The Protein SET Regulates the Inhibitory Effect of p21Cip1 on Cyclin E-Cyclin-dependent Kinase 2 Activity*

The cyclin-dependent kinase (CDK) inhibitor p21Cip1 has a dual role in the regulation of the cell cycle; it is an activator of cyclin D1-CDK4 complexes and an inhibitor of cyclins E/A-CDK2 activity. By affinity chromatography with p21Cip1-Sepharose 4B columns, we purified a 39-kDa protein, which was identified by microsequence analysis as the oncoprotein SET. Complexes containing SET and p21Cip1 were detected in vivo by immunoprecipitation of Namalwa cell extracts using specific anti-p21Cip1 antibodies. We found that SET bound directly to p21Cip1 in vitro by the carboxyl-terminal region of p21Cip1. SET had no direct effect on cyclin E/A-CDK2 activity, although it reversed the inhibition of cyclin E-CDK2, but not of cyclin A-CDK2, induced by p21Cip1. This result is specific for p21Cip1, since SET neither bound to p27Kip1 nor reversed its inhibitory effect on cyclin E-CDK2 or cyclin A-CDK2. Thus, SET appears to be a modulator of p21Cip1 inhibitory function. These results suggest that SET can regulate G1/S transition by modulating the activity of cyclin E-CDK2.

The cyclin-dependent kinase (CDK) inhibitor p21 Cip1 has a dual role in the regulation of the cell cycle; it is an activator of cyclin D1-CDK4 complexes and an inhibitor of cyclins E/A-CDK2 activity. By affinity chromatography with p21 Cip1 -Sepharose 4B columns, we purified a 39-kDa protein, which was identified by microsequence analysis as the oncoprotein SET. Complexes containing SET and p21 Cip1 were detected in vivo by immunoprecipitation of Namalwa cell extracts using specific anti-p21 Cip1 antibodies. We found that SET bound directly to p21 Cip1 in vitro by the carboxyl-terminal region of p21 Cip1 . SET had no direct effect on cyclin E/A-CDK2 activity, although it reversed the inhibition of cyclin E-CDK2, but not of cyclin A-CDK2, induced by p21 Cip1 . This result is specific for p21 Cip1 , since SET neither bound to p27 Kip1 nor reversed its inhibitory effect on cyclin E-CDK2 or cyclin A-CDK2. Thus, SET appears to be a modulator of p21 Cip1 inhibitory function. These results suggest that SET can regulate G 1 /S transition by modulating the activity of cyclin E-CDK2.
The extracellular factors that regulate mammalian cell proliferation generate intracellular signals that ultimately converge on a family of serine-threonine kinases called cyclin-dependent kinases (CDKs) 1 (1)(2)(3). CDKs, composed of regulatory cyclin and catalytic CDK subunits, are activated in a periodic manner during the cell cycle. Thus, cyclin D-CDK4 complexes are activated at mid G 1 , cyclin E-CDK2 complexes are necessary for G 1 /S transition, cyclin A-CDK2 complexes are necessary for progression of DNA replication and cyclin B-CDK1 complexes are necessary for mitosis entry (4 -6).
The activity of CDKs is regulated by the synthesis and degradation of cyclins at specific stages of the cycle, phosphorylation of specific amino acid residues of the CDK subunit, and binding of inhibitors (CKIs) that associate with cyclin-CDK complexes (1,5,(7)(8)(9). Two families of CKIs have been described: INK4 and CIP/KIP (10). The INK4 proteins (p16 INK4a , p15 INK4b , p18 INK4c , and p19 INK4d ) bind specifically to CDK4 and to its homologue CDK6 (11)(12)(13). The CIP/KIP proteins (p21 Cip1 , p27 Kip1 , and p57 Kip2 ) bind to and inhibit the activity of a wide range of cyclin-CDK complexes (14 -16). It is generally assumed that most of the biological activities of these CKIs depend on their ability to inhibit cyclin-CDKs. However, p21 Cip1 also associates with and inhibits the proliferating cell nuclear antigen (PCNA), which could be important to block DNA synthesis following DNA damage (17).
