Ras activity regulates cyclin E degradation by the Fbw7 pathway

Abstract

The Skp1–Cullin1 F-box protein–Fbw7 ubiquitin ligase regulates phosphorylation-dependent cyclin E degradation, and disruption of this pathway is associated with genetic instability and tumorigenesis. Fbw7 is a human tumor suppressor that is targeted for mutation in primary cancers. However, mechanisms other than mutation of Fbw7 may also disrupt cyclin E proteolysis in cancers. We show that oncogenic Ha-Ras activity regulates cyclin E degradation by the Fbw7 pathway. Activated Ras impairs Fbw7-driven cyclin E degradation, and, conversely, inhibition of normal Ras activity decreases cyclin E abundance. Moreover, activation of the mitogen-activated protein kinase pathway is the essential Ras function that inhibits cyclin E turnover, and activated Ha-Ras expression inhibits both the binding of cyclin E to Fbw7 and cyclin E ubiquitination. Last, we found that oncogenic Ras activity potentiates cyclin E-induced genetic instability but only when cyclin E is susceptible to degradation by Fbw7. Thus, we conclude that Ras activity regulates Fbw7-mediated cyclin E proteolysis and suggest that impaired cyclin E proteolysis is a mechanism through which Ras mutations promote tumorigenesis.

Cyclin E-Cdk2 is an evolutionarily conserved kinase that phosphorylates proteins involved in G₁ progression (e.g., p27 and Rb) (1, 2), DNA replication (e.g., NPAT and Cdt1) (3, 4, 42), and centrosome duplication (e.g., nucleophosmin and CP110) (5, 6). Although cyclin E-Cdk2 regulates many cell-cycle processes, it is not absolutely required for cell division. Most tissues in cyclin E knockout mice develop normally, and cells derived from these animals proliferate nearly normally in culture (7, 8). However, cyclin E-null mice have also revealed essential roles for cyclin E in endoreduplication and cell-cycle reentry from quiescence. Furthermore, cyclin E-null cells resist transformation by dominant oncogenes, suggesting that cyclin E has a key role in oncogenic signaling.

Many cancers express high levels of cyclin E, and cyclin E overexpression correlates with increased tumor aggression (9). Moreover, cyclin E deregulation is thought to have a fundamental role in tumorigenesis, and excess cyclin E activity impairs S-phase progression and DNA replication, hastens G₁ transit, and causes genomic instability (10–13). Thus, it is critically important that cells strictly regulate cyclin E activity.

Cyclin E activity normally oscillates, and this oscillation results from E2F-dependent transcription and ubiquitin-mediated proteolysis (9, 14). Two pathways promote cyclin E ubiquitination. Cyclin E that is not bound to Cdk2 is ubiquitinated by a pathway involving Cul3 (15). When cyclin E is bound to Cdk2, it is instead targeted for ubiquitination by the Skp1–Cullin1 F-box protein (SCF)–Fbw7 ubiquitin ligase, and this ubiquitination requires specific cyclin E phosphorylations (16, 17). Fbw7 binds to phosphorylated cyclin E and catalyzes its ubiquitination by bringing it into proximity with the remainder of the SCF ubiquitin ligase. At least four cyclin E phosphorylation sites (T62, S372, T380, and S384) regulate Fbw7-mediated cyclin E degradation, and these sites are targets of at least three kinases, including glycogen synthase kinase 3 (GSK-3) and Cdk2 (16–19). Fbw7 is a human tumor suppressor, and its loss, at least in some cultured cells, causes genetic instability associated with cyclin E deregulation (17, 20, 21). Fbw7 also targets c-Myc, Notch, and c-Jun for ubiquitination, and its mutation in cancers may affect pathways that regulate cell growth, division, and differentiation (22–25).

Several observations prompted us to investigate the impact of Ras signaling on cyclin E degradation. First, cyclin E-null cells resist Ras-mediated transformation, suggesting that cyclin E may be a downstream target of oncogenic Ras activity (7). Moreover, ectopic Ras expression in the developing Drosophila wing increases cyclin E expression without altering cyclin E mRNA abundance (26), and oncogenic Ras increases cyclin E abundance in rodent hepatocytes (27). Here, we report that Ras activity regulates cyclin E degradation by the Fbw7 pathway.

