CDK4 protein is degraded by anaphase-promoting complex/cyclosome in mitosis and reaccumulates in early G1 phase to initiate a new cell cycle in HeLa cells

CDK4 regulates G1/S phase transition in the mammalian cell cycle by phosphorylating retinoblastoma family proteins. However, the mechanism underlying the regulation of CDK4 activity is not fully understood. Here, we show that CDK4 protein is degraded by anaphase-promoting complex/cyclosome (APC/C) during metaphase-anaphase transition in HeLa cells, whereas its main regulator, cyclin D1, remains intact but is sequestered in cytoplasm. CDK4 protein reaccumulates in the following G1 phase and shuttles between the nucleus and the cytoplasm to facilitate the nuclear import of cyclin D1. Without CDK4, cyclin D1 cannot enter the nucleus. Point mutations that disrupt CDK4 and cyclin D1 interaction impair the nuclear import of cyclin D1 and the activity of CDK4. RNAi knockdown of CDK4 also induces cytoplasmic retention of cyclin D1 and G0/G1 phase arrest of the cells. Collectively, our data demonstrate that CDK4 protein is degraded in late mitosis and reaccumulates in the following G1 phase to facilitate the nuclear import of cyclin D1 for activation of CKD4 to initiate a new cell cycle in HeLa cells.

CDK4 regulates G 1 /S phase transition in mammalian cells mainly by phosphorylating retinoblastoma family proteins (RBs) 4 to release RB-bound E2F family transcription factors (1)(2)(3). D-type cyclins, including D1, D2, and D3, of which cyclin D1 has been extensively studied, are essential for the activation of CDK4 (1,4). Dysfunction of CDK4 causes cells to be arrested at the G 1 phase in ovarian cancer cells, whereas elevation of CDK4 activity results in overcoming of the G 1 phase arrest and promotes proliferation of the cells (5)(6)(7). CDK4 protein is nuclear, whereas cyclin D1 may be cytoplasmic and relocated to the nucleus for CDK4 kinase activation (8 -13). In G 0 phase cells arrested by serum starvation, the protein level of cyclin D1 is low but elevates in response to growth factor stimulation (14). Degradation or cytoplasmic retardation of cyclin D1 reduces CDK4 activity (11, 14 -16). Nuclear accumulation of cyclin D1 also regulates cell growth, migration, differentiation, and development in addition to cell cycle control (17)(18)(19). Overexpression of cyclin D1 is also linked with tumorigenesis (12, 20 -22).
Although tremendous progress has been made in investigating the function of CDK4 during the G 1 /S phase transition, many of those results were obtained from studies based on G 0 cells. In contrast, less attention has been given to the regulatory mechanisms controlling the fluctuation of CDK4 and cyclin D1 proteins and the activation of CDK4 in cycling cells (23), although deciphering these mechanisms is fundamentally important. In this work, we studied the mechanism regulating CDK4 activation in HeLa cells.

