In the present study, we report the identification of Cdc7/Dbf4 phosphorylation sites on MCM2 and determine the functional role of Cdc7/Dbf4 phosphorylation of MCM2 in the initiation of DNA replication in human cells. Previous studies demonstrated that the S phase–promoting kinase Cdc7/Dbf4 and the putative DNA replicative helicase MCM2-7 play essential roles in the initiation of DNA replication in all eukaryotes. We and others showed that Cdc7/Dbf4 selectively phosphorylates the MCM2 subunit of the MCM complex, suggesting that Cdc7/Dbf4 may be directly involved in the initiation of DNA replication by targeting MCM2 (Lei et al., 1997
; Brown and Kelly, 1998
; Jiang et al., 1999a
; Jares and Blow, 2000
). To determine the importance of Cdc7/Dbf4 phosphorylation of MCM2 in the initiation of DNA replication, we mapped Cdc7/Dbf4 phosphorylation sites in human MCM2. Our results showed that human MCM2 is phosphorylated in vivo at three major sites (Ser27, Ser41, and Ser139) and two minor sites (Ser53 and Ser108). Because Ser27, Ser41, and Ser139 in MCM2 are also specifically phosphorylated by purified Cdc7/Dbf4 in vitro and their in vivo phosphorylation was greatly reduced in Cdc7 siRNA treated cells, it strongly suggests that Ser27, Ser41, and Ser139 are in vivo Cdc7/Dbf4 phosphorylation sites in MCM2.
Sequence alignments show that Ser27, Ser41, and Ser139 of human MCM2 are highly conserved in higher eukaryotes (from Drosophila
to mammals). However, the N-terminal sequences of MCM2 between human and lower eukaryotes, such as S. cerevisiae
and S. pombe
, are highly divergent. Therefore, we could not define analogous phosphorylation sites in MCM2 from these species. Nonetheless, previous studies have demonstrated that yeast MCM2 is a critical downstream target of Cdc7/Dbf4 kinase (Lei et al., 1997
; Brown and Kelly, 1998
). Careful analysis of the surrounding sequences of Ser27, Ser41, and Ser139 in human MCM2 did not reveal consensus sites for Cdc7/Dbf4 phosphorylation. The Ser27 and Ser41 sites are similar to the Cdk phosphorylation consensus site (X-S/T-P-X-K/R), whereas the Ser139 site is similar to the casein kinase II (CKII) phosphorylation consensus site (X-S/T-X-X-D/E). Previous studies suggested that MCM2 might be a target of Cdk and Ddk (Masai et al., 2000
; Montagnoli et al., 2006
). Because phosphorylation of MCM2 was not affected in HeLa cells treated with Cdk2 siRNA (or Cdc2 siRNA, T. Tsuji and W. Jiang, unpublished observation), our results indicated that MCM2 was not directly targeted by Cdks in vivo (D). It was shown that Cdc7 belongs a subfamily of protein kinase closely related to the Cdk subfamily and the CKII subfamily (Manning et al., 2002
). These results together with ours presented in this study indicate that Cdc7 phosphorylates similar primary sequences of substrates as Cdk does. However, recognition of a ternary structure of a substrate by Cdc7 and/or Dbf4 or subcellular colocalization of a substrate with Cdc7/Dbf4 might be crucial for determining the Cdc7/Dbf4 kinase-substrate specificity. Consistent with this idea, unlike the purified MCM2 protein, a MCM2 Ser139 peptide (MRRGLLYDSDEEDEERPA) fused with GST protein could not be phosphorylated by purified Cdc7/Dbf4 in vitro (unpublished data).
