|Home | About | Journals | Submit | Contact Us | Français|
Cyclin E has been shown to have a role in pre-replication complex (Pre-RC) assembly in cells reentering the cell cycle from quiescence. The assembly of the pre-replication complex, which involves the loading of 6 MCM subunits (Mcm2–7), is a prerequisite for DNA replication. We found that cyclin E, through activation of Cdk2, promotes Mcm2 loading onto chromatin. This function is mediated in part by promoting the accumulation of Cdc7 mRNA and protein, which then phosphorylates Mcm2. Consistent with this, a phosphomimetic mutant of Mcm2 can bypass the requirement for Cdc7 in terms of Mcm2 loading. Furthermore, ectopic expression of both Cdc6 and Cdc7 can rescue the MCM loading defect associated with expression of dominant-negative Cdk2. These results are consistent with a role for cyclin E-Cdk2 in promoting the accumulation of Cdc6 and Cdc7, which is required for Mcm2 loading when cells re-enter the cell cycle from quiescence.
Pre-replication complex (pre-RC) assembly, also called DNA replication licensing, is strictly regulated to ensure that DNA replication occurs only once per cell cycle to maintain genomic stability. Sequential loading of the origin recognition complex (ORC), Cdc6, Cdt1, and six MCM subunits (Mcm2–7) to form the pre-RC normally occurs at the end of mitosis (Mendez and Stillman, 2000). Mcm2–7 are specifically recruited by Cdc6 and Cdt1 through direct protein-protein interactions (Cook et al., 2004) and by ATP hydrolysis carried out by Cdc6 (Randell et al., 2006). However, how the loading of all components is precisely regulated still remains poorly understood. Mcm2– 7 is thought to function as a DNA helicase that is required to unwind double stranded DNA for DNA replication initiation and possibly elongation (Bochman and Schwacha, 2008; Moyer et al., 2006; Nishitani and Lygerou, 2004; Pacek et al., 2006).
Studies on MEFs derived from cyclin E1,E2−/− embryos indicate that cyclin E is required for chromatin loading of Mcm2, but only during cell cycle re-entry from quiescence (Geng et al., 2003). More recently, it has been shown that cyclin E-Cdk2 activity is required to protect Cdc6 from APC-mediated ubiquitylation and proteolysis during G1, thereby potentiating chromatin loading of Mcm2–7 in human cells (Mailand and Diffley, 2005). These results are at odds with the observations of Geng and colleagues based on cyclin E1,2 −/− MEFs, where accumulation and chromatin loading of Cdc6 were not affected by the absence of cyclin E1/E2 (Geng et al., 2003). More recent studies from the same group suggest that the essential MCM-loading function of cyclin E is Cdk2-independent and doesn’t involve loading of Cdc6 and Cdt1 (Geng et al., 2007), again at odds with results reported by Mailand and Diffley (Mailand and Diffley, 2005). The relationship between cyclin E and loading of Mcm2–7 is further complicated by the observation that cyclin E deregulation can impair MCM loading in proliferating cells (Ekholm-Reed et al., 2004).
Another kinase shown to regulate MCM protein function is Cdc7-Dbf4 (Kim et al., 2003; Lei et al., 1997). Phosphorylation of Mcm4 by Cdc7-Dbf4 promotes stable binding between Mcm4 and Cdc45 in the budding yeast Saccharomyces cerevisiae and mammalian cells (Masai et al., 2006; Sheu and Stillman, 2006). This potentially explains the requirement of Cdc7 for initiation of DNA replication. Mcm2 is also a substrate of Dbf4-Cdc7 in vitro and in vivo, as shown by several groups.
In this report, we have investigated the roles of both cyclin E-Cdk2 and Cdc7-Dbf4 in chromatin loading of Mcm2 as cells reenter the cell cycle from quiescence. We found that phosphorylation of Mcm2 at Ser5 (and possibly Ser4 and Ser7) by Cdc7 kinase in vivo promotes chromatin loading of Mcm2. Furthermore, cyclin E-Cdk2 kinase activity is required for accumulation of Cdc7 during cell cycle reentry, providing an additional explanation for the essentiality of cyclin E described under these circumstances.
