Rapamycin Inhibition of the G1 to S Transition Is Mediated by Effects on Cyclin D1 mRNA and Protein Stability*

The immunosuppressant rapamycin has been shown previously to inhibit the G1/S transition in several cell types by prolonging the G1 phase of the cell cycle. This process appears to be controlled, in part, by the rapamycin-sensitive FK506-binding protein-rapamycin-associated protein-p70 S6 kinase (p70S6k) pathway and the cyclin-dependent kinases (Cdk). We now show that in serum-stimulated NIH 3T3 cells, rapamycin treatment delays the accumulation of cyclin D1 mRNA during progression through G1. Rapamycin also appears to affect stability of the transcript. The combined transcriptional and post-transcriptional effects of the drug ultimately result in decreased levels of cyclin D1 protein. Moreover, degradation of newly synthesized cyclin D1 protein is accelerated by rapamycin, a process prevented by inclusion of the proteasome inhibitor, N-acetyl-Leu-Leu-norleucinal. The overall effect of rapamycin on cyclin D1 leads, in turn, to impaired formation of active complexes with Cdk4, a process which triggers retargeting of the p27Kip1 inhibitor to cyclin E/Cdk2. In view of this novel experimental evidence, we discuss a possible mechanism for the rapamycin-induced cell cycle arrest at the G1/S transition.

Rapamycin, initially characterized as an inhibitor of G 1 cell cycle progression, has been utilized to unravel a growth factorstimulated signaling pathway leading to the preferential translation of a specific subset of mRNAs. The drug forms a stable complex with the immunophilin FK506-binding protein, which binds to a family of kinases, FK506-binding protein rapamycinassociated protein (FRAP) 1 in human cells, the target of rapamycin in yeast (reviewed in Refs. 1 and 2). FRAP is a member of phosphatidylinositol kinase-related kinases, which impinge upon cellular events as diverse as cell cycle regulation in response to stress and DNA recombination (reviewed in Refs. 2 and 3). Although poorly understood, FRAP has intrinsic kinase activity and regulates the activation of the S6 ribosomal protein kinase p70 S6k in vivo; this requires both the FRAP kinase domain and the N-terminal domains of FRAP (4). As a result of this interaction, rapamycin causes rapid inactivation of p70 S6k and dephosphorylation of S6 in vivo (reviewed in Ref. 1). Although this has only a small effect on the overall rate of translation (5)(6)(7)(8), it greatly inhibits translation of mRNAs containing a 5Ј-terminal oligopyrimidine tract (5Ј-TOP) proximal to the cap structure (reviewed in Ref. 9). The importance of p70 S6k is shown by the finding that microinjection of quiescent fibroblasts with a polyclonal antibody against p70 S6k abolished the serum-induced entry into S phase (10).
Rapamycin selectively inhibits the activation of p70 S6k , prevents cyclin-dependent kinase activation, retinoblastoma protein (pRb) phosphorylation, and G 1 progression (1,(11)(12)(13)(14)(15)(16)(17). Inhibition of cell cycle progression may be mediated via inhibition of protein synthesis or via effects on the Cdks. Regulation of Cdk activity, which is crucial for the orderly initiation and progression of the cell division cycle, involves modulation of the level of synthesis of cyclins and Cdk inhibitors (18,19). After stimulation, cells sequentially synthesize the G 1 cyclins D and E, which are rate-limiting for entry into S phase (Ref. 19 and references therein). D-type cyclins (D1, D2, and D3), expressed in a lineage-dependent fashion, are induced in response to growth factor stimulation (20). In fibroblasts, cyclin D1 is both necessary and rate-limiting for G 1 progression (21) and its overexpression has been implicated in the development of a wide range of tumors (reviewed in Refs. [22][23][24]. It forms holoenzymes with Cdk4 and Cdk6 whose activity triggers transition through a G 1 checkpoint by phosphorylation and functional inactivation of pRb (18,19,25). In turn, pRb might help to positively regulate cyclin D1 transcription, thus establishing the existence of a regulatory loop between the two gene products (26). Although abnormalities of D-type cyclin expression can be associated with increased levels of mRNA, there is also evidence for enhanced stability of cyclin D protein in human sarcoma cells (27). Little is known about the mechanisms regulating the translation of cyclin D mRNA, although there are indications for control at the level of nucleocytoplasmic transport in cells artificially overexpressing eIF-4E (28,29).
