Calmodulin Binds to p21Cip1 and Is Involved in the Regulation of Its Nuclear Localization*

p21Cip1, first described as an inhibitor of cyclin-dependent kinases, has recently been shown to have a function in the formation of cyclin D-Cdk4 complexes and in their nuclear translocation. The dual behavior of p21Cip1 may be due to its association with other proteins. Different evidence presented here indicate an in vitro andin vivo interaction of p21Cip1 with calmodulin: 1) purified p21Cip1 is able to bind to calmodulin-Sepharose in a Ca2+-dependent manner, and this binding is inhibited by the calmodulin-binding domain of calmodulin-dependent kinase II; 2) both molecules coimmunoprecipitate when extracted from cellular lysates; and 3) colocalization of calmodulin and p21Cip1 can be detectedin vivo by electron microscopy immunogold analysis. The carboxyl-terminal domain of p21Cip1 is responsible for the calmodulin interaction, since p21145–164 peptide is also able to bind calmodulin and to compete with full-length p21Cip1 for the calmodulin binding. Because treatment of cells with anti-calmodulin drugs decreases the nuclear accumulation of p21Cip1, we hypothesize that calmodulin interaction with p21Cip1 is important for p21Cip1, and in consequence for cyclin D-Cdk4, translocation into the cell nucleus.

In mammalian cells progression through the cell cycle is regulated by a family of serine/threonine protein kinases called cyclin-dependent kinases (Cdks) 1 (1)(2)(3)(4). Among these kinases, cyclin D-Cdk4 is known to have an important role in G 1 progression. Cyclin D-Cdk4 participates in the phosphorylation of the pRb family of proteins allowing expression of S phase genes and thus progression to S phase (4 -9). Cyclin D1-Cdk4 activity is regulated at different levels (1,10): synthesis and degradation of the cyclin, assembly of the complex, phosphorylation (11,12), binding of Cdk inhibitors (13), intracellular localization (14,15), and most probably association with other proteins not characterized (16). Two of the Cdk inhibitors, p21 Cip1 and p27 Kip1 , have an important but controversial role in cyclin D-Cdk4 regulation. Both proteins were first described as inhibitors of cyclin D-Cdk4, cyclin E-Cdk2, and cyclin A-Cdk2. But recently, using knockout cell lines for each one of them and double knockout, a role for p21 Cip1 and p27 Kip1 in cyclin D-Cdk4 assembly and nuclear translocation has been proposed (17,18).
Activation of cyclin D-Cdk4 occurs in response to extracellular signals that, using different receptors and transduction pathways, end in transcription and stabilization of cyclin D (19 -22), assembly of the complex (23), and its translocation to the nucleus (14) where the activating kinase and the major substrates are found. CaM is a Ca 2ϩ -binding protein and acts as transducer of the intracellular Ca 2ϩ signal. When bound to Ca 2ϩ , CaM is able to bind to other proteins (CaM-binding proteins) and regulate their activity (24 -26). By binding to proteins, CaM is able to regulate important cellular processes such as cell cycle (27)(28)(29)(30). In fact, during G 1 , CaM is essential to activate Cdk4 and thus for the phosphorylation of pRb (14,31). We have shown previously that CaM regulates nuclear translocation of Cdk4 and cyclin D1 (14). This regulation may be mediated by a CaM-binding protein present in the complex, as suggested by the fact that Cdk4 and cyclin D1 from cellular lysates both bind in a Ca 2ϩ -dependent manner to CaM-Sepharose. We show here that bacterially expressed and purified p21 Cip1 , but not p27 Kip1 , is able to bind specifically to CaM and that CaM is essential for the accumulation of p21 Cip1 in the nuclei of proliferating cells.

EXPERIMENTAL PROCEDURES
Cell Cultures-NRK cells were made quiescent by growing 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 reentry to the cell cycle, quiescent cells were trypsinized and subcultured at a lower density in fresh medium supplemented with 5% FCS. Namalwa cells were obtained from American Type Culture Collection and cultured at 1 ϫ 10 6 cells/ml in RPMI 1640 supplemented with 10% FCS.