CDK4 protein is degraded by APC/C during metaphaseanaphase (M-A) transition and reaccumulates in subsequent G 1 phase
To investigate the protein levels of CDK4 and cyclin D1 during the M-G 1 phase transition in cycling cells, we first performed immunofluorescence microscopy (IFM) in HeLa cells using specific antibodies. Surprisingly, we found that the protein level of CDK4 was decreased during the M-A transition, whereas that of cyclin D1 was unchanged (Fig. 1A). This result was confirmed in WI-38 and 3T3 cells (supplemental Fig. S1A). Then we reseeded mitotic HeLa cells collected by shake-off (Fig. 1B) and analyzed the protein levels of both cyclin D1 and CDK4 by Western blotting. We also observed that cyclin D1 protein level was stable during the G 2 /M/G 1 phases (Fig. 1C), as reported previously (24). In contrast, CDK4 protein level was decreased, and the kinase activity of CDK4 was also impaired, as indicated by unphosphorylation of RB Ser-780 (pS780-RB), a preferential phosphorylation site for CDK4 (Fig. 1C) (25). We further analyzed synchronized HeLa cells at the G 2 /M border and the indicated time points after G 2 /M release. Strikingly, we found that CDK4 protein was degraded during the M-A transition, and the phosphorylation of RB Ser-780 was decreased accordingly ( Fig. 1D and supplemental Fig. S1B). Interestingly, from middle to late G 1 phase, the protein level of CDK4 was recovered gradually, and the phosphorylation of RB Ser-780 was increased ( Fig. 1D and supplemental Fig. S1B). These results were also confirmed in 3T3 cells (supplemental Fig.  S1C). When the mitotic cells were treated with proteasome inhibitor MG132, CDK4 protein was kept at a steady high level (supplemental Fig. S1B), suggesting that CDK4 protein was degraded in mitosis via a ubiquitin-proteasomal pathway. To define the timing of ubiquitination and degradation of CDK4 proteins, we transfected HeLa cells with non-degradable GFPcyclin B1⌬90 (27,28), which sustains CDK1 activity, and arrested cells at the anaphase-telophase transition (29), treated the cells with nocodazole, and then released them into fresh medium. We observed that the protein levels of endogenous CDK4 and cyclin B1 were high in prometaphase and low in anaphase-telophase, whereas exogenous GFP-cyclin B1⌬90 was kept stable (Fig. 1E). As analyzed by microscopy, we also found that CDK4 protein was decreased during the M-A tran-sition, recovered, and was rapidly imported into the nucleus during the G 1 phase, whereas cyclin D1 was stable in the same period in all of the filmed cells ( Fig. 1F and supplemental Movies 1 and 2). In contrast, the protein levels of CDK4 and cyclin D1 and the kinase activity of CDK4 in quiescent-like cells were low but quickly increased upon serum stimulation ( Fig. 1 (G and H) and supplemental Fig. S1D). Moreover, in an in vivo ubiquitination assay, by co-expressing His-ubiquitin and GFP-CDK4 in HeLa cells followed by synchronization of the cells at the M-A transition, we also detected multiple ladder-like bands above the main band of GFP-CDK4 (Fig. 1I), indicating that CDK4 was ubiquitinated during the M-A transition. To investigate which E3 ubiquitin ligase was responsible for the ubiquitination of CDK4, we found that CDK4 interacted with Cdc20, one of the APC/C activators (Fig. 1J). By knocking down Cdc20 and Cdh1 separately, two activators for APC/C activity (26), we found that Cdc20-activated APC/C was responsible for the ubiquitination of CDK4 ( Fig. 1K and supplemental Fig.  S1E). To confirm this result, by treating Cdc20-knockdown cells with cycloheximide (CHX), we measured CDK4 degradation kinetics. The results showed that, whereas the protein level of CDK4 was decreased and eventually disappeared from 4 to 8 h with the treatment of CHX in control cells, the protein level of CDK4 in Cdc20-knockdown cells treated with CHX was stable ( Fig. 1L), indicating that Cdc20-activated APC/C did mediate the ubiquitination and degradation of CDK4. Taken together, these data demonstrate that CDK4 protein is degraded by APC/C during the M-A transition and reaccumulates in the following G 1 phase, whereas cyclin D1 protein is stable during the M-G 1 transition in HeLa cells.

