Author contributions: F.X.D., N.S., and P.J.M.: conception and design, collection and/or assembly of data, data analysis and interpretation, and manuscript writing; M.B.D., J.J.W., and Y.S.: provision of study material and final approval of manuscript.
Disclosure of potential conflicts of interest is found at the end of this article.
first published online in STEM CELLSEXPRESS March 26, 2013.
The identification of the molecular mechanisms controlling the degradation of regulatory proteins in mesenchymal stromal cells (MSC) may provide clues to promote MSC osteogenic differentiation and bone regeneration. Ubiquitin ligase-dependent degradation of proteins is an important process governing cell fate. In this study, we investigated the role of the E3 ubiquitin ligase c-Cbl in MSC osteoblast differentiation and identified the mechanisms involved in this effect. Using distinct shRNA targeting c-Cbl, we showed that c-Cbl silencing promotes osteoblast differentiation in murine and human MSC, as demonstrated by increased alkaline phosphatase activity, expression of phenotypic osteoblast marker genes (RUNX2, ALP, type 1 collagen), and matrix mineralization in vitro. Coimmunoprecipitation analyses showed that c-Cbl interacts with the transcription factor STAT5, and that STAT5 forms a complex with RUNX2, a master transcription factor controlling osteoblastogenesis. Silencing c-Cbl decreased c-Cbl-mediated STAT5 ubiquitination, increased STAT5 protein level and phosphorylation, and enhanced STAT5 and RUNX2 transcriptional activity. The expression of insulin like growth factor-1 (IGF-1), a target gene of STAT5, was increased by c-Cbl silencing in MSC and in bone marrow stromal cells isolated from c-Cbl deficient mice, suggesting that IGF-1 contributes to osteoblast differentiation induced by c-Cbl silencing in MSC. Consistent with these findings, pharmacological inhibition of STAT5 activity, or neutralization of IGF-1 activity, abrogated the positive effect of c-Cbl knockdown on MSC osteogenic differentiation. Taken together, the data provide a novel functional mechanism by which the ubiquitin ligase c-Cbl regulates the osteoblastic differentiation program in mesenchymal cells by controlling Cbl-mediated STAT5 degradation and activity. STEM Cells2013;31:1340–1349
Human and murine mesenchymal stromal cells (MSC) derived from the bone marrow stroma have the ability to differentiate into chondroblasts, osteoblasts, or adipocytes under appropriate stimulation [1–4]. The differentiation potential of these cells makes them an appropriate source for cell engineering and bone matrix regeneration [5–8]. The osteogenic differentiation of MSC is characterized by the expression of timely expressed genes such as RUNX2, alkaline phosphatase (ALP), and type I collagen (COL1A1) followed by extracellular matrix synthesis and mineralization [9, 10]. The transcription factor RUNX2 plays an essential role in osteoblastogenesis by regulating multiple genes involved in the control of the osteoblastic lineage [11–13]. Factors that promote MSC differentiation into mature osteoblasts, such as bone morphogenetic protein-2  and Wnt proteins  or fibroblast growth factor receptor signaling  act in part by regulating RUNX2 expression or transcriptional activity. RUNX2 gene transfer in murine MSC enhances the osteogenic activity of bone marrow mesenchymal cells [17, 18], indicating that this transcription factor may be a potential target for promoting osteoblast differentiation in MSC.
Up to now, most investigators have focused on how targeting genes rather than proteins to promote osteogenic differentiation in MSC. Finding ways to target essential proteins regulating the osteoblastic differentiation of MSC may be an important issue for increasing osteoblastogenesis. Here we focused on the intracellular protein degradation system because it is a key mechanism involved in the regulation of various cellular processes including cell proliferation, differentiation, and survival . Proteasome degradation involves a cascade of events requiring ubiquitin ligases that target proteins for ubiquitination and proteasome degradation [20–22]. Previous studies reported a positive effect of proteasome inhibition on osteoblast differentiation through prevention of proteasomal degradation of β-catenin , ATF4  and RUNX2 by the E3 ubiquitin ligases Smurf-1 and Schnurri-3 [25–27], which supports an important role of E3 ubiquitin ligases in the control of bone metabolism . The E3 ubiquitin ligase c-Cbl (Casitas B-lineage lymphoma) is a multifunctional adaptor protein  involved in the regulation of signaling proteins [29, 30]. The c-Cbl function is mediated by the RING finger domain which allows the recruitment of ubiquitin-conjugated enzymes (E2) and is responsible for ubiquitin ligase activity and by the N-terminal domain which interacts with the phosphotyrosine residues of multiple protein tyrosine kinases, resulting in ubiquitination and proteasome degradation of targeted proteins [31, 32]. We and others have previously shown that this E3 ubiquitin ligase controls the ubiquitination and proteasome degradation of several important signaling proteins in osteoblasts [33–38]. However, the role of c-Cbl in the control of mesenchymal osteoblast progenitor cell differentiation remains unknown.
