Initiation of chromosome DNA replication in eukaryotes is tightly regulated through assembly of replication factors at replication origins. Here, we investigated dependence of the assembly of the initiation complex on particular factors using temperature‐sensitive fission yeast mutants. The psf3‐1 mutant, a GINS component mutant, arrested with unreplicated DNA at the restrictive temperature and the DNA content gradually increased, suggesting a defect in DNA replication. The mutation impaired GINS complex formation, as shown by pull‐down experiments. Chromatin immunoprecipitation assays indicated that GINS integrity was required for origin loading of Psf2, Cut5 and Cdc45, but not Sld3. In contrast, loading of Psf2 onto origins depended on Sld3 and Cut5 but not on Cdc45. These results suggest that Sld3 functions furthest upstream in initiation complex assembly, followed by GINS and Cut5, then Cdc45. Consistent with this conclusion, Cdc7‐Dbf4 kinase (DDK) but not cyclin‐dependent kinase (CDK) was required for Sld3 loading, whereas recruitment of the other factors depended on both kinases. These results suggest that DDK and CDK regulate distinct steps in activation of replication origins in fission yeast.
Initiation of DNA replication is a dynamic reaction regulated during the cell cycle in eukaryotic cells. This is achieved through ordered assembly and disassembly of a number of replication factors onto specific chromosome loci, called replication origins (Bell and Dutta, 2002; Kearsey and Cotterill, 2003). The origin recognition complex (ORC) binds to replication origins throughout the cell cycle, at least in yeasts, although the mechanisms of recognition of origins may differ among organisms (Gilbert, 2001). When cells exit from mitosis, the mini‐chromosome maintenance (MCM) complex binds to replication origins, dependent on ORC, Cdc6/Cdc18 and Cdt1, forming pre‐replicative complexes (pre‐RCs) (Diffley et al, 1994). Activation of pre‐RCs requires further recruitment of several factors in S phase, including Cdc45 and Sld3, dependent on cyclin‐dependent protein kinase (CDK) and Cdc7‐Dbf4 protein kinase (Dbf4‐dependent kinase, DDK) (Takeda and Dutta, 2005). This process is considered to activate the DNA helicase of MCM complex, resulting in origin DNA unwinding and assembly of RPA and DNA polymerases for initiation of DNA synthesis (Bell and Dutta, 2002).
Many of the replication factors involved in activation of replication in S phase have been identified through genetic screening in budding yeast. For example, DPB11 has been isolated as a multicopy suppressor of a mutant of DPB2 that encodes the second largest subunit of DNA polymerase ε (Araki et al, 1995). Three essential factors, Sld2, Sld3 and Sld5, have been identified by screening of synthetic lethal mutations with dpb11‐1 (Kamimura et al, 1998). Interaction of Sld2/Drc1 with Dpb11/Cut5/Rad4 is required for DNA replication in budding as well as fission yeasts (Masumoto et al, 2002; Noguchi et al, 2002). Sld3 has genetic and physical interactions with Cdc45 (Kamimura et al, 1998; Nakajima and Masukata, 2002). Sld5, Psf1 (partner of Sld five 1), Psf2 and Psf3 are components of a recently discovered Go, Ichi, Nii, and San; five, one, two, and three in Japanese (GINS) complex in budding yeast and Xenopus (Kanemaki et al, 2003; Kubota et al, 2003; Takayama et al, 2003). Although all these replication factors are essential for initiation of chromosome DNA replication, precise mechanisms of assembly of individual factors or functions in DNA replication have yet to be elucidated. In budding yeast, Sld3 and Cdc45 associate with early‐firing replication origins in G1 phase in a mutually dependent manner (Kamimura et al, 2001), and the remaining factors bind to origins at the onset of S phase (Takayama et al, 2003). In contrast, fission yeast Sld3 is required for origin loading of Cdc45, but the reverse is not the case (Yamada et al, 2004). Relations of GINS with Sld3, Cut5 and Cdc45 in the process of origin binding remain unclear. In Xenopus extracts, chromatin binding of Cdc45, DNA Polε and DNA Polα depends on GINS, whereas chromatin association of Cut5 does not (Hashimoto and Takisawa, 2003; Kubota et al, 2003), suggesting that the precise processes of initiation complex formation differ among species. Thus, more information is required for understanding of the general scheme of initiation of replication.
Activation of replication origins at the onset of S phase requires two different serine/threonine kinases, CDK and Cdc7‐Dbf4 kinase (DDK). S‐phase CDK is required for tight association of Cdc45 with chromatin in budding yeast and Xenopus egg extracts (Mimura et al, 2000; Zou and Stillman, 2000). Sld2 is the only protein, so far, whose phosphorylation by CDK is required for initiation of replication (Masumoto et al, 2002; Tak et al, 2006) and this requirement is conserved in fission yeast (Noguchi et al, 2002). Phosphorylation of Sld2/Drc1 is required for the interaction with Dpb11 (Masumoto et al, 2002). On the other hand, reactions regulated by DDK have not been identified, although it phosphorylates subunits of MCM and Cdc45 (Lei et al, 1997; Weinreich and Stillman, 1999; Nougarede et al, 2000; Takeda et al, 2001). The fact that the bob1‐1/mcm5 mutation bypasses the requirement for CDC7 and DBF4 in budding yeast replication (Hardy et al, 1997) suggests, however, that MCM complex phosphorylation by DDK does play an important role in initiation of replication.
