Download pdf - Proteome analysis of differentiating human myoblasts by dialysis-assisted two-dimensional gel electrophoresis (DAGE)

RESEARCH ARTICLE

Proteome analysis of differentiating human

myoblasts by dialysis-assisted two-dimensional gel

electrophoresis (DAGE)

Florence Gonnet1, Belaid Bouazza2, Gaël Armel Millot3, 4, Simin Ziaei5, Luis Garcia5,Gillian S. Butler-Browne2, Vincent Mouly2, Jeanine Tortajada1, Olivier Danos6 andFédor Svinartchouk5

1 Laboratoire Analyse et Modélisation pour la Biologie et l’Environnement (LAMBE),Université d’Evry Val d’Essonne, CNRS UMR 8587, Evry, France

2 UMR-S 787 Inserm, Institut de Myologie, Université Pierre et Marie Curie Paris VI, Paris, France3 UMR 7147, CNRS, Institut Curie, Paris, France4 Université Pierre et Marie Curie, 4 place Jussieu, France5 GENETHON, CNRS UMR 8115, Evry, France6 Hôpital Necker – Enfants Malades, INSERM U781, Paris, France

In the present study, modifications in cytosolic expressed proteins during human myoblast dif-ferentiation were studied by dialysis-assisted 2-DE (DAGE, [1]). About 1000 spots were analysedon the 5th and 13th day of differentiation with a dynamic range of protein expression exceeding1000-fold. During myogenic differentiation, the number of nonmatching spots as well as theextent of quantitative differences between matched spots significantly increased. Over one hun-dred differentially expressed spots were excised and identified by MALDI-TOF MS. The differ-entiation-associated expression pattern of eight proteins was validated by Western blot analysis.Differential expression of several proteins was demonstrated for the first time in human myo-tubes. Interestingly, Ingenuity pathway analysis grouped 30 of these proteins into two over-lapping networks containing as principal nodes IGF-1 and tumour necrosis factor, two proteinsknown to play a crucial role in cytogenesis. Our results illustrate the large rearrangement of theproteome during the differentiation of human myoblasts and provide evidence for new partnersinvolved in this complex process.

Received: March 20, 2007Revised: July 23, 2007

Accepted: October 12, 2007

Keywords:

2-D PAGE / Differential expression / Human / Mass spectrometry / Protein identification

264 Proteomics 2008, 8, 264–278

1 Introduction

In mature muscle most satellite cells are in a quiescent state,but will become activated and proliferate in response to

extrinsic signals. The process of myogenesis from com-mitted muscle precursor cells to differentiated myotubes hasbeen studied in a number of cell and tissue culture models.As a result of these studies, key molecular aspects of thismultistep process have been elucidated [2, 3]. Previous stud-ies have mainly concentrated on the contractile proteinswhich are the major proteins expressed by the differentiatedmyotubes. However, no proteomic study of differentiation ofhuman myoblasts has been published to date and this will beessential for studies on muscle disease and ageing.

In the present study, we have applied a recently pub-lished modification of the classical 2-DE technique (dialysis-assisted 2-DE (DAGE), [1]) in order to investigate the mod-

Correspondence: Dr. Fédor Svinartchouk, GENETHON, 1 bis ruede l’Internationale – 91002 Evry, FranceE-mail: [email protected]: 133-(0)-1-69-47-28-38

Abbreviations: DAGE, dialysis-assisted 2-DE; Mb, myoblasts;MudPIT, multidimensional protein identification; M5, myotubesat the 5th day of differentiation; M13, myotubes at the13th dayof differentiation

DOI 10.1002/pmic.200700261

© 2008 WILEY-VCH Verlag GmbH & Co. KGaA, Weinheim www.proteomics-journal.com

Proteomics 2008, 8, 264–278 Cell Biology 265

ification in the cytosolic proteins which occurs during dif-ferentiation of human myoblasts isolated from skeletalmuscle biopsies and cultured in vitro. In the absence ofserum, proliferating myoblasts withdraw from the cell cycleand differentiate into polynucleated myotubes, mimickingthe first steps of muscular development and regeneration[4]. Comparative proteomic studies have been performed fordifferentiated mouse myoblasts using 2-DE [5] and multi-dimensional protein identification (MudPIT) shotgun pro-filing [6]. About 2000 proteins were identified in thesestudies. However, a surprisingly small number of proteinswas found to be regulated in the study of Tannu et al. [5]using a 2-DE-based proteomic method. Using the MudPITanalysis and the total cumulative spectral count recordedfrom each protein as a semiquantitative measure of relativeabundance, Kislinger and colleagues [5] detected changes inthe abundance of hundreds of proteins linked to cell adhe-sion, intracellular signalling, metabolism and muscle con-traction.

DAGE substantially improves spot matching and dimin-ishes quantitative variations when comparing complex pro-tein mixtures [1]. In the present study using this methodabout 1000 spots from human myoblasts and myotubes wereanalysed on the 5th and 13th days of differentiation with thedynamic range of protein expression exceeding 1000-fold.During the differentiation process the number of non-matching spots as well as the extent of quantitative differ-ences between matched spots significantly increased. Overone hundred differentially expressed spots were excised andidentified by MALDI-TOF MS and the expression of eight ofthese proteins was confirmed by Western blot analysis.

2 Materials and methods

2.1 Equipment, chemicals and antibodies

The Protean IEF cell for IEF and criterion cell for SDS-PAGE,IGP strips (11 cm 3–10 nonlinear), BioRad Bradford proteinassay, Coomassie R250, mineral oil, DTT and thiourea werepurchased from BioRad. Carrier ampholytes (3/5 and 6/11)were purchased from Amersham Pharmacia Biotech.CHAPS and Triton X-100 were purchased from Sigma.Amidosulphobetaine (ASB-14) was purchased from Calbio-chem. Complete protease inhibitor cocktail tablets were pur-chased from Roche. Benzonase was purchased from Merck.

2.1.1 Antibodies

Mouse anti-annexin I mAb, rabbit polyclonal antibodiesagainst alpha-actinin, TrxR (H-200), creatine-kinase M(N-13), factor XIII (C-20), DJ-1 (N-20), DDAH II (N-20) werepurchased from Santa Cruz (CA, USA). Mouse antidesminantibodies D33, was purchased from DAKO (Denmark).Rabbit anti-STAT1 antibodies were provided by New EnglandBiolabs (Beverly, MA, USA).

2.2 Sample preparations

2.2.1 Cell culture and harvesting

The isolation and culture of human muscle cells were carriedout in agreement with the French legislation on ethical rules.The human primary myoblast strain MCQ31 was derivedfrom a quadriceps biopsy of a stillborn boy that was born at31 wks of gestation. The myoblasts were isolated from biop-sies as described previously [7]. Briefly, the muscle was finelyminced and 25–50 explants were plated onto noncoated Petridishes (diameter 60 mm) first overnight in serum drops andthen in 2 mL of growth medium consisting of DMEM con-taining 0.1 g/L gentamicin and 20% FCS (Gibco). Serumwas selected to optimise the growth capacity of human myo-blasts and maintain myogenicity. Human myoblasts werecultivated at 377C in a humid atmosphere containing 7%CO2. Once the mononucleated cells had migrated out fromthe explants (between 2 and 3 days), they were removed bytrypsinisation (1.5% trypsin, 0.04% EDTA) and replanted at1.86103 cells/cm2 in 5 mL of the same growth medium in60 mm dishes. These human myoblasts were routinely pro-pagated in DMEM (Gibco BRL), supplemented with 20% v/vFCS (Hyclone), 100 U/mL penicillin and 100 mg/mL strep-tomycin. To induce the myoblasts to differentiate into myo-tubes, the cells were placed in the same medium withoutserum, supplemented with 10 mg/mL of insulin and 100 mg/mL of apotransferrin. All cultures were incubated at 377Cunder a humidified atmosphere of air with 7% CO2. Initialplating density was between 26103 and 56103 cells/cm2.The myogenic purity of the human myoblast cultures wasmonitored by immunocytochemistry using an antibody spe-cific for desmin.

