Cyclin-dependent kinases phosphorylate human Cdt1 and induce its degradation.

Eukaryotic cells tightly control DNA replication so that replication origins fire only once during S phase within the same cell cycle. Cell cycle-regulated degradation of the replication licensing factor Cdt1 plays important roles in preventing more than one round of DNA replication per cell cycle. We have previously shown that the SCF(Skp2)-mediated ubiquitination pathway plays an important role in Cdt1 degradation. In this study, we demonstrate that human Cdt1 is a substrate of Cdk2 and Cdk4 both in vivo and in vitro. Overexpression of cyclin-dependent kinase inhibitors such as p21 and p27 dramatically suppresses the phosphorylation of Cdt1, disrupts the interaction of Cdt1 with the F-box protein Skp2, and blocks the degradation of Cdt1. Further analysis reveals that Cdt1 interacts with cyclin/cyclin-dependent kinase (Cdk) complexes through a cyclin/Cdk binding consensus site, located at the N terminus of Cdt1. A Cdt1 mutant carrying four amino acid substitutions at the Cdk binding site dramatically reduces associations with cyclin/Cdk complexes. This mutant is not phosphorylated, fails to bind Skp2 and is more stable than wild-type Cdt1. These data suggest that cyclin/Cdk-mediated Cdt1 phosphorylation is required for the association of Cdt1 with the SCF(Skp2) ubiquitin ligase and thus is important for the cell cycle dependent degradation of Cdt1 in mammalian cells.

The replication licensing factor Cdt1 is an essential component of the pre-replication complex and is required for loading MCM proteins onto chromatin (1)(2)(3)(4)(5)(6). Cdt1 was originally isolated in fission yeast and is conserved in different organisms (1). In fission yeast, Cdt1 cooperates with another essential replication initiation regulator Cdc18 (Cdc6 homolog in Schizosaccharomyces pombe) to initiate DNA replication (4). Immunodepletion of Cdt1 in Xenopus egg extracts prevents DNA replication in an in vitro DNA replication system (7). Microinjection of a Cdt1 antibody in human cells prevents DNA replication initiation (6).
The protein levels of Cdt1 in fission yeast and in mammalian cells are tightly controlled (4,8). Human Cdt1 protein is pres-ent in G 1 phase of the cell cycle and is degraded as cells enter S phase (8). Cdt1 in higher eukaryotes binds geminin, a DNA replication inhibitor that is present in S and G 2 , and is inactivated by geminin (5). The tight control of Cdt1 protein level during cell cycle, together with the observation that overexpression of Cdt1 alone or with Cdc6/Cdc18 induces DNA rereplication (9,10), suggests that the cell cycle dependent inactivation of Cdt1 is essential for preventing inappropriate DNA replication initiation.
Cdks 1 play major roles in regulating cell cycle progression (11,12). In mammalian cells, cyclinD-Cdk4/Cdk6 complexes operate the passage through the G 1 phase of the cell cycle. Subsequently, the CyclinE/Cdk2 complex is formed and promotes the G 1 to S transition, and the CyclinA/Cdk2 complex regulates the S phase progression. In mammalian cells, the activity of Cdks can be inhibited by the interaction of cyclin/ Cdk complexes with Cdk inhibitors, such as p21 and p27 (12). It was demonstrated that the elevated Cdk activity prevents re-initiation of DNA replication in S and G2 from origins that have fired (13,14). In yeast, Cdc6 (Cdc18) are phosphorylated by Cdks and are subsequently targeted for ubiquitin-dependent proteolysis (15)(16)(17)(18). In mammalian cells, Cdc6 is also a substrate of Cdks and the phosphoryaltion regulates its subcellular localization (19,20). However, the post-translational modification of Cdt1 remains to be investigated.
We previously showed that cell cycle regulated Cdt1 degradation in mammalian cells is mediated through the SCF Skp2 ubiquitination pathway (21). Specifically, Cdt1 interacts with the F-box protein Skp2 in a phosphorylation dependent manner and the SCF Skp2 ubiquitin ligase complex targets Cdt1 for degradation. In the presence of proteasome inhibitor MG132, the phosphorylated Cdt1 species are accumulated. These studies suggest that phosphorylation of Cdt1 may be important for the controlled degradation of Cdt1 when cells enter S phase. Here, we show that cyclin/Cdk complexes interact with and phosphorylate Cdt1 and these phosphorylation modifications are required for the association of Cdt1 with Skp2 and thereby important for Cdt1 degradation.

