Calmodulin Is Essential for Cyclin-dependent Kinase 4 (Cdk4) Activity and Nuclear Accumulation of Cyclin D1-Cdk4 during G 1 *

Although it is known that calmodulin is involved in G 1 progression, the calmodulin-dependent G 1 events are not well understood. We have analyzed here the role of calmodulin in the activity, the expression, and the intracellular location of proteins involved in G 1 progression. The addition of anti-calmodulin drugs to normal rat kid-ney cells in early G 1 inhibited cyclin-dependent kinase 4 (Cdk4) and Cdk2 activities, as well as retinoblastoma protein phosphorylation. Protein levels of cdk4, cyclin D1, cyclin D2, cyclin E, p21, and p27 were not affected after CaM inhibition, whereas decreases in the amount of cyclin A and Cdc2 were observed. The decrease of Cdk4 activity was due neither to changes in its association to cyclin D1 nor to changes in the amount of p21 or p27 bound to cyclin D1-Cdk4 complexes. Calmodulin inhibition also produced a translocation of nuclear cyclin D1 and Cdk4 to the cytoplasm. This translocation could be responsible for the decreased Cdk4 activity upon calmodulin inhibition. Immunoprecipitation, calmodulin affinity chromatography, and direct binding experiments indicated that calmodulin associates with Cdk4 and cyclin D1 through a calmodulin-binding protein. The facts that Hsp90 interacts with Cdk4 and that its inhibition induced Cdk4 and cyclin in response to the activation of DNA damage-induced G 2 checkpoint has also recently been shown (64). Our results reveal a novel mechanism for Cdk4 activity regulation that may operate in response to different extracellular signals or to the activation of cell cycle checkpoints in which Ca 2 1 might be involved.

CaM-dependent protein kinases II and IV and myosin light chain kinase, phosphatases, such as calcineurin, RNA-binding proteins, such as hnRNP A2 and hnRNP C, and the autoantigen La/SSB), indicating that CaM regulates nuclear functions (1).
It is now well established that CaM is involved in the regulation of the cell cycle (2). By using expression vectors capable of inducibly synthesizing CaM sense or antisense mRNAs, it has been shown that progression through G 1 and mitosis exit is sensitive to changes in the intracellular concentration of CaM (3). Furthermore, the addition of specific anti-CaM drugs to cell cultures inhibits reentry of growth-arrested cells into the cell cycle (G 0 /G 1 transition), progression into and through the S phase, and entry and exit from mitosis (4 -10).
Recent reports also give some suggestions of the role of CaM on G 1 /S transition, although some of the results are controversial. The addition of anti-CaM drugs to NRK cells during the early G 1 inhibits the onset of DNA synthesis (9). Inhibition of the CaMKII also blocks G 1 progression in NIH 3T3 cells (11). On the contrary, calcineurin, but not CaMKII, has been shown to be essential for G 1 /S transition in Swiss 3T3 cells (12). The activation or expression of several enzymes and proteins involved in DNA replication, such as DNA polymerases ␣ and ␦ and the proliferating cell nuclear antigen, is inhibited after the addition of anti-CaM drugs during early G 1 (9,13,14). Furthermore, CaM has also been shown to be essential for retinoblastoma protein (pRb) phosphorylation (15). All of these results suggest that CaM is essential for the activation of the cell cycle regulatory machinery involved in progression from G 1 to S phase.
In mammalian cells, progression through the cell cycle is regulated by a family of serine/threonine protein kinases called cyclin-dependent kinases (Cdks). These kinases are formed by two subunits, a catalytic subunit that is present throughout the cell cycle and a regulatory subunit, called cyclin, that is present only at specific stages of the cell cycle. Cyclin D-Cdk4, cyclin E-Cdk2, cyclin A-Cdk2, and cyclin B-Cdk1 are sequentially activated during G 1 , G 1 /S transition, S phase, and mitosis, respectively (16 -19).
The activity of the cdks is regulated by several mechanisms: cyclin binding, phosphorylation, and binding to specific Cdkinhibitory proteins (20,21). Activating and inhibitory phosphorylation sites exist in the catalytic subunit (22)(23)(24). To date, seven Cdk-inhibitory proteins have been identified in mammalian cells (21). They are divided into two major families: INK4 and CIP/KIP. The INK4 family includes p16, p15, p18, and p19, which interact specifically with Cdk4 and Cdk6. Addition of p16 to cyclin D-Cdk4 complexes results in the dissociation of the kinase complex, and binding of p16 to monomeric Cdk4 prevents cyclin D association (25). The CIP/KIP family of inhibitors comprises three members: p21, p27, and p57. They have a broad range of specificity and can inhibit all of the G 1 cdk-cyclin complexes. Moreover, they have higher affinity for the cyclin-Cdk complexes than for the monomeric cdk subunit (26,27).
The major function of cyclin D-Cdk4 during G 1 is the phosphorylation of the pRb family members (28,29). The transcription factor E2F is bound to the hypophosphorylated form of pRb, but after phosphorylation of pRb, E2F is released, and then it activates transcription of S-phase-specific genes (30,31). Among the genes containing E2F-binding sites in their promoter regions are those encoding for enzymes and proteins that are essential for DNA replication, such as DNA polymerase ␣, thymidine kinase, ribonucleotide reductase, cyclin A, and cyclin E (32)(33)(34)(35)(36). The activation of cyclin D1-Cdk4 and phosphorylation of pRb is thus a limiting step for G 1 progression. Cdk2-cyclin E is also essential for G 1 /S transition. It also phosphorylates pRb and possibly some other proteins essential for the onset of DNA replication (18,37).