Methods

Antibodies and Plasmids. The following antibodies were used: 9E10-anti-myc-tag, cyclin E (HE-12, Santa Cruz Biotechnology), polyclonal phospho-T62 cyclin E (19), Cdk2 (Transduction Laboratories, Lexington, KY), FLAG (M2, Sigma), 12CA5-anti-hemagglutinin (HA)-tag, Ha-Ras (sc-520, Santa Cruz Biotechnology), p21 (Transduction Laboratories), and extracellular signal-regulated kinase (ERK)1/2 and phospho-ERK1/2 (9,102/9,121, Cell Signaling Technology, Beverly, MA). Cyclin E, Cdk2, and Fbw7 expression vectors and retroviral constructs have been described (18, 24). Human Ha-Ras cDNA was cloned from primary fibroblast mRNA after reverse transcription (Omniscript, Qiagen, Valencia, CA), and Ha-Ras point mutations were made by using the QuikChange method (Stratagene) and sequenced. Ha-Ras cDNA was inserted into pBabe to retroviral constructs. The viral Raf22W ORF (C. Der, University of North Carolina, Chapel Hill) and activated MEK1 cDNA (serine-to-glutamate substitutions at positions 218 and 222) were inserted into CS2+ with an N-terminal HA tag to generate a transfection constructs. Retroviral constructs used to stably express short hairpin (sh)RNA for Fbw7 and a control shRNA have been described (24, 28).

Cell Culture, Transient Transfection, Retroviral Transduction, and Inhibitors. The following cells were used and grown in DMEM with 10% FCS: HeLa, NIH 3T3, early passage human foreskin fibroblasts, Phoenix-Ecotropic producer cells (G. Nolan, Stanford University, Stanford, CA) and p21-null mouse embryo fibroblasts (J. Roberts, Fred Hutchinson Cancer Research Center). Retroviral transductions: Human cells were first transduced with retrovirus expressing the murine ecotropic receptor (D. Miller, Fred Hutchinson Cancer Research Center). Viral supernatants were prepared by transfecting Phoenix cells, and target cells were transduced with filtered supernatants in DMEM and 5 μg/ml polybrene. Selection was performed with puromycin (human cells, 2 μg/ml; murine cells, 10 μg/ml) or hygromycin (human, 400 μg/ml; mouse, 800 μg/ml). Transient transfections were performed by using the calcium phosphate precipitation method, except for p21-null murine embryonic fibroblasts, for which FuGENE 6 (Roche) was used. Cells were treated with the following inhibitors (all obtained from Calbiochem): U0126 (10 μM for 12 h), LY294002 (10 μM for 12 h), MG-132 (5 μg/ml for 12 h), and FTI277 (1 μM for 20 h). For in vivo ubiquitination assays, cells were treated with MG-132 24 h after transfection and lysed in Nonidet P-40 buffer containing 5 mM N-ethylmaleamide (Sigma).

Immunoblotting, Immunoprecipitation, and Kinase Assays. Cell extracts were prepared in Nonidet P-40 buffer (0.5% Nonidet P-40/10 mM Tris, pH 7.4/0.15 M NaCl), with 10 mg/ml each of aprotinin, leupeptin, and pepstatin, 50 mM sodium fluoride, and 1 mM sodium vanadate. Lysates were quantitated, electrophoresed, and transferred to poly(vinylidene difluoride) membranes. To quantify protein abundance (Fig. 2D), an infrared-emitting anti-mouse secondary antibody was used (LI-COR) with the Odyssey imaging system. For immunoprecipitations, cell lysates were incubated with primary antibody coupled to protein G–Sepharose for 90 min at 4°C. For immunoprecipitation kinase assays, cyclin E immunoprecipitates were washed; adapted to kinase buffer with 10 mM MgCl₂, 1 mM DTT, and 70 mM NaCl; and incubated with histone H1 and [γ-³²P]ATP (5 μCi per reaction; 1 Ci = 37 GBq), at 37°C for 30 min.

Fig. 2.

