The activity of human Cdt1 is negatively regulated by multiple mechanisms. This suggests that Cdt1 deregulation may have a deleterious effect. Indeed, it has been suggested that overexpression of Cdt1 can induce rereplication in cancer cells and that rereplication activates Ataxia-telangiectasia-mutated (ATM) kinase and/or ATM- and Rad3-related (ATR) kinase-dependent checkpoint pathways. In this report, we highlight a new and interesting aspect of Cdt1 deregulation: data from several different systems all strongly indicate that unregulated Cdt1 overexpression at pathophysiological levels can induce chromosomal damage other than rereplication in non-transformed cells. The most important finding in these studies is that deregulated Cdt1 induces chromosomal damage and activation of the ATM-Chk2 DNA damage checkpoint pathway even in quiescent cells. These Cdt1 activities are negatively regulated by cyclin A/Cdks, probably through modification by phosphorylation. Furthermore, we found that deregulated Cdt1 induces chromosomal instability in normal human cells. Since Cdt1 is overexpressed in cancer cells, this would be a new molecular mechanism leading to carcinogenesis.
Several molecular mechanisms contribute to the maintenance of genomic integrity. One mechanism is cell-cycle control of DNA replication to ensure that duplication of the genome is complete and occurs only once. Ataxia-telangiectasia-mutated (ATM) kinase and ATM- and Rad3-related (ATR) kinase-dependent checkpoint pathways are also important in monitoring chromosomal stresses such as DNA breaks or replication arrest, suspending cell-cycle progression and activating DNA repair (Osborn et al., 2002). Other mechanisms include the mismatch repair and excision repair pathways that remove inappropriate nucleotids from genomic DNA. Whereas disruption of the latter pathways is undoubtedly involved in genomic instability and can lead to cancer, it remains unclear whether dysfunction of the former is associated with such process.
Recent studies have elucidated the molecular mechanism by which DNA replication is cell-cycle-controlled in eukaryotic cells (Fujita, 1999; Bell and Dutta, 2003). The current concept is that, a multi-protein complex - termed the pre-replication complex (pre-RC) - is constructed during the G1 phase, beginning with the binding of the origin recognition complex (ORC) to chromosomal DNA. Cdc6 and Cdt1 proteins (Maiorano et al., 2000; Nishitani et al., 2000; Rialland et al., 2002) are recruited to the chromatin by interaction with ORC, and the resultant assembly may function as a loader for the MCM (mini-chromosome maintenance) heterohexameric complex, a putative replicative helicase (Ishimi, 1997). Once cyclin-dependent kinases (Cdks) become active at the onset of S phase, the pre-RC initiates replication (Bell and Dutta, 2003).
To prevent rereplication, re-establishment of pre-RC must be suppressed after S phase. Recent studies suggest that Cdks also play a central role in this process. Cdks prevent re-establishment of pre-RC through multiple redundant mechanisms. One of these is phosphorylation of Cdc6, leading to its degradation in yeast (Drury et al., 1997; Nishitani and Nurse, 1995) and nuclear export in mammalian cells (Saha et al., 1998; Jiang et al., 1999; Petersen et al., 1999; Fujita et al., 1999). In human cells, ORC1 is degraded after S phase, presumably depending on Cdk phosphorylation (Mendez et al., 2002; Fujita et al., 2002; Tatsumi et al., 2003). It also has been shown that the MCM complex is phosphorylated by Cdks (Hendrickson et al., 1996; Fujita et al., 1998). In budding yeast, it is necessary to block all of these pathways for induction of rereplication without inhibiting Cdk activity (Nguyen et al., 2001). In fission yeast, overexpression of Cdc6 and Cdt1 can cause rereplication (Nishitani and Nurse, 1995; Nishitani et al., 2000), although this might be a consequence of inhibition of Cdk activity by overexpression.
In cancer cells, proteins associated with DNA replication are often overexpressed. Deregulated expression of initiation factors of DNA replication could play a causative role in genetic instability and subsequent tumorigenesis. Indeed, it has been demonstrated that MCM and Cdc6 proteins are overproduced in cancer cells (Hiraiwa et al., 1997; Williams et al., 1998). However, it remains unknown how such overexpression affects normal cellular functions, and whether it actually leads to genetic instability and eventual carcinogenesis. In general, also in mammalian cells, alteration of the regulation of individual replication-initiation components alone fails to induce rereplication because Cdks prevent it through multiple redundant mechanisms. For example, several studies have shown that overexpression of the Cdc6 protein does not affect normal cell-cycle progression (Jiang et al., 1999; Petersen et al., 1999; Pelizon et al., 2000). More recently, Clay-Farrace et al. reported that Cdc6 overexpression prevents mitotic entry through Chk1 activation (Clay-Farrace et al., 2003). However the molecular mechanism involved and the reason for the discrepancy between these two results have not yet been clarified.
In higher eukaryotic cells, the geminin protein was identified as another inhibitor of pre-RC formation (McGarry and Kirschner, 1998) and shown to inhibit Cdt1 activity after S phase (Wohlschlegel et al., 2000; Tada et al., 2001). However, recent studies from our and other laboratories demonstrate that Cdt1 function is also negatively regulated through phosphorylation by cyclin A/Cdks (Nishitani et al., 2001; Sugimoto et al., 2004; Liu et al., 2004; Takeda et al., 2005). The cullin4-based ubiquitin ligase system has also been suggested to be involved in Cdt1 regulation (Zhong et al., 2003; Arias and Walter, 2006; Senga et al., 2006; Hu and Xiong, 2006; Nishitani et al., 2006). Since inhibition of Cdt1 activity after S phase is carried out by these multiple mechanisms, it is possible that deregulation of Cdt1 is a more deleterious insult than deregulation of other initiation proteins. Indeed, it has been suggested that overexpression of Cdt1 can induce rereplication in cancer cells (Vaziri et al., 2003; Nishitani et al., 2004; Saxena and Dutta, 2005). In this report, we highlight a new and interesting consequence of Cdt1 deregulation: data from several different systems all strongly indicate that unregulated Cdt1 overexpression at pathophysiological levels can induce chromosomal damage other than rereplication in non-transformed cells. The most important finding in this study is that deregulated Cdt1 induces chromosomal damage and activation of the ATM-Chk2 DNA damage checkpoint pathway in quiescent cells. Furthermore, we show that deregulated Cdt1 induces chromosomal instability in normal human cells. Since Cdt1 is overexpressed in cancer cells, this would be a new molecular mechanism leading to eventual carcinogenesis.
