to control or maternal stress
Protocol groups. The stress group follows a repeated maternal stress proto-
col between 105 and 135 dGA [10]. Maternal stress is induced by
isolating pregnant ewes in a well-lit box (3.0 m 3.0 m 1.4 m,
w l h) with no visual, tactile, or auditory contact with flock-
mates. During isolation, ewes have no access to food or water. Ewes
are isolated for 2 days per week (Monday to Friday) with at least
2 days’ recovery time between the particular isolation bouts. Dura-
tion of isolation is 3 h. Isolation is performed either between 07:00
and 10:00 h, 11:00 and 14:00 h, or 15:00 and 18:00 h. Day of the
week and time of isolation as well as isolation box are changed
randomly within the mentioned parameters to reduce habituation.
Two maternal and fetal blood samples are collected on the day
when the stress procedure is performed from the chronically
installed jugular vein and fetal arterial brachial catheters, respec-
tively: before and during isolation to estimate maternal and fetal
cortisol concentrations and degree of habituation. The sample
before the first isolation on the experimental day 1 (at 105 dGA)
serves as animal’s own baseline sample. There will be no additional
blood sampling on days between the isolations. Note that the
ability to sample fetal and maternal blood consecutively within
each animal reduces the sample size required or increases the
power of the study, respectively. This should be considered during
sample size estimations.
Blood samples in non-stressed (control) animals are taken at
the same times as in stressed animals. Maternal and fetal ECG, fetal
blood, and amniotic pressures are recorded telemetrically and con-
tinuously throughout the experiment.
2.2.2 Necropsy and Fetuses are delivered electively on 136 dGA. The ewe is killed with
Postmortem Analyses overdose of Euthanyl. The lamb can be either killed for histological
studies or weaned and monitored postnatally, as per specific proto-
col. The lamb brains are extracted and prepared into the regions of
interest for later processing. The approach will depend on the
specific study. One option is to proceed as follows. One hemi-
sphere’s brain regions of interest such as hypothalamus, hippocam-
pus, and prefrontal cortex tissues are split to be snap frozen to
perform DNA methylation analyses and for RNAlater storage to
conduct targeted or high-throughput RNA analyses. Another
hemisphere’s same brain regions are immersion-fixed and used to
perform immunohistochemistry. The remaining brain regions of
each hemisphere can be snap-frozen or immersion-fixed, respec-
tively, for future analyses.
2.2.3 Sample and Data The animals are kept together in dens with ad libitum access to food
Acquisition and water for 30 days with intermittent maternal and FHR record-
ing and sampling for maternal and fetal blood gases, metabolites,
and stress hormones. Maternal blood samples are collected from
the chronically installed jugular vein catheter regularly before and
during isolation to estimate maternal cortisol concentrations and
degree of habituation. Along with blood gas, glucose, and lactate
measurements, plasma samples can be stored at 80 (after centri-
fuging at 4000 rpm, 4 min, 4 Celsius) for later measurements of
maternal and fetal hormones such as cortisol using commercial

ELISA. Maternal and fetal ECG, and fetal arterial blood and amni-
otic pressures, is recorded telemetrically and continuously through-
out the experiment at 1000 Hz and 256 Hz, respectively.
2.2.4 Heart Rate Analysis ECG-derived heart rate (HR) and HR variability (HRV) analyses of
as a Gateway to Measuring mother and the fetus may contain signatures of stress hidden in the
Stress Exposure subtle fluctuations of HRV on millisecond timescale. To reveal such
properties of HRV, a sampling rate of 1000 Hz is required [44, 45].
HRV classifies stressed versus non-stressed elephants [46]. The
mediating mechanism by which stress exposure alters HRV proper-
ties is thought be via an activation of autonomic nervous system
(ANS).
Corticosteroid administration during pregnancy has shown to
affect the autonomic balance in utero [47–50]. This effect is tran-
sient but repeated fetal administration of betamethasone alters
nervous system maturation. There was no association between
maternal cortisol and infant DNA methylation suggesting that the
effects of maternal depression may not be mediated directly by
glucocorticoids; instead, sympathetic nervous system activity, a
component of the fetal ANS, may be the mediating pathway [51].
Aside of the sympathetic nervous system, another key contrib-
utor to the ANS activity is the vagus nerve. Fetal HRV reflects
maturation and activation of the vagus branch of the ANS involved
in sensing and control of fetal acidemia, hypoxia, and inflammation
[52–54]. In human cohorts and in fetal sheep model of human
labor and fetal inflammation, multidimensional FHR variability
analysis predicts birth outcomes [45, 54–57].
The vagus nerve influences brain function and body metabo-
lism in a pleiotropic manner [58, 59]. A new field of bioelectronic
medicine is emerging. It aims to devise therapeutic approaches
using vagus nerve stimulation (VNS) to modify the endogenous
salutory signaling of the vagus nerve [60–62]. VNS reduced sym-
pathetic tone, stress-induced anxiety behaviors, and depression
symptoms in animal models and in clinical studies [63–69]. VNS
is thought to facilitate tonic inhibition of the basolateral amygdala
by the infralimbic region of the medial prefrontal cortex, which
results in reduced fear response [63]. VNS increases CRH expres-
sion in hypothalamus [70] and CRH receptor 1 agonism increases
vagal modulation of HRV [71–73]. This reciprocal CRH–vagus
nerve circuitry provides an important diagnostic and therapeutic
link between stress and ANS.
A more integrative dimension within which to gauge the phys-
iological role of HRV involves theories such as Porges’ polyvagal
theory [74, 75] and concepts whereby HRV is seen as a proxy
signature of nonlinearly coupled interacting homeostatic (or,
rather, homeokinetic [76]) physiological systems. Within such
bidirectional autonomic and neuroendocrine interactions, informa-

tion is exchanged between the central nervous system and the
periphery which in turn facilitates the expression of the emotional,
autonomic, hormonal, and immune responses [77].
We expect that future studies in pregnant sheep model of
human stress exposure during fetal and postnatal development
will reveal novel HRV biomarkers of stress signature which will be
validated in FHR monitors. Novel therapeutic modalities to coun-
teract the lasting effects of prenatal stress exposure can be expected
from the studies refining the VNS regimens.
2.2.5 Prenatal Stress The general framework of the protocol has been published in [11]
Protocol in Pregnant Sheep and is freely accessible via the PubMed Central. This paper reports
the precise materials and steps needed for the pregnant sheep
model approach.
The stress behavioral component involves animal isolation, as
described, and is very straightforward. Two notes of caution are
worth sharing. First, as with all freely moving animal models, here it
is also important to ensure that all catheters and wires are properly
organized and secured around the animal to avoid accidental pull-
ing, chewing, or tearing of a catheter or wire.
For HRV analysis, we strongly recommend resisting the temp-
tation of using off-the-shelf software products without consulting
or, ideally, collaborating with HRV experts on the experimental
design, ECG data acquisition, and HRV analysis. Some papers fall
into this trap and produce data of doubtful validity. Two frequently
observed misconceptions should be stressed. First, HRV is not a
good source of biomarkers for sympathetic modulations of sinus
node activity. HRV is best suited to reflect more integrative, com-
plex fluctuations of ANS activity which can render excellent bio-
markers. Second, HRV does not reflect a sympathetic or vagal tone;
it may reflect fluctuations of vagal tone, but the complex mélange of
sympathetic and vagal rhythm fluctuations is difficult to untangle
through HRV alone, with some exceptions which require
specialized approach to HRV analysis not available in off-the-shelf
software [78].
VNS outside of epilepsy treatment is a nascent field with little
standardization, so a thorough literature search is advised prior to
deciding on a stimulation regimen. We refer the interested reader
to a recently published meta-analysis of the VNS approaches [62].
3 Trans- and Multigenerational Models of Prenatal Stress
3.1 Overview Since epigenetic marks are potentially heritable, the use of transge-
nerational animal models is uniquely suited to reveal mechanisms of
inheritance of epigenetic, metabolomic, and phenotypic traits.
These models offer the unique opportunity to investigate chronic
stress exposure of human populations, such as a family history of

war or poverty. For example, in a previous study we exposed preg-
nant female Long-Evans rats to transient stress [22]. Their gestat-
ing F1 daughters were again exposed to transient stress or they
remained as non-stress controls. Their gestating F2 granddaughters
were again exposed to transient stress or remained as non-stress
controls, thus generating two lineages: one of repeated, multigen-
erational stress and the other in which stress was limited to F0
parental exposure. With each generation, stress in both lineages
gradually reduced gestational length, maternal weight gain, and
maternal behavioral activity, and increased the risk of gestational
diabetes [22]. Thus both trans- and multigenerational stress expo-
sure altered gestational length, pregnancy outcomes, and offspring
health. Delayed offspring development was recognizable as early as
postnatal day 7, with the greatest effect in the F3 offspring whose
F0 parental generation experienced stress only [22]. Although
context-dependent programming occurs, a truly heritable pheno-
type became evident in the F3 generation. Thus, stress imposed on
the great-maternal generation seems to have been passed on, via the
gametes, to impede development [79].
The striking phenotype in the F3 generation suggests genuine
transgenerational epigenetic inheritance whereby the epigenetic
modifications have been passed via the gametes that have escaped
reprogramming [79, 80]. Indeed, phenotypic findings in behavior
and physiology were supported by molecular changes involving
epigenetic regulation of gene expression. Our study indicated that
the multigenerational stress altered microRNA (miRNA) expres-
sion patterns in brain and uterus of F2 mothers, including the
miR-200 family. In particular, stress led to upregulation of
miR-200b and simultaneous downregulation of miR-429. In the
uterus, both miR-200b and miR-429 were suggested to modulate
gestational length through interaction with their gene targets
Stat5b, Zeb1, and Zeb2 [81]. The data indicated that when upregu-
lated, miR-200b may suppress Stat5b, Zeb1, and Zeb2 mRNA levels
in the lineage that exposed each generation of pregnant dams to
prenatal stress (i.e., the F2 offspring from stressed F0 and F1
generations). Furthermore, stress also increased miR-181a expres-
sion in the placenta. miR-181a has been associated with preterm
birth in humans and may serve as a marker of shortened gestation
[82]. These findings suggest that the mechanisms involved in the
timing of parturition and associated behavioral and physiological
signatures may be programmed through the maternal lineage. The
identification of epigenetic signatures of preterm birth in clinically
accessible tissues, such as placenta, offers the potential for predic-
tive and preventive studies related to poor pregnancy outcomes.
It is possible that multigenerational stress may promote resil-
ience to repeated stress [83, 84]. On the other hand, it is reasonable
to expect that exposure to repeated prenatal stress throughout the
lineage may have cumulatively influenced a phenotype [85]. To test

