Identification Strategy for the Analysis of BMP2-Induced

Aug 5, 2009 - regulations have not been fully elucidated. Studies on bone morphogenic protein-2 (BMP2)-induced transdifferentiation of murine C2C12 ce...
0 downloads 0 Views 6MB Size
Download PDF
‘Two-Stage Double-Technique Hybrid (TSDTH)’ Identification Strategy for the Analysis of BMP2-Induced Transdifferentiation of Premyoblast C2C12 Cells to Osteoblast Byung-Gyu Kim,†,# Ji-Hyun Lee,† Jung-Mo Ahn,† Sung Kyu Park,‡ Ji-Hoon Cho,§ Daehee Hwang,§ Jong-Shin Yoo,| John R. Yates III,‡ Hyun-Mo Ryoo,⊥ and Je-Yoel Cho*,† Department of Biochemistry, School of Dentistry, Kyungpook National University, and 2nd BK21 program 700-422, Korea, Department of Cell Biology, The Scripps Research Institute, La Jolla, California 92014, School of Interdisciplinary Bioscience and Bioengineering, Department of Chemical Engineering, POSTECH, Pohang, 790-784, Korea, Mass Spectrometer Development team, Korea Basic Science Institute, Daejeon, Korea, and Department of Cell and Developmental Biology, School of Dentistry, Seoul National University, Seoul 110-749, Korea Received March 9, 2009

Transdifferentiation offers new opportunities in the area of cell replacement therapy; however, the molecular mechanism by which transdifferentiation occurs is not fully understood. Our understanding about the sophisticated regulations of transdifferentiation is limited yet since their comprehensive proteome regulations have not been fully elucidated. Studies on bone morphogenic protein-2 (BMP2)-induced transdifferentiation of murine C2C12 cells, a myogenic lineage committed premyoblast, to osteogenic cells can provide a full picture of the dynamic events that occur at the level of protein activity and/or expression. Here, we investigated the overall dynamic regulatory proteome associated with BMP2-induced osteoblast transdifferentiation in premyoblast C2C12 cells using a novel Two-Stage Double-Technique Hybrid (TSDTH) strategy for proteomic analysis. Here, we took the approach of a TSDTH involving phosphoproteomic analysis after a short-term treatment (stage one, 30 min) and a long-term treatment (stage two, 3 days); SILAC (Stable isotope labeling with amino acids in cell culture)-proteomics was used to map the proteins. In these experiments, a total of 1321 potential phosphoproteins were identified in stage one analysis and 433 proteins were quantified in stage two analysis. Among them, 374 BMP2-specific phosphoproteins and 54 up- or down-regulated proteins were selected. In first stage analysis, several deubiquitination enzymes including Uch-l3 as well as ubiquitination related proteins were newly identified, and its inhibitor reduced the stability of phosphorylated Smad1, and the BMP2-induced ALP levels of C2C12 cells were detected. In second stage analysis, Thrombospondin1 was identified as the highest up-regulated protein by BMP2long time stimulation and this was confirmed with immunoblot analysis. Furthermore, pathway enrichment and network analyses revealed that insulin-like growth factor (IGF) and calcium signaling pathways as well as TGFβ/BMP signaling proteins are found to be potentially involved in the early and long-term actions of BMP2. Collectively, our TSDTH is a useful simple strategy to obtain comprehensive molecular mechanism of cellular processes such as transdifferntiation. Keywords: Proteomics • transdifferentiation • BMP2 • osteoblast • phosphoproteins • SILAC • premyoblast

Introduction To recover areas of organ damage, regeneration of the correct cell types is necessary. One technique is to use stem cell therapies; * To whom correspondence should be addressed: Je-Yoel Cho, Ph.D., Deptartment of Biochemistry, School of Dentistry, Kyungpook National University, 101 Dong In-Dong, Jung-Gu, Daegu, Korea 700-422. Tel.: 82-53420-4997. Fax: 82-53-421-1417. E-mail: [email protected]. † Kyungpook National University. # Current address: Department of Protein Science, Helmholtz Zentrum Munchen, German Research Center for Environmental Health, Neuherberg, Germany. ‡ The Scripps Research Institute. § POSTECH. | Korea Basic Science Institute. ⊥ Seoul National University. 10.1021/pr900231a CCC: $40.75

 2009 American Chemical Society

however, another possibility is to replace damaged cells by the transdifferentiation of existing cells.1-4 Although the process of genetic reprogramming has been well-studied in amphibian and mammalian systems, the molecular mechanisms by which this process occurs are still not fully understood. Bone morphogenic protein-2 (BMP2) is one of the most potent bone-inducing agents of the TGFβ superfamily and was originally identified together with BMP4 and BMP7 as an inducer of ectopic bone formation when injected or implanted into rodents.5-9 Furthermore, its osteogenic transdifferentiation capability extends to fibrogenic, myogenic and adipogenic cells both in vitro and in vivo.6,10-12 BMP2 has therefore been used clinically to treat bone fractures since it increases bone mass.13,14 Journal of Proteome Research 2009, 8, 4441–4454 4441 Published on Web 08/05/2009

research articles To date, the best-characterized mediators of TGFβ signal transduction are the Smad family proteins, or Smads.15-17 The binding of BMP to its cognate serine/threonine kinase receptor results in the phosphorylation of Smads 1, 5, and 8, whereas receptors for the related factors TGFβ, nodal, and the activins primarily phosphorylate Smads 2 and 3. Receptor-mediated phosphorylation of Smad proteins occurs at two C-terminal serine residues, and triggers the translocation of phosphorylated Smads into the nucleus, where they form a complex with Smad4 to regulate the expression of hundreds of target genes.8,15,18,19 The expression and activity of Smads is modulated by a variety of receptor- and Smad-interacting proteins, including ubiquitin and phosphatases, as well as multiple proteins associated with Smad transcriptional complexes.20-25 Accumulating evidence suggests that Smad-independent regulation of TGFβ signaling also occurs through a diverse and versatile array of molecules and pathways.22,23,26-29 In recent years, highly developed proteomics technology has enabled the global screening of complex samples of proteins,30-32 which provides both qualitative and quantitative measurements of the overall proteome, including protein modifications and changes in protein expression, associated with dynamic cellular traits.33-35 However, time-course experiments remain highly demanding in terms of both cost and effort, and the selection and matching of targets for functional studies from time-course data is difficult. In an attempt to overcome these common problems in comprehensive, time-dependent proteomic analysis, we developed ‘Two Stage Double Technique Hybrid (TSDTH)’ identification strategy using the well-characterized model of BMP2-induced osteoblast transdifferentiation of premyoblast C2C12 cells. TSDTH analysis includes the application of two advanced proteomics techniques to two hypothetical stages: early (within 30 min) phosphoproteome analysis and late (3 days) SILAC-assisted quantitative proteome analysis for the process associated with BMP2-mediated transdifferentiation.

Materials and Methods Materials. Bioactive recombinant human BMP2 and recombinant human TGFβ1 were purchased from R & D Systems, Inc. SILAC Protein ID and Quantitation Kits (no. MS10033), SILAC Stable Isotopic [13C6]-L-Arginine, [13C6]-L-Lysine, Pro-Q Diamond phosphoprotein gel stain, anti-myc-HRP antibody and LipofectAMINE Plus reagent were purchased from Invitrogen (Grand Island, NY). A Bradford protein assay kit and Coomassie G250 stain was obtained from Bio-Rad (Hercules, CA). Trypsin, modified sequencing-grade, and complete protease inhibitor cocktail tablets were purchased from Roche (Mannheim, Germany). The cytochemical ALP (Alkaline phosphatase) staining kit, ammonium bicarbonate, acetonitrile, HPLC grade water and trifluoroacetic acid were purchased from Sigma-Aldrich (St. Louis, MO). Formic acid was purchased from Fluka (Switzerland). Ubiquitin aldehyde and Uch-L3 inhibitor were purchased from Merck Biosciences (Calbiochem, Darmstadt, Germany). Smad2, phospho-Smad1/5/8 antibody and Nedd4-1 antibody were purchased from Cell Signaling Technology, Inc. (Danvers, MA). Smad1 antibody was purchased from Millipore (Billerica, MA). Phospho-Erk1/2 antibody and ATP5A were purchased from Santa Cruz Biotechnology (Santa Cruz, CA). Beta actin antibody and Thrombospondin 1 antibody were purchased from Abcam (Cambridge, MA). Premixed 4442