In this study, we focused on dependence on replication factors in initiation complex formation, using a novel temperature‐sensitive mutant for the GINS component. The results suggest that Sld3, GINS and Cdc45 act sequentially in initiation complex assembly. Consistent with this model, origin association of Sld3 depends on DDK but not on CDK, whereas recruitment of GINS and Cdc45 requires both DDK and CDK. Thus, DDK and CDK regulate different steps in initiation complex assembly.
Requirement of GINS in early S phase
To elucidate the role of GINS in initiation complex formation upon onset of S phase, we isolated a temperature‐sensitive (ts) mutant, psf3‐1, by integration of randomly mutated psf3 gene into the psf3+ locus (see Materials and methods). To examine the possible defect of psf3‐1, which alters the valine residue at position 56 into aspartic acid and the isoleucine at position 159 into leucine (V56D and I159L; Supplementary Figure S1), in DNA replication, wild‐type and mutant cells grown at 25°C were shifted to 36°C and cellular DNA contents were analyzed by flow cytometry. The DNA contents of wild‐type cells remained in 2C, owing to the long G2 phase in the fission yeast cell cycle (Figure 1A). In contrast, 1C DNA cells accumulated with psf3‐1 at 1–3 h and then the DNA content gradually increased at 4–5 h (Figure 1A), showing that DNA replication did not correctly initiate or proceed. A small fraction of cells with less than 1C DNA emerged at 4–6 h, probably resulting from cell division without completion of DNA replication. Cell viability of psf3‐1 was maintained for 2 h and then slowly decreased to about 60% by 6 h (Supplementary Figure S2). These results suggest that psf3‐1 cells arrest at a reversible step in DNA replication before they go into untimely mitosis.
GINS complex formation is impaired by psf3‐1
For investigation of the effect of psf3‐1 mutation on DNA replication at the molecular level, we investigated interactions of FLAG‐tagged Psf2 with other GINS components. FLAG‐tagged derivatives of wild type (psf2‐flag) and the mutant (psf2‐flag psf3‐1) were grown at 25°C or shifted to 36°C for 2 h and proteins in the cell extracts were analyzed by immunostaining with anti‐FLAG antibody (Figure 1B, lanes 1–5). In case of wild type, a peptide migrating at around 30 kDa, slightly larger than the predicted molecular mass for Psf2‐FLAG (23 kDa), was detected (Figure 1B, top panel, lanes 2 and 3), whereas the corresponding band was not present in the untagged strain (lane 1). By immunoprecipitation (IP) with FLAG‐antibody, Psf2‐FLAG was detected in psf2‐flag but not untagged cells (Figure 1B, top panel, lanes 6–8). Immunostaining of IP products with anti‐Psf3, Psf1 and Sld5 antibodies showed that Psf3 (around 20 kDa, Figure 1B, the second panel, lanes 7 and 8), Psf1 (around 29 kDa, the third panel) and Sld5 (around 35 kDa, the bottom panel) were co‐precipitated with Psf2‐FLAG. None of these components was detected in the IP fraction from the untagged cells (Figure 1B, lane 6). These results show that all GINS components form a complex.
In the psf3‐1 mutant, Psf3 detected by immunostaining was reduced both in the permissive and restrictive temperatures compared with those in wild type (Figure 1B, second panel, lanes 2–5). Notably, Psf3 was not co‐immunoprecipitated with Psf2‐FLAG at the restrictive temperature, although decreased but significant amount of Psf3 was detected in the IP fraction at the permissive temperature (Figure 1B, second panel, lanes 7–10). These results indicate that the intact GINS complex is not formed at the restrictive temperature, probably because the psf3‐1 mutant protein is unstable or defective in interaction with the other GINS components. It should be noted, however, that the amounts of Psf1 and Sld5 recovered in the Psf2‐IP fraction from psf3‐1 were comparable to those in the wild type (lanes 7–10). These results suggest that Psf3 is not required for assembly of the remaining complex, although it is essential for the GINS function.
Integrity of the GINS complex is required for its localization at replication origins
To examine effects of the psf3‐1 mutation on association of GINS with replication origins, localization of Psf2‐FLAG on chromosomes was analyzed by chromatin immunoprecipitation (ChIP) assay. For synchronization of the cell cycle, the cold‐sensitive mutation nda3‐KM311, which is defective for β‐tubulin and causes metaphase arrest (Hiraoka et al, 1984), was employed to avoid premature inactivation of temperature‐sensitive psf3‐1 mutation, and the arrested cells were released at the restrictive temperature for psf3‐1 (Figure 2A). Hydroxyurea, which depletes dNTP pools in cells, was added to wild‐type cells at the release to detect transient localization of replication factors at replication origins more readily. In a later section, we have confirmed origin localization of factors in hydroxyurea‐free wild‐type cells (shown in Figure 6 and Supplementary Figure S3). Flow cytometry analysis showed that 1C DNA cells accumulated with both wild type (+HU) and psf3 (−HU) (Figure 2B). The viability of psf3‐1 did not decrease for 120 min under these conditions, suggesting that cells were arrested at a reversible stage (Supplementary data and Supplementary Figure S2).
