Effects of UV-B on Secondary Metabolites of St.

John’s Wort (Hypericum

perforatum L.) Grown in Controlled Environments
Melissa L. Brechner, Louis D. Albright and Leslie A. Weston
Department of Biological and Environmental Engineering, Cornell University, Ithaca,
NY, USA 2 EH Graham Centre for Agricultural Innovation, Charles Sturt University,
Wagga, NSW, Australia Received 30 August 2010, accepted 24 December 2010,
DOI: 10.1111 ⁄ j.1751-1097.2011.00904.x

ABSTRACT
The medicinal plant industry is under increasing scrutiny due to wide variance in
active ingredient (AI) concentration from values claimed on labels. Reasons for this
disparity include environmental and genotypic variation which influence AI
concentration. St. John’s wort (Hypericum perforatum) is a popular herbal remedy
which also exhibits marked variance in AI concentration among products. This study
evaluated concentration changes of three biologically active metabolites of H.
perforatum after exposure to UV light while plants were still vegetative. Treatments
were performed with 55-day-old plants grown under 400 lmol m)2 s )1 PAR for 16 h
a day. Three UV light treatments were evaluated: a single dose, a daily dose and an
increasing daily dose. Concentrations of hyperforin, pseudohypericin and hypericin
were monitored for 7 days after each treatment. A daily dose and an increasing daily
dose did not produce significantly greater increases in secondary metabolites
compared to single dose treatments. These results suggest the small but significant
transient metabolite concentration increases in H. perforatum can be induced by UV
light exposure. Information from this study can be useful in optimizing total biomass
and metabolite production in controlled environments.
INTRODUCTION St. John’s wort (Hypericum perforatum) is currently being used as
an alternative therapy to treat many medical maladies such as depression,
retroviruses and cancer. Traditional uses dating back to the ancient Greeks include
its activity as a reported antidepressant, antifungal, anti-inflammatory and
antibacterial. More recently, a key metabolite associated with St. John’s wort,
hyperforin, has been reported to possess antitumoral and antiangiogenic activity (1).
Most of the plant’s remaining beneficial uses are attributed to another important
metabolite, hypericin. A review of the medical uses and efficacy for the metabolite
hypericin may be found in Kubin et al. (2). Pseudohypericin is a secondary
metabolite that is often quantified (3–5) and although it does not have an identified
pharmacological use, it shares a common precursor with hypericin and, in general,
the concentration of the two metabolites may be found to increase in tandem.
Hypericin is produced throughout the entire plant and accumulates in dark colored
glands that may be seen without the aid of a microscope (4). Hyperforin is produced
in the chloroplasts and stored in the clear glands that surround that tissue. Chemical
structures of hypericin and hyperforin are presented in Fig. 1. The medicinal plant
industry is currently supplied with plant material produced either through field
cultivation or wild craft. Field-cultivated H. perforatum has been found to vary widely
in hyperforin and hypericin content because of annual environmental variation as
well as variation among crop locations. Factors that have been shown to clearly
affect hypericin content include drought stress, heavy metal contamination of the
soil, light intensity and nitrogen availability (4–6). Exposure to UV light is a natural
elicitor of secondary metabolite responses in higher plants. There are three
subcategories of UV light which are separated according to wavelength. UV-A has
the longest wavelengths ranging from 400 to 315 nm and the least amount of energy
at 3.1–3.93 eV per photon. UV-B ranges from 315 to 380 nm (4.43–12.4 eV per
photon) while UV-C is the shortest at 280–100 nm and has between 8.28 and 124
eV per photon. UV-C is generally filtered out by the ozone layer and is not a factor
in plant development. UV-A is less harmful to plants due to the lower energy level
per photon and is transmitted through glass, meaning that all plants grown in glass
greenhouses are exposed to UV-A radiation. UV-C radiation which can be obtained
through germicidal lamps will kill plants. Research on plant response to UV-B light
is often coupled with the decrease in total biomass production caused by UV-B
stress and attempts to predict the effect of a decreasing ozone layer. Reviews on
the general plant response to UV light can be found in Day (7) and Caldwell et al.
(8). Additionally, supplemental exposure to UV-B light has been shown to increase
the concentration of secondary metabolites in maize, basil, peanut and lettuce (9–
12). UV-B has also been associated with antifeedent properties in many higher
plants, leading to enhanced secondary product production in some plants (13). The
goal of this work is to explore the addition of practical quantities of supplemental UV-
B light to optimize UV-B exposure time ⁄ duration in order to maximize hyperforin,
hypericin and pseudohypericin content in plants grown in a controlled environment.
MATERIALS AND METHODS
Growing conditions and experimental treatments. Hypericum perforatum L. cv. New
Stem (Richter’s Herbs, Goodwood, ON) was rinsed with purified (reverse osmosis;
EC = 2 ls cm)1 ) water to remove germination inhibitors. It was triple seeded into
rockwool cubes (1550 plants m)2 center hole filled with sifted peatlite) and thinned
to one plant per cube 3 weeks after seeding, selecting for crop uniformity. Plants
were placed into one of three walk-in growth chambers (Model M1 Environmental
Growth Chambers, Chagrin Falls, OH), such that one plant per treatment per day
was sampled from each growth chamber. Initially the plants received 100 lmol m)2
s )1 for 2 weeks and after 14 days, the light intensity was increased to 400 lmol m)2
s )1 by fluorescent lamps (Sylvania, CW, VHO, Danvers, MA) for the duration of the
experiment. Rockwool cubes were transplanted 30 days after seeding into 15 cm
containers filled with MetroMix 360 (Scott’s Horticultural Products, Inc., Marysville,
CA). Planting density of St. John’s wort in each growth chamber was 36 plants m)2
. Exposure to UV-B radiation from lighting which specifically releases UV-B
irradiation (UV-B-313 bulbs; Q-Panel Co., Cleveland, OH) occurred on day 55.
During exposure, tops of the plants were 5 cm from the lamps and the intensity of
the UV light was 10 lmol m)2 s )1 (Apogee UV meter, Logan, UT). Single exposure
periods for UV-B irradiation were 10, 20, 40, 80 and 160 min. Three plants per
harvest period were exposed and samples were taken from each treatment at the
following time intervals: 12, 24, 48, 96 and 108 h. The repeated exposure treatment
received 10 min of UV light each day for 7 days. Three plants were sampled per day
and plants were sampled for 7 days starting on the initial day of exposure. An
additional experimental group (called the progressive exposure plants) received an
increasing exposure to UV-B light each day, starting with 10 min and increasing 5
min day)1 for 4 days and then by 15 min for the last 2 days as summarized in Table
1. Three plants were sampled per harvest and plants were exposed to light and
harvested for 7 days. Harvests were destructive and plants were not resampled.
Quantification of metabolites. Quantification of hypericin, pseudohypericin and
hyperforin was performed using a modification of the method of Couceiro et al. (3).
Upon harvest, 10 cm from the growing tips of the main stems of nine plants was
pooled into one mixed sample and immediately placed into an aluminum foil packet
and dropped into liquid nitrogen. One such sample was obtained for each replicate
for a total of three samples per light condition. The frozen material was ground in the
presence of liquid nitrogen into fine powder to which 4 mL of 2% (vol ⁄ vol)
dimethylsulfoxide in methanol was added to 1 g fresh weight, weighed after grinding.
The extracted solution was placed in a sonicating bath (8890 Cole Parmer) for 30
min and centrifuged at 3220 g at 4C for 15 min (5810 Eppendorf). From the
supernatant, 2 mL was filtered through a 0.2 lm Acrodisc PTFE syringe filter (Pall
Corp, East Hills, NY) and diluted two-fold with the same solvent. A portion of this
extract was placed into an amber vial for the evaluation of hyperforin and another
was placed into a clear vial for hypericin and pseudohypericin analysis (Waters Corp,
Milford, MA).

