MyoD initiates muscle differentiation and promotes skeletal myogenesis by regulating temporal gene expression. MyoD-interacting proteins induce regulatory effects, and the identification of new MyoD-binding partners may provide mechanistic insights into the regulation of gene expression during myogenesis. FHL3 is one of three members of the FHL protein family that are expressed in skeletal muscle, but its function in myogenesis is unknown. Overexpression of human FHL3 in mouse C2C12 cells retarded myotube formation and decreased the expression of muscle-specific regulatory genes such as myogenin but not MyoD. By contrast, short interfering RNA (siRNA)-mediated FHL3 protein knockdown enhanced myoblast differentiation associated with increased myogenin, but not MyoD protein expression, early during differentiation. We demonstrate that FHL3 is a MyoD-associated protein by direct binding assays, colocalisation in the nucleus of myoblasts and GST pull-down studies. Moreover, we determined that FHL3 interacts with MyoD, functioning as its potent negative co-transcriptional regulator. Ectopic expression of FHL3 in myoblasts impaired MyoD-mediated transcriptional activity and muscle gene expression. By contrast, siRNA-mediated FHL3 knockdown enhanced MyoD transcriptional activity in a dose-dependent manner. These findings reveal that FHL3 association with MyoD may contribute to the regulation of MyoD-dependent transcription of muscle genes and thereby myogenesis.
Postnatal skeletal muscle myogenesis is crucial in maintaining muscle mass during aging and essential for muscle repair, particularly in muscular dystrophies. Myoblasts are muscle precursor cells that are committed to the skeletal muscle lineage and respond to external signals leading to activation of specific gene expression, which in turn promotes a differentiated phenotype. The myogenic basic helix-loop-helix (bHLH) family of proteins govern the activation of muscle-specific gene expression and include Myf5, MyoD, myogenin and Myf6. These transcription factors direct the temporal regulation of myogenesis (Ludolph and Konieczny, 1995). Myf5 and MyoD play significant roles in committing somite-derived cells to skeletal muscle fates (Braun et al., 1994; Cossu and Borello, 1999; Buckingham et al., 2003). In addition MyoD also initiates the onset of differentiation and plays a regulatory role in governing the temporal expression of specific genes such as myogenin (Montarras et al., 2000). In turn myogenin governs myofibril formation and Myf6 regulates muscle maintenance and function (Bober et al., 1991; Buckingham et al., 2003). These four myogenic bHLH proteins, collectively called muscle regulatory factors (MRFs), form heterodimers with E proteins including E12 or E47 and bind to specific DNA sites called E-boxes found in many muscle-specific gene promoters thereby converting non-myogenic, mesenchymal cells into myoblasts (Davis et al., 1987; Choi et al., 1990; Lassar et al., 1991).
LIM proteins contain LIM motifs and regulate gene transcription and scaffold sarcomeric and signaling proteins. The LIM motif contains the consensus sequence (Cx2Cx16-23[C/H]x2-4[C/H/E]x2Cx2Cx14-21[C/H]x1-3[C/H/D/E]) but does not directly bind DNA and, instead, mediates protein-protein interactions (Kadrmas and Beckerle, 2004). The four and a half LIM domain (FHL) proteins are a family of LIM-only proteins, characterised by four complete LIM domains, preceded by an N-terminal half LIM motif. To date five mammalian family members FHL1, FHL2, FHL3, FHL4 and ACT have been identified, which act as transcriptional regulators (Fimia et al., 2000) and/or play structural roles in the actin cytoskeleton (Coghill et al., 2003; McGrath et al., 2003; Robinson et al., 2003; Samson et al., 2004; McGrath et al., 2006). FHL1, FHL2 and FHL3 are all expressed in striated muscle, and FHL3 is more highly expressed in skeletal muscle relative to cardiac tissue (Morgan and Madgwick, 1996; Chu et al., 2000).
The function of FHL1 and FHL2, but not FHL3, in skeletal myogenesis has been explored. FHL1 is significantly upregulated during cardiac and skeletal muscle hypertrophy (Morgan et al., 1995; Lim et al., 2001) and its overexpression induces α5β1-integrin-dependent hyper-elongation of myocytes and hypertrophic myosacs during differentiation of C2C12 myoblasts (McGrath et al., 2003; Robinson et al., 2003; McGrath et al., 2006). FHL1 localises at the I-band and M-line where it binds myosin-binding protein C (MyBP-C) to regulate myosin filament formation and sarcomere assembly (McGrath et al., 2006). FHL2 is the most extensively characterised of the FHL proteins (Johannessen et al., 2006) and forms a transcriptional co-activator, co-repressor, or competitor, dependent on tissue type and promoter context. FHL2 interacts with transcription factors, such as the androgen receptor (Muller et al., 2002), p300, β-catenin (Labalette et al., 2004), FOS, JUN (Morlon and Sassone-Corsi, 2003) and FOXO1 (Yang et al., 2005). FHL2 regulates cellular events including apoptosis (Scholl et al., 2000), survival (Stilo et al., 2002), proliferation (Chen et al., 2003) and differentiation (Du et al., 2002; Martin et al., 2002; Bai et al., 2005; Lai et al., 2006). Fhl2 knockout mice exhibit an exaggerated hypertrophic cardiomyopathy in response to β-adrenergic stimulation (Kong et al., 2001) and develop osteopenia as a consequence of reduced osteoblast and increased osteoclast differentiation and function (Bai et al., 2005; Gunther et al., 2005; Lai et al., 2006). In skeletal muscle, FHL2 enhances C2C12 myogenesis via the Wnt signalling pathway mediated by its interaction with β-catenin (Martin et al., 2002).