We first examined localization of Mcm6 and Psf2‐FLAG in wild‐type cells released from M phase and subsequently arrested with HU. Two sets of primers, amplifying the ars2004 locus, an efficient replication origin, and non‐ars1, 30 kb away from the origin (Figure 2C), generated similar amounts of PCR products from total cellular DNA without IP (Figure 2D, lane W as whole‐cell extract). With Mcm6‐IP, the ars2004 fragment was preferentially enriched at 60–120 min period (Figure 2D, upper bands), whereas the non‐ars1 fragment was not (Figure 2D, lower bands), consistent with G1–S‐specific localization of MCM at replication origins (Ogawa et al, 1999). IP with Psf2‐FLAG yielded amplification of the ars2004 fragment in similar periods with Mcm6‐IP (Figure 2D, lower panel). In a control experiment using immunoprecipitated DNA from untagged cells, neither fragment was significantly amplified (data not shown). IP of Psf2‐FLAG from synchronized cells without HU yielded preferential recovery of the ars2004 fragment (Supplementary Figure S3). These results show that Psf2 binds to the replication origin with a similar timing as Mcm6.
In the psf3‐1 mutant released at the restrictive temperature without HU, the ars2004 fragment was enriched at 30–120 min with Mcm6‐IP (Figure 2E, upper panel), showing that MCM was localized at the origin as in the HU‐arrested wild type. In contrast, the ars2004 fragment was not preferentially enriched by Psf2‐IP at any time point (Figure 2E, lower panel), indicating that the origin association of Psf2 was abolished by the psf3‐1 mutation. Taken together with the results shown in Figure 1B, the integrity of GINS complex is required for association with replication origin. We confirmed that localization of Psf2 at another active replication origin AT2024 (Segurado et al, 2003) was also impaired by the psf3‐1 mutation (Supplementary Figure S4).
Genetic interactions of psf3+ with initiation factors
To identify replication factors that function in close relation with Psf3, we examined suppression of temperature sensitivity of psf3‐1 by increasing expression of the gene cloned on a multicopy plasmid. On incubation at the restrictive temperature after serial dilution, the temperature sensitivity was suppressed by sld5+, psf1+ and psf2+, which encode components of GINS (Figure 3A), consistent with the finding that GINS complex formation is impaired by psf3‐1 mutation (Figure 1B). In addition, the temperature sensitivity of psf3‐1 was suppressed by drc1+, a fission yeast counterpart of budding yeast SLD2, but not significantly changed by sld3+ (Figure 3A), cut5+ or cdc45+ (data not shown). These results suggest that Drc1 acts in the same or closely related step with Psf3 in DNA replication.
It has been shown that combining two mutations of different genes results in synthetic lethality or enhanced temperature sensitivity when corresponding gene products function in the same process. To explore genetic relations of the other replication factors with Psf3, double mutants carrying psf3‐1 and a mutation of a component of the pre‐RC, the initiation complex or DNA polymerases were constructed. As shown in Figure 3B, psf3‐1 cut5‐T401 and psf3‐1 cdc20‐P7 (Polε) double mutants exhibited more severe sensitivity than parental strains, whereas the sensitivity of psf3‐1 pol1‐1 (Polα) did not significantly differ from those of single mutants. The results of similar experiments are summarized in Figure 3C. The temperature sensitivity of psf3‐1 was enhanced by sld3‐41, sna41‐928/cdc45 and cdc20‐M10 (Polε). In contrast, the sensitivity was not affected by orp1‐4 and mis5‐268/mcm6, which are components of pre‐RC, or by spp2‐9 (primase) (Figure 3C). These results suggest that Psf3 functions in a process together with Sld3, Cdc45, Cut5, which are components of the initiation complex, and DNA Polε, which may have a role in initiation complex formation rather than DNA synthesis (Feng and D'Urso, 2001).
GINS is required for Cdc45 loading but not for Sld3 loading
We have shown that Sld3 and Cdc45 exhibit different dependence with regard to origin loading (Yamada et al, 2004). To examine whether GINS is required for origin loading of Sld3 and Cdc45, ChIP assays were carried out with wild type and psf3‐1 mutant expressing FLAG‐tagged Sld3 or Cdc45. In wild‐type cells released from M phase in the presence of HU, Mcm6, Sld3 and Cdc45 associated with the ars2004 origin but not non‐origin region at 30–120 min (Figure 4A). In contrast, in psf3‐1 mutant released without HU, localization of Cdc45 at the origin was greatly reduced, whereas Mcm6 and Sld3 were localized as in wild type (Figure 4B). Absence of Cdc45 at the origin was not due to an absence of HU, because it bound to the origin in wild type incubated without HU (Yamada et al, 2004). Specific reduction of Cdc45 loading onto another efficient replication origin, AT2024, was confirmed (Supplementary Figure S4). These results show that GINS is required for recruitment of Cdc45, but not for Sld3 and Mcm6 loading. As it has been suggested that Cdc45 is required for DNA helicase activity of MCM complex (Masuda et al, 2003), we examined binding of Rpa2, the second largest subunit of RPA complex that binds to single‐stranded DNA. As expected, Rpa2 bound to ars2004 origin in HU‐arrested wild type but not in psf3‐1 (Figure 4A and B).
