Cyclin D1 expression is regulated positively by the p42/p44MAPK and negatively by the p38/HOGMAPK pathway.

We have previously shown that the persistent activation of p42/p44MAPK is required to pass the G1 restriction point in fibroblasts (Pagès, G., Lenormand, P., L'Allemain, G., Chambard, J. C., Meloche, S., and Pouysségur, J. (1993) Proc. Natl. Acad. Sci. U. S. A. 90, 8319-8323) and postulated that MAPKs control the activation of G1 cyclin-dependent complexes. We examined the mitogen-dependent induction of cyclin D1 expression, one of the earliest cell cycle-related events to occur during the G0/G1 to S-phase transition, as a potential target of MAPK regulation. Effects exerted either by the p42/p44MAPK or the p38/HOGMAPK cascade on the regulation of cyclin D1 promoter activity or cyclin D1 expression were compared in CCL39 cells, using a co-transfection procedure. We found that inhibition of the p42/p44MAPK signaling by expression of dominant-negative forms of either mitogen-activated protein kinase kinase 1 (MKK1) or p44MAPK, or by expression of the MAP kinase phosphatase, MKP-1, strongly inhibited expression of a reporter gene driven by the human cyclin D1 promoter as well as the endogenous cyclin D1 protein. Conversely, activation of this signaling pathway by expression of a constitutively active MKK1 mutant dramatically increased cyclin D1 promoter activity and cyclin D1 protein expression, in a growth factor-independent manner. Moreover, the use of a CCL39-derived cell line that stably expresses an inducible chimera of the estrogen receptor fused to a constitutively active Raf-1 mutant (ΔRaf-1:ER) revealed that in absence of growth factors, activation of the Raf > MKK1 > p42/p44MAPK cascade is sufficient to fully induce cyclin D1. In marked contrast, the p38MAPK cascade showed an opposite effect on the regulation of cyclin D1 expression. In cells co-expressing high levels of the p38MAPK kinase (MKK3) together with the p38MAPK, a significant inhibition of mitogen-induced cyclin D1 expression was observed. Furthermore, inhibition of p38MAPK activity with the specific inhibitor, SB203580, enhanced cyclin D1 transcription and protein level. Altogether, these results support the notion that MAPK cascades drive specific cell cycle responses to extracellular stimuli, at least in part, through the modulation of cyclin D1 expression and associated cdk activities.

Mammalian cells express multiple mitogen-activated protein (MAP) 1 kinases that mediate the effects of extracellular signals on a wide array of biological processes. In eukaryotic cells, three distinct MAPK cascades have been described, which appear to be linked to separate signal transduction pathways resulting in the final activation of either p42/p44 MAPK , p38/ HOG MAPK , or stress-activated protein kinases (SAPKs) also called Jun kinases (JNKs) (2). Depending on the cellular context, extracellular signals are thought to elicit a specific cellular response (proliferation/differentiation/apoptosis) through the preferential activation of one of the MAPK cascades, which have distinct spectra of substrates (3).
In most cell types, the mitogenic signal is relayed from the cytoplasm into the nucleus by the nuclear translocation of the ubiquitously expressed p42/p44 MAPK isoforms (also called ERK2 and ERK1 for extracellular regulated kinase) (4,5), resulting in activation of a range of transcription factors such as Elk1 (6 -8), c-Ets-1, and c-Ets-2 (9,10). In fibroblasts, agents that elicit a short term MAP kinase activation are not mitogenic, whereas potent mitogens that induce DNA synthesis drive long term MAP kinase activation (11). Previous studies have shown that a sustained activation of the p42/p44 MAPKs is required for fibroblasts to pass the G 1 restriction point and enter S-phase (1,12). Moreover, activation of the p42/p44 MAPK module is sufficient to stimulate early gene transcription and to reduce growth factor requirement for DNA synthesis (13,14). Therefore, the p42/p44 MAPK cascade is likely to regulate some mid-late changes in gene expression that are rate-limiting events for S-phase entry, during the G 1 progression of the cell cycle.
In contrast to p42/p44 MAPKs , which are strongly activated by growth factors and growth-promoting hormones, JNKs and p38 MAPK are poorly sensitive to growth signals, and their activation is preferentially triggered by pro-inflammatory cytokines and environmental stresses (3). The JNK cascade has been implicated in the modulation of AP-1-regulated gene expression through the phosphorylation of the proto-oncogene c-Jun (3,15,16). The two stress-activated signaling pathways can mediate the phosphorylation of the transcription factor ATF2 on residues that increase its transcriptional activity in vivo, suggesting that these two cascades may participate in the regulation of ATF2-dependent gene expression (17)(18)(19)(20). Initial reports suggested that the JNK cascade could be required for mitogenesis in fibroblasts (21). However, a more recent study indicated that activation of JNK and p38 MAPK and concurrent inhibition of p42/p44 MAPK promote the ability of nerve growth factor to induce apoptosis in PC12 cells (22). Nonetheless, JNK and p38 MAPK cascades lie on separate signaling pathways and have, in addition to their common substrates, specific cellular targets, i.e. c-Jun and MAPKAPK-2, respectively (23,24). These two cascades are thus likely to be involved in the regulation of distinct cellular functions; however, it is not clear what specific role can be attributed to one or the other stress signaling pathways at the level of cell division mechanisms.