Journal of Proteome Research • Vol. 8, No. 10, 2009

Kim et al. stacking gel solution, premixed running gel solution and RIPA lysis buffer were purchased from Elpis-Biotech (Korea). Phosphoprotein Purification and Pro-Q Diamond, CBB (Coomassie) Staining. Total cell lysates were prepared from 1 × 107 premyoblast C2C12 cells cultured in starved conditions for 2 h followed by treatment with 300 ng/mL of BMP2, 5 ng/ mL of TGFβ1 or vehicle control for 30 min. Affinity capture of phosphoproteins was performed using the phosphoprotein Purification Kit (Qiagen) according to the manufacturer’s instructions. Approximately 2 mg of total protein was adjusted to a concentration of 0.2 mg/mL in phosphoprotein lysis buffer containing 0.25% CHAPS prior to loading onto the column. After the entire sample had passed through the column, it was washed twice and the bound proteins were eluted and subsequently concentrated using nanosep ultrafiltration columns. The concentration of the enriched phosphoproteins was determined according to the Bradford assay. Finally, each sample was enriched to approximately 150-170 µg (around 8% of the total protein loaded) from the phosphorylated fractions. After electrophoresis, gels were fixed in an aqueous solution containing 40% (v/v) MeOH and 10% (v/v) CH3COOH for 30 min. To stain phosphorylated proteins with Pro-Q Diamond, the fixed gels were washed in water for 30 min, incubated with Pro-Q Diamond phosphoprotein gel stain for 3 h, and then washed in 50 mM CH3COONa-CH3COOH (pH 4.0) buffer containing 20% (v/v) CH3CN for 12 h. The phosphoproteins were imaged with excitation/emission wavelengths of 535/580 nm using a Typhoon 4900 (Amersham Biosciences). CBB staining was performed as a counter stain. SILAC Experiment. C2C12 premyoblast cells were cultured in lysine and arginine depleted advanced DMEM/F-12 (Invitrogen, Grand Island, NY) supplemented with 10% dialyzed fetal calf serum, 100 units/mL of penicillin streptomycin, and either normal or isotopically labeled Arginine (13C6-Arg) and Lysine (13C6-Lys). Cells were grown for at least 6 doublings to allow full incorporation of the labeled amino acids. After cells were washed twice with serum-free medium, one population was treated for 72 h with 300 ng/mL BMP2, whereas the other was not treated. Cells from both conditions were lysed in 1% NP40 lysis buffer containing protease inhibitors. A 1:1 mixture of control and BMP2 proteins was separated on a 12% running gel. In-Gel Tryptic Digestion. One-dimensional gel electrophoresis and Coomassie staining were performed as previously reported.36-38 In-gel digestion was also performed as described previously with slight modifications.36-38 Briefly, the excised gel slices were washed. After destaining, enough DTT solution (5 mM dithiothreitol/25 mM ammonium bicarbonate) was added to the tube to cover the gel pieces for reduction, and the tube was incubated at 60 °C for 30 min. For alkylation, IAA solution (55 mM iodoacetamide/25 mM ammonium bicarbonate) was added to the tube, and the tube was incubated at room temperature for 30 min in dark conditions. Proteins within shrunk gel slices were then digested overnight with trypsin at a substrate/enzyme ratio of 10:1 (w/w) in 25 mM ammonium bicarbonate, pH 8.0. Peptides from gel pieces were extracted by sonication for 10 min and supernatants containing peptides were transferred to new tubes. µLC-ESI-MS/MS Analysis of Phosphoproteins. LC-MS/MS analysis was carried out using Thermo Finnigan’s ProteomeX workstation LTQ linear ion trap MS (Thermo Electron, San Jose, CA) equipped with NSI sources (San Jose, CA) as described previously.38 To identify phosphorylated peptides using neutral

Comprehensive Proteome Analysis of Osteoblast Trandifferentiation loss technology, the mass spectrometer was programmed to collect MS/MS/MS (MS3) data on precursor ions that exhibited a loss upon MS/MS corresponding to phosphorylation from different charge state peptides. The data-dependent neutral loss algorithm in XCALIBUR software was used to trigger an MS3 scan when a neutral loss of 98, 49, or 32.7 Da was detected among the two most intense fragment ions. Database searching-Tandem mass spectra were extracted, and the charge state deconvoluted and deisotoped by Sorcerer 3.4 beta2 software (Sorcerer software 3.10.4, Sorcerer Web interface 2.2.0 r334 and Trans-Proteomic Pipeline 2.9.5). All MS/ MS samples were analyzed using Sequest (ThermoFinnigan, San Jose, CA; version v.27, rev. 11). Sequest was set up to search the ipiMOUSE3.29 database (IPI ver.3.29, 53 981 entries) assuming the digestion enzyme semiTrypsin. Sequest was searched with a fragment ion mass tolerance of 1.00 Da and a parent ion tolerance of 1.5 Da. The false-positive error rate (FPR) was calculated by selecting the False Positive rates decoy option in Sorcerer program to automate reversed peptide generation for decoy database strategies that accurately estimate during the data search process by excepting against a “sequence-reversed” IPI mouse database. The FPR was calculated as: FPR ) number of false peptides/(number of true peptides + number of false peptides). The iodoacetamide derivative of cysteine was specified in Sequest as a fixed modification. Oxidation of methionine, iodoacetamide derivative of cysteine, and phosphorylation of serine, threonine, and tyrosine were specified in Sequest as variable modifications. Criteria for protein identification-Scaffold (version Scaffold01_07_00, Proteome Software, Inc., Portland, OR) was used to validate MS/MS based peptide and protein identifications. Peptide or phosphopeptide identifications were accepted if they could be established at greater than 95.0% probability as specified by the Peptide Prophet algorithm.39 Protein identifications were accepted if they could be established at greater than 95.0% probability and contained at least 1 identified peptide. Protein probabilities were assigned by the Protein Prophet algorithm.40 Proteins that contained similar peptides and could not be differentiated based on MS/MS analysis alone were grouped to satisfy the principles of parsimony. After identifying proteins, each data set was carried out with subtractive analysis using ProtAn X, an in-house analytic program. For the final selection of BMP2 regulating proteins uniquely, additional subtraction with TGFβ1 data set was also performed. Phosphorylation Sites Analysis. Assignment of phosphorylation sites was also verified manually with the Scaffold program based on the stringent criteria of 95.0% probability (data presented in Supplementary Table 1). Phosphosite (http://www.phosphosite.org/) database and Scansite (http:// scansite.mit.edu/) were used for unambiguously confirming phosphorylation sites identified in this study and predicting the most likely kinases responsible for the phosphorylation sites characterized, respectively. For kinase-substrate network analysis, network building tool CYTOSCAPE-ver 2.5.2 (http://www. cytoscape.org/) was used. The data set with a list of proteins with phosphorylation sites identified and poteinitial kinases for each site were manually superimposed onto protein-protein interaction network. Quantification of SILAC Using LTQ FT-ICR Combined Census Program Analysis. MS/MS analysis of SILAC experiments were performed using a 7-T Finnigan linear quadrupole ion trap-Fourier transform (LTQ-FT) mass spectrometer

research articles

(ThermoElectron, Bremen, Germany). Briefly, for in-gel samples, the mass spectrometer was operated in the datadependent mode to automatically switch between MS and MS2 acquisition. Survey full-scan MS spectra (m/z 600-4500) were acquired in the Fourier transform ion cyclotron resonance (FT-ICR). The five most intense ions were sequentially isolated for accurate mass measurements by an ICR-FT SIM scan. They were then fragmented in the linear ion trap by collision induced dissociation. All MS/MS data were searched against the IPI mouse protein database (Version 3.23) using the SEQUEST algorithm (version 3.0, Thermo Electron, San Jose, CA) incorporated in the BioWorks software (version 3.2). Sequest was searched with a precursor ion mass tolerance of 0.01 Da, fragment ion mass tolerance of 1.00 Da and a parent ion tolerance of 1.5 Da. The database searches allowed for a modification on the cysteine residue (carboxyamidomethylation, 57 Da) and criteria of g2 peptides was applied to filtered data sets. Protein identification was made based on the corresponding peptide identification. In this study, 442 proteins (decoy matches; 5 proteins, 1.14% false positive rate) were identified using DTASelect v 2.0.13 program utilizing a decoy database strategy that set scoring threshold values based on a 1.0 false positive rate. Quantitative analysis of the filtered results was performed using Census program.41,42 Briefly, full scans were used to generate chromatograms from the m/z range surrounding both the unlabeled and labeled precursor peptides. After generation of a chromatogram reconstruction file containing the peak information and various quantitative scores, Census calculates peptide ion intensity ratios for each pair of extracted ion chromatograms and measures correlation coefficient (r), a measure of the closeness of fit between the data points of the unlabeled and labeled ion chromatograms, set to 0.5 Standard deviations, calculated for all proteins using their respective peptide ratio measurements. We then used Grubbs test (P < 0.1) to remove outliers peptide ratios. In addition, for the selection of control-only or BMP2-only detected protein (light-label or heavy-label only protein respectively), list was extracted using Trans-Proteomic Pipeline program v 2.9. To validate the data of quantitative analysis, second set analysis of SILAC and Western blotting was performed using protein samples prepared from same or independent experiments. Alkaline Phosphatase Staining. After BMP-2-induced osteogenic transdifferentiation in C2C12 cells for 72 h, cells were washed with phosphate-buffered saline twice, fixed with 2% paraformaldehyde, and stained for alkaline phosphatase according to the manufacturer’s instructions (Sigma). Immunoblot Analysis. To validate the enrichment of phosphoproteins, equal amounts of enriched phosphoproteins (treated with or without BMP2) and total cell lysates (treated with or without BMP2) were separated by SDSpolyacrylamide gel electrophoresis and blotted onto nitrocellulose membranes. Transferred membranes were incubated with phosphorylated Smad1, 5, and 8 antibody, phospho-Erk1/2 antibody, ATP5A antibody, Thrombospondin1, β-actin antibody, phospho Ser/Thr/Tyr antibody and phosphotyrosine antibody P-Tyr-100 at a 1:1000 dilution overnight at 4 °C, followed by incubation with the appropriate horseradish peroxidase-conjugated anti-IgG secondary antibodies (1:2000) for 1 h at room temperature. The level of phosphorylated Smad1 protein (3 h) induced by BMP2 in the presence or absence of ubiquitin-aldehyde or the UchJournal of Proteome Research • Vol. 8, No. 10, 2009 4443

