Human Cdt1 Lacking the Evolutionarily Conserved Region That Interacts with MCM2–7 Is Capable of Inducing Re-replication*

Replication initiation must be a carefully regulated process to avoid genomic instability caused by aberrant replication. In eukaryotic cells, distinct steps of protein loading (origin licensing) and replication activation are choreographed such that a cell can replicate only once per cell cycle. The first proteins recruited to the origins form the pre-replication complex. Of these proteins, Cdt1 is of interest, as it is the focus of several pathways to control replication initiation. It is degraded by two different pathways, mediated by the interaction of Cdt1 with proliferating cell nuclear antigen (PCNA) or with cyclin-Cdk2 and inhibited by geminin once cells are in S-phase, presumably to prevent reloading of pre-replication complexes once S-phase has begun. Although the requirement of Cdt1 in loading MCM2–7 is known, the mechanism by which overexpressed Cdt1 stimulates re-replication is unclear. In this study we have designed various mutations in Cdt1 to determine which portion of Cdt1 is important for re-replication, providing insight into possible mechanisms. Surprisingly, we found that mutants of Cdt1 that do not interact with MCM2–7 are able to induce re-replication when overexpressed. The re-replication is not due to titration of geminin from endogenous Cdt1 and is not accompanied by stabilization of endogenous Cdt1. Additionally, the N-terminal one-third of Cdt1 is sufficient to induce re-replication. The N terminus contains the PCNA- and cyclin-interacting motifs, and deletion of both motifs simultaneously in the overexpressed Cdt1 prevents re-replication. These findings suggest that exogenous Cdt1 induces re-replication by de-repressing endogenous Cdt1 through the titration of PCNA and cyclin.

DNA replication is a complex process that a cell must perform accurately to ensure the genome is copied correctly with minimal damage. Additionally, replication must be temporally regulated so that the genome is copied exactly once before division. This ensures that daughter cells will not get undesirable amplifications or deletions that may cause abnormal functions, such as the unregulated growth observed in cancer. Work in the last decade has revealed an elaborate system to ensure that DNA replication occurs once, and only once, per cell cycle. This system of replication licensing relies on the sequential recruitment of replication initiation factors to chromatin at the appropriate time and prevents association at all other times. By controlling the chromatin association of these initiation factors, cells are able to limit precisely when replication can begin and prevent it otherwise.
In higher eukaryotes, it is generally thought that the origin recognition complex recognizes origins of replication. The origin recognition complex serves as an origin marker and recruits Cdc6 and Cdt1 to chromatin. These proteins are then required to load the MCM2-7 complex. Once the MCM2-7 complex is loaded, downstream members of the preinitiation complex, including Cdc45/Sld3, GINS, MCM10, and Dpb11/Sld2, are loaded onto chromatin, eventually culminating in recruitment of the replication machinery. The recruitment of these factors is controlled by the presence or absence of cyclin-dependent kinase (CDK) 2 activity. CDK activity is low in G 1 , when the pre-replication complex (consisting of the origin recognition complex, Cdc6, Cdt1, and MCM2-7) is loaded. Increasing CDK activity is then required for further loading of the preinitiation complex members and the replication machinery (for review see Refs. 1 and 2). MCM2-7 is thought to be the replicative helicase responsible for unwinding DNA ahead of the replication machinery (3)(4)(5)(6). It is loaded in a Cdc6-and Cdt1-dependent manner, and this loading can only occur during low levels of CDK activity, in late M phase and early G 1 phase. Cdt1 has been shown to associate physically with MCM2-7 in many organisms (7)(8)(9)(10), suggesting that it directly recruits MCM2-7 to chromatin. However, Cdc6 ATPase mutants increase association of Cdt1 and decrease association of MCM2-7 with chromatin, which might suggest mutual exclusion of Cdt1 and MCM2-7 on chromatin (7,11). Either way, it has been suggested that the interaction between Cdt1 and MCM2-7 is required to load the helicase on chromatin. Recent experiments in Xenopus egg extracts show that the C-terminal region of Cdt1 is required for MCM2-7 loading and that this region is also required for the interaction with MCMs (mini-chromosome maintenance) (10). The * This work was supported by National Institutes of Health Grant R01CA60499. 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.
importance of the C-terminal one-third of Cdt1 is highlighted by the fact that this is the only area that shows sequence identity from Saccharomyces cerevisiae to humans (7,12). Once MCM2-7 is loaded, its subsequent activation (and eventual origin firing) requires CDK and CDK-like activity (for review see Ref. 2). Therefore, MCM2-7 bridges the two CDK activity states that function to license replication: its chromatin loading must occur in low CDK activity, and its activation and origin firing must occur in high CDK activity.
