Cdt1 phosphorylation by cyclin A-dependent kinases negatively regulates its function without affecting geminin binding.

The current concept regarding cell cycle regulation of DNA replication is that Cdt1, together with origin recognition complex and CDC6 proteins, constitutes the machinery that loads the minichromosome maintenance complex, a candidate replicative helicase, onto chromatin during the G(1) phase. The actions of origin recognition complex and CDC6 are suppressed through phosphorylation by cyclin-dependent kinases (Cdks) after S phase to prohibit rereplication. It has been suggested in metazoan cells that the function of Cdt1 is blocked through binding to an inhibitor protein, geminin. However, the functional relationship between the Cdt1-geminin system and Cdks remains to be clarified. In this report, we demonstrate that human Cdt1 is phosphorylated by cyclin A-dependent kinases dependent on its cyclin-binding motif. Cdk phosphorylation resulted in the binding of Cdt1 to the F-box protein Skp2 and subsequent degradation. In contrast, in vitro DNA binding activity of Cdt1 was inhibited by the phosphorylation. However, geminin binding to Cdt1 was not affected by the phosphorylation. Finally we provide evidence that inactivation of Cdk1 results in Cdt1 dephosphorylation and rebinding to chromatin in murine FT210 cells synchronized around the G(2)/M phase. Taken together, these findings suggest that Cdt1 function is also negatively regulated by the Cdk phosphorylation independent of geminin binding.

tion as a loader for the MCM heterohexameric complex, which could function as a replicative helicase (3,4). The mechanism seems basically conserved from yeast to metazoan cells, although it is more complicated in the latter (2).
Once cyclin-dependent kinases (Cdks) become active at the onset of S phase, pre-RC initiates replication accompanied by further assembly of multiple other proteins or protein complexes (1). Cdks physically interact with and phosphorylate ORC and CDC6 (5)(6)(7)(8)(9)(10). However, considering that they are no longer necessary for initiation after loading MCMs (11,12), such phosphorylation is unlikely to be involved in promotion of replication, although ORC and CDC6 could act as adapter molecules to tether Cdks to initiation sites. Prior to the DNA unwinding step, CDC45 is loaded onto chromatin, probably via physical interaction with the MCM complex (13)(14)(15). The loading is dependent on both Cdk and CDC7 kinase activities, and both these kinases phosphorylate the MCM complex (1,(13)(14)(15). However, it remains unclear how phosphorylation positively regulates this step. Once CDC45 is loaded, the DNA is unwound with the help of replication protein A (1).
To prevent rereplication, that is the reestablishment of pre-RC, rebinding of MCM needs to be suppressed during the S, G 2 , and M phases of the cell cycle. Recent studies have suggested that Cdks also play a central role in this context. Thus, Cdk activity has a bipartite function in the regulation of DNA replication. The importance of Cdk1 kinase in mammals is clearly demonstrated by the fact that its inactivation results in rebinding of MCM proteins and subsequent rereplication (16,17). Ablation of cyclin A, but not cyclin B, leads to rereplication in Drosophila tissue culture cells (18). Cdks prevent reestablishment of pre-RC through multiple redundant mechanisms (1,2). One mechanism is by phosphorylation of CDC6, leading to degradation in yeast or nuclear export in mammalian cells (5)(6)(7)19). In the latter situation, cyclin A-Cdk2 is one kinase responsible for this phosphorylation (6,7). In human cells, ORC1 is degraded after S phase, presumably depending on phosphorylation by cyclin A-Cdk2 (10,20). In budding yeast, the function of ORC2 is suppressed through Cdk phosphorylation (9). It has also been shown that the MCM complex is phosphorylated by Cdks (1,2,17). In budding yeast, it is necessary to block all three pathways for induction of rereplication without inhibiting Cdk activity (9). Also in mammalian cells, alteration of the regulation of the individual components alone, for example overexpression of the phosphorylation-deficient CDC6 mutant, fails to induce rereplication (6,7).
In metazoans, geminin has been identified as another inhibitor of pre-RC formation (21), preventing the loading of MCM proteins by binding to and inhibiting Cdt1 (22)(23)(24)(25). It appears after cells enter S phase and is destroyed during exit from mitosis (21,24). Contrary to the cyclin A case, reduction in the levels of geminin in Drosophila tissue culture cells results in partial, but not complete, rereplication (18). Therefore, Cdt1 actions might be negatively regulated by both geminin and Cdks. Alternatively interaction between Cdt1 and geminin could be under the control of Cdks. In certain cell types, Cdt1 is degraded on entry into the S phase (26), which could be directed by Cdk phosphorylation. However, the functional relationship between the Cdt1-geminin system and Cdks remains to be clarified.
