Cytokine Antibody Arrays in Biomarker Discovery and Validation

Current Proteomics, 2012, 9, 55-70 55 Cytokine Antibody Arrays in Biomarker Discovery and Validation

Ruo-Pan Huang*,1,2,3, Brett Burkholder2, Valerie Sloane Jones2, Wei-Dong Jiang1,2, Ying-Qing Mao1,2, Qiao-Lin Chen1,3 and Zhi Shi1,3

1RayBiotech, Inc, Guangzhou, China 510600; 2RayBiotech, Inc, Norcross, GA, USA 30092; 3South China Biochip Re- search Center, Guangzhou, China 510630

Abstract: Many normal physiological and disease-related pathophysiological processes are regulated by the interactions of complex signaling networks of pro- and anti-inflammatory markers, growth factors, soluble receptors and extracellular matrix proteins, as well as other cell–cell signaling proteins, which we define collectively as cytokines. Because multiple cell–cell signaling factors may be involved in a single biological process, detection of expression of multiple cytokines is essential to unraveling the mechanisms and effects of many disease processes. Cytokine antibody arrays have been developed to meet this growing demand for multiplexed protein detection. In particular, discovery and validation of disease- related protein biomarkers require high-throughput detection of many proteins simultaneously. This review will address the complexity of cytokine biology, discuss current and future antibody array technologies and explore their applications in cytokine biomarker discovery and validation for variety of human diseases. Specific examples of these applications will be presented, including the search for cytokine biomarkers related to neurological and neurodegenerative diseases (such as autism and Alzheimer’s), immunological disorders (including asthma), and various cancers.

Keywords: Antibody arrays, biomarker, cytokine, protein arrays.

INTRODUCTION THE COMPLEXITY OF CYTOKINE BIOLOGY Cytokines are among the most intensely studied protein Cytokines are typically regarded as pro- or anti- families in biomedical science, with tens of thousands of inflammatory markers, such as chemokines, or as secreted publications devoted to exploring their myriad functions and proteins that induce an immunomodulatory response. How- potential applications in biomedicine. In addition to playing ever, it is useful to consider these molecules in a broader critical roles in many normal cellular events, cytokines are context. A better definition of a cytokine may be any se- implicated in the initiation and development of almost every creted protein—distinct from non-peptide neurotransmitters major life-threatening disease. For this reason, cytokines and traditional peptide hormones—that generates or trans- have long been explored as potential disease biomarkers. mits cell–cell signals. This diverse group of molecules in- Evaluation of cytokine expression levels, ideally by cy- cludes growth and differentiation factors, pro- and anti- angiogenic factors, adipokines, and extracellular matrix pro- tokine expression profiling, is a critical step in biomarker teins (for specific examples, see Table 1). These proteins, discovery and development process. Although enzyme- collectively, are responsible for a number of physiologic linked immunosorbent assay (ELISA) is the most common processes, including inflammation [1], immune response [2]; method and the gold standard for measurement of individual [3], cell migration [4] and tissue infiltration [5], angiogenesis cytokine expression levels, the key to successfully identifying promising new biomarkers is the simultaneous detection [6], adipogenesis [7], apoptosis [8], wound healing [9], immune cell maturation [10, 11], hematopoiesis [12] and cell of multiple cytokines. growth, proliferation and differentiation of other cell types To meet this challenge, cytokine antibody arrays have [13]. been developed and, in many cases, successfully utilized in Moreover, cytokines play a prominent role in many dis- the discovery of biomarkers and key disease-related molecules. Antibody arrays enable the researchers to profile mul- eases. In cancer, for example, various cytokines can promote tumorigenesis, angiogenesis, invasion, metastasis and tumor tiple cytokine levels from trace amounts of biospecimens, an growth and proliferation [14, 15]. Cytokines mediate patho- approach which holds great promise for biomarker discov- genic inflammatory and immunomodulatory functions in ery. Other multiplex technologies, such as aptamer arrays cancer [14] and many other diseases, including auto-immune and suspension bead assays, may also be used for cytokine diseases, such as rheumatoid arthritis, psoriasis, multiple profiling; however, only planar cytokine antibody array technologies and their applications in biomarker discovery sclerosis, systemic lupus erythematosus, type 1 diabetes, Crohn’s disease and other inflammatory bowel diseases. Re- will be the focus of this review. cently, it has been suggested that chronic inflammation plays a key role in neurodegenerative diseases, such as Alz- heimer’s [16] and Parkinson’s diseases [17], insulin resis- *Address correspondence to this author at the RayBiotech, Inc. 3607 Park- way Lane, Norcross, GA 30092, USA; Tel: (770)729-2992; tance and type 2 diabetes [18], atherosclerosis [19]; [1] and Fax: (770)206-2393; E-mail: [email protected] other cardiovascular diseases [1].

Table 1. Broadly defined, cytokines are comprised of several diverse classes of proteins. This Table summarizes the major cytokine classes and presents examples from each class, as well as their associated physiologic functions.

Class / Type Examples in Class / Type Typical Physiologic Functions Immunomodulators Interleukin 6 (IL-6) Pro- and anti-inflammatory signaling Interleukins Interleukin 12 (IL-12) Growth & differentiation of immune cells ENA-78 Induce chemotaxis Chemokines (Chemoattractants) MCP-1 Regulation of inflammatory response RANTES Wound healing Interferon alpha-2 (IFN-2) Suppress viral replication Interferons Interferon gamma (FN) Activation of Natural Killer cells Tumor necrosis factor alpha (TNF) Apoptosis TNF superfamily Tumor necrosis factor beta (TNF) B-cell development CD40 ligand T-cell activation & proliferation

C-reactive Protein (CRP) Stimulate inflammatory response Complement factors Induce chemotaxis Innate immunity proteins Fibrinogen Stimulate phagocytosis Prothrombin Increase antigen presentation of pathogens Growth Factors TGF-beta 1 Promote survival, proliferation and/or differentiation of many TGF-beta superfamily Bone morphogenetic proteins (BMPs) cell/tissue types Growth & differentiation factors (GDFs) Regulation of immune response Sonic Hedgehog (Shh) Establish polarity of body axes during embryogenesis Morphogenic Wingless (Wnt) Maintenance of regenerating tissues Ciliary neurotrophic factor (CNTF) Neurotrophic Survival, differentiation and maintenance of nerve cells Neurotrophin 3 (NT-3) Colony Stimulating Factors Hematopoietic Erythropoietin Proliferation and differentiation of red and white blood cells Stem cell factor VEGF Angiogenic Regulation of angiogenesis Angiogenin Adipocytokines Adiponectin Energy balance Adipocyte-specific Leptin Glucose homeostasis Resistin Insulin sensitivity Neuropeptide Y Energy balance Neuropeptides Peptide YY Regulation of appetite Cholecystokinin (CCK) Energy balance Gastric Hormones Glucagon-like peptide 1 (GLP1) Regulation of appetite Extracellular Matrix (ECM) Proteins E-Selectin Soluble Adhesion Proteins Promote endothelial inflammation ICAM1 Metalloproteinases & Their Matrix Metalloproteinases (MMPs) Tissue remodeling & repair Inhibitors Tissue Inhibitors of MMPs (TIMPs) Angiogenesis Apoptosis Cysteine Proteases Cathepsins Bone tissue remodeling Cytokine Regulatory Proteins IGFBPs Binding Proteins Modulation of cytokine functions -2-Macroglobulin sIL-2R Soluble Receptors Modulation of cytokine functions sTNFRI

Cytokine Antibody Arrays in Biomarker Discovery and Validation Current Proteomics, 2012, Vol. 9, No. 1 57

Since cytokines play a pivotal role in the pathology of of soluble receptors is suggested by the tight regulation of many diseases, a number of approved treatments either target these processes and their conservation in human, rodent and cytokines or are recombinant cytokines themselves (Table avian species [30]. 1). Several important cytokine targets are the epithelial In sum, the complexity of cytokine biology necessitates growth factor receptor (EGFR/ErbB1) and heregulin-2 consideration of the expression of multiple cytokines. For (HER2/neu/ErbB2) receptors, which are overexpressed in this reason, multiplexed protein detection assays are a critical many tumors [20], and tumor necrosis factor alpha (TNF), tool for success in illuminating the mechanisms of cytokine- which is a key pro-inflammatory protein in many auto- driven processes. immune diseases [21]. Examples of approved recombinant cytokine biologics are interferon-alpha (IFN), which is used ANTIBODY ARRAY TECHNOLOGY to treat chronic viral hepatitis [22] and AIDS-related Ka- posi’s Sarcoma [23], or interferon-beta (IFN), which is used The complexity of cytokine networks and the limitations in treating multiple sclerosis [24]. of single-analyte detection methods (ELISA, Western blot, etc.) have prompted the development of a more highly paral- Many cytokines are pleiotropic, meaning that a single lel approach to measure multiple cytokine levels. Among the cytokine may induce a wide range of effects in various cell multi-analyte technologies that have emerged over the last and tissue types [25]. Interleukin-6 (IL-6), for example, was decade, antibody arrays are one of the most attractive and originally characterized as a factor in B-cell differentiation promising methods. [26]. Subsequent investigations have identified pleiotropic effects of IL-6 in functions as diverse as proliferation of he- The design principle of antibody arrays is usually based matopoietic stem cells, maturation of macrophages and T on either a sandwich immunoassay or a direct-labeling ap- cells, induction of acute phase response in the liver and proach. The sandwich format requires a pair of antibodies: stimulation of neuronal development and osteoclast forma- one immobilized (capture antibody) and one in solution (de- tion [27]. tection antibody) (Fig. 1A); this method has the advantages of excellent specificity and sensitivity. However, the combi- The function of cytokines is often redundant, wherein nations and the number of antibody pairs in each array are multiple cytokines have similar effects on a single cell type both limited by occasional cross-reactivity between detection [25]. For example, both leukemia inhibitory factor (LIF) and antibodies. For this reason, the practical limit on the number oncostatin M (OSM) share with IL-6 the capacity to inhibit of antibody pairs that can be included in one array panel is the growth of murine myeloid leukemia cells and the ability approximately 100. If detection of a larger number of targets to induce them to differentiate into macrophages. All of is desired, the sample can be analyzed sequentially with mul- these molecules share protein sequence homology with each tiple panels of compatible antibody pairs [31]. Alternatively, other and with IL-11, ciliary neurotrophic factor (CNTF) and larger arrays may be constructed with single-antibody detec- cardiotropin-1 (CT-1). The receptors of these cytokines also tion using the direct-labeling approach (Fig. 1B). In this share sequence homology, and signal transduction of these format, fluorescence or biotin labeling of target proteins is ligand-receptor complexes requires recruitment of a mem- employed in lieu of introducing detection antibodies [32]. brane-bound protein, gp130 [27]. Because the cross-reactivity is eliminated in the label-based Induction, modification and regulation of these extracel- format, array densities can be scaled up to a theoretically lular signals are accomplished through complex interactions unlimited size. Both the label-based method [33] and the with other cytokines. Often, the biological effects of cytoki- sandwich-based method [34] have proven valuable in the nes on a single cell depend upon multiple, simultaneous sig- discovery of disease biomarkers. nal inputs from different cytokines. The net effects of cyto- The solid supports used in the fabrication of planar cytokine–cytokine interactions can be additive, antagonistic, syn- kine antibody arrays include glass or plastic slides, ELISA ergistic, sequential, dose-dependent, intracrine, autocrine or plates, and nitrocellulose membranes. Nitrocellulose and paracrine [28]. In antigen-presenting dendritic cells, for ex- chemically activated or polymer-coated glass microscope ample, high expression levels of IL-12 are dependent on sig- slides are most commonly used supports. Membrane-based nals elicited by both CD40 ligand and interferon-gamma arrays typically have chemiluminescent readouts, making (IFN). while CD40 signaling was sufficient to induce secre- them a favorable, low cost option that is easily adaptable to tion of tumor necrosis factor alpha (TNF) and IL-8 from existing Western blot detection systems. Glass-slide–based dendritic cells [29]. arrays provide the advantage of miniaturization; the surface Additionally, the activities of cytokines are modulated by can accommodate minute amounts of capture antibody (spots a complex network of various binding proteins, inhibitors, <200 m in diameter), thereby reducing the array size and and soluble receptors. The latter appears to be a common the sample volume required [35]. Smaller array size then mechanism of regulation for many secreted cell–cell signal- allows multiple arrays to be printed on one chip, making ing proteins [30]. Cytokines exert their influence by binding them suitable for high-throughput (HT) analysis [36]. Each to membrane-associated receptors. However, many cytokine 25 mm x 75 mm glass chip can accommodate thousands of receptors exist as soluble factors, secreted from the cell by a antibody spots using current contact or non-contact array number of mechanisms, including proteolytic cleavage of the printers. extracellular domain from the cell surface (as seen with Fluorescence is by far the most commonly used readout TNF, IL-1 and M-CSF receptors) or expression of alterna- system for glass-slide–based arrays [37-40], offering high tively spliced transcripts (as seen with IL-4, Il-5 , IL-7 and signal stability and wide dynamic signal range. That being LIF receptors). The importance of the modulatory functions 58 Current Proteomics, 2012, Vol. 9, No. 1 Huang et al.