In this study, we determined the functional role of the E3 ubiquitin ligase c-Cbl in osteoblast differentiation of murine and human MSC. We focused our research on the proteins that are ubiquitinylated by c-Cbl, one of the most important mechanism by which proteins are controlled [19–21]. We provide here the first evidence that c-Cbl silencing promotes MSC osteoblast differentiation by downregulating signal transducers and activators of transcription 5 (STAT5) ubiquitination, resulting in increased STAT5-RUNX2 interaction and osteoblast gene expression. The data reveal a novel mechanism by which STAT5 activity is controlled by c-Cbl, which could be used for enhancing osteoblast differentiation in MSC for promoting bone regeneration.
MATERIALS AND METHODS
Murine pluripotent mesenchymal C3H10T1/2 cells were obtained from the ATCC (Rockville, MD, http://www.atcc.org). Additionally, we used human clonal pluripotent bone marrow stromal cells immunoselected with the murine IgM monoclonal antibody STRO-1, a cell surface antigen expressed by stromal elements in human bone marrow . The selected cells (F/STRO-1+A) show osteogenic differentiation potential under appropriate stimulation [2, 40]. Cells were routinely cultured in Dulbecco's modified Eagle's medium (DMEM, Invitrogen, Paisley, Scotland, http://www.invitrogen.com) supplemented with 10% heat inactivated fetal calf serum, 1% L-glutamine, and penicillin/streptomycin (10,000 U/mL and 10,000 μg/mL, respectively) at 37°C in humidified atmosphere containing 5% CO2 in air. Culture media were changed every 2 days.
Constructs, RNA Interference, and Reagents
The murine shc-Cbl plasmids (sh-c-Cbl1 and sh-cCbl2) were from the National RNAi Core Facility Platform of Taiwan. The human short hairpin (sh) shc-Cbl (sh-c-Cbl3 and sh-cCbl4) and sh-Control plasmid were from SantaCruz Biotechnology (SantaCruz, CA, http://www.scbt.com). Lentiviral production and virion particles containing the pLKO1-sh-c-Cbl or the control empty vector pLKO1-puro were produced as described . Cells were transduced at 50% confluence with lentiviral particles in the presence of polybrene (5–10 μg/mL) for 48 hours. Lentiviral transduction efficacy was evaluated by green fluorescent protein level under fluorescence microscopy . A two-step polymerase chain reaction (PCR) strategy was applied to generate a RUNX2 mutant using the plasmid pShuttle-CMV-RUNX2 as the initial template. Before pShuttle-CMV-RUNX2 M3 being constructed, pShuttle-CMV-RUNX2K344/345A was generated, and mutations at lysine 219 and 224 were introduced using pShuttle-CMV-RUNX2K344/345A as a template. The sense primer 5′-CCACCATGCGT ATTCCTGTAGATC-3′ was paired with different internal antisense primers to amplify the N-terminal regions with point mutations at specific residues, and antisense primer 5′-TCAAT ATGGTCGCCAAAC-3′ was paired with different internal sense primers to amplify the C-terminal region with specific point mutations. The full-length genes encoding various RUNX2 mutants were obtained using a second-step PCR using the sense primer 5′-AGATCTCCACCATGCGTATTCCTGTAGATC-3′ (the BglII site is underlined) and the antisense primer 5′-CTCGAGTCAATATGGTCGCCAAAC-3′ (the XhoI site is underlined). The RUNX2 mutant fragments obtained from the PCR amplification were digested with BglII and XhoI and subcloned into the BglII and XhoI sites of pShuttle-CMV (Stratagene, WA, http://www.stratagene.com) to obtain the pShuttle-CMV-RUNX2 M3 mutant clones. The internal primers used to generate the RUNX2 M3 mutant are listed in Supporting Information Table 1. The pShuttle-CMV-RUNX2 M3 mutant construct was verified by DNA sequencing before being transformed into BJ5183 cells to generate the recombinant adenoviral genome carrying the RUNX2 M3 mutant. In some experiments, cells were treated with the STAT5 inhibitor (Sc-355979; 75 μM) that inhibits the function of the SH2 domain of STAT5 (Santa Cruz Biotechnology), or the solvent (dimethylsulfoxyde (DMSO), 0.1%), or the insulin like growth factor-1 (IGF-1) neutralizing antibody (R&D Systems, Lille, France, http://www.rndsystems.com) used at the final concentration of 12 μg/mL.