2.2.2 Protein extractions

Cells were rinsed five times with PBS and then 56106 cellswere lysed in 1 mL of homogenisation buffer (20 mM Tris-HCl, pH 8.0, 0.32 M sucrose, 3.0 mM CaCl2, 2.0 mM Mgacetate, 0.1 mM EDTA, 1% Triton X-100, 1 mM DTT and1 mM sodium vanadate in the presence of a protease in-hibitor cocktail). The homogenate was centrifuged for10 min at 20006g to remove nuclei and then centrifugedfor 40 min at 25 0006g. The supernatant was then incu-bated with benzonase (final concentration 100 U/mL) for20 min at room temperature and precipitated by adding 1.2volume of precooled acetone for at least 4 h at 07C. Theprecipitate was collected by centrifugation at 20 0006g for10 min; the pellet was washed once with methanol andstored under methanol at –207C. Protein estimation wasperformed using the commercial Bradford reagent usingBSA as a protein standard. On the day of utilisation, thesample was collected by centrifugation and the pellet wasdried and resuspended in IEF buffer (final protein con-centration 2 mg/mL).


266 F. Gonnet et al. Proteomics 2008, 8, 264–278

2.2.3 2-DE

2-DE was performed as described previously [1]. IEF wasperformed using a Protean IEF cell and its 17 cm tray. pH 3–10 nonlinear strips (11 cm) were rehydrated for at least 6 h ina rehydration buffer consisting of 7 M urea, 2 M thiourea,1% v/v Triton X-100, 1% w/v ASB, 2% w/v CHAPS and20 mM DTT (IEF buffer). IEF for each pair of samples wasconducted on two half-strips obtained from one strip cutlongitudinally. This was done in order to avoid differences inthe IPG. The sample (100 mg of protein solubilised in 50 mLof the rehydration buffer containing 2% v/v IPG buffer 3–5and 2% v/v IPG buffer 6–11 (Amersham)) was loaded on awick cut from GF/F glass microfibre filter (Whatman) 2.5 cmlong and 3 mm wide. Long wicks required for this adaptationwere cut 2.0 cm long by 0.5 cm wide from Whatman 3 mmpaper. Anode wicks were saturated in IEF buffer supple-mented with 5 mM acetic acid and cathode wicks were satu-rated with the same buffer supplemented with 10 mMlysine/arginine. The wicks were patted dry and placed in theIEF cell in the following order: (i) anode wick with one endoverlapping the platinum wire electrode and the other endoverlapping the sample loaded wick; (ii) sample wick over-lapping with the anode end of the strip; (iii) cathode wickwith one end overlapping the platinum wire electrode andthe other end overlapping with the cathode end of the IGPstrip. In order to run the two samples under identical condi-tions this cup-loading construct was covered with a 4 mmwidth dialysis membrane (Spectra/Por, MWCO 3500, Spec-trum Laboratories) and the second sample was loaded onexactly the same cup-loading construct on the surface of themembrane. In this construct two strips are placed face-to-face with a dialysis membrane between them. IEF was per-formed using the following voltage program: 50 V constantfor 5 h, linear ramp to 4000 V over 6 h, then 8000 V for a totalof 30 000 V?h using the Protean IEF cell (BioRad). The cur-rent was limited to 50 mA per strip and the temperaturemaintained at 187C.

For SDS-PAGE, the IPG strips were incubated in equili-bration buffer containing 37.5 mM Tris-HCl (pH 8.8), 6 Murea, 4% w/v SDS, 20% glycerol and 20 mM DTT for 20 min.The equilibrated IPG strips were transferred for the seconddimension onto Criterium 10–20% linear gradient gel(BioRad). Electrophoresis was carried out at room tempera-ture using a Criterium gel system (BioRad) with runningbuffer (25 mM Tris, 192 mM glycine), containing 0.1% w/vSDS, at 150 V until the front of fast migrating ions reachedthe bottom of the gel. Upon completion of SDS-PAGE, gelswere stained for 3–4 h with gentle agitation in 0.02% CBBsolution containing 5% acetic acid and 10% ethanol and thendestained by several changes of 5% acetic acid.

2.2.4 2-D gel image analysis

A region on the gels was selected corresponding to pI 5–8and MW from 15 to about 250 kDa. We have selected this

region to avoid known problem areas in 2-DE gels becausethe goal is to generate the most reproducible data as abenchmark for further studies and because the majority ofthe mouse myoblast and myotube proteins were found on2-DE gels in these pI and MW intervals [5]. 2-D gels werecompared using Image Master Platinum software (Amers-ham). Spot detection settings were adjusted so that a max-imum of visible spots was detected. Spot volumes and per-centage of spot volumes were calculated using Image MasterPlatinum build-in options. Spot detection parameters weresettled as follows: smooth – 1, minimal area – 40 and saliency– 70. These parameters give about 900 spots in the selectedregion of 767 cm. An automated matching algorithm withfive manually settled landmarks was used to match spots onthe gels. Landmarks were selected from all regions of the geland were only assigned when a spot was clearly present in allanalysed gels. After automatic spot matching, additional spotediting and matching were performed manually. The sameperson performed all assessments across the experiment.After spot matching the following parameters were takeninto account to estimate proteomics variations during myo-blasts differentiation: percent of spots matched which isindicative of qualitative differences among gels and the nor-malised distribution of the log2 ratio of the percentage ofspot volumes for each matching pair of spots in the two gels,which reflects the quantitative differences (see below). Foreach condition, several pairs (three to four) of DAGE wereperformed until two of them showing a good spot focalisa-tion were obtained. These two pairs were then retained forfurther analyses.

2.2.5 Statistical analyses

The R environment was used for all the analyses [8]. Statis-tical significance was set at p�0.05. Pairs of 2-D gels wereperformed in duplicated for each technical (DAGE or tradi-tional) and biological (myoblasts (Mb), myotubes at the 5thday of differentiation (M5) or myotubes at the 13th day ofdifferentiation (M13)) conditions. The proportions of mat-ched spots between conditions were compared with a w2 testwithout continuity correction (Figs. 3 and 5B). The mostunfavourable pair was chosen in each test (for example, inthe comparison between Mb/M5 and Mb/M13 (Fig. 3B), thetest was performed with the two grey bars which correspondto the minimal distance between the two conditions, mean-ing that in the three other pairs, the statistical result will bemore significant). Because significant differences wereexpected, the three w2 tests were penalised using the Holmprocedure.

Variations of spot intensities between pairs of gels werestudied by a method similar to chip analysis. The log2 ratioof the percentage of spot volumes for each matching pair ofspots in the two gels was established respecting the followingrule: myoblasts derived percentage of spot volumes were inthe numerator and differentiating cell derived percentage ofspot volumes in the denominator (Mb/M5 for 5 days of dif-



ferentiation, Mb/M13 for 13 days of differentiation and Mb/Mb as a control). This means that a positive log2 ratio corre-sponds to a decrease of spot intensity during the differentia-tion and to an increase for a negative log2 ratio. Then log2ratios were centred by the median of the distribution to allowcomparisons between experiments. In Fig. 5A, the six dis-tributions of the log2 ratios were compared two by two with aKolmogorov–Smirnov test (15 tests). For the three tests be-tween duplicates, the p-values were not corrected becausenonsignificant difference was expected. For the 12 remain-ing tests, the p-values were penalised using the Holm proce-dure. The most unfavourable p-values (among four) areshown in the left part of Fig. 5A. To estimate quantitativevariations during differentiation we compared the propor-tion of matched spots for Mb/Mb and Mb/M13 gels at dif-ferent thresholds of the log2 ratio (Table 1 and Fig. 5B). Thesame statistical procedure as the one explained for Fig. 3 wasperformed. The two w2 tests were penalised using the Holmprocedure (Fig. 5B).

Table 1. Percentage of spots (mean values) beyond differentthresholds for Mb/Mb and Mb/M13 comparison

Threshold 23.0 22.0 21.6 21.0 1.0 1.6 2.0 3.0

Mb/Mb 0.18 0.42 1.32 5.67 5.9 1.57 0.72 0.12Mb/M13 0.81 2.93 5.47 15.94 16.46 7.90 4.55 1.6False positives (%) 22 14 24 36 36 20 16 8

The means of the total number of matched spots is 839.5 and 752for the Mb/Mb and Mb/M13 comparisons, respectively. Percent-age of false positive spots for each threshold was estimated asratio: (Mb/Mb)/(Mb/M13).

2.2.6 In-gel digestion

Spots of interest were manually excised and placed in 1.5 mLEppendorf tubes. They were consequently washed with500 mL of methanol, 500 mL of water, 500 mL of methanol andthen digested overnight in-gel with 12 mL (0.005 mg/mL) por-cine modified trypsin protease (Promega) in 100 mMammonium bicarbonate buffer (pH 8.0) at 377C. Trypticpeptides were directly extracted/concentrated using C18ZipTip pipette tips (Millipore), washed with 1% formic acidand eluted with a mixture of 2 mL of 80% ACN, 1% formicacid and 1 mL of saturated CHCA matrix solution in 80%ACN, 1% formic acid.

