proteomics in the chicken: tools for understanding immune responses to avian … · 2018-12-05 ·...

Proteomics in the Chicken: Tools for UnderstandingImmune Responses to Avian Diseases

S. C. Burgess1

Department of Basic Sciences, Mississippi State University, College of Veterinary Medicine,P.O. Box 6100, Mississippi State, Mississippi 39762-6100

ABSTRACT The entire chicken genome sequence willbe available by the time this review is in press. Chickenswill be the first production animal species to enter the“postgenomic era.” This fundamental structural geno-mics achievement allows, for the first time, complete func-tional genomics approaches for understanding themolecular basis of chicken normo- and pathophysiology.The functional genomics paradigm, which contrasts withclassical functional genetic investigations of one gene (orfew) in isolation, is the systematic holistic genetic analysesof biological systems in defined contexts. Context-depen-dent gene interactions are the fundamental mechanics ofall life. Functional genomics uses high-throughput large-scale experimental methods combined with statistical andcomputational analyses. Projects with expressed se-quence tags in chickens have already allowed the creationof cDNA microarrays for large-scale context-dependantmRNA analysis (transcriptomics). However, proteins arethe functional units of almost all biological processes,and protein expression very often bears no correlation to

(Key words: proteome, proteomic, mass spectrometry, two-dimensional, polyacrylamide gel electrophoresis)

2004 Poultry Science 83:552–573

INTRODUCTION

The term “proteome” was first coined by Marc Wilkinsin 1994 at the inaugural two-dimensional (2-D) gel electro-phoresis meeting in Sienna, Italy. The proteome was origi-nally defined as the protein compliment of the genome.“Proteomics” was defined as the study of the proteome(Wilkins et al., 1996). Since then, many definitions havebeen used. The proteome is context-dependent; it encom-passes time, environment, quantity, stoichiometry, andstructure including posttranslational modifications (PTM)such as glycosylations and interacting partners. The pro-teome is diverse and complex and may be infinite (Huber,2003). As a discipline, proteomics is immature, using rela-

2004 Poultry Science Association, Inc.Received for publication July 29, 2003.Accepted for publication November 11, 2003.1To whom correspondence should be addressed:burgess@cvm.

msstate.edu.

552

mRNA expression. Proteomics, a discipline within func-tional genomics, is the context-defined analysis of com-plete complements of proteins. Proteomics bridges the“sequence-to-phenotype gap;” it complements structuraland other functional genomics approaches. Proteomicsrequires high capital investment but has ubiquitous bio-logical applications. Although currently the fastest-grow-ing human biomedical discipline, new paradigms mayneed to be established for production animal proteomicsresearch. The prospective promise and potential pitfallsof using proteomics approaches to improve poultrypathogen control will be specifically highlighted. The firststage of our recently established proteomics program isglobal protein profiling to identify differentially ex-pressed proteins in the context of the commercially im-portant pathogens. Our trials and tribulations inestablishing our proteomics program, as well some ofour initial data to understand chicken immune systemfunction, will be discussed.

tively immature technologies (Patterson and Aebersold,2003) and is still rapidly evolving (Pardanani et al., 2002).

Proteomics will soon become as important to chickenresearchers as it has to human researchers. This is because,due to its biomedical utility, the chicken will imminentlyjoin those vertebrates that have representative genomessequenced (Burt and Pourquie, 2003). For the first time,complete functional genomics approaches, including pro-teomics, will be completely accessible for understanding

Abbreviation Key: 2-D = two-dimensional; CID = collision induceddissociation; CSHL = Cold Spring Harbor Laboratories; CVM = Collegeof Veterinary Medicine; DIGE = difference gel electrophoresis; ESI =electrospray ionization; FT = Fourier-transform; ICAT = isotope codedaffinity tag; ICR = ion cyclotron resonance; IEF = isoelectric focusing;IFN = interferon; IL = interleukin; IPG = immobilized pH gradient;LSBI = Life Sciences and Biotechnology Institute; MALDI = matrix-assisted laser desorption/ionization; MS = mass spectrometry; MSn =tandem mass spectrometry; MSU = Mississippi State University; NIH =National Institutes of Health; PMF = peptide mass fingerprint; pI =isoelectric point; PTM = posttranslational modification; TOF = time-of-flight.

ANCILLARY SCIENTISTS SYMPOSIUM 553

the molecular basis of chicken normo- and pathophysiol-ogy. The chicken will be the first livestock species to enterthe “postgenomic era,” and as such the chicken will be amodel for other livestock species.

A comprehensive review of every aspect of proteomicsis beyond the scope of this paper. Also, this review doesnot define proteomics simply by the use of mass spectrome-try (MS) in biochemistry (although many authors now usethe term in that context). With the exception of significantresearch that impacts proteomics, this review does not in-clude work that is not “omic” in its aims. This review has2 aims: first, to introduce proteomics and to provide basicinformation to those who are new to the field and, second,to discuss proteomics specifically in the context of immuneresponses to infectious pathogens for avian researchers andproducers. The exponential growth of proteomics (Figure1) means it is practically impossible to include every refer-ence for any particular point. Even a recent review of pro-teomics advances in 2002 alone was unable to becomprehensive in its references (Figeys, 2003). Referencesare restricted to seminal publications, representative exam-ples, or detailed reviews. This review will deal almost ex-clusively with understanding wet-laboratory proteomicstechniques and technologies and will only briefly discusscritical bioinformatics and computational biology aspectsof proteomics.

This review is written from the perspective of someonenew to proteomics who wishes to establish the technologiesin the laboratory and then to use proteomics as part of anexpanding arsenal of tools to understand disease pathogen-esis and immunity. Because so much of proteomics is tech-nology driven, and many diverse disciplines have becomedrawn together into proteomics, it is impossible to discussproteomics without the use of technical terms and acro-nyms. A novice to the field can become confused by acro-nyms. Furthermore “proteomicists” use many of theacronyms as if they were words in their own right. Gener-ally, each acronym is first defined in the text. A list ofabbreviations is also provided.

Specifically, this review discusses the proteome and pro-teomics, describes the contemporary proteomics “toolbox,”reviews the “immunoproteomics” literature, and then dis-cusses paradigms for chicken immunoproteomics. An in-troduction to chicken immunoproteomics at the College ofVeterinary Medicine Mississippi State University follows,and the review concludes with some brief personalthoughts.

THE PROTEOME AND PROTEOMICS

A basic literature search (PubMed, NCBI, and NIH)through to the end of 2002 using the search term in thetitle or abstract: “proteome OR proteomic” demonstratesthat proteomics is currently undergoing exponentialgrowth (Figure 1A). Even the number of review papers forproteomics published in 2002 averaged 7.9 per week. Fourfactors are largely fueling growth. First, the complete anno-tated genome sequences, particularly the human, allowrapid protein identification from MS data. Second, af-

FIGURE 1. The exponential growth of proteomics is illustrated bythe number refereed publications in PubMed. Total proteomics papersand proteomics reviews (A). Total immunoproteomics papers and im-munoproteomics reviews (B). The percentages of proteomics papersthat are immunoproteomic are increasing exponentially; this compareswith the percentage of genomics papers that are immunogenomic andis static (mean 4.5 ± 0.05) (C). Data derived from literature searches ofPubMed, years 1994 to 2002 inclusive; search terms are described in text.

fordable high-throughput technologies now exist to mea-sure proteins from biological samples with excellentsensitivity, accuracy, and rapidity. Third, databases, com-puter algorithms and modeling have reached the stage atwhich genomic and proteomic data can be combined togive reliable and biologically meaningful data. Finally, cellsand the organisms they make up are protein machines;understanding biology means understanding global pro-tein function. In the majority of instances when it has actu-

BURGESS554

ally been compared, there is either no correlation or verypoor correlation between quantities of mRNA encodinga protein and that protein (reviewed by Goodlet, 2003).Understanding global protein function requires quantita-tive measures of protein to be made directly and not in-ferred or assumed based, for example, on cDNAmicroarray data.

Although defined above, more detailed examination ofwhat proteome and proteomics currently mean in biologyis useful. The proteome is enormously complex. Estimatesof the total potential number of proteins produced by thehuman genome’s compliment of ∼40,000 genes are in thevicinity of 500,000 [one human gene alone produces inexcess of 1,000 proteins (Ullrich et al., 1995)]. The humangenome contains only about twice as many genes as thefruit-fly, Drosophila melanogaster; the worm, Caenorhabditiselegans; or even plants (Lander et al., 2001; Venter et al.,2001). This number of genes does not provide the complex-ity needed for vertebrate growth, maintenance, and func-tion if only a single protein is made from each gene.Complexity is provided by alternate mRNA splicing fol-lowed by co- and posttranslational modifications. Morethan 200 separate protein modifications are documented invertebrates. More than one of these modifications routinelyoccurs on most proteins (Gooley and Packer, 1997).

As implied already, this increased scale of complexitysubstantially conflicts with the “pregenomic” basic centraltenet of “one gene one protein” (Watson and Crick, 1953).We must await the genome sequence, which has alreadybeen shown to be minimized in some areas involving im-munity (Reitman et al., 1993; Gobel, 1996; Beck et al., 1999;Kaufman et al., 1999; Su et al., 1999), to get an estimate ofcomplexity scaling between genome and proteome in thechicken. Nevertheless, the mechanisms behind the in-creased complexity between the genome and the proteomeit encodes are the same in the chicken as in other verte-brates.

Proteomics departs substantially from the reductionismof classical biochemistry in that it is global by nature. Bydefinition, proteomics aims to analyze protein systems intheir entirety. In this respect, proteomics commonly oper-ates from a “discovery science” paradigm. Discovery sci-ence is designed to identify new avenues of investigationand create new hypotheses rather than being “hypothesisdriven.” Like other, more established discovery sciencetechnologies of genome sequencing and cDNA microarraytranscription analyses, proteomics is a fundamental compo-nent of a “systems biology” approach. Systems biologymay be defined as the systematic and quantitative analysesof all components of a biological system (Aebersold etal., 2000).

As with many areas in proteomics, opinion varies onexactly what constitutes a “proteomics experiment.” At oneextreme, the term may be used simply because MS wasused as one step toward identifying a protein in whatotherwise is classical reductionist biochemistry. At theother extreme, the term proteomics is used only if theexperiment surpasses a user’s definition of highthroughput. From the practical perspective of investigators

coming to proteomics from other disciplines, proteomicsmeans investing in new technologies and methodologies,rapidly producing large amounts of data, and then devel-oping new ways of analyzing this data. It is impossible forbiologists to do proteomics in isolation from engineers,mathematicians, bioinformaticians, computer scientists,and physicists at least to some degree. The complexity andsize of the proteome in a vertebrate has meant that teamsof scientists must collaborate in an organized manner. Thisneed lead to the establishment of the Human ProteomeOrganization (Hanash and Celis, 2002).

One area of consensus is that proteomics may be dividedinto 3 broad areas. The first area is termed “expression”proteomics. In expression proteomics, the relative abun-dances of proteins are measured and compared. Expressionproteomics is conceptually equivalent to differential geneexpression experiments using cDNA microarrays. The sec-ond area is “cellular” proteomics. Here, the aim is to iden-tify protein-protein interactions and to describe thecomplex interacting networks that are the components ofbiological machines. The third broad area is “structural”proteomics, in which the goal is to be able to predict thethree-dimensional structures of proteins on a genome-widescale. The basic premise is that if high resolution structuresare available for a sufficiently large number of proteins,then all other protein structures could be solved by homol-ogy (Chance et al., 2002). This review will limit itself to thefirst and second areas because they are most likely to beused initially by investigators outside of specialist facilitiesinvestigating specific poultry issues.

Proteomics Toolbox

For proteomics novices it is useful to have an inventoryof what is currently in the proteomics “toolbox.” Beforebeginning the inventory however, it is important to notetwo things. First, the obsolescence rate for proteomicsequipment is very fast. Second, unlike genomic sequencedata for which the data set may be considered linear anddirectly obtained, the data set in proteomics is nonlinearand relies on much human intervention and decision mak-ing. Correct interpretation of proteomics data requires un-derstanding how proteomics technologies work andunderstanding their limitations.

There remain many practical problems to be overcomein proteomics. One general problem in proteomics is rela-tively poor sensitivity. The dynamic range for proteins intissue is 7 to 8 orders of magnitude, and in serum thedynamic range may be up to 12 orders of magnitude (re-viewed by Anderson and Anderson, 1998). The target-spe-cific exponential amplification afforded by PCR for nucleicacids has no equivalent for proteins. Although work contin-ues toward a protein equivalent for PCR (Fredriksson et al.,2002; Gullberg et al., 2003), sensitivity issues are generallyapproached in the wet laboratory by pre-analytical fraction-ation or at the level of mass spectrometers.

Before beginning to describe the proteomics toolbox, itis important to note that the technologies themselves arewell established in other disciplines. It is their integration


toward achieving the proteomics goal that defines theirutility.

Genomes

Although not absolutely essential, a sequenced genomegreatly facilitates proteomics. However, if large amountsof genetic and expressed sequence tag data exist, it is stillpossible to do proteomics even without a sequenced ge-nome as demonstrated in the chicken by the Beynon group(Hayter et al., 2003). Simply having a complete genomicsequence is not a panacea; the genome data must be in auseable format. This generally means that at least it can beused for protein identification by matching with MS data.In short, it is best if the genome is well annotated andpresent in accessible minimally redundant databases [e.g.,Swiss-Prot (Bairoch and Apweiler, 1997) and NRDB (Na-tional Center for Biotechnology Information)]. This funda-mental requirement was immediately recognized for thechicken (Burt and Pourquie, 2003).

In addition, the genomes of pathogens are of fundamen-tal relevance to the proteomics of immunity and disease.It is axiomatic that well-annotated pathogen genome se-quences be available, in addition to the chicken genomesequence, if proteomics approaches are to be applied todeciphering chicken-pathogen interactions.

