Mcm2 and Mcm3 Each Contain a Sequence Required for Nuclear Localization of Their Respective Proteins
As a first step to understanding the CDK regulation of Mcm2-7 nucleocytoplasmic localization we first sought to identify the NLSs and NESs responsible for nucleocytoplasmic transport of the Mcm proteins. Our previous work, suggesting that Mcm proteins colocalize as a complex (Nguyen et al., 2000
), raised the possibility that NLSs or NESs on one or more of the Mcm proteins could be responsible for nuclear or cytoplasmic localization of the entire complex. We scanned the amino acid sequences of the six Mcm proteins (Mcm2-Mcm7) for the two NLS sequence motifs that are recognized and bound by the import receptor adapter, importin alpha (reviewed in Jans et al., 2000
). One of these motifs, represented by the SV40 NLS (PKKKRKV), contains a single cluster of highly basic residues, and the other motif, represented by the nucleoplasmin NLS (KRPAATKKAGQAKKKKL) contains two smaller clusters of basic residues separated by 7-22 amino acids. Four good matches to these motifs were identified on Mcm2 (residues 5-9, RRRRR; residues 150-155, RRRRRR), Mcm3 (residues 766-772, PKKRQRV), and Mcm7 (residues 199-219, RR-13aa-RRYRKK).
To determine whether any of these sequences were required for nuclear transport of the Mcm proteins, we initially examined whether these sequences were essential for cell viability. We reasoned that a sequence required for nuclear localization of any of the Mcm proteins would also be essential for viability, because the Mcm proteins must perform their essential replication function in the nucleus. We generated mutant mcm genes with alanines substituting for multiple basic residues in the identified sequences and attempted to replace the endogenous genes with these mutant genes by two-step gene replacement. Haploid strains expressing mutant Mcm proteins with alanine substitutions on Mcm2 (at amino acid residues 150-155) or on Mcm7 (at residues R214, R216, and K217) were viable and exhibited growth rates indistinguishable from wild-type strains. Thus, these sequences are not required for nuclear localization of Mcm proteins. In contrast, we could not isolate haploid strains expressing mutant Mcm proteins containing alanine substitutions in Mcm2 (at residues 5-9) or Mcm3 (at residues 766-772) (; mcm2-nls and mcm3-nls; Supplementary Figure 1). Fusing sequences encoding two tandem copies of the SV40 NLS onto the mutant mcm2-nls or mcm3-nls genes did allow isolation of gene replacement strains expressing these NLS-tagged mutant genes. Tetrad analysis confirmed that mutations in residues 5-9 of Mcm2 or 766-772 of Mcm3 resulted in inviability and that fusion of the SV40 NLS to these mutated Mcm proteins restored viability. Together these results demonstrate that Mcm2 residues 5-9 and Mcm3 residues 766-772 sequences are essential for viability and that their essential role may be to direct the nuclear localization of their respective proteins.
Figure 1. Wild-type and mutant Mcm2 and Mcm3 nucleocytoplasmic transport signals. (A) Key amino acid sequences of wild-type and mutant Mcm2 NLS and Mcm3 NLS-NES transport module. Putative NLSs and leucines in leucine-rich motif are in bold. Consensus CDK phosphorylation (more ...)