A. Sandwich B. Compeve C. Direct Label D. Label-free Method Method Method Method

Fig. (1). Signal detection in most antibody arrays relies on common labels (biotin, fluorescent dyes, or enzymes) attached to the antibody or antigen. The sandwich method (A) requires an immobilized capture antibody and a labeled, in-solution detection antibody. This method has the advantages of excellent specificity and sensitivity. The competitive method (B) relies on competitive binding between the analyte and a labelled antigen. In the direct-label method (C), the label is attached to the analyte itself, while the label-free method (D) relies on an inherent property of the analyte (such as mass per unit area or surface plasmon resonance) to produce a signal. said, assuming one has access to compatible detection sys- ers can have a transformative impact on health care. Early tem, obtaining array spot signal intensities for any of these diagnostic biomarkers can identify disease at early stages, array formats (glass slide, membrane or ELISA plate) could drastically reducing mortality rates. Predictive biomarkers be accomplished using fluorescence or chemiluminescence may also help match the individual treatment to different detection. For example, the Li-Cor Odyssey scanner (Li-Cor patients suffering from same type of disease. Prognostic Biosciences, Lincoln, Nebraska), which detects infrared (IR) biomarkers can help to predict the outcome of patients. Bio- fluorescence, is compatible with semi-quantitative glass- markers can also be critical components in drug discovery slide and membrane-based antibody arrays by substitution of and development programs by identifying new drug targets the standard reporter with an analogous IR fluor. or by serving as a surrogate marker for drug efficacy. Currently, there are several companies which manufac- Since cytokine dysregulation is implicated in numerous ture cytokine antibody arrays. The commercial availability of pathologic processes, cytokines have been extensively inves- cytokine antibody arrays greatly facilitates their widespread tigated as potential biomarkers. Many such studies have util- applications as biomarker discovery tools. See Table 3 for a ized ELISA to assess individual cytokine levels. However, summary of cytokine antibody arrays, categorized in terms because this approach greatly restrains the progress of bio- of solid support (platform), detection method and type of marker discovery, many researchers have turned to pro- immunodetecton employed (design principle). Please note teomics in an attempt to screen entire proteomes for disease- that these categories represent 3 different properties of any mediating proteins. That being said, no assay is perfect for given antibody array and can potentially be applied in any all applications, and antibody array technologies are no ex- combination (e.g., any of the detection methods listed can be ception. used with membrane-based arrays, which may be based upon A significant challenge to the development of antibody any of the three design principles.) For example, it is theo- arrays is the availability of large numbers of antibodies with retically possible to detect array signals in the ELISA-plate appropriate specificity and range of sensitivity [41-43]. Not format using fluorescence; however, at the time of this pub- only is the pool of array probes largely limited to available lication, all commercially available arrays for the ELISA- monoclonal and polyclonal antibodies, but each antibody plate format used only chemiluminescence or electrochemi- must be carefully validated for performance and compatibil- luminescence detection. This is because the available fluo- ity in microarray format, which limits the pool further. Any rescence scanners that were compatible with the ELISA- proteins detected using an immunoassay, including antibody plate format were expensive and, therefore, uncommon. arrays, would necessarily be ones that have previously been identified and for which antibodies have been produced. As BIOMARKER DISCOVERY AND DEVELOPMENT PLATFORMS such, any biomarkers identified using immunoassays will not be “novel” proteins. This underscores the need for large- Biomarkers include biological molecules, such as pro- scale development of new antibodies to a wider range of teins, DNA or mRNA, which can be measured to indicate proteins. Because traditional monoclonal and polyclonal normal biological processes, pathogenic processes, or phar- antibodies are laborious and time-consuming to produce, macologic responses to a therapeutic intervention. Biomark- recombinant antibody phage-display libraries are currently Cytokine Antibody Arrays in Biomarker Discovery and Validation Current Proteomics, 2012, Vol. 9, No. 1 59

Table 2. A number of currently approved biopharmaceutical treatments target cytokines. This table comprises a partial list of approved cytokine-targeted biopharmaceuticals, Including several drug classes, their indications and the cytokines targeted by each class. note that the indications listed represent those in the class as a whole; individual biopharmaceuticals may not be approved for all indications listed.

Cytokine Type of Drug Generic / (Brand) Names Approved Indications*

Metastatic colorectal cancer, Squamous cell carci- Anti-EGFR MAb cetuximab (Erbitux) noma (head & neck) EGFR gefitinib (Iressa) Metastatic colorectal cancer, Glioblastoma, NSCLC, EGFR inhibitor lapatinib (Tykerb) Metastatic RCC, Pancreatic cancer erlotinib (Tarceva) HER2/ ErbB2 Anti-HER2 MAb trastuzumab (Herceptin) HER2+ breast cancer sunitinib (Suvent) VEGFR VEGFR inhibitor sorafenib (Nexavar) CML, GI stromal tumors, RCC pazopanib (Votrient)

Bevacizumab (Avastin) Metastatic colorectal cancer, Glioblastoma, NSCLC, Anti-VEGF MAb VEGF Ranibizumab (Lucentis) Metastatic RCC, Macular degeneration. Anti-VEGF apatmer pegaptanib sodium (Macugen) Macular degeneration imatinib (Gleevec) PDGFR PDGFR inhibitor CML, GI stromal tumors, RCC dasatinib (Sprycel) AIDS-related Kaposi’s sarcoma, malignant mela- (Intron-A, Rebotol, Roferon, Infergen, Recombinant IFN-2a noma, various leukemias, chronic & acute Hepatitis Alferon-N) C, genital warts Recombinant IFN-2b (Pegy- Chronic & acute Hepatitis C, Acute Hepatitis B, IFN (Pegintron, Pegasys) lated) Hepatitis D TLR7 agonist (increases Genital warts, Actinic keratoses, Superficial basal expression of IFN, IFN, imiquimod (Aldara, Zyclara) cell carcinoma TNF and other cytokines) IFN Recombinant IFN (Avonex, Rebif, Betaseron) Multiple sclerosis inflizimab (Remicade) adalimumab (Humira) Rheumatoid arthritis, Psoriasis, Crohn’s disease, anti-TNF MAb golimumab (Simponi) Ulcerative colitis certolizumab (Cimzia) TNF / TNF natalizumab (Tysabri) TNF / TNF inhibitor etanercept (Enbrel) Rheumatoid arthritis, Psoriasis, Crohn’s disease TLR7 agonist (increases expression of IFN, IFN, imiquimod (Aldara) Genital and perianal warts, melanomas TNF and other cytokines) CD80/CD86 anti-CD80/CD86 MAb Abatacept (Orencia) Rheumatoid arthritis Glucagon-like Pep- Recombinant GLP1 Exenatide (Byetta) Type 2 diabetes tide-1 (GLP1) *One or more drugs in category are indicated for these conditions, but individual drugs in category may not be approved in all markets for all indications listed. being explored as an alternate source of molecular probes concentrations (10 pg/ml or less); this is particularly true for [44-46]. pro-inflammatory proteins. However, the normal serum/plasma concentrations of adiponectin are in the μg/ml Two other challenges in applying proteomics to the range, over 1 million-fold difference [48]. By contrast, anti- discovery of disease biomarkers are the complexity of hu- body-based detection assays typically have an effective con- man biological fluids and the wide range of abundance of centration range of about 1000- to 10,000-fold. Thus, opti- each of their constituents. Serum and plasma for instance, mized detection of a given set of cytokines in serum or may contain more than 100,000 proteins, many of which exist at very low abundance (pg/ml range or lower) [47]. For plasma may require separation of antibody panels according to concentrations of the various cytokines to be detected. example, there is no single dilution of serum or plasma samples that can result in detection of all cytokines present Another significant challenge to the application of anti- within the linear range of the antibodies used to detect them. body arrays is sample-dependent interference. This interfer- In plasma or serum, many cytokines are present at low pg/ml ence may be represented by high background (usually attrib- 60 Current Proteomics, 2012, Vol. 9, No. 1 Huang et al.

Table 3. This table presents multiple types of antibody array technologies grouped according to the choice of solid support (platform), the detection method employed and design principle (e.g., sandwich ELISA-based or label-based). Antibody arrays in each category have inherent advantages and disadvantages, which are summarized in this table along with the current commercial sources for arrays each category.

Major Advantages Major Disadvantages Company Names

No specialized equipment Semi-quantitative Panomics, RayBiotech, Membrane-based Easiest array platform to use Larger sample consumption R&D Systems, Whatman Robust antibody stability Difficult to adapt to automation Reduced sample consumption Laser scanner required Platform Higher density arrays possible Full Moon Biosystems, Glass-slide–based Gentel, RayBiotech, Adaptable to automation Sigma-Aldrich, Whatman, Compatible with DNA microarray

Reduced sample consumption Low density arrays Aushon, MesoScale, 96-well Plate Adaptable to automation May require specialized detector Quansys, R&D Systems High detection sensitivity Single channel Aushon, Panomics, Quan- Chemiluminescence Fast signal decay sys, RayBiotech, R&D Systems

Durable signals High background with membranes Clontech, Full Moon Bio- Visible Fluorescence Can be multi-channel Laser fluorescent scanner required systems, RayBiotech, Large dynamic range Sigma-Aldrich, Whatman

Low background on membranes IR fluorescent scanner required Panomics, Quansys, Ray- Infrared Fluorescence Detection Durable signals Limited detection resolution Biotech, R&D Systems Method No specialized equipment Low sensitivity Colorimetry Visible detection signals Limited dynamic range RayBiotech Adaptable to point-of-care Semi-quantitative

Silver-enhanced Gold High detection sensitivity High cost of detection reagents Gentel Nanoparticles Inexpensive scanner system High background

Electro- High sensitivity Expensive, dedicated scanner MesoScale chemiluminescence Low CV Low density

Highest specificity Hard to develop high density arrays Aushon, Gentel, MesoS- Lower CV Requires ELISA antibody pairs cale, Panomics, Quansys, Sandwich-based Hard to customize for new targets RayBiotech, R&D Sys- tems, Whatman Design Can develop high density arrays Higher background Principle Label-based (single Full Moon Biosystems, Easily customized for new targets Lower specificity channel) RayBiotech Higher CV