Transduction of extracellular signals culminates in the expression and assembly of different kinase holoenzymes, the cyclin-cdk (cyclin-dependent kinase) complexes, which are formed and activated at specific stages of the cell division cycle. Although the kinase-associated activity of these complexes is modulated by specific phosphorylation-dephosphorylation events on the catalytic subunits, the temporal activation of the holoenzymes is primarily dependent on the synthesis and accumulation of specific regulatory subunits, the cyclins (25). Cyclins D1, D2, and D3, in conjunction with their catalytic partners cdk4 and cdk6, appear to regulate the initial phases of G 1 progression (26 -29). In normal untransformed cells, the growth factor-dependent accumulation of cyclin D1 has been shown to be required to allow cells to pass the G 1 restriction point (27,30,31). However, cyclin D1 requirement becomes dispensable in a cell background deficient for the tumor suppressor (pRb) function (32). This observation further emphasizes the functional link between pRb and cyclin D1-associated kinase activity, which has been shown to be responsible in part for hyperphosphorylating pRb (33)(34)(35)(36)(37). Therefore, early appearance of cyclin D1 upon growth factor stimulation of resting fibroblasts plays a central role in regulating the G 0 -G 1 transition of the cell cycle.
Because of the strict requirement for a persistent p42/ p44 MAPK activation to successfully pass the G 1 restriction point, we hypothesized that these MAPKs control cyclin D1associated kinase activity. We have thus drawn our attention to cyclin D1 as a potential "nuclear sensor" of extracellular signals. Here we report that cyclin D1 expression is positively controlled by the Raf Ͼ MKK1 Ͼ p42/p44 MAPK cascade in the Chinese hamster fibroblast cell line CCL39, whereas the stress-activated p38/HOG MAPK cascade antagonizes this expression. This report that establishes a link between persistent activation of p42/p44 MAPKs and cyclin D1 expression provides a key element to the understanding of the temporal action of growth factors.

EXPERIMENTAL PROCEDURES
Materials-Highly purified ␣-thrombin (3209 NIH units/mg) was kindly provided by Dr. J. W. Fenton II (New York State Department of Health, Albany, NY). Epidermal growth factor and insulin were from Sigma. IL-1␤ was purchased from Boehringer Mannheim. The specific p38 MAPK inhibitor SB203580 was provided by SmithKline Beecham Pharmaceuticals (King of Prussia, PA). Rabbit polyclonal cyclin D1 antibody was raised against a peptide from recombinant human cyclin D1 and was a generous gift from Dr. V. Baldin (30). Mouse monoclonal antibodies against cyclin D2 and cyclin D3 were kindly provided by Dr. J. Bartek and have been previously described (38). Rabbit polyclonal antibody E1B against p42/p44 MAPK was raised against a C-terminal peptide of mouse ERK2. 2 Rabbit polyclonal antibody against MKK1 (MKK13) was described previously (5). Rabbit polyclonal antibody SAM2 against p38 MAPK was raised against a 14-residue C-terminal peptide of the human p38MAPK. 3 Monoclonal antibody against pRb was purchased from Pharmingen (Clinisciences, Paris, France), monoclonal antibody 12CA5, raised to a peptide from influenza HA1 protein, was purchased from Babco (Emeryville, CA), and monoclonal antibody M2 against the Flag epitope was from Kodak Integra Biosciences.
Transient Transfection and Luciferase Assay-CCL39 cells were transfected by the calcium phosphate method. For the cyclin D1-luciferase expression assays, CCL39 cells were seeded at a density of 150,000 cells per well in 24-well plates and co-transfected with 0.25 g of the reporter D1⌬-944 together with 0.75 g of relevant expression vector, or the corresponding empty vector. For experiments performed in exponentially growing cells, luciferase activity was measured 48 h after transfection. Normalization was achieved by co-transfecting 0.1 g of pCH110, a ␤-galactosidase reporter construct as an internal control for the transfection efficiency. For mitogen-stimulated cyclin D1-luciferase reporter expression, 1 day following the transfection cells were serum-starved for 30 h and then stimulated with 10% FCS for 18 h. Luciferase and ␤-galactosidase activities were measured according to the Promega protocol. Data are representative for at least three independent experiments performed in duplicate and are expressed as "fold increase in luciferase activity," which was calculated relative to the basal level of cyclin D1 reporter activity set to 1 unit and corrected for empty vector effects for each expression vector. Cyclin D1 protein expression analyses were performed in CCL39 cells seeded at a density of 600,000 cells per well in six-well plates, co-transfected by the calcium phosphate technique with 6 g of the selection vector pNHE3, together with 14 g of relevant expression vector. Two days following the transfection, cells were submitted to an acid-load selection (45) in the presence of the HOE694 compound to inhibit the endogenous NHE1 antiporter activity, so that only transfected cells expressing the NHE3 isoform survived (46). Cells were then allowed to recover for at least 4 h before serum starvation. Cells were lysed 3 days following the transfection, and cyclin D1 expression was monitored by immunoblotting using the antisera described above. Expression of the transfected MAPK constructs was checked for each experiment using the appropriated antibody.
Immunoblotting-Cells were lysed in SDS sample buffer (62.5 mM Tris-HCl, pH 6.8, 2.3% SDS, 10% glycerol, 5% ␤-mercaptoethanol, 0.005% bromphenol blue, 1 mM phenylmethylsulfonyl fluoride), and proteins from whole cell lysates were separated by SDS-PAGE in 10% gels or in 7.5% gels for pRb detection. For p42MAPK "shift up" experiments, the percentage of bisacrylamide was reduced to 0.07% (final), and samples were run on 12.5% SDS-polyacrylamide gels. Proteins were detected immunologically following electrotransfer onto nitrocellulose membranes as described previously (47). Horseradish peroxidase-linked goat anti-rabbit or anti-mouse immunoglobulin G (Sigma), revealed by the ECL detection system (Amersham), was used to detect the antigen-antibody complexes. Protein concentrations were measured using a modified Lowry procedure (48) with bovine serum albumin as standard. Quantitative analyses of protein levels were performed by scanning of the autoradiograms (Fugi PhosphoImager, Paris, France) and are representative of more than two independent experiments.