Ras activity regulates endogenous cyclin E abundance. (A)(Upper) HeLa cells were transduced with the indicated shRNA retroviral vectors, selected in puromycin, and subsequently transduced with either Ha-RasG12V or control vectors. After second selection, whole-cell extracts were immunoblotted (Cdk2-loading control). (Lower) Endogenous cyclin E mRNA abundance in the samples shown above as determined by quantitative real-time PCR. Error bars indicate SD from three separate measurements for each sample. (B) HeLa cells were transduced with control or Ha-RasG12V vectors and metabolic pulse–chase analysis of endogenous cyclin E protein turnover is shown. (C) Immunoblot analysis of endogenous Ha-Ras from HeLa cells treated with 1 μM FTI277 or solvent for 20 h was performed to confirm impaired Ras processing. (D)(Upper) Cells were transduced with shRNA vectors, selected in puromycin, and treated for 20 h with 1 μM FTI277 or solvent. Cells were then harvested for flow cytometry, immunoblotting, and cyclin E mRNA analyses as indicated. The cell-cycle distribution of each sample is indicated. (Lower) Cyclin E mRNA and protein abundance was quantitated in the experiments described above.

Pulse–Chase. HeLa cells (in 6-cm dishes) were transfected or transduced with the indicated vectors, starved for 30 min in met-free DMEM (ICN) containing 5% dialyzed FCS, and labeled with 1 ml of trans-³⁵S-label (300 μCi/ml, ICN) for 30 min. Dishes were washed; “chased;” at the indicated times, washed with cold PBS; and stored at -80°C before lysis and immunoprecipitation.

Phosphopeptide Mapping. Transfected HeLa cells (in 10-cm dishes) were washed in phosphate-free media (ICN) and labeled with P32-orthophosphate (1 mCi/ml) in 3 ml of phosphate-free media containing 5% dialyzed FCS for 3 h. Cell lysis and phosphopeptide mapping were performed as described (18).

Quantitative Real-Time PCR. RNA was isolated from HeLa cells (RNeasy, Qiagen), normalized, and cDNA-synthesized by oligo(dT) priming (Omniscript). Real-time PCR was performed with a PRISM 7700 sequence detector (Applied Biosystems) in triplicate 50-μl reactions containing 5 μl of cDNA, 25 μl of 2× TaqMan Universal Master mixture (Applied Biosystems), and probe/primer mixture for cyclin E (Synthegen, Houston) or GAPDH (Invitrogen). Quantitation for each probe/primer combination was determined relative to serial dilutions of HeLa cell cDNA with sequence-detection system Version 2.1 software (Applied Biosystems).

Cell Cycle Analysis. Cells were trypsinized at the indicated times, fixed in 70% ethanol, washed, and incubated in PBS containing propidium iodide and RNase A. Analyses were performed on a Becton Dickinson FACScan with multicycle software.

Micronucleation Assay. Human foreskin fibroblasts were trypsinized, washed in PBS, and nuclei fixed in 1:3 acetic acid/methanol and dropped onto slides at an approximate density of 5 × 10⁴ nuclei per slide. Nuclei were hybridized with a pan-centromeric, FITC-labeled FISH probe (ID Labs Biotechnology, London, ON, Canada) according to the manufacturer's protocol and counterstained (VECTASHIELD with DAPI, Vector Laboritories). Fluorescence microscopy was performed with an Eclipse 800 (Nikon), and micronuclei were counted.

Results

Activated Ras Inhibits Fbw7-Mediated Cyclin E Destruction. We first examined the effect of activated Ha-Ras on Fbw7-driven cyclin E degradation by coexpressing Ha-RasG12V with Fbw7 and cyclin E in NIH 3T3 and HeLa cells. Ha-RasG12V prevented Fbw7-mediated cyclin E elimination in both cell types (Fig. 1A). We directly measured the effects of Ha-RasG12V on cyclin E turnover by using metabolic pulse–chase analysis, and we found that oncogenic Ras activity increased the half-life of cyclin E in the presence of Fbw7 at ≈40–120 min but did not increase the amount of cyclin E synthesized during the labeling period (Fig. 1B and data not shown). Thus, RasG12V expression impaired Fbw7-driven cyclin E degradation.

Fig. 1.