The levels of p21 Cip1 in quiescent cells are very low, but they are up-regulated in proliferating cells and also by antimitogenic stimuli (18,19). p21 Cip1 is also induced in some cell types during senescence and terminal differentiation (20). It is transcriptionally induced by the tumor suppressor protein p53 and plays an important role in cell cycle arrest induced by the activation of the G 1 DNA damage checkpoint (21)(22)(23).
Recent evidence suggests that in addition to their role as cyclin-CDK inhibitors, p21 Cip1 and p27 Kip1 may also be activators of cyclin D-dependent kinases. p21 Cip1 could promote the assembly of active cyclin D1-CDK4 complexes and provide a localization signal for their nuclear import (24). Primary mouse embryonic fibroblasts that lack genes encoding p21 Cip1 and p27 Kip1 fail to assemble detectable amounts of cyclin D-CDK complexes, express cyclin D proteins at much lower levels, and are unable to efficiently direct cyclin D proteins to the cell nucleus (25). Restoration of CKI function reverses all three defects and thereby restores cyclin D activity to normal physiological levels.
The dual role of p21 Cip1 during the cell cycle suggests that its activity as inhibitor or activator must be highly regulated during the cell cycle and that other, still unknown, proteins might be involved in the modulation of p21 Cip1 activity. Thus, we are searching new p21 Cip1 -binding proteins by using affinity chromatography with p21 Cip1 -Sepharose 4B columns. We report here that the oncogenic protein SET binds directly to p21 Cip1 and that it reverses the inhibitory effect of p21 Cip1 on cyclin E-CDK2 activity but not on cyclin A-CDK2.

EXPERIMENTAL PROCEDURES
Cell Cultures-Lymphoblastoid cell lines Molt-4 and Namalwa were obtained from the American Type Culture Collection. They were grown at 1 ϫ 10 6 cells/ml in RPMI 1640 (Flow Laboratories) supplemented with 10% fetal calf serum and 50 g/ml of gentamicin as described previously (26).
Expression and Purification of Recombinant Proteins-The p21 Cip1 cDNA was obtained by polymerase chain reaction from a human cDNA library, sequenced, and introduced into the pGEX-KG vector at NdeI-HindIII sites. SET cDNA was a generous gift from Dr. Damuni (New York). SET cDNA was introduced into the pGEX-KG vector at NcoI-HindIII sites. p27 Kip1 cDNA was a generous gift from Dr. Massagué (New York). p27 Kip1 cDNA was introduced into the pGEX-KG vector at NdeI-XhoI sites. To obtain recombinant glutathione S-transferase (GST), GST-p21 Cip1 , and GST-SET proteins, the BL21 LysE strain of E. coli was transformed with the vectors pGEX-KG, pGEX-KG-p21 Cip1 , and pGEX-KG-SET. A single colony was grown to saturation and activated by 0.5 mM isopropyl-␤-D-thiogalactopyranoside for 16 h. To purify these fusion proteins, bacteria expressing these proteins were resuspended in NENT buffer (20 mM Tris-HCl, pH 8.0, 100 mM NaCl, 1 mM EDTA, and 0.5% Nonidet P-40). The samples were then sonicated for 10 s three times, centrifuged for 10 min at 27,000 ϫ g, and finally resuspended in NENT buffer with 0.5% Sarcosyl instead of Nonidet P-40. Supernatants were mixed and applied to glutathione-Sepharose beads. After three washes, the proteins were eluted with 20 mM reduced glutathione in 50 mM Tris-HCl, pH 9.0, and 120 mM NaCl.
p21 Cip1 -Sepharose Affinity Chromatography-The p21 Cip1 -Sepharose 4B columns were prepared by coupling 5-7 mg of purified GST-p21 Cip1 protein to 3 ml of CNBr-activated Sepharose 4B (Amersham Pharmacia Biotech). Control columns of purified GST were also constructed in the same way. Molt-4 cells were lysed in buffer A (50 mM Hepes-KOH, pH 7.6, 50 mM KCl 1 mM MgCl 2 , 1 mM EGTA, 1 mM dithiothreitol, 1 mM phenylmethylsulfonyl fluoride, and 1 g/ml aprotinin). The extracts (10 mg) were then loaded on the GST-p21 Cip1 -Sepharose or GST-Sepharose columns, and after washing in 50 vol of buffer A the bound proteins were eluted with the same buffer but containing 1 M KCl instead of 50 mM KCl.
