Multiple mechanisms regulate subcellular localization of human CDC6.

CDC6 is a protein essential for DNA replication, the expression and abundance of which are cell cycle-regulated in Saccharomyces cerevisiae. We have demonstrated previously that the subcellular localization of the human CDC6 homolog, HsCDC6, is cell cycle-dependent: nuclear during G(1) phase and cytoplasmic during S phase. Here we demonstrate that endogenous HsCDC6 is phosphorylated during the G(1)/S transition. The N-terminal region contains putative cyclin-dependent kinase phosphorylation sites adjoining nuclear localization sequences (NLSs) and a cyclin-docking motif, whereas the C-terminal region contains a nuclear export signal (NES). In addition, we show that the observed regulated subcellular localization depends on phosphorylation status, NLS, and NES. When the four putative substrate sites (serines 45, 54, 74, and 106) for cyclin-dependent kinases are mutated to alanines, the resulting HsCDC6A4 protein is localized predominantly to the nucleus. This localization depends upon two functional NLSs, because expression of HsCDC6 containing mutations in the two putative NLSs results in predominantly cytoplasmic distribution. Furthermore, mutation of the four serines to phosphate-mimicking aspartates (HsCDC6D4) results in strictly cytoplasmic localization. This cytoplasmic localization depends upon the C-terminal NES. Together these results demonstrate that HsCDC6 is phosphorylated at the G(1)/S phase of the cell cycle and that the phosphorylation status determines the subcellular localization.

Little is known regarding the regulation of DNA replication initiation in mammalian cells. In yeast, the origin recognition complex, required for replication initiation, consists of six subunits and is associated with specific DNA sequences (replicators) (1)(2)(3). In addition, other factors are required for DNA replication initiation including CDC6, CDC45, and MCM (mini-chromosome maintenance) family proteins (4). Several lines of evidence suggest that ORC, 1 CDC6, and MCM may function together as the replication initiator complex (reviewed in Refs. 5 and 6). Recently several human proteins have been identified that seem to be structural homologs of proteins known to be directly involved in DNA replication in yeast (7)(8)(9)(10)(11)(12)(13)(14).
CDC6 in Saccharomyces cerevisiae, Cdc18 in Schizosaccharomyces pombe, and XCDC6 in Xenopus are homologs and have been shown to be essential for DNA replication. For example, in S. cerevisiae the assembly of a prereplication initiation complex at origins of replication requires CDC6 (15). In both S. cerevisiae and S. pombe, the CDC6/Cdc18 protein is degraded at the G 1 /S transition after phosphorylation by cdk (16,17). In S. pombe, overexpression of Cdc18 results in re-replication without mitosis (16), whereas specific mutations in S. cerevisiae CDC6 cause over-replication of DNA (18). Together these results demonstrate the importance of CDC6 in DNA replication and suggest that stringent regulation of CDC6 protein levels, such that it is active and/or available in G 1 but destroyed soon after initiation, is crucial to ensure that DNA replication occurs precisely once per cell cycle.
The exact function(s) of CDC6/Cdc18 in mammalian cells is not known, but it is likely that it plays a role in the process of assembly of prereplication complexes and/or origin firing. The human homolog of CDC6/Cdc18, HsCDC6 binds cyclin and ORC1 similar to studies previously done with yeast, implicating HsCDC6 in DNA replication (19). However, in contrast to yeast in which CDC6 protein levels are tightly regulated during the cell cycle, levels of HsCDC6 do not decline at the onset of S phase (19 -21), suggesting an alternative mechanism of regulation of HsCDC6 at the G 1 /S transition in human cells. This idea is supported by studies from our laboratory and others that demonstrate that epitope-tagged HA-HsCDC6 is found to be nuclear in G 1 cells and cytoplasmic in S phase cells (19,21,22). We show that HsCDC6 contains two putative N-terminal NLSs and a leucine-rich C-terminal nuclear export signal (NES), LXXXLXX-LXL (reviewed in Ref. 23). Previous reports of other proteins containing both nuclear import and export motifs describe protein shuttling from one compartment to the other until one of the motifs is masked, whether by phosphorylation, association with other proteins, or other modification (24,25). Our working hypothesis is that HsCDC6 is synthesized and imported into the nucleus in an NLS-dependent manner, whereby it binds ORC and begins assembly of the prereplicative complex. Unbound (excess) HsCDC6 continues to shuttle. Upon phosphorylation of HsCDC6, possibly by cyclin A/cdk, HsCDC6 is released from the prereplicative complex, whereby it becomes available for export from the nucleus via its NES.