Cdt1 is overexpressed in tumor-derived cell lines
We examined the expression of several DNA replication-initiation proteins in human cancer cells compared with normal human cells (Fig. 1). In agreement with previous reports (Hiraiwa et al., 1997; Williams et al., 1998), MCM7 and Cdc6 proteins were overexpressed in cancer cells. Since MCM7 and Cdc6 transcription is regulated by the E2F family of transcription factors, we also determined Cdt1 and ORC1 protein levels, which are also under the control of E2F (Yoshida and Inoue, 2004). As expected, their overproduction was also observed in the cancer cells tested. On average, Cdt1 protein levels in the cancer cells were 5-20 times greater than those in normal cells. The human 293T cell line is not tumor-derived but arose after in vitro transformation by the adenovirus E1A and E1B genes. Both viral gene products inactivate RB and p53 proteins, respectively. Overexpression of initiation proteins was also observed in 293T cells. Overexpression of Cdt1 was also reported in other cancer cell lines (Xouri et al., 2004).
Transient overexpression of Cdt1, but not ORC1 or Cdc6, in 293T cells leads to chromosomal damage and activation of the ATM-Chk2 DNA damage checkpoint pathway without inducing detectable rereplication
The data above suggested that overexpression of initiation proteins affect normal cell-cycle control. Among these proteins, Cdt1 is especially interesting because its activity is suppressed through multiple independent mechanisms, suggesting that its overexpression could be a more deleterious insult than other proteins. To investigate this possibility, we first set up a transient expression system with 293T cells. Using this system, up to 80% of viable cells were found to express T7-Cdt1 (Fig. 2A). The average level of overexpressed Cdt1 was estimated, by immunoblotting, to be 20 times greater than the endogenous level (Fig. 2B,C). Functional Cdt1 is bound to chromatin/nuclear matrix fraction (Nishitani et al., 2000; Sugimoto et al., 2004). Therefore, we examined localization of overexpressed T7-Cdt1 and found it also to be mainly in chromatin/nuclear matrix fraction (Fig. 2E). We then examined cell-cycle profiles by flow cytometry, but observed no detectable difference (Fig. 2D). We also investigated cells transfected with Cdc6 or ORC1 alone, or with Cdt1 and Cdc6 (Fig. 2B, supplementary material Fig. S1A). Again, no significant effect on cell-cycle distribution was observed (Fig. 2D, supplementary material Fig. S1B). Note that, in 293T cells, p53 is functionally inactivated by E1B.
Although Cdt1 overexpression at these levels does not induce detectable rereplication, it obviously activates the ATM DNA damage checkpoint pathway (Fig. 2B,C). Immunoblotting with anti-Chk2 antibody showed the presence of phosphorylated Chk2 upon overexpression of Cdt1. Use of specific anti-phosphopeptide antibody showed phosphorylation of Chk2 at Thr68. Active ATM and ATR kinases phosphorylate Chk2 kinase at Thr68 and upregulate its activity (Matsuoka et al., 2000). Therefore, our data suggest that Cdt1 overexpression activates either ATM or ATR, or both kinases. This was also confirmed by the finding that phosphorylation of p53 at Ser15, which is also carried out by ATM and/or ATR kinases, was induced by Cdt1 overexpression. Histone H2AX is believed to be recruited to damaged chromatin sites and to be phosphorylated by ATM and/or ATR kinases (Osborn et al., 2002). We found that the levels of γ-H2AX increased significantly in cells overexpressing Cdt1. γ-H2AX forms nuclear foci around chromatin lesions, which can be detected by immunostaining (Osborn et al., 2002). Therefore, we examined 293T cells by immunostaining with anti-γ-H2AX antibody. The γ-H2AX foci were observed in cells overexpressiong Cdt1. However, the number of the foci was very different among cells; there were many cells having no or only small amounts of γ-H2AX, with some cells containing the distinct foci (supplementary material Fig. S2A). Similar focal staining was also observed when Cdt1-overexpressing Rat-1 or HFF2 cells were immunostained with anti-phosphorylated-ATM antibody after extraction with Triton X-100 (Fig. 4C, Fig. 5C, Fig. 6C,D; see below). These foci might represent the chromatin lesion induced by overexpression of Cdt1.
The ATM kinase responds primarily to DNA double-strand breaks (DSBs). This pathway can act during all phases of the cell cycle. Recently, it has been proposed that ATM is usually present as an inactive multimer and is activated by autophosphorylation at Ser1981 after DSBs or changes in the chromatin structure (Bakkenist and Kastan, 2003). Therefore, we examined ATM phosphorylation at Ser1981 with a specific antibody and, as shown in Fig. 2B,C, ATM was phosphorylated on Ser1981 in cells overexpressing Cdt1. Taken together, these data indicate that Cdt1 overexpression induces some chromosomal damage and thereby activates the ATM-Chk2 pathway in 293T cells. Such effects were not induced when Cdc6 or ORC1 were overexpressed (Fig. 2B, supplementary material Fig. S1A). Activation of the ATM-Chk2 pathway by overexpression of Cdt1 was not affected by co-expression with Cdc6 (Fig. 2B).