Proliferation and Differentiation Assays
Cells were seeded at 2.5 × 104 cells per cm2, and cell proliferation was evaluated by cell counting and incorporation of 5-bromo-2-deoxyuridine (BrdU), a thymidine analog used as reagent for cell proliferation assays, according to the manufacturer's recommendations (GE Healthcare, Amersham, U.K., http://www.gehealthcare.com). ALP activity was determined using Sigma FAST kit according to the manufacturer's recommendations (Sigma, St Louis, MI, http://www.sigmaaldrich.com). For in vitro osteogenic assay, cell culture medium was supplemented with 50 μg/mL ascorbic acid and 3 mM inorganic phosphate to allow matrix synthesis and mineralization. At the indicated time point, cells were fixed in 70% ethanol at 4°C. Matrix mineralization was evaluated by alizarin red staining, microphotographed using an Olympus microscope and quantified as described .
RNA Extraction and Quantitative Reverse Transcription PCR Analysis
Total RNA were isolated using Trizol reagent (Eurobio, Les Ulis, France, http://www.eurobio.fr) according to the manufacturer's instructions. Three microgram of total RNA from each samples were reverse transcribed with 1× RT buffer, 1 mM dNTP mix, 1× random primers, and 50 U multiscribe reverse transcriptase (Invitrogen, St Aubin, France, http://www.invitrogen.com) in a total volume of 20 μL at 37°C for 2 hours. Relative mRNA levels were evaluated by quantitative PCR (LightCycler; Roche Applied Science, Indianapolis, OH, http://www.roche-applied-science.com) using a SYBR Green PCR kit (ABGen, Courtaboeuf, France, http://www.thermoscientificbio.com) and specific primers . The primer sequences used are shown in Supporting Information Table 2. Other primers were as described . Signals were normalized to GAPDH as internal control. The relative amount of RNA was calculated by the 2−ΔΔCt method.
Western Blot and Immunoprecipitation Analyses
Total cell lysates were prepared as described . Proteins (30 μg) were resolved on 4%–12% SDS-PAGE and transferred onto nitrocellulose membranes (Millipore, Bedford, http://www.millipore.com). Filters were incubated for 2 hours in 50 mM Tris/HCl (pH 7.4), 150 mM NaCl, 0.1% (v/v) Tween-20, 0.5% (w/v) bovine serum albumin (TBST/BSA), then overnight at 4°C on a shaker with the following primary antibodies (1/500–1/1,000 in TBST/BSA): anti-mouse c-Cbl, anti-Cbl-b, anti-STAT5, anti-RUNX2, anti-ubiquitin (all from Santa Cruz), human anti-c-Cbl, anti-phospho-STAT5, anti-STAT3 and anti-phospho-STAT3 (1/1,000) (all from Abcam, Cambridge, U.K., http://www.abcam. com), and anti-β-actin (Sigma). Membranes were washed twice with TBST and incubated for 2 hours with the appropriate horseradish peroxydase (HRP)-conjugated secondary antibody (1/10,000–1/20,000 in TBST/BSA). After final washes, the signals were visualized with enhanced chemiluminescence Western blotting detection reagent (ECL, Amersham Biosciences, Piscataway, NJ, http://www.amersham.com) and autoradiographic film (X-OMAT-AR, Eastman Kodak Company, Rochester, NY, http://www.kodak.com). Densitometric analysis using ImageQuant software was performed following digital scanning (Agfa) on blots from at least three separate experiments. Representative images of immunoblots are shown. For immunoprecipitation (IP) analysis, cell lysates were prepared as for Western blot (WB) analysis, and aliquots of total protein (250 μg) were incubated overnight under weak agitation at 4°C with 2 μg specific antibody and 20 μL Dynabeads protein G (Invitrogen). Components of the bound immune complex (both antigen and antibody) were eluted from the Dynabeads and analyzed by SDS-PAGE according to the manufacturer's recommendations.