The above steps were performed under low light conditions at room temperature
with the aid of a photography darkroom red light to prevent degradation of hyperforin,
which will degrade upon exposure to UV light which exists in fluorescent lamps in
the laboratory. The clear vials were then placed 15 cm from a 100 W tungsten lamp
for 30 min to allow for the full conversion of protopseudohypericin to pseudohypericin
and protohypericin to hypericin. The three metabolites were quantified
simultaneously. A 20 lL sample of the extract was injected onto a Waters x Terra
C18 column (3.5 lm; 3.9 · 100 mm) with a C18 Waters x Terra guard column (3.9 ·
20 mm). The HPLC system utilized was a Waters 2695 Separations module with a
996 photodiode array detector with a detector range of 220–750 nm. The mobile
phases for this separation were as follows: A: 0.1% triethylammonium acetate
(Calbiochem) solution adjusted to pH 3.5 with acetic acid (Fisher Scientific,
Pittsburgh, PA); B: Acetonitrile (Fisher) adjusted to pH 3.5 with acetic acid (Fischer).
The flow rate was 1 mL min)1 with a gradient elution beginning with 50:50 (A:B) for
2 min increasing linearly to 20:80 in 12 min, isocratic at 20:80 for 3 min then linearly
increasing to 0:100 in 3 min and finally isocratic at 0:100 for 10 min after which the
flow was stopped for a total run time of 32 min. All solvents utilized were HPLC grade.
Hyperforin was quantified at 270 nm and hypericin and pseudohypericin at 588 nm.
Significant linear calibration curves were generated for hypericin, pseudohypericin
and hyperforin (standard retention times of 15, 11 and 17 min, respectively) and the
quantification of these compounds was calculated by comparison to a standard
curve. The limit of detection for hypericin was 20 ug g)1 FW. Standards were
purchased from Alexis Biochemicals (San Diego, CA).
RESULTS
Overview
Figures 2–7 show the responses of secondary metabolites in St. John’s wort shoot
tissues during and after treatment with UV-B light. The data presented have been
normalized such that the levels reported represent the increase in chemical
concentration relative to control plants harvested the same day. Visible tissue
damage to the tips of the leaves was observed 12 h after UV-B exposure in the 160
min treatment. For exposures of 40 min or greater, visible tissue damage could only
be observed after 48 h. If the plants were not re-exposed to UV-B, the new tissue
growth appeared healthy with a greater number of lateral branches. Single dose
experiment In the single dose experiment, both the dosage effect and the hours after
treatment effect were significant (P < 0.01) for all the metabolites tested. Each
metabolite showed significantly different concentrations based on the UV-B dose
administered and the concentration observed was also dependent on the elapsed
time after treatment. There was no interaction between dosage and harvest time, so
metabolite concentration levels followed the same pattern for all harvests and
dosages. This response is presented in Fig. 2 for the metabolite hyperforin, with
different letters denoting significantly different metabolite concentrations. Error bars
indicate variability between samples. The standard errors observed were greater
than expected for reasons that were unknown, although statistical analysis indicated
that the error was not significant. Raw data may be found in Brechner (14). The
highest levels of hyperforin were seen in the 40 and 80 min single dose treatments
with a maximum hyperforin concentration 2.5 times greater than the control.
Intermediate concentrations were observed in the 10 and 20 min treatments with a
maximum hyperforin concentration averaging two times greater than the control. The
lowest concentration was observed in the 160 min treatment with a maximum of 1.25
times the control values. The pattern of response by the treatments was consistent
across all harvests. Visible damage to the plants was observed 24 h after UV
application and occurred in all treatments of 40 min in length or longer. After UV
exposure was terminated, plant growth observed appeared normal. Time elapsed
after treatment was significant and all harvests could be classified into two significant
rankings for hyperforin. The highest concentrations of hyperforin were seen in the
12, 24 and 48 h harvests averaging 1.5 times control values for all three harvests
and the lowest concentrations shown at the 96 and 144 h harvests with the average
metabolite response being equal to the control values. Concerning time of exposure
to UV-B, the highest levels of pseudohypericin (Fig. 3) were also associated with the
40 and 80 min treatments. An intermediate level was found with the
20 min treatment and the lowest levels were seen with the 10 and 160 min
treatments. The harvest performed at 12 h showed the highest pseudohypericin
concentration at 2.5–3.5 times control values with intermediate concentrations
observed at 24, 48 and 96 h, with values equal to the control. The lowest
concentration was found at 144 h after exposure with metabolite concentrations at
half the control value. Similarly, hypericin (Fig. 4) showed the highest increase in
concentration compared to the control at a 40 min exposure time with a maximum at
1.8 times the control value. Intermediate concentrations were observed at 20 and 80
min exposure times, averaging 1.4 and 1.1 times the control value. The lowest
concentration was seen with the 10 and 160 min exposures, averaging 0.3 and 0.4
times the control value. The lower concentration found in the 160 min treatment may
be attributed to damage done to the plants. The time after treatment that showed the
greatest metabolite concentration was 96 h, averaging 1.2 times the control value
over the course of the experiment. The lowest concentrations were found at 12 h
after exposure and were equal to the control values. In general, a clear pattern of
metabolite formation was observed immediately following the application of UV
stress. The metabolite response, from highest to lowest concentration per g shoot
tissue, was as follows: 40, 80, 20 min and finally, 160 or 10 min. Although a
physiological explanation for this particular response pattern could not be
immediately determined, optimal UV-B exposure may result in induction of other
critical plant protective systems over time, including the likely production of free
radical scavengers required to protect plant photosynthetic apparatus and fragile
membrane structures from excessive UV exposure, thereby limiting resources
available for hypericin and hyperforin production. In addition, 160 min treatment
could result in immediate, irreparable damage to the plant and its membrane
systems. Repeated and progressive experiment UV-B exposure in the repeated and
progressive exposure models did not significantly influence hyperforin
concentrations; however, the time after treatment did have an impact on metabolite
production (Fig. 5). Assessment of metabolite concentration on day 4 showed
significantly (P < 0.01) higher levels of pseudohypericin than all other days
evaluated, with pseudohypericin averaging 1.5 times the control value.
Concentrations of pseudohypericin on days 2 and 6 were significantly lower than on
all the other days averaging 0.6 times the control. Although no clear explanation for
this pattern exists, pseudohypericin levels may peak over time and may later
degrade naturally over time without repeated UV exposures (Fig. 6). In our
experiments, it is important to note that there were no significant deviations in
temperature, light quality and quantity and nutrient solution composition during the
course of the experimental period. Additionally, the plants were not exposed to insect
or disease pressure at any point during this experiment. For pseudohypericin the
light quality treatment did not significantly impact concentrations; however, time of
harvest did have a significant impact. Highest concentrations were observed at day
4 harvest where the values averaged 1.3 times the control, intermediate
concentrations were observed at days 1, 3, 5, 6 and 7 with average values equal to
the control and lowest concentrations were seen at day 2. For hypericin (Fig. 7), the
UVB exposure was significant (P = 0.0001) with the progressive treatment providing
higher hypericin concentrations (averaging 0.1 times higher) for all the harvests than
the repeated treatment. The time of harvest after treatment was not significant.
DISCUSSION
In general, the maximum response in metabolite production in St. John’s wort tissue
was demonstrated within 24 h after the first UV-B exposure in all treatments
evaluated. The induction of secondary metabolite production by higher plant tissues
is widely recognized in response to UV treatment and other plant stressors.
Depending on the plant, the specific metabolite and the stressor, variation in
metabolite production over time is frequently observed. Furthermore, variation within
plant. Normalized Pseudohypericin (ug/gFW) Days from start of UV treatments
Continuous, 702-1260 ug/gFW Gradual, 430-1610 ug/gFW Figure 6. Normalized
pseudohypericin concentration over time for repeated and progressive dose of
supplemental UV-B experiment. Average values represent five plants. Bars
represent ±standard error. Different letters at each time represent significantly
different values. Lower-case letters represent the overall effect caused by the
harvest day and upper-case letters represent significantly different
treatments.Normalized Hypericin (ug/gFW) Days from start of UV treatment
Continuous, 15-73 ug/gFW Gradual, 18-126 ug/gFW Figure 7. Normalized hypericin
concentration over time for repeated and progressive dose of supplemental UV-B
experiment. Average values represent five plants. Bars represent ±standard error.
Different letters at each time represent significantly different values. Lower-case
letters represent the overall effect caused by the harvest day and uppercase letters
represent significantly different treatments. Photochemistry and Photobiology, 2011,
87 683 metabolite concentration over time was observed. It is hypothesized that
these metabolites are often unstable and are typically degraded and recycled in the
plant. In other studies evaluating UV-induced responses in secondary product
formation by higher plants, similar trends have been observed. In table grapes, short
to intermediate exposure to UV light has resulted in maximal production of
resveratrol, whereas longer or shorter treatments resulted in potential degradation
of metabolites over time or potential damage due to the inhibitory effects of the UV
treatment itself (15). Peanut plants that were exposed to UV-B light showed an
increase in the secondary metabolite resveratrol that was an order of magnitude
greater 12 h after exposure, compared to 3 h after exposure (11). Given the expense
of maintaining the plants under controlled conditions for additional time and limited
increases observed in metabolites with delayed harvest, it is recommended that the
UV-B light challenge be administered within 24 h prior to harvest for maximal
metabolite recovery. One exception to this in our studies was observed with the
metabolite hypericin, where the greatest enhancement in harvested St. John’s wort
tissues was determined to be 6 days after UV-B challenge. However, the final
hypericin concentration, while much greater than that in the control plants, was not
significantly greater than the values observed in plants at the time of flowering.
Physiologically, when most flowers are open on the plant and not yet senescing, up
to 10 times greater hypericin contents can be observed in harvested tissue.
Hypericin content appears to be maximal at the time of flowering and then gradually
decreases with flower senescence (data not presented). The maximum response
elicited in this experiment was shown in the single dose experiment and included a
3.7% increase in hypericin concentration compared to control plants. The repeated
and progressive treatments of UV light exposure over time did not result in
significantly greater increases in metabolite concentration in comparison with the
single dose treatment and involved more labor to complete. This treatment type was
only significant in the induction of increased hypericin concentration. To our
knowledge, this is the first reported study of impacts of repeated UV-B exposure
upon metabolite formation and retention in St. John’s wort tissues. For the purposes
of increasing secondary metabolite production, we observed that a single dose of 40
min of UV-B light followed by harvest (within 12 h after exposure) was optimal. The
40 min treatment resulted in the shortest amount of exposure time and the fastest
time to harvest. However, it must be cautioned that when comparing the natural
levels of metabolites produced when the plants are at the peak of flowering before
setting seed (on average 60 days after seeding), the increases observed in this study
(5% or less) due to UV-B light exposure do not merit harvesting the plants before full
flowering is achieved (14). Future work should include the exploration of metabolite
response specifically during the response period from 0 to 24 h after exposure to
determine if 12 h exposure indeed produces the optimal response in metabolite
production. From our studies, we would recommend utilization of a single 40 min
dose of UV-B exposure rather than multiple doses. Additionally, it would be
worthwhile to expose plants in full bloom to UV-B before harvest to determine if an
increase in metabolite production, beyond the normal increase encountered during
flowering, can be achieved. Finally, surrounding the plants on all sides with UV
irradiation rather than overhead exposure alone would also be worthwhile to
evaluate in an attempt to enhance overall metabolite production in the shortest
production period. Acknowledgements—This work was supported in part by federal
formula funds. The authors would like to thank Matthew Balestrino for helping with
data collection and quantification. I would also like to thank Dr. Donald Krizek for
consulting on the UV-B treatment methods and Roselee Harmon for assisting with
laboratory experimentation and HPLC separation.

Submerged culture conditions for the production of mycelial biomass and

antimicrobial metabolites by Polyporus tricholoma Mont.
Condições de cultivo em meio líquido para produção de biomassa e metabólitos
antibacterianos por Polyporus tricholoma Mont.

Gladys Rosane Thomé VieiraI,*; Mariane LieblII; Lorena Benathar Ballod

TavaresII; Roberta PaulertIII; Artur Smânia JúniorIII

IDepartamento de Ciências Naturais, Universidade Regional de Blumenau,

Blumenau, SC, Brasil e Programa de Pós-Graduação em Biotecnologia,
Universidade Federal de Santa Catarina, Florianópolis, SC, Brasil
IILaboratório de Engenharia Bioquímica, Departamento de Engenharia Química,
Universidade Regional de Blumenau, Blumenau, SC, Brasil
IIILaboratório de Antibióticos, Departamento de Microbiologia e Parasitologia,
Universidade Federal de Santa Catarina, Florianópolis, SC, Brasil

ABSTRACT

Basidiomycete fungi of the Polyporus genus are a source of secondary metabolites

which are of medicinal interest as antibacterial compounds. As these substances are
produced in a small amount by the fungi, the study of the cultivation conditions in
vitro that could possibly optimize their production seems of major importance. The
effects of glucose and lactose, pH and agitation on biomass concentration and on
the specific growth rate caused by the basidiomycete Polyporus tricholoma were
investigated. The initial pH (4.5, 6.5 and 8.5) was autoregulated at pH 5.5, and the
agitation increased the mycelial growth and the specific growth rate. The high
concentration of carbon sources (4%) increased biomass production. The lactose
concentration and the absence of agitation were determinant in the production of
antibacterial metabolites. The characterization of the antibacterial substance by GC-
MS indicated a major compound, isodrimenediol, produced by the fungus Polyporus
tricholoma with activity against Staphylococcus aureus.

Key-words: Polyporus tricholoma, specific growth rate, antibacterial activity,

isodrimenediol.
RESUMO

Os fungos basidiomicetos do gênero Polyporus são fonte de metabólitos

secundários de interesse medicinal como os compostos antibacterianos. Como
estas substâncias são produzidas em pequenas quantidades pelos fungos, o estudo
de condições de cultivo in vitro que otimizem sua produção, é de fundamental
importância. Os efeitos da glicose e lactose, pH e agitação na concentração da
biomassa e na velocidade específica de crescimento realizada pelo
basidiomiceto Polyporus tricholoma foram investigados. O pH inicial (4.5, 6.5 e 8.5)
foi autoregulado para pH 5.5, e a agitação aumentou o crescimento micelial e a
velocidade específica de crescimento. A maior concentração das fontes de carbono
(4%) incrementou a produção de biomassa. A concentração de lactose e a ausência
de agitação foram determinantes na produção dos metabólitos antibacterianos. A
caracterização da substância antibacteriana por CG-EM mostrou como componente
majoritário o isodrimenediol, produzido pelo fungo Polyporus tricholoma, com
atividade contra Staphylococcus aureus.

Palavras-chave: Polyporus tricholoma, velocidade específica de crescimento,

atividade antibacteriana, isodrimenediol.

INTRODUCTION

Aiming at obtaining bioactive compounds from mushrooms, several researches have

tried to cultivate medicinal mushrooms in solid artificial media for fruit body
production rather than in submerged cultures for mycelial extract (13). However,
submerged cultures have the potential for a higher mycelial production in a shorter
period of time within a reduced space if compared to cultivation in solid artificial
media. Therefore, it is essential to optimize the fermentation conditions in order to
produce biomass and a high rate of compounds with biological and pharmacological
activities.

The fungal morphology is an important parameter that influences the mycelial growth
rate and the physical properties of the fermentation broth in submerged cultures. The
rheological behavior, for instance, is closely related to the morphology and biomass
concentration, and it is exactly the broth's rheology that determines the
transportation phenomena in bioreactors, which is the key to improve the bioprocess
(21). On the other hand, the growth kinetics and the fungal morphology in broth
fermentation are highly dependent on the culture conditions, such as carbon
sources, C/N relation, initial pH and temperature, agitation intensity and aeration rate
(20). A large amount of work has reported the effects of environmental parameters
on the biomass concentration and on the yield of bioactive compounds by
mushroom, but those researches are limited to few kinds of mushrooms
(4,7,10,12,13,20,25).

Polypores constitute a large group of fungi characterized by hymenophores formed

by innumerous pores. They belong to the phylum Basidiomycota, and, along with
some fungi of the phylum Ascomycota, they are considered as an important source
of substances of medicinal interest, such as antibacterial metabolites
(27). Polyporus arcularius has been reported to synthesize norsesquiterpene
structures, such as drimanes, with antimicrobial activity against Gram-positive
bacteria like Staphylococcus aureus (8).