Peptide Sequence Analysis-To obtain an amino acid sequence of the 39-kDa protein eluted from GST-p21 Cip1 affinity column, the eluate was first separated in SDS-polyacrylamide gels. The band corresponding to the 39-kDa protein was then sliced from the gel and digested with trypsin. After digestion, the peptides were separated by high pressure liquid chromatography, and a peptide was sequenced with an automatic sequencer using protocols, reagents, chemicals, and materials from Applied Biosystems (Warrington, UK).
For kinase assays, IPs were performed as described in Ref. 28. Molt-4 cells (1 ϫ 10 8 ) were lysed in IP buffer. Lysates (1-2 mg) were immunoprecipitated with anti-cyclin E (06-459; Upstate Biotechnology) or anticyclin A (06-138; Upstate Biotechnology) antibodies. The immunoprecipitated complexes were washed in kinase buffer (25 mM HEPES-Na, pH 7.4, 5 mM MgCl 2 , and 0.5 mM dithiothreitol) and then incubated in kinase buffer containing 1 mM ATP, 3 mg of histone H1, and 10 Ci of [ 32 P]ATP for 30 min at 30°C in a final volume of 50 l. The samples were then electrophoresed, and the gels were stained with Coomassie Blue, dried, and exposed to x-ray films at Ϫ80°C. In some kinase experiments, different amounts of GST-SET, GST-p21 Cip1 , GST-p27 Kip1 , or GST were added to the incubation media as indicated.
Binding Experiments-To analyze the binding of SET to p21 Cip1 , 2.5 M GST-SET or GST in kinase buffer was loaded into a GST-p21 Cip1 -Sepharose column. After washing in 50 volumes of the same buffer, the protein bound to the column was eluted with the same buffer containing 6 M urea. The eluted protein was precipitated with 10% thrichloroacetic acid and then electrophoresed. The protein was visualized with Coomassie Blue staining.
Pull-down experiments were performed using three peptides corresponding to different regions of the p21 Cip1 molecule. The first peptide corresponded to the amino acids 17-30 (p21 [17][18][19][20][21][22][23][24][25][26][27][28][29][30], which contain the cyclin-binding domain of p21 Cip1 . The second peptide included the amino acids 28 -49 (p21 28 -49 ), a region containing an ␣-helix domain located in the middle of the bridge between the cyclin-and the CDKbinding domains. The third peptide corresponding to a region of the PCNA-binding domain comprises amino acids 145-164 (p21 145-164 ). Each of these peptides was covalently bound to a Sepharose 4B matrix. GST-SET or GST (2.5 g) in buffer B (50 mM Tris-HCl, pH 7.4, 150 mM NaCl, and 1% Triton X-100) were applied to the three different columns. After a 1-h incubation at room temperature, samples were centrifuged. The unbound fraction was collected, whereas the bound proteins were washed three times in buffer B and then solubilized with sample buffer for subsequent electrophoresis and Western blotting.
To analyze the effect of SET protein on the binding of p21 Cip1 to cyclin E-CDK2 complexes, Molt-4 cells were lysed in IP buffer and then immunoprecipitated with anti-cyclin E antibodies as described above. The immunoprecipitated complexes were resuspended in kinase buffer and incubated for 1 h at 4°C in the presence of GST-p21 Cip1 , GST-SET, or GST-p21 Cip1 and GST-SET (400 nM each). The complexes were then washed three times in kinase buffer, electrophoresed in SDS-PAGE gels, and subsequently subjected to Western blotting using specific antibodies against p21 Cip1 or SET.