On the other hand, Cdk inhibitors act stoichiometrically and oscillations in their levels can have a profound effect on cell proliferation (30 -35). The Cdk inhibitor p27 Kip1 is present at high levels in quiescent cells and down-regulated by mitogenic stimulation (17,32,36); the latter process being blocked by rapamycin (36). The level of p27 Kip1 protein appears to be modulated in part through translational control and in part via protein stability (37,38). However, the importance of p27 Kip1 regulation in the antiproliferative effect of rapamycin is a matter of some debate (34,39).
Inhibition of FRAP-p70 S6k signaling by rapamycin or wortmannin (1, 40 -42) also blocks the phosphorylation of two additional proteins (4E-BP1 and 4E-BP2), which interact with initiation factor eIF-4E and inhibit cap structure-dependent translation. Phosphorylation of 4E-BP1 disrupts its interaction with eIF-4E, liberating eIF-4E to interact with a conserved hydrophobic region of eIF-4G. Formation of this complex facilitates the binding of mRNA to the ribosome and promotes translation initiation (reviewed in Refs. 7 and 43). However, rapamycin does not prevent the phosphorylation of eIF-4E in primary T cells (5), Xenopus oocytes (5), Chinese hamster ovary cells in response to insulin (44), or NIH 3T3 cells in response to serum (8).
We have examined the role of rapamycin-sensitive pathways in the accumulation of cyclin D1 after serum-stimulation of quiescent NIH 3T3 cells. Our results demonstrate that rapamycin can delay the serum-stimulated accumulation of cyclin D1 mRNA. Although cyclin D1 mRNA contains a putative polypyrimidine tract in its 5Ј untranslated region, rapamycin did not affect the rate of synthesis of D1 protein. Rather, the drug promoted degradation of newly synthesized cyclin D1, resulting in decreased Cdk4 kinase activity, decreased phosphorylation of pRb, and increased association of p27 Kip1 with cyclin E/Cdk2.

EXPERIMENTAL PROCEDURES
Materials-Unless specified, chemicals were from Merck or Calbiochem. Media for cell culture (Dulbecco's modified Eagle's medium and fetal bovine serum) were from BioWhittaker, Inc. Radioactive isotopes were purchased by NEN Life Science Products. The glutathione Stransferase pRb (GST-Rb 1-79 ) fusion construct was kindly provided by Dr. L. Meijer (Roscoff, France).
Cell Culture and in Vivo Labeling-NIH 3T3 fibroblasts were seeded and maintained in Dulbecco's modified Eagle's medium containing 10% fetal bovine serum and penicillin/streptomycin (100 units/ml and 100 g/ml, respectively). Cells at approximately 60 -70% confluency were arrested in G 1 by serum deprivation for 36 h in Dulbecco's modified Eagle's medium containing 0.5% fetal bovine serum and re-stimulated to enter the cell cycle by adding 10% (v/v) serum for the times indicated in the figure legends. In vivo labeling of cells was carried out employing 100 Ci/ml of a [ 35 S]methionine/cysteine-labeling mix. To examine the rate of cyclin D1 synthesis, cells were pulsed for the last 30 min before harvesting in Dulbecco's modified Eagle's medium containing 10% dialyzed fetal bovine serum. To determine the protein's half-life, cells were pulsed for 2 h in the medium described above, 2 h after re-stimulation and chased for the next 4 h in the presence of an excess of unlabeled amino acids.
