Loss of Rb overrides the requirement for ERK activity for cell proliferation

Ras proteins function as a molecular switch cycling between the inactive GDP and active GTP bound state. In the GTP bound state Ras binds to and leads to the activation of downstream effectors. A number of potential effectors of Ras have been described including the Raf family of protein kinases, the Ral-GDS family of guanine nucleotide exchange factors, PI 3-kinase(Marshall, 1996) and Phospholipase C-epsilon (Kelley et al.,2001; Lopez et al.,2001; Song et al.,2001). Since many of these studies are based on over-expression studies, which of these molecules are true effectors remains to be determined. However genetic evidence from C. elegans and D. melanogaster model systems strongly supports a role for Raf and thereby the ERK-MAP kinase pathway as a downstream effector of Ras (Dickson and Hafen, 1994; Marshall,1996).

In mammalian cells, entry into the cell cycle is controlled at the G1/S boundary (reviewed in Sherr,1996). The requirement for Ras in cell cycle regulation and proliferation is now well established (Cai et al., 1990; Feig and Cooper,1988; Mulcahy et al.,1985; Stacey et al.,1991; Dobrowolski et al.,1994; Aktas et al.,1997). These studies also show that Ras activity is required at multiple points in cell cycle entry. Following growth factor stimulation of quiescent non-proliferating cells, Ras activity is required for cells to exit G0 and pass through the G1/S transition of the cell cycle(Dobrowolski et al., 1994;Feig and Cooper, 1988;Mulcahy et al., 1985). In contrast to growth factor stimulation of quiescent cells, studies in exponentially growing cells indicate that Ras is required in the G2 phase of the preceding cycle and not during G1(Hitomi and Stacey, 2001).

The key event that regulates the G1/S transition is inactivation of the retinoblastoma (Rb) protein via phosphorylation mediated by cyclin dependent kinases. In its hypo-phosphorylated state Rb is bound to the E2F transcription factor (reviewed in Weinberg,1995). This represses transcription of target genes that are required for progression through S phase of the cell cycle. Phosphorylation of Rb leads to its inactivation and de-repression of E2F facilitating DNA replication. Inactivation of Rb by phosphorylation is mediated by the sequential activation of the cyclin dependent kinases (cdks)(Sherr, 1996;Weinberg, 1995). The D-type cyclins which bind cdk4 or cdk6 and cyclin E and cyclin A which bind cdk2 are implicated in this process (Sherr,1996; Weinberg,1995).

Many studies link cyclin D1 expression to Ras signalling. The use of inducible Ha-Ras expression constructs has demonstrated that overexpression of Ha-Ras in the rat epithelial cell line IEC(Filmus et al., 1994) and BalbC-3T3 cells (Winston et al.,1996) leads to the induction of cyclin D1 expression. Conversely expression of dominant negative Ras mutants has been demonstrated to inhibit expression of cyclin D1 following serum stimulation(Aktas et al., 1997). An acceleration of G1 progression and shorter cell doubling times in asynchronously growing Ha-Ras transformed NIH3T3 cells has been shown to correlate with overexpression of cyclin D1(Albanese et al., 1995;Liu et al., 1995). These reports make a clear link between Ras expression and the induction of cyclin D1 expression in cell cycle control. Activation of ERK-MAP kinase appears to be a key event in mediating Ras-dependent regulation of cyclin D1 expression as well as cdk complex formation (Lavoie et al., 1996; Weber et al.,1997; Cheng et al.,1999). PI 3-kinase and Ral-GEF have also been postulated to regulate cyclin D1 expression and stability(Diehl et al., 1997;Muise-Helmericks et al., 1998;Gille and Downward, 1999;Henry et al., 2000).

Levels of the cdk inhibitor p27^Kip1 are high in quiescent cells. Re-entry into the cell cycle is accompanied by a decline in the levels of p27^Kip1 facilitating activation of cyclin E/cdk2 complexes(Sherr and Roberts, 1999). Expression of oncogenic Ras had also been shown to regulate the levels of p27^Kip1 (Aktas et al.,1997; Takuwa and Takuwa,1997). Studies indicate that regulation of p27^Kip1expression by the Raf-ERK pathway and PI 3-kinase is mediated by regulating the rate of translation and protein stability(Aktas et al., 1997;Kerkhoff and Rapp, 1997;Takuwa and Takuwa, 1997). A further contribution to the relief of p27^Kip1 inhibition of cdk2 complexes is made by the Mek-ERK pathway facilitating cyclin D1/cdk4 complex formation which results in sequestration of p27^Kip1 onto this complex and allows activation of cdk2 complexes(Cheng et al., 1998).