the latter, multiple hit models of stress have become useful. The
multiple hit theory by Daskalakis [86] proposes that vulnerability
to disease is enhanced with three hits, while each hit cumulatively
compromises the ability to cope with adversity. The three hits may
include (epi)genetic predisposition existing at the time of concep-
tion, perinatal environment, and later-life environment. The pre-
and postnatal environments both include an epigenetic component
through gene x environment interactions. These susceptibilities, as
suggested by Saban et al. [87], may move the individual closer
toward disease vulnerability. Thus, cumulative impacts of lifetime
stresses may compromise resilience or ability to cope with accumu-
lating physiological challenges. Consequently, as different stresses
are layered over a genetic and epigenetic inheritance, the risk for
adverse health outcomes increases.
In a line of recent studies, we demonstrated that a family
history of recurrent ancestral prenatal stress partially improves
movement performance, and that these benefits are abolished by
the experience of a second stressor, such as a focal brain injury in
adulthood [88]. The results suggest that in prenatally stressed
individuals, the experience of a brain injury or other traumatic
events may represent a second hit in a system afflicted by a dysfunc-
tional HPA axis [88]. In addition, inflammatory and psychological
stresses have additive effects on birth outcomes and lifetime health
trajectories [85]. These observations support the multiple hit the-
ory postulated by Daskalakis [86].
While the exposure of ancestors to stress is not controllable, the
recent focus on ancestral stress has shifted toward developing effec-
tive interventions that mitigate the consequences of cumulative
stress. Some attempts used dietary changes [89] while others
demonstrated that environmental enrichment (EE) represents an
effective and clinically valid therapy to mitigate neurohormonal,
behavioral, and epigenetic signatures of stress [90–92]. For exam-
ple, EE modulates HPA axis activity, stress-related behavioral pro-
files, and essential immune functions likely through epigenetic
regulation of gene expression [90–92]. Hence, beneficial endocrine
and behavioral changes induced by EE may be associated with
epigenetic reprogramming throughout the life span, thus poten-
tially overcoming ancestral and early-life adverse endocrine pro-
gramming. The epigenetic pathways, in particular miRNAs,
provide the opportunity to identify heritable predictive markers of
adverse pregnancy outcomes and potential therapeutic targets
[9, 93]. In these studies, the experimental design usually exposes
ancestrally stressed animals to an EE housing condition that resem-
bles a clinical intervention of an active and healthy lifestyle.
EE intervention represents a noninvasive therapy that produces
robust changes in neuronal morphology, endocrinology, and
behavior. EE usually consists of housing animals in an environment
that provides them with toys and a diverse diet to generate rich
cognitive, social, motor, and sensory stimulation. EE was first
studied in the 1940s by the Canadian neuropsychologist Donald
O. Hebb who brought laboratory rats to his home. He noted that
rats housed as pets in his home performed better on memory tasks
compared with rats reared in standard laboratory conditions [94].
Moreover, when reared in an EE aging rats showed thicker frontal
and occipital cortices [95]. A recent study [96] expanded these
findings by showing that social enrichment has a significant impact
on neuroplasticity and stress resiliency particularly in females. Fur-
thermore, the benefits of maternal social enrichment could be
transmitted to their unexposed female descendants [96]. The
experiments raised two generations (F0 and F1) of male and female
rats in standard and social housing conditions which were examined
for neurohormonal and molecular alterations along with changes in
four behavioral modalities [96]. The findings revealed that in addi-
tion to higher neuronal density and thickness of the prefrontal
cortex, social experience reduced the hypothalamic-pituitary-adre-
nal (HPA) axis activity in F0 rats and their F1 nonsocial housing
offspring born to social mothers [96].
The trans- and multigenerational prenatal stress models in rats
as outlined above allow studies of epigenetic mechanisms at DNA
and miRNA levels as well as of downstream cell metabolic signa-
tures in organs such as brain and blood. The discovery of epigenetic
and metabolic signatures of ancestral stress may help identify the
mechanisms of disease programming. It is likely that adult complex
disease is programmed by transgenerational inheritance of adverse
life events, and associated epigenetic markers. Thus, finding epige-
netic linkages of inter- or transgenerational programming of disease
risk may lead to new prognostic markers and preventive strategies.
These strategies are central to the development of personalized,
preventive medicine based on epigenetic signatures.
3.2 Model Design Common rat models of transgenerational inheritance use gesta-
tional restraint stress during the second half of pregnancy. Behav-
3.2.1 Trans- and ioral and neuroendocrine assessments are usually performed in the
Multigenerational Stress parental generation (F0), their offspring (F1), grand-offspring
(F2), and great-grand-offspring (F3), for both males and females.
Truly heritable epigenetic marks can be found in the F2 generation
of the male lineage and in the F3 generation of the female lineage
[80, 97]. In this model, transgenerational stress refers to stress in
only the parental generation (S in F0, SN in F1, SNN in F2, SNNN
in F3; N: no stress), and multigenerational stress refers to continu-
ous experience of maternal stress through all generations (S in F0,
SS in F1, SSS in F2, SSSS in F3) [22, 98]. Controls will consist of
non-stressed, handled animals (N, NN, NNN, NNNN). To care-
fully standardize conditions across generations, personal breeding
colony room in a dedicated facility is advised.
3.2.2 Postnatal The effectiveness of stress is determined by recording body weight,

Development food intake, postmortem brain and adrenal weight, and blood
serum markers. Litter size and sex ratios are recorded prior to
culling each litter to n ¼ 10 or less to standardize litter size effects
on maternal care and offspring development. Infrared surveillance
systems for continuous cage site recordings allow the observation
of maternal care and offspring behavior around the clock.
3.2.3 Behavior Through continuous cage-site recordings along with formal behav-
ioral assessments a wealth of behavioral data can be gained. Cage-
site observations may include prepartum maternal behavior, mater-
nal care (nest building, grooming, stereotypy, circadian rhythm,
pup retrieval), and maternal defense on lactational day (LD) 5
toward a virgin intruder female approaching the nest area. A com-
prehensive array of formal tests may assess offspring development
(P7 infantile, P21 juvenile) in sensorimotor and physical features
(skilled walking, vestibular and proprioceptive functions, body
weight, eye opening, play). Measures of maturation at P60 (adoles-
cence) and adult behavior (P90–18 months) may include five clas-
ses of behavior: (1) skilled movement (reaching, ladder rung
walking); (2) sensorimotor function (adhesive removal, von-Frey
hair test); (3) learning and memory (water maze, Ziggurat task,
Go-No Go decision-making); (4) social behavior (play fighting,
hierarchy, aggressive, defensive, and submissive behavior); and
(5) depression- and anxiety-like behavior (light/dark test, elevated
plus maze, open field, open table exploration).
3.2.4 Physiology Neuroendocrine measurements usually involve the analysis of

blood and brain samples. Blood samples can be collected rapidly
via tail vein nick or with inhalation anesthesia on days without
behavioral tests. Common measurements include blood glucose
to determine hyperglycemia and dehydration, stress plasma markers
CRH, CORT, ACTH, and sex hormone levels of estrogen and
testosterone. Puberty, estrous cycle (weekly vaginal swabs and
Shorr staining), and number of breeding sessions further determine
female stress response in the breeding cohort. Multiple comple-
mentary behavioral, physical, and endocrine hallmarks reflect the
complex manifestations of transgenerational stress across the life
span. HPA axis activity is closely linked to GR and MR density in
the hypothalamic paraventricular nucleus and hippocampus.
Immunocytochemical and stereological analysis of brain tissue
may include GR, MR, CRH, and oxytocin receptor density and
intracellular location may be determined to assess the stress
response. General morphology (cortical thickness, volume, neuron
density) can be assessed for central structures involved in stress
response and affective disorders, such as prefrontal cortex, hippo-
campus, basolateral amygdala, and paraventricular hypothalamus.
3.3 Prenatal Stress 1. Animal characteristics: 100–150-day-old timed-pregnant rats.