RESULTS
Purification of New p21 Cip1 -binding Proteins-To identify new p21 Cip1 -binding proteins, we performed affinity chromatography analysis using GST-p21 Cip1 -Sepharose 4B columns. For these experiments, we selected the lymphoblastoid cell line Molt-4, because it did not express p21 Cip1 , and thus we avoid the association of the p21 Cip1 -binding proteins of the cell extract with the endogenous p21 Cip1 . The columns were loaded with Molt-4 cell lysates and subsequently washed with loading buffer. The proteins associated with GST-p21 Cip1 were then eluted with 1 M KCl and subjected to SDS-PAGE. The eluates contained three major proteins of 39, 70, and 80 kDa and several minor proteins of a variety of sizes (Fig. 1A). These proteins were specifically bound to p21 Cip1 and not to GST, since control GST-Sepharose 4B columns did not retain them (Fig. 1A). Western blot analysis revealed that the eluates contained PCNA, cyclin A, and CDK2, indicating that the GST-p21 Cip1 -Sepharose column worked properly, since it retained known p21 Cip1 -binding proteins (Fig. 1B).
Identification of the 39-kDa p21 Cip1 -binding Protein as the Oncoprotein SET-To identify the 39-kDa p21 Cip1 -binding protein, we prepared high amounts of eluate, which were subjected to SDS-PAGE. The 39-kDa protein was sliced from the gels and then microsequenced. A sequence of nine amino acids was obtained from this protein. This sequence matched perfectly (100% identity) with the amino acids 78 -86 of the SET protein (Fig. 2). SET is a protein that has been found fused to the protein CAN in acute nonlymphocytic myeloid leukemias (29).
SET Forms Complexes with p21 Cip1 in Vivo and Binds Directly to p21 Cip1 in Vitro-To study whether SET and p21 Cip1 form complexes in vivo, Namalwa cell extracts were subjected to IP using an anti-p21 Cip1 antibody. Results showed that this antibody immunoprecipitated p21 Cip1 but also the SET protein, indicating the presence of SET-p21 Cip1 complexes in vivo (Fig. 3A).
To analyze whether SET and p21 Cip1 interact directly or through other proteins, we expressed both SET and p21 Cip1 as GST fusion proteins and performed binding analysis. GST-SET or GST was incubated in the presence of GST-p21 Cip1 -Sepharose for 1 h at 4°C as described under "Experimental Procedures." Then, after extensive washing, the protein associated with GST-p21 Cip1 was eluted with 6 M urea. As shown in Fig.  3B, GST-SET but not GST associated directly with p21 Cip1 . To determine which domain of p21 Cip1 was involved in SET interaction, pull-down experiments using three peptides corresponding to different regions of the p21 Cip1 molecule (p21 17-30 ), (p21 28 -49 ) and (p21 145-164 ) were performed as described under "Experimental Procedures." Results indicate that SET bound to the p21 145-164 peptide but not to the other two p21 Cip1 peptides (Fig. 4).
Role of SET on p21 Cip1 Activity-The direct interaction of SET with p21 Cip1 suggested that SET could modulate the function of p21 Cip1 . Since p21 Cip1 is a strong inhibitor of the activity of cyclin-CDK2 complexes, we analyzed whether SET can modulate the inhibitory effect of p21 Cip1 on cyclin E-CDK2 or cyclin A-CDK2 activities. Thus, we immunoprecipitated Molt-4 cell extracts using specific anti-cyclin E or anti-cyclin A antibodies and measured CDK2 activity in the immunoprecipitates. In both cases, CDK2 activity was inhibited by the addition of 400 nM of recombinant GST-p21 Cip1 (Fig. 5). We also observed that SET had no effect on cyclin E-or cyclin A-associated CDK2 activities at concentrations of 400 or 800 nM. However, when SET was added together with p21 Cip1 it reversed the inhibitory effect of p21 Cip1 on cyclin E-CDK2 activity but not on cyclin A-CDK2 (Fig. 5). This effect was specific, since purified GST did not reverse the p21 Cip1 -induced CDK2 inhibition.