Despite several investigators have studied different antimicrobial fractions

from Polyporus, optimal submerged culture conditions have not been extensively
determined so far (3,8,27). Then, the present study aimed at optimizing the
submerged culture conditions for the production of micelial biomass by two species
of Polyporus (P. tricholoma and P. tenuiculus). For a comparative study, a set of
fermentations was carried out.
MATERIAL AND METHODS

Organism and culture conditions

Two species of fungi were used, Polyporus tricholoma CCB-684 and Polyporus
tenuiculus CCB-685. Both of them are native from Brazil and belong to the culture
collection of basidiomycetes (CCB) of the Botanical Institute of São Paulo (Brazil).
The fungi were cultivated in potato dextrose agar (PDA) and, after being isolated,
they were maintained at 4ºC in the same media. The two fungal species were
cultivated in 100 mL of potato dextrose broth (PDB) and MALT medium (malt extract
with soy peptone broth/Difco®). The flasks of each media were inoculated with five
discs of 7mm of diameter of fresh mycelium, of the species P. tenuiculus and P.
tricholoma, on PDA in separated experiments. The flasks were incubated at 25ºC,
under aerobic conditions, in the absence of light, for a period up to 30 days and the
tests were carried out in triplicate. During this period, the biomass (g/L), pH values
and antibacterial activity were determined after every 5 days of cultivation.

Optimization of the cultivation conditions

For optimization tests only P. tricholoma was used. Twenty plugs of mycelium in
PDA were added to 400 mL of 2.4% PDB with 1% malt extract and 0.1% soy
peptone. The culture was incubated at 25ºC for 5 days. Afterwards, 10 mL of culture
were transferred to flasks containing 90 mL of the 2.4% PDB and 0.1% malt extract
(MP broth) as culture media. The cultures were incubated at 25ºC, under aerobic
condition, in the absence of light, for 15 days. During this step it was studied the
following parameters: initial pH: 4.5; 6.5; 8.5; agitation: 150 rpm and absence of
agitation; lactose: 1 and 4%. In later tests the pH value of 4,5 was maintained and
glucose in concentrations of 1 and 4% was also tested.

Determination of cell dry mass

In order to determine the fungal biomass at specific time intervals, the mycelium was
filtered through filter paper (Whatman nº 40), washed with distilled water and dried
at constant weight at 80ºC. The mycelium was placed in the desiccator and then the
mass was determined.

Specific growth rate

Aiming at determining the specific growth rate (µ), natural log of biomass (lnX) was
plotted against time (t). The slope of the line at any moment gives the specific growth
rate at each moment.

Extraction and characterization of antibacterial metabolites

The mycelium was removed through filtration and the metabolites, present in the
filtrates, were extracted with ethyl acetate, concentrated in a rotavapor and the
residue was weighted. The extract in ethyl acetate was characterized by GC-MS (70
eV), performed in a Varian Saturn 2000 GC/ MS spectrometer in split injector mode.
A CP-Sil-8CB capillary column (30 m x 0.25 mm, 0.25 mm film thickness) was
operated at 60ºC for 3 min, and then programmed for 60º-220ºC at 5 ºC/min, after
which it was kept isothermal at 220ºC for 5 min. The carrier gas was helium and the
injector temperature was of 250ºC. The components of the extract were identified by
comparison of fragmentation patterns in mass spectra with those stored on the
spectrometer database and reported in the literature. The relative percentage of
individual components was calculated from the GC peak areas.

Antibacterial activity tests

Overnight cultures of Staphylococcus aureus ATCC 25923 and Escherichia

coli ATCC 25922 strains were diluted to a final concentration of 10 8CFU/mL. The
bacterial suspensions were spread over the surface of Mueller-Hinton agar,
containing five wells of 7 mm of diameter. The wells were filled with 50 µl of both
extracts (5 mg) which were dissolved in 50 µl of dimethilsulfoxid (DMSO). The
chloramphenicol disks (30 µg) and 50 µl of DMSO were used as controls. The plates
were incubated at 37ºC for 24 hours. The reading was carried out by measuring the
growth inhibition zones, and a halo greater than 9 mm was considered positive. In
order to evaluate the minimum inhibitory concentration (MIC), the total amount of the
extracts dissolved in DMSO was diluted in Mueller Hinton broth and dispensed into
each microplate well. The bacterial suspension in a concentration of 10 8 CFU/mL
was inoculated in the plate and incubated at 37ºC for 24 hours. The reading was
carried out using an ELISA reader, at 630 nm, and it was performed with r-
iodonitrotetrazolium violet (INT) (22).

RESULTS AND DISCUSSION

Influence of the carbon and nitrogen sources on the growth of the Polyporus

To investigate the effects of nitrogen and carbon sources over mycelial growth, two
kinds of culture media were examined (Fig. 1). Neither the mycelial concentration
of P. tenuiculus, nor the growth rate were influenced by the media as long as very
low values (0,066 g/L.day) were found, if compared to the ones obtained for the other
strain (0,83 g/L.day). On the other hand, the medium containing glucose (PDB)
provided a lower growth to the strain P. tricholoma, in comparison to the MALT
medium. In fact, it was possible to notice a significant difference in the biomass
production of P. tricholoma in the inoculated media. This result reflects both the
effect of nitrogen and the specie over the mycelial growth. A significant growth
of Psathyrella atroumbonata was obtained when it was used malt extract and L-
triptofano while in the medium with sodium nitrate and ammonium sulphate the
mycelial production presented the worst result. The stimulating effect of the two
organic sources can be attributed to their carbon and nitrogen content (12). Despite
the fact that both the malt extract and the corn juice are used as nitrogen sources,
they are also carbon sources. The nitrogen content of the corn juice is of 6,7% and
the carbon content is of 36,3% in dry weight. As the fungi have efficient enzymatic
systems, they can easily break molecules more complex than glucose for their
development (14).
The maximum rate of mycelial growth (µ) of P. tenuiculus (Table 1) was similar in
both media (0,022 dya-1), as well as the cellular productivity (Px). The specific growth
velocity showed a short increase in the PDB medium, while in the medium containing
nitrogen (MALT) it was observed an exponential increase in the mycelial growth,
followed by the stationary phase, after the thirteenth day of cultivation.

Correlations between kinetics growth rate, substrate and antibacterial activity

The antibacterial activity of the two species that were used in the medium is shown
in Fig. 2. The production of antibacterial substances by P. tenuiculus was not
detected in any of the cultivation media in relation to the bacteria which was being
tested. According to the studies developed by Cabrera et al. (3), tests of the activity
of this fungus against the same bacteria that was used in this study were also
negative. The fungus P. tricholoma did not show activity against Escherichia coli in
the agar diffusion test; however, it showed activity against Staphylococcus aureus,
in both cultivation media, after the 13th day of incubation. The growth inhibition halos
of S. aureus, which point out the presence of the antibacterial substance in the fungic
extracts of the MALT medium, were higher than the ones of the PDB medium. These
results may be explained by the fact that the medium used was a source of organic
nitrogen. The soy peptone used in the production of actinomycete antibiotics has the
advantage of being slowly metabolized and, therefore, repressive ammonia salts
and aminoacids do not accumulate (17). When establishing a relation with the
phases of the mycelial growth curve, it was observed that the most pronounced
inhibition took place in the stationary phase present in the MALT medium, and the
inhibitory activity occurred during the same growth interval, independently of the
media. It can possibly be an indication that the production of the substance takes
place when the average growth velocity is low or constant. These results show that
the best growth is not always correlated with a greater production of the secondary
metabolite. Bu'Lock (2) called the period of secondary metabolite production as
idiophase due to the special nature (idiosyncratic) of the products.
The influence of pH in the synthesis of antimicrobial compounds is a factor that has
been described using different fungic species. In the present work, the pH did not
influence the production of the secondary metabolite. However, there was a
statistical difference in the case of P. tricholoma extracts when the MALT medium
was used and at 17 days and 22 days of cultivation (Fig. 2).
Culture conditions for the production of biomass and antibacterial metabolite

In the cultivation of P. tricholoma with different concentrations of the carbon sources,

pH and agitation, it was observed that there was an average adjustment of the
medium pH to values close to 5,5 (Fig. 3). Regardless the initial pH and other factors,
there was an average adjustment of the pH for final cultivation values of 5.0 and 6.0.
Some species of fungus show the characteristic of autoregulating the substrate pH,
stabilizing it at the value that is considered optimum for its development (19).

The inhibitory property of the fungus P. tricholoma was evaluated under different
cultivation conditions, against S. aureus. Fungal metabolites were extracted with
ethyl acetate and they were tested through the diffusion method. Positive results
were observed for the treatments described in Fig. 4.
The statistical analysis revealed that all factors (pH, lactose concentration and
agitation) influenced (p<0.05) the production of antimicrobial substances by P.
tricholoma, besides indicating a significant interaction between those factors.
However, the factor which showed the greatest isolated effect was the lactose
concentration. The antimicrobial activity increased with the increase in the
concentration of this carbohydrate. For fermentation processes aiming the
production of secondary metabolites, it seems better to use polysaccharides, such
as starch; oligosaccharides, such as lactose, and oils such as soy oil (1,5). An
increase in the biomass and Tremella mesenterica exopolysaccharides occurred
when there was an increase in the initial concentrations of sucrose from 1 to 6% (7).
The greater number of cultures without agitation inhibited the bacterial growth;
however, treatment 4, under agitation at 150 rpm, gave the largest halo. Some
authors report the production of pellets, which may be caused by agitation, as being
a good event to promote the production of antibiotics and other metabolites. The
production of citric and itaconic acids by Aspergillus niger was increased when the
fungus showed the morphology which is mentioned above (6). On the other hand,
disperse growth is preferable for the production of penicillin by Penicillium
chrysogenum (23), and cephalosporin by Cephalosporium acremonium (18).

By analyzing the different initial pH values used in the treatments, it is possible to

notice that they show to have an isolated effect of synergistic nature over the other
studied factors. Higher initial pH values were shown to have better effects over the
antimicrobial metabolites production, as it was demonstrated by Smânia et al. (22)
for the production of the antibacterial cinnabarin. However, an initial medium pH of
5,5 improved the production of biomass and exopolysaccharides by Grifola
frondosa (13).

Effect of agitation and carbohydrate concentration over the production of

biomass and antibacterial metabolite

The mycelial growth of P. tricholoma was higher for the cultivations under agitation
than for static cultivations. There was a significant difference (p<0.05), when
analyzing the biomass values over the time. It can be seen that the treatments with
the highest concentration of carbohydrates, T3 and T4 (glucose 4%), T7 and T8
(lactose 4%), resulted into the greatest growth, and lactose showed to be the best
source of carbon for the mycelial production (Fig. 5).

There is a higher cellular growth for treatments under agitation than for those which
are kept static. On the other hand, there was no significant interaction with the
carbohydrate concentration. The cell production was due to the homogenization and
heat transference when physical and chemical conditions are maintained
homogenous and more oxygen becomes available (16). An agitation at 50 rpm was
used to increase the production of biomass and polysaccharides by Phellinus
linteus (10), while for the culture conditions of polypore Antrodia cinnamomea an
agitation of 100 rpm increased the mycelial growth (26).

Table 2 summarizes the effect of agitation over the specific growth velocity of P.
tricholoma with carbon sources (glucose and lactose). The maximum specific growth
velocity and the productivity in cells increased with agitation.

The antibacterial activity of the extracts against the bacteria S. aureus started only
on the 14th day (Fig. 6). The statistical analysis for the 21st. day, when the largest
halos were obtained, shows that there are differences between treatments T2, T4
and T7. Those differences are connected to the best source of carbon for growth
and the one that promotes the production of secondary metabolites. Glucose, which
is an excellent source of carbon for growth, depresses the synthesis of a series of
metabolites such as actinomycin and cephalosporin. However, it does not interfere
in the production of aminoglycosides and chloramphenicol (15).
Along with glucose, other carbon sources and other medium nutrients may depress
the synthesis of antibiotics (17). A comparative study using various carbohydrates
showed that lactose is the best source of carbon for the promotion of mycelial growth
and for the production of ganoderic acid and polysaccharides in the case
of Ganoderma lucidum (25). For the cultivation conditions which were used, and
from the obtained results, it can be seen that the global physiological regulation is
very complex due to the diversity of fungi and the variety of metabolic pathways and
controls (24).

Characterization of the antibacterial substance

Spectral data of the ethyl acetate extract showed the presence of a sesquiterpenic
structure of the drimanes class (Fig. 7). The main product had a relative
concentration of 29.49% with a retention time of 32.941 min. It was present between
C20 (R.T. 32.506 min) and C21 (R.T. 34.354 min) and it was identified by the NIST 98
MS Library as Isodrimenediol (C15H26O2) (5-Hydroximethyl-1,1,4a-trimethyl-6-
methylene-decahydronaphtalen-2-ol). The characterization by EIMS (70 eV)
showed m/z (%)=238 (M+, 1); 220 (M+-H2O, 8); 202 (M+-2H2O, 12); 187 (18); 176
(16); 152 (23); 135 (100); 119 (28); 107 (63); 91 (43); 79 (39); 67 (24); 55 (13).

