Cyclin dependent kinase 5, cdk5, is a positive regulator of myogenesis in mouse C2 cells

Skeletal myogenesis is an intensely investigated and well characterized model of differentiation. This results in particular from the use of myoblast cell lines, such as the mouse C2 cell line, which are capable of inducible muscle differentiation ex vivo: following removal of mitogens, proliferative myoblasts cease to divide and through the sequential expression of muscle specific genes, fuse to produce polynucleated myotubes (Yaffe and Saxel, 1977). This process is controlled by a family of myogenic basic helix-loop-helix (bHLH) proteins which include MyoD, myf5, myogenin and MRF4. Whereas MyoD and myf5 are myogenic determination factors, already expressed in proliferating myoblasts, expression of the bHLH myogenic factor myogenin, is an early marker for myogenic differentiation (Wright et al., 1989; reviewed by Weintraub, 1993; Wright, 1992). A number of recent reports have studied the role and fate of key cell cycle regulators, cyclin dependent kinases (cdks), and their regulatory subunits, cyclins, during myogenic differentiation. It has been established that cdk1/cdc2, and cyclins A, B, C and D1 are down-regulated during myogenesis of murine myoblasts in culture (Jahn et al., 1994). In contrast cdk2 and cdk4 are still present, and cyclin D3 is upregulated, in terminally differentiated myotubes (Jahn et al., 1994; Skapek et al., 1995). The implication of cdks and cyclins in myogenesis was subsequently investigated. Overexpression of cyclin D1 both inhibits transcriptional activity of the basic helix-loop-helix myogenic factor MyoD and subsequent myogenic differentiation (Skapek et al., 1995; Rao et al., 1994). Furthermore, the cdk inhibitor p21^cip1 is upregulated during myogenesis in vivo (Parker et al., 1995) or after MyoD overexpression in cell culture (Halevy et al., 1995; Guo et al., 1995).

Another cyclin dependent kinase, cdk5, formerly called PSSALRE, highly homologous with other cdks, is present, although associated kinase activity has not yet been described, in many cell lines from various origins (Tsai et al., 1993), and cdk5 mRNA is widely expressed in human tissues and cell lines (Meyerson et al., 1992). However cdk5 protein is found mostly expressed in brain, becoming activated during neurogenesis (Tsai et al., 1993; for review see Lew and Wang, 1995; Lazaro et al., 1996), and a recent report using a dominant negative mutant of cdk5 describes a clear effect on neurite outgrowth (Nikolic et al., 1996). In human diploid fibroblasts, cdk5 is associated with cyclin D1 and D3 (Xiong et al., 1992; Bates et al., 1994) but no kinase activity has been reported for cdk5-cyclin D complexes. Recently, two specific activators of cdk5, p35^nck5a and p39^nck5ai, which are not related to cyclins, have been cloned (Lew et al., 1994; Tsai et al., 1994; Ishiguro et al., 1994; Tang et al., 1996). Both p35^nck5a and p39^nck5ai show an expression profile restricted to nervous tissue. cdk5-p35 complexes phosphorylate neurofilaments, the microtubule associated protein tau and histone H1 (for review see Lew and Wang, 1995).

While most of the recent studies on cdk5 addressed its expression and regulation during neurogenesis, its potential implication in other cell systems has not been examined. Here, we show that cdk5 protein expression and kinase activity increase markedly during the early phases of myogenic differentiation. Concomitantly with the increase in kinase activity, cdk5 undergoes differential subcellular relocalisation from the cytoplasm to the nucleus. Microinjection studies reveal that overexpression of wild-type cdk5 increases myogenic differentiation. In contrast, inhibition of cdk5 activity by overexpression of a dominant negative mutant of cdk5 clearly inhibited myogenesis. These data establish that cdk5 positively regulates myogenesis.

The mouse C2 cell line (Yaffe and Saxel, 1977) was routinely grown in growth medium: Ham’s F12/DMEM (ratio, 2:1), supplemented with 10% FCS and 2 mM glutamine. Murine fibroblasts C3H10T1/2, REF52 rat fibroblasts and L6 rat myoblasts were cultured in DMEM supplemented with 10% FCS and 2 mM glutamine. To induce differentiation and/or cell cycle arrest C2, C3H10T1/2, REF52, and L6 cells were plated at a density of 1 to 5×10⁴ cells/cm² on plastic dishes and grown for 2 days before replacing growth medium with differentiation medium (DMEM supplemented with 2% FCS and 2 mM glutamine).