One of the key players in the replication licensing system is Cdt1. Cdt1 is required for MCM2-7 loading (7,13,14). MCM2-7 loading is limited to late M and early G 1 phases in part due to inhibition of Cdt1 by several mechanisms in S-phase. Cdt1 is degraded in S-phase by two different mechanisms: CDK-dependent degradation by the SCF (Skp1/Cullin/ F-box containing) complex and PCNA-dependent degradation by the Cul4-DDB1 complex (for review see Ref. 15). It is also inhibited by geminin (16,17), a small protein that occurs at high levels during S-phase. Disrupting these mechanisms of Cdt1 inhibition can cause re-replication in a variety of systems (18 -22). This suggests that when Cdt1 is not inhibited it can stimulate re-replication. Despite the expectation that inappropriate Cdt1 activity in S-phase may cause re-replication by excessive MCM2-7 reloading, the actual role of Cdt1 in re-replication induction is unclear.
Recent work has shown that Cdt1 levels are increased in certain cancers (23). Interestingly, in 73% of cases in which p53 was also mutated, aneuploidy was observed. Our laboratory has reported similar observations; in tumor cell lines lacking p53 and overexpressing Cdt1 and Cdc6, re-replication is observed (18). Furthermore, high levels of Cdt1 appear to have a causative role in cancer. It has been shown that retrovirally activated Cdt1 can be oncogenic in mice (24) and that Cdt1 overexpression in thymocytes can cause lymphomas when p53 is also mutated. As Cdt1 overexpression causes re-replication in a variety of systems, it is possible that high levels of Cdt1 induce genomic instability and tumorigenesis by inducing re-replication. Therefore, an understanding of the mechanism of re-replication initiation is also informative for tumorigenesis and disease progression.

EXPERIMENTAL PROCEDURES
Cell Culture-293T cells, human embryonic kidney cells transformed with adenovirus oncogenes E1a and E1b and with simian virus 40 oncogene T antigen, were maintained in Dulbecco's modified Eagle's medium supplemented with 10% iron-supplemented donor calf serum and 1% penicillin/streptomycin. Standard tissue culture growth conditions and methods were used. Lipofectamine 2000 (Invitrogen) was used to transfect cells.
Cdt1 Mutant Construction-Cdt1 mutants were subcloned using PCR-based techniques in a vector with an N-terminal FLAG tag. Primer sequences used are available upon request. Because the natural nuclear localization sequence of Cdt1 is in the first 93 residues, the constructs lacking these N-terminal residues (94 -X, 163-X) have an artificial nuclear localization sequence fused to the N terminus to ensure normal localization.
Western Blotting, Immunoprecipitation, and siRNA-Western blotting was performed as described. In this case, cells were lysed in 50 mM Tris, pH 7.4, 0.2% Nonidet P-40, 150 or 300 mM NaCl, 1 mM EDTA, 10 mM NaF, 0.2 mM Na 3 VO 3 , 1 mM phenylmethylsulfonyl fluoride, 2 mM dithiothreitol, and a 1:100 protease inhibitor mixture (Sigma). Equal amounts of protein were loaded on SDS-PAGE followed by transfer to nitrocellulose for standard Western blotting. FLAG and ␤-actin antibodies were purchased from Sigma, and MCM6 and cyclin A antibodies were purchased from Santa Cruz Biotechnology. Anti-FLAGagarose beads (Sigma) were used for Cdt1 co-immunoprecipitation. Geminin antibody was described previously (16). hsCdt1 antibody was described previously (25). siRNA was performed according to common methods using RNAi MAX (Invitrogen). siRNA oligonucleotides against human geminin have been described previously (20). The Cdt1 siRNA sequence is 5Ј-GCAAUGUUGGCCAGAUCAA-3Ј.
