1.

Challenges in proteomic separation

In contrast to biochemical protein isolation and purification, which address the

needs of studying individual or group proteins, proteomics is the large-scale or
systematic study of proteins, and their structures, topologies, functions and
interrelationships. Proteomic studies require technologies that can dissect a
complex mixture of proteins for easier and more accurate analysis of its
components and their relationships. Proteomic separation consists of
technologies and methods for capture, partitioning, fractionation, depletion or
subtraction as well as enrichment of the components in each proteome. The
strategies and approaches are different from conventional methods normally
applicable for isolation, detection, and quantification of target proteins in a
complex protein mixture.
1.1.Unmet needs in proteomic separation Conventional technologies or
methods in protein isolation, separation, and purification have been well
developed in the past 30 years in life science as biochemical tools that are
useful in processing specific target proteins. In the early day of proteomic
studies, conventional methods were directly adapted as pre-fractionation
tools for meeting needs of large quantity of protein or peptide analysis by
mass spectrometry (MS) [1]. One or two-dimensional electrophoresis (1DE
or 2DE), isoelectric focusing (IEF), ion-exchange chromatography (IEC), fast
protein liquid chromatography (FPLC), high performance liquid
chromatography (HPLC), and capillary electrophoresis (CE) were relatively
mature and readily applicable [2]. For MS analyses to reach its resolution
potential and detection sensitivity, more powerful protein separation
processes were needed. The challenges specified that separation should be
robust, specific, reproducible, and amenable to processing complex protein
mixtures. Methods using a combination of 23 processes on-line or off-line
(multidimensional) and selective separation (affinity-binding and labeling)
were developed. In additional to 2DE, other methods, such as HPLCHPLC,
HPLCCE, strong cation exchange (SCX) coupled with RPLC, size exclusion
chromatography (SEC) followed by RPLC, IEF with RPLC, etc. were reported
with certain successes [1]. However, these methods are neither able to
effectively reduce the complexity of proteomic sample nor provide the
resolution that can reach deep into the LAP of a proteomic sample.
Therefore, the conventional methods, even at a multi-dimensional level,
failed to meet the necessities of MS-based proteomics analysis. Thus,
development and application of technologies and methods in affinity
separation and enrichment have become a high priority.

1.2. Plasma proteome project demands

The Plasma Proteome Project (PPP) initiated and organized in 2002 by the
Human Proteome Organization (HUPO) was the first major international
collaborative effort to dissect and analyze a challenging protein mixture,
the human plasma [3]. The major obstacle of analyzing the human plasma
proteome is the wide protein concentration range, from highly-abundant
proteins (HAP) at mg/mL to low-abundant proteins (LAP) at pg/mL (i.e. 1010
to 1012 magnitude difference) [47]. The top 1015 HAP members such as
albumin (or human serum albumin, HSA), immunoglobulins (Ig), transferrin,
etc. represent only b 0.1% of the diversity of proteins, yet constitute more
than 95% of the mass of total plasma proteins. The biomarkers for
malignant or non-malignant diseases (e.g. osteopotin, prostate-specific
antigen, various interleukins and cytokines) are usually at ng/ml to pg/ml
levels, making them like needles buried in a huge haystack [2,8]. Scientists
using 2DE and MS require tools or reagents that can specifically separate
HAP from LAP during sample preparation to dig deeper into the proteome
and detect LAP at higher sensitivity and better resolution [9]. One of the
PPP's specific aims was to evaluate advantages and limitations of many
depletion and fractionation technologies [3]. Driven by this task, various
technologies and methodologies were rapidly developed for removal or
depletion of HAP from plasma or serum samples [9].

1.3. Separation is required for other biological samples A similar situation

and need for HAP and LAP processing also exists in handling other
proteomic samples such as other clinically important body fluids, cellular
extracts, plant extracts, etc. Because albumin, IgG, 1-antitrypsin, IgA,
transferrin, and haptoglobin collectively account for 80% of the total
cerebrospinal fluid (CSF) protein content, the HAP profile in CSF resembles
that of plasma and subsequently removal of the HAP is needed for
proteomics analysis of CSF [10,11]. The proteome of whole saliva is highly
susceptible to a variety of physiological and biochemical processes. But it
still contains HAP such as salivary mucins, IgA, alpha-amylase and
lysozyme [12,13]. HAP in urine samples include albumin, immunoglobulins,
1-acid glycoprotein and uromodulin (or TammHorsfall urinary
glycoprotein) [14]. The protein concentration range is at 6 to 8 orders of
magnitude in human cells [6]. Fractionation of cell extracts by subtraction
or
enrichment is also needed to enhance 2DE and MS analyses of cellular or
subcellular proteomics. In plants, ribulose-1, 5-bisphosphate
carboxylase/oxygenase (RuBisCO) is the main
HAP. Because of its critical role in photosynthesis, RuBisCO comprises 30
60% of the total protein content in green leaf tissue, which can interfere
with determination of LAP in plant proteomics [15,16]. All of these
challenges again call for specific methods for separation of HAP,
enrichment of LAP, and detection of target proteomic samples for enabling
biomarker discovery and validation.

1.4. Most desired features

The tools or methods used for separation of complex protein mixture or
enrichment of specific groups of proteins ideally should possess the
following features:
1) High specificity and accuracy
2) Low cross-reactivity to non-target proteins (less nonspecific binding)
3) High affinity and avidity (multivalent binding power)
4) Recovery of bound proteins for analysis (good for fractionation or
partitioning)
5) High capacity (to process large volume per load)
6) Cross-species application (applicable for samples from different species)
7) Convenient use (simple and easy operation)
8) Capable of automation or streamlining (on-line with other devices)
9) High reproducibility (with acceptable coefficient of variation, CV)
10) Good reusability (fidelity after regeneration)
11) Minimal disruption to natural condition (keep sample being
representative)
12) Reasonable affordability (cost-effective)
These ideal characteristics above impose additional challenges in
technology development and validation. There are a number of approaches
to protein separation or enrichment based upon their biochemical and
biophysical features, such as molecular weight, mass, density,
hydrophobicity, surface charge, isoelectric point, tertiary structure, amino
acid sequence (epitope), etc. Conventional centrifugation, ultrafiltration,
and liquid chromatography, including dye ligands (Cibacron Blue), have
been previously used to remove albumin and other plasma proteins [17
24]. However, these processes are neither sufficiently specific nor effective
for reducing the overwhelming dynamic concentration range. Therefore,
new strategies and technologies must be developed and then applied.

2. Approaches to proteomic separation or enrichment

Specificity is a critical element in proteomic analysis. The potential for non-
specific binding in separation or enrichment of target proteins has been a
major concern for all methods or approaches designed to remove HAP or
enrich LAP. To avoid non-specific binding and ensure accurate capture of
target proteins, affinity reagents are developed for this application due
to their precise recognition of target proteins. Affinity separation and
enrichment are processes specific to capture proteins. Antibodies proteins,
peptides, and nucleotides are the biomolecules that have been used for
this purpose.

