Cyclin D1 Expression Is a Major Target of the Camp-Induced Inhibition of Cell Cycle Entry in ®Broblasts

Home , Cyclin A, Cyclin D, Cyclin D1, Cyclin E, G1 phase

Cyclin D1 expression is a major target of the cAMP-induced inhibition of cell cycle entry in ®broblasts

Gilles L'Allemain1,*, JoseÂ e N. Lavoie1,*, Nathalie Rivard1,VeÂronique Baldin1,2 and Jacques Pouyssegur1

1Centre de Biochimie, CNRS-UMR134, FaculteÂ des Sciences, Parc Valrose, 06108 NICE Cedex 02, France

We previously described in the CCL39 hamster ®broblast of cAMP on growth factor-induced activation of p42/ cell line the inhibition of DNA synthesis reinitiation by p44 MAP kinases was very transient and could not agents that elevate cyclic AMP. Here, we show that 8Br- account for the growth inhibitory action of cAMP cAMP strongly blocks both the growth factor-induced (Kahan et al., 1992; McKenzie and Pouyssegur, 1996). increase in cyclin D1 protein expression and decrease in Therefore, other cAMP targets should be considered p27KIP1 protein levels, leaving untouched the levels of and we looked for target(s) downstream of the p42/p44 cyclin D3, cdk2 and cdk4. To assess the role of cyclin MAP kinase pathway. Progression through the G1 D1 in the cAMP-mediated inhibition of DNA synthesis, phase of the cell cycle is orchestrated by complexes we overexpressed the cyclin D1 gene in CCL39 and containing a G1-speci®c cyclin and a G1-speci®c cyclin- analysed the cAMP response in stable transfectants. We dependent kinase (Cdk): among those, cyclin D family showed that the kinase activities associated to G1 cyclin- members are found associated with either cdk4 or cdk6 cdk complexes are signi®cantly more resistant to cAMP in early G1 phase, and cyclin E with cdk2 in late G1 (see in cyclin D1 transfectants than in their normal counter- for reviews Draetta, 1994; Grana and Reddy, 1995; parts, although the serum-induced p27KIP1 disparition is Reed et al., 1994; Sherr, 1993). Activity of these still cAMP sensitive in cyclin D1 overexpressors. complexes can be inhibited by the presence of a cyclin Interestingly, the mitogen-induced DNA synthesis re- kinase inhibitor (CKI), which belongs to an expanding initiation is also much less inhibited by cAMP in cyclin new family of mammalian cell cycle modulators D1 transfectants than in control cells. These data clearly comprising so far p21Cip1, p27KIP1,p57KIP2 in one sub- INK4A INK4B INK4C INK4D establish that the cAMP-inducible blockade of the G1 family, and p16 , p15 , p18 and p19 in phase of the cell cycle can be partially alleviated by the other (Lees, 1995; Sherr and Roberts, 1995 for overexpression of cyclin D1 in hamster ®broblasts, thus reviews), with p27KIP1 being recently involved in the strongly suggesting that cyclin D1 protein is one of the inhibitory process induced by cAMP in macrophages major targets for cAMP inhibitory action in ®broblasts. (Kato et al., 1994). At the G1-4S phase transition, the tumor suppressor pRb, product of the retinoblastoma Keywords: Cyclin D1; cAMP; cyclin-CDK complexes; gene, is inactivated by hyper-phosphorylation, thus p27KIP1; DNA synthesis leaving free the transcription factors E2Fs to activate E2 site-containing gene promoters (Dyson, 1994 for a review). The inactivation of pRb has recently been

shown to be performed in vivo by a series of various G1 Introduction cyclin-kinase complexes (Horton et al., 1995; Kato et al., 1993; Suzuki-Takahashi et al., 1995). Not only pRb Cyclic AMP has been described as a mediator of regulates cyclin D1 expression (Muller et al., 1994), but numerous physiological processes. Large increases in a strict relationship between cyclin D1 and pRb has cAMP concentration inside the cell are generally growth been established: if pRb function is de®cient, the cyclin inhibitory for most cell lines of mesenchymental origin. D1 requirement in the G1-4S progression becomes Accordingly, we previously published (Magnaldo et al., dispensable (Bates et al., 1994; Lukas et al., 1994a, 1989) that any sustained increase in cAMP level was 1995a). detrimental for DNA synthesis reinitiation in the Here, we addressed the question of which, if any, hamster ®broblast cell line CCL39. But the exact cell cycle regulatory protein was targeted by cAMP. A mechanism by which cAMP exerts its inhibitory action previous report from our laboratory (Magnaldo et al., remains largely unknown or somewhat controversial. 1989) targeted the cAMP growth inhibition as taking

We subsequently showed that activation of the p42/p44 place in the early G0/G1 phase, more precisely during MAP kinase pathway (ERK pathway) was an the ®rst 6 h of growth factor addition. Interestingly, obligatory step to trigger cell mitogenicity (PageÁ s et addition of cAMP agonists for the ®rst 4 h following al., 1993, 1994). This signalling cascade was indeed a growth factor stimulation delayed S phase entry by potential target for cAMP inhibition. However, we now roughly the same time period (Magnaldo et al., 1989). show that in various ®broblastic cell lines, the inhibition Thus, the cell cycle regulatory protein cyclin D1, that has been described as an essential component of an

early G1 checkpoint (Pagano et al., 1994) and as Correspondence: G L'Allemain governing cell cycle exit (Zwijsen et al., 1996), appeared 2Present address: Institut de Pharmacologie et Biologie Structurale, to be a good candidate. Indeed, cyclin D1 over- CNRS, 31077 Toulouse, France expression deregulated cell growth of normal cells (Han *The ®rst two authors contributed equally to this work Received 17 November 1996; revised 8 January 1997; accepted et al., 1995; Jiang et al., 1993; Quelle et al., 1993; 13 January 1997 Resnitzky et al., 1994) whereas antisense to this cyclin cAMP inhibits cell growth via cyclin D1 GL'Allemainet al 1982 inhibited growth of transformed cells (Zhou et al., cyclin D1, cyclin D3, cdk2, cdk4, p27KIP1 or pRb. 1995). Here, we used the 8Br-cAMP, a non-hydro- Upper panel of Figure 1 indicates that cyclin D1 lysable but diusible cAMP analog, and found that protein is not present (or barely detectable) in cyclin D1 protein expression was strongly repressed by quiescent cells but its expression could be detected as such a sustained increase in cAMP level. In order to soon as 6 h after serum stimulation, with the protein establish whether or not cyclin D1 was the major level increasing between 6 and 20 h. Interestingly, protein targeted by the inhibitory pathway triggered by 1mM 8Br-cAMP strongly blocked such an expression cAMP, we then ectopically expressed the human cyclin over all this period of time. Like in other studies D1 gene in the non-transformed hamster ®broblast cell (Lukas et al., 1995a; Matsushime et al., 1994), the line CCL39 and analysed the resulting sensitivity to appearance of a slow migrating form of cyclin D1