To examine whether GINS is required for recruitment of Cut5, Drc1 or DNA Polε, origin localization of FLAG‐tagged Cut5, Drc1 or Dpb2, which is the second largest subunit of DNA Polε, was analyzed by ChIP assay. In wild‐type cells released from M phase in the presence of HU, Cut5, Drc1 and Dpb2 were localized at the replication origin ars2004 at 60–90 min (Figure 4A). Origin‐specific localization of these factors was confirmed in the absence of HU (Supplementary Figure S3; data not shown). In contrast, localization of these factors at the origin was greatly decreased in psf3‐1 cells at the restrictive temperature in the absence of HU (Figure 4B), indicating that GINS is required for recruitment of Cut5, Drc1 and DNA Polε.
Sld3 and Cut5 but not Cdc45 are required for GINS recruitment
As GINS was required for loading of Cut5 and Cdc45 but not for Sld3, we investigated whether loading of GINS, in turn, depended on these factors. sld3, cut5 and cdc45 mutants expressing FLAG‐tagged Psf2 were released from M phase at the restrictive temperature. Results of flow cytometry analysis showed that 1C DNA cells accumulated, although <1C DNA cells appeared on longer incubation in sld3 and cut5 (Figure 5A). ChIP assays showed that localization of Psf2‐FLAG to the ars2004 origin was severely reduced in sld3 compared with the wild type, whereas Mcm6 was efficiently localized at the origins (Figure 5B, compare with Figure 2). With the cut5 mutant, origin association of Psf2‐FLAG was greatly reduced, similar to sld3 mutant (Figure 5C). In contrast, in cdc45 mutant, Psf2‐FLAG was localized at the origin at 30–120 min after release (Figure 5D). Considering these results and our previous observation that the mutant Cdc45 protein does not bind to replication origins (Yamada et al, 2004), GINS appear to bind to the origin independent of association with Cdc45. Under these conditions, origin DNA did not appear to be unwound, because Rpa2 was not detected in cdc45 mutant (Figure 5D, compare with Figure 4A).
Considering Cdc45‐independent loading of GINS together with mutually dependent loading of GINS and Cut5 (Figures 4B, 5C and D), it can be assumed that Cut5 loading is not dependent on Cdc45. In fact, Cut5‐FLAG was localized at the replication origin in cdc45 mutant released from M phase (Figure 5E). These results are consistent with the idea that GINS and Cut5 are recruited to replication origin independent of Cdc45.
DDK‐dependent and CDK‐independent origin loading of Sld3
Although both DDK and CDK are required for activation of replication origins at the onset of S phase, the specific reactions regulated by these kinases have yet to be elucidated. Dependence of Sld3, GINS and Cdc45 recruitment on DDK and CDK was examined using temperature‐sensitive DDK and CDK mutants, in which kinase activities are decreased (Nurse and Thuriaux, 1980; Booher et al, 1989; Takeda et al, 2001).
For this purpose, nda3 hsk1‐89 (DDKts) and nda3 (wild type) cells expressing Sld3‐FLAG and Cdc45‐Myc or Psf2‐FLAG were released from M‐phase block at the restrictive temperature in the absence of HU. Flow cytometry analysis showed that hsk1 cells were arrested in early S phase, whereas wild‐type cells released without HU proceeded into S phase, as shown by the appearance of >2C DNA cells (Figure 6A). In wild‐type cells, Sld3, Psf2 and Cdc45 were all preferentially localized at the ars2004 at 50–80 min, slightly later than Mcm6 (30–70 min) (Figure 6B). In hsk1 mutant, however, localization of Sld3, Psf2 or Cdc45 to the origin was severely impaired, whereas Mcm6 binding was not affected (Figure 6C). Thus, DDK is required for origin binding of Sld3, GINS and Cdc45.
However, ChIP assays with nda3 cdc2‐33 (CDKts) revealed that Sld3 was localized to the ars2004 origin with a similar timing as in wild type, whereas localization of Psf2 or Cdc45 was greatly reduced (Figure 6D). Very similar results were obtained with two different alleles, cdc2‐M26 and cdc2‐L7 (Supplementary Figure S5), suggesting that Sld3 loading does not depend on CDK. Taken together, DDK but not CDK is required for Sld3 loading, whereas loading of GINS and Cdc45 depends on both kinases. This is consistent with the ordered requirement of Sld3, GINS and Cdc45 in initiation complex formation as described above.
Distinctly ordered assembly of initiation factors at onset of S phase
Individual replication origins are activated at onset of S phase through assembly of replication factors, including Sld3, GINS, Cut5, Drc1, Cdc45, RPA and the DNA polymerase α/primase complex. This process is tightly regulated by CDK and DDK. However, in fission yeast, it has not been elucidated whether the factors assemble onto replication origins in a specific order. Our present results indicate that specific subsets of replication factors bind to replication origins when a certain factor is not functional. These observations allow us to propose specific dependence in origin loading of replication factors.