To address whether Cdt1-induced chromosomal damage is associated with replication stress, we investigated whether an ATR-Chk1 pathway is activated (Osborn et al., 2002). In contrast to Chk2, phosphorylation of Chk1 kinase on Ser345 is carried out only by ATR kinase (Zhao and Piwnica-Worms, 2001). Therefore, we examined Chk1 phosphorylation at Ser345 after Cdt1 overexpression in 293T cells, but no significant phosphorylation was observed (Fig. 2B,C). Treatment of cells with hydroxyurea or with high dose γ-irradiation (γ-IR) clearly induced Chk1 phosphorylation (supplementary material Fig. S1C), showing the ATR/Chk1 pathway to be intact in the cells used in these studies.
We recently demonstrated that cyclin A-dependent kinases phosphorylate Cdt1 and this phosphorylation may result in negative regulation of Cdt1 activity (Sugimoto et al., 2004). Therefore, we sought to determine whether the ability of overexpressed Cdt1 to induce chromosomal damage and activation of the ATM-Chk2 pathway is regulated by Cdk. We have generated a Cdt1 mutant (Cdt1 Cy) in which the conserved cyclin-binding motif is mutated and, therefore, cannot be phosphorylated by Cdk. Compared with wild-type Cdt1, the Cdt1 Cy is resistant to degradation mediated by Cdk phosphorylation and subsequent Skp2 binding (Sugimoto et al., 2004). In addition, whereas in vitro DNA-binding activity of wild-type Cdt1 is inhibited by Cdk phosphorylation, that of Cdt1 Cy is not affected (data not shown). T7-Cdt1 Cy was introduced into 293T cells like wild-type Cdt1. In spite of similar expression levels, the levels of phosphorylation of ATM at Ser1981, Chk2 at Thr68, p53 at Ser15 and H2AX at Ser139 were all higher in cells overexpressing Cdt1 Cy (Fig. 2B,C). Thus, cyclin-A-dependent kinases negatively regulate Cdt1 activity to induce chromosomal damage and ATM-Chk2 activation, probably through phosphorylation. Nevertheless, even overexpression of the Cdt1 Cy could not induce detectable rereplication (Fig. 2D).
Stable overexpression of Cdt1 at pathological levels in Rat-1 fibroblasts produces chromosomal damage and activation of ATM-Chk2 without inducing detectable rereplication
Although the above data suggest that the Cdt1-induced chromosomal damage is not associated with induction of overall genome rereplication, we could not exclude the possibility that Cdt1 overexpression induces micro rereplication that is undetectable by flow cytometry in transient assays. There is also the problem that 293T cells are in vitro transformed cells and overexpress Cdt1. Therefore, we stably introduced Cdt1 into Rat-1 cells, which are non-transformed rat fibroblast cells.
Rat-1 cells were infected with recombinant retroviruses expressing wild-type T7-Cdt1 or T7-Cdt1 Cy and several stable clones were established. From the result shown in Fig. 1, we estimate that pathophysiological levels of Cdt1 overexpression are less than 20 times those of endogenous expression. Therefore, we established several clones overexpressing various amounts of Cdt1 within this range (Fig. 3A). Similar to the situation in 293T cells, overexpressed T7-Cdt1 was found to interact with chromatin/nuclear matrix (Fig. 3D). Cdt1 overexpression obviously activated the ATM-Chk2 DNA damage checkpoint pathway also in Rat-1 cells (Fig. 3A). Since the antibody against Chk2 phosphorylated at Thr68 cannot be used to study rat cells, we monitored Chk2 phosphorylation by its reduced electrophoretic mobility. Chk2 phosphorylation was found in almost all Cdt1-overexpressing Rat-1 clones, and the relative level of Chk2 phosphorylation tended to increase as the level of Cdt1 increased. Furthermore, Chk2 phosphorylation was more prominent in cells with Cdt1 Cy. Phosphorylation of p53 at Ser15 was also observed in cells whose expression levels of Cdt1 and extent of Chk2 phosphorylation were high. Rat-1 has functional p53 protein, so expression of p21, a Cdk inhibitor, may be induced by active p53. In fact, in one Cdt1-Cy-overexpressing line, designated CA3, a high level of p21 induction was observed, although in other lines it was undetectable. The exact reason for such differential induction of p21 is unclear, but it is possible that, cells that can cancel pathways downstream of p53 were preferentially selected during establishment of the transfectants. In agreement with the results of the experiment with 293T cells, no significant phosphorylation of Chk1 Ser345 was observed in these Rat-1 transfectants. However, differing from 293T cells transiently transfected with Cdt1, only little or no γ-H2AX was detected in these Rat-1 transfectants (Fig. 3A).
We next examined the DNA content of these clones by flow cytometry, but found no obvious difference compared with parental Rat-1 cells (Fig. 3B). One could argue that rereplication occurs in only a minor population of the Cdt1-overexpressing cells. Therefore, we determined the fraction of cells with activated ATM in one representative Cdt1-Cy-overexpressing line, CB4, by immunostaining. As shown in Fig. 4C, approximately 50% of asynchronous CB4 cells showed focal staining of phosphorylated ATM. One could also argue that rereplication occurs at such a low rate in these cells that it is not detectable by flow cytometry. However, it is anticipated that, even if rereplication occurs only at a very low level during a single cell cycle, its products accumulate during long-term culture and become detectable. An argument against this possibility is that cells carrying abnormal chromosomes, resulting from rereplication, are eliminated. As described above, Cdt1-induced DNA damage occurs in at least half of the population of the Cdt1-overexpressing cells. Therefore, if these cells are eliminated, there would be many dead cells in Cdt1-overexpressed cells and the apparent growth rate of these cells would be significantly decreased. However, we found no difference in the viability of cells when we analyzed the Trypan Blue exclusion assay (∼99% viability), and no difference in their growth rates (Fig. 3C). Taken together, these data strongly indicate that stable Cdt1 overexpression at these levels produces chromosomal damage without inducing rereplication in non-transformed rat fibroblasts.