Activated Ras inhibits Fbw7-mediated cyclin E degradation. (A) HeLa cells were transfected with myc-tagged (mt)-cyclin E, Cdk2, FLAG-Fbw7α, and Ha-RasG12V plasmids as indicated and whole-cell lysates were immunoblotted. Cdk2 is shown as a loading/transfection control. (B) Cells were transfected as in A and pulse–chase analysis of Fbw7-mediated cyclin E turnover was performed in the presence or absence of activated Ha-Ras. (C) HeLa cells were transfected with plasmids expressing WT or Cdk2 nonbinding (R130A) mt-cyclin E and activated Ras as indicated, and whole-cell lysates were immunoblotted. (D) HeLa cells were transfected with equal amounts of mt-cyclin E, Cdk2 or kinase-dead Cdk2, and Ha-RasG12V plasmid as shown. (E) HeLa cells were transfected with the indicated cyclin E expression vectors ± RasG12V and immunoblotted as shown.

We next used three approaches to cripple cyclin E turnover by the Fbw7 pathway (by inhibiting regulatory cyclin E phosphorylations) and determined whether Ras increased cyclin E abundance under these conditions. We first examined a cyclin E mutant, R130A that resists Fbw7-mediated degradation because it cannot bind Cdk2 (18), and we found that Ras did not increase cyclin E R130A abundance (Fig. 1C). Next, we inhibited cyclin E phosphorylation by expressing a catalytically inactive Cdk2 mutant (kdCdk2) that protects cyclin E from degradation by Fbw7 (18), and we found that RasG12V did not increase cyclin E abundance when kdCdk2 was cotransfected (Fig. 1D). Last, we examined cyclin E mutants in which regulatory phosphorylation sites were changed to alanines. Expression of activated Ras increased the abundance of cyclin E with mutations at phosphorylation sites that do not regulate Fbw7-mediated cyclin E turnover (S58 and S75). In contrast, RasG12V had no impact on the abundance of mutants that are resistant to Fbw7 (T380A and S384A), and it reduced the impact on mutations that partially prevent Fbw7-dependent cyclin E turnover (T62A and S372A) (Fig. 1E). In summary, we found that although RasG12V impaired Fbw7-driven cyclin E degradation, it did not affect the abundance of cyclin E when it was rendered resistant to degradation by Fbw7.

Ras Activity Regulates Endogenous Cyclin E Abundance. The experiments outlined above demonstrated that activated Ras prevented Fbw7-driven cyclin E degradation. Next, we used retroviral constructs to express RasG12V in HeLa cells to determine its effect on endogenous cyclin E. In these experiments, cells were first transduced with a vector that stably expresses a shRNA that inhibits all three Fbw7 isoforms (sh-Fbw7) or a control shRNA (sh-C) (24), and then subsequently RasG12V or a control vector. RasG12V substantially increased endogenous cyclin E protein abundance in cells expressing sh-C (Fig. 2A). However, when we coexpressed sh-Fbw7, we found that RasG12V did not increase cyclin E abundance beyond that caused by the Fbw7 knockdown alone. To examine the possibility that activated Ras might increase cyclin E mRNA synthesis, we used quantitative RT-PCR to measure the effects of Ha-RasG12V on cyclin E mRNA, and found that cyclin E mRNA abundance was unaffected by RasG12V. We also determined that RasG12V expression did not alter the asynchronous cell-cycle kinetics of these cells, thus the differences in cyclin E abundance were not secondary to altered cell cycle distribution (data not shown). Last, we used metabolic pulse–chase analyses to demonstrate that RasG12V prolonged endogenous cyclin E turnover (Fig. 2B). In summary, these data indicate that Ras regulates endogenous cyclin E stability. Moreover, the finding that RasG12V did not further augment cyclin E abundance in cells in which Fbw7 was inhibited suggests that Ras and Fbw7 function within the same pathway to regulate cyclin E stability. These data are in agreement with the transfection data indicating that RasG12V failed to stabilize cyclin E that was rendered resistant to degradation by Fbw7.