2.1. Immunoaffinity-mediated proteomic separation

Among the technologies of affinity separation, immunoaffinity separation of
proteins using Immunoglobulin G (IgG) or Immunoglobulin Yolk (IgY) has
generated encouraging data [2528]. It is increasingly being accepted as
the most effective sample preparation process in plasma proteomics
studies. This technology has emerged as MS (and 2DE) have been broadly
applied to protein biomarker discovery. The purpose of this process is to
specifically remove top HAP from plasma or serum samples to achieve
broader proteome coverage. The process was given the name
immunoaffinity-based protein subtraction chromatography (IASC),
emphasizing subtraction or fractionation of HAP, not necessary depletion
[27,28]. Agilent Technologies, Bio-Rad Laboratories, EMD Biosciences,
GenWay Biotech, Millipore, Pierce Biotechnology, Sigma-
Aldrich, have developed and marketed commercial kits for separation or
subtraction of HAP. By applying immunoaffinity capture mechanism, there
have also been other novel approaches developed later to address the
needs for removing HAP in other body fluids or other samples for proteomic
studies. The technology and products were also developed for digging
deeper into the proteome and enriching LAP for biomarker discovery.

2.1.1. Two types of antibodies in immunoaffinity application

Polyclonal IgG and IgY antibodies are the two major ingredients for
immunoaffinity separation. Immobilized IgG and/or IgY have been
successfully applied to product development. These two types of
immunological macromolecules share many characteristics but also have
definitive differences and associated advantages and disadvantages (Table
1).
The differences between IgY and IgG antibodies are readily apparent. Their
features, strengths and weaknesses are directly translated and manifested
in the immunoaffinity products derived from them. It is important to choose
the appropriate system and devices for proteomic sample preparation.
Many factors need to be considered, such as the product forms, features
and specifications, experimental goals, sample type, size and number, as
well as methods and compatibilities of downstream processes. The
following are more detailed analysis of the product systems based upon IgY
or IgG or combination.

2.1.2. IgY affinity columns

Seppro, consisting of IgY microbeads, is the immunoaffinity separation
system developed by GenWay Biotech, Inc. (San Diego, California US).
Seppro system uses technology that covalently conjugates antigen
affinity-purified polyclonal IgY antibodies in an oriented fashion to
microsphere carriers in the absence of Protein A or Protein G [9,28]. The
Seppro product line is specifically designed for targeted removal of HAP
from serum or plasma [53]. The products can also be applied to samples
from CSF, urine, bronchial alveolar lavage or other complex tissue, body
fluid or cellular sources. Seppro products are available in four types of
format: bulk slurry, spin column, liquid chromatographical (LC) column and
tip. Loading capacity ranges between 15 and 500 l, depending on product
formats and column sizes (Table 2).
Seppro product line consists of a portfolio of products with specific
applications:
1) SepproIndividual IgYSpecifically for removing individual HAP targets:
Albumin, IgG, Transferrin, 1-Antitrypsin, IgA, Haptoglobin, Fibrinogen, 2-
Macroglobulin, IgM, High-Density
Lipoproteins (mainly Apo AI and Apo AII), 1-Acid Glycoprotein,
Complement C3, Low-Density Lipoproteins (mainly Apo B), and Bovine
Albumin [9,50,53,56].
2) Seppro Disposable IgY-Top3Single-use IgY microbeads for specifically
removing the three most abundant plasma or serum proteins Albumin, IgG,
and Transferrin.
3) Seppro IgY6, IgY12, and IgY14Mixtures of IgY microbeads in
optimized ratios for removing 6, 12, or 14 HAP targets in human plasma or
serum and other body fluids such as CSF [5355,5760].
4) Seppro IgY-R7, M7, IgYC12Customized for rat (R), mouse (M), cow (C)
plasma or serum samples. Although IgY antibodies are shown to have
reasonable efficiency in cross species reactivity to orthologous proteins
[50,53,60,61], rat, mouse, or cow albumin is used for developing the
species specific IgY antibodies for the R7, M7, and C12 products.
5) SepproTip-APT12 SystemAn automated system using IgY microbeads
in tip format for high-throughput (HTP) sample processing. The system
consists of SepproTip, the APT12 Magtration System (Precision System
Science USA, Inc. Livermore, California), and disposable, pre-filled buffer
cartridges. SepproTip-APT12 System allows simultaneous processing of 12
samples within 60 min in a fully automated fashion. Up to 100 plasma or
serum samples can be processed daily using the system.
6) Seppro RuBisCOIgY microbeads for specific removal of plant HAP
[16].
7) SuperMixThis product is produced using the flow through fraction of
IgY12 as group antigen and is designed to capture the moderately-
abundant proteins (MAP) in human plasma or serum after removal of HAP
(see below for more detail).
8) SuperEnrichThis is a new type of product that is designed to directly
capture and detect leakages of cellular proteins in plasma or serum
through an enrichment process. The SuperEnrich IgY microbeads were
developed using whole cell extract from cancer cells or any cell line as
mixed antigens. In contrast to the subtraction approach of Seppro
products, the strategy of SuperEnrich is to enrich LAP or potential
biomarkers in plasma or serum, assuming that detection of cellular proteins
unusually increased in blood stream provides an indication of potential
tissue damage or cell death, which can be caused by inflammation, cancer,
or other pathological or physiological processes. The IgY-microbead
products as immunoaffinity reagents for protein separation have been
demonstrated to have the following advantages: (a) high specificity to
target proteins, (b) low cross-reactivity to non-target proteins, (c) minimal
disruption of the native condition of samples, (d) simple procedure and
reproducible results, (e) high capacity, and (f) multiple-species applicability
to orthologous proteins [9,50,5361]. Seppro products has been used as
sample preparation tool in proteomic profiling and biomarker discovery
studies [54,56,58,59,61]. In a recent report, proteomics analysis of plasma
sample has revealed a new candidate biomarker that may be associated to
chronic graft-versus-host disease (cGVHD) [62].