Activation of the ATM-Chk2 checkpoint pathway by Cdt1 overexpression is cell-cycle specific
Cdt1 function is cell-cycle regulated. Therefore, the deleterious effect of Cdt1 also would be cell-cycle dependent. Accordingly, parental Rat-1 cells and Rat-1 cells expressing wild-type Cdt1 or the Cy mutant (WB4 and CB4, respectively) were arrested in quiescence (G0/G1 phase), induced to start the cell cycle, and harvested sequentially at the indicated times. Flow cytometric analysis revealed synchronous progression from G1 to S and G2/M phase (Fig. 4A,B, supplementary material Fig. S3). In parallel, whole cell lysates were analyzed by immunoblotting (Fig. 4A,B, supplementary material Fig. S3). In all the cell lines tested, Cdc6 protein levels were virtually undetectable in the quiescent state and began to increase around late G1 phase, further showing the efficacy of the synchronization procedure. The kinetics for the cell-cycle progression and Cdc6 expression in WB4 and CB4 cells were essentially the same as those in parental cells (supplementary material Fig. S3). In WB4 cells, T7-Cdt1 proteins were detectable even in quiescent cells, and were expressed continuously during G1-phase progression and degraded as the cells entered S phase. Surprisingly, ATM activation and Chk2 phosphorylation were observed in quiescent cells. Activation of the ATM pathway ceased during the G1 phase even though Cdt1 overexpression continued. The reason for the decrease in chromosomal damage and ATM pathway activation during G1 phase is unclear at present. When cells reached late S phase, activation of the ATM pathway was observed again. In contrast to wild-type Cdt1, the levels of T7-Cdt1 Cy in CB4 cells did not decrease significantly, even after S phase. The kinetics for activation of the ATM pathway by the Cy mutant were essentially the same as those for wild-type Cdt1, except for the re-activation in early S phase. This is probably due to the Cy mutant being stable even after cells enter S phase. As expected, Chk2 phosphorylation was not observed in parental Rat-1 cells throughout the cell cycle (supplementary material Fig. S3). Taken together, these data indicate that overexpression of Cdt1 protein during the cell cycle, other than in G1 phase, results in induction of chromosomal damage.
We further confirmed that ATM is activated by overexpressed Cdt1 in quiescent cells by double-immunostaining (Fig. 4C). Asynchronous or quiescent Rat-1 or CB4 cells were treated with the nucleotide analog bromodeoxyuridine (BrdU) for 6 hours and then immunostained with antibodies against BrdU and phosphorylated ATM. Under these conditions, approximately 50% of asynchronous Rat-1 cells showed positive BrdU staining, whereas only less than 2% of the quiescent cells incorporated BrdU. As mentioned above, approximately 50% of asynchronous CB4 cells showed strong staining for phosphorylated ATM. Importantly, it was clearly demonstrated that ATM is phosphorylated in approximately 50% of quiescent CB4 cells that do not incorporate BrdU.
Introduction of Cdt1 into quiescent rat fibroblasts induces chromosomal damage and activation of ATM-Chk2 without replication
Although the observation that ATM activation is detected even in quiescent Cdt1-overexpressing Rat-1 cells provides strong evidence that Cdt1-induced chromosomal damage is not associated with rereplication, it remained possible that the damage in these cells was produced by rereplication while they were still dividing. To address this issue, we introduced Cdt1 into the quiescent Rat-1 cells using lentivirus vector. Induction of quiescence was confirmed by the absence of Cdc6 protein (Fig. 5A), decrease in BrdU-positive cells (Fig. 5B), and flow cytometry analyses (data not shown). The quiescent cells were then infected with lentiviruses carrying either wild-type Cdt1, Cy mutant Cdt1 or the empty vector, and analyzed 144 hours after infection. Expression of T7-Cdt1 was confirmed by immunoblotting (Fig. 5A). Immunostaining analyses demonstrated that approximately 50-70% of these cells expressed T7-Cdt1 (Fig. 5C). Even after Cdt1 introduction, Cdc6 protein levels remained undetectable (Fig. 5A) and the number of BrdU-positive cells remained low (Fig. 5B). Nevertheless, ATM and Chk2 phosphorylation was found upon Cdt1 expression (Fig. 5A), and immunostaining analyses demonstrated that ATM Ser1981 is phosphorylated in cells expressing T7-Cdt1 (Fig. 5C).
We also developed another system with 3Y1 cells, which are non-transformed rat cells. We established a 3Y1 clone in which overexpression of the Cy mutant T7-Cdt1 is inducible by doxycycline. The Tet-on T7-Cdt1 Cy 3Y1 cells were first arrested in quiescence. Expression of T7-Cdt1 Cy was then induced in the quiescent cells, and simultaneously BrdU was added. After 48 hours post-induction, cells were analyzed. Induction of T7-Cdt1 Cy expression was confirmed by immunoblotting (supplementary material Fig. S4A). Immunostaining analyses demonstrated that approximately 45% of these cells expressed T7-Cdt1 (supplementary material Fig. S4C). Even after induction of Cdt1, Cdc6 protein levels remained undetectable (supplementary material Fig. S4A), and the number of BrdU-positive cells remained low (Fig. S4B). Nevertheless, ATM phosphorylated at Ser1981 was present in cells expressing T7-Cdt1 (supplementary material Fig. S4C).
Together, these data clearly show that deregulated Cdt1 expression in quiescent cells can induce chromosomal damage and ATM activation without replication.