cAMP of G1 cyclin-cdk complexes and of DNA correlates with the strongly induced major cyclin D1 synthesis reinitiation. band. Its identity is unknown but cannot correspond to either the cyclin D2 or D3 isoforms. Besides cyclin D1 (Baldin et al., 1993; Lukas et al., 1994b), both cyclin D2 and cyclin D3 have also been Results involved in controlling the G1 phase (Ando et al., 1993; Kiess et al., 1995; Kiyokawa et al., 1994; Lukas et al., cAMP eect on kinetics of expression of various cell 1995b; Skapek et al., 1995) with cyclin D3 mostly cycle regulatory proteins in hamster ®broblasts implicated in cell dierentiation (Kiess et al., 1995; We ®rst examined the eect of a pre-incubation with Kiyokawa et al., 1994; Skapek et al., 1995), a process 1mM 8Br-cAMP on the serum-induced protein inhibited by cyclin D1 (Skapek et al., 1995). We thus expression of dierent cell cycle regulatory proteins in checked for their presence into CCL39 cells. Upper the hamster ®broblast line CCL39. Figure 1 shows a panel of Figure 1 also shows that the cyclin D3 isoform

series of immunoblots loaded with CCL39 cell lysates was already present in G0-arrested cells but its and decorated with speci®c antibodies against either expression was only slightly induced by growth factors

cycD1

cycD3

FCS 0 6 10 15 20 6 10 15 20 (hours) 8Br-cAMP – +

pRb

cdk4

cdk2

Kip1

FCS 0610 15 20 0610 15 20 (hours) 8Br-cAMP – +

Figure 1 Eect of 8Br-cAMP on the expression levels of various cell cycle regulatory proteins in hamster ®broblasts. G0-arrested CCL39 cells (0) were activated with 10% FCS with (+) or without (7) for the indicated times before cell lysis. All protein lysates (75 mg) were separated by 10% SDS ± PAGE (except 7.5% for pRb), blotted and immunodecorated with speci®c antibodies as indicated cAMP inhibits cell growth via cyclin D1 GL'Allemainet al 1983 and resistant to the 8Br-cAMP treatment, thus D1) exhibited much higher levels of cyclin D1 excluding its putative role in cAMP inhibition. Use (respectively, ®ve and 12 times more as assessed by of four dierent antibodies, monoclonal and poly- densitometry) than their control counterparts trans- clonal, did not allow us to detect cyclin D2 isoform in fected with the empty vector (39-C, PS-C). Interest- CCL39 cells (data not shown), a situation encountered ingly, the PS-D1 clone was expressing 3 ± 4 times more in ®broblasts as well as in other cell types (Tam et al., cyclin D1 than 39-D1. Main panel of Figure 2 shows 1994). the timing of entry into S phase exhibited by these four On the lower panel of Figure 1, we analysed the cell lines after serum deprivation for 24 h followed by hyper-phosphorylation state of pRb which is, in late G1 reactivation by growth factors. Thymidine incorpora- phase, a landmark for normal cells passing the tion in resting cells was very low and essentially restriction point and entering S phase (Grana and identical both in control and D1-transfected clones, Reddy, 1995 for a review). Not less than 10 h of serum indicating that the cyclin D1 transfectants were not addition are necessary to obtain a signi®cant portion of relaxed for growth factor requirement. Data in Figure the slow migrating (i.e. hyper-phosphorylated) form of 2 also shows that the DNA synthesis induced by serum pRb. This result is in agreement with the appearance of was much higher in D1 transfectants than in control cyclin D1 at 6 h since reports indicate that cyclin D1, cells, suggesting that cyclin D1 is rate-limiting for in association with cdk4, could in part be responsible DNA synthesis in hamster ®broblasts. From the same for in vivo hyper-phosphorylation of pRb (Horton et thymidine-pulse experiment, the cyclin D1 transfectants al., 1995; Kato et al., 1993; Suzuki-Takahashi et al., appear to enter S phase sooner than their normal 1995). After serum addition, the ratio between hyper- counterparts (see Discussion). vs hypo-phosphorylated pRb, as determined by densitometric scanning, reaches 50% at 10 h. Interest- cAMP eect on cyclin D1 protein expression in cyclin ingly, 1 m 8Br-cAMP almost completely blocks the M D1 transfectants appearance of the hyper-phosphorylated inactive form of pRb. Figure 3 shows the cAMP eect in D1 transfectants vs The speci®city of cAMP action was then assessed by their normal counterparts for cyclin D1 protein analysing the levels of cdk4 or cdk2, two proteins expression either from whole cell lysates (Figure 3a) whose activity is required for G0/G1-4S phase or from anti-cdk4 immunoprecipitates (Figure 3b). progression. The same cAMP increasing treatment Whereas the serum-induced cyclin D1 expression is did not induce a signi®cant inhibition of the protein severely inhibited by 1 mM 8Br-cAMP in the control expression of cdk2 or cdk4, a result indicative of the transfectants (39-C and PS-C), the cyclin D1 protein rather speci®c cAMP action since cdk4 protein levels are unchanged in the 39-D1 and PS-D1 clones expression was already described as being aected by (Figure 3a). This result indicates that cyclin D1 TGFb, an inhibitor of DNA synthesis reinitiation in transfectants are no longer sensitive to cAMP in epithelial cells (Ewen et al., 1993). terms of cyclin D1 expression, an expected result with Finally, because variations of p27KIP1 levels were the cyclin D1 cDNA placed under the control of the described as mediating the cAMP inhibition of DNA heterologous CMV promoter. synthesis in macrophages (Kato et al., 1994), the same lysates were also checked for any variation of p27KIP1.We found that the decrease in p27KIP1 levels observed after serum addition (40% at 6 h, 50% at 10 h, 60% at 20 h as assessed by densitometry analysis) is totally blocked by 8Br-cAMP addition. Evenmore, the same analysis reveals a slight increase of 20 to 30% in cells treated by 39-C 39-D1 PS-C PS-D1 8Br-cAMP as supposed to the untreated cells. cyc D1¨