C2 cells at different stages and fibroblasts grown for 48 hours in high (10%) or low (2%) foetal calf serum, were rinsed in PBS and lysed in 1% SDS, 40 mM Tris-HCl, pH 6.8, and 7.5% glycerol (0.5 ml/100 mm dish). Adult mouse brain and muscle were lysed in a Potter homogeniser in the same buffer. After passing five times through a 26 gauge needle, protein concentrations of cell extracts were determined by the Bio-Rad Protein Assay, using bovine serum albumin as standard. After quantification, protein extracts were diluted in an equal volume of 2× Laemmli buffer and 100 μg of protein loaded per lane on 12.5% SDS-PAGE prior to transfer to nitrocellulose. Nitrocellulose sheets were saturated with 3% non-fat dried milk in PBS for one hour and incubated with primary antibody in PBS, 1% BSA for one hour. Dilutions used for primary antibody were: 1:500 for C8 (an antibody directed against the carboxy-terminal region of cdk5); 1:500 for DC17 antibody (Santa Cruz Biotechnology-Tebu, Le Perray en Yvelines, France). Anti-myogenin hybridoma supernatant (generously provided by Dr W. Wright, University of Dallas, Dallas, Texas) was diluted 1:10, monoclonal anti-troponin T antibody (Sigma) was diluted 1:2,000, Monoclonal anti-α-tubulin (DMA1A, Amersham, Les Ulis, France) was diluted 1:10,000. After several washes in PBS containing 0.01% Tween-20, horseradish peroxidase-conjugated antirabbit or anti-mouse antibody (Amersham) were added (dilution 1:10,000) for one hour. The immune signal was detected using the ECL Western Blot detection system (Amersham).

C2 cells were rinsed twice with PBS and lysed in 50 mM Tris-HCl, pH 7.4; 50 mM NaF, 150 mM NaCl, 0.4% NP40, 2 mM DTT, 2 mM NaVO₄, 10 mM β-glycerophosphate, and protease inhibitors: 4 μM leupeptin, 2 μg/ml aprotinin and 3 μM pepstatin (Fluka Chem. Corp., Ronkonkoma, NY). After passing five times through a 26 gauge needle and 30 minutes incubation at 4°C, insoluble material from the cell lysates was pelleted by 5 minutes centrifugation at 10,000 rpm. Briefly, in each experiment, 10 μl of C8 anti-cdk5 antibody (Santa Cruz Biotechnology) and 30 mg of Protein A-Sepharose 4B conjugate (Pharmacia, St Quentin Yvelines, France) were added to 600 μl of C2 cells extract (corresponding to 300 μg of protein) in lysis buffer. After 60 minutes incubation at 4°C with gentle shaking, immunoprecipitates were washed by consecutive cycles of incubation/centrifugation, twice 1 minute in 1 ml lysis buffer and twice 1 minute in 1 ml PBS. In competition experiments, diluted antibodies were preincubated for 30 minutes with 100 μM of the immunogenic peptide. Immunoprecipitates were then incubated for 40 minutes at room temperature with 25 μl of histone H1 kinase buffer containing 0.2 mg/ml histone H1 (Boerhinger Mannheim, Meylan, France), 20 mM Hepes, pH 7, 2 mM MgCl₂, 50 μM [γ−³²P]dATP) and 10 μCi [γ−³²P]dATP. Kinase reactions were stopped by boiling the immunoprecipitates for 4 minutes in an equal volume of 2× Laemmli buffer. After electrophoresis (15% SDS-PAGE), histone H1 associated radioactivity was analysed and quantitated using the ImageQuant software package on the phosphorImager system (Molecular Dynamics, Mt View, CA).