To test RUNX2 transcriptional activity, F/STRO1+A cells were seeded in 24-well plates and then cotransfected with 0.3 μg per well of 6XOSE-Luc (a Renilla luciferase expression plasmid containing six RUNX2 binding sites), phRL-SV40 (a Renilla luciferase expression plasmid without RUNX2 binding site), pShuttle-CMV-RUNX2, or pShuttle-CMV-RUNX2 M3 mutant and pSV-β-Galactosidase as a transfection control. Forty-eight hours after transfection, Renilla luciferase and β-galactosidase activities were determined sequentially using Luciferase Reporter Assay System (Promega, Madison, WI, http://www.promega.com) or β-galactosidase gene reporter assay kit (Roche, Meylan, France, http://www.roche.fr). Luciferase activity was normalized with β-galactosidase activity to avoid transfection variability and with phRL-SV40 to normalize luciferase background. The STAT5 reporter assay was conducted in F/STRO1+A cells treated or not with the STAT5 inhibitor for 24 hours, using the pGL4.52(luc2P/STAT5 RE/Hygro vector according to the manufacturer's recommendations (Promega). All results were expressed as relative luciferase units.
Isolation of Bone Marrow MSC from C-Cbl−/− Mice
Long bones from 7-weeks-old c-Cbl knockout mice  were isolated, fresh bone marrow stroma cells were collected by flushing, and RNA was extracted for reverse transcription PCR (RT-PCR) analysis as previously described .
The data are the mean ± SD and are representative of at least three experiments. The data were analyzed by Student's t test and a minimal level of p < .05 was considered significant.
C-Cbl Silencing Promotes Osteoblast Proliferation and Differentiation in MSC
We first investigated the effect of silencing c-Cbl on mesenchymal progenitor cell proliferation and osteoblast differentiation and the mechanisms involved in this effect. To this goal, we used two murine shRNA sequences (sh-c-Cbl1, sh-c-Cbl2) targeting distinct c-Cbl domains that interact with RTK or with ubiquitin (Fig. 1 A). WB analysis showed that c-Cbl protein level was decreased by the two shRNA in murine C3H10T1/2 multipotent mesenchymal cells, thus validating the assay (Supporting Information Fig. 1A, 1B). Cbl-b protein level was increased by about 30% in sh-c-Cbl-transduced murine MSC, representing possibly a compensatory effect in these cells (Supporting Information Fig. 1A, 1B). A similar effect of two human shRNA (sh-c-Cbl3, sh-c-Cbl4) targeting c-Cbl was found in human clonal multipotent bone marrow-derived F/STRO-1+A stromal cells (Fig. 1B, 1C). We found that c-Cbl silencing using these efficient shRNAs slightly increased cell number in murine C3H10T1/2 cells and human F/STRO-1+A cells (Fig. 1D, 1E). This effect was related to increased cell proliferation as determined by the BrdU assay (Fig. 1F, 1G). These results indicate that c-Cbl silencing results in a slight increase in cell proliferation in both murine and human MSC. We then investigated the effect of c-Cbl silencing on MSC osteoblast differentiation. We showed that c-Cbl silencing increased ALP staining (Fig. 2A, 2B) and activity (Fig. 2C, 2D) in both murine C3H10T1/2 and human F/STRO-1+A mesenchymal cells, reflecting an early induction of osteoblast differentiation. Consistent with this effect, c-Cbl silencing increased the expression of the phenotypic osteoblast markers RUNX2, ALP, and type 1 collagen in both cell types (Fig. 2E, 2F). This positive effect on osteoblast differentiation markers translated into increased matrix mineralization in vitro, as shown by the increased alizarin red staining of the mineralized matrix (Fig. 2G, 2H). The effect of the sh-c-Cbl used on osteogenic differentiation markers and matrix mineralization appears to broadly correlate with the amplitude of c-Cbl silencing. Specifically, the best effect on matrix mineralization which characterizes the final osteogenic program was observed with the sh-c-Cbl which had the larger inhibitory effect on c-Cbl expression in C3H10T1/2 cells (Fig. 2E, 2G). Also, when the two shRNA had a similar effect on c-Cbl levels, the stimulatory effect on osteogenic differentiation was similar, as observed in F/STRO-1+A mesenchymal cells (Fig. 2F, 2H). However, the effect of c-Cbl silencing on osteoblast markers was not always strictly related to c-Cbl levels, most probably because c-Cbl regulates the degradation of many proteins such as receptor tyrosine kinases [38, 44] that may control osteoblast differentiation directly or indirectly.