Despite the evidence indicating that CaM plays a role in G 1 progression, not much is known about the specific mechanisms regulated by CaM during this phase. Here, we analyzed the effect of the addition of an anti-CaM drug during G 1 on the activation and regulation of G 1 cyclin-cdk complexes.

EXPERIMENTAL PROCEDURES
Cell Cultures-NRK cells were made quiescent by growing them to confluence in Dulbecco's minimum essential medium supplemented with 5% FCS and then kept for 3 days in the same medium but with only 0.5% FCS. To allow them to reenter the cell cycle, quiescent cells were trypsinized and subcultured at a lower density in fresh medium supplemented with 5% FCS. Unless otherwise indicated, the anti-CaM drug W13 or the control drug W12 was added to the cell culture medium at a final concentration of 15 g/ml at 5 h after proliferative activation of the cells. Other anti-CaM drugs used were W7, J8, and calmidazolium. They were also added 5 h after proliferative activation. In vivo DNA synthesis was examined by measuring the incorporation of [ 3 H]thymidine into whole cell DNA. NRK cells (10 5 per 35-mm plate) were pulse-labeled for 1 h with [methyl-3 H]thymidine (5 Ci/mmol) (Amersham Pharmacia Biotech) at 4 Ci/ml in medium supplemented with 5% FCS. Precipitation and solubilization of DNA were done as described previously (9).
Immunoprecipitation and Kinase Assays-To detect proteins associated with cdk4, immunoprecipitations were performed as described in Ref. 40. Cells (5-10 ϫ 10 7 ) were lysed in 1 ml of Buffer A (50 mM Tris-HCl, pH 7.4, 0.1% Triton X-100, 5 mM EDTA, 250 mM NaCl, 50 mM NaF, 0.1 mM Na 3 VO 4 , 1 mM phenylmethylsulfonyl fluoride, and 10 g/ml leupeptin). 3-5 mg of protein from the lysates were incubated with 4 g of anti-Cdk4 (sc-260-R, Santa Cruz Biotechnology) or anticyclin D1 (06 -137, Upstate Biotechnology) antibodies or with 3 l of normal rabbit serum (controls) for 2 h at 4°C. Protein immunocomplexes were then incubated with 40 l of protein A-Sepharose (Pierce) for 1 h at 4°C, collected by centrifugation, and washed three times in Buffer A. Immunoprecipitated proteins were then analyzed by electrophoresis and Western blotting. A lysate from NRK cells was always loaded in the same gel in order to have a control for the mobility of each protein. To analyze coimmunoprecipitation of CaM with Cdk4 or cyclin D1, cells were lysed in Buffer B (50 mM Tris-HCl, pH 7.4, 150 mM NaCl, 0.5% Nonidet P-40, 50 mM NaF, 0.1 mM Na 3 VO 4 , 1 mM phenylmethylsulfonyl fluoride, and 10 g/ml leupeptin) and then incubated with the anti-Cdk4 antibodies as indicated above. In this case, the protein A-Sepharose used was previously blocked with a solution of bovine serum albumin (2 mg/ml). For kinase assays, immunoprecipitations were performed similarly, but 5 ϫ 10 6 cells were used per assay, and the lysates were first precleared by addition of 30 l of normal rabbit serum and 50 l of protein A-Sepharose. The immunoprecipitated complexes were washed in kinase buffer (HEPES-Na, pH 7.4, 10 mM magnesium acetate, and 1 mM dithiothreitol) and then incubated in kinase buffer containing 20 M [␥-32 P]ATP (2-4 ϫ 10 4 cpm/pmol) and 2.5 g of histone H1 (for Cdk2) or 1 g of pGST-Rb (379 -792) fusion protein (for Cdk4 kinase) for 20 min at 30°C in a final volume of 35 l (41). Then, the samples were electrophoresed on SDS-polyacrylamide gels, and the gels were stained with Coomassie Blue, dried, and exposed to x-ray films at Ϫ80°C. GST-Rb (379 -792) fusion protein expression plasmid was a kind gift from Dr. Grañ a (Philadelphia, PA).
Immunocytochemistry-Quiescent cells were grown on glass coverslips. To detect Cdk4 or cyclin D, cells were fixed in cold methanol for 2 min. For nucleolin staining, cells were fixed in 3% paraformaldehyde/ PBS (140 mM NaCl, 5 mM Na 2 HPO 4 , 1.5 mM KH 2 PO 4 , pH 7.2) for 20 min at room temperature and permeabilized with 0.2% Triton X-100 in PBS for 10 min at room temperature. In all cases, the nonspecific sites were subsequently blocked with 1% ovalbumin/PBS for 10 min at room temperature. Cells were then incubated 1 h at 37°C in a humidified atmosphere, with the specific polyclonal antibodies anti-Cdk4 (sc-260-R, Santa Cruz Biotechnology, 1:50 dilution; C-18720, Transduction Laboratories 1:20 dilution), anti-cyclin D1 (06 -194, Upstate Biotechnology, 1:50 dilution) or anti-nucleolin (1:200 dilution; a gift of Dr. Ghisolphi-Nieto). Coverslips were then washed three times (5 min each) in PBS and incubated for 45 min at 37°C with fluorescein-conjugated anti-rabbit antibody (1:100 dilution, Sigma). After two washes in PBS, coverslips were mounted on glass slides with Mowiol (Calbiochem). The specificity of the antibody response was demonstrated by the preabsorption of the antibodies with the antigenic peptide. In order to quantify nuclear Cdk4 or cyclin D1 after anti-CaM drugs treatment, peroxidaseconjugated secondary antibodies were used and visualized developing with 3,3Ј-diaminobenzidine. The ratio of absorbance to area was obtained using KS 100 Kontron Imaging System software. Nuclei were considered stained when the ratio of absorbance to area was above 0.1.