research articles l3 inhibitor was analyzed using phospho-Smad1, Smad1, and β-actin antibodies. Imunoblotting samples were separated by SDS-PAGE, transferred to nitrocellulose membranes, and then reacted with horseradish peroxidase-coupled antimouse, anti-rabbit (cell signaling) or anti-goat IgG antibody (Santa Cruze). Signals were developed with ECL-PLUS detection reagent and the membranes were exposed to X-ray film for an appropriate time and then developed. Network Analysis. We first identified KEGG pathways overrepresented by BMP2 specific proteins (375 BMP2 specific phosphoproteins and 21 proteins up-regulated in BMP2 stimulated cells above). Among the KEGG pathways, Insulin (IGF) and calcium signaling pathways including the largest number of BMP2 specific proteins (16 and 12 proteins, respectively) were then selected for network reconstruction, as well as TGFβ/BMP signaling pathway including 5 BMP2 specific proteins, which is central to BMP2 actions on the transdifferentiation of BMP2 stimulated C2C12 cells. For the three selected pathways, a network model focusing on the 33 BMP2 specific proteins was generated using their interactions obtained from KEGG and NCBI interactome databases and previous literature43-45 (see also the network legend).

Results TSDTH (Two-Stage Double Tech Hybrid) Strategy for Understanding Proteome Regulation of BMP2 Action on Osteoblast Transdifferentiation Process. Bone morphogenic protein-2 (BMP2) is one of the most potent bone-inducing agents of the TGFβ superfamily.5-9,13,14 BMP2, therefore, has been used for our two-stage identification approach, and the procedure is elaborated in Figure 1A. For short time stage analysis (30 min), our aim was to analyze the phosphoproteome to identify dynamic changes in the phosphorylation status or phosphorylation sites of enriched-phosphoproteins. In the long term stage analysis (72 h), we used the SILAC method combined with mass-spectrometry to identify quantitative changes in protein expression levels. First Stage (Short Time) Analysis of Phosphoproteins and Phosphorylation Sites. The phosphopeptide enrichment technique is very useful for identifying phosphorylation sites, but is limited by relatively low protein coverage and the difficulty in comparing or confirming results with counterparts or precursor peptides (unphosphorylated peptides). To overcome these limitations, we used a simple “phosphoprotein” enrichment method combined with a neutral loss scan technique on a mass analyzer to identify phosphorylation sites randomly as well as to increase the coverage of phosphorylated proteins. Phosphoproteins were initially enriched from premyoblast C2C12 cells treated or untreated with BMP2 for 30 min using a Phosphoprotein purification kit (Figure 1A). To verify for enrichment of Ser-, Thr-, and Tyr-phosphorylated proteins, we performed immunoblotting using several phosphospecific antibodies (Figure 1B,D) and Pro-Q diamond stained image analysis (Figure 1E) (see details in Figure 1 legend). Phosphorylated Smad1/5 (PSmad1/5) and phosphorylated-Erk1/2 were present at higher levels in the phosphoprotein-enriched (PhoE) sample than in the total cell lysate (TCL). Moreover, phosphorylated Smad8 was detected in the PhoE fraction but not in the TCL fraction (Figure 1B). Similar phosphoprotein enrichment results were obtained by Immunoblot analysis using phospho-Ser/Thr/Tyr or phospho-tyrosine specific antibodies, or 4444

Journal of Proteome Research • Vol. 8, No. 10, 2009

Kim et al. Pro-Q diamond and Coomassie stain analysis of the PhoE fraction after 1-D SDS-PAGE (Figure 1D,E). As phosphorylation of Smads and Erk1/2 is involved in serine, threonine, or tyrosine phosphorylation, we could confirm that the kit was capable of enriching for Ser-, Thr-, and Tyr-phosphorylated proteins. Subsequently, purified proteins were analyzed by 1D-SDS-PAGE and liquid-chromatography tandem mass spectrometry (GeLC-MS/MS) (Figure 1A). Tandem mass spectra were interpreted using the Sorcerer program with a SEQEST algorithm and the Scaffold program. Orthogonal filtering criteria (greater than 95.0% peptide probability and the protein probability contained at least 1 identified peptide probability) were used to establish a final data set containing 1321 potential phosphoproteins (from three independent experiments) which included 165 (12.5%) proteins that comprised a total of 605 phosphopeptides identified by neutral loss scan mode. Furthermore, 270 phosphopeptides from 80 proteins were identified together with their unphosphorylated counterparts (or precursor peptides) (Supplementary Table 1). Second Stage (Long-Term) Analysis for Changes in Protein Expression. To quantitatively analyze changes in protein expression in response to BMP2, we performed Stable isotope labeling with amino acids in cell culture (SILAC)mass spectrometry. BMP2-treated C2C12 cells were initially grown in medium containing [13C6]-L-Arginine and [13C6]-LLysine, whereas untreated cells were grown in medium containing [12C6]-L-Arginine and [12C6]-L-Lysine (Figure 1A). After 72 h stimulation, BMP2-induced osteoblast differentiation of C2C12 cells was evaluated by alkaline phosphatase staining (Figure 1C). From independently prepared samples, equal amounts of mixed proteins were analyzed by onedimensional gel electrophoresis in combination with LTQFT mass spectrometry (Figure 1A). After the tandem mass spectra were interpreted by SEQUEST using DTAselect2, the labeling efficiency of peptides and ion chromatograms were tested and quantified (Figure 1F) using the Census program. In this analysis, 97.7% (174 of 178) of the peptides were distributed within 0.5 of the error ratio and 82.4% of the peptides (2687 of 3261 total peptides) were successfully quantified (Figure 1F). Network Changes in the Interactions of Potential KinasesSubstrates in Response to Temporal BMP2 Stimulation. In the first stage analysis, a total of 259 phosphorylation sites (180 phospho-serine, 67 phospho-threonine, and 12 phosphotyrosine) were identified (Supplementay Table 1 and Figure 2). From subtractive analysis, 98 control, 52 common, and 109 BMP2-selective phosphorylation sites were identified (Figure 3A). Of these, 96 (37%) phosphorylation sites were previously confirmed using the PhosphoSite (http://www. phosphosite.org/) database (Supplementary Table 1). In addition, we analyzed potential protein kinases to identify all of the phosphorylation sites using an in silico SCANSITE program (http://scansite.mit.edu/) and CYTOSCAPE-ver 2.5.2 (http:// www.cytoscape.org/). These analyses revealed that the substrates of PRKCA-Protein kinase C R/β/γ (2.5-fold), CAMK2calmodulin dependent kinase 2 (2.4-fold), CLK2-cdc-like kinase 2 (2-fold), ATM-ataxia telangiectasia-mutated kinase (1.7-fold), and GSK3A-glycogen synthase kinase 3R (1.6-fold) were increased in BMP2-stimulated cells compared to control cells (Figure 3B and Supplementary Table 2). In contrast, the targets of PRKCD-protein kinase Cδ (3.3-fold) and PRKCM-protein kinase Cµ (1.5-fold) were decreased compared to unstimulated