Slow nuclear import and fast export retain cyclin D1 in the cytoplasm in the absence of CDK4
Next, we investigated the subcellular localizations of CDK4 and cyclin D1 in HeLa cells. Through IFM using specific antibodies, we observed that cyclin D1 was extensively localized in the cytoplasm of the paired early G 1 cells, whereas CDK4, although at a very low level, was located in the nucleus ( Fig. 2A). Along with the progress of the cell cycle, CDK4 protein was accumulated gradually and imported into the nucleus. Coincidently, cyclin D1 protein was imported into the nucleus and co-localized with CDK4 ( Fig. 2A). Through transient expres-sion of GFP-cyclin D1 and treatment of leptomycin B (LMB), an effective inhibitor for nuclear exportin CRM1 (30), we also found that GFP-cyclin D1 was mainly localized in the cytoplasm of the paired G 1 phase cells, even in the presence of LMB (Fig. 2B). Because it was reported that mutation at Thr-286 prevents cyclin D1 from binding to CRM1 and displays low nuclear export efficiency (11,31), we generated GFP-cyclin D1 T286A mutant and expressed it in HeLa cells. Interestingly, we found that this mutant also became localized in the cytoplasm within 24 h, although eventually it was imported to the nucleus in the presence of LMB by 72 h (Fig. 2B), suggesting that nuclear import of this mutant was also insufficient in early G 1 phase. By linking a nuclear localization signal (NLS) to cyclin D1 (NLS-D1-GFP) (32), we also observed that only a fraction of these fusion proteins were in the nucleus and that, when treated with LMB, all of the NLS-D1-GFPs were localized to the nucleus (Fig. 2C). Furthermore, when linking an NLS to cyclin D1 T286A -GFP (NLS-D1 T286A -GFP), we found that all of the NLS-D1 T286A -GFPs were localized in the nucleus despite the LMB treatment (Fig. 2C). Taken together, these results indicate that slow nuclear import and fast nuclear export retain cyclin D1 in the cytoplasm of the early G 1 phase cells in the absence of CDK4.

CDK4 shuttles between the nucleus and cytoplasm and promotes the nuclear import of cyclin D1
Through separate expression or co-expression of CDK4 and cyclin D1 in HeLa cells, we investigated how these proteins are transported to the nucleus. We observed that, when expressed separately, CDK4 entered the nucleus efficiently, whereas cyclin D1 was retained in the cytoplasm. When co-expressed, both CDK4 and cyclin D1 were imported and co-localized in the nucleus (Fig. 3A). In comparison, when cyclin D1 was coexpressed with CDK1 or CDK2, cyclin D1 was retained in the cytoplasm, although both CDK1 and CDK2 entered the nucleus efficiently (Fig. 3A). These results indicate that CDK4 promotes the nuclear import of cyclin D1. Because CDK6 also binds to cyclin D1 directly, we co-expressed CDK6 with cyclin D1 and found that both cyclin D1 and CDK6 were imported into the nucleus ( Fig. 3A and supplemental Fig. S2 (A and B)). These results were also confirmed by experiments done in HEK 293T and MCF-7 cells (supplemental Fig. S2C). When CDK4 was knocked down by RNA interference (RNAi), the nuclear local- Figure 1. CDK4 is degraded by APC/C-Cdc20 during M-A transition and reaccumulates from early to mid-G 1 phase. A, HeLa cells were immunostained with anti CDK4 and cyclin D1 antibodies. The early prometaphase, late prometaphase, metaphase, and anaphase/telophase cells are labeled 1, 2, 3, and 4, respectively, in a, whereas the G 2 , metaphase, anaphase, and telophase cells are labeled with 1Ј, 2Ј, 3Ј, and 4Ј in b. B, HeLa cells were synchronized at the G 1 /S transition and were then released into fresh culture medium for 10 h. Mitotic cells that were in prometaphase were shaken off, reseeded, and stained with DAPI. C, the cell lysate of the reseeded mitotic and released cells was subjected to Western blotting. D, HeLa cells were arrested at the G 2 /M border using RO3306 (9 M) and released. The lysates from these cells were analyzed by Western blotting. E, HeLa cells transfected with GFP-cyclin B1⌬90 were arrested at prometaphase and anaphase/telophase and analyzed by Western blotting. F, HeLa cells expressing GFP-CDK4 and GFP-cyclin D1 were processed for time-lapse microscopy. More than 20 cells were filmed in each of three repeated experiments, and representative movie images are shown. G, HeLa cells arrested at the quiescent-like phase were released into G 1 phase by serum at the indicated time points. The cell lysates were analyzed by Western blotting. H, quiescent-like phase to G 1 phase cells were fixed and immunostained. DNA was stained by DAPI. 200 cells were randomly counted at each time point of the three repeated experiments. Representative images are shown. I, HeLa cells were co-transfected with His-ubiquitin and GFP or GFP-CDK4 or were transfected with GFP-CDK4 alone. Transfected cells were synchronized at the G 2 /M border using RO3306 and then released to M-A transition. Lysate of cells was subjected to a pull-down assay using nickel-Sepharose beads. The isolated proteins were analyzed by Western blotting. J, HeLa cells were transfected with GFP-CDK4, and the total cell extract was used for immunoprecipitation with a GFP antibody and probed with Cdc20 and GFP antibodies. K, HeLa cells were separately transfected with negative control (NC), Cdc20, or Cdh1 RNAi double-stranded RNA followed by co-transfection with His-ubiquitin and GFP-CDK4. As negative controls, HeLa cells transfected with NC RNAi double-stranded RNA were transfected with GFP-CDK4 or co-transfected with His-ubiquitin and GFP. Transfected cells were synchronized at the G 2 /M border using RO3306 and then released to M-A transition. Lysates of cells were subjected to a pull-down assay using nickel-Sepharose beads followed by Western blotting using anti-His and -GFP antibodies. L, HeLa cells were separately transfected with negative control or Cdc20 RNAi double-stranded RNA for 72 h. The cells were then treated with or without 30 g/ml CHX for the indicated number of hours, lysed, and probed with CDK4 and ␣-tubulin antibodies. Scale bars, 10 m.

