detection technologies in proteome analysis

29
771 (2002) 3–31 Journal of Chromatography B, www.elsevier.com / locate / chromb Review Detection technologies in proteome analysis * Wayne F. Patton Proteomics Section, Biosciences Department, Molecular Probes, Inc., 4849 Pitchford Avenue, Eugene, OR 97402-9165, USA Abstract Common strategies employed for general protein detection include organic dye, silver stain, radiolabeling, reverse stain, fluorescent stain, chemiluminescent stain and mass spectrometry-based approaches. Fluorescence-based protein detection methods have recently surpassed conventional technologies such as colloidal Coomassie blue and silver staining in terms of quantitative accuracy, detection sensitivity, and compatibility with modern downstream protein identification and characteri- zation procedures, such as mass spectrometry. Additionally, specific detection methods suitable for revealing protein post-translational modifications have been devised over the years. These include methods for the detection of glycoproteins, phosphoproteins, proteolytic modifications, S-nitrosylation, arginine methylation and ADP-ribosylation. Methods for the detection of a range of reporter enzymes and epitope tags are now available as well, including those for visualizing b-glucuronidase, b-galactosidase, oligohistidine tags and green fluorescent protein. Fluorescence-based and mass spec- trometry-based methodologies are just beginning to offer unparalleled new capabilities in the field of proteomics through the performance of multiplexed quantitative analysis. The primary objective of differential display proteomics is to increase the information content and throughput of proteomics studies through multiplexed analysis. Currently, three principal approaches to differential display proteomics are being actively pursued, difference gel electrophoresis (DIGE), multiplexed proteomics (MP) and isotope-coded affinity tagging (ICAT). New multiplexing capabilities should greatly enhance the applicability of the two-dimensional gel electrophoresis technique with respect to addressing fundamental questions related to proteome- wide changes in protein expression and post-translational modification. 2002 Elsevier Science B.V. All rights reserved. Keywords: Reviews; Proteomics; Detection Contents 1. Introduction ............................................................................................................................................................................ 4 2. General protein detection methods ............................................................................................................................................ 4 2.1. Organic dye- and silver stain-based methods ..................................................................................................................... 4 2.2. Radioactive labeling methods .......................................................................................................................................... 5 2.3. Reverse stain methods ..................................................................................................................................................... 6 2.4. Fluorescence-based staining methods ............................................................................................................................... 6 2.5. Chemiluminescence-based staining methods ..................................................................................................................... 8 *Tel.: 11-541-465-8300; fax: 11-541-344-6504. E-mail address: [email protected] (W.F. Patton). 1570-0232 / 02 / $ – see front matter 2002 Elsevier Science B.V. All rights reserved. PII: S1570-0232(02)00043-0

Upload: alejandro-munoz-rojas

Post on 28-Mar-2015

150 views

Category:

Documents


2 download

TRANSCRIPT

Page 1: Detection Technologies in Proteome Analysis

771 (2002) 3–31Journal of Chromatography B,www.elsevier.com/ locate /chromb

Review

Detection technologies in proteome analysis

*Wayne F. PattonProteomics Section, Biosciences Department, Molecular Probes, Inc., 4849 Pitchford Avenue, Eugene, OR 97402-9165, USA

Abstract

Common strategies employed for general protein detection include organic dye, silver stain, radiolabeling, reverse stain,fluorescent stain, chemiluminescent stain and mass spectrometry-based approaches. Fluorescence-based protein detectionmethods have recently surpassed conventional technologies such as colloidal Coomassie blue and silver staining in terms ofquantitative accuracy, detection sensitivity, and compatibility with modern downstream protein identification and characteri-zation procedures, such as mass spectrometry. Additionally, specific detection methods suitable for revealing proteinpost-translational modifications have been devised over the years. These include methods for the detection of glycoproteins,phosphoproteins, proteolytic modifications, S-nitrosylation, arginine methylation and ADP-ribosylation. Methods for thedetection of a range of reporter enzymes and epitope tags are now available as well, including those for visualizingb-glucuronidase, b-galactosidase, oligohistidine tags and green fluorescent protein. Fluorescence-based and mass spec-trometry-based methodologies are just beginning to offer unparalleled new capabilities in the field of proteomics through theperformance of multiplexed quantitative analysis. The primary objective of differential display proteomics is to increase theinformation content and throughput of proteomics studies through multiplexed analysis. Currently, three principal approachesto differential display proteomics are being actively pursued, difference gel electrophoresis (DIGE), multiplexed proteomics(MP) and isotope-coded affinity tagging (ICAT). New multiplexing capabilities should greatly enhance the applicability ofthe two-dimensional gel electrophoresis technique with respect to addressing fundamental questions related to proteome-wide changes in protein expression and post-translational modification. 2002 Elsevier Science B.V. All rights reserved.

Keywords: Reviews; Proteomics; Detection

Contents

1. Introduction ............................................................................................................................................................................ 42. General protein detection methods ............................................................................................................................................ 4

2.1. Organic dye- and silver stain-based methods ..................................................................................................................... 42.2. Radioactive labeling methods .......................................................................................................................................... 52.3. Reverse stain methods ..................................................................................................................................................... 62.4. Fluorescence-based staining methods ............................................................................................................................... 62.5. Chemiluminescence-based staining methods ..................................................................................................................... 8

*Tel.: 11-541-465-8300; fax: 11-541-344-6504.E-mail address: [email protected] (W.F. Patton).

1570-0232/02/$ – see front matter 2002 Elsevier Science B.V. All rights reserved.PI I : S1570-0232( 02 )00043-0

Page 2: Detection Technologies in Proteome Analysis

771 (2002) 3–314 W.F. Patton / J. Chromatogr. B

2.6. Immobilized pH gradient /MALDI-TOF mass spectrometry and surface-enhanced laser desorption–ionization massspectrometry (SELDI-MS) ........................................................................................................................................................... 83. Specific detection of protein post-translational modifications ...................................................................................................... 9

3.1. Glycoprotein detection methods ....................................................................................................................................... 93.2. Phosphoprotein detection methods ................................................................................................................................... 103.3. Proteolytic modification detection methods ....................................................................................................................... 133.4. S-Nitrosylation detection method ..................................................................................................................................... 133.5. Arginine methylation detection methods ........................................................................................................................... 143.6. ADP-ribosylation detection methods ................................................................................................................................ 15

4. Detection of reporter enzymes and epitope tags ......................................................................................................................... 154.1. b-Glucuronidase ............................................................................................................................................................. 154.2. b-Galactosidase .............................................................................................................................................................. 164.3. Oligohistidine tags .......................................................................................................................................................... 174.4. Green fluorescent protein................................................................................................................................................. 17

5. Differential display proteomics: approaches and options............................................................................................................. 195.1. Difference gel electrophoresis (DIGE).............................................................................................................................. 195.2. Multiplexed proteomics (MP) .......................................................................................................................................... 235.3. Isotope-coded affinity tagging (ICAT) .............................................................................................................................. 24

6. Conclusions ............................................................................................................................................................................ 27Acknowledgements ...................................................................................................................................................................... 27References .................................................................................................................................................................................. 27

1. Introduction 2. General protein detection methods

2.1. Organic dye- and silver stain-based methodsReview articles providing overviews of signifi-

cant milestones in the development of assorted Though a range of organic dyes have been used toprotein detection techniques, as applied to gel visualize proteins in polyacrylamide gels, Coomassieelectrophoresis, have already been published in Blue dyes (R and G types) have certainly enjoyed therecent years [1–10]. After briefly updating these most popularity due to their low cost, ease of use andtopics, this review will primarily focus upon cur- unexpectedly good compatibility with downstreamrent issues in protein visualization as they directly microchemical characterization methods, such asrelate to modern proteomics investigations. Since mass spectrometry [11]. Typically, Coomassie Bluetwo-dimensional (2D) gel electrophoresis is in- R-250 dye is used in a regressive staining approachcreasingly being used in the micropreparative isola- where gels are saturated with dye that has beention of proteins destined for characterization by dissolved in an aqueous solution containing methanolmass spectrometry, it is now expected that protein and acetic acid, followed by destaining in a similarvisualization methods be validated with respect to solution devoid of dye. Since the proteins have atheir compatibility with this technology. So, in higher affinity for the dye molecules than the geladdition to sensitivity, linear dynamic range and matrix, a point is reached where the backgroundreproducibility, visualization methods should be staining is minimal but protein bands are wellfully capable of interfacing with the modern analy- labeled. Gel-to-gel reproducibility in staining issis tools of proteomics. Key issues addressed in difficult to control, however, as proteins do destain tothis review article include detection methods for varying extents along with the gel matrix, albeit withdetermining total protein, protein post-translational slower kinetics. Later, progressive staining ap-modifications and epitope-tagged fusion proteins, as proaches were introduced where proteins are gradu-well as strategies for detecting changes in protein ally stained to an endpoint, without significantprofiles by differential display proteomics. staining of the gel matrix [12]. These methods are

Page 3: Detection Technologies in Proteome Analysis

771 (2002) 3–31 5W.F. Patton / J. Chromatogr. B

predicated upon the formation of colloidal dye 2.2. Radioactive labeling methodsparticles. During staining an equilibrium is achievedbetween colloidal particles and freely dispersed dye Radiolabeling is commonly accomplished by in-

3 14 35 32 33 125in solution. The low concentration of free dye corporating H, C, S, P, P, or I into proteinspenetrates the gel matrix and preferentially stains the [21,22]. After electrophoresis, signal detection mayproteins, but the colloidal dye particle is excluded be accomplished using film, by direct autoradiog-from the gel, thus preventing matrix staining. The raphy for the g-emitting isotopes or by fluorographylimits of detection for conventional Coomassie Blue for the b-emitting isotopes. Fluorography employsand colloidal Coomassie Blue stains are 8–10 and fluorescent enhancers impregnated into the gel ma-30–100 ng, respectively. Both dyes provide a linear trix that improve detection of the radioactive emis-response with protein amount over a 10–30-fold sions through the generation of light. For certain

32 125range of concentration. Thus, the principal limita- high-energy isotopes, such as, P or I, intensify-tions associated with Coomassie Blue-based staining ing screens may be used to enhance the signal. Thesemethods is the poor detection sensitivity and small screens absorb the radioactive emissions that passlinear dynamic range obtained with the dyes. through the film, fluoresce and generate light that

Silver staining techniques are based upon saturat- exposes the film. Over the past decade, storageing gels with silver ions, washing the less tightly phosphor imagers and microchannel plate analyzersbound metal ions out of the gel matrix and reducing have begun to replace autoradiography film forthe protein-bound metal ions to form metallic silver detecting radiolabeled proteins. The newer tech-[13–15]. Most commonly, silver staining is accom- nologies offer a wider dynamic range and betterplished using silver nitrate in combination with detection sensitivity than film.formaldehyde developer in alkaline carbonate buffer Radiolabeling has been used extensively for inor using an ammonia–silver complex in combination vivo metabolic experiments. Organic dye or silverwith formaldehyde developer in citrate buffer. Silver stain methods are commonly used in conjunctionstain methods permit detection of as little as a single with radiolabeling techniques to simultaneously as-nanogram of protein, but the linear dynamic range of sess total protein expression levels along with proteinthe stain is restricted to a 10-fold range. Silver synthesis rates, as well as to provide visual land-staining methods are quite complex, multi-step pro- marks for the localization of low abundance proteinscedures that must be stopped at some arbitrary time in 2D gels [23–25]. However, the general utility ofpoint in order to avoid over development. Conse- radiolabeling as a tool for the assessment of proteinquently, gel-to-gel reproducibility is problematic and synthesis rates has been called into question recentlyvariations of 20% in spot intensity have been docu- [26–28]. Metabolically incorporating radiolabels

35mented in the literature [16]. Robotic staining de- such as [ S]methionine, using standard doses andvices offer the potential for minimizing the vari- time-courses, has been shown to induce DNA frag-ability of silver staining, but end-point stains provide mentation, elevate p53 tumor suppressor proteinconsistency without investing in expensive machin- levels, alter cell /nuclear morphology and lead to cellery [17]. Though the best silver stain methods use cycle arrest or apoptosis.aldehyde-based fixatives prior to silver impregnation, Though sensitive detection is readily achieved,this prevents subsequent microchemical analysis by radiolabeling is both hazardous and expensiveEdman sequencing or mass spectrometry. Thus, [29,30]. In 1995, the United States National Insti-alternative silver staining methods must be employed tutes of Health alone spent more than $4 millionthat omit aldehydes in the fixatives [18,19]. General- disposing of radioactive waste [31]. Additional ex-ly, detection sensitivity is poorer and background penditures were required for monitoring exposure tostaining less uniform with the modified staining radioactivity, inventory control and training labora-techniques. Additionally, destaining methods are tory personnel. During that same year, a pregnantoften employed to improve the compatibility of cancer researcher was deliberately contaminated with

32silver staining with peptide mass profiling methods at least 600 mCi of P in one of the Bethesda,[20]. Maryland NIH research laboratories, as detailed in

Page 4: Detection Technologies in Proteome Analysis

771 (2002) 3–316 W.F. Patton / J. Chromatogr. B

US National Institutes of Health Case Number 1-96- chelating agent to the transfer buffer [42–45]. Pro-033; released 09/17/97, entitled ‘‘Wrongful ad- teins stained with zinc–imidazole are also compat-ministration of P-32’’. An additional 26 staff mem- ible with Edman-based microsequencing and in-gelbers were exposed to lesser amounts of the isotope tryptic digestion followed by matrix-assisted laserfrom a water cooler at this facility. Reducing reliance desorption ionization time-of-flight (MALDI-TOF)on radioactivity in biomedical research has become a mass spectrometry for determination of peptidehigh priority in many laboratories over the past masses and sequences [41,42,46].several years as it makes economic good sense andpromotes a better quality of life by providing a saferwork environment for laboratory personnel. 2.4. Fluorescence-based staining methods

Fluorescent detection of proteins in electrophoresis2.3. Reverse stain methods gels is gaining popularity, particularly with labora-

tories engaging in large scale proteomics research.A number of reverse-staining methods for vis- Several recently developed fluorescent stains are

ualizing proteins in SDS–polyacrylamide gels have suitable for incorporation into integrated proteomicsbeen described over the years including methods that platforms for global analysis of protein regulation.employ potassium chloride [32], copper chloride The linear dynamic range of detection using fluores-[33], zinc chloride [34], cobalt acetate [34], nickel cent stains is usually superior to colorimetric orchloride [34] and zinc chloride–imidazole [35,36]. reverse stains.

