Prevention of DNA re-replication in eukaryotic cells

Abstract

DNA replication is a highly regulated process involving a number of licensing and replication factors that function in a carefully orchestrated manner to faithfully replicate DNA during every cell cycle. Loss of proper licensing control leads to deregulated DNA replication including DNA re-replication, which can cause genome instability and tumorigenesis. Eukaryotic organisms have established several conserved mechanisms to prevent DNA re-replication and to counteract its potentially harmful effects. These mechanisms include tightly controlled regulation of licensing factors and activation of cell cycle and DNA damage checkpoints. Deregulated licensing control and its associated compromised checkpoints have both been observed in tumor cells, indicating that proper functioning of these pathways is essential for maintaining genome stability. In this review, we discuss the regulatory mechanisms of licensing control, the deleterious consequences when both licensing and checkpoints are compromised, and present possible mechanisms to prevent re-replication in order to maintain genome stability.

Introduction

To ensure genome stability, DNA must be replicated once and only once during each cell cycle. Additional rounds of replication of genomic DNA, or even sections of it, within a given cell cycle would result in gene amplification, polyploidy and other kinds of genome instability, which is a hallmark of tumorigenesis (Schimke et al., 1986; Albertson, 2006; Hook et al., 2007; Cook, 2009). The complex control of DNA replication initiation in eukaryotic organisms is highly conserved in regards to the DNA licensing factors, which include the origin of replication complex (ORC) proteins, Cdc10-dependent transcript factor 1 (Cdt1), and cell division cycle 6 (Cdc6), as well as their conserved activities to help initiate DNA replication via the minichromosome maintenance (MCM) complex (Blow and Dutta, 2005; Arias and Walter, 2007; Kim and Kipreos, 2007b). It is imperative for cells to tightly regulate Cdt1 and Cdc6 activities to prevent re-initiation of DNA replication in S and G2 phases so that DNA is replicated only once during a given cell cycle (Blow and Dutta, 2005; Fujita, 2006; Arias and Walter, 2007; Hook et al., 2007). To this end, multiple distinct and redundant pathways function to control Cdt1 expression and activity, in particular mechanisms of proteolytic degradation of Cdt1 as a means to prevent Cdt1-induced re-replication (Blow and Dutta, 2005; Fujita, 2006; Arias and Walter, 2007).

It has been described that loss of the licensing control can induce re-replication and activate the cell cycle checkpoints (Hook et al., 2007; Cook, 2009). Re-replication leads to the accumulation of single-strand DNA (ssDNA) and the formation of DNA double-strand breaks (DSBs) (Davidson et al., 2006; Liu et al., 2007), and these DNA lesions activate the ataxia telangiectasia and Rad3 related (ATR) and ataxia telangiectasia mutated (ATM) checkpoint pathways (Cook, 2009). Cells with intact checkpoints can prevent further re-replication and arrest cell cycle or induce cell death, thus suppressing the harmful effects caused by re-replication (Hook et al., 2007; Liu et al., 2007). Compromising both licensing control and checkpoint pathways has been observed in various kinds of tumors, suggesting that DNA re-replication promotes genome instability and tumorigenesis, especially in the absence of functional checkpoint control.

The control of DNA replication to ensure one round of DNA replication per cell cycle

The dynamic control of DNA replication initiation is mediated by a two-step mechanism: (i) formation of the pre-replicative complex (pre-RC) comprised of ORC1–6 complex, Cdc6, Cdt1 and MCM2–7 complex in late mitosis and early G1, and (ii) activation of MCM2–7 complex to initiate origin firing and DNA replication during S phase (Bell and Dutta, 2002; Arias and Walter, 2007; Drury and Diffley, 2009). Formation of the pre-RC complex occurs as a sequential assembly of the licensing factors Cdt1 and Cdc6 onto ORC-bound chromatin and the recruitment of the MCM2–7 complex (Figure 1; Cocker et al., 1996; Donovan et al., 1997; Maiorano et al., 2000; Nishitani et al., 2000; Rialland et al., 2002; Randell et al., 2006; Chen et al., 2007). Direct association of Cdt1 with several components of the MCM2–7 complex involves both the N-terminus and C-terminus of Cdt1, where the very extreme C-terminus contains a conserved MCM complex binding motif that binds MCM6 (Yanagi et al., 2002; Ferenbach et al., 2005; Teer and Dutta, 2008; Khayrutdinov et al., 2009; Jee et al., 2010; Wei et al., 2010). Additional studies to further clarify the mechanism for Cdt1-mediated recruitment of Mcm2–7 to chromatin revealed the requirement for MCM9, which directly binds to Cdt1 and forms a stable complex to promote the interaction between Cdt1 and MCM2–7 complex. Thus, MCM9 is suggested to function as a ‘colicenser’ of DNA replication with Cdt1 (Lutzmann and Mechali, 2008). In vitro analysis suggested that the association of Cdt1 and Cdc6 with origins is reduced when MCMs are loaded onto origins (Tsakraklides and Bell, 2010).

