Comprehensive quantitative proteome analysis of 20S proteasome subtypes from rat liver by isotope...

RESEARCH ARTICLE

Comprehensive quantitative proteome analysis of 20S

proteasome subtypes from rat liver by isotope coded

affinity tag and 2-D gel-based approaches

Frank Schmidt1, 2*, Burkhardt Dahlmann3*, Katharina Janek3, Alexander Kloß3,Maik Wacker4, Renate Ackermann1, Bernd Thiede2 and Peter R. Jungblut1

1 Max Planck Institute for Infection Biology, Core Facility Protein Analysis, Berlin, Germany2 The Biotechnology Centre of Oslo, University of Oslo, Norway3 Institute of Biochemistry, Campus Charité Mitte, Charité-Universitätsmedizin-Berlin, Berlin, Germany4 Institute for Human Genetics, University Clinic Charité, Berlin, Germany

Quantitative protein profiling is an essential part of proteomics and requires technologies thataccurately, reproducibly, and comprehensively identify and quantify proteins. Over the pastyears, many quantitative proteomic methods have been developed. Here, 20S proteasome sub-types isolated from rat were compared by four approaches based on the combination of isotope-coded affinity tag (ICAT), 2-DE, LC and ESI and MALDI MS: (i) 2-DE, (ii) ICAT/2-DE MALDI-MS, (iii) ICAT/LC-ESI-MS, (iv) ICAT/LC-MALDI-MS. A definite qualitative advantage of 2-DEgels was the separation of all known protein species, the identification of cysteine sulfoxide of a-4(RC6-IS) and N-terminal acetylation of several subunits. Furthermore, quantitative differencesbetween the standard subunits b-2, and b-5 and their immunosubunits were only detected by2-DE image analysis revealing a higher replacement of standard- by immuno-b-subunits in sub-type IV. It was obvious that for relative quantification only protein spot and mass peaks with acertain level of intensity displayed acceptable values of SD. However, ICAT in conjunction withLC/MALDI-MS was the most accurate method for quantification. The experimental data of thisinvestigation are accessible via http://www.mpiib-berlin.mpg.de/2D-PAGE/.

Received: December 21, 2005Revised: April 7, 2006

Accepted: April 28, 2006

Keywords:

20S Proteasome subtypes / Isotope-coded affinity tag / MS / MS-Screener / Quantita-tive proteomics

4622 Proteomics 2006, 6, 4622–4632

1 Introduction

A major goal of proteomics is the global qualitative andquantitative analysis of proteins in defined biological sys-tems. Staining of gels and stable isotope tagging are the most

commonly used approaches for relative quantification ofproteins. Quantification of protein spots using 2-DE [1] isusually based on statistical analysis performed on sets of gelsstained with visible dyes via image analysis software. Alter-natively, the DIGE technique based on fluorescent labelingwas developed for quantitative proteome analyses using2-DE [2]. Quantitative proteomics via stable isotope taggingand MS [3–5] include metabolic, enzymatic and chemicalmethods. Metabolic stable isotope labeling of amino acidsduring cell culture is a very promising method for quantita-tive proteomics of cultured cells [6]. However, recently stable

Correspondence: Dr. Peter R. Jungblut, Core Facility ProteinAnalysis, Max Planck Institute for Infection Biology, Charitéplatz1, D-10117 Berlin, GermanyE-mail: [email protected]: 149-30-28460507

Abbreviations: ICAT, isotope-coded affinity tag; LIMS, laboratoryinformation management system * These authors contributed equally to this work.

DOI 10.1002/pmic.200500920

© 2006 WILEY-VCH Verlag GmbH & Co. KGaA, Weinheim www.proteomics-journal.com

Proteomics 2006, 6, 4622–4632 Animal Proteomics 4623

isotope labeling of amino acids during cell culture was alsoapplied to relative quantification of mouse brain proteins byculture-derived isotope tags. The stable isotope labeling ofamino acids during cell culture labeled culture cells weremixed with mouse brain samples and exploited as internalstandard [7]. Enzymatic labeling of proteins during trypticdigestion in H2O

16 and H2O18 generates peptides with a

mass difference of 4 Da between two samples and can beused for relative quantification [8]. Furthermore, many dif-ferent chemical methods were developed for quantitativeanalysis. The most employed chemical labeling methodsinclude isotope-coded affinity tags (ICAT) [9], isotope-codedprotein label [10], global internal standard technology [11],and isobaric tags for relative and absolute quantification [12].Due to the increasing number of methods for protein quan-tification, the advantages and drawbacks of different ap-proaches must be evaluated. For this purpose, the smallproteasome complex was selected as model system becauseof its simple and well-known protein composition.

The 26S proteasome consists of a 20S core protease par-ticle and two 19S regulatory particles. The 20S proteasome iscomposed of seven different a-subunits and seven differentb-subunits with a heterodimeric configuration: a-1-7, b-1-7,b-1-7, a-1-7 [13]. Beside serving as a docking site for protea-some regulators the alpha subunits form a pore for substrateentrance, since the proteolytically active sites formed bythree of the beta subunits are located inside of the barrel-shaped 20S proteasome. A mechanism to adapt the proper-ties of proteasomes for optimal antigen processing is thereplacement of the three active sites containing b-subunitsby so-called immunosubunits, b1i, b2i, and b5i, the synthe-sis of which is induced by interferon-g. Thus, two majorsubpopulations of 20S proteasomes exist, immunoprotea-somes and standard proteasomes (constitutive or house-holdproteasomes). These proteasome-subpopulations disclosednot only the existence of an additional third proteasome-subpopulation, that we have designated intermediate-typeproteasomes, but lead also to the discovery that each sub-population splits off into several subtypes with differentenzymatic properties [14]. Due to their different net chargesthese subtypes can be separated by ion exchange chroma-tography. As all 20S proteasomes basically contain the samesubunit stoichiometry, it is interesting to find out differencesbetween proteasome subtypes of the same subpopulation.