Label-based (dual chan- Same as single-channel Same as single-channel, except: Clontech, Sigma-Aldrich nel) Can only compare 2 samples utable to non-specific binding of sample proteins to the array laborious fractionation techniques (2-dimensional gel elec- surface) or matrix effects. In the context of an immunoassay, trophoresis or liquid chromatography) to deplete high- the latter is a general term describing disruption of antigen- abundance proteins. However, these separation techniques specific interactions by one or more components of the sam- result in low reproducibility and sensitivity [49]; [50]. This ple itself. For example, the epitope on a target molecule may drawback, along with the low detection sensitivity of be blocked because of interaction of the target protein with MS/MS, has hampered the success of these traditional pro- another protein in the sample or an auto-antibody to the same teomic methods in identifying valuable biomarkers. Yet an- target. However, the challenge of matrix effects and dispa- other difficulty is the necessity to subsequently develop im- rate concentrations of cytokines are not unique to antibody munoassays to validate those proteins identified. arrays; these challenges plague more traditional methods of The emergence of antibody microarrays 11 years ago proteomics as well [47]. [51] has sought to address these limitations. Following the Until now, the majority of proteomic research has been same concept as the DNA microarray, cytokine antibody undertaken using the classical methods of single or tandem arrays are an affinity-based proteomic method employing mass spectrometry (MS/MS), which must be preceded by antibodies as probes instead of nucleic acids. Cytokine anti- Cytokine Antibody Arrays in Biomarker Discovery and Validation Current Proteomics, 2012, Vol. 9, No. 1 61 body arrays are now widely recognized as a reliable and ro- validation into a predictive matrix of 18 markers able to clas- bust methodology for mining complex biomarkers. In addi- sify patients as having Alzheimer’s. Then the authors used tion to eliminating the need for separation and depletion two test sets to evaluate the predictive value of the 18 mark- techniques, the high sensitivity of antibody probes allows for ers. HT and high-specificity detection at pg/ml levels. Since an- The first test set consisted of plasma samples from 42 tibody arrays by necessity use well-characterized antibodies Alzheimer’s patients, 11 with other types of dementia (OD) as capture agents, immunoassays for individual targets are and 39 non-dementia controls. Each sample was tested with easily implemented for validation. Thus, biomarker discov- the cytokine antibody arrays to simultaneously measure 120 ery can advance seamlessly and rapidly from discovery cytokine levels and used a prediction analysis of microarray phase to validation phase. Cytokine antibody arrays offer the (PAM) scoring matrix based upon the expression of the 18 additional advantages of much higher detection sensitivities, markers to classify the patients. Using this method, they lower cost, and ease of use compared with the MS-based were able to correctly identify the clinical diagnosis of these approach. For these reasons, cytokine antibody array tech- patients with 89% accuracy (90% agreement for those with nology has been widely used in biomarker discovery pro- Alzheimer’s, 88% agreement for non-Alzheimer’s). grams. Specific examples of such applications are discussed in the following sections. The second test set consisted of plasma from 47 pre- symptomatic patients (patients with mild cognitive impair- CYTOKINE ANTIBODY ARRAYS IN NEUROLOGI- ment (MCI)). While not every patient with MCI will ulti- CAL & NEURODEGENERATIVE DISEASE BIO- mately progress to Alzheimer’s, MCI can be an early warn- MARKER DISCOVERY ing sign of Alzheimer’s for some patients. The authors tested archived plasma samples from patients with symptoms of Cytokine antibody arrays can provide insight into com- MCI who were followed longitudinally. Some were subse- plex disease processes. Unfortunately, for many neurological quently diagnosed with Alzheimer’s, and some with other and neurodegenerative diseases, very few details of the dis- types of dementia. Using the same PAM scoring method, ease process are fully understood. One example of the use of they analyzed the results of these tests to see which of the 47 antibody arrays to explore the molecular basis of disease is patient’s cytokine expression profiles predicted an Alz- an investigation of the role of neuroinflammation in autism heimer’s diagnosis. The authors were able to predict which [52]. Autism is a neuro-developmental disorder resulting in a patients were ultimately diagnosed with Alzheimer’s with wide range of symptoms which include repetitive behavior 81% accuracy (91% agreement with Alzheimer’s patients and impaired social interactions. Although the etiology of and 72% for non-Alzheimer’s patients). Moreover, the autism is still enigmatic, an inflammatory mechanism had plasma samples were taken up to 6 years before clinical di- long been suspected. To investigate the correlation between agnosis with Alzheimer’s. These results suggest that the 18- symptoms of autism and neuronal inflammation, Vargas, marker panel may be useful as an early screening test for et al., used a cytokine antibody array to screen for inflamma- Alzheimer’s. tory factors in the CSF and brain lysates of autistic patients compared with age-matched normal controls. This seminal report not only describes the discovery of a blood test for Alzheimer’s, but also illustrates a broad strat- The results showed consistently higher expression of egy for screening for disease biomarkers, providing a road- MCP-1 and TGF1, among other cytokines, in various re- gions of the brains of autistic patients. This study was the map for future biomarker investigations using antibody array first to implicate neuroinflammation in the etiology of technologies. autism, and the results gave a clear indication of which pro- Dr. Wyss-Coray’s group continues to pioneer the use of teins might be mediating it. Without a cytokine screening antibody arrays and novel methods of data analysis in search tool such as the cytokine antibody arrays, researchers would of biomarkers for neurodegenerative diseases. In a recent have had to correctly select individual antibodies for West- investigation by Wyss-Coray and Philipp Jäger, protein ex- ern blots or ELISA to find the key factors of neuroinflamma- pression profiles for 776 secreted cytokine signaling factors tion. By using the predetermined panel of cytokine antibod- were screened in plasma of patients with sporadic Alz- ies in the array, the authors were able to efficiently screen heimer’s compared to those of healthy controls. Statistical their precious biospecimens for inflammatory mediators. In evaluation between the two groups using Significance so doing, they shone a light into a previously sealed black Analysis of Microarray (SAM) algorithms revealed differen- box. tial expression of 42 proteins among Alzheimer’s patients The utilization of antibody arrays to detect biomarkers is including several previously associated with Alzheimer’s, well-illustrated in a landmark study by Wyss-Coray at Stan- e.g., amyloid precursor protein (APP) and apo-lipoprotein E ford on the key plasma markers of Alzheimer’s disease [53]. (ApoE). Their analysis also included the use of an elastic net In this study the authors describe how they used cytokine regression algorithm (eNet) to identify even more potential antibody arrays to discover plasma biomarkers specific for plasma biomarkers. Further, they used connectivity and Alzheimer’s. Their training set consisted of plasma samples pathway analysis to identify the biological significance of from 43 patients with a previous Alzheimer’s diagnosis (AD) the potential biomarkers, finding that most of them are in- and 40 age- and sex-matched non-demented controls (NDC). volved in signaling pathways related to TNF, TNF or an- Results of the antibody array for each sample revealed sali- giogenesis (personal communication). ent differences in levels of the 120 proteins detected by the arrays. These differences were scored and distilled by cross- CYTOKINE ANTIBODY ARRAYS IN IMMU- NOLOGICAL DISEASE BIOMARKER DISCOVERY 62 Current Proteomics, 2012, Vol. 9, No. 1 Huang et al.

The human immune system is comprised of the innate gastrointestinal tract from mouth to anus. It affects millions and adaptive immune systems, both of which function to of individuals worldwide and causes a wide variety of symp- protect the individual from bacterial, viral, or parasitic toms such as abdominal pain, diarrhea, vomiting, or weight pathogens. The innate immune system comprises physical loss. While the exact cause of CD is unknown, overactive barriers (such as skin and mucosa) and phagocytic cells immune response to environmental and genetic factors or (such as macrophages, granulocytes, dendritic cells, and microorganisms is thought to be the main reason for chronic natural killer cells). Upon activation, this system reacts rap- inflammation. Cytokines and chemokines play an important idly by releasing a variety of inflammatory cytokines. The role in CD [59]. With a homemade 78-cytokine antibody adaptive immune system is mediated through T- or B- array, Kader, et al., noticed that the serum levels of four cy- lymphocytes. Upon activation, T-cells can selectively prolif- tokines (PLGF, IL-7, TGF1, and IL-12p40) were elevated erate into CD8+ cytotoxic effector cells, CD4+ T helper cells in CD patients in clinical remission versus active disease and long-lived memory cells, while B-cells develop into an- [60]. A 23-plex cytokine antibody array was also applied to tibody producing plasma cells. There are many distinct sub- monitor the peripheral immune response in CD patients dur- populations of CD4+ T helper cells with different cytokine ing remission induction therapy with Infliximab, an anti- profiles and distinct effector functions. Th1 cells produce TNF antibody [61]. IFN , IL-2 and TNF and promote cell-mediated immune Cytokine antibody arrays have been reported in many responses involved in tissue damage and infection against publications as a useful tool to study the complexity of im- intracellular parasites. Th2 cells produce IL-4, IL-5, IL-10 munodeficiency diseases. Through analyzing the profile of and IL-13 which are responsible for strong antibody produc- 120 cytokines, chemokines and growth factor in human CSF, tion, eosinophilic inflammation, allergy and clearance of Meeker, et al., showed that an increase of inflammatory cy- helminthic infections. Th17 cells produce IL-17, IL-17F, IL- tokines and chemokines in HIV-infected patients correlated 21 and other inflammatory cytokines and play a role in host poorly with neurological status [62]. However, the severity defense against extracellular pathogens as well as in auto- of neurological disease correlated very well with the decline immune disorders. Cytokine antibody arrays have not only in growth factors, especially neurotrophin-3 (NT-3), indicat- been used to study the gate-keeper cells (dendritic cells) of ing the loss of neuroprotection in HIV-infected patients. the immune system [54], but have also been extensively used in all the three categories of human immunological disor- Antibody arrays have also been used in the cell mediated ders: autoimmunity, immunodeficiency and hypersensitiv- type IV hypersensitivity such as graft-versus-host disease ities. (GVHD) [63]. GVHD is a common complication of alloge- neic bone marrow transplantation in which functional im- Autoimmune diseases affect 3-8% of the world popula- mune cells in the transplanted marrow recognize the recipi- tion, and comprise a wide variety of systemic or organ- ent as "foreign" and mount an immunologic attack. GVHD specific inflammatory diseases characterized by abnormal can also occur with solid tissue transplants if the solid tissue activation of immune cells to target self-tissues. The spec- is rich in lymphatic cells. While no validated biomarkers trum of autoimmune diseases includes rheumatoid arthritis exist for the diagnosis of acute GVHD, Paczesny, et al., re- (RA), systemic lupus erythematosus (SLE), multiple sclero- ported that a panel of 4 serum proteins (soluble IL-2 receptor sis, systemic sclerosis (SS), type 1 diabetes, and psoriasis. (IL-2R), soluble TNF receptor I (TNFRI), IL-8, and he- The precise pathogenesis of most of the autoimmune dis- patocyte growth factor (HGF)), identified by screening 120 eases is still poorly understood, with disease diagnosis and serum proteins with antibody array, were able to confirm the classification relying mainly on clinical examination com- diagnosis of GVHD and provide prognostic information in- bined with traditional laboratory test and imaging studies. dependent of GVHD severity. Inflammatory cytokines and chemokines appear to play a central role [55]. CYTOKINE ANTIBODY ARRAYS IN ASTHMA BIO- SLE is a severe chronic autoimmune connective tissue MARKER DISCOVERY disease, characterized by nuclear autoantibodies, formation Asthma is a disease characterized by remodeling of air- of immune complexes and systemic vasculitis. There are way epithelium, which permanently obstructs airflow and by currently no specific biomarkers for SLE diagnosis, classifi- chronic inflammation of the small and medium airways. This cation, and monitoring of disease activity. Using protein mi- inflammation contributes to hyperresponsive bronchocon- croarray platform, Bauer, et al., identified a group of 30 in- striction upon exposure to irritants or allergens, leading to an terferon-regulated chemokines in the SLE subjects [56]. RA asthma attack. Currently, there is no single serological bio- is characterized by chronic inflammation of the joints and marker used in routine clinical asthma diagnosis and progno- subsequent loss of function. Hueber, et al., reported a panel sis. However, multiple serum proteins such as eosinophil of pro-inflammatory serum cytokines (IL-1 , IL-6, IL-13, cationic protein (ECP), PARC (CCL18), TARC, HC gp-39, IL-15, and TNF ) for early RA detection [57]. Through and fibrinogen have been described as potential biomarkers analysis of synovial fluid from early onset RA with antibody in pulmonary research [64]. Fibroblast growth factors (FGF), arrays, Raza, et al., found that a profile of T-cell derived HGF, and stem cell growth factor beta (SCGF) have also cytokines was significantly elevated when compared with showed promise in a cytokine antibody array study [65]. established RA patients, indicating the an active role of this Because of the importance of inflammation in the patho- cell type in disease pathogenesis [58]. physiology of asthma, several studies have focused on in- Crohn's disease (CD) is a chronic inflammatory bowel flammatory biomarkers of asthma in both the sputum and disease (IBD) of the intestines that may affect any part of the exhaled breath condensates of affected individuals. Cytokine Antibody Arrays in Biomarker Discovery and Validation Current Proteomics, 2012, Vol. 9, No. 1 63