Figure 1.

Regulation of DNA replication during a cell cycle. At the end of the mitosis and G1, Geminin is degraded, and Cdt1 and Cdc6 recruit MCM2–7 to ORC-bound origins to establish the pre-RC complex. At the onset of S-phase, CDKs and Cdc7/Dbf4 kinase (DDK) establish the pre-IC complex by recruiting Cdc45 and the GINS complex to MCM2–7. With Cdc45 and GINS as accessory factors, MCM2–7 unwinds DNA, followed by recruitment of replication machinery to start DNA replication. As MCM2–7 moves away from origins, pre-RCs are disassembled. In S and G2, Geminin is expressed to inhibit Cdt1 activity and Cdt1 is also subject to proteasomal degradation, while Cdc6 is exported to cytoplasm. These mechanisms prevent reassembly of pre-RC complex on already fired origins and subsequent re-replication within one cell cycle.

Although the MCM2–7 complex is recruited to form the pre-RC complex in G1, activation of MCM proteins by the Cyclin-Dependent Kinase (CDK), Cdk2, and the Dbf4-Dependent Kinase (DDK), Cdc7, to establish the pre-initiation complex (pre-IC) does not actually occur until the onset of S phase (Bell and Dutta, 2002; Arias and Walter, 2007; Drury and Diffley, 2009). At the G1/S transition, Cdk2 and Cdc7-mediated phosphorylation events, along with the activities from other players such as MCM10, are required for recruiting Cdc45 and GINS onto MCM2–7, which activate MCM2–7 and promote its DNA helicase activities (Wohlschlegel et al., 2002; Pacek and Walter, 2004; Gambus et al., 2006; Moyer et al., 2006). Subsequent recruitment of RPA, DNA polymerase α, RFC, PCNA, and DNA polymerase δ initiates DNA replication (Waga and Stillman, 1998). In addition to its role to recruit MCM2–7 to chromatin for establishing pre-RCs, recent studies demonstrate that Cdt1 also participates in the activation of MCM2–7 complex by directly associating with Dbf4 of the Cdc7–Dbf4 complex contributing to the recruitment of Cdc45 to MCM2–7 (Ballabeni et al., 2009). Furthermore, Cdt1 was shown to stimulate MCM2–7 helicase activity through in vitro gel shift assays combined with DNA helicase assay, suggesting a role for Cdt1 to promote efficient DNA unwinding during replication (You and Masai, 2008). Once DNA replication is initiated, the pre-RCs are disassembled as MCM2–7 moves away from the origins, releasing Cdt1 and Cdc6 from the origins, which are subject to proteolytic degradation and nuclear export [(Aparicio et al., 1997; Blow and Dutta, 2005; Arias and Walter, 2007), Figure 1]. These mechanisms are thought to help prevent loading of de novo MCM2–7 complex onto the already fired origins to reassemble pre-RCs leading to DNA re-replication (Blow and Dutta, 2005; Arias and Walter, 2007; Xouri et al., 2007).

The regulation of replication licensing factors during the cell cycle

In order to maintain proper licensing control, various pathways are utilized to tightly regulate the expression and activity of the licensing factors during the cell cycle so that unscheduled DNA licensing is prevented in S and G2 after DNA replication is initiated (Blow and Dutta, 2005; Arias and Walter, 2007; Cook, 2009). In Saccharomyces cerevisiae, Cdc6 is degraded, and both Cdc6 and Orc6 are inhibited by binding to Clb proteins (Drury et al., 1997; Mimura et al., 2004; Wilmes et al., 2004), while Cdt1 is primarily inhibited by nuclear export with MCM2–7 in a CDK-dependent manner (Nguyen et al., 2000, 2001; Arias and Walter, 2007). In contrast, Schizosaccharomyces pombe utilizes proteolytic degradation to regulate both Cdc6 and Cdt1 in S and G2, in which Cdt1 is degraded through the Cul4–Ddb1–Cdt2 E3 ubiquitin ligase complex (Jallepalli et al., 1997; Gopalakrishnan et al., 2001; Hu and Xiong, 2006; Ralph et al., 2006). Proteolytic degradation of Cdt1 is a conserved regulatory mechanism that extends to higher eukaryotes including Caenorhabditis elegans, Drosophila, Xenopus, and mammals (Arias and Walter, 2007; Cook, 2009).