Protocol in Pregnant 2. Behavioral and neuroendocrine assessments are performed in
Rat (Multi- and the parental generation (F0), their offspring (F1), grand-
Transgenerational offspring (F2), and great-grand-offspring (F3), for both
Stress): Materials and males and females.
Methods 3. Generating a maternal lineage of trans- and multigenerational
prenatal stress:
(a) Pregnant dams of the F0 generation need to be exposed to
external stress from GD12 to 18 (F0-S). External stress
usually consists of a semi-random sequence of daily
restraint stress (20 min) and forced swimming (5 min)
from gestational day (GD) 12–18. To carefully standard-
ize conditions across generations, a personal breeding
colony room in a dedicated facility is advised. Stress by
shipment of animals or environmental factors should be
avoided.
(b) One stressor is given in the morning hours, and the other
one in the afternoon. In this mode, the pregnant daugh-
ters and granddaughters are treated according to two
different protocols: S ¼ stress, N ¼ non-stressed control
conditions.
(c) Transgenerational stress (TPS): Prenatally stressed females
F1-S are bred with naı̈ve F1-N males to produce F2-SN
and F3-SNN pups. The F2 and F3 dams do not undergo
stress. This creates a truly transgenerational model in
which the F3 generation is not directly exposed to stress
[97].
(d) Multigenerational stress (MPS): Stress is applied to each
generation (S, SS, SSS, SSSN). Generations of
non-stressed rats (F0-F3) serve as control.
4. Phenotype assessment. Postnatal development.
(a) Measure the effectiveness of stress by changes in body
weight, gestational length, food intake, postmortem
brain and adrenal weight, uterine implantation sites, and
blood serum markers of stress.
(b) Record litter size, pup weight, and sex ratios prior to
culling each litter to n ¼ 10 or less to standardize litter
size effects on maternal care and offspring development.
5. Maternal behavior.
(a) Use infrared cameras to continuously record cage sites
along with formal behavioral assessments. Cage-site obser-
vations may include prepartum maternal behavior, mater-
nal care (nest building, grooming, stereotypy, circadian
rhythm, pup retrieval), and maternal defense on lactational
day (LD) 5 toward a virgin intruder female approaching

the nest area.
6. Offspring development.
(a) P7 infantile and P21 juvenile stages are assessed for senso-
rimotor and physical features: skilled walking, vestibular
and proprioceptive functions, body weight, eye
opening, play.
(b) P7–60 should assess
l Affective, social, and sensorimotor functions: vestibu-
lar and proprioceptive functions, skilled movement,
play, anxiety-like
l Open-field exploratory behaviors and physical features:
body weight, eye opening, tooth eruption
(c) Assess adult behaviors by
l Skilled movement (reaching, ladder rung walking)
l Learning and memory (water maze, Ziggurat task,
Go-No Go decision-making)
l Social behavior (play fighting, aggressive, defensive,
and submissive behavior)
l Depression- and anxiety-like behavior (light/dark test,
elevated plus maze, open field, open table exploration)
For more details of comprehensive behavioral test procedures
refer to Whishaw and Kolb [99].
7. Sample collection: Blood samples can be collected before
breeding (baseline), on gestational day 18 and lactational day
1 from the dams.
(a) Sample collection may include blood, brain, lung, liver,
spleen, reproductive tissues (cervix, uterine horns), fetal
membranes, and placenta.
(b) DNA and RNA may be extracted with purification kits for
miRNA, DNAm, and mRNA analyses as desired.
(c) Neuroendocrine measurements may include blood and
brain samples.
(d) Collect blood samples rapidly via tail vein nick or with
inhalation anesthesia on days without behavioral tests.
l Common measurements include
(i) Blood glucose to determine hyperglycemia and
dehydration
(ii) Stress plasma markers CRH, CORT, and ACTH
(iii) Sex hormone levels of estrogen and testosterone
8. Measure physiological correlates of stress by changes in body
weight and food intake.
9. Immunohistochemistry of stress response axes.

(a) Quantify GR, MR, CRH, and oxytocin receptor density
and intracellular location to assess the stress response.
(b) Assess the general morphology (cortical thickness, vol-
ume, neuron density) for central structures involved in
stress response and affective disorders, such as prefrontal
cortex, hippocampus, basolateral amygdala, and paraven-
tricular hypothalamus.
4 Synthesis and Conclusion
We reviewed the animal models in pregnant rat and sheep reprodu-