Studies on cyclin E1,E2−/− mice show that cyclin E is critical for DNA replication licensing upon cell cycle re-entry possibly through regulating chromatin loading of Mcm2–7 (Geng et al., 2003). We hypothesized that cyclin E-Cdk2 might mediate this process by directly phosphorylating Mcm2–7 subunits. Therefore, we determined potential cyclin E-Cdk2 phosphorylation sites by phosphorylating immunoaffinity purified Mcm2–7 complexes in vitro using recombinant cyclin E-Cdk2. TERT-immortalized human diploid fibroblasts (IHFs) were arrested in G0 by simultaneous contact inhibition and serum starvation and then released to re-enter the cell cycle by replating at low density in medium with serum. Roscovitine, a CDK inhibitor, was then added immediately to prevent in vivo phosphorylation of Mcm proteins by cyclin E-Cdk2. Using this strategy, IHFs arrest prior to S phase and show a pattern of protein expression typical of G0 or early G1 (Supplemental Fig. 1a & 1b). Furthermore, loading of Mcm2 is significantly reduced (Supplemental Fig 1b), suggesting that CDK activity is required for MCM protein loading onto chromatin. MCM protein complexes were then immunoaffinity purified from the nuclear fraction of roscovitine-arrested IHFs. Purified MCM complexes were incubated with ATP and recombinant cyclin E1-Cdk2 and then subjected to MuDPIT (multi-dimensional protein identification technology) analysis to detect phosphorylation sites. We found that S27 of MCM2 and S365 of Mcm7 were strongly phosphorylated. S4, 5, 7, 13, and 381 of Mcm2, T94 of Mcm4, and S121 of Mcm7 were weakly phosphorylated (data not shown; note that based on the spectral obtained, we could not determine whether S4, S5 and S7 or a subset of these residues was phosphorylated). In the current study, we focus on phosphorylation of Mcm2 (sites shown in Fig. 1a). Whereas S13, S27 and S381 conform to a minimal CDK consensus, surprisingly, S4, S5, and S7 do not, although in vitro phosphorylation was carried out with purified cyclin E-Cdk2. However, these sites are conserved in human, mouse, and Xenopus, suggesting that they are functionally important (Fig. 1a).
Phosphosite Mcm2 mutations were created to determine the role(s) of phosphorylation of the sites identified. Although the goal of these studies was to analyze Mcm2 phosphorylation during cell cycle reentry, we initially examined behavior of the phosphosite mutants in 293A cells due to the facility of this approach. HA-tagged Mcm2 wild-type (WT) or mutants were stably introduced into 293A cells by retroviral transduction. The total level of wild-type Mcm2 is only elevated slightly compared to controls (Fig. 1b). Chromatin loading was then analysed using mimosine to arrest cells at the G1/S boundary, (Fig. 1c). Mutating the two aminoterminal putative CDK sites to alanine (13/27A) had no effect on relative level of chromatin loading. However mutation of the three aminoterminal non-CDK consensus sites (4/5/7A) alone or in conjunction with the two CDK sites conferred a severe defect in chromatin loading. Conversely, the 4/5/7D triple phosphomimetic mutant was more efficiently loaded than wild-type.
Immunoprecipitation experiments were carried out to determine if phosphosite alanine mutants defective in chromatin loading were able to interact with other MCM subunits. Phosphosite alanine or aspartate mutations did not affect the ability of Mcm2 to form complexes with other MCM subunits based on efficiency of coprecipitation (Fig. 1d) suggesting that amino-terminal phosphorylation of Mcm2 is not a prerequisite for MCM complex formation.
Since the S4, S5, S7 cluster appeared to be of interest with respect to chromatin loading of Mcm2, we sought to determine the kinase that phosphorylates these residues. Although these residues were identified through an in vitro kinase assay using recombinant cyclin E1-Cdk2, based on the lack of a CDK consensus, it is unlikely that any of these serines is a direct target of cyclin E1-Cdk2. We assumed therefore, that phosphorylation of these residues was carried out by a contaminating kinase that copurified with the MCM protein complexes isolated from IHF cultures. Since Cdc7-Dbf4 has been reported to prefer serines followed by acidic residues, we tested whether these aminoterminal residues might be phosphorylated by Cdc7 (see Fig. 1a). Recombinant Mcm2 13/27A and 4/5/7/13/27A were reacted in vitro using recombinant cyclin E1-Cdk2, cyclin A2-Cdk2, or Cdc7-Dbf4. S13 and S27 were mutated to eliminate background due to phosphorylation of these residues by Cdk2. Phosphorylation of Mcm2 4/5/7/13/27A by Cdc7-Dbf4 was reduced ~20% compared to Mcm2 13/27A (Fig. 2a). Phosphorylation by cyclin E1-Cdk2 or cyclin A2-Cdk2 was minimal under these conditions and not affected by the 4/5/7A mutation (Fig. 2a). Similar results were obtained when wild-type Mcm2 was compared to the 4/5/7A mutant (Supplemental Fig. 2). Therefore, it is likely that Cdc7 phosphorylates one or more residues in the aminoterminal cluster.