Immunoaffinity fractionation of HAP reduces the wide dynamic range of

protein concentration and enables better detection of LAP. However,
immediately following removal of HAP, the next level of abundant proteins,
MAP, become an obstacle towards accessing LAP. Therefore, isolation of
MAP is a new challenge for effectively and accurately detecting and
analyzing LAP. GenWay Biotech has further developed Super-Mix column to
separate MAP from LAP. The IgY antibodies in SuperMix column are
generated by using plasma proteins depleted of top-12 HAP (a flow-through
fraction from an IgY12 column) as immunogens and affinity ligands. The
resulting antibodies are mostly against MAP. The SuperMix column is shown
to effectively separate MAP from LAP as designed. In a case study,
SuperMix column was applied to partitioning the flow-through fraction of
IgY12, which resulted in a bound/eluted fraction (designated MAP fraction)
and the flow-through fraction (designated LAP fraction). Unfractionated and
sequentially fractionated samples using IgY12 and Super-Mix columns were
analyzed by SDS-PAGE and 2DE. The data demonstrate that SuperMix
column removes a group of proteins that are the main abundant proteins in
the flow-through fraction of IgY12. Studies by using SDS-PAGE coupled with
LC/MS/MS [63,64] or SCX/LC-MS/MS [65] show that
tandem IgY 12 and SuperMix system enabled identification of over 500 MAP
and LAP, including proteins with normal reported concentrations ranging
from ~100 pg/mL to 100 ng/mL. The advantage of this technology is that it
offers an effective means for digging deeper into complex biofluid
proteomes with high reproducibility and without significant reduction in
analytical throughput because the separations only generate two fractions
(the flow-through and bound
fractions) for analyses, which is unlike other fractionation approaches that
typically produce many fractions for analyses.

2.1.3. IgG affinity columns

Multiple Affinity Removal System(MARS) is a system developed by Agilent
Technologies (Santa Clara, California US). In contrast to SepproIgY
products, this product line initially consists of an affinity resin using
polyclonal IgG antibodies. The first product contains a mixture of the
antibodies against six human serum HAP (albumin, IgG, transferrin, 1-
antitrypsin, haptoglobin, and IgA). The Fc regions of antibodies are
attached to polymeric beads through protein A cross-linking. This linkage
provides easy access of proteins to the affinity-binding sites. The MARS
beads are available in both spin cartridge and high performance liquid
chromatography (HPLC) column formats. It has been shown that depletion
efficiency was ~99%, and average recovery of spiked tumor markers was
78% [70]. Later additions to the
product line includes a Human-7 column able to deplete the original six
HAP plus fibrinogen [59], a Mouse-3 column for depletion of murine
albumin, immunoglobulin, and transferrin.

The most recent product in this line is Human-14 that targets original six
HAP plus fibrinogen, 2-macroglobulin, 1-acid glycoprotein, IgM,
apolipoprotein Al, apolipoprotein All, complement C3, and transthyretin,
approximately 94% of total plasma proteins [71]. The new MARS Human-14
product was described to contain not only polyclonal IgG antibodies but
also
recombinant protein ligands (Affibody Molecules, refer to Section 2.2).
The depletion efficiency is 95% to 99% of the 14 HAP. The MARS columns
were reported to be reproducibly used for more than 200 runs, the highest
recycling times among the similar products. The product information is
summarized in Table 2.

As described above, MARS product line contains the following formats:

1) MARS Human-6 Spin Cartridge and LC column
2) MARS Human-7 Spin Cartridge and LC column
3) MARS Mouse-3 Spin Cartridge and LC column
4) MARS Human-14 Spin Cartridge and LC column

MARS Human-6 was the first product of this kind entering the market.
Application of this product has led to profiling of proteome in human
plasma, CSF, and other biological fluids as
reported in many cases [52,63,6670]. In one of applications, researchers
reported results that a combination of MARS based immune depletion, 3D
LC separation and MS/MS analysis allowed 10 potential biomarkers to be
identified from the 3000 identified proteins [66]. More data on MARS
Human-14 are to be further reported.

2.1.4. Mixed antibody columns By using mixed antibodies and recombinant

affinity fragments,
Sigma-Aldrich (St. Louis, Missouri US) developed and commercialized
another type of immunoaffinity separation product. The ProteoPrep
Immunoaffinity Albumin and IgG Depletion Kit is designed to specifically
remove albumin and IgG from human serum (2550 l). The ProteoPrep
immune affinity medium in the prepacked spin columns is a mixture of two
beaded mediums containing recombinantly expressed, small single-chain
antibody ligands, resulting in low nonspecific binding and high capacity.
This kit is also effective in depleting albumin and IgG from mouse and
guinea pig serum. The further-developed ProteoPrep 20 immunodepleting
technology employs a mixture of antibodies against 20 human plasma HAP
(Table 2). The antibodies include polyclonal IgG and single-chain antibodies
conjugated to a sphere support via linkages designed to reduce non-
specific binding. It can remove up to 98% of the plasma protein mass in a
single passage of sample through the bed and N 99% in a second pass. The
average depletion efficiency for the 20 proteins is 99.3% [7274]. The
ProteoPrep 20 Spin Column has a loading capacity of 8 l of a plasma or
serum sample and the LC Column
can accommodate a plasma or serum sample of 100 l. More specifically,
in internal and carefully-controlled tests [72], the results indicated that
each spin column removed the 20 HAP with an average depletion of 99.6%,
when 108 l plasma depletions were concentrated and depleted twice.
This depletion enabled a 38-fold and a 3-fold increase, respectively, in the
load of low-abundance proteins compared to the sample without depletion
and depletion of just 6 proteins.
This enrichment consequently enabled the identification of several low-
abundance proteins that could not be detected either in the non-depleted
serum or the 6-protein depleted serum.
The ProteoPrep 20 product was reported to be reusable for at least 100
times. The ProteoPrep 20 product removes the largest number of HAP in
one step when compared to other HAP-subtraction products. More
experimental data on its application for enabling proteome profiling and
biomarker discovery have yet to further emerge.

2.2. Protein ligand-based affinity separation

This type of product is derived from natural or recombinant proteins that
have specific recognition and binding capacity of certain target proteins. In
addition to the recombinant single chain antibody fragments that
specifically bind HSA or IgG as used in ProteoPrep products [72], bacterial
proteins (Protein A or G) are ideal ligands for immunoglobulins. Each of
these bacterial proteins specifically binds to the Fc region of IgG [75,76].
Recombinant protein L from Peptostreptococcus magnus binds
immunoglobulins primarily through kappa light-chain interactions without
interfering with the antigen binding site of immunoglobulins. Recombinant
protein scaffold based upon the Fc binding structure of Protein A, known as
Affibody, is also useful for the purpose of protein capture and separation
[77]. Lectins are sugar-binding proteins which are highly specific for their
sugar moieties [78] and lectin-based affinity columns have also been
developed to specifically capture, enrich, or separate glycoproteins [79,80].