Microinjection studies using the neutralising Y13-259 antibody(Furth et al., 1982) against Ras show that the requirement for Ras signalling to promote cell cycle entry is altered in Rb^-/- cells(Mittnacht et al., 1997;Peeper et al., 1997). Interestingly these experiments defined a difference between cells that are asynchronously growing and cells that have been rendered quiescent and then treated with growth factors to stimulate cell cycle entry. Regardless of whether they express Rb or are Rb null, cells require Ras activation in order to leave G0. In contrast microinjection of the Y13-259 antibody into asynchronously growing cells revealed that in comparison to wild-type MEFs,Rb-null MEFs have a reduced requirement for Ras activity(Mittnacht et al., 1997;Peeper et al., 1997). This reveals two interesting points. Ras activity is critical for both exit from G0 and for cells to make the G1/S transition. However with the loss of Rb, cells are released from the requirement for Ras for the G1/S transition. The observation that proliferation of asynchronously growing Rb^-/- MEFs is much less affected by blocking Ras function raises the question whether Rb^-/- cells have a reduced requirement for Ras dependent signalling pathways such as the ERK-MAP kinase signalling pathway. We find that Rb^-/- cells no longer depend on ERK signalling for re-entry into the cell cycle from G0, or for asynchronous growth. Consistent with this phenotype is the observation that the main role of ERK-MAP kinase pathway is to regulate cyclin D1 expression and associated cdk activity rather than the regulation of cyclin E and cyclin A expression.

Mouse embryo fibroblasts (MEFs) were prepared from embryos at 12.5 days post-coitus (dpc) as described by D'Abaco and Olson(D'Abaco and Olson, 2000). Cell lines were maintained in Dulbecco's modified Eagle's medium (DMEM) containing 10% fetal calf serum (FCS) and all experiments were conducted using early passage (<6) MEFs. For BrdU experiments cells were seeded at 2500 cells/well on 13 mm coverslips in DMEM supplemented with 10% FCS. The following day, cells were washed once in DMEM supplemented with 0.5% FCS and then maintained in this medium for 24-48 hours. Subsequently cells were treated with inhibitor(s) for 60 minutes then stimulated with 10% (v/v) serum. U0126 (Promega) was used at 20 μM, with DMSO as the vehicle control. LY294002 (Biomol Research Laboratories) was used at 20 μM with ethanol as the vehicle control For growth curve experiments cells were plated at a density of 10,000/plate in 5 cm² plates in DMEM containing 10% FCS. Following an overnight incubation cells were treated with inhibitor(s) as described above.

To monitor proliferation, following treatment with inhibitor(s) cells were pulsed with bromodeoxyuridine (BrdU) (Amersham Pharmacia Biotech) at the time of serum stimulation for 12-16 hours. Cells were fixed and stained for immunofluorescence as follows. Cells were rinsed once in phosphate-buffered saline (PBS) and fixed in 4% paraformaldehyde for 15 minutes at room temperature. After rinsing twice in PBS cells were permeabilized in 0.2%Triton X-100/0.1 mM glycine for 30 minutes at room temperature. Cells were rinsed several times in PBS and then blocked by incubation in 5% FCS in PBS(PBS/FCS) for 30 minutes at room temperature. The cells were then treated with RNase free DNase I (Roche) at 1-2 units/μl for 60 minutes at 37°C. After washing three times in PBS/FCS cells were incubated with anti-BrdU rat monoclonal antibody (diluted to 5 μg/mL in PBS/BSA) for 1 hour at room temperature. After washing cells three times in PBS/BSA cells were incubated with fluorescent secondary antibody (Jackson Immunoresearch Laboratories) to detect incorporated BrdU and DAPI (Sigma) to detect total nuclei in each field. In addition a Texas-Red-X-conjugated phalloidin stain (Molecular Probes) was used to detect cells. Secondary antibodies and phalloidin were diluted at 1:200, DAPI was diluted to a working concentration of 1 μg/ml in PBS/FCS. Cells were then incubated for 1 hour at room temperature in a light-proof container. After rinsing three times in PBS and once in water,coverslips were mounted onto glass slides by inverting onto 5-10 μl of moviol mountant containing 0.1% para-phenylenediamine as an antiquenching agent. Stained preparations were examined with a Bio-rad MRC confocal imaging system in conjuction with a Nikon Diaphot epifluorescence microscope.