The experiments described above indicate that exogenous oncogenic Ras caused endogenous cyclin E protein to overaccumulate. Next, we examined whether physiologic Ras activity also regulates endogenous cyclin E abundance by using the pharmacologic farnesyl transferase inhibitor FTI277 (29) to inhibit endogenous Ras activity in cells expressing either sh-Fbw7 or sh-c. To eliminate the possibility that changes in cyclin E abundance might be secondary to altered cell-cycle distribution, we determined the maximal concentration of FTI227 that could be used without grossly altering cell-cycle kinetics (1 μM) and verified that Ras processing was inhibited (Fig. 2C) (30). We found that HeLa cells treated with 1 μM FTI227 for 24 h had normal amounts of cyclin E mRNA and unaltered cell-cycle kinetics but exhibited a 50% reduction in the amount of cyclin E protein (Fig. 2D). However, when cells were first transduced with sh-Fbw7, FTI277 treatment did not reduce cyclin E protein abundance. Thus, augmented or reduced Ras activity increased or decreased endogenous cyclin E abundance respectively, and modulation of cyclin E abundance did not occur when Fbw7 was inhibited by RNA interference.

Ras Stabilizes Cyclin E by Activation of the Mitogen-Activated Protein Kinase (MAPK) Pathway. Ras activates multiple signal transduction pathways, including the MAPK cascade, the PI3 kinase (PI3K) pathway, and the Ral guanine nucleotide dissociation stimulator-regulated family of Rho GTPases (31). We used several approaches to determine which of these activities was important for Ras-mediated cyclin E stabilization. We first used specific drug inhibitors of either the MAPK or PI3K pathways and found that in HeLa cells, MAPK inhibition by the MEK inhibitor U0126 blocked Ras-mediated cyclin E stabilization, but that the PI3K inhibitor LY29042 did not (Fig. 3A). Next, we used retroviral vectors to express Ras mutants that selectively activate each downstream pathway (32). We found that RasG12V/G35S, which activates the MAPK pathway, increased endogenous cyclin E abundance as well as RasG12V, but that the Ras mutants that selectively activate PI3K and Ral guanine nucleotide dissociation stimulator did not (Fig. 3B). Last, we directly tested the role of MAPK activation in cyclin E turnover by expressing activated downstream MAPK components (Raf and MEK). In both cases we found that MAPK pathway activation prevented cyclin E turnover by Fbw7, and inhibition of cyclin E turnover was reversed by pharmacologic MEK inhibition with U0126 (Figs. 3 C and D). Thus, MAPK activation is sufficient to prevent cyclin E turnover and is the pathway through which activated Ras prevents cyclin E degradation.

Fig. 3.

Ras stabilizes cyclin E via activation of the MAPK pathway. (A) HeLa cells were transfected with mt-cyclin E, Cdk2, and activated Ha-Ras plasmids as indicated. At 24 h after transfection, cells were treated with MEK inhibitor (U0126, 10 μM), PI3K inhibitor (LY294002, 10 μM), or solvent for 12 h before harvesting and immunoblotting. MAPK activity is indicated by phospho-ERK immunoblotting. (B) HeLa cells were transduced with constructs expressing Ha-Ras mutants (V12 alone constitutively activates Ras; V12/S35 activates the MAPK cascade; V12/G37 activates the Ral guanine nucleotide dissociation stimulator pathway; V12/C40 activates the PI3K signaling cascade). Whole-cell lysates were immunoblotted for endogenous cyclin E. (C and D) 293T and HeLa cells were transfected with mt-cyclin E, Cdk2, FLAG Fbw7α, and activated Raf or MEK-expressing constructs as shown. The indicated samples were treated with MEK inhibitor overnight before preparing whole-cell extracts for immunoblotting.

Activated Ras Does Not Inhibit Regulatory Cyclin E Phosphorylations. Phosphorylation of c-Myc and c-Jun by MAPKs regulate their degradation (33, 34). Thus, we thought it would be likely that Ras activity impaired cyclin E turnover via altering cyclin E phosphorylation. We first used orthophosphate labeling and phosphopeptide mapping and found that Ras did not alter phosphorylation at any of the regulatory C-terminal sites (S372, S384, and T380) within cyclin E (Fig. 4A). Moreover, no additional cyclin E phosphorylations that could be detected by phosphopeptide mapping resulted from RasG12V expression. We then used phosphorylation-specific anti-cyclin E antibodies that specifically recognize regulatory phosphorylations at T62, T380, and S384, and we found that steady state phosphorylation of these sites was not altered by RasG12V expression (Fig. 4A and data not shown). These data indicate that RasG12V does not significantly alter the phosphorylation of the four sites known to regulate cyclin E degradation by Fbw7.

Fig. 4.