2.2.1. Protein A or G columns

Protein A is a 4060 kDa surface protein originally found in the cell wall of
the bacteria Staphylococcus aureus. Protein G is an immunoglobulin-
binding protein expressed in group C and G Streptococcal bacteria much
like Protein A, but with differing specificities. It is a 65-kDa (G148 protein G)
and a 58 kDa (C40 protein G) cell surface protein. The native molecule also
binds albumin. In addition to native Protein A or G, other immunoglobulin-
binding bacterial proteins such as recombinant Protein A/G and Protein L
are all commonly used to purify, immobilize or detect immunoglobulins.
Each of these immunoglobulin-binding proteins has a different antibody
binding profile in terms of the portion of the antibody that is recognized
and the species and type of antibodies it will bind. Protein A binds with high
affinity to human IgG1 and IgG2 as well as mouse IgG2a and IgG2b. Protein
G has strong binding to human IgG1, IgG2, IgG3 and IgG4, as well as
mouse IgG1, IgG2 and IgG3. Protein A is generally preferred for rabbit,
pig, dog and cat IgG. Protein G has better binding capacity for a broader
range of mouse and human IgG subclasses. Protein A/G is a recombinant
fusion protein that includes the IgG binding domains of both Protein A and
Protein G. Therefore, Protein A/G is ideal for binding the broadest range of
IgG subclasses from rabbit, mouse, human and other mammalian samples.
As a group, the immunoglobulins represent the second most abundant
proteins in the plasma or serum. Protein A, G or A/G -based affinity
separation has been applied to the removal of Ig proteins, together with the
dye-ligand (Cibacron Blue) for depleting HSA [8188]. Advantage of using
this type of products is relatively low cost and high capacity. However, for
most MS or 2DGE-based analysis, removal of two HAP is not enough.
2.2.2. Affibody ligands
This is a type of recombinant protein ligand developed by mimicking or
complementing the structures and binding activities of an antibody [89
91]. Examples are trinectin [92,93], anticalins [91], and affibody ligands
[94]. A class of affinity ligands denoted affibody ligands were designed by
randomization of 13 solvent-accessible surface residues of a stable alpha-
helical bacterial receptor domain Z, derived from Protein A [9498]. In
function, affibody ligands mimic monoclonal antibodies. Compared to
antibodies, the dissimilarity of affibody ligands lies in its small size. Affibody
ligands have a molecular weight of 6 kDa, compared to the molecular
weight of 150 kDa for antibodies. In spite of its small size, the binding site
of affibody ligands is similar to that of an antibody. These types of ligands
also have robust physical properties; able to withstand a broad range of
analytical conditions, including extreme pH and elevated temperature.
Their unique C-terminal cysteine residue facilitates direct conjugation to
matrices. Affibody Molecules were developed and commercialized y
Affibody AB (Stockholm, Sweden). Their products have been applied to HAP
removal [77,99]. Affibody Molecules to transferrin were used for depletion
of the protein from plasma and CSF. Specificity was demonstrated using
biosensor technology and dot blot analysis as transferrin was the only
protein recovered by affinity chromatography from human plasma.
Affibody-mediated capture of transferrin, combined with IgG- and HSA-
depletion, was reported. For human plasma, 85% of the total transferrin
content in the samples was depleted after only two cycles of transferrin
removal, and for CSF, 78% efficiency was obtained in single-step depletion
[77]. In another study, Affibody Molecules with specificity towards human
HSA, IgG, transferrin, and transthyretin were combined in an affinity
column. In addition, polyclonal antibodies against cystatin C were coupled
to chromatographic beads and packed in a separate column. Reproducible
and efficient removal of the five target proteins was observed. The
proportion of depleted proteins were estimated to be 99%, 95%, 74%, 92%
and 83% for HSA, IgG, transferrin, transthyretin and cystatin C,
respectively. SDS-PAGE and MS were used to monitor and identify proteins
in native CSF, depleted CSF samples and the captured fractions. Enhanced
identification of lower abundant components was observed in the depleted
fraction, in terms of more detected peptides per protein [99].

2.2.3. Lectin affinity columns

Lectins are 60100 kDa glycoproteins known for their ability to agglutinate
(clump) erythrocytes in vitro. Lectins are found in most types of beans,
such as soybeans [78].While the function of lectins in plants is believed to
be the binding of glycoproteins on the surface of parasitic cells, their role in
animals involves binding of soluble extracellular and intercellular
glycoproteins. Lectins were used to capture or enrich certain serum
glycoproteins [100102]. In a study, five lectins, Concanavalin A (Con A),
Wheat germ agglutinin (WGA), Jacalin, Lentil lectin (LCA), and peanut lectin
(PNA), have been selected to individually capture different glycoproteins
from human serum. It has been shown that Con A predominately
recognizes
alpha-mannose [103], which is very common in N-linked glycans. WGA
recognizes N-acetyl-glucosamine (GlcNAc) and was also found to have
affinity to sialic acid [104]. The specificity of Jacalin lectin is to galactosyl
(-1,3) N-acetylgalactosamine (GalNAc) and has been used to capture
Olinked glycoproteins [105]. LCA has the specificity like Con A but with
lower affinity [106], although it has a useful affinity for branched fucose.
The specificity of PNA is the same as Jacalin lectin, but the affinity is
affected by sialic acid associated with galactose [105]. The results showed
that each individual lectin captured a different subset of glycoproteins from
serum, and Con A, WGA and Jacalin enriched larger amounts of
glycoproteins than LCA and PNA, although overlaps in specificity were
observed [107]. In order to capture and study a significant part of the
human serum glycoproteome, a multi lectin affinity column containing
ConA, WGA, and Jacalin lectin was designed based on a consideration of
common N-linked and O-linked glycan structures present in serum proteins.
By using the multi-lectin affinity column (M-LAC), 10% of human serum
proteins were found to be glycosylated (w/w basis). If the serum sample
pre-depleted of the six serum HAP (albumin, IgG, IgA, antitrypsin,
transferrin, and haptoglobin), which together count for 80% of serum
proteins, the M-LAC was shown to capture 50% glycosylated proteins (w/w)
in the 6 HAP-depleted sample [107]. The M-LAC was shown to enhance
digging deeper into the proteome following a pre-fractionation of plasma
HAP with immunoaffinity separation and coupled with RPLCMS/MS or nano
LC-MS/MS [108,109]. In another study, a panel of six lectins was used in the
capture step for discovery of glycoprotein biomarkers in serum samples
from prostate cancer and hepatocellular carcinoma subjects. The
enrichment effect of lectin-based capture with pretreatment of
immunoaffinity-based HSA and IgG depletion was compared to a process
without the pretreatment. A workflow was developed that demonstrates
the usefulness of arly validation in biomarker discovery using lectin-based
capture in combination with immunoaffinity-based separation technology
[110].