Potent
Mitogens Promote Long-term p42/p44 MAPK Activation, Cyclin D1 Expression, and pRb Phosphorylation-D-type cyclin expression profile was analyzed in whole cell extracts of G 0 -arrested Chinese hamster fibroblasts (CCL39), by Western blot technique using specific antibodies. In G 0 -arrested CCL39 cells, cyclin D1 protein expression was barely detectable (Fig.  1). However, serum stimulation of cells led to a dramatic accumulation of cyclin D1 protein that became easily detectable at 6 h poststimulation and increased until cells entered S-phase (Fig. 1A). As previously shown in other fibroblastic cell lines (30,49), the protein expression level of cyclin D1 in CCL39 cells was not cell cycle-modulated and only a modest peak of accumulation was observed in late G 1 (12-16 h poststimulation). Cyclin D2 and D3 expression were also analyzed in CCL39 cells using specific monoclonal antibodies. In contrast to cyclin D1, for that expression was strictly dependent on the presence of mitogens, cyclin D3 was present in resting cells and its level was poorly increased by growth factors (Fig. 1A). Cyclin D2 was not detectable in either quiescent or growth factor-stimulated CCL39 cells. The hyperphosphorylation of pRb was analyzed for each time point following serum stimulation of resting cells. A specific antibody able to detect the active hypophosphorylated form of pRb (lower band) as well as the inactive hyperphosphorylated form of the protein (upper band) was used. In G 0 -arrested cells, pRb was exclusively found in its active hypophosphorylated state, suggesting that cyclin D3, although expressed does not contribute to pRb phosphorylation in this cell system and thus, is not a limiting factor for progression through G 1 -phase of the cell cycle. However, pRb inactivation (hyperphosphorylated form) became apparent at 10 -12 h poststimulation, when cyclin D1 expression was maximal (Fig. 1A, lower panel) and when the first cells enter S-phase. 5 The results suggest that cyclin D1 is the major mitogen-regulated D-type cyclin identifiable in this cell system and, therefore, that its associated kinase activity is likely to participate in pRb inactivation by phosphorylation.
Previous studies demonstrated that in fibroblasts only potent mitogens that drive the G 0 to S-phase transition can maintain a long term activation of the p42/p44 MAPK cascade (11). Initial experiments to link this signal transduction pathway to mitogen regulation of cyclin D1 expression revealed a tight correlation between the extent of cyclin D1 accumulation and the ability of various agents to induce sustained p42 MAPK activation. Cells were serum-starved for 24 h and stimulated for 9 h in the presence of various agonists, and whole cell extracts were analyzed for cyclin D1 protein levels, pRb phosphorylation, and p42 MAPK activity. The polyclonal anti-p42 MAPK antibody used detected three electrophoretically distinct forms of the kinase: a fast-migrating band representing the inactive unphosphorylated form of the endogenous p42 MAPK , an intermediate band that corresponds to the phosphorylated and active enzyme, and an upper, slow-migrating form that corresponds to cross-reactivity of the antibody with the p44 MAPK isoform (Fig. 1B, lower panel). Cyclin D1-associated kinase activity that is mainly dependent on the level of cyclin D1 was also evaluated in each sample by analyzing pRB phosphorylation. The ability of the various mitogens to induce cyclin D1 expression and pRb hyperphosphorylation was closely related to their ability to maintain p42 MAPK activities throughout G 1 progression of the cell cycle (Fig. 1B). Only potent mitogens, such as whole serum and thrombin, which potently initiate DNA synthesis in arrested CCL39 cells, were able to promote high levels of cyclin D1 accumulation and allow pRb phosphorylation (appearance of a minor slow-migrating band at 9 h post-stimulation) (Fig. 1B, middle panel). Whereas a slight effect was observed with a weakly mitogenic combination of epidermal growth factor plus insulin, neither factor alone was able to promote long term p42 MAPK activation, cyclin D1 expression, and pRb phosphorylation, suggesting an interdependent relationship between the long lasting activation phase of p42/p44 MAPKs , cyclin D1 expression, and pRb inactivation in CCL39 fibroblasts.