Kinetics of S phase entry in cyclin D1 transfectants To examine precisely the role of the cyclin D1 in the inhibitory action of cAMP, transfection experiments of the human cyclin D1 cDNA were then conducted both in the hamster ®broblast line CCL39 and in its mutant derivative PS120, which lacks a functional Na+/H+ exchange protein (PouysseÂ gur et al., 1984). Thus, the Na+/H+ antiporter activity was used as the selective marker in co-transfection experiments (see Materials and methods section). From independently transfected populations of hamster ®broblasts, two clones exhibit- ing dierent levels of cyclin D1 were isolated, and then Figure 2 Kinetics of S phase entry in parental and cyclin D1 expressing cells. Quiescent 39-C ( ~), 39-D1 (~), PS-C (&) and analysed for their kinetics of S phase entry. Among 27 PS-D1 (&) cells were stimulated with 10% FCS for the indicated colonies, 39-D1 and PS-D1 clones, representative of times. At each time-point, the incorporation of radiolabelled each transfected population (respectively CCL39 and thymidine ([3H]thymidine, 4 mCi/ml) into DNA was performed PS120), were chosen for expressing dierent levels of for 30 min during the last half-hour of serum incubation. TCA- insoluble counts are the mean of duplicates. A parallel experiment cyclin D1 in resting cells. An anti-cyclin D1 immuno- with 1 h pulses of [3H]thymidine gave identical results. Inset: blot (Figure 2, inset) made with lysates from resting Anti-cyclin D1 immunoblot loaded with 50 mg of protein lysates cells showed that both D1 transfectants (39-D1, PS- extracted from quiescent cells cAMP inhibits cell growth via cyclin D1 GL'Allemainet al 1984 a

cyc D1¨

FCS –+ + – + + –+ + – + + Br-cAMP –+ – – + – –+ – – + – PS-C PS-D1 39-C 39-D1

b IgG ¨

cyc D1 ¨ PS-C 39C 39D1 PSD1 PS-C 39C 39D1 PSD1 PS-C 39C 39D1 PSD1 FCS – + + Br-cAMP – – + Figure 3 Eect of 8Br-cAMP on serum-induced cyclin D1 protein expression and cyclin D1-CDK4 complex formation in parental vs cyclin D1 expressing cells. Quiescent 39-C, 39-D1, PS-C and PS-D1 cells were (+) or not (7) stimulated with serum for 12 h in the presence (+) or in the absence (7)of1mM8Br-cAMP before lysis. (a) An anti-cyclin D1 immunoblot from whole cell lysates is shown. Amounts of protein loaded were 140 mg for PS-C and PS-D1 lanes, and 25 mg in 39-C and 39-D1 lanes. Note that the upper band is only detected when the main cyclin D1 band is strongly induced. Parallel immunoblots made from cells stimulated with FCS during 6, 8 or 15 h gave essentially the same pattern of inhibition. (b) Lysates from 100 mm dishes (2 mg protein) were immunoprecipitated with anti-cdk4 antibodies then immunoblotted with anti-cyclin D1 antibodies

On Figure 3b, immunoprecipitation experiments 39D1 39-C PSD1 PS-C with anti-cdk4 antibodies are shown for the four transfectants, and their content analysed for the KIP1 presence of the cyclin D1 protein. Although no cyclin 8Br-cAMP ––+–+– –+–––+ D1-cdk4 association was detected in quiescent cells, much more cyclin D1-cdk4 complex was present in D1 0 FCS FCS 0 0 FCS 0FCS transfectants than in control transfectants, suggesting Figure 4 Eect of 8Br-cAMP on serum-induced p27KIP1 protein that cyclin D1 was rate-limiting for complex forma- expression in parental vs cyclin D1 expressing cells. Quiescent 39- tion. We thus identi®ed by co-immunoprecipitation the C, 39-D1, PS-C and PS-D1 cells were (FCS) or not (0) stimulated with serum for 15 h in the presence (+) or in the absence (7)of KIP1 cdk protein that mainly associated with cyclin D1 in 1mM 8Br-cAMP before lysis. An anti-p27 immunoblot hamster ®broblasts as being cdk4. Interestingly, such loaded with 70 mg of proteins from whole cell lysates is shown. an association was resistant to 1 mM 8Br-cAMP Parallel immunoblots made from cells (parental vs D1 treatment in both the cyclin D1 transfectants and transfectants) stimulated with FCS during 25 h gave essentially the control transfectants, thus suggesting that the the same pattern of sensitivity to cAMP mechanism of complex formation was not targeted by cAMP.

p27KIP1 protein levels both in the control transfectants KIP1 cAMP eect on p27 protein expression in D1 and in the cyclin D1 transfectants. Thus, in D1 transfectants transfectants, the variations in p27KIP1 levels stay Two dierent regulatory mechanisms can explain the sensitive to cAMP (Figure 4) in spite of the growth factor-induced variations in p27KIP1 levels: constitutive levels of cyclin D1 protein and high levels activation of its degradation (Pagano et al., 1995) of cyclin D1/cdk4 activity (see next section). These data and inhibition at the translational level of its synthesis indicate that cyclin D1 can be induced without a prior (Agrawal et al., 1996; Hengst and Reed, 1996). reduction in the p27KIP1 levels, a result suggesting that Whatever the mechanism of regulation involved in the growth factor-induced variations in cyclin D1 and the stabilization of p27KIP1 levels by cAMP, we p27KIP1 levels are regulated independently. compared the sensitivity of p27KIP1 to cAMP in the four cell transfectants 39-C, PS-C, 39-D1 and PS-D1. Eect of cAMP on FCS-induced cyclin-associated kinase KIP1 Figure 4 shows that the resting level of p27 protein activities in D1 transfectants is roughly the same in these four cell lines. This result suggests that cyclin D1 overexpression does not To analyse more precisely the mechanism of inhibitory KIP1 in¯uence the steady-state level of p27 in G0 state. action of cAMP, we then compared in cyclin D1 The same ®gure also demonstrates that serum addition transfectants vs control cells the levels of cyclin D1- elicits the p27KIP1 disappearance in the four cell lines. associated kinase activity. After an immunoprecipita- More interestingly, treatment with 1 mM 8Br-cAMP tion with anti-cyclin D1 antibodies, cell lysates were appears to prevent the serum-induced reduction in tested for their capacity to phosphorylate the cAMP inhibits cell growth via cyclin D1 GL'Allemainet al 1985 exogenous substrate pRb-GST. In G0-arrested cells, the inhibition), as compared to control transfectants (80 to incorporation of [32P] into pRb-GST was undetectable 100% inhibition). in control cells and barely detectable in D1 transfectants, whereas 12 h of serum addition triggered high cAMP eect in late G phase of cyclin D1 transfectants D1 kinase activity in PS-D1 cells (Figure 5), the extent 1 of which appearing as a function of the expression level Considering that mid- and late-G1 cyclin dependent of cyclin D1 protein (compare PS-D1 to 39-D1). kinase activities were becoming resistant to cAMP Interestingly, 1 mM 8Br-cAMP treatment induced very inhibition in D1 transfectants (see Figures 5 and 6), low inhibition on cyclin D1-associated activity in D1 we examined in late G1 phase the phosphorylation state transfectants (25% in 39-D1 and 12% in PS-D1 as of pRb along with the cyclin A status in the four cell compared to 95% and 90% in their corresponding transfectants. Densitometry scanning from results control transfectants). These results indicated that the shown in Figure 7a indicated that after 12 h of serum cyclin D1-associated kinase activity induced by serum addition, the ratio between the hyper- and the hypo- in the cyclin D1 transfectants is almost fully resistant phosphorylated forms of pRb is approximately 50 : 50 to the cAMP treatment. whatever the cell line. Interestingly, the serum-induced Because cyclin D1 is also known to form complexes hyper-phosphorylation of pRb was fully blocked by with cdk2 and cdk6, and cdk4 to associate with 8Br-cAMP only in control transfectants. Truly, a cyclins D2 and D3 (Sherr and Roberts, 1995 for signi®cant presence of hyper-phosphorylated pRb form review), we tested in D1 transfectants the sensitivity is still detected in 8Br-cAMP-treated D1 transfectants, a of cdk4 activity to the same 8Br-cAMP treatment. Figure 6a showed that kinase activities exhibited by anti-cdk4 immunoprecipitates from D1 transfectants stimulated 12 h with serum, are as resistant to cAMP (30% and 15% inhibition for respectively 39-D1 and PS-D1) as those exhibited by anti-cyclin D1 immunocomplexes issued from the same cells (compare with Figure 5). Because cyclin E and cyclin D1 were recently described as being located on distinct although interdependent pathways (Ohtsubo et al., 1995; Resnitzky et al., 1995; Roussel et al., 1995), we then asked, whether or not the enzymatic activity of cyclin E-cdk2 was sensitive to cAMP in the cyclin D1 transfectants. The results, shown in Figure 6b and c, unambiguously indicated that anti-cdk2 and anti-cyclin E immunocomplexes exhibited kinase activities much more elevated in D1 transfectants than in control transfectants. Interestingly, the enzymatic activities of both immunocomplexes were only slightly inhibited by 1mM 8Br-cAMP in D1 transfectants (25 to 50%