EXPERIMENTAL PROCEDURES
Plasmids-Myc-tagged full-length human Cdt1 and various Cdt1 deletion mutants were described previously (21). The mutant carrying a deletion of residues 62 to 72 was generated by ligating PCR products encoding residues 1-62 (AgeI-NheI) and 72-546 (NheI-XhoI) with the vector pCDNA3␤ (22) that carries the sequence encoding the Myc epitope after AgeI and XhoI digestion. The NAAIRS mutant was generated by ligating the PCR product encoding the residues 1-64 followed by the amino acids NAAIRS included in the 3Ј-primer (ATCGATCGG-ATGGCGGCGTTTGGCCTGGCCTGGTCGCGTCCGGGG; underline, * This work was supported by a start-up fund from The Scripps Research Institute (TSRI). 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 U.S.C. Section 1734 solely to indicate this fact. PvuI site; bold, encode NAAIRS) to that encoding the residues 71-546 proceeded with the amino acids RS included in the 5Ј-primer (TACGA-TCGCGGCTGTCGGTGGACGAGGTTTCCAGC; underline, PvuI site and RS) after PvuI digestion.
Cell Culture-All cell lines were purchased from American Type Culture Collection and cultured in Dulbecco's modified Eagle's medium supplemented with 10% fetal bovine serum.
Immunoprecipitation, Immunoblotting, and Antibodies-Immunoprecipitation and Western blot analysis were performed as described previously (23).
Protein phosphatase treatment was conducted by incubating cell lysates with 500 units of -phosphatase (New England Biolabs) for 1 h at 30°C.
Transfection and Retroviral Infection-Transient transfections were carried out by the calcium phosphate method. 15 g of the indicated plasmids and 1 g of CMV-GFP plasmid were used for each transfection.
Recombinant retroviruses were generated as described and infected cells were selected by drug treatment (24). pBABE-Puro retroviral vector and 293T-derived packaging cell line LinXA were used in this study.
Kinase Assay-FLAG-Cdk2 or FLAG-Cdk4 was expressed in 293T cells after transient transfection and purified by anti-FLAG immunoprecipitation. The Cdk2 and Cdk4 kinases were eluted by the FLAG peptide in the Cdk kinase reaction buffer (25 mM Tris-Cl, pH 7.5, 150 mM NaCl, 10 mM MgCl 2 , 1 mM dithiothreitol). Equal amounts of eluted Cdk kinases were added to GST-fused full-length Cdt1 immobilized on glutathione-Sepharose beads. Kinase reactions were carried out in a 20-l volume at 30°C for 30 min with 40 M ATP, 1 l of [␥-32 P]ATP (2000 Ci/mmol), and 20 M protein kinase A inhibitor. 5 M Cdk inhibitor roscovitine was added in the reactions served as negative controls. After boiling the samples in Laemmli sample buffer, the proteins were resolved by SDS-PAGE and the incorporation of [ 32 P]phosphate was revealed by autoradiography.

Human Cdt1 Is a Substrate of Cdks in Vivo-In yeast, Cdc6
and Cdc18 are Cdk substrates and Cdk-mediated phosphorylation triggers Cdc6/Cdc18 degradation (15)(16)(17)(18). To examine whether Cdks phosphorylate Cdt1 in mammalian cells, we overexpressed Cdk2 and Cdk4 with or without co-expression of cyclinD1, cyclinE and cyclinA in 293T cells. Transient transfection of Cdk2 and Cdk4 induced phosphorylation of endogenous Cdt1, as revealed by increased density and a more dramatic shift of the slower migrating Cdt1 bands (Fig. 1A). Co-transfection of cyclinA, cyclinE, or cyclinD1 along with Cdks further enhanced Cdt1 phosphorylation. phosphatase treatment of cell lysate abolished the slower migrating bands, confirming that the shift of Cdt1 bands was caused by Cdkmediated phosphorylation.