In the past, 20S proteasomes were characterized by the2-DE/MS approach in different organisms and tissues.Kuckelkorn and co-workers identified different protein spe-cies compositions of the 20S proteasomes from mouse thy-mus, small intestine, colon and liver [15]. Furthermore, areference map of the human 20S proteasome purified fromerythocytes [16] was established. In addition, the 20S protea-some of chicken [17], yeast [18] and Trypanosoma brucei [19]were characterized with regard to proteome analyticalaspects. All studies showed a diversification of proteasomalproteins into protein species, which are defined by their dif-ferent chemical structures [20]. This may be caused by PTM,

isoforms, alternative splicing, truncation or point mutation.PTMs such as N-terminal acetylation [21, 22], phosphory-lation [18, 23, 24], and glycosylation [25, 26] were alreadyidentified in proteasomes. Quantitative proteome analysis tocompare erythrocytes and U937 cancer cells of human 20Sproteasomes was performed using the ICAT reagent in con-junction with 2-DE/MALDI-MS [27]. Moreover, dynamics ofthe 20S proteasome was studied using chicken fed withstable isotope labeled valine and 2-DE/MS analysis [17].

Here, standard 2-DE patterns for 20S proteasome sub-types from rat were established and made available withinthe Proteome Database System for Microbial Research atMax Planck Institute for Infection Biology (http://www.mpiib-berlin.mpg.de/2D-PAGE/). Moreover, 20S pro-teasome subtypes II and IV purified from rat liver were ana-lyzed to compare different proteome approaches because theminor complex protein composition facilitates the analysis.Furthermore, the proteasomal subtypes were applied tostudy protein species. For this purpose, 2-DE in conjunctionwith image analysis using TopSpot, and the cleavable form ofthe cICAT reagent in combination with 2-DE/MALDI-MS,LC-ESI-MS or LC-MALDI-MS were applied to compare thefour approaches and to reveal quantitative and qualitativedifferences between the proteasome subtypes and its proteinspecies. Moreover, the well-defined differences betweenimmunoproteasomes and standard proteasomes can beexploited to examine the different approaches for quantita-tive proteomics.

2 Material and methods

2.1 Preparation of total 20S proteasome

20S Proteasomes were purified from rat liver (Wistar rats,180 g body weight, Harlan-Winkelmann, Borchen, Ger-many) [14]. Briefly, liver tissue was homogenized in a 3-foldvolume w/v of 20 mM Tris/HCl/1 mM EDTA/ 1 mM NaN3/1 mM DTT, pH 7.0 (TEAD buffer). The homogenate wascentrifuged at 15 000 g for 20 min and the supernatant sub-jected to ammonium sulphate fractionation. Proteins pre-cipitating between a saturation of 45–65% with respect to(NH4)2SO4 were dissolved in TEAD buffer and then dialyzedagainst the same buffer. The dialyzed protein solution wasapplied onto a column (5615 cm) of DEAE-Sephacel equili-brated with TEAD buffer. Bound proteins were eluted by alinear increasing gradient of 0–400 mM NaCl dissolved inTEAD buffer. To detect the proteasome in the eluate, thecollected fractions were tested with the fluorogenic substrateSuc-LLVY-MCA as described in [14]. Proteasome containingfractions were pooled and then subjected to gel filtration on acolumn of Sepharose 6B (3680 cm) equilibrated in TEADbuffer. Fractions containing 20S proteasome were combinedand further purified by chromatography on Mono Q (HR 5/5) in conjunction with an FPLC system (Amersham Bio-sciences, Freiburg, Germany). The proteasome was eluted


4624 F. Schmidt et al. Proteomics 2006, 6, 4622–4632

from the column with an increasing NaCl/TEAD gradient(0–300 mM). Then the proteasome containing fractions werepooled, (NH4)2SO4 added to a final concentration of 1.2 Mwith respect to the salt and subjected to a Phenyl-Superosecolumn (HR 5/5). From this column proteasomes wereeluted with a linear decreasing gradient (1.2–0.0 M) of(NH4)2SO4 dissolved in the TEAD and then extensively dia-lyzed against TEAD buffer.

2.1.1 Preparation of proteasome subtypes

20S proteasome subtypes were separated on Mini Q (PC 3.2/3) in conjunction with a SMART system (Amersham Bio-sciences) [14]. Fractions containing a specific proteasomesubtype were pooled, diluted with the TEAD buffer and re-chromatographed under the same conditions. Re-chroma-tography was repeated once again to obtain well-separatedand homogenous 20S proteasome subtypes. In total, fifesubtypes were isolated and subtype II and IV were used forfurther investigations.

2.1.2 cICAT reagent-labeling and chromatography

Before labeling, 10 mg of each subtype were run on a 1-D geland stained with CBB G-250. The gel was scanned and ana-lysed by TopSpot to quantify the protein bands of the sub-types. As a result of the accurate quantification, 70 mg of bothproteasome subtypes were dissolved in 100 mL buffer(6 M Urea, 0.05% SDS, 5 mM Tris pH 8.3, 5 mM EDTA) andwere reduced with 5 mM triscarboxyethylphosphine for30 min at 377C. After addition of 175 nmol light cleavablecICAT reagent (Applied Biosystems) to subtype II and theheavy one to subtype IV (approximately 0.5 nmol cICAT/mgof protein; final cICAT concentration: 1.4 mM), proteinswere incubated for 90 min in the dark at room temperaturewith gentle stirring. After incubation, DTT was added to afinal concentration of 10 mM to quench residual freereagents. Labeled samples were mixed, diluted 4-fold withwater, so that the final urea concentration was 1.5 M anddigested with 1:25 trypsin/protein (Promega, Madison, WI)over night at 377C. The resulting peptide mixture was com-bined and purified by cation exchange and avidin columnchromatography as described in the manufacture manual toremove the remaining cICAT-reagent and other contami-nants and to separate the peptides.