In the example of Kim, et al., the authors used a cytokine Cancer is a disease driven by both genetic and epigenetic antibody array to examine the expression profiles of cytoki- factors, leading to uncontrolled cellular growth, which mani- nes in the sputum of asthma patients compared to non- fests in altered signaling pathways including cell cycle de- asthmatic control patients [66]. They found that, among oth- regulation, inhibition of apoptosis, blockage of differentia- ers, expression of chemokines GRO (CXCL1), Eotaxin-2 tion, enhancement of angiogenesis, invasion and metastasis. (CCL24) and PARC (CCL18) were significantly higher in Crosstalk among these pathways greatly influences the bio- asthma patients. In addition, the authors noted that PARC logical behavior of oncogenesis, i.e., its initiation, promo- levels were highly correlated with the relative number of tion, progression, metastasis and recurrence. Often these eosinophils in sputum samples. Elevated eosinophil count is changes include changes in tumor microenvironment and known to be associated with asthma, so the high correlation resulting local inflammation [14]; changes that may be re- of PARC with eosinophil counts suggested a role for this vealed by alterations in protein profiling in sera and plasma chemokine in the pathogenesis of asthma. of cancer patients. Therefore, evaluating changes in cellular protein profiling and identifying cancer-associated proteins In another investigation of the role of inflammation in between cancer and non-cancer patient samples could pro- asthma, antibody arrays were used to analyze cytokine ex- vide useful information for cancer diagnosis and treatment. pression patterns in exhaled breath condensates from normal The goal for cancer proteomics is to identify cancer-specific and asthmatic individuals [67]. Since asthma is known to be an inflammatory disease, it was no surprise that inflamma- biomarkers for specific cancer types or specific cancer stages or responses for particular treatments. tory markers were expressed at higher levels in asthmatic patients. What was surprising was the correlation of the ex- Since cancer is a heterogeneous disease, it is unlikely that pression of inflammatory markers with patient symptoms. a single biomarker will detect all cancers of a particular or- The authors compared the expression patterns with standard gan with high specificity and sensitivity. For example, pros- measures of patient symptoms, including forced expiratory tate–specific antigen (PSA) is a biomarker for prostate can- volume in 1 second (FEV1) and hyperresponsiveness to irri- cer with high sensitivity yet low specificity. CA-125 (also tant challenge. RANTES (CCL5) expression correlated to known as Mucin 16) is a biomarker for ovarian cancer with FEV1, and TNF and TGF1 expression correlated to airway high sensitivity and low specificity. It is widely agreed that hyperresponsiveness. Moreover, there was a linear correla- high specificity is the key performance issue for a successful tion between the two: the more severe the patient’s symp- cancer screening marker. Due to the prevalence rate of many toms, the greater the concentration of the cytokines detected. cancer types, even a small false positive rate in a screening This same group followed up on this study by proving that biomarker assay will translate a large number of costly and contamination by saliva in exhaled breath condensates was unnecessary diagnostic procedures as well as unwarranted not responsible for the cytokines detected in these experi- stress for the patient. Therefore, rather than using a single ments [68] and extended their investigation of cytokine ex- marker, a multi-marker assay may be better suited for poten- pression in exhaled breath condensates of asthmatic patients tial cancer screening with high sensitivity and specificity. in response to standard therapies. Antibody array technology has emerged as a promising In one of these subsequent investigations, cytokine anti- platform for cancer biomarker discovery due to its unique body arrays were used to compare cytokine profiles of asth- advantages, of high-throughput quantification, low sample matic patients before (at baseline) and after commencement consumption, ease of use, and cost-effectiveness [70, 71]. of corticosteroid therapy [69]. They then performed a statis- Antibody arrays have been reported as cancer biomarker tical analysis of the pre- and post-treatment cytokine expres- discovery platforms for several cancers, including ovarian sion profiles in exhaled breath condensates. Patients with cancer, prostate cancer, and breast cancer. increased levels of IL-4 and RANTES along with decreased Among females in the US, breast cancer represents the levels of IP-10 at baseline were much more likely to respond leading cancer in incidence and the second deadliest cancer. to treatment, as demonstrated by improvement in FEV , a 1 Due to the heterogeneity of breast cancer, the search for measure of airway obstruction. Also, improvement in FEV 1 multi-marker panels, which may lead to more accurate and following steroid therapy correlated well with decreased reliable diagnosis, prognosis and treatment monitoring, has expression levels of IL-4 and RANTES. Taken together, this drawn much attention. There are a number of reports in study identified elevated IL-4 and RANTES as biomarkers which antibody array technology has been used to identify indicative of predicted positive response to steroidal therapy such multi-marker signatures. Yeretssian, et al., identified treatment and reduced levels of IL-4 and RANTES as bio- unique protein expression profiles in breast-cancer derived markers of successful treatment of symptoms by steroidal cell lines [72]. Hudelist, et al., identified a group of bio- therapy. This illustrates two of the important uses of biomarkers by comparing protein expression in malignant and markers in personalized medicine: prediction of response to adjacent normal breast tissue from the same cancer patients treatment and monitoring progress during treatment. In a [73]. Antibody arrays were also used to identify biomarkers separate publication, this same group identified lower ex- for monitoring treatment response. Vazques-Martin, et al., pression of RANTES in exhaled breath condensates in re- reported the identification of an HER2-induced “cytokine sponse to a combination of steroidal and leukotriene modi- signature” in breast cancer. Using cytokine antibody arrays, fier therapy, particularly in patients with comorbid allergic expression of 42 cytokines and growth factors were com- rhinitis [67]. pared in conditioned media from MCF-7/HER2-18 (MCF-7 breast cancer cell line over-expressing HER2) and MCF- CYTOKINE ANTIBODY ARRAYS IN CANCER BIO- 7/Neo control cells. Two biomarkers were identified with MARKER DISCOVERY 64 Current Proteomics, 2012, Vol. 9, No. 1 Huang et al. increased expression in MCF-1/Her2-18 cells (IL-8 and creased expression in both post-vaccination time points. Fur- GRO). The findings may represent novel biomarkers in thermore, using cytokine antibody arrays, a panel of 4 pro- monitoring breast cancer responses to endocrine treatments teins were identified as markers to distinguish vaccine re- and/or HER2-targeted therapy [74]. Taken together, these sponders from non-responders (CXCL11, L-selectin, MCP-1 studies strongly suggest that antibody array technology has and soluble gp130). The results may represent a novel shown great promise in the discovery and development of method for monitoring vaccine treatment efficacy in cancer breast cancer biomarkers. patients. Ovarian cancer represents the third most frequent cancer SUMMARY AND FUTURE PROSPECTS and is one of the leading causes of cancer death among females in the US and Europe. Due to a lack of obvious symp- Application of Cytokine Antibody Arrays in Biomarker toms, almost 80% of ovarian cancer patients are diagnosed at Discovery in Many Diseases later stages. Unfortunately, the 5-year survival rate for patients with clinically advanced ovarian cancer is only 15% to Disease biomarkers are among the most actively re- 20%. By contrast, the 5-year survival rate for patients with searched topics in biomedicine. Examples of molecular bio- stage I disease is over 90%. Therefore, it is of utmost impor- markers include DNA, mRNA, microRNA, proteins and tance to discover and develop biomarkers for ovarian cancer metabolites. Since the completion of the Human Genome screening and early detection. Currently, CA-125 and imag- Project, many researchers have looked for genetic markers ing are the two most common tests for ovarian cancer suitable for diagnosing a disease, assessing a patient’s risk of screening. However, using a cutoff of 30 to 35 units/mL, developing a given disease, or predicting the patient’s re- serum CA-125 has demonstrated a specificity of only 50% to sponse to a specific treatment. Inborn errors in metabolism 60% and a sensitivity of >98% for early-stage ovarian cancer and other rare genetic disorders, such as cystic fibrosis and [75, 76]. Multiple studies have reported identification of ova- sickle-cell anemia, can be definitively diagnosed by the pres- rian cancer biomarkers using antibody array technology. ence a single gene or allele. Using biotin-label–based antibody arrays, we screened the Additionally, protein functions may be influenced by serum expression profiles of 507 proteins in serum samples post-translational processing (such as glycosylation. phos- from 47 patients with ovarian cancers, 33 patients with be- phorylation or proteolytic cleavage) or interactions with nign ovarian masses and 39 healthy, age-matched controls. A other proteins, all of which can be detected using antibody panel of protein expression showed significant difference array technologies. Thus, characterizing the proteome may between normal and cancer (P<0.05). By classification give a more realistic view of the biological status of the pa- analysis and split-point score analysis of these two groups, a tient than genomic markers or gene expression [78]. six-marker panel of proteins (IL-2R, endothelin, osteopro- tegerin, vascular endothelial growth factor D (VEGF-D) and Although protein biomarkers can be found in many dif- betacellulin (BTC)) could be used to distinguish ovarian ferent types of patient specimens, blood samples have great- cancer patients from normal subjects [33]. Those studies est potential for discovery of diagnostic and predictive bio- strongly suggest that antibody array technology has great markers. Blood perfuse all tissues of the body and carries not promise in the discovery and development of ovarian cancer only plasma-specific proteins but also proteins derived from biomarkers. other tissues. The proteins circulating in the blood are in a state of dynamic change, reflecting the health status of the Prostate cancer is the most common cancer, with the sec- individual. By contrast, an individual’s genome is stable and ond highest mortality rate of any cancer in the US. Strategies unchanging. Thus, changes in proteins (and perhaps metabo- to inhibit prostate cancer metastasis include targeting both lites) may provide more valuable information for patient’s tumor-induced osteoblastic lesions and underlying osteoclas- treatment than genomic profiles. tic activities. To identify the mechanism by which prostate cancer induces osteoclastic activity, Lu, et al., profiled cyto- Since the first demonstration that cytokine antibody ar- kine expression in conditioned media from primary prostate rays could detect proteins in complex biological samples epithelial cells, prostate cancer cells (LNCaP) and its deriva- with high sensitivity and specificity [79], cytokine antibody tives (C4-2B and PC3 cells) with an antibody array detecting arrays have been widely used in biomarker discovery and expression 174 cytokines. A panel of cytokine markers with development programs, including the application of bio- increased expression in cancer and cancer derived cells was markers as drug targets, or as predictive, diagnostic, or prog- identified (MCP-1, IL-6, IL-8, GRO, ENA-78, CXCL16). nostic tools. Cytokine antibody arrays have been success- The results may provide novel therapeutic targets for treat- fully used with many different biological sample sources, ment of prostate-induced bone metastasis. including serum [80-83], urine [84-86], plasma [87, 88], bronchoalveolar lavage [89], gingival crevicular fluid [90], In order to find biomarkers for monitoring cancer vaccine synovial fibroblast supernatants [91], CSF [92], sputum [93]; efficacy, Toh, et al., measured expression profiles of 507 [94], tears [95], laser capture-microdissected tissue [96], cell proteins using biotin-label-based antibody array in cultures lysates [97], tissue lysates [98], conditioned medium [99], of dendritic cells obtained as from colorectal cancer patients, liver cyst fluids [100], saliva [101], breast interstitial fluid both before vaccination and at 2 months and 4 months post- [102] and many others. vaccination [77]. When samples were compared from post- vaccination (month 2 and 4) and pre-vaccination, 58 and 47 In addition to those diseases and applications mentioned differentially expressed proteins were identified, respec- elsewhere in this review, cytokine antibody arrays have also tively. Among them, 28 proteins showed consistent in- been used to identify biomarkers for chronic obstructive pulmonary disease [93, 94], transplantation [86], endome- Cytokine Antibody Arrays in Biomarker Discovery and Validation Current Proteomics, 2012, Vol. 9, No. 1 65 triosis [104], irritable bowel syndrome [105], periodontal label-free technologies including biosensors of carbon diseases [106], chronic kidney disease [103], cystic fibrosis nanowires and nanotubes, which detect changes in conduc- [82], ulcerative colitis [83], renal systemic lupus [87], re- tivity on the surface upon binding, can also be used in pro- sponse to anti-retroviral therapy for AIDS [88], many other tein array detection. types of cancers [107-110] and many other diseases [111]. Without the interference of tagging agents, label-free multiplexed protein arrays will ensure detection of proteins Biomarkers in Clinical Applications with native conformations and functionalities. Furthermore, Recently, NIH has made biomarker discovery and valida- label-free technologies will provide quantitative data such as tion a cornerstone of its support of translational research, affinity constants and kinetic parameters, among others. To meant to bridge the gap between bench science and clinical date, most studies of label-free detection technology have applications, and it has devoted considerable resources to made use of ideal model systems for proof-of-concept ex- support it. However, the translation of biomarkers into more periments. However, as detection sensitivities improve and effective patient care and better outcomes remains a chal- equipment costs are reduced, this trend should gradually lenge. There are still many obstacles to developing clinically change in the years to come and label-free techniques may useful biomarkers, including technical challenges associated be used for the detection and study of more general biologi- with pre-analytical variables and validating potential bio- cal interactions. markers in larger and unbiased clinical tests. The ability to Since in label-based or label-free systems only one anti- simultaneously detect multiple proteins greatly facilitates the body is used, specificity of the antibodies employed is para- biomarker discovery progress. Indeed, the discovery of bio- mount, and cross-reactivity of antibodies to other proteins markers using the antibody array technology has increased tremendously in recent years. becomes a greater concern than with sandwich-based detection. Therefore, the development of systems suitable for con- One of major advantage of antibody array-based bio- clusive validation of biomarkers will certainly increase the marker discovery and development is the seamless transition reliability of assay. MS-based antibody arrays represent one of biomarker discovery to validation. Many biomarker vali- possible solution. The limitations of MS-based approach are dation processes are performed by industry and may not be low sensitivity and difficulty in quantification. To overcome published in the literature. An unknown percentage of bio- this problem, a surface plasmon resonance coupled to MS markers fail in the validation stage; therefore, it is difficult to (SPR-MS) array platform has been developed. In this format, estimate how many biomarkers have been validated after the antibody arrays are used to capture the corresponding targets discovery phase. Furthermore, the FDA has approved very from biological samples. The nature of captured protein is few biomarkers for clinical use. Diagnostics employing a further validated by MS and quantified by SPR [123, 124]. single biomarker may lack the specificity needed for FDA Such MS-based antibody arrays offer dual specificity, which approval. However, the use of multiple biomarkers in a panel is resulted from by the affinity of antibody and the protein may improve the overall positive predictive value. Pharma- mass readout. Furthermore, the MS aspect of the analysis ceutical companies are increasingly interested in developing provides simultaneous detection of protein isoforms, modifi- companion biomarker tests to pre-screen patients to predict cations and interaction in a single step. The combination of whether a given patient will respond to a given therapeutic (personalized medical) or to monitor the patient’s response high-content and high-throughput of antibody arrays, label- to treatment. free and real time detection of SPR and protein-specific mass spectra of MS provides an ideal system for simultaneous With the advent of new and improved protein chip tech- quantitative measurement of multiple protein levels, differ- nologies, it is foreseeable that the application of protein ar- ent isoforms of a protein and protein modifications with high rays in biomarker program will accelerate translation of specificity. Although at this moment, the detection sensitiv- biomarkers into clinical practice. ity of this approach is still much lower than label-based approach and sandwich-based approach, and sophisticated Array Detection Technology equipment is needed. As the advance of technology, the The increased use of protein and antibody arrays (includ- combined approach represents the future of profiling of mul- ing cytokine antibody arrays) in biomarker discovery pro- tiple proteins, particularly in the biomarker discovery phase. grams has spurred the advancement of various detection technologies. With improved sensitivities now being New Technologies on the Horizon achieved by sandwich-based and label-based techniques, they continue to be the preferred method of detection in cy- One way to speed up biomarker discovery and validation tokine antibody arrays. Because paired antibodies are re- would be to develop microfluidic devices, which would required in sandwich-based detection technology and labeling quire only a few microliters of samples. Moreover, microflu- agents or tags can introduce artifacts in label-based detection idic techniques are amenable to automated testing, allowing technology, (masking epitopes or altering protein functional- for high-throughput applications. Dr. Heath’s group have ity), the use of label-free detection in antibody arrays is gain- developed a platform which can detect a dozen cytokines ing attention. Many new label-free technologies are emerg- with high specificity and sensitivity [125]. Their system has ing in this field [112-114], including surface plasmon reso- demonstrated detection of cytokine expression in a single nance (SPR) [115], planar waveguide technology [116], elec- cell. Due to the rapidity of the assay and its low sample con- trochemical detection [117], quartz crystal microbalance sumption, microfluidic devices may have applications in (QCM) [118-120] and microcantilevers [121, 122]. Other clinical diagnosis and managing patients’ care. 66 Current Proteomics, 2012, Vol. 9, No. 1 Huang et al.

