(Received for publication, May 8, 1996, and in revised form, October 14, 1996)
From the Division of Cardiovascular Research, St. Elizabeth's Medical Center, Tufts University School of Medicine, Boston, Massachusetts 02135 and Department of Biochemistry, Case Western Reserve University, School of Medicine, Cleveland, Ohio 44106
During skeletal myogenesis, cell cycle withdrawal accompanies the expression of the contractile phenotype. Here we show that ectopic expression of each D-type cyclin is sufficient to inhibit the transcriptional activation of the muscle-specific creatine kinase (MCK) gene. In contrast, ectopic expression of cyclin A or cyclin E inhibits MCK expression only when they are co-expressed with their catalytic partner cyclin-dependent kinase 2 (Cdk2). For each of these conditions, myogenic transcriptional inhibition is reversed by the ectopic co-expression of the general Cdk inhibitor p21. Inhibition of MCK expression by cyclins or cyclin-Cdk combinations correlates with E2F activation, suggesting that the inhibition is mediated by the overall Rb-kinase activities of the Cdk complexes. In support of this hypothesis, a hyperactive mutant of Rb was found to partially reverse the inhibition of MCK expression by cyclin D1 and by the combination of cyclin A and Cdk2. These data demonstrate that the inhibition of myogenic transcriptional activity is a general feature of overall Cdk activity which is mediated, at least in part, by an pocket protein/E2F-dependent pathway. MCK promoter activity is also inhibited by ectopic E2F1 expression, but this inhibition is not reversed by the co-expression of p21. Analyses of a series of E2F1 mutants revealed that the transcriptional activation, leucine zipper, basic, and cyclin A/Cdk2-binding domains are dispensable, but the helix-loop-helix region is essential for myogenic inhibition. These data demonstrate that myocyte proliferation and differentiation are coordinated at the level of E2F and that these opposing activities are regulated by different E2F domains.
Irreversible cell cycle withdrawal is a key component of myogenic differentiation, but little is known about the interplay between the myogenic transcription factors and the cell cycle regulatory proteins. Previous studies have shown that the retinoblastoma susceptibility gene product (Rb)1 has an essential role in maintaining the postmitotic state of differentiated myotubes (1, 2). Rb is hypophosphorylated upon terminal differentiation and this state of dephosphorylation is maintained when differentiated myotubes are re-exposed to high mitogen media. Hypophosphorylated Rb inactivates the E2F transcription factor which is essential for the expression of genes required for DNA synthesis (3). Myocytes from Rb-deficient mice can differentiate into myotubes, but these myotubes can resynthesize DNA upon mitogen stimulation (2). The phosphorylation status of Rb is regulated, at least in part, by the activity of cyclin-dependent kinases including Cdk2 and Cdk4 (4). Recently, the general Cdk inhibitor p21 was shown to be dramatically induced during skeletal muscle differentiation, and a high level of p21 expression is sustained when myotubes are re-exposed to high mitogen media (5, 6, 7). p21 is induced early during the differentiation program (8), and its expression is critical for myocyte viability (9). In myotubes, p21 is the predominant inhibitory subunit of the Cdk complex (7, 10). Thus p21 is likely to function to inhibit the phosphorylation of Rb or other Rb family members during myocyte differentiation.
Surprisingly, it has been reported that overexpression of cyclin D1, but not cyclins A, B, D2, D3, or E, can block myogenesis as detected by the transcriptional activation of an MCK-reporter construct in transiently transfected cultures of differentiating C2C12 cells (11). More recently it was reported that the cyclin D1-mediated inhibition of myogenic transcriptional activation could be reversed by the co-expression of the Cdk inhibitors p21 or p16 (12). These data have led to the proposal that cyclin D1 down-regulation may be a nodal point in the coordination of myocyte differentiation and cell cycle activity. Further, it was proposed that myogenic inhibition by cyclin D1 occurs through an Rb-independent pathway because the overexpression of cyclin A or cyclin E, the regulatory subunits of the Cdk2 Rb-kinase, do not inhibit myogenesis (12).