Microinjection-Quiescent cells were seeded onto glass coverslips at 50 -60% confluence and activated as described above. Coverslips were transferred into Dulbecco's minimum essential medium (HEPES modification) (Sigma) 5% FCS fresh medium just before injection, (5-8 h after proliferative activation). The CaM-dependent protein kinase II fragment 290 -309 (42) aqueous solution was mixed with molecular weight 155,000 rhodamine-dextran (Sigma) in PBS, to act as an injection marker (final concentrations, 240 M peptide and 1% dextran). Microinjection of 1% dextran in PBS solution was used as a control. Cytoplasmic injections were carried out with a Zeiss automated injection system equipped with an Eppendorf 5246 transjector. After injection coverslips were placed into fresh Dulbecco's minimum essential medium 5% FCS medium and incubated for 1 h before fixation and immunocytochemical processing as described above to measure nuclear staining of Cdk4. To measure the labeling index with BrdUrd, coverslips were incubated until 21 h after proliferative activation, BrdUrd (3 g/ml) was added to the medium at 19 h, and cells were fixed in ethanol/acetic acid (95:5) for 30 min. BrdUrd was detected with monoclonal anti-bromodeoxyuridine antibody (Amersham Pharmacia Biotech) according to the manufacturer's procedures using fluorescein-conjugated anti-mouse antibody (dilution 1:100, Boehringer Mannheim) as secondary antibodies. Microinjected cells were detected by the rhodamine cytoplasmic signal, and BrdUrd or cdk4 nuclear staining was detected by the fluorescein signal. Quantitation of nuclear cdk4 staining was performed with the KS 100 Kontron Imaging System software. The "meangreen" (mean densitometric value of channel green) of each nucleus was recorded. Nuclei were considered stained when meangreen was above 100 grey units. The results represent the average of at least two separate experiments. Minimum totals of 80 and 40 cells were measured for peptide/dextran-and dextran-injected cells, respectively.
Calmodulin-Sepharose Affinity Chromatography-Cells (2 ϫ 10 7 ) were lysed in 500 l of Buffer C (50 mM Tris-HCl, pH 7.5, 100 mM NaCl, 1% Triton X-100, 1 mM dithiothreitol, 200 M Na 3 VO 4 , 0.1 mM phenylmethylsulfonyl fluoride, 1.5 M pepstatin, 2 g/ml leupeptin, 4 g/ml aprotinin) containing 5 mM EGTA. Solubilized proteins (1 mg) were loaded to a 300-l DEAE-Sephacel column preequilibrated with Buffer C. Proteins not bound to the column were collected. Under these conditions, endogenous CaM remains bound to the column, whereas Cdk4 is found in the flow-through fraction. An adequate volume of 100 mM CaCl 2 was added to this fraction to obtain a final concentration of 7 mM CaCl 2 and then was mixed with 300 l of CaM-Sepharose 4B preequilibrated with Buffer C containing 2 mM CaCl 2 . The resin was subsequently washed with the same buffer, and finally, the CaMBPs were eluted three times with 300 l of Buffer C containing 5 mM EGTA. Eluted fractions were subsequently analyzed by Western blot. For the pull-down experiments, flow-through from the DEAE-Sephacel (containing 5 mM EGTA) was incubated for 2 h at 4°C with 40 l of CaM-Sepharose or Sepharose alone in the same buffer (5 mM EGTA) or after the addition of 7 mM Ca 2 Cl. Samples were washed five times with lysis buffer C containing either EGTA or free Ca 2ϩ and then extracted directly with SDS-polyacrylamide gel electrophoresis Laemmli loading buffer and analyzed by Western blot.

Inhibition of CaM during Early G 1 Blocked pRb Phosphorylation and Cdk4 and Cdk2
Activation-When NRK cells were activated to proliferate from quiescence, they reenter the cell cycle and start DNA synthesis at 12 h, reaching a maximum at 18 -20 h (9, 43). Addition of different anti-CaM drugs 5 h after activation (first half of G 1 ) inhibited the entrance of the cells into S phase, the activation of DNA polymerases ␣ and ␦, and the expression of proliferating cell nuclear antigen (9,13,14). Addition of the same anti-CaM drugs later in G 1 had much less effect. We first analyzed whether this inhibitory effect was reversible. The anti-CaM drug used was W13. W12 was used as a control because it is chemically very similar to W13 but it has much lower affinity for CaM. W13 has been extensively used to inhibit CaM in cell cultures, and it is known to be highly specific at the dosages used in this work (6,9,13,14). W13 (15 g/ml) or W12 (15 g/ml) was added to the medium 5 h after proliferative activation of the cells. Drugs were removed 2 h later, and the cells were incubated with fresh media containing 5% FCS. Thymidine incorporation during 1 h was measured in nontreated and W13-and W12-treated cells at 18 h, 20 h, 22 h, and 24 h. Nontreated and W12-treated cells showed a maximum of thymidine incorporation at 20 h, whereas W13-treated cells reached the same level of thymidine incorporation but with a maximum at 22 h (Fig. 1). Thus, inhibition of CaM for 2 h at mid G 1 produced a reversible G 1 arrest, and cells synchronously proceed into S phase upon removal of W13 with a delay of 2 h. One of the key events for G 1 to S phase progression is Cdk4 and Cdk2 activation and consequently pRb hyperphosphorylation. Thus, the phosphorylation status of this protein upon anti-CaM drug addition was analyzed. The drug was always added 5 h after proliferative activation of the cells. Thus, they were added after the peak of Fos expression (data not shown).
