Cellular Checkpoint Mechanisms Monitoring Proper Initiation of DNA Replication*

The most basic function of the eukaryotic cell cycle is to duplicate accurately the vast amount of DNA in the chromosomes and then segregate the copies into two daughter cells. These cell cycle events are initiated sequentially at the appropriate times during the cell cycle by successive activation of different types of complexes of cyclin and cyclin-dependent kinase (Cdk). The integrity of each cell cycle event is monitored by checkpoint mechanisms that prevent initiation of the later event until the earlier event is complete (1). Chromosomes have to be replicated not only accurately but also once and only once per cell cycle. Therefore, replication initiation must be under strict control so that origins do not fire more than once in a cell cycle. Studies in different organisms show that eukaryotic cells have developed redundant control mechanisms that prevent re-replication. Checkpoint proteins are activated when re-replication is induced. This suggests a potential role for checkpoint proteins in regulation of the cell cycle and DNA replication in response to re-replication. In addition, recent studies show that perturbation of replication initiation proteins arrests the cell cycle in G1 with low Cdk activity in higher eukaryotes, suggesting the existence of a checkpoint mechanism that monitors replication-competent origins at G1/S. In this review, we explore cellular checkpoint mechanisms that ensure genome integrity by monitoring proper initiation of DNA replication.

The most basic function of the eukaryotic cell cycle is to duplicate accurately the vast amount of DNA in the chromosomes and then segregate the copies into two daughter cells. These cell cycle events are initiated sequentially at the appropriate times during the cell cycle by successive activation of different types of complexes of cyclin and cyclin-dependent kinase (Cdk). 1 The integrity of each cell cycle event is monitored by checkpoint mechanisms that prevent initiation of the later event until the earlier event is complete (1). Chromosomes have to be replicated not only accurately but also once and only once per cell cycle. Therefore, replication initiation must be under strict control so that origins do not fire more than once in a cell cycle. Studies in different organisms show that eukaryotic cells have developed redundant control mechanisms that prevent re-replication. Checkpoint proteins are activated when re-replication is induced. This suggests a potential role for checkpoint proteins in regulation of the cell cycle and DNA replication in response to re-replication. In addition, recent studies show that perturbation of replication initiation proteins arrests the cell cycle in G 1 with low Cdk activity in higher eukaryotes, suggesting the existence of a checkpoint mechanism that monitors replication-competent origins at G 1 /S. In this review, we explore cellular checkpoint mechanisms that ensure genome integrity by monitoring proper initiation of DNA replication.

Initiation of DNA Replication
In eukaryotic cells, initiation of DNA replication is achieved by an ordered assembly of protein complexes at origins of replication. This process consists of two steps, licensing of origins in late M or early G 1 phase and firing of origins at different times during S phase. Origin licensing is mediated by the assembly of pre-replicative complexes (pre-RCs) on origins. At the center of the pre-RC is the origin recognition complex (ORC), a six-subunit complex (Orc1-Orc6), which binds origins of replication (2). During late M and early G 1 phase, Cdt1 and Cdc6 bind origins in an ORC-dependent manner and then cooperatively recruit a putative helicase, Mcm2-7 complex, to origins (2) (Fig. 1). Once pre-RCs are assembled, origins are licensed for replication in the subsequent S phase and ready to fire. A key event for origin firing, the second step of the initiation process, is loading of Cdc45 on origins (Fig. 1). This step is triggered by the action of two kinases, cyclin-Cdk and Cdc7-Dbf4. Although cyclin-Cdk complex works as a global activator for S phase, Cdc7-Dbf4 acts as an activator in the initiation step at the individual origins. Following Cdc45 loading on origins, the single-stranded DNA (ssDNA)-binding protein, RPA, and the primase-DNA polymerase ␣ complex are loaded on chromatin to initiate DNA replication (2).