FHL3 also possesses intrinsic transactivation and repressor activity in a cell-type and gene-specific manner. FHL3 interacts with transcription factors, such as CREB (Fimia et al., 2000), BKLF/KLF3 and CtBP2 (Turner et al., 2003), PLZF (McLoughlin et al., 2002), and MZF-1 (Takahashi et al., 2005). FHL3 also forms a complex with CDC25B2 phosphatase (Mils et al., 2003) and ERK2 (Purcell et al., 2004). In myoblasts the only known function of FHL3 is to destabilise actin bundles by preventing α-actinin crosslinking (Coghill et al., 2003); however, its transcriptional targets in skeletal muscle have not been reported.
The function of FHL3 in skeletal myogenesis is currently unknown and forms the basis of this study. Using FHL3 overexpression and siRNA-mediated protein knockdown during C2C12 myogenesis, we demonstrate that FHL3 negatively regulates C2C12 myotube formation. FHL3 forms a complex with MyoD inhibiting its transcriptional activity and regulates the expression of genes such as muscle creatine kinase (Ckm, also known as and hereafter referred to as MCK) and myogenin. These studies have identified FHL3 as a new regulator of MyoD-dependent myoblast differentiation.
FHL3 overexpression retards C2C12 cell differentiation
FHL3 is most highly expressed in skeletal muscle, however, its function in myogenesis is unknown. C2C12 cells are a well-defined model for studying myogenic differentiation because, when cultured under conditions that promote differentiation, they fuse to form mature myotubes (Yaffe and Saxel, 1977). To determine whether FHL3 overexpression affected myoblast differentiation, C2C12 myoblast cell lines that were transfected to stably overexpress hemagglutinin-tagged FHL3 (HA-FHL3) or vector at low levels were generated in duplicate by pooling G418-resistant clones (pool 1 or 2, respectively). FHL3 immunoblot analysis revealed a 33 kDa immunoreactive polypeptide, whose expression levels were greater (∼1.4-fold) in cell lines overexpressing HA-FHL3 relative to Vector (Fig. 1A,B). HA-immunoblot analysis revealed that the HA-FHL3 construct was expressed at all times during C2C12 cell differentiation (Fig. 1C). Significantly, the myotubes formed in HA-FHL3 C2C12 cell lines appeared consistently and reproducibly smaller and thinner, with less nuclei per myotube relative to Vector cell lines after 120 hours of differentiation (Fig. 1D). The pan actin antibody, which recognises all forms of actin was used as a protein loading control throughout this study. Whereas α-skeletal and α-cardiac actin levels may increase, and β-actin and γ-actin levels decrease during C2C12 cell differentiation (Bains et al., 1984), the total actin protein pool remains a constant with ∼18% of total cellular protein for confluent C2C12 cells and ∼17% for differentiating C2C12 cells (Babcock and Rubenstein, 1993).
To provide a quantitative analysis of C2C12 myoblast differentiation into multinuclear myotubes, specific aspects of myoblast differentiation using well-defined parameters, including the differentiation and fusion indices, were assessed. At the onset of differentiation, ∼50% of C2C12 myoblasts express myogenin and myosin heavy chain (MHC) proteins (Miller, 1990; Conejo and Lorenzo, 2001; Conejo et al., 2001), and form myocytes, which subsequently fuse into multi-nucleated myotubes. The remaining myoblasts form myogenin and MHC-negative reserve cells (Yoshida et al., 1998). The differentiation index is a measure of the proportion of all nuclei, that are localised within myocytes and myotubes, providing a score of myoblasts differentiating into myofibres (Sabourin et al., 1999; Erbay et al., 2003). Normally, this does not exceed ∼50% owing to the presence of reserve cells (Yoshida et al., 1998). For these experiments myocytes and myotubes were identified by MHC-positive staining, and nuclei by propidium-iodide staining. HA-FHL3 C2C12 myotubes exhibited a significantly lower differentiation index (Fig. 1E), than Vector myotubes at 96 hours and 120 hours of differentiation, but not prior to these time points. Myocyte cell fusion was calculated as the average number of nuclei per MHC-positive cell (Sabourin et al., 1999). At 96-120 hours of differentiation, HA-FHL3 cell lines exhibited decreased cell fusion relative to cells transfected with Vector (Fig. 1F).
Myoblast differentiation was also assessed by analysis of the temporal expression of the muscle-specific proteins MyoD and myogenin, standardised to total actin levels as a protein loading control. HA-FHL3 myotubes consistently showed a ∼50% decrease in myogenin protein at 24 hours of differentiation relative to Vector cells, as assessed by anti-myogenin immunoblot analysis, however, MyoD levels appeared relatively unaffected (Fig. 2A,B). In addition, the proportion of myogenin-positive nuclei in HA-FHL3 myocytes was reduced by ∼50% at 24 hours of differentiation relative to Vector cell lines (Fig. 2C,D). Collectively, these studies reveal that FHL3 expression delays myogenin expression, and retards myoblast differentiation and cell fusion into multinucleated myotubes, whereas expression of MyoD is relatively unaffected.
FHL3 is expressed throughout differentiation of C2C12 cells and its siRNA-mediated knockdown accelerates C2C12 cell differentiation
Since we had observed that FHL3 gain-of-function inhibits myoblast differentiation and fusion, we next determined the effect of extinguishing the expression of endogenous FHL3 using siRNA technology. Expression of either of two siRNA oligonucleotides (#2 and #3) targeting the sequence of GenBank entry NM_010213 (mouse Fhl3 mRNA), significantly reduced FHL3 protein levels at all times during differentiation in C2C12 myoblasts compared with a non-silencing control siRNA, as assessed by FHL3 immunoblot analysis (Fig. 3A,B). In control-siRNA-transfected cultures, anti-FHL3 western blots revealed that endogenous FHL3 protein was expressed at all stages of C2C12 cell differentiation, with levels increasing threefold at 48 hours of differentiation (Fig. 3B). FHL3-siRNA-transfected cells exhibited an early increase in myocyte numbers after 24 hours of differentiation, correlating with the accelerated formation of visibly larger myotubes containing increased numbers of nuclei at 48-72 hours (Fig. 3C, images shown are at 72 hours of differentiation). We also noticed an increased differentiation index. For example, cultures transfected with the FHL3 siRNAs #2 and #3 had already reached a differentiation index of ∼50%, whereas cells transfected with control siRNA achieved ∼32% at 72 hours of differentiation (Fig. 3D). To assess cell fusion, the average number of nuclei per MHC-positive cell was determined (Sabourin et al., 1999). Significantly, after 72 hours of differentiation cells transfected with the FHL3 siRNAs exhibited a more than threefold increase in cell fusion compared with control-siRNA-transfected cells (Fig. 3E). Hence, siRNA-mediated knockdown of FHL3 accelerates myoblast differentiation and myotube formation, the opposite phenotype to FHL3 overexpression.