Ser5 matches the Cdc7 consensus more closely than other serines in the aminoterminal cluster because it is followed by an acidic residue (Cho et al., 2006; Montagnoli et al., 2006). Therefore, GST fusions were prepared for residues 2–24 of Mcm2 corresponding to wild-type, S5A and S4/5/7A, purified from E. coli and phosphorylated in vitro using recombinant Cdc7-Dbf4. Mutation of Ser5 almost completely eliminated phosphorylation of this fragment as did mutation of all three serines in the cluster (Fig. 2b). Therefore Ser5 appears to be the primary aminoterminal target of Cdc7 in Mcm2. Studies by Cho et al. also show that Ser5 can be phosphorylated by Cdc7 kinase in vitro (Cho et al., 2006) but that Ser4 and Ser7 can also be phosphorylated albeit with lesser efficiency. It may be that efficient phosphorylation of Ser4 requires prior phosphorylation of Ser5 to generate an acidic phosphorylated residue at the +1 position. Our detection of Ser5 (and possibly Ser4 and/or Ser7) phosphorylation of Mcm2 purified from mammalian cells and incubated with recombinant cyclin E1-Cdk2 probably resulted from contamination of the purified MCM protein complexes with endogenous Cdc7-Dbf4.
A phosphosite specific (pSer5) antibody was used to determine if Ser5 is phosphorylated in vivo. The characterization of this antibody is shown in Supplemental Fig. 3. When 293A extracts were analyzed using pSer5 antibody, a signal could be detected in the chromatin fraction but not in the detergent soluble fraction, whereas total Mcm2 partitioned equally between these fractions (Supplemental Fig. 3c). These data suggest that Mcm2 phosphorylated on Ser5 is preferentially partitioned into the chromatin bound fraction, possibly because phosphorylation of Ser5 promotes chromatin loading.
To confirm that Cdc7 is the kinase responsible for phosphorylating Ser5 in vivo, Mcm2 was introduced into U2OS cells by transient transfection followed by transduction with adenoviruses expressing either empty vector or kinase-dead dominant-negative Cdc7 (Cdc7-dn). Cdc7-dn efficiently blocks phosphorylation of Ser5, suggesting that it is a direct target of Cdc7 in cells (Supplemental Fig. 3d).
If phosphorylation of aminoterminal serines of Mcm2 is important for chromatin loading but is not defective in binding other MCM subunits, then overexpression of Mcm2–4/5/7A might be expected to impair loading of heterologous MCM subunits. Overexpression of Mcm2 4/5/7A in 293A cells by transient transfection decreased chromatin loading of endogenous Mcm2 (Fig. 3). This was confirmed by a reduction of pS5 signal, a marker for wild-type Mcm2, in the chromatin fraction and suggests that by sequestering other MCM subunits, Mcm2–4/5/7A prevents chromatin loading of endogenous Mcm2. Additionally, loading of endogenous Mcm4 was impaired (Fig. 3), consistent with sequestration of MCM complexes in a non-loadable state. Conversely, overexpressed phosphomimetic mutant, Mcm2 4/5/7D, loaded efficiently onto chromatin and promoted loading of Mcm4 (Fig. 3). These data suggest that Mcm2 has an active role in recruiting other MCM subunits onto chromatin and that aminoterminal phosphorylation of Mcm2 is important for this function. Interestingly, even though transient transfection of Mcm2 led to significant elevation in steady state intracellular levels of Mcm2, this was restricted to the detergent-soluble fraction (Fig. 3), suggesting that chromatin loading of Mcm2 is saturable near the level observed in non-transfected cells. Under these saturation conditions, Ser5-phosphorylated Mcm2 accumulated in the detergent soluble fraction, suggesting that Mcm2 can be phosphorylated on Ser5 prior to chromatin loading.
Having shown that phosphorylation of aminoterminal serines regulates chromatin loading of Mcm2 in cycling cells, we returned to the question of the role of Cdk2 and now additionally Cdc7 in cell cycle re-entry from quiescence. We chose to use TERT-immortalized human mammary epithelial cells (IME cells), which can easily be rendered quiescent and then stimulated to re-enter the cell cycle. Synchronization time course studies indicated that Mcm2–7 loads onto chromatin from 12–16 hours after release from G0 (Fig. 4a). Cdc6 and Cdc7 accumulate and load onto chromatin with similar kinetics (Fig. 4a). Furthermore cyclin E1 protein and associated kinase activity begin to accumulate at 8 hours after release from quiescence, prior to Mcm2–7 loading and Cdc6 and Cdc7 accumulation (Fig. 4a, b). In a parallel time course carried out in the presence of roscovitine, in order to block Cdk2 activity, Cdc6 and Cdc7 did not accumulate and Mcm2–7 did not load onto chromatin, suggesting that cyclin E-Cdk2 activity regulates both Cdc6 and Cdc7 accumulation, possibly explaining the failure to load MCM complexes. It has already been shown that Cdk2 activity is required for stabilization of Cdc6 in G1 (Mailand and Diffley, 2005).