cing the effects of stress during fetal development, rather than
periconceptually [100] or during embryonic development. We
also considered the transgenerational effects of fetal programming.
This is important, as it appears established now that a combination
of genetic and a plethora of yet to be fully defined epigenetic
mechanisms can convey not only single, but also multi- and trans-
generational memory of acute or chronic insults endured during
fetal development. This adaptive or maladaptive physiological phe-
nomenon is also referred to as the concept of the developmental
origins of health and disease (DoHAD). The framework of psycho-
neuroimmunology provides an attractive framework within which
to consider the pleiotropic effects of prenatal stress on mother and
the offspring.
The protocols from pregnant rat and sheep models presented in
this review are complementary in that they provide the foundation
for extensive testing of fetal and postnatal effects of prenatal stress
with possibilities to manipulate the intrauterine or postnatal envir-
onments and directly measure fetal, newborn, and maternal physi-
ological responses. This has translational relevance for the
identification of predictive and diagnostic biomarkers of prenatal
stress as well as the development of therapeutic approaches to
mitigate the “stressed brain” phenotype in mother and offspring.
We hope that such studies will help devise clinical strategies
breaking the vicious cycle of multigenerational memory of stress
exposures to promote healthy development and successful aging in
mothers and their offspring.
References
1. van den Bergh BRH, van den Heuvel MI, mental health: the influence of maternal stress
Lahti M, Braeken M, de Rooij SR, in pregnancy. Neurosci Biobehav Rev
Entringer S, Hoyer D, Roseboom T, 2. Abbasi J (2017) The paternal epigenome
Raikkonen K, King S, Schwab M (2017) Pre- makes its mark. JAMA 317:2049–2051
natal developmental origins of behavior and 3. Mychasiuk R, Harker A, Ilnytskyy S, Gibb R
(2013) Paternal stress prior to conception
alters DNA methylation and behaviour of prior and during pregnancy. Neurosci Biobe-
developing rat offspring. Neuroscience hav Rev
241:100–105 14. Barker DJ (1998) In utero programming of
4. Diz-Chaves Y, Pernia O, Carrero P, Garcia- chronic disease. Clin Sci (Lond) 95:115–128
Segura LM (2012) Prenatal stress causes 15. Gluckman PD, Hanson MA, Spencer HG
alterations in the morphology of microglia (2005) Predictive adaptive responses and
and the inflammatory response of the hippo- human evolution. Trends Ecol Evol
campus of adult female mice. J Neuroinflam- 20:527–533
mation 9:71 16. Gluckman PD, Hanson MA (2006) The
5. Diz-Chaves Y, Astiz M, Bellini MJ, Garcia- developmental origins of health and disease.
Segura LM (2013) Prenatal stress increases Early Life Origins Health Dis:1–7
the expression of proinflammatory cytokines 17. Van den Bergh BR (2011) Developmental
and exacerbates the inflammatory response to programming of early brain and behaviour
LPS in the hippocampal formation of adult development and mental health: a conceptual
male mice. Brain Behav Immun 28:196–206 framework. Dev Med Child Neurol 53(Suppl
6. Slusarczyk J, Trojan E, Glombik K, 4):19–23
Budziszewska B, Kubera M, Lason W, 18. Lahiri DK, Maloney B, Zawia NH (2009) The
Popiolek-Barczyk K, Mika J, Wedzony K, LEARn model: an epigenetic explanation for
Basta-Kaim A (2015) Prenatal stress is a vul- idiopathic neurobiological diseases. Mol Psy-
nerability factor for altered morphology and chiatry 14:992–1003
biological activity of microglia cells. Front
Cell Neurosci 9:82 19. Hertzman C (1999) The biological embed-
ding of early experience and its effects on
7. Boersma GJ, Tamashiro KL (2015) Individual health in adulthood. Ann N Y Acad Sci
differences in the effects of prenatal stress 896:85–95
exposure in rodents. Neurobiol Stress
1:100–108 20. Crews D, Gillette R, Scarpino SV,
Manikkam M, Savenkova MI, Skinner MK
8. Weinstock M (2017) Prenatal stressors in (2012) Epigenetic transgenerational inheri-
rodents: effects on behavior. Neurobiol Stress tance of altered stress responses. Proc Natl
6:3–13 Acad Sci U S A 109:9143–9148
9. Babenko O, Kovalchuk I, Metz GA (2015) 21. Porter TF, Fraser AM, Hunter CY, Ward RH,
Stress-induced perinatal and transgenera- Varner MW (1997) The risk of preterm birth
tional epigenetic programming of brain devel- across generations. Obstet Gynecol 90:63–67
opment and mental health. Neurosci
Biobehav Rev 48:70–91 22. Yao Y, Robinson AM, Zucchi FC, Robbins
JC, Babenko O, Kovalchuk O, Kovalchuk I,
10. Rakers F, Frauendorf V, Rupprecht S, Olson DM, Metz GA (2014) Ancestral expo-
Schiffner R, Bischoff SJ, Kiehntopf M, sure to stress epigenetically programs preterm
Reinhold P, Witte OW, Schubert H, Schwab birth risk and adverse maternal and newborn
M (2013) Effects of early- and late-gestational outcomes. BMC Med 12:121
maternal stress and synthetic glucocorticoid
on development of the fetal hypothalamus- 23. Christiaens I (2012) Chronic maternal stress
pituitary-adrenal axis in sheep. Stress and genetic variants in the etiology of sponta-
16:122–129 neous preterm birth. Department of Physiol-
ogy, University of Alberta, Edmonton, AB
11. Burns P, Liu HL, Kuthiala S, Fecteau G,
Desrochers A, Durosier LD, Cao M, Frasch 24. Christiaens I, Hegadoren K, Olson DM
MG (2015) Instrumentation of near-term (2012) Adverse childhood experiences are
fetal sheep for multivariate chronic associated with spontaneous preterm birth: a
non-anesthetized recordings. J Vis Exp 105: case-control study. Reprod Sci 19:183A
e52581 25. Shapiro GD, Fraser WD, Frasch MG, Seguin
12. Rakers F, Rupprecht S, Dreiling M, JR (2013) Psychosocial stress in pregnancy
Bergmeier C, Witte OW, Schwab M (2017) and preterm birth: associations and mechan-
Transfer of maternal psychosocial stress to the isms. J Perinat Med 41:631–645
fetus. Neurosci Biobehav Rev 26. Christian LM (2015) Stress and immune
13. Braun K, Bock J, Wainstock T, Matas E, function during pregnancy: an emerging
Gaisler-Salomon I, Fegert J, Ziegenhain U, focus in mind-body medicine. Curr Dir
Segal M (2017) Experience-induced transge- Psychol Sci 24:3–9
nerational (re-)programming of neuronal 27. Raju TNK, Buist AS, Blaisdell CJ, Moxey-
structure and functions: impact of stress Mims M, Saigal S (2017) Adults born pre-
term: a review of general health and system-
specific outcomes. Acta Paediatr 38. Monteleone MC, Adrover E, Pallares ME,
106:1409–1437 Antonelli MC, Frasch AC, Brocco MA
28. Adrover E, Berger MA, Perez AA, Tarazi FI, (2014) Prenatal stress changes the glycopro-
Antonelli MC (2007) Effects of prenatal stress tein GPM6A gene expression and induces epi-
on dopamine D2 receptor asymmetry in rat genetic changes in rat offspring brain.
brain. Synapse 61:459–462 Epigenetics 9:152–160
29. Berger MA, Barros VG, Sarchi MI, Tarazi FI, 39. Baier CJ, Pallares ME, Adrover E, Katunar
Antonelli MC (2002) Long-term effects of MR, Raisman-Vozari R, Antonelli MC
prenatal stress on dopamine and glutamate (2014) Intrastriatal 6-OHDA lesion differen-
receptors in adult rat brain. Neurochem Res tially affects dopaminergic neurons in the ven-
27:1525–1533 tral tegmental area of prenatally stressed rats.
30. Barros VG, Berger MA, Martijena ID, Sarchi Neurotox Res 26:274–284
MI, Perez AA, Molina VA, Tarazi FI, Anto- 40. Baier CJ, Katunar MR, Adrover E, Pallares
nelli MC (2004) Early adoption modifies the ME, Antonelli MC (2012) Gestational
effects of prenatal stress on dopamine and restraint stress and the developing dopaminer-
glutamate receptors in adult rat brain. J Neu- gic system: an overview. Neurotox Res
rosci Res 76:488–496 22:16–32
31. Katunar MR, Saez T, Brusco A, Antonelli MC 41. Debeir T, Ginestet L, Francois C, Laurens S,
(2010) Ontogenetic expression of dopamine- Martel JC, Chopin P, Marien M, Colpaert F,
related transcription factors and tyrosine Raisman-Vozari R (2005) Effect of intrastria-
hydroxylase in prenatally stressed rats. Neuro- tal 6-OHDA lesion on dopaminergic innerva-
tox Res 18:69–81 tion of the rat cortex and globus pallidus. Exp
32. Carboni E, Barros VG, Ibba M, Silvagni A, Neurol 193:444–454
Mura C, Antonelli MC (2010) Prenatal 42. Rakers F, Bischoff S, Schiffner R, Haase M,
restraint stress: an in vivo microdialysis study Rupprecht S, Kiehntopf M, Kuhn-Velten
on catecholamine release in the rat prefrontal WN, Schubert H, Witte OW, Nijland MJ,
cortex. Neuroscience 168:156–166 Nathanielsz PW, Schwab M (2015) Role of
33. Silvagni A, Barros VG, Mura C, Antonelli catecholamines in maternal-fetal stress trans-
MC, Carboni E (2008) Prenatal restraint fer in sheep. Am J Obstet Gynecol 213(684):
stress differentially modifies basal and stimu- e681–e689
lated dopamine and noradrenaline release in 43. Dreiling M, Bischoff S, Schiffner R,
the nucleus accumbens shell: an ‘in vivo’ Rupprecht S, Kiehntopf M, Schubert H,
microdialysis study in adolescent and young Witte OW, Nathanielsz PW, Schwab M,
adult rats. Eur J Neurosci 28:744–758 Rakers F (2016) Stress-induced decrease of
34. Pallares ME, Baier CJ, Adrover E, Monte- uterine blood flow in sheep is mediated by
leone MC, Brocco MA, Antonelli MC alpha 1-adrenergic receptors. Stress
(2013) Age-dependent effects of prenatal 19:547–551
stress on the corticolimbic dopaminergic sys- 44. Durosier LD, Green G, Batkin I, Seely AJ,
tem development in the rat male offspring. Ross MG, Richardson BS, Frasch MG
Neurochem Res 38:2323–2335 (2014) Sampling rate of heart rate variability
35. Pallares ME, Adrover E, Baier CJ, Bourgui- impacts the ability to detect acidemia in ovine
gnon NS, Monteleone MC, Brocco MA, fetuses near-term. Front Pediatr 2:38
Gonzalez-Calvar SI, Antonelli MC (2013) 45. Li X, Xu Y, Herry C, Durosier LD, Casati D,
Prenatal maternal restraint stress exposure Stampalija T, Maisonneuve E, Seely AJ,
alters the reproductive hormone profile and Audibert F, Alfirevic Z, Ferrazzi E, Wang X,
testis development of the rat male offspring. Frasch MG (2015) Sampling frequency of
Stress 16:429–440 fetal heart rate impacts the ability to predict
36. Barros VG, Duhalde-Vega M, Caltana L, pH and BE at birth: a retrospective multi-
Brusco A, Antonelli MC (2006) Astrocyte- cohort study. Physiol Meas 36:L1–L12
neuron vulnerability to prenatal stress in the 46. Vezina-Audette R, Herry C, Burns P,
adult rat brain. J Neurosci Res 83:787–800 Frasch M, Chave E, Theoret C (2016) Heart
37. Adrover E, Pallarés ME, Baier CJ, Monte- rate variability in relation to stress in the Asian
leone MC, Giuliani FA, Waagepetersen HS, elephant (Elephas maximus). Can Vet J
Brocco MA, Cabrera R, Sonnewald U, 57:289–292
Schousboe A (2015) Glutamate neurotrans- 47. Dawes GS, Serra-Serra V, Moulden M, Red-
mission is affected in prenatally stressed off- man CW (1994) Dexamethasone and fetal
spring. Neurochem Int 88:73–87 heart rate variation. Br J Obstet Gynaecol
101:675–679
48. Mulder EJ, Derks JB, Visser GH (1997) 58. Pavlov VA, Tracey KJ (2012) The vagus nerve
Antenatal corticosteroid therapy and fetal and the inflammatory reflex—linking immu-
behaviour: a randomised study of the effects nity and metabolism. Nat Rev Endocrinol
of betamethasone and dexamethasone. Br J 8:743–754
Obstet Gynaecol 104:1239–1247 59. Pavlov VA, Tracey KJ (2015) Neural circuitry
49. Senat MV, Minoui S, Multon O, and immunity. Immunol Res 63:38–57
Fernandez H, Frydman R, Ville Y (1998) 60. Pavlov VA, Tracey KJ (2017) Neural regula-
Effect of dexamethasone and betamethasone tion of immunity: molecular mechanisms and
on fetal heart rate variability in preterm clinical translation. Nat Neurosci 20:156–166
labour: a randomised study. Br J Obstet 61. Borovikova LV, Ivanova S, Zhang M, Yang H,
Gynaecol 105:749–755 Botchkina GI, Watkins LR, Wang H,
50. Derks JB, Mulder EJ, Visser GH (1995) The Abumrad N, Eaton JW, Tracey KJ (2000)
effects of maternal betamethasone administra- Vagus nerve stimulation attenuates the sys-
tion on the fetus. Br J Obstet Gynaecol temic inflammatory response to endotoxin.
102:40–46 Nature 405:458–462
51. Braithwaite EC, Kundakovic M, Ramchan- 62. Kwan H, Garzoni L, Liu HL, Cao M,
dani PG, Murphy SE, Champagne FA Desrochers A, Fecteau G, Burns P, Frasch
(2015) Maternal prenatal depressive symp- MG (2016) VNS in inflammation: systematic
toms predict infant NR3C1 1F and BDNF review of animal models and clinical studies.
IV DNA methylation. Epigenetics Bioelectron Med 3:1–6
10:408–417 63. Caliskan G, Albrecht A (2013) Noradrenergic
52. Frasch MG, Muller T, Wicher C, Weiss C, interactions via autonomic nervous system: a
Lohle M, Schwab K, Schubert H, Nathanielsz promising target for extinction-based expo-
PW, Witte OW, Schwab M (2007) Fetal body sure therapy? J Neurophysiol 110:2507–2510
weight and the development of the control of 64. Clancy JA, Mary DA, Witte KK, Greenwood
the cardiovascular system in fetal sheep. J JP, Deuchars SA, Deuchars J (2014)
Physiol 579:893–907 Non-invasive vagus nerve stimulation in
53. Frasch MG, Muller T, Weiss C, Schwab K, healthy humans reduces sympathetic nerve
Schubert H, Schwab M (2009) Heart rate activity. Brain Stimul 7:871–877
variability analysis allows early asphyxia detec- 65. George MS, Ward HE Jr, Ninan PT,
tion in ovine fetus. Reprod Sci 16:509–517 Pollack M, Nahas Z, Anderson B, Kose S,
54. Durosier LD, Xu A, Matushewski B, Cao M, Howland RH, Goodman WK, Ballenger JC
Herry C, Batkin I, Seely A, Ross M, Richard- (2008) A pilot study of vagus nerve stimula-
son BS, Frasch MG (2013) Neural signature tion (VNS) for treatment-resistant anxiety
of cerebral activity of the fetal cholinergic anti- disorders. Brain Stimul 1:112–121
inflammatory pathway derived from heart rate 66. Liu RP, Fang JL, Rong PJ, Zhao Y, Meng H,
variability. FASEB J 27 Ben H, Li L, Huang ZX, Li X, Ma YG, Zhu B
(1 Supplement):926–911 (2013) Effects of electroacupuncture at auric-
55. Liu D, Diorio J, Tannenbaum B, Caldji C, ular concha region on the depressive status of
Francis D, Freedman A, Sharma S, unpredictable chronic mild stress rat models.
Pearson D, Plotsky PM, Meaney MJ (1997) Evid Based Complement Alternat Med
Maternal care, hippocampal glucocorticoid 2013:789674
receptors, and hypothalamic-pituitary-adrenal 67. O’Keane V, Dinan TG, Scott L, Corcoran C
responses to stress. Science 277:1659–1662 (2005) Changes in hypothalamic-pituitary-
56. Durosier LD, Herry CL, Cortes M, Cao M, adrenal axis measures after vagus nerve stimu-
Burns P, Desrochers A, Fecteau G, Seely AJ, lation therapy in chronic depression. Biol Psy-
Frasch MG (2015) Does heart rate variability chiatry 58:963–968
reflect the systemic inflammatory response in 68. Pena DF, Childs JE, Willett S, Vital A, McIn-
a fetal sheep model of lipopolysaccharide- tyre CK, Kroener S (2014) Vagus nerve stim-
induced sepsis? Physiol Meas 36:2089–2102 ulation enhances extinction of conditioned
57. Lun Liu H, Garzoni L, Herry C, Durosier fear and modulates plasticity in the pathway
LD, Cao M, Burns P, Fecteau G, from the ventromedial prefrontal cortex to
Desrochers A, Patey N, Seely AJ, Faure C, the amygdala. Front Behav Neurosci 8:327
Frasch MG (2016) Can monitoring fetal 69. Ylikoski J, Lehtimaki J, Pirvola U, Makitie A,
intestinal inflammation using heart rate varia- Aarnisalo A, Hyvarinen P, Ylikoski M (2017)
bility analysis signal incipient necrotizing Non-invasive vagus nerve stimulation reduces
enterocolitis of the neonate? Pediatr Crit
Care Med 17(4):e165–e176
sympathetic preponderance in patients with from pregnancies complicated by preeclamp-