The three reverse stain techniques that have been Nile Red dye has excitation maxima of about 300used most extensively are the potassium chloride, and 540 nm and an emission maximum of roughlycopper chloride and zinc chloride methods. Copper 640 nm [47]. The detection sensitivity of 20–100 ngchloride staining is slightly more sensitive, while for Nile Red staining is slightly better than detectionpotassium chloride staining is significantly less sensi- obtained with Coomassie Blue staining, and the dyetive than Coomassie Blue staining [37]. Zinc chlo- is compatible with standard Edman sequencing andride–imidazole staining is the most sensitive reverse- immunodetection procedures [48–51]. Nile Redstain developed to date [38,39]. In the presence of staining is not stable, however, and the dye is proneimidazole, free or weakly bound zinc ions are readily to precipitation in aqueous solution, consequentlyprecipitated as zinc–imidazole, while tightly bound making it more difficult to handle than commercial-ions associated with the proteins are refractory to ized fluorescent stains. The photolabile nature ofprecipitation. This generates clear protein zones on a Nile Red staining makes quantitation of proteins bysemi-opaque gel background. Careful attention to the image analysis difficult. Nile Red dye can be used toduration of staining is necessary as over-staining of pre-stain gels prior to electroblotting them to poly-gels can obscure bands and on occasions complex vinylidene difluoride (PVDF) membrane [50,51].patterns of negative and positive staining are ob- SYPRO Red, SYPRO Orange and SYPROtained using zinc–imidazole [38]. The limit of Tangerine dyes can detect proteins in SDS–poly-detection for the zinc–imidazole stain is roughly acrylamide gels using a simple, one-step staining10–20 ng per band for protein separated by SDS– procedure that requires 30–60 min to complete andpolyacrylamide gel electrophoresis, 40–80 ng per does not involve a destaining step [52–55]. As littleband for protein separated on native polyacrylamide as 4–8 ng of protein can be detected, rivaling thegels, and 100–500 ng per band for proteins separated sensitivity of rapid silver staining techniques andin carrier ampholyte or immobilized pH gradient colloidal Coomassie Blue staining methods. Since allisoelectric focusing gels [38,39]. three dyes excite at roughly 300 nm, documentation

Proteins visualized in zinc–imidazole-stained gels of stained gels can readily be achieved by photo-can be evaluated by densitometry, though the linear graphy on a standard laboratory UV transilluminator.dynamic range for protein detection is restricted to Alternatively, the dyes may be quantified with10–100 ng [40,41]. Zinc–imidazole-stained gels can commercially available CCD camera-based imagebe electroblotted or electroeluted by addition of a analysis workstations or a variety of laser scanners,

Page 5: Detection Technologies in Proteome Analysis

771 (2002) 3–31 7W.F. Patton / J. Chromatogr. B

providing a linear dynamic range of three orders ofmagnitude. In addition to their UV excitation peak,SYPRO Orange, Red and Tangerine dyes may beexcited by visible light at approximately 470, 550and 490 nm, respectively. The three dyes emitmaximally at 570, 630 and 640 nm, respectively.Thus, SYPRO Orange and Tangerine dyes are wellsuited for imaging systems with 473 nm second-harmonic generation (SHG) or 488 nm argon-ion(Ar) lasers as well as those that use 470 nm blue-LEDs or 460 nm blue fluorescent light sources.SYPRO Red dye is most suitable for 532-nm fre-quency-doubled neodymium–yttrium–aluminum–garnet (Nd–YAG) laser sources, 543-nm helium–neon (He–Ne) laser sources or 550-nm green-lighttransilluminators [56].

Both SYPRO Orange and SYPRO Red dyesrequire 7% acetic acid in the staining solution, whichis problematic when electroblotting, electroeluting ormeasuring enzyme activity is indicated. SYPROTangerine dye is an environmentally benign alter-native to conventional protein stains that does notrequire solvents such as methanol or acetic acid foreffective protein visualization [55]. Instead, proteinscan be stained in a wide range of buffers, includingphosphate-buffered saline or simply 150 mM sodiumchloride. Since proteins may be stained usingSYPRO Tangerine dye without the need for harshfixatives, they are readily eluted from gels or used inzymographic assays, provided that SDS presentduring electrophoresis does not inactivate them.SYPRO Tangerine dye is also suitable for stainingproteins in gels before they are transferred to mem-branes by electroblotting. Using a 7% acetic acidstaining solution, SYPRO Tangerine dye-stained gelshave a slightly higher background fluorescence than

Fig. 1. Fluorescent staining compared with silver staining ofSYPRO Orange and SYPRO Red dyes. Thus, the two-dimensional gels. Proteins from a cultured fibroblast cell linelatter dyes are preferred choices for routine staining (Rat-1) were fractionated by 2D gel electrophoresis and thenin a fixative solution. stained using SYPRO Ruby protein gel stain (top image) or silver

stain (bottom image), as previously described [9,58]. The com-SYPRO Ruby dye is a proprietary ruthenium-parison demonstrates similar detection sensitivity between thesebased metal chelate stain, which side-by-side com-techniques. It should be noted that the silver stain method

parisons demonstrate is as sensitive as the best silver employed in this comparison is optimized for protein detection,stains available (Fig. 1) [57,58]. SYPRO Ruby dye but not compatible with subsequent analysis by mass spec-allows one-step, low background staining of proteins trometry. Mass spectrometry-compatible silver stains are signifi-

cantly less sensitive than the mass spectrometry-friendly fluores-in polyacrylamide gels without resorting to lengthycent stain. In addition, the silver stain provides only qualitativedestaining steps. The linear dynamic range of the dyeinformation about the proteins due to its narrow 10-fold linear

extends over three orders of magnitude, thus surpas- dynamic range, while the fluorescent stain provides quantitativesing both silver and Coomassie blue stains in per- capabilities extending over 3-logs. Figure courtesy of Drs. Birteformance. The dye can be excited using a standard Schulenberg and Jill Hendrickson, Molecular Probes.

Page 6: Detection Technologies in Proteome Analysis

771 (2002) 3–318 W.F. Patton / J. Chromatogr. B

300-nm UV transilluminator or using imaging sys- of proteins in complex biological specimens with atems equipped with 450-, 473-, 488- or even 532-nm detection sensitivity that is competitive with fluores-lasers. Though more sensitive than SYPRO Orange, cence stain, reverse-stain or silver stain methods.SYPRO Red and SYPRO Tangerine dyes, optimal The specific detection of an antigen directly instaining is somewhat slower, requiring about 3 h to SDS–polyacrylamide gels has also been demonstra-complete. Similar to colloidal Coomassie Blue stain ted recently, obviating the need to electroblot gelsbut unlike silver stain, SYPRO Ruby dye is an [60]. The method is simply based upon old pro-end-point stain. Thus, staining times are not critical cedures for immunodetection in gels that were inand staining can be performed over night without routine use prior to Towbin’s seminal paper ongels overdeveloping. electroblotting [61]. The availability of highly sensi-

Side-by-side comparisons of silver staining (minus tive chemiluminescence methods in recent years hasglutaraldehyde) and SYPRO Ruby dye staining have made the in gel technique sensitive enough forbeen performed, and SYPRO Ruby stain was found certain applications.to be superior with respect to possessing a muchbroader linear dynamic range of quantitation (1000-fold), ability to visualize a 20% greater number of 2.6. Immobilized pH gradient /MALDI-TOF massproteins, better reproducibility in quantifying the spectrometry and surface-enhanced laserlowest intensity proteins in a 2D gel profile, and desorption–ionization mass spectrometry (SELDI-ability to identify lower amounts of protein with MS)higher sequence coverage by peptide mass profiling(75–100 fmol) [57]. This clearly demonstrates that Laser desorption of proteins by direct MALDI-SYPRO Ruby dye is superior to silver staining for TOF mass spectrometry-based surface scanning ofmodern proteomics applications. carrier ampholyte isoelectric focusing gels, immobil-

ized pH gradient isoelectric focusing gels, native2.5. Chemiluminescence-based staining methods polyacrylamide gels, and SDS–polyacrylamide gels

has been demonstrated, with sub-picomolar detectionWhile commonly employed for the detection of sensitivities achieved [62–66]. In a preferred em-

proteins after electroblotting, chemiluminescence had bodiment, ‘‘virtual’’ 2D gels may be created bynot been used to directly detect proteins in poly- desorbing proteins directly from immobilized pHacrylamide gels until very recently [59,60]. A gradient gels using MALDI-TOF mass spectrometry,chemiluminescent acridan phosphate labeling reagent in effect substituting mass spectrometry for SDS–(maleimide derivative) was used to pre-derivatize polyacrylamide gel electrophoresis [65,66]. Analyti-bovine serum albumin prior to subjecting the protein cal data may be presented as a computer-generatedto SDS–polyacrylamide gel electrophoresis [59]. image that is similar to a classical 2D gel inGels were then treated with acidic and basic buffer appearance. The ‘‘virtual’’ 2D gel approach appearscontaining a peroxide, which initiated a rapid gene- to detect lower-molecular-mass proteins (,30 kDa)ration of light. Roughly 1–650 ng of serum albumin better but higher-molecular-mass proteins (.30 kDa)could be detected by the procedure. This should be more poorly than conventional 2D gels [65]. Cur-considered a best-case scenario, however, as serum rently, a number of artifacts must be contended withalbumin is unusual in containing a very large number using the method, including horizontal streaks due toof cysteine residues (35 total) when reduced. De- variations in baseline slope, artifacts caused by thetection sensitivity for proteins with average numbers presence of protein multimers and matrix adducts onof cysteine residues would be closer to 10 ng, or the the proteins that generate duplicate or triplicatelevel of sensitivity achieved by colloidal Coomassie spots, higher molecular masses than predicted due toBlue stain. Currently, the chemiluminescence de- difficulty in desorbing proteins from the gel matrix,tection method has only been validated for feasibility and difficulty in quantifying the amounts of theusing a single model protein. It remains to be seen proteins due to ion suppression phenomenon, a well-whether the labeling technique will permit detection known problem associated with mass spectrometry

Page 7: Detection Technologies in Proteome Analysis

771 (2002) 3–31 9W.F. Patton / J. Chromatogr. B

[65,66]. The procedure is also currently quite slow, least three times as many proteins as genes [70].requiring a day to run the gel, 2 days to dry it down Post-translational modifications modulate the func-and 2 days to acquire spectra [65]. tion of proteins and thus directly impact their

Surface-enhanced laser desorption / ionization capacity to participate in cellular regulatory events.(SELDI) involves retaining proteins on a solid-phase Mass spectrometry has become a powerful re-chromatographic surface (ProteinChip Array) and search method for the characterization of proteindirect detection of the retained proteins by time-of- post-translational modifications, since modificationsflight mass spectrometry (TOF-MS) [67–69]. Typi- either increase or decrease the molecular mass of thecally, cationic, anionic, hydrophobic, hydrophilic, affected amino acid residue [71]. Generally, somemixed charge, or immobilized metal ion chemistries method for the rapid proteome-wide screening ofare employed on SELDI ProteinChip surfaces, as proteins with respect to the presence or absence of awell as antibodies, DNA, enzymes or receptors. particular modification is required prior to micro-Surfaces are incubated with a crude cell lysate or characterization by mass spectrometry. Currently,biological fluid, and then washed at various strin- heavily modified proteins, such as highly glyco-gency levels to remove nonspecifically associated sylated proteins, can prove difficult to identify bycomponents. After this sample purification, the re- standard mass spectrometry-based methods such astained components are evaluated by TOF-MS. peptide mass profiling, since derivatized proteinsQuantitation of specifically captured TNF-a over a cannot be matched to calculated peptide masses100-fold concentration range has been demonstrated, derived from genome databases. Finally, the pres-though more complex samples are likely to be ence of certain post-translational modifications, suchsusceptible to ion suppression phenomenon [68]. as phosphorylation and sialoglycosylation, results inThough excellent for rapid profiling of minute peptide signal suppression, complicating identifica-samples, the ability to identify proteins from SELDI tion of peptides bearing these groups.profiles remains challenging, and conventional pep- The detection of three prominent protein post-tide mapping must be employed [69]. translational modifications: glycosylation, phos-

phorylation and proteolytic modifications are sum-marized below. In addition, three lesser known post-

3. Specific detection of protein post-translational translational modifications, S-nitrosylation, argininemodifications methylation and ADP-ribosylation are discussed.

Proteomic investigations regarding certain of theseThough genomics provides comprehensive data- post-translational modifications are only just begin-

bases of sequence information, DNA and mRNA ning, as new methodologies allowing their detectionprovide no information concerning the activities and have only recently been introduced.post-translational modifications of proteins. Theelucidation of protein post-translational modifications 3.1. Glycoprotein detection methodsis perhaps the most important justification forproteomics as a scientific endeavor. The number of Protein glycosylation is a key post-translationaldocumented protein co- and post-translational modi- modification relevant to a range of biological phe-fications now exceeds 400 (http: / / abrf.org / index- nomenon. For example, the metastatic spread of.cfm/dm.home). The most common modifications tumor cells in malignant progression is known to beinclude phosphorylation, glycosylation, lipidation, a major cause of cancer mortality. Protein glycosyla-sulfation and proteolytic modifications. These chemi- tion is increasingly being recognized as one of thecal modifications of proteins are crucial to modu- most prominent biochemical alterations associatedlating their function, but are not directly coded for by with malignant transformation and tumorigenesisgenes. Even in the simplest self-replicating organism, [72–75]. Not only is the total amount of protein-Mycoplasma genitalium, there are 24% more pro- associated carbohydrate of relevance to cancer, butteins than genes due to post-translational modifica- alterations in the branching patterns of the attachedtions, and in humans there are estimated to be at glycans also play important roles in malignancy.