The p42/p44 MAPK Cascade Positively Regulates Cyclin D1 Expression in Fibroblasts-To investigate the role of the p42/ p44 MAPK signaling pathway in the regulation of cyclin D1 expression more directly, we used previously characterized expression constructs to modulate either positively or negatively the endogenous p42/p44 MAPK activity (see "Experimental Procedures"). Cyclin D1 transcription was monitored by transfecting CCL39 cells with a previously cloned fragment of the human cyclin D1 promoter fused to the luciferase reporter gene (D1⌬-944) (39), together with the relevant constructs. The results showed that the p42/p44 MAPK activity strongly affects cyclin D1 transcription in CCL39 cells. When a constitutively activated form of MAPK kinase (MKK1-SS/DD) was expressed, a large increase in luciferase expression (6 -10-fold) could be detected in exponentially growing cells, when compared with the luciferase expression in control cells transfected with the empty vector ( Fig. 2A). This dramatic up-regulation of the cyclin D1 reporter expression in cells overexpressing the constitutively active MKK1 mutant likely resulted from a higher p42/p44 MAPK activity in these cells, since co-expression of the MAPK phosphatase (MKP-1), a dual specificity phosphatase shown to be able to inactivate p42/p44 MAPKs (42) 1. A, time course of D-type cyclin expression and pRb phosphorylation in CCL39 cells. CCL39 were serum-starved for 24 h and then stimulated with 20% FCS for the indicated period (hours). Equal amounts of whole cell lysates were separated by SDS-PAGE, and proteins were electrotransfered onto nitrocellulose. Western blot detection of cyclin D1, cyclin D3, and pRb proteins was performed using the appropriate specific antibodies. B, cyclin D1 accumulation, pRb phosphorylation, and p42 MAPK activity in G 0 -arrested CCL39 stimulated with various agonists. Cells were serum-starved for 24 h and then stimulated for 9 h in the presence of various agonists as indicated. Equal amounts of whole cell lysates were separated by SDS-PAGE, and Western blot detection was performed using either antibody against cyclin D1, pRb, or p42 MAPK , following electrotransfer of total proteins onto nitrocellulose. erase expression ( Fig. 2A). No significant effects of these plasmids were observed in the same experiments on the ␤-galactosidase internal control reporter expression. A positive effect on the cyclin D1-luciferase expression was also measured in G 0 -arrested cells expressing the constitutively active form of MKK1. In these cells, activation of the p42/p44 MAPKs in the absence of mitogen led to an increase of cyclin D1-luciferase expression that was even higher than the level measured in serum-stimulated cells expressing the wild type MKK1 (Fig.  2B). However, the reporter gene expression in these cells was still inducible by serum to the same extent as in control cells (2-4-fold). Conversely, when the p42/p44 MAPK cascade was blocked by expression of inhibitory constructs, cyclin D1-luciferase expression was strongly reduced in exponentially growing cells. Expression of dominant-negative p44 MAPK (p44 MAPK -T192A) or MKP-1 produced a 45-80% inhibition of cyclin D1luciferase expression, respectively ( Fig. 2A). As expected, expression of the negative regulatory MKK1-S222A construct inhibited cyclin D1-luciferase expression to a lesser extent (30%), correlating with a partial inhibition of endogenous MKK1 activity in cells expressing this mutated form of MKK1 (40). Moreover, the serum-stimulated luciferase expression was strongly inhibited (80%) in cells expressing the negative regulatory constructs, as compared with control cells expressing the reporter gene alone (Fig. 2B). These results thus demonstrate that the p42/p44 MAPK signal transduction pathway positively regulates cyclin D1 transcription in fibroblasts.
Previous studies have shown that an increase in the level of cyclin D1 messenger RNA does not always lead to an increase in cyclin D1 protein levels in transfected cells, suggesting that post-transcriptional processes might also play an important role in the regulation of cyclin D1 expression (50 -52). We thus examined whether the modulation of cyclin D1 transcription by the p42/p44 MAPK pathway also resulted in modification of cyclin D1 protein levels in CCL39 fibroblasts. The effects of the same constructs (MKK1, MKK1-SS/DD, or MKP-1) on the level of endogenous cyclin D1 protein were analyzed in CCL39 cells. To do so, the expression plasmids were co-transfected with a selection vector that encodes an amiloride-resistant Na ϩ /H ϩ exchanger isoform, NHE3. Nontransfected cells were selec- tively eliminated by acid-load selection, as described under "Experimental Procedures," and cyclin D1 protein levels were analyzed in whole cell extracts of resistant cells by Western blots using a specific antibody. In exponentially growing cells expressing the constitutively active MKK1 mutant, a higher level of cyclin D1 could be detected as compared with cyclin D1 expression in control cells transfected with either the empty vector or the wild type MKK1 (Fig. 2C). This increase of cyclin D1 protein level in MKK1-expressing cells kept in serum-supplemented medium was associated with an overall increase in the amount of the hyperphosphorylated form of pRb; however, the relative total amount of pRb expression was also increased, keeping constant the ratio between the hyper-and hypophosphorylated forms. Conversely, inactivation of p42/p44 MAPKs by expression of the MKP-1 markedly reduced cyclin D1 protein expression below the level detected in control cells (Fig. 2C). The significant inhibition of cyclin D1 protein expression in MKP1-overexpressing cells also correlated with a 45% decrease in the amount of the hyperphosphorylated form of pRb, suggesting that the resulting low level of cyclin D1 in these cells was limiting for cyclin D1-associated kinase activity. When similar experiments were performed in resting cells, a strong positive effect on cyclin D1 expression could be detected upon activation of the p42/p44 MAPK pathway. As shown in Fig. 2D, cyclin D1 was barely detectable in serum-starved control cells, whereas in serum-starved MKK1-SS/DD expressing cells, a growth factor-independent expression of cyclin D1 was detected. Furthermore, this increase of cyclin D1 protein level mediated by expression of the constitutively active MKK1 mutant was comparable with the level measured in thrombinstimulated control cells (see quantitations in the legend of Fig.  2D). Conversely, when the p42/p44 MAPK activity was blocked by the presence of an elevated level of MKP-1, thrombin-induced accumulation of cyclin D1 was severely impaired, being inhibited by as much as 51%, as compared with the level measured in cells transfected with the corresponding empty vector (Fig. 2D). Altogether, these results strongly suggest an important contribution of the p42/p44 MAPK signal transduction pathway in the regulation of cyclin D1 expression in response to growth signals, both at the level of transcription and protein synthesis.