C-Cbl Silencing Increases STAT5 Interaction with RUNX2 in MSC
We next investigated the mechanisms that are involved in the increased osteoblast differentiation induced by c-Cbl silencing in MSC. We postulated that c-Cbl may modulate RUNX2 transcriptional activity, an essential mechanism regulating osteoblastogenesis . We found that RUNX2 mRNA level was significantly increased in c-Cbl-silenced C3H10T1/2 cells, although it was less increased in c-Cbl-silenced F/STRO1+A cells (Fig. 2E, 2F), possibly because of distinct mechanisms/signaling pathways involved in human compared to murine MSC differentiation. STAT5, a member of the STATs family , was found to interact with RUNX2 in lymphocytes . In this cellular context, STAT5 and RUNX proteins mutually inhibit their transcriptional activity . Whether STAT5 and RUNX2 may interact in MSC to regulate osteoblast differentiation was unknown. We therefore investigated STAT5 interaction with RUNX2 in human F/STRO1+A cells used as a model. Our IP-WB analyses showed that RUNX2 interacts with STAT5 and that STAT5 interacts with c-Cbl in F/STRO+A cells (Fig. 3A, 3B), suggesting that these proteins form a complex in MSC. We then determined by IP-WB analysis whether c-Cbl silencing may impact STAT5–RUNX2 interaction in F/STRO1+A cells. We found that c-Cbl silencing decreased STAT5 associated with c-Cbl (Fig. 3B). Quantification of the immunoblots indicates that the c-Cbl level associated with STAT5 decreased by 50% and 70% with the sh-c-Cbl3 and sh-c-Cbl4, respectively, compared to sh-Control. c-Cbl silencing concomitantly increased STAT5 associated with RUNX2 (Fig. 3A). Quantification of the immunoblots confirmed that STAT5 associated with RUNX2 was increased by c-Cbl silencing (Fig. 3A), suggesting that STAT5–RUNX2 interaction is controlled by c-Cbl. We then hypothesized that c-Cbl may interact with STAT5 ubiquitination to modulate its protein level. To investigate whether c-Cbl may regulate STAT5 ubiquitination, we performed IP-WB analysis in F/STRO1+ cells. We found that c-Cbl silencing markedly decreased STAT5 associated with ubiquitin, indicating that STAT5 protein level is regulated by c-Cbl-mediated ubiquitination (Fig. 3C). This effect was confirmed in murine C3H10T1/2 cells (Supporting Information Fig. 1C). To further investigate the functional interaction between c-Cbl, STAT5 and RUNX2, we determined whether c-Cbl silencing can impact RUNX2 and STAT5 protein levels in human MSC. The analysis of multiple WBs showed that c-Cbl silencing increased RUNX2 and STAT5 protein levels in F/STRO1+A cells (Fig. 3D, 3E). These results indicate that STAT5 and RUNX2 protein levels are controlled by c-Cbl in MSC.