Here we show that HsCDC6 is phosphorylated by cyclin/cdks in vitro via association with cyclin/cdks through the Cy motif and becomes dephosphorylated in G 1 and phosphorylated at the G 1 /S transition in vivo. A functional Cy motif and intact phosphorylation sites are required for the cytoplasmic displace-* This work was supported by National Institutes of Health Grant CA60499. The costs of publication of this article were defrayed in part by the payment of page charges. This article must therefore be hereby marked "advertisement" in accordance with 18  ment of HsCDC6 in S phase. Furthermore, conversion of the cyclin/cdk-targeted serines (SPXK) to phosphate-mimicking aspartic acid resulted in nearly exclusive cytoplasmic localization of the protein during G 1 phase when cdk is inactive. Together these results indicate that dephosphorylated HsCDC6 localizes to the nucleus through the action of either of two NLSs, is phosphorylated at the G 1 /S border, whereby HsCDC6 relocates to the cytoplasm, possibly through the unmasking of the NES.

EXPERIMENTAL PROCEDURES
Binding and Competition Studies-The conditions for competition of binding reactions with peptides PS100 and PS101 was described previously (26,27). PS100 (ACRRLFGPVDSE) is derived from the cdk inhibitor p21, whereas PS101 (ACRRLKKPVDSE) has a mutated Cy motif (RRLFG). In vitro transcription and translation of radiolabeled cyclin A was performed using Promega TnT-coupled rabbit reticulocyte lysate system. [ 35 S]cyclin A was incubated with glutathione agarose beads coated in GST or GST-CDC6 in buffer A7.4 containing 100 mM NaCl at 37°C for 10 min in the presence or absence of 75 M PS100 or PS101. Proteins bound to beads were separated on an SDS-PAGE gel and visualized by fluorography.
Cell Culture and Synchronization-HeLa, U2OS, and 293T cells were grown in 10% donor calf serum in Dulbecco's modified Eagle's medium supplemented with penicillin and streptomycin. In cell synchronization studies, HeLa cells were grown to 40 -60% confluence in 100-mm dishes and mimosine (200 M) or hydroxyurea (10 mM) were added for 20 -24 h prior to harvesting. Cells were phosphate-starved in phosphate-free medium containing mimosine or hydroxyurea for 2 h, medium containing [ 32 P]ortho-phosphate (300 Ci) was added for 3 h, and the cells were harvested. Cells were lysed using 100 mM NaCl, 50 mM Tris (pH 8.0), 5 mM EDTA, 0.1% Nonidet P-40, 50 mM NaF, 1 mM Na 3 VO 4 , 0.1 mM phenylmethylsulfonyl fluoride, and protease inhibitors and DNaseI and RNase. Immunoprecipitations were performed with the antibody BWH81, which was generated against full-length HsCDC6. The samples were subjected to SDS-PAGE (10% acrylamide gel) and transferred to nitrocellulose. The nitrocellulose blot was directly exposed to film. The Western blot was later probed with the antibody BWH81 (1:1000 dilution).
Transfections-U2OS cells were grown to 30 -50% confluence on chamberslides and transfected using 0.5 g of plasmid DNA, 1.5 l of LipofectAMINE Plus reagent (Life Technologies, Inc.), and 4.5 l of LipofectAMINE for 6 -12 h. In leptomycin B (LMB) experiments, 200 nM LMB and 10 ng/ml cycloheximide (to block protein synthesis) were added to cells 36 h post-transfection. Cells were fixed in 3% paraformaldehyde/2% sucrose in PBS for 10 min., permeabilized in Triton/bovine serum albumin in PBS for 5 min, and blocked in 5% donor calf serum/ PBS for 30 min. Anti-HA antibody 12CA5 (1:100 in 5% donor calf serum) was used as the primary antibody and goat anti-mouse fluorescein isothiocyanate-conjugated antibody (1:2000) (DAKO) was used as the secondary antibody. Immunofluorescence images were acquired at a magnification of ϫ100.
U2OS cells grown to 30 -50% confluence and transfected in 100-mm dishes were trypsinized, fixed in 70% ethanol, and incubated in PBS containing 50 g/ml propidium iodide, 0.05% Nonidet P-40, and 10 g/ml RNaseA for 1 h. Cells were analyzed using the FACScan system and Modfit software.