tinnitus. Acta Otolaryngol 137(4):426–431 sia and preterm labor. Reprod Sci 18:46–56
70. Hosoi T, Okuma Y, Nomura Y (2000) Elec- 83. Faraji J, Soltanpour N, Lotfi H, Moeeini R,
trical stimulation of afferent vagus nerve Moharreri AR, Roudaki S, Hosseini SA,
induces IL-1beta expression in the brain and Olson DM, Abdollahi AA, Soltanpour N,
activates HPA axis. Am J Physiol Regul Integr Mohajerani MH, Metz GAS (2017) Lack of
Comp Physiol 279:R141–R147 social support raises stress vulnerability in rats
71. Farrokhi CB, Tovote P, Blanchard RJ, Blan- with a history of ancestral stress. Sci Rep
chard DC, Litvin Y, Spiess J (2007) Corta- 7:5277
gine: behavioral and autonomic function of 84. Faraji J, Soltanpour N, Ambeskovic M, Zuc-
the selective CRF receptor subtype 1 agonist. chi FCR, Beaumier P, Kovalchuk I, Metz GAS
CNS Drug Rev 13:423–443 (2017) Evidence for ancestral programming
72. Porges SW (1995) Cardiac vagal tone: a phys- of resilience in a two-hit stress model. Front
iological index of stress. Neurosci Biobehav Behav Neurosci 11:89
Rev 19:225–233 85. Olson DM, Severson EM, Verstraeten BS, Ng
73. Porges SW (2009) The polyvagal theory: new JW, McCreary JK, Metz GA (2015) Allostatic
insights into adaptive reactions of the auto- load and preterm birth. Int J Mol Sci
nomic nervous system. Cleve Clin J Med 76 16:29856–29874
(Suppl 2):S86–S90 86. Daskalakis NP, Bagot RC, Parker KJ, Vinkers
74. Porges SW, Furman SA (2011) The early CH, de Kloet ER (2013) The three-hit con-
development of the autonomic nervous sys- cept of vulnerability and resilience: toward
tem provides a neural platform for social understanding adaptation to early-life adver-
behavior: a polyvagal perspective. Infant sity outcome. Psychoneuroendocrinology
Child Dev 20:106–118 38:1858–1873
75. Porges SW (2007) The polyvagal perspective. 87. Saban KL, Mathews HL, DeVon HA, Janusek
Biol Psychol 74:116–143 LW (2014) Epigenetics and social context:
76. Seely AJ, Macklem PT (2004) Complex sys- implications for disparity in cardiovascular dis-
tems and the technology of variability analysis. ease. Aging Dis 5:346–355
Crit Care 8:R367–R384 88. Zucchi FC, Yao Y, Ilnytskyy Y, Robbins JC,
77. Taylor AG, Goehler LE, Galper DI, Innes KE, Soltanpour N, Kovalchuk I, Kovalchuk O,
Bourguignon C (2010) Top-down and Metz GA (2014) Lifetime stress cumulatively
bottom-up mechanisms in mind-body medi- programs brain transcriptome and impedes
cine: development of an integrative frame- stroke recovery: benefit of sensory stimula-
work for psychophysiological research. tion. PLoS One 9:e92130
Explore (NY) 6:29–41 89. Li Y, Saldanha SN, Tollefsbol TO (2014)
78. Frasch MG, Muller T, Hoyer D, Weiss C, Impact of epigenetic dietary compounds on
Schubert H, Schwab M (2009) Nonlinear transgenerational prevention of human dis-
properties of vagal and sympathetic modula- eases. AAPS J 16:27–36
tions of heart rate variability in ovine fetus 90. Gapp K, Bohacek J, Grossmann J, Brunner
near term. Am J Physiol Regul Integr Comp AM, Manuella F, Nanni P, Mansuy IM
Physiol 296:R702–R707 (2016) Potential of environmental enrich-
79. Migicovsky Z, Kovalchuk I (2011) Epigenetic ment to prevent transgenerational effects of
memory in mammals. Front Genet 2:28 paternal trauma. Neuropsychopharmacology
41:2749–2758
80. Zucchi FC, Yao Y, Metz GA (2012) The
secret language of destiny: stress imprinting 91. McCreary JK, Erickson ZT, Metz GA (2016)
and transgenerational origins of disease. Front Environmental enrichment mitigates the
Genet 3:96 impact of ancestral stress on motor skill and
corticospinal tract plasticity. Neurosci Lett
81. Renthal NE, Chen CC, Williams KC, Gerard 632:181–186
RD, Prange-Kiel J, Mendelson CR (2010)
miR-200 family and targets, ZEB1 and 92. McCreary JK, Erickson ZT, Hao Y,
ZEB2, modulate uterine quiescence and con- Ilnytskyy Y, Kovalchuk I, Metz GA (2016)
tractility during pregnancy and labor. Proc Environmental intervention as a therapy for
Natl Acad Sci U S A 107:20828–20833 adverse programming by ancestral stress. Sci
Rep 6:37814
82. Mayor-Lynn K, Toloubeydokhti T, Cruz AC,
Chegini N (2011) Expression profile of 93. Rodgers AB, Morgan CP, Leu NA, Bale TL
microRNAs and mRNAs in human placentas (2015) Transgenerational epigenetic pro-
gramming via sperm microRNA recapitulates
effects of paternal stress. Proc Natl Acad Sci U 97. Skinner MK (2008) What is an epigenetic
S A 112:13699–13704 transgenerational phenotype?: F3 or F2.
94. Hebb DO (1947) The effect of early experi- Reprod Toxicol 25:2–6
ence on problem solving at maturity. Am 98. Ambeskovic M, Soltanpour N, Falkenberg
Psychol 2:737–745 EA, Zucchi FCR, Kolb B, Metz GAS (2017)
95. Diamond MC, Johnson RE, Protti AM, Ancestral exposure to stress generates new
Ott C, Kajisa L (1985) Plasticity in the behavioral traits and a functional hemispheric
904-day-old male rat cerebral cortex. Exp dominance shift. Cereb Cortex
Neurol 87:309–317 27:2126–2138
96. Faraji J, Soltanpour N, Lotfi H, Nejad- 99. Whishaw IQ, Kolb B (2004) The behavior of
Ghorban F, Moharrerie A, Roudaki S, Hos- the laboratory rat: a handbook with tests.
seini SA, Abdollahi AA, Soltanpour N, Oxford University Press, New York
Moeeini R, Metz GA S (Submitted) Social 100. Heijmans BT, Tobi EW, Stein AD, Putter H,
experience promotes brain health and stress Blauw GJ, Susser ES, Slagboom PE, Lumey
resiliency in mothers and their unexposed LH (2008) Persistent epigenetic differences
offspring associated with prenatal exposure to famine
in humans. Proc Natl Acad Sci U S A
105:17046–17049
INDEX
A Autophagosomes............................................................. 14
Autophagy ...................................................................7, 15
Adaptation ..................................................... v, 3–5, 8, 13,
14, 37, 38, 41, 46–49, 51, 56, 58, 59, 78, 79, 82, B
158, 201, 239, 240, 260, 261, 319, 359, 370
Aerobic exercise........................................... 154, 210, 211 Balance...................................................................... v, 4, 5,
Aging ........................................................ 4–11, 174, 181, 7–9, 11–17, 77, 79–81, 83, 84, 94, 118, 129, 130,
224, 250, 367, 370 145, 158, 177, 188, 357, 363
Allergic diseases ............................................................. 185 Batch PLS ...................................................................... 107
Allostasis ............................................................... 260, 261 Behavioral .................................................................. vi, 22,
Alzheimer’s disease ................................................ v, 6, 10, 64, 65, 69, 82, 88–92, 122, 126, 127, 130, 134,
17, 49, 186, 288, 301 136, 138, 154, 165, 205, 221–229, 231–252,
Amygdala ...............................................80, 363, 368, 371 263, 264, 267–271, 273–276, 356, 359, 368
Anhedonia .................................... 82, 239, 240, 297, 356 Bioactives ....................................................................... 222
Animal model ...................................................... 122, 150, Biobehaviors .................................................................. 252
156, 197, 223, 261, 262, 301, 342, 353, 356, Biological response........................................................ 172
364, 370 Biomarkers............................................................ v, 4, 7–9,
Anorexia................................................................ 228, 231 11–13, 16–18, 58, 65, 70, 78–80, 83, 102, 103,
Anthropology .............................................................. v, 55 106, 110–112, 155, 175, 177, 181–186, 357,
Antibiotic ...................................... 25, 26, 44, 46, 49, 344 364, 370
Antibody .............................................................. 176, 185, Biopsychosocial ..............................................................v, 4
201, 269, 310, 311, 313–323, 328–336 Blood-brain barrier (BBB)...........................................263,
Antidepressant ......................................................... 81, 83, 266, 269, 272, 273, 295–298
140, 154, 238 B-lymphocytes ...................................................... 309, 311
Antigen ............................................................15, 62, 117, Body mass index (BMI).................................59, 106, 112
176, 185, 210, 262, 268, 269, 310, 311, Brain........................................................................ v, 4, 22,
317–319, 322, 327–329, 331–334, 336, 338 41, 56, 79, 89, 122, 163, 172, 195, 222,
Anxiety .................................................................... v, 4, 18, 259–263, 265–270, 272–276, 291, 341, 353
46, 77, 78, 89, 90, 92, 223, 227, 228, 233–240,
C
245, 262, 263, 268, 275, 315–317, 356, 363, 368
Apoptosis ............................................................. 7, 10, 15, Cancer.................................................................. v, vi, 5, 7,
79, 149, 187, 263, 264, 272, 274 9, 11, 14, 17, 28, 49, 50, 70, 172, 174, 181, 183,
Astrocytes .................................7, 16, 298, 301, 342, 345 184, 186–188, 222, 315
Atherosclerosis .........................6, 9, 13, 50, 62, 146, 186 Cannabinoid system ............................................. 179, 181
Attention ..............................................................v, 5, 6, 8, Carcinogenesis................................................................. 15
89, 90, 94, 124, 125, 132–134, 165, 248, 354 Cardiovascular diseases ............................v, 8, 13, 17, 354
Attenuation.......................................................... 196, 200, Caregiving ................................................... 311, 314, 315
201, 274, 289, 291, 293 Catecholamine......................................... 24–27, 149, 183
Autoantigens ........................................................ 260, 261 Central nervous system (CNS)............................... 13, 16,
Autoimmune diseases ................................................. vi, 5, 25, 82, 89, 125, 163, 260, 262–264, 266–274,
181, 183–186, 188, 189, 260, 261 276
Autoimmunity ............................................................7, 15, Chemogenetics................................................ vi, 195–207
28, 48, 51, 172, 174, 181, 183–186, 259–263, Chemokines.......................................................... 172, 259
265–276 Chemotaxis.....................................................13, 157–161
Autonomic nervous system (ANS) ............................... 25, Cholinergic signaling ...................................... vi, 341–350
47, 260, 297, 342, 363 Chromium release assay ........................................... vi, 209
https://doi.org/10.1007/978-1-4939-7828-1, © Springer Science+Business Media LLC, part of Springer Nature 2018
377
PSYCHONEUROIMMUNOLOGY: METHODS AND PROTOCOLS
378 Index