Recently, Dr. Wang at Stanford University invented an CCL = CC chemokine ligand innovative magnetic nanosensor detection technology to CD = Crohn’s disease measure multiple serum proteins including cytokine from diverse biological samples [126]. In this platform an array of CML = Chronic myelogenous leukemia giant magnetoresistive (GMR) sensors is used to detect pro- CNTF = Ciliary neurotrophic factor tein binding events by surface-bound antibodies. In the sandwich-based format, the capture antibody is bound to the CSF = Cerebrospinal fluid sensor and the detection antibody is tagged with a superpar- CT-1 = Cardiotropin-1 amagnetic nanoparticle. Under an external magnetic field, the nanoparticles become magnetized and can be detected by CXCL = CXC chemokine ligand the underlying GMR sensor. As demonstrated by the authors, ECP = Eosinophil cationic protein this platform offers several important advantages over the previous technologies. It is insensitive to matrix effects due EGFR = Epithelial growth factor receptor to heterogeneity in ionic strength, pH, temperature and auto- ELISA = Enzyme-linked immunosorbent assay fluorescence. Protein concentrations as low as 50 attomolar -18 ENA-78 = Epithelial-derived neutrophil-activating (10 ) can be detected, making it at least 1000 times more Peptide 78 sensitive than ELISA. The authors claimed that magnetic nanosensors are at least comparable to the most sensitive eNet = Elastic net regression algorithm biosensors. Furthermore, it can achieve detection dynamic FEV1 = Forced expiratory volume in 1 second ranges over six orders of magnitude. FGF = Fibroblast growth factor Current technologies of fabrication of protein and antibody arrays produce typical spot sizes of 100-200 microns GI = Gastrointestinal (μm). Thus, applications of these are limited to sample vol- GLP1 = Glucagon-like peptide 1 umes of at least 50-100 μL. To reduce the volumes of reagents and samples used in these arrays and incubation times, GMR = Giant magnetoresistive many new technologies are under development to fabricate gp130 = Glycoprotein 130 the arrays with very small spot sizes (usually from 50 nm to 400 nm). Among them are dip pen lithography [127], por- GRO = Growth-regulated oncogene phyrin-based photocatalytic lithography [128], imprint li- GVHD = Graft versus host disease thography [129], block co-polymer formation [130]. HC gp-39 = Human cartilage glycoprotein-39 (HC gp- In summary, cytokine antibody arrays have been emerg- 39) ing as an effective tool in biomarker discovery. As the technology matures, more biomarkers will be discovered and HGF = Hepatocyte growth factor validated using cytokine antibody arrays. HER2 = Heregulin-2 (ErbB2)

ACKNOWLEDGEMENTS HT = High-throughput IBD = Inflammatory bowel disease This work was supported by NIH/NCI grant 1R41HD065360-01 (RPH). We would like to express our IFN = Interferon thanks for the support of the leading scientist project for IL = Interleukin Guangzhou economic development district (2009L-P180), Guangzhou leading talent entrepreneurial venture IL-2R = Interleukin 2 receptor alpha (LCY201111), Guangdong innovative research and devel- IP-10 = Interferon gamma-induced protein 10 opment team (201001s0104659419) and research grants from the Guangzhou economic development district LIF = Leukemia inhibitory factor (2010Q—P450). MAb = Monoclonal antibody CONFLICT OF INTEREST MCF-7 = Michigan Cancer Foundation 7 (breast cancer cell line) None declared. MCI = Mild cognitive impairment ABBREVIATIONS MCP-1 = Monocyte chemotactic protein 1 AIDS = Acquired immunodeficiency syndrome M-CSF = Macrophage colony stimulating factor AD = Alzheimer’s disease MS = Mass spectrometry ApoE = Apo-lipoprotein E NSCLC = Non-small–cell lung cancer APP = Amyloid precursor protein NDC = Non-demented controls BMP = Bone morphogenetic protein NT-3 = Neurotrophin-3 BTC = Betacellulin OD = Other types of dementia CA-125 = Carbohydrate/cancer antigen 125 OSM = Oncostatin M Cytokine Antibody Arrays in Biomarker Discovery and Validation Current Proteomics, 2012, Vol. 9, No. 1 67