Activation of the p42/p44 MAPK Cascade Is Sufficient to Promote Cyclin D1 Accumulation in CCL39-⌬Raf-1:ER Cells-The contribution of the p42/p44 MAPK cascade on mitogen-induced cyclin D1 protein synthesis was next examined, using a CCL39derived cell line (CCL39-⌬Raf-1:ER) expressing an estradioldependent human Raf-1 protein kinase (44). In this cell line, the ⌬Raf-1:ER chimera is activated in response to estradiol, thereby activating MKK1 and then p42/44 MAPKs . Previous characterization of CCL39-⌬Raf-1:ER cells has shown that addition of estradiol to serum-starved cells stimulates p42/ p44 MAPKs within minutes. The p42/44 MAPK activity increases for up to 1 h, thus reaching a level comparable to the maximal p42/44 MAPK activation measured in serum-stimulated CCL39 control cells, and remains elevated in the presence of estradiol. 4 We used this cell system to directly measure the specific contribution of Raf Ͼ MKK1 Ͼ p42/p44 MAPK cascade on cyclin D1 protein expression. CCL39-⌬Raf-1:ER cells were serumstarved for 24 h and then stimulated with either 10 M estradiol or 10% FCS for varying periods. Whole cell extracts were analyzed for cyclin D1 protein expression and the pRb phosphorylation state by Western blots using specific antibodies. The results showed that estradiol-treated CCL39-⌬Raf-1:ER cells accumulated cyclin D1 protein to a level comparable with that stimulated in the same cells by 10% serum (Fig. 3). Moreover, cyclin D1 accumulation followed a similar time course in estradiol-treated cells when compared with serum-stimulated cells, indicating that activation of the Raf pathway alone (leading to p42/44 MAPK activation) initiated positive regulatory signals responsible for cyclin D1 protein synthesis. Nevertheless, the Raf pathway alone was not sufficient to promote activation of the D1-associated kinase activity, since neither pRb phosphorylation (Fig. 3, lower panels) nor in vitro cdk4 kinase activity (data not shown) could be detected in estradiol-stimulated CCL39-⌬Raf-1:ER cells.
The p38/HOG MAPK Cascade Antagonizes Mitogen-induced Expression of Cyclin D1 in Fibroblasts-In addition to the p42/p44 MAPK pathway, at least two other MAPK cascades are implicated in the transduction of external stimuli in mammalian cells, the p38/HOG MAPK and the JNKs (2). We next examined whether other MAPK pathways could have some regulatory effects on cyclin D1 expression. Previous studies indicated that the proto-oncogene c-jun can increase expression of a cyclin D1 promoter-controlled luciferase reporter in co-transfection experiments, suggesting a positive regulatory effect of the JNK pathway on cyclin D1 transcription (39,53). However, the effect on protein expression has not been shown. The role of the p38 MAPK cascade in gene expression is less clear. We thus focused on the putative effect of the p38 MAPK cascade on cyclin D1 transcription and protein expression in CCL39 cells. Cotransfection experiments were performed, using plasmids encoding different members of this cascade: the p38 MAPK kinase (MKK3), the constitutively active MKK3 mutant (MKK3-S189G/T193G called MKK3-Glu), the p38 MAPK , and the dominant-negative p38 MAPK (p38 MAPK -TY/AF), together with the cyclin D1-luciferase reporter construct. Surprisingly, overexpression of MKK3, which resulted in a higher basal level of p38 MAPK activity in CCL39 cells, did not activate cyclin D1 promoter; rather cyclin D1-luciferase expression was reduced by 33% in exponentially growing cells, when compared with cells transfected with the empty vector (Fig. 4A). A stronger inhibitory effect was detected when a wild type form of the p38 MAPK was expressed together with MKK3, as if endogenous p38 MAPK was limiting. In contrast, expression of dominantnegative p38 MAPK -TY/AF together with MKK3 totally abolished the MKK3-mediated inhibition of cyclin D1 luciferase expression, indicating that the negative effect on cyclin D1 transcription results from increased MKK3 activity. Expression of the constitutively active MKK3 mutant (MKK3-Glu) elicited a more pronounced inhibition of cyclin D1-luciferase expression (62%) than the wild type MKK3, a result further indicating that the inhibitory effect was dependent on MKK3 activity. When assays were performed in G 0 -arrested cells, basal luciferase expression of the cyclin D1 promoter-controlled FIG. 3. Time course of cyclin D1 accumulation and pRb phosphorylation in estradiol or serum-stimulated CCL39-⌬Raf-1:ER cells. CCL39-⌬Raf-1:ER cells were serum-starved for 24 h and stimulated with either 10 M estradiol or 10% FCS for varying periods as indicated (hours). Cells were then lysed in SDS sample buffer, and equal amounts of whole cell lysates were separated by SDS-PAGE in 10% gels (cyclin D1) or in 7.5% gels (pRb). Cyclin D1 and pRb protein levels were analyzed by Western blotting using specific antibodies, following electrotransfer of total proteins onto nitrocellulose.
reporter was also reduced (Fig. 4B). Furthermore, increasing p38 MAPK activity by expression of MKK3 resulted in a marked inhibition of the serum-induced cyclin D1-luciferase expression in resting cells. This result suggests that the p38 MAPK cascade could be involved in the negative regulation of cyclin D1 transcription and thus antagonize the mitogen-dependent stimulation of cyclin D1 transcription mediated, at least in part, by the p42/p44 MAPK cascade.
This opposing effect of the p38 MAPK cascade, as compared with the positive regulatory effect of the p42/p44 MAPK pathway, was also detected on endogenous cyclin D1 protein expression in CCL39. The effects of MKK3 and p38 MAPK were monitored following co-transfection of the relevant constructs together with the selection vector encoding an amiloride-resistant Na ϩ /H ϩ exchanger isoform, which allowed us to monitor cyclin D1 protein levels in a population of cells enriched for the transfected genes. Cyclin D1 expression was measured in whole cell lysates of resistant cells by immunoblotting using a specific antibody. The results showed that cyclin D1 protein levels were reproducibly decreased by 40 -50% in cells expressing the MKK3 construct when compared with control cells transfected with the corresponding empty vector (Fig. 4C). This inhibition of cyclin D1 expression was associated with a slight decrease in the amount of the slow-migrating hyperphosphorylated form of pRb in the same cell extracts, together with a corresponding increase in the amount of the hypophosphoryl-ated form, suggesting that the MKK3-mediated interfering effect on cyclin D1 expression was affecting cyclin D1/cdk4 -6 activity (Fig. 4C). Even though this modification of the pRb phosphorylation state was not pronounced, we could reproducibly quantitate 25% decrease of the slow-migrating form. Similarly, the thrombin-induced accumulation of endogenous cyclin D1 was inhibited by 55% upon co-expression of the p38 MAPK together with its activator, MKK3, when compared with the level of thrombin-induced cyclin D1 in cells transfected with the empty vectors (Fig. 4D).