No consistent differences in MyoD protein levels (standardised to total actin) were detected following siRNA-mediated FHL3 knockdown compared with control siRNA during C2C12 cell differentiation (Fig. 4A). The expression of MyoD in the nucleus of C2C12 myoblasts transfected with FHL3 siRNA oligonucleotides was also examined by indirect immunofluorescence, with no significant differences observed (data not shown). siRNA-mediated FHL3-knockdown-accelerated myogenesis was associated with a 2.7-fold increase in myogenin protein levels at 24 hours of differentiation (Fig. 4A,B), levels not observed until 48 hours of differentiation in control siRNA cultures. Beyond 24 hours of differentiation, myogenin levels were more variable and, as such, no significant reproducible differences were observed for later time points. In addition, the proportion of myogenin-positive nuclei in cells transfected with FHL3 siRNA was increased after 24 hours of differentiation by 1.7-fold compared with that detected in control siRNA transfected cells (Fig. 4C,D).
Given FHL3 was reported to bind actin and destabilise actin filaments (Coghill et al., 2003), sarcomere formation was also evaluated in cells with altered FHL3 expression. FHL3-overexpressing C2C12 cells demonstrated a reduction in myotubes displaying visible Z-line formation relative to Vector cell lines at 120 hours of differentiation (data not shown). siRNA-mediated FHL3 knockdown induced a temporal increase in the numbers of myotubes displaying Z-line formation (data not shown). However, in FHL3-overexpressing cell lines, no disruptions were observed in the F-actin cytoskeleton, as evaluated by phalloidin staining. Furthermore, FHL3 did not colocalise to actin-rich sites of Z- and M-line myofibrillogenesis in wild-type C2C12 cell myotubes (data not shown), suggesting the myogenic phenotypes induced by FHL3-altered expression are distinct from its actin regulatory effects.
Ectopic FHL3 overexpression rescues FHL3 siRNA-mediated phenotype
siRNA may induce non-specific off-target gene effects unrelated to the downregulation of the gene of interest (Jackson and Linsley, 2004). To confirm the phenotypes induced by FHL3 siRNA were specific to FHL3, a rescue of the siRNA FHL3 phenotypes was attempted by overexpression of HA-FHL3. The FHL3 siRNA oligonucleotides used in this study were designed to target mouse mRNA and exhibit several base-pair differences with the human FHL3 sequence; therefore, human HA-FHL3 should be resistant to degradation by mouse FHL3 siRNA oligonucleotides. Stable cell lines overexpressing human HA-FHL3 were transfected with mouse FHL3 siRNA oligonucleotides. FHL3 protein expression as assessed by FHL3 immunoblot analysis in the stable cell lines was significantly reduced by FHL3 siRNA transfection, relative to control siRNA-transfected cells (Fig. 5A). By contrast, anti-HA immunoblot analysis revealed that expression of 33 kDa recombinant human HA-FHL3 was not affected by mouse FHL3 siRNA transfection (Fig. 5A). C2C12 cells stably overexpressing Vector or HA-FHL3, were transfected with FHL3 or control siRNA oligonucleotides, differentiated for 72 hours and stained with anti-MHC antibody and propidium iodide. HA-FHL3 cells transfected with control siRNA demonstrated a more immature myofibre phenotype, with thinner myofibres showing decreased multi-nucleation relative to Vector-transfected myofibres. By contrast, siRNA FHL3 #2 or #3 transfection in Vector cell lines resulted in accelerated myotube formation as described above. However, HA-FHL3 overexpressing cells transfected with siRNA FHL3 #2 or #3 exhibited myotube formation at similar rates and levels to control siRNA transfected cells (Fig. 5B). The relative level of myogenin protein after 24 hours of differentiation, as assessed by anti-myogenin immunoblot analysis, revealed rescue of the elevated myogenin protein levels in FHL3 siRNA cells following HA-FHL3 overexpression (Fig. 5C,D). In addition, the number of myogenin-positive nuclei following 24 hours of differentiation was also decreased (data not shown). By this analysis overexpression of HA-FHL3 rescued the accelerated myotube development mediated by siRNA FHL3 #2 or #3. Therefore, HA-FHL3 expression antagonises the FHL3-siRNA-induced phenotype and FHL3 negatively regulates myogenin protein expression early in differentiation.
FHL3 forms complexes with MyoD and other bHLH proteins
These results prompted us to ask whether FHL3 inhibits myogenesis by regulating muscle gene expression. Given that we demonstrated MyoD protein expression was unchanged by FHL3 overexpression or siRNA-mediated knockdown, although expression of its downstream target gene myogenin was altered, we investigated whether FHL3 forms a complex with MyoD, thereby regulating the transcription of its target genes and, hence, myoblast differentiation. FHL proteins exhibit varying levels of intrinsic activation potential (Fimia et al., 2000). Evidence of in vivo complex formation between FHL3 and MyoD was initially demonstrated by mammalian two-hybrid analysis. The interaction of MyoD and FHL3 was quantified by the activation of a luciferase reporter containing Gal4-binding sites in the promoter. To this end a Gal4-promoter luciferase reporter was transactivated by Gal4DBD-FHL3 and the effect of MyoD overexpression determined. Under these conditions the intrinsic transactivation domain of MyoD was used (Fig. 6A). These results confirmed that MyoD and FHL3 interact in vivo, because the luciferase activity from cells expressing both proteins was increased ∼eightfold compared with cells expressing Gal4DBD-FHL3 protein plus empty vector. In control studies all recombinant proteins used in these assays were expressed intact, as determined by immunoblot analysis (Fig. 6B,C).