An ERK2 rabbit polyclonal antibody generated against a C-terminal ERK2 peptide (no. 122) was used for immunoprecipitation experiments(Leevers and Marshall, 1992). A phospho-specific antibody to ERK1 and 2 (clone MAP-YT) was purchased from Sigma. The antibody that detects the hypo and hyper phosphorylated forms of the Retinoblastoma protein (14001A), was purchased from PharMingen. Anti-cyclin D1 (72-13G), cyclin E (M20) and cyclin A (H-432) antibodies were purchased from Santa Cruz Biotechnology. An anti-p27^Kip1 (clone 57)antibody was purchased from Transduction Laboratories.

Cells were washed twice in ice-cold PBS and lysed in ELB buffer (50 mM NaCl, 100 mM HEPES pH 7.0, 10 mM EDTA and 0.2% Triton-X-100) containing 20 mMβ-glycerol phosphate, 20 mM sodium fluoride, 10 mM sodium vanadate, 1 mM DTT, 1 mM phenylmethylsulfonyl fluoride, 10 μg/ml aprotinin and 10 μg/ml leupeptin. Cell debris was cleared by centrifugation at 18,000 g for 10 minutes. Protein concentrations were determined using the BCA protein assay reagent (Pierce).

Immunoblotting was performed using 25-30 μg of cleared lysate. For detection of the Rb protein 250 μg of cleared lysate was loaded. Prior to loading Laemmli buffer was added to each sample, followed by heating to 100°C for 3-5 minutes. Samples were resolved by electrophoresis through denaturing SDS-polyacrylamide gels. Proteins were transferred to polyvinylidene difluoride (PVDF) (Millipore) membrane for western blotting. Membranes were probed with an appropriate dilution of primary antibody for 1-2 hours at room temperature followed by incubation with a horseradish peroxidase(HRP) conjugated secondary antibody (Bio-Rad) for 45 minutes at room temperature. Proteins were visualised using the enhanced chemiluminescence(ECL) reagent (Amersham).

Cells were lysed in lysis buffer as described above. For each sample 50μg of cleared lysate was immuopprecipitated with the 122 antibody coupled to protein-A-Sepharose beads for 90 minutes at 4°C. Immuno-complexes were washed twice in lysis buffer and twice in kinase buffer (30 mM Tris pH 8.0, 20 mM MgCl₂ and 2 mM MnCl₂) containing 10 mMβ-glycerol phosphate, 1 mM DTT, 1 mM phenylmethylsulfonyl fluoride, 10μg/ml aprotinin and 10 μg/ml leupeptin. Kinase reactions were conducted in a final volume of 30 μl kinase buffer containing 10 μM ATP (Sigma), 5μCi [γ-³²P]ATP Amersham) and 0.25 mg of myelin basic protein (Sigma) as the substrate per reaction. After a 30 minute incubation at 30°C reactions were stopped by the addition Laemmli buffer followed by heating to 100°C for 3-5 minutes. Proteins were then transferred to PVDF membrane and ERK2 activity was detected by PhosphoImager (Molecular Dymamics).

Cells were lysed in lysis buffer as described above. For each sample 500μg of cleared lysate was immuoprecipitated with antibodies coupled to protein-A-Sepharose beads for 90 minutes at 4°C. Immuno-complexes were washed twice in lysis buffer and twice in kinase buffer (50 mM HEPES pH 7.4,10 mM MgCl₂, 10 mM MnCl₂) containing 10 mMβ-glycerol phosphate, 1 mM DTT, 1 mM phenylmethylsulfonyl fluoride, 10μg/ml aprotinin, 10 μg/ml leupeptin and 1 mM protein kinase inhibitor(Sigma). For cyclin D1/cdk4 associated kinase assays a GST-fusion construct encoding a C-terminal fragment (amino acids 763-928) of Rb(Zarkowska and Mittnacht,1997) was used at 0.5 μg per reaction. For cyclin E/cdk2 kinase assays histone H1 was used as the substrate (a gift from G. Goodwin, ICR) at 2μg per reaction. Kinase reactions were conducted in a final volume of 25μl kinase buffer containing 50 μM ATP and 5 μCi[γ-³²P] ATP. After a 15 minute incubation at 30°C reactions were stopped by the addition of SDS-PAGE sample buffer. Proteins were then transferred to PVDF membrane and kinase activity was detected by PhosphoImager.