In this report, we demonstrate that Cdt1 is phosphorylated by cyclin A-dependent kinases (cyclin A-Cdk1 and cyclin A-Cdk2) dependent on its cyclin-binding motif. Cdk phosphorylation resulted in Cdt1 binding to the F-box protein Skp2 (27), a component of the Skp1-Cdc53-F-box protein (SCF) ubiquitin ligase complex, and subsequent degradation. In contrast, in vitro DNA binding activity of Cdt1 was reduced by the phosphorylation. However, geminin binding to Cdt1 was not affected by the phosphorylation. Finally we show that Cdk1 inactivation results in Cdt1 dephosphorylation and rebinding to chromatin in murine FT210 cells. These findings suggest that Cdt1 function is negatively regulated by Cdk phosphorylation independent of geminin binding.

Construction of Expression Vectors for Human Cdt1 and Geminin
and Murine Skp2-Cloning of human Cdt1 and geminin cDNAs and construction of mammalian expression vectors (based on pcDEB⌬ or pcDEB⌬-T7) were described previously (26). To generate a mutant Cdt1 (Cdt1 Cy) in which Arg-68, Arg-69, and Leu-70 were all converted into Ala, site-directed mutagenesis was performed with oligonucleotide CA-GGCCACCGGCCCGCGCTGCAGCGCGGCTGTCGGTGGACGAG using Mutan-Super Express Km kit (Takara), and the product of the resultant cDNA was confirmed by sequencing. For bacterial expression of glutathione S-transferase (GST)-Cdt1, the Cdt1 cDNA was introduced into pGEX-6P-1 (Amersham Biosciences), and for in vitro production of His-geminin with bacterial lysate, the geminin cDNA was inserted into pIVEX2.4b Nde (Roche Applied Science). Plasmid pcDNA3-HA-Skp2 (27), in which hemagglutinin (HA)-tagged murine Skp2 cDNA is transcribed from the T7 promoter, was kindly provided by Dr. K. Nakayama (Kyushu University).
Preparation of Cyclin-Cdks Using Recombinant Baculoviruses-GST-murine cyclin A and GST-murine cyclin B baculoviruses were provided by Dr. H. Masai (Tokyo Metropolitan Institute of Medical Science), and human Cdk1 and Cdk2 baculoviruses were purchased from Orbigen (San Diego, CA). At 48 h after infection (multiplicity of infection of 2 for GST-cyclins and 5 for Cdks) of 2 ϫ 10 8 High Five cells with recombinant baculoviruses, extracts were prepared with 5 ml of lysis buffer (50 mM Tris-HCl, pH 7.4, 100 mM NaCl, 0.1% Nonidet P-40, 10% glycerol, 1 mM dithiothreitol (DTT)) containing multiple protease inhibitors (Sigma). The lysates were clarified by centrifugation and filtration, and GST-cyclin-Cdk complexes were purified with 1.5 ml of glutathione-Sepharose beads (Amersham Biosciences). The beads were then resuspended in thrombin cleavage buffer (50 mM Tris-HCl, pH 7.4, 150 mM NaCl, 2.5 mM CaCl 2 , 10% glycerol) and digested with 20 units of thrombin (Amersham Biosciences) for 45 min at room temperature to allow the cyclin-Cdks to be eluted. The purity of the cyclin-Cdk complexes was confirmed by SDS-PAGE followed by silver staining and immunoblotting. The approximate concentrations of the obtained cyclin-Cdks were estimated to be as follows: 1 ng/l for cyclin A-Cdk1 and cyclin B-Cdk1 and 10 ng/l for cyclin A-Cdk2.
Production of Recombinant GST-Cdt1, His-geminin, and HA-Skp2 Proteins-GST-Cdt1 and a GST-Cdt1 Cy mutant were produced in Escherichia coli strain BL-21, purified on glutathione-Sepharose beads, eluted with 20 mM glutathione in elution buffer A (200 mM NaCl, 20 mM Tris-HCl, pH 7.4, 1 mM DTT, 0.05% Triton X-100), and dialyzed. Histagged Geminin proteins were synthesized using an in vitro transcription-translation system with bacterial extracts (RTS 500, Roche Applied Science) according to the manufacturer's instructions, purified on nickel affinity gel (Sigma), eluted with elution buffer B (50 mM sodium phosphate, pH 8.0, 250 mM NaCl) containing 250 mM imidazole, and dialyzed. HA-tagged Skp2 proteins were synthesized by in vitro transcription-translation with rabbit reticulocyte lysate (TNT T7 quick coupled transcription/translation system, Promega) according to the manufacturer's instructions.
In Vitro Binding Assays-For assaying the association of Cdt1 with cyclin-Cdks, a 3-g aliquot of GST-Cdt1 was incubated with 10 -50 ng of cyclin-Cdks in binding buffer A (200 mM NaCl, 20 mM Tris-HCl, pH 7.4, 1 mM DTT, 0.05% Triton X-100, 5% glycerol) containing 1 M calyculin A, 10 mM glycerophosphate, and 1 mM phenylmethylsulfonyl fluoride at 4°C for 60 min. Then the GST fusion proteins were collected on glutathione beads, and the beads were washed three times with binding buffer A. The bound proteins were eluted with 25 l of 1ϫ sample buffer and analyzed by immunoblotting.