Immunoprecipitation-Immunoprecipitations were performed as described elsewhere (32). Cells (5-10 ϫ 10 7 ) were lysed in 1 ml of buffer A (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). 3-5 mg of protein from the lysates was incubated with 5 g of polyclonal anti-p21 Cip1 (sc-397, Santa Cruz), 5 g of monoclonal anti-p21 Cip1 (0P64, Calbiochem), or 5 g of monoclonal anti-CaM for 2 h at 4°C. As a control, an equal amount of a monoclonal nonrelated antibody or 1 l of normal rabbit serum was used. Protein immunocomplexes were then incubated with 40 l of protein A-Sepharose (for the polyclonal antibodies) or protein G-Sepharose (for the monoclonal an-* This work was supported by Commisión Interministerial de Ciencia y Tecnología (Spain) Grants SAF97-0069 and SAF98-0014 and National Institutes of Health Grant CA75205. The costs of publication of this article were defrayed in part by the payment of page charges. This article must therefore be hereby marked "advertisement" in accordance with 18   tibodies) for 1 h at 4°C, collected by centrifugation, washed three times in buffer A, and solubilized in Laemmli sample buffer. Immunoprecipitated proteins were then analyzed by electrophoresis and Western blotting. A cell lysate was always loaded in the same gel as a control for the mobility of each protein.
Pull-downs with Purified Proteins-All recombinant proteins were obtained as GST fusion proteins: p27 Kip1 -GST was obtained by digesting pET21a-p27 (gift of Dr. Massagué, Memorial Sloan-Kettering Cancer Center, New York) with NdeI-XboI and ligating into pGEX-KG (33); cyclin D1-GST was obtained by digesting pET3d-cyclin D1 (34) with NcoI-HindIII and ligating into pGEX-KG; p21 Cip1 -GST was obtained by polymerase chain reaction amplification of pBS-hp21 (gift of Dr. Massagué, Memorial Sloan-Kettering Cancer Center, New York) and cloning into pGEX-KG in NdeI-XboI sites; Cdk4-GST was obtained by polymerase chain reaction amplification of a B-cell cDNA library using a forward initial primer and a reverse terminal primer and cloning into pGEX-KG in BamHI-HindIII sites. GST fusion proteins were expressed in BL21 pLysE Escherichia coli strain and purified by gluthatione-Sepharose chromatography. CaM was purified from bovine brain, as described elsewhere (35). CaM and p21 145-164 peptide were bound to BrCN-activated Sepharose 4B, as indicated by the manufacturer. For CaM-Sepharose pull-downs, 1 g of purified protein was incubated for 1 h at 4°C with 20 l of CaM-Sepharose (1:1 v/v) or Sepharose alone in a buffer B (50 mM Tris-HCl, pH 7.4, 150 mM NaCl, 1% Triton X-100) containing either 1 mM EGTA or 0.1 mM CaCl 2 . After centrifugation, unbound proteins were collected and bound proteins washed three times with buffer B and eluted directly with SDS-polyacrylamide gel electrophoresis Laemmli loading buffer, electrophoresed, and analyzed by Coomassie Blue staining or Western blotting. For peptide competition experiments, CaM-Sepharose was preincubated with 2 nmol of the peptides in 200 l of buffer B (containing only 50 mM NaCl) for 1 h at 4°C. To determine binding of CaM to p21 145-164 peptide, 1 g of purified CaM was incubated for 1 h at 4°C with 20 l of p21 145-164 -Sepharose (1:1 v/v) or Sepharose alone in buffer B (containing 400 mM NaCl) containing either 1 mM EGTA or 0.1 mM CaCl 2 . After centrifugation, unbound protein was collected and bound protein washed three times with buffer B (last wash with 100 mM NaCl) and eluted directly with SDS-polyacrylamide gel electrophoresis Laemmli loading buffer, electrophoresed, and analyzed by Coomassie Blue staining.
Gel Electrophoresis and Immunoblotting-Proteins were electrophoresed in SDS-polyacrylamide gels essentially as described (38). After electrophoresis the proteins were either stained with Coomassie Blue or transferred to Immobilon-P strips for 2 h at 60 V. In the case of CaM analysis 2 mM EGTA was added in the loading buffer, and in the stacking and resolving gel solutions, transfer was for 1 h at 60 V, and Immobilon-P was then treated with 0.2% glutaraldehyde in TBS (20 mM Tris-HCl, pH 7.5, 150 mM NaCl) for 45 min and washed twice in TBS. The sheets were preincubated in TBS containing 0.05% Tween 20 and 5% defatted milk powder for 1 h at room temperature and then incubated for 1 h at room temperature in TBS, 0.05% Tween 20, 1% BSA, and 0.5% defatted milk powder containing antibodies against: Cdk4 (sc-260-R, Santa Cruz, 0.5 g/ml), cyclin D (06 -450, Upstate Biotechnology, Inc., 2 g/ml), p21 Cip1 (sc-397, Santa Cruz, 2 g/ml), and p27 Kip1 (1:500 dilution; gift from Dr. Massagué, Memorial Sloan-Kettering Cancer Center, New York) polyclonal antibodies or CaM monoclonal antibody. After washing in TBS, 0.05% Tween 20 (three times, 10 min each), the sheets were incubated with peroxidase-coupled secondary antibody (1:1000 dilution) (Bio-Rad) for 1 h at room temperature. After incubation, the sheets were washed twice in TBS, 0.05% Tween 20 and once in TBS and the reaction visualized by ECL (Amersham Pharmacia Biotech).