CDK4 accumulation initiates the new cell cycle
ization of cyclin D1 was significantly decreased (Fig. 3

, B-D).
To understand whether the nuclear accumulation of cyclin D1 was due to CDK4-promoted nuclear import or CDK4-mediated nuclear retention, we linked an NLS to CDK4-Myc (NLS-CDK4-Myc) and expressed this fusion protein in HeLa cells. The results showed that the forced nuclear import of NLS-CDK4-Myc, which does not weaken the binding of CDK4 with cyclin D1, impaired the nuclear localization of cyclin D1 (Fig. 3, A, E, and F), indicating that disrupting nucleocytoplasmic shuttling of CDK4 reduced the nuclear import of cyclin D1. Taken together, these data demonstrate that CDK4 shuttles between nucleus and cytoplasm, and this shuttling facilitates the nuclear import of cyclin D1.

Direct binding of CDK4 with cyclin D1 is required for CDK4mediated nuclear import of cyclin D1
Next, we tested how CDK4 mediates the nuclear translocation of cyclin D1. First, by co-expressing a CDK4 kinase-dead mutant CDK4 D158N with cyclin D1 in HeLa cells, we observed that both CDK4 D158N and cyclin D1 were imported into the nucleus (Fig. 4A), indicating that the kinase activity of CDK4 was not required for their nuclear import. In contrast, by coexpressing GFP-cyclin D1 with CDK4 fused with an NES sequence of PKI-␣ and a Myc tag (CDK4-NES-Myc), we observed that GFP-cyclin D1 was co-localized with this NEScontaining CDK4 in cytoplasm. Interestingly, when the cells were treated with LMB, both GFP-cyclin D1 and CDK4-NES-Myc were co-localized in the nucleus (Fig. 4B). We also constructed another two mutants, CDK4 K22A and CDK4 3A (K22A/ R24A/D25A), that bind to cyclin D1 weakly (Fig. 4C) (33) and co-expressed each of them with cyclin D1. We observed that CDK4 K22A , but not CDK4 3A , was able to partially target cyclin D1 to the nucleus (Fig. 4D).
It has been reported that the residues Lys-112 and Lys-114 within the cyclin box of cyclin D1 are important for its interaction with CDK4, although it is ambiguous which one is pivotal (34,35). In this work, we generated the mutants cyclin D1 K112E , cyclin D1 K114E , and D1 K112E/K114E and co-expressed each of them with CDK4 in HeLa cells. Through co-immunoprecipitation assays, we found that cyclin D1 K112E and cyclin D1 K112E/ K114E did not bind to CDK4, whereas cyclin D1 K114E retained about 30% of binding capability with CDK4 compared with the wild type (Fig. 4E). Through IFM, we observed that only GFPcyclin D1 K114E was co-localized with CDK4 in the nucleus (Fig.  4F). We also generated a CDK4 construct containing an NLS and a nucleolus localization signal (NoLS) (NLS-NoLS-CDK4-Myc) and co-expressed it with each of the cyclin D1 mutants. We observed that only cyclin D1 K114E was co-localized with NLS-NoLS-CDK4-Myc in the nucleus and the nucleoli, whereas most of cyclin D1 K112E and cyclin D1 K112E/K114E were retained in cytoplasm without co-localization with NLS-NoLS-CDK4-Myc (Fig. 4G). By establishing a tet-on system (36) to induce the expression of GFP-cyclin D1 K112E/K114E by tetracycline and time-lapse microscopy, we also observed that GFPcyclin D1 K112E/K114E was