The Ras-neutralising antibody Y13-259(Furth et al., 1982) has been shown to have little effect on proliferation when microinjected into asynchronously growing Rb-null cells suggesting that these cells have a reduced requirement for Ras signalling(Mittnacht et al., 1997;Peeper et al., 1997). It is well established that activation of ERK-MAPK pathway is Ras-dependent(Marshall, 1995). Therefore we compared the ability of serum growth factors to drive quiescent Wt and Rb^-/- MEFs into S-phase in the presence of U0126, an inhibitor of Mek activation (Favata et al.,1998). Following serum starvation for 20-24 hours cells were rendered quiescent as judged by their failure to incorporate BrdU(Fig. 1A,B). Serum stimulation for 16-20 hours resulted in efficient cell cycle entry and DNA replication(Fig. 1A,B) whereas treatment of Wt MEFs with U0126 blocked cell cycle entry. Treatment of Rb^-/-MEFs with U0126 did not result in a block of cell cycle entry(Fig. 1A,B). Similar results were obtained with PD098059 (Dudley et al., 1995), another inhibitor of Mek activation (data not shown). As Rb^-/- MEFs appear to have a reduced requirement for ERK signalling we went on to study the role of PI 3-kinase which has also been shown to be an effector of oncogenic Ras(Rodriguez-Viciana et al.,1994; Rodriguez-Viciana et al., 1997; Rodriguez-Viciana et al., 1996; Kodaki et al.,1994). Analysis of proliferation in the presence of LY294002, an inhibitor of PI 3-kinase activation (Davies et al., 2000) revealed that unlike the Wt cells the Rb^-/- MEFs are able to enter the cell cycle(Fig. 1B). However, treatment with the combination of inhibitors results in a block in cell cycle entry in both cell lines (Fig. 1B).

Since the requirement for ERK activation for cell proliferation was abrogated in Rb^-/- cells, we investigated whether cells lacking other negative regulators of cell cycle entry had altered signalling requirements. Fig. 2 shows that MEFs that do not express the cyclin dependent kinase inhibitor(s) (CDKi)p21^Waf1/Cip1, p27^Kip1 or p16^Ink4 all required ERK activation for cell cycle entry. Thus it is only the loss of Rb that alleviates the requirement for ERK signalling.

Lee et al. have shown that Rb^-/- cells have elevated levels of Ras-GTP during the G1 phase of the cell cycle(Lee et al., 1999). In addition, Alessi et al. have shown that PD098059 does not block ERK activation at high levels of growth factor stimulation(Alessi et al., 1995). Higher levels of active Ras in Rb^-/- cells might result in higher levels of ERK activation that cannot be blocked by U0126. Hence the inability of U0126 to block cell cycle entry in Rb^-/- cells could be a consequence of higher levels of ERK activation following growth factor stimulation. We therefore investigated whether ERK activity was different in Rb^-/- compared with Wt MEFs. Analysis of ERK activation(Fig. 3) by in vitro kinase assays (Fig. 3a,c) and western blotting for (active) phospho-ERK (Fig. 3b,d) revealed that the basal level of ERK activation was the same in quiescent Rb^-/- and Wt MEFs and that it was stimulated to the same degree with similar duration in both Rb^-/- and Wt MEFs. Furthermore, Fig. 3 shows that U0126 inhibited ERK activation to a similar degree in both cell lines. Thus the failure of U0126 to block cell proliferation of Rb^-/- cells is a consequence of a reduced requirement for ERK signalling rather than the inability of U0126 to block ERK activation in these cells.