Bioinformatics and Computational Biology

Bioinformatics and computational biology are critical toproteomics in acquiring, storing, interpreting, and analyz-ing data as well as building testable models and hypothe-ses. So that these terms do not appear to be nebulous catch-alls, and because bioinformatics and computational biologyare so essential to proteomics, they will be defined. Thedefinitions proposed by the US National Institutes ofHealth (Huerta et al., 2000) are as follows.

Bioinformatics: research, development, or applicationof computational tools and approaches for expandingthe use of biological, medical, behavioral or health data,including those to acquire, store, organize, archive, ana-lyze, or visualize such data.

Computational biology: the development and applica-tion of data-analytical and theoretical methods, mathe-matical modeling, and computational simulationtechniques in the study of biological, behavioral, andsocial systems.

Companies have emerged that offer specific computersolutions for particular proteomics problems. Universitydepartments have also invested much effort into devel-oping computing solutions, which are freely available overthe World Wide Web. Some of the basic concepts behindbioinformatics and computational biology for specific tech-nologies will be given below when appropriate. In theinterest of space and to provide a primer from which read-ers may then expand their own knowledge, only databasesmost immediately useful to avian proteomics will be de-scribed. Data storage at the investigator’s laboratory willthen be specifically discussed because of its importance.

Many more or less specialized databases, which aresearchable repositories of data for proteomics, exist aroundthe world (reviewed by Various, 2003). More detailed re-views of databases are available from the reading list inthe Appendix. The main nucleotide sequence databases arethe EMBL Nucleotide Sequence Database (Europe), Gen-Bank (USA), and the DNA Database of Japan (DDBJ; Japan).These 3 databases are repositories for nucleotide sequencedata, and all exchange data daily. Translations of the se-quences in these databases, which are then digested insilico, are used to compare with MS data for protein identi-fication. The Swiss-Prot database is the major curated pro-tein database. It is notable for 3 reasons. First, all data isannotated, which means that known features such as PTM,functional domains, and structure and disease associationsare included. Second, it is minimally redundant, meaningthat it is easier to get all the known data in one search.Third, it is cross-referenced with many other databasescontaining the 3 types of sequence-related data (nucleicacid sequences, protein sequences, and protein tertiarystructures). This means that moving to other databasesis a straightforward process. The TrEMBL database is asupplement of Swiss-Prot that contains all translations ofEMBL nucleotide sequence entries that are not yet inte-grated into Swiss-Prot.

The amounts of data generated in proteomics are expo-nentially larger than that for genomics. As discussed above,the proteome is immensely complex, and there may belarge inherent potential variation in proteomics data. Thosefamiliar with cDNA microarrays will be aware that similarissues face transcriptome analyses (Bhanot et al., 2003).The minimum information about a microarray experiment(MIAME) guidelines that some journals are adopting asa standard before publishing data outline the minimuminformation requirements for unambiguously interpretingmicroarray data and to subsequently allow independentverification of this data at a later stage if required (Brazmaet al., 2001). Similar initiatives are just beginning in pro-teomics.

Two groups are currently addressing the issues of pro-teomics community standards for data representation tofacilitate data comparison, exchange, and verification andto define what is necessary for standard representation ofthe methods use and data generated in proteomics experi-ments (Kaiser, 2002; Taylor et al., 2003). These issues needto be resolved not only for the implementation of proteomerepositories but also to optimize protocols in the laboratoryand for others to understand an investigator’s results. Asoutlined above, human logic is needed in proteomics; logi-cal interpretation cannot be done without understandingthe technologies and knowing the exact settings and pro-tocols.

The terminology for data pertaining to proteomics resultsis metadata, i.e., “data about data.” From the perspectiveof a biological proteomicist, metadata includes samplepreparation methods, liquid chromatography gradients,mass spectrometer settings, and database search algorithmsettings. Of course, these examples are data for a massspectrometrist. Although this may appear intuitive and

BURGESS556

the sort of thing that would normally be written into thematerials and methods section of a publication, the amountof metadata is exponentially greater in proteomics than formost other disciplines. Therefore, metadata is exponentiallyharder to track in proteomics than in more reductionistbiological research. Regardless, keeping, storing, and pro-viding access to proteomics metadata will be a serious issueand is predicted to affect publishing proteomics results inthe future.

Electrophoresis-Based Proteomics

Electrophoresis-based proteomics may be done in 1 or2 dimensions usually using SDS-PAGE, but there are alsoliquid and gel-based microcapillary systems (capillary elec-trophoresis and capillary electrochromatography). Regard-less, the aim is to deconvolute a complex protein or peptidemixture by taking advantage of the physicochemical prop-erties of the proteins or peptides. The principles are thesame no matter what physical appearance the separationsystem has. This review focuses on PAGE. From a proteo-mics perspective, one-dimensional SDS PAGE may be use-ful for subcellular protein complexes or when combinedwith other technologies [see isotope coded affinity tag(ICAT) below].

Two-dimensional (2-D) PAGE is the electrophoresis-based technology most commonly associated with proteo-mics (Figures 2 and 6). 2-D PAGE was first described in 1974(MacGillivray and Rickwood, 1974). But it only becamecommonly used during the late 1980s and early 1990s whenmicroanalytical techniques, which were able to definitivelyidentify proteins from these gels, became available (re-viewed by Rabilloud, 2002). 2-D PAGE separates proteinsin 2 dimensions. The first dimension is separation by iso-electric point (pI) in a polyacrylamide gel, which has a pHgradient, in an electric field (Figure 2, A and B). This processis called isoelectric focusing (IEF). The pI is the specific pHat which the net charge of the protein is zero. Proteins maybe separated by their pI because they are amphoteric (i.e.,at a given pH the net charge of a protein is positive, nega-tive, or zero). Proteins are positively charged at pH valuesbelow their pI and negatively charged at pH values abovetheir pI. A protein that is positive in the pH range of thegel that it is currently in will migrate toward the cathodeand visa versa (Figure 2B). Once the protein enters the pHat which it has no net charge, it stops migrating. Notably,changes in PTM (such as phosphorylation) will cause achange in pI without a discernable change in mass detect-able in the second dimension (Figure 2C).

Originally IEF gels were cast in the investigators’ labora-tories. Carrier ampholytes, which are small soluble mole-cules with high buffering capacity near their pI, were usedto make the pH gradients. When voltage is applied acrossthe carrier ampholyte mixture in the gel, the carrier ampho-lytes with the lowest pI (most negative) migrate to theanode and visa versa. However, these gels had a numberof disadvantages. Now, IEF gels are available commerciallyin a plastic backed format (Figure 2A) for using the alterna-tive technique of immobilized pH gradients [IPG (Bjellqvist

FIGURE 2. Photograph of silver-stained immobilized pH gradient(IPG) strip, from 2-dimensional (2-D) PAGE analysis of neoplasticallytransformed T lymphocytes from Marek’s disease lymphoma (nonlinearpH range 3 to 11, 11 cm; BioRad, www.biorad.com), after soelectricfocusing (IEF) and second dimension electrophoresis (A). Many bandscan be seen, and these bands are well resolved, demonstrating satisfac-tory IEF. But protein remains in the IPG strip, demonstrating that notall proteins will migrate into the second dimension SDS-polyacrylamidegel. This occurs in all 2-D PAGE experiments. Schematic representationof IEF (B). Proteins are loaded into the IPG strip, and, because proteinsare amphoteric, they carry charges relative to the pH of the environmentthey are in (i). An electric current is passed through the IPG, and IEFoccurs (ii) (i.e., the proteins migrate to the region of the IPG where pHis such that they have zero net charge) (iii). Second dimension SDSPAGE (C). The IPG strip is laid on top of the SDS polyacrylamide gel,again in an electric current. Those proteins that will enter the seconddimension will migrate relative to their mass (i), so-called “spot trains”occur when posttranslational modifications (PTM) alter the charge of asubset of the molecules of a given protein but alter the mass undetectably(e.g., phosphorylation) (ii); some proteins will not enter the seconddimension (dotted line), whereas those that do, will do so in directproportion to the amount in the IPG strip (providing the basis of quanti-fication by 2-D PAGE) (iii). PTM are not all or none events; multiplespecies of a single protein are seen as spot trains (iv).

et al., 1982)]. IPG are created by covalently incorporatinga gradient of acidic and basic buffering groups into thepolyacrylamide gel at the time it is cast. These IPG IEFgels are less variable and much easier to handle then thehomemade IEF gels, resulting in less variation betweengels. IEF strips are commercially available in many pHranges and in linear and nonlinear formats. The second


dimension separation is based simply on molecular weight(MW) using an SDS polyacrylamide gel on which the firstdimension is laid. Precast linear and nonlinear commercialgels are available to reduce variation.

The most recent manifestation of 2-D PAGE is fluores-cence 2-D difference gel electrophoresis [DIGE (Unlu et al.,1997)]. The advantage of 2-D DIGE is that proteins from 2different sources may be analyzed and compared in a singlegel. This greatly improves experiment accuracy and repeat-ability. Briefly, proteins are harvested, and total protein isquantified. Equal amounts of proteins from the 2 differentsources are then conjugated to 2 different amine reactivedyes. A third dye is used to normalize the gel; a mixtureof equal amounts of protein from each source is labeledwith this dye. The dyes are designed such that proteinscommon to both samples have the same relative mobilityregardless of the dye used to tag them. Equal amounts oftotal protein tagged with each of the 3 dyes are then mixedprior to 2-D PAGE. Post-electrophoresis, the gel is imagedinto its 3 component images. The images are superimposed,and differences are analyzed by proprietary software.

Regardless of the 2-D PAGE/DIGE method used, samplepreparation is critical. In short, the starting material must besolubilized and chemically reduced (to destroy disulphidebonds at cistine residues), and the reduced covalent bondsmust be alkylated to prevent disulphide bond reformation.Sample preparation requires a lot of optimization and vali-dation. This optimization and validation must be doneusing the 2-D PAGE/DIGE system that will finally be usedfor the actual experiment. It is worth reiterating that sampleoptimization is fundamental to meaningful 2-D PAGE/DIGE. Unfortunately, sample optimization is an expen-sive exercise.

After finally getting to the stage of actually being ableto run a 2-D PAGE/DIGE experiment, the experimentermust somehow identify differentially expressed proteinspots across the gels. This is not a problem for 2-D DIGEassuming the experimenter has access to proprietary soft-ware and specialized equipment. For the more commonlyused 2-D PAGE, a number of staining techniques may beused. There is not room here to describe each in detail. Inshort the 2 most common current methods are Coomassiebrilliant blue staining and silver staining. Staining is yetanother source of variation in 2D-PAGE, and so protocolsmust be carefully optimized. Silver has the advantage be-cause it is approximately 50 to 100 times more sensitivethan Coomassie brilliant blue staining (silver detects ∼103

copies/cell, which is ∼0.01 ng/mm of gel; Coomassie de-tects ∼105 copies/cell, which is 10 ng/mm of gel). However,Coomassie brilliant blue staining is more quantitative thansilver staining. Automatic gel stainers are available to helpwith the issue of consistent staining. Another critical pointis that whatever staining protocol is chosen, it must becompatible with MS (although obvious, this point may beoverlooked by naive investigators).

The next step is computer-assisted spot differentiation.All commercially available computer programs use the pix-els generated by the gel-imaging instrument and softwareto measure spot size and density. The first thing commer-

cially available programs basically do is to match all of thegels in a particular experiment using some form of landmarking. Gel matching is critical because gels distort dur-ing fixation and staining; no 2 gels end up the same shape.Next, algorithms normalize each gel in the set using aparticular criterion such as background intensity. Finally,after manually establishing sensitivity thresholds, the pro-grams will help operators identify spots expressed differen-tially in one set of biological conditions but not the other.However, as indicated above, 2-D PAGE is extremely vari-able. Even experienced and skilled 2-D PAGE “artisans”may get up to 30% variation between gels from the sameparent sample run in parallel (S. Vazquez, Waters LifeSciences, Milford, MA, personal communication). Cur-rently in my laboratory we analyze all gels, run in parallelwithin a biological condition, for variation. We considergood intracondition variation to be 10 to 15%; anythinggreater and the experiment is rerun from the beginning(i.e., from sample solubilization).

The objective of all proteomics is to have definitive pro-tein identification. To do this for 2-D PAGE/DIGE, proteinspots must be cut from the gels, which may be done roboti-cally. The resulting gel plugs are digested with proteasewhile in gel (most commonly using trypsin). The derivedpeptides are desalted and measured by MS, and the re-sulting spectra are compared with databases (see below).

The main advantages of 2-D electrophoresis over theelectrophoresis-free methods discussed below are that it isan easy method for protein quantification, and it is cur-rently the best method for analyzing PTM. The main draw-backs with 2-D PAGE/DIGE are (1) that the methodsrequire much optimization; (2) the first dimension IEFstrips have limited protein capacity, and not all proteinswill get into the IEF gel; (3) not all proteins will migratefrom the first to the second dimension gel (Figure 2A); (4)the methods are difficult to automate although not impossi-ble (automation is critical given proteomics, by definition,must be high throughput); (5) resolution and sensitivitymay be problematic; and (6) 2-D PAGE/DIGE also haverelatively poor dynamic range in terms of concentration(104 compared with 106 for an average cell). Very large,very small, very basic (pI >11), very acidic (pI < 3), andvery hydrophobic (e.g., cell membrane) proteins are alsoproblematic.

An interesting recent departure from classical one-di-mensional and 2-D PAGE, which has particular merit whenused in combination with matrix-assisted laser desorption/ionization (MALDI) time-of-flight (TOF) TOF MS, is massspectrometric imaging of IPG gels and creation of virtual2-D gels (Walker et al., 2001). To create these virtual 2-Dgels, the IPG strip (after IEF) is loaded directly into aMALDI TOF or MALDI TOF TOF. Not only can a 2-D mapbe made, but also all proteins that have been loaded intothe IPG strip are accessible to mass spectrometric analysis(unlike 2-D PAGE). If a MALDI TOF TOF is used, tandemMS (MSn) fragmentation spectra can be produced directly.