persistently cytoplasmic throughout the cell cycle (supplemental Movie 3). Collectively, these results demonstrate that only Lys-112 on cyclin D1, but not Lys-114, is essential for direct binding with CDK4 and also that this direct binding is required for the nuclear import of cyclin D1. We further tested whether another part of the cyclin D1 molecule is involved in its nuclear import. By co-expressing the GFP-or Myc-tagged cyclin box, C and N terminus deletion mutants of cyclin D1 with CDK4, we found that, no matter whether the mutant without cyclin box was expressed alone or co-expressed with CDK4, this mutant remained in the cytoplasm (supplemental Fig. S2, D and E). In contrast, when the mutants with either C or N terminus deletion were expressed alone, they were retained in cytoplasm, but, once co-expressed with CDK4, they entered the nucleus (supplemental Fig. S2F), indicating that neither the N nor C terminus of cyclin D1 contributes to its nuclear import. Taken together, we conclude that the Lys-112 residue of cyclin D1 is required for the interaction of cyclin D1 with CDK4 and the nuclear import of cyclin D1 in HeLa cells.

Stable binding of D-type cyclins with CDK4 enhances their nuclear localization
Because physical interaction between CDK4 and cyclin D1 promotes the nuclear import of cyclin D1, we decided to test whether the stability of CDK4-cyclin D1 complex influences the subcellular localization of cyclin D1 in HeLa cells. Because P16 INK4 disrupts the interaction by competing with cyclin D1 for binding with CDK4 and p21 CIP1 /p27 KIP1 directly binds and stabilizes cyclin D1-CDK4 complex (37,38), we investigated the influence of P16 INK4 and p21 CIP1 /p27 KIP1 on the complex stability of CDK4-cyclin D1 and the subcellular localization of cyclin D1. Through a co-immunoprecipitation assay, we found that the stability of cyclin D1-CDK4 complex was compromised in the cells expressing GFP-p16 INK4 . In contrast, the stability of the cyclin D1-CDK4 complex was strengthened in the cells expressing GFP-p21 CIP1 or -p27 KIP1 (Fig. 5A). IFM showed that cyclin D1 was mostly localized in the cytoplasm in GFP-p16 INK4 -expressing cells and in the nucleus without GFP-p16 INK4 expression (Fig. 5B). On the contrary, cyclin D1 was localized exclusively in the nucleus in GFP-p21 CIP1 -or GFP-p27 KIP1 -expressing cells (Fig. 5B). Western blot analysis also confirmed that cyclin D1 was mainly cytoplasmic in p16 INK4expressing cells (Fig. 5C). Because the other two D-type cyclins, D2 and D3, can also bind to CDK4 directly, by co-expressing these proteins in HeLa cells, we tested whether D2 and D3 cyclins are also imported into the nucleus along with CDK4. We found that, as a control, cyclin E was persistently in the nucleus, whereas cyclin D2 and D3 were cytoplasmic when expressed alone and entered the nucleus when co-expressed with CDK4 (supplemental Fig. S3A). These results indicate that cyclins D1, D2, and D3 share a very similar mechanism for their nuclear import and retention. Taken together, we conclude that a stable binding between D-type cyclins and CDK4 enhances their nuclear localization in HeLa cells.

