Multiple Ras Effector Pathways Contribute to G1Cell Cycle Progression*

The involvement of Ras in the activation of multiple early signaling pathways is well understood, but it is less clear how the various Ras effectors interact with the cell cycle machinery to cause G1 progression. Ras-mediated activation of extracellular-regulated kinase/mitogen-activated protein kinase has been implicated in cyclin D1 up-regulation, but there is little extracellular-regulated kinase activity during the later stages of G1, when cyclin D1 expression becomes maximal, implying that other effector pathways may also be important in cyclin D1 induction. We have addressed the involvement of Ras effectors from the phosphatidylinositol (PI) 3-kinase and Ral-GDS families in G1 progression and compared it to that of the Raf/mitogen-activated protein kinase pathway. PI 3-kinase activity is required for the expression of endogenous cyclin D1 and for S phase entry following serum stimulation of quiescent NIH 3T3 fibroblasts. Activated PI 3-kinase induces cyclin D1 transcription and E2F activity, at least in part mediated by the serine/threonine kinase Akt/PKB, and to a lesser extent the Rho family GTPase Rac. In addition, both activated Ral-GDS-like factor and Raf stimulate cyclin D1transcription and E2F activity and act in synergy with PI 3-kinase. Therefore, multiple cooperating pathways mediate the effects of Ras on progression through the cell cycle.

Upon reentry of cells into the cell cycle and throughout the G 1 phase, mitogenic signals are integrated through the GTPase Ras. Inhibition of Ras function by microinjection of Ras-neutralizing antibodies or by inducible expression of dominant negative Ras arrests cycling cells in G 1 and prevents growth factor stimulated cells from leaving G 0 to reenter the cell cycle (1)(2)(3)(4). The requirement of Ras function during the G 0 /G 1 transition seems to be conserved from yeast, in which it is necessary for spore germination (5). Microinjection of anti-Ras antibodies has demonstrated a requirement for Ras up to 2 h prior to S phase entry (6,7).
Overexpression of mutationally activated Ras leads to cellular transformation and a shortening of the G 1 phase of the cell cycle (8). However, transformation of cells by activated Ras requires other genetic changes, as only immortal cells that have lost cell cycle checkpoints, such as those imposed by cyclin-dependent kinase (cdk) 1 inhibitors, retinoblastoma protein, and p53, can be transformed by Ras alone. The expression of activated Ras in primary cells leads to cell cycle arrest via up-regulation of cdk inhibitors and p53. The resulting phenotype resembles that of cellular senescence (9). In mouse embryo fibroblasts (10,11) and rat Schwann cells (12), the Ras effector Raf appears to be sufficient to mediate this effect via induction of p21 Waf1/Cip1 . Thus, Ras exerts both positive and negative effects on cell growth, depending on cellular context.
Recent work from this and other laboratories has shown that Ras is able to interact with multiple effector enzymes, including the Raf protein kinases, the Ral-GDS family of guanine nucleotide exchange factors for Ral, and type I phosphoinositide 3-kinases (13). The functions of multiple Ras effectors are required for cellular transformation; in addition to Raf, both the PI 3-kinase and the Ral-GDS pathways cooperate to achieve efficient transformation of immortalized cells (14,15). Therefore, several Ras effector pathways may interface with the cell cycle machinery.
Ras has been implicated in the positive regulation of the cyclin D 1 promoter. The conditional expression of oncogenic Ras induces cyclin D 1 protein production in growth factordeprived cells (16). However, the resulting cyclin D 1 /cdk4 complexes may remain inactive in the absence of growth factors (17). Activation of Raf-ER, a conditionally active fusion of Raf to the hormone binding domain of the estrogen receptor, and conditional expression of an activated form of the MAP/ERK kinase, MEK, reproduce the effect of oncogenic Ras on cyclin D 1 , and under certain conditions this signal can be blocked by co-expression of a MAP kinase phosphatase or dominant-interfering mutants of ERK. Therefore, these Ras effects have been attributed to the sequential activation of Raf kinase, MEK, and ERK (18 -20). However, there are circumstances under which ERK activity may be dispensable for cyclin D 1 production as the MEK inhibitor PD 98059 does not inhibit serum-induced cyclin D 1 expression in IIC9 fibroblasts (21).