Hyperphosphorylated pRb has a reduced mobility in SDSpolyacrylamide gels; thus, it can easily be distinguished from the hypophosphorylated form (44). In quiescent cells, pRb was hypophosphorylated. At 5 h, the phosphorylated form was already detected, and at 20 h, in cells treated with control drug (W12), almost all pRb was hyperphosphorylated ( Fig. 2A). In W13-treated cells analyzed at 20 h, the hypophosphorylated form was the most abundant, indicating that the inhibition of CaM blocked the phosphorylation of pRb.
To analyze whether the blockage of pRb phosphorylation by CaM inhibitors was due to the inhibition of Cdk4 or Cdk2, their activities upon W13 addition were measured. Quiescent cells did not show any Cdk4 or Cdk2 activity (Fig. 2B). Cdk4 was already activated 5 h after proliferative stimulation, and its activity was maintained high during G 1 (up to 10 h). Cdk2 was activated later than Cdk4. Its activity was low at 5 h, increased at 7 h, and was high at 15 h (S phase). The activities of both Cdk4 and Cdk2 in W13-treated cells were much lower than in Cells harvested at 20 h were treated at 5 h after activation with W12 or W13 (15 g/ml). 50 g of proteins from the lysates were separated on an 6% SDS-polyacrylamide gel, electroblotted, and subjected to Western blot as indicated under "Experimental Procedures." The antibody recognized the hyperphosphorylated (pRb-P) and the hypophosphorylated (pRb) forms of pRb. B, effect of anti-CaM drugs on the activity of Cdks. Quiescent (Q) NRK cells were activated to proliferate, and at the times indicated in the figure, they were harvested and lysed. The cells harvested at 7, 10, or 15 h were treated with 15 g/ml of W12 or W13 at 5 h after serum addition. Cell lysates were immunoprecipitated with antibodies against Cdk4 or Cdk2. Immunoprecipitations using normal rabbit serum (NRS) were performed as controls. After immunoprecipitation, the kinase activity was analyzed as described under "Experimental Procedures." To analyze the activity of Cdk4, a fragment of pRb was used as substrate, whereas to detect Cdk2 activity, the substrate used was histone H1.
control cells at all the times studied (Fig. 2B).
Effect of CaM Inhibition on the Levels of Cyclins and Cdks-To study how CaM could regulate the activities of Cdks, we first measured the amount of cyclins and Cdks by Western blot analysis in extracts from quiescent and proliferating cells treated with either W12 or W13. Cdk4 and Cdk2 were detected in quiescent cells. After proliferative activation, the amount of Cdk4 was unchanged, whereas that of Cdk2 increased at 5 h and then remained constant until 20 h. The addition of W13 to the cultures at 5 h did not affect the amount of either Cdk4 or Cdk2 (Fig. 3).
We further analyzed the levels of the different G 1 cyclins in W13-treated cells. As shown in other cellular types, the levels of cyclins D1 and D2 were low, although detectable, in quiescent cells (45,46). After serum addition, the amount of both cyclins increased slightly, showing a faint peak at 5-7 h. Cyclin D3 was not detected in these cells by Western blot analysis. Addition of W13 to the cultures at 5 h after activation did not have any effect on the amount of any of the D type cyclins (Fig.  3).
Cyclins E and A were not detectable in quiescent cells. Maximal levels of cyclin E were observed at 10 -15 h, corresponding to G 1 -S transition, whereas cyclin A showed a large increase at 15-20 h. The treatment with W13 did not affect the amount of cyclin E, whereas the amount of cyclin A was clearly decreased (Fig. 3).
Thus, the addition of W13 to cultures at 5 h after activation induced a decrease of the amount of cyclin A, whereas the levels of Cdk4, Cdk2, cyclin D, and cyclin E were not affected. The decrease of cyclin A could be responsible for the lower Cdk2 activity observed at 15 h in W13-treated cells (Fig. 2B).
The two-dimensional pattern of Cdk4 on two-dimensional electrophoretic gels upon W13 treatment was also analyzed. As shown in Fig. 4, several spots (pI 5.3-5.9) were recognized by anti-Cdk4 antibodies. Although changes in the intensity of several spots were observed at 5 and 7 h after proliferative activation, no significant changes between W12-and W13-treated cells were observed.
CaM Inhibition Did Not Alter the Amount of p21 and p27-Cells respond to different inhibitory stimuli by increasing the amount of the Cdk-inhibitory proteins, which bind to cyclin-Cdk4 and cyclin-Ckd2 complexes and inhibit its activity (23). Thus, we also analyzed here the possibility that W13 treatment induced an increase in the amount of p21 or p27. p21 was low in quiescent cells; it increased upon serum addition, and the levels remained constant between 5 and 15 h after activation. In contrast, the amount of p27 was high in quiescent cells and decreased upon serum addition. The treatment of the cells with W13 did not affect the levels of either inhibitor (Fig. 3).