Because initiation factors are assumed to execute their functions at replication origins, loading onto origins is the essential process. We have shown origin loading of GINS and Cut5 to depend on Sld3, whereas Sld3 binds to origins independently. Furthermore, recruitment of Cdc45 requires Sld3 and GINS (Figure 4), but not vice versa. These results are consistent with our previous observation that Sld3 binds to replication origins independent of Cdc45 in sna41/cdc45 and nda4/mcm5 mutants (Yamada et al, 2004), and indicate that loading of Sld3 is the most upstream reaction required for assembly of initiation complex at the beginning of S phase (Figure 7A, Step 1). GINS and Cut5 are then recruited onto origins depending on Sld3 in a mutually dependent manner (Step 2). Depending on all these factors, Cdc45 is recruited (Step 3). Replication origins may be unwound only after all these factors have assembled, consistent with the previous observation that Cdc45 is required for DNA helicase activity of MCM complex (Masuda et al, 2003).
One may argue that apparent independence in origin loading using conditionally lethal mutants may result from some remaining function of the mutant protein. However, Sld3 was loaded onto origins in psf3‐1 mutant, in which neither GINS complex formation nor loading of Psf2 was detected at the restrictive conditions (Figures 2 and 4). This demonstrates that Sld3 can bind to origins independently from association of GINS, although a possibility that GINS assists Sld3 loading without tight binding to origins cannot be excluded. Consistently, Sld3 was localized at replication origin in the cdc2 mutants, in which GINS did not associate with origins (Figure 6).
In budding yeast arrested in G1 phase in the presence of α‐factor, both Cdc45 and Sld3 associate with the early firing replication origins in a mutually dependent manner, whereas GINS does not bind to origins (Figure 7B, Step 1) (Kamimura et al, 2001). The primary association of Cdc45, called the ‘bound’ state, is independent of CDK activity (Aparicio et al, 1999), whereas the association of Cdc45 becomes a tightly ‘engaged’ state depending on GINS, Dpb11, Sld2 and CDK (Takayama et al, 2003) (Figure 7B, Step 2). These observations suggest that budding yeast Cdc45 have two distinct association states before and after GINS loading. It has recently been shown that Sld3 does not associate or travel with replication fork components once DNA synthesis initiates in budding yeast, whereas both GINS and Cdc45 remain associated with MCM (Gambus et al, 2006; Kanemaki and Labib, 2006). These results are consistent with the idea that Sld3, GINS and Cdc45 have distinct roles in initiation of replication. Possible roles of Sld3 in recruitment of GINS and Cut5 might be conserved between budding yeast and fission yeast, although fission yeast Sld3 may not depend on Cdc45 in the interactions with pre‐RC. The apparent requirement of Sld3 for Cdc45 loading could be a consequence of sequential assembly as described above. However, it may also be possible that Sld3 may participate in the loading of Cdc45 in fission yeast as shown in budding yeast, because tight genetic interactions were observed between sld3+ and cdc45+ (Nakajima and Masukata, 2002). The conserved role of Sld3 may be a kind of molecular chaperon that mediates stepwise formation of MCM–GINS–Cdc45 complex for active DNA helicase.
Regulation of initiation complex assembly by DDK and CDK
Although the requirement of CDK and DDK for initiation of replication is conserved in eukaryotes, the order of execution points is somewhat controversial and may differ among species. CDK function has been shown to be required before DDK in budding yeast (Nougarede et al, 2000), whereas DDK must execute its function before CDK for replication in Xenopus egg extracts (Walter, 2000). In this study, we found that Sld3 was not loaded onto origins in the hsk1‐89 mutant but bound to origins in three different cdc2 mutants, where the CDK activity is below the level required for loading of Psf2, Cut5 and Cdc45 (Figure 6 and Supplementary Figure S5). Therefore, loading of Sld3, which is the most upstream reaction in initiation factor assembly, is regulated by DDK but not by CDK in fission yeast, although this conclusion does not exclude the possibility that pre‐assembled CDK‐dependent complex may be recruited to origins depending on DDK. This may occur for late firing origins. In fission yeast, a large number of replication origins may fire rather stochastically in a domain at the beginning of S phase (Dai et al, 2005; Patel et al, 2006), although a small number of late‐firing origins have been identified (Kim and Huberman, 2001; Yompakdee and Huberman, 2004). In contrast, the timing of firing of individual origin appears to be distinctly determined in budding yeast (Friedman et al, 1997; Yamashita et al, 1997). Requirement of DDK before CDK in the initiation complex assembly in fission yeast and Xenopus egg extracts may be relevant to activation of a large number of origins at the beginning of S phase.
For Sld3 loading, DDK may phosphorylate pre‐RC components or Sld3 itself. MCM subunits are the efficient substrates for DDK phosphorylation in vitro and in vivo (Weinreich and Stillman, 1999; Masai and Arai, 2002). In addition, requirements of Cdc7 and Dbf4 are bypassed by the bob1/mcm5 mutation in budding yeast (Hardy et al, 1997), suggesting the possibility that certain conformational change in the MCM complex may be caused by DDK phosphorylation. Although the molecular mechanisms of interaction between pre‐RC and Sld3 are not yet completely understood, interactions of Sld3 with Mcm2, Mcm4 and Mcm6, which are likely substrates for DDK phosphorylation, might play important roles in origin loading of Sld3 (S Azuma and H Masukata, unpublished results).