It has been suggested that low levels of Raf kinase activity promote proliferation by inducing cyclin D 1 and strong Raf activity inhibits cell cycle progression through production of cdk inhibitors (9 -12). However, it is also possible that the divergent effects of Ras on proliferation are mediated by distinct effector proteins, which could be subject to independent regulation by cross-talk with other signaling pathways. Therefore, we investigated which of the known Ras effectors would interface with the G 1 cell cycle machinery. We find that activation of PI 3-kinase is required for cyclin D 1 protein expression and S phase entry in fibroblasts. The PI 3-kinase pathway also contributes to the activation of E2F-dependent transcription underlining its importance for progression through G 1 . An activated form of Ral-GDS-like factor (Rlf), which has recently been implicated in the transcriptional induction of the c-fos proto-oncogene, also strongly activated cyclin D 1 transcription. Thus, multiple pathways in addition to Raf/MAP kinase mediate the effects of Ras on cyclin D 1 expression and E2F transcriptional activity in order to drive G 1 progression.

MATERIALS AND METHODS
Plasmids-The reporter plasmid E2ACAT and has been described previously (31). The luciferase reporter pGL2-lucD 1 contains 1.8 kilobase pairs of the human cyclin D 1 promoter, including the TATA box inserted into the SmaI site of pGL2 (Promega) and was provided by R. Assoian. Expression vectors for Ral-GDS-like factor (52) and RalA (53) have been described. RalA F39L was generated using the mutagenic oligonucleotide primer CATGTACGATGAGCTCGTGGAGGACTATG on pMT3-RalA using standard procedures.
The activated pSG5 gag-Akt/PKB (54) and dominant-negative Akt/ PKB (55) constructs were kindly provided by B. Burgering. Expression plasmids for activated PI 3-kinase (25), V12 Ras, V12 Rac, and V12 Cdc42 have been described (15). Raf CAAX DD is an activated form of c-Raf-1 in which Raf is localized to the membrane by a farnesylation sequence from H-Ras and also by mutation of two tyrosine residues, 340 and 341, to aspartic acids; it was kindly provided by Raichard Marais and Chris Marshall.
Cell Culture and Transfections-NIH 3T3 cells were maintained in 10% calf serum and starved for 48 h in 0.15% calf serum where indicated. LY294002 was used at 20 M unless indicated otherwise. The final concentration of PD 98059 (Biomol) was 30 M, and rapamycin was used at 50 nM. Cells were preincubated with inhibitors for 10 min prior to addition of calf serum to 10%. LipofectAMINE was used to transfect 1.5 10 5 cells with 0.5 g of each plasmid. Cells were lysed 36 h after transfection. Chloramphenicol acetyltransferase activity was determined by enzyme-linked immunosorbent assay (Roche Molecular Biochemicals), and luciferase activity was measured using Promega luciferin as substrate. The results were corrected using CH110-expressed (Amersham Pharmacia Biotech) ␤-galactosidase activity. Figs. 5 and 6 show the means of four to seven experiments, and error bars represent S.E. The expression of all constructs were confirmed by immunoblotting: in the data presented in Figs. 5 and 6, any given construct was expressed at similar levels in different assay points in the same experiment.
Cell Cycle Analysis-In order to determine cell cycle distribution, NIH 3T3 cells were trypsinized washed with phosphate-buffered saline and fixed in 70% ethanol at 4°C while vortexing and stored at Ϫ20°C. Cells were washed twice in phosphate-buffered saline and treated with 100 g/ml ribonuclease for 5 min at room temperature. Cells were stained with propidium iodide (50 g/ml) and analyzed by flow cytometry using 488 nm excitation.
Antisera-The cyclin D 1 polyclonal antiserum was a gift from G. Peters. An anti-RB monoclonal antibody was used for immunoprecipitation (Pharmingen), and Rb was detected using a polyclonal serum (Santa Cruz). Anti-phospho-ERK was purchased from Promega. Ras was detected using anti-pan Ras antibody-4 (Calbiochem).