Since we observed a reduced requirement for ERK signalling in Rb-null MEFs,we were interested to investigate whether human tumour cell lines lacking Rb had a reduced requirement for ERK signalling. We compared tumour cell lines,which are deficient for Rb expression to cells lines of similar origin, which are wild-type for Rb. To facilitate these experiments we used two cell lines derived from glioma (SF295 Rb wild-type and SF539 Rb deficient) and two cell lines derived from non-small cell lung carcinoma (H1299 Rb wild-type and H2009 Rb deficient). Analysis of SF295 and H1299 (Rb wild-type) revealed that treatment with U0126 blocked proliferation of asynchronously growing cells(Fig. 4A and B, respectively). In contrast the Rb-deficient tumour cell lines, SF539 and H2009, continued to proliferate in the presence of U0126. Western blotting for phospho-ERK(Fig. 4C) showed that U0126 inhibited ERK activation in SF539 and H2009(Fig. 4C). These data show that human tumour cell lines that lack Rb, like MEFs that are deficient for Rb expression, have a reduced requirement for ERK signalling.

The G1/S transition is facilitated by the inactivation of the retinoblastoma protein. At the molecular level, inactivation of pRb¹⁰⁵ correlates with the hyper-phosphorylation of a number of Ser/Thr sites (Mittnacht,1998; Weinberg,1995) mediated by the G1 cyclin dependent kinases (cdks)(Sherr, 1996;Weinberg, 1995). Using a pan Rb antibody in a western blot we were able to probe for endogenous hypo(active) and hyper (inactive) forms of pRb as judged by altered electrophoretic mobility (Fig. 5). Upon serum deprivation only the active faster migrating(hypo-phosphorylated) form of pRb¹⁰⁵ was detected(Fig. 5, lane 1) correlating with the quiescent state of the cells. On serum stimulation the inactive slower migrating (hyper-phosphorylated) form of retinoblastoma was also detected (Fig. 5, lane 2). In contrast, when cells were treated with U0126 prior to serum stimulation we did not detect the hyper-phosphorylated of Rb(Fig. 5, lanes 3). When the cells were treated with either LY294002 or the combination of both inhibitors we were also unable to detect the hyper-phosphorylated form of Rb(Fig. 5, lanes 4,5).

The G1/S boundary is marked by expression of cyclin D1 and cdk4 activity(Sherr, 1996). Therefore we examined expression of cyclin D1 and cdk4 activity in MEFs treated with U0126(Fig. 6a,Fig. 7A). Serum starved cells had low levels of cyclin D1 (Fig. 6a) that was induced upon serum stimulation(Fig. 6a). Treatment of cells with U0126 partially blocked induction of cyclin D1 expression in both the Wt and Rb-deficient MEFs. To determine whether this decrease in cyclin D1 expression affected cdk4 activity, cdk4 was immunoprecipitated from MEF cell lysates and assayed for in vitro kinase activity using a C-terminal fragment of Rb (amino acids 763-928) fused to GST as substrate(Zarkowska and Mittnacht,1997). Serum starved MEFs had low cdk4 kinase activity(Fig. 7A), while serum stimulation lead to an increase in cdk4 activity(Fig. 7A). Treatment with U0126 resulted in a marked decrease in cdk4 kinase activity in both the Wt and Rb^-/- MEFs (Fig. 7A). In agreement with previous studies,(Lavoie et al., 1996;Weber et al., 1997;Cheng et al., 1999), these results support the model that activation of ERK is required for cyclin D1 expression, cdk4 complex formation and kinase activity. We consistently observed that treatment with U0126 did not completely abrogate serum stimulated cyclin D1 expression or cdk4 activity although it did block entry into S-phase of WT cells (Fig. 1). Since the PI 3-kinase pathway has also been shown to regulate cyclin D1 expression at both the transcriptional(Gille and Downward, 1999),and postranslational level (Diehl et al.,1997; Diehl et al.,1998) we investigated how inhibition of PI 3-kinase affected cyclin D1 expression and cdk4 kinase activity. These experiments revealed that inhibition of PI 3-kinase with LY294002 lead to a partial block of cyclin D1 expression (Fig. 6a) and cdk4 activity (Fig. 7A). Interestingly inhibition of both the ERK and PI 3-kinase pathways by treatment with U0126 and LY294002 completely ablated serum stimulated cyclin D1 expression (Fig. 6a) and cdk4 activity (Fig. 7A). These results show that both the ERK and PI 3-kinase pathways contribute to cyclin D1 expression and that complete inhibition of cyclin D1 expression and associated cdk4 activity requires inhibition of ERK and PI 3-kinase signalling.