The isodrimenediol, which was here isolated from P. tricholoma, was described by
Fleck at al., when obtained from the species Polyporus arcularius (8), as a
biosynthesis intermediate of the drimanes norsesquiterpenes of this fungus. The
isolation of these compounds from a fungal strain which also produces other
drimane-type structures, indicates that these compounds may be biosynthesized via
isodrimenediol as an intermediate. The total extract was used to evaluate the
minimum inhibitory concentration (MIC) and the result for the S. aureus was of 1,0
mg/mL. The moderate activity of isodrimenediol and 7-drimano-3,11,12-triol against
Gram-positive bacteria such as Staphylococcus aureus SG511, and yeasts such
as Sporobolomyces salmonicolor 549 was reported by Fleck et al. (8). Recent
studies have revealed new sesquiterpenes, the isocriptoporic H and I acids obtained
from P. arcularius, while the criptoporic H acid and its methyl ester have been found
in Polyporus ciliatus (3). There is the hypothesis that the three terpene alcohols,
grouped in the farnesane group, act on the cellular membrane of S. aureus (11).
Isodrimenediol is an intermediate of the biosynthesis of terpenes of the
bicyclofarnesane group (9). Since the chemical structures of the two groups are
similar, there is a possibility that the action mechanism of isodrimenediol over S.
aureus follows the model described for the farnesane group.

In conclusion, the results indicate that the production of biomass of P. tricholoma in

MALT medium is higher than the production of biomass of P. tenuiculus. The
antibacterial activity was observed for the Gram-positive Staphylococcus aureus in
the stationary phase, with the MIC of 1,0 mg/mL. The agitation and the higher
concentration of carbohydrates incresead the biomass, but the treatments which are
kept static, with 4% of lactose, showed the best antibacterial activity.
ACKNOWLEDGEMENTS

We hereby thank Marina Capelari and Adriana Gugliotta, researchers of the São
Paulo Institute of Botany, for kindly providing the fungus species used in the study.
Isolation and screening for plant growth-promoting (PGP) actinobacteria
from Araucaria angustifolia rhizosphere soil

Isolamento e seleção de actinobactérias promotoras do crescimento de

plantas de solo rizosférico de Araucaria angustifolia

Rafael Leandro Figueiredo de VasconcellosI; Mylenne Calciolari Pinheiro da

SilvaII; Carlos Marcelo RibeiroII; Elke Jurandy Bran Nogueira CardosoIII,*
IUSP/ESALQ Programa de Pós-Graduação em Solos e Nutrição de Plantas, C.P.
09 13418-900 Piracicaba, SP Brasil
IIUSP/ESALQ Programa de Pós-Graduação em Microbiologia Agrícola
IIIUSP/ESALQ Depto. de Ciência do Solo Laboratório de Microbiologia do Solo, C.P.

09 13418-900 Piracicaba, SP – Brasil

ABSTRACT

Actinobacteria are capable of playing several different roles in soil ecosystems.

These microorganisms affect other organisms by producing secondary metabolites
and are responsible for the degradation of different complex and relatively
recalcitrant organic compounds. In our survey of actinobacteria isolated from the
rhizosphere of Araucaria angustifolia, five culture media (AI, WYE, YCED, MSSC
and LNMS) were compared for their effectiveness in isolating these microorganisms.
When summing up all the isolates randomly obtained, we got 103 isolates. After
isolation, the phosphate-solubilizing ability and the "in vitro" production of indole-
acetic acid and chitinases were evaluated. The AI medium was ineffective for
actinobacteria isolation, when it was compared with the other four culture media.
Indole-acetic acid and chitinase were produced by respectively 36% and 24% of the
strains tested. However, only 2% of the 103 strains presented some phosphate-
solubilizing ability. These results demonstrate the biotechnological potential of these
microorganisms.

Key words: Brazil Pine, selective culture media, indole-acetic acid, chitinase,
phosphate solubilization
RESUMO

Actinobactérias são capazes de desenvolver diferentes funções no ecossistema

edáfico. Esses microrganismos, além de interagir com outros grupos de
microrganismos e plantas, ao produzir metabólitos secundários, são responsáveis
pela degradação de diferentes compostos orgânicos. Com intuito de facilitar os
estudos envolvendo actinobactérias encontradas em sistemas florestais, cinco
meios de cultura (AI, WYE, YCED, MSSC e LNMS) foram avaliados quanto à
eficiência em isolar estes microrganismos. Além disso, foi analisada a capacidade
dos diferentes isolados em solubilizar fosfato de cálcio, produzir ácido indol-acético
e quitinases. Dos cinco meios de cultura testados, somente o AI foi ineficiente em
isolar actinobactérias. A produção de ácido indol-acético e quitinases foi observada
em 36% e 24% dos isolados analisados, respectivamente. Contudo, apenas 2% dos
103 isolados foram capazes de solubilizar fosfato de cálcio. Estes resultados
comprovam o potencial biotecnológico desses microrganismos.

Palavras-chave: Pinheiro Brasileiro, meios de cultura seletivos, ácido indol-acético,

quitinase, solubilização de fosfato

Introduction

Araucaria angustifolia is a tree found in the Brazilian Pine Forest that belongs to the
Atlantic Forest Biome. This tree has socio-economic and environmental importance
(Lima and Capobianco, 1997), and its seeds are consumed by the fauna and are
also appreciated by human beings (Carvalho, 1994). Due to the intense
deforestation, less than 2% of the original Araucaria forest in Brazil is still remaining
(Guerra et al., 2002). Therefore, A. angustifolia is classified as a critically
endangered species, and studies involving its preservation and management,
including the study of associated soil microorganisms, are urgently required.

An important group of microorganisms, that is almost unexplored in Brazil, comprises

the actinobacteria (Streptomycetes) (Araújo et al., 2000; Gomes et al., 2000; Sousa
et al., 2008). These filamentous organisms have a high G + C, propagate themselves
by spores and grow through the soil in the form of hyphae (Stackebrandt et al., 1997).
The actinobacteria are widely distributed in nature, being commonly found in soil,
especially in the rhizosphere (Merzaeva and Shirokikh, 2006; Sousa et al., 2008).
By means of their secondary metabolites, these microorganisms can influence plant
development producing plant hormones, increasing the availability of mineral
nutrients and also excreting antibiotics or toxins that act in the biological control of
pathogens in the rhizosphere (Crawford, 1978; Hamdali, 2008; Miyadoh, 1993).
They are well known for their extensive production of compounds with phosphate-
solubilizing ability and compounds with high biotechnological potential such as
chitinases, and antibiotics (Gomes et al., 2000; Hamdali et al., 2008; Miyadoh, 1993).

The first objective of this study was to evaluate the effectiveness of selective culture
media for the isolation of actinobacteria from the A. angustifolia rhizosphere.
Moreover, the potential of the obtained isolates to promote plant growth and
biocontrol was analyzed through the evaluation of indole-acetic acid and chitinase
production. The phosphate-solubilizing ability was also tested.

Material and Methods

The rhizospheric soil adhered to roots was sampled in the native forest of A.
angustifolia , in the state of São Paulo, Brazil, located in the Mantiqueira mountain
range (22°44' S; 45º30' W, altitude 1,450 m). Sampling was performed around nine
trees. For each tree, three subsamples of roots with adhering rhizospheric soil were
collected and homogenized, resulting nine composite samples.
The actinobacteria were isolated from rhizospheric soil by the serial dilution
technique in Petri dishes, containing one of the culture media used by Crawford et
al. (1993): AI (Actinomycetes Isolation Agar: K2HPO4, 0.5 g; MgSO4.7H2O, 0.1 g;
FeSO4.7H2O, 1.0 mg; sodium caseinate, 2 g; asparagine, 0.1 g; sodium propionate
4 g; glycerol, 5 g; agar, 15 g; distilled water, 1000 mL), YCED (Casamino Acids Yeast
Extract Glucose Agar: yeast extract, 0.3 g; casamino acids, 0.3 g; D-glucose, 0.3 g;
agar, 20 g; distilled water, 1000 mL), WYE (Water Yeast Extract Agar: yeast extract,
0.25 g; K2HPO4, 0.5 g; agar, 20 g; distilled water, 1000 mL), LNMS (Low-Nutrient
Mineral Salts-Agar: yeast extract, 0.1 g; K2HPO4, 2 g; NaCl, 0.2 g; MgSO4.7H2O,
0.005 g; CaCO3, 0.05 g; FeSO4.7H2O, 0.01 g; soluble starch, 0.1 g; agar, 20 g;
distilled water, 1000 mL) and MSSC (Mineral Salts Starch Casein Agar: K 2HPO4, 2
g; NaCl, 2 g; MgSO4.7H2O, 0.05 g; CaCO3, 0.02g; FeSO4.7H2O, 0.01; soluble starch,
10 g; KNO3, 2 g; casein, 0.3 g; agar, 20 g; distilled water, 1000 mL).

The dilutions (101 to 10-5) were plated in duplicate, and the plates were incubated at
28°C for 10 days to allow for spore formation. After incubation, dry, powdery and
small colonies with extremely slow growth were counted and selected randomly from
mixed plate culture and transferred by streaking to a new plate with the same culture
medium for purification. It is possible that actinobacterial colonies belonging to
different families that do not show these characteristics, and could not be
distinguished from other bacteria by macroscopic visualization were missed. The
pure colonies were transferred to plates containing ISP2 culture medium (Pridham
et al., 1957) (10 g glucose, 5 g malt extract and 5 g yeast extract with the pH adjusted
to 7.2, 15 g agar and 1000 mL distilled water), and incubated at 28°C for one week.
Subsequently, each isolate was stored in 40% glycerol. The isolates were replicated
every 3 months.

All strains isolated from A. angustifolia soil were tested for the production of indole-
acetic acid with a colorimetric method described by Bric et al. (1991), with an
adaptation that consisted in reading absorbance values in microplates. Each isolate
was inoculated in test tubes containing 3 mL of Luria Bertani (LB) culture medium,
supplemented with 1 g L1 of the precursor of indole-acetic acid (L-tryptophan), and
incubated at 28°C for 5 days under constant agitation. Afterwards, 1.5 mL of each
homogenized culture was transferred to microtubes and centrifuged at 9500 g for
two minutes and 100 mL of the supernatant were transferred to microplate wells,
following the addition of 100 mL of the Salkowski reagent, consisting of 1 mL of 0.5
mol L1 FeCl3 and 49 mL of 35% HClO4.

Thirty minutes after the addition of the Salkowski reagent, the readings were carried
out at 530 nm, using a microplate reader Cary® 50 (Varian. Walnut Creek. CA. USA).
A reddish-pink color of the samples indicates the indole-acetic acid production.

To establish a standard curve a commercial solution of IAA was used.

The phosphate-solubilizing ability of the actinobacteria was evaluated by a modified

method described by Katznelson and Bose (1959), using inorganic P (CaHPO4) as
phosphate source. The culture medium was composed of 10 g glucose, 20 g agar,
2 g asparagine, 0.5 g MgSO4, 0.1 g NaCl, 0.1 g KCl, 0.5 g yeast extract, 1000 mL
distilled water and the pH was adjusted to 7.0 with 1M NaOH, before autoclaving.
Fifty mL of sterile 10% K2HPO4solution and 100 mL of sterile 10% CaCl2 solution
were added aseptically to the medium at 50°C and then the medium was dispensed
into the petri plates. The reaction between CaCl2 and K2HPO4 produced
CaHPO4 that turned the medium turbid and permitted the easy detection of
dissolution haloes around the colonies. The strains were transferred, in duplicate, to
petri dishes, each one being considered as an experimental unit. They were
incubated at 28°C for 10 days, and the phosphate solubilizers showed a clear
dissolution zone around the colonies.

The ability of the isolates to produce chitinases was assessed using a culture
medium with chitin as unique carbon source as described by Hsu and Lockwood
(1975): 4 g Chitin, 0.7 g K2HPO4, 0.3 g KH2PO4, 0.5 g MgSO4.5H2O, 0.01 g
FeSO4,7H20. 0.001 g ZnSO4, 0.001 g MnCl2, 20 g agar and 1000 mL distilled water.

After autoclaving, the pH was adjusted to 8.0, using sterilized 5N NaOH. The isolates
that formed a clear zone around the colony were considered as chitinase producers.
The statistical design was completely randomized with nine A. angustifolia trees as
replicates and two plates per dilution for each tree. Data were log-transformed and
were submitted to analysis of variance (ANOVA).Mean comparison was made using
the Tukey test (p < 0.01) (SAS 2002).