Mechanism for Preventing Re-replication
Re-replication is prevented by strict control of the licensing step. The central player in this process is cyclin-Cdk (3). High Cdk activity prevents reinitiation of DNA replication by blocking reassembly of pre-RCs during S, G 2 , and M phases. Therefore, pre-RC formation is restricted to G 1 phase where Cdk activity is low. Inhibition of pre-RC assembly by the same kinase that triggers replication initiation explains how origins are prevented from refiring until the kinase is inactivated at the end of M phase. Components of pre-RC are substrates of Cdk, and their phosphorylation has negative effects on new pre-RC assembly (Fig. 1). In yeasts, phosphorylation of Cdc6 by Cdk promotes its ubiquitinylation by SCF complex and degradation by proteasome (4 -9). Phosphorylation of ORC subunits by Cdk prevents chromatin-loading of Mcm2-7 (10,11). Nuclear export of Mcm2-7 and Cdt1 is also promoted by Cdk in Saccharomyces cerevisiae, leading to their exclusion from the nucleus (12,13). In higher eukaryotes, some of these mechanisms seem to be conserved. Phosphorylation of metazoan Cdc6 promotes its nuclear export in S phase instead of degradation, although a fraction of Cdc6 remains nuclear (14 -19). Cdk-dependent phosphorylation of human Cdt1 triggers its ubiquitinylation and subsequent degradation by proteasomes in S phase (20 -22). Xenopus Mcm4 has been shown to be a substrate of Cdk, and hyperphosphorylated Mcm complexes cannot load on chromatin (23). These facts suggest that Cdk-mediated inhibition of pre-RC formation is a conserved mechanism to suppress re-replication in eukaryotes.
Higher eukaryotes have evolved a Cdk-independent mechanism to prevent re-replication. Geminin, first identified as a substrate of anaphase-promoting complex (APC), is an inhibitor of pre-RC formation (24) (Fig. 1). Geminin binds Cdt1 and inhibits chromatin loading of Mcm2-7 complexes (25,26). Protein levels of geminin are high during S, G 2 , and M phases, where pre-RC assembly is restricted. Upon activation of APC in M phase, geminin is degraded to allow new pre-RC assembly for the next S phase. Once APC becomes inactive in S phase, geminin is stabilized and inhibits further formation of pre-RCs.
In many organisms, perturbation of the mechanisms preventing new pre-RC formation after S phase entry leads to re-replication. In fission yeast, overexpression of a Cdc6 homolog, Cdc18, is sufficient to cause re-replication (27,28). In contrast, overexpression of Cdc6 does not simply result in rereplication in budding yeast (4,29). To induce re-replication in this organism, at least three mechanisms preventing re-replication have to be relieved simultaneously: Cdc6 degradation, nuclear export of Mcm2-7, and ORC phosphorylation (10). Interestingly, overexpression of Cdt1, but not Cdc6, is sufficient to induce re-replication in human cells (30). This appears to be through the accelerated Cdt1 function in pre-RC formation because co-expression of geminin suppresses re-replication induced by Cdt1 (30). In Caenorhabditis elegans inactivation of Cul-4 causes Cdt1 accumulation and results in re-replication (31). Disappearance of the worm Cdt1 during S phase was impaired in the absence of Cul-4, indicating an important role for Cul-4 in regulation of Cdt1 protein levels. Re-replication phenotype was rescued by a hemizygote of Cdt1, supporting the idea that re-replication was caused by increased Cdt1 protein levels. It is worth noting that Cdt1 was not overexpressed ectopically in this experiment. Impairment of Cdt1 degradation during S phase caused massive re-replication, suggesting a central role for Cdt1 degradation in preventing re-replication. Consistent with the major role of Cdt1 control in preventing re-replication in higher eukaryotes is that RNAi against geminin causes re-replication in human and Drosophila cells (32)(33)(34). This fact suggests that geminin is not just a backup mechanism to restrict Cdt1 function after S phase entry. Cdt1 protein is still degraded in geminin RNAi-treated cells (32)(33)(34), supporting the notion that geminin is critical to suppress rereplication before Cdt1 is degraded. The major role of Cdt1 regulation in the re-replication control in higher eukaryotes is clearly different from the situation in fission yeast, where overexpression of Cdc18 but not Cdt1 is sufficient to override rereplication control. It appears that the amount of Cdt1 protein became more critical in re-replication control in higher eukaryotes. This is consistent with the fact that higher eukaryotes have evolved geminin as a regulator of Cdt1 function.