FACS-Cells were prepared as described previously (18). Cells were cotransfected with farnesylated EGFP and gated for GFP-positive cells to ensure analysis was limited to the Cdt1transfected cells. Alternately, cells were transiently transfected with a puromycin resistance marker, and 48 h of puromycin treatment was used to enrich transfected cells. The analysis was carried out on a BD Biosciences FACS Calibur using Cellquest and FloJo software.
Chromatin Fractionation-Chromatin fractionation protocols used to observe MCM7 and Cdt1 chromatin loading were described previously (Refs. 26 and 27, respectively). Briefly, the MCM7 chromatin loading was observed as follows. Cells (2 ϫ 10 6 ) were lysed in 100 l of CSK buffer (10 mM Pipes, pH 7.0, 100 mM NaCl, 300 mM sucrose, 3 mM MgCl 2 ) containing 0.5% Triton X-100, 1 mM ATP, and 1 mM Na 3 VO 4 . Lysates were incubated on ice for 20 min and then centrifuged at 1500 rpm for 5 min at 4°C. Supernatant (S1) was removed, and pellets were washed with 1 ml of lysis buffer and centrifuged again. Pellets were incubated in 100 l of lysis buffer containing 1 mM CaCl 2 and 120 units of micrococcal nuclease (Worthington) for 10 min at 37°C and centrifuged. Supernatant (S2, chromatinbound fraction) was removed, and pellets were washed with 1 ml of lysis buffer and centrifuged again. Pellets were boiled in 100 l of 1ϫ sample buffer (P2) (26).
The Cdt1 chromatin loading was determined as follows. Cells were resuspended (4 ϫ 10 7 cells/ml) in Buffer A (10 mM HEPES, pH 7.9, 10 mM KCl, 1.5 mM MgCl 2 , 0.34 M sucrose, 10% glycerol, 1 mM dithiothreitol, a 1:100 protease inhibitor mix (Sigma), 0.1 mM phenylmethylsulfonyl fluoride). 0.1% Triton X-100 was added, and the cells were incubated for 5 min on ice. Nuclei were collected in pellet 1 (P1) by low-speed centrifugation (4 min, 1,300 ϫ g, 4°C). The supernatant (S1) was further clarified by high-speed centrifugation (15 min, 20,000 ϫ g, 4°C) to remove cell debris and insoluble aggregates. Nuclei were washed once in Buffer A and then lysed in Buffer B (3 mM EDTA, 0.2 mM EGTA, 1 mM dithiothreitol, protease inhibitors as described above). Insoluble chromatin was collected by centrifugation (4 min, 1,700 ϫ g, 4°C), washed once in Buffer B, and centrifuged again under the same conditions. The final chromatin pellet (P3) was resuspended in Laemmli buffer and sheared by passing it through a syringe. To release chromatin-bound proteins by nuclease treatment, cell nuclei (P1) were resuspended in Buffer A plus 1 mM CaCl 2 and 0.2 unit of micrococcal nuclease (Sigma). After incubation at 37°C for 2 min, the nuclease reaction was stopped by the addition of 1 mM EGTA. Nuclei were collected by low-speed centrifugation and lysed as above (27).

RESULTS
Recent evidence has suggested that the C terminus of Xenopus and mouse Cdt1 interacts with the MCM2-7 complex. We first sought to confirm this result in human cells. A variety of truncations and internal deletions were made in the regions defined in mouse (9) and Xenopus (10) systems to be important for the interaction with MCM2-7. To determine the analogous amino acids in hsCdt1 (Homo sapiens) (546 amino acids), a multiple alignment was performed using ClustalW (28). In Xenopus egg extracts, xlCdt1⌬573-620 (X. laevis) did not bind MCM2-7. The analogous hsCdt1 residues (499 -546) were therefore deleted. Yeast two-hybrid and in vitro pulldown experiments with mouse Cdt1 suggested that mmCdt1-(407-477) (Mus musculus) was important for direct interaction with mmMCM6. Interestingly, mmCdt1⌬520 -577 (a deletion of the extreme C terminus) seemed to interact more strongly with mmMCM6, suggesting that this region may be inhibitory for MCM binding. We synthesized the analogous deletions of the internal required region and C-terminal inhibitory region in human Cdt1: hsCdt1⌬395-465 and hsCdt1⌬508 -546. Additional internal deletions were also made based on a secondary structure prediction using the GOR4 algorithm (29). The fourth-from-last helix (hsCdt1⌬444 -465), the third-from-last helix (hsCdt1⌬495-505), and the final helix (hsCdt1⌬518 -546) were each deleted. These mutations are summarized in Fig. 1.