Western Blotting, Immunoprecipitations, and Protein Kinase Assay-Cell extracts were prepared at the indicated times by Dounce homogenization in ice cold Buffer A (50 mM Tris-HCl, pH 7.5, 120 mM NaCl, 20 mM NaF, 1 mM EDTA, 6 mM EGTA, 15 mM sodium pyrophosphate, 30 mM p-nitrophenyl phosphate, 1 mM benzamidine, 0.1 mM phenylmethylsulfonyl fluoride, 1% Nonidet P-40). Extracts were centrifuged at 18,000 ϫ g for 10 min at 4°C. Protein concentration was determined with a BCA protein assay kit (Pierce). To analyze extracts, equal amounts of protein (5 g) were resolved by SDS-PAGE and transferred to polyvinylidene difluoride membrane (Millipore). The resulting membranes were probed with anti-cyclin D1, Cdk4, or p27 Kip1 polyclonal antibody (Santa Cruz Biotechnology), a polyclonal serum specific for p70 S6k , or with a monoclonal antibody to pRb (Pharmingen) and revealed using the ECL system (Amersham Pharmacia Biotech). To analyze proteins present in the cyclin-Cdk complexes and monitor kinase activity, mouse polyclonal antisera to Cdk2 (Upstate Biotechnology, Inc.) or Cdk4 (Santa Cruz Biotechnology) were used. Extractions and immunoprecipitations were carried out in Buffer B (25 mM Tris-HCl, pH 7.5, 60 mM ␤-glycerophosphate, 15 mM MgCl 2 , 15 mM EGTA, 0.1 mM NaF, 15 mM p-nitrophenyl phosphate, 1 mM dithiothreitol, 0.1 mM phenylmethylsulfonyl fluoride, 0.1% Nonidet P-40) using equal amounts of total protein (300 g). Antibodies were immobilized on protein A-Sepharose beads, and the resin washed with 3 ϫ 1 ml of ice-cold Buffer B and 1 ml of ice-cold Buffer C (50 mM Tris-HCl, pH 7.5, 10 mM MgCl 2 , 1 mM dithiothreitol. For kinase assays, the resin was resuspended in a total volume of 25 l of Buffer C, containing 5 M [ 32 P]ATP (specific activity, 50 Ci/nmol), and 10 g of GST-Rb 1-79 fusion protein as substrate. GST-Rb 1-79 phosphorylation was analyzed by SDS-PAGE and visualized by autoradiography. A Cdk2 activity assay was performed as described (45) using 50 M p70 S6k 408 -427 peptide as substrate. Incorporation of [ 32 P]phosphate onto the peptide substrate was determined by spotting 20-l aliquots on P81 paper (Whatman) (45). For Western blotting, recovered proteins were eluted with Laemmli sample buffer and resolved by SDS-PAGE.
Northern Blotting-Small scale RNA preparation was performed as described (46), and its quality assessed by examining the 18 S/28 S ratio. Oligo-primed labeling of mouse cyclin D1 was carried out following standard procedures, and membranes were incubated with 2 ϫ 10 6 cpm of the probe. Signals were revealed and quantified using a Phos-phorImager (Molecular Dynamics).

RESULTS
Cyclin D1 mRNA Levels Are Affected by Rapamycin-Expression of cyclin D1 mRNA is down-regulated in mitogendeprived cells and induced upon stimulation of fibroblasts with serum, epidermal growth factor, platelet-derived growth factor and basic fibroblast growth factor (47,48). To dissect the mechanism by which rapamycin regulates cyclin D1 expression, we have analyzed the transcriptional activation of cyclin D1 mRNA. NIH 3T3 cells were arrested in G 0 by serum starvation for 36 h and restimulated with 10% (v/v) serum in the presence or absence of rapamycin. Total RNA was purified at different times of activation and the level of cyclin D1 mRNA was estimated by Northern blot analysis. Hybridization of the products with a mouse specific cDNA probe showed that serum increased cyclin D1 mRNA levels and treatment of cells with rapamycin caused a delay in the accumulation as well as a decrease in the absolute amount of cyclin D1 transcript (Fig.  1A). To distinguish between effects of rapamycin on transcription and the stability of the mRNA, quiescent NIH 3T3 fibroblasts were stimulated as described above and treated either in the presence or the absence of actinomycin D. The results obtained indicate that the stability of cyclin D1 mRNA is clearly affected by rapamycin (Fig. 1B).
Rapamycin Inhibits Serum-stimulated Cyclin D1 Protein Expression-Examination of the 5Ј-untranslated region (5Ј-UTR) of the human cyclin D1 gene suggested the presence of a putative polypyrimidine tract at the start site of transcription (49). This element, comprising a cytidine residue at the cap site followed by 4 uninterrupted pyrimidines, is similar to sequences demonstrated to confer translational control in a cell type-and sequence context-dependent manner (9,50). To address the question of whether D-type cyclin mRNAs are under such selective translation control, we examined the role of the FRAP-p70 S6k signaling pathway on the expression of cyclin D1 protein. Fig. 2A shows that serum increased the expression of cyclin D1 protein within 4 -6 h following addition. This was greatly reduced and delayed by treatment of cells with rapamycin, in a manner similar to that reported for ribosomal proteins in T lymphocytes (51). Under these conditions, p70 S6k activity was completely inhibited (see Ref. 45 and below). To exclude the possibility that the observed drop in cyclin D1 protein was nonspecific, the effect of rapamycin on the expression of Cdk4 was also examined. Fig. 2B shows that rapamycin did not influence the level of expression of this protein. Similar results were obtained for p70 S6k (see below). Pulse-labeling experiments with [ 35 S]methionine/cysteine, indicated that at each time point examined, the initial rate of cyclin D1 synthesis was unchanged by the presence of rapamycin (data not shown). Taken together, these data indicate that the p70 S6k signaling pathway is not required for enhanced synthesis of cyclin D1 and that the putative 5Ј-TOP sequence in the mRNA may not have a role in vivo.