To directly examine the role of residues 5-9 of Mcm2 and 766-772 of Mcm3 in the localization of their respective proteins, we used fluorescence microscopy to examine the subcellular distribution of mutant Mcm2 and Mcm3 fused to GFP during the G1 phase of the cell cycle (). We previously reported that wild-type Mcm2, 3, 4, 6, and 7 tagged with GFP are fully functional when expressed as the sole copy of their respective Mcm proteins, indicating that these fusion proteins can complex with other Mcms (Nguyen et al., 2000
). In , the fusion proteins were expressed in addition to the endogenous wild-type MCM
genes. The latter supported the viability of these cells but did not interfere with the fluorescence analysis of the GFP fusion proteins. As expected, GFP fusions to wild-type Mcm2 or Mcm3 accumulated in the nucleus during G1 phase. In contrast, Mcm2-nls-GFP and GFP-Mcm3-nls were distributed throughout the cytoplasm. This mislocalization could be rescued by fusing two tandem copies of the SV40 NLS to the mutant fusion proteins; both Mcm2-nls-GFP-SVNLS2
-GFP-Mcm3-nls were constitutively nuclear. These results directly demonstrate that residues 5-9 of Mcm2 and residues 766-772 of Mcm3 are required for the nuclear localization of their respective proteins. Hence, we refer to these residues as the Mcm2 NLS and Mcm3 NLS, respectively. Our results corroborate previously published results in S. cerevisiae
, implicating the Mcm3 residues in nuclear localization of Mcm3 (Young et al., 1997
), and in S. pombe
, implicating N-terminal Mcm2 residues in nuclear localization of Mcm2 (Pasion and Forsburg, 1999
The Mcm2 and Mcm3 NLSs Are Each Required for Localization of the Mcm2-7 Complex
We have previously shown that fusion of two tandem SV40 NLSs to any Mcm subunit promotes the constitutive nuclear localization of both that subunit and each of the other Mcm subunits (Nguyen et al., 2000
). Hence, we suspected the lethality arising from mutation of the Mcm2 or Mcm3 NLSs could be rescued by fusing the SV40 NLS to other Mcm subunits. To examine this possibility, we reattempted two-step gene replacement of MCM2
, respectively, in haploid strains containing two tandem copies of the SV40 NLS fused to other Mcm proteins (Supplementary Figure 1). We could successfully replace MCM2
when the SV40 NLSs were fused to Mcm3, Mcm4, Mcm5, or Mcm6 and could successfully replace MCM3
when the SV40 NLSs were fused to Mcm2, Mcm4, Mcm5, or Mcm6. Furthermore, the mutant Mcm2-GFP or GFP-Mcm3 in these strains displayed constitutive nuclear localization (unpublished data). These results indicate that mislocalization of the mutant Mcm2-GFP or GFP-Mcm3 can be rescued by ensuring the nuclear localization of one other Mcm subunit. These results further suggest that mutating the Mcm2 or Mcm3 NLS disrupts the nuclear localization of all other Mcm subunits; otherwise an Mcm subunit that could retain its nuclear localization would have rescued the NLS mutations without requiring fusion to the SV40 NLS.
To directly examine whether the Mcm2 or Mcm3 NLS is required for the nuclear localization of other Mcm subunits, we performed a set of experiments exemplified by the one shown in . In this experiment, we examined the effect of mutating the NLS in Mcm2 on the nuclear import of Mcm7-GFP during the transition from G2/M to G1 phase. Because the NLS mutation is lethal, we complemented the mutant mcm2-nls
gene with an MCM2
gene encoding a conditionally degraded version of the Mcm2 protein (mcm2-td
). Mcm2-td is targeted for ubiquitin-mediated degradation by both raising the temperature to 37°C and shifting cells into galactose containing media to induce the E3 ubiquitin ligase Ubr1; under these restrictive conditions there is no detectable Mcm2-td after 60 min (Labib et al., 2000
). In this conditionally complemented strain we could examine how the NLS-defective Mcm2 protein directs the localization of Mcm7-GFP after Mcm2-td is degraded. In parallel, we generated control strains where either the wild-type MCM2
gene or the suppressed mutant gene mcm2-nls-GFP-SVNLS2
was introduced into the mcm2-td