We then introduced HA-tagged wild-type and mutant alleles of Mcm2 in IME cells to determine the role of phosphorylation of the aminoterminal serine cluster upon cell cycle reentry from quiescence. HA-Mcm2 wild-type stably expressed in IME cells using retrovirus transduction caused only a slight increase in total Mcm2 level compared to an empty vector control population (Fig. 4c). Chromatin loading of wild-type and mutant HA-Mcm2 was then examined in IMEs after release from G0 for 15 hours (Fig. 4d). The Mcm2 4/5/7A and 5A mutants were severely defective in chromatin loading during cell cycle reentry (Fig. 4d), similar to results with mimosine-arrested 293A cells, while phosphomimetic mutants Mcm2 4/5/7D and 5D were not defective in chromatin loading (Fig. 4d). Consistent with phosphorylation of the aminoterminal serine cluster promoting chromatin loading, Ser5 phosphorylation was detected in the chromatin bound fraction of Mcm2 during cell cycle reentry (Fig. 4e). These data suggest that Cdc7, which accumulates at the time of chromatin loading of Mcm2 (Fig. 4a), phosphorylates Mcm2, thus enabling it load onto chromatin to form pre-replication complexes during cell cycle re-entry.
We next sought to determine whether Cdk2 and Cdc7 regulate phosphorylation of Mcm2 and Mcm2 loading onto chromatin. IME cells were arrested in G0 and then infected with an adenovirus expressing a kinase-dead dominant negative allele of Cdc7 (Cdc7-dn) prior to stimulation and cell cycle re-entry. Overexpression of Cdc7-dn impaired chromatin loading of Mcm2 without affecting loading of Cdc6 (Fig. 5a). Therefore, Cdc7-Dbf4 has a role in chromatin loading of Mcm2 during cell cycle reentry.
In order to determine whether phosphorylation of Mcm2 by Cdc7-Dbf4 is sufficient to promote chromatin loading, we tested the ability of Mcm2 phosphomimetic mutants to load onto chromatin in the absence of Cdc7 activity. Increasing multiplicities of Cdc7-dn were transduced into G0 IME cells stably transduced with retroviruses expressing wild-type and phosphomimetic Mcm2 (Fig. 5b). Upon stimulation of transduced cells to reenter the cell cycle, Mcm2 4/5/7D conferred little if any rescue of Mcm2 loading (data not shown). Thus phosphorylation of these sites, although necessary, is not sufficient for efficient chromatin loading. However, Cdc7-Dbf4 has been reported to phosphorylate other amino terminal serines, including Ser27, Ser 40, Ser41, and Ser53 (Cho et al., 2006; Montagnoli et al., 2006; Tsuji et al., 2006). A phosphomimetic Mcm2 mutant substituting 7 aminoterminal putative Cdc7 phosphorylation sites for aspartate (S4/5/7/27/40/41/53D) almost completely rescued the loss of Mcm2 loading in the absence of Cdc7-Dbf4 activity (Fig. 5b). Notably, this phosphomimetic mutant partially rescued loading of Mcm4, consistent with the idea that Mcm2 has a regulatory role in the loading of pre-assembled MCM complexes, also suggested by data in Fig. 3.
Cdc6 and Cdc7 protein levels are dramatically reduced in IME cells stimulated to reenter the cell cycle in presence of the CDK inhibitor roscovitine (Fig 4a). In addition, G0 IME cells transduced with adenovirus expressing a kinase-dead allele of Cdk2 (Cdk2-dn) (van den Heuvel and Harlow, 1993) and stimulated to reenter the cell cycle are impaired in accumulation of Cdc6 and Cdc7 (Fig. 6a). Furthermore, expression of Cdk2-dn impairs chromatin loading of Mcm2 (Fig. 6b, first two lanes). The failure to accumulate Cdc6 when Cdk2-dn is expressed can be explained by the requirement for Cdk2 activity to stabilize Cdc6 during G1 (Mailand and Diffley, 2005). In order to determine whether failure to accumulate Cdc7 was caused by protein instability, as is the case for Cdc6, or a defect in Cdc7 mRNA accumulation, RNA from Cdk2-dn expressing IME cells was analyzed during cell cycle reentry. Compared to controls, Cdk2-dn expressing cells failed to accumulate Cdc7 mRNA (Fig. 6c), accounting for defect in Cdc7 protein expression. Interestingly, there was also a defect in Cdc6 mRNA accumulation, most likely contributing to the defect in Cdc6 protein accumulation. On the other hand, there was no defect in accumulation of the mRNA encoding Dbf4, the activating subunit of the Cdc7 kinase holoenzyme. Moreover, there was no destabilization of Cdc7 protein based on cycloheximide chase experiments (data not shown). In order to determine if cyclin E is required for accumulation of Cdc7, G0 IME cells were transfected with siRNAs targeting cyclin E1 and E2 and then stimulated to reenter the cell cycle. Simultaneous silencing of cyclin E isoforms blocked Mcm2 loading and prevented accumulation of both Cdc6 and Cdc7 (Fig. 6d). Therefore cyclins E1 and E2 are most likely the activators of Cdk2 required to accumulate Cdc6 and Cdc7 during cell cycle reentry.