Stable overexpression of Cdt1 induces chromosomal damage and activation of ATM-Chk2 without rereplication, and leads to chromosomal instability in normal human cells
Chromosomal damage induced by deregulated expression of Cdt1 should lead to chromosomal instability. We tested this possibility in HFF2 cells, normal human fibroblasts immortalized by telomerase. It is known that the frequency of karyotype abnormalities is low in telomerase-expressing human fibroblasts (Morales et al., 1999). HFF2 cells were infected with the high-titer wild-type T7-Cdt1, Cy mutant T7-Cdt1, or control retroviruses and selected. At 2 weeks after infection, overexpression of the introduced T7-Cdt1 was detected (Fig. 6A). The average level of the overexpression was 10-15 times greater than the endogenous level (Fig. 6B). As expected, phosphorylation of ATM on Ser1981 was induced by Cdt1 overexpression, which was more prominent in HFF2 cells with Cdt1 Cy (Fig. 6A,B). Immunostaining analysis further demonstrated that ATM is activated in 25-35% of asynchronous Cdt1-overexpressing HFF2 cells (Fig. 6C). Nevertheless, no obvious difference in DNA content and growth rate was detected compared with the control cells (supplementary material Fig. S5). We then examined whether chromosomal damage and checkpoint activation are also observed in the quiescent Cdt1-expressing HFF2 cells. Asynchronous or quiescent cells were treated with BrdU for 8 hours and then double-immunostained (Fig. 6C,D). Under these conditions, ∼ 40% of asynchronous cells were BrdU-positive, whereas less than 2% of the quiescent cells incorporated BrdU. As expected, ATM was phosphorylated in 25-35% of quiescent Cdt1-overexpressing HFF2 cells that did not incorporate BrdU. All of these findings are essentially the same as those observed in Rat-1 cells.
At 2 months after infection, cells were subjected to karyotypic analysis to estimate chromosomal stability (Spruck et al., 1999). As shown in Fig. 7A, most HFF2 cells have a normal diploid karyotype with 46 chromosomes. This chromosome pattern was not changed by infection with control retroviruses. However, HFF2 cells that overexpressed T7-Cdt1 accumulated an increased proportion of aneuploid cells, and overexpression of T7-Cdt1 Cy appeared to induce aneuploidy at a frequency, higher than overexpression of wild-type T7-Cdt1 (Fig. 7A). We performed more detailed karyotypic analysis on at least 40 metaphase spreads for each cell line. Several representative results are shown in Fig. 7B. Cdt1 overexpression in HFF2 cells caused elevated frequencies of chromosomal losses and gains, although no significant increase in the frequency of chromosomal translocations, deletions or breakage was found.
Cdt1 overexpression induces rereplication with reduced Cdk1 activity in 293T cells when the cells are arrested in M phase by nocodazole
Since Cdks prevent rereplication through multiple redundant mechanisms, the result that Cdt1 overexpression cannot induce rereplication in normal cells might be expected. For example, ORC1 protein is also degraded after S phase, and overexpressed Cdt1 could not compensate for ORC1 function. Nevertheless, Cdt1 overexpression could sensitize cells to rereplication, especially in cancer cells. Thus, we were interested in whether rereplication occurs when cells overexpressing Cdt1 suffer additional stress.
When cells are treated with nocodazole, the spindle-checkpoint is activated and cells are arrested at metaphase. In such cells, the levels of ORC1 and Cdt1 proteins recover by unknown mechanism(s) (Fujita et al., 2002; Sugimoto et al., 2004), which can sensitize cells to rereplication (Di Leonardo et al., 1997). Therefore, we examined the effect of Cdt1 overexpression on nocodazole-treated cells. For these studies, 293T cells were transfected with T7-Cdt1, treated with nocodazole and then analyzed by immunoblotting and flow cytometry (Fig. 8A,B). Cells transfected with control vector showed a typical profile for cells arrested in G2/M phase and about 8% of cells had a DNA content that was >4N. When T7-Cdt1 was overexpressed, the percentage of cells with >4N DNA increased approximately twofold. Also, in this assay, the T7-Cdt1 Cy acted as a gain-of-function mutant because the percentage increased further. Overexpression of Cdc6 did not confer such a rereplication phenotype. These data clearly indicate that Cdt1 deregulation predisposes cells to rereplication.
To gain some insight into the molecular mechanisms of Cdt1-induced rereplication, we investigated several proteins that might be involved in this phenomenon (Fig. 8A,C). Cdk1 inactivation in G2/M phase results in rebinding of MCM proteins to chromatin and subsequent rereplication (Itzhaki et al., 1997; Fujita et al., 1998). Therefore, we investigated Cdk1 activity in Cdt1-overexpressing cells. We first examined inhibitory phosphorylation of Cdk1 at Tyr15 and found that it increased significantly with Cdt1 overexpression. We also examined vimentin phosphorylation on Ser55, which is catalyzed by Cdk1 kinase (Fujita et al., 1998). Vimentin Ser55 phosphorylation decreased with Cdt1 overexpression, consistent with the increased Cdk1 phosphorylation at Tyr15. The Cdt1 Cy mutant was more potent than wild-type Cdt1 in induction of Cdk1 inactivation. Activation of the ATM-Chk2 pathway by overexpressed Cdt1 was also evident in metaphase-arrested 293T cells. Therefore, we thought that diminished Cdk1 activity might result from activation of the ATM-Chk2 pathway. To test this possibility, shRNA targeting the ATM gene was stably expressed in 293T cells. As shown in Fig. 8C, expression of ATM protein was specifically reduced by approximately 90%. The 293T cells with silenced ATM protein were then transfected with T7-Cdt1, and treated with nocodazole (Fig. 8C). ATM silencing caused a considerable reduction in phosphorylation of Chk2 at Thr68, providing further evidence that the phosphorylation is mainly carried out by ATM. Surprisingly, we could not detect significant recovery of Cdk1 activity on ATM silencing, as measured by phosphorylation of Cdk1 at Tyr15 and phosphorylation of vimentin. The explanation of these results is not clear at present. One possibility is that the residual ATM-Chk2 activity is sufficient to suppress Cdk1 activity. Another possibility is the involvement of stress-responsive p38 MAP kinase. It has been suggested that p38 MAP kinase is activated by ultraviolet (UV)-light-induced DNA damage, which thereby inhibits the Cdc25/Cdk cascade independently of the ATM/ATR pathways (Bulavin et al., 2001). Therefore, it is interesting that we found that p38 MAP kinase is activated on Cdt1 overexpression and that this activation is not affected by ATM silencing (data not shown). As expected from a limited effect of ATM silencing on Cdk1 inhibition, ATM silencing only partially affected the rereplication phenotype induced by overexpression of Cdt1 (Fig. 8D). Therefore, it remains to be clarified how Cdt1 overexpression inhibits Cdk1 activity in nocodazole-treated 293T cells and whether this inhibition contributes to the rereplication.