CDK4/CDK6 knockdown results in cell-cycle arrest at the G 1 phase
Knowing that accumulation of CDK4 in cells facilitates nuclear import of cyclin D1 and activation of CDK4 kinase in the nucleus to promote the M/G 1 /S transition, we wanted to verify whether knocking down CDK4 would arrest post-mitotic cells in G 1 phase due to cytoplasmic retention of cyclin D1 and low CDK4 kinase activity. To do this, we knocked down CDK4 with siRNA and simultaneously expressed GFP-cyclin D1 in HeLa cells. The results showed that cyclin D1 could not be imported to the nucleus in CDK4-knockdown cells (Fig. 6 (A  and B) and supplemental Movies 4 and 5). We also combined CDK4/6 knockdown with tet-on expression of wild-type GFPcyclin D1 or mutant GFP-cyclin D1 K112E/K114E . Through time- lapse microscopy, we confirmed that only a small portion of wild-type cyclin D1 was able to enter the nucleus, whereas the total proteins of mutant cyclin D1 K112E/K114E were restrained in the cytoplasm in CDK4/CDK6-knockdown cells (Fig. 6C and supplemental Movies 6 and 7). We also examined the phosphorylation status of RB Ser-780 and found that CDK4 knockdown resulted in the accumulation of unphosphorylated RB at Ser-780 (supplemental Fig. S3B), indicating that the cells were arrested in the G 1 phase (39). Taken together, these results demonstrate that, to initiate a new cell cycle, the post-mitotic HeLa cells need to accumulate CDK4 to gain the CDK4 kinase activity.

Discussion
Collectively, in this work, we show that periodic degradation of CDK4 is directly related to the activation of CDK4 in HeLa cells. The CDK4 protein level declines dramatically during metaphase-anaphase transition, and this decline results in a dramatic loss of CDK4 kinase activity in late M and early G 1 phases. Importantly, the protein level of CDK4 increases during the following early and middle G1 phase, and accordingly, the kinase activity of CDK4 is recovered to enable the phosphorylation of RB again from middle to late G 1 phase. Although we currently do not know why CDK4 is degraded in late mitosis, we suspect that this degradation may be crucial for a cell to decide to either continue cycling or arrest at G 0 phase. Cyclin D1 is regarded as a nuclear protein (40), although it is localized in the cytoplasm at certain stages of the cell cycle. Because cyclin D1 does not possess a recognized NLS, it may be co-imported into the nucleus with other factors. In this work, we have identified that CDK4 protein shuttles between the nucleus and cytoplasm and promotes the nuclear import of cyclin D1. CDK4 may enter the nucleus without cyclin D1, but cyclin D1 alone cannot enter the nucleus. Due to the degradation of CDK4 in mitosis, cyclin D1 will be retained in the cytoplasm in early G 1 phase, although its protein level remains unchanged. Therefore, CDK4 reaccumulation after mitosis is also required for nuclear translocation of cyclin D1.
Based on our results and previously reported data, we propose a working model for the role of CDK4 in initiating a new cell cycle in HeLa cells (Fig. 7). In this model, CDK4 protein is degraded during the metaphase-anaphase transition in a proteasome-dependent manner, whereas cyclin D1 protein remains stable and is retained in the cytoplasm. CDK4 protein accumulates from the next early to mid-G 1 phase and shuttles between the nucleus and the cytoplasm to facilitate the nuclear import of cyclin D1 and the M/G 1 /S transition (Fig. 7). When the divided cells lack nutrients, such as during serum starvation, they do not synthesize CDK4 and remain in the G 0 phase. Accordingly, cyclin D1 is degraded to a basal level in these G 0 cells. Once supplied with nutrients, these G 0 cells begin to synthesize CDK4 and cyclin D1, as well as other essential factors, and initiate a new cell cycle starting from the G 0 phase (G 0 /G 1 /S transition) (Fig. 7). In conclusion, this work reveals a CDK4 degradation/reaccumulation-dependent mechanism for initiating a new cell cycle in HeLa cells. Our findings may have important implications, potentially helping us to understand cell cycle control and tumorigenesis.