Raf-Ras Binding Domain Pull-out Assays-GST-Raf-Ras binding domain (56) was expressed in Escherichia coli BL21, and bacteria were lysed in 1% Triton X-100, 50 mM HEPES-NaOH, pH 7.5, 100 mM NaCl, 5% glycerol, 0.2 M phenylmethylsulfonyl fluoride, 10 g/ml aprotinin, and 1 mM benzamidine. The suspension was aliquoted, snap-frozen, and stored at Ϫ70°C. Prior to use, aliquots were thawed and sonicated. The fusion protein was then purified on glutathione-Sepharose beads. NIH 3T3 cells were lysed in the same lysis buffer containing 10 mM MgCl 2 but lacking glycerol. The cleared lysates were incubated with beads in lysis buffer containing 10 mM MgCl 2 for 20 min at 4°C. Beads were washed once in cold phosphate-buffered saline containing 5 mM MgCl 2 and 0.1% Triton X-100. Bound Ras was quantified by Western blotting.
Akt/PKB Kinase Assays-Akt/PKB kinase was immunoprecipitated from NIH 3T3 cell lysates, and immune complex kinase assays were performed using histone H2B as a substrate as described previously (25).

RESULTS
Sustained Ras and Akt/PKB Activation during G 1 -Ras function has been shown to be required throughout G 1 up to the restriction point, after which cells become committed to enter S phase. In order to determine Ras activity at later stages in G 1 , we made use of the high affinity interaction of the active GTPase with the Ras binding domain of Raf kinase (22). Quiescent NIH 3T3 cells were stimulated with serum, and GTPbound Ras was purified based on its ability to bind to recombinant fusion protein containing the Ras binding domain of Raf. Ras activation was found to be biphasic with a strong second mid-G 1 peak of activity that was maximal at about 2-4 h after exit from quiescence (Fig. 1A). Whereas the first peak of Ras activity up to about 30 min after stimulation correlated with ERK phosphorylation, little ERK activity could be detected at 4 h ( Fig. 1B) (see also Ref. 22). This is likely to be due to the induction of MAP kinase phosphatases as a result of the early phase of growth factor receptor and Ras activation (23).
One of the functions in cell cycle progression attributed to Ras in cells released from quiescence is the induction of cyclin D 1 protein as a consequence of ERK activation. However, cyclin D 1 expression became detectable only at 4 h after stimulation and became maximal at 6 -8 h (see Fig. 3, A and B), consistent with previous reports (24). Thus, cyclin D 1 induction correlates with the second phase of Ras activity, which is not associated with ERK activation.
The serine/threonine kinase Akt/PKB is activated by phosphatidylinositol 3,4,5-trisphosphate (PIP 3 ), which is generated upon activation of the Ras effector PI 3-kinase. In order to determine whether the time course of Akt/PKB activation is consistent with a role in G 1 progression, its kinase activity was determined at various times during the G 1 phase. As shown in Fig. 1C, Akt/PKB activity was sustained during G 1 , and its period of activity encompassed the second peak of Ras activation.
PI 3-kinase Is Required for G 1 Progression-In order to determine which Ras effector pathways are required for G 1 progression, we made use of chemical inhibitors for MEK (PD 98059) and PI 3-kinase (LY 294002). Because it is known that LY 294002 also inhibits some relatives of PI 3-kinase, such as TOR, the specific TOR inhibitor rapamycin was also used.
FIG. 1. Activation of Ras and Akt/PKB during G 1 . NIH 3T3 cells were serum-starved for 36 h and released from quiescence by addition of calf serum to 10%. A, activation state of endogenous Ras. The level of GTP-bound Ras was measured by extracting lysates from various stages during G 1 on a GST-Raf Ras binding domain (RBD) affinity matrix. B, the activation state of ERK/MAP kinase from cell lysates from various stages during G 1 was assessed by analyzing its phosphorylation state using a phosphospecific antiserum. C, the activity of Akt/PKB was analyzed by immunoprecipitation from NIH 3T3 cell lysates and measurement of kinase activity using histone H2B as a substrate.