In contrast to Sld3 recruitment, loading of GINS, Cut5 and Cdc45 requires both DDK and CDK (Figure 6). At present, Sld2 is the only known substrate whose phosphorylation by CDK is required for initiation of replication. Interactions between Dpb11 and Sld2, and their origin loading require phosphorylation of Sld2 (Masumoto et al, 2002; Tak et al, 2006). The essential role of phosphorylation of Drc1/Sld2 appears to be conserved in fission yeast (Noguchi et al, 2002). GINS binds to replication origins in a mutually dependent manner with Cut5. Thus, dependence of GINS loading on CDK could be due to its dependence on Cut5 (Figure 5).
In budding yeast, activation of early or late origins is regulated by S‐CDK, Dbf4/Cdc7 (DDK) and Rad53, an effector kinase in checkpoint regulation (Tanaka and Nasmyth, 1998). Chromatin association of Dbf4/Cdc7 during S phase and its kinase activity are downregulated by Rad53 and Cdc45 loading is implicated to be a target step of this regulation (Weinreich and Stillman, 1999). In fission yeast, because DDK function is required for Sld3 loading (Figure 6), possible reduction of DDK activity under checkpoint activation may prevent Sld3 loading. Therefore, Sld3 loading seems to be a key step for regulation of replication in fission yeast. Although a homolog of Sld3 is yet to be discovered in higher eukaryotes, a certain factor should play a role in recruitment of GINS, Cut5 and Cdc45 in initiation complex formation. Amino‐acid sequences of Sld3 in higher eukaryotes may be diverse from those in budding and fission yeasts, which have only 24% similarity (Nakajima and Masukata, 2002).
Roles of GINS in DNA replication and genome stability
The results of co‐precipitation of GINS components revealed that the psf3‐1 mutation impaired GINS complex formation (Figure 1B), possibly owing to reduction of the interaction between Psf3 and the other GINS components. This possibility is consistent with the finding that temperature sensitivity of the psf3‐1 was suppressed by copy number increase of sld5+, psf1+ and psf2+ (Figure 3A). The mutations in psf3‐1 alter two amino‐acid residues (V56D and I159L; Supplementary Figure S1). As the I159L is a change into a similar amino acid, the V56D might be responsible for the defect. Interestingly, other psf3 mutants with similar phenotypes with the psf3‐1 carry amino‐acid changes in the second exon near the Val56 (Supplementary Figure S1). Thus, the region around the second exon appears to be involved in interactions with other GINS components.
A complex containing Psf1, Psf2 and Sld5 but lacking Psf3 appears to be formed in the psf3‐1 mutant (Figure 1B). Similar complexes lacking Psf3 were detected in budding yeast and Xenopus egg extracts, although such partial complexes are not functional in DNA replication (Kanemaki et al, 2003; Kubota et al, 2003; Takayama et al, 2003). Here, we showed that Psf3 is required for recruitment of GINS to replication origins (Figure 2). A possible interpretation of this requirement is that the interaction of Psf3 with pre‐RC components or Sld3 is required for GINS loading. We have detected the interactions of Psf3 with MCM subunits in yeast two‐hybrid assay (S Azuma and H Masukata, unpublished results) and such interactions may facilitate association of GINS with replication origins.
Loading of GINS depends on Sld3 and Cut5 (Figure 5), suggesting tight relation between these factors. This is consistent with genetic interactions between psf3+ with sld3+ and cut5+ (Figure 3). GINS might interact directly or indirectly with Sld3 and Cut5. We have detected the interactions of GINS components with Sld3 and Cut5 in yeast two‐hybrid assays (S Azuma and H Masukata, unpublished results). However, co‐precipitation of GINS components with Sld3 or Cut5 in synchronized cells with or without crosslinking has not been detected under our experimental conditions (data not shown). Possible interactions of GINS with Sld3 or Cut5 may be transient or fragile during the process of initiation complex formation. In budding yeast, GINS components interact with Sld3 and Dpb11/Cut5 in two‐hybrid assays (Takayama et al, 2003). However, Sld3 or Dpb11 is not found in the complex containing GINS, MCM and Cdc45 during S phase (Gambus et al, 2006; Kanemaki and Labib, 2006). These results suggest that Sld3 and Cut5/Dpb11 may act specifically for initiation step, probably for recruitment of GINS and Cdc45 to replication origins. In contrast, GINS, which interacts with DNA polymerase ε (Takayama et al, 2003; Seki et al, 2006), may act as a mediator between MCM helicase and DNA polymerases. Alternatively, GINS may function as accessory components for MCM helicase, as GINS is required for maintaining interaction between MCM and Cdc45 after initiation of replication (Gambus et al, 2006).
Some fraction of psf3‐1 cells enter mitosis without completion of DNA replication, as shown by the appearance of <1C DNA cells after accumulation of 1C DNA cells in the present study (Figure 1). A similar phenotype has been observed for sld3‐10, cut5‐T401 and hsk1‐89 (Saka et al, 1994; Takeda et al, 2001; Nakajima and Masukata, 2002). On the other hand, neither cdc45 allele causes abnormal mitosis when arrested at a restrictive temperature (Miyake and Yamashita, 1998; Uchiyama et al, 2001). These lines of evidence suggest that assembly of Sld3, GINS and Cut5 onto origins in Step 2 (Figure 7A) may play a role in preventing unscheduled cell division. Loading of Cut5 to replication origins may be important for its role in DNA damage checkpoint or for regulation of cytokinesis (Norden et al, 2006). Alternatively, GINS may be involved in mitosis or cell division, because psf2+ on a multicopy plasmid suppresses the defect of the bir1/cut17 mutation, which causes defects in chromosome segregation and spindle attachment (Huang et al, 2005).