293T cells were grown to 40 -60% confluence and transfected using LipofectAMINE and LipofectAMINE Plus reagent. After 48 h proteosome inhibitor MG 1 32 (1 M) (Calbiochem) was added for 1 h, and cells were harvested (described above). The anti-T7 antibody (Novagen) was used for the immunoprecipitation of T7-ORC1 and associated proteins.
Samples were separated by SDS-PAGE and transferred to nitrocellulose. Western blotting was performed using anti-Myc (9E10) and anti-T7 antibodies.
Truncation mutations were generated using restriction enzyme sites within the HsCDC6 nucleotide sequence, blunt-ending with Klenow and religating to the blunt-ended NheI site in the vector sequence as follows: D4 -364T corresponds to the XbaI site, D4-462T corresponds to the BglII site, and D4 -488T corresponds to the BspMI site. The open reading frame of HsCDC6 was cloned into the EBG vector (EBG-CDC6) using the BamHI site.

RESULTS
HsCDC6 Is Phosphorylated at the G 1 -S Transition-Our laboratory and others have published previously that HsCDC6 localizes to the nucleus during the G 1 phase and to the cyto-  (1) preimmune control. B, a Western blot of that described above using an anti-HsCDC6 antibody (BWH81).

FIG. 2. Cy motif on HsCDC6 is necessary for interaction with cyclin A in vitro and in vivo. a, pull-down experiment of [ 35 S]cyclin A (Cyc A) using glutathione agarose beads coated with GST or GST-CDC6. In vitro transcribed and translated [ 35 S]cyclin A is incubated
with beads in the presence or absence of Cy peptide PS100 or Cy mutant peptide PS101. Lane 1 represents 0.1ϫ cyclin A input. b, 293T cells were transfected with EBG-HsCDC6 or EBG-HsCDC6-Cy mutants. The GST-tagged CDC6 and associated proteins were bound to glutathione beads, separated by SDS-PAGE, and immunoblotted for cyclin A, cdk2, and GST (lower three panels). The top two panels show 0.1ϫ of the protein input into the glutathione binding assay. GSH, glutathione.
plasm during S phase of the cell cycle (19,21,22,28). Because the activity of many regulatory proteins is modulated by phosphorylation via specific kinases and phosphatases, we examined the phosphorylation level of endogenous HsCDC6 during G 1 and S phase in HeLa cells, a human cervical carcinoma cell line. Autoradiography revealed that endogenous HsCDC6 is dephosphorylated in G 1 -blocked cells (mimosine) and phosphorylated in S phase (hydroxyurea) and asynchronously growing FIG. 3. Subcellular localization of HsCDC6. A, U2OS cells were transfected with HA-tagged HsCDC6, ⌬Cy, A4, and D4 plasmid constructs. B, indirect immunofluorescence was performed on transfected cells using an anti-HA antibody (12CA5 mouse monoclonal antibody) followed by a fluorescein isothiocyanate-conjugated goat anti-mouse antibody. For each plasmid construct, Ͼ50 transfectants were counted per experiment. All transfections were repeated at least once. 4,6-diamidino-2-phenylindole staining shows all nuclei in the field, whereas anti-HA staining shows HA-CDC6 only in the transfected cells. cells (Fig. 1A). Western analysis of the nitrocellulose blot reveals similar protein levels of HsCDC6 in asynchronous and mimosine-and hydroxyurea-treated cells, indicating that the low level of phosphorylation observed in mimosine-treated cells is not caused by the absence of endogenous protein levels (Fig.  1B). These results demonstrate that HsCDC6 is phosphorylated as cells enter S phase.
HsCDC6 Associates with Cyclin/cdks through the Cy Motif-The N-terminal one third of HsCDC6 contains one Cy motif (an RRLVF cyclin-docking motif). To determine the importance of the Cy motif in the association of HsCDC6 with cyclin/cdks, competition studies were performed using a peptide containing a Cy motif (PS100) and another peptide with mutations in the Cy motif (PS101) as a negative control. GST-CDC6 specifically binds in vitro 35 S-radiolabeled cyclin A produced by in vitro transcription and translation. The Cy peptide competitively inhibited this association of HsCDC6 with cyclin A as compared with the mutant Cy peptide (Fig. 2a). To determine directly whether a functional Cy motif on HsCDC6 was required for its association with cyclin/cdk in vivo, the Cy motif on HsCDC6 was independently mutated by two different mutations, CDC6⌬Cy1 (RRLV Ͼ GGSV) and CDC6⌬Cy1-2 (RRLV Ͼ RRVG), using a polymerase chain reaction-based approach (described under "Experimental Procedures"). Each mutant plasmid was sequenced and subjected to in vitro transcription and translation to eliminate the presence of inadvertent proteintruncating mutations (data not shown). Wild type or CDC6⌬Cy was expressed in mammalian 293T cells fused with a GST epitope, and the proteins were isolated from cell extracts by affinity chromatography with glutathione agarose beads. The wild-type HsCDC6 co-precipitated cellular cyclin A and cdk2, whereas this cyclin/cdk binding was not observed with the CDC6⌬Cy (Fig. 2b). Therefore, mutation of the Cy motif specifically disrupts HsCDC6 association with cyclin/cdks in vivo.