2.1.3 cICAT reagent-labeling for 2-DE

The proteasome subtypes were labeled as described above.Then, 140 mL of the pooled samples were diluted 1:6 with ice-cold (2207C) acetone (volume/volume) and mixed on ice.Afterwards, the sample was precipitated for 10 min on iceand centrifuged at 13 000 rpm for 10 min. The dried pelletwas mixed with 95 mL of 95% TFA and 5 mL Cleave B(standard protocol Applied Biosystems) and incubated for

2 h at room temperature to cleave the cICAT linker. Afteragitation the sample was dried in the speed vac for 2-DEpreparation.

2.1.4 2-DE

The dried proteasomal proteins were dissolved in 9 M urea,70 mM DTT, 2% carrier ampholytes Servalyte 2-4 (Serva,Heidelberg, Germany) and protease inhibitors (TLCK, Leu-peptin, E64, Pepstatin A; 25 mM). The carrier ampholyte iso-electric focusing method was combined with SDS-PAGE.Gels with the size of 23630 cm [28] or 1068 cm [29] wereproduced. 25 mg of proteasomal proteins were applied to thelarge gel and 5 mg to small gels and stained with CBB G-250.Three independent 20S proteasome subtype preparationswere analysed and compared by quantitative image analysesusing TopSpot (http://www.mpiib-berlin.mpg.de/2D-PAGE/). Due to the equal staining intensity of most of the proteinbands in the 1-D gel, the main spots of each subtype with anintensity difference lower than 60.15 were used for normal-ization.

2.2 MALDI-MS

Proteins separated by 2-DE were identified by PMF after in-gel digestion [30]. A Voyager Elite MALDI-TOF mass spec-trometer and/or a 4700 Proteomics Analyzer equipped with a4700-Explorer software (Version 2.0) (both from Applied Bio-systems) and an ULTRAFLEX II (Bruker Daltonics, Bremen,Germany) were used with a mass accuracy of 30 ppm afterinternal calibration using peptides occurring due to autolysisof trypsin as markers. The samples were analyzed in the MSmode (for generation of peptide mass fingerprints) as well asin the TOF/TOF mode (for fragmentation analysis of thethree most intense peaks). MS-spectra generated by theVoyager Elite MALDI-TOF were transformed into a peak listby using the software GRAMS, where all significant peakswere labeled manually. The peak lists of the MS and MS/MSspectra from the 4700 Proteomics Analyzer were created bythe peak-to-mascot script of the 4700 Explorer TM applyingthe following peak filter settings: mass range: 500–4000Da (MS), from 60 to precursor minus 20 Da (MS/MS); peakdensity: up to 10 peaks per 200 Da (MS), up to 20 peaks per100 Da (MS/MS); minimal signal-to-noise ratio: 20 (MS), 3(MS/MS); minimal area: 1000 (MS), 0 (MS/MS); maximalnumber of peaks/spot: 100 (MS), maximal number of peaks /precursor: 100 (MS/MS). No smoothing was applied and thepeaks were not de-isotoped. MS-spectra generated by theULTRAFLEX II were transformed into peak list by using thesoftware Flexanalysis (Bruker Daltonics). The following peakfilter settings were used: mass range: 700–4000 Da (MS),minimal signal-to-noise ratio: 10 (MS). MS/MS spectra weregenerated manually, where smoothing and baseline subtrac-tion were applied. Protein identifications were achieved bydatabase comparisons with a proteasome rat subset of Swis-sprot database (Release 45.2, 32 entries) using the in-house



search algorithm MASCOT (Version 1.9) [31]. A tolerance of50 ppm and one missed cleavage was allowed. Moreover, thevariable modifications N-acetylation, Met-oxidation, propion-amide on Cys and pyroglutamic-acid on N-terminal Gln wereallowed. To identify tagged proteins, the variable modifica-tions cICAT-light and cICAT-heavy were selected instead ofpropionamide. Proteins were considered as identified if thesequence coverage was higher than 30% and the probabilityscore higher than 60.

2.3 LC-ESI-iontrap-MS/MS

The purified and dried cysteinyl peptides were dissolved in10 mL buffer A (0.1% formic acid in water). Fifteen micro-liters of sample were loaded onto an LCPackings PepmapC18 column (length 15 cm, i.d. 75 mm) and run at a flowrate of 200 nL per minute for a duration of 210 min. An elu-tion gradient ranging from 0 to 30% solvent B from 0 to180 min. and 30 to 95% solvent B from 180 to 210 min. wasapplied. Solvent A was 0.1% v/v FA in water, solvent B was0.1% v/v FA in ACN. Peptide identification by collision-induced dissociation was carried out by data dependent pre-cursor ion selection using the dynamic exclusion option onan LCQ ion trap mass spectrometer (Thermo Finnigan,Wal-tham, MA, USA). Each measurement was repeated threetimes with three independently prepared samples. The MS/MS spectra were submitted to a protein sequence database(20S proteasome rat, subset of Swissprot database(Release 45.2, manually annotated) using the TURBO-SEQUEST (running on Xcalibur 1.2) with a peptide toler-ance of 2.5 Da and a fragment tolerance of 1.0 Da wereallowed. Furthermore, the fixed modification of cysteines bycICAT and a possible oxidation of methionines were con-sidered. For peptide identifications, only tryptic cysteinylpeptides and peptides with Xcorr score higher than 2 andRsp equal to 1 were accepted. The abundance ratios of ICATpeptide pairs were calculated with a mass tolerance of 1 Dausing XPRESS [32]. The most intense peptide pairs of thesubtypes with an intensity difference smaller than 60.1 wereused for the normalization.