b ¨

Figure 6 Eect of 8Br-cAMP on serum-induced cyclin E- associated, CDK2 and CDK4 kinase activities in parental and cyclin D1 expressing cells. Quiescent 39-C, 39-D1, PS-C and PS- Figure 5 Eect of 8Br-cAMP on serum-induced cyclin D1 D1 cells were (+) or not (7) stimulated with 10% FCS either for associated kinase activity in parental vs cyclin D1 expressing 12 h (a), or for 15 h (b and c), in the presence (+) or in the cells. Quiescent 39-C, 39-D1, PS-C and PS-D1 cells were (+) or absence (7)of1mM 8Br-cAMP. Kinase activity of immuno- not (7) stimulated with 10% FCS for 12 h in the presence (+) or complexes generated with speci®c immunoprecipitating antibodies in the absence (7)of1mM 8Br-cAMP. For each cell line, the is shown on a for cdk4 antibodies, on c for cyclin E antibodies, kinase activity associated with anti-cyclin D1 immunocomplexes on b for cdk2 antibodies. Parallel data were obtained in was determined. (a) shows an autoradiography of the gel; (b) independent experiments. Note that the very slight inhibitory presents the corresponding counts. The data presented are typical eect of cAMP which is observed in PS-D1 cells, is also detected of at least three experiments with ®ve times less of PS-D1 cell lysates cAMP inhibits cell growth via cyclin D1 GL'Allemainet al 1986 a 39C PS-C 39-D1 PS-D1

pRb-PP ¨ pRb-P ¨ FCS –++–++–++–++ Br-cAMP –+––+––+––+–

b 39-C PS-C 39-D1 PS-D1

cyclin A ¨

Br-cAMP –+–+–+–+ FCS Figure 7 Eect of 8Br-cAMP on serum-induced hyper-phosphorylation of pRb and cyclin A expression in parental and cyclin D1 expressing cells. Quiescent 39-C, 39-D1, PS-C and PS-D1 cells were (+) or not (7) stimulated with serum for 12 h in the presence (+) or in the absence (7)of1mM 8Br-cAMP for the same time before lysis. (a) anti-pRb immunoblot loaded with 80 mg proteins. The hypo- (lower band) and hyper-(upper band) phosphorylated forms of pRb are indicated by arrows; note that immunoblots of cells stimulated with FCS for 16 h gave the same inhibitory pattern for each cell line. (b) anti-cyclin A immunoblot loaded with 80 mg proteins. Note that cyclin A protein in G0 arrested cells is not detectable Figure 8 Eect of 8Br-cAMP on serum-induced DNA synthesis reinitiation in parental and cyclin D1 expressing cells. 24 h-serum deprived cells are stimulated with 10% serum for 18 h with or without 1 mM 8Br-cAMP. DNA synthesis reinitiation is measured as the incorporation of tritiated thymidine (2.5 mCi/ml) during the result consistent with the phenotype of resistance to last hour of serum addition (0.86104 c.p.m./mg for 39-C, 4 4 cAMP already exhibited in early and mid-G1 phases by 1.9610 c.p.m./mg for 39-D1, 2.3610 c.p.m./mg for PS-C and these cells. The anti-cyclin A immunoblot shown on 2.96104 c.p.m./mg for PS-D1). Data are plotted as % of Figure 7b clearly demonstrates that the expression levels inhibition induced by incubation with 1 mM 8Br-cAMP. Note that the same cAMP treatment is less inhibitory on the of cyclin A is much more intense in D1 transfectants cumulative incorporation of [3H]thymidine over 24 h, but the than in control cells after being challenged with serum. inhibition ratio between D1 transfectants and control cell lines Noteworthy, addition of 1 mM 8Br-cAMP totally (respectively, 39-D1 vs 39-C and PS-D1 vs PS-C) stays blocks cyclin A expression in control transfectants but qualitatively the same has no eect in D1 transfectants. Thus, the early resistance to cAMP developed by cells overexpressing

cyclin D1 is also detected in late G1 for cyclin A expression. Considering the fact that cyclin A gene Discussion promoter carries E2F box, this result is in agreement with the presence of hyper-phosphorylated pRb in The growth inhibitory action of cAMP is known for cAMP-treated D1 transfectants (Figure 7a). more than 20 years (Pastan et al., 1975) and yet the mechanism by which this second messenger exerts its inhibitory action has remained unknown or highly cAMP eect on DNA synthesis reinitiation in cyclin D1 controversial. Recently two interesting mechanisms transfectants have been proposed for its growth negative action. To analyse if the phenotype of resistance to cAMP In the ®rst one, it was proposed that the p42/p44 would be also detected later for S phase entry as it was MAPK cascade was the target of this inhibition in