The interactions between c-Cbl, STAT5, and RUNX2 described above suggest that c-Cbl silencing may act in part through STAT5-mediated modulation of RUNX2 in MSC. To investigate this hypothesis, we analyzed the effect of c-Cbl silencing on STAT5 activity. The activation and nuclear translocation of STAT proteins are dependent on SH2–phosphotyrosine interactions, resulting in dimerization, nuclear translocation, binding to specific DNA response elements, recruitment of co-activator(s), and transcriptional induction of target genes . We therefore analyzed whether c-Cbl silencing resulted in alteration in STAT5 phosphorylation in F/STRO1+A cells. We found that silencing c-Cbl increased total and phosphorylated STAT5 protein levels. Moreover, STAT5 phosphorylation induced by c-Cbl silencing was inhibited by the STAT5 inhibitor that inhibits the function of the SH2 domain of STAT5, confirming a role of c-Cbl on STAT5 activity (Fig. 4A, 4B). Because STAT3 was reported to control osteoblast differentiation in MSC , possibly via RUNX2 expression , we determined whether c-Cbl silencing may also affect STAT3 in F/STRO1+A cells. We found that silencing c-Cbl had no effect on total STAT3 protein level. Phospho-STAT3 level was only slightly increased (30%–40%) by c-Cbl silencing, possibly as a result of the expected increase in RTK signaling induced by c-Cbl silencing in MSC  (Supporting Information Fig. 2A, 2B). This indicates that, in contrast to STAT5, the total STAT3 level is not affected by c-Cbl silencing in these cells.
To investigate whether c-Cbl controls STAT5-mediated RUNX2 activity, we performed a functional analysis of RUNX2 transcriptional activity. As shown in Figure 4C, c-Cbl silencing increased RUNX2 reporter activity in basal conditions and under RUNX2 overexpression in F/STRO1+A cells. More importantly, when cells were transfected with a RUNX2 M3 construct bearing lysine (potential ubiquitination site) to arginine or alanine mutant (M3: K219, 224A; K344, 345R) and thereby codes for a protein that may no longer be modified by c-Cbl, the transcriptional activity was not enhanced compared to the RUNX2 construct, as it would be expected in case of a functional c-Cbl-RUNX2 interaction. Moreover, RUNX2 reporter expression induced by c-Cbl silencing was abrogated in the presence of the STAT5 inhibitor (Fig. 4C). These results strongly indicate that c-Cbl silencing impacts RUNX2 transcriptional activity by acting on STAT5 rather than on RUNX2 ubiquitination in MSC. To analyze whether c-Cbl may directly affect STAT5 transcriptional activity, we performed a STAT5 RE-LUC reporter assay in F/STRO1+A cells. The data showed that silencing c-Cbl increased STAT5 transactivation, an effect that was abrogated by the STAT5 inhibitor (Supporting Information Fig. 3A), indicating a direct link between c-Cbl and STAT5 activity. To determine whether STAT5 activation is required for RUNX2 activity, we analyzed STAT5–RUNX2 interaction in the presence of the STAT5 inhibitor which inhibits the function of the SH2 domain of STAT5. Our IP-WB analysis showed that the increased STAT5–RUNX2 interaction induced by c-Cbl silencing in F/STRO1+A cells was abolished in the presence of the STAT5 inhibitor (Supporting Information Fig. 3B), indicating that STAT5 activation is required to interact with Runx2.