2.3. Peptide-based affinity subtraction or enrichment

Different from antibody or single-chain antibody, a peptide is much smaller
and has unique advantages in proteomic subtraction or enrichment.
Effective depletion of albumin using a new peptide-based affinity medium
was developed and reported [111]. The albumin-binding capacity is at least
14 mg/mL of gel. The material may be reused hundreds of times after a
simple regeneration step involving NaOH, with full retention of specificity
and capacity. The material was tested with human and monkey plasma and
serum and rat serum, and has been used to deplete liter volumes of human
plasma. Combinatorial libraries of Fab fragments and synthetic peptides
have been used widely in screening and capturing target proteins since
early 90s [112,113]. The power of this technological approach is to provide
exhaustive binding diversity and access to nearly all target proteins in a
complex mixture. Solid phase combinatorial libraries of peptides can be
synthesized via a modified Merrifield approach by using the well-known
separate-recombine-assemble method. The process yields a situation
where each bead comprises multiple copies of the same peptide with all
beads mixed together. The peptide ligand is grafted throughout the bead's
porous structure and can reach a concentration of about 10 to 15 pmole
per bead, ensuring a maximum binding capacity ranging from 5 to 10
mg/mL of settled bead bed. Total binding capacity for protein partners per
single bead is thus around 13 ng depending on the molecular mass. Each
single bead, therefore, has millions of copies of a single unique ligand and
has a different ligand from every other bead [114]. In other methods of
sample preparation, such as immunoaffinity separation, target proteins are
diluted as they flow through the column. In contrast, peptide-based affinity
capture is based upon binding of all protein targets. Immobilized libraries of
combinatorial peptide ligands were applied to purify and concentrate
essentially all of the components of a
complex mixture on ligands synthesized on individual beads. No depletion
or pre-fractionation of the starting material is performed before it is
incubated with the library, and no a prior knowledge of the target protein or
of the ligand to which it binds is required [115]. There is no limit to the
amount of sample that can be applied to the library; therefore the sample
an be loaded until the library captures enough target protein for detection.
There are two types of products currently on the market.

2.3.1. ProteoMiner kit

Commercialized by Bio-Rad Laboratories (Hercules, California US), the
technology is based upon a combinatorial hexapeptides library (called
ProteoMiner), consisting of dozens of millions of hexapeptides capable of
interacting with most, if not all, proteins in any given proteome. In theory,
each unique hexapeptide binds to a unique protein sequence. Because the
bead capacity limits binding capacity, high-abundance proteins quickly
saturate their ligands and excess protein is washed out during the
procedure. In contrast, low-abundance proteins are concentrated on their
specific ligands, thereby decreasing the dynamic range of proteins in the
sample. When analyzed in downstream applications, the number of
proteins detected is dramatically increased. ProteoMiner has been tested
against a number of human biological fluids, such as sera, urine,
cerebrospinal fluid as well as against cell lysates (e.g., platelets, red blood
cells) with interesting results [116124]. This technology was proposed as
one of the effective methods for biomarker discovery and validation [125].

2.3.2. ProSpectrum Libraries

Developed and commercialized by Prolias (Laytonsville, Maryland US),
ProSpectrum Libraries, has a similar technological approach and
mechanism as ProteoMiner. ProSpectrum Libraries consist of millions of
peptide ligands synthesized on microscopic chromatography resin beads.
The ligands are synthesized by solid-phase peptide synthesis, resulting in
millions of beads, each bearing millions of copies of a single ligand, and
each bead theoretically different from every other bead. These beads, each
of which has a protein binding capacity in the nanomolar range, are
incubated with the starting material. The proteins and other components of
the starting material, including viruses, bacteria and chemicals bind to
individual beads according to the standard laws of affinity interactions
[126,127].

Combinatorial peptide library techniques represent a promising and

powerful alternative to immunoaffinity separation as a type of affinity
enrichment methodology. In addition to the advantages discussed above,
the products based upon the combinatorial peptide libraries have the
following features:

1) One product can be used for various samples and species

2) Large capacity, therefore more proteins can be detected
3) Ideal for digging deeper into the proteome with less number of samples
4) Single-use and disposable, which a void cross-contamination

More advantages will be identified after more data of experiments and

applications are published. The weakness or limitation of the technologies
have not been well characterized and reported. In addition to the further
characterization and understanding of reproducibility, elution of bound
materials, and being quantitative in enrichment, the following are some
concerns based upon limited published data:
1) The benefits of the technology may not be obtained if the sample size is
small. In many cases of clinical or animal model studies, the sample size is
a limiting factor.
2) Based on available data, the detection of spiked samples was at ~5
ug/ml, which can be achieved by other affinity depletion methods.
3) The product is currently available in spin column format, which is
relatively easy to introduce variation when processing multiple samples. It
may not be an ideal product if hundreds of samples need to be processed.

2.4. Chemical affinity tagging or enrichment

This approach has been widely used in proteomic sample preparation and
analysis. It utilizes knowledge in chemical bonding and application of
affinity-binding reagents. Some of the methods utilize traditional
biochemical affinity reagents such as dye ligands. Some apply the old
chemicals for new uses in proteomics [128]. A chemical bond is the specific
force in the attractive interactions between atoms and molecules, which
confers specific association and stability between binding chemical
compounds and target molecules [129]. Chemical affinity bindings to
proteins are based upon various types of chemical bonding that can be
non-covalent or covalent bonding between chemical reagents and target
proteins.

2.4.1. Dye-ligand affinity columns

The representative one is Cibacron Blue dye affinity reagent. It is a
polycyclic anionic ligand that interacts with proteins through a specific
interaction, acting as a mimic of NAD+ and NADP+, or through non-specific
electrostatic, hydrophobic, and other forces. Cibacron Blue has been long
used as an affinity reagent for protein purifications, including albumin and
antibodies [17,130134]. This strategy of removing albumin is still being
used in proteomic analyses due to its relatively low cost [85,135139].
Successful application of Cibacron Blue in biomarker discovery was also
reported [135]. Other small molecules have been designed (e.g. mimetic
dyes) and demonstrated greater specificity than Cibacron blue. In human
serum, albumin constitutes 5070% of the total protein and IgGs constitute
1025% [3]. Cibacron Blue ligand together with Protein A/G were developed
and commercialized as affinity-depletion kits by several companies. At the
early stage of proteomic sample preparation, these products were widely
used for HSA and IgG depletion from plasma or serum samples
[82,84,85,87]. Comparative studies indicate that using antibody affinity
ligands for HSA and IgG results in more specific depletion compared to the
traditional Cibacron blue/Protein A or G depletion methods
[52,66,67,69,140]. Because of this demonstrated specificity, the trend is
now towards the use of immunoaffinity media for most proteomic analyses.