Higher eukaryotes also negatively regulate Cdt1 via Geminin, which binds to and sequesters Cdt1 on chromatin during S and G2, thus inhibiting Cdt1 association with MCM2–7 and preventing pre-RC reassembly within one cell cycle (Wohlschlegel et al., 2000; Tada et al., 2001; Lee et al., 2004; Maiorano et al., 2004). The significance of Geminin-mediated Cdt1 inhibition is evident by data demonstrating that loss of Geminin alone is sufficient to induce DNA re-replication (Mihaylov et al., 2002; Zhu et al., 2004; Melixetian et al., 2004; Li and Blow, 2005; Kerns et al., 2007). Interestingly, additional studies also suggest a positive role of Geminin to promote Cdt1-mediated Mcm2–7 chromatin loading in G1. Although exact mechanisms which regulate these ‘licensing permissive’ Cdt1–Geminin activities have yet to be fully elucidated, structure analysis of the Cdt1–Geminin interaction reveal a dynamic, complex regulation in which the stoichiometry of the Cdt1–Geminin complex determines its activity to recruit Mcm2–7 for licensing or to inhibit pre-RC formation (Lutzmann et al., 2006; De Marco et al., 2009). It is proposed that elevated Geminin levels in S-phase convert the Cdt1–Geminin complex to a licensing-defective state (De Marco et al., 2009).

While Cdt1 is negatively regulated by proteolysis and Geminin binding, positive regulation of Cdt1 occurs through E2F-mediated transcriptional activation during the cell cycle (Karakaidos et al., 2004; Yoshida and Inoue, 2004) and through acetylation by HDAC11 and HBO1 (human acetylase binding to Orc1) (Iizuka et al., 2006; Miotto and Struhl, 2008; Glozak and Seto, 2009). Acetylation of Cdt1 positively modulates its activity to facilitate licensing (Miotto and Struhl, 2010) as well as prevents Cdt1 from ubiquitination and subsequent proteasomal degradation (Glozak and Seto, 2009).

Cdt1 degradation during the cell cycle

The evolution of multiple distinct and redundant pathways for Cdt1 proteolysis suggests the importance for inactivating Cdt1 function during S phase and G2 to prevent re-replication (Fujita, 2006; Arias and Walter, 2007; Hook et al., 2007). One degradation pathway is through the SCF–Skp2 E3 ubiquitin ligase complex, in which Cdt1 proteolysis is regulated by Cdk activity. In humans, Cdk2 and Cdk4 have both been shown to interact with Cdt1 on its N-terminus Cy motif (residues 67–69) and phosphorylate Cdt1 at residue Thr-29, thus recruiting the SCF–Skp2 complex to Cdt1 for inducing Cdt1 degradation during S and G2 (Figure 2; Li et al., 2003, 2004; Kondo et al., 2004; Sugimoto et al., 2004; Nishitani et al., 2006). Cdt1 mutants defective for Skp2 interaction or Cdk-mediated phosphorylation were found to still undergo degradation in S phase (Takeda et al., 2005; Nishitani et al., 2006), leading to the discovery of an additional pathway for Cdt1 proteolysis utilizing the Cul4–Ddb1–Cdt2 E3 ubiquitin ligase complex (Hu et al., 2004; Takeda et al., 2005; Arias and Walter, 2006; Jin et al., 2006; Nishitani et al., 2006; Senga et al., 2006; Kim and Kipreos, 2007a).

Figure 2.

Degradation of Cdt1 during the cell cycle and after DNA damage. In S phase, Cdt1 degradation occurs by the SCF–Skp2 and by the PCNA–Cul4–Ddb1–Cdt2 pathways. Cyclin A/Cdk2 associates with Cdt1 on its Cy motif and phosphorylates Cdt1 at residue Thr-29, followed by recruitment of the SCF–Skp2 E3 ubiquitin ligase complex. PCNA binds to Cdt1 on its PIP motif, leading to recruitment of Cul4–Ddb1–Cdt2 E3 ubiquitin ligase complex. Cdt2 recognizes the degron motif (D) adjacent to the PIP box on Cdt1, which allows for binding of Cdt1 with Cdt2 thus promoting E3 ligase activity of the Cul4–Ddb1–Cdt2 complex. Following DNA damage by UV/IR irradiation, PCNA–Cul4–Ddb1–Cdt2 mediates Cdt1 degradation on chromatin in a similar manner as S-phase-induced degradation.

Cdt1 degradation through the Cul4–Ddb1–Cdt2 E3 ubiquitin ligase complex is mediated by proliferating cell nuclear antigen (PCNA) (Figure 2; Arias and Walter, 2006; Higa et al., 2006; Hu and Xiong, 2006; Nishitani et al., 2006; Senga et al., 2006). In the study using Xenopus extracts, direct binding of PCNA to Cdt1 only occurs on chromatin when PCNA is loaded onto chromatin during DNA replication, and thus Cdt1 degradation is coupled with DNA replication (Arias and Walter, 2006; Havens and Walter, 2009). The association of Cdt1 with PCNA involves the PCNA-interacting protein (PIP) motif of the N-terminus of Cdt1, and mutations within this highly conserved region stabilize Cdt1 protein levels and induce re-replication (Arias and Walter, 2006; Hu and Xiong, 2006; Nishitani et al., 2006; Senga et al., 2006). The interaction of PCNA with Cdt1 recruits the Cul4–Ddb1–Cdt2 complex to Cdt1, which contains a ‘degron’ motif located four residues downstream of the Cdt1 PIP motif that is recognized by Cdt2 and targets Cdt1 for degradation (Havens and Walter, 2009). Depletion of any one of the Cul4–Ddb1–Cdt2 complex members leads to Cdt1 stabilization and re-replication (Jin et al., 2006; Lovejoy et al., 2006; Hall et al., 2008). It is proposed that PCNA-mediated degradation of Cdt1 through Cul4–Ddb1–Cdt2 provides a means for rapid degradation of chromatin-bound Cdt1 during S to prevent re-replication (Arias and Walter, 2006, 2007).