Cell culture and transfection
HeLa, HEK 293T, MCF-7, WI-38, or NIH3T3 cell lines were maintained at 37°C in Dulbecco's modified Eagle's medium containing 10% fetal bovine serum (FBS), 100 units/ml penicillin, and 100 g/ml streptomycin in the presence of 5% CO 2 . WI38 cells were cultured in Eagle's minimum essential medium with Earle's balanced salts containing 10% FBS and 1% nonessential amino acids in the presence of 5% CO 2 . To express exogenous proteins, the cells were transiently transfected via the calcium phosphate or liposome techniques according to the manufacturer's instructions.

Plasmids and antibodies
Cyclin D1 cDNA was cloned from MCF-7 cells by PCR with relevant oligonucleotides. Upstream sense primer contains a BglII site and the cyclin D1 initiation codon, whereas the downstream antisense prime is complementary to the 3Ј-end of the cyclin D1 sequence containing a SalI site but having a deleted cyclin D1 termination codon. The PCR product was gel-puri-

CDK4 accumulation initiates the new cell cycle
fied and digested with BglII and SalI restriction endonucleases before ligation to pcDNA3.1myc vector that had been digested with BamHI and XhoI restriction endonucleases and pEGFP-C1 vector that had been digested with BglII and SalI restriction endonucleases. Other genes were cloned by the same method with corresponding oligonucleotides. A point mutation was constructed by a PCR-based technique with relevant oligonucleotides. All constructs were verified by DNA sequence analysis (supplemental Table 1).

Western blotting and immunoprecipitation assay
HeLa or 3T3 cells were washed twice with calcium/magnesium-free PBS and lysed in lysis buffer (0.5% Nonidet P-40, 50 mM HEPES, 150 mM NaCl, 1 mM EDTA, 1 mM DTT, 1 mM NaF, 0.1 mM NaVO 4 , 0.2 mM PMSF, and 10 g/ml protein inhibitors).  GFP-tagged proteins were immunoprecipitated with an anti-GFP antibody. The protein samples were denatured and separated by 10% SDS-polyacrylamide gel electrophoresis and transferred onto nitrocellulose membranes. The membranes were probed with primary antibodies for 2 h at 37°C, and the positive bands were detected by using the specific secondary antibody conjugated with horseradish peroxidase followed by a chemiluminescence detection system.

RNAi
HeLa cells were transfected with siRNA especially designed to silence the CDK4 or CDK6 gene. Two oligomers of CDK4specific siRNA were synthesized with the following sequences: 5Ј-GCAGCACUCUUAUCUACAU-3Ј and 5Ј-GCAGAGAU-GUUUCGUCGAA-3Ј. Two oligomers of CDK6-specific siRNA were synthesized with the following sequences: 5Ј-GCAGAAAUGUUUCGUAGAA-3Ј and 5Ј-GCAAAGACC-UACUUCUGAA-3Ј. 72 h after transfection, the cells were subjected to immunofluorescence or nuclear and cytoplasmic fractionation.