To further investigate whether Cdt1 is phosphorylated by Cdks, we overexpressed Cdk inhibitors p21 and p27 and examined Cdt1 phosphorylation. 293T cells were transfected with p21, p27, or the vector and lysed before and after MG132 treatment. In the vector-transfected cells, endogenous Cdt1 exhibited a doublet even without MG132 treatment (Fig. 1B,  first lane). The top band can be abolished by phosphatase treatment (Fig. 1B, second lane). In the presence of MG132, the Cdt1 species with even slower mobility accumulated. phosphatase treatment indicated that the slower migrating species of Cdt1 was the phosphorylated form of Cdt1. When p21 or p27 was expressed, in the absence of MG132, the top band of the doublet present in the vector-transfected cells is diminished. In the presence of MG132, the expression of p21 or p27 dramatically decreased the amount of slower migrating species of Cdt1, and the lower band was accumulated. These results demonstrate that overexpression of p21 and p27 inhibits Cdt1 phos- phorylation and suggest that Cdk activities are required for the phosphorylation of Cdt1 in mammalian cells.
Cdk2 and Cdk4 Phosphorylate Cdt1 in Vitro-To more directly test whether Cdt1 is a substrate of Cdks, we performed an in vitro kinase assay. 293T cells were transfected with FLAG-Cdk2, FLAG-Cdk4, or vector. Cdk2 and Cdk4 were purified by anti-FLAG immunoprecipitation and eluted by the FLAG peptide. GST-fused full-length Cdt1 was used as a substrate. As shown in Fig. 1C, Cdt1 was phosphorylated by Cdk2 and Cdk4 in vitro, and the phosphorylation was inhibited in the presence of Cdk inhibitor roscovitine.
The C-terminal 168 Residues Are Important for Cdk-mediated Phosphorylation of Cdt1-To examine the regions of Cdt1 that can be phosphorylated by Cdks in vivo, we co-transfected 293T cells with cyclinA/Cdk2 or cyclinE/Cdk2 and the Myctagged full-length Cdt1 or Cdt1 C-terminal deletion mutants. Overexpression of cyclinA/Cdk2 and cyclinE/Cdk2 retarded the migration of the full-length Cdt1-(1-546) and the fragment 1-417 but not the fragments 1-378 and 1-285 (Fig. 1D).
-Phosphatase treatment abolished the migration shift (data not shown), confirming that overexpression of cyclinA/Cdk2 and cyclinE/Cdk2 induces phosphorylation of the full-length Cdt1 and fragment 1-417. These results suggest that the Cterminal 168 residues that are absent in the Cdt1 fragment 1-378 are important for the Cdk-mediated phosphorylation of Cdt1, and the region between residues 378 and 417 may contain major Cdk phosphorylation sites or other elements that are required for Cdk-mediated Cdt1 phosphorylation.

Inhibition of Cdk Activities Suppresses the Association of Cdt1 with Skp2 and Blocks the Degradation of Cdt1-We
showed that the interaction of Cdt1 and Skp2 is phosphorylationdependent. To test whether Cdk-mediated phosphorylation of Cdt1 is important for this association, we transfected 293T cells with Cdk inhibitors, p21 and p27. The interaction of Cdt1 and Skp2 was examined by co-immunoprecipitation of Skp2 with Cdt1 ( Fig. 2A). Overexpression of p21 and p27 dramati-cally reduced the association of Cdt1 with Skp2. This suggests that Cdk activities are required for Cdt1 to bind Skp2.
The SCF Skp2 -dependent ubiquitination pathway plays an important role for Cdt1 degradation (21). As inhibition of Cdk kinase activities suppresses the association of Cdt1 with Skp2, we tested whether Cdk activities are important for Cdt1 degradation. Overexpression of p21 and p27 led to accumulation of endogenous Cdt1 (Fig. 2B). The stability of Cdt1 was measured. As shown in Fig. 2C, Cdt1 was significantly more stable when p21 and p27 were overexpressed. Taken together, these data suggest that the phosphorylation of Cdt1 by cyclin/Cdk complexes triggers the association of Cdt1 with Skp2, thereby inducing Cdt1 degradation.

Cdt1 Interacts with Cyclin/Cdk2 and Cyclin/Cdk4 Complexes through a Cyclin Binding Consensus Site at the N Terminus of
Cdt1-Co-immunoprecipitation revealed that Cdt1 interacted with cyclinE/Cdk2 and cyclinA/Cdk2 complexes, and Cdk4 when they were overexpressed in 293T cells (Fig. 3, B-D; vector versus full-length Cdt1 (FL)).