Results and Discussion

For isolation and quantification of the actinobacteria in rhizosphere soil of Araucaria

forests, the most adequate dilution of a ten-fold dilution series was 10-3, resulting an
average of ten colonies per plate. However, the number of contaminants was
greater, when comparing with the 10-4 dilution. The UFC number of actinobacteria
detected in this study is in line with Crawford et al. (1993) who reported a greater
amount of actinobacteria in rhizosphere in comparison with surrounding soil. Media
containing antibiotics were not used in our study because there are reports about
their action also against the actinobacteria themselves (Goodfellow and Williams,
1983).

Among the four culture media evaluated, the AI medium does not seem to be
favorable for most of these bacteria, since only very few colonies were formed. This
culture medium is widely used (Anzai et al., 2008; Gomes et al., 2001) but,
apparently, its effectiveness for isolation has not been previously compared with that
of other media. The AI medium has sodium propionate in its composition that can
inhibit bacteria and can also reduce the number of actinobacteria (Pereira et al.,
1996). The other four culture media were equivalent and brought satisfactory results
(Figure 1).
Of the 103 actinobacteria strains tested, 37 isolates (36%) were able to produce
indole-acetic acid. Six actinobacteria strains were outstanding (A4-8, L4-3, M4-6,
Y71, Y4-1, Y4-2), with a production above 40 µg mL-1(Figure 2); it is more common
to find values around 5 to 10 mg mL1 referred to in the literature (Hynes et al., 2008),
although much higher values also have been found. Microorganisms that produce
auxins, as indole-acetic acid, are widely distributed in soil (Akbari et al., 2007; Hynes
et al., 2008). Sousa et al. (2008) already stated that actinobacteria have a good
potential as indole-acetic acid producers in Brazil. These auxins stimulate the growth
and development of plants (Sousa et al., 2008). Therefore, they may be useful in the
management of forest species, mainly in nurseries or during transplantation to the
field.

Of the 103 isolates, 24% were able to produce chitinases. Chitinases may affect the
growth of pathogenic fungi, degrading their cell wall, which is composed primarily of
chitin. They have been used successfully in the control of plant diseases (Crawford
et al., 1993; El-Tarabily et al., 2000; Toledo and Cardoso, 1975). However, only three
strains (A4-1, Y2-3 and Y7-3) or 2% of all isolates showed phosphate-solubilizing
ability in vitro.

The microorganisms that interact with A. angustifolia have been relegated for many
years. Only lately some findings in this area have been reported, especially by our
group. This is the first time that the actinobacteria of the A. angustifolia rhizosphere
attracted the attention of microbiologists interested in finding PGP bacteria. Our
study was helpful to show the importance of this pine tree for researchers that work
with actinobacteria in Brazil. In fact, the culture media used had implications not only
in the isolation of actinobacteria but also to emphasize the great biotechnological
potential of these microorganisms.
Metabolite variability in Caribbean sponges of the genus Aplysina

Monica Puyanaa * Joseph Pawlikb James Blumc William Fenicald

aDepartmento de Ciencias Biologicas y Ambientales, Universidad de Bogota Jorge
Tadeo Lozano, Bogota, Colombia
bDepartment of Biology and Marine Biology and Center for Marine Science,

University of North Carolina Wilmington, NC, USA

cDepartment of Mathematics and Statistics, University of North Carolina Wilmington,

NC, USA
dCenter for Marine Biotechnology and Biomedicine, Scripps Institution of

Oceanography, UC San Diego, La Jolla, CA, USA

ABSTRACT

Sponges of the genus Aplysina are among the most common benthic animals on
reefs of the Caribbean, and display a wide diversity of morphologies and colors.
Tissues of these sponges lack mineralized skeletal elements, but contain a dense
spongin skeleton and an elaborate series of tyrosine-derived brominated alkaloid
metabolites that function as chemical defenses against predatory fishes, but do not
deter some molluscs. Among the earliest marine natural products to be isolated and
identified, these metabolites remain the subject of intense interest for commercial
applications because of their activities in various bioassays. In this study, crude
organic extracts from 253 sponges from ten morphotypes among the
species Aplysina archeri,Aplysina bathyphila,Aplysina cauliformis,Aplysina
fistularis,Aplysina fulva,A. insularis, and Aplysina lacunosa were analyzed by liquid
chromatography–mass spectrometry (LC–MS) to characterize the pattern of intra-
and interspecific variabilities of the twelve major secondary metabolites present
therein. Patterns across Aplysina species ranged from the presence of mostly a
single compound, fistularin-3, in A. cauliformis, to a mixture of metabolites present
in the other species. These patterns did not support the biotransformation hypothesis
for conversion of large molecular weight molecules to smaller ones for the purpose
of enhanced defense. Discriminant analyses of the metabolite data revealed strong
taxonomic patterns that support a close relationship between A. fistularis,A.
fulva and A. insularis, while two morphotypes of A. cauliformis (lilac creeping vs.
brown erect) were very distinct. Two morphotypes of A. lacunosa, one with hard
tissue consistency, the other soft and thought to belong to a separate genus
(Suberea), had very similar chemical profiles. Of the twelve metabolites found
among samples, variation in fistularin-3, dideoxyfistularin-3 and hydroxyaerothionin
provided the most predictive influence in decreasing order. Except for one
morphotype, weak relationships were found from within-morphotype analyses of
metabolite concentrations as a function of geographic location (Florida, N Bahamas,
S Bahamas) and depth (<10 m, 10–20 m, >20 m). Our data suggest that metabolite
profiles are strongly influenced by sponge phenotype rather than by the diverse
microbiome which many Aplysina species share.

Keywords Chemical ecology; Coral reef; Biotransformation; Marine natural

products; Chemical defense; Induced defense

INTRODUCTION

Sponges now dominate the benthic community of most Caribbean coral reefs,
particularly when one considers overall biomass and biodiversity; integrates it across
the reef community to deep water, and considers the communities within the reef
framework (Diaz and Rützler, 2001; Pawlik, 2011; Villamizar et al., 2013). Sponges are important to the
overall ecology of coral reefs for many reasons: they are very efficient filter feeders,
providing an important link in benthic-pelagic coupling (Southwell et al., 2008), they appear
to be capable of absorbing dissolved organic carbon as a food source ( de Goeij et al.,
2013), and their bodies provide shelter for large numbers of invertebrates and fishes
(Westinga and Hoetjes, 1981; Henkel and Pawlik, 2005). Sponges are aggressive competitors for
space (Aerts, 1998) and are primary agents of carbonate bioerosion on coral reefs
(Zundelevich et al., 2007).

Of all invertebrates, sponges have yielded the largest number of secondary

metabolites that have been isolated and described by marine natural products
chemists (Blunt et al., 2013 and previous reviews). Many of these compounds have potent
pharmacological activities, including anti-tumor, anti-fungal, anti-viral, and anti-
bacterial effects (Laport et al., 2009; Newman and Cragg, 2014). Because these compounds are
often structurally complex or are present at high concentrations in sponge tissues,
they are likely to be metabolically expensive to make and store, and to have some
adaptive purpose. By far, the most commonly hypothesized defensive role for these
compounds is predator deterrence, but antifouling, anti-overgrowth, antimicrobial
and UV-protective functions have also been proposed (Pawlik, 2011).

Our understanding of the ecology of sponges on Caribbean reefs has changed

substantially over the past 20 years. Before the mid-1990s, the conventional view of
sponge ecology on Caribbean reefs was that there was no top-down (predator)
control of sponge populations, because only a few fish species spread their
predatory activities over a wide variety of sponge species (Randall and Hartman, 1968).

Subsequent research on the chemical defenses of sponges demonstrated the

importance of predation, with sponges falling into three categories: preferred species
that are grazed completely from the reef and restricted to refugia inside the reef
framework, palatable species that are common on the reef and tolerate grazing
because they heal, grow or reproduce fast enough to persist, and defended species
that are protected from predation by chemical defenses (Pawlik, 2011). A conceptual
model of sponge ecology on Caribbean reefs, based on an understanding of sponge
chemical defenses and resource trade-offs with growth or reproduction, was
validated in cross-Caribbean surveys of sponge communities on heavily overfished
and less-fished coral reefs (Loh and Pawlik, 2014). As part of those surveys, it was
established that sponges of the genus Aplysina are among the most common
chemically defended species, with Aplysina cauliformis the single most abundant
species on Caribbean coral reefs (Loh and Pawlik, 2014).

Caribbean sponges of the genus Aplysina are strongly chemically defended against
sponge-eating fishes (Pawlik et al., 1995; Loh and Pawlik, 2014). Interestingly, these defenses do
not extend to certain nocturnal and cryptic molluscan predators, such as cowries,
which leave behind grazing trails and pits on the surfaces of several species,
including Aplysina fistularis (Pawlik and Deignan, 2015). Ecologically relevant bioassays
have also been used to demonstrate that the secondary metabolites of sponges of
this genus have potent antimicrobial (Kelly et al., 2005) and allelopathic effects (Pawlik et al.,
2007). The metabolites responsible for these biological activities were among the
earliest marine natural products described, and they came from Aplysina species
collected in the Mediterranean (Fattorusso et al., 1972) and the eastern Pacific (Andersen and
Faulkner, 1973) in the quest for new drugs from the sea. Since then, over 100
halogenated tyrosine alkaloids have been described from Aplysina species
worldwide (Lira et al., 2011), with considerable speculation about their importance in
ecological interactions and use in human applications (Niemann et al., 2015).

Comparative studies of the inter- and intraspecific variations in secondary

metabolites of sponges are uncommon, yet they can provide important information
about taxonomic relationships (Erpenbeck and van Soest, 2007) and may yield important clues
about the importance of microbial symbionts in metabolite production (Ozturk et al., 2013).
In this study, we compared the brominated alkaloid profiles of organic extracts of
253 individual sponges of the genus Aplysina (six species and ten morphotypes), a
taxon that has been difficult to characterize on the basis of morphology because of
the absence of a mineralized skeleton (Zea et al., 2014). Our comparisons were designed
to investigate the pattern of compound variation within and between species,
between depth zones, and between geographic regions.

MATERIALS AND METHODS

Table 1 provides the relevant information regarding the sites of sponge collection.
Samples were collected in the Bahamas Islands during the following months:
September 1998, July-August 1999, July-August 2000 and March 2001. Samples
were collected in the Florida Keys, USA, in May 2000. Categorization of sponge
morphotypes, and subsequent discriminant analysis, was based on the most up-to-
date taxonomic determination found in Zea et al. (2014), which includes photographs of
the variation in morphologies seen for morphotypes based on location.

Table 1 Sites from which sponges were collected for this study. Sites were
subdivided into three regions, Northern Bahamas, Southern Bahamas and Florida
Keys.
Site Region GPS coordinates
Acklins Island S Bahamas 22.186474, −74.300231
Andros Island N Bahamas 24.618540, −77.683952
Bimini Island N Bahamas 25.667746, −79.319229
Black Rock S Bahamas 22.662327, −74.023134
Cat Island S Bahamas 24.123313, −75.513862
Cay Lobos S Bahamas 22.381946, −77.591482
Cay Santo Domingo S Bahamas 21.719539, −75.762152
Chub Cay N Bahamas 25.393290, −77.880825
Conch Reef Florida Keys 24.946651, −80.456086
Egg Island N Bahamas 25.498649, −76.898396
Eleuthera Point S Bahamas 24.606181, −76.154849
Little San Salvador-Pinnacles S Bahamas 24.588934, −75.974433
Long Cay S Bahamas 22.562957, −74.393792
North Key Largo Dry Rocks Florida Keys 25.130833, −80.292017
North North Key Largo Dry Rocks Florida Keys 25.136778, −80.288621
Pickles Reef Florida Keys 24.984737, −80.413676
Porpoise Rocks N Bahamas 25.140337, −77.135227
San Salvador S Bahamas 24.062287, −74.545403
Stirrup Cay N Bahamas 25.822565, −77.932398
Sweetings Cay N Bahamas 26.556550, −77.880492
Three Sisters Reef Florida Keys 25.021933, −80.396383

Whenever possible, whole sponges were gently removed from the bottom using
latex gloves. Larger or thicker sponge individuals were sliced at their base with a
sharp knife. Sponge samples were individually placed in sealable plastic bags
underwater, keeping track of the site, depth and location of collection. Fresh samples
were placed directly into a −20 °C freezer and kept frozen until extraction.
For extraction, frozen sponges were cut in 0.5 inch cubes and placed in a graduated
cylinder with the extraction solvent mixture to determine the sponge volume. In the
case of large specimens, either half or a longitudinal slice of the sponge tube was
cut and extracted. Rope-shaped species were sliced. Sponges were exhaustively
extracted with a 1:1 DCM:MeCN mixture. This solvent mixture was selected based
on its extraction efficiency and the fact that acetonitrile does not favor the formation
of artifacts as methanol does. The volume of solvent was always proportional to the
volume of sponge extracted, roughly 10 ml of solvent per 1 ml of sponge. Sponges
were allowed to extract for several hours in the freezer. Each batch of solvent was
decanted off and filtered through diatomaceous earth (Celite). Solvents were
removed by rotary evaporation and extracts reduced to an aqueous layer. Initially,
the water layer was additionally extracted with butanol to prevent loss of very polar
metabolites. The aqueous layer was exhaustively extracted with EtOAc (ethyl
acetate). Analysis using thin layer chromatography (TLC) and Diode Array HPLC
confirmed that the butanol layer did not contain significant concentrations of
secondary metabolites. All the major and minor metabolites were retained in the
EtOAc extract.