While both SCF–Skp2 and PCNA–Cul4–Ddb1–Cdt2 mediate degradation of Cdt1 in S phase redundantly, Nishitani et al. elegantly demonstrated the distinct functions of these pathways during the cell cycle. Utilizing Cdt1 mutants specifically defective for SCF–Skp2 binding (Cdt1 Cy motif mutant) and for Cul4–Ddb1–Cdt2 association (Cdt1 PIP motif mutant) in cells synchronized for early S, mid-S, and G2, they showed that Cul4–Ddb1–Cdt2 functions to degrade Cdt1 only in S phase, whereas SCF–Skp2 acts in both S and G2 (Nishitani et al., 2006). Thus, SCF–Skp2 and PCNA–Cul4–Ddb1–Cdt2 function independently of one another to degrade Cdt1, and SCF–Skp2 is specifically mediated by Cdk-dependent cell cycle regulation (Hook et al., 2007).

Additional means to negatively regulate licensing factors include Cul4-mediated nuclear export of Cdc6 through CKI-1 (Kim et al., 2007), caspase-3-mediated cleavage of Cdc6 (Pelizon et al., 2002), and APC/C^Cdh1-mediated proteolysis of Cdt1 (Sugimoto et al., 2008). While APC/C^Cdh1-mediated ubiquitination and degradation of Cdt1 has been demonstrated to occur in both in vivo and in vitro systems (Sugimoto et al., 2008), the exact biological contributions of this pathway has yet to be fully elucidated with respect to the SCF–Skp2 and PCNA–Cul4–Ddb1–Cdt2 pathways.

The regulation of Cdt1 following DNA damage

In addition to degradation during the cell cycle, Cdt1 proteolysis in response to DNA damage has also been demonstrated in a number of organisms (Higa et al., 2003; Ralph et al., 2006; Hall et al., 2008; Cook, 2009). In mammalian cells, Cdt1 degradation upon DNA damage caused by both UV and IR has been shown to be predominantly mediated by the PCNA–Cul4–Ddb1–Cdt2 pathway, although some evidence also suggests involvement of the SCF–Skp2 pathway in the UV-induced Cdt1 degradation (Higa et al., 2003; Hu et al., 2004; Kondo et al., 2004; Ralph et al., 2006; Hall et al., 2008). Depletion of PCNA, PCNA inhibition by p21, and the Cdt1 mutants defective for PCNA binding all lead to stabilized Cdt1 levels under conditions of UV damage (Arias and Walter, 2006; Hu and Xiong, 2006; Senga et al., 2006). As well, depleting Cul4, Ddb1, or Cdt2 suppresses Cdt1 degradation from UV and IR (Higa et al., 2003, 2006; Hu et al., 2004; Jin et al., 2006). As shown in Figure 2, it was proposed that PCNA–Cul4–Ddb1–Cdt2-mediated, damage-induced degradation of Cdt1 occurs when Cdt1 associates with PCNA on chromatin, thus providing a means for rapid destruction of Cdt1 on DNA in response to DNA damage (Arias and Walter, 2007; Hook et al., 2007).

Further analysis will be needed to address the mechanisms of how damage-induced Cdt1 degradation occurs outside of S-phase. It was described that IR induces Cdt1 degradation in G1, and this degradation also requires Cul4 activity and PCNA (Higa et al., 2003, 2006). Since PCNA chromatin loading is usually associated with replication forks, more detailed analysis will be needed to address how Cdt1 is degraded through the PCNA–Cul4–Ddb1–Cdt2 pathway when IR is induced in G1. It is generally believed that PCNA–Cul4–Ddb1–Cdt2-mediated Cdt1 degradation is independent of checkpoints (Arias and Walter, 2007). IR-induced Cdt1 proteolysis in mammalian cells does not depend on the ATM-Chk2 and ATR-Chk1 pathways, but Kondo et al. described that UV-induced Cdt1 degradation is caffeine-sensitive, suggesting that ATR/ATM may be involved (Higa et al., 2003; Kondo et al., 2004). In fission yeast, damage-induced Cdt1 turnover is also dependent on Ddb1 and Cdt2, but independent of checkpoint genes Rad3 and Cds1 (ATR and Chk2 homologues) (Ralph et al., 2006). It still remains possible that multiple pathways, which can be checkpoint-dependent or -independent and act in a cell-cycle-regulated manner, are involved in damage-induced Cdt1 degradation.