Cdt1-geminin interaction was assessed as follows. GST-Cdt1 (0.6 g) and His-geminin (0.2 g) were subjected to phosphorylation reaction as described above except that radiolabeled ATP was omitted. The mixtures were diluted with binding buffer A, and binding reactions were carried out at 4°C for 60 min. Then the GST-Cdt1 and associated His-geminin proteins were analyzed as described above.
Cdt1-Skp2 interaction was assessed as follows. GST-Cdt1 (0.6 g) was subjected to phosphorylation reactions as described above with or without the cyclin-Cdks, mixed with the HA-Skp2 produced by in vitro translation in 50 l of reaction mixture, diluted with binding buffer B (100 mM NaCl, 20 mM Tris-HCl, pH 7.4, 0.1% Triton X-100) containing 10 mM glycerophosphate and 5 mM NaF, and incubated at 4°C for 60 min. Then the GST-Cdt1 and associated Skp2 proteins were collected, washed with binding buffer A, and analyzed.
DNA binding activity of Cdt1 was examined as follows. GST-Cdt1 (0.6 g) was subjected to reaction with or without the cyclin-Cdks as above, diluted with binding buffer B, and incubated with 30 l of double-stranded DNA cellulose beads (Amersham Biosciences) at 4°C for 90 min. Then the beads were washed three times with binding buffer A, and the associated proteins were analyzed by immunoblotting.
Transfection, Immunoprecipitation, and Immunoblotting-293T cells were grown in Dulbecco's modified Eagle's medium with 8% fetal calf serum. The indicated plasmids (6 g) were transiently transfected into 4 ϫ 10 6 cells in 100-mm dishes with TransIT-293 reagent (Mirus, Madison, WI) according to the manufacturer's instructions. When indicated, cell were synchronized in G 2 /M phase with 50 ng/ml nocodazole treatment for 14 -18 h.
After 48 h, cells were lysed on ice in 1 ml of Nonidet P-40 buffer (150 mM NaCl, 1% Nonidet P-40, 10 mM Tris-HCl, pH 7.4) containing multiple protease inhibitors, and the soluble fraction was separated by centrifugation. Aliquots of the lysates were immunoprecipitated with anti-T7 antibody-or control IgG-fixed beads (Novagen), and the beads were washed four times with 1 ml of NET gel buffer (50 mM Tris-HCl, pH 7.4, 150 mM NaCl, 0.1% Triton X-100, 1 mM EDTA). The immunoprecipitates were eluted with the elution buffer (100 mM glycine-HCl, pH 2.5) and subjected to immunoblotting.
Immunoblotting was performed as described previously (20). Antibody binding was visualized using the ECL system (Amersham Biosciences).
Establishment of Rat-1 Cells Stably Expressing T7-tagged Cdt1-Rat-1 cells (normal rat fibroblasts) were grown in Dulbecco's modified Eagle's medium with 8% fetal calf serum. Recombinant retroviruses expressing wild type T7-tagged human Cdt1 or T7-Cdt1 Cy were prepared, and cells were infected with the viruses. Cells were then selected with hygromycin B and cloned. Representative lines expressing wild type T7-Cdt1 (designated WB4) or T7-Cdt1 Cy (designated CB3) were established and used for further study. The levels of T7-Cdt1 in these cells were about 20 times those of endogenous Cdt1. Details of establishment and characterization of these cells will be published elsewhere. 2 When indicated, cells were synchronized in S phase with 2.5 mM hydroxyurea treatment for 18 h.
-Phosphatase Treatment of Cdt1-Geminin Complexes Immunoprecipitated from 293T Cells-Cdt1-geminin complexes were immunoprecipitated with anti-T7 antibody from 293T cells transfected with Cdt1 and T7-geminin and synchronized in G 2 /M phase with nocodazole. After washing, the purified immunoprecipitates were treated with 100 units of -phosphatase (New England Biolabs) in 50 l of phosphatase buffer (50 mM Tris-HCl, pH 7.8, 5 mM DTT, 2 mM MnCl 2 , 1 mM phenylmethylsulfonyl fluoride) for 30 min at 30°C or left untreated in the same buffer. Then proteins released into supernatants or remaining on the beads were separated by centrifugation and subjected to immunoblotting.