To test the hypothesis that cyclin D1 functions uniquely to coordinate cell cycle with myogenic transcription through an Rb-independent mechanism, we analyzed the effects of additional cell cycle regulatory molecules on the same MCK transcriptional assay that was employed in these previous studies (11, 12). Here we report that in addition to cyclin D1, the overexpression of other D-type cyclins (D2 and D3) or the combinations of Cdk2 and cyclin A or cyclin E also inhibit MCK transcriptional activation, and that this inhibition is reversed by the co-expression of p21. MCK transcription is inversely correlated with expression from an E2F-reporter gene construct, and a hyperactive Rb mutant partially reverses the cyclin-Cdk-mediated inhibition of MCK transcription. Collectively these data indicate that the inhibition of myogenic transcription by cell cycle components can be mediated by a pocket protein/E2F pathway. Myogenic transcription is also inhibited by the ectopic expression of the E2F1 transcription factor. Surprisingly, mutations in the activation, leucine zipper, basic, and cyclin A-Cdk2 binding domains of E2F1 have no effect on its ability to inhibit myogenic transcription, but this activity was abolished by mutations in the E2F1 helix-loop-helix region. Taken together, these data demonstrate that the inhibition of myogenesis is a general feature of cyclin-Cdk activity and that myogenic differentiation is coordinated with cell cycle activity at the level of E2F.
C2C12 myocytes and 10T1/2 fibroblasts were maintained in DMEM supplemented with 20% fetal bovine serum (growth medium). Myogenic differentiation was initiated by shifting subconfluent cultures into DMEM supplemented with 2% horse serum (differentiation medium).
The (650)MCK-Luc plasmid that contains the rabbit MCK promoter and
enhancer was described previously (13, 14). All expression plasmids
used in these studies were under the control of the CMV promoter/enhancer. The expression plasmids for cyclin D1, D2, D3, A, E,
and Cdk2 were from L. Zhu and E. Harlow. The E2F1 expression vectors
utilized the CMV promoter/enhancer (15, 16). E2F1(E113) has glutamate
substituted for arginine at position 113; E2F1(E120) has glutamate
substituted for lysine at position 120; E2F1(E138) has glutamate
substituted for phenylalanine at position 138; and E2F1(E177) glutamate
substituted for isoleucine at position 177. E2F1(d113-120) was derived
by deleting amino acid positions 113 through 120, and E2F1(d87) was
derived by deleting amino acids 1 through 87 (15). The deletion
mutations E2F1(1-241), E2F1(1-196), and E2F1(1-127) were described
in Qin et al. (16). The expression plasmids for p21,
pCDNAIII-p21, was from A. Dutta (17). K. Wills provided the
expression plasmids for hyperactive and inactive Rb, p56 and p56/H209,
respectively. The (E2F)x4-E1bTATA-Luc reporter plasmid was constructed
by subcloning the PvuII/SacI fragment from
(E2F)x4-E1BTATA-CAT (18, 19) into the SmaI/SacI
site of pGL2-Basic plasmid (Promega Inc.). The plasmid MCK E-box-Luc, that has the luciferase gene downstream from the E1b TATA element and
four copies of the high affinity E-box element from the MCK enhancer,
was provided by A Yee. In all transfection experiments the plasmid
pSV2-A.P, that has the alkaline phosphatase gene under the control of
the SV40 promoter and enhancer (20), was used to control for
differences in transfection efficiencies.
Plasmid transfections were performed by the calcium-phosphate
procedure. For transfections in Fig. 1, calcium-DNA precipitates were
incubated with cells in growth medium for 12 h, and then cells
were switched to DMEM supplemented with 20% fetal bovine serum for 1 day. Cells were either harvested and assayed for luciferase and
alkaline phosphatase activities or switched to differentiation medium
(DMEM supplemented with 2% horse serum). At indicated times after
incubation in low mitogen differentiation medium, cells were harvested
and assayed for reporter expression. Luciferase and alkaline
phosphatase activities were determined as described previously (21).
Luciferase (Promega) and alkaline phosphatase (CSPD chemiluminescent
substrate; Tropix) activities were measured with a Berthold Lumat
LB9501 luminometer. The amount of DNA was kept constant in each set of
transfection experiments by adding the required amount of Rc/CMV
plasmid (Promega) as described in Guo et al. (7). All
transfection experiments were performed in triplicate. For transient
transfections in other figures, transfections were performed as above
except that 12 h after transfection, cells were switched to fresh
DMEM, 20% fetal bovine serum for 1 day, and then switched to low
mitogen differentiation medium for 2 days. Cells were harvested, and
luciferase/alkaline phosphatase activities were assayed as described
above.