Experimental cells expressing Mcm7-GFP, Mcm2-nls, and Mcm2-td under permissive conditions for Mcm2-td (rich medium containing raffinose at 25°C) were arrested in metaphase with nocodazole. Once arrested, we induced degradation of Mcm2-td by shifting the cells to restrictive conditions (adding galactose at 37°C). After 30 min, the cells were released from the nocodazole arrest into an α-factor G1 arrest, still under restrictive conditions. Mcm proteins normally enter the nucleus during this G2/M to G1 phase transition, but Mcm7-GFP failed to accumulate in the nucleus of these cells (). In contrast, Mcm7-GFP strongly accumulated in the nucleus of the control strains expressing either Mcm2 or Mcm2-nls-SVNLS2 (). Moreover, when cells were maintained at permissive conditions for the Mcm2-td proteins, Mcm7-GFP accumulated in the nucleus in all three strains (unpublished data). These results indicate that the Mcm2 NLS is required for the nuclear localization of Mcm7 in G1 phase. Similar experiments with Mcm3 () demonstrate that the Mcm3 NLS is also required for nuclear localization of Mcm7. Finally, by repeating these experiments with GFP fused to other Mcm subunits, we have been able to show that the Mcm2 NLS is required for nuclear localization of Mcm3 and Mcm4, and the Mcm3 NLS is required for nuclear localization of Mcm2 (unpublished data). Taken together, these results suggest that the Mcm2 and Mcm3 NLSs are required, not just for nuclear localization of their respective proteins, but of the entire Mcm2-7 complex.
Together the Mcm2 and Mcm3 NLS Are Sufficient for Strong Nuclear Localization Activity
The classical definition of an NLS is a sequence that is both necessary and sufficient for directing the nuclear localization of proteins. To determine whether the Mcm2 and Mcm3 NLSs are sufficient to direct the nuclear localization of proteins, we fused them to three tandem copies of GFP (GFP3
). With a combined molecular mass of 82 kDa, these tandem GFPs are larger than the 60-kDa upper limit for proteins to diffuse through nuclear pores and therefore must be actively transported to enter and exit the nucleus (reviewed in Weis, 2003
). These fusion proteins were placed under the control of the regulatable GAL10
promoter. To ensure that the observed localization was not inherited from a previous stage of the cell cycle, yeast cells containing these fusion constructs were first arrested in G1 phase or G2/M phase before the fusion proteins were induced by galactose.
When the SV40 NLS was fused to GFP3, strong nuclear localization was observed. In contrast, sequences containing the Mcm2 NLS (amino acids 1-17) or the Mcm3 NLS (amino acids 760-789), and hereafter referred to as NLS2 and NLS3, respectively, showed only a very weak ability to localize GFP3 to the nucleus at either stage of the cell cycle (). Hence, individually, these two sequences are insufficient to confer robust nuclear localization on a heterologous protein. This conclusion is consistent with the observation that neither NLS is sufficient to direct the Mcm2-7 complex into the nucleus in the absence of the other. Together, however, the NLS2 and NLS3 strongly directed GFP3 into the nucleus in both G1 and G2/M phases (). Thus, the weak Mcm2 and Mcm3 NLSs can functionally act together as a single strong NLS. Using a larger segment spanning the Mcm3 NLS (amino acids 746-789), which increases the spacing between the Mcm2 and Mcm3 NLSs (to 30 amino acids), also resulted in strong composite NLS activity.
Our findings differ from the previous study in S. cerevisiae
that identified the Mcm3 NLS (Young et al., 1997
). In that study, the sequence was reported to be not only necessary for nuclear localization of Mcm3, but also sufficient for nuclear localization of a heterologous protein. That conclusion, however, was based on a slightly different segment spanning the Mcm3 NLS (amino acids 755-781 vs. our segment of amino acids 760-789) examined in combination with 50 amino acids of the Leu2 protein. When we fused that Mcm3 segment without the Leu2 segment to our tandem GFP reporter, we still observed poor NLS activity relative to the SV40 NLS or the combined NLS2-NLS3 (unpublished data). Thus, our examination of an isolated Mcm3 NLS indicates that this NLS only has weak activity.