Cell-cycle synchronization
HeLa cells were synchronized at the G 1 /S phase transition by double thymidine blocks. Briefly, cells were treated with 2.5 mM thymidine for 16 h, released for 10 h, and incubated again for an additional 14 h in thymidine-containing medium. Cells blocked in the G 1 /S boundary were released for 8 h in complete medium and subsequently treated with 0.25 g/ml nocodazole for an additional 6 h to obtain prometaphase cells. Prometaphase cells were then released and treated with 5 mM MG132 for 2 h to obtain metaphase cells. Quiescent G 0 to G 1 3T3 cells were achieved by serum starvation for 72 h and release as described previously (41).

IFM
HeLa, HEK 293T, MCF-7, or WI-38 cells grown on coverslips in Petri dishes were fixed with 4% paraformaldehyde for 15 min at room temperature and permeabilized with 0.2% Triton X-100 in PBS, pH 7.4, on ice for 2 min. The cells were sealed with Mowiol containing 1 g/ml DAPI. For indirect immunofluorescence, cells were incubated with the respective primary antibodies for 1 h at 37°C. After washing, the cells were successively incubated with the secondary antibodies conjugated with TRITC or FITC for 1 h at room temperature, and the cells were sealed with Mowiol containing 1 g/ml DAPI.

Nuclear and cytoplasmic cell fractionation
HeLa cells were washed twice with cold low-penetration buffer (20 mM HEPES (pH 7.8), 5 mM potassium acetate, 0.5 mM MgCl 2 , 0.5 mM DTT, 1 mM PMSF, 20 units/ml aprotinin, 5 mg/ml leupeptin, and 0.4 mM NaF). After incubation on ice for 15 min, cells were collected with a scraper. 50 l of cell suspension mix was incubated with 50 l of 2ϫ loading buffer (12.5 mM Tris, pH 6.8, 2% SDS, 2 mM DTT, 20% glycerine, 5% ␤-mercaptoethanol, and 0.2% bromphenol blue) to obtain whole-cell sample. The rest of the cell suspension was added to a Dounce homogenizer. Nuclei were pelleted at 8,000 rpm for 5 min, and the supernatant (cytoplasmic extracts) was centrifuged at 13,200 rpm for 5 min. Nuclei were washed three times with 1 ml of PBS and suspended in 50 l of sample buffer to obtain nuclear sample.

Cell line generation and cell culture
Tet-on cell lines were generated according to the supplier's instructions (plasmids and drugs from Invitrogen) as we described previously (36). HeLa cells were transfected with pcDNA6/TR, and cultured in the presence of 10 g/ml blasticidin for 2 weeks. Positive clones were selected for further transfection of pcDNA4/TO-GFP-tagged wild-type cyclin D1 and mutant cyclin D1 K112E/K114E and cultured in the presence of 200 g/ml zeocin for an additional 2 weeks. Again, positive clones were selected and cultured for use. GFP-tagged wildtype cyclin D1 and mutant cyclin D1 K112E/K114E were expressed when 1 g/ml tetracycline was added.

Living cell imaging and microscopy
HeLa cells were cultured on dishes with glass bottoms. The dynamics of the targeting proteins during the cell cycle were recorded by spinning disc microscopy with a heated culture chamber (37°C, 5% CO 2 ) and a Nikon ϫ60 (numeric aperture 1.43) Plan-apo oil objective.

Statistical analyses
HeLa cells transfected with double-stranded siRNA and RFP-H2B for 72 h were immunostained with antibody against cyclin D1. The cells with obviously less cyclin D1 in the nucleus were counted. Each experiment was repeated three times, and error bars in Fig. 3

represent S.D.
Author contributions-C. Z., Q. J., H. C., and X. X. conceptualized the study; H. C., X. X., G. W., B. Y., G. W., and G. X. performed the experiments; J. J. L., Q. J., and H. Y. Z. discussed the project and analyzed the data. C. Z., Q. J., H. C., and X. X. wrote the manuscript; all authors approved the final version of the manuscript.