Individual compounds used as standards for quantification were purified by a

combination of standard chromatographic techniques. Crude extracts were
subjected to flash column chromatography using reverse phase C-18 Silica as
stationary phase and eluted with a gradient of decreasing polarity of water/MeCN to
100% MeCN with final elution with methanol (MeOH) and DCM. Most of the
brominated metabolites eluted in one or two fractions, requiring further separation
by reversed phase (C-18) HPLC using as mobile phases mixtures of water and
MeCN. Pure compounds were stored under nitrogen in the freezer. Pure compounds
were analyzed by LC-MS and 1H and 13C NMR spectroscopy in order to confirm their
structures comparing with data reported in the literature.

For each sample, the metabolite composition in the crude extract was analyzed by
using Diode Array LC-MS (HP1090 and HP1100). In preparation for LC-MS, crude
extracts were passed through a reverse phase C-18 cartridge and eluted with 1:1
MeCN/water in order to remove fats, sterols and pigments in samples. The remaining
compounds were resuspended in MS grade MeOH for analysis. Separation was
achieved using a Dynamax reverse phase C-18 analytical column (60 Å, 4.6 mm ID
× 250 mm L) with a water-MeCN (10-100%) gradient acidified with 0.1% acetic acid
at a rate of 0.7 ml/min for 28 min. Compounds were detected at 254 nm and identified
by comparing their retention times and UV and MS spectra with those of authentic
standards.

The bromine in these compounds gives a very distinctive fragmentation pattern, due
to differences in the isotopes giving specific masses of fragment ions and
characteristic mass differences between the molecular ion and fragment ions. When
bromine is present in the molecule, there are n-1 fragment ions depending on how
many bromine atoms are in the molecule. Fragment ions are separated by 2 mass
units and the major one (>98%) corresponds to the molecular ion ( Pretsch et al., 1989).
Quantification of individual compounds in the crude extracts was accomplished by
the external standard method, with standard calibration curves for each compound.
Once the total amount of a compound in the sample was determined, the total
amount (mass) of each compound in the crude extract was estimated and divided
by the total volume of sponge extracted to obtain the amount of compound in
milligrams per milliliter of sponge tissue extracted (mg/ml).

Differences among Aplysina morphotypes, locations, and depths were determined

using linear discriminant functions. The sets of potential discriminating variables
included concentrations of individual metabolites and the proportion of a given
metabolite relative to the total metabolite concentration. Dimension reduction
techniques on predictor sets were performed via principal components reductions
on each set which employed rotations of the Crawford-Ferguson type (Crawford and
Ferguson, 1970). Each set of potential discriminating variables was introduced in a
forward selection manner and error rates were compared to determine the quality of
the classifier. Error rates were assessed based on the leave-one-out, or jackknife,
cross-validation technique (Seber, 2004). All analyses were conducted with SAS/STAT and
SAS/GRAPH software (SAS, 2002). Codes used for the morphotypes of Aplysina were as
follows: ARC = Aplysina archeri, BAT = Aplysina bathyphila, CAK = A. cauliformis -
brown erect, CAN = A. cauliformis - lilac creeping, FIA = A. fistularis - aggregata, FIN
= A. fistularis - typical tube form, FUL = Aplysina fulva, INS = A. insularis, LAH
= Aplysina lacunosa - hard tissue, LAS = A. lacunosa - soft tissue.

RESULTS AND DISCUSSION

Patterns of metabolite variability among ten morphotypes of Aplysina spp.

Analyses, using LC-MS of the crude organic extracts from tissue samples of 253
sponges from ten morphotypes among the species A. archeri,A. bathyphila,A.
cauliformis,A. fistularis,A. fulva, A. insularis, and A. lacunosa, provided some
interesting general patterns of the distribution of the twelve most abundant
metabolites (Fig. 1). Total concentration of brominated alkaloids was distinctly
highest in A. archeri (11.5 ± 1.3 mg ml−1) and lowest in the soft morph of A.
lacunosa (1.0 ± 0.1 mg ml−1), with the remaining morphotypes ranging 2.4-4.3 mg
ml−1(Fig. 2). The relative percentage of total brominated alkaloid metabolites that
consisted of the low molecular weight (LMW) compounds (1-3, Fig. 1) was highly
variable across morphotypes, with a high of 44.6% for A. cauliformis (brown, erect)
and a low of 5.7% for A. archeri, with remaining morphotypes ranging from 8.7 to
26.6% (Fig. 2). Among the ten morphotypes, tissue extracts of some yielded no
measurable amount of some of the twelve metabolites; in particular, compounds 3-
8, and 10-12 were consistently absent from some morphotypes, while compound 9
was consistently present (Dataset available at
http://people.uncw.edu/pawlikj/2015PuyanaData.xlsx). While tissue extracts of most
morphotypes yielded mixtures of some subset of the twelve metabolites, those of A.
cauliformis (lilac, creeping) consisted primarily of fistularin-3 (9).
Fig. 1 Structures of the tyrosine-derived brominated alkaloids from Caribbean
sponges of the genus Aplysina.
Fig. 2 Total metabolite concentration (black bars) and percentage of low molecular
weight compounds (1-3; gray bars) in crude organic extracts of all samples for each
of the 10 morphotypes. ARC = A. archeri, BAT = A. bathyphila, CAK = A.
cauliformis – brown erect, CAN = A. cauliformis – lilac creeping, FIA = A. fistularis –
aggregata, FIN = A. fistularis – typical tube form, FUL = A. fulva, INS = A. insularis,
LAH = A. lacunosa – hard tissue, LAS = A. lacunosa – soft tissue.

Discriminant analyses and taxonomic boundaries

A combination of methods was employed to determine the simplest model that could
distinguish to the greatest degree among the ten morphotypes of Aplysina. Both raw
values and proportions of metabolites were considered as potential predictors, in
addition to a variety of rotations of principal components reductions of each. Each
potential set of predictors was used in construction of linear discriminant functions in
a forward selection manner to determine the smallest set of predictors that classified
morphotypes with a suitably small error rate (initial targets of 75% correct cross-
validated results within each morphotype along with 75% overall). Similar analyses
were conducted for depth and location in place of morphotype. Initial results from
efforts to classify samples within four morphotypes consisting of A. fistularis (FIN), A.
fulva (FUL), A. insularis (INS), and A. fistularis forma aggregata (FIA) resulted in
frequent misclassifications; however, nearly all such errors remained within this set
of four morphotypes. Therefore, these were combined into a single category for
subsequent analyses (FIS). A similar phenomenon occurred with the hard and soft
morphotypes of A. lacunosa (LAH and LAS), and these were combined for further
analyses as well (LAC).

Discriminant analyses of the six remaining taxonomic categories revealed that the
proportions of three compounds, fistularin-3 (9), dideoxyfistularin-3 (12) and
hydroxyaerothionin (5), provided the most predictive influence in decreasing order
(Table 2, Fig. 3). While FIS has a 71% within-class success rate, this is partly driven
by the limited sample size for BAT; if the prior information for the discriminant
functions is adjusted to reflect relative sample sizes, the within-class success rate
rises to 83% for FIS and 87% for CAK. One primary mode of separation that
manifests itself is the relative concentration of dideoxyfistularin-3 (12), because three
taxonomic groups (A. archeri,A. cauliformis (lilac, creeping) and A. lacunosa) lack
this metabolite (Fig. 4). These three taxonomic groups were easily separable on the
concentrations of fistularin-3 (9) and hydroxyaerothionin (5), with the concentration
of the former exceeding 50% in nearly all cases for A. cauliformis (lilac, creeping)
while ranging 10-40% for A. lacunosa and below 10% for A. archeri (Fig. 4).

Table 2 Cross-validation summary of linear discriminant analysis of the

concentration of the 3 most predictive metabolites, fistularin-3 (9), dideoxyfistularin-
3 (12) and hydroxyaerothionin (5) in the most cohesive morphotype categories
of Aplysina : ARC = A. archeri, BAT = A. bathyphila, CAK = A. cauliformis (brown
erect), CAN = A. cauliformis (lilac creeping), FIS = A. fistularis(includes aggregata,
fulva, insularis), LAC = A. lacunosa (includes hard and soft morph). In this analysis,
100% of samples of ARC, BAT and CAN are classified correctly, while only 71% of
FIS samples are classified correctly.
Observations/percent classified into category
Actual category ARC BAT CAK CAN FIS LAC Total
24 0 0 0 0 0 24
ARC
100.00 0.00 0.00 0.00 0.00 0.00 100.00
0 6 0 0 0 0 6
BAT
0.00 100.00 0.00 0.00 0.00 0.00 100.00
0 3 38 0 6 0 47
CAK
0.00 6.38 80.85 0.00 12.77 0.00 100.00
0 0 0 57 0 0 57
CAN
0.00 0.00 0.00 100.00 0.00 0.00 100.00
1 7 5 0 66 14 93
FIS
1.08 7.53 5.38 0.00 70.97 15.05 100.00
3 0 2 0 0 21 26
LAC
11.54 0.00 7.69 0.00 0.00 80.77 100.00
28 16 45 57 72 35 253
Total
11.07 6.32 17.79 22.53 28.46 13.83 100.00
Error rates
ARC BAT CAK CAN FIS LAC Total
Rate 0.0000 0.0000 0.1915 0.0000 0.2903 0.1923 0.1124
Fig. 3 Three-dimensional graph of proportional metabolite concentration of fistularin-
3 (9), dideoxyfistularin-3 (12) and hydroxyaerothionin (5) for the six remaining
morphotypes ARC = A. archeri, BAT = A. bathyphila, CAK = A. cauliformis – brown
erect, CAN = A. cauliformis – lilac creeping, FIS = A. fistularis (combined FIA, FIN,
FUL, INS), LAC = A. lacunosa (combined LAH, LAS). See legend of Fig. 2 for other
morphotype codes.
Fig. 4 Two-dimensional graph of proportional metabolite concentration of fistularin-
3 (9) and hydroxyaerothionin (5) for ARC, CAN and LAC with boundaries indicated
for dideoxyfistularin-3 (12) set to zero. ARC/LAC boundary, solid line; LAC/CAN
boundary, dashed line. See legend of Fig. 3 for morphotype codes.

Discriminant analyses revealed surprisingly little differentiation within morphotypes

as a function of location or depth. One notable exception was for A.
cauliformis (brown, erect), for which there were greater concentrations of fistularin-
3 (9) for samples from Florida than for N or S Bahamas combined (Fig. 5). Separating
this morphotype into Florida and Bahamas subsets provided clearer separation of
them from both A. fistularis and A. bathyphila (Fig. 5). Aside from this exception,
neither location nor depth generated distinctions that were as clear as those that
were apparent across morphtypes. For example, when comparing the three depth
ranges within morphotypes, error rates below 30% could not be achieved, even
when four metabolites were used.
Fig. 5 Two-dimensional graph of proportional metabolite concentration of fistularin-
3 (9) and hydroxyaerothionin (5) for BAT, CAK and FIS with data for CAK separated
into two sets by location; samples from Florida and samples from the Bahamas (N
and S combined). See legend of Fig. 3 for morphotype codes.