2.2.7 MS and database interrogation

The total peptide eluate (3 mL) was applied dropwise to aMALDI target. MALDI-TOF MS was performed on aVoyager-DE STR™ (Applied Biosystems). All spectra wereacquired in a positive reflector mode (20 kV acceleratingvoltage) with 225 ns delayed extraction. Typically 300shots were recorded per sample from different crystals on

the MALDI target, 50 shots per crystal. Trypsin autolyticpeptides were used to internally calibrate each deisotopedspectrum to a mass accuracy within 20 ppm. Peak listswere generated by Data Explorer software (Applied Bio-systems, version 4.0.0.0 of 18-Oct-2000). After calibrationpeptide mass list was filtered for trypsin autolytic pep-tides and keratines (if any). Tryptic monoisotopic peptidemasses were searched by using Aldente software (version11/09/2006) (http://www.expasy.org/tools/aldente) in theUniProtKB/Swiss-Prot database (release 50.7 of 19-Sep-2006, 258 126 protein entries) and UniProtKB/TrEMBLdatabase (release 33.7 of 19-Sep-2006, 3 209 448 entries)with the following parameters: human/mouse species,one missed cleavage site and mass tolerance setting of20 ppm. Partial chemical modification such as oxidationof methionine was taken into consideration for the quer-ies. Highest confidence identifications have statisticallysignificant search scores, are consistent with the gelregion from which the protein was excised (MW and pI)and account for the extent of sequence coverage and thenumber of peptides matched (minimum of four). Onlyproteins with the highest statistical search scores arepresented in the paper. In some cases where MS spectrapermit the identification of two proteins with nearly thesame probability and similar MW and pI both proteinsare presented in the paper. Where it was possible splicevariants of the same proteins were distinguished bymanual comparison of their MS spectra and respectiveposition on the gel. When such identification wasimpossible, all variants are presented in the paper. Thesame procedure was applied to distinguish single proteinmember of a multiprotein family.

2.2.8 Western blot

Twenty micrograms of the same protein extracts that wereused for 2-DE analysis were analysed by Western blot asdescribed [9]. Two analyses were performed for each proteinand results are presented as mean 6 SD.

2.2.9 Ingenuity pathway analysis

Differentially expressed proteins were analysed by the Inge-nuity pathway analysis software to monitor possible globalproteomic regulation during myoblast differentiation. TheIngenuity pathway analysis program (https://analysis.ingenuity.com) uses a knowledge base derived from the lit-erature to relate gene products, based on their interactionand function. This software is designed to identify dynami-cally generated biological networks, global canonical path-ways and global functions. Basically, the identified proteinswith their fold change and corresponding Swiss-Prot acces-sion numbers are uploaded as an Excel spreadsheet file intothe Ingenuity software. Ingenuity then uses these data tonavigate the Ingenuity Pathways Knowledge Base and extractan overlapping network(s) between the candidate proteins. A



score better than 2 is usually attributed to a valid network(the score represent the log probability that this network wasfound by random chance).

3 Results

In the present study, we utilised the DAGE approach tocompare the proteomes of a proliferating human myoblastcell strain MCQ31 and of myotubes at different times of dif-ferentiation. The human primary myoblast cell strain,MCQ31, was chosen for studies in order to limit contamina-tion of primary human myoblasts by fibroblasts [10]. Whencultivated in growth medium proliferating MCQ31 cellsgrow as mononuclear flattened cells in a monolayer. Whencultivated cells were placed in the differentiation mediumthey acquired an elongated morphology and fused to formmultinucleated myotubes (Fig. 1). These changes are clearlyvisible on the 5th and 13th day in differentiation medium. By13 days the majority of the myogenic cells are included inmultinucleated myotubes and no more cell fusion occurs,indicating the terminal differentiation of the cells.

3.1 Intrasample variations: DAGE versus traditional

workflow

Protein extracts were prepared from proliferating MCQ31Mb, M5, and fully differentiated myotubes (13 days of dif-ferentiation; M13) and were separated by DAGE. Since theDAGE method, in its present state, permits comparison ofonly two samples, we performed pairwise analysis of gels forMb/Mb, Mb/M5 and Mb/M13 samples. Two pairs of gelswere produced for each comparison.

Protein extracts from myoblasts were analysed by DAGEand traditional workflow approaches in order to compare re-producibility of the methods. Analysis of the 2-D imagespermitted the identification of about 900 protein spots in theselected region of the gels (Fig. 2) of which 98% could bematched on the second gel of the pair when the experimentis performed by DAGE (Fig. 3A). This precise matching isdue to the identical distribution of spots along the pI axis(Fig. 2). Small differences in positioning of the correspond-ing spots along a molecular weight axis between the gelscannot be eliminated by this method (Fig. 2 and results notshown). Comparison of two pairs of gels (Mb (5) vs. Mb (6)and Mb (7) vs. Mb (8), Fig. 3A) performed by traditionalworkflow, gives 85% spot matching that is significantly lowercompare to DAGE method (p,0.001). Because intrasamplevariability found by the DAGE method was very similar forprotein extracts from two different cell lines (mouse C2C12 [1]and human MCQ31, Fig. 3A) we conclude that experimentalintrasample variations estimated by DAGE method are lim-ited to the values presented in Fig. 3A. Consequently, moreimportant variations can be attributed to a variability insample preparation or, when comparing different biologicalsamples, to the differences in protein expression.

Figure 1. Differentiation of MCQ31 cells in culture. (A) Proliferat-ing myoblasts and (B) 13 days in the differentiation mediumstained with an antibody again desmin. Desmin staining wasperformed as described in Section 2.

3.2 Differential expression of cytosolic proteins

during human myoblast differentiation

Protein extracts prepared from M5 and M13 were analysedby comparing with protein extracts from Mb. This analysiswas based on the comparison between Mb/Mb pairs versusMb/M5 pairs and Mb/Mb versus Mb/M13 pairs performed byDAGE method. The time points were chosen because thereare clear signs of differentiation starting at day 5 and becauseday 13 is the end point of MCQ31 differentiation in vitro.

Figure 4 shows a comparison of Mb versus M13. Thesame region used in Mb/Mb comparison was selected foranalysis. The percentage of matched spots was significantlydiminished for myoblasts versus 5 days and myoblasts versus13 days as compare to myoblasts versus myoblasts analysis



Figure 2. Representative pairsof 2-D gels of protein extractsfrom human MCQ31 myoblasts(A and B). Samples wereresolved by DAGE method andthe proteins were stained withCoomassie R-250. The boxedareas from gel A and B corre-spond to the regions analysedfor spot matching and areenlarged in (C) and (D), respec-tively.

Figure 3. Analysis of spot matching under different experimentalconditions. Pairs of gels were performed in duplicates using (A)either DAGE or the traditional workflow on myoblasts (Mb)extracts or (B) DAGE on cytoplasmic extracts prepared at differ-ent stages of myoblast differentiation (M5, 5 days in differentia-tion media; M13, 13 days in differentiation media). The propor-tion of matched spots is represented in grey for the first pair ofgels and in black for the second. Statistical analyses were per-formed as described in Section 2. In (A), the ratio of matchedspots/total number of spots correspond to the following gelsfrom left to right: Mb (1)/Mb (2) (817/837), Mb (3)/Mb (4) (862/882),Mb (5)/Mb (6) (840/979) and Mb (7)/Mb (8) (830/968). In (B), theratio of matched spots/total number of spots correspond to thefollowing gels from left to right: Mb (1)/Mb (2) (817/837), Mb (3)/Mb (4) (862/882), Mb (9)/M5 (1) (789/928), Mb (10)/M5 (2) (890/1006), Mb (11)/M13 (1) (781/996) and Mb (12)/M13 (2) (723/946).

(Fig. 3B). In the pairs Mb/M5 and Mb/M13 gels, a mean of87 and 77% of spots were matched, respectively (Fig. 3B). In

some cases, in order to confirm spot matching, matchedspots were excised from both gels and identified by MALDI-MS. In all cases the alignment along the pI axis was exact.During differentiation the percentage of matched spotsdiminished, reflecting changes in the pattern of proteinexpression (Figs. 3, 4). The average number of unmatchedspots is 17 for the Mb/Mb comparison, whereas for Mb/M5this number increased up to 100 and for Mb/M13 up to 170spots. We then asked the question: What is the probabilitythat these 170 proteins are really differentially expressed be-tween myoblasts and myotubes? Taking into considerationthe large number of analysed spots, we reasoned that theprobability of “false positive” could be simply estimated asthe ratio 17/170, or about 10%.