RESULTS
Cyclin A-dependent Kinases, but Not Cyclin B-Cdk1, Bind to Human Cdt1 through the Cyclin-binding Motif-To explore functional associations between Cdt1 and Cdks, we first assessed physical interactions in vivo by immunoprecipitation with anti-T7 tag antibody after transfection of T7-tagged Cdt1 into 293T cells (Fig. 1). Cyclin A, Cdk1, and Cdk2, but not cyclin B, proteins were specifically co-precipitated with T7-Cdt1. Cdt1 has a putative cyclin-binding motif (7,8,30,31) in the N terminus. To test its role, we created a T7-Cdt1 mutant in which the conserved Arg-68, Arg-69, and Leu-70 (Cy motif) had been substituted with alanines (T7-Cdt1 Cy). When T7-Cdt1 Cy was introduced into 293T cells and analyzed by immunoprecipitation as above, no or reduced co-precipitation of cyclin A, Cdk1, and Cdk2 was evident (Fig. 1). These data suggest that Cdt1 may interact with cyclin A-Cdk1 and cyclin A-Cdk2 kinases through the Cy motif. We sought to co-precipitate cyclin A-Cdks with endogenous Cdt1 but failed to do so (data not shown). Therefore, physical interaction between Cdt1 and cyclin A-Cdks may not be very strong. However, as described below, further studies clearly demonstrate that this interaction plays a crucial role in phosphorylation and regulation of Cdt1 by cyclin A-dependent kinases.
We next set up an in vitro assay system. Bacterially produced purified wild type GST-Cdt1 and GST-Cdt1 Cy fusion proteins were incubated with cyclin A-Cdk1, cyclin A-Cdk2, and cyclin B-Cdk1 complexes generated in the baculoviral system and purified. Then the GST fusion proteins were collected on glutathione beads, and associated cyclin-Cdks were analyzed by immunoblotting. As shown in Fig. 2A, cyclin A-Cdk1 and cyclin A-Cdk2, but not cyclin B-Cdk1, efficiently bound to GST-Cdt1, whereas no specific binding was noted to GST-Cdt1 Cy or to GST alone (data not shown) under the same conditions. Lower bands observed both in cyclin A and cyclin B samples may be truncated forms nonspecifically cleaved by thrombin, but the kinase complexes have sufficient activities as mentioned below. Taken together, these data clearly demonstrate that cyclin A-Cdk1 and cyclin A-Cdk2 kinases directly bind to Cdt1 through the Cy motif.
Cyclin A-dependent Kinases Phosphorylate Cdt1 Depending on Their Binding to the Cyclin-binding Motif-The obtained data suggested that Cdt1 could be a good substrate for cyclin A-dependent kinases. Therefore, we performed an in vitro phosphorylation assay with [␥-33 P]ATP. When histone H1 proteins were used as the substrate, all three cyclin-Cdks demonstrated similar phosphorylation under our assay conditions (Fig. 2B). However, the A-and B-type cyclin-Cdk complexes varied significantly in their ability to phosphorylate Cdt1. Both cyclin A-Cdk1 and cyclin A-Cdk2 were about 6 -8-fold more active toward GST-Cdt1 than the cyclin B-Cdk1 (Fig. 2B). Furthermore GST-Cdt1 phosphorylation by the cyclin A-dependent kinases specifically retarded its electrophoretic mobility. As expected, this efficient phosphorylation required binding of the cyclin A-dependent kinases to Cdt1 via the Cy motif (Fig. 2, A and B). It was also notable that phosphorylation efficacy of Cdt1 by the two cyclin A-dependent kinases was virtually the same and that similar electrophoretic retardation was observed after the phosphorylation, suggesting that their phosphorylation sites in Cdt1 may be very similar in vitro.
We then asked whether Cdt1 is phosphorylated by cyclin A-dependent kinases in vivo as well as in vitro. However, a problem to be overcome in this context is that Cdt1 itself is degraded during the S, G 2 , and M phases (26) when cyclin A-dependent kinases are active. Especially given that Cdt1 phosphorylated by cyclin A-dependent kinases is targeted for degradation (see below), we anticipated that it would be difficult to analyze this in vivo. One possibility for dealing with phosphorylated Cdt1 may be to treat cells with proteasome inhibitors. Previously it was shown that phosphorylated and thus more slowly migrating Cdt1 becomes detectable following proteasome inhibitor treatment (26,32). Another possibility may be to synchronize cells in the G 2 /M phase through activation of a spindle checkpoint pathway with nocodazole treatment (26), although the mechanism is unknown. Indeed we also detected a slow migrating form of Cdt1 in 293T cells synchronized in the G 2 /M phase with nocodazole that disappeared with -phosphatase treatment ( Fig. 3B; for details, see below). If such phosphorylation were carried out by cyclin Adependent kinases, it would be expected to be lacking with Cdt1 Cy. To test this, 293T cells were transfected with either wild type T7-Cdt1 or T7-Cdt1 Cy and synchronized around G 2 /M phase with nocodazole treatment. Then whole cell lysates were prepared and analyzed by immunoblotting with anti-T7 antibody. As expected, the hyperphosphorylated and slow migrating form of Cdt1 was observed with the wild type but was undetectable with Cdt1 Cy (Fig. 2C). Taken together, these data lead us to conclude that Cdt1 is phosphorylated by cyclin A-Cdks in vivo.