Effect of CaM Inhibition on the Formation of Cdk4
Complexes-The possibility that CaM was involved in the formation of the cyclin D1-Cdk4 complexes was also analyzed here. Thus, cellular lysates immunoprecipitated with anti-cyclin D1 were electrophoresed and electroblotted, and the membranes were incubated with antibodies against cyclin D1, Cdk4, p27, and p21. As shown in Fig. 5, cyclin D1 present in NRK serum starved cells was already complexed with cdk4. In those complexes, high amounts of p27 were present, whereas almost no p21 was detected. At 7 h after activation, higher amounts of cyclin D1-cdk4 complexes were observed. Furthermore, the amount of p21 associated to cyclin D1 increased, whereas that of p27 decreased. The effect of CaM inhibition at 5 h after proliferative activation on the cyclin D1 complexes at 7 h was analyzed. As shown in Fig. 5, the amounts of Cdk4, p27, and p21 bound to cyclin D1 were not modified by the inactivation of CaM by W13-treatment or upon addition of the control drug W12. No change in the complexes was observed when immunoprecipitations were performed using anti-Cdk4 antibodies (data not shown).
CaM Is Essential for the Nuclear Accumulation of Cdk4 and Cyclin D1-Cdk4 and cyclin D1 are present in the nucleus during G 1 (47), which is in agreement with the nuclear location of the pRb family of proteins, the best known Cdk4 substrates. Immunocytochemical analysis indicated that the proportion of quiescent cells with nuclear staining for Cdk4 or for cyclin D1 was low (as also shown for cyclin D1 in Ref. 47). The percentage was higher at 5 h after proliferative activation, and at 7 h, almost all the nuclei were labeled in control cells (W12-treated cells and nontreated cells) (Fig. 6A). After W13 treatment, the percentage of cells at 7 h with Cdk4 and cyclin D1 nuclear FIG. 3. Effect of anti-CaM drug treatment on the levels of Cdks and cyclins. Quiescent (Q) cells were activated to proliferate by serum addition and, at the indicated times, lysed and subjected to Western blot analysis as indicated under "Experimental Procedures." Where indicated, W12 or W13 (15 g/ml) was added 5 h after serum addition. 30 g of proteins from the lysates were separated on SDS-polyacrylamide gels and transferred onto membranes as described under "Experimental Procedures." The blots were then incubated with the antibodies against the proteins indicated in the figure. For Cdk4, Cdk2, cyclin D1, and cyclin D2 (cycD1 and cycD2) analysis, the proteins were resolved on 10% SDS-polyacrylamide gels; for the detection of cyclin E and cyclin A (cycE and cycA), the proteins were separated on 8% SDS-polyacrylamide gels; and for p21 and p27, the proteins were separated on 12% SDS-polyacrylamide gels.

FIG. 4. Two-dimensional electrophoretic pattern of Cdk4. Quiescent cells (Q), cells at 5 h after activation (5 h), and cells at 7 h (7 h)
treated with W12 or W13 were lysed, and the proteins were separated by isoelectrofocusing (IEF). Then, they were electrophoresed on 10% SDS-polyacrylamide gels and transferred onto membranes as described under "Experimental Procedures." The blots were incubated with anti-Cdk4 antibodies.
staining decreased dramatically to the values of quiescent cells (Figs. 6A and 7). The decrease of Cdk4 and cyclin D1 in the nucleus correlated with an increase in the cytoplasmic reactivity with both antibodies (Fig. 7), indicating that CaM was needed to maintain Cdk4 and cyclin D1 in the nucleus. Similar results were obtained using two different anti-Cdk4 antibodies and normal skin human fibroblasts (data not shown). The effect of CaM inhibition on the nuclear localization of Cdk4 and cyclin D1 was reversible because when W13 was removed 2 h after its addition and the cells were incubated for 2 h more in a medium without the drug, Cdk4 reentered the cell nucleus (90% of nuclei had a ratio of absorbance to area Ն0.1). In order to ensure the specificity of CaM inhibition, nuclear location of cdk4 was analyzed upon treatment of the cells with different anti-CaM drugs. As shown in Fig. 6B, the four anti-CaM drugs used, W13, W7, J8, and calmidazolium, induced in a doseresponse manner a decrease in the nuclear staining of cdk4. Microinjection of the CaM-binding domain of CaMKII (CaMKII 290 -309 ) was also performed in order to inhibit CaM. Results showed a clear inhibition of DNA synthesis and Cdk4 nuclear accumulation after microinjection of CaMKII 290 -309 peptide: 21 Ϯ 7% of CaMKII 290 -309 peptide-injected cells were positive for BrdUrd versus 50 Ϯ 2% of control cells; and 42 Ϯ 10% of the CaMKII 290 -309 peptide-injected cells had intense Cdk4-labeled nuclei (meangreen Ն100) versus 85 Ϯ 3% of the control cells.
We also tried to analyze the amount of Cdk4 and cyclin D in purified nuclei, but unfortunately, with all the methods we have used (either with non-detergent-containing buffer or hypotonic buffers), more than 70% of both proteins leaked out from the nucleus and were found in the soluble fraction. Thus, it was impossible to analyze the effect of CaM inhibition on nuclear accumulation of cyclin D and Cdk4 by cell fractionation.