for G1 phase progression, we assessed the DNA various ®broblast cell lines or in smooth muscle cells. synthesis reinitiation in the same four cell lines by Indeed, activation of PKA has been shown (Burgering measuring the incorporation of radiolabelled thymidine and Bos, 1995, for a review) to inhibit Raf-1, the for 1 h after 17 h of serum stimulation. Results shown upstream kinase of the MAPK module. Although this in Figure 8 indicated that 1 mM 8Br-cAMP inhibited action was satisfying in regard of the essential role serum-induced DNA synthesis by respectively 75 and played by the MAPK cascade in cell proliferation 80% in PS-C and 39-C control transfectants, but by (PageÁ s et al., 1993), we showed that this inhbition was only 30 and 60% in PS-D1 and 39-D1. These results too transient in nature to account for the growth indicated that the inhibitory action of cAMP on DNA inhibitory action of cAMP (McKenzie and Pouysse- synthesis can be partially alleviated by increasing cyclin gur, 1996). D1 levels in hamster ®broblasts. Evenmore, every The second and perhaps more appealing mechanism cyclin D1 expressing clone isolated during this study was the demonstration that cAMP-mediated growth exhibited a lower sensitivity to cAMP than any of the arrest is caused by an elevation of p27Kip1, a CK1

control transfected clones isolated in parallel (data not known to block cell cycle progression by targeting G1 shown). Thus, cyclin D1 expression truly constitutes cyclin-dependent kinases. This action was reported to one of the major steps targeted by cAMP in its operate in cAMP-treated macrophages (Kato et al., inhibitory pathway on DNA synthesis. 1994). cAMP inhibits cell growth via cyclin D1 GL'Allemainet al 1987 In the present study, we have shown that the al., 1996; N Rivard and J Pouyssegur, manuscript in concentration of cAMP (8Br-cAMP) known to preparation). We proposed that G1 cyclin-dependent antagonize cell cycle re-entry in response to growth kinases, once activated, could initiate the degradation factors has, in fact, a strong inhibitory action on cyclin of p27Kip1 by the proteasome machinery (Lavoie et al.,

D1 induction. This repression was similarly observed 1996b), thus acting as `sensor' of G1 progression. upon prostaglandin E1 treatment, a hormone which Consistent with this proposal, cells derived from activates adenylyl cyclase, and reproduced in all cell p27Kip1-de®cient mice are still sensitive to TGF b and types analysed; the immortalized hamster lung fibro- rapamycin (Nakayama et al., 1996). With such Kip17/7 blast line CCL39, secondary cultures of hamster or cells now available, it would be very interesting to mouse ®broblasts, rat vascular smooth muscle cells determine as well their sensitivity to cAMP. (data not shown). Similar eects have been previously One of the most striking observations from the reported in other cell types (Cocks et al., 1992; Gagelin present study is that 8Br-cAMP still prevents p27Kip1 et al., 1994; Sewing et al., 1993; Vadiveloo et al., 1996; degradation in cyclin D1-transfectants in which the G1 Winston et al., 1996). As expected from such a cyclin block has been partially released as shown by pRb D1 repression, the resulting cyclin D1-associated cdk4 hyperphosphorylation, cyclin A induction and thymi- and cdk2 activities, as well as pRb hyperphosphoryla- dine incorporation (see Figures 7 and 8). If our tion, were blunted by 8Br-cAMP. We also noticed that previous hypothesis is proven to be correct, then the 8Br-cAMP prevented the growth factor-mediated situation of p27Kip1 increasing in response to cAMP degradation of p27Kip1 and even led to a slight increase represents an intriguing feature. Based on results (up to 30%) after a 24 h treatment (see Discussion showing that 8Br-cAMP is no longer able to block later). the serum-induced p27Kip1 degradation in hamster To test whether inhibition of cyclin D1 expression ®broblasts deprived in protein kinase A (N Rivard was the cause of this pleiotypic eect exerted by cAMP and F McKenzie, personal communication), we know Kip1 in mid and late G1 phase, we generated cell lines that cAMP exerts its eect on p27 through PKA. constitutively expressing cyclin D1. These clones, in Recently, PKA was shown to phosphorylate, at least in which cyclin D1 is now insensitive to cAMP action, are vitro, p27Kip1 (Yin Sun, personal communication). still capable of G0 arrest upon serum-deprivation. Through either this action or the possible PKA- Stably expressing cyclin D1 increased cdk4 and cdk2 mediated phosphorylation of some components of the activities, thus suggesting that cyclin D1 was rate- proteasome, the growth signal initiating p27Kip1 limiting. The apparent shortening of G0/G1 phase seen degradation should be prevented. In this regard, in Figure 2 might simply re¯ect a better synchrony for analysing the p27Kip1 regulation in cells for which cell cycle re-entry. Interestingly, in these cyclin D1- cAMP is a growth promoting agent would be very expressing cells, 8Br-cAMP failed to decrease signifi- informative. Accordingly, one could speculate that in cantly both cyclin D1-associated and cdk4 activities. Swiss 3T3 or in FRTL5 cells for example, the addition Kip1 However, kinasic activities in late G1 such as cyclin E- of cAMP will induce the degradation of p27 . associated and total cdk2 activities were slightly In conclusion, this report has demonstrated that, in reduced by 8Br-cAMP, but comparably much less addition to an elevation of p27Kip1, cAMP-mediated than in control cells (30 ± 40% inhibition instead of growth inhibition is triggered via a suppression of 80% in control cells). As a consequence, the serum- cyclin D1 induction. Interestingly, these two separate induced pRb hyperphosphorylation was only slightly actions are complementary since both cooperate aected by cAMP in D1-expressing cells. Inhibition of towards the inhibition of G1 cyclin-dependent kinases. DNA synthesis reinitiation, although still sensitive to We propose that overexpression of cyclin D1 behaves 8Br-cAMP addition in cyclin D1-transfectants, was as an `anti p27Kip1 factor': generating higher amounts of severely attenuated in particular in the high expressor cyclin D1/cdk4 complexes will allow titration of the PS-D1 (25% inhibition instead of 75% in control cells). inhibitor, thus explaining why cyclin D1 transfectants These results clearly demonstrate that cyclin D1 is a can grow with undegraded levels of p27Kip1. From this critical target of cAMP-mediated growth inhibition, conclusion, we could anticipate that p27Kip1-de®cient although not unique. The mechanism by which PKA ®broblasts are less, but still inhibited by cAMP, antagonizes growth factor-mediated cyclin D1 induc- whereas a constitutive expression of cyclin D1 in tion is yet unknown. At least, this negative action is these cells should totally release the inhibition. not at all mediated through an attenuation of the p42/ Preliminary experiments using anti-sense p27Kip1 ap- p44 MAPK activity, a signalling cascade crucial for proach are in support of this conclusion. cyclin D1 induction (Lavoie et al., 1996a). Another interesting point is the control of p27Kip1 levels by cAMP. As originally reported in macrophages Materials and methods (Kato et al., 1994), agents increasing cAMP have a remarkable eect in preventing the mitogenically- Cell culture and cell transfection induced p27Kip1 down-regulation. This action certainly represents an attractive mechanism for cAMP- Cell culture The CCL39 cell line was from the American mediated cell growth. However, we recently hypothe- Type Culture Collection (Bethesda, USA). The CCL39- mutant derivative PS120 was selected previously (Pouys- sized (Rivard et al., 1996) that this inhibition of p27Kip1 seÂ gur et al., 1984) as being totally deprived in Na+/H+ degradation merely re¯ects the consequence of a G1 exchange activity, a biochemical characteristic used in our block. Indeed this feature is not restricted to cAMP, laboratory to select positive clones after co-transfection but observed in various conditions leading to growth experiments with the reporter Na+/H+ antiporter gene inhibition such as treatment with rapamycin, TGF b, (NHE3) and the gene of interest (PageÁ s et al., 1993, 1994). lovastatin or high cell density cultures (Vadiveloo et Cells were cultivated in DME medium supplemented with cAMP inhibits cell growth via cyclin D1 GL'Allemainet al 1988