Immunoblotting
To induce myogenic differentiation, proliferating C2C12 myoblasts were switched to low mitogen differentiation media (DMEM supplemented with 2% horse serum) for 2 days. Cells were lysed in buffer (50 mM Tris, pH 8.0, 150 mM NaCl, 5 mM EDTA, 0.5% Nonidet P-40, 1 µg of leupeptin/ml, and 1 mM phenylmethylsulfonyl fluoride) and passaged through G21 needles several times. Lysates were cleared of insoluble materials by centrifugation at 4 °C and 12,000 × g for 15 min. Protein concentrations of lysates were determined by using Bradford assay (Bio-Rad). 50 µg of cell lysates were loaded into 12% SDS-PAGE gel after boiling in SDS sample loading buffer. Following electrophoresis, proteins were transferred to nylon membranes (Millipore). Membranes were blocked in buffer A (1 × PBS, 0.2% Tween 20, 5% dry milk) overnight at 4 °C and then incubated with anti-Cdk2 or anti-Cdk4 antibodies (1:200 dilution in 1 × PBS, 0.2% Tween 20, and 2% dry milk) (Santa Cruz Biotechnology) for 3 h at room temperature. After incubation the membranes were washed three times with buffer B (1 × PBS, 0.2% Tween 20, and 2% dry milk) (10 min each). Then membranes were incubated with secondary antibodies (horse peroxidase-conjugated goat anti-rabbit IgG) for 45 min at room temperature. Membranes were then washed three times with buffer B and three times with buffer C (1 × PBS, 0.2% Tween 20). Membranes were developed by using the enhanced ECL chemiluminescence reagents (Amersham Corp.).
MCK transcriptional activity is enhanced by the co-expression of p21 in 10T1/2 fibroblasts that are transiently transfected with a MyoD expression plasmid (12). To further investigate this property of p21, C2C12 myocytes were transfected with a luciferase reporter plasmid containing the rabbit MCK promoter/enhancer in the presence or absence of a p21-expression plasmid. MCK transcriptional activity was analyzed in replicating myoblast cultures and at different times following the induction of differentiation by mitogen deprivation. As shown in Fig. 1, MCK promoter activity was enhanced by ectopic p21 expression in the high mitogen growth medium (GM) or 1 day following the shift to differentiation medium (DM). However, MCK expression was not enhanced by p21 at the 2-day time point, when the endogenous levels of this protein are high (8, 9).
Various cyclins were compared for their ability to inhibit MCK
transcription in differentiating myocytes. As shown in Fig. 2, and consistent with previous reports (11, 12),
co-expression of cyclin D1 substantially inhibited MCK promoter
activity, but the co-expression of cyclin A or cyclin E had no effect.
Further experiments compared the myogenic inhibitory activity of the
three D-type cyclins that function as activating subunits for Cdk4 and Cdk6. In contrast with previous reports (11, 12), these analyses revealed an inhibition of MCK transcription by all three D-type cyclins, although cyclin D3 and D2 expression plasmids were less potent
inhibitors (Fig. 3A). In each case
co-expression of p21 reversed the inhibition of MCK promoter activity
by the D-type cyclin (Fig. 3B).
To determine the mechanism by which the D-type cyclins, cyclin A, and
cyclin E differentially affects myogenic transcription, Western blot
analyses were performed to determine the relative levels of Cdk2 and
Cdk4 catalytic subunits in myoblasts and myotubes. During myogenesis
the level of the Cdk4 protein (catalytic partner for D-type cyclins)
remained constant, but there was a marked decrease in the level of the
Cdk2 protein (catalytic partner for cyclins A and E) (Fig.
4A). Thus, we tested whether differences in
the expression patterns of Cdk2 and Cdk4 could contribute to the
differential effects of their cognate cyclins on myogenic transcription. As shown in Fig. 4B, MCK transcription was
inhibited by the expression plasmid combinations of Cdk2 and cyclin A
or Cdk2 and cyclin E, but transfections of the Cdk2 plasmid alone had
little or no effect on MCK transcription. Furthermore, the inhibition
of MCK transcription by the combinations of cyclins A or E and Cdk2 was
reversed by the co-expression of p21. These data were corroborated by
demonstrating similar regulatory behavior using 10T1/2 cells
transfected with a MyoD expression vector and a minimal MCK E-box
reporter plasmid (not shown) Collectively, these data provided an
initial indication that the inhibition of MCK promoter activity is
related to the overall extent Cdk activity within the cell.
Inverse Correlation between MCK Transcription and E2F Activation
Phosphorylation of Rb by the Cdks results in the
activation of the E2F transcription factor (3). To test whether E2F is activated under the same conditions that inhibit myogenesis, a reporter
plasmid for E2F activity, (E2F)x4-E1bTATA-Luc, was co-transfected into
C2C12 cells with various combinations of expression plasmids for
cyclins, Cdks and p21. As shown in Fig. 5, E2F
transcriptional activity was stimulated by the ectopic expression of
cyclin D1 or the combination of Cdk2 with either cyclin A or cyclin E,
and this activation was blocked by the co-expression of p21.