Electrophoresis-Free ProteomicsBecause of the restrictions inherent in electrophoresis-

based proteomics methods, some proteomicists are devel-

BURGESS558

oping electrophoresis-free proteomics. Their immediateand easiest aims are to improve on throughput, sensitivity,and dynamic range and the ability to analyze large, small,basic, acidic, and hydrophobic proteins compared withelectrophoresis-based methods. Their more difficult aimsare to equal, and then improve upon, the ability of electro-phoresis-based proteomics methods to quantify proteinsand to identify PTM.

Like electrophoresis-based proteomics, most electropho-resis-free proteomics relys on methods of deconvolutingbiological samples prior to identification. This deconvolu-tion is primarily based on multidimensional HPLC. A pri-mary initial problem was to determine how to do HPLCat a scale appropriate for samples taken directly ex vivosuch that they provided correct concentrations of analytesfor MS. To accomplish this, the HPLC columns (containingthe stationary phase that initially binds the sample to beanalyzed) and the buffer flow rates have been miniaturizedto microscales or even nanoscales (Shen et al., 2002). Com-monly in the proteomics literature the words “high perfor-mance” are omitted from the acronym, and “LC” issimply used.

The HPLC may be done off line or directly inline witha mass spectrometer fitted with an electrospray ionization(ESI) source (described below). Depending on whether themultidimensional LC is targeted toward entire proteomesor molecular machines, integrated proteomics methods thatuse multidimensional LC have been called multidimen-sional protein identification technology [MudPIT (Wolterset al., 2001)] or direct analysis of large protein complexes(Link et al., 1999). Reverse phase and strong cation ex-change are generally used, although other affinity methodsor size exclusion may be used. Many extensive reviews ofthe technical application of HPLC to proteomics have beenwritten in the last 3 yr, and readers are referred to thosefor further detail.

MS

Until now this review has discussed tools that deconvo-lute the proteome. However, mass spectrometers, associ-ated computer algorithms and soft ionization techniquesare critical to definitively identify proteins and their PTMand thus for the expansion and evolution of proteomics.Although initially daunting to many classically trained bi-ologists, mass spectrometers are conceptually very simple.Mass spectrometers do one thing with great sensitivity andprecision—they measure the masses of very small things.Mass spectrometers have been considered exquisitely sen-sitive weigh scales that can distinguish between differentisotopes of the same compounds. So why is the abilityof mass spectrometers to weigh very small things veryaccurately so critical to proteomics?

Humans define things based on their physical properties.For example, biological molecules may be defined by whatthey do (e.g., enzymes) or what they bind (e.g., binding aspecific monoclonal antibody). For proteins, which are theproducts of all genes, the primary core structure is aminoacids joined by covalent peptide bonds. Any protein of a

FIGURE 3. Representative in silico peptide mass fingerprints (PMF),after trypsin digest, of 2 different hypothetical proteins with the sameamino acid composition (MAREKSKILLSCHICKS and CHICKSKILL-SMAREKS both have molecular weights of 1,930 Da ). Despite havingthe same mass, the proteins have different PMF.

given weight can only be composed of a finite number andcombination of specific amino acids. Of course, the finalweight of any protein can be variably decreased or in-creased by PTM. As a primary protein sequence gets larger(i.e., contains more amino acids) the possibility of it havinga unique weight increases. Theoretically all that shouldneed to be done to identify the protein would be to weighits molecules very accurately. However, the weight of intactproteins almost never definitively identifies that protein.First, a small protein with a lot of PTM may have, bychance, the same weight as a large protein with no PTM.Second, 2 different proteins may contain exactly the sameamino acids in the same ratios, although the amino acidsequence is different.

To solve the problem outlined above, 2 techniques areused. The first is peptide mass mapping by endopeptidasedigestion (Figure 3); the second is peptide fragmentation,most commonly by collision-induced dissociation (CID;Figures 4 and 5). Endopeptidase digestion is most oftendone with trypsin, although many enzymes have beenused. The concept is that particular endopeptidases cleavepeptide bonds in specific places (analogous to restrictionenzymes cleaving DNA). Trypsin cleaves on the carboxyterminus of lysine and arginine (unless proline follows, inwhich case, trypsin will not cleave). For a given proteinthis produces a specific pattern of cleavage products [i.e.,a peptide mass fingerprint (PMF)] of a protein (Figure 3).The PMF is used to search databases containing translationsof all open reading frames in the search set. This may bethe genome of a single species, or all sequences of a related


FIGURE 4. Schematic representation of collision-induced dissocia-tion (CID) using the hypothetical protein CHICKSKILLSMAREKS. (A)From the complex mixture of peptides created by the trypsin digest (i)all but the peptide of interest are removed from the ion trap (ii). Theremaining clonal peptide population (all ILLSMAR) (iii) is bombardedwith an inert gas (e.g., argon) (iv). CID occurs at the peptide bonds,and b and y ions are formed (v). The b and y ion series for ILLSMARis shown (B).

group of species, or all DNA sequences in EMBL/Gen-bank/DDBJ. The translated open reading frames are thendigested in silico with the enzyme used in the laboratoryto produce a database of all potential in silico PMF. Thepossibility remains of different peptides having the sameamino acids in a different sequence and, therefore, the samemass. But the possibility of this occurring across the entireprotein is extremely low. Unfortunately, because of thephysicochemistry of soft ionization, PMF peptide coverageof any protein is nowhere near 100%. Often, instead of adefinitive match from the database, 2 or more possibilitiesmay exist.

To solve the problem of differentiating between peptideswith the same mass but different sequences, molecules ofthe peptide of interest are physically chosen by the massspectrometer for fragmentation analysis. When fragmenta-tion is done by CID it may be called MSn (n = the numberof generations of fragment ions analyzed). The abbreviationMSn is applied to processes that analyze beyond the initialions (MS) to the fragment ions (MS2 or MS/MS) and thento subsequent generations of fragment ions (MS3, MS4, and

FIGURE 5. Representation of an in silico collision-induced dissocia-tion (CID) fragmentation pattern for peptide ILLSMAR but with b andy ion series separated (A) and diagrammatic representation of how areal CID spectrum (without any other ion species) could appear (B).

so on). MS2 is most commonly done and will be describedin detail. MS>2 is becoming more common with the intro-duction of de novo sequencing algorithms. In MS2 the pep-tide molecules enter a collision chamber containing an inertgas. As these peptides collide with the gas molecules, theyare physically fragmented. Each molecule is fragmented inone place (Figure 4). The most common but, unfortunately,not the only fragmentation site is the amide bond of thepolyamide backbone to produce b ions (the amino acidsn-terminal to the fragmentation) and y ions (the aminoacids c-terminal to the fragmentation). The masses of allof the pieces of all of the fragmented molecules are thencollected. Among this extremely complicated spectra arethe b and y ion series. The first b ion (b1) represents thefirst amino acid, and it will have a corresponding y ion(y1) and so on (although b1 is often not present) (Figure5). All corresponding b and y ions add to the parentalpeptide mass. A few extremely skilled people can solvethese complex spectra in a matter of only an hour or soper spectrum. Writing from personal experience, I can attestthat the rest of us can now very gratefully rely on computeralgorithms (Yates et al., 1995; Perkins et al., 1999; Gras andMuller, 2001) to solve CID spectra.

Having briefly summarized MS concepts for proteomics,it is now time to describe the machines. A detailed technical

BURGESS560

FIGURE 6. Laser capture microdissection (LCM) of chicken bursaof Fabricius follicles and stroma (tissue fixed as for immunohistochemis-try but not stained; collaboration with G. T. Pharr, Mississippi StateUniversity). Outlined areas to be microdissected before LCM (A). Tissueremaining after microdissecetion (B) and captured tissue (C).

description of each type of mass spectrometer will not begiven; it is only necessary for proteomicists to have anunderstanding of which types of machine do what and thelimitations of each. To reiterate, MS data must be interpre-ted. The data appears to be similar regardless of how it isobtained. However, the absolute minimum critical meta-data set is the mass spectrometer’s sensitivity; its massaccuracy; its resolution, whether you are dealing with mo-noisotopic masses or average masses; and which PTM (es-pecially those imparted by the chemistry used in samplepreparation) are taken into account by the computer algo-rithm. Investigators have to invest time to learn how toanalyze their data. Relying on a final gene “hit,” as aninvestigator might be tempted to do with DNA sequencedata, is extremely risky. The investigator must start fromthe mass spectra. Detailed texts to help with data interpreta-tion are necessary and widely available

There are 3 basic components to all mass spectrometers:an ion source that ionizes the molecules of interest, a massanalyzer that differentiates the ions according to their mass-to-charge ratio, and a detector that measures the ion beamcurrent. Each of these elements exists in many forms, andmany combinations have been and are being used to pro-duce a wide variety of mass spectrometers with specializedcharacteristics. Only the ionization source and the massanalyzers will be reviewed.

Two ionization sources are primarily used, MALDI andESI for which K. Tanaka and J. B. Fenn shared the 2002Nobel Prize in chemistry, respectively. In MALDI, eitherintact protein or a peptide digest is first cocrystallized ontoa metal plate with a large molar excess of a matrix com-pound. This matrix is an ultraviolet-absorbing weak or-ganic acid (e.g., sinapinic acid or a-cyano-4-hydroxycinnamic acid for proteins or peptides, respec-tively). Ultraviolet laser radiation of this analyte-matrixmixture results in the formation of a plasma and thenvaporization of the matrix, which carries the analyte withit into the vapor phase. The matrix is a proton donor andreceptor, and it ionizes the analyte. The main advantagesof MALDI MS are high sensitivity and high throughputwhen automated and the ability to archive samples on theMALDI plate for re-analysis at a later time.

In contrast to MALDI, ESI generates ions directly fromsolution. In the presence of a strong electric field, a finespray of highly charged droplets is produced. As the drop-let decreases in size, the electric charge density on its surfaceincreases. The mutual repulsion between like charges onthis surface becomes so great that it exceeds the forces ofsurface tension, and ions begin to leave the droplet througha “Taylor cone.” These ions are electrostatically directedinto the mass analyzer. Because ions are formed in solutionby ESI, inline liquid chromatography MS is now routine.The ions next travel into the mass analyzer. The mass ana-lyzer is used to separate ions within a selected range ofmass-to-charge ratios.

Quadrupole ion traps are the first major class of massanalyzers, often simply abbreviated as “Q.” The quadru-pole mass analyzers have 4 precisely parallel rods with adirect current voltage and a superimposed radio-frequency


potential. Scanning one preselected radio-frequency fieldeffectively scans one mass range. Because they toleraterelatively poor vacuums, quadrupoles are suited to ESI inwhich the ions are produced under atmospheric pressureconditions. In quadrupole ion traps, ions are trapped in aradio-frequency quadrupole field. The ions are then ejectedand detected as the radio-frequency field is scanned. Oneion species may be isolated by ejecting all others from thetrap. The isolated ion species can then be subjected to CID.

The second class of mass analyzers includes the TOFanalyzers that are commonly used with MALDI. The TOFanalysis is based on accelerating a set of ions through aflight tube to a detector with a standard force. All ionshave the same energy but a different mass so the ions reachthe detector at different times. The smaller ions reach thedetector first because of their greater velocity; larger ionstake proportionately longer. Therefore, the analyzer iscalled “time-of-flight,” and mass is determined by the ions’time of arrival. MALDI TOF machines fragment peptidesby postsource decay directly on the MALDI target after thelaser power is significantly increased. However, postsourcedecay does not provide the quality of fragmentation dataprovided by CID, and up to 10 separate spectra of differentmass regions may need to be obtained and combined.MALDI TOF TOF machines have 2 flight tubes separatedby a fragmentation chamber so fragmentation analysis byCID can be done. The high energy CID in MALDI TOFTOF results in side-chain loss fragment ions, which allowthe differentiation of the isomers leucine and isoleucine(Yergey et al., 2002).

The third class of mass analyzers is the Fourier-transform(FT) ion cyclotron resonance (ICR) mass analyzers. FT-ICR has extremely high resolution, sensitivity, and massaccuracy as well as the ability to do MSn. FT-MS is basedon the principle of charged particles orbiting in the presenceof a magnetic field. While the ions are orbiting, a radio-frequency signal is used to excite them. As a result of thisradio-frequency excitation, the ions produce a detectableimage current on the cell in which they are trapped. Thetime-dependent image current is then Fourier-transformedto obtain the component frequencies of the different ions,which correspond to their mass-to-charge ratios. Until nowthe cost of FT-ICR machines was prohibitive for all but afew laboratories. However, competitively priced FT-ICRmachines will be available after August 2003. These ma-chines are coupled to ion traps that further increase sensi-tivity.

Like 2-D PAGE/DIGE, certain proteins are problematicfor MS analysis alone or in combination with HPLC. Mostof the problem is that salts and detergents are incompatiblewith MALDI and ESI. This incompatibility conflicts withtraditional protein isolation protocols. Cell membrane pro-teins are typically difficult to isolate, and these moleculesare of critical importance. However, a great deal of workis ongoing to solve the cell-membrane protein problem.This work is targeted at global expression analysis (Wu etal., 2003) and at identifying specific proteins (Arnott et al.,2002). The latter is suggested to be a replacement for West-

ern blotting of proteins, and for this reason the phrase massWestern analysis was coined.