Mutation of the Cy Motif of HsCDC6 Results in Exclusively Nuclear Localization-To determine the effect of the Cy mutation on subcellular localization and cell cycle progression, immunohistochemical studies and FACS analysis were performed. Hemagglutinin-tagged HsCDC6 containing the Cy mutation (HA-CDC6⌬Cy1) was transfected into U2OS cells for 48 h. With wild-type HA-HsCDC6, ϳ60% of the transfected cells contained HA-HsCDC6 exclusively in the nucleus, whereas the other ϳ40% expressed the protein in the cytoplasm. In the case of HA-CDC6⌬Cy, Ͼ95% of the cells viewed had the protein exclusively in the nucleus (Fig. 3). To determine whether this localization is an effect of G 1 arrest, because we demonstrated previously that G 1 cells express HsCDC6 exclusively in the nucleus (19), FACS analysis was performed on U2OS cells co-transfected with HA-CDC6⌬Cy and a marker plasmid that expresses farnesylated green fluorescent protein (pEGFP-F, CLONTECH). Results of this experiment demonstrated that the transfected cells (GFP-positive) proliferate in an asynchronous manner and are not arrested in G 1 (Fig. 4E), whereas control cells transfected with plasmid encoding p21 result in G 1 arrest (Fig. 4F). This indicates that the nuclear localization is the result of the Cy mutation and not indirectly caused by cell cycle arrest in G 1 .
Nonphosphorylatable HsCDC6 Remains in the Nucleus-The results of the Cy mutation experiment support the idea that the dephosphorylated form of HsCDC6 is constrained to the nucleus, whereas the phosphorylation of HsCDC6 by cyclin/cdks may be necessary for release from the nucleus. To test the hypothesis that dephosphorylated HsCDC6 is restricted to nuclear localization, U2OS cells were transfected with a plasmid encoding HA-tagged HsCDC6 with mutations of serines 45, 54, 74, and 106 to alanines (A4), and the subcellular localization was scored. Transfected cells showed Ͼ95% nuclear localization (Fig. 3), demonstrating that nonphosphorylatable HsCDC6 remains in the nucleus. FACS analysis of HsCDC6 A4 transfectants demonstrated no significant changes in cell cycle distribution (Fig. 4C).
There are two possible mechanisms that may account for this result: 1) the conformation of dephosphorylated HsCDC6 allows an NLS to be accessible for nuclear import, or 2) association with another protein(s) transports dephosphorylated HsCDC6 into the nucleus. To distinguish between these possibilities, we examined HsCDC6 amino acid sequence for NLS motifs. We found two candidate NLSs: one bipartite NLS sequence and a second potential NLS, based on sequence similarity to the c-Myc NLS (29). First, we investigated the function of the putative bipartite NLS (Lys-Gln-Gly-Lys 80 -Lys 81 ) by using polymerase chain reaction mutagenesis. We constructed a plasmid that contains the mutation of lysine 80 to alanine and deletion of lysine 81 (A4⌬80 -81) on the constitutively nuclear A4 plasmid. Cells transfected with A4⌬80 -81 were processed for immunofluorescence and scored for cytoplasmic staining. A FIG. 5. CDC6 has two NLSs. U2OS cells were transfected with HA-tagged HsCDC6 and mutant HsCDC6 constructs. After 48 h, the cells were treated with or without leptomycin B, to block nuclear export, and cycloheximide (CHX), to block new protein synthesis in the cytoplasm, for 20 min. Indirect immunofluorescence was performed with an anti-HA antibody, and the cells were scored for nuclear and cytoplasmic localization. WT, wild type. significant number of cells transfected with A4⌬80 -81 showed a redistribution from the nucleus to the cytoplasm (58% nuclear) as compared with the exclusively nuclear staining of A4 (Fig. 5). The second NLS candidate, similar to the putative c-Myc NLS, was mutated (Arg 66 Lys 67 Arg 68 to alanines A4⌬66 -68), and the CDC6 protein was examined for subcellular distribution. This mutation resulted in 61% nuclear location. Mutation of both of these putative NLSs on the A4 CDC6 plasmid (A4⌬66 -68, 80 -81) resulted in a 90% cytoplasmic localization (Fig. 5). Therefore, nuclear localization depends on either of the two NLSs on HsCDC6.