If transcription of Cdc6 and Cdc7, as well as stabilization of Cdc6, are downstream of Cdk2 in the context of cell cycle reentry, then ectopic expression of a mutationally stabilized allele of Cdc6 (Cdc6-D4-NLS) and wild-type Cdc7 might bypass the requirement for Cdk2 in terms of Mcm2 loading. We therefore transduced G0 IME cells with an adenovirus expressing Cdc6-D4-NLS (Delmolino et al., 2001) and/or Cdc7 and found that expression of either could partially rescue chromatin loading of Mcm2 in the absence of Cdk2 kinase activity (Fig. 6b). However, expression of both Cdc6 and Cdc7 fully rescued chromatin loading of Mcm2 in the presence of dominant-negative Cdk2. These data suggest that cyclin E-Cdk2 promotes chromatin loading of Mcm2, and presumably pre-RC complex assembly, by stimulating transcription of both Cdc6 and Cdc7, as well as by stabilizing Cdc6 (Mailand and Diffley, 2005). Consistent with this, we showed that a phosphomimetic mutant of Mcm2 could be loaded onto chromatin in the absence of both Cdk2 and Cdc7 kinase activities if Cdc6 was supplied (Fig. 6e). Therefore, phosphorylation of Mcm2 is the critical Cdk2-dependent function of Cdc7 in the context of Mcm2 loading onto chromatin whereas synthesis and maintenance of Cdc6 is the only other essential Cdk2-dependent function.
If phosphorylation of Mcm2 by Cdc7 is a pre-requisite for chromatin loading, then phosphorylated Mcm2 should be detectable in the soluble fraction and should accumulate in the soluble fraction if chromatin loading is impeded. We showed in Fig. 3 that Ser5 phosphorylated Mcm2 could be detected in the soluble fraction when Mcm2 was overexpressed by transient transfection in 293A cells. Similarly, Ser5-phosphorylated endogenous Mcm2 can be detected in both the chromatin associated and insoluble fractions of IME cells reentering the cell cycle (Fig. 7). When chromatin loading of Mcm2 is blocked by expression of dominant-negative Cdk2 (Cdk2-dn) but phosphorylation of Mcm2 is maintained by ectopic expression of Cdc7, Ser5-phosphorylated Mcm2 is reduced in the chromatin fraction but accumulates in the soluble fraction. These data indicate that Mcm2 can be phosphorylated on Ser5 by Cdc7 prior to loading onto chromatin and, along with the data presented using Mcm2 phosphomimetic mutants, suggest that phosphorylation of Mcm2 by Cdc7 is a prerequisite for chromatin loading.
We show here by treatment with roscovitine, a Cdk inhibitor, or by expression of a dominant negative allele of Cdk2, that Cdk2 activity is required for chromatin loading of Mcm2 (and presumably other MCM complex subunits) during cell cycle re-entry. The role of Cdk2 in this context appears to potentiate the accumulation of Cdc6 and Cdc7, both of which are required for MCM loading. It has been previously reported that Cdk2 activity is required for stabilization of Cdc6 in G0/G1 by preventing its APC-mediated ubiquitylation and degradation (Mailand and Diffley, 2005). However, we report here that Cdk2 activity is also required for accumulation of both Cdc6 and Cdc7 mRNAs. The cyclins utilized for this function are cyclin E1 and cyclin E2 since their simultaneous RNAi-mediated silencing prevents accumulation of both Cdc6 and Cdc7. Although we do not know the basis for Cdk2 dependency for Cdc6 and Cdc7 mRNA accumulation, one likely possibility is through regulation of the Rb-E2F transcriptional regulatory system (Coqueret, 2002; Harbour and Dean, 2000; Sun et al., 2007). Both the Cdc6 and Cdc7 promoters contain E2F binding sites and have been shown to be stimulated by E2F family transcription factors (Humbert et al., 2000; Kim et al., 1998; Leone et al., 1998; Yan et al., 1998). On the other hand, cyclin E-Cdk2 activity has been shown to reverse the repression exerted by pRb on E2F transcription factors (DeGregori et al., 1995; Harbour et al., 1999). Therefore, cyclin E-Cdk2 kinase activity may be required for efficient pRb phosphorylation and concomitant transcription of Cdc6 and Cdc7 mRNAs. These relationships, however, don’t explain why cyclin E1 or E2 are required for MCM loading specifically during cell cycle re-entry but not in cycling cells. It is possible that in cycling cells cyclin D-Cdk4/6 kinases provide sufficient activity for inactivating pRb, but not during cell cycle reentry, or that more E2F activity needs to be derepressed during cell cycle reentry. Consistent with this, E2F3 alone is sufficient in cycling cells, but E2F3 and E2F1 are required to support cell cycle re-entry (Kong et al., 2007).