From the standpoint of chemotaxonomic differentiation (Erpenbeck and van Soest,

2007), discriminant analysis of this metabolite dataset provides some intriguing
results. The cross-classification of the five morphotypes in the A. fistularis complex
supports the general opinion among taxonomists that these morphotypes are very
closely related, to the point that two or more growth forms can be seen on the same
individual sponge (Zea et al., 2014). Contrarily, there was a clear difference in metabolite
pattern between the two morphotypes of A. cauliformis, which argues for taxonomic
distinction. An even more interesting case is that of A. lacunosa, for which taxonomic
separation of the hard and soft morphotypes has been proposed. Zea et al.

(2014) considered the soft morph to be a different genus (Suberea), yet the metabolite
profiles of these two morphotypes were not readily separable in the present study. It
should be noted, however, that total metabolite concentrations in the tissues of the
soft morph were less than half those of the hard morph (Fig. 2), which may be due
to the difference in the density (hence, water content) of the tissues of these two
morphs.
Biotransformation hypothesis

The pattern of brominated alkaloids across morphotypes of Aplysina from Caribbean

coral reefs has a bearing on the hypothesis that these metabolites are part of an
activated chemical defense system for this genus of sponges. Originally proposed
for the Mediterranean species Aplysina aerophoba on the basis of experiments done
with freeze-dried sponge samples and cell-free extracts, the biotransformation
hypothesis posits that damage to sponge tissues results in the rapid, enzyme-
catalyzed transformation of large molecular weight metabolites (e.g., 4-12) to smaller
molecules, primarily dibromocycloheadienone (1) and aeroplysinin 1 (2), which were
suggested to have greater defensive capacities against pathogens or predators
(Teeyapant and Proksch, 1993). This optimized defense concept appears to have been
modeled on established activated defense systems known for cyanogenic and
cruciferous terrestrial plants, and several subsequent studies by the same research
group have supported the hypothesis (Niemann et al., 2015). However, the hypothesis was
rigorously tested in the field (Bahamas) on living specimens of A. archeri and A.
insularis, and no evidence of conversion of large to small molecules was observed
(Puyana et al., 2003).

The present study similarly does not support the biotransformation hypothesis for
Caribbean species of Aplysina. Tissue samples across all ten morphotypes were
prepared using the same methods, yet the percentage of low molecular weight
molecules present in extracts of these samples varied significantly, but was highly
consistent by morphotype (Fig. 2). If the proposed activated defense system was
common to species across the genus, one would expect very low levels of low
molecular weight metabolites in rapidly prepared extracts of sponge tissue
regardless of morphotype, but this was not the case, with nearly 50% of metabolites
in extracts of A. cauliformis(brown, erect) present in the low molecular weight
category (Fig. 2). It seems unlikely that the complicated defense mechanism
proposed for A. aerophoba (Niemann et al., 2015) would have evolved independently for
that species and not for several others within the larger genus.
Microbial contribution to secondary metabolite production

Like all verongid sponges, species in the genus Aplysina have tissues that are
perfused with symbiotic microorganisms, placing them in the high microbial
abundance (HMA) sponge category (Gloeckner et al., 2014). Whether or not the sponge
microbiome is responsible for the production or alteration of secondary metabolites
found in the sponge tissue has been an issue of great interest to natural products
chemists and sponge biologists alike (Unson et al., 1994). Metabolite synthesis and
alteration has been a particular subject of study for the Mediterranean species A.
aerophoba, with X-ray microanalysis evidence suggesting that the sponge is mostly
or entirely responsible for metabolite production (Turon et al., 2000). Subsequent studies
of A. aerophoba have linked individual brominated alkaloids with specific microbial
phylotypes (Sacristán-Soriano et al., 2011), but the issue remains unresolved for this species
(Sacristán-Soriano et al., 2015).

The cross-species survey of metabolites from the tissues of

Caribbean Aplysina species provided in this study tends to support sponge-derived,
rather than microbiome-derived brominated alkaloids. If the microbiome was
substantially involved with either the production or alternation of brominated
alkaloids, the predicted outcome of this survey would have been considerably more
metabolite variation within morphotypes, particularly as a function of location and
depth, yet this level of variation was not observed. Depth is a particularly important
factor, because Aplysina species are known to contain cyanobacteria that are
variably important in providing sponge nutrition (Erwin and Thacker, 2008), and it would be
expected that the influence of photosymbionts would change with light intensity as a
function of depth, but this was not the case. Instead, metabolite profiles were highly
correlated to morphotype, which suggests that the sponge cells are primarily
responsible for the production of the chemical defenses present in sponge tissue.
Nevertheless, it remains to be determined whether some metabolite changes are
subject to the activities of highly conserved components of the microbiome (Sacristán-
Soriano et al., 2015).
Amphibian Chemical Defense: Antifungal Metabolites of the Microsymbiont
Janthinobacterium lividum on the Salamander Plethodon cinereus

Robert M. Brucker & Reid N. Harris & Christian R. Schwantes & Thomas N. Gallaher
& Devon C. Flaherty & Brianna A. Lam & Kevin P. C. Minbiole

Received: 17 July 2008 /Revised: 24 August 2008 /Accepted: 25 September 2008 /

Published online: 24 October 2008 # Springer Science + Business Media, LLC 2008

Abstract

Disease has spurred declines in global amphibian populations. In particular, the

fungal pathogen Batrachochytrium dendrobatidis has decimated amphibian diversity
in some areas unaffected by habitat loss. However, there is little evidence to explain
how some amphibian species persist despite infection or even clear the pathogen
beyond detection. One hypothesis is that certain bacterial symbionts on the skin of
amphibians inhibit the growth of the pathogen. An antifungal strain of
Janthinobacterium lividum, isolated from the skin of the red-backed salamander
Plethodon cinereus, produces antifungal metabolites at concentrations lethal to B.
dendrobatidis. Antifungal metabolites were identified by using reversed phase high
performance liquid chromatography, high resolution mass spectrometry, nuclear
magnetic resonance, and UV-Vis spectroscopy and tested for efficacy of inhibiting
the pathogen. Two metabolites, indole3-carboxaldehyde and violacein, inhibited the
pathogen’s growth at relatively low concentrations (68.9 and 1.82 μM, respectively).
Analysis of fresh salamander skin confirmed the presence of J. lividum and its
metabolites on the skin of host salamanders in concentrations high enough to hinder
or kill the pathogen (51 and 207 μM, respectively). These results support the
hypothesis that cutaneous, mutualistic bacteria play a role in amphibian resistance
to fungal disease. Exploitation of this biological process may provide longterm
resistance to B. dendrobatidis for vulnerable amphibians and serve as a model for
managing future emerging diseases in wildlife populations.
Keywords Janthinobacterium lividum . Plethodon cinereus. Batrachochytrium
dendrobatidis. Beneficial bacteria . Violacein . Indole-3-carboxaldehyde

Introduction

Nearly one third of amphibian species are threatened with extinction, due in part to
the spread of the emerging infectious disease chytridiomycosis (Berger et al. 1998;
Stuart et al. 2004; Rachowicz et al. 2006). This disease is linked to the decline of
many frog populations, especially in tropical areas not affected by habitat loss (Lips
1999; Lips et al. 2006). Chytridiomycosis is caused by the chytrid fungus
Batrachochytrium dendrobatidis and can lead to tissue erosion, hyperkeratosis,
moderate hyperplasia, weight loss, and death (Berger et al. 1998). Some species,
such as the bullfrog Rana catsbeiana and the Eastern red-backed salamander
Plethodon cinereus, persist despite infection (Daszak et al. 2004; unpublished data).
Interestingly, survival in the presence of the pathogen can vary within a species;
such is the case for Rana muscosa, the mountain yellowlegged frog (Briggs et al.
2005; Woodhams et al. 2007b). An explanation of the differences in susceptibility
remains elusive but may involve differences in innate immune responses
(Woodhams et al. 2007a). An additional hypothesis is that a mutualism exists
between the amphibian host and cutaneous antifungal bacteria symbionts and is
effective on some individuals and species, but not in others (Fig. 1; Harris et al. 2006;
Lauer et al. 2007, 2008; Woodhams et al. 2007b). We suspected that the inhibitory
effects of the antifungal symbionts are due to secondary bacterial metabolites as
opposed to resource competition. Previously, we tested the hypothesis of inhibitory
metabolite production with a strain of Lysobacter gummosus isolated from the skin
of the salamander P. cinereus. The bacterium inhibited the fungus in challenge
assays and produced the antifungal metabolite 2,4-diacetylphloroglucinol in vitro at
concentrations that could inhibit the pathogen (Brucker et al. 2008). Continuing our
investigations of antifungal metabolites, a bacterium closely related to
Janthinobacterium lividum based on DNA sequence similarity was also examined.
This bacterium is commonly found on P. cinereus as well as the four-toed
salamander, Hemidactylium scutatum, and it strongly inhibited pathogenic fungi in
vitro (Lauer et al. 2007). We hypothesized that it, too, produces antifungal
metabolites that act to reduce or prevent colonization of B. dendrobatidis on
amphibians. Furthermore, we sampled the skins of wild salamanders to determine
whether antifungal metabolites were present in inhibitory concentrations. Methods
and Materials Isolation and Culturing of J. lividum As per published method (Harris
et al. 2006; Lauer et al. 2007, 2008; Brucker et al. 2008), individuals of the
salamander species P. cinereus were rinsed twice to remove transient bacteria and
swabbed with sterile swabs. Pure bacterial cultures were obtained, genomic DNA
was extracted, and a portion of the 16S rRNA gene was amplified by using 357F and
907R primers (Lauer et al. 2007). DNA obtained from the pure culture of J. lividum
was sequenced by Agencourt (Beverly, MA, USA). A consensus sequence of
approximately 1,400 bp was obtained by aligning the forward and reverse
sequences’ amplicons. The sequence was compared with the reference organisms
by BLAST search (http://www.ncbi.nlm.nih.gov/ blast) using the GenBank database
in order to confirm the identification of J. lividum (99% match). Co-cultures of J.
lividum and B. dendrobatidis (JEL 215 strain) were grown on 1% tryptone agar plates
and incubated for 48 h. The plates were checked for the presence of a zone of fungal
inhibition around the bacterial colony to reconfirm the bacterium’s antifungal
properties (Lauer et al. 2007). Additionally, 100 ml co-cultures were made in liquid
1% tryptone medium to obtain adequate amounts of metabolites for chemical
analysis (Brucker et al. 2008). Concurrently, cocultures were compared to
monocultures of the organisms to discriminate metabolites that are not produced by
J. lividum (unpublished data). To inoculate the cultures, J. lividum was transferred
from agar plates into 100 ml of 1% tryptone and
1 ml of a solution containing zoospores (the motile infective stage of B.
dendrobatidis; 1–2×106 zoospores/ml). All cultures were incubated for 72 h in a Lab-
Line incubator shaker at room temperature (∼23°C). After incubation, cultures were
centrifuged at 5,000 rpm for 12 min. The supernatant was kept frozen until organic
extraction; the pellet was discarded. Organic Extraction and Antifungal Metabolite
Isolation To extract organic metabolites from the cultures, the supernatant was
thawed and extracted ×4 with a 1/3 volume of ethyl acetate (EtOAc). The combined
organic layers were dried over Na2SO4, filtered, and evaporated in vacuo. The crude
samples (3–15 mg) were kept frozen at −20°C, under nitrogen, until further analysis.
A total of eight samples were extracted. Each crude sample was dissolved in high
performance liquid chromatography (HPLC)-grade methanol (2–3 ml). RP-HPLC
analysis (Agilent Technologies, 1200 series, Wilmington, DE, USA) was then used
to determine the retention times of the antifungal standards as well as to analyze the
components of bacterial samples. The HPLC diode array detector was programmed
to record absorbance at 220, 270, and 310 nm. Samples were injected (50 μl) into
the HPLC equipped with a C18 reverse phase column (5 μm; 4.6×150 mm; Agilent
Technologies, Wilmington, DE, USA) and eluted at 1 ml/min. The initial eluent, 10%
acetonitrile/water (v/v, both acidified with 0.1% acetic acid), ran for 2 min. This was
followed by a linear gradient to 100% acetonitrile (acidified with 0.1% acetic acid),
over an 18-min period. This final solvent was eluted for another 3 min. Fractions
were collected, and those that demonstrated antifungal activity were purified further
and identified. Nuclear Magnetic Resonance and High Resolution Mass
Spectrometry Analysis of Antifungal Metabolites 1 H nuclear magnetic resonance
(NMR) spectra were performed in CDCl3 (indole-3-carboxaldehyde) and DMSO-d6
with an aliquot of CDCl3 (violacein) on a Bruker Advance 600 MHz NMR
spectrophotometer. Samples were sent to the Mass Spectrometry Facility at Harvard
University (Cambridge, MA, USA) for high resolution mass spectrometry (HR-MS)
analysis (Agilent 6210 Time-of-flight LC/MS, ESI source, positive mode).
Confirmation of Antifungal Metabolites and Their Inhibitory Activity To determine the
minimum inhibitory concentration (MIC) of the compounds required to inhibit the
growth of B. dendrobatidis, an in vitro assay was performed according to the
procedures previously detailed (Brucker et al. 2008). In brief, concentration of
metabolites was varied in 96-well microtiter plates, and fungal growth was
determined by measuring optical density. The positive control wells contained the
fungus with the solvent DMSO; the negative control wells contained heat-killed
zoospores. An exception to our previous published procedure was the reduction of
the solvent DMSO to 1% by volume. Indole-3- carboxaldehyde treatments and
controls were replicated 16 times, while the violacein assay treatments and controls
were replicated eight times. Statistical Analysis A Dunnett’s t test was used to test
the null hypothesis of “no inhibition” by comparing the amount of fungal growth in
each metabolite concentration treatment to the positive control treatment that had
fungi without metabolites. Pair-wise comparisons were made with Tukey’s HSD
procedure, which holds experiment-wide error at a maximum of 0.01. Extraction of
Metabolites from Skin Samples Seven wildcaught P. cinereus were collected from
the James Madison University Arboretum (Harrisonburg, VA, USA) during the
months of November and December of 2007. Each individual was handled
separately with sterilized gloves and housed overnight, in separate sterile containers
at 14° C. Salamanders were euthanized with gaseous CO2. A 2.3–4.4-cm2 portion
of skin was excised from the shoulders to the hips, and the surface area of the skin
was determined by using the program ImageTool 3.0 (distributed by the University
of Texas Health Science Center at San Antonio, Texas. May 2002). The skin
components were then extracted with HPLC-grade methanol (4×5 ml). The organic
solvent was evaporated, and the resulting sample was then constituted with 200 μl
of HPLC-grade methanol before being injected onto the HPLC, as per above.
Approximate concentrations were determined by computing a standard curve of the
metabolites. Animals were collected by permit from the Virginia Division of Game
and Inland Fisheries. Our animal care protocol was approved by JMU’s Animal Care
and Use Committee. Sampling for Naturally Occurring J. lividum Sampling and
analysis of cutaneous bacteria follow methods published in detail by us elsewhere
(Lauer et al. 2007). Briefly, transient bacteria were removed from salamander skin
by rinsing them twice in sterile water and then swabbing their ventral and lateral
sides with a wet sterile cotton swab (Harris et al. 2006; Lauer et al. 2007, 2008). The
presence of J. lividum was tested by using polymerase chain reaction or denaturing
gradient gel electrophoresis (DGGE; see published methods (Lauer et al. 2007)).
Typically one bacterial species is indicated by one band (sequence type) on DGGE
gels, although heterogeneity in ribosomal RNA copy number in some species can
cause a species to appear as two or more bands. In one lane, we placed the
amplification products from a pure culture of J. lividum. Other lanes of the DGGE
contained a culture-independent assay of bacterial diversity on salamander skins. If
one band in these lanes migrated to the same position as the band from the pure
culture of J. lividum, then we concluded that J. lividum was present on the skin.
Evaluating Mucus Depth of P. cinereus The mucous depth of P. cinerus was
calculated by using three salamanders collected from local populations in order to
determine metabolite concentration on salamander skins. The salamanders were
sedated and euthanized using carbon dioxide. Skin samples were removed using a
double-headed scalpel, which was assembled by using two size 10 scalpel blades
taped parallel to each other. The resulting double-headed scalpel had a parallel 1–
1.5 mm span between the blades and was used to cut and remove three 1–1.5-mm
thick cross-sections, per salamander, at three locations: on the tail, above the hind
legs, and below the fore limbs. Scalpel blades were replaced between each cut.
Each crosssectional sample was rotated 90° and immediately laid transversely onto
a glass microscope slide. To avoid distortion of the sample from the weight of the
cover slip, two platforms were cemented on either side of the sample comprised of
7–10 cover slips, with a final cover slip spanning the sample field. Drops of a 1%
saline solution were added to the sample to prevent dehydration of the sample. Two
to three drops of methylene blue were added to the saline solution to provide contrast
for the background and translucent mucus layer. The slide was placed on an inverse
microscope within half an hour of construction. Five out of the nine slides had
undamaged, viable, cross-sectional samples. From these five samples, the depth of
the mucus layer was measured at five random locations per sample, using
microscopy measurement software (Nikon Imaging Systems Elements, Tokyo,
Japan). Results Compounds in the culture medium produced by J. lividum that were
fractioned around 9.07 min and 11.15 min via HPLC (“Electronic Supplementary
Material” SI Fig. 1) inhibited fungal growth. Further HPLC purification of these
bioactive fractions led to the isolation of two antifungal metabolites whose structures
were identified as violacein and indole-3-carboxaldehyde (Fig. 2) via NMR and HR-
MS analysis (“Electronic Supplementary Material” SI Table 1). Optical density was
used to assess fungal growth, and inhibitory growth assays against B. dendrobatidis
were performed to determine the MIC for each of the metabolites. Increasing
concentrations of violacein and indole-3-carboxaldehyde reduced fungal growth
(violacein: F=73.32, df=1).
Discussion