Other and Emerging Technologies

So far the most common proteomics technologies usedby established proteomics laboratories have been reviewed.It is worthwhile to briefly alert readers to other technologiesused for proteomics research and also emerging technolog-ies that will become important to the chicken community.Generally when technologies that already exist are appliedto proteomics, they are scaled up and are done in highthroughput. An archetypal example is the large-scale yeast2-hybrid screen in the yeast Saccharomyces cerevisiae, whichresulted in the detection of 957 putative interactions thatinvolved 1,004 proteins (Uetz et al., 2000). In more direct“guilt-by-association” approaches, large-scale coprecipita-tion technologies may be used. Two models for this ap-proach, again done in S. cerevisiae, are coprecipitation afterlarge-scale Flag epitope tagging (Ho et al., 2002) and tan-dem affinity purification tagging (Gavin et al., 2002). Tan-dem affinity purification tagging has been used on a smallerscale to analyze multiprotein complexes (Tasto et al., 2001)as have antibodies (Sanders et al., 2002).

Three emerging proteomics technologies are quantitativemethods for electrophoresis-free proteomics, protein/pep-tide arrays, and imaging MS. Quantitative methods forelectrophoresis-free proteomics rely on the sensitive abilityto distinguish small differences in mass. The general princi-pal underlying all of these quantitative MS methods is tolabel, either in vitro or in vivo, the proteins from cells underone set of conditions with heavy isotopes and those fromanother set of conditions with light isotopes. One exampleis ICAT (Gygi et al., 1999). In ICAT experiments, equalmasses of total protein from the 2 conditions to be com-pared are labeled with a heavy or a light tag. These proteinsare then digested with a protease, and the tagged peptidesare isolated using affinity columns. When analyzed by ESIMS, the same peptides from each sample are equally ion-ized. Thus their relative concentration can be directly com-pared. The defined difference in mass allows identificationof each peptide in the mass spectra.

Arrays are now being developed in which thousands ofproteins, peptides, or antibodies are spotted as microarrayson glass or plastic slides. Analogous to cDNA microarrays,protein arrays (1) profile the relative abundance of proteinsin complex mixtures (protein expression profiling), (2) de-tect interactions (antibody-protein, protein-protein, pro-tein-DNA, protein-small molecule), and (3) determineprotein activities. Bound proteins are detected by addingantibodies tagged with fluorescent molecules or by chemi-cally labeling all the proteins in the sample before analysis.The signal intensity represents the amount of bound pro-tein. Protein expression arrays, like their genetic counter-parts, generate signatures for different cell types andtissues. For functional arrays, different proteins are ana-lyzed for the molecules they bind. The proteins in question,rather than the antibodies recognizing them, are arrayedon a slide. Protein binding information can also be obtained

BURGESS562

by using a small molecule microarray, in which thousandsof different small synthetic molecules are arrayed on theslide. These small molecule arrays are primarily drug dis-covery tools but can also be used as research tools to definenovel reagents that interfere with the function of a protein.

Imaging MS is a technique that uses MS directly to imagetissue sections (Stoeckli et al., 1999; Chaurand and Caprioli,2002). Tissue sections, cut as for any histological analysis,are analyzed directly using a MALDI TOF mass spectrome-ter. Imaging MS is the MS equivalent of immunohistochem-istry except that global protein profiles are produced.Although in its infancy, imaging MS has been used formapping and imaging of biomolecules and drugs directlyfrom tissue sections and produces mass spectral signaturesfor tissues in health and disease.

Immunoproteomics

Like everything else in the “omic” world, proteomicswhen applied to immunology and immunity has acquiredits own term, “immunoproteomics” (Jungblut, 2001). Thisterm, however, is recent and is not adequate on its own toidentify the literature concerning the proteomics of immu-nology and immunity in literature searches. A literaturesearch (PubMed) for publications up until the end of 2002used the search specific term in the title or abstract [(pro-teome OR proteomic) AND (immune OR immunology ORimmunity OR vaccine OR leukocyte OR lymphocyte ORcytokine) NOT plant)]. From this search it is clear that therewas a lag time of just 1 yr from the time of first publicationof the term proteome (Wasinger et al., 1995) to its firstpublication in connection with immunology (Qi et al., 1996)(Figure 1B). For both proteomics generally and immuno-proteomics specifically, approximately one-quarter of pub-lished papers are reviews (mean/year = 26.6 and 24.0 %,respectively). Like proteomics in general, an exponentialmodel best fits the increase in immunoproteomics publica-tions. Finally, immunoproteomics and immunogenomics/transcriptomics {search term in title or abstract: [(cDNAOR microarray OR transcriptome OR transcriptomic ORgenomic) AND (immune OR immunology OR immunityOR vaccine OR leukocyte OR lymphocyte OR cytokine)NOT plant)]}, on average, compose ∼4% of the total proteo-mics and genomics/transcriptomics literature. However,compared with immunogenomics/transcriptomics, whichhas stablized at about 4%, immunoproteomics publicationsare increasing exponentially (Figure 1C). All the indicationsare that proteomics will significantly impact immunob-iology.

Immunoproteomics is in its infancy. Many proteomicstools that to date have not been published in an immuno-proteomics context undoubtedly soon will be. As of June2003, there were only 85 primarily immunoproteomics re-search publications (using the above criteria) out of1,485,197 immunology publications [search term in titleor abstract: (immune OR immunology OR immunity ORvaccine OR leukocyte OR lymphocyte OR cytokine)NOT plant].

Proteomics relies on a set of platform technologies andhas wide-ranging applications in many disciplines. In gen-eral, however, proteomics provides at least one of the fol-lowing:1. Definition of genomes by distinguishing putative open

reading frames that actually encode proteins fromthose that do not and by identifying protein speciesthat cannot be predicted from the structure of the ge-nome or transcriptome alone.

2. Maps of protein expression and differential proteinexpression to obtain signatures in, and for, specificcontexts and to establish comparative databases.

3. Definition of molecular machines, interacting proteincomplexes, and so on up to the much larger scale ofquantitative modeling global interacting proteinnetworks.

4. Predictions of molecular structure, utility, or functionbased on experimentally defined knowledge primarilyusing bioinformatic modeling.

The paradigm for this review is that proteomics is valu-able for defining disease pathogenesis and immunity; pro-teomics will provide insights into and suggest novelhypotheses for improving health. The immunoproteomicsliterature can arbitrarily be separated into general groupsthat (1) define immune system function and dysfunctionfrom primary structure, to higher order structure, to aglobal level; (2) define pathogens from primary structure,to higher order structure, to a global functional molecularlevel; (3) identify immunogens and epitopes particularlyin the context of the genetic basis of host disease resistanceand susceptibility and also identify virulence and attenua-tion determinants of pathogens and related vaccine strains;and (4) define the molecular machines used in host immu-nity and pathogen immune evasion. Although there issome intersection, most immunoproteomics publicationspredominantly describe mapping and databases; the nextmost common immunoproteomics publications predomi-nantly describe the functions of molecular machines andhigher multiprotein complexes. Publications that define theimmunogenome and primarily use bioinformatics are leastrepresented. 2-D PAGE and MALDI TOF MS are the mostcommon techniques used to date.

Two papers from the same group, using mouse fetalthymus cDNA libraries, define some of the elastic capacitythe genome provides the immune system. Using in vitrotranslation methods that allow PTM, their approach al-lowed identification of rare protein species that otherwisewould not have been detected by traditional proteomicstechnologies. Multiple protein species were derived fromsingle RNA. These multiple protein species might have bedue to PTM or resulted from programmed degradation.The group concluded “that in most instances expressedgenes yield transcript(s) that translate into several, andoften very numerous families of polypeptide species.” With2-D PAGE, the proteomic profiles of the clonal polypeptidefamilies were mapped in terms of mass, charge, multipleproducts, and appearance in 2D PAGE (Lefkovits et al.,2001; Kettman et al., 2002).


Proteome maps aimed at understanding gene expressionin immune system development, physiology, disease, andpharmaco-therapeutics have been established but are notyet comprehensive. Databases for normal human CD19+B, CD4+ T-helper, and CD8+T-killer lymphocytes wereestablished by Vuadens et al. (2002). Although the B-lym-phocyte proteome segregated from the T-lymphocyte pro-teome, and the CD8+ T-lymphocyte proteome segregatedfrom CD4+ lymphocyte proteome, no spots were B-lym-phocyte specific. The ability of 2-D page to discriminatebetween cells with known differences is encouraging, butthe absence of known B-lymphocyte-specific markers inthe discriminatory proteomes (CD19 itself and the rest ofthe surface immunoglobulin complex) exemplifies one lim-itation of 2-D PAGE. A proteome database of primary hu-man T-helper cells was established using 2-D PAGE andMALDI (Nyman et al., 2001). Although ∼1,500 protein spotswere detected in the resulting 2-D gels with silver staining,and ∼2,000 spots with autoradiography, only 91 were iden-tified by PMF at the time of publication. Joubert-Caron etal. (2000) used in vitro EBV-immortalized lymphocytes toproduce an annotated reference map of human B-lymph-oblastoid cell proteins.

Moving more toward function, murine B-lymphocytesactivated by lipopolysaccharide have been mapped (Freyet al., 2000), as has have murine T lymphocytes activated inthe presence of the immunosuppressive drug cyclosporin A(Mascarell et al., 2000) and also reactivating aftercyclosporin A removal (Truffa-Bachi et al., 2000). B lympho-cytes were shown to utilize sequential waves of function-ally related proteins to anticipate antibody secretion (vanAnken et al., 2003). Proteomics identified caspase activationto be important to the phenomenon of oral tolerance ofCD4 T lymphocytes. In addition to caspase-3 expression,the tolerant CD4 T cells unregulated anti-apoptotic factorsand could not form normal T-cell receptor signaling com-plexes (Kaji et al., 2003). Wang et al. (2002) mapped theeffect of a novel potential immunomodulator on spleno-cytes and identified, among other proteins, upregulation ofthe cytokines interleukin (IL)-1, IL-2, and interferon (IFN)-γ.Formation of mixed disulfides between glutathione andthe cysteines of some proteins (glutathionylation) was sug-gested to be a mechanism of global regulation of proteinfunctions by redox status using T lymphoblasts that un-dergo glutathionylation during oxidative stress. Nonreduc-ing 2-D PAGE was used, and labeled proteins weredetected by phosphorimaging then identified by MS [redoxproteomics (Fratelli et al., 2002)].

Cytokines play critical roles in immunomodulation, andproteomics has revealed ubiquitin-conjugating enzymes tobe regulated by IFN-α. Ubiquitin conjugation is a rate-limiting step in antigen presentation and therefore theupregulation of ubiquitin-conjugating enzymes by IFNmay contribute to the enhanced antigen presentation bymacrophages (Nyman et al., 2000). IFN-γ affects fresh hu-man bladder transitional cell carcinoma culture proteomes,but any potential utility for this cytokine in bladder cancermanagement is still to be defined (Aboagye et al., 1998).

The innate immune system is important for controllingpathogens immediately after infection or even at the timeof pathogen entry. Le Naour et al. (2001) profiled changesin transcriptome and proteome during differentiation andmaturation of monocyte-derived dendritic cells by bothDNA microarrays and proteomics. Novel genes associatedwith dendritic cell differentiation/maturation and PTM ofspecific proteins were identified to be part of the differentia-tion and maturation processes. Notably, there was littleconcordance between cDNA and proteomics results, show-ing that to get a complete understanding of how the systemworks transcriptomics and proteomics were needed. Inwork targeting the role innate immune cells play in theoutcome of disease secondary to pathogen infection, a pro-teomics signature from macrophages predicted HIV-1 wasassociated with cognitive impairment (Luo et al., 2003).Subcellular organelles operate as machines in their ownright. At this subcellular level, the molecules by whichdendritic cell exosomes modulate their adjuventicity weredescribed by Thery et al. (2001), and human butyrophilinin milk fat globules were suggested to provide breast-fedinfants with immune molecules by Cavaletto et al. (2002).Lower vertebrates and insects are used as comparativemodels to identify innate immune molecules and antimi-crobial peptides. Protease inhibitors have been mappedfrom the skin secretions of the crawfish frog, Rana areolata(Ali et al., 2002).

Not all immune responses and immune and nonimmunecell interactions are beneficial. Immunopathologies are im-portant causes of disease. Proteome mapping has impli-cated IFN-γ, IL-1, and IL-6 in the pathogenesis ofinflammatory bowel disease (Barcelo-Batllori et al., 2002)and IFN-γ in human skin aging (Gromov et al., 2003). Thedeleterious effects of IL-1β on islets of Langerhans in arat model of diabetes were mapped at a proteomic level,including identifying changes in PTM (Larsen et al., 2001).During work aimed at identifying the mechanisms under-lying cancer metastasis, a potential role for IL-18 was sug-gested (Jiang et al., 2003). IL-18 was present in a highlymetastatic cell line but absent in a poorly metastatic cellline. Both cell lines were derived from the same parentalcell line. The association of IL-18 with metastasis was sup-ported by IL-18 sense/IL-18 antisense experiments.