2.4 LC-MALDI-MS

The purified and dried cysteinyl peptides were dissolved in10 mL 0.1% formic acid in water and separated in an AgilentnanoLC system (Walbronn, Germany) equipped with a1100 Series nanoflow pump with micro-vacuum degasserand a RP trapping column (trapping column: Zorbax 300SB-C18, 5 mm, 560.3 mm, 5/pkv). Peptide separation was per-formed on Zorbax 300SB-C18 column (3.5 mm,150 mm675 mm) with linear gradient elution (eluent A,0.5% formic acid in water and 5% ACN; eluent B: 0.5% for-mic acid in 100% ACN). Fractions were collected by aninterval of 2 min, which corresponded to a sample volume of0.6 mL, and spotted on a 9662 array MALDI-target. Thedried sample fractions were dissolved with 0.5 mL of a-cyano-

4-hydroxycinnamic acid at the target. MALDI spectra wereacquired using the 4700 Proteomics Analyzer and the sixmost intense peaks were selected for MS/MS analysis usingthe LC-MALDI analysis program. The peak lists were createdfrom the raw data as described above. The peptide identifi-cations were achieved by database comparisons with a pro-teasome rat subset with a tolerance of 50 ppm for the pre-cursor and 0.5 Da for the MS/MS fragments. Moreover, thevariable modifications cICAT-light and cICAT-heavy, Met-oxidation, pyroglutamic-acid on N-terminal Gln and onemissed cleavage were allowed. Peptides were considered tobe identified if the probability based mowse score was higherthan 30 and the peptides were modified by cICAT. None-theless, all MS/MS spectra were finally re-evaluated manu-ally. Each measurement was repeated three times with threeindependently prepared samples.

2.5 Quantification of MALDI peaks by MS-Screener

The MALDI-MS spectra were transformed in t2d files by the4700 Explorer program and post-processed by the DataExplorer (Applied Biosystems). The generated binary pktpeak lists were exported to MS-Screener [33] (http://www.mpiib-berlin.mpg.de/2D-PAGE/) to eliminate all sys-tematic contamination peaks. The remaining masses werefiltered for cICAT specific peptide pairs (9.031, 18 062) usinga peak tolerance of 0.01 Da (MS-Screener analysis tool). Themost intense peptide pairs of the subtypes with an intensitydifference smaller than 60.1 were used for the normal-ization.

2.6 Laboratory information management system

(LIMS)

The workflow described above required a suitable system forintegration and management of raw and processed experi-mental data. These issues were addressed by a LIMS incombination with the implemented SQL*LIMS™ ProteomicsSolution (Applied Biosystems), customized for our proteom-ics laboratory. For the protein identifications, MS peak listswere submitted to search database engines such as MASCOTand SEQUEST. The queries were performed from theSQL*LIMS™ or by using the search engine front-end inter-face. Peak lists and protein/peptide identification and quan-tification results were stored in the SQL*LIMS™ database andcould be queried by the Query Builder (Oracle, RedwoodCity, California).

3 Results and discussion

3.1 Experimental course of actions

In order to obtain structural and quantitative differences in20S proteasome subtypes, we applied four different methodsand compared their specific results (Fig. 1). Each measure-



Figure 1. To investigate 20S proteasome subtypes, the total 20S proteasome was separated into subtypes by anion exchange chroma-tography. A 2-DE gel of subtype II and IV was prepared and quantified by image analysis using TopSpot. Furthermore, both subtypes werelabeled by cleavable cICAT and combined. One part of the pool was digested by trypsin followed by avidin purification and cysteinyl pep-tides were analyzed by LC/ESI-MS/MS and LC/MALDI-MS/MS. Second part of the pool was separated by 2-DE and proteins were analyzedby PMF. To compare all results, protein identification obtained by LC-ESI-MS/MS (SEQUEST), LC/MALDI-MS/MS, ICAT/2-DE/MALDI-MSand 2-DE/MALDI-MS (all by MASCOT), as well as relative quantifications received by TopSpot, Xpress or MS-Screener, were stored in LIMSand prompted by SQL statement.

ment was repeated three times with three independentlyprepared samples. First, we started to analyze the total20S proteasome from the rat liver by 2-DE/MALDI-MS tocreate a reference gel (Fig. 2). To investigate 20S proteasomesubtypes, the total 20S proteasome was separated into fivedifferent subtypes by anion exchange chromatography [14].For further investigations, subtypes II and IV were baseline

separated by anion exchange (Fig. 1). At first, a 2-DE gel ofeach subtype was prepared and quantified by image analysisusing TopSpot. In addition the subtypes were labeled bycleavable cICAT and combined. One part of the pool wasdigested with trypsin followed by avidin purification. Thepurified cysteinyl peptides were then analyzed by LC/ESI-MS/MS and LC/MALDI-MS/MS. The remaining part was



Figure 2. 2-DE map of rat 20S proteasome (subtype II) purified from liver cells.In the first dimension, the proteins were separated by IEF using the carrier ampholyte IEF method (pH 4–11) and in the second dimensionby SDS-PAGE. The gel was stained with CBB G-250 and all labeled spots were identified by MALDI-TOF-MS and database search. Differentprotein species of the subunits were marked with additionally numbers, starting from the most intense spot. Spots from constitutive20S subunits (framed by a rectangle) were exchanged with their immunoform in the subtype IV (framed by octagon).

separated by 2-DE and spots were analyzed by PMF. In orderto compare all four methods, the protein identificationsobtained by LC-ESI-MS/MS (SEQUEST), LC/MALDI-MS/MS, ICAT/2-DE/MALDI-MS and 2-DE/MALDI-MS (all byMascot), as well as the relative quantifications received byTopSpot, Xpress or MS-Screener, were stored in LIMS andcompared via SQL statement (Fig. 1).