Cdt1 Phosphorylation by Cyclin A-dependent Kinases Does Not Affect the Binding to Geminin-It has been suggested that inhibition of Cdt1 function by geminin is not associated with FIG. 1. In vivo interaction of Cdt1 with various cell cycle-regulating proteins and the role of the Cy motif. After transfection into 293T cells, T7-tagged wild type Cdt1 or the cyclin-binding motifmutated Cdt1 (Cdt1 Cy) were immunoprecipitated with anti-T7 antibody beads, and immunoprecipitates were subjected to immunoblotting with the indicated antibodies to detect Cdt1-associated proteins. Ten percent of the input sample (input) was analyzed with the precipitate (IP) to show the precipitation efficiency. cont, control antibody.
Cdk activity (25). On the other hand, ablation of Cdk1 kinase leads to rebinding of MCM and subsequent rereplication (16,17), leaving a possible regulatory role for Cdks in the Cdt1geminin interaction. To this end, we examined whether Cdk phosphorylation alters the affinity of Cdt1 for geminin in vitro. GST-Cdt1 and His-geminin produced by bacterial extracts and purified were mixed and subjected to kinase reactions with the three cyclin-Cdks. Under our experimental conditions, geminin was also phosphorylated by the Cdks although inefficiently (data not shown). The mixtures were then diluted with buffer and subjected to precipitation with glutathione beads. Remarkable amounts of His-geminin were bound to GST-Cdt1, and the efficacy was not changed by Cdt1 phosphorylation by the cyclin A-dependent kinases as confirmed by mobility shifts (Fig. 3A). No His-geminin was co-precipitated with GST alone (Fig. 3A).
Therefore, it is unlikely that Cdt1 phosphorylation by Cdks affects the binding to geminin.
The notion was further supported by different approaches as follows. 293T cells were transfected with expression vectors encoding Cdt1 and T7-geminin and synchronized at G 2 /M with nocodazole. Cell lysates were then prepared, immunoprecipitated with anti-T7 antibodies, and analyzed by immunoblotting with anti-Cdt1 and anti-T7 antibodies. In G 2 /M phase 293T cells, the slow migrating form of Cdt1 was found in addition to the normally migrating form, and both were similarly co-precipitated with T7-geminin (Fig. 3B). On treatment of the immunoprecipitates with -phosphatase, the slow migrating form disappeared (Fig. 3B), consistent with it being Cdt1 phosphorylated by cyclin A-dependent kinases. The treatment resulted in no detachment of Cdt1 from T7-geminin (Fig. 3B). Another

FIG. 2. Cyclin A-Cdk complexes specifically bind to and phosphorylate human Cdt1 depending on the cyclin-binding motif.
A, bacterially produced GST-Cdt1 was incubated with cyclin A-Cdk1, cyclin A-Cdk2, or cyclin B-Cdk1 purified from baculovirus-infected cells. Then GST fusion and associated proteins were collected on glutathione beads and subjected to Coomassie Blue staining (top panel) or to immunoblotting with mixtures of the anti-cyclin A and anti-cyclin B antibodies (middle panel) or the anti-Cdk1 and anti-Cdk2 antibodies (lower panel). Ten percent of the input sample (I) was analyzed with the precipitate (P). B, GST-Cdt1 or histone H1 were phosphorylated by the cyclin-Cdk complexes in the presence of 10Ci of [␥-33 P]ATP and processed for SDS-PAGE. The gels were stained with Coomassie Blue, dried, and analyzed with bioimaging analyzer BAS 2500 to measure incorporated radioactivity. C, 293T cells were transfected with either wild type T7-Cdt1 or T7-Cdt1 Cy and synchronized around G 2 /M phase with nocodazole treatment. Then the whole cell lysates were prepared and analyzed by immunoblotting with anti-T7 antibody. CBB, Coomassie Brilliant Blue; WT, wild type. approach taken was to investigate whether T7-Cdt1 Cy, which cannot be phosphorylated by cyclin A-dependent kinases as above, could bind to geminin in vivo. As shown in Fig. 1, no apparent difference in association with geminin was observed between wild type Cdt1 and Cdt1 Cy. Together these data clearly show that the interaction between Cdt1 and geminin is not affected by Cdk phosphorylation.

Association of Cdt1 with the F-box Protein Skp2 Requires Phosphorylation by Cyclin A-dependent Kinases and Leads to
Cdt1 Degradation-At least in certain cell types, the levels of Cdt1 proteins are decreased as they enter S phase (26). This appears to be due to the ubiquitin-proteasome system (26). Very recently it has been shown that Cdt1 binds to Skp2 depending on its phosphorylation, although the kinase responsible has not yet been identified (32). Independently we also found Skp2 to be co-precipitated with T7-Cdt1 ( Fig. 1; for details, see below). Skp2 is an F-box protein mediating the ubiquitination of target proteins by the SCF ubiquitin ligase complex (27). Therefore, it is presumable that ubiquitination of Cdt1 by the SCF Skp2 complex may lead to its degradation after S phase (32). We considered that Cdt1 phosphorylation required for Skp2 binding might be achieved by cyclin A-dependent kinases and investigated this possibility by in vitro binding assay. Recombinant Skp2 proteins were obtained by in vitro translation with rabbit reticulocyte lysate and subjected to binding assays with GST-Cdt1 treated with or without cyclin-Cdks. When GST-Cdt1 proteins were unphosphorylated or phosphorylated with cyclin B-Cdk1, no binding to Skp2 was observed (Fig. 4A). In clear contrast, significant binding was detected when GST-Cdt1 was hyperphosphorylated with cyclin A-dependent kinases (Fig. 4A). The reason for the relatively inefficient association observed is unclear at present, but considering that in vivo interaction between them detected by immunoprecipitation appears more efficient (see below), other components of SCF complex such as Skp1 would be required for the efficient recognition of Cdt1 by Skp2. Another problem is that cyclin A-Cdk2 phosphorylation of GST-Cdt1 appears to enhance binding to Skp2 more efficiently than cyclin A-Cdk1 phosphorylation. This finding proved repeatable in multiple experiments and thus suggests the possibility that there are differences in some phosphorylation site(s) in Cdt1 when phosphorylated by the two kinases, affecting its affinity for Skp2. If this were the case, it would be of interest and should be addressed in the future.