The G1-S phase of the cell cycle is regulated by cyclin E and cyclin A associated cdk2 activity (Sherr,1996). We therefore investigated whether inhibition of ERK activation using U0126 would inhibit the expression of cyclin E and cyclin A(Fig. 6b,c). In Wt MEFs expression of cyclin E and A was unaffected by inhibition of the ERK pathway,whilst interestingly the level of cyclin E/cdk2 activity was low. In contrast inhibition of PI 3-kinase blocked expression of cyclin E and A(Fig. 6b,c) and activation of cyclin E/cdk2 activity in Wt MEFs. The lack of cdk4 and cdk2 activity that we observed in Wt MEFs upon treatment with either U0126 or LY294002 correlates with the absence of the inactive hyperphosphorylated form of Rb and the growth arrest that we observe. These data show that ERK signalling is required to regulate expression of cyclin D1 but does not impinge on the expression of cyclin E or cyclin A. This is in contrast to PI 3-kinase signalling which our studies show impinges on all 3 G1/S regulating cyclins and associated cyclin D1/cdk4 and cyclin E/cdk2 activity.

As previously described (Herrera et al., 1996), Rb^-/- cells have elevated levels of cyclin E and cyclin A both in serum starved and serum stimulated conditions(Fig. 6b,c). As was observed in Wt MEFs, treatment with U0126 had no effect on the levels of cyclin E or cyclin A in Rb^-/- cells (Fig. 7B,C). Stimulation of Rb^-/- cells with serum induces a higher level of cyclin E/cdk2 activity in comparison to the Wt MEFs. Whilst treatment with U0126 had no effect on cyclin E expression in Rb^-/-MEFs, analysis of cyclin E/cdk2 activity by immunoprecipitation of cyclin E from cell lysates indicated that treatment with U0126 had a small but consistent decrease in cyclin E associated kinase activity(Fig. 7B). However, the remaining kinase activity was greater than or equal to that observed in serum stimulated Wt MEFs in the absence of inhibitors. This high residual cyclin E/cdk2 activity in U0126-treated Rb^-/- MEFs, coupled with the knowledge that Rb^-/- cells do not require cyclin D1 associated kinase activity for cell cycle progression(Lukas et al., 1995) may explain why Rb^-/- cells can proliferate when the ERK pathway is inhibited.

Analysis of PI 3-kinase signalling in Rb^-/- MEFs revealed that unlike Wt cells inhibition of PI 3-kinase had no effect on cyclin E or cyclin A expression levels (Fig. 6b,c). The combination of U0126 and LY294002 lead to a small ihhibition in the level of both cyclin E and cyclin A. Interestingly, we did not observe complete inhibition in expression levels(Fig. 6b,c). Analysis of cyclin E/cdk2 activity revealed that despite a negligible effect on the levels of cyclin E, treatment with LY294002 and the combination of inhibitors lead to a marked inhibition of kinase activityin the Rb^-/- MEFs(Fig. 7B).

The level of the expression of the cdk inhibitor p27^Kip1 is known to be important determinant for cell cycle entry at least in part as a regulator of cdk2 activity (Sherr and Roberts, 1999). In an effort to study the mechanism by which cdk2 activity is regulated as a consequence of ERK and PI 3-kinase signalling, we examined how inhibition of these pathways would affect p27^Kip1expression levels. Consistent with previous studies(Slingerland and Pagano, 2000)removal of growth factors led to an accumulation of p27^Kip1(Fig. 7C) that was downregulated upon serum restimulation in both Wt and Rb-deficient cells(Fig. 7C). Analysis of cells treated with either U0126 or LY294002 revealed that following serum stimulation, the levels of p27^Kip1 were still downregulated in both Wt and Rb^-/- MEFs (Fig. 7C). However, in the presence of both U0126 and LY294002 serum stimulation was unable lead to the downregulation of p27^Kip1 in Wt and Rb^-/- MEFs (Fig. 7C). These results show that either ERK or PI 3-kinase signalling can regulate the levels of p27^Kip1 suggesting that redundancy exists between these two signalling pathways for the regulation of p27^Kip1 levels.