Re-replication Activates G 2 /M Checkpoint
Integrity of DNA replication during S phase is monitored by checkpoint mechanisms in eukaryotes. If replication forks are stalled during S phase, mitotic entry is blocked by G 2 /M checkpoint until DNA replication is complete. ssDNA created at stalled forks recruits a checkpoint kinase ATR (ATM-and Rad3-related) and Rad17-Rfc2-5, which in turn loads the Rad9-Rad1-Hus1 complex, to initiate the checkpoint pathway (35). Activated ATR stimulates activation of a checkpoint kinase, Chk1, which transduces checkpoint signal to the cell cycle machineries. One of the critical targets of this checkpoint is a dual specificity phosphatase, Cdc25C, which is sequestered in cytoplasm after phosphorylation by Chk1. As a result, the phosphorylation of mitotic Cdk at Tyr-15 is increased, and entry into mitosis is inhibited (35). Recently, it was also reported that Cdc6 plays a role in coupling of DNA replication and the following mitosis in human cells (36). DNA damage by UV or alkylating agents such as methyl methanesulfonate activates a similar checkpoint pathway to create time to repair those lesions before mitotic entry (35). On the other hand, damage response to double-stranded DNA breaks (DSBs) is primarily mediated by another checkpoint kinase, ATM (ataxia telangiectasia-mutated), and results in the activation of Chk2 in mammalian cells (35).
Activation of checkpoint proteins was observed in re-replicating cells (30,(32)(33)(34). Re-replication induced by Cdt1 overexpression or geminin RNAi results in the activation of Chk1 and Chk2 in human cells (30,32,34). Chk1 activation after geminin depletion was also reported in Drosophila and Xenopus cells (33,37). There are two possible mechanisms for checkpoint activation after re-replication. The first possibility is that re-replication produces DSBs and activates ATM-mediated DNA damage checkpoint. DSBs could be generated by abnormal tension of aberrant structure of re-replicated DNA. Alternatively, double-stranded DNA ends mimicking DSBs might be created when the second replication fork catches up with the previously formed fork with unligated Okazaki fragments. Although DSBs were not directly detected using a comet assay in Cdt1-overexpressed human cells, 2 activation of Chk2, which preferentially mediates signals from DSBs, suggests that rereplication creates low levels of DSBs and results in activation of the DNA damage checkpoint. Consistent with this is foci formation of Rad51, a protein involved in homologous recombination during DSB repair, in re-replicating cells (34). The second possibility is that re-replication causes fork stalling and activates checkpoint proteins. Re-replication eventually ceases after geminin depletion (32), suggesting that forks are stalled after re-replication. Indeed, ssDNA, which is sensed by the ATR-Chk1 pathway after fork perturbation, was detected in re-replicating cells (34).
Checkpoint proteins activated in response to re-replication by geminin RNAi cause G 2 /M arrest in Xenopus and human cells (32,34,37). Overexpression of a dominant-negative form of Chk1, treatment with an inhibitor of Chk1 (UCN-01), or depletion of Chk1 by RNAi can abolish the cell cycle arrest (32,34,37). These data suggest that Chk1 activation is critical for inhibition of mitotic entry in re-replicating cells. Abrogation of Chk1 function results in a decrease in re-replicating cells after geminin RNAi (32,33). Without checkpoint activation, re-replicating cells undergo premature mitosis and subsequent cell death (32,34). Cdc25C was phosphorylated at Ser-216 and sequestered in cytoplasm after re-replication in human cells (32). Overexpression of a mutant Cdc25C lacking a Chk1 phosphorylation site overcomes the G 2 /M arrest caused by geminin depletion in Xenopus, suggesting that Cdc25C is a critical Chk1 target in the G 2 /M checkpoint pathway in response to re-replication (37). Consistent with this, the cell cycle arrest is accompanied by accumulation of cyclin B1 and an increase in inhibitory phosphorylation at Tyr-15 of Cdk1 (32, 34, 37). Over-2 Y. Machida and A. Dutta, unpublished data.

FIG. 1. Initiation of DNA replication in eukaryotes.
In G 1 phase, ORC recruits Cdc6, Cdt1, and Mcm2-7 complexes to origins of DNA replication to form pre-RCs. Loading of Cdc45, a key step of origin firing, is triggered by two kinases, cyclin E-Cdk2 and Cdc7-Dbf4. Once origins are unwound, RPA and various DNA polymerases are loaded on origins to initiate DNA replication. During S phase, Cdk-mediated phosphorylation of pre-RC components prevents re-formation of pre-RCs on origins that have already fired. Geminin is stabilized upon S phase entry and inhibits pre-RC formation by interaction with Cdt1. expression of Cdk1AF, a mutant Cdk1 lacking inhibitory phosphorylation sites, bypasses the cell cycle arrest in Xenopus after re-replication (37). In summary, re-replication results in activation of the G 2 /M checkpoint pathway to inhibit mitotic entry with overreplicated chromosomes (Fig. 2).