3.2 2-DE/MALDI-MS

Each of the detected spots in the 2-DE gel (Fig. 2) was cut out,subjected to in-gel digestion, and analyzed by MALDI-MS.All a- and b-subunits were identified including immuno-subunits b-1i, b-2i and b-5i. With the exception of a-1, all ofthe subunits resulted in two or more protein species. Inter-estingly, some protein species of the a-type proteins showeddifferences in Mr and pI. In contrast, b-type protein speciesshifted only in pI. Quantitative 2-DE gel image analysisshowed that replacement of b1 by b1i and b5 by b5i hasoccurred to a larger extent in subtype IV as compared tosubtype II. Although b2i is also more prominent in sub-type IV than in subtype II this replacement seems to havetaken place predominantly at the expense of b2.2. (Table 1).In general, spots with intensities higher than about 100 (thesum of the gray levels of all pixels belonging to the spot)clearly enabled a better SD (0.02–0.15) than faint spots (0.03–0.70). Considering the spots with higher intensities, all sub-units were presented equally in both subtypes (Table 1), withthe exception of b-3.1/2 (0.88) and a-3.3 (0.73), which areslightly underrepresented in subtype IV. Furthermore, ourinvestigations of PTMs showed that most of the protein spe-cies of subunits a-1, a-4, a-5, a-6, a-7, b-3, and b-4 wereacetylated at their N-terminus, the origin of which, however,remains to be established. Na-acetylation, for instance, maybe a common PTM in proteasomal subunits since it has alsobeen detected by other investigators [21, 22]. The subunitsa-2 and a-3 might be N-acetylated, as previously described byKimura et al. [22], but these acetylations could not be con-

firmed by MALDI-MS, due to the low masses of the trypticpeptides (548.2 and 434.2 for a-2 and 3). Furthermore, thecomparison of the peptide mass fingerprint of a-4.1 and a-4.2 revealed two additional masses for a-4.2 at 2317 and2333 Da (Fig. 3). The MALDI-MS/MS spectrum of the pre-cursor ion mass of 2317 Da was determined with a fully-ion series and b2-4-ions as the sequence 61KICOX-

ALDDNVCPAMAFAGLTADR81 with COX displaying a cyste-ine sulfonic acid and CPA cysteine modified by acrylamide(Fig. 4). The high intensity of the mass peak of the y18 ion at1924.88 Da supported the identification of cysteine sulfonicacid because preferential cleavage of the amide bond at the C-terminal side of the oxidized cysteine was already describedusing an ion trap mass spectrometer [34]. In addition, thesecond additional mass of a-4.2 compared to a-4.1 at 2333 Dacan easily be explained by the oxidation of methionine atposition 71. In contrast to it, the peaks 2340 and 2356 deter-mined in a-4 (Fig. 3a) are the corresponding peptides whereboth cysteines are modified by acryl amide. The determinedsequence matched to the splice isoforms RC6-IS a-4 protea-somal subunit from rat (Swiss-Prot entry P48004-1). Inter-estingly, only the described oxidized peptide form could bedetermined with the fragmentation spectrum, but not thepossible other form with the same mass, where the modifiedcysteines are exchanged (acrylamide at position 63 and cys-teine sulfonic acid at position 70). In addition to the fact thatthe cysteine sulfonic acid was not observed randomly withinthe peptide, this cysteine modification was not described tobe generated by sample preparation and gel electrophoresis[35]. Thus, a biological role of the detected modified cysteinemight be feasible, because oxidation of cysteine plays animportant role in protein structure and function [36]. More-over, cysteines from all subunits were modified by acryl-amide, except a-2, b-1i, b-2i and b-5i. Interestingly, thesecysteine modifications are not always consistent in all pro-tein species of a subunit. To give an example, whereas thepeptide LTPIHDHIFCCR of the subunit b-1.1 was modifiedby either one or two acryl amides (1525.69, 1596.67), the



Table 1. Normalized protein and protein species ratios of rat proteasomal subunits (IV/II) determined by four different quantificationapproaches

Proteinspecies

2-DE Averagespot intensity

2-DETopSpota)

cICAT/2-DEMALDI-MSa) b)

cICAT/LCESI-MSa) b)

cICAT/LCMALDI-MSa) b)

Cysteinyl peptides(500–3000 Da)ChemScore .5

a–1.1 186 0.91 6 0.05 – 1.12 6 0.10 1.14 6 0.05 4a–2.1 210 1.06 6 0.09 – 0.95 – – 1a–2.2 23 0.88 6 0.20 – 6 1a–3.1 238 0.99 6 0.04 1.10 – 2a–3.2 76 0.96 6 0.04 – 6 1.00 6 0.34 0.77 6 0.01 2a–3.3 90 0.73 6 0.14 1.45 – 2a–4.1 128 1.05 6 0.06 1.26 –

1.17 – 1.01 6 0.10

1a–4.2 78 0.93 6 0.19 1.02 – 6 1a–4.3 162 0.92 6 0.02 – 1a–4.4 41 1.17 6 0.17 – 1a–5.1 97 1.06 6 0.03 1.22 6 0.13 1.00 6 0.10 – 3a–5.2 114 0.94 6 0.04 1.22 6 0.09 6 3a–6.1 31 1.40 6 0.21 –

1.00 6 0.14 1.02 6 0.02

3a–6.2 73 1.02 6 0.12 0.98 – 3a–6.3 49 1.29 6 0.13 1.17 – 6 3a–6.4 123 1.06 6 0.17 1.10 6 0.10 3a–6.5 28 0.53 6 0.12 – 3a–7.1 201 1.02 6 0.08 1.15 –

1.07 – 1.06 6 0.011

a–7.2 36 1.22 6 0.16 1.29 – 6 1a–7.3 15 0.5 – – 1

b–1.1 131 0.67 6 0.09 0.59 6 0.10 0.80 6 0.03 0.64 6 0.05 4b–1.2 4 n.d. 0.59 – 6 4b–2.1 121 1.05 6 0.04 – – – 2b–2.2 13 0.41 6 0.59 – 6 2b–3.1/2 196 0.88 6 0.06 – 0.97 6 0.02 0.99 6 0.00 2b–4.1 193 1.04 6 0.09 – 0.99 6 0.05 0.95 6 0.01 3b–4.2 20 0.90 6 0.19 – 6 3b–5.1 96 0.42 6 0.20 – – – 2b–5.2 13 n.d. – 6 2b–6.1 202 1.05 6 0.15 0.71 6 0.12

0.95 – 0.95 6 0.022

b–6.2 24 1.06 6 0.43 1.29 – 6 2b–7.1 193 1.04 6 0.11 –

0.92 6 0.20 –1

b–7.2 20 0.72 6 0.34 – 6 1b–1i 68 5.09 6 0.71 – – – 0b–2i 21 1.40 6 0.48 – – – 3b–5i 73 1.29 6 0.70 – – – 2

a) 6SD; –SD could not be calculated; n.d. not detected (ratio could not be calculated due to low spot intensities)b) – no cysteine containing peptide detected or not enough cysteine peptides for calculating a SD

same peptide was not found in b-1.2. Furthermore, themodification of methionine to sulphoxide and pyroglutamic-acid on N-terminal Gln were detected in almost all proteins.A detailed overview of all modified peptides found in thisstudy is given in the Supplementary Table.