Indirect immunofluorescence experiments were performed with C2 cells. After brief washing in PBS, cells were fixed in 3.7% formalin in PBS for 5 minutes, extracted in −20°C acetone for 1 minute and incubated in PBS-BSA 1% for 5-10 minutes. The cells were then stained with rabbit polyclonal anti-cdk5, C8 antibody (SantaCruz Biotechnology) diluted 1:50 in anti-myogenin hybridoma supernatant (a kind gift from W. Wright) for 60 minutes at 37°C. Alternatively, cells stained for cdk5 with C8 were co-stained with anti-troponin T monoclonal antibody (Sigma, St Quentin Fallavier, France) diluted 1/200. After washing in PBS, biotinylated goat anti-rabbit IgG (Amersham) and fluorescein-conjugated anti-mouse antibody (Cappel) diluted 1:100 and 1:40, respectively, in PBS-BSA were incubated for an additional 60 minutes. Thereafter, cells were washed in PBS, and incubated with Texas red-conjugated streptavidin (Amersham) diluted 1:200 in PBS for 30 minutes. Cells were stained for DNA using Hoechst (Sigma). In competition experiments, the diluted antibodies were preincubated for 30 minutes with 100 μM of the immunogenic peptide. Coverslips were washed and mounted for microscopy as previously described (Vandromme et al., 1992).

After immunofluorescence treatment, cells were observed on an Axiophot microscope (Carl Zeiss, Le Pecq, France) using a planapochromat ×40 objective. Fluorescent images were either directly photographed using a Kodak DCS420 professional digital color camera and images were acquired under Adobe photoshop as TIFF format, or analysed on a confocal laser scanning microscope (CLSM): dualchannel detection was performed using the Leica CLSM (Leica, Heidelberg, FRG) equipped with a Krypton-Argon ion laser using two major emission lines at 488 nm for FITC excitation and 568 nm for rhodamine or Texas-red excitation, respectively. A ×40 or a ×63 planapochromat lens was used and untreated images were directly transferred from the Leica Motorola 68040 to a Silicon Graphics IRIS Indigo workstation (RS3000) (Silicon Graphics, Mt View, CA). Images were deconvoluted, remapped and converted to SGI raster format using convert file (S. Guihem and N. Lamb, unpublished). Fluorescence measurements were made as follows: images from stained cells were acquired on the CLSM with a fixed value for pinhole size, laser power and photomultiplier sensitivity. These values were such that the brightest images values were within the maximum threshold level (12 bit). Figures using image files from either the Kodak camera or the CLSM were assembled completely under SGI showcase 3.21 and printed directly as postscript files using a Kodak Colorease thermal sublimation printer.

Human full length cdk5, a kind gift from Dr L.-H. Tsai, was subcloned into PJ3Ω expression vector (Morgenstern and Land, 1990) containing the SV40 promoter as a SacI-ClaI fragment to obtain PJ3-cdk5wt plasmid. Aspartic acid in position 144 was changed into an asparagine by oligonucleotide directed mutagenesis using the mutagenic primer 5′GAAATTGGCTAATTTTGGCCTGGCC3′. The selection primer used, 5′GCTTCGATCGCCGACTCTAGAGG3′, inactivates the unique SalI site in a non coding sequence of PJ3-cdk5wt plasmid. Mutagenesis was carried out with the Transformer Site-directed Mutagenesis kit (Clontech, Palo Alto, CA, USA) using the conditions recommended by the manufacturer. Colonies were screened with the mutagenic primer and positive clones were checked by sequencing by the di-deoxy termination procedure with primer ACATCCG-GTGGGCGGTACC, using the T7 Sequencing kit (Pharmacia).

In order to obtain expression of recombinant proteins in mouse C2 cells, the cells were microinjected with either 0.1 μg/μl of DNA of PJ3-cdk5wt, PJ3-cdk5dn, or empty PJ3Ω plasmids prepared by caesium chloride gradient centrifugation. Control plasmid PJ3Ω was coinjected with an inert rabbit antibody (1 mg/ml) to serve subsequently as a marker in identifying injected cells. After microinjection cells were returned to the incubator 10 hours before transfer to differentiation medium (2% foetal calf serum in DMEM). After 48 hours or 72 hours of differentiation, cells were fixed and stained as described above for cdk5, myogenin and troponin T. Injected cells were recorded as cdk5 overexpressing cells (for PJ3-cdk5wt or PJ3-cdk5dn injections) or inert marker antibody containing cells (for PJ3Ω injections). Injections were performed in an area delimited before the injection with a scalpel. PJ3-cdk5dn, PJ3-cdk5wt, and control PJ3Ω plasmid injections were performed on different areas of the same dishes.