While this manuscript was being prepared for submission, Montagnoli et al. (2006)
reported identification of MCM2 phosphorylation sites by S phase–regulating kinases. They found that MCM2 was phosphorylated in vitro by Cdks at Ser13, Ser27, and Ser41, by Ddk at Ser40, Ser53, and Ser108, and by CKII at Ser139. The basis for the discrepancies between their results and ours is currently unclear. One possibility is that the purified Ddk used in their and our in vitro phosphorylation studies might have different specific activity. In our study, highly purified Cdc7/Dbf4 strongly phosphorylated MCM2 in vitro, whereas kinase-dead Cdc7kd/Dbf4 did not (Supplementary Figure S1). These results ruled out the possibility that a contaminating kinase(s) copurified with the Cdc7/Dbf4. We showed that Ser27, Ser41, and Ser139 of MCM2 were specifically phosphorylated by Cdc7/Dbf4 but not by Cdc7kd/Dbf4 in vitro and that phosphorylation of these sites was also detected in vivo. In contrast, phosphorylation of Ser26 and Ser40 in MCM2 by Cdc7/Dbf4 was not detected in vitro and phosphorylation of these sites was not detected in vivo either (Supplementary Figure S2). Although the amino acid sequences around Ser26 and Ser40 are the same (LTSSPGR). Montagnoli et al. (2006)
could only detect Cdc7/Dbf4 phosphorylation of Ser40, but not Ser26. In addition, Montagnoli et al.
showed that MCM2 was phosphorylated by CKII but not Cdc7/Dbf4 in vitro. However, they also showed that phosphorylation of Ser139 in vivo was affected by serum starvation, suggesting that phosphorylation of Ser139 was regulated in vivo. Because the activity of CKII is not regulated during the cell cycle, they were not sure if CKII is responsible for Ser139 phosphorylation in vivo (see discussion section in Montagnoli et al., 2006
). In contrast, we clearly showed that MCM2 Ser139 was phosphorylated by Cdc7/Dbf4 in vitro and phosphorylation of this site in vivo is required for chromatin recruitment of MCM complex and the initiation of DNA replication during the cell cycle (, , and ). Taken together, our results strongly indicated that Ser27, Ser41, and Ser139 are Cdc7/Dbf4 phosphorylation sites.
We examined chromatin recruitment and phosphorylation of chromatin-bound MCM2 by Cdc7/Dbf4 using immunoblotting, immunofluorescence, and high-speed automated cell-imaging analyses with antibodies specific against MCM2 and Cdc7/Dbf4 phosphorylated MCM2. We found that MCM2 was not phosphorylated by Cdc7/Dbf4 in G1 and unphosphorylated MCM2 accumulated on chromatin. Chromatin-bound MCM2 was phosphorylated by Cdc7/Dbf4 during G1/S and early S phase, which coincided with the initiation of DNA replication. In late S and G2/M, phosphorylated MCM2 gradually disassociated from chromatin. Previous studies showed that, unlike other essential DNA replication proteins such as RPA and PCNA, MCM proteins fail to colocalize with sites of DNA replication through S phase (Todorov et al., 1994
; Romanowski et al., 1996
; Dimitrova et al., 1999
; Laskey and Madine, 2003
). The results suggest that either MCM complex does not preferentially colocalize with sites of DNA replication or MCM complex could not be detected at replication sites by certain antibodies due to conformational changes induced by post-translational modification, such as phosphorylation. We show that, like MCM2, chromatin-bound Cdc7/Dbf4 phosphorylated MCM2 does not colocalize with replication foci during G1/S and S phase, indicating that failure of detection of MCM2 at replication foci is not due to the use of antibodies that cannot recognize phosphorylated MCM2. Thus, our results, together with previous studies, indicate that MCM complex does not preferentially colocalize with sites of DNA replication (Todorov et al., 1994
; Romanowski et al., 1996
; Dimitrova et al., 1999
; Laskey and Madine, 2003
). How could the DNA replicative MCM helicase fail to colocalize with sites of DNA replication? Recently, Laskey and Madine (2003)
proposed a rotary pumping model to explain the discrepancy. They suggested that MCM helicase could work at a distance from replication forks as a rotary motor that pumps DNA along its helical axis by simple rotation, such that the movement resembles that of a threaded bolt through a nut.