The effect of W13 treatment on the intracellular location of nucleolin and hnRNP A2 was also analyzed. Nucleolin contains a classical nuclear localization sequence (NLS); thus its transport to the nucleus is mediated by importin (48). HnRNP A2 has an M9 sequence that mediates its entrance to the nucleus via binding to transportin (49). Both proteins are known to shuttle in and out of the nucleus (48,50). As shown in Fig. 7, CaM inhibition by W13 did not affect the nuclear location of either nucleolin or hnRNP A2.
A role of intact cytoskeleton in the nuclear transport of proteins that do not contain a canonical NLS, such as protein kinase C, has been shown (51). Thus, we analyzed the effect of actin filaments and microtubules disruption on nuclear localization of Cdk4. Cytochalasin D (1 g/ml) or nocodazole (1 g/ml) treatment 5 h after proliferative activation did not have any effect on the nuclear staining of Cdk4 (data not shown).
Calcineurin inhibition by cyclosporin A (0.6 -2.4 g/ml) or CaMKII inhibition by KN93 (5-10 M) treatment of the cells 5 h after proliferative activation did not have any effect on nuclear localization of Cdk4 or cyclin D1 (data not shown), indicating that these CaMBPs were not involved in cyclin D1-Cdk4 nuclear accumulation during G 1 .
Cyclin D1 and Cdk4 Bind CaM-In order to study whether the effect of CaM on the nuclear location of cyclin D1-Cdk4 was mediated by a CaM-binding protein associated to the complex, the binding of Cdk4 and cyclin D from cellular extracts to a CaM-affinity column was analyzed. Cell lysates containing EGTA were first applied to a DEAE-Sephacel column in order to eliminate endogenous CaM. Whereas Cdk4 appeared in the flow-through of the DEAE-Sephacel column, CaM was eluted at 600 mM salt (data not shown). After Ca 2ϩ addition, the DEAE-Sephacel flow-through was mixed with CaM-Sepharose resin. As shown in Fig. 8A, Cdk4 and cyclin D1 were able to bind to the CaM-Sepharose in the presence of Ca 2ϩ and eluted FIG. 6. Effect of anti-CaM drug treatment on the nuclear location of Cdk4 and cyclin D1. Quiescent NRK cells (Q) were subcultured on coverslips and at the same time activated to proliferate by serum addition. At 5 h after activation, the different anti-CaM drugs were added to the cultures, and at 7 h, cells were fixed and immunostained with anti-Cdk4 and anti-cyclin D1 as described under "Experimental Procedures" using peroxidase-conjugated secondary antibodies. The percentage of labeled nuclei was determined by immunocytochemistry and quantified by an Olympus image analysis. Nuclei were considered stained when the ratio of optical density to area was higher than 0.1. The results presented here are the mean of three different experiments. For each experiment, more than 200 nuclei were analyzed. A, filled bars indicate the percentage of cycD1-labeled nuclei, and empty bars indicate the percentage of Cdk4-labeled nuclei. W12 and W13 were added to a final concentration of 15 g/ml. B, % of Cdk4 labeled nuclei was measured after addition of W12, W13, W7, J8, and calmidazolium (CMZ) at the indicated concentrations or after no drug addition (C).
FIG. 5. Effect of anti-CaM drug addition on the association of Cdk4, p21, and p27 to cyclin D. Quiescent NRK cells (Q) were activated to proliferate, and at 5 h, W12 or W13 was added to the medium. Two hours later, the cells were harvested and processed for immunoprecipitation as indicated under "Experimental Procedures" with anti-cyclin D1 (cycD1) (ϩ) or with normal rabbit serum (Ϫ) as a control. The immunoprecipitates were electrophoresed on 12% SDSpolyacrylamide gels and transferred onto membranes. The blots were incubated with antibodies against cyclin D1, Cdk4, p21, and p27. 7 h, cells lysed at 7 h without previous drug addition; W12 and W13, cells lysed at 7 h previously treated at 5 h with 15 g/ml of W12 or W13. specifically when EGTA was added. None of the proteins bound to a control Sepharose column. Bacterially expressed and purified Cdk4-GST and cyclin D1-GST were not found in the EGTA eluted fraction when applied to the CaM-Sepharose column under the same condition. To further prove that binding of cdk4 and cyclin D1 to CaM was Ca 2ϩ -dependent, pulldown experiments were performed. The flow-through of the DEAE-Sephacel column (in a buffer containing 5 mM EGTA) was divided and precipitated with CaM-Sepharose in the presence of either EGTA or free Ca 2ϩ . Bound proteins were eluted with SDS-containing buffer. A replica of the Ca 2ϩ -containing sample was also precipitated with control Sepharose. As shown in Fig. 8B, the binding of cdk4 and cyclin D1 to the CaM-Sepharose was much higher in the presence of Ca 2ϩ than of EGTA, and no binding was observed with the control Sepharose. In addition, CaM was found to immunoprecipitate with Cdk4 and with cyclin D1 (Fig. 8C). Hsp90, a CaM-binding protein that has recently been shown to associate with Cdk4 (52-54) was also detected in the EGTA-eluted fraction of the CaM-affinity column (Fig. 8A). Furthermore, in NRK cells, Hsp90 and Cdk4 also associate one to each other because Cdk4 was detected by Western blot in anti-Hsp90 immunoprecipitates from NRK lysates (Fig. 8C).