To investigate the potential involvement of cdk5 in myogenic differentiation, we initially looked for the presence of cdk5 protein in adult mouse muscle and C2 myoblasts (Yaffe and Saxel, 1977) by western blot analysis using the DC17 monoclonal antibody (Tsai et al., 1994). Because cdk5 is reportedly highly expressed in mammalian brain (Lew and Wang, 1995), we used an adult mouse brain extract as a control for cdk5 expression.

As shown in Fig. 1A, cdk5 can be identified by western blotting in both adult mouse skeletal muscle and C2 myoblasts, where a single band is recognised, with a molecular mass of 31 kDa similar to that observed in mouse brain extracts. We next examined if changes in cdk5 expression could be observed during the course of myogenic differentiation using the C2 cell line, a well defined model for ex vivo differentiation: these cells proliferate as myoblasts in high serum concentration and can be induced to differentiate by reducing the serum concentration from 10% to 2% (see Materials and Methods). Under our culture conditions, C2 myoblasts fuse into myotubes within 60-72 hours with an efficiency of 70 to 80% (Vandromme et al., 1992). The expression of cdk5 protein during the course of C2 cell differentiation was investigated by western blot analysis of protein extracts taken at different times after serum reduction. To follow the process of differentiation, we monitored for the induction of myogenin, which marks the entry of myoblasts into the differentiation pathway (Wright et al., 1989), and troponin T expression, a skeletal muscle marker.

As shown in Fig. 1B, cdk5 is already present in myoblasts (lane P, top) whereas no myogenin and troponin T expression can yet be detected (lane P, middle and bottom). An increase in cdk5 expression occurs as soon as 12 hours after switching cells into the differentiation medium and the highest level of expression is reached after 48 hours of differentiation, at which time densitometric scanning of the ECL signal shows a 4-fold increase over the level in proliferative myoblasts. By this time myogenin is also expressed at its maximal level showing that the majority of the cells are committed to the differentiation pathway. After 72 hours of differentiation, cdk5 levels are slightly reduced in terminally differentiated myotubes (75% of maximal level as checked by densitometry scanning of the ECL blot, data not shown), although remaining at higher levels than in proliferative cells. To address the question of a specific link between the cdk5 upregulation and the process of differentiation, we examined if similar changes occurred in another myogenic cell line, rat L6 myoblasts (Yaffe, 1968). As a control, to show that these effects did not simply reflect a response to low serum medium culture conditions, we examined cdk5 expression in two non-myogenic lines: C3H10T1/2 and REF52 fibroblasts. After plating, the different cell lines were allowed to grow to 80% confluence in 10% foetal calf serum, before switching into differentiation medium, containing only 2% serum. Total cell extracts (without a centrifugation clearing step) were made after 48 hours in differentiation medium, a time at which maximum increases in cdk5 levels were observed in C2 cells (Fig. 1B). As shown in Fig. 1C, an increase in cdk5 expression occurs after low serum treatment in myogenic L6 cells. In contrast, exposure to low serum did not result in any increase in cdk5 expression in either C3H10T1/2 or REF52 fibroblasts (rather a slight decrease is detectable in C3H101/2). Total removal of serum, to ensure complete growth arrest of 10T1/2 fibroblasts, led to a clear decrease in cdk5 levels. These changes in cdk5 protein levels were not due to variations in sample loading as confirmed by parallel blotting for tubulin (Fig. 1C). All the western blotting analyses were also confirmed using the C8 anti-cdk5 polyclonal antibodies (data not shown). Together, these data strongly argue for a direct link between the induction of skeletal muscle differentiation and the up-regulation of cdk5 protein expression.

We next assessed if cdk5 proteins expressed during differentiation had associated histone H1 kinase activity. For these experiments, we immunoprecipitated cdk5 protein from proliferative or differentiating C2 cells at different stages of myogenesis. As shown in Fig. 2A and B, extracts of C2 cells immunoprecipitated with anti-cdk5 C8 antibodies display low histone H1 kinase activity during proliferation and at terminally differentiated stages. However, immunoprecipitated cdk5 H1 kinase activity increased notably during differentiation, between 24 and 48 hours, reaching a maximum level at 48 hours (17-18-fold greater than the level in proliferating myoblasts), after which it decreases to a level approximately 3-fold higher than that observed in proliferative myoblasts. Preincubation of the cdk5 C8 antibody with the immunogenic peptide prior to use in immunoprecipitation completely abolished the immunoprecipitation of H1 kinase activity (data not shown). The cell extracts used in these immunoprecipitation experiments are the same as those analysed for cdk5, troponin T and myogenin by western blot in Fig. 1B and western blot analysis of the immunoprecipitates showed a similar profile for cdk5 to that found in total cell extracts (data not shown).