Osteoblast Differentiation Induced by C-Cbl Silencing Involves STAT5 and IGF-1
To further determine the functional role of STAT5 on mesenchymal cell differentiation induced by c-Cbl silencing, STAT5 was inhibited using a pharmacological inhibitor, and osteoblast differentiation marker expression was analyzed. We found that inhibition of STAT5 activity abrogated the stimulatory effect of c-Cbl shRNA on ALP staining (Fig. 5A), matrix mineralization (Fig. 5B), and osteoblast phenotypic gene expression in F/STRO1+A cells (Fig. 5C, 5D). This effect was not restricted to human MSC since similar effects on ALP staining, matrix mineralization, and osteoblast gene expression were observed in murine C3H10T1/2 cells (Supporting Information Fig. 4A–4C). Consistent with these results, c-Cbl silencing increased the expression of more mature osteoblast genes such as bone sialoprotein and osteocalcin in F/STRO1+ cells, and this effect was abrogated by STAT5 inhibition (Supporting Information Fig. 5A). In contrast to osteoblastic genes, c-Cbl silencing had no effect on adipocyte gene markers and reduced chondrocyte genes in these cells, suggesting a preferential effect on the orientation toward osteoblast rather than chondrocyte or adipocyte differentiation (Supporting Information Fig. 5B, 5C). The effect of c-Cbl silencing on SOX9 expression was partially abolished by STAT5 inhibition which also increased PPARγ2, aP2, and LPL mRNA levels (Supporting Information Fig. 5C), which is consistent with a role of STAT5 in c-Cbl-mediated orientation towards the osteogenic pathway. Overall, these data support the concept that Cbl silencing preferentially promotes osteoblast differentiation in MSC at least in part through increased STAT5 functional activity.
One established target gene of STAT5 signaling is IGF-1 [51, 52], a major growth factor involved in osteoblastogenesis . We thus hypothesized that one mechanism by which STAT5 activation induced by c-Cbl silencing may promote MSC osteoblast differentiation is through IGF-1 expression. To test this hypothesis, we analyzed the expression of IGF-1 in response to c-Cbl silencing. As shown in Figure 6A, c-Cbl silencing increased IGF-1 mRNA expression in murine C3H10T1/2 cells, and this effect was fully abolished by pharmacological inhibition of STAT5. Similar effects were observed in human F/STRO-1+A cells (Fig. 6B), further suggesting that c-Cbl silencing promotes osteoblast differentiation both in human and mouse MSC in part through STAT5-mediated IGF-1 expression. To confirm this finding, we performed a functional assay on matrix mineralization which characterizes the final osteogenic program. We found that the induction of matrix mineralization by c-Cbl silencing in F/STRO-1+A cells was blunted by a IGF-1 neutralizing antibody (Fig. 6C). Consistent with this finding, inhibition of IGF-1 abolished the promoting effect of c-Cbl silencing on osteoblast phenotypic genes and c-Myc, a Wnt-responsive gene (Supporting Information Fig. 6A–6C). These functional assays support a role of IGF-1 in MSC differentiation induced by c-Cbl silencing. To further investigate the functional link between c-Cbl, STAT5, and IGF-1 in vivo, we analyzed the changes in IGF-1 mRNA expression in bone marrow stromal cells isolated from c-Cbl knockout mice . As shown in Figure 6D, bone marrow cells from c-Cbl−/− mice express higher IGF-1 mRNA levels compared to wild-type mice, confirming the in vitro data in c-Cbl silenced MSC. Taken together, our data indicate that c-Cbl silencing in mesenchymal cells leads to increased STAT5 and STAT5-RUNX2 activity, causing increased IGF-1 expression and MSC osteogenic differentiation (Fig. 6E).
Our results indicate that the E3 ubiquitin ligase c-Cbl regulates osteoblast differentiation in mesenchymal cells by governing STAT5 level and RUNX2 transcriptional activity. We first showed that c-Cbl silencing promoted the osteogenic differentiation program in murine and human MSC, as demonstrated by the increased ALP activity, osteoblast gene expression, and matrix mineralization. c-Cbl silencing also increased cell proliferation in murine and human MSC. One possible explanation for this apparent antagonism is that some cells may be proliferating while others are not and are actually induced to differentiate by c-Cbl silencing. Additionally, we found that Cbl-b level was increased in MSC under c-Cbl silencing, which may compensate for the c-Cbl downregulation in some MSC. As other proteins, transcription factors are regulated by protein degradation . Importantly, we found that c-Cbl silencing increased the level of the transcription factor STAT5 as a result of decreased STAT5 ubiquitination, suggesting that STAT5 is targeted by c-Cbl for protein degradation. This finding is likely to have important implications in MSC since STAT5 is an essential transcription factor controlling cell differentiation . The function of STAT5 molecules is known to be highly dependent on the interaction with cofactors . We found that STAT5 forms a functional complex with RUNX2, an essential transcription factor involved in MSC osteoblast differentiation. Our data also indicate that the increased STAT5 induced by c-Cbl silencing increased STAT5–RUNX2 interaction in MSC. This effect was functional since it resulted in increased RUNX2 transcriptional activity and enhanced expression of genes involved in osteoblast differentiation. Thus, although the effects of shCbl on RUNX2 transcriptional activity may not be exclusively mediated via STAT5, the present results indicate that c-Cbl silencing promotes osteoblast differentiation in MSC in part by increasing STAT5–RUNX2 interaction and transcriptional activity. This concept is consistent with the recent report indicating that STAT5 activation promotes osteogenic differentiation in murine MSC . While STAT5 was reported to interact with RUNX proteins in lymphocytes , our study is the first to report a functional STAT5–RUNX2 interaction and a subsequent positive impact on osteoblast differentiation in MSC.