To test the interaction of these hsCdt1 deletions with MCM2-7, we overexpressed FLAG-tagged versions of the proteins in human 293T cells. hsCdt1 was then immunoprecipitated using an anti-FLAG antibody, and hsMCM6 levels were examined by Western blotting. Wild type hsCdt1 was able to pull down endogenous hsMCM6, reproducing earlier results in different systems from other laboratories ( Fig. 2A). However, none of the deletion mutants pulled down detectable hsMCM6, indicating that a large portion of the C terminus of hsCdt1 is important for the interaction with hsMCM6. Both the wild type and deletion mutants of hsCdt1 were able to pull down geminin, indicating the proteins are likely to be folded correctly. Interestingly, both ⌬499 -546 and ⌬508 -546 did not interact with hsMCM6. In previous studies, the ⌬499 -546 Xenopus equivalent did not bind to MCM2-7 (10), but the mouse ⌬508 -546 equivalent bound more strongly (9), suggesting that the extreme C terminus of Cdt1 was inhibitory for MCM binding. Our data agrees with the Xenopus data, as the extreme C terminus appears to be required for MCM interaction. It is possible that the Cdt1-MCM6 interaction in mice is slightly different from that in other organisms.
To eliminate the possibility that these deleted regions are also important for a non-MCM binding function of hsCdt1, we  Each set of point mutants was made in highly conserved residues in the regions deleted previously (see Fig. 1). FLAG-tagged versions of these point mutants were expressed in 293T cells and pulled down with anti-FLAG. Although wild type hsCdt1 interacted with hsMCM6, we did not observe an interaction between any of our point mutants and hsMCM6 (Fig. 2B), suggesting that these mutations interrupt the binding between the two proteins. Again, the hsCdt1 point mutants all bind to geminin. As only 4 -5 amino acids have been changed in each mutant, the chance of disrupting another activity of Cdt1 in all three cases is decreased.
We tested the re-replication initiation activity of these point mutants by overexpressing them in human 293T cells. Our laboratory has shown previously that by either overexpressing hsCdt1 (18) or relieving its inhibition via knockdown of gemi-nin (20,30), re-replication can be induced in human cells. As expected, transient overexpression of wild type full-length hsCdt1 induced Ͼ4N DNA content as determined by FACS (Fig. 3A). In addition, overexpression of each of the MCM noninteracting point mutants also induces re-replication (Fig. 3A). The percentage of re-replicating cells after hsCdt1 point mutant overexpression is similar or higher than wild type hsCdt1 (quantitated in Fig. 3B). This result is very surprising, as these point mutants of hsCdt1 do not appear to interact with hsMCM6. This suggests that although Cdt1 is required for MCM2-7 loading onto chromatin, the interaction between these proteins is not required for re-replication induction.
Because the MCM non-interacting point mutants of Cdt1 were able to induce re-replication, we examined the ability of C-terminal deletion mutants to do the same. In previous Xenopus experiments, a C-terminal truncation neither interacted with MCM2-7 nor functioned in replication licensing. The analogous human mutant, residues 1-498, as well as two mutants with larger C-terminal deletions, residues 1-389 and 1-370, was tested for its ability to induce re-replication. Again, surprisingly, these MCM non-interacting mutants caused rereplication (Fig. 3, C and D). This further supports the notion that direct interaction between the overexpressed Cdt1 and MCM2-7 is not required for re-replication. Additionally, these results indicate that the entire C-terminal region, although the only part conserved from S. cerevisiae to humans, is not required to induce re-replication, suggesting that re-replication may not depend on the normal licensing role of Cdt1.