Mcm3 Contains a Crm1-dependent NES, Which Cooperates with the Mcm2 and Mcm3 NLSs to Form a Cell Cycle-regulated Transport Module
We have previously shown that the Mcm2-7 complex undergoes net nuclear export when cells activate Cdc28 kinase activity (Nguyen et al., 2000
). Because each subunit is too large to diffuse through the nuclear pore, we suspected the complex contains nuclear export signals on one or more subunits. The most recognizable NES motif identified to date is a leucine-rich motif that recruits the nucleocytoplasmic export receptor Crm1 (Fornerod et al., 1997
; Fukuda et al., 1997
; Ossareh-Nazari et al., 1997
; Stade et al., 1997
) and is exemplified by the NESs of the HIV REV and PKIα (LQLP-PLERLTL and LALKLAGLDI respectively; Fischer et al., 1995
; Wen, et al., 1995
). One such motif (LQRRLQLGL; aa 834-842) is present in a 67-amino acid segment (aa 790-856) immediately C-terminal of the Mcm3 NLS segment. To test the export activity of this potential NES, we added it to a GFP3
reporter construct containing a composite Mcm2-Mcm3 NLS. In this new construct () the adjacent Mcm3 NLS and NES are derived from one contiguous 111 amino acid segment of Mcm3 (aa 746-856).
When the resulting fusion protein was induced in exponentially growing cells, it displayed cell cycle-regulated localization that was reminiscent of the CDK regulation of Mcm2-7 localization (Labib et al., 1999
; Nguyen et al., 2000
). The protein was distributed throughout the cytoplasm of budded cells, which contain active Cdc28 kinase, and was predominantly nuclear in unbudded G1 phase cells, which contain little or no active Cdc28 kinase. Moreover, the fusion protein was strongly nuclear in G1 cells arrested with α-factor and was distributed throughout the cytoplasm in G2/M cells arrested with nocodazole (, two left panels). The cytoplasmic localization in nocodazole-arrested cells was dependent on the leucine-rich motif (Figures and ).
To demonstrate that this cytoplasmic localization was due to net nuclear export during passage through the cell cycle, we introduced the GFP3
fusion construct into cells containing a leptomycin B-sensitive allele of CRM1, crm1-T539C
(Neville and Rosbash, 1999
). Cells expressing the fusion protein were released from a G1 arrest into a G2/M arrest, either in the presence or absence of 100 ng/ml leptomycin B. By the time both cultures had completed S phase (60 min), the GFP3
fusion protein was distributed throughout the cytoplasm in the absence of leptomycin B, but remained strongly nuclear in its presence (). Similarly, addition of leptomycin B to exponentially growing cells expressing the fusion protein resulted in constitutive nuclear localization of the protein (unpublished data). Thus, the redistribution of this fusion protein from nucleus to cytoplasm is indeed due to nuclear export and is dependent on the Crm1 export receptor as well as the leucine-rich motif. We henceforth refer to the 67-amino acid segment downstream of the Mcm3 NLS as the Mcm3 NES or NES3. Importantly, the Mcm2 NLS and the contiguous Mcm3 NLS and NES behave as a minimal transport module (NLS2-NLS3NES3) that recapitulates the cell cycle-regulated localization of the entire Mcm2-7 complex.
The Mcm3 NES Promotes the Nuclear Export of Mcm Subunits
We next examined whether the Mcm3 NES promotes nuclear export in the context of the full Mcm2-7 complex. To do this, we generated a haploid strain in which the wild-type endogenous MCM3 gene was replaced by a mutant GFP-Mcm3-nes gene, which contains alanine substitutions at the leucine residues of the Mcm3 leucine-rich motif (AQR-RAQAGA). The ability to generate such a strain indicates the Mcm3 leucine-rich motif does not perform an essential function. As a control, we generated a GFP-MCM3 strain, which expresses a wild-type fusion protein. Both experimental and control strains divided at identical rates and contained a similar distribution of cells throughout the cell cycle by both budding indices and flow cytometry (unpublished data). However, the mutant GFP-Mcm3-nes could be detected in the nuclei of virtually all uninucleate budded cells, whereas GFP-Mcm3 could only be detected in the nuclei of 40% of small budded cells and 0% of uninucleate large budded cells (). Similar results were observed if GFP was fused to Mcm7 instead of Mcm3 (), suggesting that the Mcm3 NES also promotes the nuclear export of other Mcm subunits.