3.3 Quantitative differences

The level of protein expression on a gel, calculated as a per-cent of the volume, varied from 0.007 to 13 (more than 1000-fold difference). To estimate quantitative changes in proteinexpression during myoblast differentiation, we comparedvariations in matched spot intensities for Mb/Mb, Mb/M5and Mb/M13 gels. To do this, we built up distributions of thelog2 ratio of the percentage of spot volumes for each match-ing pair of spots in the corresponding gel pairs. We thenobtained distributions which were centred with the medianto allow comparison between experiments (Fig. 5A and Sec-tion 2). Comparison of the distributions of the log2 ratio forMb/Mb, Mb/M5 and Mb/M13 gels shows that the propor-tion of high and low ratios is significantly increased during



Figure 4. Representative pairsof 2-D gel images of proteinextracts from myoblasts (A, C)and 13 day myotubes (B, D).Only the analysed regions asdefined in Fig. 2 are shown. Thelocations of spots identified byMALDI-TOF (Table 2) are indi-cated in C and D.

myoblast differentiation. The percentage of matched spotsbeyond different thresholds of median centred log2 ratios isindicated for Mb/Mb and Mb/M13 in Table 1. At any thresh-old, this percentage is higher for Mb/M13 compared to Mb/Mb gels, indicating that quantitative variations accumulatedduring differentiation. Of the 752 matched spots (averagefrom two DAGE experiments) for the Mb/M13 comparison,41 protein spots were found to be up-regulated during dif-ferentiation at a threshold of 21.6 (corresponding to a three-fold increase) and 59 proteins were found to be down-regu-lated at a threshold of 11.6 (corresponding to a three-folddecrease). Those numbers are significantly higher than thenumbers observed with Mb/Mb (Fig. 5B). In order to esti-mate the proportion of false positives among the numbers ofproteins found beyond thresholds in Mb/M13, we comparedproportion of spots beyond a certain threshold for Mb/Mband Mb/M13 gels. For example for threshold 11.6, this pro-portion is four times higher when comparing Mb/M13 versusMb/Mb (Table 1). Consequently, we estimated that the rate offalse positive for this ratio is about 20%.

3.4 Identification of differentially expressed proteins

Some proteins found differentially expressed in both pairs ofMb/M13 gels were identified by MALDI-MS (Table 2). Weorganised these 105 proteins into functional categories inorder to reveal changes in protein expression linked to intra-cellular signalling, metabolism, structure, protein synthesisand protease activity.

3.5 Validation of selected data by Western blotting

Expression of some proteins, identified by MALDI-MS, wasconfirmed by Western blotting. Eight proteins were selectedfor the analysis: two with a specific expression in myotubes,three with an enhanced expression in myotubes and threewith a decreased expression in myotubes (Fig. 6). Proteinextracts from two independent experiments of myoblast dif-ferentiation were fractionated by denaturing SDS-PAGE andanalysed by Western blotting (Fig. 6). Changes in theexpression level of seven proteins were fully consistent with



Figure 5. Quantitative variationsof matched spots during myo-blasts differentiation. Statisticalanalyses were realised as descri-bed in Section 2. (A) Distributionof the log2 ratio of percentage ofspot volume of matched spotscentred with the median. Theproportion of high and low ratiosis significantly increased duringmyoblast differentiation. (B) Thepercentages of log2 ratiosbeyond the 21.6 and 11.6thresholds are higher in Mb/M13than in Mb/Mb gel pairs. Fromleft to right, the percentages cor-respond to the gels Mb (1)/Mb (2)(14/817 above, 18/817 below), Mb(3)/Mb (4) (8/862 above, 8/862below), Mb (11)/M13 (1) (40/781above, 77/781 below) and Mb(12)/M13 (2) (42/723 above, 43/723 below).

Figure 6. (A) Western blot analysis ofselected proteins displaying differentialexpression. Samples from two inde-pendent experiments of myoblasts dif-ferentiation were used in the analysis.Lanes from 1 to 3 correspond to the firstexperiment (1, myoblats, 2–9 days ofdifferentiation, 3–13 days of differentia-tion); 4 and 5 to the second experiment(4, myoblasts, 5–13 days of differentia-tion). The same quantity of proteins wasloaded in each lane. (B) Comparison ofprotein expression by DAGE and West-ern blot analysis. Mean values for West-ern analysis was calculated for proteinextracts from myoblasts and myotubeson the day 13 of differentiation (lanes 1and 3 for the first experiment; lanes 4and 5 for the second one).

the DAGE data (factor XIII, creatine kinase, DJ-1, STAT 1,TrxR, annexin 1 and alpha-actinin). For DDAH II no differ-ence was found by Western blotting while DAGE showedincreased expression of this protein during differentiation.To some degree discrepancy in quantification by DAGE and

Western analysis could be attributed to protein modificationsthat change the pI of proteins resulting in multiple spots notvisible in Western analysis. Overestimation of STAT1 proteinin myotube gels can be attributed to contamination of thecorresponding spot (72T) by the Gelsolin precursor (Table 2).



Table 2. The proteins identified by MALDI-TOF MS from human Mb and M13

Spot no. Protein name (Swiss-Prot accession no.) TheoreticalMW/pI

Sequencecoverage/no. of peptidesmatched/(% of totalpeptides no.)

Score Z-score Foldincrease/decrease(average6 SD)

Metabolism

1T, 2T, 3T Beta enolase (P13929) 46 855/7.7 39/19/(46) 26.2 1322.7 100.004T Aldehyde dehydrogenase 1A1 (P00352) 54 862/6.3 21/9/(31) 8.1 478.45 100.005T Dihydropyrimidinase-related protein-3 (Q14195) 61 964/6.0 27/13/(41) 22 978.9 100.006T Dihydropyrimidinase-related protein-1 (Q14194) 62 184/6.6 18/9/(33) 5.5 145.5 100.007T GDP-L-fucose synthetase (Q13630) 35 893/6.1 19/15/(21) 3.00 1026.8 100a)

8T GST theta 2 (P30712) 27 507/6.0 23/6/(22) 5.1 875.9 100.009T ECH1 protein (Q96EZ9) 35 816/8.2 20/7/(21) 8.9 325.4 100.0010T, 11T, 12T Creatine kinase, M chain (P06732) 43 101/6.8 51/19/(70) 47.9 7191.4 45.87 6 6.0713T Enoyl-CoA hydratase, mitochondrial precursor (P30084) 31 372/8.3 48/13/(45) 16 3299.4 6.25 6 0.2614T Aldo-keto reductase family 1 member C2 (P52895) 36 736/7.1 55/14/(52) 43.1 4097.3 5.53/10015T Ubiquinol-cytochrome-c reductase complex core protein

(P31930)52 619/5.9 48/19/(38) 57.7 1372.7 3.51 6 2.31

16T (1) 6-Phosphogluconolactonase (O95336)(2) ER protein ERp29 (P30040)

(1)27 546/5.7(2)25 852/6.1

64/13/(34)38/10/(26)

(1) 42.4(2) 10.7

(1) 3309.1(2) 2940.6

3.31 6 0.08

17T Alcohol dehydrogenase [NADP1] (P14550) 36 573/6.3 51/18/(60) 90 5329.5 2.60 6 0.6718T GST Mu 3 (P21266) 26 428/5.4 36/7/(44) 16 1540.1 2.95 6 0.0119T Aldose reductase (P15121) 35 854/6.5 54/15/(44) 13.1 926 2.77 6 0.1620T DPYSL3 protein 73 909/6.1 37/23/(50) 72.5 2850.7 3.17 6 0.7021T Isocitrate dehydrogenase [NADP] cytoplasmic (O75874) 46 659/6.5 58/26/(59) 110.7 8086.4 2.43 6 0.0422T Dihydropyrimidinase-related protein 2 (Q16555) 62 299/6.0 58/27/(47) 99.7 4195.7 1.88 6 0.7223T Galactokinase (P51570) 42 272/6.1 39/17/(43) 34.9 2159 1.78 6 0.8224T Adenosylhomocysteinase (P23526) 47 584/5.9 35/18/(49) 25.3 3207.2 2.34 6 0.0125T Pyruvate kinase muscle isozyme M1/M2 (P14618) 57 930/7.6 24/13/(29) 13.8 459 3.85 6 2.1626M Tryptophanyl-tRNA synthetase (P23381) 53 166/5.8 36/14/(28) 33.3 751.7 –27T 3-Hydroxyisobutyryl-coenzyme A hydrolase isoform 1