When rapamycin was added to cells 4 h after serum-stimulation, only a partial decrease in the accumulation of cyclin D1 protein was observed (35% versus 55%) (Fig. 2C). These data are in agreement with earlier reports (13,52) that cells already in G 1 are less responsive to the inhibitory effects of rapamycin.
Rapamycin Stimulates the Rate of Degradation of Cyclin D1 Protein-The data above indicate that rapamycin affected cyclin D1 at both the transcriptional and post-transcriptional level. To examine other possible effects of rapamycin that may influence the expression of cyclin D1 protein, we have considered the rate of cyclin D1 polypeptide degradation. To directly compare half-lives, quiescent NIH 3T3 cells were stimulated in the presence or absence of rapamycin, pulsed with [ 35 S]methionine/cysteine labeling mix and chased in the presence of an excess of cold methionine/cysteine. The data presented in Fig. 3 show that the turnover rate of cyclin D1 was clearly accelerated upon treatment of cells with rapamycin. To further substantiate this finding, we carried out studies with the proteasome inhibitor, N-acetyl-Leu-Leu-norleucinal (LLnL). As described above, rapamycin decreased the accumulation of cyclin D1 protein. However, this was prevented by the simultaneous addition of LLnL (Fig. 4A). This compound did not influence proteins with a longer half-life, such as Cdk4 (Fig. 4B) and p70 S6k (Fig. 4C). To rule out possible effects of LLnL on cyclin D1, which might be independent of inhibition of the proteasome, the compound was administered to cells either alone or in combination with rapamycin and the levels of cyclin D1 were examined 8 h post-stimulation. Fig. 4D shows that LLnL affected solely protein degradation.
Reduced Levels of Cyclin D1 Affect Cdk4 Activity and pRb Phosphorylation-The cause of the G 1 /S block observed after administration of rapamycin to cells in culture has been attributed to a number of distinct events. Among those is an overall inhibition of protein synthesis due to decreased ribosome assembly (51), the lack of formation of active cyclin E-Cdk2 complexes (14) and the prevention of p27 Kip1 elimination from the cyclin E-Cdk2 complex (36). To assess whether the rapamycininduced degradation of cyclin D1 has any physiological effect on Cdk4 activity and G 1 progression in these cells, we have assayed Cdk4 activity using a GST-Rb 1-79 fusion protein as substrate. The increase of Cdk4 activity that occurred within 4 -6 h after serum stimulation was prevented by the addition of rapamycin to the cells (Fig. 5A). In addition, we analyzed the phosphorylation status of endogenous pRb by separation of cell extracts by SDS-PAGE and Western blotting with serum specific for pRb protein. As indicated by the characteristic shift in mobility upon phosphorylation and inactivation, pRb phosphorylation was increased after serum stimulation of cells (Fig.  5B). This mobility shift was prevented by rapamycin, suggesting a lack of pRb phosphorylation in these cells and indicative of a decrease in Cdk4 activity. Western blot analysis of immunoprecipitated Cdk4 indicated that inhibition of Cdk4 activity was not caused by a decrease in Cdk4 protein levels ( Fig. 5C) but was because of lack of the kinase regulatory partner cyclin D1 (Fig. 6A).