We also analyzed the effect of SET on the inhibition of cyclin E-CDK2 or cyclin A-CDK2 by p27 Kip1 . Thus, IPs obtained using anti-cyclin E or anti-cyclin A antibodies were assayed for CDK2 activity in the presence of p27 Kip1 or p27 Kip1 plus SET. CDK2 activity was inhibited by the addition of 400 nM GST-p27 Kip1 ,

FIG. 3. In vivo and in vitro interaction of SET and p21 Cip1 . A,
Namalwa cell extracts were immunoprecipitated with a monoclonal anti-p21 Cip1 antibody or with a nonrelated monoclonal antibody, which was used as a control (C). The immunoprecipitates were then subjected to Western blotting (WB) using anti-SET antibodies. A Western blot of the total extract using anti-SET antibodies was also performed. B, purified GST-SET or GST was loaded on a GST-p21 Cip1 -Sepharose column, and after extensive washing the protein bound to the columns was eluted. The eluates were then subjected to SDS-PAGE and visualized with Coomassie Blue staining.

FIG. 4. Identification of the SET-binding domain of p21 Cip1 .
Three different peptides corresponding to different regions of the p21 Cip1 molecule (p21 17-30 , p21 28 -49 , and p21 145-164 ) were covalently bound to a Sepharose 4B matrix. Pull-down experiments with purified GST-SET (A) or GST (B) were then performed as described under "Experimental Procedures." The bound (B) and not bound (N) fractions were electrophoresed and subsequently subjected to Western blotting using specific anti-SET or anti-GST antibodies. but the activity was not reversed by the addition of the same amount of SET (Fig. 6). Thus, SET specifically modulates the inhibitory function of p21 Cip1 .
SET Binds to Cyclin E-CDK2 Complexes in the Presence and in the Absence of p21 Cip1 -We also analyzed whether SET associates with cyclin E-CDK2 complexes and if this binding depends on the presence of p21 Cip1 . Thus, we performed IP experiments on Molt-4 cell extracts using anti-cyclin E antibodies. The immunoprecipitates were then incubated with p21 Cip1 alone, SET alone, or with both p21 Cip1 and SET. Results showed that p21 Cip1 associated with cyclin E-CDK2 complexes either in the presence or in the absence of SET (Fig. 7). These results indicate that proteins of these complexes other than p21 Cip1 are able to bind SET.

DISCUSSION
The CDK inhibitor p21 Cip1 negatively regulates cell cycle progression and enforce a cell cycle arrest when overexpressed in the cells (15,30). These effects are due to the binding of p21 Cip1 to cyclin-CDK complexes, which leads to the inhibition of CDK activity and to the association of p21 Cip1 with PCNA, which blocks DNA synthesis (31). It has been also shown that p21 Cip1 and p27 Kip1 are necessary for certain processes that positively regulate cell cycle progression: cyclin D assembly with CDK4, its stability, and its nuclear localization (24,25). More recently, it has been shown that calmodulin binds to p21 Cip1 in a Ca 2ϩ -dependent manner and that it regulates its nuclear localization but also that of cyclin D1 and CDK4 (32,33). The p21 Cip1 functions are performed by different domains of the protein. The NH 2 -terminal domain contains two regions, which specifically bind to cyclins and CDKs. The carboxylterminal domain has a sequence that interacts with PCNA and calmodulin and contains an NLS region and possibly a second cyclin A-binding sequence (32,34,35).
Thus, it appears that p21 Cip1 has a dual role in the regulation of cell cycle progression as activators of CDK4 and as inhibitors of CDK2 activity. To perform these different functions, the activity of p21 Cip1 must be highly regulated during the cell cycle. One possible mechanism of regulation may be through additional components associated with cyclin-CDK-CKI complexes. Consistent with this idea is the observation of high molecular weight cyclin-CDK-CKI complexes in different cellular types (36). Although some of the proteins associated with these complexes are known, some others still remain unidentified (37,38).