As an independent approach to probe the role of the p38 MAPK cascade, we used a chemical compound, the SB203580, which has been described recently as a specific inhibitor for p38 MAPK , but without inhibitory action on p42/p44 MAPKs and JNKs (54,55). When the endogenous p38 MAPK activity was inhibited by pretreatment of CCL39 with SB203580, the cyclin D1-luciferase expression was reproducibly enhanced by 2-3-fold (Fig.  5A). Furthermore, treatment of MKK3-expressing cells with SB203580 reversed the inhibition of cyclin D1-luciferase expression, a result that indicates that the MKK3-mediated inhibition of cyclin D1 transcription results from an increased p38 MAPK activity. A 2-3-fold enhancing effect of SB203580 on the level of endogenous cyclin D1 protein expression was also observed in exponentially growing cells as well as in resting cells in absence of mitogens (Fig. 5B). Altogether, these results strongly suggest that the p38 MAPK cascade may exert a nega- CCL39 were co-transfected with D1⌬-944 and expression vectors containing members of the p38 MAPK cascade, as indicated. A, luciferase activity was measured 48 h after the transfection and was normalized to the ␤-galactosidase activity by co-transfecting the pCH110 reporter. B, transfected cells were serum-starved for 30 h and stimulated with 10% FCS for 18 h. The fold increase in luciferase activity was calculated relative to the basal level of cyclin D1-luciferase, which was set to 1 unit and corrected for empty vector effects. Data are representative of at least three independent experiments. C and D, immunoblots of whole cell lysates from transfected cells. CCL39 were co-transfected with pNHE3 and expression constructs containing members of the p38 MAPK cascade or the appropriate empty vector (EV) as indicated. Transfected cells were selected 48 h after the transfection by applying an acid-load selection. C, cells were kept in the exponential growing phase in 7.5% serum-supplemented medium. D, cells were allowed to recover after the acid-load selection treatment in DMEM 7.5% FCS for at least 4 h before being starved for 24 h and stimulated in the presence of 1 unit/ml thrombin for 9 h. Equal amounts of whole cell lysates were separated by SDS-PAGE and proteins were detected by Western blots using specific antibodies against either cyclin D1 or pRb. Protein levels were quantitated by scanning of the autoradiograms. tive control on cyclin D1 expression.
A single external stimulus can lead to the simultaneous activation of multiple MAPK signal transduction pathways (2). However, the magnitude and time course of the response observed for each pathway is greatly dependent upon the nature of the stimulus. In order to determine if the opposing effects of p42/p44 MAPK and p38 MAPK activation on cyclin D1 expression were physiologically relevant, we designed experiments to measure the differential effects on cyclin D1 in response to appropriate stimuli. ␣-Thrombin mainly activates p42/ p44 MAPKs , whereas IL-1␤ drives a strong p38 MAPK activation with little effect on p42/p44 MAPKs . When these two MAPK pathways were simultaneously activated by the co-addition of ␣-thrombin and IL-1␤ to resting CCL39 cells, the cyclin D1 protein level obtained after a 9-h stimulation was attenuated by 25-30% as compared with the level measured in cells exposed to thrombin alone (Fig. 6). Based on co-transfection experiments shown above, the reduced expression of cyclin D1 obtained upon simultaneous activation of p42/p44 MAPKs and p38 MAPK likely resulted from opposite regulatory effects of these two signaling pathways. Therefore, this result strongly indicates that the differential regulation of cyclin D1 by the p42/p44 MAPK and the p38 MAPK cascades could have some physiological relevance.

DISCUSSION
Entry into the cell cycle upon growth factor stimulation requires fine coordination of events from the membrane to the nucleus. To pass the restriction point, mammalian cells must be continuously "fired" by signals during the first 6 -8 h that precede the onset of DNA replication (56). At least two types of Ser/Thr protein kinases play a determinant role during this critical period of G 0 /G 1 progression: (i) the MAP kinase isoforms p42/p44 MAPKs (ERK2/ERK1) and (ii) the G 1 cyclin-cdk complexes, particularly cyclin D1/cdk4. Specific inactivation of either of these two protein kinase signaling systems results in specific growth arrest in G 1 (1,12,27,30,31).
We previously demonstrated that persistent activation of p42/p44 MAPKs is critical for the commitment of cell cycle entry in fibroblasts (5,11). Only potent mitogens are capable of stimulating a sustained kinase activation of p42/p44 MAPKs and the concomitant nuclear translocation of these enzymes (5). The long-term activation (that represents 10 -30% of the peak activity measured 5 min following mitogen addition) usually persists for several hours before declining to barely detectable levels when cells enter and progress through S-phase (5,11). 6 In marked contrast to p42/p44 MAPKs , cyclin D1/cdk4 activity in CCL39 fibroblasts emerges rather late in G 1 -phase and increases progressively as the cells approach and pass through S-phase. 5 Based on this temporal difference in activation and on the mitogenic regulatory roles of these two protein kinase signaling systems, we postulated that the p42/p44 MAPKs control the activation of cyclin D1/cdk4-cdk6, the first complex to be activated in early G 1 . However, mechanisms that govern G 1cyclin/cdk activation are complex. First, de novo synthesis of the regulatory cyclin subunit is required, and this step is certainly the most limiting step in G 0 -arrested cells. Second, appropriate sites on the catalytic subunit must be phosphorylated by the activating kinase CAK and dephosphorylated by the specific cdc25 phosphatase (57). Third, multiple G 1 -cdk inhibitors (CKIs) must be blunted for the kinase activity to emerge (58). It is possible that p42/p44 MAPKs control many of these steps acting on the cyclin D1/cdk4-cdk6 activation. The present study focused only on the first level of regulation that concerns the growth factor-sensitive cyclin D1 induction.