Page 8: Detection Technologies in Proteome Analysis

771 (2002) 3–3110 W.F. Patton / J. Chromatogr. B

Glycosylation changes in human carcinomas appear 1,3,6-trisulfonate requires heating at 60 8C for 45–60to contribute to the malignant phenotype observed min, and limits of detection sensitivity range from 50downstream of certain oncogenic events [72]. to 200 ng [84]. A sensitive green-fluorescent glyco-

Glycoproteins can be detected by autoradiography protein-specific stain, referred to as Pro-Q Emerald3 14after incorporation of H or C sugars into cultured 300 dye, was recently reported, that allows detection

cells or tissues [76–78]. Though sensitive detection of glycoproteins on PVDF membranes or directly inis achieved, radiolabeling is slow, hazardous and polyacrylamide gels [85]. Pro-Q Emerald 300 dyeexpensive. Nonradioactive detection of glycoproteins may be conjugated to glycoproteins by a periodicis usually performed using a periodic acid Schiff acid Schiff’s (PAS) mechanism using very mildbase procedure that employs the dye, acid fuchsin reaction conditions at room temperature. Pro-Q(pararosanaline) [79,80]. The principal limitation of Emerald 300 dye-labeled gels may be post-stainedthis colorimetric method is that it has poorer de- with SYPRO Ruby dye, allowing sequential two-tection sensitivity than Coomassie Blue staining, thus color detection of glycosylated and nonglycosylatedmaking the technique difficult to implement in the proteins (Fig. 2). As little as 300 pg of a1-acidcontext of modern proteomics research. Similarly, glycoprotein (40% carbohydrate) and 1 ng of glucoseAlcian Blue has been used as a colorimetric stain for oxidase (12% carbohydrate) or avidin (7% carbohy-glycoproteins [81,82]. Sensitivity of this method is drate) are readily detected in gels after staining withsimilar to the acid fuchsin dye method and neutral Pro-Q Emerald 300 dye, and a 500–1000-fold linearglycoproteins, such as horseradish peroxidase and dynamic range of detection is obtained using thetransferrin, are poorly detected by the staining meth- labeling method [85]. A UV transilluminator isod. required to detect the stain, but recently Pro-Q

Dansyl hydrazine is a fluorescent alternative to Emerald 488 dye was introduced for detection ofacid fuchsin dye that can be used for the detection of glycoproteins using laser-based gel scanners.glycoproteins in polyacrylamide gels [83]. Glycopro- A general antibody-based method for detectingteins are subjected to periodate oxidation in gels and glycoproteins involves periodic acid oxidation of thethe resulting aldehydes are condensed with dansyl carbohydrate moieties, reaction with digoxigeninhydrazine to form hydrazones. These are in turn hydrazide, and detection with alkaline phosphatase-reduced to stable hydrazine derivatives with sodium conjugated anti-digoxigenin antibody [87]. Similarborohydride. The coupling reaction requires incubat- methods have also been devised using biotin hy-ing gels in acidified dimethylsulfoxide at 60 8C for 2 drazide and alkaline phosphatase-conjugated orh and this harsh treatment has undoubtedly prevented horseradish peroxidase-conjugated streptavidinwidespread adoption of the method. Very high con- [88,89]. The methods are unsuitable for glycoproteincentrations of dye are required for the dansyl hy- detection in gels, however, and the introduction ofdrazine labeling procedure (0.5–2.0 g / l) and newer contaminating reporter proteins complicates targetmethods offer better specificity and sensitivity of identification by peptide mass profiling. Methods fordetection. the determination of carbohydrates in proteins after

Three new fluorescence-based methods for detect- gel electrophoresis by covalent derivatization withing glycoproteins have been introduced recently [84– fluorophores such as 2-aminobenzoic acid have been86]. All techniques are based upon periodic acid introduced, but require band excision and proteinSchiff base chemistry. Fluorescein semicarbazide hydrolysis in strong acids at high-temperature [90].hydrazide and 8-aminonaphthalene-1,3,6-trisulfonate Pulsed amperometric detection procedures employ-have been employed for the detection of glycopro- ing similarly harsh reaction steps have also beenteins electroblotted to PVDF membranes, but the devised [91].techniques are not suitable for detection of glycopro-teins directly in polyacrylamide gels [84,86]. To 3.2. Phosphoprotein detection methodsdate, the fluorescein semicarbazide labeling methodhas not been characterized extensively [86]. How- Though the study of reversible protein phosphoryl-ever, the method employing e-aminonaphthalene- ation reactions has become central to biochemistry

Page 9: Detection Technologies in Proteome Analysis

771 (2002) 3–31 11W.F. Patton / J. Chromatogr. B

plays a central role in biological regulation, and thisis especially true in the context of carcinogenesis.The role of protein phosphorylation cascades inmalignant transformation is exceedingly well estab-lished. Malignant transformation occurs throughsuccessive mutations in specific cellular genes, lead-ing to the activation of oncogenes and the inactiva-tion of tumor suppressor genes [94–96]. The cellulargenes that are altered are normally involved inmaintaining cellular homeostasis through participa-tion in signaling cascades that are tightly regulated tomaintain the functional integrity of the cell [96].Oncogenic signaling pathways lead to the activationof tyrosine or serine / threonine kinases. Tyrosinekinases that are often found to be mutated in humantumors include the cytoplasmic kinases Src and Abl,as well as the integral plasma membrane kinasesEGFR and PDGFR [96]. Serine / threonine kinasesthat are mutated or activated in human cancerinclude Raf, Akt and Tpl-2 [96]. These proteinkinases are involved in signaling pathways that leadto cell proliferation, or inhibition of apoptosisthrough activation of transcription factors, inhibitionof pro-apoptotic molecules, or deregulation of cellcycle control. A tyrosine phosphatase associated withmetastasis in colorectal cancer has recently beenidentified [97].Fig. 2. Serial dichromatic detection of glycosylated and un-

Phosphoproteins are most often detected by auto-glycosylated proteins in the same polyacrylamide gel. Detection of32 33glycoproteins using Pro-Q Emerald 488 glycoprotein stain kit radiography after incorporation of P or P into

(Molecular Probes) (Top image). This PAS-based chemistry cultured cells or after incorporation into subcellulargenerates green fluorescent bands in gels. Detection of total fractions by protein kinases [98]. Such approachesprotein profile using SYPRO Ruby protein gel stain (Molecular

are restricted to a limited range of biological materi-Probes) (Bottom image). This colloidal metal chelate stainingals, such as tissue culture samples. Needless to say,procedure generates orange–red fluorescent bands in gels. Both

images were collected using an FLA-3000 laser-based gel scanner analysis of clinical samples would require in vivo(Fuji Film Scientific Imaging). Lanes 1–3: 2-fold serial dilution of labeling of patients, which is not feasible. Severalbroad range molecular mass markers containing the glycoprotein alternatives to radiolabeling have also been de-ovalbumin (3% carbohydrate). Lanes 5–13: 2-fold serial dilutions

veloped over the years. Phosphoproteins can beof CandyCane molecular mass markers (Molecular Probes) con-detected by immunoblotting [99,100]. Though hightaining a mixture of glycosylated and unglycosylated proteins.

Figure courtesy of Ms. Courtenay Hart, Molecular Probes. quality antibodies to phosphotyrosine are commer-cially available, antibodies that specifically recognize

since its discovery in the 1950s, by Nobel Prize phosphoserine and phosphothreonine residues havewinners Drs. Edwin Krebs and Edmond Fischer, been more problematic, often being sensitive tophosphoproteomics, the systematic parallel analysis amino acid sequence context. Immunoblotting is bestof the phosphorylation status of large sets of proteins performed after blocking unoccupied sites on theinvolved in the regulatory circuitry of cells and solid-phase support with protein solutions, whichtissues, is a relatively new field destined to drive interferes with microchemical analysis. Removal ofresearch in the post-genomics era for many years to the antibody and stain require relatively harsh treat-come [92,93]. Reversible protein phosphorylation ments (i.e., heating to 65 8C, incubation with 0.1%

Page 10: Detection Technologies in Proteome Analysis

771 (2002) 3–3112 W.F. Patton / J. Chromatogr. B

SDS and 1 mM DTT). This also poses problems with antibody. Though methods have recently been intro-subsequent use of the protein for sequencing and duced to directly quantify the relative abundance ofmass spectrometry. phosphoproteins in two different samples by mass

The cationic carbocyanine dye ‘‘Stains-All’’ (1- spectrometry through culturing different cell popula-15 14ethyl-2-[3-(3-ethylnaphtho[1,2d] thiazolin-2-ylidene)- tions in N-enriched and N-enriched medium, the

2-methylpropenyl]-naphtho [1,2d]thiazolium bro- linear dynamic range of such methods has explicitlymide) stains RNA, DNA, phosphoproteins and cal- been demonstrated over only a 10-fold range [109].cium-binding proteins blue while unphosphorylated In addition, an isotope-coded affinity tag approach toproteins are stained red [101,102]. Stains-All is not the differential quantitation of phosphopeptides byroutinely used to detect phosphoproteins due to poor b-elimination and subsequent addition of biotinyl-specificity and low sensitivity. Stains-All is at least iodoacetamidyl-3,6-dioxaoctanediamine containing10 times less sensitive than Coomassie Brilliant Blue either four alkyl hydrogen or four alkyl deuteriumand several orders of magnitude less sensitive than atoms has been reported [111]. Ion suppression32P autoradiography. phenomena associated with mass spectrometry pre-

Phosphoproteins may be specifically detected in vents stoichiometric comparison of different phos-gels through alkaline hydrolysis of phosphate esters phoproteins by such techniques.of serine or threonine, precipitation of the released Recently, two methods for the chemical derivati-inorganic phosphate with calcium, formation of an zation and enrichment of phosphopeptides have beeninsoluble phosphomolybdate complex and then vis- described [112,113]. The first method requiresualization of the complex with a dye such as Methyl proteolytic digestion of the sample, reduction andGreen, Malachite Green or Rhodamine B [103,104]. alkylation of cysteine residues, N-terminal and C-The Methyl Green dye approach has recently been terminal protection of the peptides, formation ofcommercialized as the GelCode Phosphoprotein de- phosphoramidate adducts at phosphorylated residuestection kit (Pierce Chemical, Rockford, IL). The by carbodiimide condensation with cystamine, cap-detection sensitivity of the staining method is con- ture of the phosphopeptides on glass beads coupledsiderably poorer than Coomassie Blue staining, with to iodoacetate, elution with trifluoroacetic acid and80–160 ng of phosvitin, a protein containing roughly evaluation by mass spectrometry [112]. The second100 phosphoserine residues, being detectable by the method involves oxidation of cysteine residues withcommercialized kit. The staining procedure is fairly performic acid, alkaline hydrolysis to induce b-complex (involving seven different reagents) and elimination of phosphate groups from phosphoserinealkaline hydrolysis requires heating gels to 658C, and phosphothreonine residues, addition ofwhich causes the gel matrix to hydrolyze and swell ethanedithiol, coupling of the resulting free sulf-considerably. Since phosphotyrosine residues are not hydryl residues with biotin, purification of phos-hydrolyzed by the alkaline treatment, proteins phos- phoproteins by avidin affinity chromatography,phorylated at this amino acid residue escape de- proteolytic digestion of the eluted phosphoproteins, atection by the method. Gels may be destained and second round of avidin purification and then analysisthen post-stained with colloidal Coomassie Blue dye by mass spectrometry [113]. Both procedures werein order to visualize the total protein profile. validated using the phosphoprotein b-casein. Both

Several instrument-based methods are available procedures identified the monophosphorylatedfor the determination of protein phosphorylation such trypsin peptide fragment from the test protein, but

31as P NMR [105,106], mass spectrometry [71,107– both failed to detect the tetraphosphorylated peptide109] and protein sequencing [110]. While these fragment [112,113].procedures accurately characterize the phosphoryla- Fluorescence detection methods appear to offer thetion status of proteins and peptides, they are unsuit- best solution to global protein quantitation inable for high-throughput screening of phosphorylated proteomics. However, currently, there is no satisfac-substrates. The techniques are generally used after a tory method for the specific and reversible fluores-phosphoprotein has been identified by autoradiog- cent detection of gel-separated phosphoproteins fromraphy or immunoblotting with anti-phosphotyrosine complex samples. Derivatization and fluorophore

Page 11: Detection Technologies in Proteome Analysis

771 (2002) 3–31 13W.F. Patton / J. Chromatogr. B

labeling of phosphoserine residues by blocking free the samples are combined and the proteins separatedsulfhydryl groups with iodoacetate or performate, on a single 2D gel. Specific inhibition of particularalkaline b-elimination of the phosphate residue, cysteine proteases is readily identified through com-addition of ethanedithiol, and reaction of the re- parison of the resulting red-fluorescent and blue-sulting free sulfhydryl group with 6-iodoacetamido- fluorescent images. Gels can subsequently be stainedfluorescein has been demonstrated in capillary elec- with a high sensitivity total protein stain, the proteintrophoresis using laser-induced fluorescence detec- isoforms of interest excised and then identified bytion and similar reactions have been performed on standard mass spectrometry-based techniques.protein microsequencing membranes [114,115]. Nei- Tools for the specific proteome-wide detection ofther publication demonstrates detection of phos- natural substrates of proteolytic enzymes are notphoproteins directly in gels, but the detection of currently available. Individual proteins can be incu-cysteine residues by similar chemistry is shown, bated with specific proteases to determine whetherindicating that the pre-derivatization of phosphopro- they are potential substrates, but the physiologicalteins by this chemistry should be feasible [115]. One relevance of such measurements is not assured.problem with the approach is that a delicate balance Standard immunodetection techniques may be usedmust be struck between the base and the to monitor the proteolysis of specific proteins afterethanedithiol in order to achieve elimination of the electroblotting, as long as the epitope used forphosphate group from the serine residue and addition detection is not proteolysed. In the realm of 2D gelof the ethane dithiol to the resulting dehydroalanine electrophoresis, spot-by-spot identification of theresidue without hydrolysis of the peptide backbone majority of proteins resolved in a particular sample[114]. Though the quest for a Coomassie Blue-like generally leads to the identification of subsets ofstain allowing the simple and selective detection of proteins related to one another through proteolyticphosphoproteins in gels dates back to the 1960s, this processing events.holy grail remains lost for the time being.