RUNX2 expression is known to be regulated at multiple levels  including protein degradation. Notably, the E3 ubiquitin ligases Smurf-1 and Schnurri-3 were found to associate with RUNX2 and to mediate its ubiquitination and degradation by the proteasome [25, 26]. RUNX2 exhibits a putative interaction site with c-Cbl, and it was suggested that Cbl-b may control RUNX2 protein levels at the post-translational event . In our studies, however, Cbl-b was upregulated by c-Cbl silencing and our coimmunoprecipitation studies revealed no detectable interaction between c-Cbl and RUNX2 in MSC (data not shown). Moreover, we found that the RUNX2 M3 mutant that abrogates the recruitment of c-Cbl did not affect RUNX2 transcriptional activity induced by c-Cbl silencing in MSC, further indicating that the increased osteoblast differentiation induced by c-Cbl silencing may not involve a direct interaction of c-Cbl with RUNX2. In contrast, we found that c-Cbl silencing controls STAT5 activity, which resulted in increased IGF-1 levels, suggesting that IGF-1 may mediate the c-Cbl-dependent MSC osteogenic differentiation. Growth hormone (GH)/IGF-1 signaling is known to play a major role in osteoblastogenesis and bone formation [53, 57, 58]. The actions of GH in osteoblasts are in part mediated by STAT5 activation which upregulates IGF-1 expression [51, 59]. IGF-1 signaling in turn activates RUNX2 activity and promotes osteoblast differentiation in MSC [60–62]. Our finding that c-Cbl silencing increased IGF-1 expression in vitro and in vivo suggests that IGF-1 may functionally contribute to the observed induction of osteoblast differentiation. A role of IGF-1 in MSC differentiation induced by c-Cbl silencing is supported by our finding that both phenotypic genes and matrix mineralization promoted by c-Cbl silencing were abolished by IGF-1 neutralization. These findings demonstrate a functional role of Cbl-mediated regulation of STAT5 and IGF-1 in the control of the osteoblast differentiation program in MSC and warrant further investigations on the molecular mechanisms by which STAT5 activity controls MSC differentiation.
Our functional studies on the role of the E3 ubiquitin ligase c-Cbl show that c-Cbl ablation activates the osteogenic differentiation program in mesenchymal cells. This effect is in part mediated by mechanisms involving decreased c-Cbl-mediated STAT5 ubiquitination, enhanced STAT5-RUNX2 interaction, and increased IGF-1 level and osteoblast gene expression. This supports an important role of Cbl in the regulation of mesenchymal cell differentiation, which may serve as a basis to design strategies targeted to c-Cbl to increase MSC osteogenic differentiation in situations where osteoblastogenesis is compromised.
This work was supported in part by a grant from the Agence Nationale de la Recherche (ANR-09-BLAN-0402-01 Promotos) to P.J.M. N.S. was a recipient of a Ph.D. Award from the Ministère de la Recherche (Paris, France). We thank Dr. Roland Baron (Harvard Dental School of Medicine, USA) for sharing bones from c-Cbl knockout mice. M.B.D. is currently affiliated with Service Odontologie Bretonneau, AP-HP, INSERM U781, Université Paris Descartes, Paris, France.
DISCLOSURE OF POTENTIAL CONFLICTS OF INTEREST
The authors indicate no potential conflicts of interest.