Transfections with Cdk2, cyclin A, or cyclin E alone had no effect on
E2F activity. These data revealed a good correlation between the
inhibition of myogenesis and the activation of the E2F transcription
factors. They also suggest that the E2F factors may mediate the
inhibition of MCK transcription by the cyclin-Cdk complexes. Therefore,
E2F1 was directly tested for its ability to inhibit MCK transcription. MCK promoter activity was repressed by ectopic E2F1 expression, but
this repression was not reversed by the co-expression of p21 (Fig.
6).
A Hyperactive Rb Mutant Can Reverse the Cyclin-Cdk- and E2F-mediated Inhibition of MCK Promoter Activity
The potential
involvement of Rb in MCK promoter repression by the cyclin-Cdks was
investigated using hyperactive Rb mutant which lacks the N-terminal 378 amino acids. This mutant, referred to as p56, repressed E2F
transcriptional activity in C2C12 cells (Fig.
7A) and in other cell
types.2 The specificity of this effect was
indicated by the inactivity of the p56/H209 variant that has
phenylalanine substituted for cysteine at amino acid position 706. This
mutation gives rise to an Rb protein that is defective in
phosphorylation and oncoprotein binding (22, 23), and it inactivates
the p56 Rb construct with regard to its ability to repress
transcription from E2F sites (Fig. 7A). The hyperactive form
of Rb reversed the inhibition of MCK promoter activity by E2F1, but the
p56/H209 variant was ineffective (Fig. 7B). Similarly, p56,
but not p56/H209, partially reversed the inhibition of MCK promoter
activity by cyclin D1 or by the combination of Cdk2 and cyclin A (Fig.
7C).
The Helix-Loop-Helix Domain of E2F1 Is Essential for the Inhibition of Myogenic Transcription
E2F1 mutants were analyzed to identify
the structural motifs required for the inhibition of myogenic
transcription. Initially we tested the E2F1 mutant (1-284) which lacks
the C-terminal transcriptional activation domain. This mutant could not
activate the expression of an E2F-reporter construct (Fig.
8) or promote S phase entry when overexpressed in
serum-deprived cells (16). However, the E2F1(1-284) mutant, like
wild-type E2F1, effectively inhibited the ability of MyoD to
transactivate a reporter construct containing four E box elements from
the MCK enhancer (Fig. 8). Further, this E2F1 mutant which lacks the
Rb-binding domain could inhibit myogenic transcriptional activity in
the presence of the hyperactive p56 Rb construct (not shown). These
data indicate that cell proliferation and myogenic differentiation are
coordinated at the level of E2F and that different functional domains
within E2F regulate each process.
Additional E2F1 mutants were examined to determine the region required
for the inhibition of myogenic transcription. In addition to mutants
that lack the transactivation domain, mutant E2F(1-196), which is
missing the leucine (24, 25), and mutant E2F(d87), which is missing the
cyclin A/Cdk2 binding domain (26), were also effective at inhibiting
MyoD transcriptional activity (Fig. 9). However,
myogenic transcription was not inhibited by mutant E2F1(1-127), which
lacks the transactivation domain, the leucine zipper and the putative
helix-loop-helix region (15) (Fig. 9). A series of mutations were
analyzed to characterize the involvement of the basic and
helix-loop-helix regions in the inhibition of myogenic transcription.
The ability of E2F1 to inhibit myogenic transcription was not affected
by an 8-amino acid deletion in the basic region or by mutations at
positions 113 or 120 that change basic amino acids to glutamate
residues (Fig. 10). However, the MyoD-inhibitory
activity of E2F1 was abolished by mutations that substitute glutamate
residues in the helix-loop-helix region (Fig. 10). Similar results were
obtained using either 10T1/2 cells transfected with a MyoD expression
vector and a minimal MCK E-box reporter plasmid (Figs. 8, 9, 10) or using
C2C12 cells transfected with the larger MCK enhancer/promoter construct
(not shown). Taken together, these data demonstrate that only the
helix-loop-helix domain is required for the myogenic inhibitory
activity of E2F1.
Here we investigated the molecular links between cell cycle activity and myogenic differentiation. These analyses extend the studies of Rao et al. (11), which showed that the overexpression of cyclin D1, but not cyclins A, B, or E, represses MCK transcriptional activation, and the studies of Skapek et al. (12), which showed that the cyclin D1-mediated repression is overcome by the ectopic expression of Cdk inhibitors. Based upon these prior observations it was proposed that cyclin D1 functions uniquely to coordinate cell cycle withdrawal and myogenic differentiation through an Rb-independent mechanism (12). However, this hypothesis is not supported by the data from this current study. Here it is shown that the inhibition of myogenic transcription is a function shared by many cyclins or cyclin-Cdk combinations. Further, these data indicate that the inhibitory effects of the cyclin-Cdk complexes are largely mediated by Rb, or a related pocket proteins, and that the E2F transcription factor plays a central role in coordinating myogenic differentiation and cell cycle activity.