3.3 cICAT/2-DE/MALDI-MS

With this approach all 17 proteins were identified by PMF.Considering the quantitative data, 7 proteins (consisting of15 species) were calculated due to the presence of cysteinyl

peptide pairs in their PMFs. In comparison to the non-cICATlabeled 2-DE pattern, the proteasomal subunitsa-5,a-6,a-7,b-1, b-2 and b-1i showed additional spots shifted in the Mr and/or pI (a-5, a-6 and b-1i) (data not shown). However, the re-placement of b-1 by b-1i in subtype IVcould be detected by themeans of 10% standard deviation. In addition to the 0.59 folddownregulation of both protein species of b-1.1, 4 proteasomeprotein species were equally, eight slightly over and oneslightly under-represented in subtype IV (Table 1). The com-parison with the ratios of the 2-DE-TopSpot evaluation showswith one exception (a-3.3) a good confirmation within 20%.



Figure 3. Comparison of thepeptide mass fingerprint froma-4.1 (a) and a-4.2 (b) generatedby an ULTRAFLEX II mass spec-trometer. Detection of two addi-tional mass peaks at m/z 2317.08and 2333.08 for a-4.2 in compar-ison to a-4.1, which correspondto the peptide 61KICOXALDD-NVCPA MAFAGLTADAR81 con-taining cysteine sulfonic acid(63) and propionamido-cyste-ine (70) and its methionine oxi-dized analogon. Mass peaks atm/z 2340.09 and 2356.09 areassigned to the correspondingpeptides where both cysteinesare modified by acrylamide.

Figure 4. Verification of cysteineoxidation by MALDI-MS/MS.MALDI-MS/MS cysteine sulfonicacid (63) and propionamido-cys-teine (70) of the peptide 61KICOX-

ALDDNVCPAMAFAGLTADAR81

from a–4.2. The complete y-se-ries is shown as well as the b-ions 2, 3, 4. The spectrum wasgenerated with an ULTRAFLEX IImass spectrometer.

3.4 ICAT/LC/ESI-MS/MS

The most common quantification techniques besides thecomparative 2-DE image analysis exploit proteins taggedwith different isotopes identified and quantified by LC/ESI-MS/MS. However, the identification and quantification oftagged cysteinyl peptides in a complex mixture representonly a relative number and do not distinguish protein species[37]. Therefore, the LC/ESI-MS/MS results showed only one

ratio as the sum of all protein species for each identifiedproteasomal protein (Table 1). By using this approach, 12 ofthe 17 proteins were identified and quantified. These identi-fications include all constitutive subunits but no immunosubunit, due to the limited number of detectable cysteinylpeptides or low amounts. Considering the quantificationdata, the observed ratio for b-1 corresponds to the expecteddifference due to the substitution of b-1 to b-1i in subtype IV(0.80). The remaining quantitative data obtained for each



subtype showed variations from 0.92 (b-7) up to 1.17 (a-4) insubtype IV (Table 1). However, considering these data moreclosely the standard deviations varied from 0.02 (b-3) up to0.34 (a-3). One reason was that multiple forms of cICATlabeled peptide pairs were generated, due to the oxidation ofmethionine and the cICAT reagent. Moreover, the fact thatone peptide can occur in multiple charges increased thetrouble by calculating the correct ratio.

3.5 cICAT/LC/MALDI-MS/MS

In the recent years, offline-LC separation has been linked toMALDI instruments because of the improved fragmentationtechnology. Using this approach, 9 of the 17 subunits wereidentified and quantified (Table 1). More than 700 MS peakswere detected, before all common matrix peaks were elimi-nated by MS-Screener. 460 of these masses showed a massshift of 9.031 (one cICAT), 18.062 (two cICAT), and 15.99(Met-ox), using the MS-Screener cICAT analysis tool. About240 of these precursor masses fitted with the in silico data-base, which includes 313 predicted cysteinyl peptides in themolecular range of 500 to 3000 Da of the 20S proteasome,whereas one missed cleavage site, N-terminal acetylation,pyroglutamic-acid on N-terminal glutamine and oxidation ofmethionine were allowed for the calculation. Approximately40% of these could be confirmed by MALDI-MS/MS andwere used for quantification. Examining these results moreclosely, it could be noticed that the ratios differ concerningtheir peak intensities. cICAT peak pairs with intensity lowerthan 1000 counts showed a higher standard deviation (15%)in contrast to peaks with higher intensities (,5%) (data notshown). Furthermore, it seemed that the relationship be-tween the intensities of a mass and their ratio do not have alinear character. As a result, in this study we have only usedpeak pairs with intensities higher than 1000 counts to calcu-late the relative ratios. In total, six proteasomal proteinsshowed an equal value for both subtypes. Furthermore, therelative amount of b-1 (1:0.64) and a-3 (1:0.77) were slightlydecreased in contrast to a-1 (1:1.14). PTMs were not detectedusing this approach.

3.6 Comparison of all four approaches

Considering the identification, with the 2-DE-based approa-ches all 17 proteasomal proteins were identified includingthe three immuno forms. Moreover, 19 additionally proteinspecies with different Mr and/or pI properties could bedetected (Table 1). In contrast, using the cICAT/LC-basedapproaches only 9 (MALDI-MS/MS) and 12 (ESI-MS/MS)proteins were identified and none of the identified cysteinylpeptide pairs contained evidence to the presence of proteinspecies. Additionally, immuno forms were not detected al-though two of them have more than two cysteinyl peptides(b-2i, b-5i). Moreover, we found many masses in the MSspectra which fit to predicted tryptic peptides of the 17 sub-units, but we were not able to verify all of them by MS/MS.