Comprehensive Proteome Analysis of Osteoblast Trandifferentiation

research articles

Figure 1. ‘Two-Stage Double-Technique Hybrid (TSDTH)’ identification strategy. (A) Schematic diagram of the TSDTH strategy. (B) First stage (30 min)-phosphoproteomics. Phosphoproteins enriched using the phospho-enrichment column from BMP2 treated or nontreated C2C12 cells. The enrichment of phosphoproteins was validated. Phosphorylated Smad1/5 and 8 and Erk were detected by Immunoblot using their specific phospho-antibodies. β-Actin was used as a negative control. A dot indicates the β-actin band that was not completely removed before the detection of p-Smads on the same blot. TCL, total cell lysate; PhoE; phosphoprotein enriched. (C) Second stage (at 72 h)-SILAC (Stable isotope labeling with amino acids in cell culture). C2C12 cells were grown in medium containing 12C6-Arg and 12C6-Lys or 13C6-Arg and 13C6-Lys medium for 6 doubling times (6 days). After 6 days, C2C12 cells were stimulated with BMP2 or incubated in basal media for 3 days. BMP2-induced transdifferentiation of C2C12 cells was confirmed by alkaline phosphatase staining. (D) Total levels of enriched phosphoproteins were checked using a phospho Ser/Thr/Tyr antibody (P-S/T/Y) or the phosphotyrosine antibody p-Tyr-100. (E) After SDS-PAGE analysis, pro-Q diamond staining (left panel) was followed by detection with an image analyzer (middle panel). The Pro-Q-detected phosphoproteins are shown as black bands, and the Coomassie-stained bands are shown in blue (right panel). Dual marker (Bio-Rad) was used as a loading marker and as a positive (large arrow) and negative control (small arrow) for phosphoproteins. (F) SILAC labeling efficiency test; error ratio distribution for labeling efficiency. After tandem mass spectra were interpreted by SEQEST using DTAselect 2, ion chromatograms were quantified using the Census program. The graph shows a plot of the error ratio and the number of peptides. Ion chromatograms were quantified by the Census program. A total of 2687 out of 3261 peptides were quantified, indicating a labeling efficiency of 82.4%.

cells. Several proteins including PI3KR1-phosphoinositide-3kinase, regulatory subunit 1, MAPK14-mitogen-activated protein kinase 14 (p38 kinase), and Src homology 2 group were newly identified as being involved in BMP2 signaling. In total, the number of substrates of potential kinases was increased by approximately 1.6-fold upon BMP2-stimulation compared to the number in control cells. Interestingly, the potential activities of CSNK2B, casein kinase 2b; CDC2, cyclin-dependent kinase 2; CDK5, cyclin-dependent kinse 5; MAPK3, mitogenactivated protein kinase 3; PRKACG, protein kinase, cAMPdependent, catalytic (gamma); AKT1, v-akt murine thymoma viral oncogene homologue 1; CSNK1G2, casein kinase 1,

gamma 2; and PRKCE, protein kinase Cε were maintained in response to BMP2 stimulation, although the expressions of their substrates were changed (Figure 3B and Supplementary Table 2). These results show that during BMP2-induced transdifferentiation of C2C12 cells the phosphorylation status of many proteins is controlled by several specific kinase pathways as well as those that are normally activated during proliferation. Statistical Analysis of Two-Stage Profiles of Identified Proteins. For stringent identification of BMP2-selective phosphoproteins, we performed stepwise subtractive analysis (Figure 4A). At first, proteins from the entire experimental set, including 956 proteins from 3 control sets and 1073 proteins Journal of Proteome Research • Vol. 8, No. 10, 2009 4445

research articles

Kim et al.

Figure 2. Phosphorylation site analysis of stage one phosphoproteins. Mass spectra of phosphopeptides from identified proteins. (A) Phosphorylation of Dock7 on Ser-80 was detected as a unique site in control cells. (B) Phosphorylation of Bat2 on Thr-609 was specific to BMP2-treated cells. (C) Phosphorylation of Abcf1 on Ser-107 was detected in control and BMP2-treated cells. (D) Ser-126 phosphorylation of Abcf1 was specific to BMP2-treated cells. S+80 (*) or T+80 (*) represent the addition of one phosphate group (80 Da) to the indicated serine or threonine residue, respectively.

from 3 BMP2 sets, were filtered according to orthogonal filtering criteria (greater than 95.0% peptide probability and protein probability contained at least 1 identified peptide). Next, we used an in-house informatics tool, ProtAn X, for grouping and comparative semiquantitative analysis of uniquely identified proteins (Figure 4A). From this analysis, 401 BMP2selective phosphoproteins and 264 phosphoproteins with a 1.5fold increase in the number of peptides were selected from the control (Figure 4A). Finally, 375 phosphoproteins were identified as BMP2-selective by additive subtraction from a TGFβ1 set of phosphoproteins with a low stringency criteria of g1 peptide (Figure 4A and Supplementary Table 3). TGFβ1 is in the same superfamily as BMP2; however, its biological role is different as TGFβ1 inhibits the progression of differentiation andthematurationofmyoblasts,osteoblasts,andadipocytes.8,16,18 For the peptide and protein identification, we set the 5% of false positive (FP, 95% probability) rate. If we use 1% of FP, important information to an identification of phosphorylation sites could be missed due to intrinsic feature of phosphopeptides that have been shown to a relatively lower score than normal peptide in mass analysis. Instead of using 1% FP, we therefore applied additional subtractive analysis with TGFβ1 set (identified with one peptide but more lower stringency 4446

Journal of Proteome Research • Vol. 8, No. 10, 2009

criteria (80%) than the case of Control or BMP2 set) to first subtracted BMP2 data for more stringent selection of data. The other reason we used 5% FP rate for the protein identification in the beginning is because false positives would be removed in the subsequent analyses of functional enrichment analysis and network modeling. SILAC analysis quantified 433 out of 437 uniquely identified proteins (Figure 4B). Through mass spectra analysis, we not only confirmed the efficient incorporation of 13C6-Arg and 13C6Lys, but also established that the quantified peptide ratios were consistent with their respective protein levels (see detail in Figure 7). Of these 433 proteins, 21 proteins were increased more than 1.5-fold in BMP2-stimulated cells, including wellknown osteoblast differentiation related-gene, Thrombospondin 1 (Thbs1, increased 5.26-fold);46,47 second highly upregulated is an Aspartyl/asparaginyl beta-hydroxylase (Asph) (Supplementary Table 4), of which knockout mouse showed facial dysmorphology and defect in hard palate formation.48 Classification and Reconstruction of BMP2-Selective Phosphoproteins and Up-Regulated Proteins. Generally, the phosphorylation of proteins plays important roles in cellular signaling. We divided the identified BMP2-selective phosphoproteins into two groups: enzymes and substrates. From a total

Comprehensive Proteome Analysis of Osteoblast Trandifferentiation

research articles

Figure 3. Diagram depicting a prediction of the distribution of potential kinases that alter phosphorylation sites in response to BMP2 stimulation in C2C12 cells. (A) The diagram represents the comparison between identified phosphorylation sites in the first stage analysis. A total of 259 phosphorylation sites were identified including 98 control, 52 common, and 109 BMP2-unique sites. (B) Proposed early BMP2 signaling event. The proposed network change reflects identified phosphorylation site analysis data. Graphical visualization of the interaction mode among phosphorylation sites identified proteins and their potential kinases. Proteins in gray ovals or black ovals indicate substrates (proteins with identified phosphorylation sites) or potential kinases, respectively. Nodes indicate connection of substrates and their potential kinases.

of 375 BMP2-selective phosphoproteins, 81 (21.7%) were identified as enzymes, and these were further classified according to their biological function (Figure 5A). The major biological functions of the proteins identified include 20 (25%) energy metabolism related proteins, 18 (22%) protein metabolism related, 16 (20%) RNA regulation related, 14 (17%) kinases, 9 (11%) ubiquitination related, 7 (9%) phosphatases, and 4 (5%) DNA regulation related (Figure 5A). SILAC analysis classified 40 up-regulated proteins according to their molecular function (Figure 5B, left panel) or subcellular localization (Figure 5B, right panel). In this analysis, 9 structural proteins (22%) such as Actr2 (Actin-like protein 2), Twf1 (Twinfilin, actin-binding protein, homologue 1), and Thbs1 (Thrombospondin 1) were identified. Moreover, their average expression level was about 2.23-fold higher in BMP2 stimulated C2C12 cells. In contrast, 11 proteins including Ptgfrn and Rnh1 were down-regulated less than 1.5-fold by BMP2 stimulation, and 13 proteins were found only in unstimulated cells (Supplementary Table 4). Deubiquitin Related Enzymes Might Be Involved in BMP2 Induced Transdifferentiation of C2C12 Cells. Like many other post-translational modifications, ubiqutination and/or deubiquitination are key factors in cell fate determination in response to extracellular signals.24,49 In first stage analysis,

several BMP2-selective ubiquitin carboxyl-terminal hydrolases were identified, including Usp4, Usp9x, Usp15, Usp16, and Uch-l3 (Supplementary Table 5). These findings suggest that BMP2-selective deubiquitination enzymes might have a positive role in BMP2-mediated signaling, through increased R-Smad stability or activity. To determine whether deubiquitinating enzymes were involved in BMP2-induced transdifferentiation, C2C12 cells were first pretreated with ubiquitin aldehyde, a potent and specific inhibitor of multiple ubiquitin hydrolases, and then stimulated with BMP2 for 3 or 72 h. In the presence of ubiquitin aldehyde, BMP2 increased the levels of p-Smad1 (Figure 6A) and decreased ALP activity (Figure 6B). Recently, it was shown that blocking Uch-l3 activity leads to an increase in the ubiquitination of the epithelial sodium channel (ENaC), which is a well-known substrate of Nedd4-1 and Nedd4-2.50,51 To determine whether Uch-l3 was involved in Smad1 ubiquitination, C2C12 cells were treated with 4, 5, 6, 7-tetrachloroindan-1, 3-dione, a Uch-l3 inhibitor, and BMP2-induced Smad1 levels were analyzed. Treatment with the Uch-l3 inhibitor decreased p-Smad1 levels and ALP activity in a dosedependent manner as compared to control (untreated) cells (Figure 6). These results suggest that deubiquitination plays Journal of Proteome Research • Vol. 8, No. 10, 2009 4447

research articles

Kim et al.