Unregulated Cdt1 overexpression can induce chromosomal damage and ATM activation without rereplication
Vaziri et al. reported that overexpression of Cdt1 can induce rereplication in cancer cells (Vaziri et al., 2003). In addition, they proposed that rereplication activates ATM/ATR kinase-dependent checkpoint pathways. Nishitani et al. also reported that overexpression of Cdt1 at very high level can induce rereplication in 293T cells, which are in vitro transformed cancer-like cells (Nishitani et al., 2004). In this report, we provide new and important results on the effect of Cdt1 deregulation: data from several different systems with non-transformed cells all strongly indicate that unregulated Cdt1 overexpression at pathophysiological levels can induce chromosomal damage other than rereplication. The most important finding in these studies is that deregulated Cdt1 induces chromosomal damage and activation of the ATM-Chk2 DNA damage checkpoint pathway in quiescent cells. We first observed that ATM-Chk2 is still activated when Rat-1 and HFF2 cells that overexpress Cdt1 are arrested in quiescence (Fig. 4, Fig. 6D). Although it would be possible that the damage was produced by rereplication when quiescent cells were still dividing, this was excluded by the fact that introduction of Cdt1 into quiescent Rat-1 cells results in chromosomal damage and ATM-Chk2 activation without inducing replication (Fig. 5).
The data obtained from exponentially growing, Cdt1-overexpressing cells are also consistent with this notion. In both Rat-1 and HFF2 cells that stably overexpress Cdt1, activation of an ATM-Chk2 DNA damage checkpoint pathway occurs in a significant fraction of cells (Fig. 3, Fig. 4C, Fig. 6A,C). Nevertheless, no change in DNA content was detected by flow cytometry, even after long-term culture (Fig. 3B, supplementary material Fig. S5B). There was also no sign that cells with damaged chromatin are eliminated (Fig. 3C, supplementary material Fig. S5A). One could argue that rereplication occurs at a very low rate, remaining undetectable by flow cytometry even after cells repeat mitosis. One way to detect rereplication at very low levels might be detailed karyotypic analysis. Although we do not know exactly what chromosomal abnormalities result from rereplication, there are some indications. In budding yeast, rereplication induced by disrupting the multiple prevention mechanisms leads to chromosomal breakage (Green and Li, 2005); similar breakage is observed when geminin-silenced cancer cells are treated with a Chk1 inhibitor (Melixetian et al., 2004). Therefore, the frequencies of chromosomal translocation, deletion or breakage might be increased in cells with rereplication. It is also possible that rereplication simply results in broadened and/or bizarrely shaped chromosomes. However, no such findings were observed in Cdt1-overexpressing HFF2 cells (Fig. 7B). Rather, elevated frequencies of chromosomal losses and gains were observed. It is possible, but very unlikely, that only one or two chromosomes are completely rereplicated and the others are not affected. Furthermore, chromosomal losses cannot be explained by rereplication. Overall, the results of detailed karyotypic analysis seem to be consistent with the conclusion that Cdt1 overexpression in HFF2 does not induce rereplication. We cannot completely exclude the possibility that Cdt1 induces rereplication in cycling cells, which is rapidly repaired and becomes undetectable. However, taking the fact into account that Cdt1 can induce chromosomal damage and ATM activation in quiescent cells, it is very possible that unregulated Cdt1 can also induce chromosomal damage other than rereplication in cycling cells. However, such chromosomal damage is not observed with ORC1 or Cdc6. The fact that Cdt1 deregulation leads to a more deleterious insult than other initiation proteins seems compatible with the fact that its function is strictly controlled through multiple mechanisms.
As mentioned above, Vaziri et al. and Nishitani et al. reported that overexpression of Cdt1 can induce rereplication (Vaziri et al., 2003; Nishitani et al., 2004). A probable explanation for the apparent difference between the results reported here and by the groups mentioned above might be that Vaziri et al. and Nishitani et al. overexpressed Cdt1 at very high levels in cancer cells. Cancer cells constitutively overexpress replication initiation factors such as ORC1, Cdc6, Cdt1 and MCM (Fig. 1) and, therefore, might be sensitive to Cdt1-induced rereplication. Indeed, whereas Cdt1 overexpression at the levels that are approximately 20 times greater than the endogenous level does not induce detectable rereplication in 293T cells (Fig. 2), rereplication can be induced in the same cells when Cdt1 is overexpressed at levels approximately 100 times greater than the endogenous level (Nishitani et al., 2004) (Nishitani et al., personal communication). It has been reported that silencing of geminin induces rereplication attributable to Cdt1 deregulation (Melixetian et al., 2004; Zhu et al., 2004). However, cancer cells have been used in most of these studies. In this regard, recent results that excess Cdt1 induces rereplication in Xenopus egg extracts, could be caused by a large amount of maternal initiation proteins in the eggs (Arias and Walter, 2005; Li and Blow, 2005; Maiorano et al., 2005; Yoshida et al., 2005). We agree that, under certain condition, Cdt1 deregulation can induce rereplication. Indeed, Cdt1 overexpression also predisposes cells to rereplication in our experimental system (Fig. 8). Our results highlight a new aspect concerning the consequences of Cdt1 deregulation. Vaziri et al. reported that induction of rereplication by deregulated Cdt1 in cancer cells is prevented by the presence of functional p53 (Vaziri et al., 2003). Also in our studies, induction of rereplication by overexpression of Cdt1 and nocodazole treatment is observed in cancer-like 293T cells (Fig. 8) - which have suppressed p53 function, but not with Rat-1 cells - which possess functional p53 (data not shown). However, rereplication by geminin depletion in cancer cells is observed regardless of the p53 status. The reason for the difference is not clear at present, but geminin has other role(s) than inhibiting Cdt1 (Del Bene et al., 2004; Luo et al., 2004; Seo et al., 2005) that might contribute to the difference.