The data obtained with the in vitro assays indicate that Cdt1 phosphorylation by cyclin A-dependent kinases, which is dependent on the Cy motif, is required for the Skp2 binding. It was therefore predicted that wild type Cdt1, but not Cdt1 Cy, interacts with Skp2 in vivo. As shown in Fig. 1, wild type T7-Cdt1 efficiently co-immunoprecipitated Skp2 from 293T cell extracts, while T7-Cdt1 Cy could not, indicating that Cdt1 phosphorylation by cyclin A-dependent kinases also plays a crucial role in the Skp2 binding in vivo.
If recognition of the phosphorylated Cdt1 by Skp2 in fact results in its degradation after S phase, then the Cdt1 Cy mutant would be resistant to this. To test this prediction, we established Rat-1 cells stably expressing wild type T7-Cdt1 or

FIG. 3. Cdt1 phosphorylation by cyclin A-Cdk complexes does not affect its binding affinity for geminin. A, in vitro assay for
Cdt1-geminin binding. GST-Cdt1 and His-geminin synthesized by in vitro translation with bacterial lysates were subjected to phosphorylation reactions without radiolabeled ATP. Then GST fusion and associated proteins were collected on glutathione beads and subjected to immunoblotting with anti-Cdt1 or anti-geminin antibodies. As a negative control, GST and His-geminin were subjected to the same assay without a phosphorylation reaction. Ten percent of the input sample (I) was analyzed with the precipitate (P). B, phosphatase treatment of immunopurified Cdt1-geminin complexes. Cdt1-geminin complexes were immunoprecipitated with anti-T7 antibody from 293T cells transfected with Cdt1 and T7-tagged geminin and synchronized in G 2 /M phase with nocodazole. The purified immunoprecipitates were treated with -phosphatase or left untreated. Then proteins released into supernatants (S) or remaining on the beads (P) were separated by centrifugation and subjected to immunoblotting with anti-T7 or anti-Cdt1 antibodies. cont., control; Ab, antibody; PPase, phosphatase.

FIG. 4. Association of Cdt1 with Skp2 is dependent on phosphorylation by cyclin A-dependent kinases and results in its degradation.
A, in vitro binding of Skp2 to Cdt1 phosphorylated by cyclin A-dependent kinases. GST-Cdt1 was subjected to reaction with or without the cyclin-Cdks and mixed with HA-tagged Skp2 proteins synthesized by in vitro transcription-translation with rabbit reticulocyte lysate. Then the GST-Cdt1 and associated Skp2 proteins were collected on glutathione beads and analyzed by immunoblotting with anti-Cdt1 or anti-HA antibodies. Ten percent of the input sample (I) was analyzed with the precipitate (P). B, phosphorylation-deficient Cdt1 Cy mutant is partially resistant to S phase degradation. Rat-1 cells stably expressing wild type T7-Cdt1 or its Cy mutant were established. Whole cell lysates were prepared from the cells either synchronized in S phase by hydroxyurea treatment or left asynchronous and subjected to immunoblotting with anti-T7, anti-p27, or anti-actin antibodies. As, asynchronous; HU, hydroxyurea. its Cy mutant as described under "Experimental Procedures." Whole cell lysates were prepared from cells either synchronized in S phase with hydroxyurea or left asynchronous and then immunoblotted with anti-T7 antibody. The samples were also immunoblotted with anti-actin antibody for normalization. The levels of wild type T7-Cdt1 protein in the S phase cells were reduced to ϳ5% of those in the asynchronous cells (Fig. 4B). In contrast, the levels of T7-Cdt1 Cy protein in the S phase cells were reduced only to ϳ25% of those in the asynchronous cells (Fig. 4B). We also examined p27, a Cdk inhibitor, as an example of another protein that is degraded in an Skp2-dependent manner in S phase (27). In both cell lines, p27 protein levels were similarly reduced by about 50% by hydroxyurea treatment (Fig. 4B). These data indicate that Skp2 binding to Cdt1 phosphorylated by cyclin A-dependent kinases causes Cdt1 degradation in S phase. However, the data also demonstrate that there are other pathways involved in Cdt1 degradation.