Reduced Levels of Cyclin D1 Facilitate Retargeting of p27 Kip1 to Cdk2-The inhibitor protein p27 Kip1 has been found to associate with both Cdk4 and Cdk2 via the regulatory partner, cyclins D and E, respectively (19,23). The distribution of p27 Kip1 among different cyclin-kinase complexes has been predicted to maintain a correct balance between their active forms (19,53,54). To determine whether inhibition of cyclin D1 accumulation in these cells affected the association of p27 Kip1 with Cdk4 or Cdk2, cells were stimulated with serum in the presence or absence of rapamycin and Cdk4 or Cdk2 complexes Serum-starved NIH 3T3 fibroblasts were stimulated with complete medium in the absence or presence of 20 nM rapamycin (rap) (A). Total RNA was extracted at the indicated time points and analyzed by Northern blot hybridization employing a mouse-specific cDNA probe to cyclin D1 as described under "Experimental Procedures" (upper panel). Before hybridization the membrane was stained with methylene blue to confirm equal loading of each sample (lower panel). In B, cells were treated as in A, with the exception that 2 g/ml actinomycin D was added 1 h post-stimulation. Total RNA was extracted and analyzed as in A. Open and closed circles denote presence or absence of rapamycin, respectively. Signals were revealed and quantified using a PhosphorImager (Molecular Dynamics). recovered by immunoprecipitation. The antisera employed were highly specific and showed no cross-reactivity with other Cdks (data not shown). Treatment of cells with rapamycin resulted in a decrease in the association of p27 Kip1 with Cdk4 (Fig. 6B). Conversely, complexes of Cdk2 and cyclin E appeared to be enriched in p27 Kip1 upon treatment of cells with rapamycin (Fig. 6B). Accordingly, rapamycin caused an almost 8-fold inhibition of Cdk2 activity (Table I). These data suggest that, apart from any direct effect on phosphorylation of pRb, the decrease in cyclin D1 protein levels facilitates retargeting of p27 Kip1 to cyclin E-Cdk2 complexes and in turn inhibition of the associated kinase activity. This further promotes the G 1 /S block in the cell cycle.

DISCUSSION
Treatment of quiescent cells with growth factors or hormones induces re-entry into the cell cycle, accompanied by an increase in the synthesis of RNA and protein and culminating in a doubling of protein mass prior to DNA synthesis and cell division (reviewed in Refs. 43 and 55). Physiological regulation of protein synthesis is almost always exerted at the level of polypeptide chain initiation, mediated in part by the availability of eIF-4E to participate in the initiation process and the phosphorylation of ribosomal protein S6 (reviewed in Ref. 43).
Here we set out to investigate the contribution of rapamycinsensitive signaling pathways in conveying external growth signals to the cell cycle regulatory machinery. Recent studies showed that regulated translation functions in modulating the activity of Cdks in mammalian cells (2,38,53). One possible target for regulation is cyclin D1, whose synthesis is induced during the delayed early response to growth factor addition to cells. The D-type cyclins are the first cyclins synthesized during the cell cycle and complex with Cdk4 and Cdk6. As such, they are detected in mid G 1 and function as pRb kinases, overcoming the growth-suppressive effects of pRb and thus allow progression through G 1 /S (18,21,47,56). It has been postulated that this class of cyclins represents the link between extracellular signals and the cell cycle machinery (20). Cyclin D1 activity appears to be responsible for activation of cyclin E and cyclin A, both partnered by Cdk2, which are active at the G 1 /S boundary and during S phase, respectively (57)(58)(59)(60)(61).
We observed that cyclin D1 mRNA and protein (Figs. 1A and 2A) were increased after a lag period of 2-4 h after stimulation of quiescent NIH 3T3 fibroblasts to proliferate. Rapamycin treatment significantly affected the level of cyclin D1 transcript, apparently causing a delay in its accumulation. This can be attributed to effects on either transcription or stability of the mRNA. The former possibility could not be assessed because of low template activity. With regard to the latter, inspection of the 3Ј-UTR of cyclin D1 mRNA did not reveal the presence of AU-rich sequences, which account for rapid degradation of interleukin-3 mRNA by a rapamycin-sensitive mechanism (62). However, measurement of cyclin D1 mRNA stability showed a substantial effect of rapamycin on the transcript's half-life (Fig.  1B). This indicates that novel rapamycin-sensitive pathways FIG. 4. The proteasome inhibitor, LLnL, rescues cyclin D1 degradation induced by rapamycin. Serum-starved NIH 3T3 fibroblasts were treated as described in the legend to Fig. 2, except that the proteasome inhibitor LLnL (50 M) was added together with rapamycin at the time of restimulation. Extracts were prepared at the times indicated and subjected to SDS-PAGE and immunoblot analysis, employing antisera specific to cyclin D1 (A), Cdk4 (B), or p70 S6k (C). D, extracts obtained from synchronized NIH 3T3 cells 8 h post-stimulation in the presence or absence of 20 nM rapamycin, 50 M LLnL, or a combination of the two were resolved and probed with an antiserum specific to cyclin D1.