Quiescent NIH 3T3 cells were preincubated with inhibitor and then stimulated with serum in the presence of inhibitor. Fig. 2A shows the cell cycle distribution at 14 and 20 h after serum treatment. Whereas both PD 98059 and rapamycin caused delays in S phase entry, many cells treated with either inhibitor were able to initiate DNA synthesis by 14 h. However, in the presence of LY 294002, cells had not progressed into S phase at all after 14 h. At 20 h, more than 75% of the LY 294002-treated cells were still in G 0/1 , compared with well under for 50% for control cells and those treated with PD 98059 or rapamycin. In order to determine the effectiveness of the drug treatments, cell lysates were immunoblotted with antiphospho-Akt or anti-phospho-ERK antibodies. Levels of the activity of these downstream effectors of PI 3-kinase and Raf were less than 5% of untreated serum controls for Akt in the presence of 20 M LY294002 or 15% for ERK in the presence of 30 M PD98059 at all time points (data not shown). Fig. 2B shows the effect of the PI 3-kinase inhibitor on unsynchronized cells. The presence of LY 294002 for 48 h induced a level of arrest similar to that observed in growth factor deprived cells. The half-maximal inhibition of G 1 progression was achieved at 4 M LY 294002, which is similar to the IC 50 for inhibition of insulin stimulated Akt/PKB activity (Fig. 2C) (25). These results suggest that PI 3-kinase activity is essential for the progression of cells through the G 1 phase of the cell cycle.
Effect of PI 3-kinase and MEK Inhibition on Endogenous Cyclin D 1 Expression-One of the earliest effects of serum stimulation leading to cell cycle reentry in fibroblasts is the expression of cyclin D 1 protein. In order to determine the effects of the various inhibitors on cyclin D 1 expression, NIH 3T3 cells were treated with inhibitors as above, and the expression of cyclin D 1 was analyzed. Fig. 3A shows that the presence of the PI 3-kinase inhibitor LY 294002 severely inhibited cyclin D 1 expression at all time points studied. The presence of the MEK inhibitor PD 98059 led to a slightly reduced expression of cyclin D 1 relative to uninhibited cells at 7 h, but there was no effect at later time points. The TOR inhibitor rapamycin caused some decrease in cyclin D 1 expression at 7 and 14 h, but by 20 h, this effect was negligible. At all times, rapamycin was less inhibitory for cyclin D 1 expression than was LY 294002, suggesting that the effects of the PI 3-kinase inhibitor were not connected with inhibition of the target of rapamycin. The PI 3-kinase inhibitor LY 294002 thus inhibits one of the earliest events in G 1 .
In order to determine the length of time in G 1 during which PI 3-kinase activity is required for cyclin D 1 expression, the inhibitor was added at different times after release of cells from starvation by serum treatment. Fig. 3B shows that about 8 h after serum treatment, cyclin D 1 expression becomes independent of PI 3-kinase activity but that inhibition of PI 3-kinase at any time during the first 6 h of serum treatment results in inhibition of cyclin D 1 expression. Inhibition of MEK through the use of PD98059 has only a small effect on cyclin D 1 expression, which is lost if inhibitor treatment is delayed until 4 h after serum addition, suggesting that the low levels of ERK activity in late G 1 are not important for cyclin D 1 expression. Fig. 3C shows that the concentration of rapamycin used in this experiment was sufficient to inhibit phosphorylation of p70 S6K , the downstream target of TOR activity. The presence of LY 294002 did not lead to detectable p70 S6K inhibition, reflecting the fact that although PI 3-kinase contributes to the regulation of p70 S6K , there are multiple pathways involved including some utilizing protein kinase C that are not sensitive to inhibition of PI 3-kinase. Therefore, p70 S6K does not seem to be required for S phase entry in fibroblasts, in agreement with previous observations (26). The effects of drug treatments on earlier time points after serum stimulation were also studied (Fig. 3D). In agreement with the data shown in Fig. 3A, LY 294002 caused a profound delay in induction of cyclin D 1 , whereas PD 98059 causes only a slight delay.
Akt/PKB Activation Increases Levels of Endogenous Cyclin D 1 -The Akt/PKB kinase is believed to be activated by PI 3-kinase generated PIP 3 through a sequential mechanism. The binding of the lipid to its PH domain recruits the enzyme to the plasma membrane, where it is phosphorylated at Thr-308 and Ser-473. A kinase, termed PDK1, that can phosphorylate Thr-308 only when Akt/PKB is bound to PIP 3 has been purified and cloned. PDK1 is itself dependent on PIP 3 for its activity (27,28). The identity of the Ser-473 kinase is not known at present. Thus, PIP 3 is required for at least two steps in the Akt/PKB activation mechanism, the first of which can be overcome by constitutively targeting the kinase to the membrane, as is observed in its viral counterpart v-Akt, in which the viral gag sequence is myristylated. Constitutively membrane-localized Akt/PKB is thus partially, but not completely, resistant to the effects of PI 3-kinase inhibitors.