Do FHL3 and MyoD interact directly, without the cooperation or requirement for other muscle-specific proteins, or do they form part of a multi-protein complex and interact indirectly? To address this question, we evaluated whether purified recombinant GST-FHL3 directly interacts with His-tagged MyoD. Recombinant His-MyoD and GST-FHL3 were coexpressed in Escherichia coli, and bacterial lysates were incubated with glutathione Sepharose and washed extensively. In control studies, the muscle LIM protein (MLP) GST fusion protein (GST-MLP) was coexpressed with His-MyoD, because previous studies have shown that these species interact directly (Kong et al., 1997). GST-FHL3 did bind full-length His-MyoD as well as His-tagged C-terminally truncated MyoD breakdown fragments, however, the interaction between GST-FHL3 and MyoD was consistently less, under these in vitro binding conditions, than the complex between GST-MLP and His-MyoD (Fig. 6D).
To further confirm an interaction between FHL3 and MyoD, a GST pull-down assay was undertaken using purified GST or GST-FHL3 protein produced in E. coli and partially purified by coupling to glutathione Sepharose beads. C2C12 cells were differentiated for 24 hours and cell lysates mixed with GST-FHL3-coupled or GST-coupled beads. A 24-hour time point was selected because differentiation-dependent transactivation of the myogenin gene occurs by then (Shimokawa et al., 1998). Bound, supernatant and lysate input fractions were immunoblotted with rabbit anti-MyoD or anti-GST antibodies as a loading control. A 40 kDa polypeptide corresponding to endogenous MyoD was specifically pulled down by GST-FHL3 but not GST (Fig. 6E). This polypeptide was not detected by immunoblot analysis using non-immune rabbit antibody, purified GST or GST-FHL3 probed with anti-MyoD antibody, which had not been incubated with C2C12 cell lysates (data not shown). Hence, endogenous MyoD protein specifically associates with recombinant GST-FHL3 protein, confirming an interaction between these proteins.
bHLH transcription factors that are upregulated in response to myogenic signals include MyoD, myogenin, Myf5, and MRF4. MRFs form heterodimers with E proteins, such as E12 or E47, and bind to specific E-box DNA sites in many muscle-specific gene promoters (Lassar et al., 1991). The GST pull-down assay outlined above was repeated and samples were immunoblotted with antibodies against other related bHLH proteins, such as Myf5, myogenin and E47. All these proteins were detected in complex with purified recombinant GST-FHL3 but not GST (Fig. 6F), suggesting that FHL3 either binds a common structural motif in these proteins or a common adapter molecule; however, it might bind MyoD itself, which in turn associates with E47 as previously reported (Lassar et al., 1991).
To further determine whether endogenous FHL3 and MyoD indeed interact in vivo, we performed a series of colocalization studies. Endogenous FHL3 colocalised in the nucleus of myoblasts with endogenous MyoD but was also observed in the cytoplasm (Fig. 6G). The relative level of nuclear FHL3 compared with cytoskeletal FHL3 varied in C2C12 cells – with predominant nuclear FHL3 staining observed in sub-confluent growing myoblasts – but was absent from the nuclei of myotubes, suggesting the nuclear localization of FHL3 is regulated by cell density, serum growth factors and/or the degree of differentiation (data not shown). Therefore, our data support the contention that FHL3 and MyoD interact in the nucleus of myoblasts to regulate the onset of differentiation.
FHL3 suppresses MyoD transcriptional activation
As we had demonstrated that FHL3 and MyoD form a complex, and FHL3 overexpression or siRNA-mediated depletion was associated with altered expression of a MyoD-dependent gene, myogenin, we postulated that FHL3 regulates MyoD-dependent gene transcription. To address this issue a series of luciferase assays were carried out. In the first assay the effect of FHL3 overexpression on the transcriptional activity of a Gal4DBD-MyoD fusion protein on a Gal4-promoter was investigated. C2C12 myoblasts were seeded at a sub-confluent density and co-transfected with various combinations of plasmids expressing Gal4DBD, Gal4DBD-MyoD, Vector (HA-vector) or HA-FHL3. Following transfection, myoblasts were maintained in growth media for 48 hours and luciferase activity was assayed. Gal4DBD-MyoD coexpression with Vector transactivated the luciferase reporter gene approximately 18-fold and this activity was specifically repressed approximately threefold by coexpression of HA-FHL3 (Fig. 7A), consistent with the contention that FHL3 suppresses MyoD transcriptional activity. As FHL3 regulates C2C12 cell differentiation, these assays were repeated in C2C12 cells differentiated for 48 hours. Gal4DBD-MyoD, coexpressed with Vector, activated the luciferase reporter gene sixfold (compared with Gal4DBD), and coexpression of HA-FHL3 significantly inhibited MyoD reporter gene activity by approximately twofold (Fig. 7B). Expression of intact recombinant proteins from pCMX-Gal4DBD and pCGN constructs was verified in C2C12 myoblasts by immunoblot analysis (Fig. 7C,D).