Another form of proteomics mapping, but one specificto immunoproteomics, is identification of target antigensin natural immune responses for rational vaccine design(Nilsson, 2002). For antibodies, mapping is done by West-ern blotting 2-D polyacrylamide gels with immune seraand has been called serological proteome analysis (Eschen-brenner et al., 2002). The antigens are identified either afterprotease digestion of spots cut from nonimmunoblottedmatched gels or by protease digestion of spots cut directlyfrom the nitrocellulose/polyvinylidene difluoride mem-brane. Identifying antigens recognized by CD8+ T-killerlymphocytes and CD4+ T-helper lymphocytes by MS waspioneered by Hunt et al. (1992a,b) in the preproteomicsera. The peptides recognized by CD8+ and CD4+ T lym-phocytes bind to MHC classes I and II, respectively. Theconcept behind identifying T-lymphocyte antigens is to

BURGESS564

purify MHC class I or II through immunoaffinity assaysand then to elute the bound peptides. The peptides arethen identified by MSn. Since the sequencing of pathogengenomes, antigen mapping has become more accessible.Proteomic B- and T- lymphocyte antigen mapping has beenused to identify immunogens in allergy (Ou et al., 2001;Beyer et al., 2002; Yu et al., 2003), autoimmunity (Ochi etal., 2002; Robinson et al., 2002; Thebault et al., 2002; Bohringand Krause, 2003; Orru et al., 2003; Stulik et al., 2003),cancer (Prasannan et al., 2000; Brichory et al., 2001; Knezevicet al., 2001; Murphy et al., 2001; Paweletz et al., 2001; Sree-kumar et al., 2001; Vercoutter-Edouart et al., 2001; Benven-uti et al., 2002; Brown and Nazmi, 2002; Celis et al., 2002;Petricoin et al., 2002; Rai et al., 2002), and infectious diseases(Hendrickson et al., 2000; Buranda et al., 2001; Covert etal., 2001; Holtappels et al., 2002; Montigiani et al., 2002;Veith et al., 2002; Cullen et al., 2003; Guina et al., 2003;Yatsuda et al., 2003)

In conjunction with a high-throughput version of a classi-cal immune T-lymphocyte function test, proteomics wasused to map the T-lymphocyte antigens of Mycobacteriumtuberculosis (Covert et al., 2001). Subcellular protein frac-tions of M. tuberculosis were resolved by 2-D liquid phaseelectrophoresis. These fractions were then tested in lym-phocyte stimulation assays using splenocytes from micepreviously infected with M. tuberculosis. Thirty individualimmunogenic proteins contained in the 2-D liquid phaseelectrophoresis fractions were identified by liquid chroma-tography MS and MSn; many represented previously de-fined antigens, but 17 were novel T-lymphocyte antigens.

So far only the identification of antigens that are naturallyimmunogenic (or at least have peptides bound to MHCclass I and II) have been discussed. However, this does notaddress the issue of hidden antigens to which no pre-existing immune response may be detected (Thoren et al.,2002; Yatsuda et al., 2003). The proteome of all pathologiesis rich in potential immunogens, and differential proteomemapping has been done for many pathologies, especiallycancer. Some examples of cancer proteomics that relate toimmunology are identified above. The differential pro-teome mapping of pathogens will be further discussed.The proteomes of many human and some animal patho-gens have been mapped, and many more will be. Not onlywill pathogen proteome maps confirm (and modify) eachgenome’s annotation, but these proteomes will also differ-entiate conserved from stage-specific proteomes. They willalso define internal, membrane-associated and secreted,proteomes. Each proteome is of potential interest for vac-cine design.

The impact of pathogen proteomes will be greater thanjust identifying potential vaccine antigens. Many of themechanisms of pathogen virulence and vaccine attenuationare not understood. Nor are such mechanisms detectableat the genetic level because they are hidden in the “noise”of variation that is in (especially the larger) polymorphicpathogen genomes. Differential proteome analysis hasidentified pathogen virulence and attenuation mechanisms(Ferrari et al., 1999). Specific mechanisms of pathogen im-mune evasion have also been described using proteomics.

Group A Streptococcus has a secreted protein with homol-ogy to the α-subunit of Mac-1, a leukocyte β2 integrinrequired for innate immunity, which inhibits opsonopha-gocytosis and thus evades human innate and acquired im-munity (Lei et al., 2001).

One of the goals of proteomics is to describe the func-tional protein complexes. Proteomics has been used to de-fine the detergent-resistant membrane skeleton fromneutrophil plasma membranes and lipid rafts (Nebl et al.,2002). Lipid rafts are important for localized signal trans-duction. Cascades of adaptor protein and kinase recruit-ment to lipid rafts are critical during receptor activation.In other work, IgM has been shown to be associated withthe Sp-α (CD5 antigen-like) molecule. Sp-α is a memberof the scavenger receptor cysteine-rich superfamily of pro-teins, and other members of the superfamily influence im-mune cell fate. The authors propose a role for Sp-α in IgMhomeostasis (Tissot et al., 2002).

Last, 2 papers exemplify primary bioinformatics ap-proaches to immunoproteomics. Ristori et al. (2000) investi-gated whether molecular mimicry affects the shaping ofthe helper T-cell repertoire. They used an algorithm thatmeasures the probability of mimicry between epitopes ofknown immunogenicity. Self or nonself proteomes werederived from human or microbial data in Swiss-Prot. Theauthors conclude that “mimicry, rather than complicatingself-nonself discrimination, assists in the shaping of theimmune repertoire and helps define the defensive or autor-eactive potential of a T cell.” This model fits with the labora-tory data in which an inflammatory environment is neededto break immune tolerance. The authors claim that, beinga predictor of epitope immunogenicity, their algorithm isrelevant to vaccine design. As part of an iterative researchprocess, this claim is easily testable.

A computational approach was taken in human cancerimmunology aimed to understand the correlation betweenthe immunogenicity of peptides derived from tumor-asso-ciated antigens relative to the similarity level to the host’sproteome (Mittelman et al., 2002). The authors used thebreast and prostate cancer-associated HER-2/neu antigenand computationally defined similarity with the host’s pro-teome. These epitopes were then experimentally tested forreactivity using human polyclonal sera from patients withbreast or prostate cancer. Peptide sequences with mediumor low similarity to the human proteome were preferen-tially recognized by breast and prostate cancer patient sera.Perhaps not surprisingly, the authors concluded that “lowlevel of sequence similarity to the host’s proteome mightbe an important factor in shaping the pool of B cell epi-topes.” The authors claim “new directions for the applica-tion of computational biology to anti-cancer vaccineresearch” and to have provided “experimental strategiesapplicable to the identification of functionally relevant epi-topes within disease-associated-proteins.” Although trueto some degree, the approach outlined takes no account ofIgM or T-cell receptor avidity or affinity to HER2/neu inits natural context, antigen expression levels (important inHER2/neu), the natural availability of CD4+ T lymphocytehelp, and assumes only that central, but not peripheral,


FIGURE 7. Representative 2-dimensional (2-D) PAGE gel from neoplastically transformed T lymphocytes from Marek’s disease lymphoma [A;silver-stained, nonlinear pH range 3 to 11, 11 cm immobilized pH gradient (IPG), linear SDS-PAGE; Criterion, BioRad, www.biorad.com; collaborationwith M. Parcells, University of Arkansas]. The gel shown was produced from the IPG strip in Figure 3. Analysis for differentially expressed spotsbetween two types of Marek’s disease lymphoma, caused by different pathotypes of Marek’s disease herpesvirus (B). Three replicate 2-D PAGEgels were run for each pathotype. Criteria for differential expression was >2-fold difference for the most similar gels, between groups of gels, fromeach pathotype. Differentially expressed spots are marked by “X”. Graphs show absolute expression as [silver-stain intensity/background intensityimmediately adjacent to spot (or corresponding pixels from gels without spots as appropriate)]; units are parts per million (analysis softwarePDQUEST, BioRad, www.biorad.com) Zoom view of dashed boxes (B) showing spots to be cut out for processing for MS analysis (C).

BURGESS566

FIGURE 8. MALDI-TOF peptide mass fingerprint analysis of nondifferentially expressed spot cut from the 2-dimensional (2-D) PAGE gel inFigure 7. The spot was digested in gel by using sequencing-grade modified trypsin (Promega, www.promega.com/default.asp). The MS data (A)is of high resolution as demonstrated by the isotopes of the mono-isotopic peak at 1,243.7760, increasing by almost exactly 1 Da per isotope (inset).Protein identified as (nonmuscle) chicken tropomyosin 1, obtained from database search using the Mascot search engine and database (Perkins etal., 1999). Matching peptides are underlined in sequence.

tolerance is important when establishing primary immu-nity or enhancing secondary immunity. It would be inter-esting to test the authors’ model with autoimmune models(the flipside of tumor immunity) and models of molecularmimicry of host epitopes in infectious disease.

Paradigms for Chicken Immunoproteomics

The background information above leads us to a discus-sion of immunoproteomics in the chicken. To date, withone notable exception from the Beynon group at the Uni-


FIGURE 9. Representative microcapillary liquid chromatography electrospray ionization (ESI) tandem mass spectrometry (MS2) analysis ofbacterial protein (Haemophilus spp.; collaboration with M. Lawrence, Mississippi State University) after trypsin digestion (as above). Data frompool of different proteins analyzed en mass (ProteomeX LCQ DecaXP, Bioworks analysis software, Thermofinnigan, San Jose, CA). (A) After settingappropriate statistical stringencies, the first protein was identified by the first of its peptides (PGIVIGK) at ∼5.8 min of ion current (arrowed intop panel), at scan 169, as ribosomal protein S3 (rpS3, middle and bottom panels). Overall, rpS3 was identified by 6 peptides above thresholdstringency (underlined), i.e., 24.26% coverage (B). The b-y ion series for PGIVIGK is shown (C).

versity of Liverpool Veterinary School, UK (Hayter et al.,2003), no wet-laboratory proteomics work in the chickenhas been published. No chicken immunoproteomics workhas been published at all.

Four points are immediately obvious from immunopro-teomics in other species when considering immunoproteo-

mics in the chicken. First, sequenced pathogen genomes, inaddition to a sequenced chicken genome, are fundamental.Pathogen genome sequencing is relatively well supportedby the USDA and will almost certainly continue to beso. The major impediment to chicken proteomics (i.e., nothaving a chicken genome sequence) will soon be removed

BURGESS568

(Burt and Pourquie, 2003). Second, the proteome is massive,and proteomics has ubiquitous applications. Thus, anyimaginative list of possibilities for immunoproteomicscould never be comprehensive and would be of little practi-cal use in this review. Third, the exponential growth inproteomics provides a wealth of data in other species fromwhich to plan an immunoproteomics strategy for thechicken. Fourth, the evidence from the Human ProteomeOrganization suggests that strategic alliances between in-vestigators on a large scale will be critical to efficientchicken proteomics. The small and closely knit avian immu-nology community may provide an excellent nidus for asimilar organization.

The initial priority for chicken proteomics is to identifythe genes in the genome sequence (Burt and Pourquie,2003), i.e., to do the initial annotation. The value of a ge-nome sequence to a research community is limited by howwell it is annotated (Stein, 2001). The human genome proj-ect model is to first annotate the genome, using humanexperts, based on information in the public domain thento take advantage of data mining and automatic annotationthat has been developed from the human proteomics initia-tive (O’Donovan et al., 2001). Initially, annotation work willrapidly progress because of existing chicken DNA andcDNA sequences as well as expressed sequence tags. Infor-mation available from the published literature on alreadyknown chicken genes, and from obviously homologousgenes from other species, will facilitate initial annotation.After this stage, however, approaches that analyze realproteins will become critical.

Basic proteome maps of cells and tissues in health anddisease and as those of pathogens will be fundamental todefining the chicken genome’s functional capacity. In thechicken, for which the immunology research communityis much smaller than that of the human, it may be logical tocombine basic immunoproteome mapping with differentialproteome mapping to compare health with disease. Theimmune system has inherent plasticity. It is the means ofsense and defense against infectious microorganisms andparasites; immunity has co-evolved with pathogens. Theimmune system must be flexible; no single pathogen cantotally evade immunity and replicate completely un-checked if the host species is to remain extant. If proteomemaps are to help to define the capacity of a genome, thenthe diverse array of pathogens, which elicit the huge yetoften subtle diversity of host immunity, must provide ex-cellent data for a complete immunoproteome.

The chicken is established as an excellent biomedicalmodel for immunology research. From the accidental in-vention of attenuated vaccines by Pasteur to the first de-scription of graft vs. host reactions, the first definitiveassociation of specific MHC haplotype with resistance andsusceptibility to pathogens, the discovery of B cells andinterferon, the first successful vaccines against a cancer,and the first prenatal vaccines, immunology is indebted to

2Arcturus, http://www.arctur.com/.

the chicken (Davison, 2003). The chicken is beginning toprove itself to be uniquely positioned to elucidate criticalsequences in vertebrate genomes (Burt and Pourquie, 2003).

Although an excellent biomedical model, the main objec-tive for chicken immunologists is to improve poultryhealth, production, and welfare. Although chicken immu-nologists almost never care about the individual, and cer-tainly would never treat an individual chicken, the aimsof chicken immunologists are similar to the human immu-nologist. Like humans, chickens live in large dense commu-nities, and, like humans, intensive vaccination is critical inthe poultry industries to control infectious disease. Vaccina-tion is likely to become even more important as the politicalpressure to remove in-feed antibiotic growth promotersincreases. Unfortunately, although poultry vaccines to vi-ruses, bacteria, and even parasites are extremely successful,they also select pathogens to evade the immunity inducedby these vaccines, thereby increasing pathogen virulence.In addition, the chicken industries are afflicted with ubiqui-tous immunosuppressive viruses (e.g., chicken anemia vi-rus, infectious bursal disease virus, Marek’s disease virus,avian reovirus, and a cluster of retroviruses). Immunopro-teomics offers a means of directly identifying novel vaccineimmunogens and potential immunogens as well as under-standing disease pathogenesis.

Not all chickens are hatched immunologically equal. Im-munogenetics is extremely important to chicken produc-tion. There are extreme differences in the abilities of chickengenotypes to mount immune responses to both pathogensand vaccines. Furthermore, chickens do different things.Tens of billions of broiler chickens annually worldwide arekilled at approximately 6 wk of age for meat. Only hun-dreds of elite broiler breeders provide the genetics for thesebroilers. Hundreds of millions of chickens are laying tableeggs worldwide at any time (USDA, National AgriculturalStatistics Service). The broiler and egg industries are deal-ing with completely different genotypes. The most obviousdifferences between broiler and layer chickens is in statureand shape. The chickens in each industry live to differentages under different conditions, eat different food, andsuffer from different diseases. The ancestor breed of alldomestic chickens, the Red Jungle Fowl, will have its ge-nome sequenced first. One-fold genome sequences of repre-sentative broiler and layer genomes are proposed (Burtand Pourquie, 2003). However, broiler and layer chickenshave been intensively selected to survive in different patho-gen environments. Their immunogenetics are different;their immunoproteomes are probably even more different.The existence of inbred chicken lines may simplify proteo-mics analyses as it has genomic analyses.