Several such measurements of immunoprecipitable cdk5 H1 kinase activity in the course of C2 cell differentiation all confirmed a low activity in myoblasts, a peak of activity between 36 and 48 hours of differentiation and lower activity in fused myotubes (60 to 90 hours). These data show that the increase in cdk5 protein during myogenic differentiation is accompanied by a marked increase in cdk5 activity, which suggests that cdk5 may be involved in myogenic differentiation.

The expression and cellular distribution of cdk5 were also investigated by indirect immunofluorescence using affinity-purified rabbit anti-cdk5 C8 antibody. The stage of differentiation was assessed in parallel by double immunofluorescence using either mouse monoclonal anti-troponin T or antimyogenin antibodies. As shown in Fig. 3, proliferative myoblasts show a low level, essentially cytoplasmic staining pattern for cdk5 in proliferative cells (Fig. 3A). cdk5 becomes redistributed and can be stained in the nucleus as early as after 12 hours of differentiation (Fig. 3D). This increased nuclear staining is visible before the synthesis of troponin T (Fig. 3D-E), since cells which are still negative for this myogenic marker after 12 hours of differentiation have clearly defined nuclear cdk5 staining. To analyse more precisely these early changes in cdk5 staining, the cellular distribution of cdk5 and myogenin were examined by confocal scanning of 0.4 μm sections during the early stages of myogenesis (Fig. 4). In myoblasts cdk5 is clearly predominantly cytoplasmic although some staining is already detectable in the nucleus (Fig. 4A). After moving to differentiation medium, cells show a specific increase in staining of the nuclear compartment (Fig. 4C) together with relatively unchanged levels of cytoplasmic staining. This apparent increase in nuclear staining also occurs before the expression of myogenin (a field without myogenin expression is shown at 12 hours, to show the early pattern of cdk5 distribution). After 24 hours of differentiation, the nuclear staining for cdk5 clearly increases to become predominant (Figs 3G, 4E). At this time, as differentiation proceeds, there appears to be a correlation between the extent of cdk5 staining and the presence of myogenin or troponin T. Those cells which show the highest levels of troponin T staining after 24 of differentiation also show intense nuclear and cytoplasmic staining for cdk5 (Fig. 3G and H, see arrows). This is also the case for myogenin, where a close correlation exists between the levels of cdk5 staining and staining for this marker (Fig. 4E-F, see arrows). The same cytoplasmic to nuclear change in cdk5 staining was observed with monoclonal antibody DC17 (data not shown). Importantly, at all stages of differentiation, the signal for cdk5 staining in both the cytoplasmic and nuclear compartments was completely abolished when the C8 antibody was pre-incubated with antigenic peptide prior to use (Fig. 4, see right panel for stage 24 hours). These data show that, in addition to changes in expression and kinase activity, cdk5 undergoes specific alterations in cytolocalization as cells transit through early differentiation.

To examine the potential role of cdk5 during myogenesis, we over-expressed either wild-type (cdk5wt) or dominant negative mutant forms (cdk5dn) of the protein (mutated in the ATP-binding site; see Materials and Methods, and Van den Heuvel and Harlow, 1993) in proliferating myoblasts and, following induction of differentiation, estimated the extent of differentiation in overexpressing cells. cdk5dn shows similar affinities to the wild type for the binding of regulatory subunits (Tsai et al., 1994; J.-B. Lazaro, unpublished observations) and inhibits neurite outgrowth when overexpressed in differentiating neuronal cells (Nikolic et al., 1996). The immunoreactivity of cdk5dn protein was neither modified nor reduced, as shown by microinjection of cdk5wt and cdk5dn encoding plasmids (respectively, PJ3-cdk5wt and PJ3-cdk5dn) in C2 cells and staining of overexpressed proteins using anti-cdk5 antibodies (see Fig. 5).