MyoD binds the E-boxes (CANNTG) of the gene encoding MCK with high affinity to activate gene transcription (Chakraborty et al., 1991). The ability of FHL3 to regulate MyoD-dependent transcription of the MCK-promoter luciferase reporter pGL3-MCK (Novitch et al., 1999) was investigated. The basal reporter activity – i.e. 1 relative luciferase unit, was arbitrarily defined as 1 RLU – in Flag-vector transfected cells was decreased by approximately threefold by expression of HA-FHL3, relative to Vector. This change in the basal level of reporter activity is probably due to FHL3 regulation of endogenous MyoD, although FHL3 regulation of other transcription factors cannot be excluded, because the full-length MCK promoter (in pGL3-MCK) contains many binding elements, including MEF2 (Amacher et al., 1993). The MCK-luciferase reporter gene was further transactivated by expression of MyoD-Flag with Vector and decreased significantly following coexpression with HA-FHL3 (Fig. 7E). Therefore, FHL3 inhibits MyoD-dependent transcription of target genes.
siRNA-mediated FHL3 knockdown increases MyoD transcription of luciferase reporter genes
To further verify that FHL3 regulates MyoD transcriptional activity, the Gal4-based luciferase assay outlined above was repeated in C2C12 cells undergoing differentiation under conditions of siRNA-mediated FHL3 knockdown. Gal4DBD-MyoD activity increased 1.5-fold and 2.7-fold by siRNA FHL3 #2 or #3, respectively (Fig. 8A). FHL3 immunoblot analysis was performed on lysates to confirm FHL3 protein knockdown (Fig. 8B). The level of FHL3 protein significantly decreased with transfection of FHL3 siRNA, relative to control siRNA and negatively correlated with the increased Gal4DBD-MyoD transcriptional activity, suggesting FHL3 regulates MyoD-dependent gene transcription in a dose-dependent manner.
To verify whether FHL3 regulates MyoD transcriptional activity during C2C12 cell differentiation on native gene targets, the MCK promoter reporter was used as a reporter construct under conditions of FHL3 knockdown mediated by transfection of siRNA. The basal reporter activity (Flag-vector and control siRNA co-transfected, arbitrarily defined as 1 RLU) was significantly increased approximately threefold and tenfold fold in cells transfected with siRNA FHL3 #2 or #3, respectively. This change in the basal level of reporter activity is probably due to the regulation of endogenous MyoD by FHL3, although the regulation of FHL3 on other transcription factors can, again, not be excluded. Basal MCK promoter activity was also increased sixfold with MyoD-Flag expression and was further significantly increased twofold and sixfold upon transfection with siRNA FHL3 #2 and #3 oligonucleotides, respectively (Fig. 8C). The relative level of FHL3 siRNA protein knockdown negatively correlated with the magnitude of MyoD (recombinant or endogenous) transcriptional activity, suggesting that FHL3 dose dependently regulates MyoD transcriptional activity (Fig. 8D).
We have demonstrated here that FHL3, a relatively uncharacterised member of the FHL protein family, inhibits MyoD transcriptional activity, providing new insights into the molecular mechanisms governing gene expression during myogenesis. We have shown that FHL3 functions as a new regulator of myofibre formation. Evidence from FHL3 overexpression and siRNA protein knockdown has revealed that altered expression of this LIM protein affects both the differentiation and fusion of myoblasts. Other C2C12 cell sub-populations, such as reserve cells and/or myoblasts, were assessed using specific markers for CD34, Pax7, MyoD and PCNA, which showed no differences in the levels and/or localization of these markers following altered FHL3 expression (data not shown). These results suggest that FHL3 functions in myoblast differentiation along the myofibre lineage. We propose that the altered sarcomere assembly detected in myotubes following FHL3 overexpression or siRNA-mediated depletion was a consequence of altered gene transcription, as indicated by temporal changes in myogenin protein levels. We were unable to detect any changes in MyoD protein expression following FHL3 overexpression or siRNA-mediated depletion; however, induction of myogenin protein following 24 hours of differentiation was altered and negatively correlated with the FHL3 protein levels. MyoD and myogenin are essential to skeletal myogenesis. MyoD-null primary myoblasts demonstrate delayed differentiation in vitro (Yablonka-Reuveni et al., 1999) and myogenin is required for cell fusion and myotube formation (Hasty et al., 1993; Nabeshima et al., 1993; Hashimoto and Ogashiwa, 1997). Downregulation of either MyoD or myogenin during C2C12 cell myogenesis causes differentiation and fusion defects, a similar phenotype to the defects observed in cells overexpressing FHL3 (Hashimoto and Ogashiwa, 1997; Dedieu et al., 2002). Owing to the experimental limitations of the C2C12 skeletal muscle cell line, it was impossible to determine whether FHL3 overexpression completely blocked or merely altered the rate of differentiation – the latter phenotype is exemplified by MyoD gene-targeted deletion. For example, MyoD gene deletion delays myogenesis of branchial arches, tongue, limbs and diaphragm during embryogenesis, and the timing of trunk and abdominal wall musculature formation. However, these MyoD-null mice display only minor skeletal muscle changes in adult life (Kablar et al., 1998).
FHL3 has been shown previously to repress gene transcription in non-muscle cell types (Turner et al., 2003; Takahashi et al., 2005). As described here in myoblasts, FHL3 negatively regulates MyoD-dependent gene transcription, whereas siRNA knockdown of FHL3 enhanced MyoD transcriptional activity. The FHL3-mediated effects demonstrated by luciferase assays are most likely the consequence of a transcriptional complex between FHL3 and MyoD, as shown by GST pull-down, direct protein-protein binding and colocalization studies. Both MLP and the FHL3-family member FHL2 bind bHLH proteins, and alter bHLH heterodimerization with E proteins to regulate their transcriptional activity. MLP binds MyoD and myogenin to enhance heterodimerization with co-activator E proteins and promote C2C12 cell myogenesis, whereas FHL2 retards HAND1 transcriptional activity in cardiac tissue by repressing heterodimerization between Hand1 and E proteins (Arber et al., 1994; Kong et al., 1997; Hill and Riley, 2004; Lu et al., 2004). The FHL3 overexpression phenotype was reminiscent of the phenotype induced by Id overexpression which blocks MyoD and E12 heterodimerization and, thereby, retards MyoD transcriptional activity (Jen et al., 1992). Hence, it is likely that FHL3 also regulates the heterodimerization and/or activity of MyoD-bHLH dimers. In this regard it is noteworthy that FHL3 forms a complex with other bHLH proteins, such as Myf5, myogenin and E47. However, the muscle-differentiation phenotype mediated by FHL3 in retarding myotube formation is unique and different to that mediated by other FHL proteins or unrelated LIM proteins, such as MLP. For example, FHL1 overexpression results in myosac formation, whereas its siRNA-mediated depletion impairs the formation of myosin thick filaments associated with reduced incorporation of myosin-binding protein C into the sarcomere (McGrath et al., 2006). FHL2 overexpression in C2C12 mouse myoblasts results in increased myogenic differentiation and accelerated myotube formation by binding to β-catenin to reduce the expression of LEF/TCF (Martin et al., 2002). MLP overexpression promotes myogenesis (Arber et al., 1994; Kong et al., 1997).