To search for regions in Cdt1 that mediate the cyclin/Cdks binding, a series of Cdt1 deletion mutants were co-expressed with HA-cyclinA/FLAG-Cdk2, HA-cyclinE/FLAG-Cdk2, or cy-clinD1/FLAG-Cdk4 in 293T cells. Deletion of the N-terminal 52 residues significantly reduced Cdt1 binding with different cyclin/Cdk complexes (Fig. 3, B-D). Further deletions from the N terminus caused even more dramatic effects on the binding. The C-terminal deletion mutants were capable to bind cyclin/ Cdk complexes with same efficiency as wild-type Cdt1. These data suggest that the association between Cdt1 and cyclin/Cdk complexes is mediated by the N-terminal portion of Cdt1.
A cyclin/Cdk binding consensus site 65 PARRRLRL 72 is located at the N-terminal part of Cdt1 ( Fig. 3A; Refs. 25 and 26)). Deletion of residues 62-72 covering the consensus site abolished or dramatically reduced binding of Cdt1 with cyclinA/ Cdk2, cyclinE/Cdk2, and Cdk4 (Fig. 3, B-D). To minimize the conformational disruption caused by the deletion, we replaced FIG. 2. The kinase activities of Cdk complexes are important for the association of Cdt1 with Skp2 and for the degradation of Cdt1. A, overexpression of p21 and p27 inhibits the association of Cdt1 with Skp2. 293T cells were transfected with p21, p27, or the vector. The association of Skp2 with Cdt1 was demonstrated by anti-Cdt1 immunoprecipitation, followed by anti-Skp2 Western blot analysis (left). Similar Skp2 protein levels in each sample were detected by Western blot analysis using cell lysates (right). The expression of p21 and p27 was confirmed by Western blot analysis (data not shown). B, Cdt1 protein level increases when p21 or p27 is overexpressed. 293T cells were transfected with p21 or p27 and lysed after 48 h. Cdt1 protein level was examined by Western blot analysis. Actin was used as a loading control. C, overexpression of p21 or p27 inhibits Cdt1 degradation. 293T cells were transfected with p21, p27, or vector. Cells were lysed at the indicated time after CHX (100 M) treatment. Cdt1 protein levels were visualized by anti-Cdt1 Western blot analysis. The expression of p21 and p27 was confirmed by Western blotting (data not shown). Actin was used as a loading control. Quantitative analysis was performed by the measurement of Cdt1 band intensities normalized by the band intensities of actin at each time point. The percentage of the remaining of Cdt1 after CHX treatment was plotted over time (hour).
four amino acids in the consensus binding site with the sequence NAAIRS (Fig. 3A), a flexible linker that can assume an ␣-helical or ␤-strand conformation (27). The Myc-Cdt1 NAAIRS mutant failed to bind cyclinA/Cdk2, cyclinE/Cdk2, and Cdk4, suggesting that the Cdk binding consensus site at the N terminus of Cdt1 mediates the association of Cdt1 with Cdks.
The Interaction of Cdt1 with Cdk Complexes Is Important for Cdt1 Phosphorylation-To test whether the interactions of Cdt1 and Cdk complexes are important for the phosphorylation of Cdt1, we generated stable 293 cell lines that express the Myc-tagged wild-type Cdt1 and the Myc-NAAIRS Cdt1 mutant by retroviral infection. Myc-tagged Cdt1 migrates more slowly than endogenous Cdt1, thus the anti-Cdt1 Western blot analysis shows the tagged and the endogenous Cdt1 species as two separate bands. As illustrated in Fig. 4A, treatment with MG132 induced a migration shift of Myc-Cdt1-WT and endogenous Cdt1, which can be abolished by phosphatase treatment. However, no obvious phosphorylation of Myc-Cdt1 NAAIRS mutant was observed after MG132 treatment, although at the same time the endogenous Cdt1 was phosphorylated. These data demonstrate that the Cdt1 NAAIRS mutant defective in Cdk binding is not phosphorylated efficiently in vivo and suggest that the association of Cdt1 with Cdk complexes is important for Cdt1 phosphorylation.
The Cdt1 Mutant Defective in Cdk Association Does Not Bind to Skp2 and Is More Stable than Wild-type Cdt1-As the Cdt1 NAAIRS mutant is defective in phosphorylation, we examined whether this mutant was able to bind Skp2. Immunoprecipitation of Skp2 from the 293 stable cell lines that expressed Myc-Cdt1-WT or Myc-Cdt1-NAAIRS showed that endogenous Cdt1 and Myc-Cdt1-WT, but not the Myc-Cdt1-NAAIRS mutant, interacted with Skp2 (Fig. 4B, left). Anti-Myc immunoprecipitation also demonstrated that Myc-Cdt1-WT, but not the Myc-Cdt1-NAAIRS mutant, associated with endogenous Skp2 (Fig. 4B, middle).