A symbiotic relationship between cutaneous bacteria and a host amphibian is

plausible; metazoans evolved with bacteria and mutualistic associations are
common in animals (McFall-Ngai 1999). A fascinating example of cutaneous
bacteria that benefit their host can be found in leaf cutter ants, Acromyrmex (Currie
et al. 2006). The farming ants house bacteria, Pseudonocardia, which feed in the
hosts’ cuticle, and in return produce metabolites that deter an unwanted fungal
species, Escoviopsis, from invading their subterranean garden (Currie et al. 2006).
Our studies indicate that a similar mutualism might exist in amphibians, although
whether associations will be found between specific amphibians species and specific
bacterial species remains an open question (Harris et al. 2006; Lauer et al. 2007,
2008; Woodhams et al. 2007b; Brucker et al. 2008; Fig. 1). The bacterium J. lividum
is likely to be a mutualistic partner with red-backed salamanders. The two
metabolites produced by J. lividum and found on salamander skin, indole-3-
carboxaldehyde and violacein (Fig. 2; “Electronic Supplementary Material”

SI Table 1), are known agents of chemical defense although indole-3-

carboxaldehyde is not a reported product of J. lividum. The fungus Epichloë festucae
employs indole-3-carboxaldehyde in its endophytic relationship with many different
host plants, defending the host against fungal pathogens (Yue et al. 2000). This
metabolite also has antibacterial characteristics against several human pathogens
in vitro (Gutierrez-Lugo et al. 2005). Violacein biosynthesis is a cell density-
dependent factor, controlled via quorum sensing for the bacterial species
Chromobacterium violaceum. Violacein is a known metabolite of J. lividum as well
as several marine surfaceassociated bacteria (Durán and Menck 2001; Matz et al.
2008). Violacein has a broad bioactivity profile including antibacterial, antiviral, and
antitumoral activity (Durán and Menck 2001). Neither compound, however, has been
reported to protect vertebrates in the wild from an emerging infectious disease. The
concentrations of indole-3-carboxaldehyde and violacein on the skin of the
salamanders (Table 1) approached or exceeded the MICs determined in vitro (Fig.
3). The concentration of indole-3-carboxaldehyde on the skin of one individual was
just below the MIC required to completely inhibit B. dendrobatidis growth (51 vs. 68.9
μM). The violacein MIC was exceeded in all three salamanders that had violacein,
with one individual having a concentration 114 times greater than the in vitro MIC.
Comparatively, 2,4-diacetylphloroglucinol was not detected on the skin of the wild-
caught salamanders, although there is a significant difference in concentrations
required to inhibit B. dendrobatidis in vitro (MIC=136 μM; Brucker et al. 2008).
Concurrently, the bacterial diversity of the salamanders was examined with DGGE,
specifically screening for J. lividum (“Electronic Supplementary Material” SI Fig. 1).
Three of the four individuals that had indole-3-carboxaldehyde or violacein tested
positive for J. lividum. This result does not exclude the possibility of low densities of
J. lividum on the fourth individual, nor does it exclude the presence of alternative
symbionts playing a similar role. J. lividum was not detected on the individuals
lacking these metabolites. Together, these data suggest that the metabolites are
related directly to the bacterial symbiont and further support the hypothesis that
symbiotic bacteria play a significant role in host resistance. Host salamanders have
variations in the diversity of bacteria, both antifungal and non-antifungal, on their skin
(Lauer et al. 2007, 2008), and this variation in bacteria populations may determine
the type and amount of antifungal compounds available. We have evidence from
laboratory experiments that amphibians with adequate microbially produced
antifungal metabolites on their skin persist with B. dendrobatidis infection while those
with low concentrations die (unpublished data). It is an open question why some
amphibians do not have adequate antifungal metabolites on their skin, but perhaps
there are costs as well as benefits to maintaining skin bacteria that produce such
metabolites. The presence or absence of beneficial bacteria and the antifungal
metabolites they produce could account for the variation in infection and decline
among populations and amphibian species. Some species house beneficial bacteria
on their skin due to their physiology, behavior, environment, or any combination of
the three. Physiologically, it is likely that there are strong selective pressures, such
as antimicrobial peptides (Simmaco et al. 1998; Rollins-Smith et al. 2002) and
glycoproteins (Ducklow and Mitchell 1979; Brizzi et al. 2002) on the skins of
amphibians that influence the type of bacteria that are found there. Behaviorally,
some amphibian species congregate densely, perhaps aiding in the spread of
beneficial and pathogenic microbes; similarly, some care for their eggs and may,
therefore, impart microbes to their embryos and young (Harris et al. 2006).
Environmentally, factors such as plants, soils, and water can determine the type of
bacteria available to the amphibian. In addition, anthropogenic effects may alter
microbial species compositions (Belden and Harris 2007). These factors likely
contribute to the differences observed in amphibian vulnerability to chytridiomycosis.
Our research suggests two modalities for amphibians affected by chytridiomycosis:
(a) application of antifungal metabolites onto infected amphibians in laboratory
environments; (b) application of probiotic bacteria that already exist on amphibians
and that produce antifungal metabolites in natural environments. The use of
chemical treatments for chytridiomycosis is precedented in amphibian husbandry
(Young et al. 2007) although they cannot provide long-term protection from the
disease. However, the application of J. lividum itself as a bio-augmentation or
probiotic treatment is potentially self propagating and may provide long-term
protection from the disease as seen in agriculture (Haas and Degafo 2005). Should
the anti B. dendrobatidis properties of the bacteria be extended to susceptible
species, then it is possible that species declines due to the pathogen could be
avoided. A study focusing on the beneficial antifungal bacteria of R. muscosa
populations that are persisting with B. dendrobatidis has suggested that the entire
population does not need to have beneficial bacteria to allow persistence
(Woodhams et al. 2007b). By inoculating probiotic bacteria onto the amphibian host,
B. dendrobatidis resistance may be selfperpetuating in the species and perhaps
other amphibians through communal interactions, for example in hibernacula. This
method could prove superior to the continual application of antifungal chemicals,
which could have a negative effect on nontarget species such as beneficial fungi, is
not self-perpetuating, and is economically difficult to sustain. Further studies of the
metabolites being produced by other antifungal bacteria from amphibians should
yield a diverse suite of antifungal metabolites, perhaps including novel antifungal
compounds that could be used in amphibian husbandry as well for treatment of
human pathogens. Additional research is needed to understand the microbial
ecology of amphibian skin and to use that understanding to develop strategies to
manage chytridiomycosis in nature.
Acknowledgements

This work was supported by a Research Corporation Cottrell College Science Award
(KPCM), the Thomas F. Jeffress and Kate Miller Jeffress Memorial Trust (KPCM), a
National Science Foundation Research in Undergraduate Institutions grant 0640373
(RNH), and James Madison University. HR-MS analyses were conducted by the
Mass Spectrometry Facility at Harvard University (Cambridge, MA, USA). The
authors would like to thank Dr. Nancy Staub for mucus depth discussions and Dr.
Grace Wyngaard for a critical reading of the manuscript.