Electron Microscopy-To study colocalization of p21 Cip1 and CaM, NRK cells were grown on transwell filters. Small pieces of the filters were cryofixed by protection against a copper block cooled by liquid nitrogen (Ϫ196°C) using a Cryoblock (Leica) according to description (36). Freeze-substitution was performed in a homemade cryosystem (37), using acetone containing 0.5% of uranyl acetate, for 3 days at Ϫ90°C. At day 4 the temperature was slowly increased, 5°C/h, to reach Ϫ50°C. At this temperature the samples were rinsed in acetone and were infiltrated and embedded in Lowicryl HM20 as follows: 50% Lowicryl in acetone for 4 h, 75% Lowicryl in acetone overnight, 100% Lowicryl overnight and 100% fresh resin for 4 h. After infiltration the resin blocks were made and polymerized with UV lamps. Ultra-thin sections were picked up on Formvar-coated gold grids in order to carry out the immunocytochemical labeling. To study colocalization Cdk4 or cyclin D1 with CaM-pelleted cells were fixed in 3% paraformaldehyde and 0.1% glutaraldehyde and 0.1 M phosphate buffer for 1 h at room temperature, dehydrated according to the PLT technique (progressive low-ering temperature) and embedded in Lowicryl K4M. Ultra-thin sections were mounted in Formvar-carbon-coated gold grids and processed for immunocytochemical labeling as follows. After two 10-min washes with PBS (140 mM NaCl, 5 mM Na 2 HPO 4 , 1.5 mM KH 2 PO 4 , pH 7.2), grids were incubated for 5 min with blocking solution (0.02 M glycine, 0.1% Tween 20, 0.1% BSA in PBS) and then incubated for 1 h at room temperature in the same solution containing either anti-Cdk4, anti-CycD1, or anti-p21 Cip1 polyclonal antibodies (1:50 dilution) together with anti-CaM monoclonal antibody (1:500 dilution). After washing, grids were incubated 1 h at room temperature with the same solution but containing gold-coupled secondary antibodies (1:35, British Bio-Cell). Following that grids were sequentially washed with 0.1% BSA in PBS (10 min), PBS (5 min), and distilled water (30 min). Finally samples were contrasted with 2% acetate uranyl in water during 30 min (samples processed by conventional PLT) or 5 min (samples processed by slam freezing method).
Immunocytochemistry-For p21 Cip1 staining, cells grown in coverslips 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. Cells were then incubated for 1 h at 37°C in a humidified atmosphere, with the specific polyclonal antibody anti-p21 Cip1 (sc-397, Santa Cruz, 20 g/ml). Coverslips were then washed three times (5 min each) in PBS and incubated for 45 min at 37°C with fluoresceinconjugated anti-rabbit antibody (dilution 1:100, Sigma). After two washes in PBS, coverslips were mounted on glass slides with Mowiol (Calbiochem) and observed in a confocal microscope.

RESULTS
In Vitro Association of p21 Cip1 with CaM-We had shown previously that Cdk4 and CycD1 from cellular lysates could bind to CaM in a Ca 2ϩ -dependent manner. To determine whether one of the known proteins of the Cdk4-CycD1 complex was able to bind directly to CaM, we tested the binding of bacterially expressed and purified Cdk4-GST, CycD1-GST, p27 Kip1 -GST and p21 Cip1 -GST to CaM-Sepharose. Pull-down experiments were performed as indicated in the methods and demonstrate that p21 Cip1 -GST bound to CaM-Sepharose in the presence of Ca 2ϩ but not in its absence (Fig. 1). CycD1-GST and p27 Kip1 -GST, however, were found only in the fraction not bound to the column either in the presence or in the absence of Ca 2ϩ , and a small amount of Cdk4-GST was bound to the CaM-Sepharose in the presence of Ca 2ϩ . Preincubation of CaM-Sepharose with 2 nM (total volume 200 l) of the CaM-binding domain of the CaMKII (CaMKII 290 -309 fragment) inhibited the binding of p21 Cip1 -GST to CaM, indicating that p21 Cip1 -GST association with CaM was specific (Fig. 2B).