Assembly of initiation machinery and its regulation are complex and elaborate mechanisms in eukaryotes. As components involved in various chromosome functions such as chromosome cohesion, segregation and cell division are likely to be recruited to appropriate locations on chromosomes before or during S phase, the mechanisms of initiation of replication may have been established in conjunction with these other functions. Significance of stepwise assembly of the initiation complex might be establishment of molecular links between replication machinery and other chromosome functions before actual DNA synthesis. Further investigation of molecular interactions between replication factors and those involved in other chromosome functions may be required for better understanding of genome integrity.
Materials and methods
Strains and media
Yeast strains used in this study are listed in Table I. Fission yeast strains were cultured in complete YE medium (0.5% yeast extract and 3% glucose) and minimal EMM minimum (Moreno et al, 1991), all solid media containing 2% agar.
Construction of plasmids
A 1.0 kb genomic fragment containing the psf3+ gene was PCR‐amplified from Schizosaccharomyces pombe HM17 (h−, wild type) genome DNA, using primers 5′‐AAAGGATCCCCTGTTGATGAAACTTGGCG‐3′ and 5′‐AAAGGATCCGCGTCTGATTATAGACAACC‐3′ and the BamHI fragment was cloned into pBluescript KS+ to obtain pBS‐Psf3. A multicopy plasmid pXB940‐Psf3 carrying psf3+ was made by insertion of the BamHI fragment into pXB940, an ARS plasmid containing the fission yeast origin ars2004. pXB940‐Sld5, pXB940‐Psf1 and pXB940‐Psf2 were constructed by a similar procedure as described for psf3+, using the following primers: 5′‐AAAGGATCCGCTTAAACGACGCCAGTCTC‐3′ and 5′‐AAAGGATCCGTCTTGACAGAAGTCCC‐3′ for pXB940‐Sld5; 5′‐AAAGGATCCCCCTGTCCGATGAAGCAAGC‐3′ and 5′‐AAAGGATCCGGATGCCGAATAAACAGTCC‐3′ for pXB940‐Psf1; and 5′‐AAAGGATCCCGATATTGGCTGAGCGAAGG‐3′ and 5′‐AAAGGATCCCTGCCACACCGTGGAGAACG‐3′ for pXB940‐Psf2. For all constructions, the nucleotide sequences were confirmed after PCR cloning.
Epitope tagging of Psf2
The psf2+ gene was tagged with DNA sequences encoding six histidine residues and 3xFLAG epitopes using PCR amplification, essentially as described previously (De Antoni and Gallwitz, 2000). The integration cassette was amplified from pU6H3VSV with primers containing 5′‐TGATGGACAGAATGCGCAAAATTGTTCAAGTTTCCCAAGAAGAACACCACCATCATCATCACGG‐3′, and 5′‐AGGATTCATTATCATTATTTTTAAAGTACATCATCCACATGATAACTATAGGGAGACCGGCAGAT‐3′, subsequently transformed to HM17 (h−, wild type). Transformants grown on YE‐geneticin (100 μg/ml) plates were analyzed by PCR and Western blotting to confirm psf2‐3flag. The resultant strain, HFY108, grew as wild type, suggesting the FLAG‐tagged Psf2 to be functional.
Isolation of psf3 mutants
For construction of the mutating plasmid, the BamHI fragment of psf3+ was cloned into pUCTL1 (Obuse et al, 1996), and the XbaI fragment of 1.7 kb ura4+ was inserted at the XbaI site downstream of psf3‐ORF, resulting in pTL‐ura4‐Psf3. Mutating PCR was carried out with ampli‐Taq Gold (Perkin‐Elmer) in a 100 μl reaction mixture containing 1.5 mM MgCl2 (Perkin‐Elmer) and 0.5 mM MnCl2 using 5′‐ATGCGTCTGATTATAGACAACC‐3′ and 5′‐AACCTGTTGATGAAACTTGGCG‐3′ as primers. PCR products purified by phenol/chloroform extraction followed by ethanol precipitation were used for transformation of HM 83 (h+ura4‐D18). Among ura+ transformants grown on MM plates at 25°C, temperature‐sensitive mutants at 36°C were collected, and integration of the psf3‐ura4+ fragment at the psf3+ locus was confirmed by PCR.
Cell cycle synchronization
To synchronize the cell cycle, derivatives carrying nda3‐KM311 were incubated at 20°C for 4 h to arrest at M phase and then released at 36°C, the restrictive temperature for the psf3‐1 and other mutants, except for cdc45/sna41‐928 (37.5°C).