Damage-induced Cdt1 degradation may also contribute to the prevention of DNA re-replication. Upon DNA damage, checkpoints are activated and consequently Cdk activities are reduced so that cell cycle is arrested, meanwhile reduced Cdk activity in G2 may allow reassembly of pre-RCs on already fired origins (Hayles et al., 1994; Dahmann et al., 1995; Itzhaki et al., 1997). Damage-induced Cdt1 degradation is thus important for the prevention of pre-RC reassembly on already fired origins. In addition, damage-induced Cdt1 degradation may also be involved in the suppression of replication initiation in response to DNA damage, as Cdt1 is important for promoting the activation of MCM2–7 at the onset of DNA replication in addition to its licensing roles (You and Masai, 2008; Ballabeni et al., 2009). Consistently, it was described that IR-induced reduction of Cdt1 levels reversely correlates with S-phase recovery (Higa et al., 2003). Taken together, multiple mechanisms of Cdt1 regulation by proteasomal degradation during the cell cycle and following DNA damage all function to ensure proper licensing control and proper DNA replication.

Cdt1 overexpression leads to DNA re-replication

DNA re-replication can be induced when the regulation of licensing control factors such as Cdt1 and Cdc6, or the control of Cdk activities is impaired during the cell cycle (Blow and Dutta, 2005; Machida et al., 2005; Arias and Walter, 2007). Cdt1 overexpression alone can induce re-replication in higher eukaryotes (Vaziri et al., 2003; Maiorano et al., 2005). Likewise, depletion of the Cdt1-binding inhibitor Geminin is also sufficient for re-replication (Tada et al., 2001; Mihaylov et al., 2002; Melixetian et al., 2004; Li and Blow, 2005). One mechanism by which Cdt1 overexpression can induce re-replication is thought to be through its activity to recruit MCM2–7 onto chromatin and promote pre-RC reassembly (Cook et al., 2004). Geminin inhibits Cdt1 by preventing its interaction with MCM2–7 (Wohlschlegel et al., 2000), and thus inactivation of Geminin allows non-degraded Cdt1 species in S and G2 to reassemble pre-RCs causing re-replication. However, one study showed that the Cdt1 mutants deleted for the C-terminal MCM2–7 binding sites are still capable to induce re-replication (Teer and Dutta, 2008). Although it is proposed that the binding of these Cdt1 mutants with PCNA and Cdks through the N-terminus of Cdt1 may titrate away PCNA and Cdks to interact with endogenous Cdt1 causing re-replication (Teer and Dutta, 2008), additional complexity and intricacies may be involved by which Cdt1 overexpression leads to re-replication.

DNA re-replication induces the activation of checkpoint pathways

Since DNA re-replication would inevitably lead to genome instability, mechanisms to suppress re-replication as a means to prevent potentially harmful genome instabilities caused by re-replication would be beneficial to the cell. It has been addressed by a number of laboratories that cell-cycle checkpoint is activated when re-replication is induced. In S. cerevisiae, combining mutations in ORC, Cdc6, and MCM2–7 causes DNA re-replication and activates checkpoints (Archambault et al., 2005). In the Xenopus cell free system, a caffeine-sensitive phosphorylation of Chk1 occurs when re-replication is induced (Li and Blow, 2005). In mammalian cells, it was initially observed by Dr. Dutta's laboratory that overexpression of Cdt1 induces activation of the ATM/ATR-dependent checkpoint (Vaziri et al., 2003).

One important biological function of re-replication-induced checkpoint activation is to arrest cell cycle or to eliminate cells with over-replicated DNA by apoptosis or senescence (Figure 3). Re-replicating yeast cells show a discrete cell cycle arrest (Archambault et al., 2005). Similarly, in Xenopus egg extracts, extensive re-replication leads to head-to-tail collision of re-replication forks with normal replication forks resulting in DNA fragmentation and DSB formation, which activates checkpoint pathways and blocks further cell cycle progression (Davidson et al., 2006). In mammalian cells, H2AX phosphorylation is readily observed when re-replication is induced by Cdt1 overexpression or Geminin inactivation (Melixetian et al., 2004; Zhu et al., 2004; Zhu and Dutta, 2006; Liu et al., 2007), suggesting that re-replication also causes DSB accumulation in mammalian cells. Depending on the cellular background, activation of the ATM and ATR-checkpoint pathways induces G2/M arrest, senescence or apoptosis upon DNA re-replication [(Vaziri et al., 2003; Melixetian et al., 2004; Zhu et al., 2004) and Truong and Wu, unpublished results]. The BRCA1-mediated Fanconi anemia pathway is also suggested to be involved in the G2/M checkpoint activation (Zhu and Dutta, 2006).

Figure 3.