Figure 4. Schematic diagram of the proteomic approach and overall results. (A) Three paired samples were analyzed independently and then compared for the sum of the control and BMP2 sets, respectively. All of the data from the experiments were filtered using the Sorcerer and Scaffold program, followed by grouping and comparative analysis using ProtAn X. Finally, 375 BMP2 selective phosphoproteins remained after the additive subtraction of a TGFβ1 data set with a low stringent criteria of g1 peptides (B). Proteins from the second stage analysis were quantified using the Census program. A total of 433 out of 437 uniquely identified proteins were quantified. Of these, 40 proteins were found to be increased more than 1.5-fold in BMP2-stimulated cells. Otherwise, 24 proteins were downregulated less than 1.5-fold by BMP2 stimulation.

Thrombospondin-1 Is a Potential Marker for BMP2 InducedOsteoblast Transdifferentiation of C2C12 Cells. In second stage analysis, Throbospondin-1 (Thbs1) was identified as most upregulated protein by BMP2 long-term stimulation in C2C12 premyoblast cells. Figure 7A shows that an ion chromatogram coincided with the quantification results of the Census program. Western blot analysis of Thbs1 was also coincident with that of SILAC analysis (Figure 7B). Western blot of Nedd4-1 was generally correlated with SILAC data, but ATP5A was not coincident. Nineteen proteins were also detected only in heavily labeled BMP2-treated sample (Supplementary Table 4). In this confirmation assay, it seems that Thrombospondin-1 is a good biomarker for the monitoring of the transdifferentiation of C2C12 cells to osteoblast-like cells in 3 days treatment of BMP.

and 8) and deubiquitin related enzymes (Nedd4-1 and Nedd4-2); and (3) calcium signaling pathway including several Ca2+-binding proteins (Fkbp10, Hsp90b1 and Thbs1 up-regulated in BMP2 stimulated cells by SILAC), signaling molecules (Plcb3, Prkaca, Prkaca, Ttn, and Calm1), and calcium transporters (Cacna1h, Asph, Atp2a2). The two pathways other than TGFβ/BMP signaling pathway have been previously reported to be closely involved in osteogenesis.52-55 The network shows the BMP2 specific proteins (red nodes in Figure 8) and their involvement in the three pathways, suggesting their potential roles in actions of BMP2: the squared red nodes represent BMP2 specific phosphoproteins identified from the first stage analysis (short-term) while the circled red nodes represent up-regulated proteins after BMP2 treatment in the second stage analysis (longterm).

Network Analysis Reveals Association of BMP2 Specific Proteins with Transdifferentiation of C2C12 Premyoblasts. To investigate association of BMP2 specific proteins (375 BMP2 specific phosphoproteins and 40 proteins up-regulated in BMP2 stimulated cells above) with the transdifferentiation of C2C12 premyoblasts into osteoblasts, we reconstructed a biological network (Figure 8; Materials and Methods) describing several pathways closely related to the BMP2 stimulated transdifferentiation: (1) IGF signaling pathway including several signaling molecules (Ppp1ca, Calm1, Pka family), and several proteins related to fatty acid metabolism (Hadha, Acaa2, Cpt2, Me1 and Acsl4); (2) TGFβ/BMP signaling pathway including Smad molecules (Smads 1, 2, 3, 5,

The network also shows interactions among the three pathways via Nedd4-1, calmodulin (Calm1), and calcineurinNFAT-Fra-2. Importantly, Nedd4-1, one of the 40 up-regulated proteins by BMP2 stimulation, interacts with IGF-I receptor45 and Smad1, a critical molecule in TGFβ/BMP2 signaling, which suggests the potential importance of Nedd4-1 as a modulator in the activities of the two pathways. Second, calmodulin was reported to inhibit Smad2 and activates Smad1 simultaneously, which can induce the osteoblast differentiation by inhibiting calcineurin-NFAT signaling that negatively regulates Fra-2 which has an essential role in osteoblast differentiation.44,56 Finally, the network shows several BMP2 specific proteins that inhibit Smad2 (Nedd4-2) and Smad3 (Junb, Ybx1 and Tgfb1i1),

important roles in fate determination during BMP2-induced osteoblast transdifferentiation of premyoblast C2C12 cells.

4448

Journal of Proteome Research • Vol. 8, No. 10, 2009

Comprehensive Proteome Analysis of Osteoblast Trandifferentiation

research articles

Figure 5. Classification of phosphoproteins and up-regulated proteins selective to BMP2 stimulation. Pie charts of functional and/or subcellular location categories for BMP2 specific phosphoproteins (first stage) and up-regulated proteins (second stage). (A) A total of 375 BMP2 selective phosphoproteins were divided into two potential categories of enzymes and substrates. Next, the enzymes were classified according to their biological function. (B) BMP2 up-regulated proteins were classified by their molecular function (left Venn diagram) and subcellular localization (right Venn diagram).

Figure 6. Influence of deubiquitination enzymes broad spectrum and Uch-l3 specific inhibitor on the BMP2 induced transdifferentiation of C2C12 cells. (A) After pretreatment with ubiquitin aldehyde, a potent and specific inhibitor of multiple ubiquitin hydrolases, or Uch-l3 inhibitor for 1 h, BMP2 induced phospho-Smad1 levels (at 3 h) were analyzed by Western blot using phospho-Smad1, 5, and 8 specific antibodies (Ser-463, Ser-465). (B) After pretreatment with ubiquitin aldehyde or Uch-l3 inhibitor for 1 h, ALP staining was preformed to measure the activity of early osteoblast differentiation (at 3 days) induced by BMP2. Journal of Proteome Research • Vol. 8, No. 10, 2009 4449

research articles

Kim et al.

Figure 7. Quantitative second stage analysis of total protein. Quantification of the protein ratios of peptide doublets. (A) Peptide pair derived from ubiquitin-activating E1, with a ratio of 1:1, indicating that this protein is not altered by BMP2 stimulation for 72 h. (B) Peptide pair derived from prohibitin-2, which was decreased 1.45-fold in response to BMP2 stimulation. (C) Peptide pair derived from thrombospondin 1, which was strongly increased (5.26-fold) by BMP2 stimulation (upper panel). Triply charged peptide with two heavy lysine residues as a result of one trypsin miscleavage (bottom panel). Ion chromatograms were quantified using the Census program. The ratios represent heavy divided by light. (D) Immunoblot analysis of selected proteins identified by SILAC-MS. SILAC data was validated by immunoblot using anti-thrombospondin-1 (Abcam), -Nedd4-1 (BD Science), -p-Smad1 (Cell Signaling), -ATP5a (Santa Cruz), and -β-actin antibodies.

thus, negatively modulating TGFβ signaling that competes with BMP2 signaling.43,57 Collectively, network analysis thus suggests close association of BMP2 specific proteins with the transdifferentiation of C2C12 premyoblasts by showing modulation of Smad molecules via the interactions among the pathways and the BMP2 activated modulators such as Nedd4.

Discussion Understanding the Molecular Mechanism of the Transdifferentiation Process. There is increasing interest in treating organ or tissue failure with stem cell-based therapies. Indeed, mesenchymal stem cells or progenitor cells are under investigation for a number of therapeutic applications. Recently, it was suggested that transdifferentiation might be applied to cell replacement therapies.1,4,58 For example, neural stem cell differentiation into dopaminergic neurons is difficult; however, bone marrow stem cells obtained for autologous transplant from adult tissue could be transdifferentiated into dopaminergic neurons that could be used to treat Parkinson’s disease.59,60 At the protein level, the cell fate exchange is reached in response to a change in the activity or expression of specific sets of proteins. We suggest that reprogramming events at the molecular level could be divided into two major stages de4450

Journal of Proteome Research • Vol. 8, No. 10, 2009

pending on the stimulation time of the transdifferentiation signals. The first stage is ‘the remodeling time of already existing proteins by triggering’ that might occur within about 30 min to 1 h, when there is little or no de novo protein synthesis. In this period, cell fate maintenance related proteins would rapidly decrease in activity and/or stability, while the activity and/or stability of the positive regulators of transdifferentiation would be increased through conformational changes and/or post-translational modifications (PTM). In addition, since it is difficult to obtain reproducible results from nonpurified cell populations of bone marrow cultures, we selected the premyoblast C2C12 mouse cell line and studied osteoblast transdifferentiation in response to BMP2 as a model system for establishing an efficient large-scale analysis of cellular signaling. TSDTH Analysis Is a Useful Approach for Studying the Comprehensive Regulatory Proteome during Cell Signaling. Over the past few years, a number of techniques have been reported for the enrichment of phosphorylated proteins or peptides from cells or tissues.31,61-63 Subfractionation techniques have led to the identification of potential phosphorylation sites, and immunoprecipitation can be used to determine the identity of specific phosphoproteins.64,65 Phosphotyrosine (pTyr)-specific antibodies have been developed to study ty-