Chronic experimentation...............................81, 227, 249 Endotoxin................................................... 163, 164, 297,
Circadian rhythms .............................. vi, 4, 174, 177, 179 298, 300, 301, 303, 304, 343
Co-expression patterns ................................................. 103 Environment.........................................................v, 16, 22,
Cognition ....................................... 58, 77, 222, 240, 297 29, 38–40, 44, 48, 51, 56, 58, 59, 61, 62, 92, 97,
Cognitive function ...........................................8, 240, 248 126, 133, 201–203, 223–225, 227, 235, 250,
Complement........................................................ 6, 10, 67, 252, 259, 261, 263, 356, 359, 366, 370
259, 266, 272, 328 Enzyme-linked immunosorbent assay
Complex adaptive systems (CAS) ...................... v, 3–7, 78 (ELISA).................................................... 103, 104,
Computational modeling ...................................... 96, 140 156, 329–331, 336–338, 347, 363
Coronary artery disease (CAD) ..................................... 13 Epidemiology ....................................................... 146, 151
Cortisol ................................................................ 8, 25, 47, Epilepsy................................................................. 341, 364
59, 61, 64, 66–68, 71, 88, 89, 92, 94, 96, 97, 102, Epinephrine (E)....................................................... 24, 25,
104, 107–109, 111, 115, 117, 124, 125, 130, 89, 92, 94, 124, 125, 132, 133, 138, 140
132–134, 138–140, 173, 174, 177, 179, 181, Escherichia coli ......................................................... 24, 25,
183, 260, 298, 362, 363 27, 28, 304, 346
C-reactive protein (CRP) ..........................................8, 12, Evolution ..................................................................45–48,
59, 62, 66, 67, 81, 156, 157, 175, 177 50, 58, 92, 112, 113, 126, 173, 223
Culture..................................................................... 22, 23, Exercise response ................................................. 103, 109
42, 51, 56, 57, 62, 69, 102, 103, 107–109, 117, Experience ............................................................. v, vi, 22,
212–215, 218, 219, 342, 345–347, 349 39–44, 48, 50, 62, 64, 65, 70, 121, 154, 155, 165,
Cyclooxygenase-2 (COX-2) ......................................... 301 228, 239, 262, 315, 344, 350, 353, 355, 365–367
Cytokine network................................................ 177, 181, Exploration .............................................................. 39, 58,
183–186, 189, 269 70, 88, 97, 228, 229, 231, 232, 234–239, 241,
Cytokine profiles .................................................. 117, 328 243–247, 263, 268, 338, 368, 370
Cytokines ............................................................... 4, 6–10,
12, 16, 47, 66, 78–80, 83, 101, 102, 106, 107, F
109, 117, 118, 145, 150, 156–158, 163, 164, Fatigue ..................................................................... 61, 79,
174–177, 180, 181, 184–186, 188, 189, 222,
82, 90, 92, 94, 101, 103, 114, 125, 132–134,
228, 259, 263, 266, 269, 271, 276, 297, 301, 148, 154–156, 210, 213, 218, 222, 228, 231,
311, 347 249, 251, 297, 300
Fear ........................................................................ 66, 236,
D
243, 244, 363
Dementia ....................................272, 311, 314, 315, 322 Feedback ................................................................... v, 4, 5,
Depression .............................................................. v, 4, 57, 18, 89, 91, 92, 94, 96, 97, 117, 122, 123, 138
77, 88, 121–141, 145, 174, 222, 261, 297, 315, Flow cytometry ................................................... 199, 212,
363 216, 333, 334, 346
Developmental Origin of Health and Disease Fluorine-18-labeled 2-deoxy-D-glucose
(DOHaD) ................................................. 354, 370 (FDG) ....................................................... 298–300
Diabetes ..................................................... v, 4–6, 8, 9, 11, Food consumption........................................................ 231
17, 18, 29, 46, 49, 50, 62, 66, 148, 365 Functional MR imaging (fMRI) ......................... 298, 299
Diagnostic classification ......................103, 110, 113, 363
Dimension .........................................................39, 43, 44, G
51, 87, 104, 356, 358, 363 Gamma-aminobutyric acid (GABA) .........................7, 14,
DNA methylation ........................................................... 48
89, 90, 92, 94, 122–125, 130, 132–134, 136, 138
Dopamine6, 11, 25, 78, 82, 88, 89, 94, 96, 97, 124, 296, Glutamate ..................................................... 7, 14, 89, 94,
303, 355 124, 125, 132, 134, 135, 138, 139, 272, 356
Drug-resistant ............................................................... 341
Glutathione ...............................................................11, 83
Dynamical medicine............................................. 3, 15, 18 G protein-coupled receptor(GPCR) ..................... 24, 196
Gulf war illness (GWI)............................. v, 101–118, 210
E
Gut microbiota and metabolites ......................... 162–164
Ecoimmunology.............................................................. 57
Emotional disturbances ................................................ 272 H
Endocrinology................................. v, 22, 23, 29, 58, 366
Heart failure (HF)............................vi, 14, 145, 148, 157
PSYCHONEUROIMMUNOLOGY : METHODS AND PROTOCOLS
Index 379
Heart-gut axis................................................................ 162 Interleukin-6 (IL-6).......................................... 13, 61–63,
Heart rate (HR) .........................................................4, 12, 66, 67, 79, 102, 103, 107, 109, 111, 117, 150,
13, 45, 79, 149, 151, 342, 363 155–157, 163, 175–177, 180, 181, 184, 185,
Hemagglutination inhibition assay (HAI)......... 330–333, 188, 269, 272, 297, 311
338, 339 Interventions .....................................................14, 80, 81,
Hepatitis B............................................................. 59, 313, 88, 89, 123, 125, 129, 130, 136–139, 154, 288,
316–318, 321, 322 296, 298, 299, 320, 322, 323, 327, 366
Hippocampus ........................................................ 83, 267,
272, 356, 368, 371 K
Homeostasisv, 4, 5, 10, 12–14, 24, 77, 82, 94, 122, 126, K562 target cells ........................................................... 214
132, 158, 259–261, 354 Kynurenine ......................................................... 78, 80, 81
Homeostatic regulation .................................90, 135, 139
Hormones...................................................................8, 13, L
16, 22, 23, 25, 26, 47, 58, 59, 65, 67, 68, 71, 81,
88, 89, 101, 105, 118, 123, 124, 130, 132, 138, Learning............................................................... 8, 9, 222,
139, 172, 173, 177, 180, 181, 261, 267, 269, 228, 233, 235, 239, 240, 243, 244, 246–248,
271, 272, 276, 320, 322, 355, 356, 362, 368, 370 263, 268, 273, 274, 368, 370
Host-microbiota.............................................................. 46 Life history theory .......................................................... 56
Human biological variation ............................................ 56 Life style.................................................................. 18, 137
Human Microbiome Project (HMP)............................. 23 Light cycle ................................................. 5, 12, 225, 234
Humidity .............................................................. 212, 225 Locomotor activity.............................................. 228, 229,
Hypothalamic-pituitary-adrenal (HPA) axis............ vi, 14, 235, 236, 244, 268, 274
18, 47, 61, 77, 81, 94, 117, 138, 155, 173, 175, Lupus ................................................................... 184, 261,
180, 181, 183, 226, 272, 366–368 263, 264, 266, 267, 269–276
Lymphocytes ........................................................... 24, 63,
I 107, 117, 158, 172, 173, 175–177, 179, 181,
183, 184, 186–189, 209, 211, 216, 217, 263,
Imbalance .................................................................v, vi, 3,
268, 310
6–9, 11–16, 18, 28, 77–83, 107, 109, 122, 356 Lymphocyte-to-monocyte ratio ..................................... 45
Immune ....................................................................v, vi, 8,
21, 37, 55, 79, 88, 101, 139, 148, 172, 173, 195, M
209, 221, 259, 288, 309, 327, 341, 364
Immunity ............................................................ vi, 22, 29, Macrophage............................................................. 10, 25,
49, 51, 55, 59, 67, 78, 79, 157, 158, 176, 184, 172, 173, 175–177, 179, 182, 184, 186–189,
187, 195, 222, 309, 311, 315, 318, 323, 328 266, 273, 301, 328, 342
Immunoactivation................................................ 177, 181 Macular degeneration ..................................................... 11
Immunoassays ......................................................... 58, 68, Magnetic resonance (MR) ..................294, 298, 368, 371
104, 328, 333, 338 Major depressive disorder (MDD)..........................80–83,
Immunobehaviors ............................................... 222, 224, 122, 140, 155
225, 227, 234, 250 Maze .................................................................... 224, 233,
Immunocompetence ....................................................... 47 235–237, 241, 244–248, 263, 356, 368, 370
Immunosenescence ....................................................... 315 Melatonin ............................................................. 172, 180
Immunosuppression ........................................... 172, 173, Memory ......................................................................9, 89,
177, 181, 183, 270 90, 124, 125, 163, 176, 222, 228, 232, 240–248,
Immunotherapy ............................................................ 187 268, 274, 311, 367, 368, 370
Infectious diseases ......................................................7, 38, Metabolic syndrome (MetS) ............................... 6, 11, 28
47, 49, 50, 57, 60, 69, 310, 311, 323, 328 Microarray ....................................................334, 336–338
Inflammation ..................................................... v, vi, 3–18, Microbiome ........................................................ 23, 28, 46
50, 59, 61, 62, 66, 67, 71, 77, 81, 117, 145, 150, Microbiota ................................................................v, vi, 4,
154, 155, 157, 158, 163, 164, 249, 261, 266, 21, 24, 30, 46, 162–164
268, 270–272, 274, 296–301, 304, 348, 363 Microbiota-gut-brain (MGB) axis .............................4, 18
Influenza ................................................... 49, 59, 66, 297, Microglia.......................................................................6, 7,