Chicken Immunoproteomicsat Mississippi State University Collegeof Veterinary Medicine

A unique situation allowed me to begin to develop ouravian immunoproteomics program at Mississippi StateUniversity (MSU) College of Veterinary Medicine (CVM)in January 2002. MSU invested heavily in biotechnology


with a particular emphasis on proteomics. In 2001, theMSU Life Sciences and Biotechnology Institute (LSBI) wasestablished and funded to be a core facility with proteomicsequipment, training, and seed funding. The MississippiFunctional Genomics Network was established and fundedby the National Institutes of Health (NIH). The State ofMississippi is heavily reliant on agriculture for its income,and the poultry industry is the primary income earner.MSU strongly supports Mississippi’s agriculture indus-tries, and one of the missions of the CVM is to develop aninternationally recognized poultry research program. Atthe same time, rumors of NIH funding the sequencing ofthe chicken genome were circulating. My personal back-ground was in genetic resistance to poultry disease. Proteo-mics in the chicken was a natural fit.

When starting with an empty laboratory, the biggesthurdle to entering proteomics is money. The MSU invest-ment to establish the LSBI has helped greatly. The nextmost significant problem is finding proteomicists. Suchpeople are a new breed and certainly do not exist for animalresearch. We are training ourselves. This has its advantagesin that we can try to emulate the best, and we can dowhatever we want with the equipment. However, thelearning curves are extremely steep. The first stage of theproteomics arm of my research program was to decideon technologies. We are taking both electrophoretic andnonelectrophoretic approaches, using both MALDI-TOFand microcapillary liquid chromatography ESI MSn. Thereis a very large amount of competition in the rapidly devel-oping proteomics marketplace, which can lead to a lot ofconfusion regarding technologies and instrumentation.

Sample preparation is the single most critical issue inproteomics. We first optimized sample preparation, whichsounds straightforward, but the only way to optimize sam-ple preparation is to use the proteomics equipment, whichalso needs to be optimized at the same time. Because ourparadigm is to work directly ex vivo when possible, whichmeans using less tissue than most proteomics experiment-ers use, we next scaled down. Also tissues are heteroge-neous in vivo; to enable us to analyze pure proteomesdirectly ex vivo, CVM purchased a laser capture microdis-section microscope2, (Figure 6, representative example oflaser capture microdissected tissue). Since starting our wet-lab experiments in May of 2002, we have optimized samplepreparation, 2-D PAGE using MS-compatible silver (Figure7) and Coomassie brilliant blue stains, MALDI-TOF PMF(Figure 8), and liquid chromatrography ESI MSn (Figure9). We have optimized methods for various chicken cellsand tissues and bacteria including methods for analyzingmembrane proteins (data not shown; collaborations withB. Gao, M. Lawrence, MSU; and G. T. Pharr, MSU). Wehave had to learn a great deal about computer programsfor interpreting our 2-D PAGE data (Figure 7), how to setup a MALDI-TOF, offline and inline 2-D microcapillarychromatrography systems, and how to interpret both rawMS spectra. We are now beginning our real experiments.

ConclusionsProteomics is exciting. I was extremely fortunate to be

selected as a student at the inaugural Proteomics course at

Cold Spring Harbor Laboratories (CSHL) in November2002. I have never been in such a high-energy learningenvironment as this course. The main things I learned atCSHL are that there is no single definition of proteomics;the future of proteomics involves extremely high technol-ogy equipment, which has a very fast rate of obsolescence;proteomics requires multidisciplinary collaboration andlarge amounts of money; trained proteomicists will becomemore common, but at this time they are very rare.

Specifically for chicken immunoproteomics we need toattract highly trained and competent scientists. Proteomicsalso needs to become accessible to scientists outside theMS world. It is critical that chicken immunoproteomicsgrant application and manuscript reviewers are comfort-able with the size and scope of proteomes as well as proteo-mics techniques and technologies and the capabilities andlimitations of these.

Finally, proteomics is simply an organized collection ofplatform technologies with a definite goal of defining andunderstanding proteomes. The biological questions remainfundamentally important. Proteomics merely provides aset of tools to use for solving these biological questions.

ACKNOWLEDGMENTS

Much of what is in this review was learned at the inaugu-ral proteomics course at CSHL (New York, November2002). I am grateful to have been in the company of theother 15 students from around the world on that 15-d (andnight!) course. I am indebted to the instructors P. Andrews(University of Michigan), A. Link (Vanderbilt University),and J. LaBaer (Harvard University); the teaching assistantsP. Braun (Harvard University), S. Vazquez (University ofMichigan), and J. Jennings (Vanderbilt University); andall of the visiting instructors for catapulting me into theproteomics world. I am also thankful to J. McPherson(Washington University), M. Karpuj (UCSD), and S. Zhang(Columbia University) who were my lab group paddrtnersand were both inspirational and educational in many ways.My entry into proteomics has been facilitated by the CVMand the LSBI at MSU. I thank particularly A. Wood andT. Pechan (MSU) for helpful discussion and also the formerfor paying for me to attend the CSHL proteomics course.I thank A. Shack (MSU) who has worked extremely hardto perfect 2-D PAGE; we are extremely collaborative. Inthe 18 mo that I have been at MSU, a number of collabora-tions have evolved. Some of that data shown and men-tioned are a result of these collaborations. The followingpeople are collaborators who have material in this reviewor have contributed specific suggestions for poultry immu-noproteomics: B. Gao (MSU), L. Hansen (MSU), B. Holmes(MSU), M. Kidd (MSU), M. Lawrence (MSU), M. Parcells(University of Arkansas), T. Pharr (MSU), and C. Weaver(MSU). Finally, I thank J. Ainsworth, J. Burgess, D. McGee,and M. Lawrence (MSU) for editorial input. This work hasbeen supported by seed grants from MSU, CVM, and LSBIand by the USDA. MAFES journal number J10381.

BURGESS570

ACKNOWLEDGMENTS

Much of what is in this review was learned at the inaugu-ral Proteomics course at Cold Spring Harbor Laboratories,New York, November, 2002. I am grateful to have been inthe company of the other 15 students from around theworld on that 15 day (and night!) course. I am indebtedto the instructors P. Andrews (University of Michigan),A. Link (Vanderbilt University) and J. LaBaer (HarvardUnivesity); the teaching assistants P. Braun (Harvard Uni-versity), S. Vazquez (University of Michigan) and Ms. J.Jennings (Vanderbilt University), and all of the visitinginstructors for catapulting me into the proteomics world.I am also thankful to J. McPherson (Washington Univer-sity), M. Karpuj (UCSD) and S. Zhang (Columbia Univer-sity) who were my lab group partners and were bothinspirational and educational in many ways. My entry intoproteomics has been facilitated by the CVM and the LSBI,Mississippi State University. I thank particularly A. Woodand T. Pecham (MSU) for helpful discussion and also theformer for paying for me to attend the CSHL proteomicscourse. I thank A. Shack (MSU) who has worked extremelyhard to perfect 2-D PAGE. We are extremely collaborative.In the 18 months that I have been at MSU a number ofcollaborations have evolved. Some of that data shown andmentioned is a result of these collaborations. The followingpeople are collaborators who have material in this review orhave contributed specific suggestions for poultry immuno-proteomics: B. Gao (MSU), L. Hansen (MSU), B. Holmes(MSU), M. Kidd (MSU), M. Lawrence (MSU), M. Parcells(University of Arkansas), T. Pharr (MSU) and C. Weaver(MSU). Finally, I thank J. Ainsworth (MSU), J. Burgess, D.McGee (MSU) and M. Lawrence (MSU) for editorial input.This work has been supported by seed grants from MSU,CVM, LSBI and by the USDA. MAFES journal number:J10381

REFERENCES

Aboagye, E. O., A. B. Kelson, M. Tracy, and P. Workman. 1998.Preclinical development and current status of the fluorinated2-nitroimidazole hypoxia probe N-(2-hydroxy-3,3,3-trifluoro-propyl)-2-(2-nitro-1-imidazolyl) acetamide (SR 4554, CRC 94/17): A non-invasive diagnostic probe for the measurementof tumor hypoxia by magnetic resonance spectroscopy andimaging, and by positron emission tomography. AnticancerDrug Des. 13:703–730.

Aebersold, R., L. E. Hood, and J. D. Watts. 2000. Equipping scien-tists for the new biology. Nat. Biotechnol. 18:359.

Ali, M. F., K. R. Lips, F. C. Knoop, B. Fritzsch, C. Miller, and J.M. Conlon. 2002. Antimicrobial peptides and protease inhibi-tors in the skin secretions of the crawfish frog, Rana areolata.Biochim. Biophys. Acta 1601:55–63.

Anderson, N. L., and N. G. Anderson. 1998. Proteome and proteo-mics: new technologies, new concepts, and new words. Electro-phoresis 19:1853–1861.

Arnott, D., A. Kishiyama, E. A. Luis, S. G. Ludlum, J. C. Marsters,Jr., and J. T. Stults. 2002. Selective detection of membraneproteins without antibodies: a mass spectrometric version ofthe Western blot. Mol. Cell Proteomics 1:148–156.

Bairoch, A., and R. Apweiler. 1997. The SWISS-PROT proteinsequence database: its relevance to human molecular medicalresearch. J. Mol. Med. 75:312–316.

Barcelo-Batllori, S., M. Andre, C. Servis, N. Levy, O. Takikawa,P. Michetti, M. Reymond, and E. Felley-Bosco. 2002. Proteomicanalysis of cytokine induced proteins in human intestinal epi-thelial cells: Implications for inflammatory bowel diseases. Pro-teomics 2:551–560.

Beck, S., D. Geraghty, H. Inoko, L. Rowen, et al. 1999. Completesequence, and gene map of a human major histocompatibilitycomplex. The MHC sequencing consortium. Nature 401:921–923.

Benvenuti, S., R. Cramer, J. Bruce, M. D. Waterfield, and P. S. Jat.2002. Identification of novel candidates for replicative senes-cence by functional proteomics. Oncogene 21:4403–4413.

Beyer, K., L. Bardina, G. Grishina, and H. A. Sampson. 2002.Identification of sesame seed allergens by 2-dimensional pro-teomics and Edman sequencing: seed storage proteins as com-mon food allergens. J. Allergy Clin. Immunol. 110:154–159.

Bhanot, G., Y. Louzoun, J. Zhu, and C. DeLisi. 2003. The impor-tance of thermodynamic equilibrium for high throughput geneexpression arrays. Biophys. J. 84:124–135.

Bjellqvist, B., K. Ek, P. G. Righetti, E. Gianazza, A. Gorg, R. West-ermeier, and W. Postel. 1982. Isoelectric focusing in immobi-lized pH gradients: principle, methodology and someapplications. J. Biochem. Biophys. Methods 6:317–339.

Bohring, C., and W. Krause. 2003. Immune infertility: towards abetter understanding of sperm (auto)-immunity: The value ofproteomic analysis. Hum. Reprod. 18:915–924.

Brazma, A., P. Hingamp, J. Quackenbush, G. Sherlock, et al. 2001.Minimum information about a microarray experiment (MI-AME)-toward standards for microarray data. Nat. Genet.29:365–371.

Brichory, F., D. Beer, F. Le Naour, T. Giordano, and S. Hanash.2001. Proteomics-based identification of protein gene product9.5 as a tumor antigen that induces a humoral immune re-sponse in lung cancer. Cancer Res. 61:7908–7912.

Brichory, F. M., D. E. Misek, A. M. Yim, M. C. Krause, T. J.Giordano, D. G. Beer, and S. M. Hanash. 2001. An immuneresponse manifested by the common occurrence of annexinsI and II autoantibodies and high circulating levels of IL-6 inlung cancer. Proc. Natl. Acad. Sci. USA 98:9824–9829.

Brown, R. E., and R. K. Nazmi. 2002. The Reed-Steinberg cell:molecular characterization by proteomic analysis with thera-peutic implications. Ann. Clin. Lab. Sci. 32:339–351.

Buranda, T., G. P. Lopez, P. Simons, A. Pastuszyn, and L. A. Sklar.2001. Detection of epitope-tagged proteins in flow cytometry:fluorescence resonance energy transfer-based assays on beadswith femtomole resolution. Anal. Biochem. 298:151–162.

Burt, D., and O. Pourquie. 2003. Genetics. Chicken genome—Science nuggets to come soon. Science 300:1669.

Burt, D., and O. Pourquie. 2003. International Chicken GenomeWorkshop. Final Report. Cambridge, Hinxton Genome Cam-pus.http://www.chicken-genome.org/events/icgwr2003.html. Accessed July 2003.

Cavaletto, M., M. G. Giuffrida, D. Fortunato, L. Gardano, G. Della-valle, L. Napolitano, C. Giunta, E. Bertino, C. Fabris, and A.Conti. 2002. A proteomic approach to evaluate the butyrophilingene family expression in human milk fat globule membrane.Proteomics 2:850–856.

Celis, J. E., P. Celis, H. Palsdottir, M. Ostergaard, P. Gromov, H.Primdahl, T. F. Orntoft, H. Wolf, A. Celis, and I. Gromova.2002. Proteomic strategies to reveal tumor heterogeneityamong urothelial papillomas. Mol. Cell Proteomics 1:269–279.

Chance, M. R., A. R. Bresnick, S. K. Burley, J. S. Jiang, C. D. Lima,A. Sali, S. C. Almo, J. B. Bonanno, J. A. Buglino, S. Boulton,H. Chen, N. Eswar, G. He, R. Huang, V. Ilyin, L. McMahan,U. Pieper, S. Ray, M. Vidal, and L. K. Wang. 2002. Structuralgenomics: A pipeline for providing structures for the biologist.Protein Sci. 11:723–738.