Here it is shown that the overexpression of each D-type cyclin is sufficient to inhibit MCK promoter activity. However, the forced expression of cyclins A or E have little if any effect on MCK transcriptional activation (Fig. 2) (11, 12). To investigate the differential effects of these cyclins on myogenic transcriptional activation, the expression patterns of their cognate catalytic subunits were compared. During myogenesis the expression of the Cdk2 protein is markedly down-regulated, but Cdk4 levels do not detectably change (Fig. 4A) (also see Refs. 7, 11, and 12). Thus, we reasoned that cyclin A or cyclin E overexpression might inhibit myogenesis only when they are co-expressed with their catalytic partner Cdk2. On the other hand, overexpression of the D-type cyclins might be sufficient to block MCK transcription because their catalytic partner, Cdk4, is not down-regulated during myogenesis. Consistent with this hypothesis are the observations that MCK promoter activity is inhibited, albeit with different efficiencies, by the transfection of cyclin D1, D2, or D3 expression vectors (Fig. 3). MCK promoter activity was also inhibited by combinations of expression vectors for cyclin A and Cdk2 or cyclin E and Cdk2, but not by the transfection of these individual plasmids (Figs. 2 and 4B). In all cases, the inhibition of MCK transcription was overcome by the co-expression of p21. Collectively, these data indicate MCK transcriptional inhibition is dependent on the overall cell cycle activity, rather than the action of an individual cell cycle component (i.e. cyclin D1). This hypothesis is consistent with the observation of an inverse correlation between MCK promoter expression and E2F transcriptional activity in cells transfected with different combinations of these cell cycle factors (Fig. 5).
Vectors expressing hyperactive and inactive mutants of Rb were utilized to test the potential role of Rb (or a related pocket protein) as a mediator of myogenic transcriptional inhibition. Ectopic expression of the hyperactive (p56), but not an inactive (p56/H209), Rb mutant partially reversed the MCK transcriptional inhibition that resulted from the overexpression cyclin D1 or the combination of Cdk2 and cyclin A (Fig. 7C). These data further indicate that MCK transcription is inhibited by overall cell cycle activity and that this inhibition is mediated by the Rb/E2F pathway. This notion is further supported by the finding that E2F1 overexpression inhibits MCK transcription (Fig. 6) and overall myogenic differentiation (27). MCK promoter inhibition by E2F1 is reversed by the co-expression of the hyperactive Rb mutant, but not by p21 (Figs. 6 and 7); presumably because E2F1 acts at a step that is downstream from p21 (28).
The data from this study are consistent with the model depicted in Fig.
11. Various cyclins, through interactions with their catalytic subunits, inhibit the transcription of muscle-specific genes
via the phosphorylation of Rb or other pocket proteins. This
phosphorylation leads to the activation of the E2F family of
transcription factors, which induce the expression of essential S-phase
genes and also inhibit the expression of muscle-specific genes through
an unknown mechanism. The induction of p21 during myogenesis, a
broad-specificity Cdk inhibitor, enhances differentiation by blocking
the action of multiple cyclin-Cdk complexes. Finally, E2F also
activates the expression of cyclin A and cyclin E (29), but this
pathway appears to have a minimal effect on MCK transcription because
the inhibition of MCK promoter activity by forced E2F1 expression is
not reversed by p21.
The data presented here suggest that E2F functions to coordinate the opposing cellular fates of proliferation and differentiation during myogenesis. Support for the hypothesis comes from the finding that different domains within E2F1 function to promote cell proliferation and inhibit myogenic differentiation (Fig. 8). The transcriptional activation, leucine zipper and helix-loop-helix domains are required for the expression of S phase genes and cell cycle progression. However, only the helix-loop-helix region of E2F1 is essential for the myogenic inhibitory activity, while other regions of the protein are dispensable (Figs. 10 and 11). The inhibition of myogenic transcription by E2F1 overexpression may result from its ability to inactivate E proteins that are essential for myogenic differentiation. Further analyses on the mechanism of E2F-mediated inhibition of myogenic differentiation will be of interest.
We thank Drs. K. Wills, L. Zhu, E. Harlow, A. Yee, A. Dutta, and W. G. Kaelin, Jr., for plasmids.