2.4.2. Metal or inorganic ligands

Chemical approaches for determining the abundance and post-translational
modification (PTM) of proteins in complex proteomes have been mainly
focused on phosphoproteins through phosphate-binding with metal or
inorganic ligands [128,141]. The identification of proteins via gel-
electrophoresis coupled with LCMS/MS is not always sufficient to interpret
biological function, because many of the naturally occurring proteins are
post-translationally modified. More than 85% of identified proteins were
previously known to be phosphorylated,
which is a mechanism that regulates a large array of cellular biochemical
pathways of the biological system. Traditionally, the study of
phosphoprotein structurefunction relationships used classical protein
chemistry approaches that involved protein purification, peptide mapping,
and the identification of the phosphorylated peptide regions and sites by N-
terminal sequence analysis. Proteomic approaches require the
development of specific strategies to preferentially enrich phosphoproteins.
One of them is through covalent modifications that incorporate affinity tags
using physicochemical properties of phosphoaminoacids [141]. The
Isotope-Coded Affinity Tag (ICAT) method is an approach to specifically
label, enrich, and quantify target proteins in complex proteomes. The
chemical probes consist of three general elements: a reactive group
capable of labeling a defined amino acid side chain (e.g. iodoacetamide to
modify cysteine residues), an isotopically coded linker, and a tag (e.g.
biotin) for the affinity isolation of labeled proteins/peptides. Both solution-
phase and solid-phase ICAT methods were reported [142147]. Several
other chemical reagents have been developed and applied to measuring
the phosphorylation state of proteins in complex proteomes [148]. These
methods were shown to be more effective than metabolic radio-labeling
with inorganic 32P-phosphate [149,150]. Another type of metal tagging
method, called immobilized metal affinity chromatography (IMAC), has also
been developed and applied to phosphoprotein studies [151159].
Phosphopeptides are often present in small amounts and need selective
isolation or enrichment before identification. One method employs a two-
dimensional column setup, with titanium oxide (TiO2)-based solid-phase
material (Titan sphere) as the first dimension and reverse-phase material
as the second dimension. Phosphopeptides were separated from non-
phosphorylated peptides by trapping them under acidic conditions on a
TiO2 pre-column. Non-phosphorylated peptides break through and are
trapped on a reverse-phase precolumn after which they are analyzed by
nanoflow LCESIMS/MS. Subsequently, phosphopeptides were desorbed
from the TiO2 column under alkaline conditions, re-concentrated onto the
reverse-phase pre-column, and analyzed again by nanoflow LCESIMS/MS.
The selectivity and practicality of using TiO2 re-columns for trapping
phosphopeptides were demonstrated via the analysis of a model peptide
RKISASEF, in a 1:1 mixture of a non- and a mono-phosphorylated form. A
sample of 125 fmol of the phosphorylated peptide could be easily isolated
from the non-phosphorylated peptide with a recovery above 90% [160].
The method was developed further with an improved procedure for using
TiO2 microcolumns that significantly enhanced
binding selectivity of TiO2 toward phosphorylated peptides, thereby
enabling phosphorylated peptide characterization from low femtomole level
phosphorylated proteins. This procedure
was shown to be more selective and higher sensitive than the IMAC method
[161]. In addition, the TiO2 purification was fast (typically less than 5 min
per sample) and can be used in combination with high performance liquid
chromatography coupled to either MALDIMSMS or ESIMSMS. For better
selectivity, applying phospho-specific antibodies for detection was reported
successful [162165], although the approach is an immunoaffinity-based
capture. Chemical affinity-mediated enrichment and detection of
phosphoproteome in human cancer cell (U937) has led to identification of
two cancer-related phosphoproteins implicated in intracellular hormones
signaling which are dramatically altered in the course of monocyte to
macrophage differentiation [166].

2.4.3. Hydrazide-based glycoprotein capture

The enrichment of N-glycopeptides is of particular interest for
characterizing the plasma proteome. The changes in abundance and the
alternations in glycan composition of plasma proteins and cell surface
proteins have been shown to correlate with cancer and other disease
states. In fact, numerous clinical biomarkers and therapeutic targets are
glycosylated proteins, such as the prostate-specific antigen for prostate
cancer, and CA125 for ovarian cancer. N-glycosylation (the carbohydrate
moiety is attached to the peptide backbone via asparagine residues) is
particularly prevalent in proteins that are secreted and located on the
extracellular side of the plasma membrane, and are contained in various
body fluids (e.g., blood plasma) [167]. More importantly, because the N-
glycosylation sites generally fall into a consensus NXS/T sequence motif in
which X represents any amino acid residue except proline [168], this motif
can be used as a sequence tag prerequisite to aid in confident validation of
N-glycopeptide identifications. Based on the conjugation of glycoproteins to
a solid support using hydrazide chemistry, a method was developed for the
selective isolation, identification and quantification of peptides that contain
N-linked carbohydrates through steps of stable isotope labeling of
glycopeptides and specific release of formerly N-linked glycosylated
peptides via peptide-Nglycosidase F (PNGase F) [169]. The recovered
peptides were then identified and quantified by MS/MS. This approach was
expanded by combining immunoaffinity subtraction and glycoprotein
capture followed by strong cation exchange (SCX) fractionation, and
reverse-phase capillary liquid chromatography coupled to tandem mass
spectrometry (LCMS/MS) [170]. A total of 2053 different N-glycopeptides
were confidently identified, covering 303 non-redundant N-glycoproteins.
This enrichment strategy significantly improved detection and enabled
identification of a number of low abundance proteins, exemplified by
interleukin-1 receptor antagonist protein (approximately 200 pg/mL),
cathepsin L (approximately 1 ng/mL), and transforming growth factor beta
1 (approximately 2 ng/mL). Another approach for large-scale identification
of N-glycosylated proteins from a complex biological sample, termed
isotope-coded glycosylation-site-specific tagging (IGOT), was based on
peptide-N-glycosidase-mediated incorporation of a stable isotope tag, 18O,
specifically into the N-glycosylation site [171]. The 18O-tagged peptides
were then identified by multidimensional liquid chromatographymass
spectrometry (LCMS)-based technology. The application of this method to
the characterization of N-linked high-mannose and/or hybrid-type
glycoproteins from an extract of Caenorhabditis elegans proteins allowed
the identification of 250 glycoproteins, including 83 putative
transmembrane proteins, with the simultaneous determination of 400
unique N-glycosylation sites. As an effective approach and diversified
alternative, chemical affinity-based protein tagging or enrichment has been
further developed. In one approach, a protein mixture is either first labeled
with an affinity tag and then digested or first digested and then labeled. In
both cases, labeled peptides are subsequently enriched by an affinity
chromatography step, so that ideally only the tagged peptides remain
[172]. Recently, a new and general methodology was described where
sample components containing a chemical moiety of interest are first
selectively labeled with perfluoroalkyl groups, and the entire sample is then
applied to a perfluoroalkyl-silylated porous silicon (pSi) surface. Due to the
unique hydrophobic and lipophobic nature of the per fluorinated tags,
unlabeled sample components are readily removed using simple surface
washes, and the enriched sample fraction can then directly be analyzed by
desorption/ionization on silicon mass spectrometry (DIOSMS) [173].
Chemical affinity tagging and enrichment was also used for proteomics
studies of post-translational modifications in plants. These include affinity-
based enrichment methods, immobilized metal affinity chromatography
and immunoprecipitation used for phosphorylation and ubiquitination
studies, respectively, and the phase partitioning approach for
glycosylphosphatidylinositol modification studies [174]. More developments
along this approach are ongoing. For additional information, please refer to
related reviews [175179]. Label-free method was also recently developed,
which may be a complementary approach to enhance chemical affinity
tagging or enrichment methodology [180].