We have shown that Rb^-/- MEFs are capable of cell cycle entry in the presence of U0126 and LY294002 in a synchronised cell growth assay(Fig. 1). While we observe residual cyclin E/cdk2 activity in Rb^-/- MEFs treated with U0126,treatment with LY294002 results in complete loss of kinase activity. Due to this surprising result we were interested to assess the capacity of Rb^-/- cells to proliferate in a long term growth assay in the presence of these inhibitors (Fig. 8). As expected treatment of asynchronously growing Wt MEFs with U0126 retarded proliferation over a 4 day period(Fig. 8a). In contrast,Rb^-/- MEFs were capable of sustained growth over a 4 day growth period in the presence of U0126 (Fig. 8a), which was comparable to the growth kinetics of untreated MEFs(Fig. 8a,b). Quiescent Rb^-/- MEFs stimulated with serum in the presence of LY294002 entered DNA synthesis but after 24 hours entered apoptosis. Furthermore treatment with LY294002 led to inhibition of proliferation through apoptosis of asynchronous Rb^-/- MEFs over a 4 day period(Fig. 8b)

It is now well established that active Ras is required for mitogenic signalling in a variety of systems (Feig and Cooper, 1988; Mulcahy et al., 1985; Dobrowolski et al.,1994; Stacey et al.,1991). Recently the role of Ras in cell cycle progression has been more clearly defined (Mittnacht et al.,1997; Peeper et al.,1997). These studies showed that the requirement for Ras signalling is reduced in asynchrononously growing Rb-null MEFS(Mittnacht et al., 1997;Peeper et al., 1997). However Peeper et al., showed that Ras signalling is still required for quiescent Rb-null MEFs to exit G0 and enter G1(Peeper et al., 1997). The observation that Ras signalling is still required for quiescent Rb-null MEFs to enter the cell cycle is consistent with the observation that these cells do not have a reduced serum requirement for proliferation(Herrera et al., 1996). These results imply that Ras signalling is required at two points: first, to leave G0 and second for the Rb-dependent G1-S transition. The observation that the requirement for Ras signalling is still required for Rb-null cells to leave G0 led us to investigate whether these cells require activation of the ERK-MAPK pathway for cell cycle entry. This pathway is a major downstream effector of Ras (Marshall, 1995) and has been strongly implicated in cell cycle entry particularly for the induction of cyclin D1 expression (Cheng et al.,1998; Lavoie et al.,1996; Weber et al.,1997).

When quiescent serum starved Wt MEFs were treated with the Mek inhibitor U0126, serum restimulation did not lead to S-phase entry. In contrast Rb-null MEFS were able to exit the G0 state, enter the cell cycle and proliferate when Mek activity is inhibited. This inability of U0126 to block cell cycle entry in Rb-null cells is not a consequence of the inhibitor failing to block Mek1/2 and thereby ERK1/2 activation in these cells because ERK1/2 activation was similarly inhibited in both Wt and Rb-null cells. U0126 and PD098059, another Mek inhibitor, have been shown to also block Mek5 in the ERK5/BMK1 pathway(Kamakura et al., 1999). Since the ERK5/BMK1 pathway has been implicated in cell cycle control(Kato et al., 1998) these results could imply that loss of Rb may also abrogate the requirement for ERK5 activation. However we have been unable to reproducibly measure ERK5 kinase activity and therefore do not know whether it is activated in the cell system we have used. These results show that while Rb-null cells still require Ras signalling to exit G0 they do not require ERK1/2 activation. The role of the ERK1/2 pathway is to bring about the inactivation of Rb and the G1-S transition rather than the Ras dependent step in G0 exit. While oncogenes have long been known to affect cell signalling pathways, these results show for the first time that loss of a tumour suppressor gene can alter the requirement for the MAP kinase signalling pathway. However the requirement for ERK signalling was unaffected in MEFs null for p21^Waf1/Cip1, p27^Kip1 or p16^Ink4 which are also tumour suppressor genes linked to cell cycle control. Thus it is only the loss of Rb that alleviates the requirement for ERK signalling.

We were interested to try and understand the mechanism that abrogates the requirement for ERK signalling in Rb-null cells. These cells do not depend on cyclin D1/cdk activity cdk (Lukas et al.,1995) arguing that they no longer depend on ERK signalling for cyclin D1/cdk activity. However Rb-null cells are known to require growth factor stimulation so they might require ERK signalling for other cell cycle events (Herrera et al., 1996). Although U0126 treatment reduced cyclin E/CDK2 activity by approximately 50%in Rb^-/- MEFS (without affecting cyclin E expression), the serum stimulated level of E/CDK2 activity after U0126 treatment in Rb^-/-MEFs was comparable to that in serum-stimulated untreated Wt MEFs. This suggests that this level of E/CDK2 activity is sufficient for proliferation. Rb-null human tumour cell lines — unlike their Rb Wt counterparts— also proliferated in the presence of U0126 despite complete inhibition of ERK activation. We conclude that Rb-null cells do not depend on ERK signalling for proliferation because the requirement for cyclin D dependent kinase activity is abrogated by loss of Rb and cells have elevated cyclin E/Cdk2 activity. This increased activity probably results from deregulated cyclin E transcription caused by the de-repression of E2F that is associated with loss of Rb (Herrera et al.,1996).