Very large transcriptional complexes that involve FHL3 have been previously reported (Takahashi et al., 2005), suggesting that additional proteins facilitate or participate in MyoD:FHL3 complex. Some candidate adapter proteins based on published FHL3 and other FHL family member binding partners may include histone-associated proteins, such as NFY or HDACS (Yang et al., 2005; Takahashi et al., 2006), p300 (Labalette et al., 2004) and CREB (Fimia et al., 2000). CREB is of particular significance because it forms a complex with MyoD and is also an FHL3-binding partner in other cells; however, this has not been shown in muscle (Fimia et al., 2000; Magenta et al., 2003; Kim et al., 2005).
We propose a model whereby FHL3 in the nucleus of myoblasts complexes with MyoD to repress differentiation (Fig. 9). The nuclear localization of FHL3 is regulated by integrin engagement, Rho activity, serum, cell density and cytoskeletal dynamics (Muller et al., 2002; Coghill et al., 2003), suggesting FHL3 connects cell signalling through these pathways, directly to a master control of myogenesis, MyoD. Once appropriate differentiation conditions are met, FHL3 is actively excluded from the nucleus and, thus, is unable to interact with MyoD and regulate its transcriptional activity. Consistent with this model, FHL3 is absent from the nucleus of developing myotubes and FHL3 localises to the Z-line of mature skeletal muscle (Li et al., 2001; Coghill et al., 2003).
In summary, FHL3 protein knockdown and overexpression during C2C12 cell myogenesis have demonstrated a functional role for FHL3 in negatively regulating myotube formation. The selective recruitment of interacting proteins such as FHL3, which negatively regulates MyoD transcriptional activity, may represent a mechanism by which MyoD temporally regulates gene-specific activation and, thereby, myogenesis.
Materials and Methods
Restriction and DNA-modifying enzymes were obtained from Promega (Australia), MBI Fermentas or New England Biolabs. DNA oligonucleotides were purchased from Geneworks (Australia) or Micromon (Australia). E. coli TOP10 were purchased from Invitrogen (Australia). Big Dye Version 3.1 sequencing terminators were supplied by PE applied systems (Australia). All other reagents were purchased from Sigma-Aldrich or BDH Chemicals unless otherwise stated.
Plasmids and cloning
pCGN (Tanaka and Herr, 1990) was a kind gift from Tony Tiganis (Monash University, Melbourne, Australia). pGEX-5x1 was purchased from Amersham Biosciences. pCGN-FHL3, pEGFP-C2-FHL3, pGEX-5x1-FHL3 have been described previously (Coghill et al., 2003). pTrio12R-HGN has been described previously (McGrath et al., 2006) and pTrio12R-HA-FHL3-GN was generated by cloning FHL3 cDNA from pCGN-FHL3 into the XbaI site downstream of the HA-tag. pCMX-Gal4DBD and G5E1b-LUC (Muller et al., 2000) were kind gifts from Roland Schule (University of Freiberg, Freiberg, Germany). pHRLTK was purchased from Promega. pGL3-MCK (Novitch et al., 1999) was a kind gift of Andrew B. Lassar (Harvard Medical School, Boston, MA). The complete mouse MyoD sequence was amplified with flanking AscI sites from pEMSV-MyoD (Davis et al., 1987), a gift from Edna Hardeman (Children's Medical Research Institute, Sydney, Australia), with the primers 5′-cctggcgcgccagatggagcttctatcgccg and 3′-cctggcgcgccataagcacctgataaatcgc, based on gene bank entry NM_010866 nucleotides 192-209 and 1128-1145 respectively, and cloned via pCR-Blunt (Invitrogen) into the pEFBOS-Flag vector (Mizushima and Nagata, 1990), a gift from Tracey Wilson (Walter and Eliza Hall Institute of Medical Research, Melbourne, Australia). The MyoD coding sequence was cloned from pEFBOS-MyoD-Flag into the EcoRI site of pCMX-Gal4DBD to generate pCMX-Gal4DBD-MyoD. The FHL3 sequence was cloned from pEGFP-C2-FHL3 into pCMX-Gal4DBD (EcoRI site), to generate pCMX-Gal4DBD-FHL3. pET-30a(+) was purchased from Novagen, and pET-30a(+)-MyoD was generated by shuttling the MyoD sequence from pCR-Blunt using in frame flanking EagI sites in the cloning vector into the NotI site of pET-30a(+). The identity and fidelity of the PCR products were confirmed by sequencing (Micromon).