The results reported here are in apparent conflict with those reported by Geng and colleagues (Geng et al., 2003). Utilizing cyclin E1/E2−/− MEFs, these authors found that although there was no defect in Cdc6 accumulation upon cell cycle reentry there was a defect in Mcm2 loading. They did not investigate the accumulation of Cdc7, but they attribute the defect in Mcm2 loading to a direct kinase-independent function of cyclin E (Geng et al., 2007). With respect to Cdc6 accumulation, the discrepancy between both our data and those reported by Mailand and Diffley (Mailand and Diffley, 2005) versus those reported by Geng et al. (Geng et al., 2003), may be attributable to a difference between germline ablation (Geng) and acute loss of function (the two other studies). There is ample evidence that germline loss of function can confer different phenotypes than acute loss of function in somatic cells, e.g. (Boulet et al., 2004; Sage et al., 2003). With respect to the apparent discrepancy regarding the requirement for Cdk2 activity, it is difficult to compare data directly since Geng and colleagues used neither a Cdk2 inhibitor nor a dominant-negative allele of Cdk2. Instead they used an allele of cyclin E1 that was defective in activating Cdk2 but sufficient for Mcm2 loading (Geng et al., 2007). However, based on the data presented, it would be difficult to conclusively rule out a residual level of Cdk2 kinase activity supported by this mutant cyclin.
Until this report, the role attributed to Cdc7 has been firing of origins containing assembled pre-replication complexes (Bousset and Diffley, 1998; Kim et al., 2003; Lei et al., 1997). The relevant substrates are thought to be specific subunits of the MCM heterohexamer (Francis et al., 2009; Jiang et al., 1999; Masai et al., 2000; Masai et al., 2006; Sheu and Stillman, 2006). However, in cells re-entering the cell cycle from quiescence, Cdc7 appears to have an additional role specifically in MCM protein loading. We have shown that phosphorylation of sites near the aminoterminus of Mcm2 is necessary and sufficient to promote Mcm2 loading. This interpretation is supported by the behavior of serine to alanine substitution mutations at these aminoterminal sites, which cannot load onto chromatin in the presence of Cdc7 kinase and that of phosphomimetic mutations at these sites, which allow Mcm2 to load even in the absence of Cdc7 kinase activity. The phosphorylation status of these aminoterminal sites also appears to have a role in cycling cells, as phosphorylation-site mutations affect Mcm2 loading in mimosine-arrested 293A cells. If Cdc7 is responsible for phosphorylation of these sites in cycling cells, its role may not have been detected due to the kinetics of pre-replication complex assembly. Normally pre-replication complex assembly begins at telophase of the previous mitosis (Mendez and Stillman, 2000) and it is possible that the Cdc7-mediated phosphorylation of the Mcm2 aminoterminus occurs during or prior to mitosis. Re-entry into the cell cycle from quiescence must pose the unique problem of requiring these phosphorylations de novo in a low-kinase environment. In this context, cells may depend on activation of E2F-mediated transcription to accumulate Cdc7, explaining in part the special requirement for cyclin E-Cdk2 during cell cycle reentry.