One possible explanation for the induction of re-replication by MCM non-interacting point mutants of Cdt1 is that the overexpressed Cdt1 is simply titrating geminin away from endogenous Cdt1, thereby activating it. We addressed this possibility by overexpressing the central domain of Cdt1 that contains a geminin-interacting region: hsCdt1-(163-370). This FLAG-tagged fragment is able to pull down equal amounts of geminin compared with wild type and point mutant hsCdt1, indicating it is able to titrate geminin just as efficiently (Fig. 4A). However, this fragment is not able to induce re-replication compared with wild type hsCdt1 (Fig. 4B); to ensure proper localization, this fragment and others with large N-terminal deletions have an artificial nuclear localization sequence added to the N terminus. This leads us to believe that the re-replication stimulation observed after overexpression of hsCdt1 is not due to geminin titration.
To further demonstrate that Cdt1 overexpression does not simply titrate geminin away from endogenous Cdt1, siRNA was used to knock down geminin followed by Cdt1 overexpression. Although re-replication is induced after geminin siRNA in some cell lines (20, 31), we did not observe significant re-replication after geminin siRNA in 293T cells (Fig. 4, C and D, FLAG  lanes). Taken on its own, this observation suggests that geminin depletion from endogenous Cdt1 is not sufficient to induce re-replication in this cell line. Wild type Cdt1 was also overexpressed in control or geminin siRNA-treated cells. Geminin depletion in Cdt1-overexpressing cells had very little effect on the percentage of re-replicating cells. (Fig. 4, C and D) Thus, Cdt1 overexpression induces re-replication by a mechanism other than titration of geminin.
It is possible that overexpressed Cdt1 may titrate another inhibitor of re-replication. CDK activity is also important for inhibition of re-replication as has been demonstrated recently (30). As Cdt1 contains a cyclin (Cy) binding motif that is required for interaction with cyclins, titration of cyclins from endogenous targets (including endogenous Cdt1) could cause re-replication. To test this, we deleted the Cy motif in both full-length Cdt1 and Cdt1-(1-370) (Fig. 4, E and F). Although the ⌬Cy mutants do not interact with cyclin A (Fig. 4G) (32), they are still able to induce re-replication, suggesting that titration of cyclins is not required for the observed re-replication following Cdt1 overexpression.
It is also interesting to note that although the 163-370 fragment of Cdt1 did not induce re-replication, the 1-370 fragment did. This indicates that a critical function of the N terminus of Cdt1 is required for re-replication. To further examine this phenomenon, we overexpressed the 163-546 fragment, in which only the N terminus is deleted. Surprisingly, we found that this fragment induced an S-phase arrest (Fig. 5, A and B). A slightly shorter N-terminal truncation, fragment 94 -546, also induced an S-phase arrest. Interestingly, if we disrupted MCM binding by overexpressing a Cdt1 mutant lacking the N terminus and lacking the C terminus or having C-terminal point mutations, the S-phase arrest was lost, and cells behaved like the control FLAG transfection (Fig. 5, A and B). This indicates not only that the N terminus has functions important for normal activity of Cdt1 but also that the cell cycle is inhibited when this fragment is absent. The cell cycle inhibition can be rescued when MCM interaction is disrupted, suggesting that N-terminally deleted Cdt1 inhibits the cell cycle by stably associating with and perturbing the normal function of MCM2-7.
As the Cdt1 N terminus seemed to be required for induction of re-replication, we wondered whether the N terminus was sufficient for re-replication. Cdt1-(1-163) was overexpressed and was able to induce re-replication nearly as well as fulllength Cdt1 (Fig. 5, C and D). This result suggests that the N terminus is sufficient for re-replication induction and strongly suggests that re-replication stimulation does not occur via the normal function of Cdt1. However, the N terminus does con-

HsCdt1 Induces Re-replication Without MCM Interaction
tain both motifs required for Cdt1 degradation. We have shown that removal of the Cy motif (required for CDK/SCF-mediated destruction of Cdt1) did not affect re-replication. It is possible that the N terminus is titrating away factors required for the PCNA/CUL4/DDB1-mediated degradation of Cdt1. We tested this by overexpressing Cdt1-(6 -546). As this fragment lacks part of the PCNA-interacting protein box motif, PCNA-mediated degradation is disrupted (25). This fragment was still able to induce re-replication, suggesting that titration of the PCNAmediated degradation apparatus alone is not required to induce re-replication (Fig. 5, C and D).