To confirm this role for the Mcm3 NES, we compared the distribution of GFP-Mcm3 with mutant GFP-Mcm3-nes protein in cells synchronously released from an α-factor arrest into a nocodazole arrest (). Both flow cytometry and budding indices confirmed that cell cycle progression was not affected by mutation of the Mcm3 NES. As expected, at the beginning of the time course (, 0 min), when all cells were unbudded G1 cells, the GFP fusion proteins of both experimental and control strains were strongly nuclear. However, 60 min after release from α-factor, when most cells were small-budded and in late S phase, there was a dramatic difference. In the GFP-MCM3 strain, few of the small budded cells (5/100 cells) retained detectable nuclear levels of the fusion protein, whereas in the GFP-mcm3-nes strain, residual nuclear accumulation of the fusion protein could be detected in almost all small budded cells (122/123). Similar, but less striking differences were observed at 50 and 70 min after release from α-factor arrest (unpublished data). These data indicate that Mcm3 NES promotes the nuclear export of Mcm3. Eventually, as cells remained at the nocodazole arrest and became large-budded (, 100 min), nuclear accumulation of the GFP fusion proteins gradually became undetectable (0/71 for GFP-Mcm3 and 1/54 for GFP-Mcm3-nes), indicating that the NES is not absolutely required for the net nuclear export of Mcm3. Nonetheless, the NES is essential for the timely export of Mcm3, and without this timely export, Mcm3 is not effectively cleared from the nuclei of cycling cells (). A very similar delay in nuclear export was observed in a GFP-Mcm3 crm1-T539C leptomycin B-sensitive strain, if leptomycin B was added upon release from alpha factor arrest (unpublished data). These results suggest that the Mcm3 NES functions through Crm1 in the full Mcm2-7 complex as it does in the GFP3 fusion protein.
Nuclear export of Mcm7-GFP is also delayed in a mcm3-nes strain relative to an MCM3 strain (). Fifty minutes after release from α-factor arrest, 42% (56/134) of the small budded MCM3 cells retained barely detectable nuclear accumulation of Mcm7-GFP, whereas almost 90% (94/110) of the small budded mcm3-nes cells retained residual nuclear accumulation of Mcm7-GFP. Similar but smaller differences could be seen at 40 and 60 min after release from α-factor (unpublished data). By 70 min, however, most cells were large-budded, and almost no wild-type (0/77) or mutant (3/172) large budded cells exhibited detectable nuclear accumulation. Again, the MCM7-GFP crm1-T539C strain showed a similar delay in Mcm7 nuclear export (unpublished data). Thus, although there appears to be another partially redundant export signal(s) for the Mcm2-7 complex, we conclude that the Mcm3 NES functions as a Crm1-dependent export signal for at least two subunits of the complex. Also, because the export defect in the crm1-T539C mutant pheno-copies the export defect of the mcm3-nes mutant, it appears that the partially redundant export signal(s) may function through a different export receptor besides Crm1.