(Q53GA8)43 320/8.40 30/12/(43) 22.8 1913.4 3.96 6 1.53

28T ATP synthase D chain, mitochondrial (O75947) 18 491/5.2 62/8/(35) 16.7 2374.2 1.95 6 0.0429T Adenine phosphoribosyltransferase (APRT) (P07741) 19 608/5.8 40/6/(20) 7.9 1060.3 0.86 6 0.0130M Transaldolase (P37837) 37 540/6.4 36/16/(35) 21.7 512.46 0.63 6 0.0831M Glucosamine-fructose-6-phosphate aminotransferase

(Q06210)76 627/6.4 27/18/(38) 35.8 1146.67 0.60 6 0.10

32M ATP synthase beta chain, mitochondrial precursor (P06576) 56 560/5.3 66/31/(63) 201.1 6984.2 0.59 6 0.34

Cell structure

33T CKAP4 protein (Q504S5) 58 205/5.2 41/20/(65) 57 4537.58 Albumin34T Macrophage capping protein (P40121) 38 518/5.9 25/11/(44) 12.4 1597.1 3.54 6 1.3635T Septin-9 (Q9UHD8) 64 649/7.2 30/17/(35) 31.7 1249 2.49 6 0.7036T Septin-2 (Q15019) 41 487/6.2 61/18/(41) 59.9 4338.2 2.31 6 0.6137T, 38T Gelsolin (P06396) 82 958/5.72 39/27/(60) 65 3582.7 2.26 6 0.2139M (1) Tubulin alpha-6 chain (Q9BQE3)

(2) Tubulin beta-2 chain (P07437)(1)49 895/5.0(2)49 670/4.8

45/15/(25)33/14/(24)

(1) 30.6(2) 24.7

(1) 1665.6(2) 763.8

1.01 6 0.50

40M Tubulin alpha-6 chain (Q9BQE3) 49 895/5.0 38/12/(50) 16.6 3363.8 2.59 6 1.8241M Actin, cytoplasmic 2 (P63260M) and/or actin, cytoplasmic 1

(beta-actin) (P60710M)41 793/5.3 41/15/(43) 26.8 1761.3 0.62a)

42M Caldesmon, splice variant 2 (Q05682-2) 65 707/6.4 28/15/(38) 18.5 911.6 0.65 6 0.1443M Procollagen-lysine,2-oxoglutarate 5-dioxygenase 2; splice

isoform 2 (O00469-2)87 097/6.2 27/22/(44) 52.2 1345.2 0.57 6 0.12

44M (1) Filamin C (Q14315)(2) filamin A (Q60FE6) (9 common ions)

(1)290 955/5.7(2)277 501/5.7

21/44/(56)14/26/(33)

(1) 109.1(2) 29.6

(1) 973.4(2) 263.8

0.26a)

45M, 46M (1) Alpha-actinin 4 (O43707)(2) alpha-actinin 1 (P12814) (16 common ions)

(1)104 855/5.3(2)103 056/5.3

47/45/(62)47/30/(41)

(1) 206.9(2) 66.5

(1) 2385.8(2) 766.8

0.20 6 0.01



Table 2. Continued




47M LIM and SH3 domain protein 1 (Q14847) 29 717/6.6 40/13/(37) 15.4 981.78 0.0148M Vimentin (P08670) 53 520/5.06 78/44/(69) 251.8 6735.5 0.01

Cell defensive

49T, 50T Peroxiredoxin 6 (P30041) 24 903/6.0 74/15/(58) 55.9 6707 6.39 6 1.8751T Nicotinamide N-methyltransferase (P40261) 29 574/5.6 59/14/(41) 23.3 3296.3 3.77 6 1.1352T Heat-shock protein beta-1 (HSP 27) (P04792) 22 782/5.9 77/14/(48) 49.2 10 447.6 2.94 6 0.2253T Peroxiredoxin 2 (P32119) 21 760/5.67 44/11/(58) 20.6 4527.3 2.48 6 0.6454T Superoxide dismutase (Cu-Zn) (P00441) 15 804/5.7 67/7/(18) 7 915.5 2.28 6 1.1055M Protein disulphide-isomerase A6 (Q15084) 46 170/5.0 45/16/(48) 40.9 3415.3 0.44 6 0.1256M DNA damage binding protein 1(Q16531) 126 966/5.1 26/26/(67) 41 2366.2 0.26 6 0.1157M 78 kDa glucose-regulated protein (P11021) 70 477/5.0 58/41/(76) 262 5203.4 0.16 6 0.13

Signal transduction

58M Thyroid receptor-interacting protein 13 (Q15645) 48 550/5.72 26/9/(36) 12.3 553.5 0.0159M Mitogen-activated protein kinase 1 (ERK-2) (P28482) 41 390/6.5 27/12/(28) 10.9 362.5 0.0160T Toll-interacting protein (Q9H0E2) 30 282/5.7 21/4/(67) 2.5 0 100.0061T Adenylyl cyclase-associated protein 2 (CAP 2) (P40123) 52 824/6.0 35/17/(35) 23.7 1158.9 4.92 6 2.0562T Stathmin (oncoprotein 18) (Q96CE4) 17 303/5.8 31/5/(38) 2.5 846.7 2.44 6 0.7563T AH receptor-interacting protein (O00170) 37 664/6.1 28/8/(35) 11.1 1118.5 3.47 6 2.1564T Prohibitin (P35232) 29 804/5.6 51/10/(32) 22.5 2036.7 2.78 6 0.4565T Guanine nucleotide-binding protein G(I)/G(S)/G(T) beta sub

unit 1 (P62873)37 245/5.6 21/10/(31) 13 951.3 2.55 6 1.29

66T DJ-1 protein (oncogene DJ1) (Q99497) 19 891/6.3 39/8/(22) 14.2 602.5 2.85 6 0.7767T Thioredoxin-dependent peroxide reductase (P30048) 21 468/5.78 40/8/(27) 9.6 1754 3.35 6 0.1268T Annexin A1 (P04083) 38 715/6.6 49/13/(50) 38.7 7871.7 2.60 6 0.4669T NG,NG-dimethylarginine dimethylaminohydrolase 2

(DDAH II) (O95865)29 643/5.66 46/12/(41) 29 1427.2 2.23 6 0.59

70T Guanine nucleotide-binding protein G(I)/G(S)/G(T) betasubunit 2 (P62879)

37 199/5.6 25/11/(42) 16.5 1099.2 2.06 6 0.26

71M Ras-related protein Rab-2A (P61019) 23 546/6.1 52/11/(52) 21.9 2345.1 1.03 6 0.2172T (1) Gelsolin precursor (P06396)

(2) Signal transducer and activator of transcription1-alpha/beta (STAT1) (P42224)

(1)82 958/5.7(2)87 333/5.7

35/23/(37)30/19/(31)

(1) 50.1(2) 59.8

(1) 1194.1(2) 2173.3

0.72 6 0.08

73M Signal transducer and activator of transcription 1-alpha/beta(STAT1) (P42224)

87 333/5.7 32/19/(41) 18.4 948.6 0.72 6 0.09

74M Thioredoxin reductase 1, cytoplasmic (TrxR) (Q16881) 54 461/6.1 49/26/(45) 58.7 2128.6 0.64 6 0.1975M (1) cAMP-dependent protein kinase type I -alpha

regulatory subunit (P10644)(2) Thioredoxin domain containing protein 5 (Q5TCQ0)

(1)42 982/5.3

(2)47 628/5.6

43/17/(25)

30/12/(18)

(1) 43.0

(2) 16.0

(1) 897.08

(2) 707.59

0.53 6 0.45

76M Apolipoprotein A-I (P15497)a) 27 549/5.4 70/17/(29) 85.5 5014.2 0.06 6 0.0777M Rho GDI 1 (P52565) 23 076/5.0 61/13/(54) 25.3 2520.3 0.0178M Serine/threonine protein phosphatase PP1-beta catalytic

subunit (P62140)37 055/5.9 29/9/(20) 5 217.5 0.01a)

Contractile apparatus

79T Myosin light polypeptide 6B (P14649) 22 764/5.6 55/16/(59) 27.1 5075.3 100.0080T Calponin-3 (Q15417) 36 414/5.7 28/9/(48) 3.6 195.8 100.00

Proteasome and protease activity

81T Cathepsin D (P07339) 37 852/5.6 28/9/(48) 27.3 1312.1 8.01 6 5.4282T Ubiquitin carboxyl-terminal hydrolase isozyme L1 (UCH-L1)

(P09936)24 825/5.3 56/12/(29) 17.6 1158.6 3.58 6 0.20

83T Leukocyte elastase inhibitor (P30740) 42 741/5.91 49/20/(40) 58.7 2650.7 2.46 6 0.15