Because Hsp90 regulates the intracellular trafficking of several proteins (55), we analyzed whether inhibition of Hsp90 by the antibiotic geldanamycin had the same effect on intracellular location of cyclin D1-Cdk4 as CaM inhibition. Geldanamycin (2 M) addition to NRK cells 5 h after proliferative activation for a period of 2 h produced a clear reduction in nuclear Cdk4 and cyclin D1 (Fig. 9), whereas the total amount of Cdk4 analyzed by Western blot was not altered (data not shown).

DISCUSSION
Although it is well accepted that CaM has a role in the progression of G 1 phase and in G 1 -S transition, until now, the specific G 1 steps regulated by CaM remain unclear. However, recent reports indicate that the inhibition of CaM during early G 1 blocks the phosphorylation of pRb (15) and the expression of proteins involved in DNA replication (9), suggesting that the role of CaM in G 1 progression is to regulate the activity of the kinases that phosphorylate pRb. In fact, we show here that CaM is essential for Cdk4 and Cdk2 activation. The role of CaM is not to regulate the expression of either Cdk4 or cyclin D1, the assembly of cyclin D1-Cdk4 complexes, or the binding of either A, binding of cyclin D1-Cdk4 to a calmodulin affinity column. Cellular extracts were loaded to a DEAE-Sephacel column as described under "Experimental Procedures," and the flow-through was applied to CaM-affinity Sepharose (CaM-Seph.) or to a control Sepharose column (Seph.) in the presence of free Ca 2ϩ . After extensive washing in the presence of free Ca 2ϩ , the bound proteins were specifically eluted with EGTA. Equivalent volumes of all the fractions were subjected to 10% SDS-polyacrylamide gels and analyzed by Western blotting using anti-Cdk4, anticyclin D1, and anti-Hsp90 antibodies. PC, precolumn; FT, flow-through; W 1 and W 2 , two last washes; E 1 and E 2 , EGTA, first and second specific elutions. B, flow-through of the DEAE column was incubated with CaM-Sepharose (CaM-Seph.) in the presence of EGTA or free Ca 2ϩ or with control Sepharose (Seph.) in the presence of free Ca 2ϩ . After extensive washing, bound proteins were eluted with Laemli SDS-electrophoresis loading buffer. One-third of the unbound proteins and all of the bound proteins were resolved by SDS-polyacrylamide gel electrophoresis, and the presence of cdk4 and cycD1 was analyzed by Western blot. C, lysates of asynchronously proliferating NRK cells were immunoprecipitated as indicated under "Experimental Procedures" with anti-Hsp90, anti-Cdk4, anti-cyclin D1, or normal rabbit serum (NRS). The presence of Cdk4 in the Hsp90 immunoprecipitates was analyzed by Western blot. The presence of CaM in the Cdk4 and cycD1 immunoprecipitates was analyzed by Western blot. Cell lysates prior to immunoprecipitation were also loaded in the gels (L).
FIG. 9. Immunolocalization of Cdk4 and cyclin D1 after geldanamycin treatment. Quiescent NRK cells were subcultured on coverslips and at the same time activated to proliferate by serum addition. At 5 h after activation, 2 M geldanamycin (GA) dissolved in Me 2 SO or 0.1% Me 2 SO alone (C) was added; 2 h later, the cells were fixed and immunostained with anti-Cdk4 (cdk4) or anti-cyclin D1 (cycD) antibodies as described under "Experimental Procedures." p21 or p27, but to maintain cyclin D1-Cdk4 complexes in the cell nucleus. Because these complexes are not in the nucleus after CaM inhibition, pRb cannot be phosphorylated; consequently, the expression of E2F-depending genes is blocked, and the cell cycle stops before G 1 -S transition. The inhibition of CaM strongly decreased the levels of cyclin A and cdc2, proteins encoded by genes containing E2F-binding sites in their promoter regions, whereas in contrast, the levels of Cdk4, Cdk2, cyclin D1, p21, and p27, proteins the genes of which do not contain E2F-binding sites, were not affected by CaM inhibition. Although cyclin E also contain E2F-binding sites in its promoter region, its expression is not reduced by W13 treatment, indicating that a pRb phosphorylation-independent pathway of cyclin E expression exists in these cells.
Very little is known about the mechanisms of transport of Cdks and cyclins from the cytoplasm to the nucleus and vice versa. However, it is known that Cdk7 (the catalytic subunit of Cdk-activating kinase) is a nuclear protein that contains a classical NLS sequence that is responsible for its translocation from cytoplasm into the nucleus (56,57). The intranuclear location of cyclin A depends of its association with Cdk2, although none of these proteins has NLS sequences (58). Cyclin D1 does not have a putative NLS, whereas Cdk4 has a potential classical NLS; thus, its import into the nucleus could be mediated by importin.
In quiescent cells, Cdk4 and cyclin D1 were located in the cytoplasm, but at 5-7 h after proliferative activation, both proteins were already in the nuclei. Inhibition of CaM at 5 h with W13 induced a release of Cdk4 and cyclin D from the nucleus to the cytosol in mid-G 1 , thus not allowing phosphorylation of pRb by cyclin D1-Cdk4. The general import of proteins into the nucleus was not affected upon anti-CaM drug addition, because the transport of nucleolin (mediated by importin) and hnRNP A2 (mediated by transportin) was not modified by the treatment. These results strongly suggest that CaM participates in the retention of cyclin D1-Cdk4 but does not regulate a general mechanism of nucleocytoplasmic transport.