Based on the experiments reported here, phosphorylation of Ser4, 5, and 7 are essential but not sufficient for Mcm2 chromatin loading in the absence of Cdc7 kinase activity upon cell cycle reentry. Others have published additional Cdc7 specific sites in the aminoterminal region of Mcm2 based both on in vitro and in vivo experiments (Cho et al., 2006; Montagnoli et al., 2006; Tsuji et al., 2006). Unfortunately, these reports do not arrive at a consensus concerning which sites are specifically targeted by Cdc7 in vitro or in vivo. Our own studies overlapped with but did not completely concur with any of the published studies. Therefore, we created a phosphomimetic mutant converting the seven most aminoterminal phosphorylation sites attributed to Cdc7 by our analysis and those in the three published papers to aspartate: Ser4, 5, 7, 27, 40, 41, 53D. This mutant is capable of completely bypassing the requirement for Cdc7 with respect to chromatin loading (Fig. 5b), suggesting that extensive phosphorylation of the Mcm2 aminoterminus is required. Only one of the studies carried out a functional analysis concerning the implications of Mcm2 phosphorylation by Cdc7 (Tsuji et al., 2006). They showed that a triple alanine substitution at positions Ser27, 41 and 139 was incapable of supporting DNA replication, but did not affect chromatin loading. Taken together, all of these results suggest that the extreme aminoterminal Cdc7 phosphorylation sites on Mcm2 (Ser4, 5, 7) are of primary importance for chromatin loading and only a subset of the other reported aminoterminal sites is additionally required.
In vitro studies have suggested that Mcm2 is an inhibitor of a DNA helicase composed of Mcm4, Mcm6 and Mcm7 (Ishimi et al., 2001). However, genetic studies in yeast and drosophila clearly indicate an essential and positive role of Mcm2 in DNA replication (Treisman et al., 1995; Yan et al., 1991). More recently, however, studies of a larger GINS-Cdc45-Mcm2–7 complex as well as of the MCM hexamer alone confirms that Mcm2 is an essential component of a putative replicative DNA helicase in vitro (Bochman and Schwacha, 2008; Moyer et al., 2006). In light of our findings, it is likely therefore that Mcm2 via aminoterminal phosphorylation contributes to recruiting the other MCM subunits to chromatin, at least in the context of cell cycle reentry.
Immortalized human fibroblasts (IHFs, a gift from J. Shay), 293A cells, and 293 phoenix ampho cells were cultured in DMEM medium (Gibco) supplied with 10% newborn calf serum (Gemini Bio.), penicillin, streptomycin, and glutamate (Gibco). Immortalized human diploid mammary epithelials (IMEs, a gift from J. Shay) were grown in MCDB131 (Gibco) supplied with insulin (Sigma), bovine pituitary extract (Hammond), holo-transferrin (Sigma), hydrocortisone (Sigma), recombinant human EGF (Invitrogen), 1% fetal bovine serum, penicillin, streptomycin, and glutamate (Gibco). Cells were grown in an atmosphere of 5% CO2 and at 37 °C. Sf9 cells were cultured in Ex-Cell 401 supplied with penicillin, streptomycin, and glutamate (Gibco) at 27 °C. Reagents used for cell culture were roscovitine (LC Labs), mimosine (Sigma), thymidine (Sigma), MG132 (Sigma), and G418 (Gibco).
IME cells were synchronized by growth until confluence, subsequent incubation in the absence of serum and growth factors for 2 days, and then release into medium with 1% serum and complete growth factors for the indicated times. 293A cells were synchronized using either mimosine or thymidine, as indicated.
The following antibodies were used: anti-Mcm2, 4, 6, and 7 and Orc2 were obtained from B. Stillman, anti-Cdc6 (180.2, Santa Cruz), anti-Cdk2 (M2, Santa Cruz), anti-cyclin E1 (HE12, Ascites), anti-HA (12CA5, ascites), anti-Cdc7 (DCS-342, MBL), anti-β actin (Sigma) anti-pRb (C15, Santa Cruz). Anti-pSer5 of MCM2 was made against MCM2 N-terminal peptide (MAESpSESFTMAC-amide) conjugated with Keyhole Limpet Hemocyanin (QCB; Hopkinton, MA).
For mapping of in vitro phosphorylation sites, endogenous MCM complexes were purified from nuclear extract of IHFs and phosphorylated by cyclin E-Cdk2 complexes purified from insect cells. See Supplementary Materials for protocol details.
Either MCM complexes purified from nuclear extract or recombinant proteins were incubated with recombinant kinases in 50 mM Hepes, pH 7.4, 15 mM MgCl2, 25 mM beta-glycerol phosphate, pH 7.4, 1 mM EGTA, 0.05% NP-40 and cold ATP (10mM, Sigma) or γ-32P-ATP (MP Biomedicals). Reactions were incubated at 30°C for 30 minutes and then stopped by adding SDS-PAGE sample buffer and boiling or addition of TCA/acetone for mass spectroscopy analysis.
Cyclin E specific monoclonal HE172 was used to immunoaffinity purify cyclin ECdk2 complexes from cell extract. Immune complexes were then bound to GammaBind G Sepharose. After washing 3 times with NETN buffer (50 mM Tris–HCl pH 7.5, 150 mM NaCl, 1 mM EDTA, 0.1% Nonidet P-40), cyclin E-bound beads were used for histone H1 kinase assays.