Inhibition of Re-replication by Checkpoint Proteins
Several lines of evidence suggest a role for checkpoint pathways in blocking re-replication. Treatment of cells with an ATM/ATR inhibitor, caffeine, or with a Chk1 inhibitor, UCN-01, promotes re-replication by Cdt1 overexpression in human cells, 3 suggesting that checkpoint proteins have an inhibitory effect on re-replication. Deletion of one of the checkpoint proteins, Rad17, in human cancer cells results in re-replication (38), indicating that checkpoint pathways involving Rad17 monitor and suppress spontaneous re-replication. Interestingly, overexpression of Cdt1 and Cdc6 causes re-replication in p53-deficient but not p53ϩ human cells (30). In p53ϩ cells, p53 is stabilized, and its downstream targets including p21 and PIG3 are induced (Fig. 2). Overexpression of Mdm2, an E3 ubiquitin ligase for p53, in p53ϩ cells allows re-replication by Cdt1 and Cdc6 overexpression, suggesting that p53 has a role in preventing re-replication. The best candidate for re-replication block among p53 targets is p21. It is possible that p21 suppresses re-replication through inhibition of Cdk2, which is required for new origin firing (Fig. 2). p53 may also induce apoptosis through PIG3 induction (Fig. 2). On the other hand, geminin depletion causes re-replication regardless of p53 status (32,34). The difference could be due to the overexpression of Cdc6 in the experiments that induced p53 or due to a direct role of geminin in directing the checkpoint enzymes to p53 so that when geminin is depleted, the checkpoint enzymes fail to phosphorylate and stabilize p53.
Another possible pathway by which checkpoint proteins could suppress re-replication is through activation of intra-S phase checkpoint. One of the important functions of intra-S phase checkpoint is to prevent firing of late origins in response to fork stalling or DNA damage during S phase (35). Like other type of checkpoints, the intra-S phase checkpoint pathway includes activation of ATM/ATR and Chk1/Chk2 checkpoint kinases. Genetics in yeasts and biochemistry using Xenopus extracts have accumulated evidence that Cdc7-Dbf4 is one of the targets of the intra-S phase checkpoint pathway to prevent further firing of origins (35). Given that re-replication is a process with repeated firing of the same origins in an S phase, it is possible that the intra-S phase checkpoint pathway is utilized to suppress further reinitiation of replication when re-replication is detected (Fig. 2). To test if checkpoint proteins are involved in preventing re-replication, it will be interesting to see whether checkpoint mutant strains of Schizosaccharomyces pombe or S. cerevisiae tend to have more re-replication after overexpression of Cdc18 or mutations that relieve the Cdk-mediated repression of Orc, Cdc6, and Mcm.
If checkpoint proteins can suppress re-replication, why can they not suppress re-replication after Cdt1 overexpression and geminin depletion? One possibility is that Cdt1 and geminin are targets of checkpoint pathways in response to re-replication; therefore Cdt1 overexpression and geminin depletion override the checkpoint effect. Recently, it was reported that Cdt1 is rapidly degraded by the ubiquitin-proteasome pathway after DNA damage (39). If Cdt1 degradation is one of the mechanisms to suppress re-replication, it could explain why cells cannot suppress re-replication caused by Cdt1 overexpression. Similarly it is possible that geminin is a target of the checkpoint pathway and plays a role in suppressing further re-replication. This model could explain why geminin depletion results in rereplication despite the activation of checkpoint proteins. Overexpression of Cdt1 or geminin depletion is not a normal situation and might cause too robust re-replication that cannot be suppressed by cellular checkpoint mechanisms. Therefore it will be interesting to see whether checkpoint proteins have roles in preventing spontaneous re-replication during a normal S phase by a careful study of cells with inhibitors of checkpoint enzymes.