To determine whether there was residual nuclear localization activity, LMB, an inhibitor of nuclear export, was added to transfected cells (Fig. 5). In the transfections with mutations in the individual NLSs, A4⌬66 -68 and A4⌬80 -81, an increased nuclear localization was seen, suggesting residual nuclear localization activity when either NLS was mutated alone. However, in cells transfected with the combined mutation A4⌬66 -68,80 -81, there was less than 25% nuclear localization in response to LMB, suggesting that NLS activity was confined to these two signals.
Cyclin/cdk Phosphorylation of HsCDC6 Is Responsible for Cytoplasmic Localization-The loss of the cyclin binding motif restricts HsCDC6 localization to the nucleus. Because the absence of a Cy motif is predicted to diminish the efficiency with which cyclin/cdks dock on HsCDC6 and phosphorylate target serines, we proposed that the phosphorylation of HsCDC6 is important for cytoplasmic localization. To test this hypothesis, U2OS cells were transfected with a plasmid encoding HAtagged HsCDC6 with mutations of serines 45, 54, 74, and 106 to aspartates (D4) to mimic the negative charge of a phosphate. Immunofluorescence studies of transfected cells demonstrate cytoplasmic staining in Ͼ95% cells (Fig. 3). FACS analysis of cells co-transfected with D4 and a GFP marker plasmid showed no significant change in cell cycle distribution when compared with mock transfected cells (Fig. 4D), indicating that cytoplasmic localization of D4 is not a result of S phase arrest. To precisely define the regulatory serine(s) involved in cytoplasmic localization, we individually mutated serines 54 and 74 to aspartates (Asp 54 and Asp 74 ). To prevent the phosphorylation of other cyclin/cdk targeted serines in HsCDC6, which may contribute to the cytoplasmic displacement of Asp 54 , serines 45, 74, and 106 were converted to alanines (ADAA) (Fig. 6), and similar mutations were added to the mutant Asp 74 (AADA). These plasmids were transfected into U2OS cells and cells were scored for cytoplasmic localization. Mutation of either serine 54 or 74 alone to aspartates resulted in nearly exclusive cytoplasmic localization, thus defining that phosphorylation of either of these serines is sufficient for nuclear exclusion.
Phosphorylation Does Not Inactivate the CDC6 NLS-Because serine 54 and 74 are located near the two NLSs, it is possible that the negative charge acquired by the phosphorylation of these residues (or mutation to aspartates) causes cytoplasmic localization by inactivating the NLS. Addition of LMB to cells transfected with D4 CDC6 (Fig. 5), however, increased nuclear localization to ϳ50% of cells. This nuclear localization suggests that the NLS is still active in the phosphorylated HsCDC6 (or in HsCDC6 with phospho-mimicking mutations).
The Nuclear Export Signal of HsCDC6 Is Located in the C Terminus-Another model to explain the phosphorylation-dependent subcellular distribution of HsCDC6 is that the phosphorylation of key serines results in a conformational change that unmasks a NES. To identify the location of a potential NES(s), truncations of the constitutively cytoplasmic HsCDC6D4 were transfected into U2OS cells and scored for localization. Truncations at amino acids 364 (D4 -364T) and 462 (D4 -462T) resulted in predominantly nuclear localization as compared with full-length D4, which is cytoplasmic (Fig. 7). However, D4 -488T demonstrated cytoplasmic localization in Ͼ90% cells, suggesting the possibility that an NES exists in residues 462-488. Sequence analysis of HsCDC6 predicts a highly hydrophobic and classical NES in this region, ILVCSLMLLIRQLKI.
Although this study clearly demonstrates a role for phospho-

FIG. 7. C-terminal truncations of HsCDC6 indicate that an NES is localized near the C terminus of CDC6.