7.5% FCS (Gibco, France) in a humidi®ed 5% CO2 provided by Dr Bernard Ducommun (IPBS, CNRS, atmosphere. Cells were rendered quiescent by a 24 h Toulouse, France) and used at the 1 : 1000 dilution in WB serum deprivation treatment. and at the 1 : 500 dilution in IP. Anity puri®ed anti-cyclin A antibodies (1 : 300 in WB) Expression vectors The cDNA coding for the rat NHE3 was a kind gift of Dr Sobczack-Thepot, HoÃ pital Necker, isoform of the Na+/H+ antiporter gene (Orlowski et al., Paris. 1992) was inserted (Fafournoux et al., 1994; Noel et al., Whenever available, antibodies were tested in hamster 1996) into the pECE vector (InVitrogen). The human ®broblasts for their speci®city in WB and in IP by cyclin D1 cDNA was previously cloned (Baldin et al., comparison with non-immune sera. We also veri®ed that 1993) into the pX vector (InVitrogen), and thus placed anti-cyclin D1 or anti-cdk4 Abs, and anti-cyclin E or anti- under the control of the heterologous CMV promoter and cdk2 Abs were able to sustain kinasic activity towards enhancer. respectively the exogenous substrates pRb-GST or H1 Transfection CCL39 and PS120 cells were transfected histone. with the calcium phosphate precipitation method as previously described (PageÁ s et al., 1993, 1994). 50% con¯uent cells in 100 mm dishes were co-transfected Immunoprecipitation overnight with 2 mgoftheNa+H+ antiporter sequence Cells were cultivated in 100 mm dishes until con¯uence, (NHE3 being used as the selector gene for both PS120 and then rendered quiescent by serum starvation for 24 h and CCL39 cells) along with 18 mg of the cyclin D1 expression re-stimulated before harvesting. Cell lysis was performed vector (gift of Dr VeÂ ronique Baldin, IPBS, CNRS, by a 10 min incubation on ice with 1 ml/dish of lysis buer Toulouse, France). For CCL39 cells that express endogen- (150 mM NaCl, 1 mM EDTA, 40 mM Tris, pH 7.6+1% ous NHE1, this exchanger was inhibited by the speci®c Triton X100) supplemented with protease inhibitors inhibitor HOE694 (a kind gift of Dr W Scholz, Hoescht (0.1 mM PMSF, 1 mg/ml pepstatin, 10 mg/ml aprotinin, AG, Frankfurt, Germany) which speci®cally blocks the 10 mg/ml leupeptin) and phosphatase inhibitors (0.1 mM endogenous NHE1 isoform with little interference with NaVO ,20mM para-nitro phenyl phosphate, 40 mM b- exogenous NHE3 (Counillon et al., 1993). As a result, most 4 glycerophosphate). Lysates were then cleared by centrifu- colonies expressing a functional NHE3 protein by virtue of gation (10 000 g, 10 min) before a 2 h incubation at 48C their resistance to intracellular acidi®cation would express with protein A-Sepharose (Sigma, France) pre-incubated cyclin D1. Stable transfectants were obtained after three to for 1 h with immune sera. Each antibody having a dierent four rounds of H+ suicide selection over 10 to 12 days of titer, a 1 : 50 to 1 : 250 dilution was used for maximal culture. Then clones were picked and recloned for immunoprecipitation. Then, immunocomplexes were ®rst puri®cation. washed four times with ice-cold lysis buer, and three Immunoblotting Equal amounts of protein from whole cell times with ice-cold kinase buer (20 mM PNPP, 10 mM