311, 313–315, 317–322, 328, 331, 333, 338 10, 16, 266, 272, 276, 301, 303, 342, 343,
Innate immunity ............................................................. 67 345–349
Insomnia .............................................................. v, 5–7, 17 Mind-body............................................................... v, 3, 77
PSYCHONEUROIMMUNOLOGY: METHODS AND PROTOCOLS
380 Index
miRNAs .................................................................. 16, 366 O
Mitochondria...................................................... 13, 14, 45
Molecular imaging ........................................................ 288 Obesity......................................................... v, 6, 7, 11–13,
Monte Carlo algorithm.......................................... 92, 126 17, 18, 28, 46, 63, 78, 80, 181, 222
Mood .................................................................55, 61, 62, Odors ...................................................226, 228, 232, 252
65, 66, 71, 79, 80, 89, 90, 92, 94, 103, 121, 132, Opioid system.............................................. 172, 179, 183
134, 136, 138, 145, 272, 297, 321 Opsins ..................................................196, 197, 202, 205
Motor activity.................................................................. 46 Opsonization ................................................................. 328
Mouse ......................................................................vi, 197, Optogenetics ................................................... vi, 195, 206
202, 205, 221–229, 231–252, 262–264, 272 Oxidant ............................................................... 15, 79, 83
Multi-hit ...................................................... 342, 345, 366
P
Multiplex ......................................................... vi, 327–339
Multi-stability ....................................................... 122, 133 Parkinson’s disease (PD) ................................................ 11
Multivariate data acquisition ........................................ 345 Partial least squares (PLS) .................................. 104, 105,
Myocardium ................................................ 147, 149, 150 107, 109–111, 116
Peripheral nervous system (PNS)........... 22, 89, 125, 262
N Personality ............................................................. 5, 6, 77,
Natural killer cell cytotoxicity (NKCC) ............. 209–212, 122, 314, 321, 322
215–218 Phagocytosis .................................................................. 328
Natural killer cells................................................. 209–219 Photons................................................288, 289, 291–294
Natural selection ................................................ 47–51, 57 Physiology .............................................................vi, 8, 11,
Nerve stimulation ...............................341, 343–345, 363 22, 25, 41, 45, 48, 50, 58, 59, 61, 67, 88, 92, 94,
Neurodegeneration ......................................................6, 9, 97, 103, 105, 122–124, 174, 188, 224–227, 269,
267, 269, 270, 272 342, 343, 353, 354, 363, 365, 366, 368, 370
Neurodegenerative diseases ........... 6, 8, 9, 186, 207, 222 Pineal gland .........................................172, 173, 179–181
Neuroendocrine .........................................................8, 15, Populations......................................................... vi, 24, 25,
21–30, 81, 82, 101, 102, 105, 106, 112, 118, 42, 50, 57–63, 66, 105, 129, 145, 146, 151, 153,
172–174, 177, 209, 260, 261, 276, 364, 365, 163, 176, 181, 197, 288, 310, 314, 319, 321,
368–370 322, 327, 346, 356, 365
Neuroimmune ..................................................... v, vi, 4–6, Positive psychology ...................................................87, 94
16, 17, 21–30, 77–84, 145–165, 171–189, Positron emission tomography (PET) ............................vi,
221–226, 231, 233, 239–241, 243–245, 248, 287–301, 303, 304
250, 345, 353 Prenatal stress (PS)..................................................vi, 342,
Neuroimmunology ............................................. 195, 196, 353–370
342, 346, 350 Prevention ....................................................................4, 7,
Neuroimmunomodulation (NIM)............................... 172 14, 15, 79, 83
Neuroinflammation................................................. 79, 83, Proinflammatory ........................................................8–10,
266, 272, 288, 299, 301, 343, 345 13, 14, 16, 78–81, 83, 145, 150, 156,
Neuropeptide Y (NPY) 79, 89, 102, 104, 107–109, 115, 157, 164, 272
124, 125, 132, 138, 181 Pseudomonas aeruginosa ................................................. 25
Neuroprotection ........................................................... 6, 9 Psoriasis............................................................... 7, 16, 184
Neuroscience .......................................... vi, 195, 196, 299 Psychiatric............................................................... v, 7, 13,
Neurotransmitters ......................................................4, 24, 78–80, 82–84, 174, 261, 276, 298, 355
25, 28, 78–80, 82, 83, 88, 89, 92, 101, 105, Psychology....................................................................... 55
122–126, 130, 132, 134, 138, 139, 165, 172, Psychoneuroimmunology (PNI).............................v, vi, 3,
222, 273, 276, 296, 299, 355 11, 13, 16, 17, 21, 22, 37–41, 55–71, 77, 83,
Neutralization ................................................................... 4 154–165, 209, 221–229, 231–252, 287–301,
Noise .................................................................... 226, 227, 303, 304, 309–311, 313–323, 327, 328, 333,
235, 252, 292, 296, 301, 303 338, 342, 343, 350, 353
Nonlinearity.......................................................... 4, 5, 363
Q
Normalization ............................................. 250, 293, 348
Nuclear imaging ................................................... 287, 288 Quality control ....................................103, 217, 336, 347
Quality of life............................................... 103, 145, 153
PSYCHONEUROIMMUNOLOGY : METHODS AND PROTOCOLS
Index 381
R Stroke......................................................... 7, 14, 149, 301
Surgery........................................................ 197, 200–202,
Redox............................................ 6, 7, 16, 18, 78, 79, 83 344, 356
Regression model ................................................. 111, 157 Sympathetic nervous system (SNS) ....................... 22, 24,
Regulatory logic .......................................... 123, 130, 134 147, 150, 155, 363
Regulatory metasystem........................................ 259–261 Synchronization ............................................................ 174
Renin-Angiotensin-Aldosterone System Systemic diseases .............................................. v, 172, 274
(RAAS)...................................................... 147, 149 Systemic lupus erythematosus (SLE)..........................184,
Reverse engineering .................................... 88, 89, 92, 94 185, 260–263, 265–269, 271, 274, 276
RNAseq........................................................ 347, 348, 350 Systems biology....................................... v, vi, 3, 9, 17, 77
Robustness................................................... 4, 5, 130, 338 Systems medicine .......................................v, 3, 15, 18, 84
RT-PCR ......................................................................... 348
T
S
T cells .................................................................51, 59, 78,
Salmonella typhimurium................................................. 25 80, 82, 158, 172, 173, 187, 210, 266, 273, 276,
Schizophrenia ..................................................... 78, 82, 83 311, 318
Selective serotonin re-uptake inhibitor Temperature ............................................................ 62, 66,
(SSRI) .................... 122, 123, 130, 134, 136–140 68, 200, 215, 225, 247, 356, 358
Self-organization ........................................................... 4, 5 Th1/Th2 ratio .............................................................. 158
Self-regulatory ................................................................... 4 Thymus .......................................172, 173, 263, 311, 318
Serotonin ................................................................. 80, 82, Toll-like receptors (TLR) ............................................... 15
89, 90, 92, 94, 122, 124, 125, 130, 132–134, Transcriptome .......................................79, 343, 347, 348
138, 139 Translocator protein (TSPO) .............299, 301, 303, 304
Sickness behavior................................................. 117, 222, Trauma.................................................................... 45, 301
228, 229, 231–233, 248, 260, 261, 297 Tumor ................................................................3, 4, 8, 10,
Signaling .............................................................. v, vi, 7, 9, 16, 150, 173, 174, 177, 183, 184, 187, 188, 209,
11, 12, 14–16, 18, 26, 59, 64, 78, 79, 82, 88, 226, 228, 297
90–92, 103, 107, 125–127, 130, 132, 135, 136,
157, 196, 222, 226, 263, 275, 341, 363 U
Single nucleotide polymorphisms (SNP).................59, 61
Skin7, 16–18, 23, 25, 29, 45, 64, 68, 69, 202, 203, 206, Umwelt .........................................................40–44, 47, 48
264, 344, 359, 360
V
Sleep......... 5–8, 12, 17, 79, 81, 121, 155, 222, 231, 297
Social support......................................313, 315, 318–322 Vaccination ................................................................ vi, 62,
Spatiotemporal ....................................4, 5, 9, 43, 77, 342 66, 304, 309–311, 313–323, 327
Stress ...................................................................v, 6, 7, 21, Vaccine ................................................................ vi, 44, 47,
46, 55, 78, 87, 122, 162, 181, 195, 210, 225, 49, 59, 66, 69, 310, 311, 313–320, 327–339
267, 309, 342, 353 Vagotomy ............................................................. 343, 344
Stress management..................................................vi, 123, Vagus nerve .................................. 46, 298, 341–344, 363
136, 138, 322
Stressors .................................................................vi, 5, 38, W
64, 78, 131, 136, 138, 209, 224, 239, 252,
Well-being ............................................................... 62, 71,
259–261, 263, 311, 319, 355, 356, 358, 359,
87–98, 222, 223, 227, 240
366, 369
Stress response......................................................v, 4, 7, 8, Y
10, 16, 18, 61, 77, 79, 89, 163, 210,
226, 368, 371 Yin-Yang ........................................................... v, 3, 77, 78