The Cdt1 NAAIRS mutant is more stable than wild-type Cdt1. 293 cell lines that stably express Myc-Cdt1-WT or Myc-Cdt1-NAAIRS were treated with cycloheximide (CHX) and subsequently lysed at different time points. Anti-Myc immunoblot-ting showed that Myc-Cdt1-NAAIRS degraded more slowly than Myc-Cdt1-WT (Fig. 4C, top). Consistently, anti-Cdt1 Western blot indicated that the Myc-Cdt1-NAAIRS mutant, but not Myc-Cdt1, was more stable than endogenous Cdt1 (Fig.  4C, bottom). These results suggest that Cdt1 phosphorylation is required for Cdt1 to associate with Skp2 and therefore important for the degradation of Cdt1.

DISCUSSION
The Cdt1 protein levels fluctuate during the cell cycle and proteolysis appears to be a primary mechanism for the periodic accumulation of Cdt1 (8). MG132 treatment stabilizes Cdt1 and leads to the accumulation of the phosphorylated species of Cdt1, suggesting that phosphorylation of Cdt1 may play an important role for targeting Cdt1 for degradation (21). Here we provide evidence that Cdt1 is a substrate of Cdk2 and Cdk4 both in vivo and in vitro. Cdt1 interacts with multiple cyclin/ Cdk complexes, including cyclinE/Cdk2, cyclinA/Cdk2, and Cdk4 complexes. Inhibition of Cdk activities suppresses Cdt1 phosphorylation and leads to the stabilization of Cdt1. Furthermore, a NAAIRS Cdt1 mutant defective in Cdk binding is no longer phosphorylated and exhibits greater stability than wildtype Cdt1. These results suggest that cyclin/Cdk complexes phosphorylate Cdt1 and these phosphorylation events are important for Cdt1 degradation. The contribution of each individual cyclin/Cdk complex to this process is currently under investigation.
We have shown that the degradation of Cdt1 is mediated at least in part by the SCF Skp2 -dependent ubiquitination pathway (21). The interaction of Cdt1 and Skp2 is phosphorylation-dependent. Our study demonstrates that the association of Cdt1 with Skp2 can be abolished by the inhibition of Cdt1 phosphorylation when p21 or p27 is overexpressed or when a Cdk binding mutant (NAAIRS) is used. These results are consistent with a model that Cdt1 is phosphorylated by Cdk complexes as cells pass through G 1 and enter S phase. Consequently, the phosphoryalted Cdt1 associates with the SCF Skp2 complex through Skp2 and is degraded through ubiquitination pathway.

FIG. 3. Cdt1 interacts with cyclin/ Cdks through the Cdk binding consensus site at the N terminus of Cdt1.
A, a schematic drawing of full-length Cdt1 (FL) and deletion mutants. The sequence from residues 61-72 covering the Cdk binding consensus site is shown. The amino acids that are changed by substitutions to NAAIRS are indicated. B-D, Cdt1 interacts with different cyclin/Cdk complexes. Myc-tagged Cdt1 full-length (FL) and deletion or NAAIRS mutants were co-transfected with HA-CycA/ FLAG-Cdk2, HA-CycE/FLAG-Cdk2, and CycD1/FLAG-Cdk4 in 293T cells. Myc-Cdt1 wild-type and mutant proteins were immunoprecipitated with an anti-Myc antibody (9E10). The interactions were visualized by Western blot analysis using anti-HA (HA11) and anti-FLAG (M2) antibodies. The expression of full-length Cdt1 or Cdt1 mutants were shown by anti-Myc Western blot using cell lysates.
Cdt1 is a replication licensing factor that is required for the initiation of DNA replication (4 -6). In addition, the levels of Cdt1 need to be tightly regulated to prevent origin re-firing, as overexpression of Cdt1 in p53-deficient cells leads to DNA re-replication (10). Under the phosphorylation control by Cdk complexes, Cdt1 becomes a substrate of SCF Skp2 and degrades when cells enter S phase. This regulation allows Cdt1 to accumulate only in G 1 and thus prevents the formation of prereplication complex after DNA replication starts, which is important for the prevention of DNA re-replication. Therefore, Cdk-mediated and cell cycle-regulated Cdt1 degradation may be an important mechanism ensuring strict control of DNA replication and the maintenance of genome stability.