The S-phase checkpoint prevents re-replication caused by disruption of the licensing control. Cdt1 is required for initial assembly of pre-RC complex, and Cdt1 overexpression leads to reassembly of pre-RC complexes on already fired origins. Uncoupling of MCM unwinding from DNA synthesis at the initiation of re-replication results in ssDNA accumulation, which in turn activates the S-phase checkpoint (Liu et al., 2004). DSBs are accumulated when stalled re-replication forks are collapsed or when re-replicating forks collide with other forks. In cells with functional checkpoint regulation, activated S-phase checkpoint inhibits re-replication, so that re-replication is suppressed or prevented. Activated checkpoints also induce cell cycle arrest, as well as apoptosis and senescence when the amount of re-replication-induced DNA lesions exceeds cellular repair capacity. Compromised checkpoints, combined with disrupted licensing control, lead to overt re-replication which results in genome instability and tumorigenesis.

ATR-mediated checkpoint pathway plays an important role in preventing DNA re-replication when the licensing control is defective

Intriguingly, when the licensing control is disrupted in mammalian cells by Cdt1 overexpression, extensive re-replication occurs only in certain tumor cell lines, whereas in primary cell lines and a number of tumor cell lines (re-replication non-permissive), significant re-replication is not induced while checkpoints are activated (Vaziri et al., 2003; Tatsumi et al., 2006; Liu et al., 2007).

ATM and ATR are two related checkpoint kinases involved in detecting abnormal DNA structures and initiating checkpoint activation (Abraham, 2001; Shiloh, 2001). However, these two kinases respond to different DNA damage signals. Whereas ATM is activated by DSBs, ATR is activated by ssDNA accumulation at stalled replication forks (Cimprich and Cortez, 2008). Significantly, when Cdt1 is overexpressed in primary cell lines and in those tumor cell lines that are resistant to Cdt1-induced re-replication, inactivation of ATR and its downstream kinase Chk1 leads to extensive re-replication, while inactivation of ATM or Chk2 does not promote such re-replication (Liu et al., 2007). Impaired expression of Rad17 or ATRIP, required for ATR activation (Cimprich and Cortez, 2008), both showed a similar effect to permit Cdt1-induced re-replication as seen with ATR deficiency. These observations suggest that the ATR-mediated S-phase checkpoint acts as a surveillance barrier to prevent DNA re-replication when the licensing control is impaired in the ‘re-replication non-permissive’ cells lines. This S-phase checkpoint-mediated suppression of DNA re-replication confers another layer of protection on top of the licensing control, ensuring one round of DNA replication per cell cycle (Figure 3).

Further studies showed that when the licensing control is impaired, ATM is activated but its activation occurs later than that of ATR (Liu et al., 2007). Soon after Cdt1 overexpression, RPA-bound ssDNA is accumulated, activating ATR prior to the detection of extensive re-replication, whereas ATM activation occurs when DSBs are detected at much later stages. This observation is consistent with the hypothesis that ATR is activated when the initial steps of DNA re-replication are detected, and this activation in turns suppresses further DNA re-replication.

Given the critical roles of the S-phase checkpoint in the suppression of DNA re-replication, what are the signals triggering the activation of the S-phase checkpoint upon loss of replication licensing? When Cdt1 is overexpressed or Geminin expression is lost, Cdt1 re-loads MCM proteins onto replication forks and reassembles pre-RCs in S-phase (Figure 3). Inhibition of re-replication initiation by overexpressing p27 or by inhibiting Cdc7 kinase activity prevents S-phase checkpoint activation (Liu et al., 2007), suggesting that re-replication initiation is required for S-phase checkpoint activation, while Cdt1-mediated reassembly of pre-RCs is not sufficient. Further analysis demonstrated that at the onset of re-replication, MCM-mediated DNA unwinding is uncoupled from DNA synthesis, causing ssDNA accumulation at re-replication forks [(Liu et al., 2007), Figure 3]. These RPA-bound ssDNA serve as initial signals to activate the ATR-mediated S-phase checkpoint, which is supported by the observation that suppressing MCM helicase activity prevents ssDNA accumulation and attenuates the checkpoint activation (Liu et al., 2007). Re-replication-induced uncoupling of DNA unwinding and DNA synthesis is likely due to the helicase activity that exceeds the rate or capacity of DNA polymerases to synthesize DNA, as re-replication is not a scheduled event. These studies suggest a critical mechanism to activate the S-phase checkpoint at the initiation of re-replication.

ATR-mediated S-phase checkpoint targets different downstream effector proteins to mediate the inhibition of DNA re-replication. Activated ATR directly acts on DNA replication machinery to inhibit re-replication through phosphorylating replication factors such as RPA2 and MCM2, or indirectly suppresses DNA replication by modulating the activities of Rb or p53 (Vaziri et al., 2003; Lee et al., 2007; Liu et al., 2007). Prevention of DNA re-replication thus requires the intact ATR pathways including sensors, transducers, and effectors. Therefore, it is reasonable to predict that defects in one or more components of the ATR-dependent S-phase checkpoint pathway, which often occurs in tumor cells, would cause loss of re-replication suppression function and allow extensive re-replication upon loss of replication licensing control (Figure 3). This provides explanations as to why overexpression of Cdt1 or inactivation of Geminin causes extensive re-replication in certain tumor cell lines, but not in others and in primary cell lines.