Polyclonal antibodies against Psf1, Psf3, Sld5 and Rpa2
To express His6‐tagged Psf3 peptides in Escherichia coli, the open reading frame of psf3+ lacking introns was PCR‐amplified from a cDNA library (donated by H Nojima) using 5′‐AAAGGATCCCATATGCATCATCATCATCATCATGATTACTATGATATTGAT‐3′ and 5′‐AAAGGATCCCGTCTCGCAGCATGCAATC‐3′ as primers. The PCR products digested with BamHI were cloned into pBluescript KS+ and the sequence was confirmed. The psf3+ (without introns) fragment was cloned into pET21a, to obtain pET21a‐His6‐Psf3. pET21a‐His6‐Psf1 and pET21a‐His6‐Sld5 were constructed similarly except that the psf1+ and sld5+ genes were amplified from the genomic DNA using respective pairs of primers: 5′‐AAAGGATCCTCATAACTTAGCCAAAAAACCTTGGG‐3′ and 5′‐AAAGGATCCCATATGGAGAATGGTTTGGGGAATCG‐3′; and 5′‐AAAGGATCCATATGGAGTGGGATGCTGAT‐3′ and 5′‐AAACGATCCTATATTAGTGAGACAACACC‐3′. pET21a‐His6‐Rpa2, containing the ssb2+, which encodes the second largest subunit of fission yeast RPA, was constructed from pET19b‐rSpSSB (Ishiai et al, 1996). The His6‐tagged N‐terminus of ssb2+ was amplified by using 5′‐AAAGGATCCCATATGCATCATCATCATCATCATGCTTATGATGCTTTTGGC‐3′ and 5′‐AAAGGATCCTTCAACGCCATCTATCTT‐3′ as primers, and the NdeI–EcoRI fragment together with EcoRI–HindIII fragment from pET19b‐rSpSSB was inserted into pET21a.
His6‐tagged Psf1, Psf3 or Sld5 polypeptides were expressed in E. coli BL21 (DE3), purified as recommended by the manufacturer (Qiagen) and used to immunize rabbits (Hokudo Inc.). Anti‐Psf3 polyclonal antibodies were purified as recommended by the manufacturer using a Hi‐Trap NHS‐activated column (Pharmacia).
Immunoprecipitation and immunostaining
For preparation of yeast cell extracts, a method of Moreno et al (1991) was used. S. pombe haploid cells (0.5–1 × 109 cells) were suspended in 0.4 ml of HB buffer (25 mM MOPS (pH 7.2), 15 mM MgCl2, 15 mM EGTA, 60 mM β‐glycerophosphate, 15 mM p‐nitrophenylphosphate, 0.1 mM Na3VO4, 1 mM dithiothreitol, 1 mM phenylmethylsulfonyl fluoride, 20 μg of leupeptin per ml, 20 μg of aprotinin per ml) and disrupted with glass beads using a bead beater (BioSpec Products). The supernatant (0.4 ml) after centrifugation of the extracts at 12 000 g for 15 min was incubated with 25 μl of magnetic beads (Dynal) associated with mouse anti‐FLAG antibody at 4°C for 2 h, and the beads were washed with HB buffer four times and then proteins were eluted with 50 μl of SDS buffer (120 mM Tris–HCl (pH 6.8), 4% SDS, 20% glycerol, 10% β‐mercaptoethanol) at 95°C for 5 min. Immunoblotting and immunostaining were carried out as described (Takahashi and Masukata, 2001). Antibodies were used at 1/3000 dilution for anti‐FLAG (mouse IgG1) and 1/200–1/500 dilution for anti‐Psf1, Psf3 and Sld5 antibodies (rabbit IgG).
Chromatin immunoprecipitation assay
ChIP assays were performed as described previously (Yamada et al, 2004). PCR amplification with ampli‐Taq Gold (Perkin‐Elmer) was performed in 20 μl of PCR buffer containing 1.5 mM MgCl2 (Perkin‐Elmer) and 0.5–1 μl of immunoprecipitated DNA or total DNA as the template with two or three sets of primers: 5′‐CTTTTGGGTAGTTTTCGGATCC‐3′ and 5′‐ATGAGTACTTGTCACGAATTC‐3′ for ars2004; and 5′‐TCGAAGATCCTACCGCTTTC‐3′ and 5′‐CTTGCGCTGAAGCTTTAGTAAAAG‐3′ for the non‐origin region, as described previously (Yamada et al, 2004). PCR products were separated in 2.5% agarose gels and visualized with 0.5 μg/ml ethidium bromide.
Supplementary data are available at The EMBO Journal Online (http://www.embojournal.org).
Supplementary Figure S1 [emboj7601347-sup-0001.pdf]
Supplementary Figure S2 [emboj7601347-sup-0002.pdf]
Supplementary Figure S3 [emboj7601347-sup-0003.pdf]
Supplementary Figure S4 [emboj7601347-sup-0004.pdf]
Supplementary Figure S5 [emboj7601347-sup-0005.pdf]
We thank Drs Toshiki Tsurimoto, Hiroyuki Araki and Tatsuro Takahashi for critical reading of the manuscript and Mitsuhiro Yanagida, Teresa S Wang, Jerard Hurwitz, Hisao Masai, Katsunori Tanaka, Hiroshi Nojima and Gennaro D'Urso for plasmids and strains. This study was supported in part by a Grant‐in‐aid from the Ministry of Education, Science, Technology, Sports, and Culture, Japan, to HM.
- Copyright © 2006 European Molecular Biology Organization