We report here that the oncoprotein SET associates with p21 Cip1 in vivo and in vitro. Results also indicate that SET modulates the p21 Cip1 inhibitory effect on cyclin E-CDK2 but not on cyclin A-CDK2 in vitro. SET was first identified as a gene that was fused to the CAN gene in a patient with acute undifferentiated leukemia, apparently as a result of a translocation (29). The SET-CAN fusion gene generates a transcript encoding a chimeric SET-CAN protein of 155 kDa. The SET protein shows similarity with the yeast nucleosome assembly protein NAP-I and is widely expressed in human and mouse tissues (29,39,40). It is phosphorylated in serine residues in vivo and it is mainly located in the nucleus.
Not much is known about the cellular role of SET, although several possible functions have been proposed. SET interacts specifically with B-type cyclins but not with cyclin A, although the functional significance of this interaction has not been elucidated (41). Moreover, SET has been found to be identical to template-activating factor I, a host protein necessary for DNA replication of the adenovirus genome (42). A long acidic domain in the C-terminal region of SET is essential for template-activating factor I activity (42). SET has been also identified as a potent inhibitor of the protein phosphatase PP2A (43). Since PP2A has been implicated in the regulation of cell cycle progression (44), it has been suggested that the SET-CAN gene fusion, which occurs in acute nonlymphocytic myeloid leukemia, may lead to altered regulation of PP2A activity and thus contribute to leukemogenesis (43).
The known functions of SET clearly relate it with the control of cell cycle progression, although the steps regulated by SET remain unclear. Our results support the hypothesis that SET is a cell cycle regulator because of its ability to modulate p21 Cip1 function, specifically on cyclin E-CDK2 activity. It is assumed that cyclin E-CDK2 activity is essential for the triggering of DNA replication (1). Thus, our results suggest that SET could be involved in the control of G 1 /S transition. Our results can be related with the reported effect of template-activating factor I (SET) on DNA replication in the adenovirus. The stimulatory effect of template-activating factor I on DNA synthesis could be due to the inactivation of p21 Cip1 inhibitory function on cyclin E-CDK2 activity needed to trigger DNA synthesis. However, this possibility remains to be demonstrated.
The mechanism by which SET regulates cyclin E-CDK2 activity has not been elucidated. We showed that SET did not block the binding of p21 Cip1 to cyclin E-CDK2 complexes. Moreover, we observed that SET was able to bind to cyclin E-CDK2 complexes in the absence of p21 Cip1 . This binding did not produce any alteration of cyclin E-CDK2 activity; thus, the SETinduced reversion of the inhibition of CDK2 activity by p21 Cip1 cannot be due to an increase in CDK2 activity induced by SET. The most likely explanation for the effect of SET is that its association with the cyclin E-CDK2-p21 Cip1 complexes induces a conformational change in any of these proteins that overrides the inhibition of CDK2 activity by p21 Cip1 . We have not identified the protein of the complex that binds SET in the absence of p21 Cip1 , but work in that direction is under way in our laboratory.
We found that the SET-binding domain of p21 Cip1 is located at the carboxyl-terminal region. This fact is consistent with the evidence we report here showing that the binding of SET to p21 Cip1 did not block the association of p21 Cip1 with cyclin E-CDK2 complexes, since this association is mediated by p21 Cip1 regions located at the NH 2 terminus. Interestingly, the carboxyl-terminal domain contains a region, which binds PCNA and calmodulin (32,35). Thus, although we still do not know the functional significance of this finding it suggest that SET could also modulate the action of p21 Cip1 on PCNA and the role of calmodulin on p21 Cip1 -cyclin D-CDK4 complexes.
Although the functional relevance of the results reported here in the cell cycle progression in vivo are unknown, we can hypothesize that SET can regulate cell cycle progression at the G 1 /S transition. We can also speculate that the CAN-SET chimeric protein produced in acute nonlymphocytic myeloid leukemia may impair the normal regulation of p21 Cip1 by wild type SET and that this process could also contribute to leukemogenesis.