3. Affinity-mediated proteomics: retrospective and prospective

Affinity technology and application are essential for life science research,
because the majority of biomedical and biopharmaceutical studies are
directly or indirectly dependent upon specific recognition and interactions
within a biological process. Biological assays, diagnostic kits or therapeutic
products are also based on binding affinity between a reagent or a
biomolecule and a target. Affinity technologies are particularly important
for proteomic research, since protein studies under a scale-up condition or
through a systemic approach require more specific and accurate target
capture, separation, enrichment, detection and quantification.

3.1. Separomics: a science of proteomic separation

The purpose of protein separation in proteomic study is different from that
in conventional protein studies in which the focus is to isolate or purify one
or a few individual proteins for biochemical, functional or structural
analysis. In contract, protein separation in proteomic studies aims to collect
large number of proteins in a given biological system for analyzing
proteome composition, proteinprotein interaction,
Fig. 1 Technology positioning and workflow technologies and methods in
proteomic sample preparation are categorized into two major approaches:
subtraction and enrichment. Their relative positions, inter-relationship, and
possible connections to downstream fractionation processes and mass
spectrometry analyses are illustrated. How to select proper technologies
and combine them to form a best fit workflow depends upon the needs and
purposes of each proteomic study. An independent process for validating
and analyzing proteomic studies is also depicted. Certain approaches are
listed as example of applications. In addition to AMIGAP, HPA, SISCAPA, and
those conventional assay methods, an online bioinformatics and database
tool for optimizing subtraction approach, Affymex [213], is listed.

and differential change in concentration under diseased conditions.

Although the underlying mechanisms for many methods are same or
similar to those used in traditional protein biochemistry studies, the
principle, strategies, tools, and processes for proteomic separation are
distinct. The concept of Separomics (originally called Seppromics) was
proposed for describing and defining the technologies, processes,
requirements, standards, and applications for proteomic separation,
including fractionation and enrichment [53]. he most significant challenge
in this new field is how to reduce the complexity or vast dynamic range of
protein
concentration in a given proteome, such as plasma or serum, so that the
downstream proteomic analysis, mainly mass spectrometry, can be carried
out effectively. The technologies and methods reviewed in this paper are all
aimed to conquer this challenge. From this review, it is evident that the
technologies or methods of immunoaffinity subtraction, affinity separation,
and affinity enrichment play critical roles in sample preparation for
enabling the final stage of MS-based analyses. However, each method is
only one of many steps in the entire
process of proteomic analysis (Fig. 1). If we put these technologies or
methods in a given environment such as the human plasma proteome,
which contains a dynamic concentration range of 1012 magnitude, their
relative positions in access to proteins at different concentrations can be
viewed in a bigger picture (Fig. 2). It is clear that each technology or
approach has its own area of function and boundary of action. Systematic
understanding of each technology in this field will assist researchers to
design an integrated process to achieve
the ultimate goal of their studies. In accordance with the concept of an
interactome [181], which describes proteinprotein interactions and their
relationship it other components in a proteomic context, Separomics
emphasizes observing and preserving the natural, interactive, or functional
relationships of the members in the complex protein mixture during the
separation or partitioning process of the protein mixture. In this
consideration, separation, subtraction, fractionation or enrichment of the
protein targets in a solution or complex mixture are designed and
organized in a global and systematic manner. The process accounts for the
natural or structural (hierarchical) relationship of the components in a
protein mixture or proteome to best retain the original biological or
physiological conditions for high quality, representative, and reproducible
sample preparation. This best meets the purpose of specific, accurate and
reliable analysis of the proteins and related components solated from a
complex mixture.
Fig. 2 Technology mapping in a plasma proteome context affinity
separation and enrichment technologies discussed in this review are put
into the context of human plasma proteome, which has a dynamic
concentration range of 10111012 magnitude (from mg/L to pg/mL,
depicted at the top). Each technology or method has its own position and
scope of action. It is clear that subtraction technologies have more specific
functional areas and those for enrichment and detection appear to be
covering the full range of the proteome. The decision to choose the proper
technology, single or in combination, depends upon the needs of different
experiments and the proficiency of different researchers. The application
results may vary accordingly.
3.2. Affinity-mediated capture and detection
Affinity separation and enrichment are only a part of the entire picture of
Affinity-Mediated Proteomics. Affinity binding strategies and approaches
have also been developed and applied to protein identification,
quantification, and characterization. Since these methods use an approach
to capturing and analyzing protein targets independent from those
described above, they are particularly important for validation of the
biomarker or drug target discovered by the
methods of MS-based detection and analysis. AMIGAP (Antibody-Mediated
Identification of Genes And Proteins) is one such approach, which utilizes
gene or protein domain-specific antibodies for screening and validating
biomarkers and drug targets [34]. The idea is to develop
domain-specific antibody libraries based on available genomic information
or on the peptides identified in proteomics studies for further validating the
protein target and providing additional information on the target such as its
distribution, subcellular location, expression profile, and relation to
diseases. This approach was further developed and reduced to practice via
the project of Human Protein Atlas (HPA) [182]. The present version of the
atlas is based on the analysis of 3014 antibodies to 2618 human proteins
and contains more than 2.8 million high-resolution immunohistochemistry
images generated by tissue microarrays and manually annotated by
certified pathologists [183]. The atlas also contains confocal microscopy
images of three human cell lines using four fluorescent probes to give more
detailed data on subcellular localization [184]. Both monospecific
polyclonal and monoclonal antibodies based upon affinity proteomics is
viewed as a part of affinity application for life science [185]. SISCAPA
(Stable Isotope Standards and Capture by Anti-Peptide Antibodies) is
another example of utilizing antibody libraries to standardize MS analysis
on peptides in proteomic studies [186]. The method describes
measurements of plasma digests with enrichment of peptides by antibody
capture, followed by the mass spectrometer as a second antibody that
has absolute structural specificity.
Affinity-mediated capture and detection can go further beyond the
boundary of antibodies. Peptide-based affinity reagents have become more
and more important as affinity reagents. In addition to the peptides for
protein subtraction and combinatorial peptide libraries for protein
enrichment [111,114127], there are many cases where peptides were
successfully developed for protein capture and measurement [187189].
Affinity reagents based on protein scaffolds
and antibody mimicking, such as trinectin, anticalins, affibody, single-
domain antibody fragments (VHHs or Nanobodies and CaptureSelect),
etc were demonstrated to be useful in proteomic separation, enrichment,
and detection [9194,176,190192]. Aptamers are nucleic acid or peptide
fusion species that have been engineered through repeated rounds of in
vitro selection or equivalently, SELEX (systematic evolution of ligands by
exponential enrichment)
to bind to various molecular targets such as small molecules, proteins,
nucleic acids, and even cells, tissues and organisms [193196]. The
technologies in affinity-mediated proteomics are
summarized in Table 3, where the technologies and methods are divided
into three categories and listed chronologically according to their year of
publication.