Comprehensive Proteome Analysis of Osteoblast Trandifferentiation

research articles

Figure 8. Reconstruction of phosphoproteins and up-regulated proteins selective to BMP2 stimulation. A biological network describing IGF and calcium signaling pathways, overrepresented by the BMP2 specific proteins in first- and second-stage analysis and TGFβ/BMP pathway as well as the interactions among the pathways. The squared red nodes represent BMP2 selective phosphoproteins, while the circled red nodes represent BMP2 up-regulated proteins.

rosine-phosphorylated proteins from cell extracts and this might be due to the capturing efficiency of the antibodies used in the enrichment process, since anti-pTyr antibodies typically have a higher affinity than other anti-phospho antibodies. However, although pTyr plays a critical role in receptormediated signaling pathways, its occurrence is very low (approximately 0.05%). The most common sites of phosphorylation are on serine (pSer, ∼90%) and threonine (pThr, ∼10%). In the current study, we used a phosphoprotein purification kit to enrich serine, threonine, and tyrosine phosphorylated proteins from cell lysates (Figure 1B,D). For the quantitative analysis of enriched phosphoproteins, a combination of SILAC and IMAC of phosphopeptides may be an alternative strategy; however, when phosphopeptide enrichment is done by IMAC, most phosphoproteins may be identified with very low sequence coverage due to very low number of phosphopeptide identification (average only 1-2 per protein) and consequently with low confidence score (because of its intrinsic labile character of phospho-group on mass analysis). The rate of general difficulties for accurate protein identification will be increased when SILAC technology is combined with IMAC enrichment because only very small number of comparison sets of phosphopeptides (those phosphorylated at the same position should be compared) can be quantified. We therefore have used phosphoprotein enrichment by phosphoprotein affinity column, 1D SDS-PAGE protein fractionation, and MS protein identification with neural loss

mode in MS3 to obtain more abundant identification of phosphoproteins. In this way, we could also obtain some of phosphopeptides which were able to be confirmed with their precursor (un- or dephosphorylated) peptides at the same time. This approach enabled us to identify a total 605 phosphopeptides from both control and BMP2 sets, of which 180 were phosphorylated on serine (69%), 67 were phosphorylated on Thr (26%), and 12 were phosphorylated on Tyr (5%) (Supplementary Table 1). In 375 BMP2 specific phosphoproteins set, 259 phosphorylation sites were identified, and this data, in combination with in silico analysis, provided us with a more detailed view of BMP2-mediated signaling (Figure 3B). Many of the 375 BMP2-specific phosphoproteins identified in the current study had not previously been identified as BMP2 targets, with the exception of Cdk2, JunB, and HDAC1.37,66,67 These novel findings are therefore likely to have implications for the design of clinical approaches that target BMP2-specific signaling cascades. In second stage analysis, changes in the expression levels of existing proteins and the expression of new proteins involved in the transdifferentiation process were analyzed. We used stable isotope labeling with amino acids in cell culture (SILAC), combined with mass-spectrometry (MS) to identify quantitative changes in these proteins. SILAC-MS is a useful method for the quantitative analysis of changes in protein levels in response to growth factor stimulation in cell culture systems.68,69 In the current study, we used a double labeling system of heavy Journal of Proteome Research • Vol. 8, No. 10, 2009 4451

research articles 13

13

C6-Arg and C6-Lys, and successfully quantified 433 proteins with a peptide efficiency of 82.40% (Figure 1F and 4B). Approximately 56% (18 of 32) of the proteins, excluding faintly or heavily labeled peptides of the proteins from a second data set, which were derived from the same protein sample but independently prepared, matched the results of the first data set (Supplementary Table 4). The levels of filaments and extracellular matrix-related proteins were increased by BMP2, suggesting that cytoskeletal remodeling and the formation of new extracellular matrix components precede the transdifferentiation of premyoblast C2C12 cells into osteoblast-like cells. Our results demonstrate that SILAC-MS is a useful method for understanding the molecular mechanism of transdifferentiation. Key Regulators of TGFβ/BMP Signaling Predicted from Network Analysis of BMP2 Specific Proteins. To further understand the potential roles of these BMP2-selective proteins in TGFβ/BMP signaling, we reconstructed a biological network describing the two overrepresented pathways and TGFβ/BMP pathways and their interactions (Figure 8). The network shows the BMP2 specific proteins (red nodes in Figure 8) and their involvement in the three pathways, suggesting their potential roles in each pathway: the squared red nodes represent BMP2selective phosphoproteins identified from the first stage analysis (short-term), while the circled red nodes represent upregulated proteins after BMP2 treatment in the second stage analysis (long-term). The network also shows interactions (or cross-talks) among these three pathways via Nedd4-1, calmodulin (Calm1), and calcineurin-NFAT-Fra-2. Most importantly, Nedd4-1, one of the 40 up-regulated proteins by BMP2 stimulation, interacts with IGF-I receptor45 and Smad1, a critical molecule in TGFβ/BMP2 signaling, which suggests the potential importance of Nedd4-1 as a modulator in the activities of the two pathways. Second, calmodulin was reported to inhibit Smad2 and activate Smad1 simultaneously, which can induce the osteoblast differentiation by inhibiting calcineurinNFAT signaling that negatively regulates Fra-2 which has an essential role in osteoblast differentiation.44,56 The network further shows several BMP2-selective proteins that inhibit Smad2 (Nedd4-2) and Smad3 (Junb, Ybx1 and Tgfb1i1), thus, negatively modulating TGFβ signaling that competes with BMP2 signaling.43,57 Collectively, these findings from network analysis suggest that TGFβ/BMP2 signaling associated with osteoblast differentiation might be regulated by the activation/ inhibition of Smad molecules that are being modulated by the interactions among the pathways and the modulators such as Nedd4 shown in the network. We have shown that the levels of phosphorylated-Smad1 could be regulated in part by deubiquitination enzymes acting on BMP2 (Figure 6A). Recently, evidence of the importance of deubiquitination enzymes as target molecules in various biological processes has been reported.70-73 These results suggest that the inactivation of Nedd4 and activation of deubiquitination enzymes such as Uchl-3 might improve BMP2-induced osteoblast transdifferentiation (Figures 6 and 8). Collectively, our ‘Two-Stage Double-Technique Hybrid (TSDTH)’ identification strategy is useful to understand the dynamic molecular mechanism of cellular processes, and these enable us to obtain new insights of BMP/TGFβ signaling, and specific ubiquitination-deubiquitination enzymes that are involved in BMP2 induced transdifferentiation of C2C12 cells. 4452

Journal of Proteome Research • Vol. 8, No. 10, 2009

Kim et al.