(Methods in Molecular Biology 1781) Qing Yan-Psychoneuroimmunology-Springer New York - Humana Press (2018)

Загружено:

Сведения о документе

Оригинальное название

Авторское право

Доступные форматы

Поделиться этим документом

Поделиться или встроить документ

Параметры публикации

Этот документ был вам полезен?

Это неприемлемый материал?

Авторское право:

Доступные форматы

(Methods in Molecular Biology 1781) Qing Yan-Psychoneuroimmunology-Springer New York - Humana Press (2018)

Загружено:

Авторское право:

Доступные форматы

Methods in

Molecular Biology 1781

Qing Yan Editor

For further volumes:

Methods and Protocols

ISSN 1064-3745 ISSN 1940-6029 (electronic)

© Springer Science+Business Media, LLC, part of Springer Nature 2018

Printed on acid-free paper

As a multidisciplinary area, the biopsychosocial models in psychoneuroimmunology (PNI)

Santa Clara, CA, USA Qing Yan

PART I STRESS AND IMMUNITY: BIOPSYCHOSOCIAL MODELS

1 Stress and Systemic Inflammation: Yin-Yang Dynamics in Health

PART II TECHNOLOGIES AND METHODS IN PSYCHONEUROIMMUNOLOGY STUDIES

11 Application of Chemogenetics and Optogenetics to Dissect

MARTA C. ANTONELLI  Facultad de Medicina, Instituto de Biologı́a Celular y Neurociencia

de Recherche en Reproduction Animale (CRRA), University of Montreal, St-Hyacinthe,

ERIC C. SHATTUCK  Department of Anthropology, University of Texas at San Antonio,

Stress and Immunity: Biopsychosocial Models and Clinical

Stress and Systemic Inflammation: Yin-Yang Dynamics

1 Yin-Yang Dynamics in the Complex Adaptive Systems: A Conceptual Framework

Studies in psychoneuroimmunology (PNI) would provide better

identifying systems-based biomarkers and targets for more effective

rhythms may affect the complex networks associated with sleep

Health conditions Yin-Yang factors/interactions References

Health conditions Yin-Yang factors/interactions References

In the following sections, the Yin and Yang characteristics will

2 The Yin and Yang in Sleep and Insomnia

disturb the temporal patterns of the elements including leptin and

3 Yin-Yang Dynamics in Aging and Neurodegenerative

The neuroendocrine-immune plasticity enables stress responses and

the neuronal plasticity, leading to neurobehavioral problems and

treatment strategies for a broad range of diseases related to learning

influence the neuronal functions, while the imbalances between the

4 Yin-Yang Dynamics in Obesity, Diabetes, and Metabolic Syndrome

PNI evidences have suggested inflammation as the transitional

weight gain. The maladaptation to chronic stress may influence

for the early diagnosis of diabetic complications especially diabetic

5 Stress, Yin-Yang Imbalances, and Cardiovascular Diseases

PNI studies have revealed the important impacts of psychophysio-

may lead to the damages in some mitochondria that can be seques-

6 Inflammation and Yin-Yang Imbalances in Cancer

More and more evidences have emphasized the important roles of

of more effective prevention and therapies in dynamical systems

have been found meaningful as the potential biomarkers for oxida-

7 Stress, Inflammation, and Yin-Yang Imbalances in Skin Diseases

in the regulation of proliferation and differentiation in psoriatic

In conclusion, PNI and systems biology evidences as discussed

Environmental Factors Light/Dark Cycles

Interactions Neuroimmune Functions Obesity

the imbalances in lifestyle and psychological conditions, such as

Intersections Between Neuroimmune and Microbiota

Key words Microbiota, Neuroendocrine, Stress, Immunity, Health

The integration of biologic systems while complex is believed to

Psychoneuroimmunology (596 Total Citations)

to infectious and inflammatory disease [1, 3, 8, 9]. Subsequently,

3 Neuroendocrine Effects on Microbiota and Pathogenic Species

The significance of microbial endocrinology has benefited from

Keyword Terms (Microbiota and Neurohormones)

Citations per year

Since 1982, research defining the interplay between neuroen-

3.1 Catecholamine As a constituent of the sympathetic nervous system, catecholamine

bronchiseptica and others [34–36]. Catecholamines have also been

Host Transferrin, lactoferrin and Ferritin

MARTA C. ANTONELLI Facultad de Medicina, Instituto de Biologı́a Celular y Neurociencia

ERIC C. SHATTUCK Department of Anthropology, University of Texas at San Antonio,