To determine which domains of p21 Cip1 could be involved in CaM interaction, three different peptides of the p21 Cip1 molecule ( Fig. 2A) were synthesized and assayed for competition in the p21 Cip1 binding to CaM-Sepharose. As shown in Fig. 2B, preincubation of CaM-Sepharose with 2 nmol of the p21 145-164 peptide (total volume 200 l) inhibited the binding of p21 Cip1 , while the other peptides had no effect. To further test that this peptide was the CaM-binding domain of p21 Cip1 , the peptide was covalently bound to Sepharose and pull-down experiments FIG. 1. p21 Cip1 binding to CaM-Sepharose. Bacterially expressed and purified GST-fused p21 Cip1 , p27 Kip1 , cyclin D1 (CycD1), or Cdk4 (1 g of each) were incubated with CaM-Sepharose (CaM-Seph) or Sepharose alone (Seph), in the presence of either Ca 2ϩ or EGTA. Unbound proteins were collected, and after extensive washing bound proteins were eluted with SDS containing buffer as indicated under "Experimental Procedures." Bound and not bound proteins were electrophoresed and stained with Coomassie Blue. p21 Cip1 Interaction with CaM with purified CaM performed in the presence or absence of free Ca 2ϩ . As shown in Fig. 2C, CaM was able to bind to p21 145-164-Sepharose, and the binding was Ca 2ϩ -dependent.
Coimmunoprecipitation of p21 Cip1 and CaM-The association between p21 Cip1 and CaM was also analyzed by immunoprecipitation (Fig. 3). When lysates from Namalva cells were immunoprecipitated with anti-CaM monoclonal antibody, p21 CIP1 was detected by Western blotting in the immunoprecipitates. Equally, CaM was detected in the immunoprecipitates using monoclonal anti-p21 Cip1 antibodies. Interestingly, no CaM was found to coimmunoprecipitate with p21 Cip1 when the antibodies used were the polyclonal anti-p21 Cip1 directed against the carboxyl terminus of the p21 Cip1 molecule (data not shown).
Colocalization of p21 Cip1 and CaM Analyzed by Electron Microscopy-To support the hypothesis that interaction of CaM with p21 Cip1 could have a physiological role, in vivo colocalization was analyzed by double immunogold labeling. Cryofixation of proliferative activated NRK cells was achieved by slam freezing and, as indicated under "Experimental Procedures," samples were incubated with anti-CaM monoclonal antibody and anti-p21 Cip1 polyclonal antibody and processed for electron microscopy. As shown in Fig. 4A, aggregates of CaM and p21 Cip1 were detected in the nucleus and in the cytoplasm. Colocalization of Cdk4 and cyclin D1 in the cell nucleus with CaM was also observed in samples processed by PLT method (Fig. 4B).
Anti-CaM Drug Addition Inhibits Nuclear Accumulation of p21 Cip1 -To analyze whether, as for cyclin D1 and Cdk4, CaM was essential for nuclear accumulation of p21 Cip1 , the anti-CaM drug W13 (15 g/ml), or the control drug W12 (15 g/ml), was added to NRK cells 5 h after proliferative activation. After 2 h of incubation with the drugs, p21 Cip1 intracellular distribution was analyzed by immunocytochemistry. As shown in Fig. 5, W13 addition caused a decrease in the nuclear staining for p21 Cip1 compared with the control drug, in parallel with an increase in the cytoplasmic staining. Similar results were obtained with other anti-CaM drugs such as J8 (7 g/ml), W7 (10 g/ml), or calmidazolium (5 M) (data not shown). DISCUSSION Although association of p21 Cip1 with Cdk4-cyclin D1 complexes has been described in a great variety of cells, its role in the regulation of Cdk4 activity is still controversial. While an inhibitory role was first proposed (38,39), recently it has been shown that this association is necessary for Cdk4-cyclin D1 assembly and its nuclear translocation (17,18). The diverse functionality of this protein may depend on the number of molecules bound to each Cdk4-cyclin D complex or to the association with other regulatory proteins. We show here that CaM binds to p21 Cip1 and so may be one of the molecules regulating its activity. Evidence for this is as follows: 1) purified p21 Cip1 is able to bind in vitro to CaM; 2) both molecules coimmunoprecipitate when extracted from cell lysates; and 3) colocalization of CaM and p21 Cip1 can be detected in vivo by electron microscopy immunogold analysis. Quiescent NRK cells were activated to proliferate, and 5 h after activation W12 or W13 (15 g/ml) was added to the medium. 2 h latter cells were fixed and p21 Cip1 immuodetection performed as indicated under "Experimental Procedures." p21 Cip1 Interaction with CaM On the contrary, although Cdk4 and cyclin D1 coimmunoprecipitate (14) and colocalize by electron microscopy with CaM, when purified, they do not bind (cyclin D1) or bind very little (Cdk4) to CaM. This indicates that their association with CaM depends on another protein. Interestingly, p27 Kip1 , which has some structural similarity to p21 Cip1 , is clearly not a CaMbinding protein. This points to a different regulation of p21 Cip1 and p27 Kip1 functions.