Acknowledgment. This work was supported by the Korea Science and Engineering Foundation (KOSEF) grant funded by the Korea Government (MOST) (Grant No. M10641000106-07N4100-10610 and R13-2008-009-01002), and by the Korea Research Foundation Grant funded by the Korean Government (MOEHRD) (KRF-2007-313-E00493). Supporting Information Available: Supplementary Table 1, summary of identified phosphorylation sites and their potential kinases in first stage analysis. Supplementary Table 2, potential protein kinases for the proteins with identified phoshporylation sites. Supplementary Table 3, list of BMP2selective phosphoproteins (increased g1.5 fold). Supplementary Table 4, list of up- or down-regulated proteins in response to BMP2 (second stage of 72 h stimulation); Second set data are added to the result of first set data. Supplementary Table 5, selected list of decreased and increased ubiquitination- or proteasome-related phosphoproteins after BMP2 stimulation. This material is available free of charge via the Internet at http://pubs.acs.org. References (1) Slack, J. M. Metaplasia and transdifferentiation: from pure biology to the clinic. Nat. Rev. Mol. Cell Biol. 2007, 8 (5), 369–78. (2) Orkin, S. H.; Zon, L. I. Hematopoiesis and stem cells: plasticity versus developmental heterogeneity. Nat. Immunol. 2002, 3 (4), 323–8. (3) Phinney, D. G.; Prockop, D. J. Concise Review: Mesenchymal Stem/ Multi-Potent Stromal Cells (MSCs): the state of transdifferentiation and modes of tissue repairscurrent views. Stem Cells 2007, 25 (11), 2896–2902. (4) Batts, S. A.; Raphael, Y. Transdifferentiation and its applicability for inner ear therapy. Hear. Res. 2007, 227 (1-2), 41–7. (5) Urist, M. R. Bone: formation by autoinduction. Science 1965, 150 (698), 893–9. (6) Chen, D.; Zhao, M.; Mundy, G. R. Bone morphogenetic proteins. Growth Factors 2004, 22 (4), 233–41. (7) Imamura, T.; Maeda, S.; Hayashi, M. [Regulation of BMP signaling]. Clin. Calcium 2006, 16 (5), 738–44. (8) Nishimura, R.; Hata, K.; Ikeda, F.; Matsubara, T.; Yamashita, K.; Ichida, F.; Yoneda, T. The role of Smads in BMP signaling. Front. Biosci. 2003, 8, s275–84. (9) Kim, Y. J.; Lee, M. H.; Wozney, J. M.; Cho, J. Y.; Ryoo, H. M. Bone morphogenetic protein-2-induced alkaline phosphatase expression is stimulated by Dlx5 and repressed by Msx2. J. Biol. Chem. 2004, 279 (49), 50773–80. (10) Kitisin, K.; Saha, T.; Blake, T.; Golestaneh, N.; Deng, M.; Kim, C.; Tang, Y.; Shetty, K.; Mishra, B.; Mishra, L. Tgf-Beta signaling in development. Sci STKE 2007, 2007 (399), cm1. (11) Roussa, E.; Wiehle, M.; Dunker, N.; Becker-Katins, S.; Oehlke, O.; Krieglstein, K. Transforming growth factor beta is required for differentiation of mouse mesencephalic progenitors into dopaminergic neurons in vitro and in vivo: ectopic induction in dorsal mesencephalon. Stem Cells 2006, 24 (9), 2120–9. (12) Gunther, T.; Schule, R. Fat or bone? A non-canonical decision. Nat. Cell Biol. 2007, 9 (11), 1229–1231. (13) Khosla, S.; Westendorf, J. J.; Oursler, M. J. Building bone to reverse osteoporosis and repair fractures. J. Clin. Invest. 2008, 118 (2), 421–8. (14) Zachos, T.; Diggs, A.; Weisbrode, S.; Bartlett, J.; Bertone, A. Mesenchymal stem cell-mediated gene delivery of bone morphogenetic protein-2 in an articular fracture model. Mol. Ther. 2007, 15 (8), 1543–50. (15) Heldin, C. H.; Miyazono, K.; ten Dijke, P. TGF-beta signalling from cell membrane to nucleus through SMAD proteins. Nature 1997, 390 (6659), 465–71. (16) Miyazawa, K.; Shinozaki, M.; Hara, T.; Furuya, T.; Miyazono, K. Two major Smad pathways in TGF-beta superfamily signalling. Genes Cells 2002, 7 (12), 1191–204. (17) Miyazono, K.; Maeda, S.; Imamura, T. BMP receptor signaling: transcriptional targets, regulation of signals, and signaling crosstalk. Cytokine Growth Factor Rev. 2005, 16 (3), 251–63. (18) Massague, J.; Blain, S. W.; Lo, R. S. TGFbeta signaling in growth control, cancer, and heritable disorders. Cell 2000, 103 (2), 295– 309.

Comprehensive Proteome Analysis of Osteoblast Trandifferentiation (19) Attisano, L.; Wrana, J. L. Signal transduction by the TGF-beta superfamily. Science 2002, 296 (5573), 1646–7. (20) Asano, Y.; Ihn, H.; Yamane, K.; Kubo, M.; Tamaki, K. Impaired Smad7-Smurf-mediated negative regulation of TGF-beta signaling in scleroderma fibroblasts. J. Clin. Invest. 2004, 113 (2), 253–64. (21) Chen, C.; Matesic, L. E. The Nedd4-like family of E3 ubiquitin ligases and cancer. Cancer Metastasis Rev. 2007, 26 (3-4), 587–604. (22) Sapkota, G.; Alarcon, C.; Spagnoli, F. M.; Brivanlou, A. H.; Massague, J. Balancing BMP signaling through integrated inputs into the Smad1 linker. Mol. Cell 2007, 25 (3), 441–54. (23) Ogunjimi, A. A.; Briant, D. J.; Pece-Barbara, N.; Le Roy, C.; Di Guglielmo, G. M.; Kavsak, P.; Rasmussen, R. K.; Seet, B. T.; Sicheri, F.; Wrana, J. L. Regulation of Smurf2 ubiquitin ligase activity by anchoring the E2 to the HECT domain. Mol. Cell 2005, 19 (3), 297– 308. (24) Datto, M.; Wang, X. F. Ubiquitin-mediated degradation a mechanism for fine-tuning TGF-beta signaling. Cell 2005, 121 (1), 2–4. (25) Yamashita, M.; Ying, S. X.; Zhang, G. M.; Li, C.; Cheng, S. Y.; Deng, C. X.; Zhang, Y. E. Ubiquitin ligase Smurf1 controls osteoblast activity and bone homeostasis by targeting MEKK2 for degradation. Cell 2005, 121 (1), 101–13. (26) Lin, X.; Duan, X.; Liang, Y. Y.; Su, Y.; Wrighton, K. H.; Long, J.; Hu, M.; Davis, C. M.; Wang, J.; Brunicardi, F. C.; Shi, Y.; Chen, Y. G.; Meng, A.; Feng, X. H. PPM1A functions as a Smad phosphatase to terminate TGFbeta signaling. Cell 2006, 125 (5), 915–28. (27) Knockaert, M.; Sapkota, G.; Alarcon, C.; Massague, J.; Brivanlou, A. H. Unique players in the BMP pathway: small C-terminal domain phosphatases dephosphorylate Smad1 to attenuate BMP signaling. Proc. Natl. Acad. Sci. U.S.A. 2006, 103 (32), 11940–5. (28) Chen, H. B.; Shen, J.; Ip, Y. T.; Xu, L. Identification of phosphatases for Smad in the BMP/DPP pathway. Genes Dev. 2006, 20 (6), 648– 53. (29) Zhu, H.; Kavsak, P.; Abdollah, S.; Wrana, J. L.; Thomsen, G. H. A SMAD ubiquitin ligase targets the BMP pathway and affects embryonic pattern formation. Nature 1999, 400 (6745), 687–93. (30) Park, H. W.; Shin, J. S.; Kim, C. W. Proteome of mesenchymal stem cells. Proteomics 2007, 7 (16), 2881–94. (31) Smolka, M. B.; Albuquerque, C. P.; Chen, S. H.; Zhou, H. Proteomewide identification of in vivo targets of DNA damage checkpoint kinases. Proc. Natl. Acad. Sci. U.S.A. 2007, 104 (25), 10364–9. (32) Colland, F.; Jacq, X.; Trouplin, V.; Mougin, C.; Groizeleau, C.; Hamburger, A.; Meil, A.; Wojcik, J.; Legrain, P.; Gauthier, J. M. Functional proteomics mapping of a human signaling pathway. Genome Res. 2004, 14 (7), 1324–32. (33) Domon, B.; Aebersold, R. Mass spectrometry and protein analysis. Science 2006, 312 (5771), 212–7. (34) Gingras, A. C.; Gstaiger, M.; Raught, B.; Aebersold, R. Analysis of protein complexes using mass spectrometry. Nat. Rev. Mol. Cell Biol. 2007, 8 (8), 645–54. (35) Kocher, T.; Superti-Furga, G. Mass spectrometry-based functional proteomics: from molecular machines to protein networks. Nat. Methods 2007, 4 (10), 807–15. (36) Kim, B. G.; Kim, H. J.; Park, H. J.; Kim, Y. J.; Yoon, W. J.; Lee, S. J.; Ryoo, H. M.; Cho, J. Y. Runx2 phosphorylation induced by fibroblast growth factor-2/protein kinase C pathways. Proteomics 2006, 6 (4), 1166–74. (37) Jeon, E. J.; Lee, K. Y.; Choi, N. S.; Lee, M. H.; Kim, H. N.; Jin, Y. H.; Ryoo, H. M.; Choi, J. Y.; Yoshida, M.; Nishino, N.; Oh, B. C.; Lee, K. S.; Lee, Y. H.; Bae, S. C. Bone morphogenetic protein-2 stimulates Runx2 acetylation. J. Biol. Chem. 2006, 281 (24), 16502–11. (38) Park, H. J.; Kim, B. G.; Lee, S. J.; Heo, S. H.; Kim, J. Y.; Kwon, T. H.; Lee, E. B.; Ryoo, H. M.; Cho, J. Y. Proteomic Profiling of Endothelial cells in human lung cancer. J. Proteome Res. 2008, 7 (3), 1138–50. (39) Keller, A.; Nesvizhskii, A. I.; Kolker, E.; Aebersold, R. Empirical statistical model to estimate the accuracy of peptide identifications made by MS/MS and database search. Anal. Chem. 2002, 74 (20), 5383–92. (40) Nesvizhskii, A. I.; Keller, A.; Kolker, E.; Aebersold, R. A statistical model for identifying proteins by tandem mass spectrometry. Anal. Chem. 2003, 75 (17), 4646–58. (41) Venable, J. D.; Wohlschlegel, J.; McClatchy, D. B.; Park, S. K.; Yates, J. R., III. Relative quantification of stable isotope labeled peptides using a linear ion trap-Orbitrap hybrid mass spectrometer. Anal. Chem. 2007, 79 (8), 3056–64. (42) Park, S. K.; Venable, J. D.; Xu, T.; Yates, J. R. A quantitative analysis software tool for mass spectrometry-based proteomics. Nat Methods 2008, 5 (4), 319–22. (43) Higashi, K.; Inagaki, Y.; Fujimori, K.; Nakao, A.; Kaneko, H.; Nakatsuka, I. Interferon-gamma interferes with transforming growth factor-beta signaling through direct interaction of YB-1 with Smad3. J. Biol. Chem. 2003, 278 (44), 43470–9.