Mcm3 Is a Substrate of Cdc28 Kinase
We and others have previously shown that Cdc28 kinase activity promotes the net nuclear export of Mcm proteins (Labib et al., 1999
; Nguyen et al., 2000
), raising the possibility that this regulation is through direct phosphorylation of the Mcm2-7 complex. A scan of the amino acid sequences of all six Mcm proteins for the full consensus CDK phosphorylation site (S/T)-P-X-(K/R; Nigg, 1993
) only identified two sites on Mcm4 and five sites on Mcm3. The sites on Mcm3 were of particular interest because they are all located within the Mcm3 portion of the NLS-NES transport module (). Four sites flank the basic region of the Mcm3 NLS, and the fifth site is adjacent to the leucine-rich motif of the Mcm3 NES. Two additional sites that satisfy a more degenerate CDK phosphorylation site consensus ((S/T)-P) are positioned between the basic region and leucine-rich motif. For the experiments discussed below, we generated mutations in these sites that substitute alanine for the phosphoacceptor serine or threonine of these consensus sites.
shows that recombinant Clb2-Cdc28 kinase can phosphorylate purified GST-Mcm3 in vitro, confirming previous reports of Mcm3 phosphorylation by purified Clb2-Cdc28 and Clb5-Cdc28 kinases (Ubersax et al., 2003
; Loog and Morgan, 2005
). Importantly, GST-Mcm3-cdk5A and GST-Mcm3-cdk7A, which contain mutations in the five full CDK consensus sites and all seven potential CDK sites, respectively (), were both poorly phosphorylated. We also examined Mcm3 phosphorylation in vivo by metabolically labeling cells with 32
P-orthophosphate and observed that Mcm3 displayed significantly more phosphorylation than Mcm3-cdk5A and Mcm3-cdk7A (). Together these results suggest that the Mcm3 portion of the NLS-NES transport module is a target of Cdc28 kinase in vitro and in vivo.
Figure 7. In vitro and in vivo phosphorylation of Mcm3 is dependent on the consensus CDK phosphorylation sites in the Mcm3 NLS3NES3 module. (A) GST-Mcm3 is phosphorylated by Cdc28-Clb2 kinase in vitro. In vitro kinase reactions were performed with purified Cdc28-Clb2 (more ...)
The Mcm3 CDK Consensus Phosphorylation Sites Regulate the Transport Activity of the NLS-NES Module
We next asked whether Cdc28 phosphorylation of the Mcm3 NLS3NES3 segment regulates the activity of the NLS-NES transport module. We examined the effect of mutating the Mcm3 consensus CDK phosphorylation sites on the localization of the GFP3 fusion protein containing the transport module. We introduced alanine substitutions in all five full consensus CDK sites or just the four sites flanking the Mcm3 NLS (). Representative cells from exponentially growing cultures expressing the wild-type or mutant fusion proteins are shown in . As described earlier, the wild-type fusion protein displayed a range of subcellular distributions from nuclear to cytoplasmic depending on the cell cycle position of individual cells. In contrast, both mutant proteins were constitutively nuclear. These results show that Cdc28 promotes the net nuclear export of the GFP3 fusion protein through phosphorylation of the NLS-NES transport module. They suggest that phosphorylation acts as a switch that flips the activity of the NLS-NES transport module from directing net nuclear import to directing net nuclear export.
Phosphorylation of the NLS-NES Transport Module Promotes the Net Nuclear Export of the Mcm2-7 Complex
To determine whether Cdc28 regulation of the NLS-NES transport module contributes to the cell cycle-regulated export of the entire Mcm2-7 complex, we investigated the effect of mutating the CDK consensus sites of the transport module in the endogenous MCM3
gene. We first examined three strains where the wild-type MCM3
gene was replaced by GFP-mcm3-cdk4A, GFP-mcm3-cdk5A
, or GFP-mcm3-cdk7A
(). At a metaphase arrest imposed by nocodazole, all three strains displayed partial nuclear retention of their mutant GFP-Mcm3 (), in contrast to wild-type GFP-Mcm3, which showed no such retention (, GFP-Mcm3 and Nguyen et al., 2000
). This inability to fully export the mutant GFP-Mcm3 proteins was also observed in exponentially growing cells. (first two panels and accompanying bar graphs) shows a quantitative analysis of the nuclear localization of GFP-Mcm3-cdk4A and GFP-Mcm3 in unbudded, small budded, and uninucleate large budded cells. GFP-Mcm3-cdk4A persists in the nuclei of small budded and uninucleate large budded cells while GFP-Mcm3 is disappearing. These observations show that the CDK consensus sites in the Mcm3 NLS-NES module are required for the efficient cytoplasmic localization of Mcm3.