Activated Ras inhibits Fbw7-dependent cyclin E ubiquitination in vivo.(A)(Left) Cyclin E phosphopeptide maps were generated from ³²P-orthophosphate-labeled mt-cyclin E isolated by immunoprecipitation from extracts of control or RasG12V-expressing HeLa cells. The identities of the peptide spots were determined in ref. 18. (Right) Equalized amounts of immunoprecipitated cyclin E obtained from control or RasG12V-expressing HeLa cells were immunoblotted by using an antibody that detects phosphorylated threonine 62 (19). (B) HeLa cells were transfected with mt-cyclin E, Cdk2, and either activated Ha-Ras or control plasmid as indicated. Histone H1 kinase activity was measured from cyclin E immunoprecipitates (myc-E), which were also immunoblotted along with total lysates (TL) as indicated. (C) p21-null mouse embryo fibroblasts were transfected with mt-cyclin E and Cdk2 plasmids and, where indicated, FLAG-Fbw7α and Ha-RasG12V plasmids. Whole-cell lysates were subjected to the immunoblot analyses as indicated. (D) NIH 3T3 cells were transfected with control plasmid alone or cyclin E and, where indicated, FLAG-Fbw7α, β, and RasG12V-expressing vectors. Cells were treated with MG-132 24 h later, and whole-cell extracts were prepared. Fbw7 complexes were immunoprecipitated with anti-FLAG antibody, and the amount of bound cyclin E revealed by immunoblotting. (E) NIH 3T3 cells were transfected with mt-cyclin E, Cdk2, and where indicated, HA-ubiquitin, FLAG-Fbw7α, and activated Ras (Left) or Raf (Right) plasmids. Cells were treated with MG-132 as shown and polyubiquitinated cyclin E species were detected by 9E10 immunoblotting. (F) (Upper) NIH 3T3 cells were transfected with mt-cyclin D and, where indicated, HA-ubiquitin and activated Ras, treated with MG-132, and cyclin D1-ubiquitin conjugates were resolved by 9E10 immunoblotting. (Lower) Whole-cell lysates were blotted with anti-HA to identify the total pool of ubiquitinated proteins in control or activated Ras-expressing cells.

When cyclin E-Cdk2 is bound to either p21 or p27, it is catalytically inactive, and the cyclin E in these complexes is underphosphorylated and stable (35). Because activated Ras induces p21 expression (36, 37), we considered the possibility that RasG12V prevented cyclin E turnover by increasing the amount of p21 bound to cyclin E-Cdk2. However, this was not the case. First, although we observed that activated Ras increased p21 expression, its net effect was elevated, rather than inhibited, cyclin E kinase activity (Fig. 4B). Next, we directly tested the role of p21 induction by performing cyclin E turnover assays in p21-null murine fibroblasts. We found that RasG12V prevented cyclin E degradation equally well in p21-null mouse embryo fibroblasts (Fig. 4C). Therefore, although activated Ras increased p21 abundance, p21 induction was not the mechanism through which it stabilized cyclin E.

Activated Ras Inhibits Fbw7-Dependent Cyclin E Ubiquitination in Vivo. We next examined whether Ras activity altered the binding of cyclin E to Fbw7. In these experiments, we treated cells with the proteasome inhibitor MG-132 to allow stable interactions between cyclin E and full-length Fbw7. We found that RasG12V reduced the amount of cyclin E that coprecipitated with both Fbw7α and Fbw7β, suggesting that activated Ras stabilized cyclin E by preventing its association with Fbw7 in vivo (Fig. 4D). Because the Fbw7 isoforms exhibit unique subcellular localizations (28), we also determined whether Ras altered the subcellular localization either cyclin E or Fbw7, and we found that activated Ras expression caused neither protein to mislocalize (data not shown).