research articles

(44) McCabe, L. R.; Banerjee, C.; Kundu, R.; Harrison, R. J.; Dobner, P. R.; Stein, J. L.; Lian, J. B.; Stein, G. S. Developmental expression and activities of specific fos and jun proteins are functionally related to osteoblast maturation: role of Fra-2 and Jun D during differentiation. Endocrinology 1996, 137 (10), 4398–408. (45) Vecchione, A.; Marchese, A.; Henry, P.; Rotin, D.; Morrione, A. The Grb10/Nedd4 complex regulates ligand-induced ubiquitination and stability of the insulin-like growth factor I receptor. Mol. Cell. Biol. 2003, 23 (9), 3363–72. (46) Ueno, A.; Miwa, Y.; Miyoshi, K.; Horiguchi, T.; Inoue, H.; Ruspita, I.; Abe, K.; Yamashita, K.; Hayashi, E.; Noma, T. Constitutive expression of thrombospondin 1 in MC3T3-E1 osteoblastic cells inhibits mineralization. J. Cell Physiol. 2006, 209 (2), 322–32. (47) Sherbina, N. V.; Bornstein, P. Modulation of thrombospondin gene expression during osteoblast differentiation in MC3T3-E1 cells. Bone 1992, 13 (2), 197–201. (48) Dinchuk, J. E.; Henderson, N. L.; Burn, T. C.; Huber, R.; Ho, S. P.; Link, J.; O’Neil, K. T.; Focht, R. J.; Scully, M. S.; Hollis, J. M.; Hollis, G. F.; Friedman, P. A. Aspartyl beta -hydroxylase (Asph) and an evolutionarily conserved isoform of Asph missing the catalytic domain share exons with junctin. J. Biol. Chem. 2000, 275 (50), 39543–54. (49) Gao, M.; Karin, M. Regulating the regulators: control of protein ubiquitination and ubiquitin-like modifications by extracellular stimuli. Mol. Cell 2005, 19 (5), 581–93. (50) Fotia, A. B.; Ekberg, J.; Adams, D. J.; Cook, D. I.; Poronnik, P.; Kumar, S. Regulation of neuronal voltage-gated sodium channels by the ubiquitin-protein ligases Nedd4 and Nedd4-2. J. Biol. Chem. 2004, 279 (28), 28930–5. (51) Butterworth, M. B.; Edinger, R. S.; Ovaa, H.; Burg, D.; Johnson, J. P.; Frizzell, R. A. The deubiquitinating enzyme UCH-L3 regulates the apical membrane recycling of the epithelial sodium channel. J. Biol. Chem. 2007, 282 (52), 37885–93. (52) Canalis, E.; Gabbitas, B. Bone morphogenetic protein 2 increases insulin-like growth factor I and II transcripts and polypeptide levels in bone cell cultures. J. Bone Miner. Res. 1994, 9 (12), 1999–2005. (53) Celil, A. B.; Campbell, P. G. BMP-2 and insulin-like growth factor-I mediate Osterix (Osx) expression in human mesenchymal stem cells via the MAPK and protein kinase D signaling pathways. J. Biol. Chem. 2005, 280 (36), 31353–9. (54) Siddappa, R.; Martens, A.; Doorn, J.; Leusink, A.; Olivo, C.; Licht, R.; van Rijn, L.; Gaspar, C.; Fodde, R.; Janssen, F.; van Blitterswijk, C.; de Boer, J. cAMP/PKA pathway activation in human mesenchymal stem cells in vitro results in robust bone formation in vivo. Proc. Natl. Acad. Sci. U.S.A. 2008, 105 (20), 7281–6. (55) Ohta, Y.; Nakagawa, K.; Imai, Y.; Katagiri, T.; Koike, T.; Takaoka, K. Cyclic AMP enhances Smad-mediated BMP signaling through PKA-CREB pathway. J. Bone Miner. Metab. 2008, 26 (5), 478–84. (56) Zayzafoon, M. Calcium/calmodulin signaling controls osteoblast growth and differentiation. J. Cell. Biochem. 2006, 97 (1), 56–70. (57) Wang, H.; Song, K.; Sponseller, T. L.; Danielpour, D. Novel function of androgen receptor-associated protein 55/Hic-5 as a negative regulator of Smad3 signaling. J. Biol. Chem. 2005, 280 (7), 5154–62. (58) Satija, N. K.; Gurudutta, G. U.; Sharma, S.; Afrin, F.; Gupta, P.; Verma, Y. K.; Singh, V. K.; Tripathi, R. P. Mesenchymal stem cells: molecular targets for tissue engineering. Stem Cells Dev. 2007, 16 (1), 7–23. (59) Silani, V.; Corbo, M. Cell-replacement therapy with stem cells in neurodegenerative diseases. Curr. Neurovasc. Res. 2004, 1 (3), 283–9. (60) Dezawa, M. [Insights into auto-transplantationsthe unexpected discovery of transdifferentiation systems in bone marrow stromal cells and its application]. Brain Nerve 2007, 59 (5), 503–8. (61) Chen, Y.; Choong, L. Y.; Lin, Q.; Philp, R.; Wong, C. H.; Ang, B. K.; Tan, Y. L.; Loh, M. C.; Hew, C. L.; Shah, N.; Druker, B. J.; Chong, P. K.; Lim, Y. P. Differential expression of novel tyrosine kinase substrates during breast cancer development. Mol. Cell. Proteomics 2007, 6 (12), 2072–87. (62) de la Fuente van Bentem, S.; Hirt, H. Using phosphoproteomics to reveal signalling dynamics in plants. Trends Plant Sci. 2007, 12 (9), 404–11. (63) Wissing, J.; Jansch, L.; Nimtz, M.; Dieterich, G.; Hornberger, R.; Keri, G.; Wehland, J.; Daub, H. Proteomics analysis of protein kinases by target class-selective prefractionation and tandem mass spectrometry. Mol. Cell. Proteomics 2007, 6 (3), 537–47. (64) Kratchmarova, I.; Blagoev, B.; Haack-Sorensen, M.; Kassem, M.; Mann, M. Mechanism of divergent growth factor effects in mesenchymal stem cell differentiation. Science 2005, 308 (5727), 1472–7.

Journal of Proteome Research • Vol. 8, No. 10, 2009 4453

research articles (65) Kruger, M.; Kratchmarova, I.; Blagoev, B.; Tseng, Y. H.; Kahn, C. R.; Mann, M. Dissection of the insulin signaling pathway via quantitative phosphoproteomics. Proc. Natl. Acad. Sci. U.S.A. 2008, 105 (7), 2451–6. (66) Miyazaki, H.; Watabe, T.; Kitamura, T.; Miyazono, K. BMP signals inhibit proliferation and in vivo tumor growth of androgen-insensitive prostate carcinoma cells. Oncogene 2004, 23 (58), 9326–35. (67) Chalaux, E.; Lopez-Rovira, T.; Rosa, J. L.; Bartrons, R.; Ventura, F. JunB is involved in the inhibition of myogenic differentiation by bone morphogenetic protein-2. J. Biol. Chem. 1998, 273 (1), 537–43. (68) Wang, X.; Huang, L. Identifying dynamic interactors of protein complexes by quantitative mass spectrometry. Mol. Cell. Proteomics 2008, 7 (1), 46–57. (69) Waanders, L. F.; Hanke, S.; Mann, M. Top-down quantitation and characterization of SILAC-labeled proteins. J. Am. Soc. Mass Spectrom. 2007, 18 (11), 2058–64.

4454

Journal of Proteome Research • Vol. 8, No. 10, 2009

Kim et al. (70) Lindner, H. A. Deubiquitination in virus infection. Virology 2007, 362 (2), 245–56. (71) Chen, Z. J. Ubiquitin signalling in the NF-kappaB pathway. Nat. Cell Biol. 2005, 7 (8), 758–65. (72) Sridhar, V. V.; Kapoor, A.; Zhang, K.; Zhu, J.; Zhou, T.; Hasegawa, P. M.; Bressan, R. A.; Zhu, J. K. Control of DNA methylation and heterochromatic silencing by histone H2B deubiquitination. Nature 2007, 447 (7145), 735–8. (73) Oestergaard, V. H.; Langevin, F.; Kuiken, H. J.; Pace, P.; Niedzwiedz, W.; Simpson, L. J.; Ohzeki, M.; Takata, M.; Sale, J. E.; Patel, K. J. Deubiquitination of FANCD2 is required for DNA crosslink repair. Mol. Cell 2007, 28 (5), 798–809.

PR900231A