To examine the effect of the CDK consensus site mutations on the net nuclear export of other Mcm proteins, MCM3 was replaced by mcm3-cdk5A in MCM2-GFP, MCM4-GFP, and MCM7-GFP strains. At a nocodazole arrest, partial nuclear retention of the GFP fusion protein was observed in all three mcm3-cdk5A strains, in contrast to the full cytoplasmic distribution observed in the congenic MCM3 strains (). Similarly, in exponentially growing cells, the Mcm-GFP fusion proteins in these strains were never fully cleared from the nucleus. Partial nuclear retention of these three Mcm-GFP proteins was also observed in both nocodazole-arrested and exponentially growing cells when mcm3-cdk5A was replaced by mcm3-cdk4A strains ( and unpublished data). These results suggest that the CDK consensus sites in the Mcm3 portion of the NLS-NES transport module are required for efficient nuclear export of each Mcm protein. We conclude that Cdc28 phosphorylation of NLS-NES module promotes the net nuclear export of the Mcm2-7 complex.
Because some nuclear export of Mcm proteins was still observed in mcm3-cdk4A, mcm3-cdk5A, and mcm3-cdk7A strains, it appears that phosphorylation of the Mcm3 NLS-NES module is not the sole mechanism by which Cdc28 promotes the net nuclear export of Mcm2-7. Whatever the additional mechanism, it presumably requires one or more NES(s) in the Mcm2-7 complex. To determine whether the Mcm3 NES contributes to this residual export, we replaced the MCM3 ORF with mcm3-cdk4A-nes in GFP-MCM3 and MCM7-GFP strains. The mutant Mcm3 expressed in these strains contains alanine substitutions in both the four CDK consensus sites flanking the NLS and the leucine-rich repeat of the NES (). The GFP-Mcm fusion proteins in these strains were strongly nuclear throughout the cell cycle, indicating that the combination of CDK consensus site and NES mutations in Mcm3 could completely abrogate the net nuclear export of Mcm3 and Mcm7 (, last panels and graph). These results provide further evidence of the importance of the NLS-NES transport module in the regulation of Mcm protein localization.
Phosphomimic Mutation of Mcm3 Promotes the Net Nuclear Export of Mcm Proteins and Impairs Replication Initiation
We have shown that phosphorylation of the CDK consensus sites in the NLS-NES transport module is necessary for the efficient net nuclear export of the Mcm2-7 complex after G1 phase. To examine whether phosphorylation of these CDK consensus sites is sufficient to promote this export during G1 phase, we mutated the phosphoacceptor residues of all five full consensus sites to aspartic acid or glutamic acid to mimic constitutive phosphorylation of these sites (). This mutant mcm3
, was substituted for the wild-type endogenous MCM3
gene by two-step gene replacement in MCM2-GFP, GFP-MCM3, MCM4-GFP
, or MCM7-GFP
strains. As described earlier (, and Labib et al., 1999
; Nguyen et al., 2000
), these GFP fusion proteins normally concentrate in the nucleus during G1 phase. In the mcm3-cdk5ED
mutant background, in contrast, much of these proteins were redistributed to the cytoplasm at a G1 phase arrest (). Some residual nuclear localization was still observed, although this is not surprising given that 1) phosphomimic mutations only partially resemble phosphorylated residues, 2) CDK phosphorylation of sites beyond the CDK consensus sites on Mcm3 may be necessary for full exclusion of the Mcm2-7 complex, and 3) the loading of Mcm2-7 onto chromatin makes these proteins refractory to cytoplasmic redistribution. Despite these limitations, the phosphomimic mutations on Mcm3 were sufficient to promote significant, albeit incomplete, export of the Mcm2-7 complex in G1 phase.