Table 2. Continued




84T Proteasome subunit, alpha type, 1 (Q53YE8) 29 597/6.15 54/12/(40) 22.3 3632.7 3.07 6 0.8985T Proliferation-associated protein 2G4 (Q9UQ80) 43 655/6.1 46/13/(27) 21 774 1.19 6 0.1086M 26S proteasome non-ATPase regulatory subunit 2 (Q13200) 100 198/5.08 22/18/(56) 28.8 1158.6 0.01a)

87M Nonidentified 0.01a)

Intracellular trafficking

88T Tumour susceptibility gene 101 protein (Q99816) 43 945/6.1 33/13/(48) 18 1687.9 100.0089M ERGIC-53 protein precursor (P49257) 54 208/5.8 37/19/(33) 32 737.6 –90M Vesicular integral-membrane protein VIP3 (Q12907) 35 515/6.07 47/15/(31) 33.5 1388.4 0.01

Protein synthesis

91T Eukaryotic translation initiation factor 3 subunit 3 (O15372) 39 931/6.1 45/11/(41) 20.8 1853.3 100.0092T Elongation factor Tu (P49411) 45 044/6.32 60/25/(64) 95 8302.8 1.41 6 0.2593M Elongation factor 2 (P13639) 95 206/6.41 34/26/(50) 43.3 1082.20 0.69 6 0.29

Others

94T Ferritin light chain (P02792) 20 020/5.5 55/11/(44) 18.9 5236.3 100.0095T, 96T, 97T Serotransferrin (transferrin) (P02787) 77 051/6.8 34/23/(50) 38.5 1225 100.0098T Biliverdin reductase A precursor (P53004) 33 429/6.1 43/11/(50) 23.7 4846.2 4.68/10099T, 100T Coagulation factor XIII A chain (P00488) 83 268/5.7 38/25/(57) 47 1826.8 100.00101T Lupus La protein (P05455) 46 836/6.7 37/13/(41) 16.000 885.50 1.01 6 0.31102T (1) Synaptic vesicle membrane protein VAT-1 (Q99536)

(2) Solute carrier family 25 member 24, isoform 1(Q6NUK1) Q9P129

(1) 41 920/5.9(2) 53 338/6.0

44/13/(28)23/11/(23)

(1) 8.8(2) 3.6

(1) 500.6(2) 369.27

2.25 6 0.66

103T Solute carrier family 25 member 24, isoform 1 (Q6PJJ9);Idem (2) Q6NUK1 (102T), only one amino acid difference

53 354/6.0 48/27/(61) 129 3770.7 1.81 6 0.17

104T SAM domain and HD domain-containing protein 1 (Q9Y3Z3) 72 199/6.7 18/8/(23) 7.4 217.9 1.55 6 0.31105M Chloride intracellular channel protein 1 (O00299) 26 791/5.1 59/16/(42) 39.2 3109.6 0.97 6 0.18106M Iron-responsive element binding protein 1 (P21399) 98 397/6.2 40/33/(61) 146.5 4042.40 0.94 6 0.20107M DNA replication licensing factor MCM7 (P33993) 81 307/6.1 50/32/(63) 207.1 5767 0.45 6 0.09108M Heterogeneous nuclear ribonucleoprotein (P61978) 50 976/5.4 27/10/(26) 8.1 1179 0.17 6 0.09109M, 110M Serum albumin (P02769)a) 69 394/5.8 34/24/(46) 48.60 2495.60 0.05 6 0.03111M Serotransferrin (Q29443)a) 78 339/8.6 26/17/(29) 8.4 320.64 0.01112M DEAD-box protein 3, X-chromosomal (O00571) 73 111/6.7 33/23/(51) 41.2 1843.7 0.01113M EH-domain containing protein 2 (Q9NZN4) 61 160/6.0 45/25/(61) 101.4 3682.70 0.08 6 0.09114M Annexin A5 (Annexin V) (P08758) 35 805/4.9 78/26/(74) 198.2 13 837.9 0.01

Minimal detection

115D Nonidentified 0.74a)

116D Lamin A/C (Q5TCJ3) 65 134/6.4 42/22/(51) 89.4 927.5 0.38 6 0.03117D Syntaxin binding protein 1 (P61764) 68 735/6.3 19/13/(43) 16.8 336.6 1.13 6 0.31

a) The corresponding pair of spots on the second gel pair could not be identified.M and T following after numbers indicate gel taken for MALDI-TOF analysis: M, myoblasts and T, myotubes. All spots D (minimal detection)present less than 0.008% of total proteins on the gel. Spots were quantified as a percent of total volume and mode of their regulation inresponse to myogenic differentiation is indicated as a ratio (expression in myotubes/myoblasts). Average values and SD of the ratio fromtwo gels are presented in the table. When a spot in the M13 gel has no counterpart in the Mb gel this ratio is equal to 100 (specific expres-sion in myotubes) and for specific expression in Mb this ratio is equal to 0.01. The MS spectra of some spots allow us to identify two pro-teins with nearly the same probability (spots 72T and 75M). In some cases, several spots correspond to the same protein, probablyreflecting PTMs (spots 10T, 11T, 12T; 95T, 96T, 97T; 39M, 40M etc.). In the case when we were unable to distinguish between different splicevariants or members of a multiprotein family all variants are presented in the table (spots 41M, 44M, 45M, 117D), see also Section 2. In a fewcases, we were unable to measure the response to myogenic differentiation (the counterpart of spot 33T was masked by albumin; spots26M and 89M are not separated on the M13 gel and the corresponding spot on this gel contains both proteins).



3.6 Sensitivity of the approach

It is interesting to be able to calculate the minimal proteinabundance in the cell that can be detected and identifiedusing the DAGE – MALDI-TOF approach. We made thisestimation for the protein extracts used in our experimentsin the following way. The percentage of spot volume variedinside the gels from 12.3 (ACTB protein) to 0.007. Only afew proteins had a percentage of spot volume lower than0.007: nine in the gel for myoblasts and 15 in the gel formyotubes. Four proteins with the spot volume between0.007 and 0.008 were excised from the gel and, using theprocedure described in Section 2, three of them wereunambiguously identified (spots 115D, 116D and 117D inTable 2). Because 100 mg of protein lysates (corresponds to26106 myoblasts) were loaded on the gel, it is possible toestimate the sensitivity of this approach: about 7 ng ofprotein can be detected and identified. For a protein with amolecular weight of 70 kDa this is equal to 0.1 pM (or36104 molecules per cell).

3.7 Network analysis

To analyse the overlapping network function between pro-teins whose expression were found to be altered duringmyoblast differentiation, we uploaded the set of proteinsfrom Table 2 with their accession numbers and fold differ-ence in the expression level into the Ingenuity pathwayanalysis software. Eighty-nine proteins of the 105 submittedwere recognised by the Ingenuity software as being eligiblefor overlapping network creation. From a total of nine over-lapping networks proposed by the software, six had a score ofmore than 16 (i.e., score.2 is significant; it represents thelog of the probability that the network was found by chance).The most relevant functions extracted from these networkswere related to protein synthesis, cell growth and cell signal-ling. Very interestingly the first two ranking overlappingnetworks contained as the principal nodes IGF-1 (score 28;Fig. 7) and tumour necrosis factor (TNF superfamily, mem-ber 2; score 22) proteins, known to play a crucial role inmyogenesis [11–16]. Interestingly IL13, which was recentlydescribed as a factor recruiting reserve myogenic cell for dif-ferentiation and fusion [16], appears as the principal node inthe network with score of 18 (Fig. S1 in Supporting Infor-mation).

4 Discussion

Systematic comparison of changes in the protein composi-tion in proliferating and differentiating human myoblastsshould provide clues for wide studies of mechanisms thatunderlie the formation and physiopathology of skeletalmuscles. Recently two groups have compared the level ofprotein expression in mitotic and differentiated murineC2C12 myoblasts by 2-D PAGE and by MudPIT [6]. Total,

membrane- and nuclear-enriched proteins, and phospho-proteins were compared by 2-D PAGE. Surprisingly, a rela-tively small number of proteins (75 of 2000) appeared to bemodulated during differentiation. Kislinger comparednuclear-enriched protein fractions from proliferating myo-blasts and differentiating myotubes at different time pointsby MudPIT shotgun profiling methodology. About 1800 pro-teins were identified in this study and changes in relativeprotein abundance were detected for hundreds of proteinslinked to cell adhesion, intracellular signalling, metabolismand muscle contraction. Large-scale RNA-expression analy-sis of human skeletal myoblast differentiation have also beenreported [10]. Although the transcriptome analysis of myo-genesis was instructive, it is clear that the molecular mech-anisms involved in myogenic differentiation cannot beunderstood from the analysis of the transcriptome alone [17].Results of proteomic studies of human myoblast differentia-tion, to our knowledge, are not yet available for the scientificcommunity. This sort of database will be important if wewant to investigate alterations in protein expression causedby gene defects in skeletal muscle or modifications in proteinexpression which occurs during normal physiological pro-cesses such as muscle growth and muscle ageing.