PAM = Prediction Analysis of Microarray [10] Schluns, K.S. and Lefrancois, L. Cytokine control of memory T- cell development and survival. Nat. Rev. Immunol., 2003, 3, 269- PARC = Pulmonary and activation-regulated 279. Chemokine [11] Parrish-Novak, J.; Dillon, S.R.; Nelson, A.; Hammond, A.; Spre- cher, C.; Gross, J.A.; Johnston, J.; Madden, K.; Xu, W.; West, J.; PDGFR = Platelet-derived growth factor receptor Schrader, S.; Burkhead, S.; Heipel, M.; Brandt, C.; Kuijper, J.L.; Kramer, J.; Conklin, D.; Presnell, S.R.; Berry, J.; Shiota, F.; Bort, PLGF = Placental growth factor S.; Hambly, K.; Mudri, S.; Clegg, C.; Moore, M.; Grant, F.J.; Lof- PSA = Prostate–specific antigen ton-Day, C.; Gilbert, T.; Rayond, F.; Ching, A.; Yao, L.; Smith, D.; Webster, P.; Whitmore, T.; Maurer, M.; Kaushansky, K.; Holly, QCM = Quartz crystal microbalance R.D. and Foster, D. Interleukin 21 and its receptor are involved in NK cell expansion and regulation of lymphocyte function. Nature, RA = Rheumatoid arthritis 2000, 408, 57-63. [12] Majka, M.; Janowska-Wieczorek, A.; Ratajczak, J.; Ehrenman, K.; RANTES = Regulated upon activation, normal T-cell Pietrzkowski, Z.; Kowalska, M.A.; Gewirtz, A.M.; Emerson, S.G. Expressed, and secreted and Ratajczak, M.Z. Numerous growth factors, cytokines, and chemokines are secreted by human CD34(+) cells, myeloblasts, RCC = Renal cell carcinoma erythroblasts, and megakaryoblasts and regulate normal hematopoiesis in an autocrine/paracrine manner. Blood, 2001, 97, 3075- SAM = Significance Analysis of Microarray 3085. SCGF = Stem cell growth factor-beta [13] Goldring, M.B. and Goldring, S.R. Cytokines and cell growth control. Crit. Rev. Eukaryot. Gene Expr., 1991, 1, 301-326. SLE = Systemic lupus erythematosus [14] Mantovani, A.; Allavena, P.; Sica, A. and Balkwill, F. Cancer- related inflammation. Nature, 2008, 454, 436-444. SPR = Surface plasmon resonance [15] Joyce, J.A. and Pollard, J.W. Microenvironmental regulation of metastasis. Nat. Rev. Cancer, 2009, 9, 239-252. SS = Systemic sclerosis [16] Akiyama, H.; Barger, S.; Barnum, S.; Bradt, B.; Bauer, J.; Cole, TARC = Thymus and activation-regulated Chemo- G.M.; Cooper, N.R.; Eikelenboom, P.; Emmerling, M.; Fiebich, B.L.; Finch, C.E.; Frautschy, S.; Griffin, W.S.; Hampel, H.; Hull, kine M.; Landreth, G.; Lue, L.; Mrak, R.; Mackenzie, I.R.; McGeer, TGF1 = Transforming growth factor beta-1 P.L.; O'Banion, M.K.; Pachter, J.; Pasinetti, G.; Plata-Salaman, C.; Rogers, J.; Rydel, R.; Shen, Y.; Streit, W.; Strohmeyer, R.; Too- TLR7 = Toll-like receptor 7 yoma, I.; Van Muiswinkel, F.L.; Veerhuis, R.; Walker, D.; Web- ster, S.; Wegrzyniak, B.; Wenk, G. and Wyss-Coray, T. Inflamma- TNF = Tumor necrosis factor tion and Alzheimer's disease. Neurobiol. Aging, 2000, 21, 383-421. [17] Hirsch, E.C. and Hunot, S. Neuroinflammation in Parkinson's dis- TNFRI = Tumor necrosis factor receptor 1 ease: a target for neuroprotection? Lancet Neurol., 2009, 8, 382- 397. VEGF = Vascular endothelial growth factor [18] Wellen, K.E. and Hotamisligil, G.S. Inflammation, stress, and VEGF-D = Vascular endothelial growth factor diabetes. J. Clin. Invest., 2005, 115, 1111-1119. [19] Ross, J.S. and Fletcher, J.A. The HER-2/neu oncogene in breast VEGFR = Vascular endothelial growth factor receptor cancer: prognostic factor, predictive factor, and target for therapy. Stem Cells, 1998, 16, 413-428. REFERENCES [20] Mendelsohn, J. and Baselga, J. Status of epidermal growth factor receptor antagonists in the biology and treatment of cancer. J. Clin. [1] Hansson, G.K. Inflammation, atherosclerosis, and coronary artery Oncol., 2003, 21, 2787-2799. disease. N. Engl. J. Med., 2005, 352, 1685-1695. [21] Feldmann, M. and Maini, R.N. Lasker Clinical Medical Research [2] Trinchieri, G. Interleukin-12: a proinflammatory cytokine with Award. TNF defined as a therapeutic target for rheumatoid arthritis immunoregulatory functions that bridge innate resistance and anti- and other autoimmune diseases. Nat. Med., 2003, 9, 1245-1250. gen-specific adaptive immunity. Annu. Rev. Immunol., 1995, 13, [22] Hoofnagle, J.H. Therapy of acute and chronic viral hepatitis. Adv. 251-276. Intern. Med., 1994, 39, 241-275. [3] Van, P.L. and Abbas, A.K. Homeostasis and self-tolerance in the [23] Lane, H.C.; Kovacs, J.A.; Feinberg, J.; Herpin, B.; Davey, V.; immune system: turning lymphocytes off. Science, 1998, 280, 243- Walker, R.; Deyton, L.; Metcalf, J.A.; Baseler, M. and Salzman, N. 248. Anti-retroviral effects of interferon-alpha in AIDS-associated Ka- [4] Moser, B.; Wolf, M.; Walz, A. and Loetscher, P. Chemokines: posi's sarcoma. Lancet, 1988, 2, 1218-1222. multiple levels of leukocyte migration control. Trends Immunol., [24] Goodin, D.S.; Frohman, E.M.; Garmany, G.P.Jr.; Halper, J.; Li- 2004, 25, 75-84. kosky, W.H.; Lublin, F.D.; Silberberg, D.H.; Stuart, W.H. and van [5] Engelhardt, E.; Toksoy, A.; Goebeler, M.; Debus, S.; Brocker, E.B. den, N.S. Disease modifying therapies in multiple sclerosis: report and Gillitzer, R. Chemokines IL-8, GROalpha, MCP-1, IP-10, and of the Therapeutics and Technology Assessment Subcommittee of Mig are sequentially and differentially expressed during phase- the American Academy of Neurology and the MS Council for specific infiltration of leukocyte subsets in human wound healing. Clinical Practice Guidelines. Neurology, 2002, 58, 169-178. Am. J. Pathol., 1998, 153, 1849-1860. [25] Ozaki, K. and Leonard, W.J. Cytokine and cytokine receptor plei- [6] Belperio, J.A.; Keane, M.P.; Arenberg, D.A.; Addison, C.L.; otropy and redundancy. J. Biol. Chem., 2002, 277, 29355-29358. Ehlert, J.E.; Burdick, M.D. and Strieter, R.M. CXC chemokines in [26] Hirano, T.; Taga, T.; Nakano, N.; Yasukawa, K.; Kashiwamura, S.; angiogenesis. J. Leukoc. Biol., 2000, 68, 1-8. Shimizu, K.; Nakajima, K.; Pyun, K.H. and Kishimoto, T. Purifica- [7] MacDougald, O.A. and Mandrup, S. Adipogenesis: forces that tip tion to homogeneity and characterization of human B-cell differen- the scales. Trends Endocrinol. Metab., 2002, 13, 5-11. tiation factor (BCDF or BSFp-2). Proc. Natl. Acad. Sci. USA, 1985, [8] Janes, K.A.; Albeck, J.G.; Gaudet, S.; Sorger, P.K.; Lauffenburger, 82, 5490-5494. D.A. and Yaffe, M.B. A systems model of signaling identifies a [27] Kishimoto, T.; Akira, S.; Narazaki, M. and Taga, T. Interleukin-6 molecular basis set for cytokine-induced apoptosis. Science, 2005, family of cytokines and gp130. Blood, 1995, 86, 1243-1254. 310, 1646-1653. [28] Audet, J. Adventures in time and space: Nonlinearity and complex- [9] Werner, S. and Grose, R. Regulation of wound healing by growth ity of cytokine effects on stem cell fate decisions. Biotechnol. Bio- factors and cytokines. Physiol. Rev., 2003, 83, 835-870. eng., 2010, 106, 173-182. [29] Snijders, A.; Kalinski, P.; Hilkens, C.M. and Kapsenberg, M.L. High-level IL-12 production by human dendritic cells requires two signals. Int. Immunol., 1998, 10, 1593-1598. 68 Current Proteomics, 2012, Vol. 9, No. 1 Huang et al.

[30] Heaney, M.L. and Golde, D.W. Soluble cytokine receptors. Blood, [52] Vargas, D.L.; Nascimbene, C.; Krishnan, C.; Zimmerman, A.W. 1996, 87, 847-857. and Pardo, C.A. Neuroglial activation and neuroinflammation in [31] Hsu, H.Y.; Wittemann, S. and Joos, T.O. Miniaturized parallelized the brain of patients with autism. Ann. Neurol., 2004, 57, 67-81. sandwich immunoassays. Methods Mol. Biol., 2008, 428, 247-261. [53] Ray, S.; Britschgi, M.; Herbert, C.; Takeda-Uchimura, Y.; Boxer, [32] Kusnezow, W.; Banzon, V.; Schroder, C.; Schaal, R.; Hoheisel, A.; Blennow, K.; Friedman, L.F.; Galasko, D.R.; Jutel, M.; Kary- J.D.; Ruffer, S.; Luft, P.; Duschl, A. and Syagailo, Y.V. Antibody das, A.; Kaye, J.A.; Leszek, J.; Miller, B.L.; Minthon, L.; Quinn, microarray-based profiling of complex specimens: systematic J.F.; Rabinovici, G.D.; Robinson, W.H.; Sabbagh, M.N.; So, Y.T.; evaluation of labeling strategies. Proteomics, 2007, 7, 1786-1799. Sparks, D.L.; Tabaton, M.; Tinklenberg, J.; Yesavage, J.A.; Tibshi- [33] Huang, R.; Jiang, W.; Yang, J.; Mao, Y.Q.; Zhang, Y.; Yang, W.; rani, R. and Wyss-Coray, T. Classification and prediction of clini- Yang, D.; Burkholder, B.; Huang, R.F. and Huang, R.P. A biotin cal Alzheimer's diagnosis based on plasma signaling proteins. Nat. label-based antibody array for high-content profiling of protein ex- Med., 2007, 13, 1359-1362. pression. Cancer Genom. Proteomics, 2010, 7, 129-141. [54] Jensen, S.S. and Gad, M. Differential induction of inflammatory [34] Ray, S.; Britschgi, M.; Herbert, C.; Takeda-Uchimura, Y.; Boxer, cytokines by dendritic cells treated with novel TLR-agonist and cy- A.; Blennow, K.; Friedman, L.F.; Galasko, D.R.; Jutel, M.; Kary- tokine based cocktails: targeting dendritic cells in autoimmunity. J. das, A.; Kaye, J.A.; Leszek, J.; Miller, B.L.; Minthon, L.; Quinn, Inflamm.(Lond)., 2010, 7, 37. J.F.; Rabinovici, G.D.; Robinson, W.H.; Sabbagh, M.N.; So, Y.T.; [55] Kunz, M. and Ibrahim, S.M. Cytokines and cytokine profiles in Sparks, D.L.; Tabaton, M.; Tinklenberg, J.; Yesavage, J.A.; Tibshi- human autoimmune diseases and animal models of autoimmunity. rani, R. and Wyss-Coray, T. Classification and prediction of clini- Mediators. Inflamm., 2009, 2009, 979258. cal Alzheimer's diagnosis based on plasma signaling proteins. Nat. [56] Bauer, J.W.; Baechler, E.C.; Petri, M.; Batliwalla, F.M.; Crawford, Med., 2007, 13, 1359-1362. D.; Ortmann, W.A.; Espe, K.J.; Li, W.; Patel, D.D.; Gregersen, [35] Spisak, S.; Tulassay, Z.; Molnar, B. and Guttman, A. Protein mi- P.K. and Behrens, T.W. Elevated serum levels of interferon- crochips in biomedicine and biomarker discovery. Electrophoresis, regulated chemokines are biomarkers for active human systemic 2007, 28, 4261-4273. lupus erythematosus. PLoS. Med., 2006, 3, e491. [36] Chandra, H. and Reddy, P.J. and Srivastava, S. Protein microarrays [57] Hueber, W.; Tomooka, B.H.; Zhao, X.; Kidd, B.A.; Drijfhout, and novel detection platforms. Expert. Rev. Proteomics, 2011, 8, J.W.; Fries, J.F.; van Venrooij, W.J.; Metzger, A.L.; Genovese, 61-79. M.C. and Robinson, W.H. Proteomic analysis of secreted proteins [37] Wingren, C. and Borrebaeck, C.A. Antibody-based microarrays. in early rheumatoid arthritis: anti-citrulline autoreactivity is associ- Methods Mol. Biol., 2009, 509, 57-84. ated with up regulation of proinflammatory cytokines. Ann. Rheum. [38] Chandra, H. and Reddy, P.J. and Srivastava, S. Protein microarrays Dis., 2007, 66, 712-719. and novel detection platforms. Expert. Rev. Proteomics, 2011, 8, [58] Raza, K.; Falciani, F.; Curnow, S.J.; Ross, E.J.; Lee, C.Y.; Akbar, 61-79. A.N.; Lord, J.M.; Gordon, C.; Buckley, C.D. and Salmon, M. Early [39] Kingsmore, S.F. Multiplexed protein measurement: technologies rheumatoid arthritis is characterized by a distinct and transient and applications of protein and antibody arrays. Nat. Rev. Drug synovial fluid cytokine profile of T cell and stromal cell origin. Ar- Discov., 2006, 5, 310-320. thritis Res. Ther., 2005, 7, R784-R795. [40] Borrebaeck, C.A. and Wingren, C. High-throughput proteomics [59] Fiorino, G.; Rovida, S.; Correale, C.; Malesci, A. and Danese, S. using antibody microarrays: an update. Expert. Rev. Mol. Diagn., Emerging biologics in the treatment of inflammatory bowel dis- 2007, 7, 673-686. ease: what is around the corner? Curr. Drug Targets, 2010, 11, [41] Alhamdani, M.S.; Schroder, C. and Hoheisel, J.D. Oncoproteomic 249-260. profiling with antibody microarrays. Genome Med., 2009, 1, 68. [60] Kader, H.A.; Tchernev, V.T.; Satyaraj, E.; Lejnine, S.; Kotler, G.; [42] Ayoglu, B.; Haggmark, A.; Neiman, M.; Igel, U.; Uhlen, M.; Kingsmore, S.F. and Patel, D.D. Protein microarray analysis of dis- Schwenk, J.M. and Nilsson, P. Systematic antibody and antigen- ease activity in pediatric inflammatory bowel disease demonstrates based proteomic profiling with microarrays. Expert. Rev. Mol. Di- elevated serum PLGF, IL-7, TGF-beta1, and IL-12p40 levels in agn., 2011, 11, 219-234. Crohn's disease and ulcerative colitis patients in remission versus [43] Borrebaeck, C.A. and Wingren, C. Design of high-density antibody active disease. Am. J. Gastroenterol., 2005, 100, 414-423. microarrays for disease proteomics: Key technological issues. J. [61] Kato, K.; Fukunaga, K.; Kamikozuru, K.; Kashiwamura, S.; Hida, Proteomics, 2009, 72, 928-935. N.; Ohda, Y.; Takeda, N.; Yoshida, K.; Iimuro, M.; Yokoyama, Y.; [44] Dexlin, L.; Ingvarsson, J.; Frendeus, B.; Borrebaeck, C.A. and Kikuyama, R.; Miwa, H. and Matsumoto, T. Infliximab therapy Wingren, C. Design of recombinant antibody microarrays for cell impacts the peripheral immune system of immunomodulator and surface membrane proteomics. J. Proteome. Res., 2008, 7, 319-327. corticosteroid naive patients with Crohn's disease. Gut. Liver, 2011, [45] Kwong, K.Y.; Baskar, S.; Zhang, H.; Mackall, C.L. and Rader, C. 5, 37-45. Generation, affinity maturation, and characterization of a human [62] Meeker, R.B.; Poulton, W.; Markovic-Plese, S.; Hall, C. and Rob- anti-human NKG2D monoclonal antibody with dual antagonistic ertson, K. Protein changes in CSF of HIV-infected patients: evi- and agonistic activity. J. Mol. Biol., 2008, 384, 1143-1156. dence for loss of neuroprotection. J. Neurovirol., 2011, 17, 258- [46] Pansri, P.; Jaruseranee, N.; Rangnoi, K.; Kristensen, P. and Ya- 273. mabhai, M. A compact phage display human scFv library for selec- [63] Paczesny, S.; Krijanovski, O.I.; Braun, T.M.; Choi, S.W.; Clouth- tion of antibodies to a wide variety of antigens. BMC Biotechnol., ier, S.G.; Kuick, R.; Misek, D.E.; Cooke, K.R.; Kitko, C.L.; Wey- 2009, 9, 6. and, A.; Bickley, D.; Jones, D.; Whitfield, J.; Reddy, P.; Levine, [47] Anderson, N.L. and Anderson, N.G. The human plasma proteome: J.E.; Hanash, S.M. and Ferrara, J.L. A biomarker panel for acute history, character, and diagnostic prospects. Mol. Cell Proteomics, graft-versus-host disease. Blood, 2009, 113, 273-278. 2002, 1, 845-867. [64] Wadsworth, S.; Sin, D. and Dorscheid, D. Clinical update on the [48] Hotta, K.; Funahashi, T.; Arita, Y.; Takahashi, M.; Matsuda, M.; use of biomarkers of airway inflammation in the management of Okamoto, Y.; Iwahashi, H.; Kuriyama, H.; Ouchi, N.; Maeda, K.; asthma. J. Asthma Allergy, 2011, 4, 77-86. Nishida, M.; Kihara, S.; Sakai, N.; Nakajima, T.; Hasegawa, K.; [65] Patil, S.P.; Wisnivesky, J.P.; Busse, P.J.; Halm, E.A. and Li, X.M. Muraguchi, M.; Ohmoto, Y.; Nakamura, T.; Yamashita, S.; Hana- Detection of immunological biomarkers correlated with asthma fusa, T. and Matsuzawa, Y. Plasma concentrations of a novel, adi- control and quality of life measurements in sera from chronic asth- pose-specific protein, adiponectin, in type 2 diabetic patients. Arte- matic patients. Ann. Allergy Asthma Immunol., 2011, 106, 205-213. rioscler. Thromb. Vasc. Biol., 2000, 20, 1595-1599. [66] Kim, H.B.; Kim, C.K.; Iijima, K.; Kobayashi, T. and Kita, H. Pro- [49] Chandra, H.; Reddy, P.J. and Srivastava, S. Protein microarrays tein microarray analysis in patients with asthma: elevation of the and novel detection platforms. Expert. Rev. Proteomics, 2011, 8, chemokine PARC/CCL18 in sputum. Chest, 2009, 135, 295-302. 61-79. [67] Matsunaga, K.; Yanagisawa, S.; Ichikawa, T.; Ueshima, K.; Aka- [50] Ray, S.; Mehta, G. and Srivastava, S. Label-free detection tech- matsu, K.; Hirano, T.; Nakanishi, M.; Yamagata, T.; Minakata, Y. niques for protein microarrays: prospects, merits and challenges. and Ichinose, M. Airway cytokine expression measured by means Proteomics, 2010, 10, 731-748. of protein array in exhaled breath condensate: correlation with [51] MacBeath, G. and Schreiber, S.L. Printing proteins as microarrays physiologic properties in asthmatic patients. J. Allergy Clin. Immu- for high-throughput function determination [see comments]. Sci- nol., 2006, 118, 84-90. ence, 2000, 289, 1760-1763. Cytokine Antibody Arrays in Biomarker Discovery and Validation Current Proteomics, 2012, Vol. 9, No. 1 69