3.4. S-Nitrosylation detection method3.3. Proteolytic modification detection methods

Nitric oxide has been implicated in a variety ofGel-based methods for the detection of proteolytic physiological and pathophysiological conditions, in-

activity are too numerous to enumerate in this review cluding vasodilation, proliferation, differentiation,article [116,117]. Zymographic assays for serine NMDA receptor activation, septic shock, atheroscler-proteinases, cysteine proteases, aspartate proteases osis, apoptosis and hypoxia-induced pulmonary hy-and metalloproteinases have been developed over the pertension [119,120]. The intracellular reaction ofyears. For example, a fluorescence-based protease the higher oxides of nitric oxide (NO , N O ,2 2 3

2assay was recently developed for the detection of OONO ) with reactive cysteine groups leads to thedifferent members of the papain family of cysteine generation of protein S-nitrosothiols. Like proteinproteases within a complex cell lysate [118]. Four phosphorylation, protein S-nitrosylation could poten-different fluorescent BODIPY dye conjugates of the tially represent a fundamental mechanism for thegeneral cysteine protease inhibitor trans-epoxysuc- reversible post-translational control of protein activitycinyl-L-leucylamido (4-guanidino)butane were syn- and by extension cellular functions [119–121].thesized, allowing multiplexed analysis of papain The difficulty in devising specific gel-based assaysfamily proteases. Covalent modification of the pro- for particular post-translational modifications shouldteases through the active site cysteine residue allows be readily apparent from the preceding discussion ofanalysis of changes in enzyme activity. For example, phosphorylation detection systems. To date, radio-

32a specimen can be divided in two, and one sample labeling with P provides the bulk of the data onincubated with the red-fluorescent BODIPY 588/616 protein phosphorylation, with anti-phosphotyrosineconjugate. The other sample can then be incubated antibodies and most recently monitoring 80-Da shiftswith a specific protease inhibitor, along with the in peptide mass by mass spectrometry serving asblue-fluorescent BODIPY 493/503 conjugate. Next, very useful adjunct detection technologies. Until

Page 12: Detection Technologies in Proteome Analysis

771 (2002) 3–3114 W.F. Patton / J. Chromatogr. B

recently, the evaluation of protein S-nitrosylation in grams [123,124]. A consensus sequence for argininebiological systems has been hampered by the lack of methylation has been identified, the RGG box RNA-a detection technology to conveniently monitor this binding motif, allowing identification of numerouslabile post-translational modification. Radioactive proteins as potential substrates of arginine methyla-isotopes of nitrogen or oxygen, the atoms forming tion [125]. Other arginine methylation motifs havethe nitric oxide adduct are not available, and the half been identified as well, such as RXR, DRG and GRAlife of protein S-nitrosylation is in the few second to [126,127]. Unlike S-nitrosylation, arginine methyla-few minute range. Protein S-nitrosylation is not tion is essentially an irreversible post-translationalstable enough to survive the rigors of denaturing modification, as the generated aminoalkyl bond isSDS–polyacrylamide gel electrophoresis, so while quite stable and no dimethylarginine demethylase hasanti-S-nitrosocysteine antibodies have been gener- been identified to date [126]. Arginine methylationated and applied in immunohistochemistry studies, has been noted in proteins involved in signal trans-they are of little use in the context of proteomics duction, nuclear RNA processing, ribosome biogen-investigations. esis, nucleocytoplasmic transport, and gene silencing

Recently, the Biotin Switch method was estab- [125,128].lished to allow detection of protein S-nitosylation As with protein S-nitrosylation, investigations intoafter SDS–polyacrylamide gel electrophoresis and the role of protein arginine methylation have beenelectroblotting [122]. The procedure allows specific fettered somewhat by a lack of technologies facilitat-biotinylation of S-nitrosylated cysteine residues in ing the detection of this post-translational modi-proteins. First, free thiols are blocked by incubation fication on a proteome-wide scale [129]. Yeast cells,with a thiol-specific methiolating agent (methyl in contrast to the majority of other cell types,methanethiosulfonate or MMTS) in the presence of actively import the biological methyl donor S-sodium dodecyl sulfate to ensure access to buried adenosylmethionine, allowing radiolabeling of pro-cysteine residues. The methiolating agent does not teins in intact cells under physiological conditions

3react with nitrosothiols or disulfide bonds under the [128]. Consequently, radiolabelling with [methyl- H]reaction conditions employed. Excess MMTS is S-adenosyl-L-methionine, followed by electropho-removed using a spin column or by acetone precipi- resis and fluorography, is a commonplace approachtation. Next, nitrosothiol bonds are selectively de- to the study of arginine methylation in this organismcomposed to thiols with ascorbate. Finally the free [125,128]. In eukaryotic organisms lacking an activethiol groups are reacted with the sulfhydryl-specific import mechanism, such as mammals and insects,biotinylation reagent, N-[6-(biotinamido)hexyl]-39- cultured cells are incubated in medium containing(29-pyridyldithio)propionamide (biotin-HPDP). The protein synthesis inhibitors and then in methionine-

3labeled proteins may then be separated by SDS– free medium containing [methyl- H] S-adenosyl-L-polyacrylamide gel electrophoresis, electroblotted methionine for several hours [130,131]. Conventionaland detected by standard immunoblotting methods gel electrophoresis and fluorography are then em-using anti-biotin antibodies or streptavidin. Labeled ployed to detect the modified proteins.proteins may be selectively enriched using strep- Mass spectrometry is playing an increasinglytavidin affinity chromatography as well. The Biotin important role in the characterization of arginineSwitch method has permitted new proteomic in- methylation sites [126,127,132–134]. For example,vestigations into the fundamental role of protein the 24-kDa high-molecular-mass fibroblast growthS-nitrosylation in cellular signalling [121,122]. factor-2 has been demonstrated to contain seven to

eight dimethylated arginine residues (28 Da per3.5. Arginine methylation detection methods residue) using a combination of MALDI-TOF mass

spectrometry and Edman sequencing [127]. SuchArginine methylation was first noted more than 30 studies are performed on a protein by protein basis,

years ago, and study of this post-translational modi- with candidates suspected to contain the modifica-fication has benefited from renewed interest due to tion.the completion of several genome sequencing pro- An immunological approach to the global de-

Page 13: Detection Technologies in Proteome Analysis

771 (2002) 3–31 15W.F. Patton / J. Chromatogr. B

tection of arginine methylation has recently been tion and nonenzymatic mono-ADP ribosylationreported that does not require radiolabeling of speci- [139]. Mono-ADP ribosylation is confirmed ifmens [129]. At the heart of the procedure are labeled proteins that have been electroblotted andmonoclonal antibodies specific for mono- and treated with snake venom phosphodiesterase, gener-

32asymmetric dimethylated arginine residues (Abcam, ate [ P]59-AMP, as identified by TLC or HPLCCambridge, UK). Proteins are first immuno- analysis. Nonenzymatic mono-ADP ribosylation canprecipitated with antibodies to methylarginine, the be excluded as a possibility if incubation of samples

32immunoprecipitated proteins are separated by SDS– with [ P]ADP ribose does not cause labeling of thepolyacrylamide gel electrophoresis, transferred to proteins.membranes by electroblotting and specific proteins An alternative approach to detecting ADPare then detected by immunoblotting or total protein ribosylation is the use of anti-ADP ribose antibodiesstaining. Unfortunately, while the anti-methylar- by standard immunoblotting methods [143]. Sinceginine antibodies are suitable for immunoprecipita- ADP ribose-conjugated serum albumin results in 59-tion, they do not appear to be suitable for Western AMP-specific antibodies due to hydrolysis, an ADPblot analysis, leaving assignment of the post-transla- ribose analog resistant to enzymatic degradation attionally modified proteins somewhat tentative. Con- the pyrophosphate group was coupled to bovinetrol experiments are suggested in order to validate serum albumin in order to generate an antigenthe assignment of the post-translational modification appropriate for the production of anti-ADP ribose[129]. Recombinant protein substrates may be antibodies. The antibody is reported to detect as littlemethylated in vitro using S-adenosyl-L-methionine as 1 pmol of ADP ribosylated proteins. Additionally,and used to monitor the effectiveness of the immuno- 1,6-etheno NAD, a fluorescent analog of NAD, mayprecipitation step. Furthermore, preincubation of the be employed to measure ADP ribosylation [144–antibodies with unmethylated or in vitro methylated 146]. The fluorophore excites maximally at 292 nmprotein as competitors can serve as an additional and emits maximally at 413 nm.control for the specificity of the immunoprecipita-tion.

4. Detection of reporter enzymes and epitopetags

3.6. ADP-ribosylation detection methodsSeveral recombinant DNA-based strategies for

Mono-ADP ribosylation is a reversible post-trans- appending affinity tags and reporter enzymes tolational modification of proteins catalyzed by en- proteins have been developed as tools for proteinzymes that transfer the ADP-ribose moiety of NAD purification and detection. These include fusion tagsto specific amino acid residues, especially arginine of maltose-binding protein, hemagglutinin, b-gal-residues, on protein substrates [135]. Among the actosidase, b-glucuronidase, green fluorescent pro-known protein substrates for ADP-ribosyltransferases tein, glutathione S-transferase, c-myc oncoprotein,are small RAS-like membrane GTP-binding proteins. FLAG peptide or oligohistidine peptide. b-Galacto-Nonenzymatic ADP ribosylation of cysteine residues sidase, b-glucuronidase, the oligohistidine peptidein certain proteins, such as nuclear GTP-binding tag and green fluorescent protein are discussed inproteins has also been observed [136]. ADP ribosyla- more detail below:tion is known to regulate membrane trafficking andactin cytoskeleton arrangement [137]. 4.1. b-Glucuronidase

Mono-ADP ribosylation of proteins is typicallymonitored by incubating cells or membranes with The Escherichia coli b-glucuronidase-encoded

32[ P]NAD [138–142]. Protein labeling in the pres- gene, gusA, is commonly used as a reporter gene for32ence of [ P]NAD may arise from several different nonvertebrate organisms that minimally express endo-

reactions, however. Two important ones to consider genous b-glucuronidase activity such as lower andin experiments of this type are poly-ADP ribosyla- higher plants, most bacteria, yeast, fungi and insects

Page 14: Detection Technologies in Proteome Analysis

771 (2002) 3–3116 W.F. Patton / J. Chromatogr. B

[147]. As plants express significant amounts of b- SYPRO Tangerine dye must intercalate into SDSgalactosidase endogenously, b-glucuronidase is an micelles for protein detection while measuring b-especially useful reporter enzyme for molecular glucuronidase activity requires prior removal of SDSbiology studies focusing upon these organisms. from the gels [55]. Combining ELF 97 b-D-glucuro-Naturally, b-glucuronidase protein levels can be nide with SYPRO Ruby dye does permit convenientdetermined by standard immunoblotting procedures dichromatic detection of b-glucuronidase activityusing anti-GUS antibody [148,149]. By measuring and total protein profiles using a single gel, however.b-glucuronidase activity directly in polyacrylamide Unlike SYPRO Tangerine dye, SYPRO Ruby dyegels, time-consuming immunoblotting steps and cost- interacts directly with basic amino acids in proteins.ly antibodies are avoided, however. b-Glucuronidase This electrostatic binding mechanism allows stainingactivity can be detected directly in gels using of proteins after renaturation of enzyme activitycolorimetric detection systems such as naphthol-AS- through removal of SDS from the gels [147].BI b-D-glucuronide coupled to Fast Garnet GBC,6-bromo-2-naphthyl b-D-glucuronide coupled to Fast 4.2. b-GalactosidaseBlue B or 5-bromoindol-3-yl b-D-glucuronide incombination with nitroblue tetrazolium [116,150]. The b-galactosidase gene (lacZ) is commonlyGels are incubated in the stain solution until colored used as a reporter enzyme for the evaluation of genebands appear and then fixed in an acidified aqueous regulation, the detection of recombinant fusionsolution of 7% acetic acid or 10% trichloroacetic genes, monitoring cell transfection efficiencies andacid. The fluorogenic substrate 4-methylumbelliferyl the detection of protein–protein interactions by theb-glucuronide is more commonly used to measure two-hybrid system [152]. In a parallel manner asb-glucuronidase activity in gels [116]. Typically, the with b-glucuronidase, b-galactosidase protein levelssubstrate is applied to the gel surface soaked into may be determined by standard immunoblottingfilter paper. After incubating the gel at 37 8C for 90 procedures [153]. This enzyme may also be detectedmin, the filter paper is removed and fluorescent directly in gels using a number of assays based uponbands are detected using long wavelength ultraviolet enzyme activity [116]. Similar to b-glucuronidase,(UV-A) illumination in combination with a 420-nm colorimetric and fluorogenic substrates for b-galacto-long-pass emission filter. The hydrolytic product of sidase, including 5-bromo-4-chloro-3-indolyl b-D-this substrate does not precipitate at the site of galactopyranoside, alizarin-b-D-galactoside, p-naph-enzyme activity, however, making the compound tholbenzein-b-D-galactoside, cyclohexenosculetin-b-less than ideal for localization studies in poly- D-galactoside, 4-methylumbelliferyl b-D-galacto-acrylamide gels due to a tendency to diffuse after a pyranoside, 3-carboxyumbelliferyl b-D-galacto-period of time. Additionally, the emission of the pyranoside, resorufin-b-D-galactopyranoside orproduct, 4-methylumbelliferone, is maximal at 456 fluorescein mono-b-D-galactopyranoside, have beennm, which overlaps with the emission of polyester developed for detecting b-galactosidase activity inbackings commonly used to support pre-cast poly- agar-based colonies or in gel-based zymographyacrylamide gels. An analogous fluorogenic substrate, [116,154–157]. In an alternate coupled enzymefluorescein mono-b-D-glucuronide was recently em- detection system, gels are incubated in lactose, whichployed in microchip capillary electrophoresis for is hydrolyzed to D-glucose and D-galactose by b-rapid enzyme assay [151]. galactosidase [158]. Galactose dehydrogenase sub-

A fluorescent serial staining method based upon sequently converts D-galactose to D-galactono-1,4-detection of total protein after SDS–polyacrylamide lactone, utilizing the cofactor NAD. The resultinggel electrophoresis using SYPRO Tangerine dye in NADH is coupled to a phenazine methosulfatesaline solution, followed by enzyme renaturation in (PMS)/methyl thiazolyl blue (MTT) chromogenic0.1% Triton X-100 and detection of b-glucuronidase system to form the colored product formazan. Final-activity using ELF 97 b-D-glucuronide was described ly, in yet another method, salicyl b-D-galactosiderecently, but unfortunately, simultaneous detection of generates D-galactose and salicylic acid when hydro-total protein and enzyme activity was not possible as lyzed by b-galactosidase. Addition of terbium chlo-

Page 15: Detection Technologies in Proteome Analysis

771 (2002) 3–31 17W.F. Patton / J. Chromatogr. B

ride and EDTA produces a ternary luminescent ing immunoblotting, reporter enzymes or secondarylanthanide chelate complex with excitation maximum detection reagents. Pro-Q Sapphire 365 oligohis-of roughly 280 nm and emission maximum of 515 tidine gel stain emits maximally at 450 nm (bluenm [159]. fluorescence) when illuminated with 300–365 nm

ultraviolet light. Pro-Q Sapphire 488 oligohistidine4.3. Oligohistidine tags gel stain emits maximally at 515 nm (green fluores-

cence) when illuminated with visible light from aThe oligohistidine domain is a transition metal- laser-based gel scanner equipped with a 470–488-nm

binding peptide sequence comprising a string of four laser source. Roughly 25–65 ng of oligohistidine-to 10 histidine residues. Recombinant proteins are tagged fusion protein is detectable using either stain.often fused to oligohistidine affinity tag sequences to After recording the fluorescence signal from theaid in their purification by immobilized nickel-ion Pro-Q Sapphire dyes, gels may be post-stained withaffinity chromatography. The detection of oligohis- the red-fluorescent SYPRO Ruby protein gel stain intidine-tagged fusion proteins after polyacrylamide order to reveal the total protein profile (Fig. 3).gel electrophoresis typically requires many time-consuming steps, including the transfer of the gel to 4.4. Green fluorescent proteina membrane, the blocking of unoccupied sites on themembrane with protein or detergent solutions, incu- Wild-type green fluorescent protein (GFP) frombation with an oligohistidine-binding agent (primary the jellyfish Aequorea victoria, as well as its differ-antibody or biotin-nitrilotriacetic acid), incubation ently colored variants, are useful tools commonlywith a secondary detection agent (antibody–reporter employed to monitor gene expression and proteinenzyme conjugate, streptavidin–reporter enzyme localization when expressed as chimeric fusion pro-conjugate), and incubation with a visualization re- teins [169]. Wild type GFP displays excitation max-agent (colorimetric, fluorogenic or chemiluminescent ima of 395 and 475 nm and an emission maximumreagent). Specifically, biotinylated nitrilotriacetic of 509 nm. Mutants of the GFP protein have beenacid has been used in combination with streptavidin– devised with visible excitation maxima of 490 nm,alkaline phosphatase or streptavidin–horseradish per- providing excellent compatibility with fluoresceinoxidase conjugates and colorimetric or chemilumi- filter sets and argon ion laser sources. From anescent substrates to allow detection of the oligohis- proteomics perspective, it should be noted that thetidine-tagged fusion proteins after electroblotting introduction of a single GFP reporter gene, used to[160–162]. Additionally, direct reporter enzyme– monitor cell metabolism, can profoundly effect anitrilotriacetate–nickel conjugates have been used to host of cellular processes, including carbon metabo-detect oligohistidine-tagged fusion proteins on elec- lism, protein synthesis and protein translation [170].troblots [163,164]. Likewise, colloidal gold–nitrilo- Using pH 4–7 isolectric focusing gels and 12% Ttriacetic acid conjugates have been used to detect SDS–polyacrylamide gels, it has been demonstratedoligohistidine-containing fusion proteins on blots that the introduction of GFP into Escherichia coliafter a silver enhancement step [165]. Lastly, though strain JM105 using pBluescript expression vectoroligohistidine is not very immunogenic, several high (Stratagene, La Jolla, CA) leads to the expression ofaffinity monoclonal antibodies specific to the epitope 68 new proteins (including 50S ribosomal proteinhave been produced that allow detection of fusion L9, serine methylase, aerobic glycerol-3-phosphateproteins by standard immunoblotting techniques dehydrogenase and amino acid ABC transporter[166–168]. binding protein) and the loss of 44 others (including