lysates were loaded and separated through 8.5% SDS ± MgCl2,1mM DTT, in 30 mM HEPES, pH 7.4), before PAGE. Gels were then equilibrated in transfer buer performing the kinase assay. (50 mM Tris-HCl, 180 mM glycine) and electrotransferred onto nitrocellulose (Amersham, France) for 2 h at 50 volts or overnight at 150 mA. Transfer quality was checked by Kinase assay staining with Ponceau S. Western blots were blocked for Kinase reaction was started by incubating the washed 1 h at 378C with 10% low fat milk in PBS+0.1% Triton immunocomplexes at 308C in the presence of the X100 (Sigma, France) before decoration for 2 ± 3 h at room corresponding in vitro substrate and of [32P]ATP (Amer- temperature (RT) with speci®c antibodies, then washed in sham France) at 20 ± 100 mM,1±5mCi/assay. In order to PBS+0.1% Triton X100 (4610 min). This was sequen- get reasonable counts, higher speci®c radioactivity was tially followed by 1 h incubation (RT) with goat either used for cyclin D1-associated and cdk4 enzymatic anti-rabbit (for polyclonal Abs) or anti-mouse (for mAbs) activities. The kinasic substrates used were either histone IgG horseradish peroxidase conjugated (Sigma, France). H1 (Boehringer, Mannheim, Germany) for cyclin E- Peroxydase activity was revealed using an enhanced associated and cdk2 activities, or pRb-GST for cyclin chemiluminescence (ECL) kit (Amersham, France). D1-associated and cdk4 activities. After 30 min, the reaction was stopped by addition of hot Laemmli's Antibodies source and characterization buer. Radiolabelled substrates were separated from immunocomplexes by SDS ± PAGE (10% for histone H1, Mouse monoclonal antibodies against pRb were purchased 7.5% for pRb-GST), and autoradiographed, then cut out from Pharmingen (Clinisciences, Paris, France) and used at and counted. The duration time of the kinase assay was the 1 : 200 dilution in Western blotting experiments (WB). chosen after veri®cation that the incorporation of [32P] on The previously characterized mouse monoclonal antibodies histone H1 was linear over at least 45 min. against cyclin D1 (DCS-6), cyclin D2 (DCS-3, DCS-5), cyclin D3 (DCS-22) and against p27KIP1 (DCS-72) (Lukas et al., 1995a, 1994b; Rivard et al., 1996) were generous gifts Puri®cation of pRb-GST from Dr Jiri Bartek (Danish Cancer Society, Copenhagen, Denmark). Monoclonal DCS-6 and DCS-22 Abs were used pRb-GST construct in pGEX-2T was a kind gift of Dr at the 1 : 1000 dilution in WB; anti-cyclin D2 Abs (DCS-3 Joan MassagueÂ (Memorial Sloan-Kettering Cancer Center, and DCS-5) were tested from 1 : 1000 to 1 : 100 dilution. New York). Overnight cultures of JM109 were diluted Rabbit polyclonal antibodies for cyclin D1, cyclin E, and 1 : 10 in fresh culture broth for 2 h then IPTG (0.1 mM) cdk2 were from Dr VeÂ ronique Baldin (IPBS, CNRS, supplemented for 3 h before lysis in ice-cold TNEN buer Toulouse, France). The polyclonal peptide antibodies were (20 mM Tris, pH 7.6, 150 mM NaCl, 5 mM EDTA, 0.5% raised against respectively the recombinantly produced NP40). Lysates were sonicated then cleared by centrifuga- human cyclin D1 protein, the N-terminus part (13 aa) of tion before applying them on Agarose-GSH column the human cyclin E sequence, or the C-terminus part (11 aa) (Pharmacia, France) pre-equilibrated in TNEN buer. of the human cdk2 sequence. Polyclonal Ab against cyclin D1 After extensive washing in TNEN buer, the pRb-GST was used at the 1 : 2000 dilution in WB and at the 1 : 250 protein was eluted with three volumes of column of 50 mM dilution in immunoprecipitation experiments (IP). Polyclonal Tris, pH 8.0+10 mM glutathione, then aliquoted at anti-cdk4 against the human C-terminus sequence was kindly 7808C. 1 mg of pRb-GST was used per assay. cAMP inhibits cell growth via cyclin D1 GL'Allemainet al 1989 DNA synthesis reinitiation phosphate; pRb: retinoblastoma protein; SDS: sodium dodecyl sulfate. Cells were grown in 24-well dishes until con¯uence, then serum-deprived for 24 h before stimulation with mitogen for various times. All these steps were performed either in DME-Ham (1 : 1) medium (Figure 8), or in the Acknowledgements thymidine-poor DME medium alone for the thymidine- The authors wish to warmfully thank Dr Jiri Bartek pulse experiment (Figure 2) to ensure higher speci®c (Danish Cancer Society, Copenhagen, Denmark) for the radioactivity. The reaction was stopped by rinsing the mouse monoclonal antibodies against cyclins D1, D2, D3 KIP1 cells four times with 5% TCA. Then, the cells were and against p27 . Drs Marcel DoreÂ e (CRBM, CNRS, harvestedin0.1M NaOH, and the radioactivity Montpellier, France) and Jean-Marie Darbon (HoÃ pital incorporated in the TCA-insoluble fraction was counted Purpan, Toulouse, France) are thanked for rabbit poly- by liquid scintillation. clonal anti-cdk2 antibodies, Dr Bernard Ducommun (IPBS, CNRS, Toulouse, France) for the anti-cdk4 antibodies, Dr Joan MassagueÂ (Memorial Sloan-Kettering Abbreviations Cancer Center, New York) for the pRb-GST construct and 8Br-cAMP: 8 bromo-cyclic AMP; a.a.: amino acid; cAMP: Dr Charles Sherr (Department of Tumor Cell biology, St cyclic adenosine mono-phosphate; CDK: cyclin-dependent Jude Children's Research Hospital, Memphis, Tennessee) kinase; CKI: cyclin-dependent kinase inhibitor; DME for various antisera against cdk4 and cyclins D. Reading medium: Dulbecco's modi®ed Eagle medium; DTT: dithio- the manuscript and fruitful discussions with Drs Fergus threitol; ERK: extracellularly regulated kinase; FCS: fetal McKenzie and Vjekoslav Dulic were greatly appreciated. calf serum; GST: glutathione S transferase; IPTG: This work was supported by grants from CNRS (Centre isopropyl-thio-galactoside; MAP kinase: Mitogen-Acti- national de la Recherche scienti®que), INSERM (Institut vated Protein kinase; MEK: MAP and ERK Kinase; National de la SanteÂ et de la Recherche MeÂ dicale) and PAGE: polyacrylamide gel electrophoresis; PBS: phos- from l'Association pour la Recherche contre le Cancer phate buer saline; PKA: protein kinase A; PMSF: phenyl (grants n: 1021). JoseÂ eNLavoiereceivedafellowshipfrom methyl sulfonyl ¯uoride; PNPP: para nitro phenyl the Medical Research Council of Canada.