To determine whether the decreased association of cyclin E with Fbw7 caused by RasG12V resulted in decreased cyclin E ubiquitination, we established an in vivo assay for Fbw7-dependent cyclin E ubiquitination in NIH 3T3 cells. In this assay, cotransfection of Fbw7 with cyclin E, Cdk2, and HA-ubiquitin causes increased cyclin E-ubiquitin conjugates that are visualized by treating cells with the proteasome inhibitor MG-132. However, when activated Ras was coexpressed with cyclin E and Fbw7, cyclin E ubiquitin conjugation was substantially reduced (Fig. 4E Left). Similar results were obtained when activated Raf, rather than Ras, was expressed, indicating that MAPK pathway activation impairs cyclin E ubiquitination by Fbw7 (Fig. 4E Right). Importantly, Ras activation reduces but does not eliminate cyclin E ubiquitination, and this finding is consistent with our data that Ras impairs, but does not completely abrogate, cyclin E degradation by Fbw7. We also determined whether the decreased cyclin E ubiquitination resulted from a general decrease in protein ubiquitination caused by RasG12V. However, cyclin D1 ubiquitination, as well as the conjugation of HA-ubiquitin with total cellular proteins, was unaffected by RasG12V (Fig. 4F). Thus, Ras specifically impairs cyclin E turnover and ubiquitination, rather than causing globally impaired ubiquitin conjugation.

Activated Ras Cooperates with Cyclin E to Induce Genetic Instability. Because Fbw7 degrades the catalytically active fraction of cyclin E, we predicted that coexpression of RasG12V with cyclin E would produce biologic effects consistent with elevated cyclin E activity. Moreover, activated Ras should not further augment the activity of cyclin E that is resistant to Fbw7 (e.g., cyclin E-T380A). We used two biological assays for cyclin E hyperactivity to test these predictions. We first examined the cell-cycle kinetics of NIH 3T3 cells transduced with retroviruses expressing either cyclin E or cyclin E-T380A with or without RasG12V. Cyclin E overexpression caused cells to accumulate in S-phase, and this accumulation was exaggerated with cyclin E-T380A (Table 1). Importantly, coexpression of activated Ras with cyclin E led to a further increase in S-phase that closely approximated that seen with cyclin E-T380A alone, but it did not further increase the S-phase fraction of cells expressing cyclin E-T380A. Moreover, the S-phase fractions mirrored the relative amount of cyclin E-associated kinase activity seen in these cell populations (Fig. 5A).

Table 1. Activated Ras cooperates with cyclin E to produce S-phase accumulation.

Vector	S-phase, %	SD
pBabe	14	0.3
Cyclin E+ pBabe	37	4
Cyclin E+ Ras G12V	57	2
Cyclin E(T380A)+ pBabe	54	2
Cyclin E(T380A)+ Ras G12V	54	0.7

NIH 3T3 cells were transduced with the indicated cyclin E or control retroviral vectors, followed by a second transduction with Ha-RasG12V-expressing or control vectors. Asynchronously proliferating cell cultures were harvested after selection and fixed for flow cytometry. Cell-cycle distributions were measured for triplicate samples, and S-phase populations with SD are shown.

Fig. 5.

Activated Ras cooperates with cyclin E to produce excess kinase activity and genetic instability. (A) Cyclin E immunoprecipitates from cells shown in A were used to measure the amount of cyclin E-associated histone H1 kinase activity in these sampes. (B) Early passage human foreskin fibroblasts were transduced with retroviral constructs as indicated. After selection, nuclei were isolated and fixed for microscopic analysis. Micronuclei were identified by DAPI staining and by fluorescence in situ hybridization with a FITC-labeled pan-centromeric probe. Numbers of micronuclei were then compared among the indicated cell populations after counting 300 normal nuclei (nuclei were counted three times from separate slides for a total of 900 for each sample) and expressed as ratios to normal nuclei. Error bars indicate SD for each sample.

Cyclin E Deregulation Induces Genetic Instability (12, 13). We previously found that cyclin E-T380A expression caused cytogenetic abnormalities in primary human fibroblasts that greatly exceeded that seen with WT cyclin E. These data indicated that in primary human cells, Fbw7-mediated cyclin E degradation must be disabled in order for cyclin E to generate genetic instability. Rajagopalan et al. (20) recently reported a rapid assay for cyclin E-induced genome instability based on the formation of micronuclei that are detectable in interphase cells by DAPI or centromeric FISH probes (Fig. 5B). Thus, we determined whether RasG12V and cyclin E cooperatively induce micronucleation in early passage primary human foreskin fibroblasts (which were used to avoid Ras-induced senescence; ref. 38). Consistent with our previous studies, cyclin E-T380A-induced micronucleation greatly exceeded that caused by WT cyclin E (which was barely above background). Remarkably, we found that RasG12V increased the amount of micronucleation caused by cyclin E to a level very close to that caused by cyclin E-T380A. Moreover, RasG12V did not further significantly increase the amount of micronucleation caused by cyclin E-T380A (Fig. 5B). Therefore, activated Ras cooperatively induces genetic instability with cyclin E, and this cooperativity depends on impaired Fbw7-mediated cyclin E proteolysis.