Two novel fluorophore–nitrilotriacetic acid conju- sulfate starvation-induced protein 7, histidine-bind-gates, Pro-Q Sapphire 365 and Pro-Q Sapphire 488 ing periplasmic protein and glucose-permease IIAoligohistidine gel stains (Molecular Probes, Eugene, component) [170]. Numerous quantitative proteinOR), allow for the fluorescence detection of oligohis- changes were observed as well.tidine-tagged fusion proteins directly in sodium Recent studies indicate that GFP may be isolateddodecyl sulfate–polyacrylamide gels, without requir- with high yield and purity employing serial copper

Page 16: Detection Technologies in Proteome Analysis

771 (2002) 3–3118 W.F. Patton / J. Chromatogr. B

fusion proteins, the two-step chromatography pro-cedure for GFP is relatively straightforward andlikely to be applicable in many instances to GFP-fusion proteins as well. In the context of electro-phoretic separations methodologies, the native fluo-rescence of GFP has had limited utility, as its signalis destroyed by conventional denaturing SDS–poly-acrylamide gel electrophoresis. Consequently, over-expressed GFP fusion proteins are typically detectedby Coomassie Blue staining, silver staining or byimmunoblotting techniques using anti-GFP antibo-dies [172–175]. GFP may be visualized directly inSDS–polyacrylamide gels if a nonreducing samplebuffer is used to prevent complete protein denatura-tion [176,177].

A GFP-protein A fusion protein has been con-structed for use as a single-step detection reagent forWestern blotting [178,179]. As with any directfluorophore conjugate of an antibody, protein A orstreptavidin, sensitivity is limited by the overallbrightness of the probe, and reporter enzyme con-structs employing alkaline phosphatase, horseradishperoxidase or b-galactosidase are generally muchmore sensitive due to the signal amplification af-forded by rapid turnover of large numbers of sub-Fig. 3. Serial dichromatic detection of oligohistidine-tagged re-strate molecules.combinant proteins and total proteins in the same polyacrylamide

gel. Detection of oligohistidine-containing fusion proteins using Certain specialized applications of GFP fluores-Pro-Q Sapphire 365 oligohistidine gel stain kit (Molecular Probes) cence in gel electrophoresis have been reported. For(top image). This nitrilotriacetate-based detection system gener-

instance, a GFP–calmodulin fusion protein has beenates blue fluorescent bands in gels. Detection of total proteingenerated, permitting rapid, nonradioactve detectionprofile using SYPRO Ruby protein gel stain (Molecular Probes)of interacting proteins in gel overlay assays (Far-(bottom image). This colloidal metal chelate staining procedure

generates orange–red fluorescent bands in gels. Lanes: (1) broad Western blots) [176]. Proteins are fractionated byrange molecular mass markers; (2) 6Xhis protein ladder (Qiagen); conventional SDS–polyacrylamide gel electropho-(3) protein lysate containing hexahistidine-tagged urate Oxidase

resis, electroblotted to poly(vinylidene difluoride)(Pierce Corporation); (4–6) serial dilutions of Escherichia coli(PVDF) membrane, blocked with 2% nonfat milklysate containing the recombinant oligohistidine fusion protein,protein and incubated with 2–5 mg/ml of the GFPOSCP; (7–9) serial dilutions of Escherichia coli lysate containing

the recombinant oligohistidine fusion protein, a-subunit of ATP- fusion protein over night. After washing the mem-ase; (10–12) serial dilutions of Escherichia coli lysate con- brane several times, as little as 100 ng of cal-taining the recombinant oligohistidine fusion protein, d subunit of

modulin-binding proteins may be visualized by UVATPase; (13) bovine serum albumin. Both images were collectedepi-illumination. Detection sensitivity using biotin-using A LumiImager CCD-camera-based gel scanner (Rocheylated calmodulin, horseradish peroxidase-conju-Biochemicals). Figure courtesy of Ms. Courtenay Hart, Molecular

Probes. gated streptavidin and luminol was found to bevastly superior to that obtained by the direct GFP–

(II) and zinc (II) immobilized metal-ion affinity calmodulin fusion protein, though the conventionalcolumns, using diafiltration between the two chroma- detection method required more steps.tography steps [171]. Though more complex than the In another example, native gel electrophoresis andstandard nickel (II) immobilized metal-ion affinity affinity capillary electrophoresis with laser-inducedcolumn used for isolation of oligohistidine-tagged fluorescence have been employed to monitor the

Page 17: Detection Technologies in Proteome Analysis

771 (2002) 3–31 19W.F. Patton / J. Chromatogr. B

interaction of a GFP-cyclophilin fusion protein well [182]. Images of the 2D gels are acquired using(‘‘bait’’ protein) with cyclosporin A [177]. The as many as three different excitation /emission filters,native gel band-shift assay was performed using and the ratio of the differently colored fluorescentTris–borate–EDTA electrophoresis gels and fluores- signals is used to find protein differences among thecence signal was visualized using a UV transil- samples. DIGE allows two to three samples to beluminator /CCD camera system. Roughly 1 mg of separated under identical electrophoretic conditions,purified recombinant protein was required to obtain a simplifying the process of registering and matchingband of moderate intensity with the approach. The the gel images. Most often, the technology is used toaffinity capillary electrophoresis band-shift assay examine differences between two samples (e.g.,allowed detection of the protein–protein interaction drug-treated versus control cells or diseased versusin phosphate buffer without addition of detergent. healthy tissue). Though seductive in concept, itConsequently, weak and moderate interactions could should be noted that a few problems afflict DIGEbe monitored using 60-fold lower amounts of protein technology, especially when endeavoring to performand significantly shorter analysis times. Subsequent- quantitative proteomics analyses.ly, both band shift assays were applied to extracts of One requirement of DIGE is that only |1–2% ofDrosophila melanogaster embryos, and a putative the lysine residues in the proteins be fluorescentlyhigh affinity endogenous ligand of cyclophilin was modified, so that the solubility of the labeled proteinsdetected. Since GFP is larger than its fusion partner, is maintained during electrophoresis [180]. Theit is necessary to establish in control experiments to solubility of the proteins is decreased considerablyensure that any observed interactions arise from the when higher degrees of substitution are employed,bait component of the fusion protein. resulting in the detection of many fewer protein spots

on 2D gels. This limited covalent labeling step setsthe lower limit for fluorescence detection to 1–2% of

5. Differential display proteomics: approaches its intrinsic limit, given an alternate fluorescenceand options imaging technology that does not have this solubility

constraint. In experimental terms, optimized DIGEProteomics is often based upon the comparison of technology is capable of detecting about half as

different protein profiles. The central objective of many proteins as conventional silver staining, ordifferential display proteomics is to increase the about four times as many proteins as colloidalinformation content of proteomics studies through Coomassie Blue staining [181]. It is almost certainmultiplexed analysis. Currently, three principal ap- that the proteins missed by DIGE staining representproaches to differential display proteomics are being the lowest abundance species in the sample. DIGEactively pursued, difference gel electrophoresis technology is more sensitive than modified silver(DIGE), multiplexed proteomics (MP) and isotope- stain formulations optimized for mass spectrometry,coded affinity tagging (ICAT). The basic principles though SYPRO Ruby dye staining detects 40% moreof these three methods in the context of 2D gel protein spots than the Cy dyes [183].electrophoresis are outlined in Figs. 4, 5 and 7. While matched in charge, the dye-labeled proteins

produced by DIGE technology migrate with slightly5.1. Difference gel electrophoresis (DIGE) higher mass than the bulk of the unlabeled protein,

due to the addition of the fluorophores [180,181].Succinimidyl esters of the cyanine dyes, Cy2, Cy3 This produces registration errors between the labeled

and Cy5, may be employed to fluorescently label as and unlabeled proteins in 95% of the spots, andmany as three different complex protein populations consequently complications arise when spots are toprior to mixing them together and running them be excised and analyzed by mass spectrometry [181].simultaneously on the same 2D gel using a method As a result, gels must be post-stained to aid inreferred to as difference gel electrophoresis (DIGE) protein cutting for spot identification by mass spec-(Fig. 4) [180,181]. Recently, a similar procedure trometry [181,183,184]. Regrettably, since colloidalutilizing Alexa Fluor dyes has been demonstrated as Coomassie Blue staining is less sensitive than the

Page 18: Detection Technologies in Proteome Analysis

771 (2002) 3–3120 W.F. Patton / J. Chromatogr. B

Fig. 4. Schematic illustration of the difference gel electrophoresis (DIGE) technology platform [180]. Two different samples are derivatizedwith two different fluorophores, combined and then run on a single 2D gel. Proteins are detected using a dual laser scanning device or xenonarc-based instrument equipped with different excitation /emission filters in order to generate two separate images. The images are thenmatched by a computer-assisted overlay method, signals are normalized, and spots are quantified. Differences in protein expression areidentified by evaluation of a pseudo-colored image and data spreadsheet. DIGE technology can maximally evaluate three different samplesusing Cy2-, Cy3- and Cy5-based chemistries.

DIGE technology, 40% of the proteins of interest aware of that can effect sensitivity of detection iscannot be identified directly from the Coomassie that fluoresence cross-talk between excitation /emis-Blue-stained gel [181]. Consequently, SYPRO Ruby sion band-pass filters may cause Cy5-labeled pro-dye has been adopted as the post-stain dye of choice teins to emit when illuminating at the optimal Cy3for DIGE analysis [183,184]. Another issue to be excitation wavelength [185].

Page 19: Detection Technologies in Proteome Analysis

771 (2002) 3–31 21W.F. Patton / J. Chromatogr. B

Fig. 5. Schematic illustration of the multiplexed proteomics (MP) technology platform [85]. Two different samples are run on two different2D gels. Gels are stained for a particular functional attribute, such as glycosylation, and then imaged. Gels are then stained for total proteinexpression using SYPRO Ruby protein gel stain and imaged again. The images resulting from glycoprotein staining and the images resultingfrom total protein staining are then matched by computer-assisted rubber sheeting and overlay methods, signals are normalized (if necessary)and spots are quantified. Differences in glycosylation and protein expression are identified by evaluation of pseudo-colored images and dataspreadsheets. Registration of total protein profiles with glycosylation patterns is readily accomplished by direct superimposition of imagesfrom the same gel. MP technology can evaluate an almost limitless number of 2D gels with respect to two or three different attributes(protein expression, post-translational modification, drug-binding capability, drug-metabolizing capability).

Page 20: Detection Technologies in Proteome Analysis

771 (2002) 3–3122 W.F. Patton / J. Chromatogr. B

It has been demonstrated recently that 20% absent from normal tissue and annexin I absent fromchanges in very abundant proteins and 800% changes tumor tissue. In a study comparing an ErbB-2-re-in scarce ones can be quantified reliably at the 90% ceptor tyrosine kinase transformed luminal epithelialconfidence level using DIGE technology [181]. Thus, cell line to its untransformed counterpart, 135 pro-though the DIGE technology can be applied to the teins with a 1.3-fold average difference in abundancedetection of qualitative changes in protein expres- were identified. Though many of these proteins weresion, quantitative changes in lower abundance pro- identified by mass spectrometry, none of the expres-teins are likely to be overpowered by dye-dependent sion changes were confirmed by an independentphotophysical effects. A concrete illustration of the method, such as Western blotting.qualitative nature of DIGE technology may be found Another important issue with respect to the DIGEwith regard to the study of protein changes in heart technology is that it is predicated upon an overlymitochondria of knockout mice lacking sarcomeric simplistic tenet that all of the information required inmitochondrial creatine kinase [186,187]. Mitochon- an analysis can be distilled down into a single 2D geldrial creatine kinase plays a central role in cellular using a simple control versus treated state ex-energetics of muscle, catalyzing the reversible trans- perimental design. However from a biology perspec-fer of the phosphoryl group between phosphocreatine tive, the analysis of drug dose–response curves orand ADP [188,189]. Deletion of such a vital enzyme the kinetics of changes in tissue /cell proteomesis expected to have proteome-wide repercussions, as require that multiple 2D gels be quantified. Thecells are forced to adapt to the metabolic crisis. assessment of different stages of cancer progressionConventional Western blot analysis has demonstrated and epidemiological studies of global protein expres-loss of the target kinase, as well as 2–4-fold changes sion patterns in populations also require many morein a number of proteins, including mitochondrial than the two to three sample capacity of the DIGEcytochrome c oxidase, inorganic phosphate carrier, technique. Therefore, multiple 2D gels almost alwaysadenine nucleotide translocator and voltage-depen- must be quantified and referenced to one another,dent anion channel proteins, consistent with the regardless of the method employed to detect theimpaired cardiac energetics observed in these ani- proteins. This necessity to quantify multiple 2D gelsmals [187]. Proteome-wide analysis of the mito- effectively contradicts the whole premise behind thechondria by DIGE technology confirmed the com- DIGE detection technology. This shortcoming hasplete absence of sarcomeric mitochondrial creatine been recognized by some, and an innovative methodkinase in the knockout mice, as well as a minor for linking multiple 2D gels generated by DIGEdifference in the precursor form of aconitase, but technology has recently been devised [183]. Cy2-surprisingly, no other large-scale changes in heart labeled proteins are used as internal standards withmitochondrial proteins were revealed by the tech- all Cy3- and Cy5-labeled sample pairs to aid innique [186]. Presumably, the quantitative changes quantitative comparisons between gels in a database.identified in this model system by Western blotting Even if the biological problem can realistically becould not be differentiated from background fluctua- framed as a simple binary experimental /controltions associated with the DIGE labeling methodolo- study, it should be noted that DIGE technologygy. cannot reliably identify differences between two

Recent DIGE studies comparing the difference samples using a single 2D gel. In an analysis of sixbetween normal and cancerous tissue were more two-color DIGE gels comparing toxin-treated tosuccessful in demonstrating changes in protein ex- control liver homogenates, a total of 138 proteinpression levels [183,184]. In a study comparing changes were identified [181]. Of these 138 differ-esophageal carcinoma cells with normal epithelial ences, however, only 4.4% (eight differences) werecells, statistical analysis revealed that 56 proteins observed in all six gels. Thirty percent (41 differ-were up-regulated by more than 3.5-fold and 39 were ences) were finally nominated as real changes indown-regulated by more than 4-fold [184]. Two protein abundance, as they occurred in three or morechanges, protein gp96 and annexin I, were confirmed gels. The authenticity of the changes was not cor-by Western blotting. It should be noted that both roborated using any independent measurement tech-validated changes were qualitative, with gp96 being nique such as Western blotting, however.