In the present proof of principle study, we have analysedprotein extracts from primary human myotubes and myo-blasts using the DAGE method. This method substantiallyimproves spot matching and diminishes quantitative varia-tions when comparing complex protein mixtures [1]. Thepossibility of unambiguous protein matching using theDAGE method permitted us to analyse about 1000 spots onsmall gels (1169 cm). Due to the smaller spot size, small gelshave an advantage in sensitivity. The estimated sensitivity ofthe approach is several tens of thousand molecules of proteinper cell. Our method of protein preparation results in theextraction of soluble cytosolic and to some extent membraneproteins. During myoblast differentiation this set of proteinsundergoes substantial modulation. Variations in proteinexpression level can be classified into two categories: qualita-tive (protein spots detected in one cellular type only) andquantitative (variation in volume for the same spot; Figs. 3, 5).Attention has to be paid when proteins with low expressionlevels are classified into one of these categories. Because of thedetection limit of 0.007% of spot volume, the counterpart ofthe spot with low volume could be lost. We found 100 spotsspecifically expressed in the Mb/M5 comparison and 170spots in the Mb/M13 comparison. A significant portion ofmatched proteins underwent qualitative variations (Table 1).For example, 0.72 and 4.55% of matched spots varied by morethan four-fold (threshold 12) when comparing Mb/Mb andMb/M13, respectively (Table 1). If a large number of spots areanalysed and V1 is the proportion of matched spots varyingmore than X-fold for Mb/Mb comparison, and V2 is that pro-portion in Mb/M13 comparisons, then the value (1 2 V1/V2)will be the probability for a variable spot, varying more than X-fold, to represent a true difference. For example, for four-folddifference this value is equal to (1 – 0.72/4.55) = 0.84.



Figure 7. Ingenuity pathway network (score 28) obtained on a set of differentially expressed proteins during myoblast differentiation (Mbvs. M13). Proteins with their fold change and corresponding Swiss-Prot accession numbers were uploaded as an Excel spreadsheet file intothe Ingenuity software (http://www.ingenuity.com). Proteins with a coloured background correspond to the uploaded proteins, while otherinteracting proteins with a clear background were added by Ingenuity. IGF-1, insulin-like growth factor 1; FGF-19, fibroblast growth factor19; ENO3, enolase 3; CKM, creatine kinase; LASP1, LIM and SH3 protein 1; ALDHA1A1, aldehyde dehydrogenase 1; AFR1, alpha fetoproteinregulation 1; ACO1, aconitase 1; FTL, ferritin, light polypeptide; YWHAZ, tyrosine 3-monooxygenase/tryptophan 5-monooxygenase acti-vation protein; AHCY, S-adenosylhomocysteine hydrolase; MAP3K5, mitogen-activated protein kinase kinase kinase 5; ECHS1, enoylcoenzyme A hydratase; AIP, aryl hydrocarbon receptor interacting protein; WEE1, WEE1 homolog; PARK7, Parkinson disease (DJ-1);ATP5B, ATP synthase, beta polypeptide; PKM2, pyruvate kinase; EEF2, eukaryotic translation elongation factor 2; JUN, v-jun sarcoma virus17 oncogene homolog; FLNC, filamin C, gamma; MYCN, v-myc myelocytomatosis viral related oncogene; CKAP4, cytoskeleton-associatedprotein 4; UCHL1, ubiquitin carboxyl-terminal esterase L1; TUFM, Tu translation elongation factor. EIF3S#: eukaryotic translation initiationfactor 3, where # means corresponding subunit (3, 8, 10 theta etc.) Interactions found between these proteins are indicated by differentarrows.

––– : Binding; : Acts on; –––: Direct interaction; - - - -: Indirect interaction; : Kinase; : Peptidase; : Enzyme; : Cytokine;

: Transporter; : Transcriptional regulator; : Translational regulator; : Other

Some of differentially expressed proteins in myoblastsversus myotubes on the 13th day of differentiation wereidentified by MALDI-MS (Table 2). As expected the majorityof the identified proteins are cytosolic, mitochondrial ormembrane proteins. On the 13th day of differentiation someproteins known to be highly expressed during myoblast dif-ferentiation were identified (myosin, creatine-kinase, calpo-

nin-3, muscle-specific beta enolase, [18]), and others knownfor their low expression in myotubes disappeared (alpha-actinin 4; vimentin [19]). Moreover, in complete agreementwith the exit of cells from the cell cycle during myoblast dif-ferentiation, expression of a subset of proteins involved incellular proliferation (c-Crk1; DNA replication licensing fac-tor MCM7; mitogen-activated protein kinase 1) were con-



siderably decreased in the 13th day myotube cultures.Expression of other genes known to be involved in growtharrest such as prohibition [20] and tumour susceptibilitygene 101 [21, 22] increased. It is interesting to note that wewere able to detect in the gel at the 13th day of differentiationapotransferrin added to the differentiation medium at100 mg/mL.

Of 105 proteins identified by MALDI-TOF in this studyand 106 proteins identified as differentially expressed duringmouse C2C12 cells differentiation by [5], we found ten pro-teins that were identified in both studies. The expression be-haviour of five of them (Thioredoxin-dependent peroxidereductase, annexin A1, vimentin, myosin light chain 1,annexin A5) is in agreement with our data, while theexpression of four others was discordant (ATP synthase betachain, LIM and SH3 domain protein, 78 kDa glucose-regu-lated protein, calponin-3). More recent comparative prote-omic studies of differentiated mouse myoblasts using Mud-PIT shotgun profiling [6] detected changes in the abundanceof hundreds from 1800 identified proteins. Among 39 com-mon proteins found in this study and detected by Tannu et al.[5], ten (26%) showed significant difference in expressionpattern. The low level of correlation between our results andthose of Tannu et al. could be explained by the difference incellular models used for the experiments.

Finally, we would like to describe modulations in expres-sion of some proteins that have not been described before inhuman muscle differentiation.

4.1 Rho GDP-dissociation inhibitor 1 (Rho GDI 1)

This protein is specific for myoblasts and was not detectedafter 13 days of differentiation. A crucial role of the Rhofamily proteins in muscle differentiation has been shownpreviously using the C2C12 cell model, in which it regulatesthe expression of the myogenin and MEF2 genes [23].Dominant negative forms of Rho family proteins and RhoGDI suppressed the transcription of muscle-specific genes,while mutationally activated forms of Rho family proteinsstrongly activated transcription of muscle-specific genes.Moreover, C2C12 cells overexpressing RhoGDI are no longerable to differentiate into myotubes [23, 24].

4.2 STAT1

The JAK/STAT signalling pathway mediates both anti-proliferative responses following interferon stimulation andcellular proliferation in response to cytokines and growthfactors [25]. Decrease in STAT1 expression could indicate adecreased ability of myotubes to respond to growth factorsinvolved in cell proliferation.

4.3 Heterogeneous nuclear ribonucleoprotein K

Expression of hnRNP K decreased significantly during dif-ferentiation. Heterogeneous nuclear ribonucleoprotein K is

one of the major pre-mRNA binding proteins. This proteinmay play a role in specific splicing. Functional analysisshowed that hnRNP K is a component of the enhancer com-plex that promotes chicken beta-tropomyosin exon 6A splic-ing through the wild-type S4 sequence [26].

4.4 Stathmin

Increase in the level of expression of this protein duringmuscle differentiation can be attributed to the early postnatalorigin of the MCQ31 cells. Stathmin is a 19 kDa, highlyconserved cytosolic phosphoprotein involved in the coordi-nation and relay of diverse signals controlling cell growthand differentiation. Although stathmin can induce micro-tubule depolymerisation in interphase, the functions medi-ated by its interphase activity are not well understood.Recently, it was shown that stathmin can physically interactwith p27Kip1 and regulate cell motility [27]. It is abundant inthe inner cell mass of the early rat embryo at the blastulastage and is most highly expressed during the perinatal pe-riod in all tissues, including skeletal muscle [28]. In the C2myogenic cell line, stathmin expression is detected only atincreased cell density when myoblasts tend to align and dif-ferentiate in vitro toward myotubes [29] which is in good cor-relation with our observations.