In general, activation of the DNA damage checkpoint pathway delays cell-cycle progression. However, in our system, using normal cells, cell-cycle progression was not altered by Cdt1 overexression. Cdt1 gives rise to chromosomal damage only after S phase, not during G1 phase. Presumably, activation of the ATM-Chk2 pathway by Cdt1 overexpression at levels we examined is relatively mild, so that the cell-cycle progression is apparently unchanged. Alternatively, the negative effect on cell-cycle progression could be compensated by the growth-stimulating effect of Cdt1 overexpression itself.
At present, the mechanism by which Cdt1 overexpression damages chromatin is unclear. One possibility is that Cdt1 overexpression induces DNA DSBs. Induction of γ-H2AX in 293T cells transiently transfected with Cdt1 suggests the presence of DSBs. When estimated by immunoblotting, however, the extent of H2AX phosphorylation induced by wild-type Cdt1 in 293T cells is no more than that by 0.4 Gy γ-IR and that by Cy mutant Cdt1 is no more than 0.8 Gy (compare Fig. 2C with supplementary material Fig. S1D). When using Ser343 phosphorylation of NBS1 (Osborn et al., 2002) as another indicator, roughly the same result was obtained (supplementary material Fig. S2B). Therefore, even granted that overexpressed Cdt1 induces DSBs, the number may be no more than that induced by ∼0.8 Gy γ-IR. Exposure to 1 Gy ionizing radiation causes only 30-40 DSBs in the genomic DNA of a human diploid cell, physical detection of which is very difficult even if using pulsed-field gel electrophoresis (Cedervall et al., 1995; Ruiz de Almodovar et al., 1994). We sought to clarify whether Cdt1 overexpression induces DNA DSBs using a TUNEL assay, but so far failed to obtain significant results (data not shown). On the other hand, as discussed above, detailed karyotypic analysis shows no sign of chromosomal breakage (Fig. 7B). In addition, in Rat-1 clones stably overexpressing Cdt1, only a little or no γ-H2AX was detected (Fig. 3A). It has been proposed that ATM is activated not only by DNA strand breaks but also by inappropriate changes in the chromatin structure (Bakkenist and Kastan, 2003). Therefore, another possibility is that the presence of Cdt1 in the quiescent state or after S phase inappropriately changes the chromatin architecture, either directly or indirectly through recruiting other protein(s). As components of the MCM complex-loading machinery, ORC and Cdc6 proteins use their ATPase activity as does replication factor C, a loader for proliferating cell nuclear antigen (Bell and Dutta, 2003). However, it remains unclear how Cdt1 acts during MCM loading. Alteration of chromatin structure by Cdt1 could be related to its physiological role.
Deregulated Cdt1-expression-induced chromosomal damage is a new molecular mechanism leading to genetic instability and possibly carcinogenesis
Here, we demonstrate that deregulated Cdt1 induces chromosomal instability in normal human cells. It has been reported that Cdt1 overexpression enables murine NIH3T3 cells to form tumors in nude mice (Arentson et al., 2002). Rat-1 cells are sensitive to transformation by several oncogenes. However, we observed no transformed phenotype in Cdt1-overexpressing Rat-1 cells (data not shown). This difference might simply be attributable to a cell-type-specific response. However, our data indicate that deregulated Cdt1 overexpression impacts cells by inducing chromosomal damage and instability. Thus, we prefer another explanation, i.e. deregulated Cdt1 does not lead to acute oncogenic transformation, but to chronic chromosomal damage and instability that can eventually results in carcinogenesis.
If the above explanation is true, an important question that needs to be addressed is: Why is the ATM-Chk2 pathway not activated in cancer-derived cell lines in which Cdt1 is constitutively overexpressed (Fig. 1). There is no compelling answer at present but one explanation is `adaptation and selection' during the continuous proliferation. Our data with Rat-1 cells provide some support for this idea. While establishing Rat-1 cells stably expressing exogenous Cdt1, several clones exhibiting different responses to Cdt1 overexpression were obtained. In some of these cells, the p21 Cdk inhibitor was induced by the activated ATM-Chk2-p53 pathway, but in others it was not. The latter would have a selective growth advantage. During establishment of a cancer-derived cell line in which Cdt1 is overexpressed, the population that can negate its deleterious effects by additional changes would become predominant.
Aneuploid cells can be generated by a variety of mechanisms ranging from DNA damage to defects in chromosomal segregation. Deregulated Cdt1 in normal human fibroblasts caused elevated levels of and a higher frequecy in chromosomal losses and gains, suggesting that it affects processes involved in the faithful segregation of chromosomes. In Drosophila, it is known that activated Chk2 kinase impairs centrosome function (Takada et al., 2003). Therefore, it is possible that, in addition to Cdt1-induced chromosome damage itself, activation of Chk2 kinase contributes to the chromosomal instability observed here. However, it also remains possible that deregulated Cdt1 affects chromosomal segregation by unknown mechanism(s), independent of the chromosomal damage and checkpoint activation.
Materials and Methods
Production of expression vectors and shRNA vectors
Human Cdt1, Cdc6 and ORC1 cDNAs were inserted into mammalian expression vector pcDEBdelta (Sugimoto et al., 2004; Fujita et al., 1999), retroviral expression vector pCLMSCVhyg, which is a derivative of pMSCVhyg (Clonetech) and contains the CMV/LTR fusion promoter, and lentiviral expression vector pCSII-CMV (Miyoshi, 2004). For silencing ATM, the 19-nucleotide sequence corresponding to ATM cDNA nucleotides 604-622 (underlined) was expressed as shRNA as follows. Two oligonucleotides (ATM-S, 5′-GAT CCC CGC TGA TTG TAG CAA CAT ACT TCA AGA GAG TAT GTT GCT ACA ATC AGC TTT TTG GAA A-3′; and ATM-AS, 5′-AGC TTT TCC AAA AAG CTG ATT GTA GCA ACA TAC TCT CTT GAA GTA TGT TGC TAC AAT CAG CGG G-3′) were annealed and introduced into the pSI-MSCVhyg-H1R retroviral expression vector, in which shRNA is transcribed from the H1 promoter and the 3′ LTR is inactivated by an internal deletion. The retroviral expression vector and packaging vector pCL-10A1, or the lentiviral expression vector and packaging vectors pCMVR8.2DVPR (provided by I. S. Y. Chen, UCLA, Los Angeles, CA) and pHCMV-VSV-G (Miyoshi, 2004) were co-transfected into 293FT cells to generate infectious viruses. At 60 hours post transfection, viral supernatants were harvested.