Page 21: Detection Technologies in Proteome Analysis

771 (2002) 3–31 23W.F. Patton / J. Chromatogr. B

Radiolabeling is probably the only method that protein gel stain as the foundation staining technolo-can compete with fluorescence in terms of linear gy. Differential display comparisons are made bydynamic range and detection sensitivity. However, computer, using image analysis software, such asradio-imaging procedures for differential display Compugen’s Z3 program [190]. All gels are imaged

125 131such as the I / I multi-photon detection (MPD) using the same excitation /emission filter sets andmethod devised by BioTraces (Herndon, VA) suffer resulting images are then automatically matched,from the same experimental design limitations as the with the option of adding some manual anchor pointsfluorescence DIGE technology. The inherent slow- to facilitate the process. Any two images can then beness of radioactive 2D gel imaging also severely re-displayed as a single pseudo-colored map, permit-limits this technology. Currently, 2.5 days are re- ting the visual inspection of differences in an identi-quired to collect an image with attomole detection cal manner as performed with DIGE technology. Insensitivity. Throughput might be improved in a addition, quantitative information can readily beproteomics work environment by employing multiple obtained in tabular form, with differential expressionimaging devices, but the instruments are expensive data calculated. One of the strengths of the platformto manufacture, costing around US $100 000 apiece. is that similar profiles, such as total protein patterns,By contrast, a typical 2D gel fluorescently labeled are matched by computer from different gels, whilewith a total-protein dye, such as SYPRO Ruby dissimilar ones, such as total protein patterns andprotein gel stain, may be imaged in 0.1–2 s on a glycoprotein patterns, are matched by computer fromwide range of imaging platforms, approximately the same gel. Thus, though few landmarks are shared200 000 times faster than radiolabel imaging. between total protein profiles and the assayed func-

tional property, the images may be matched bysuperimposition anyway.

5.2. Multiplexed proteomics (MP) A recently published study demonstrates thefeasibility of the MP platform, describing methods to

The multiplexed proteomics (MP) platform is assay for the presence of glycans, followed by thedesigned to allow the parallel determination of detection of the total protein profile [85]. Pro-Qprotein expression levels as well as certain functional Emerald 300 dye, a sensitive green-fluorescentattributes of the proteins, such as levels of glycosyla- glycoprotein-specific stain is employed, having non-tion, drug-binding capabilities or drug-metabolizing overlapping spectral properties with the red-fluores-capabilities (Fig. 5). The MP technology platform cent total protein stain, SYPRO Ruby dye [85]. Thisutilizes the same fluorophore to measure proteins hydrazide dye may be affixed to glycoproteins usingacross all gels in the database, and employs additional a periodic acid Schiff’s (PAS) reaction under veryfluorophores with different excitation and/or emis- mild coupling conditions. Proteins may then be post-sion maxima to accentuate specific functional attri- stained with SYPRO Ruby dye, allowing sequentialbutes of the specimen. As stated previously, bio- two-color detection of glycosylated and non-logical investigations usually require large numbers glycosylated proteins, an unprecedented new capa-of 2D gels and methods are required that facilitate bility in proteomics research.the rigorous and reliable matching of proteins across Other functional attributes of protein samples thata database. A clear advantage of MP technology over may be detected in 2D gels with subsequent post-DIGE technology is that it provides a much broader staining by SYPRO Ruby dye include the detectionbandwidth of information. A limitless number of gels of b-glucuronidase activity using ELF 97 b-D-gluc-may be evaluated with respect to a number of uronide, the detection of penicillin-binding proteinsdistinct attributes. using fluorescent Bocillins, and the detection of

With the MP platform, a set of 2D gels is oligohistidine-tagged proteins using Pro-Q Sapphirefluorescently stained and imaged to reveal some dye [147,191]. These examples illustrate four areasfunctional attribute of the proteins, such as drug- in which the MP platform can find application:binding capability, or a particular post-translational analysis of post-translational modifications, en-modification. Then, protein expression levels are zymatic activity, small ligand binding and epitoperevealed in the same gels using SYPRO Ruby tags. Other procedures for detecting hydrolase activi-

Page 22: Detection Technologies in Proteome Analysis

771 (2002) 3–3124 W.F. Patton / J. Chromatogr. B

ty, and cell surface proteins are actively under pairs does vary in composition and concentration anddevelopment. thus quantification of the peptides occurs in different

matrix environments [197]. The deuterated and non-5.3. Isotope-coded affinity tagging (ICAT) deuterated peptides often separate from each other

during liquid chromatography, and are ionized atIsotope-coded affinity tag (ICAT) peptide labeling different times in different matrixes. This can strongly

is predicated upon distinguishing between two popu- impact the ionization efficiency of peptides across alations of proteins using isotope ratios [192–196]. chromatographic peak, leading to errors in quantita-Isotope-coded affinity tags are reagents consisting of tion. In addition, the ICAT tag is a relatively largea protein-reactive group, an ethylene glycol linker molecule compared to the size of some small pep-region and a biotin tag. Currently, the commercial- tides and it has been suggested that the tag mayized ICAT reagent employs a reactive functionality interfere with peptide ionization, leading to compli-specific for the thiol group of cysteine residues in cations in the interpretation of mass spectra [198].proteins and peptides. Two different isotope tags are One significant obstacle to the routine im-generated by using linkers that contain either eight plementation of ICAT technology in the study ofhydrogen atoms (d0, light reagent) or eight biological systems is that proteolytic digests of cellsdeuterium atoms (d8, heavy reagent). A reduced or tissues are exceedingly complex, with literallyprotein mixture from one protein specimen is deriva- hundreds of thousands of peptides generated in atized with the isotopically light version of the ICAT sample [196]. Though isolation of cysteine-contain-reagent, while the other reduced protein specimen is ing peptides reduces sample complexity perhaps 10-derivatized with the isotopically heavy version of the fold, other fractionation methods are usually requiredICAT reagent. Next, the two samples are combined, to reduce the number of peptides delivered to theand proteolytically digested with trypsin or Lys-C to mass spectrometer further. Recently, the microsomalgenerate peptide fragments. The combined sample is fraction of control versus differentiated HL-60then fractionated by avidin affinity chromatography, human myeloid leukemia cells were evaluated by thethus retrieving only the cysteine-containing peptides. ICAT method [199]. After labeling with heavy andThe affinity-purified mixture of peptides contains light ICAT reagents and combining the samples, theroughly 10-fold fewer peptides than the original peptide mixture was subjected to cation-exchangesample, thus simplifying analysis. The peptides may chromatography, followed by avidin affinity chroma-be further separated by microcapillary reversed- tography, a multidimensional chromatographic ap-phase liquid chromatography, followed by mass proach. Finally, peptides were separated and ana-spectrometry. The ratio of the isotopic molecular lyzed by LC–ESI-MS. A total of 491 proteins weremass peaks that differ by 8 Da, as revealed by mass unambiguously identified and the ratio of eachspectrometry, provides a measure of the relative protein between control and differentiated culturesamounts of each protein from the original samples. was determined.Typically, nanoscale liquid chromatography–electro- One weakness of the current ICAT method is thatspray ionization mass spectrometry (LC–ESI-MS) it requires that the proteins contain cysteine residuesprovides quantitative information based upon the flanked by appropriately spaced protease cleavagerelative abundances of the heavy and light peptides, sites [198]. The problem has recently been high-while nanoscale liquid chromatography combined lighted in the study of the multi-subunit membranewith electrospray ionization tandem mass spec- protein, Escherichia coli F –F –ATP synthase1 0

trometry (LC–ESI-MS–MS) provides qualitative [200]. This enzyme consists of distinct extramem-information based upon the peptide molecular mass branous and transmembranous components of knownand amino acid sequence information [196]. subunit stoichiometry, and thus provides a good

Though quantitation by mass spectrometry is often model for the combination of hydrophobic andunreliable, due to ion suppression phenomena, ICAT hydrophilic proteins normally encountered inreagents are chemically identical and differ in mass proteomics studies. The membrane-associated F0

by only eight neutrons. Nevertheless, it has been component is composed of three subunits, a, b, andsuggested that the background of co-eluting peptide c, polypeptides with five, one and two transmem-

Page 23: Detection Technologies in Proteome Analysis

771 (2002) 3–31 25W.F. Patton / J. Chromatogr. B

brane domains, respectively. Labeling with the cys-teine-reactive fluorophore, monobromobimane, failedto detect four of the eight subunits of ATP synthase(Fig. 6). None of the membrane-embedded proteinsin the F complex were visualized. Genome-wide0

analysis of cysteine distribution in Escherichia colidemonstrates that the cysteine-labeling problem doesnot arise from a peculiar choice of model proteincomplex. In the Escherichia coli genome, the num-ber of proteins containing no cysteine residues ishigh (|10–15%), obviating the use of any cysteine-specific technology as a total-protein indicator.

This cysteine-labeling problem may be overcomeby devising different ICAT reagents, reactive forother amino acid residues in proteins. Indeed, severalalternative isotope tags have been devised that labelall proteolytic fragments in a sample. A method

14based upon the growth of cells in either N- or15N-enriched medium has been described for quan-tifying changes in protein expression by 2D gelelectrophoresis and mass spectrometry [199]. Like-

16 18wise, regular water (H O) and heavy water (H O)2 2

have been employed as the solvent during Glu-Cproteolysis of samples, leading to the incorporation

18 16of two O or two O atoms in the C-terminalmoiety of each proteolytic fragment [201]. Thisresults in a 4-Da difference in mass between pairedpeptides. Acetate (d0) and trideuteroacetate (d3)have been employed to acetylate primary aminogroups in peptides [202]. Peptide digests are sub-sequently simplified using copper-immobilized metalaffinity chromatography to selectively capture his-tidine-containing peptides. Alternatively, lectin af-finity chromatography is employed to selectivelyanalyze glycoproteins. Similarly, methyl esterifica- Fig. 6. Subunits of F F –ATP synthase revealed by two different1 0

detection methods [200]. (A) Subunits revealed using a basiction of aspartate and glutamate residues using regularamino acid binding dye, SYPRO Ruby protein gel stain. (B)methanol (d0) or trideuteromethanol (d3) has beenSubunits detected using a thiol-reactive fluorophore, monobromo-

used as an isotope tagging strategy [203]. Finally, bimane. Thiol-reactive chemistries, such as monobromobimanestandard 1,2-ethanedithiol (d0) and tetraalkyl deuter- and ICAT labeling do not globally detect all proteins in a cell orated 1,2-ethanedithiol (d4) have been employed to tissue. Roughly 10–15% of the polypeptides escape detection.

Secreted proteins and certain plant lectins tend to have a highermeasure differences between O-phosphorylation sitesthan average number of cysteine residues, while polypeptides thatin samples using b-elimination chemistry [111]. Theare components of multi-subunit protein complexes tend to have

pendant sulfhydryl group is then reacted with biotin lower than average number of cysteine residues. Figure courtesyiodoacetamidyl-3,6-dioxaoctanediamine. Just as with of Dr. Birte Schulenberg, Molecular Probes.the original cysteine-reactive ICAT reagent, thephosphate-labeling procedure allows enrichment oftarget peptides using avidin affinity chromatography. technology can be combined into a single differential

Recently, it has been demonstrated that the advan- display platform (Fig. 7) [194]. This method closelytages of 2D gel electrophoresis and ICAT labeling parallels the DIGE methodology, with some im-

Page 24: Detection Technologies in Proteome Analysis

771 (2002) 3–3126 W.F. Patton / J. Chromatogr. B

Fig. 7. Schematic illustration of the iosotope-coded affinity tagging (ICAT) technology platform [194]. Two different samples arederivatized with two different ICAT reagents; heavy (deuterated) and light (normal). The samples are combined and then run on a single 2Dgel. Proteins are detected using a sensitive mass spectrometry-compatible protein stain, such as modified silver stain or SYPRO Ruby proteingel stain. Proteins are excised from the gel, enzymatically digested in the gel matrix, extracted and analyzed by mass spectrometry. Proteinsare identified by peptide mass fingerprinting using a single stage mass spectrometer or by collision-induced dissociation of selected peptidesand database sequence searching. Differences in protein expression between individual proteins in the two samples are determined byevaluation of the ratio of signal intensities for the isotopically normal and heavy forms of a specifically tagged peptide. As illustrated, everyspot in the profile must be evaluated by mass spectrometry and cysteine-containing peptides identified. Alternatively, three gels can be runcontaining heavy isotope-labeled material, light isotope-labeled material and a mixture of the two. The first two gels are used to generatedifference maps using a sensitive protein stain and computer-based matching software. The third gel is then used to excise differencesidentified from the first two gels.