Several stable cell lines expressing gag-Akt/PKB were generated and pooled to investigate the regulation of cyclin D 1 expression in the presence of enhanced Akt/PKB activity. Fig.  4A demonstrates that gag-Akt/PKB expression results in a partial rescue of cyclin D 1 expression in the presence of LY 294002 (compare with Fig. 3A). At 7 h of serum treatment, LY 294002 did reduce cyclin D 1 expression, but the effect was much less marked at 14 h, and by 20 h the amount of cyclin D 1 was not much different to that in uninhibited cells. Intriguingly, the gag-Akt/PKB transfected cells expressed low levels of cyclin D 1 in the absence of growth factors. Therefore, we determined the levels of cyclin D 1 expression in individual clones expressing gag-Akt/PKB. All cell lines showed cyclin D 1 expression in the absence of growth factors, albeit at lower levels than serum-starved V12 Ras-expressing cells (Fig. 4B). Thus, Akt/ PKB activity may mediate some, but not all, of the effects of V12 Ras on cyclin D 1 expression.
Cyclin D 1 Transcription Is Controlled by Multiple Ras Effectors-Previous work on the regulation of cyclin D 1 transcription has shown that the promoter is growth factor-regulated and can be activated by oncogenic mutants of Ras (16,17,29). As we were unable to detect any differences in cyclin D 1 protein stability in the presence of LY 294002 or upon expression of activated Akt/PKB (data not shown), we investigated the regulation of cyclin D 1 transcription in response to the expression of various Ras effectors. We used a reporter construct encompassing 1.8 kilobases proximal to the transcriptional start site of human cyclin D 1 . As shown in Fig. 5A, V12 Ras induced transcription from the reporter 4-fold. Various dominant negative alleles of Ras effectors were used to determine their contribution to this effect. Dominant negative mutants of Ral (N28 Ral) and Akt/PKB (PKB-CAAX, dn Akt) both strongly reduced activated Ras induction of the cyclin D 1 reporter. Al-FIG. 3. PI 3-kinase is required for cyclin D 1 expression. Quiescent NIH 3T3 cells were either left unstimulated (left lanes) or treated with growth factors for the indicated times. A, endogenous cyclin D 1 expression in the presence of inhibitors was analyzed at various times after growth factor stimulation. B, endogenous cyclin D 1 expression in quiescent cells treated with serum at time zero, and then with LY294002 or PD98059 after the indicated interval. Cells were lysed 12 h after serum addition. C, the ratio of p70 S6K phosphorylated species in the presence of the various inhibitors was analyzed by their differential mobility after growth factor stimulation. Shown is an immunoblot of p70 S6K species in whole cell lysates. D, endogenous cyclin D 1 expression in the presence of inhibitors was analyzed at shorter times after growth factor stimulation. though activated Rac (V12 Rac), a putative PI 3-kinase target, was also able to activate the cyclin D 1 reporter construct as described previously (30), a dominant negative form of the Rac (N17 Rac) gave only a minor inhibition of V12 Ras-induced transcription, suggesting that Rac does not function downstream of Ras in cyclin D 1 regulation. The level of V12 Ras expression was checked and found not to vary when cells were co-transfected with dominant negative effectors. In addition, as a control, c-Jun N-terminal kinase activity was checked in transfections done in parallel to Fig. 5A: N17 Rac caused 70% inhibition of the activity of epitope tagged c-Jun N-terminal kinase-1 co-expressed with V12 Ras. This compares well with the level of inhibition of c-Jun N-terminal kinase activity by N17 Rac reported by others, indicating that the dominant negative Rac was effective on other pathways. Activated forms of the various effectors of Ras were used to determine their effects on cyclin D 1 transcription. On their own, activated PI 3-kinase as well as gag-Akt/PKB both gave signals of comparable strength to that observed upon expression of an activated form of Raf kinase (Fig. 5). Because the different Ras effectors activate distinct downstream signaling cascades, we investigated whether they would cooperate to activate cyclin D 1 transcription. Another distinct Ras effector, Rlf, was also included in this analysis. Whereas expression of activated forms of Raf, Rlf, and PI 3-kinase on their own gave 2.5-3-fold inductions, the combined effect of Raf and Rlf was additive (Fig. 5B). Combined expression of PI 3-kinase and Raf as well as PI 3-kinase and Rlf synergized in cyclin D 1 induction, suggesting that they use independent pathways to activate D 1 transcription.