Interaction domains of p21 Cip1 with Cdks and cyclins reside in the amino-terminal half of the molecule (Cyc 21-26; Cdk 49 -71). These regions are precisely the ones well conserved between p21 Cip1 and p27 Kip1 . The second half of the molecule contains diverse nuclear localization signal (NLS) motif (139 -142, 160 -163, 140 -162, 141-158), the PCNA binding site (124 -164), and a cyclin A inhibitory binding domain (Fig. 1C) (40 -43). In order to analyze the possible binding domain of p21 Cip1 to CaM, three different peptides were synthesized and assayed for their ability to compete for the binding to CaM. Among the peptides analyzed, the one corresponding to amino acids 145-164 prevented the binding of p21 Cip1 to CaM. Furthermore CaM was able to bind to a p21 145-164 peptide-Sepharose column in a Ca 2ϩ -dependent manner. This peptide contains basic and hydrophobic residues that, plotted in a helical wheel diagram, show an amphipathic distribution. Thus, this region corresponds most probably to the CaM-binding domain of p21 Cip1 . This sequence is not present in the p27 Kip1 molecule, in agreement with the results presented which show that p27 Kip1 does not bind to CaM.
As mentioned above, apart from the inhibitory role of p21 Cip1 in Cdk-cyclin complexes, it has also been shown that p21 Cip1 is important for the assembly of Cdk4-cyclin D1 and for its translocation to the cell nucleus (17,18,44). La Baer et al. (17) showed that cells transfected with a truncated p21 Cip1 containing the amino-terminal half of the molecule were able to form complexes with Cdk4 and cyclin D1, but these were not found in the cell nucleus. This suggested that p21 Cip1 provides the nuclear localization signal for the nuclear import of the complex. Interestingly, in p21 Ϫ /p21 Ϫ cells Cdk4 and cyclin D1 were found in the nucleus, most probably because p27 Kip1 was then able to bind to Cdk4 and cyclin D1 and provide the NLS signal. In double p21 Cip1 and p27 Kip1 knockouts the number of cells with nuclear Cdk4 and cyclin D1 was greatly reduced (18). Our finding that functional inhibition of CaM during G 1 inhibited nuclear accumulation of p21 Cip1 opens the possibility that binding of CaM to p21 Cip1 exposes an NLS of p21 Cip1 , so allowing its nuclear translocation. Previous results showing that CaM inhibition was also blocking nuclear accumulation of Cdk4-cyclin D1 (14), lead us to suggest that CaM, through the binding to p21 Cip1 , activates the translocation of Cdk4-cyclin D1-p21 Cip1 complexes to the cell nucleus in response to a Ca 2ϩ signal. However, this possibility remains to be demonstrated.
We have shown previously that inhibition of HSP90, which is a CaM-binding protein that is associated with Cdk4, also inhibits nuclear translocation of Cdk4 and cyclin D1 (14). CaM may be essential in two steps leading to the nuclear translocation of Cdk4-cyclin D1 complexes: first, through the action of HSP90 allowing a proper Cdk4 folding (45,46) and binding to cyclin D1 and p21previously, and second, binding to p21 Cip1 , so allowing its import to the nucleus. Alternatively, the binding of CaM to p21 Cip1 may regulate another function of p21 Cip1 , which is not related to the nuclear import of the complexes. The fact that the CaM-binding domain of p21 Cip1 is located inside the PCNA-binding domain suggests that CaM may modulate the binding of p21 Cip1 to PCNA and thus its ability to inhibit DNA replication.
Although further studies are necessary to elucidate the role of CaM binding to p21 Cip1 , the new findings presented here, showing the in vitro interaction, the coimmunoprecipitation, and the in vivo colocalization of CaM and p21 Cip1 , open up the possibility of the regulation of p21 Cip1 functions in response to a Ca 2ϩ signal and in a different way from p27 Kip1 .