To investigate the role of cdk5 during myogenesis, we independently microinjected these two plasmids, PJ3-cdk5wt and PJ3-cdk5dn into growing C2 myoblasts and monitored myo-genesis through expression of myogenin and troponin T. Cells were microinjected 10 hours before serum withdrawal to allow the plasmids sufficient time to express and subsequently fixed after 48 or 72 hours of differentiation (58 or 82 hours after microinjection). The overexpression of cdk5 was followed by indirect immunofluorescence using C8 antibody. To quantify the extent of differentiation we stained for either myogenin or troponin T expression in parallel. These two time points were chosen because: (1) 48 hours of differentiation allow us to observe both an increase or an inhibition in the extent of myo-genesis; and (2) 72 hours of differentiation allow us to better investigate for an effect on cell fusion as well as on expression of muscle markers.

Figs 5 and 6 shows representative examples of cells from both experiments. Injected cells are highlighted by arrows. The expression of cdk5wt and cdk5dn proteins is shown in Figs 5C,F,I and 6B,E. Clearly, in all cases, cdk5 proteins are over-expressed to a level significantly higher than that in surrounding uninjected cells. Quantitative analysis of the levels of expression in injected and surrounding cells show that cells injected with PJ3-cdk5wt or PJ3-cdk5dn over-express 6- to 10-fold the levels of cdk5 found typically in non-injected cells (data not shown). For this reason, endogenous cdk5 staining is only weakly visible in the micrograph shown. Fig. 5B,E and H shows the expression of troponin T in the injected (and surrounding) cells and Fig. 6C and F shows the expression of myogenin. To determine the effect of cdk5wt and cdk5dn on the expression of myogenic markers, we counted the percentage of cells overexpressing either form of cdk5 that were stained positively for myogenin or troponin T. A summary of these data is presented in Fig. 7. The controls were carried out by injecting empty PJ3-Ω plasmid together with the inert injection marker IgGs and counting the percentage of myogenin and troponin T expression in injected cells (Fig. 7). These values were very similar to the levels of myogenic marker expression in non-injected cells.

As shown in Fig. 7 overexpression of cdk5wt significantly increased the ratio of cells expressing myogenin or troponin T after 48 hours when compared to the percentage of expression of these markers in cells injected with a control plasmid. This effect is not visible after 72 hours of differentiation when both the control and PJ3-cdk5wt injected cells have attained the maximum extent of differentiation. It should be noted that when PJ3-cdk5wt was microinjected into sparse proliferative C2 myoblasts and the cells kept in growth medium for 48 hours, no myogenic marker expression was induced (data not shown). This result indicates that cdk5 alone is not sufficient to drive proliferative myoblasts into the differentiation pathway and additional events are required outside of cdk5. In the complementary experiments (see Fig. 7 and Fig. 5D-I and Fig. 6D-F), overexpression of the dominant negative mutant of cdk5, cdk5dn, in C2 myoblasts resulted in a significant inhibition of both myogenin and troponin T expression in C2 cells induced to differentiate. In comparison to cells injected with the control plasmid, the expression of myogenin and troponin T in cells injected with the PJ3-cdk5dn was reduced 3- and 5-fold, respectively, after 48 hours of differentiation in comparison to cells injected with control plasmid (Fig. 7). Figs 5 and 6 show that whereas troponin T and myogenin are detected in cdk5wt overexpressing cells (Figs 5B and 6B, respectively), cdk5dn overexpressing cells failed to express these myogenic markers (Fig. 5E and H for troponin T and Fig. 6E for myogenin). This inhibitory effect of the dominant negative mutant of cdk5 appears even more clearly after 72 hours of differentiation where 90% inhibition is observed compared to the level of troponin T expression in control or PJ3-cdk5wt injected cells (Fig. 7B, effect illustrated in Fig. 5G-I). However, when myoblasts where injected with cdk5dn only after they had been moved into differentiation medium, i.e when they had already started to differentiate, no significant inhibitory effect was observed on muscle marker expression nor on myoblast fusion in cdk5dn overexpressing cells. Together these data imply that cdk5 is a positive modulator of myogenesis.