We found that among the D-type cyclins, cyclin D1 is the only detectable early G 1 -cyclin to be regulated by growth factors in 6  A, CCL39 cells were co-transfected with D1⌬-944 and MKK3, or the empty vector (EV). One day after transfection, cells were exposed to 10 M of SB203580 for 16 h, and luciferase activity was measured. Data are representative of at least three independent experiments. The fold increase in luciferase activity was calculated relative to the basal expression level of D1⌬-944, which was set to 1 unit and corrected for empty vector effects. D1-luc, D1-luciferase. B, cyclin D1 immunoblots of CCL39 whole cell lysates. Exponentially growing (upper panel) or thrombin-stimulated resting CCL39 cells (lower panel) were exposed to 10 M SB203580 for 9 h, and whole cell lysates were separated by SDS-PAGE in 10% gels. Cyclin D1 protein levels were detected by Western blotting and quantitated by scanning of the autoradiograms. Data are the means Ϯ S.E. for at least three independent experiments and are expressed as fold increase of cyclin D1 protein levels in cells incubated with SB203580 as compared with the level measured in untreated control cells. Exponentially growing cells, 2.03 Ϯ 0.15; G 0 -arrested cells, 2.6 Ϯ 0.38; thrombin-stimulated cells, 1.15 Ϯ 0.09. FIG. 6. IL-1␤ attenuates the thrombin-induced expression of cyclin D1 in CCL39 cells. CCL39 cells were serum-starved for 24 h and then stimulated in the presence of either 0.1 unit/ml thrombin or 0.1 unit/ml thrombin together with 20 ng/ml IL-1␤ for 9 h. Equal amounts of whole cell lysates were separated by SDS-PAGE in 10% gels, and cyclin D1 protein levels were determined on Western blots and quantitated by scanning of the autoradiograms. The thrombin-induced cyclin D1 expression was inhibited by 26.5 Ϯ 2.06% in the presence of IL-1␤. Percent IL-1␤ inhibition is the mean Ϯ S.E. for at least two independent experiments and was calculated relative to the level measured in cells stimulated with ␣-thrombin only. Cyclin D1 immunoblot is shown. CCL39 cells. Furthermore, we demonstrated that the p42/ p44 MAPK cascade controls cyclin D1 expression in response to growth signals thus contributing to the regulation of S-phase entry. Such a modulation of cyclin D1-associated kinase activity will participate in pRb inactivation (pRb hyperphosphorylation). As a result, E2F-regulated gene expression is expected to be suppressed in cells where p42/p44 MAPK activity is inhibited. In fact, not only cyclin D1 and its associated kinase activity were blocked in cells expressing either the dominant-negative p44 MAPK or the MKP-1, but also the cdk2 activity was inhibited (data not shown). This result is consistent with the fact that cdk2 activation in late G 1 /early S-phase is dependent upon transcription and synthesis of its regulatory subunits, cyclin E and cyclin A, which in turn require E2F release for their expression, as many other inducible genes at the G 1 /S boundary (59 -62). Athough preliminary data suggest that MKP-1 could have the ability to inactivate other members of the MAPK family, in particular the JNKs, 7 a similar inhibition of cyclin D1 expression, pRb hyperphosphorylation and cdk2 activity were observed using the dominant-negative p44 MAPK mutant, a more specific interfering construct. Considering previous studies that have shown that inhibition of the p42/ p44 MAPK cascade blocks DNA synthesis and cell proliferation (1), we conclude that the requirement of this signaling pathway for S-phase entry may thus rely in part on its essential function as a positive regulator of cyclin D1 expression.