Human Mcm2 cDNA was amplified from a plasmid (a gift from Kazusa DNA Research Institute) by PCR using primers that fused an HA tag to the N-terminus of the protein and then cloned to BamHI and SnaBI sites of pBabe-Neo. pBabe-Neo-HA-MCM2 was used for subsequent site-directed mutagenesis. All mutants were confirmed by DNA sequencing. For recombinant protein expression using the baculovirus system, cDNAs were cloned into pFast or pFastBac1 (Invitrogen). For transient transfections in 293A cells, cDNAs were cloned into vector pcDNA3.
Cells (either 293A or IME) were harvested using trypsin and pellets were lysed by incubating in complete cytoskeleton (CSK) buffer (20 mM Hepes, pH 7.4, 100 mM NaCl, 3 mM MgCl2, 300 mM Sucrose, 0.1 % NP-40) for 15 minutes on ice. Detergent soluble fractions were obtained as supernatants after low speed centrifugation (3000 × g) at 4 °C. Pellets rinsed with complete CSK buffer for 10 min on ice were recentrifuged to obtain a chromatin-enriched fraction. Pellets were then sonicated for 5 seconds in CSK buffer or treated with DNAse I (Roche) in CSK buffer for 30 minutes (as indicated in the figure) and subjected to high-speed centrifugation (16,000 × g). The post-sonication or DNAse treatment supernatant was designated the chromatin-bound fraction.
HA-MCM2 WT or mutants in pBabe-Neo were transfected into the packaging cell lines 293 phoenix ampho by the calcium phosphate method for retrovirus production. Harvested retrovirus was then transduced into 293A or IME cells. Transduced cells were selected in medium with G418 (Gibco). After 10–14 days of selection, cells were used for chromatin loading assays.
The adenovirus expressing dominant negative Cdk2 (Cdk2-dn) was made based on the dominant negative allele described by van den Heuvel and Harlow (van den Heuvel and Harlow, 1993). The adenoviruses expressing wild-type and dominant negative Cdc7 (Cdc7-dn) were obtained from Dr. X. Wu (The Scripps Research Institute). Cdc6-D4-NLS (Delmolino et al., 2001) was cloned from a plasmid obtained from A. Dutta into an adenovirus vector and packaged into a recombinant adenovirus. Amplification of adenovirus was carried out using 293A cells. Pure virus stocks were prepared by banding in CsCl gradients at 30000 rpm for 12 hours at 4°C. Adenovirus bands were collected and dialyzed against TBS. Adenovirus stocks were stored at −80°C in TBS with 10% glycerol.
Proteins were purified either from either baculovirus infected insect cells or from E. coli. See Supplemental Materials for protocol details.
Detergent soluble and chromatin-bound fractions were prepared without phosphatase inhibitors. Then, calf intestinal phosphatase (NEB) was added and incubated for 30 minutes at 37 °C. The reaction was stopped by adding SDS-PAGE sample buffer and boiling.
Overexpression of wild-type Mcm2 or mutants in 293A or U2OS cells was accomplished by transient transfection using Lipofectamine 2000 (Invitrogen). Transfection mix was removed after 12 hours after which cells were then incubated in complete medium for 12 hours and synchronized at G1-S using 0.5 mM mimosine for 24 hours before harvesting (293A cells). In the case of U2OS cells, cells were harvested 24 hours after transfection without inhibitors.
IME cells were kept in complete medium until confluence. Then, siRNA targeting GFP and cyclin E1/E2 (siGenome; Dharmacon) were transfected using Lipofectamine 2000 for 48 hours in the absence of serum and growth factors. Cells were released for 15 hours and then harvested for chromatin loading assays.
This work was supported by NIH grant CA78343 to S.I.R. L.-C. C, L.K.T. and J.A.W. were supported by postdoctoral fellowships from the Susan G. Komen Breast Cancer Foundation, the Pew Latin American Fellows Program and the American Cancer Society, respectively. J.R.Y was supported by NIH P41 RR011823. We are grateful to Dr. B. Stillman, Dr. X. Wu, Dr. A Dutta, Dr. D. Tedesco and Dr. F. van Drogen for sharing materials and to Dr. T. Tsuji and Dr. W. Jiang for sharing preliminary results and materials used in this project.
Publisher's Disclaimer: This is a PDF file of an unedited manuscript that has been accepted for publication. As a service to our customers we are providing this early version of the manuscript. The manuscript will undergo copyediting, typesetting, and review of the resulting proof before it is published in its final citable form. Please note that during the production process errors may be discovered which could affect the content, and all legal disclaimers that apply to the journal pertain.