The present study reveals several new findings: firstly, that the previously named ‘neuronal cdc2 like kinase’, cdk5, is present in adult mouse muscle; secondly, that both the expression and kinase activity of cdk5 are increased during the course of myo-genesis in murine C2 myoblasts. Thirdly, cdk5 protein is relocated into the nucleus during myogenesis and, finally, over-expression of cdk5 protein in myoblasts enhances the number of cells undergoing myogenesis whereas abolition of cdk5 kinase activity through overexpression of a dominant negative mutant markedly inhibits myogenesis in C2 cells. These results represent the first demonstration of a positive effect by a cyclin-dependent protein kinase, cdk5, on the process of myogenesis. Given its role in neurogenesis, the presence of cdk5 in mouse skeletal muscle led us to investigate its possible implication in myogenesis. The increased expression and activity of cdk5 that we observed during the course of myogenesis differs markedly from the pattern for other cdks. cdc2/cdk1 mRNA is down regulated and cdk2 and cdk4 mRNA levels remain constant during the course of myogenesis of C2C12 cells, a subclone of C2 cells (Jahn et al., 1994; Kiess et al., 1995). At the protein level cdk4 is slightly downregulated during myogenesis of C2C12 cells (Andrés and Walsh, 1996). Contrasting with these results, our data suggest a unique role for cdk5 during myogenic differentiation.

Interestingly, the passage of cdk5 into the nucleus occurs during very early stages of differentiation when myogenic markers like troponin T or even myogenin are not yet detectable. Moreover, at later stages (24-36 hours), simultaneous immunofluorescence staining for cdk5 and either troponin T or myogenin shows that cells positive for the myogenic markers contain high levels of both nuclear and cytoplasmic cdk5. As such, the increase in cdk5 protein level is clearly linked with the appearance of myogenic markers when observed at a single cell level. cdk5 kinase activation appeared closely temporally related to the increased expression of cdk5, both peaking after 36-48 hours of differentiation, which is after the initial entry of the protein into the nucleus during early myogenesis. However, whereas the level of cdk5 protein goes up by 3- to 4-fold (as assessed by densitometry scanning of the ECL blot signals), the H1-kinase activity of cdk5 shows a 17-fold increase after 48 hours of differentiation (see Fig. 2B), indicating that an activation of cdk5 kinase takes place at this time (approximately 4-fold). In addition, as confirmed at later times of differentiation, the nuclear localisation of cdk5 and its activation are not correlated: by 72 hours of differentiation, in differentiated myotubes, the predominant nuclear localisation of cdk5 is accompanied with a low level of kinase activity. Therefore two phases for the regulation of cdk5 can be distinguished during myogenesis: (1) an early partial recruitment of cdk5 into the nucleus; and (2) an increase in cdk5 expression with a sharp increase in activation of cdk5 protein kinase concomitant with the appearance of differentiation markers. When compared to the different stages of myogenic differentiation defined in parallel with myogenin and p21 expression during the course of C2C12 myoblast differentiation (Andrés and Walsh, 1996), cdk5 nuclear recruitment must occur in the earliest phase of myogenesis since it is observed before myogenin expression.

To examine the potential role of cdk5 in myogenesis, we chose to inhibit or over-express cdk5 through plasmid microinjection into C2 myoblasts and monitor the extent of differentiation via staining for expression of early and later myogenic markers, myogenin and troponin T, respectively. Inhibition of cdk5 was achieved by overexpression of a dominant negative mutant form of the kinase, cdk5dn. This mutant bears a substitution of aspartic acid at position 144 to asparagine, shown to yield an inactive kinase (Van den Heuvel and Harlow, 1993). In cAMP dependent protein kinase, the homologous residue is necessary for the phosphate transfer (Taylor et al., 1993). The cdk5dn mutant associates with p35, the brain specific activator of cdk5, but fails to form an active kinase (Tsai et al., 1994) and we observed that cdk5dn and cdk5wt associate similarly with cyclin D1 (data not shown). More recently, it was reported that overexpression of this same mutant in primary cultures of developing neurons inhibited neurite outgrowth, establishing that cdk5 is required for neurogenesis (Nikolic et al., 1996). When overexpressed in cells, cdk5dn is thought to compete with endogenous cdk5 for association with positive regulatory subunits and as such inhibit cdk5 kinase activity. Although one may hypothesize that cdk5dn could block interactions of other cdks with cyclin subunits, thus perturbating the total cyclincdk pool rather than just the cdk5 activity, this possibility can be ruled out by the following data: (1) Van den Heuvel and Harlow (1993) have shown that overexpression of this same cdk5dn mutant in different cell lines did not perturb their cell cycle (whereas the same authors found that cdk2dn and cdc2dn did block the cell cycle). (2) No activity has been shown for cdk5wt associated with a different cyclin subunit and yet we observed that overexpression of cdk5wt in myoblasts did not inhibit myogenesis, but enhanced it (see below). Our data show that overexpression of cdk5dn markedly inhibited myogenesis as assessed by the expression of both myogenin and troponin T. Interestingly, we have observed no effect of cdk5 overexpression on p21, assessed by immuno-staining for p21 in cdk5 overexpressing cells (data not shown). Since we have observed that cdk5dn and cdk5wt bind cyclin D1 equally, their opposite effects on myogenesis exclude the possibility that their effects are due to non-specific squelching of cyclin subunits. Thus, specific inhibition of cdk5 activity led to a dramatic decrease in the rate of myogenesis.