Different Ras Effectors Can Activate E2F-dependent Transcription-Because the presence of the PI 3-kinase inhibitor LY 294002 delayed cyclin D 1 expression (Fig. 3) and Rb hyperphosphorylation (data not shown) we reasoned that the PI 3-kinase activity may be required for the release of E2F activity from Rb-mediated repression. Therefore, we used an E2F reporter derived from the adenovirus E2 promoter to assess E2F transcriptional activity (31). Fig. 6 shows that the V12 Ras-induced increase of E2F activity can be partially suppressed by coexpression of dominant-negative constructs that block the function of the Ras effectors Ral-GDS (N28 Ral) and PI 3-kinase (⌬p85). Co-expressing a dominant-interfering construct of Akt/ PKB also reduced V12 Ras-dependent induction of E2F. Activated forms of PI 3-kinase and gag-Akt/PKB led to a pro-nounced response. The PI 3-kinase response was reduced upon co-expression of dominant-interfering Akt/PKB indicating that Akt/PKB is the predominant PI 3-kinase effector required for cell cycle progression; dominant negative Akt/PKB did not reverse the effects of activated Raf or Rlf. The observation that the Akt/PKB-mediated increase in E2F activity could be largely suppressed by co-expression of cdk inhibitor p16 suggests that PI 3-kinase acts by up-regulating cyclin D 1 to affect E2F.
As dominant negative Ral (N28 Ral) was able to partially suppress the V12 Ras-induced E2F induction, activated forms of Ral as well as Rlf were also tested. L72 Ral, an activated form of Ral, only led to a minor induction. A mutant of Ral with a decreased affinity for nucleotide that retained GTPase activity, L39 Ral was also examined. The increased exchange of hydrolyzed GDP leads to a rapid reactivation of this mutant; this approach has recently been shown to lead to strong activation of Cdc42 (32). L39 Ral led to a small but reproducible increase in E2F activity, comparable to that observed upon expression of activated Raf. These results suggest that part of the activation mediated by V12 Ras is mediated by a pathway involving Ral function, but that a greater contribution comes from the PI 3-kinase and Akt/PKB pathway. DISCUSSION Expression of oncogenic forms of Ras proteins leads to induction of cell cycle progression, causing exit of quiescent cells from G 0 and passage through at least G 1 and S phases in most cell types (4). In recent years, it has become clear that activated Ras proteins are capable of engaging a number of families of downstream effector proteins, thereby triggering in parallel the activation of several signaling systems, most notably the Raf/ MAP kinase pathway, the Ral-GDS pathway, and the PI 3-kinase pathway (13). The best characterized of these, the Raf/ MAP kinase pathway, has been known for some time to influence the expression of cell cycle regulatory proteins, but the possible involvement of the other two pathways in cell cycle control is much less well understood. This paper explores the influence of the different Ras effector systems on cell cycle regulation, in particular attempting to determine whether the function of particular Ras pathways can be correlated with the control of different check points, or whether the activity of the different pathways is integrated at each check point.
One of the earliest changes that can been seen in the expression of cell cycle regulatory proteins following release of fibro- blasts from quiescence using serum is an increase in the expression of cyclin D 1 . Expression of activated Ras will cause this in the absence of serum (8,16,17). At least in part, this is caused by activation of the Raf/MAP kinase pathway: in a variety of cell types, specific activation of Raf or MEK alone has been reported to stimulate expression of cyclin D 1 (11,18,19,33). The data in Fig. 5 show that in NIH 3T3 cells, activated Raf is able to induce cyclin D 1 reporter transcription, albeit less strongly than activated Ras. Perhaps surprisingly, activated forms of the other two Ras effectors studied here, Rlf and PI 3-kinase ␣ catalytic subunit (p110), both induce cyclin D 1 transcription at least as efficiently as does Raf. Furthermore, some combinations of effectors, especially those involving PI 3-kinase, appear to act synergistically. It is therefore likely that transcription of the cyclin D 1 gene can be induced via a number of different pathways downstream of Ras and that high efficiency induction, particularly in response to physiological stimuli, may require the function of more than one pathway, with PI 3-kinase possibly being particularly important.