Inhibition of DNA re-replication by the S-phase checkpoint pathway is conserved in various organisms. In S. cerevisiae, deficiency of Mec1 (the ATR homologue) and Rad17 leads to significantly more extensive re-replication when the licensing control is impaired (Archambault et al., 2005). Similarly, Chk1 activation antagonizes Cdt1-induced re-replication in Xenopus nuclear extracts (Li and Blow, 2005).

It was described that Cdt1 and Cdc6 are destabilized when re-replication is induced by Geminin inactivation or Cdt1 overexpression (Hall et al., 2008). This feedback regulation minimizes the extent of re-replication by proteolysis of the licensing factors, thereby protecting genome stability. It was demonstrated that this re-replication-induced Cdt1 degradation requires the PCNA-binding site of Cdt1 and Cul4–Ddb1–Cdt2 ubiquitin ligase, suggesting that it uses the same pathway for S-phase- and damage-induced Cdt1 ubiquitination and degradation (Hall et al., 2008). Presumably, this PCNA–Cul4–Ddb1–Cdt2 degradation pathway is independent from the ATR/ATM checkpoint pathways (Higa et al., 2003; Arias and Walter, 2005). We thus propose that accumulated ssDNA upon re-replication may recruit PCNA chromatin binding, triggering the activation of Cul4–Ddb1–Cdt2 pathway to degrade Cdt1. Re-replication-induced degradation of Cdc6 is less clear, but it is known that this degradation requires the Huwe1 ubiquitin ligase (Hall et al., 2008). These observations suggest that in addition to the re-replication suppression function mediated by the ATR-checkpoint pathway, an ATR-independent mechanism involving direct degradation of the licensing factors may also have a role in limiting the extent of re-replication.

DNA re-replication and DSB repair

DSBs are accumulated when re-replication is induced (Figure 3). This can be caused by collapse of stalled re-replication forks or collision of new re-replication forks with existing replication/re-replication forks (Figure 3; Davidson et al., 2006; Liu et al., 2007). ATM is activated when DSBs are detected upon loss of the licensing control. In addition to its role in inducing cell cycle arrest, apoptosis and senescence, ATM may also be involved in promoting DSB repair when re-replication is induced. Inactivation of the Mre11 complex results in accumulation of more DSBs, suggesting that this complex likely participates in the repair of DSBs caused by DNA re-replication (Lee et al., 2007). Yeast mutants impaired in the licensing control and Rad52 function are synthetic lethal (Archambault et al., 2005), implying that HR-mediated DSB repair may be involved in repairing re-replication-associated DSBs to maintain cell viability. Despite these observations, the exact mechanism of how DSBs are repaired during re-replication is still not clear.

Suppression of re-replication plays critical roles in preventing genome instability when mutations and defects are present in the replication control pathways. However, such mechanisms may be more important for a normal cell to cope with mistakes at replication onset and/or during replication. For instance, reassembly of preRCs may occur at one or more origins at fault in a normal cell. In response to such mistakes, inhibition of re-replication immediately after DNA unwinding would limit re-replication to a minimal extent, possibly right after the synthesis of RNA/DNA primers or small stretches of DNA at origins. Under this circumstance, checkpoint-activated repair pathways would be able to remove these limited duplicated sequences and repair re-replication-associated lesions, so that DNA lesions caused by transient loss of DNA replication control are fixed and a normal cell cycle can be restored. Checkpoint-induced cell cycle arrest, apoptosis, or senescence would be only induced when re-replication-associated DNA lesions exceed the repair capacity of a normal cell.

DNA re-replication and tumorigenesis

Deregulated overexpression of Cdt1 and Cdc6 was observed in various tumor samples. In a set of analysis, 75 cases of non-small cell lung carcinomas and adjacent healthy lung tissue were examined for the expression levels of Cdt1 and Cdc6 (Karakaidos et al., 2004). Strikingly, overexpression of Cdt1 and Cdc6 (more than 4-fold) was observed in 43 and 50% of neoplasms, respectively. Co-overexpression of Cdt1, Cdc6, and E2F1 is common. It has been shown that E2F1 and E2F2 are important transcription activators of Cdt1 and Cdc6 (Hateboer et al., 1998; Yan et al., 1998; Karakaidos et al., 2004; Yoshida and Inoue, 2004), and overexpression of E2F family members is thus likely a contributing mechanism for Cdt1 and Cdc6 overexpression in tumors. Other studies revealed that gene amplification is another source leading to Cdt1 and Cdc6 overexpression (Liontos et al., 2007). Overexpression of Cdt1 and/or Cdc6 has also been documented in mantle cell lymphoma, colon cancer and head-and-neck carcinomas (Karakaidos et al., 2004; Pinyol et al., 2006; Liontos et al., 2007). In mouse models, Cdt1 overexpression predisposes for malignant transformation (Seo et al., 2005). These studies suggest that deregulated overexpression of Cdt1 and Cdc6 and its associated re-replication are closely linked to tumorigenesis.