FIG. 5. Rapamycin-induced cyclin D1 degradation results in a decreased Cdk4 activity and pRb phosphorylation. Serumstarved NIH 3T3 fibroblasts were treated as described in the legend to Fig. 2, and extracts were prepared at the times indicated. A, cell extracts containing equal amounts of protein were subject to immunoprecipitation with serum specific to Cdk4, as described under "Experimental Procedures." Cdk4 activity was assayed in immunocomplex assay using GST-Rb 1-79 fusion protein as substrate followed by SDS-PAGE and autoradiography. B, phosphorylation of endogenous pRb was visualized after separation of cellular proteins by SDS-PAGE and Western blotting. The slower and faster migrating bands correspond to hyper-and hypophosphorylated pRb, respectively. C, cell extracts were fractionated as described above and the level of Cdk4 protein was determined by Western blotting.  are possibly involved in regulation of cyclin D1 mRNA levels.
The effect of rapamycin on the transcript level resulted in a decreased and delayed accumulation of cyclin D1 protein ( Fig.  2A). On the other hand, inhibition of FRAP-p70 S6k signaling did not influence the level of other proteins, i.e. Cdk4 and p70 S6k (Figs. 2B and 4, B and C), for the duration of our experiments. This finding suggests that the effect of rapamycin during transition through G 1 was not because of overall inhibition of protein synthesis. This was confirmed by the observation that polysome formation was unaffected, and the overall rate of translation was only inhibited by 20% at 8 h after serum stimulation, 2 despite inhibition of the phosphorylation of 4E-BP1 within 1 h of rapamycin treatment. Analysis of the 5Ј-UTR of the human cyclin D1 gene suggested the presence of a putative tract of oligopyrimidines (TOP) proximal to the cap structure (49). This element was demonstrated to confer translational control in a cell type-and sequence context-dependent manner (9,50). Rapamycin is capable of potently and selectively inhibit this regulatory mechanism (9). Contrary to observations involving ribosomal protein S6, which contains a functional 5Ј-TOP sequence (63,64), rapamycin did not induce a shift of cyclin D1 mRNA to the untranslated population and the initial rate of its synthesis was unchanged by the presence of rapamycin. This indicates that the putative 5Ј-TOP sequence in cyclin D1 is not functional in vivo. On the other hand, pulse-chase experiments revealed that the protein's half-life was decreased upon treatment of cells with rapamycin, suggesting an effect of the drug on cyclin D1 stability (Fig. 3). This finding was confirmed by studies on inhibition of protein degradation by the proteasome inhibitor LLnL. The compound rescued cyclin D1 levels indicating that rapamycin reduced the accumulation of cyclin D1 protein by promoting its degradation (Fig. 4). For the time being, the mechanism behind this is not understood but stability of cyclin D1 protein appears to be important in regulating the abundance of this protein in human sarcoma cells (27).
Once active, cyclin-Cdk complexes are responsive to a second control mechanism, exemplified by antiproliferative signals such as transforming growth factor ␤ (65-67). This is mediated by the synthesis (38,53) and activity of Cdk-inhibitor proteins (68,69) such as p27 Kip1 . This protein is present at high levels in resting fibroblasts and decreases after mitogenic stimulation (53,70). It is postulated that removal of p27 Kip1 from cyclin E-Cdk2 complexes is an essential step for S phase entry. In some cell types, cyclin D1 sequesters p27 Kip1 , providing an alternative mechanism for the activation of cyclin E-Cdk2 and cell cycle progression (32,54,66,71). Inhibition of FRAP-p70 S6k signaling has been shown to prevent the removal of p27 Kip1 from cyclin E-Cdk2 complexes (36). Consequently, we have analyzed the association of p27 Kip1 with cyclin D1 and cyclin E. Rapamycin decreased the association of p27 Kip1 with cyclin D1-Cdk4 complexes but increased its association with cyclin E-Cdk2 (Fig. 6). Our data suggest that in NIH 3T3 cells the low levels of cyclin D1 induced by rapamycin compromise the formation of active Cdk4 complexes. This in turn has two major consequences: (i) it obstructs Cdk4-dependent pRb phosphorylation, the release of active E2F, and probably the synthesis of cyclins E and A; (ii) it causes redistribution of p27 Kip1 to the cyclin E-Cdk2 complex and inhibition of Cdk2 activity. These events, along with additional effects on inhibition of translation of other 5Ј-TOP containing mRNAs, ultimately cooperate in delaying the G 1 /S transition of the cell cycle.