Studying endogenous cyclin D 1 expression in NIH 3T3 cells released from serum starvation appears to support this hypothesis (Fig. 3). Treatment of quiescent cells with the PI 3-kinase inhibitor LY 294002 results in a considerable delay in the expression of cyclin D 1 following serum stimulation. This effect is noticeable if the inhibitor is added up to 6 h after the addition of serum. By contrast, inhibition of the Raf/MAP kinase pathway by the use of the MEK inhibitor PD 98059 results in a much more modest delay, if any, in cyclin D 1 expression. This indicates that the function of PI 3-kinase signaling pathways is more important in the serum regulation of cyclin D 1 protein levels than is the Raf/MAP kinase pathway. Others have reported that LY 294002 inhibits the ability of insulin-like growth factor I to induce cyclin D 1 expression in MCF-7 cells (34), but that PD 98059 inhibits platelet-derived growth factorinduced cyclin D 1 induction in Chinese hamster embryo fibroblasts (21). However, certain caveats exist: although PD 98059 is thought to be a highly specific inhibitor of MEK, LY 294002 is known to inhibit several enzymes other than PI 3-kinase, particularly the more distant relatives of this kinase superfamily such as TOR and DNA-dependent protein kinase (35,36). No highly specific inhibitors of PI 3-kinase have been identified. It is therefore important that the inhibitory effects of LY 294002 on cyclin D 1 expression and cell cycle progression occur at doses that correlate with inhibition of PI 3-kinase activity (Fig. 2) and that the specific TOR inhibitor, rapamycin does not have strongly inhibitory effects on cyclin D 1 expression (Fig. 3). As well as being regulated at the transcriptional level, cyclin D 1 expression may also be controlled posttranscriptionally, allowing for other factors to influence the endogenous protein level that are not apparent in reporter transcription assays. Indeed, it has recently been reported that Akt/PKB is required for up-regulation of translation of cyclin D mRNA (37) and that Akt/PKB acts through GSK-3␤ to stabilize cyclin D 1 protein (38).
A number of signaling pathways have been characterized downstream of PI 3-kinase, including the Rho-family GTPase Rac and the serine/threonine kinase Akt/PKB. Activated forms of both these proteins are able to induce cyclin D 1 reporter expression, and dominant negative Akt/PKB, and to a lesser extent dominant negative Rac, inhibits activated Ras-induced cyclin D 1 expression. Rac induction of cyclin D 1 transcription has been reported previously (30). There is therefore evidence that multiple pathways acting downstream of PI 3-kinase can contribute to cyclin D 1 induction. When endogenous cyclin D 1 levels are studied in cells stably expressing activated Ras or activated Akt/PKB, it is clear that whereas V12 Ras causes strong serum-independent expression of cyclin D 1 , activated Akt/PKB is capable of only a weak induction (Fig. 4), and Rlf-CAAX and V12 Rac do not cause any detectable induction (data not shown). In the case of expression of endogenous cyclin D 1 , multiple synergizing pathways acting downstream of Ras may therefore be particularly important, again emphasizing the possibility that regulation of cyclin D 1 protein levels occurs at a number of levels in addition to the regulation of cyclin D 1 gene transcription. Stable expression of gag-Akt/PKB, a membrane-localized activated form of Akt/PKB, can only partially abrogate the inhibitory effect of LY 294002: this may indicate that other pathways downstream of PI 3-kinase, such as Rac, play an important role here or that gag-Akt/PKB is not fully independent of PI 3-kinase activity. Although the need for PIP 3 to translocate Akt/PKB to the membrane is supplanted by the myristylation signal on the gag fusion, gag-Akt/PKB still requires phosphorylation by the upstream kinases PDK1 and PDK2, which are at least in part dependent on PI 3-kinase activity themselves (39).