4.5 Factor XIII

At the 13th day of differentiation there is a substantialincrease in factor XIII expression. Factor XIII belongs to asuperfamily of transglutaminases (TGases2) that catalyse theCa21-dependent crosslinking of proteins via the g-carbox-imide group of glutamine residues and the e-amino group oflysine. Although upregulation of factor XIII mRNA has beenfound in some human muscular dystrophies, its expressionwas undetectable in normal muscles by gene chip array andwas attributed to activated dendritic cell infiltration in dys-trophic muscle [30]. Others studies indicated that synapseformation and maintenance of the neuromuscular junction(NMJ) involved components of the coagulation cascade dur-ing development. It has been shown that either factor XIII ortTG are localised at NMJs in a developmentally regulatedfashion [31]. It is possible that expression of factor XIII byMCQ31 cells and the absence of its expression in adultmuscle reflect changes in transglutaminase subtypes used atdifferent stages of muscle ontogenesis.

The authors wish to acknowledge financial support from theAFM (Association Francaise contre les Myopathies), INSERM,Université Pierre et Marie Curie, CNRS, CPLD and theMYORES Network of Excellence, contract 511978, from the Eu-ropean Commission 6th Framework Programme. We thank Dr.Susan CURE for manuscript editing.

The authors have declared no conflict of interest.



5 References

[1] Danos, O., Svinartchouk, F., Dialysis-assisted two-dimen-sional gel electrophoresis. Electrophoresis 2006, 27, 3475–3479.

[2] Shi, X., Garry, D. J., Muscle stem cells in development,regeneration, and disease. Genes Dev. 2006, 20, 1692–1708.

[3] Dhawan, J., Rando, T. A., Stem cells in postnatal myogen-esis: molecular mechanisms of satellite cell quiescence,activation and replenishment. Trends Cell. Biol. 2005, 15,666–673.

[4] Decary, S., Mouly, V., Hamida, C. B., Sautet, A. et al., Repli-cative potential and telomere length in human skeletalmuscle: implications for satellite cell-mediated gene ther-apy. Hum. Gene Ther. 1997, 8, 1429–1438.

[5] Tannu, N. S., Rao, V. K., Chaudhary, R. M., Giorgianni, F. etal., Comparative proteomes of the proliferating C(2)C(12)myoblasts and fully differentiated myotubes reveal thecomplexity of the skeletal muscle differentiation program.Mol. Cell. Proteomics 2004, 3, 1065–1082.

[6] Kislinger, T., Gramolini, A. O., Pan, Y., Rahman, K. et al.,Proteome dynamics during C2C12 myoblast differentiation.Mol. Cell. Proteomics 2005, 4, 887–901.

[7] Edom-Vovard, F., Mouly, V., Barbet, J. P., Butler-Browne, G.S., The four populations of myoblasts involved in humanlimb muscle formation are present from the onset of primarymyotube formation. J. Cell Sci. 1999, 112, 191–199.

[8] R Development Core Team, R Foundation for StatisticalComputing, Vienna, Austria 2006.

[9] Rosa Santos, S. C., Dumon, S., Mayeux, P., Gisselbrecht, S.,Gouilleux, F., Cooperation between STAT5 and phosphati-dylinositol 3-kinase in the IL-3-dependent survival of a bonemarrow derived cell line. Oncogene 2000, 19, 1164–1172.

[10] Sterrenburg, E., Turk, R., t Hoen, P. A., van Deutekom, J. C. etal., Large-scale gene expression analysis of human skeletalmyoblast differentiation. Neuromuscul. Disord. 2004, 14,507–518.

[11] Patel, K., Macharia, R., Amthor, H., Molecular mechanismsinvolving IGF-1 and myostatin to induce muscle hyper-trophy as a therapeutic strategy for Duchenne musculardystrophy. Acta Myol. 2005, 24, 230–241.

[12] Noguchi, S., The biological function of insulin-like growthfactor-I in myogenesis and its therapeutic effect on musculardystrophy. Acta Myol. 2005, 24, 115–118.

[13] Guttridge, D. C., Signaling pathways weigh in on decisionsto make or break skeletal muscle. Curr. Opin. Clin. Nutr.Metab. Care 2004, 7, 443–450.

[14] Stewart, C. E., Rittweger, J., Adaptive processes in skeletalmuscle: molecular regulators and genetic influences. J.Musculoskelet. Neuronal Interact. 2006, 6, 73–86.

[15] Li, Y. P., Reid, M. B., Effect of tumor necrosis factor-alpha onskeletal muscle metabolism. Curr. Opin. Rheumatol. 2001,13, 483–487.

[16] Jacquemin, V., Butler-Browne, G. S., Furling, D., Mouly, V.,IL-13 mediates the recruitment of reserve cells for fusionduring IGF-1-induced hypertrophy of human myotubes. J.Cell Sci. 2007, 120, 670–681.

[17] Boutz, P. L., Chawla, G., Stoilov, P., Black, D. L., MicroRNAsregulate the expression of the alternative splicing factornPTB during muscle development. Genes Dev. 2007, 21, 71–84.

[18] Fougerousse, F., Edom-Vovard, F., Merkulova, T., Ott, M. O.et al., The muscle-specific enolase is an early marker of hu-man myogenesis. J. Muscle Res. Cell Motil. 2001, 22, 535–544.

[19] Deschenes, I., Chahine, M., Tremblay, J., Paulin, D., Puy-mirat, J., Increase in the proliferative capacity of humanmyoblasts by using the T antigen under the vimentin pro-moter control. Muscle Nerve 1997, 20, 437–445.

[20] Mishra, S., Murphy, L. C., Nyomba, B. L., Murphy, L. J., Pro-hibitin: a potential target for new therapeutics. Trends Mol.Med. 2005, 11, 192–197.

[21] Krempler, A., Henry, M. D., Triplett, A. A., Wagner, K. U., Tar-geted deletion of the Tsg101 gene results in cell cycle arrestat G1/S and p53-independent cell death. J. Biol. Chem. 2002,277, 43216–43223.

[22] Wagner, K. U., Krempler, A., Qi, Y., Park, K. et al., Tsg101 isessential for cell growth, proliferation, and cell survival ofembryonic and adult tissues. Mol. Cell Biol. 2003, 23, 150–162.

[23] Takano, H., Komuro, I., Oka, T., Shiojima, I. et al., The Rhofamily G proteins play a critical role in muscle differentia-tion. Mol. Cell Biol. 1998, 18, 1580–1589.

[24] Charrasse, S., Comunale, F., Grumbach, Y., Poulat, F. et al.,RhoA GTPase regulates M-cadherin activity and myoblastfusion. Mol. Biol Cell. 2006, 17, 749–759.

[25] Mukherjee, T., Hombria, J. C., Zeidler, M. P., Opposing rolesfor Drosophila JAK/STAT signalling during cellular prolifer-ation. Oncogene 2005, 24, 2503–2511.

[26] Expert-Bezancon, A., Le Caer, J. P., Marie, J., Heterogeneousnuclear ribonucleoprotein (hnRNP) K is a component of anintronic splicing enhancer complex that activates the splic-ing of the alternative exon 6A from chicken beta-tropomyo-sin pre-mRNA. J. Biol. Chem. 2002, 277, 16614–16623.

[27] Iancu-Rubin, C., Atweh, G. F., p27(Kip1) and stathmin sharethe stage for the first time. Trends Cell Biol. 2005, 15, 346–348.

[28] Koppel, J., Boutterin, M. C., Doye, V., Peyro-Saint-Paul, H.,Sobel, A., Developmental tissue expression and phyloge-netic conservation of stathmin, a phosphoprotein asso-ciated with cell regulations. J. Biol. Chem. 1990, 265, 3703–3707.

[29] Balogh, A., Mege, R. M., Sobel, A., Growth and cell density-dependent expression of stathmin in C2 myoblasts in cul-ture. Exp. Cell Res. 1996, 224, 8–15.

[30] Chen, Y. W., Zhao, P., Borup, R., Hoffman, E. P., Expressionprofiling in the muscular dystrophies: identification of novelaspects of molecular pathophysiology. J. Cell Biol. 2000,151, 1321–1336.

[31] Citron, B. A., Gregory, E. J., Steigerwalt, D. S., Qin, F., Fest-off, B. W., Regulation of the dual function tissue transgluta-minase/Galpha(h) during murine neuromuscular develop-ment: gene and enzyme isoform expression. Neurochem.Int. 2000, 37, 337–349.