Cdt1 DNA Binding Activity Is Reduced by Phosphorylation-Murine Cdt1 binds to DNA in a sequence-, strand-and conformation-independent manner (33). Given that this is an important property for Cdt1 function as an MCM loader, it would be expected that Cdt1 DNA binding activity is regulated by Cdk phosphorylation as well as by geminin binding (33). We therefore investigated whether human Cdt1 also has DNA binding activity and, if so, whether it is affected by Cdk phosphorylation. To this end, bacterially produced GST-Cdt1 was subjected to phosphorylation with the cyclin-Cdks as above, and using double-stranded DNA cellulose beads, the DNA binding activity was compared with the untreated case. While no specific binding of GST alone to DNA cellulose was observed (data not shown), the GST-Cdt1 demonstrated significant DNA binding, which was unaltered by phosphorylation with cyclin B-Cdk1 (Fig. 5). In contrast, GST-Cdt1 phosphorylated by cyclin A-dependent kinases had remarkably reduced affinity for DNA (Fig. 5).
Inhibition of Cdk1 Activity Results in Cdt1 Dephosphorylation and Rebinding to Chromatin in Murine FT210 Cells-We finally examined the in vivo role of Cdk1 in Cdt1 phosphorylation using murine Cdk1 kinase temperature-sensitive mutant FT210 cells (17,19,29). The cells were incubated with nocodazole for 16 h at permissive or non-permissive temperatures for synchronization at the G 2 /M phase, and whole cell lysates, Triton X-100-extractable fractions, and nuclear pellets containing chromatin and non-chromatin nuclear structures were prepared. Immunoblotting with monoclonal antibody 4A4 specific for Cdk1-phosphorylated vimentin confirmed the Cdk1 inactivation at the non-permissive temperature (data not shown) as described previously (17). The samples were then immunoblotted with anti-Cdt1 antibodies, showing the presence of slow migrating hyperphosphorylated murine Cdt1 in the soluble fraction from the G 2 /M phase FT210 cells cultured at the permissive temperature. In contrast, in cells cultured at the nonpermissive temperature, Cdt1 was converted to the normally migrating form and became detectable mainly in the nuclear pellet fraction (Fig. 6), demonstrating Cdt1 dephosphorylation and rebinding to chromatin/nuclear matrix in the absence of Cdk1 activity. Consistent with the data we previously reported (17), reassociation of MCM7 with chromatin was also observed in cells lacking Cdk1 activity (Fig. 6). These data indicate that Cdt1 phosphorylation by Cdk1 kinase plays a crucial role in suppression of its functions in vivo.

Cyclin A-dependent Kinases Phosphorylate Cdt1 through
Their Binding to the Cyclin-binding Motif-Our data demonstrate that cyclin A-dependent kinases efficiently phosphorylate Cdt1. This phosphorylation needs stable association of the cyclin-kinase complexes through the Cy motif of Cdt1. CDC6, which is ORC-dependently recruited onto chromatin and is required for MCM loading like Cdt1, is also phosphorylated by cyclin A-dependent kinases (6 -8) and similarly harbors a conserved cyclin-binding motif, which is in fact required for cyclin-Cdk complex binding (8). However, the cyclin A-dependent kinases need not stably interact with CDC6 via the Cy motif to efficiently phosphorylate the protein (8), indicating that another site in CDC6 may mediate the specific but relatively unstable interaction with the cyclin A-Cdk. This is in contrast to what we have observed here for Cdt1. Since the Cy-mutated Cdt1 we made is actually refractory to Cdk phosphorylation, it will be useful to explore how the phosphorylation influences the Cdt1 function in vivo.
It was found that both cyclin A-Cdk1 and cyclin A-Cdk2 are similarly active toward Cdt1 phosphorylation, indicating that cyclin A targets the kinase complexes. In the mammalian somatic cell cycle, cyclin A-Cdk2 is believed to be the S phasepromoting Cdk (34 -36), while cyclin B-Cdk1 is the mitotic Cdk (37). In addition, cyclin A-Cdk1 is active during S and G 2 phases (34). However, its biological functions remain to be fully addressed. Ablation of Cdk1 kinase activity results in rebinding of MCM proteins and subsequent rereplication (16,17). On the other hand, ablation of cyclin A, but not cyclin B, leads to rereplication in Drosophila tissue culture cells (18). Taken together, the data suggest a crucial role of cyclin A-Cdk1 in prevention of rereplication. Indeed cyclin A-Cdk1 can phosphorylate Cdt1 as efficiently as cyclin A-Cdk2, leading to suppression of some Cdt1 functions as shown here. It has been reported that in mammalian cells CDC6 is phosphorylated by cyclin A-Cdk2, resulting in its nuclear export and canceling its functions (6, 7). Cyclin A-Cdk1 could execute the same function, although this has yet to be investigated. Very recently it was FIG. 5. Cdt1 phosphorylation by cyclin A-Cdk complexes diminishes its affinity for double-stranded DNA in vitro. GST-Cdt1 was subjected to phosphorylation and loaded onto double-stranded DNA cellulose beads. Then the beads were washed three times, and the bound proteins were analyzed by immunoblotting with anti-Cdt1 antibodies. Forty percent of the input sample (I) was analyzed with the precipitate (P).