However, the cICAT-LC-based methods showed some posi-tive properties concerning the attended analysis time andprotein amount. Whereas the time effort for cICAT-LC-basedmethods did not exceed 20 h for one experiment includingdata analysis, the 2-DE based ones were much more timeconsuming (14 days). Moreover, the required proteinamount for 2-DE based identifications (3 mg) is far beyondthe cICAT-LC-based (200 ng) methods (Table 2). Consideringthe dynamic range more closely, 2-DE is better than its com-mon reputation. In this study we have identified proteinsfrom extremely faint spots with at least 15 counts (a-7.3), upto spots with intensities higher than 230 (a-3.1) (Fig. 2). Thisis equal to factor 16. In contrast to it, all cICAT approachesfailed to identify proteins of lower concentrations, such as b-2, b-2i, b-5 and b-5i (less than 130 counts in 2-DE), althoughthese proteins have at least two cysteinyl peptides withChemScore higher than five. Moreover, we found that thedetection of cICAT modified cysteinyl peptides is not only aquestion of sample amounts and the presence of cysteines inthe sequence, it is more dependent on the chemical proper-ties of the peptides. To give an example, the spots b–2.1 andb–4.1 have similar spot intensities and at least two cysteinylpeptides with a ChemScore higher than five; however thecICAT peptides of b-2.1 (MAAVSVFQAPVGGFSFDNCR,LVSEAIAAGIFNDLGSGSNIDLCVISK) were not found, incontrast to the peptides from b–4.1 (NLADCLR, CLEELQK,ILLLCVGEAGDTVQFAEYIQK). Here, all peptides weredetected. Interestingly, all mentioned cysteinyl peptides werefound in the 2-DE-MS approach in its acryl amide modifiedforms. This fact indicates that there are no steric hindrancesfor chemical modifications of the cysteines (SupplementaryTable).

The comprehensive and reliable identification of proteincompounds in a mixture was a pre-requisite to analysequantitative differences. In our investigations all four meth-ods showed the expected ratio of 1:1 between the subtype IIand IV after normalization. However, we observed thatquantification by 2-DE image analysis and cICAT/2-DE/MALDI-MS displayed a higher SD for faint spots (Table 1).The cICAT/2-DE approach displayed problems by matchingthe spots to their equivalents in the normal 2-DE gel. In thiscase additional spots were detected which prohibited to dis-tinguish between regular protein species and artificial ones(data not shown). Regarding the SDs for relative quantifica-tion, cICAT/LC/MALDI-MS was the most accurate approachwhich allowed the detection of minimal changes in 20S pro-teasome subtypes (5%). Considering the detection of PTMs,only the 2-DE in combination with MALDI-MS/MS allowedthe identification of biological relevant modifications such asN-terminal acetylation and cysteine sulfoxidation as shownin a-4.2. The Supplementary Table gives a first view onmodified amino acids found in proteasomal proteins. Theelucidation of the individual PTM of each spot is a challeng-ing task, which prerequisites a 100% sequencing coveragefor each protein species, which is possible in principle, asshown by Okkels et al. [38] for ESAT6 proteins of



Table 2. Comparison of the four different quantification approaches investigated to analyse rat 20S proteasomalproteins with a complexity of 17 proteins diversified in 37 protein species

Criteria\Methods cICAT/LC/ESI-MS/MS

cICAT/LC/MALDI-MS/MS

cICAT/2-DE/MALDI-MS

2-DE/MALDI-MS

Time effort ,5 h ,20 h ,14 day ,14 daySensitivity 200 ng 200 ng 30 mg 30 mgProtein coverage 70% 53% 100% 100%Protein species coveragea) – – 41% 92%Quantitative Standard deviation 620% 65% n.a. 620%Sequence coverage proteins 10–20% 10–20% 30–70% 30–70%PTMs No No Yes Yes

a) related to quantification

Mycobacterium tuberculosis. The efforts needed for such adetailed study are outlined in Hoehenwarter et al. [39] andare outside the scope of the present investigation, where wefocused on the comparison of the four different quantifica-tion methods.

4 Concluding remarks

We have used two 20S proteasome subtypes to compareidentification and quantification results of gel-free and gel-based proteomic approaches. In general, the 2-DE basedmethods were able to identify and quantify at the proteinspecies level in contrast to cICAT/LC-approaches. Even onthe protein level, only 70% (LC/ESI) and 53% (LC/MALDI)of the predicted 17 proteins were identified by LC-MS/MS.Moreover, quantitative differences between constitutive andimmunoproteins were only detected by 2-DE image analysiswhich clearly shows that rat liver 20S proteasome subtype IVhas a higher content of immuno b-subunits compared tostandard b-subunits than subtype II. However, a more com-plex protein mixture would increase the error in spot quan-tification due to the higher number of different proteinswithin one spot [33]. Furthermore, a clear advantage of 2-DEwas the detection of PTMs. On the other hand, ICAT-LCbased methods were faster to perform and less material wasneeded. In particular, ICAT combined with LC-MALDIseemed to be the best approach for relative quantification ofthe proteins. However, although many peptide masses cor-responding to proteasomal proteins were found by LC-MALDI, only a limited number of MS/MS were of sufficientevidence for unequivocal identification of the proteins. TheICAT-2-DE approach showed major disadvantages in com-parison to the other approaches because multiple spots weregenerated from single proteins. In summary, 2-DE ap-proaches are a good choice to study biological systems on theprotein species level, whereas the complementary ICAT-LCapproaches simplify the analysis by reducing the complexityto the protein level.

This work was supported by Bundesministerium für Bildungund Forschung (Germany) Grant 031U107A/207A and by aDFG grant (Da 146–6) to B.D.