Chaurand, P., and R. M. Caprioli. 2002. Direct profiling and im-aging of peptides and proteins from mammalian cells andtissue sections by mass spectrometry. Electrophoresis23:3125–3135.


Covert, B. A., J. S. Spencer, I. M. Orme, and J. T. Belisle. 2001.The application of proteomics in defining the T cell antigensof Mycobacterium tuberculosis. Proteomics 1:574–586.

Cullen, P. A., D. A. Haake, D. M. Bulach, R. L. Zuerner, and B.Adler. 2003. LipL21 is a novel surface-exposed lipoprotein ofpathogenic Leptospira species. Infect. Immun. 71:2414–2421.

Davison, T. F. 2003. The immunologists’ debt to the chicken. Br.Poult Sci. 44:6–21.

Eschenbrenner, M., M. A. Wagner, T. A. Horn, J. A. Kraycer, C.V. Mujer, S. Hagius, P. Elzer, and V. G. DelVecchio. 2002.Comparative proteome analysis of Brucella melitensis vaccinestrain Rev 1 and a virulent strain, 16M. J. Bacteriol.184:4962–4970.

Ferrari, G., H. Langen, M. Naito, and J. Pieters. 1999. A coatprotein on phagosomes involved in the intracellular survivalof mycobacteria. Cell 97:435–447.

Figeys, D. 2003. Proteomics in 2002: A year of technical develop-ment and wide-ranging applications. Anal. Chem. 75:2891–2905.

Fratelli, M., H. Demol, M. Puype, S. Casagrande, et al. 2002.Identification by redox proteomics of glutathionylated proteinsin oxidatively stressed human T lymphocytes. Proc. Natl.Acad. Sci. USA 99:3505–3510.

Fredriksson, S., M. Gullberg, J. Jarvius, C. Olsson, K. Pietras, S.M. Gustafsdottir, A. Ostman, and U. Landegren. 2002. Proteindetection using proximity-dependent DNA ligation assays.Nat. Biotechnol. 20:473–477.

Frey, J. R., M. Fountoulakis, and I. Lefkovits. 2000. Proteomeanalysis of activated murine B-lymphocytes. Electrophoresis21:3730–3739.

Gavin, A. C., M. Bosche, R. Krause, P. Grandi, et al. 2002. Func-tional organization of the yeast proteome by systematic analy-sis of protein complexes. Nature 415:141–147.

Gobel, T. W. F. 1996. The T-dependent immune system. Pages31–46 in Poultry Science Symposium Series. T. F. Davison, T. R.Morris and L. N. Payne, ed. Carfax Publishing, Abingdon, UK.

Goodlet, D. 2003. Correlation of mRNA and protein expression.Pages 58–63 in Analyzing Gene Expression: a Handbook ofMethods: Possibilities and Pitfalls. S. Lorkowski and P. Cullen,ed. Wiley-VCH Weinheim, New York.

Gooley, A., and N. Packer. 1997. Post-translational modifications.Pages 65–91 in Proteome Research: New Frontiers in Func-tional Genomics. Principles and Practice. M. R. Wilkins, K. L.Williams, R. D. Appel, and D. F. Hochstrasser, ed.. Springer,New York.

Gras, R., and M. Muller. 2001. Computational aspects of proteinidentification by mass spectrometry. Curr. Opin. Mol. Ther.3:526–532.

Gromov, P., G. L. Skovgaard, H. Palsdottir, I. Gromova, M. Oster-gaard, and J. E. Celis. 2003. Protein profiling of the humanepidermis from the elderly reveals up-regulation of a signatureof interferon-γ-induced polypeptides that includes manga-nese-superoxide dismutase and the p85 β-subunit of phospha-tidylinositol 3-kinase. Mol. Cell Proteomics 2:70–84.

Guina, T., S. O. Purvine, E. C. Yi, J. Eng, D. R. Goodlett, R.Aebersold, and S. I. Miller. 2003. Quantitative proteomic analy-sis indicates increased synthesis of a quinolone by Pseudomo-nas aeruginosa isolates from cystic fibrosis airways. Proc. Natl.Acad. Sci. USA 100:2771–2776.

Gullberg, M., S. Fredriksson, M. Taussig, J. Jarvius, S. Gustafsdot-tir, and U. Landegren. 2003. A sense of closeness: proteindetection by proximity ligation. Curr. Opin. Biotechnol.14:82–86.

Gygi, S. P., B. Rist, S. A. Gerber, F. Turecek, M. H. Gelb, andR. Aebersold. 1999. Quantitative analysis of complex proteinmixtures using isotope-coded affinity tags. Nat. Biotechnol.17:994–999.

Hanash, S., and J. E. Celis. 2002. The human proteome organiza-tion: a mission to advance proteome knowledge. Mol. CellProteomics 1:413–414.

Hayter, J. R., D. H. Robertson, S. J. Gaskell, and R. J. Beynon.2003. Proteome analysis of intact proteins in complex mixtures.Mol. Cell Proteomics 2:85–95.

Hendrickson, R. C., J. F. Douglass, L. D. Reynolds, P. D. McNeill,D. Carter, S. G. Reed, and R. L. Houghton. 2000. Mass spectro-metric identification of mtb81, a novel serological marker fortuberculosis. J. Clin. Microbiol. 38:2354–2361.

Ho, Y., A. Gruhler, A. Heilbut, G. D. Bader, et al. 2002. Systematicidentification of protein complexes in Saccharomyces cerevisiaeby mass spectrometry. Nature 415:180–183.

Holtappels, R., D. Thomas, J. Podlech, and M. J. Reddehase. 2002.Two antigenic peptides from genes m123 and m164 of murinecytomegalovirus quantitatively dominate CD8 T-cell memoryin the H-2d haplotype. J. Virol. 76:151–164.

Huber, L. A. 2003. Is proteomics heading in the wrong direction?Nat. Rev. Mol. Cell Biol. 4:74–80.

Huerta, M., G. Downing, F. Haseltine, B. Seto, and Y. Liu. 2000.NIH working definition of bioinformatics and computationalbiology . Bioinformatics Definition Committee, NIH, Washing-ton DC.

Hunt, D. F., R. A. Henderson, J. Shabanowitz, K. Sakaguchi, H.Michel, N. Sevilir, A. L. Cox, E. Appella, and V. H. Engelhard.1992a. Characterization of peptides bound to the class I MHCmolecule HLA-A2.1 by mass spectrometry. Science 255:261–1263.

Hunt, D. F., H. Michel, T. A. Dickinson, J. Shabanowitz, A. L.Cox, K. Sakaguchi, E. Appella, H. M. Grey, and A. Sette. 1992b.Peptides presented to the immune system by the murine classII major histocompatibility complex molecule I-Ad. Science256:1817–1820.

Jiang, D., W. Ying, Y. Lu, J. Wan, Y. Zhai, W. Liu, Y. Zhu, Z. Qiu,X. Qian, and F. He. 2003. Identification of metastasis-associatedproteins by proteomic analysis and functional exploration ofinterleukin-18 in metastasis. Proteomics 3:724–737.

Joubert-Caron, R., J. P. Le Caer, F. Montandon, F. Poirier, M.Pontet, N. Imam, J. Feuillard, D. Bladier, J. Rossier, and M.Caron. 2000. Protein analysis by mass spectrometry and se-quence database searching: a proteomic approach to identifyhuman lymphoblastoid cell line proteins. Electrophoresis21:2566–2575.

Jungblut, P. R. 2001. Proteome analysis of bacterial pathogens.Microbes Infect. 3:831–840.

Kaiser, J. 2002. Proteomics. Public-private group maps out initia-tives. Science 296:827.

Kaji, T., S. Hachimura, W. Ise, and S. Kaminogawa. 2003. Proteomeanalysis reveals caspase activation in hyporesponsive CD4 Tlymphocytes induced in vivo by the oral administration ofantigen. J. Biol. Chem.

Kaufman, J., S. Milne, T. W. Gobel, B. A. Walker, J. P. Jacob, C.Auffray, R. Zoorob, and S. Beck. 1999. The chicken B locus isa minimal essential major histocompatibility complex. Nature401:923–925.

Kettman, J. R., C. Coleclough, J. R. Frey, and I. Lefkovits. 2002.Clonal proteomics: one gene-family of proteins. Proteomics2:624–631.

Knezevic, V., C. Leethanakul, V. E. Bichsel, J. M. Worth, V. V.Prabhu, J. S. Gutkind, L. A. Liotta, P. J. Munson, E. F. Petricoi-n,III, and D. B. Krizman. 2001. Proteomic profiling of the cancermicroenvironment by antibody arrays. Proteomics 1:1271–1278.

Lander, E. S., L. M. Linton, B. Birren, C. Nusbaum, et al. 2001.Initial sequencing and analysis of the human genome. Nature409:860–921.

Larsen, P. M., S. J. Fey, M. R. Larsen, A. Nawrocki, H. U. Andersen,H. Kahler, C. Heilmann, M. C. Voss, P. Roepstorff, F. Pociot,A. E. Karlsen, and J. Nerup. 2001. Proteome analysis of interleu-kin-1 β -induced changes in protein expression in rat islets ofLangerhans. Diabetes 50:1056–1063.

Lefkovits, I., J. R. Kettman, and J. R. Frey. 2001. Proteomic analysisof rare molecular species of translated polypeptides from amouse fetal thymus cDNA library. Proteomics 1:560–573.

BURGESS572

Lei, B., F. R. DeLeo, N. P. Hoe, M. R. Graham, S. M. Mackie, R.L. Cole, M. Liu, H. R. Hill, D. E. Low, M. J. Federle, J. R. Scott,and J. M. Musser. 2001. Evasion of human innate and acquiredimmunity by a bacterial homolog of CD11b that inhibits opso-nophagocytosis. Nat. Med. 7:1298–1305.

Le Naour, F., L. Hohenkirk, A. Grolleau, D. E. Misek, P. Lescure,J. D. Geiger, S. Hanash, and L. Beretta. 2001. Profiling changesin gene expression during differentiation and maturation ofmonocyte-derived dendritic cells using both oligonucleotidemicroarrays and proteomics. J. Biol. Chem. 276:17920–17931.

Link, A. J., J. Eng, D. M. Schieltz, E. Carmack, G. J. Mize, D. R.Morris, B. M. Garvik, and J. R. Yates, 3rd. 1999. Direct analysisof protein complexes using mass spectrometry. Nat. Biotech-nol. 17:676–682.

Luo, X., K. A. Carlson, V. Wojna, R. Mayo, T. M. Biskup, J. Stoner,J. Anderson, H. E. Gendelman, and L. M. Melendez. 2003.Macrophage proteomic fingerprinting predicts HIV-1-associ-ated cognitive impairment. Neurology 60:1931–1937.

MacGillivray, A. J., and D. Rickwood. 1974. The heterogeneity ofmouse-chromatin nonhistone proteins as evidenced by two-dimensional polyacrylamide-gel electrophoresis and ion-ex-change chromatography. Eur. J. Biochem. 41:181–190.

Mascarell, L., J. R. Frey, F. Michel, I. Lefkovits, and P. Truffa-Bachi. 2000. Increased protein synthesis after T cell activationin presence of cyclosporin A. Transplantation 70:340–348.

Mittelman, A., A. Lucchese, A. A. Sinha, and D. Kanduc. 2002.Monoclonal and polyclonal humoral immune response to ECHER-2/NEU peptides with low similarity to the host’s pro-teome. Int. J. Cancer 98:741–747.

Montigiani, S., F. Falugi, M. Scarselli, O. Finco, R. Petracca, G.Galli, M. Mariani, R. Manetti, M. Agnusdei, R. Cevenini, M.Donati, R. Nogarotto, N. Norais, I. Garaguso, S. Nuti, G. Saletti,D. Rosa, G. Ratti, and G. Grandi. 2002. Genomic approach foranalysis of surface proteins in Chlamydia pneumoniae. Infect.Immun. 70:368–379.

Murphy, K. J., K. R. Nielson, and K. H. Albertine. 2001. Defininga molecularly normal colon. J. Histochem. Cytochem.49:667–668.

Nebl, T., K. N. Pestonjamasp, J. D. Leszyk, J. L. Crowley, S. W.Oh, and E. J. Luna. 2002. Proteomic analysis of a detergent-resistant membrane skeleton from neutrophil plasma mem-branes. J. Biol. Chem. 277:43399–43409.

Nilsson, C. L. 2002. Bacterial proteomics and vaccine develop-ment. Am. J. Pharmacogenomics 2:59–65.

Nyman, T. A., S. Matikainen, T. Sareneva, I. Julkunen, and N.Kalkkinen. 2000. Proteome analysis reveals ubiquitin-conjugat-ing enzymes to be a new family of interferon- a-regulatedgenes. Eur. J. Biochem. 267:4011–4019.

Nyman, T. A., A. Rosengren, S. Syyrakki, T. P. Pellinen, K. Rauta-joki, and R. Lahesmaa. 2001. A proteome database of humanprimary T helper cells. Electrophoresis 22:4375–4382.

Ochi, H., I. Horiuchi, N. Araki, T. Toda, T. Araki, K. Sato, H.Murai, M. Osoegawa, T. Yamada, K. Okamura, T. Ogino, K.Mizumoto, H. Yamashita, H. Saya, and J. Kira. 2002. Proteomicanalysis of human brain identifies a-enolase as a novel autoan-tigen in Hashimoto’s encephalopathy. FEBS Lett. 528:197–202.

O’Donovan, C., R. Apweiler, and A. Bairoch. 2001. The humanproteomics initiative (HPI). Trends Biotechnol. 19:178–181.

Orru, S., I. Caputo, A. D’Amato, M. Ruoppolo, and C. Esposito.2003. Proteomics identification of Acyl-acceptor and Acyl-do-nor substrates for transglutaminase in a human intestinal epi-thelial cell line: implications for coeliac disease. J. Biol. Chem.