Culturing and differentiation of C2C12 myoblasts
Murine C2C12 myoblasts (ATCC, Manassas, VA) were maintained in growth media (DMEM supplemented with 20% foetal calf serum, 2 mM L-glutamine, 100 units/ml penicillin and 0.1% streptomycin). To induce differentiation experiments, myoblasts at 100% confluency were incubated in DMEM supplemented with 2% horse serum, 2 mM L-glutamine, 100 units/ml penicillin, and 0.1% streptomycin for up to 120 hours. Differentiation media was replaced every 48 hours. For all siRNA transfections, myoblasts were seeded, grown and differentiated in media without antibiotics. Growth media containing 1.5 mg/ml G418-sulphate (AG Scientific) was used for C2C12 cell lines stably expressing Vector (HA-vector) and HA-FHL3. G418-sulphate was removed during differentiation experiments.
Transfection of plasmid DNA and/or siRNA oligonucleotides
Transfection-quality plasmid DNA was prepared using Qiagen (Germany) midiprep kits and transfected using Lipofectamine (Invitrogen) according to the manufacturer's protocol. HP GenomeWide siRNA Oligonucleotides: Mm_FHL3_2_HP siRNA (SI01002890), Mm_FHL3_3_HP siRNA (SI01002897), and control (non-silencing) siRNA (1022076) oligonucleotides (Qiagen) were transfected into C2C12 myoblasts as previously described (McGrath et al., 2006). Plasmid and siRNA co-transfection was performed using Lipofectamine 2000 (Invitrogen) according to the manufacturer's protocol.
Stable overexpression of HA-vector C2C12 myoblast cell lines and HA-FHL3 in C2C12 cells
pTrio12R-HGN and its derivatives autonomously replicate and express a tri-cistronic mRNA encoding HA or HA-fusion proteins, low amounts of glutathione S transferase (GST) and neomycin resistance in mammalian cells (McGrath et al., 2006). pTrio12R-HGN (HA-vector) and pTrio12R-HA-FHL3-GN (HA-FHL3) constructs were stably transfected into C2C12 myoblasts and selected in 1.5 mg/ml G418-sulphate for 3 weeks. Individual G418 resistant myoblasts for each transfected construct were not isolated, but rather were pooled as a heterogeneous population (which equalise upon cell fusion), to limit the background variability of individual clones and to increase the passaging life-time of the C2C12 cell lines, as previously reported (Jen et al., 1992). C2C12 myoblast cell lines stably expressing Vector (HA-vector) or HA-FHL3 were created in duplicate (Pool 1 and Pool 2) and designated as either Vector (1) and Vector (2) or HA-FHL3 (1) and HA-FHL3 (2), respectively.
Immunofluorescent labeling of C2C12 cells
Immunofluorescence staining of C2C12 cells has been previously described (McGrath et al., 2003). Briefly, 2×105 stably transfected C2C12 myoblasts were seeded onto 22×22 mm fibronectin-coated coverslips in six-well plates, cultured for 24 hours in growth media, then switched to differentiation media. Alternatively, 5×104 C2C12 myoblasts were seeded on fibronectin-coated coverslips in growth media, transfected the following day with siRNA oligonucleotides, cultured for 48 hours in growth media, then switched to differentiation media. Cells grown on coverslips were fixed, permeabilised, blocked and incubated with propidium iodide at ∼25 μg/well (Sigma-Aldrich) or To-Pro-3 at a dilution of 1:3000 (Molecular Probes) and antibodies.
Western blot analysis
C2C12 cells were washed twice in PBS and scraped into HEPES lysis buffer (10 mM HEPES pH 8, 10 mM KCl, 0.1 mM EDTA, 0.2% NP-40, Roche complete mini protease cocktail inhibitor tablet). Samples were rocked for 1 hour at 4°C, then pelleted for 10 minutes at 16,000 g at 4°C. Alternatively, extracts were prepared using Passive Lysis Buffer as part of the Dual Luciferase Reporter Assay Kit (Promega). Protein concentration was determined using BioRad DC protein assay kit. 25-50 μg of lysate was analysed by SDS-PAGE and western blotting.
Antibodies include anti-pan actin Ab-5 clone ACTN05 (1:3333 dilution western blotting-WB, NeoMarkers anti-E47 N-649 (1:100 WB, Santa Cruz Biotechnology), anti-FHL3 (1:30 immunofluorescence-IF and 1:200 WB) (Coghill et al., 2003), anti-Flag polyclonal (1:500 WB, Sigma-Aldrich), anti-Gal4(DBD) clone RK5C1 (1:100 WB, Santa Cruz Biotechnology), anti-GST polyclonal (1:1000 WB, Amersham Biosciences), anti-HA.11 clone 16B12 (1:5000 WB, Covance), anti-polyHistidine Clone HIS-1 (1:3000 WB, Sigma-Aldrich), anti-MHC MF20 (1:20 IF, University of Iowa, Developmental Studies Hybridoma Bank), anti-Myf5 C-20 (1:100 WB, Santa Cruz Biotechnology), anti-myogenin F5D (1:100 WB, Santa Cruz Biotechnology), anti-myogenin M-225 (1:100 IF and WB, Santa Cruz Biotechnology), anti-MyoD 5.8A (1:50 IF, Imgenex Corporation), anti-MyoD C-20 (1:100 WB, Santa Cruz Biotechnology), anti-MyoD M-318 (1:100 WB, Santa Cruz Biotechnology), donkey anti-goat-HRP 1:10,000 WB (Chemicon), donkey anti-mouse (H+L)-Alexa Fluor-594 (1:600 IF, Molecular Probes), donkey anti-rabbit (H+L)-Alexa Fluor-647 (1:600 IF, Molecular Probes), sheep anti-mouse (H+L)-FITC (1:400 IF, Chemicon), sheep anti-mouse-HRP (1:10,000 WB, Chemicon), sheep anti-rabbit (H+L)-FITC (1:400 IF, Chemicon), sheep anti-rabbit-HRP (1:10,000 WB, Chemicon),
Direct FHL3:MyoD protein interaction
The E. coli strain BL21 DE3 Gold Codon Plus RP (Stratagene) were co-transformed with pGEX-5x1 derivatives (Amersham Biosciences) and pET-30a(+) derivatives (Novagen), grown to an OD600 of ∼0.4 at 37°C in LB supplemented with 50 μg/ml ampicillin, 50 μg/ml chloramphenicol, 25 μg/ml kanamycin and 10 μM ZnSO4. Recombinant protein production was induced with 0.1 mM IPTG for 2 hours at 33°C. Cell pellets were collected and resuspended in 2.5 ml of media and lysed with 0.25 ml of PopCulture lysis reagent, 2.5 μl of Lysonase bioprocessing reagent (Novagen), and 5 μl of protease inhibitors P8849 (Sigma Aldrich). GST fusion proteins were bound to glutathione-Sepharose 4B resin (Amersham Biosciences) followed by extensive washing in Tris-Saline pH 8 containing 1% Triton X-100, then Tris-saline pH 7.4. GST proteins were eluted with 10 mM reduced glutathione in 50 mM Tris.Cl pH 8, and immediately stored at 4°C in SDS reducing-buffer prior to immunoblotting.