FIG. 6. Cdk1 inactivation in murine Cdk1 kinase temperaturesensitive mutant FT210 cells results in Cdt1 dephosphorylation and rebinding to chromatin/nuclear matrix. FT210 cells were incubated with nocodazole for 16 h at permissive (32°C) or non-permissive (39°C) temperatures. Whole cell extracts (Total), Triton X-100extractable fractions (Triton S), and nuclear chromatin/matrix fractions (Triton P) were prepared from the cells and immunoblotted with anti-Cdt1 or anti-MCM7 antibodies. The samples were also subjected to SDS-PAGE followed by Coomassie Blue staining for detection of core histones.
shown that Cdk2 is dispensable for somatic cell cycling (38), while cyclin A is essential (36). Therefore, the cyclin A-Cdk1 could play important roles in both positive and negative regulation of replication. On the other hand, although our data demonstrate that cyclin A-dependent kinases play a crucial role in phosphorylation and regulation of Cdt1, we cannot exclude the possibility that cyclin B-Cdk1 might also function in vivo.
Cdt1 Phosphorylation by Cyclin A-dependent Kinases Does Not Affect the Binding to Geminin-In the absence of Cdk1 kinase activity, geminin is insufficient to completely suppress rereplication (17,18), suggesting the possibility that geminin binding to Cdt1 is regulated by Cdk phosphorylation. However, our present data clearly indicate that this is not the case. Such aspect of regulation of geminin binding to Cdt1 might be favorable under certain conditions during the cell cycle; even when Cdk activity is down-regulated by the checkpoint mechanism in cells undergoing DNA damage, reformation of pre-RC could be prevented (1). However, during normal cell cycling, both Cdk phosphorylation and geminin may play indispensable roles in prohibition of rereplication.
Inhibitory Effects of Cdt1 Phosphorylation by Cyclin A-dependent Kinases-Cdk activity has a bipartite function in the cell cycle regulation of eukaryotic DNA replication, promoting replication and preventing rereplication (1,2). Candidate molecules that should be phosphorylated by the Cdks to stimulate replication may include MCM and Sld2 proteins (1,2,39). It has been reported that the Cdks physically interact with and phosphorylate ORC and CDC6 (5-10). However, considering that they are no longer required for initiation reaction after loading MCM in an in vitro DNA replication system with Xenopus egg extracts (11,12), such phosphorylation is likely to be involved in negative regulation. Indeed, in mammalian somatic cells, ORC1 protein is degraded after S phase (10,20), and CDC6 is, at least partly, excluded from the nuclei through phosphorylation by Cdks (5)(6)(7)19). It has also been shown that unphosphorylatable mutant CDC6 can support DNA replication in egg extracts (40).
The current concept is that Cdt1, together with ORC and CDC6 proteins, forms the machinery that loads MCM heterohexameric complexes onto prereplication chromosomal DNA (1,2). By analogy with ORC and CDC6, Cdt1 might also be dispensable after loading MCM. Indeed it has been found that Cdt1 protein levels fall after S phase in certain cell types (26,32). Therefore, it is very conceivable that Cdt1 phosphorylation by cyclin A-dependent kinases negatively regulates its function after S phase. Our data presented here support this notion. First, Cdt1 phosphorylated by cyclin A-dependent kinases is targeted for binding to Skp2 and subsequently degraded. It should also be noted, however, that other pathways exist that are involved in Cdt1 degradation after S phase (41). Prevention of Cdt1 function may be carried out not only by regulation at the protein level but also through effects on activity. In nocodazole-treated and spindle checkpoint-activated cells, levels of Cdt1 protein recover by unknown mechanisms (26), but they remain detached from chromatin as shown here. Also in a DNA replication system with egg extracts, Cdt1 is dissociated from chromatin after S phase without degradation (22). Since Cdk1 inactivation leads to Cdt1 dephosphorylation and rebinding accompanied by MCM reassociation, inhibition of the Cdt1 chromatin rebinding may depend on the Cdk1 phosphorylation. Our finding that Cdt1 phosphorylation by cyclin A-dependent kinases reduces its affinity to DNA provides one possible explanation for such inhibition. Overall the data lead us to conclude that Cdt1 phosphorylation by cyclin A-dependent kinases plays crucial roles in negative regulation of its function after S phase. Indeed we have observed that the ability of the Cymutated Cdt1 to induce activation of the ATM (ataxia telangiectasia-mutated) checkpoint pathway and subsequent rereplication in the overexpressed cell is higher than the wild type despite similar binding affinity for geminin. 2