References

AgrawalD,HauserP,McPhersonF,DongF,GarciaAand Lavoie JN, L'Allemain G, Brunet A, Muller R and Pledger WJ. (1996). Mol. Cell Biol., 16, 4327 ± 4336. Pouyssegur J. (1996a). J. Biol. Chem., 271, 20608 ± 20616. Ando K, Ajchenbaum CF and Grin JD. (1993). Proc. Natl. Lavoie JN, Rivard N, L'Allemain G and Pouyssegur J. Acad. Sci. USA, 90, 9571 ± 9575. (1996b). Prog. Cell Cycle Res., 2, 49 ± 58. Baldin V, Lukas J, Marcote MJ, Pagano M and Draetta G. Lees E. (1995). Curr. Opin. Cell Bio., 7, 773 ± 780. (1993). Genes Dev., 7, 812 ± 821. Lukas J, Bartkova J, Rohde M, Strauss M and Bartek J. Bates S, Parry D, Bonetta L, Vousden K, Dickson C and (1995a). Mol. Cell. Biol., 15, 2600 ± 2611. Peters G. (1994). Oncogene, 9, 1633 ± 1640. Lukas J, Bartkova J, Welcker M, Petersen OW, Peters G, Burgering BM and Bos JL. (1995). Trends Biochem. Sci., 20, Strauss M and Bartek J. (1995b). Oncogene, 10, 2125 ± 18 ± 22. 2134. Cocks BG, Vairo G, Bodrug SE and Hamilton JA. (1992). J. Lukas J, Muller H, Bartkova J, Spitkovsky D, Kjerul AA, Biol. Chem., 267, 12307 ± 12310. Jansen DP, Strauss M and Bartek J. (1994a). J. Cell. Biol., Counillon L, Scholz W, Lang HJ and Pouyssegur J. (1993). 125, 625 ± 638. Molec. Pharm., 44, 1041 ± 1045. Lukas J, Pagano M, Staskova Z, Draetta G and Bartek J. Draetta GF. (1994). Curr. Opin. Cell. Biol., 6, 842 ± 846. (1994b). Oncogene, 9, 707 ± 718. Dyson N. (1994). J. Cell. Sci. Suppl., 18, 81 ± 87. Magnaldo I, Pouyssegur and Paris S. (1989). FEBS Lett, 245, Ewen ME, Sluss HK, Whitehouse LL and Livingston DM. 65 ± 69. (1993). Cell, 74, 1009 ± 1020. Matsushime H, Quelle DE, Shurtle SA, Shibuya M, Sherr Fafournoux P, Noel J and Pouyssegur J. (1994). J. Biol. CJ and Kato JY. (1994). Mol. Cell. Biol., 14, 2066 ± 2076. Chem., 269, 2589 ± 2596. McKenzie FR and Pouyssegur J. (1996). J. Biol. Chem,, 271, Gagelin C, Pierre M and Toru DD. (1994). Biochem. 13476 ± 13483. Biophys. Res. Commun., 205, 923 ± 929. Muller H, Lukas J, Schneider A, Warthoe P, Bartek J, Eilers Grana X and Reddy EP. (1995). Oncogene, 11, 211 ± 219. M and Strauss M. (1994). Proc. Natl. Acad. Sci. USA, 91, Han EK, Sgambato A, Jiang W, Zhang YJ, Santella RM, 2945 ± 2949. Doki Y, Cacace AM, Schieren I and Weinstein IB. (1995). Nakayama K, Ishida N, Shirane M, Inomata A, Inoue T, Oncogene, 10, 953 ± 961. Shishido N, Horii I, Loh DY and Nakayama K. (1996). Hengst L and Reed SI. (1996). Science, 271, 1861 ± 1864. Cell, 85, 707 ± 720. Horton LE, Qian Y and Templeton DJ. (1995). Cell Growth Noel J, Roux D and PouysseÂ gur J. (1996). J. Cell Sci., 109, Dier., 6, 395 ± 407. 929 ± 939. Jiang W, Kahn SM, Zhou P, Zhang YJ, Cacace AM, Infante Ohtsubo M, Theodoras AM, Schumacher J, Roberts JM and AS, Doi S, Santella RM and Weinstein IB. (1993). Pagano M. (1995). Mol. Cell Biol., 15, 2612 ± 2624. Oncogene, 8, 3447 ± 3457. Orlowski J, Kandasamy RA and Shull GE. (1992). J. Biol. Kahan C, Seuwen K, Meloche S and PouysseÂ gur J. (1992). J. Chem., 267, 9331 ± 9339. Biol. Chem., 267. 13369 ± 13375. Pagano M, Tam SW, Theodoras AM, Beer RP, Del SG, Kato J, Matsushime H, Hiebert SW, Ewen ME and Sherr CJ. Chau V, Yew PR, Draetta GF and Rolfe M. (1995). (1993). Genes Dev., 7, 331 ± 342. Science, 269, 682 ± 685. Kato JY, Matsuoka M, Polyak K, Massague J and Sherr CJ. Pagano M, Theodoras AM, Tam SW and Draetta GF. (1994). Cell, 79, 487 ± 496. (1994). Gene Dev., 8, 1627 ± 1639. Kiess M, Gill RM and Hamel PA. (1995). Oncogene, 10, PageÁ s G, Brunet A, L'Allemain G and PouysseÂ gur J. (1994). 159 ± 166. EMBO J., 13, 3003 ± 3010. Kiyokawa H, Richon VM, Rifkind RA and Marks PA. (1994). Mol. Cell. Biol., 14, 7195 ± 7203. cAMP inhibits cell growth via cyclin D1 GL'Allemainet al 1990 PageÁ s G, Lenormand P, L'Allemain G, Chambard JC, Sherr CJ. (1993). Cell, 73, 1059 ± 1065. Meloche S and PouysseÂ gur J. (1993). Proc. Natl. Acad. Sherr CJ and Roberts JM. (1995). Genes Dev., 9275, 1149 ± Sci. USA, 90, 8319 ± 8323. 1163. Pastan I, Johnson GS and Anderson WB. (1975). Ann. Rev. Skapek SX, Rhee J, Spicer DB and Lassar AB. (1995). Biochem., 44, 491 ± 522. Science, 267, 1022 ± 1024. PouysseÂ gur J, Sardet C, Franchi A, L'Allemain G and Paris Suzuki-Takahashi I, Kitagawa M, Saijo M, Higashi H, S. (1984). Proc. Natl. Acad. Sci. USA, 81, 4833 ± 4837. Ogino H, Matsumoto H, Taya Y, Nishimura S and Quelle DE, Ashmun RA, Shurtle SA, Kato JY, Bar SD, Okuyama A. (1995). Oncogene, 10, 1691 ± 1698. Roussel MF and Sherr CJ. (1993). Genes Dev., 7, 1559 ± Tam SW, Theodoras AM, Shay JW, Draetta GF and Pagano 1571. M. (1994). Oncogene, 9, 2663 ± 2674. Reed SI, Bailly E, Dulic V, Hengst L, Resnitzky D and Vadiveloo PK, Vairo G, Novak U, Royston AK, Whitty G, Slingerland J. (1994). J. Cell Sci., 18, 69 ± 73. Filonzi EL, Cragoe Jr EJ and Hamilton JA. (1996). Resnitzky D, Gossen M, Bujard H and Reed SI. (1994). Mol. Oncogene, 13, 599 ± 608. Cell. Biol., 14, 1669 ± 1679. Winston J, Dong F and Pledger WJ. (1996). J. Biol. Chem., Resnitzky D, Hengst L and Reed SI. (1995). Mol. Cell. Biol., 271, 11253 ± 11260. 15, 4347 ± 4352. Zhou P, Jiang W, Zhang YJ, Kahn SM, Schieren I, Santella Rivard N, L'Allemain G, Bartek J and PouysseÂ gur J. (1996). RM and Weinstein IB. (1995). Oncogene, 11, 571 ± 580. J. Biol. Chem., 271, 18337 ± 18341. Zwijsen RML, Klompmaker R, Wientjens EBHGM, Kristel Roussel MF, Theodoras AM, Pagano M and Sherr CJ. PMP, van der Burg B and Michalides RJAM. (1996). Mol. (1995). Proc. Natl. Acad. Sci. USA, 92, 6837 ± 6841. Cell. Biol., 16, 2554 ± 2560. Sewing A, Burger C, Brusselbach S, Schalk C, Lucibello FC and Muller R. (1993). J. Cell. Sci., 104, 545 ± 555.