To investigate whether deregulated overexpression of Cdt1 and Cdc6 is an active driving force for tumorigenesis rather than a mere reflection of increased proliferation rate in tumors, the expression levels of Cdt1 and Cdc6 was examined at different cancerous and precancerous stages of tumorigenesis of lung, colon, and head-and-neck cancer (Liontos et al., 2007). Overexpression of Cdt1 and Cdc6 was detected at mRNA levels by 2-fold in the hyperplasia when compared with the adjacent normal tissues, while protein levels of Cdt1 and Cdc6 are elevated at least 4-fold in dysplasia and carcinomas. No correlation of elevated expression of the proliferation marker Ki-67 with that of Cdt1 and/or Cdc6 was detected (Liontos et al., 2007). It is thus hypothesized that overexpression of Cdt1 and/or Cdc6 promotes DNA re-replication, leading to genome instability and DNA damage responses. The activated checkpoints serves as an anti-tumor barrier, which induces senescence and apoptosis to eliminate cells with severe DNA lesions associated with DNA re-replication. Loss of p53 and/or other genome caretaker genes abrogated the antitumor barriers and resulted in tumorigenesis. Consistently, loss of p53 was frequently observed in the tumors with overexpressed Cdt1 and Cdc6 (Karakaidos et al., 2004; Pinyol et al., 2006). In the mouse transgenic model, overexpression of Cdt1 in thymocytes exhibited normal T-cell development, but developed thymic lymphoblastic lymphoma when crossed to p53 null mice (Seo et al., 2005).

Activation of checkpoints at the early stages of tumorigenesis is not only limited to the cancer development that is associated with overexpression of Cdt1 and Cdc6. It has been described that activation of DNA damage checkpoint response is a general consequence of oncogene activation (Gorgoulis et al., 2005; Venkitaraman, 2005; Bartek et al., 2007). Various oncogenes, such as H-RasV12, cyclin E, and Cdc25A, induce unscheduled DNA replication or DNA re-replication at pre-cancerous stages of tumorigenesis, generating signals to activate DNA damage checkpoint response, and subsequent loss of checkpoint barriers leads to tumorigenesis (Bartkova et al., 2005, 2006; Di Micco et al., 2006). These studies suggest that loss of replication control is a common phenomenon at the initiation of oncogenesis, which is not only caused by mutations in the replication licensing pathway. On the other hand, at least in some cases, oncogene-induced deregulation of DNA replication occurs through modulating the activities of replication licensing factors. For instance, expression of the oncogenic cyclin D1 mutation P287A led to increased Cdt1 expression and DNA re-replication. The human esophageal carcinoma-derived cell lines TE3 and TE7, harboring the cyclin D1-P287A mutation express Cdt1 5–6-fold higher than the control cell lines due to disruption of the Cul4–Ddb1–Cdt2-mediated Cdt1 degradation pathway (Aggarwal et al., 2007). It was also shown that overexpression of oncogenes H-RasV12 caused increases in Cdc6 protein levels (Di Micco et al., 2006). Therefore, DNA re-replication is an integral aspect of tumorigenesis, which activates cellular anti-tumor barriers while inducing genome instability. Subsequent accumulation of mutations in DNA damage checkpoint pathways would abrogate the cellular anti-tumor barriers and lead to oncogenesis.

Unanswered questions

DNA re-replication is tightly associated with tumorigenesis, which highlights the importance to understand the mechanisms underlying the prevention of re-replication. Despite recent significant progress, many questions remain unanswered. For instance, endoreduplication occurs during normal development to form megakarocytes and trophoblastic cells (Zybina and Zybina, 2005; Deutsch and Tomer, 2006). It is not clear how cells distinguish this normal developmental endoreduplication from abnormal re-replication and how checkpoints respond to them differently. In addition, activation of ATR-mediated S-phase checkpoint suppresses re-replication, but initial DNA re-replication is needed to activate this checkpoint. It is expected that low levels of DNA re-replication and short stretches of re-replicated DNA are present in cells despite the suppression of re-replication by the checkpoints. The mechanisms involved in removing these initial re-replicated DNA to maintain genome stability are not clear. Furthermore, loss of the licensing control by Cdt1 overexpression or Geminin inactivation results in chromosomal instability. It is necessary to elucidate the mechanisms of how DNA re-replication and its associated DNA lesions lead to generation of specific chromosomal abnormalities commonly present in tumors, such as chromosomal translocation, gene amplification and aneuploidy. Clarifying the interplay of DNA re-replication control, checkpoint activation and repair of re-replication-associated DNA lesion would be fundamental to address these questions.

Funding

This work was supported by the NIH R01 Grant CA102361 and NIH R01 Grant GM080677 to X.W., and the NIH Training Grant DK007022-30 to L.T.