3.3. Application improvement and development

The technologies and methods described above are useful in certain
applications, but none of them are perfect in meeting the needs of all
proteomic studies. Immunoaffinity separation or subtraction is currently
viewed as one of the most effective proteomic approaches to studying
complex protein mixtures [55,197201]. Successes of applying
immunoaffinity separation to identifying potential biomarkers were
reported in quite a few studies [56,57,63,66,67,70,202,203]. However,
there were also cases where protein subtraction was not completely
satisfactory. Often, adjusting experimental conditions, such as sample
dilutions, loading amount, buffer pH, chromatography program setting, can
correct the issues of column leakage or incomplete subtraction. Different
binding and washing conditions of immunoaffinity
columns can also make significant differences in protein separation or
partitioning. For example, addition of denaturing reagents or detergents in
the wash buffer can directly affect
proteinprotein interactions and result in different protein separation.
Nevertheless, this issue can be empirically addressed by optimizing and
standardizing a particular separation process based upon the research
needs and downstream processes. Sample dilution and target protein loss
through the immunodepletion process can be a concern [204]. There was a
report that albumin, as a serum carrier protein, naturally binds various
proteins in circulation, which may cause unwanted loss of target proteins
during the process of removing albumin [205]. If the association of the
target protein with albumin or other abundant proteins is specific and tight,
then the capture of albumin or the specific abundant proteins can be used
as an enrichment process of affinity proteomics. It is important to
understand and control the protein partitioning effect by collecting and
measuring each fraction in the separation process. Collected fractions can
be further analyzed for co-captured biomarkers, enabling target associated
proteomic analyses, such as Albuminomics, Haptoglobinomics,
Apolipoproteinomics, etc. Albuminomics was successfully used for
analysis of albumin-associated peptides and proteins from ovarian cancer
patients [206]. This emphasis on analyzing co-captured biomaterials
distinguishes the concepts of separation from depletion.
Immunoaffinity separation as a single step in proteomic sample preparation
is often not enough. Compatible downstream processes incorporating the
immunoaffinity subtraction process with other enrichment or fractionation
methods prior to final analysis have been developed. There are many
examples of using immunoaffinity subtraction as the first step
of sample preparation, following with further fractionation using 1DE, 2DE,
IEF, SCX, RP-HPLC, PF 2D, etc., prior to LCMS/MS analysis [5355,197
199,207209]. There were also reports on combination of immunoaffinity
subtraction with other enrichment methods, such as lectins column (M-LAC)
for better resolution and coverage of MS analysis of glycoproteins
[210,211]. In another study by applying HAP depletion, glycoprotein
enrichment and stable isotope labeling, a potential biomarker,
angiotensinogen, was found to be present at high levels in patients with
obesity plus diabetes and hypertension [202]. In general, the ideal
approach is to obtain most data through the simplest process. This is
particularly important when dealing with large number of samples, which is
often the case for biomarker discovery studies. Choosing the
sample preparation methods that can be incorporated into semi- or fully-
automated system will greatly reduce sample to-sample variations and
improve reproducibility [55,65]. On the other hand, in-depth analysis via
multiple steps, parallel approaches, and quantitative analysis can lead to
reduction in complexity of a given proteomic sample and obtaining more
comprehensive data [199,212].

3.4. Trends of technology development

In the future, proteomic separation and enrichment will continue to be one
of the most applied tools in proteomic studies. Driven by the demands for
methods and processes that are more specific, sensitive, accurate,
reproducible, and effective, it can be foreseen that the following frontiers
are advancing:
1) Further development of HAP and MAP subtraction, where Seppro IgY-
HAP-SuperMix coupled system is one example of trying to capture and
fractionate the MAP immediately after removal of HAP in plasma [65].
2) Optimizing combinatorial processes by coupling immunoaffinity
subtraction of HAP with other enrichment methods such as M-LAC or
ProteoMiner, or other conventional separation methods such as 2DE, SCX,
RP-HPLC, etc. [53,55,202].
3) Further exploring applications of cross-board enrichment methods
(ProteoMiner and ProSpectrum Libraries) to the field of biomarker discovery
and possible clinical assay enablement [124,126].
4) Development of tissue or cell specific protein affinity columns for
detecting cellular or tissue leakages in plasm or other body fluids, where
SuperEnrich poineered along this direction.
5) Further developing protein labeling, tagging, specific binding chemical
affinity ligands, to enable more robust and inexpensive processes for
proteomic enrichment.
6) Development of automation, multiplex, and high-throughput
technologies to reduce human manipulation and improve reproducibility of
a proteomic separation and/or enrichment process. Basically, all LC
products for proteomic subtraction or enrichment are on-line and
semiautomated. More closed and completed automation systems are
needed. SepproTip system is the first product available that is fully
automated and can process 12 samples of immunoaffinity subtraction
within 1 h.
7) Better approaches or methods that can significantly reduce the dynamic
range and complexity of proteomic samples. Innovative technology and
revolutionary approaches await
to be invented, tested, and developed.
8) Further development of affinity reagent standards and database tools
such as AMIGAP [34], HPA [183], SISCAPA [186], Affymex [213], Human
Proteinpedia [63], etc., promoting standardized and comparable proteomic
analyses across different research groups and laboratories.
9) From discovery to practice, affinity-mediated proteomics is expected to
enable further development of biomedicine [214]. Quality sample
preparation and clinical relevance of the affinity technologies and
standards will further drive the applications of proteomics to biomedical
research, early disease diagnostics and drug development [215218].

4. Closing remarks
Affinity-mediated protein separation science is an integral part of
proteomics. Affinity separation and enrichment are the rate-limiting steps
in the process of proteomic analyses and
studies. Developments in the science and technologies involved are
unveiling and fostering a new discipline: Separomics, which requires
putting protein separation or enrichment into a context of proteomics and
systems biology. It will play an increasing role in life science and biomedical
research and development. Further challenges and unsatisfied needs in
proteomics are how to utilize the affinity strategy and technology to reduce
the sample complexity and simplify a complex process or system for
enabling specific, sensitive, accurate, reproducible, and effective protein
analyses. The technology platforms and tools of proteomic separation or
enrichment correspondingly need to be robust, automated, multiplexed,
high-throughput, reliable, and inexpensive.

Acknowledgements
The authors sincerely thank Michael J. Zhang for his efforts and
contributions in preparing the references and revising the manuscript of
this article. The authors also sincerely thank the reviewers for their helpful
comments and suggestions.