GST pull-down from mammalian cell lysates
Recombinant GST and GST-FHL3 protein were produced by auto-induction of E. coli BL21 DE3 pLysS (Novagen) transformed with pGEX-5x1 (Amersham Biosciences) or pGEX-5x1-FHL3 in Overnight Express Media (Novagen), supplemented with 100 μg/ml ampicillin, and purified using 0.1 ml per 1 ml of culture of PopCulture lysis reagent, 1 μl per 1 ml of culture of Lysonase bioprocessing reagent (Novagen), 2 μl per 1 ml of culture of protease inhibitors P8849 (Sigma Aldrich) and glutathione-Sepharose 4B beads (Amersham Biosciences) according to manufacturer's protocols. Bound recombinant GST and GST-FHL3 protein were washed and stored at 4°C as a 1:5 resin slurry in Tris-saline pH 7.4 containing Roche complete mini cocktail inhibitor tablets and P8849 liquid inhibitors. The concentration of bound GST or GST-FHL3 was quantified by Coomassie-stained SDS-PAGE. C2C12 myoblasts were differentiated for 24 hours and lysates prepared in HEPES lysis buffer with 0.2% or 1% NP-40, then pre-cleared for 2 hours with glutathione-Sepharose 4B resin. 5 μg of GST or GST-FHL3 were incubated for 2-4 hours, rocking at room temperature with C2C12 cell lysates. Resin-bound GST or GST-FHL3 pellets were then washed six times in Tris-saline pH 7.4 or Tris-saline pH 7.4 including 1% Triton X-100 and immunoblotted.
C2C12 myoblasts were seeded in 12-well plates at 2.5×104 cells per well for growth condition assays or 1×105 cells per well for differentiation assays. Cells were cultured for 24 hours in growth media, and transfected with the following plasmids and/or siRNA oligonucleotides in various combinations at the amount indicated.
Mammalian two hybrid: pCMX-Gal4DBD or pCMX-Gal4DBD-FHL3 (50 ng) and pEFBOS-Flag or pEFBOS-MyoD-Flag (500ng), with G5E1b-LUC firefly luciferase reporter vector (500 ng) and pHRLTK renilla luciferase reporter (50 ng). Gal4 FHL3 Overexpression: pCMX-Gal4DBD or pCMX-Gal4DBD-MyoD (50 ng) and pCGN or pCGN-FHL3 (500 ng) with G5E1b-LUC firefly luciferase reporter vector (500ng) and pHRLTK renilla luciferase reporter (50 ng). MCK FHL3 Overexpression: pEFBOS-Flag or pEFBOS-MyoD-Flag (50 ng) and pCGN or pCGN-FHL3 (1 μg), along with pGL3-MCK (50 ng) and pHRLTK renilla luciferase reporter (50 ng). Gal4 FHL3 knockdown: pCMX-Gal4DBD or pCMX-Gal4DBD-MyoD (50 ng) and control siRNA, siRNA FHL3 #2 or siRNA FHL3 #3 (12.5 pmol), with G5E1b-LUC firefly luciferase reporter vector (500 ng) and pHRLTK Renilla luciferase reporter (50 ng). MCK FHL3 knockdown: pEFBOS-Flag or pEFBOS-MyoD-Flag (50 ng) and control siRNA, siRNA FHL3 #2 or siRNA FHL3 #3 (12.5 pmol), with pGL3-MCK (50 ng) and pHRLTK renilla luciferase reporter (50 ng). For growth conditions, transfected myoblasts were maintained in growth media for 48 hours post transfection and harvested. For differentiation conditions, transfected myoblasts were maintained in growth media for 24 hours, switched to differentiation media for 48 hours, then harvested. Lysates were prepared and assayed using the `dual luciferase reporter assay kit' (Promega) according to manufacturer's protocol and analysed on a BMG Labtech Fluostar Optima plate reader. Luciferase values were adjusted for background luminescence and normalised to Renilla luciferase activity to adjust for transfection efficiency. Relative luciferase units were standardised as a fold relative to either Gal4DBD, Vector (HA-vector) and Flag-vector, or control siRNA and Flag-vector expressing samples, as indicated.
Image and statistical analysis
For confocal image based cell counts, a minimum of 4 random fields were scanned for each slide and measured/counted using the public domain ImageJ software (version 1.34 NIH) (Abramoff et al., 2004). For quantitative western blot analysis films were scanned and the band signal intensities determined using ImageJ software. The densitometry values were expressed as a fold level relative to the control, and standardised to corresponding total actin densitometry values obtained from the same sample. Statistical analysis was performed using the unpaired student's t-test. P values of <0.05 were considered significant.
The authors thank Mark Prescott, Lisa Ooms and Anne Kong for helpful suggestions, Monash MicroImaging and Protein Express for technical assistance. This work was funded from a grant from the NHMRC Australia number 284272.
- Accepted February 14, 2007.
- © The Company of Biologists Limited 2007