To understand how the C-terminally deleted Cdt1 that does not interact with MCM induces re-replication, we examined whether it was loaded on chromatin (Fig. 6A). The S1 fraction represents the soluble fraction after low-salt, mild-detergent lysis (data not shown). It is clarified by a high-speed spin to yield S2. After hypotonic buffer treatment of P1 (thought to be nuclei), soluble fraction S3 and pellet P3 are recovered. Proteins bound to chromatin would be released by micrococcal nuclease (MNase) from P3 to the S3 fraction. Orc2, a control chromatin-binding protein, shifts from P3 to S3 with MNase treatment. Interestingly, a fraction of the overexpressed Cdt1 deletion mutants are present in the chromatin-bound protein fraction just as Orc2 (Fig.  6A).
Because these mutant Cdt1 proteins stimulate re-replication and can be found on chromatin, we wondered whether they stimulated MCM loading on chromatin. A more stringent chromatin fractionation protocol was used to better visualize MCM chromatin loading. In this method, the S1 fraction contains proteins solubilized by medium-salt, medium-detergent buffer. P1 is then treated with MNase, releasing chromatin-bound proteins into the soluble S2 fraction. Proteins not solubilized by MNase treatment remain in the insoluble P2 fraction. We examined the MCM7 levels in the S2 fraction and found that the levels appear to be equal after Cdt1-(1-370) overexpression (Fig. 6B). However, MCM is released from chromatin as cells pass through S-phase, so a confounding factor is that more of the re-replicating cells are in S-phase and post-S-phase. Indeed, after Cdt1-(1-370) overexpression the percentage of G 1 cells is decreased to about half that of control cells (Fig. 6C), and chromatin-bound MCM levels are expected to be low. Nevertheless, the levels of chromatin-bound MCM are equal in the two populations, implying that chromatin loading of MCM is increased by the exogenous overexpressed Cdt1 mutant.
It is therefore possible that the exogenous Cdt1 that cannot interact with MCM is overactivating endogenous Cdt1 to load cellular MCM onto chromatin. We tested this by knocking down endogenous Cdt1 using siRNA in cells overexpressing FLAG-Cdt1 (residues 1-498), which does not interact with MCM. The overexpressed exogenous FLAG-Cdt1 was not detectably decreased by the siRNA, whereas the endogenous Cdt1 was decreased (Fig. 6E). Intriguingly, selective decrease of the endogenous Cdt1 decreases the Cdt1 (residues 1-498)stimulated re-replication (Fig. 6D). This indicates that overexpressed Cdt1 requires the presence of endogenous Cdt1 to induce re-replication and may induce re-replication by overactivating endogenous Cdt1 to load cellular MCM.
Earlier we observed that titration of PCNA or cyclin A by exogenous Cdt1 was not required for re-replication; preventing the titration of either cyclin A or PCNA did not disrupt re-replication induction. But what if titration of either one of these proteins, PCNA or cyclin A, alone were sufficient to induce re-replication? If this were true, then overexpressing a Cdt1 fragment mutated for both PCNA and cyclin interaction would not be able to stimulate re-replication. This possibility was tested by overexpressing the double mutant Cdt1-(6 -370)⌬Cy. Although the Cdt1-(6 -370) mutant induced re-replication, Cdt1-(6 -370)⌬Cy did not, indicating that titrating at least one of the two Cdt1-interacting proteins is required to induce rereplication (Fig. 7, A and B).