[68] Ichikawa, T.; Matsunaga, K.; Minakata, Y.; Yanagisawa, S.; [86] Hu, H.; Kwun, J.; Aizenstein, B.D. and Knechtle, S.J.; Noninvasive Ueshima, K.; Akamatsu, K.; Hirano, T.; Nakanishi, M.; Sugiura, detection of acute and chronic injuries in human renal transplant by H.; Yamagata, T. and Ichinose, M. Possible impact of salivary in- elevation of multiple cytokines/chemokines in urine. Transplanta- fluence on cytokine analysis in exhaled breath condensate. Anal. tion, 2009, 87, 1814-1820. Chem. Insights, 2007, 2, 85-92. [87] Paczesny, S.; Krijanovski, O.I.; Braun, T.M.; Choi, S.W.; Clouth- [69] Matsunaga, K.; Yanagisawa, S.; Ichikawa, T.; Akamatsu, K.; ier, S.G.; Kuick, R.; Misek, D.E.; Cooke, K.R.; Kitko, C.L.; Wey- Koarai, A.; Hirano, T.; Sugiura, H.; Minakata, Y. and Ichinose, M. and, A.; Bickley, D.; Jones, D.; Whitfield, J.; Reddy, P.; Levine, Molecular mechanism of the additive effects of leukotriene modi- J.E.; Hanash, S.M. and Ferrara, J.L. A biomarker panel for acute fier in asthmatic patients receiving steroid therapy. Allergol. Int., graft-versus-host disease. Blood, 2009, 113, 273-278. 2009, 58, 89-96. [88] Wu, J.Q.; Wang, B.; Belov, L.; Chrisp, J.; Learmont, J.; Dyer, [70] Huang, R.P. An array of possibilities in cancer research using cyto- W.B.; Zaunders, J.; Cunningham, A.L.; Dwyer, D.E. and Saksena, kine antibody arrays. Expert. Rev. Proteomics, 2007, 4, 299-308. N.K. Antibody microarray analysis of cell surface antigens on [71] Huang, R.P.; Yang, W.; Yang, D.; Flowers, L.; Horowitz, I.R.; CD4+ and CD8+ T cells from HIV+ individuals correlates with Cao, X. and Huang, R. The promise of cytokine antibody arrays in disease stages. Retrovirology, 2007, 4, 83. the drug discovery process. Expert. Opin. Ther. Targets, 2005, 9, [89] Bogatkevich, G.S.; Ludwicka-Bradley, A.; Highland, K.B.; Hant, 601-615. F.; Nietert, P.J.; Singleton, C.B.; Feghali-Bostwick, C.A. and Sil- [72] Yeretssian, G.; Lecocq, M.; Lebon, G.; Hurst, H.C. and Sakanyan, ver, R.M. Impairment of the antifibrotic effect of hepatocyte V. Competition on nitrocellulose-immobilized antibody arrays: growth factor in lung fibroblasts from African Americans: possible from bacterial protein binding assay to protein profiling in breast role in systemic sclerosis. Arthritis Rheum., 2007, 56, 2432-2442. cancer cells. Mol. Cell Proteomics, 2005, 4, 605-617. [90] Sakai, A.; Ohshima, M.; Sugano, N.; Otsuka, K. and Ito, K. Profil- [73] Hudelist, G.; Pacher-Zavisin, M.; Singer, C.F.; Holper, T.; Kubista, ing the cytokines in gingival crevicular fluid using a cytokine anti- E.; Schreiber, M.; Manavi, M.; Bilban, M. and Czerwenka, K. Use body array. J. Periodontol., 2006, 77, 856-864. of high-throughput protein array for profiling of differentially ex- [91] Andreas, K.; Lubke, C.; Haupl, T.; Dehne, T.; Morawietz, L.; pressed proteins in normal and malignant breast tissue. Breast Can- Ringe, J.; Kaps, C. and Sittinger, M. Key regulatory molecules of cer Res. Treat., 2004, 86, 281-291. cartilage destruction in rheumatoid arthritis: an in vitro study. Ar- [74] Vazquez-Martin, A.; Colomer, R. and Menendez, J.A. Protein array thritis Res. Ther., 2008, 10, R9. technology to detect HER2 (erbB-2)-induced 'cytokine signature' in [92] Meeker, R.B.; Poulton, W.; Markovic-Plese, S.; Hall, C. and Rob- breast cancer. Eur. J. Cancer, 2007, 43, 1117-1124. ertson, K. Protein changes in CSF of HIV-infected patients: evi- [75] Lowe, K.A.; Shah, C.; Wallace, E.; Anderson, G.; Paley, P.; McIn- dence for loss of neuroprotection. J. Neurovirol., 2011, 17, 258- tosh, M.; Andersen, M.R.; Scholler, N.; Bergan, L.; Thorpe, J.; Ur- 273. ban, N. and Drescher, C.W. Effects of personal characteristics on [93] Bafadhel, M.; Saha, S.; Siva, R.; McCormick, M.; Monteiro, W.; serum CA125, mesothelin, and HE4 levels in healthy postmeno- Rugman, P.; Dodson, P.; Pavord, I.D.; Newbold, P. and Brightling, pausal women at high-risk for ovarian cancer. Cancer Epidemiol. C.E. Sputum IL-5 concentration is associated with a sputum eosi- Biomarkers Prev., 2008, 17, 2480-2487. nophilia and attenuated by corticosteroid therapy in COPD. Respi- [76] Jacobs, I.J. and Menon, U. Progress and challenges in screening for ration, 2009, 78, 256-262. early detection of ovarian cancer. Mol. Cell Proteomics, 2004, 3, [94] Miller, M.; Cho, J.Y.; Pham, A.; Ramsdell, J. and Broide, D.H. 355-366. Adiponectin and functional adiponectin receptor 1 are expressed by [77] Toh, H.C.; Wang, W.W.; Chia, W.K.; Kvistborg, P.; Sun, L.; Teo, airway epithelial cells in chronic obstructive pulmonary disease. J. K.; Phoon, Y.P.; Soe, Y.; Tan, S.H.; Hee, S.W.; Foo, K.F.; Ong, S.; Immunol., 2009, 182, 684-691. Koo, W.H.; Zocca, M.B. and Claesson, M.H. Clinical benefit of al- [95] Leonardi, A.; Sathe, S.; Bortolotti, M.; Beaton, A. and Sack, R. logeneic melanoma cell Lysate-Pulsed autologous dendritic cell Cytokines, matrix metalloproteases, angiogenic and growth factors vaccine in MAGE-positive colorectal cancer patients. Clin. Cancer in tears of normal subjects and vernal keratoconjunctivitis patients. Res., 2009, 15, 7726-7736. Allergy, 2009, 64, 710-717. [78] Engwegen, J.Y.; Gast, M.C.; Schellens, J.H. and Beijnen, J.H. [96] Tadros, Y.; Ruiz-Deya, G.; Crawford, B.E.; Thomas, R. and Abdel- Clinical proteomics: searching for better tumour markers with Mageed, A.B. In vivo proteomic analysis of cytokine expression in SELDI-TOF mass spectrometry. Trends Pharmacol. Sci., 2006, 27, laser capture-microdissected urothelial cells of obstructed uretero- 251-259. pelvic junction procured by laparoscopic dismembered pyeloplasty. [79] Huang, R.P.; Huang, R.; Fan, Y. and Lin, Y. Simultaneous detec- J. Endourol., 2003, 17, 333-336. tion of multiple cytokines from conditioned media and patient's [97] Reverter, J.L.; Nadal, J.; Fernandez-Novell, J.M.; Ballester, J.; sera by an antibody-based protein array system. Anal. Biochem., Ramio-Lluch, L.; Rivera, M.M.; Elizalde, J.; Abengoechea, S.; 2001, 294, 55-62. Guinovart, J.J. and Rodriguez-Gil, J.E. Tyrosine phosphorylation [80] Britschgi, M. and Wyss-Coray, T. Blood protein signature for the of vitreous inflammatory and angiogenic peptides and proteins in early diagnosis of Alzheimer disease. Arch. Neurol., 2009, 66,161- diabetic retinopathy. Invest. Ophthalmol. Vis. Sci., 2009, 50, 1378- 165. 1382. [81] Shafer, M.W.; Mangold, L.; Partin, A.W. and Haab, B.B. Antibody [98] Kuusniemi, A.M.; Lapatto, R.; Holmberg, C.; Karikoski, R.; array profiling reveals serum TSP-1 as a marker to distinguish be- Rapola, J. and Jalanko, H. Kidneys with heavy proteinuria show fi- nign from malignant prostatic disease. Prostate, 2007, 67, 255-267. brosis, inflammation, and oxidative stress, but no tubular pheno- [82] Srivastava, M.; Eidelman, O.; Jozwik, C.; Paweletz, C.; Huang, W.; typic change. Kidney Int., 2005, 68,121-132. Zeitlin, P.L. and Pollard, H.B. Serum proteomic signature for cystic [99] Traktuev, D.O.; Merfeld-Clauss, S.; Li, J.; Kolonin, M.; Arap, W.; fibrosis using an antibody microarray platform. Mol. Genet. Me- Pasqualini, R.; Johnstone, B.H. and March, K.L. A population of tab., 2006, 87, 303-310. multipotent CD34-positive adipose stromal cells share pericyte and [83] Kader, H.A.; Tchernev, V.T.; Satyaraj, E.; Lejnine, S.; Kotler, G.; mesenchymal surface markers, reside in a periendothelial location, Kingsmore, S.F. and Patel, D.D. Protein microarray analysis of dis- and stabilize endothelial networks. Circ. Res., 2008, 102, 77-85. ease activity in pediatric inflammatory bowel disease demonstrates [100] Nichols, M.T.; Gidey, E.; Matzakos, T.; Dahl, R.; Stiegmann, G.; elevated serum PLGF, IL-7, TGF-beta1, and IL-12p40 levels in Shah, R.J.; Grantham, J.J.; Fitz, J.G. and Doctor, R.B. Secretion of Crohn's disease and ulcerative colitis patients in remission versus cytokines and growth factors into autosomal dominant polycystic active disease. Am. J. Gastroenterol., 2005, 100, 414-423. kidney disease liver cyst fluid. Hepatology, 2004, 40, 836-846. [84] Liu, B.C.; Zhang, L.; Lv, L.L.; Wang, Y.L.; Liu, D.G. and Zhang, [101] Walt, D.R.; Blicharz, T.M.; Hayman, R.B.; Rissin, D.M.; Bowden, X.L. Application of antibody array technology in the analysis of M.; Siqueira, W.L.; Helmerhorst, E.J.; Grand-Pierre, N.; Oppen- urinary cytokine profiles in patients with chronic kidney disease. heim, F.G.; Bhatia, J.S.; Little, F.F. and Brody, J.S. Microsensor ar- Am. J. Nephrol., 2006, 26, 483-490. rays for saliva diagnostics. Ann. N. Y. Acad. Sci., 2007, 1098, 389- [85] Rovin, B.H.; Song, H.; Hebert, L.A.; Nadasdy, T.; Nadasdy, G.; 400. Birmingham, D.J.; Yung, Y.C. and Nagaraja, H.N. Plasma, urine, [102] Celis, J.E.; Moreira, J.M.; Gromova, I.; Cabezon, T.; Ralfkiaer, U.; and renal expression of adiponectin in human systemic lupus Guldberg, P.; Straten, P.T.; Mouridsen, H.; Friis, E.; Holm, D.; erythematosus. Kidney Int., 2005, 68, 1825-1833. Rank, F. and Gromov, P. Towards discovery-driven translational research in breast cancer. FEBS J., 2005, 272, 2-15. 70 Current Proteomics, 2012, Vol. 9, No. 1 Huang et al.