A role for the CIP/KIP family of Cdk inhibitors on the cyclin D1-Cdk4 assembly and translocation to the nucleus has been proposed (59). Because CaM inhibition does not alter either Cdk4 and cyclin D1 assembly or the amount of p21 and p27 bound to those complexes, CaM action on nuclear accumulation of cyclin D1-Cdk4 must be downstream of the action of these inhibitors. Most recently, a residue of cyclin D1 (threonine 156) has been shown to be essential for the nuclear import of cyclin D1-Cdk4 but not essential for the assembly of the complex (60). Thus, in agreement with our results, assembly is not sufficient for nuclear translocation of cyclin D1-Cdk4. One possibility is that this cyclin D1 residue could be essential for the association of cyclin D1-Cdk4 with a CaMBP involved in the nuclear location of the complex.
It is worth mentioning that the inhibition of CaM also produced a strong decrease in Cdk4 activity. This decrease was not produced by a reduction in the amounts of Cdk4 and cyclin D1, nor was it due to either alterations in the association of Cdk4 to cyclin D1 or an increase in the levels of p27 and p21 bound to the complexes. An increase in the levels of inhibitors of the Ink4 family cannot be involved in the decrease of Cdk4 activity because they would interfere in the binding of Cdk4 to cyclin D1, and we did not find modifications of this association. Thus, the presence of cyclin D1-Cdk4 in the nucleus may be essential to the maintenance of the kinase activation. Because Cdkactivating kinase is located in the nucleus, when Cdk4 moves to the cytoplasm after CaM inhibition, this kinase cannot be phosphorylated by Cdk-activating kinase, and it then remains inactive. However, the two-dimensional pattern of Cdk4 observed after W13-treatment was similar to that observed in W12treated cells, suggesting, although not completely proving, that the phosphorylation state of Cdk4 is not altered when CaM is inhibited. It is also possible that the association of cyclin D1-Cdk4 with other proteins, or the phosphorylation of other proteins of the complex, could be essential for kinase activation in the nucleus or its inhibition in the cytoplasm. Mahony et al. (61) have recently shown that active cdk6 complexes are predominantly nuclear in T cells. Those complexes have a molecular mass determined by gel filtration chromatography of 170 kDa and contain cdk6, cyclin D, and other nonidentified proteins (61).
We propose two ways to explain how CaM inhibition induces the release of Cdk4 and cyclin D1 from the nucleus: 1) the kinase complex is moving continuously between the cytoplasm and the nucleus, and CaM is essential for its entrance into the nucleus; 2) the kinase complex is retained in the nucleus in a CaM-dependent manner. In relation to the first possibility, a GTP-independent and Ca 2ϩ /CaM-dependent nuclear protein import that would function at high cytoplasmic Ca 2ϩ concentrations has recently been described (62). According to those results, import of all NLS containing protein should be affected by CaM inhibition under these conditions, whereas in our model, nuclear accumulation of nucleolin was not modified. Related to the second possibility, it should be emphasized that CaM-binding proteins and CaM itself are present in the nucleus of proliferating cells (1). Thus, CaM could bind to cyclin D1-Cdk4 complexes, or alternatively, CaM could participate in a posttranslational modification (for example, phosphorylation or dephosphorylation) of cyclin D1-Cdk4 complexes, essential to retain these complexes in the nucleus. An example of such a mechanism has been recently reported for the nuclear location of the transcription factor NF-AT4. The presence of NF-AT4 in the nucleus depends on its association with calcineurin, a CaMdependent phosphatase. CaM/calcineurin associates with NF-AT4 and keep it in a dephosphorylated state essential for its nuclear location (63).
We have shown that Cdk4 and cyclin D1 from a cell lysate are able to bind to CaM. Because purified Cdk4 and cyclin D1 do not bind CaM, they may form a complex with a CaMBP. We have analyzed the role of three different CaMBPs on the nuclear localization of cyclin D1-Cdk4: CaMKII, calcineurin, and Hsp90. Whereas CaMKII and calcineurin inhibition does not have any effect on nuclear localization of cyclin D1-Cdk4, Hsp90 inhibition by geldanamycin has the same effect as CaM inhibition. This, together with the facts that Hsp90 associates to Cdk4 also in NRK cells and that Hsp90 participates in the intracellular trafficking of several proteins (55), indicates that Hsp90 is a good candidate for being the CaMBP involved in the regulation of the nuclear accumulation of cyclin D1-Cdk4. It is also possible that Hsp90 is only associated to Cdk4, not to cyclin D1-Cdk4 complexes, and that Hsp90 is essential for nuclear accumulation of cdk4 in step previous to the one revealed using the anti-CaM drugs. Experiments to definitively determine the CaM-dependent steps and to identify the CaMBPs involved in nuclear location of cdk4 are under way in our laboratory.
Thus, we demonstrate here that CaM regulates the presence of cyclin D1-Cdk4 in the nucleus and consequently the phosphorylated state of pRb and G 1 progression. Inhibition of nuclear accumulation of cyclin B1 and thus cdc2 activity in response to the activation of DNA damage-induced G 2 checkpoint has also recently been shown (64). Our results reveal a novel mechanism for Cdk4 activity regulation that may operate in response to different extracellular signals or to the activation of cell cycle checkpoints in which Ca 2ϩ might be involved.