Page 25: Detection Technologies in Proteome Analysis

771 (2002) 3–31 27W.F. Patton / J. Chromatogr. B

portant improvements. It is also quite similar to the function of the biological state of the cell or tissue14 15 14N/ N labeling procedure [109]. Unlike the N/ are being developed utilizing electrophoresis, chro-15N labeling scheme, the ICAT method is a post- matography, and/or mass spectrometry as key com-isolation isotopic labeling approach that does not ponents of the detection platforms. Methods for therequire cells to be cultured in specialized media. differential display analysis of protein populationsProteins from two different samples are labeled with that parallel the capabilities of mRNA-based ap-heavy and light ICAT reagents, combined and then proaches are actively under development. Gel elec-separated by 2D gel electrophoresis. The gel-sepa- trophoresis remains a formidable analytic tool thatrated proteins are detected with a sensitive protein has found wide application in a variety of fields,stain, excised, treated with protease and identified by ranging from toxicology, and cancer discovery re-peptide mass profiling. Additionally, selected pep- search to quality control and clinical screening oftides can be evaluated further using collision-induced samples. Though efforts to displace gel-based elec-dissociation (CID) and sequence database searching. trophoretic techniques with alternatives, such asAs with DIGE, the ICAT labeling decreases the solution-phase, microarray, capillary electrophoresiselectrophoretic mobility of proteins, due to addition and microfluidic devices are underway, gel electro-of 422 Da at every cysteine residue. Since cysteine phoresis-based technology remains attractive due toresidues are modified with a neutral ICAT group, the its simplicity, reliability, higher information contentisoelectric point is preserved for all but the most and ready accessibility to researchers [204].basic proteins. While DIGE requires minimal label-ing with the hydrophobic cyanine dyes, ICAT label-ing must be performed quantitatively and to comple- Acknowledgementstion in order to avoid generating molecular massladders of spots with varying degrees of substitution The author thanks Richard Haugland, Joseph[184,185]. This is readily accomplished using excess Beechem, Tom Steinberg, Kiera Berggren, Cour-labeling reagent and since the ICAT reagent is tenay Hart, Birte Schulenberg, Karen Martin, Bradrelatively hydrophilic, migration problems do not Arnold, Mark Lim, Negin Shojaee and Lily Lam forarise during electrophoresis. Of course, the biotin numerous contributions to the development of metalcomponent of the tag is superfluous in this applica- chelate detection strategies over the past 8 years.tion and smaller cysteine-reactive reagents could beemployed, as long as a normal and deuteratedspecies are used. One important application of ICAT

Referencesdifferential display in 2D gels is for the assessmentof the relative abundances of protein isoforms that

[1] M. Dunn, Methods Mol. Biol. 112 (1999) 313.arise from post-translational modification. This uti-[2] N. Matsui, D. Smith-Beckerman, L. Epstein, Methods Mol.lizes a similar strategy as originally employed with

Biol. 112 (1999) 307.14 15the N/ N labeling procedure [109]. [3] C. Merril, in: A. Chrambach, M. Dunn, B. Radola (Eds.),Advances in Electrophoresis, Vol. 1, VCH Publishers, NewYork, NY, 1987, p. 111.

[4] W. Patton, Biotechniques 28 (2000) 944.6. Conclusions[5] W. Patton, Electrophoresis 21 (2000) 1123.[6] W. Patton, in: S. Pennington, M. Dunn (Eds.), Proteomics:

The pace of technological innovations in protein From Protein Sequence To Function, BIOS Scientific Pub-detection has markedly accelerated in the past 4 lishing, Oxford, 2001, p. 65.

[7] W. Patton, M. Lim, D. Shepro, Methods Mol. Biol. 112years. Undoubtedly, this is partly due to a shift in the(1999) 331.focus of the research enterprise from genomics to

[8] T. Rabilloud, Anal. Chem. 72 (2000) 48A.proteomics, spurred by the completion of the human[9] T. Rabilloud, Methods Mol. Biol. 112 (1999) 297.

genome project. Methods for the global analysis of [10] P. Wirth, A. Romano, J. Chromatogr. A 698 (1995) 123.protein abundance, protein post-translational modi- [11] S. Fazekas de St. Groth, R. Webster, A. Datyner, Biochim.fications and changes in these two parameters as a Biophys. Acta 71 (1963) 377.

Page 26: Detection Technologies in Proteome Analysis

771 (2002) 3–3128 W.F. Patton / J. Chromatogr. B

[12] V. Neuhoff, R. Stamm, H. Eibl, Electrophoresis 6 (1985) [44] L. Castellanos-Serra, C. Fernandez-Patron, E. Hardy, V.427. Huerta, Electrophoresis 17 (1996) 1564.

[13] R. Switzer, C. Merril, S. Shifrin, Anal. Biochem. 98 (1979) [45] L. Castellanos-Serra, C. Fernandez-Patron, E. Hardy, H.231. Santana, V. Huerta, J. Protein Chem. 16 (1997) 415.

[14] B. Oakley, D. Kirsch, N. Morris, Anal. Biochem. 105 (1980) [46] L. Castellanos-Serra, W. Proenza, V. Huerta, R. Moritz, R.361. Simpson, Electrophoresis 20 (1999) 732.

[15] C. Merril, D. Goldman, S. Sedman, M. Ebert, Science 211 [47] J. Daban, M. Samso, S. Bartolome, Anal. Biochem. 199(1981) 1437. (1991) 162.

[16] M. Quadroni, P. James, Electrophoresis 20 (1999) 664. [48] J. Daban, S. Bartolome, M. Samso, Anal. Biochem. 199[17] P. Sinha, M. Poland, M. Schnolzer, T. Rabilloud, Proteomics (1991) 169.

7 (2001) 835.[49] A. Bermudez, J. Daban, J. Garcia, E. Mendez, Biotechniques

[18] A. Shevchenko, M. Wilm, O. Vorm, M. Mann, Anal. Chem.16 (1994) 621.

68 (1996) 850.[50] F. Alba, A. Bermudez, J. Daban, Electrophoresis 22 (2001)

[19] J. Yan, R. Wait, T. Berkelman, R. Harry, J. Westbrook, C.399.

Wheeler, M. Dunn, Electrophoresis 21 (2000) 3666.[51] F. Alba, J. Daban, Electrophoresis 19 (1998) 2407.

[20] F. Gharahdaghi, C. Weinberg, D. Meagher, B. Imai, S.[52] T. Steinberg, L. Jones, R. Haugland, V. Singer, Anal.

Mische, Electrophoresis 20 (1999) 601.Biochem. 239 (1996) 223.

[21] A. Link, Methods Mol. Biol. 112 (1999) 285.[53] T. Steinberg, R. Haugland, V. Singer, Anal. Biochem. 239

[22] W. Springer, in: J. Walker (Ed.), The Protein Protocols(1996) 238.

Handbook, Humana Press, Totowa, NJ, 1996, p. 161.[54] T. Steinberg, H. White,V. Singer, Anal. Biochem. 248 (1997)[23] L. Kuhn, J. Kettman, I. Lefkovits, Electrophoresis 10 (1989)

168.708.[55] T. Steinberg, W. Lauber, K. Berggren, C. Kemper, S. Yue, W.[24] L. Luo, P. Wirth, Electrophoresis 14 (1993) 127.

Patton, Electrophoresis 21 (2000) 497.[25] J. Westbrook, J. Yan, R. Wait, M. Dunn, Proteomics 1 (2001)[56] F. Alba, A. Bermudez, J. Daban, Electrophoresis 22 (2001)370.

399.[26] J. Yeargin, M. Hass, Curr. Biol. 5 (1995) 423.[57] M. Lopez, K. Berggren, E. Chernokalskaya, A. Lazarev, M.[27] V. Hu, D. Heikka, P. Dieffenbach, L. Ha, FASEB J. 15

Robinson, W. Patton, Electrophoresis 21 (2000) 3673.(2001) 1562.[58] K. Berggren, E. Chernokalskaya, T. Steinberg, C. Kemper,[28] V. Hu, D. Heikka, FASEB J. 14 (2000) 448.

M. Lopez, Z. Diwu, R. Haugland, W. Patton, Electrophoresis[29] K. Hofer, W. Hughes, Radiat. Res. 47 (1971) 94.21 (2000) 2509.[30] A. Kassis, S. Adelstein, Radiat. Res. 84 (1980) 407.

[59] H. Akhaven-Tafti, R. DeSilva, K. Sugioka, R. Handley, A.[31] J. Steinberg, J. NIH Res. 8 (1996) 30.Schaap, Luminescence 16 (2001) 187.[32] D. Hager, R. Burgess, Anal. Biochem. 109 (1980) 76.

[60] S. Desai, B. Dworecki, E. Cichon, Anal. Biochem. 297[33] C. Lee, A. Levin, D. Branton, Anal. Biochem. 166 (1987)(2001) 94.308.

[61] H. Towbin, T. Staehelin, J. Gordon, Proc. Natl. Acad. Sci.[34] J. Dzandu, J. Johnson, G. Wise, Anal. Biochem. 174 (1988)USA 76 (1979) 4350.157.

[62] R. Ogorzalek Loo, C. Mitchell, T. Stevenson, S. Martin, W.[35] M. Ortiz, M. Calero, C. Fernandez-Patron, C. Patron, L.Hines, P. Juhasz, D. Patterson, J. Peltier, J. Loo, P. Andrews,Castellanos, E. Mendez, FEBS Lett. 296 (1992) 300.Electrophoresis 18 (1997) 382.[36] C. Fernandez-Patron, L. Castellanos-Serra, P. Rodriguez,

[63] R. Ogorzalek Loo, J. Loo, P. Andrews, Methods Mol. Biol.Biotechniques 12 (1992) 564.112 (1999) 473.[37] L. Adams, K. Weaver, Appl. Theoret. Electrophoresis 1

[64] J. Loo, J. Brown, G. Critchley, C. Mitchell, P. Andrews, R.(1990) 279.Ogorzalek Loo, Electrophoresis 20 (1999) 743.[38] C. Fernandez-Patron, L. Castellanos-Serra, E. Hardy, M.

[65] A. Walker, G. Rymar, P. Andrews, Electrophoresis 22 (2001)Guerra, E. Estevez, E. Mehl, R. Frank, Electrophoresis 19933.(1998) 2398.

[66] R. Ogorzalek Loo, J. Cavalcoli, R. VanBogelon, C. Mitchell,[39] L. Castellanos-Serra, E. Hardy, Electrophoresis 22 (2001)J. Loo, B. Moldover, P. Andrews, Anal Chem. 73 (2001)864.4063.[40] M. Ferreras, J. Gavilanes, J. Garcia-Segura, Anal. Biochem.

[67] T. Hutchens, T. Yip, Rapid Commun. Mass Spectrom. 7213 (1993) 206.(1993) 576.[41] N. Matsui, D. Smith, K. Clausner, J. Fichmann, L. Andrews,

[68] M. Merchant, S. Weinberger, Electrophoresis 21 (2000)C. Sullivan, A. Burlingame, L. Epstein, Electrophoresis 181164.(1997) 409.

[69] F. Von Eggeling, K. Junker, W. Fiedler, V. Wollscheid, M.[42] C. Fernandez-Patron, M. Calero, P. Collazo, J. Garcia, J.Durst, U. Claussen, G. Ernst, Electrophoresis 22 (2001)Madrazo, A. Musacchio, F. Soriano, R. Estrada, R. Frank, L.2898.Castellanos-Serra, Anal. Biochem. 224 (1995) 203.

[70] A. Abbott, Nature 402 (1999) 715.[43] E. Hardy, H. Santana, A. Sosa, L. Hernandez, C. Fernandez-Patron, L. Castellanos-Serra, Anal. Biochem. 240 (1996) [71] M. Larsen, P. Roepstorff, Fresenius J. Anal Chem. 366150. (2000) 677.

Page 27: Detection Technologies in Proteome Analysis

771 (2002) 3–31 29W.F. Patton / J. Chromatogr. B

[72] J. Dennis, M. Granovsky, C. Warren, Biochim. Biophys. [107] M. Affolter, J. Watts, D. Krebs, R. Aebersold, Anal.Acta 1473 (1999) 21. Biochem. 223 (1994) 74.

[73] A. Litynska, M. Przybylo, D. Ksiazek, P. Laidler, Acta [108] P. Liao, J. Leykam, P. Andrews, D. Gage, J. Allison, Anal.Biochim. Pol. 47 (2000) 427. Biochem. 219 (1994) 9.

[74] S. Hakomori, Cancer Res. 56 (1996) 5309. [109] Y. Oda, K. Huang, F. Cross, D. Cowburn, B. Chait, Proc.[75] A. Kobata, Glycoconjugate J. 15 (1998) 323. Natl. Acad. Sci. USA 96 (1999) 6591.[76] D. Ivanov, M. Antonova, J. Stoilov, S. Zacharieva, E. [110] R. Aebersold, J. Watts, H. Morrison, E. Bures, Anal.

Gavazova, S. Ivanov, Cancer Lett. 125 (1998) 227. Biochem. 199 (1991) 51.[77] M. Bramwell, J. Cell Sci. 88 (1987) 335. [111] M. Goshe, T. Conrads, E. Panisko, N. Angell, T. Veenstra,[78] N. Chandra, M. Spiro, R. Spiro, J. Biol. Chem. 273 (1998) R. Smith, Anal. Chem. 73 (2001) 2578.

19715. [112] H. Zhou, J. Watts, R. Aebersold, Nat. Biotechnol. 19 (2001)[79] R. Zacharius, T. Zell, J. Morrison, J. Woodlock, Anal. 375.

Biochem. 30 (1969) 148. [113] Y. Oda, T. Nagasu, B. Chait, Nat. Biotechnol. 19 (2001)[80] P. Devine, J. Warren, Biotechniques 8 (1990) 493. 379.[81] A. Wardi, W. Allen, Anal. Biochem. 48 (1972) 621. [114] P. Fadden, T. Haystead, Anal. Biochem. 225 (1995) 81.[82] A. Wardi, G. Michos, Anal. Biochem. 49 (1972) 607. [115] K. Hsi, S. O’Neill, D. Dupont, P. Yuan, Anal. Biochem 258[83] A. Eckhardt, C. Hayes, I. Goldstein, Anal. Biochem. 73 (1998) 38.

(1976) 192. [116] G. Manchenko, in: Handbook of Detection of Enzymes On[84] F. Che, J. Song, L. Yu, K. Wang, Q. Xia, Biotechniques 30 Electrophoretic Gels, CRC Press, Boca Raton, FL, 1994, p.

(2001) 1272. 259.[85] T. Steinberg, K. Pretty On Top, K. Berggren, C. Kemper, L. [117] R. Beynon, J. Bond, in: Proteolytic Enzymes, A Practical

Jones, Z. Diwu, R. Haugland, W. Patton, Proteomics 1 Approach, IRL Press, Oxford, England, 1989, p. 25.(2001) 841. [118] D. Greenbaum, A. Baruch, L. Hayrapetian, Z. Darula, A.

[86] N. Packer, L. Keatinge, in: S. Pennington, M. Dunn (Eds.), Burlingame, K. Medzihradszky, M. Bogyo, Molecular andProteomics; From Protein Sequence To Function, BIOS Cellular Proteomics. Papers in Press. Published 11 Sep-Scientific Publishers, Oxford, 2001, p. 257. tember 2001 as Manuscript T100003-MCP200.

[87] I. Bouchez-Mahiout, C. Doyen, M. Lauriere, Electrophoresis [119] J. Stamler, O. Jaraki, J. Osborne, D. Simon, J. Keaney, J.20 (1999) 1412. Vita, D. Singel, R.Valeri, J. Loscalzo, Proc. Natl. Acad. Sci.