Cyclin D 1 is only the earliest of the cell cycle regulators to be affected by Ras. Many other points of regulation of the cell cycle by Ras are possible at later times. A later event that is studied here is the regulation of E2F transcription factors released from Rb following its phosphorylation. When an E2F target sequence derived from the adenovirus E2 promoter was used as a reporter of E2F activity following activation of different Ras pathways, a pattern related to, but distinct from, that found for cyclin D 1 transcription was seen (Fig. 6). What is especially noticeable is that activated Akt/PKB and PI 3-kinase are particularly strong inducers of E2F activity, more so than Ras itself. Because Akt/PKB and PI 3-kinase induction of cyclin D 1 expression is less strong than Ras, this suggests that Akt/PKB may act through other pathways in addition to cyclin D 1 to control E2F activity. A very likely candidate is the regulation of the cyclin-dependent kinase inhibitor p27 Kip1 , the expression of which is suppressed following serum treatment of quiescent cells in a manner requiring Ras activity and is also downregulated in Ras transformed cells. There is lack of agreement as to whether the Raf/MAP kinase pathway can lead to reduction in p27 expression, with some reports indicating that it can (11,33) and others that it cannot (19). LY 294002 has been found to reduce the ability of growth factors to reduce p27 levels (40,41), although others report that PD 98059 can inhibit the ability of activated Ras to down-regulate p27 expression (42). Further investigation will be required to determine the signaling connections between activated Ras and induction of p27 degradation.
The picture emerging of Ras regulation of the cell cycle is one in which multiple signaling pathways are activated by Ras, which then act together synergistically at several different control points. Broadly similar conclusions were reached by Yang et al. (43), using effector loop mutants of activated Ras, Ser-35, Gly-37, and Cys-40, which show some selectivity in the downstream effector pathways that they activate (14). In addition, activation of PI 3-kinase alone has been shown to promote cell cycle entry in some cell lines (44). In this scheme, it is important to distinguish between the situation in which transformed cells express a constitutively activated Ras oncogene and that of normal cells that use endogenous wild type Ras protein to transduce signals from various extracellular growth stimuli. Strong activation of the Raf, Ral-GDS, PI 3-kinase, Akt/PKB, or Rac pathways by overexpression of the activated effectors, or of oncogenic Ras itself, may be sufficient to stimulate events that would not normally be activated by physiological activation of each pathway. As shown in Fig. 1, serum stimulation of NIH 3T3 cells leads to biphasic activation of endogenous Ras protein, as has been reported previously (22). Activation of endogenous MAP kinase and Akt/PKB does not correlate at all well with the time course of Ras activation; this is presumably because several other pathways are involved in the regulation of both these enzymes, including protein kinase C and calcium-dependent pathways in MAP kinase activation (45), induction of specific phosphatases in MAP kinase inactivation (46), and tyrosine kinases in the control of PI 3-kinase upstream of Akt/PKB (47). Under normal physiological conditions, endogenous Ras is only one of several signaling molecules influencing the activities of these enzymes, although complete removal of Ras function through the expression of dominant negative Ras protein or introduction of neutralizing antibodies may cause catastrophic failure of the pathway. Similar considerations hold true for the influence of the pathways downstream of Ras on the regulation of cell cycle check points.
During normal regulation of the cell cycle, early activation of the MAP kinase pathway, in part through endogenous Ras, may initiate events leading to induction of cyclin D expression. At later points in progression through G 1 , other pathways also controlled in part by endogenous Ras, such as Ral-GDS, PI 3-kinase, Akt/PKB, and Rac, may continue to provide stimulatory signals to cyclin D expression, even after MAP kinase activity has returned to basal levels. This requirement for activation of multiple pathways during normal growth regulation may ensure that a cell has been exposed to range of growth stimuli before it is able to progress through the cell cycle (48,49). A key to the potent oncogenicity of mutant Ras is that it can activate multiple pathways continuously that may well normally contribute to cell cycle progression in a sequential manner. Multicellular organisms may guard against the oncogenic potential of Ras by mounting an anti-proliferative response to continuous strong activation of Ras; this can be in the form of induction of expression of cyclin-dependent kinase inhibitors p21 Waf1/Cip1 and p16 INK4A (9,11,12) or the induction of apoptosis in some cell types (50,51).