Titrating the destruction mechanisms seems to overactivate Cdt1. As both pathways degrade Cdt1, the inhibition of these pathways may stabilize endogenous Cdt1. This was tested by examining the endogenous Cdt1 levels after overexpressing Cdt1-(1-370), which induces re-replication. We transiently transfected cells with a Cdt1-overexpressing plasmid and a puromycin resistance plasmid, and 24 h after transfection we subjected them to puromycin selection for 48 h to enrich the Cdt1-transfected cells. 48 h of puromycin selection is enough to kill untransfected cells, and indeed we could see robust re-replication when all surviving cells were examined at this stage (Fig. 7D). However, endogenous Cdt1 levels were not increased after overexpression of Cdt1 (Fig. 7C), indicating that endogenous Cdt1 may be overactivated without an obvious increase in the level of total Cdt1.

DISCUSSION
Genomic stability is of paramount importance to a cycling cell. To protect genomic integrity, cells have evolved a complex regulatory mechanism to ensure the genome is copied exactly once per cell cycle. Cdt1 regulation is an important aspect of this mechanism. Its destruction and inhibition in S-phase prevents re-replication, protecting the genome. Perturbing this inhibition by overexpressing Cdt1 or preventing its degradation or geminin-mediated inhibition causes re-replication. However, the way in which Cdt1 overexpression causes re-replication is not clear. In this study we have shown that mutations in Cdt1 that disrupt physical interaction with MCM2-7 are still able to induce re-replication, suggesting that re-replication induction does not require this interaction or even the evolutionarily conserved part of Cdt1.
Although our results indicate that large regions of human Cdt1 may not be required for re-replication initiation, C-terminal deletions in Xenopus egg extracts are not able to function in licensing normal replication (10). This suggests that Cdt1 does not induce re-replication in human cells via its normal replication initiation function. Indeed, the N-terminal one-third of Cdt1 alone is able to stimulate re-replication, despite the fact that two-thirds of the protein (including the highly conserved C terminus) is missing. This result is rather puzzling and seems to contradict expectations that Cdt1 overexpression might cause re-replication by inappropriately recruiting MCM2-7 to chromatin via protein-protein interactions during S-phase.
If the stable, direct interaction between the C terminus of Cdt1 and MCM2-7 is not required for re-replication, how does exogenous Cdt1 function? One possibility is that the N-terminal part of exogenous Cdt1 is bridged to MCM2-7 via another protein and still functions to physically recruit MCM2-7 to chromatin despite the lack of the C-terminal interaction with MCM. It was reported recently that cyclin E protein, but not its  MARCH 14, 2008 • VOLUME 283 • NUMBER 11 associated CDK activity, is required for MCM2-7 loading in mouse but is not required for Cdt1 loading (33). The authors also found that cyclin E binds to both Cdt1 and MCM2-7, which would make it a candidate bridge protein linking Cdt1 to MCM2-7. However, ⌬Cy mutants lacking the cyclin-interacting motif still support re-replication, indicating that the cyclin E-Cdt1 interaction is not required to mediate an indirect interaction of Cdt1 with MCM (Fig. 4, E and F), although the Cy motif may play a different role in inducing re-replication (see below).

HsCdt1 Induces Re-replication Without MCM Interaction
There are two additional candidate proteins that could interact with Cdt1 and promote MCM loading. Cdt1 has been shown to physically interact with Cdc6 in several organisms. In addition, a recent paper reported that human Orc6 co-immunoprecipitates with Cdt1 (34). In theory, geminin could disrupt either of these interactions (35). However, we have found that the N-terminal portion of Cdt1-(1-163), although still able to induce re-replication, is dispensable for the interaction between Cdt1 and Cdc6 (data not shown). It is possible Orc6 is somehow involved, but we have been unable to co-immunoprecipitate Orc6 with Cdt1. 3 If Cdt1 does not induce re-replication in a positive fashion through its normal replication initiation activity, perhaps it acts in a negative fashion by inhibiting an S-phase inhibitor. Cdt1 interacts well with (and is inhibited by) the S-phase inhibitor protein geminin. It is possible that exogenous Cdt1 titrates geminin away from endogenous Cdt1, overactivating it. However, expression of the central domain of Cdt1 did not induce re-replication, despite the observation that it was able to titrate geminin (Fig. 4, A and B). Therefore, titration of geminin alone is not sufficient to induce re-replication. Exogenous Cdt1 could also titrate the destruction machinery away from endogenous Cdt1, preventing its degradation. In fact, a Cdt1 mutant missing both the PCNA and cyclin A binding motifs is not able to induce re-replication (Fig. 7, A and B). Titrating only one of the two mechanisms is sufficient for exogenous Cdt1 to stimulate re-replication. This explains the observation that mutating one interaction motif did not disrupt re-replication mediated by overexpressed Cdt1; titrating the other pathway (PCNA or cyclin A) was sufficient for re-replication. Only when titration of both PCNA and cyclin A was disrupted was re-replication prevented. This means that in this system, the inhibitory activity of both PCNA and cyclin A is required to prevent re-replication, suggesting that the two pathways are not redundant; they are both absolutely required for inhibiting endogenous Cdt1.