This study revealed that not only the p42/p44 MAPK cascade is required for cyclin D1 transcription and protein synthesis, but it is sufficient by itself. Expression of a constitutively active MAP kinase kinase produced an increase in cyclin D1 protein level in absence of any other growth signal, which was equivalent to the expression level measured in cells stimulated with a strong mitogen. More importantly, the use of a CCL39-derived cell line expressing an estrogen-dependent human Raf-1 protein kinase (CCL39-⌬Raf-1:ER) demonstrated that the exclusive activation of the Raf Ͼ MKK1 Ͼ p42/p44 MAPK cascade was able to induce cyclin D1 protein expression to the same magnitude and with an identical time course as that induced by serum in these cells. Even though stimulation of the Raf pathway was sufficient to induce cyclin D1 accumulation, it could not promote pRb phosphorylation or cdk2 kinase activation (data not shown). As a result, no significant estradiol-induced DNA synthesis could be detected in this system. 8 The apparent increase in the overall pRb hyperphosphorylation in cells expressing the constitutively active MKK1 mutant in the presence of serum may thus likely result from a cooperative effect between the p42/p44 MAPK cascade and other signaling pathways. Interestingly, we could detect a similar increase in the overall amount of pRb expression (and thus in the pRb hyperphosphorylated form), associated with a marked increase in the level of cyclin D1 expression in stable-CCL39 transfectants expressing a high constitutive level of the MKK1-SS/DD mutant. 9 The use of CCL39-⌬Raf-1:ER-derived cell line allowed us to discriminate between secondary versus immediate effects of p42/p44 MAPK activation in a synchronized population of cells. We thus conclude that although activation of the p42/p44 MAPK cascade is necessary and sufficient for cyclin D1 expression, other signals are required to promote the activation of cyclin D1/cdk4 -6 complex and thus the hyperphosphorylation of pRb to ensure passage through S-phase. However, activation of the Raf Ͼ MKK1 Ͼ p42/p44 MAPK by estradiol treatment led to a decrease in growth factor requirement for DNA synthesis, a result in agreement with our previous finding in cells expressing the constitutively active MKK1 mutant (13). Interestingly, a CCL39-derived cell line expressing high levels of human cyclin D1 exhibited a similar higher sensitivity to growth factors. 5 In marked contrast to the positive action of p42/p44 MAPK activation, the p38 MAPK signaling pathway exerted a negative effect on cyclin D1 expression. Activation of the p38 MAPK cascade led to a significant decrease in cyclin D1 transcription and, conversely, inhibition of this signaling pathway by the specific inhibitor SB203580 had an opposite enhancing effect. Although less pronounced, this inhibitory effect of the p38 MAPK on cyclin D1 expression was also visualized at the level of protein synthesis and reproducibly detected using different strategies (either expression of MKK3, p38 MAPK , or pretreatment with SB203580). In addition, IL-1␤, a physiological inducer of p38 MAPK activation, showed a similar antagonizing effect on thrombin-induced cyclin D1 accumulation in CCL39 cells. However, treatment of CCL39 fibroblasts with IL-1␤ also increases the JNK activity. 10 The fact that the JNK cascade has been shown to promote cyclin D1 transcription and cell proliferation in fibroblasts (21,53) may explain the poor inhibitory effect of IL-1␤ on DNA synthesis in CCL39 cells. 11 However, the results presented here strongly suggest that the p38 MAPK -mediated inhibition of cyclin D1 expression may have more dramatic effects in cell systems where IL-1␤ negatively regulates cell division. Interestingly, it has been recently shown that hypophosphorylation of pRb could mediate the G 0 /G 1 growth arrest induced by IL-1␤ in human A375-C6 melanoma cells (63).
The molecular mechanisms underlying the negative regulation of cyclin D1 expression by p38 MAPK is not clear. The transcription factor ATF2 is substrate for p38 MAPK and, therefore, may account for the effect of this signaling pathway. The cyclin D1 promoter contains multiple regulatory elements (TRE, E2F, Oct, SP1, CRE) and some uncharacterized elements that may also play a role in transcription of the gene (39). Thus cyclin D1 expression may be responsive to a large set of transcription factors. In addition, multiple MAPK cascades appears to be implicated in the regulation of the promoter activity (this study) (39,53), some of that modulating the activity of a common substrate. This is the case for Elk-1, a substrate for the three distinct mammalian MAPKs, which can therefore integrate signals from multiple MAPK cascades in response to extracellular stimuli (64). The transcription factor ATF2 is also a common substrate for p38 MAPK and JNKs (17,18,20). However, only the JNK cascade can activate c-Jun (16,(65)(66)(67). Therefore, activation of a specific MAPK cascade is likely to produce a differential effect on gene expression. Finally, there are many mechanisms for the regulation of the respective MAPK signal transduction pathways and both positive and negative cross-talk between these MAPK cascades probably exists, which could explain the interfering effect of the p38 MAPK cascade on cyclin D1 expression.
Although the data obtained here strongly suggest that the p42/p44 MAPK and the p38 MAPK exert their regulatory action on cyclin D1 expression at the level of the transcriptional machinery, they do not exclude a possible additional post-transcriptional regulation. It has been suggested that the p42/p44 MAPK cascade could up-regulate translation initiation of specific genes in response to insulin (68). A recent study has shown that overexpression of eIF-4E, which is released upon phosphorylation of PHAS-1 protein by p42/p44 MAPKs , increased both cyclin D1 mRNA and protein in resting fibroblasts (52). There is also evidence suggesting a role for p38 MAPK in the control of gene expression by post-transcriptional regulation of specific gene transcripts (54). Modulation of cyclin D1 expression by the MAPK cascades may thus result from a combination of both transcriptional and post-transcriptional events.
We thus conclude that MAPK signal transduction pathways play essential and differential roles in the regulation of cyclin D1 expression, thus establishing a link between receptor-coupled intracellular signaling and cell cycle machinery. The positive regulatory role of the p42/p44 MAPK cascade further emphasizes its essential function as a positive regulator of cell proliferation in response to growth signals. In contrast, the negative effect of the p38 MAPK cascade on cyclin D1 expression suggests that this signaling pathway may be detrimental to cell growth. A recent study has reported that activation of p38 MAPK together with the JNK cascade is critical for the induction of apoptosis in PC12 cells (22). This is in agreement with a possible negative function of the p38 MAPK signal transduction pathway in cell division mechanisms. Interestingly, in CCL39 cells, growth factor removal leads to rapid inactivation of the p42/p44 MAPKs (11,69,70), whereas the p38 MAPK activity seems to increase with the duration of starvation, 12 correlating with the abrogation of cyclin D1 expression. It thus seems that in contrast to the p42/p44 MAPK cascade that controls the G 0 -to G 1 -phase transition in fibroblasts and enhances cell differentiation in PC12 cells (14), the p38 MAPK cascade could be implicated in the maintenance of cell quiescence in fibroblasts and in promotion of programmed cell death in differentiated cellular systems.
While this manuscript was in preparation for submission, data were presented showing, in agreement with the results presented here, a positive modulation of cyclin D1 transcription by the p42 MAPK (53). This positive effect has been shown to be mediated by the transcription factor c-Ets-2 and dependent on the presence of a putative Ets-like binding domain on the proximal region of the human cyclin D1 promoter.