[88] D. O’Shannessy, P.Voorstad, R. Quarles, Anal. Biochem. 163 USA 89 (1992) 7674.(1987) 204. [120] P. Lane, G. Hao, S. Gross, Science’s STKE (2001) http: /

[89] E. Bayer, H. Ben-Hur, M. Wilchek, Anal. Biochem. 161 /stke.sciencemag.org /cgi /content / full /OC sigtrans;2001/]

(1987) 123. 86/ re1[90] K. Anumula, P. Du, Anal. Biochem. 275 (1999) 236. [121] J. Hoffmann, J. Haendeler, A. Zeiher, S. Dimmeler, J. Biol.[91] E. Zdebska, J. Koscielak, Anal. Biochem. 275 (1999) 171. Chem. 276 (2001) 41383.[92] D. Stern, Exp. Mol. Pathol. 70 (2001) 327. [122] S. Jaffrey, S. Snyder, Science’s STKE (2001) http: /[93] N. Ahn, K. Resing, Nat. Biotechnol. 19 (2001) 317. / stke.sciencemag.org /cgi /content / full /OC sigtrans;2001/

][94] D. Grander, Med. Oncol. 15 (1998) 20. 86/pl1.5.[95] R. Weinberg, in: J. Brugge, T. Curran, F. McCormick (Eds.), [123] W. Paik, S. Kim, J. Biol. Chem. 243 (1968) 2108.

Origins of Human Cancer; A Comprehensive Review, Cold [124] A. Niewmierzycka, S. Clarke, J. Biol. Chem. 274 (1999)Spring Harbor Laboratory Press, Cold Spring Harbor, NY, 814.1991, p. 1. [125] A. McBride, V. Weiss, H. Kim, J. Hogle, P. Silver, J. Biol.

[96] C. Tsatsanis, D. Spandidos, Int. J. Mol. Med. 5 (2000) 583. Chem. 275 (2000) 3128.[97] S. Saha, A. Bardelli, P. Buckhaults, V. Velculescu, C. Rago, [126] J. Smith, K. Rucknagel, A. Schierhorn, J. Tang, A. Nemeth,

B. St. Croix, K. Romans, M. Choti, C. Lengauer, K. Kinzler, M. Linder, H. Herschman, E. Wahle, J. Biol. Chem. 274B. Vogelstein, Science 294 (2001) 1343. (1999) 13229.

[98] G. Guy, R. Phillip, Y. Tan, Electrophoresis 15 (1994) 417. [127] S. Klein, J. Carroll, Y. Chen, M. Henry, P. Henry, I.[99] D. Moore, B. Sefton, in: Current Protocols in Molecular Ortonowski, G. Pintucci, R. Beavis, W. Burgess, D. Rifkin,

Biology, Vol. 3, Wiley, New York, 1995, p. 18.0.3. J. Biol. Chem. 275 (2000) 3150.[100] H. Kaufmann, J. Bailey, M. Fussenegger, Proteomics 1 [128] P. Zobel-Thropp, J. Gary, S. Clarke, J. Biol. Chem. 273

(2001) 194. (1998) 29283.[101] A. Dahlberg, C. Dingman, A. Peacock, J. Mol. Biol. 41 [129] K. Mowen, M. David, Science’s STKE 2001 http: / / stke.s-

(1969) 139. ciencemag.org /cgi /content / full /OC sigtrans;2001/93/pl1]

[102] M. Green, J. Pastewka, A. Peacock, Anal. Biochem. 56 [130] Q. Liu, G. Dreyfuss, Mol. Cell. Biol. 15 (1995) 2800.(1973) 43. [131] R. Desrosiers, R. Tanguay, J. Biol. Chem. 263 (1988) 4686.

[103] J. Cutting, T. Roth, Anal. Biochem. 54 (1973) 386. [132] L. Belyanskaya, P. Gehrig, H. Gehring, J. Biol. Chem. 276[104] I. Debruyne, Anal. Biochem. 133 (1983) 110. (2001) 18681.[105] M. Brauer, B. Sykes, Methods Enzymol. 107 (1984) 36. [133] B. Schurter, S. Koh, D. Chen, G. Bunick, J. Harp, B.[106] K. Swiderek, K. Jaquet, H. Meyer, C. Schachtele, F. Hanson, A. Henschen-Edman, D. Mackay, M. Stallcup, D.

Hofmann, L. Heilmeyer, Eur. J. Biochem. 190 (1990) 575. Aswad, Biochemistry 40 (2001) 5747.

Page 28: Detection Technologies in Proteome Analysis

771 (2002) 3–3130 W.F. Patton / J. Chromatogr. B

[134] B. Strahl, S. Briggs, C. Brane, J. Caldwell, S. Koh, H. Ma, [165] J. Hainfeld, W. Liu, C. Halsey, P. Freimuth, R. Powell, J.R. Cook, J. Shabanowitz, D. Hunt, M. Stallcup, C. Allis, Struct. Biol. 127 (1999) 185.Curr. Biol. 11 (2001) 996. [166] H. Zentgraf, M. Frey, S. Schwinn, C. Tessmer, B. Wil-

[135] B. Weng, W. Thompson, H. Kim, R. Levine, J. Moss, J. lemann, Y. Samstag, I. Velhagen, Nucleic Acids Res. 23Biol. Chem. 274 (1999) 31797. (1995) 3347.

[136] M. Cohen-Armon, I. Hammel, Y. Anis, S. Homburg, D. [167] E. Pogge von Strandmann, C. Zoidl, H. Nakhei, B. Holewa,Noya, J. Biol. Chem. 271 (1996) 26200. R. Pogge von Strandmann, P. Lorenz, L. Klein-Hitpass, G.

[137] P. Randazzo, Z. Nie, K. Miura, V. Hsu, Science’s STKE Ryffel, Protein Eng. 8 (1995) 733.2000 http: / / stke.sciencemag.org /cgi /content / full /OC sig- [168] P. Lindner, B. Guth, C. Wulfing, C. Krebber, B. Steipe, F.]trans;2000/59/ re1 Muller, A. Pluckthun, Methods 4 (1992) 41.

[138] S. Prasad, J. Walent, A. Dritschilo, Biochem. Biophys. Res. [169] M. Chalfie, Y. Tu, G. Euskirchen, W. Ward, D. Prasher,Commun. 204 (1994) 772. Science 263 (1994) 802.

[139] M. Di Gorolamo, M. Silletta, M. De Matteis, A. Braca, A. [170] V. Hatzimanikatis, L. Choe, K. Lee, Biotechnol. Prog. 15Colanzi, D. Pawlak, M. Rasenick, A. Luini, D. Corda, Proc.

(1999) 312.Natl. Acad. Sci. USA 92 (1995) 7065.

[171] Y. Li, A. Agrawal, J. Sakon, R. Beitle, J. Chromatogr. A[140] L. Ronnov-Jessen, O. Petersen, Electrophoresis 17 (1996)

909 (2001) 183.1776.

[172] S. Ericksson, E. Raivio, J. Kukkonen, K. Eriksson, C.[141] H. Phan, C. Boissard, M. Pessah, K. Regnauld, S. Emani,Lindqvist, J. Virol. Methods 59 (1996) 127.C. Gespach, G. Rosselin, Biochem. Biophys. Res. Com-

[173] D. Richards, C. Stathakis, R. Polakowski, H. Ahmadzadeh,mun. 271 (2000) 86.N. Dovichi, J. Chromatogr. A 853 (1999) 21.[142] G. Fritz, I. Just, B. Kaina, Int. J. Cancer 81 (1999) 682.

[174] C. Gordon, V. Khalaj, A. Ram, D. Archer, J. Brookman, A.[143] T. Meyer, H. Hilz, Eur. J. Biochem. 155 (1986) 157.Trinci, D. Jeenes, J. Doonan, B. Wells, P. Punt, C. van den[144] A. Giovane, C. Balestrieri, L. Quagliuolo, L. Servillo,Hondel, G. Robson, Microbiology 146 (2000) 415.FEBS Lett. 191 (1985) 191.

[175] C. Nemetz, R. Reichhuber, R. Schweizer, P. Hloch, M.[145] V. Hingorani, Y. Ho, J. Biol. Chem. 263 (1988) 19804.Watzele, Electrophoresis 22 (2001) 966.[146] B. Klebl, D. Pette, Anal. Biochem. 239 (1996) 145.

[176] N. Garameszegi, Z. Garameszegi, M. Rogers, S. DeMarco,[147] C. Kemper, T. Steinberg, L. Jones, W. Patton, Electro-E. Strehler, Biotechniques 23 (1997) 864.phoresis 22 (2001) 970.

[177] S. Kiessig, J. Reissmann, C. Rascher, G. Kullertz, A.[148] S.-M. Yu, Y.-H. Kuo, G. Sheu, Y.-J. Sheu, L.-F. Liu, J. Biol.Fischer, F. Thunecke, Electrophoresis 22 (2001) 1428.Chem. 266 (1991) 21131.

[178] T. Aoki, Y. Takahashi, K. Koch, H. Leffert, H. Watanabe,[149] S. Ho, W. Tong, S. Yu, Plant Physiol. 122 (2000) 57.FEBS Lett. 384 (1996) 193.[150] K. Yoshida, N. Iino, I. Koga, K. Kato, Anal. Biochem. 58

[179] T. Aoki, M. Miyashita, H. Fujino, H. Watanabe, Biosci.(1974) 77.Biotechnol. Biochem. 64 (2000) 1547.[151] D. Starkey, A. Han, J. Bao, C. Ahn, K. Wehmeyer, M.

[180] M. Unlu, M. Morgan, J. Minden, Electrophoresis 18 (1997)Prenger, H. Halsall, W. Heineman, J. Chromatogr. B.2071.Biomed. Sci. Appl. 762 (2001) 33.

[181] R. Tonge, J. Shaw, B. Middleton, R. Rowlinson, S. Rayner,[152] D. Nazarenko, S. Dertinger, T. Gasiewicz, Biotechniques 30J. Young, F. Pognan, E. Hawkins, I. Currie, M. Davison,(2001) 776.Proteomics 1 (2001) 377.[153] R. Duncan, J. Hewitt, P. Weston, Biochem. J. 205 (1982)

[182] F. Von Eggeling, A. Gawriljuk, W. Fiedler, G. Ernst, U.219.Claussen, J. Klose, I. Romer, Int. J. Mol. Med. 8 (2001)[154] B. Burke, F. Regnier, Electrophoresis 22 (2001) 3744.373.[155] M. Eggertson, D. Craig, Biomed. Chromatogr. 13 (1999)

[183] S. Gharbi, P. Gaffney, A. Yang, M. Zvelebil, R. Cramer, M.516.Waterfield, J. Timms, Mol. Cell. Prot. (2002) in press.[156] A. James, K. Chilvers, J. Perry, L. Armstrong, F. Gould,

[184] G. Zhou, H. Li, D. DeCamp, S. Chen, H. Shu, Y. Gong, M.Appl. Environ. Microbiol. 66 (2000) 5521.Flaig, J. Gillespie, N. Hu, P. Taylor, M. Emmert-Buck, L.[157] A. James, J. Perry, K. Chilvers, I. Robson, L. Armstrong,Liotta, E. Petricoin, Y. Zhao, Mol. Cell. Prot. (2002) inK. Orr, Lett. Appl. Microbiol. 30 (2000) 336.press.[158] M. Ho, S. Povey, D. Swallow, Am. J. Hum. Genet. 34

[185] J. Minden, A. Waggoner (2000), US patent 6,043,025.(1982) 650.[186] F. Kernec, M. Unlu, W. Labeikovsky, J. Minden, A.[159] R. Evangelista, A. Pollak, E. Templeton, Anal. Biochem.

Koretsky, Physiol. Genomics 6 (2001) 117.197 (1991) 213.[187] A. de Groof, F. Oerlemans, C. Jost, B. Wieringa, Muscle[160] E. Hochuli, S. Piesecki, Methods 4 (1992) 68.

Nerve 24 (2001) 1188.[161] D. O’Shannessy, K. O’Donnell, J. Martin, M. Brigham-[188] K. Saupe, M. Spindler, J. Hopkins, W. Shen, J. Ingwall, J.Burke, Anal. Biochem. 229 (1995) 119.

Biol. Chem. 275 (2000) 19742.[162] S. McMahan, R. Burgess, Anal. Biochem. 236 (1996) 101.[163] C. Botting, R. Randall, Biotechniques 19 (1995) 362. [189] V. Veksler, A. Kuznetsov, K. Anflous, P. Mateo, J. van[164] L. Jin, X. Wei, J. Gomez, M. Datta, A. Birkett, D. Peterson, Deursen, B. Wieringa, R. Ventura-Clapier, J. Biol. Chem.

Anal. Biochem. 229 (1995) 54. 270 (1995) 19921.

Page 29: Detection Technologies in Proteome Analysis

771 (2002) 3–31 31W.F. Patton / J. Chromatogr. B

[190] Z. Smilansky, Electrophoresis 22 (2001) 1616. [198] P. Haynes, J. Yates, Yeast 17 (2000) 81.[191] K. Gee, H. Kang, T. Meier, G. Zhao, L. Blaszcak, Electro- [199] D. Han, J. Eng, H. Zhou, R. Aebersold, Nat. Biotechnol. 19

phoresis 22 (2001) 960. (2001) 946.[192] S. Gygi, B. Rist, S. Gerber, F. Turecek, M. Gelb, R. [200] K. Berggren, E. Chernokalskaya, M. Lopez, J. Beechem, W.

Aebersold, Nat. Biotechnol. 17 (1999) 994. Patton, Proteomics 1 (2001) 54.[193] R. Aebersold, B. Rist, S. Gygi, Ann. NY Acad. Sci. 919 [201] C. Henry, Chem. Eng. News 2 (2001) 47.

(2000) 33. [202] J. Ji, A. Chakraborty, M. Geng, X. Zhang, A. Amini, M.[194] M. Smolka, H. Zhou, R. Aebersold, Mol. Cell. Proteomics. Bina, F. Regnier, J. Chromatogr. B 745 (2000) 197.

(2002) in press. [203] D. Goodlett, A. Keller, J. Watts, R. Newitt, E. Yi, S.[195] M. Smolka, H. Zhou, S. Purkayastha, R. Aebersold, Anal. Purvine, J. Eng, P. von Haller, R. Aebersold, E. Kolker,

Biochem. 297 (2001) 25. Rapid Commun. Mass. Spectrom. 15 (2001) 1214.[196] M. Moseley, Trends Biotechnol. 19 (2001) S10. [204] J. Hille, A. Freed, H. Watzig, Electrophoresis 22 (2001)[197] R. Zhang, C. Sioma, S. Wang, F. Regnier, Anal. Chem. 73 4035.

(2001) 5142.