The 293T, HeLa, Caski, SiHa, C33A and Rat-1 cells were cultured in Dulbecco's modified Eagle's medium (DMEM) supplemented with 8% fetal bovine serum (FBS). Human normal cells, HCK (human cervical keratinocyte), HDK (human dermal keratinocyte), HSAEC1 (human small airway epithelial cell), HBEC1 (human brain endothelial cell) and HFF2 (human foreskin fibroblast) were immortalized by introduction of human telomerase reverse-transcriptase. They were cultured in keratinocyte-SFM (Invitrogen) medium. HFF2 cells were cultured in DMEM supplemented with 8% FBS.
Transfection and infection
For transient transfection, 4×106 293T cells were seeded onto 100-mm dishes and transfected with 5.5 μg of the pcDEBdelta-based vectors and 0.5 μg of pMSCVpuro (Clontech) with Trans-IT293 reagent (Mirus). At 24 hours after transfection, cells were selected with 2 μg/ml puromycin for 2 days. Cells then were released in fresh medium with or without nocodazole (100 ng/ml) for 24 hours. For silencing ATM, 293T cells were infected with the shRNA retroviruses and selected in 200 μg/ml hygromycin B.
For establishment of Rat-1 cells stably expressing T7-Cdt1, cells were infected with the retroviruses, selected with 200 μg/ml hygromycin B and cloned. To investigate the effect of Cdt1 overexpression on chromosome stability, 1×105 HFF2 cells (population doublings 100-120) were infected with the high-titer retroviruses (∼5×105 colony-forming units per milliliter), and selected with 200 μg/ml hygromycin B for 10 days.
Synchronization of Rat-1 and HFF2 cells into quiescence was achieved by contact inhibition followed by serum starvation in DMEM supplemented with 0.1% FBS for 48 hours. For re-entry into the cell cycle, quiescent cells were released into DMEM supplemented with 15% FBS. To express T7-Cdt1 in quiescent Rat-1 cells, the cells were infected with lentiviruses, then cultured in DMEM supplemented with 0.1% FBS and harvested 144 hours post-infection.
Cell-cycle and karyotype analysis
Cells were stained with propidium iodide and analyzed with a Becton Dickinson FACS Calibur. HFF2 cells were treated with colcemid (20 ng/ml) for 2 hours, incubated in hypotonic buffer, and fixed with Carnoy's fixative. The chromosomes were then Giemsa-stained.
Preparation of polyclonal rabbit antibodies against human Cdt1, Cdc6, MCM7 and ORC1 was described previously (Sugimoto et al., 2004). Anti-Chk1 (number 06-965), anti-Chk2 (clone 7), anti-phosphorylated Chk1 (Ser345, number 2341), anti-phosphorylated Chk2 (Thr68, number 2661), anti-phosphorylated ATM (Ser1981, number 4526), anti-phosphorylated p53 (Ser15, number 9284), anti-phosphorylated H2AX (Ser139, number 07-164), anti-phosphorylated Cdk1 (Tyr15, number 9111), anti-laminA/C (number 2032) and anti-phosphorylated NBS1 (Ser343, number 3001) were purchased from Cell Signaling Technology. Anti-p53 (Ab-6) and anti-Ras (Ab-4) were from Oncogene Research Products, anti-ATM (2C1) was from Gene Tex, anti-phosphorylated-vimentin55 (clone 4A4) was from MBL (Japan), anti-phosphorylated ATM (Ser1981, 600-401-400) was from Rockland Inc., and anti-T7 tag was from Novagen.
Cell fractionation to prepare chromatin- or nuclear-matrix-bound fraction was performed with modified CSK buffer containing 0.1% Triton X-100 (0.1% Triton X-100 mCSK buffer) as described previously (Fujita et al., 1999; Sugimoto et al., 2004).
To immunostain T7-Cdt1 and γ-H2AX, transfected 293T cells were seeded onto coverslips, fixed with 3.7% formaldehyde, and then permeabilized with Triton X-100. The secondary antibody for T7 staining was goat anti-mouse IgG antibody conjugated with Alexa Fluor-488, and that for γ-H2AX was goat anti-rabbit IgG antibody conjugated with Alexa Fluor-594 (Molecular Probes). For double-immunofluorescence experiments, newly synthesized DNAs were labeled by incubating cells with 10 μM BrdU. Cells were washed with ice-cold PBS and extracted with 0.1% Triton X-100 mCSK buffer at 4°C for 10 minutes. Cells were then fixed with formaldehyde for 20 minutes on ice and incubated for 2 hours with the anti-phosphorylated ATM antibody. The samples were further incubated for 2 h with the secondary goat anti-rabbit IgG antibody conjugated with Alexa Fluor-594 and re-fixed. The cells were then treated for 20 minutes with 2N HCl to expose the incorporated BrdU and stained with Alexa Fluor-488-conjugated anti-BrdU mouse monoclonal antibody (Roche).
We thank H. Miyoshi and I. S. Y. Chen for providing materials. We are also grateful to M. Itoh and T. Tsuji for technical assistance and Y. Hanada and A. Noguchi for secretarial work. This work was supported in part by a grant to M.F. from the Ministry of Education, Science, Sports and Culture of Japan. N.S. is supported by JSPS research fellowship.
Supplementary material available online at http://jcs.biologists.org/cgi/content/full/119/15/3128/DC1
- Accepted May 3, 2006.
- © The Company of Biologists Limited 2006