Although our data show that cdk5 is implicated in myogenesis, cdk5wt overexpression alone was unable to trigger myogenic differentiation in growing C2 cells, implying that the activation/inactivation of other pathways are required for the induction of differentiation. However, we observed a clear increase in the extent of myogenin and troponin T expression in cells overexpressing cdk5wt after 48 hours of differentiation, compared to control injections (see Fig. 7). Because cdk5dn and cdk5wt have opposite effects on differentiation and the same properties in binding subunits, it implies that the inhibition of myogenesis by cdk5dn is due to the loss of kinase activity. The activation of cdk5 during myogenesis may occur by interaction of cdk5 with positive regulators. It is known that cdk5 can associate with cyclins D1 and D3 in several cultured cells and with p35^nck5a and p39^nck5ai in brain (Bates et al., 1994; Xiong et al., 1992; Tsai et al., 1994; Tang et al., 1996). However, no histone H1 or pRb kinase activity has been detected in complexes formed between cdk5 and D type cyclins. Additionally, the inhibition of myogenesis that we observed with cdk5dn suggests that it recruits an endogenous activator (cyclin or p35 like protein) in a similar way to that in which overexpression of the mutant inhibits neurite outgrowth by recruiting the neuronal activator p35 (Nikolic et al., 1996). Thus an important question remains: the identity of the myogenic cdk5 regulatory subunit(s).

Another essential question to elucidate the pathway in which cdk5 acts in myogenesis is the identification of the protein substrates phosphorylated by cdk5. The brain cdk5-p35 proline directed kinase phosphorylates many residues at the same sites as cdc2-cyclin B kinase (Lew and Wang, 1995). For example, neurofilaments and the microtubule associated protein Tau are phosphorylated in vitro, in a similar manner by these two kinases, although in mature postmitotic neurones cdc2 is absent when phosphorylation of neurofilament and Tau occurs (Kobayashi et al., 1993; Lew et al., 1992a,b). Interestingly, one report has described the phosphorylation of the giant cytoskeletal protein titin during myogenesis at KSP motifs sharing sequence similarity to motifs phosphorylated in neurofilaments (Gautel et al., 1993). Although this question was not addressed in this report, an attractive possibility is that cdk5 may be responsible for such a phosphorylation activity. More recently, a report by Tajbakhsh et al. (1994), described a population of myogenic cells derived from the mouse neural tube, which coexpress both neurone and muscle-specific markers, raising the possibility of overlapping regulatory pathways between myogenesis and neurogenesis.

Together, our data show for the first time the up-regulation in expression and activity and corresponding changes in subcellular localisation of cdk5 during the course of myogenesis and demonstrate clearly that cdk5 is a positive regulator of the differentiation process. The novelty of these findings opens up numerous attractive directions for future investigation.

We thank Dr Woody E. Wright for his generous gift of F5D antimyogenin antibody. We thank Dr Li-Huei Tsai and Dr Ed Harlow for providing cdk5 encoding plasmid, and Dr Daniel Fisher for critical reading of the manuscript. J.-B. L. thanks Simon Gallas for initial advice on dominant negative mutant strategy. We gratefully acknowledge the support of Association Française contre les Myopathies (A.F.M.) and Association pour la Recherche contre le Cancer (grant no. 1306).