G 1 /S Checkpoint Preventing Premature Entry into S Phase
Events in the eukaryotic cell cycle are ordered into dependent steps in which the initiation of late events is dependent on the completion of early events. This dependence is not due to an intrinsic feature of the events themselves but to checkpoint mechanisms (1). The clearest example of the checkpoint is the dependence of mitotic entry upon completion of DNA replication (Fig. 3). Cells defer activation of mitotic cyclin-Cdk until DNA replication is complete. Another example is the spindle checkpoint, which ensures that anaphase does not begin until all chromosomes are connected to the spindle and aligned on the metaphase plate (40) (Fig. 3). By analogy to these mechanisms, an important question will be whether S phase entry is dependent on pre-RC formation in G 1 (Fig. 3). Because Cdk activation prevents further formation of pre-RCs once cells enter S phase, it is reasonable to expect that there might be a licensing checkpoint, a mechanism ensuring that Cdk becomes active only after enough pre-RCs are formed. Without such a mechanism, premature activation of Cdk in G 1 phase could trigger replication initiation from fewer origins. Indeed, deregulation of Cdk in G 1 phase was shown to impair efficient pre-RC formation and cause chromosome instability in yeast and human cancer cells (41,42).
Studies in yeasts and Drosophila cells suggest that there is no such checkpoint mechanism in G 1 . In yeasts, null mutants of 3 W. Zhu and A. Dutta, unpublished data.
FIG. 2. Checkpoint pathways activated by re-replication. Rereplication activates ATM/ATR, and the signal is transduced by Chk1/ Chk2. Cdc25C is sequestered in cytoplasm after phosphorylation by Chk1/Chk2, and as a result, mitotic entry is blocked through Cdk1 inhibition. Activation of p53 induces downstream targets including p21 and PIG3. p21 may play a role in re-replication block by inhibition of Cdk2, which is required for new origin firing. Induction of PIG3 may induce apoptosis in response to re-replication. A possible pathway, which blocks re-replication through Cdc7-Dbf4 inhibition, is also shown.
FIG. 3. Cell cycle checkpoints. Checkpoint pathways suppress initiation of the next events by inhibiting cell cycle regulators until completion of previous events. G 2 /M checkpoint inhibits cyclin B-Cdk1 until DNA replication is complete. Spindle checkpoint prevents activation of APC until all chromosomes are attached to spindles. Recent data suggest that there is a checkpoint-like mechanism at G 1 /S, which blocks activation of cyclin E-Cdk2 until enough pre-RCs are formed.
pre-RC members, Cdc6/Cdc18, Cdt1, and Orc1, are defective in DNA replication initiation and therefore lethal (43)(44)(45)(46)(47). Examination of the spores lacking Cdc18, Cdt1, and Orc1 revealed that these cells undergo mitosis without DNA replication leading to cells with fractional DNA content and fragmented chromosome. Because generation of checkpoint signal requires formation of DNA replication forks, mutants of replication initiation genes cannot generate the inhibitory signal necessary to stop entry into mitosis. In contrast, conditional yeast mutants of replication initiation proteins show a different phenotype in the restrictive conditions. Haploid cells carrying a temperature-sensitive mutation in Cdc18 arrest with a 2C DNA content (44). Pulse-field gel electrophoresis revealed that these cells have not completed replication. This result suggests that the cells could initiate DNA replication from a fraction of origins, but they could not complete DNA replication. Mitotic entry was successfully inhibited in these cells because cells have created enough replication forks to generate a checkpoint signal. Therefore, it seems that premature entry of cells into mitosis is dependent on the extent of impairment of initiator protein function in yeasts. In Drosophila, a null mutant of DUP, a Cdt1 homolog, is also defective in DNA replication (48). Cells are arrested in an abnormal mitosis with overcondensed and fragmented chromosomes. Importantly, these cells are positive for a mitotic marker, phosphohistone H3, suggesting that DUP Ϫ cells undergo premature mitosis. This is similar to the null phenotype of initiation protein mutants in yeasts. A similar result was obtained in geminin-overexpressing Drosophila cells. These cells are deficient in DNA replication and undergo premature mitosis and apoptosis (49). In addition chromosome condensation was abnormal in Drosophila mutants of replication proteins Orc2, Orc5, Mcm4, and proliferating cell nuclear antigen (50,51). Interestingly, the situation seems different in higher eukaryotes. Orc2 RNAi in human cells impairs DNA replication and arrests them in G 1 phase with low cyclin E-Cdk2 activity. 4 Blocking pre-RC assembly by overexpression of a stable form of geminin in primary cells affects DNA replication, and Cdk is inhibited based on the appearance of hypophosphorylated pRB (52). These results suggest that there is a mechanism that prevents Cdk activation until enough replication-competent origins are formed in G 1 . At this moment, it is not clear whether pre-RC formation is a signal for S phase entry. The Cdk inhibitor, p27, is stabilized after Orc2 depletion and associated with the cyclin E-Cdk2 complex in human cells. 4 In Xenopus egg extracts, replication-coupled degradation of Xic1, a homolog of p27, was reported (53,54). Xic1 was degraded at origins of replication, and this degradation was coupled with replication initiation. These results suggest a possible role for p27 as an effector in this checkpoint-like response to prevent progression of the cell cycle when there are insufficient initiator proteins.

Concluding Remarks
Initiation of DNA replication is under strict control and coordinated with other cell cycle processes by checkpoint pathways. Regulation of DNA replication and cell cycle progression is critical for maintenance of genome integrity. Therefore, an interesting future question will be whether impairment of cellular mechanisms that normally ensure proper initiation of DNA replication is a potential cause of human tumors. For example, sporadic re-replication might contribute to gene amplification or polyploidy seen in many types of cancer cells. High levels of cyclin E, reduced levels of p27, and impairment of pRB seen in several cancers might override the checkpoint at the G 1 /S transition and cause deletion or loss of chromosomes.
Continued study of the molecular network of the cell cycle regulators, replication machinery, and checkpoint proteins will help us understand how perturbation of replication control can cause genome instability in cancer cells.