[103] Liu, B.C.; Zhang, L.; Lv, L.L.; Wang, Y.L.; Liu, D.G. and Zhang, [117] Pei, R.; Cheng, Z.; Wang, E. and Yang, X. Amplification of anti- X.L. Application of antibody array technology in the analysis of gen-antibody interactions based on biotin labeled protein- urinary cytokine profiles in patients with chronic kidney disease. streptavidin network complex using impedance spectroscopy. Bio- Am. J. Nephrol., 2006, 26, 483-490. sens. Bioelectron., 2001, 16, 355-361. [104] Hou, Z.; Sun, L.; Gao, L.; Liao, L.; Mao, Y. and Liu, J. Cytokine [118] Rickert, J.; Brecht, A. and Gopel, W. Quartz crystal microbalances array analysis of peritoneal fluid between women with endometri- for quantitative biosensing and characterizing protein multilayers. osis of different stages and those without endometriosis. Biomark- Biosens. Bioelectron., 1997, 12, 567-575. ers, 2009, 14, 604-618. [119] Renken, J.; Dahint, R.; Grunze, M. and Josse, F. Multifrequency [105] Lembo, A.J.; Neri, B.; Tolley, J.; Barken, D.; Carroll, S. and Pan, evaluation of different immunosorbents on acoustic plate mode H. Use of serum biomarkers in a diagnostic test for irritable bowel sensors. Anal. Chem., 1996, 68, 176-182. syndrome. Aliment. Pharmacol. Ther., 2009, 29, 834-842. [120] Nakanishi, K.; Ikebukuro, K. and Karube, I. Determination of [106] Sanchez-Carbayo, M.; Socci, N.D.; Lozano, J.J.; Haab, B.B. and cyanide using a microbial sensor. Appl. Biochem. Biotechnol., Cordon-Cardo, C. Profiling bladder cancer using targeted antibody 1996, 60, 97-106. arrays. Am. J. Pathol., 2006, 168, 93-103. [121] Wu, G.; Ji, H.; Hansen, K.; Thundat, T.; Datar, R.; Cote, R.; Ha- [107] Vazquez-Martin, A.; Colomer, R. and Menendez, J.A. Her-2/neu- gan, M.F.; Chakraborty, A.K. and Majumdar, A. Origin of induced "cytokine signature" in breast cancer. Adv. Exp. Med. Biol., nanomechanical cantilever motion generated from biomolecular in- 2008, 617, 311-319. teractions. Proc. Natl. Acad. Sci. USA, 2001, 98, 1560-1564. [108] Sanchez-Carbayo, M. Dissecting cancer serum protein profiles [122] Wu, G.; Datar, R.H.; Hansen, K.M.; Thundat, T.; Cote, R.J. and using antibody arrays. Methods Mol. Biol., 2008, 428, 263-287. Majumdar, A. Bioassay of prostate-specific antigen (PSA) using [109] Li, C.; Simeone, D.M.; Brenner, D.E.; Anderson, M.A.; Shedden, microcantilevers. Nat. Biotechnol., 2001, 19, 856-860. K.A.; Ruffin, M.T. and Lubman, D.M. Pancreatic cancer serum de- [123] Cui, Y.; Wei, Q.; Park, H. and Lieber, C.M. Nanowire nanosensors tection using a lectin/glyco-antibody array method. J. Proteome. for highly sensitive and selective detection of biological and Res., 2009, 8, 483-492. chemical species. Science, 2001, 293, 1289-1292. [110] Lukesova, S.; Kopecky, O.; Vroblova, V.; Hlavkova, D.; Andrys, [124] Li, Y.; Yang, H.H.; You, Q.H.; Zhuang, Z.X. and Wang, X.R. C.; Moravek, P. and Cermakova, E. Determination of angiogenic Protein recognition via surface molecularly imprinted polymer factors in serum by protein array in patients with renal cell carci- nanowires. Anal. Chem., 2006, 78, 317-320. noma. Folia Biol.(Praha)., 2008, 54, 134-140. [125] Ma, C.; Fan, R.; Ahmad, H.; Shi, Q.; Comin-Anduix, B.; Chodon, [111] Huang, R.P. Cytokine antibody arrays: a promising tool to identify T.; Koya, R.C.; Liu, C.C.; Kwong, G.A.; Radu, C.G.; Ribas, A. and molecular targets for drug discovery. Comb. Chem. High Through- Heath, J.R. A clinical microchip for evaluation of single immune put. Screen., 2003, 6, 769-775. cells reveals high functional heterogeneity in phenotypically simi- [112] Espina, V.; Woodhouse, E.C.; Wulfkuhle, J.; Asmussen, H.D.; lar T cells. Nat. Med., 2011, 17, 738-743. Petricoin, E.F. III. and Liotta, L.A. Protein microarray detection [126] Gaster, R.S.; Hall, D.A.; Nielsen, C.H.; Osterfeld, S.J.; Yu, H.; strategies: focus on direct detection technologies. J. Immunol. Mach, K.E.; Wilson, R.J.; Murmann, B.; Liao, J.C.; Gambhir, S.S. Methods, 2004, 290,121-133. and Wang, S.X. Matrix-insensitive protein assays push the limits of [113] Angenendt, P. Progress in protein and antibody microarray tech- biosensors in medicine. Nat. Med., 2009, 15, 1327-1332. nology. Drug Discov. Today, 2005, 10, 503-511. [127] Lee, K.B.; Park, S.J.; Mirkin, C.A.; Smith, J.C. and Mrksich, M. [114] Ray, S.; Chandra, H. and Srivastava, S. Nanotechniques in pro- Protein nanoarrays generated by dip-pen nanolithography. Science, teomics: current status, promises and challenges. Biosens. Bioelec- 2002, 295, 1702-1705. tron., 2010, 25, 2389-2401. [128] Blattler, T.; Huwiler, C.; Ochsner, M.; Stadler, B.; Solak, H.; Vo- [115] Nelson, B.P.; Grimsrud, T.E.; Liles, M.R.; Goodman, R.M. and ros, J. and Grandin, H.M. Nanopatterns with biological functions. Corn, R.M. Surface plasmon resonance imaging measurements of J. Nanosci. Nanotechnol., 2006, 6, 2237-2264. DNA and RNA hybridization adsorption onto DNA microarrays. [129] Truskett, V.N. and Watts, M.P. Trends in imprint lithography for Anal. Chem., 2001, 73, 1-7. biological applications. Trends Biotechnol., 2006, 24, 312-317. [116] Pawlak, M.; Schick, E.; Bopp, M.A.; Schneider, M.J.; Oroszlan, P. [130] Kumar, N. and Hahm, J.I. Nanoscale protein patterning using self- and Ehrat, M. Zeptosens' protein microarrays: a novel high per- assembled diblock copolymers. Langmuir, 2005, 21, 6652-6655. formance microarray platform for low abundance protein analysis. Proteomics, 2002, 2, 383-393.

Received: October 13, 2011 Revised: January 05, 2012 Accepted: January 10, 2012