However, as Cdt1 is stabilized in S-phase only when both PCNA/Cul4 and cyclin A/Cul1-mediated pathways are inhibited (25), it seems surprising that titrating one alone could activate endogenous Cdt1. In fact, no increase in endogenous Cdt1 level is observed following overexpression of Cdt1-(1-370) (Fig. 7C). It is possible that titrating both degradation pathways stabilizes a subpopulation of endogenous Cdt1, thereby overactivating it. An increase in this special subpopulation will not be apparent when examining total cellular Cdt1 or even the chromatin-associated Cdt1 (data not shown). The other possibility is that Cdt1 is inhibited simply by the stable association with cyclin A and PCNA. The exogenous Cdt1 needs to titrate either of these proteins to disrupt inhibition by stable association, allowing the overactivation of endogenous Cdt1 without extensive stabilization. Although it is also possible that titration of the two destruction mechanisms stabilizes or activates another positive replication factor, endogenous Cdt1 is required for rereplication (Fig. 6D), so it is the likely target for overactivation. We imagine a model in which exogenous Cdt1 titrates PCNA and/or cyclin A away from endogenous Cdt1. This disrupts the normal inhibition of Cdt1 during S-phase, causing an inappropriate increase in MCM2-7 loading and resulting in re-replication (Fig. 7E).
Our results suggest that sustained interaction between overexpressed Cdt1 and MCM2-7 is detrimental to DNA replication if the degron in the N-terminal 94 residues of Cdt1 is deleted. We have shown previously that overexpression of nondegradable Cdt1 delays progression through S-phase in HeLa cells (36), and this result is reproduced in 293T cells, where Cdt1-(94 -546) or Cdt1-(163-546) caused a pronounced S-phase accumulation. This accumulation was greatly diminished upon introduction of mutations that disrupt the stable association of MCM2-7 with Cdt1 (RPLVF or ⌬499-C). Thus, 3 A. Dutta and T. Abbas, unpublished observations. although Cdt1 is required to recruit MCM2-7 to origins in G 1 , and although S. cerevisiae Cdt1 (TAH11) is in a stable complex with MCM2-7 (7), persistent interaction between non-degradable Cdt1 and MCM2-7 in S-phase leads to problems in cell cycle progression.
The role of Cdt1 in re-replication is interesting and important. In higher eukaryotes, many pathways converge upon Cdt1 to prevent re-replication, including degradation in S-phase by two distinct pathways and inhibition by geminin. Cdt1 levels are often elevated in tumors (23), and overexpression of Cdt1 in mice can lead to tumors (24). This guilt by association and causality suggests that high levels of Cdt1 may be one pathway to cancer progression. Increased levels of Cdt1 in tissues would induce re-replication, eventually resulting in increasing genomic instability and oncogenic potential. We have shown that this induction of re-replication can occur in a variety of mutants of Cdt1, many of which would not be expected to be functional for replication initiation. Additionally, we do not see any more re-replication when wild type Cdt1 is overexpressed compared with a C-terminal truncation, which supports the notion that this exogenous Cdt1 is not inducing re-replication through its normal function. One might therefore imagine that mutations allowing overexpression of Cdt1 in cells would be oncogenic even if other loss-of-function mutations existed in the central or C-terminal domains of the gene. Because re-replication is seen despite the failure to see an increase in the level of endogenous full-length Cdt1, the other implication from the study is that Cdt1 may be overactivated and lead to genome instability in some cancers where there is not an overt increase in Cdt1 protein.