All six MCM proteins (MCM2-7) are members of the diverse AAA ATPase family, and it is known that the purified recombinant yeast and Xenopus
MCM2-7 heterohexameric complex has ATPase activity (Tye, 1999
; Schwacha and Bell, 2001
; Davey et al., 2003
; You et al., 2003
). We examined if Cdc7/Dbf4 phosphorylation of MCM2 affects the formation, conformation, and/or ATPase activity of MCM2-7 complex in vitro. We show that MCM2, its nonphosphorylatable mutant MCM2A or its phosphomimetic mutant MCM2E could copurify with MCM3-7 as a heterohexameric complex from insect Sf9 cells using a baculovirus expression system. These results indicate that phosphorylation of MCM2 does not affect MCM complex formation. However, we show that the MCM2-7 complex or the MCM2E-7 complex displays much higher ATPase activity in vitro than the MCM2A-7 complex. Previous studies showed that the purified yeast globular heterohexamer MCM2-7 complex did not yield any detectable enzymatic or nucleotide binding activities (Adachi et al., 1997
). In contrast, the purified ring-shaped (MCM4,6,7)2
subcomplex displayed ATP-dependent DNA helicase activity (Sato et al., 2000
). Therefore, it was proposed that activation of helicase activity of heterohexamer MCM 2-7 complex in vivo at replication origins requires a series of post-translational modifications, such as phosphorylation, of MCM proteins that convert the globular inactive heterohexamer MCM2-7 complex conformation to the ring-shaped active heterohexamer MCM 2-7 complex conformation (Tye and Sawyer, 2000
). Our unpublished results suggested Cdc7/Dbf4 phosphorylation of MCM2 might induce a conformational change of MCM2-7 complex as it was assayed by a protease partial digestion analysis (T. Tsuji and W. Jiang, unpublished observation and Herbig et al., 1999
; Mizushima et al., 2000
). Thus, alteration of the confirmation of MCM2-7 complex by Cdc7/Dbf4 phosphorylation of MCM2 could be one of the critical steps for activation of the ATPase-coupled helicase activity of MCM2-7 complex. Further studies, such as electron microscopic observation and the determination of the structure of MCM2–7 complex (Sato et al., 2000
; Fletcher et al., 2003
), are needed to define how Cdc7/Dbf4 phosphorylation of MCM2 regulates the conformational change and the enzymatic activity of MCM2-7 complex in the future.
We provide direct evidence that Cdc7/Dbf4 phosphorylation of MCM2 is required for the initiation of DNA replication in human cells. Consistent with previous biochemical and genetic studies in various organisms (Todorov et al., 1994
; Forsburg et al., 1997
; Kubota et al., 1997
; Lei et al., 1997
), we show that suppression of MCM2 expression by MCM2 siRNA inhibits DNA replication in HeLa cells, indicating that MCM2 is essential for DNA replication in human cells. The inhibition of DNA replication in MCM2 siRNA-treated cells can be rescued by coexpression of MCM2 or its Cdc7/Dbf4 phosphomimetic MCM2E mutant but not its Cdc7/Dbf4 nonphosphorylatable MCM2A mutant. These results, together with the result that Cdc7/Dbf4 phosphorylation of MCM2 is important for the ATPase activity of MCM2–7 complex in vitro, demonstrate that Cdc7/Dbf4 phosphorylation of MCM2 is essential for the initiation of DNA replication in human.
On the basis of our results, together with the current model for the initiation of DNA replication (Bell and Dutta, 2002
; Mendez and Stillman, 2003
; Forsburg, 2004
), we propose a model for regulation of the initiation of DNA replication by Cdc7/Dbf4 phosphorylation of MCM2 in human cells. In G1, when Cdc7/Dbf4 activity is at a minimum, MCM2 is not phosphorylated by Cdc7/Dbf4. Unphosphorylated MCM2 together with other MCMs (MCM2-7 complex) is recruited to chromatin, presumably to the replication origins by the DNA replication loading factors Cdc6 and Cdt1 to establish pre-RCs. During G1/S and early S, activation of Cdc7/Dbf4 results in phosphorylation of chromatin-bound MCM2. Phosphorylation of MCM2 in the MCM2-7 complex by Cdc7/Dbf4 induces the conformational change of the complex and activates its helicase activity, which is essential for DNA replication. MCM2-7 helicase works at a distance from the replicative forks, pumping DNA along its helical axis by ATPase-coupled rotation. In late S and G2/M, MCM2-7 complex dissociates from chromatin, presumably by additional post-translational modifications to prevent DNA rereplication.