Like the ERK pathway PI 3-kinase signalling has been shown to be required for quiescent cells to enter the cell cycle(Gille and Downward, 1999;Jones et al., 1999;Jones and Kazlauskas, 2001). Furthermore PDGF dependent PI 3-kinase signalling is known to occur at G0/G1 and mid to late G1, interestingly however entry into S-phase depends on PI 3-kinase activity during the mid to late phase of G1(Jones et al., 1999;Jones and Kazlauskas, 2001). Therefore we were interested to investigate the effect of inhibition of PI 3-kinase signalling on cell cycle progression in Rb-null cells. In contrast to Wt MEFS, treatment of quiescent Rb-null cells with serum in the presence of the PI 3-kinase inhibitor LY294002 did not inhibit S-phase entry. However unlike inhibition of the ERK pathway, cells could not proliferate long term when PI 3-kinase was inhibited and died through apoptosis. At the molecular level treatment with LY294002 partially blocked induction of cyclin D1 expression, cdk4 activity but in contrast to U0126 also blocked expression of cyclin E, A and cyclin E/cdk2 activity. Thus, cyclin D1 expression is responsive to both ERK and PI 3-kinase dependent signalling, but cyclin E and cyclin A only requires PI 3-kinase signalling. Together these data are consistent with a role for PI 3-kinase in regulating the G1/S transtion that is overcome in the absence of the Rb check-point. However Rb-null cells are still dependent on PI 3-kinase signalling for survival.

Downregulation of the CDKI p27^Kip1 is a key element for the movement of cells out of G0 and into S-phase(Malek et al., 2001;Slingerland and Pagano, 2000). Our data show that in both Wt and Rb-null cells p27^Kip1 degradation occurs after serum stimulation in the presence of either U0126 to block ERK signalling or LY294002 to block PI 3-kinase signalling but not in the presence of both inhibitors. Similar results were found with human tumour cell lines(data not shown). These results demonstrate that there is redundancy between these signalling pathways for regulation of p27^Kip1. Cell cycle analysis of MEFS revealed that in the presence of both inhibitors, S-phase entry is inhibited regardless of Rb status. Our data predicts that it is the block of p27^Kip1 degradation that occurs when both pathways are inhibited that underlies the block in cell cycle entry that we observe. The residual levels of cyclin D1 and CDK4 activity that persist after treatment with either U0126 or LY294002 suggest that these low levels of the cyclin D1/CDK4 complex may be important to sequester p27^Kip1 and permit cell cycle entry in Rb^-/- cells(Cheng et al., 1998). The mechanism which underlies the accumulation of p27^Kip1 is still to be elucidated. However is has been shown that Ras signalling pathways can control expression of p27^Kip1 by regulating the rate of translation and protein stability (Aktas et al.,1997; Kerkhoff and Rapp,1997; Takuwa and Takuwa,1997).

Since the Ras protein is a critical transducer of proliferative signals, it is not surprising that a high percentage of some human cancers contain activating Ras gene mutations (Bos,1989). Most oncogenic Ras containing tumours that have been examined contain Wt Rb (Horowitz et al.,1990; Kashii et al.,1994) and would be predicted to be sensitive to Mek inhibitors. However where Rb is lost for example in non-small-cell lung cancers our data predict that inhibition of the Mek pathway will not have an anti-proliferative effect. Thus effective therapeutic use of signal transduction inhibitors may require knowledge of the genetic alterations to the cell cycle machinery for different tumours.

C.J.M. is a Gibb Life Fellow of the Cancer Research UK. This work was supported by Glaxo Welcome and Cancer Research UK. We thank M. Coleman and E. Sahai for invaluable discussion and advice on the manuscript. We are grateful to S. Mittnacht, M. Olson and R. Wilson for advice and support. We thank T. Jacks and K. F. Macleod for the gift of the Rb-knockout mouse.