Ou, K., T. K. Seow, R. C. Liang, B. W. Lee, D. L. Goh, K. Y. Chua,and M. C. Chung. 2001. Identification of a serine proteaseinhibitor homologue in Bird’s Nest by an integrated proteo-mics approach. Electrophoresis 22:3589–3595.

Pardanani, A., E. D. Wieben, T. C. Spelsberg, and A. Tefferi. 2002.Primer on medical genomics. Part IV: Expression proteomics.Mayo Clin. Proc. 77:1185–1196.

Patterson, S. D., and R. H. Aebersold. 2003. Proteomics: the firstdecade and beyond. Nat. Genet. 33(Suppl.):311–323.

Paweletz, C. P., L. A. Liotta, and E. F. Petricoin, 3rd. 2001. Newtechnologies for biomarker analysis of prostate cancer progres-sion: Laser capture microdissection and tissue proteomics.Urology 57:160–163.

Perkins, D. N., D. J. Pappin, D. M. Creasy, and J. S. Cottrell. 1999.Probability-based protein identification by searching sequencedatabases using mass spectrometry data. Electrophoresis20:3551–3567.

Petricoin, E. F., A. M. Ardekani, B. A. Hitt, P. J. Levine, V. A.Fusaro, S. M. Steinberg, G. B. Mills, C. Simone, D. A. Fishman,E. C. Kohn, and L. A. Liotta. 2002. Use of proteomic patternsin serum to identify ovarian cancer. Lancet 359:572–577.

Petricoin, E. F., III, D. K. Ornstein, C. P. Paweletz, A. Ardekani,P. S. Hackett, B. A. Hitt, A. Velassco, C. Trucco, L. Wiegand,K. Wood, C. B. Simone, P. J. Levine, W. M. Linehan, M. R.Emmert-Buck, S. M. Steinberg, E. C. Kohn, and L. A. Liotta.2002. Serum proteomic patterns for detection of prostate can-cer. J. Natl. Cancer Inst. 94:1576–1578.

Prasannan, L., D. E. Misek, R. Hinderer, J. Michon, J. D. Geiger,and S. M. Hanash. 2000. Identification of β-tubulin isoformsas tumor antigens in neuroblastoma. Clin. Cancer Res.6:3949–3956.

Qi, S. Y., A. Moir, and C. D. O’Connor. 1996. Proteome of Salmo-nella typhimurium SL1344: identification of novel abundantcell envelope proteins and assignment to a two-dimensionalreference map. J. Bacteriol. 178:5032–5038.

Rabilloud, T. 2002. Two-dimensional gel electrophoresis in proteo-mics: old, old fashioned, but it still climbs up the mountains.Proteomics 2:3–10.

Rai, A. J., Z. Zhang, J. Rosenzweig, M. Shih Ie, T. Pham, E. T.Fung, L. J. Sokoll, and D. W. Chan. 2002. Proteomic approachesto tumor marker discovery. Arch. Pathol. Lab. Med.126:1518–1526.

Reitman, M., J. A. Grasso, R. Blumenthal, and P. Lewit. 1993.Primary sequence, evolution, and repetitive elements of theGallus gallus (chicken) β-globin cluster. Genomics 18:616–626.

Ristori, G., M. Salvetti, G. Pesole, M. Attimonelli, C. Buttinelli, R.Martin, and P. Riccio. 2000. Compositional bias and mimicrytoward the nonself proteome in immunodominant T cell epi-topes of self and nonself antigens. FASEB J. 14:431–438.

Robinson, W. H., L. Steinman, and P. J. Utz. 2002. Protein andpeptide array analysis of autoimmune disease. Biotechniques(Suppl.):66–69.

Sanders, S. L., J. Jennings, A. Canutescu, A. J. Link, and P. A. Weil.2002. Proteomics of the eukaryotic transcription machinery:Identification of proteins associated with components of yeastTFIID by multidimensional mass spectrometry. Mol. Cell Biol.22:4723–4738.

Shen, Y., R. Zhao, S. J. Berger, G. A. Anderson, N. Rodriguez,and R. D. Smith. 2002. High-efficiency nanoscale liquid chro-matography coupled on-line with mass spectrometry usingnanoelectrospray ionization for proteomics. Anal. Chem.74:4235–4249.

Sreekumar, A., M. K. Nyati, S. Varambally, T. R. Barrette, D.Ghosh, T. S. Lawrence, and A. M. Chinnaiyan. 2001. Profilingof cancer cells using protein microarrays: Discovery of novelradiation-regulated proteins. Cancer Res. 61:7585–7593.

Stein, L. 2001. Genome annotation: from sequence to biology. Nat.Rev. Genet. 2:493–503.

Stoeckli, M., T. B. Farmer, and R. M. Caprioli. 1999. Automatedmass spectrometry imaging with a matrix-assisted laser de-sorption ionization time-of-flight instrument. J. Am. Soc. MassSpectrom. 10:67–71.

Stulik, J., L. Hernychova, S. Porkertova, O. Pozler, L. Tuckova,D. Sanchez, and J. Bures. 2003. Identification of new celiacdisease autoantigens using proteomic analysis. Proteomics3:951–956.

Su, C., I. Jakobsen, X. Gu, and M. Nei. 1999. Diversity and evolu-tion of T-cell receptor variable region genes in mammals andbirds. Immunogenetics 50:301–308.


Tasto, J. J., R. H. Carnahan, W. H. McDonald, and K. L. Gould.2001. Vectors and gene targeting modules for tandem affinitypurification in Schizosaccharomyces pombe. Yeast 18:657–662.

Taylor, C. F., N. W. Paton, K. L. Garwood, P. D. Kirby, et al. 2003.A systematic approach to modeling, capturing, and dissemi-nating proteomics experimental data. Nat. Biotechnol.21:247–254.

Thebault, S., D. Gilbert, M. Hubert, L. Drouot, N. Machour, C.Lange, R. Charlionet, and F. Tron. 2002. Orderly pattern ofdevelopment of the autoantibody response in (New ZealandWhite x BXSB)F1 lupus mice: Characterization of target anti-gens and antigen spreading by two-dimensional gel electro-phoresis and mass spectrometry. J. Immunol. 169:4046–4053.

Thery, C., M. Boussac, P. Veron, P. Ricciardi-Castagnoli, G. Ra-poso, J. Garin, and S. Amigorena. 2001. Proteomic analysis ofdendritic cell-derived exosomes: a secreted subcellular com-partment distinct from apoptotic vesicles. J. Immunol.166:7309–7318.

Thoren, K., E. Gustafsson, A. Clevnert, T. Larsson, J. Bergstrom,and C. L. Nilsson. 2002. Proteomic study of non-typableHaemophilus influenzae. J. Chromatogr. B Analyt. Technol.Biomed. Life Sci. 782:219–226.

Tissot, J. D., J. C. Sanchez, F. Vuadens, A. Scherl, J. A. Schifferli,D. F. Hochstrasser, P. Schneider, and M. A. Duchosal. 2002.IgM are associated to Sp-α (CD5 antigen-like). Electrophoresis23:1203–1206.

Truffa-Bachi, P., I. I. Lefkovits, and J. Rudolf Frey. 2000. Proteomicanalysis of T cell activation in the presence of cyclosporin A:immunosuppressor and activator removal induces de novoprotein synthesis. Mol. Immunol. 37:261.

Uetz, P., L. Giot, G. Cagney, T. A. Mansfield, et al. 2000. A compre-hensive analysis of protein-protein interactions in Saccharo-myces cerevisiae. Nature 403:623–627.

Ullrich, B., Y. A. Ushkaryov, and T. C. Sudhof. 1995. Cartographyof neurexins: more than 1000 isoforms generated by alternativesplicing and expressed in distinct subsets of neurons. Neuron14:497–507.

Unlu, M., M. E. Morgan, and J. S. Minden. 1997. Difference gelelectrophoresis: a single gel method for detecting changes inprotein extracts. Electrophoresis 18:2071–2077.

Various. 2003. Computational methods and bioinformationaltools. Pages 769–904 in Analysing Gene Expression: a Hand-book of Methods, Possibilities, and Pitfalls. Vol. 2. S. Lorkowskiand P. Cullen, ed. S. Wiley-VCH, New York.

van Anken, E., E. P. Romijn, C. Maggioni, A. Mezghrani, R. Sitia,I. Braakman, and A. J. Heck. 2003. Sequential waves of func-tionally related proteins are expressed when B cells preparefor antibody secretion. Immunity 18:243–253.

Veith, P. D., G. H. Talbo, N. Slakeski, S. G. Dashper, C. Moore,R. A. Paolini, and E. C. Reynolds. 2002. Major outer membraneproteins and proteolytic processing of RgpA and Kgp of Porph-yromonas gingivalis W50. Biochem. J. 363:105–115.

Venter, J. C., M. D. Adams, E. W. Myers, P. W. Li, et al. 2001.The sequence of the human genome. Science 291:1304–1351.

Vercoutter-Edouart, A. S., X. Czeszak, M. Crepin, J. Lemoine, B.Boilly, X. Le Bourhis, J. P. Peyrat, and H. Hondermarck. 2001.Proteomic detection of changes in protein synthesis inducedby fibroblast growth factor-2 in MCF-7 human breast cancercells. Exp. Cell Res. 262:59–68.

Vuadens, F., D. Gasparini, C. Deon, J. C. Sanchez, D. F. Hochstras-ser, P. Schneider, and J. D. Tissot. 2002. Identification of specificproteins in different lymphocyte populations by proteomictools. Proteomics 2:105–111.

Walker, A. K., G. Rymar, and P. C. Andrews. 2001. Mass spectro-metric imaging of immobilized pH gradient gels and creationof “virtual” two-dimensional gels. Electrophoresis 22:933–945.

Wang, Y. Y., K. H. Khoo, S. T. Chen, C. C. Lin, C. H. Wong,and C. H. Lin. 2002. Studies on the immuno-modulating andantitumor activities of Ganoderma lucidum (Reishi) polysac-

charides: functional and proteomic analyses of a fucose-con-taining glycoprotein fraction responsible for the activities.Bioorg. Med. Chem. 10:1057–1062.

Wasinger, V. C., S. J. Cordwell, A. Cerpa-Poljak, J. X. Yan, A. A.Gooley, M. R. Wilkins, M. W. Duncan, R. Harris, K. L. Williams,and I. Humphery-Smith. 1995. Progress with gene-productmapping of the Mollicutes: Mycoplasma genitalium. Electro-phoresis 16:1090–1094.

Watson, J. D., and F. H. C. Crick. 1953. Genetical Implications ofthe structure of deoxyribonucleic acid. Nature 171:964–967.

Wilkins, M. R., J. C. Sanchez, A. A. Gooley, R. D. Appel, I. Hum-phery-Smith, D. F. Hochstrasser, and K. L. Williams. 1996.Progress with proteome projects: why all proteins expressedby a genome should be identified and how to do it. Biotechnol.Genet. Eng. Rev. 13:19–50.

Wolters, D. A., M. P. Washburn, and J. R. Yates, 3rd. 2001. Anautomated multidimensional protein identification technologyfor shotgun proteomics. Anal. Chem. 73:5683–5690.

Wu, C. C., M. J. MacCoss, K. E. Howell, and J. R. Yates. 2003. Amethod for the comprehensive proteomic analysis of mem-brane proteins. Nat. Biotechnol. 21:532–538.

Yates, J. R., III, J. K. Eng, A. L. McCormack, and D. Schieltz. 1995.Method to correlate tandem mass spectra of modified peptidesto amino acid sequences in the protein database. Anal. Chem.67:1426–1436.

Yatsuda, A. P., J. Krijgsveld, A. W. Cornelissen, A. J. Heck, andE. de Vries. 2003. Comprehensive analysis of the secreted pro-teins of the parasite Haemonchus contortus reveals extensivesequence variation and differential immune recognition. J. Biol.Chem. 278:16941–16951.

Yergey, A. L., J. R. Coorssen, P. S. Backlund, Jr., P. S. Blank, G.A. Humphrey, J. Zimmerberg, J. M. Campbell, and M. L. Vestal.2002. De novo sequencing of peptides using MALDI/TOF-TOF. J. Am. Soc. Mass Spectrom. 13:784–791.

Yu, C. J., Y. F. Lin, B. L. Chiang, and L. P. Chow. 2003. Proteomicsand immunological analysis of a novel shrimp allergen, Penm 2. J. Immunol. 170:445–453.

APPENDIX

Further Reading

Conn, P. M. 2003. Handbook of Proteomic Methods. HumanaPress, Totowa, NJ.

James, P. 2001. Proteome Research: Mass Spectrometry. Springer,New York.

Kannicht, C. 2002. Posttranslational Modifications of Proteins:Tools for Functional Proteomics. Humana Press, Totowa, NJ.

Kinter, M., and N. E. Sherman, N. E. 2000. Protein Sequencingand Identification Using Tandem Mass Spectrometry. JohnWiley, New York.

Liebler, D. C. 2002. Introduction to Proteomics: Tools for the NewBiology. Humana Press, Totowa, NJ.

Link, A. J. 1999. 2-D Proteome Analysis Protocols. Humana Press,Totowa, NJ.

Lorkowski, S., and P. Cullen. 2003. Analysing Gene Expression:A Handbook of Methods, Possibilities and Pitfalls. Wiley-VCH,New York.

Pennington, S. R., and M. J. Dunn. 2001. Proteomics: From ProteinSequence to Function. Bios, Springer-Verlag, New York.

Simpson, R. J. 2003. Proteins and Proteomics: A Laboratory Man-ual. Cold Spring Harbor Laboratory Press, Cold Spring Harbor,New York.

Westermeier, R., and T. Naven. 2002. Proteomics in Practice: aLaboratory Manual of Proteome Analysis. Wiley-VCH, NewYork.

proteomics in the chicken: tools for understanding immune responses to avian … · 2018-12-05 ·...

Documents