5 References

[1] O’Farrell, P. H., J. Biol. Chem. 1975, 250, 4007–4021.

[2] Uenlue, M., Morgan, M. E., Minden, J. S., Electrophoresis1997, 18, 2071–2077.

[3] Fenn, J. B., Mann, M., Meng, C. K., Wong, S. F., Whitehouse,C. M., Science 1989, 246, 64–71.

[4] Tanaka, K., Waki, H., Ido, Y., Akita, S. et al., Rapid Commun.Mass Sp. 1988, 2, 151–153.

[5] Karas, M., Hillenkamp, F., Anal. Chem. 1988, 60, 2299–301.

[6] Ong, S. E., Blagoev, B., Kratchmarova, I., Kristensen, D. B. etal., Mol. Cell. Proteomics 2002, 1, 376–386.

[7] Ishihama, Y., Sato, T., Tabata, T., Miyamoto, N. et al., Nat.Biotechnol. 2005, 23, 617–621.

[8] Staes, A., Demol, H., Van Damme, J., Martens, L. et al., J.Proteome Res. 2004, 3, 786–791.

[9] Gygi, S. P., Rist, B., Gerber, S. A., Turecek, F. et al., Nat. Bio-technol. 1999, 17, 994–999.

[10] Schmidt, A., Kellermann, J., Lottspeich, F., Proteomics 2005,5, 4–15.

[11] Chakraborty, A., Regnier, F. E., J. Chromatogr. A 2002, 8, 173–184.

[12] Ross, P. L., Huang, Y. N., Marchese, J. N., Williamson, B. etal., Mol. Cell. Proteomics. 2004, 3, 1154–1169.

[13] Bochtler, M., Ditzel, L., Groll, M., Huber, R., Annu. Rev. Bio-phys. Biomol. Struct. 1999, 28, 295–317.

[14] Dahlmann, B., Ruppert, T., Kuehn , L., Merforth, S., Kloetzel,P.-M., J. Mol. Biol. 2000, 303, 643–653.

[15] Kuckelkorn, U., Ruppert, T., Strehl, B., Jungblut, P. R. et al., J.Exp. Med. 2002, 195, 983–990.

[16] Claverol, S., Burlet-Schiltz, O., Girbal-Neuhauser, E., Gairin,J. E., Monsarrat, B., Mol. Cell. Proteomics 2002 , 8, 567–578.

[17] Hayter, J. R., Doherty, M. K., Whitehead, C., McCormack, H.et al., Mol. Cell. Proteomics 2005, 4, 1370–1381.

[18] Iwafune, Y., Kawasaki, H., Hirano, H., Electrophoresis 2002,23, 329–338.



[19] Huang, L., Jacob, R. J., Pegg, S. C., Baldwin, M. A. et al., J.Biol. Chem. 2001, 27, 28327–28339.

[20] Jungblut, P., Thiede, B., Zimny-Arndt, U., Muller, E. C. et al.,Electrophoresis. 1996, 17, 839–47.

[21] Tokunaga, F., Aruga, R., Iwanaga, S., Tanaka, K. et al., FEBSLett. 1990, 263, 373–375.

[22] Kimura, J., Takaoka, M., Tanaka, S., Sassa, H. et al., J. Biol.Chem. 2000, 7, 4635–4639.

[23] Mason, G. G. F., Hendil, K. B., Rivett, A. J., Eur. J. Biochem.1996, 238, 453–462.

[24] Bose, S., Stratford, F. L. L., Broadfoot, K. I., Mason, G. G. F.,Rivett, A. J., Biochem. J. 2004, 378, 177–184.

[25] Schmid, H. P., Vallon, R., Tomek, W., Kreutzer-Schmid, C. etal., Biochimie 1993, 75, 905–910.

[26] Sümegi, M., Hunyadi-Gulyás, È., Medzihradszky, K. F.,Udvardy, A., Biochem. Biophys. Res. Commun. 2003, 3,1284–1289.

[27] Froment, C., Uttenweiler-Joseph, S., Bousquet-Dubouch, M.P. et al., Proteomics 2005, 5, 2351–2363.

[28] Klose, J., Kobalz, U., Electrophoresis 1995, 16, 1034–1059.

[29] Jungblut, P. R., Seifert, R., J. Biochem. Biophys. Methods1990, 21, 47–58.

[30] Thiede, B., Hohenwarter, W., Krah, A., Mattow, J. et al.,Methods 2005, 35, 237–247.

[31] Perkins, D. N., Pappin, D. J., Creasy, D. M., Cottrell, J. S.,Electrophoresis 1999, 20, 3551–3567.

[32] Han, D. K., Eng, J., Zhou, H., Aebersold, R., Nature Bio-technol. 2001, 19, 946–951.

[33] Schmidt, F., Schmid, M., Mattow, J., Facius, A. et al., J. Am.Soc. Mass Spectrom. 2003, 14, 943–956.

[34] Wang, Y., Vivekananda, S., Men, L., Zhang, Q., J. Am. Soc.Mass Spectrom. 2004, 15, 697–702.

[35] Hamdan, M., Galvani, M., Righetti, P. G., Mass Spectrom.Rev. 2001, 20, 121–141.

[36] Jacob, C., Giles, G. I., Giles, N. M., Sies, H., Angew. Chem.Int. Edit. 2003, 42, 4742–4758.

[37] Schmidt, F., Donahoe, S., Hagens, K., Mattow, J. et al., Mol.Cell. Proteomics 2004, 3, 24–42.

[38] Okkels, L. M., Muller, E. C., Schmid, M., Rosenkrands, I. et al.,Proteomics 2004,4, 2954–2960.

[39] Hoehenwarter, W., Ackermann, R., Zimny-Arndt, U., Kumar,N. M., Jungblut, P. R., Amino Acids 2006, in press.


Comprehensive quantitative proteome analysis of 20S proteasome subtypes from rat liver by isotope...

Documents

Transcript of Comprehensive quantitative proteome analysis of 20S proteasome subtypes from rat liver by isotope...