Such incomplete export may account for the ability to isolate mcm3-cdk5ED
strains, as complete exclusion of Mcm2-7 from the nucleus during G1 phase would presumably be lethal. Nonetheless, the phosphomimic mutations clearly compromised the cell cycles of these cells. Although exponentially growing liquid cultures of wild-type GFP-MCM3
control strains doubled every 90 min, GFP-mcm3-cdk5ED
mutant strains doubled every 140-150 min. Analysis of microcolonies derived from individually plated GFP-mcm3-cdk5ED
cells not only confirmed that a significant percent of these cells divide much slower than wild-type GFP-MCM3
cells, but also established that 10-12% of the mutant cells generated fully arrested microcolonies after <3-4 divisions (Supplementary Figure 2A). Hence, although a mutant cell lineage could be propagated, considerable inviability was experienced every generation. Flow cytometry and budding indices of exponentially growing GFP-mcm3-cdk5ED
cells showed that they spend a greater proportion of their cell cycle in G2/M phase relative to GFP-MCM3
cells (Supplementary Figure 2B). The simplest interpretation of these observations is that poor nuclear accumulation of Mcm2-7 during G1 phase compromises the initiation of DNA replication, making it difficult to complete a full S phase in a timely manner and triggering a checkpoint delay or arrest at G2/M phase. Many known replication initiation mutants, including the originally isolated mcm
mutants, display a similar accumulation of G2/M cells (Gibson et al., 1990
; Hennessy et al., 1991
; Foiani et al., 1994
; Merchant et al., 1997
; Jacobson et al., 2001
To determine whether replication initiation is indeed disrupted in the GFP-mcm3-cdk5ED mutant, we examined the rate of plasmid loss during multiple generations of nonselective growth (). Failure to initiate DNA replication on a plasmid will enhance its intrinsic loss rate. In control MCM3 and GFP-MCM3 strains, the plasmid YCp50 was lost at a rate of 3.0 and 4.4% per generation, respectively. In the GFP-mcm3-cdk5ED strain this rate was increased to 18.5%. Increased plasmid loss rates due to defective initiation can often be suppressed by increasing the number of origins on the plasmid. The plasmid pJW1112, which contains an additional seven tandem copies of the H4ARS origin inserted into Ycp50, specifically reduced the plasmid loss rate in the GFP-mcm3-cdk5ED strain (unpublished data), indicating that the elevated plasmid loss seen in the GFP-mcm3-cdk5ED strain is due to defective replication initiation.
This defect in initiation could be due to mislocalization of the Mcm2-7 complex or to localization-independent effects of these mutations on the initiation function of the complex. To distinguish between these possibilities, we fused the mcm3-cdk5ED to two tandem copies of the SV40 NLS (SVNLS2-mcm3-cdk5ED). This NLS fusion restored strong nuclear localization of Mcm2-GFP, GFP-Mcm3, Mcm4-GFP, and Mcm7-GFP in G1 phase () and restored liquid culture doubling times and microcolony expansion to wild-type levels (unpublished data). In addition, this fusion restored plasmid loss rates in the SVNLS2-GFP-mcm3-cdk5ED strain to wild-type levels (). We conclude that the replication initiation defect observed in the GFP-mcm3-cdk5ED is primarily due to mislocalization of Mcm proteins in G1 phase. Considering the incomplete extent of this mislocalization arising from the phosphomimic mutations, these results suggest that bona fide ectopic phosphorylation of the Mcm3 portion of the NLS-NES module in G1 phase would severely impair replication initiation in S phase and that the normal CDK regulation of Mcm2-7 localization contributes significantly to restraining reinitiation of DNA replication during and after S phase.