Discussion

We have shown that Ras activity regulates cyclin E turnover via the Fbw7 pathway. We used several methods to determine that activated Ras affects the stability of cyclin E only when it is susceptible to Fbw7-mediated degradation. Mutations in either cyclin E or Cdk2 that prevented cyclin E phosphorylation also prevented cyclin E stabilization by RasG12V. We also disrupted the Fbw7 pathway by using shRNA, and again we found that Ras did not stabilize cyclin E under these conditions. Whereas excess Ras activity augmented cyclin E abundance, pharmacologic Ras inhibition decreased endogenous cyclin E abundance in an Fbw7-dependent manner. These data are consistent with a model in which the amount of Ras activity in cells functions as a rheostat that regulates cyclin E abundance through its effects on Fbw7-mediated cyclin E degradation.

Because Ras activates multiple downstream pathways, we attempted to dissect Ras signaling to determine which pathway was essential for cyclin E accumulation, and we found that MAPK activation is the critical Ras function that modulates cyclin E stability. We examined the mechanisms underlying Ras-mediated cyclin E stabilization, and we found that expression of RasG12V reduced cyclin E ubiquitination driven by Fbw7 in vivo, as well as the amount of cyclin E that is bound to Fbw7 in proteasome-inhibited cells. These data suggest that Ras regulates cyclin E stability is by altering the physical interaction between Fbw7 and cyclin E. However, unlike for the cases of c-Myc or c-Jun, Ras does not alter cyclin E phosphorylation on any of its known regulatory sites in vivo, nor does purified MAPK phosphorylate cyclin E or Fbw7 in vitro (data not shown). Additional studies are underway to further explore the possibility that oncogenic Ras activity may cause unique cyclin E modifications that alter its interactions with Fbw7.

We next examined the effects of activated Ras with two assays that reflect the biologic consequences of excess cyclin E activity (cell-cycle kinetics and genomic instability). In both cases, activated Ras increased the activity of cyclin E to the same levels observed with cyclin E-T380A, but did not potentiate cyclin E-T380A activity. These observations led us to conclude that by preventing active cyclin turnover via Fbw7, activated Ras essentially converts WT cyclin E into nondegradable cyclin E with respect to its biologic activity.

Normal cells have at least two pathways that protect them against excess cyclin E activity. The first is the Fbw7 pathway, which ensures that cyclin E in active complexes is rapidly degraded, and the second is the p53-p21 pathway, which is induced by excess cyclin E, and limits its kinase activity. Full oncogenic cyclin E activity likely requires that both of these safeguards be inactivated, and one consequence of Ras mutations in tumors may be that the Fbw7-mediated turnover pathway for cyclin E is disabled. Thus, Ras might cooperate with deregulated cyclin E during tumorigenesis, and this cooperation has been shown in several systems. Activated Ras collaborates with cyclin E in the transformation of primary rodent fibroblasts (39, 40), and cyclin E-T380A is not more active than WT cyclin E in this assay, consistent with our findings that Ras does not further increase the biologic activity of cyclin E that is not sensitive to Fbw7. Activated Ras also cooperates with cyclin E during T-cell lymphomagenesis in mice. Mice harboring a CD2-cyclin E transgene do not spontaneously develop lymphomas, and despite high levels of transgene expression, the thymuses in these animals do not have excessive cyclin E activity. However, these animals are hyper-sensitive to carcinogens, and Ras mutations were found in a substantial fraction of T cell lymphomas induced by MNU (41). Interestingly, all of these tumors had high levels of cyclin E-associated kinase activity, consistent with the notion that Ras activation stabilized the pool of active cyclin E. Last, Sicinski and coworkers (7) found that cyclin E-null mouse embryo fibroblasts were highly resistant to transformation by Ras. Therefore, cyclin E activity appears essential for Ras-mediated transformation, at least in cultured rodent cells. However, the role of altered cyclin E degradation in Ras-mediated tumorigenesis has not yet been fully determined.