U2OS cells were transfected with HsCDC6, D4, and D4-truncated plasmid constructs. After 48 h, the cells were fixed and permeabilized, and indirect immunofluorescence was performed as described in the Fig. 6 legend. Transfected cells were scored for nuclear (% NUC) and cytoplasmic (% CYTO) localization. WT, wild type. rylation in cytoplasmic localization, the exact mechanism of nuclear export could require either 1) phosphorylation alone of HsCDC6 or 2) phosphorylation and continued association between cyclin/cdk. To distinguish between these two possibilities, U2OS cells were transfected with HA-tagged HsCDC6D4 also containing a Cy1-2 mutation that does not disrupt the NLS (Fig. 6). Transfected cells show predominantly cytoplasmic staining similar to HsCDC6D4, indicating that phosphorylation alone without continued association with cyclin/cdk is sufficient for cytoplasmic localization.
HsCDC6-ORC1 Interaction-We have previously described HsCDC6-ORC1 interactions (19). To determine the role of ORC1 binding to nonphosphorylated and phosphorylationmimicking HsCDC6 mutants, A4 and D4, respectively, T7tagged ORC1 was co-transfected with Myc-tagged HsCDC6-A4 or -D4 into 293T cells. Cell lysates were immunoprecipitated using an anti-T7 antibody and analyzed for co-immunoprecipitation of Myc-A4 and Myc-D4 by Western blot (Fig. 8) using an anti-Myc antibody. ORC1 binds similar levels of A4 and D4, suggesting that the binding of ORC1 to HsCDC6 is not regulated by the phosphorylation status of HsCDC6.

DISCUSSION
Our laboratory and others (19,21,22) demonstrated previously that subcellular localization of HsCDC6 is cell cycle-dependent, whereby HsCDC6 is nuclear in G 1 phase and cytoplasmic in S phase. We also showed cell cycle-specific association of cyclin A/cdk2 with HsCDC6 (19). We show here that endogenous HsCDC6 is phosphorylated at the G 1 /S border, a finding temporally consistent with cyclin A expression, origin firing, and the redistribution of HsCDC6 to the cytoplasm. In addition, we investigate the role of this phosphorylation in HsCDC6 localization through mutational analysis of 1) the four putative cyclin/cdk phosphorylation sites, 2) the cyclindocking motif, Cy, 3) N-terminal nuclear localization signals, and 4) a C-terminal nuclear export signal. It was reported previously that HsCDC6 is phosphorylated on serines 54, 74, and 106 by cyclin/cdks (21,22). We have similar results 2 and extended the analysis to show that the Cy motif of HsCDC6 improves the K m of a CDC6-based peptide substrate of cdks by 100-fold (30).
The results described here indicate a role for phosphorylation and NLS in the subcellular localization of HsCDC6. Both A4 and ⌬Cy mutants are limited to the nuclei only when the NLS are intact. However, disruption of each individual putative NLS results in a significant shift toward cytoplasm, 42 and 39% as compared with Ͻ10%. Mutation of both sites results in nearly complete inhibition of nuclear import. The addition of LMB, a CrmI nuclear export inhibitor, further demonstrates an incomplete block to nuclear import in individual NLS mutations. Because LMB did not significantly increase nuclear localization of A4⌬66 -68, 80 -81, we can eliminate the possibility that these mutations merely enhance nuclear export and thereby increase cytoplasmic localization. The results presented here support the idea that HsCDC6 utilizes either of two NLSs to gain access to the nucleus.
We show here that mutation of either serine 54 or 74 to phosphate-mimicking aspartates results in cytoplasmic localization of HsCDC6. This is not an effect of blocking nuclear import, because we see nuclear localization of the D4 mutant when cells are treated with LMB, suggesting D4 cycles in and out of the nucleus. In addition, we see nuclear staining in cells transfected with truncated D4 mutants (D-364T and D-462T), suggesting that the NLS sequences are functional despite the localized increase in negative charges mimicking phosphorylation. Also, we show that phosphorylation and export of HsCDC6 does not depend upon a continued association with cyclin A/cdk, because the elimination of the cyclin-interacting Cy motif on the ⌬D4 mutant does not abolish cytoplasmic localization.
In addition, we identify a leucine-rich C-terminal NES. On the basis of reports of other proteins containing both NLS and NES motifs, shuttling from one compartment to the other continues to occur until one of the motifs is masked whether by phosphorylation, association with other proteins, or other modification. Our working hypothesis is that HsCDC6 is synthesized and imported into the nucleus in an NLS-dependent manner, whereby it binds ORC and begins assembly of the prereplicative complex. Upon phosphorylation, possibly by cyclin A/cdk, HsCDC6 is released from the prereplicative complex and is exported from the nucleus via its NES. It is possible that conformational changes induced by phosphorylation of HsCDC6 allow the unmasking of the C-terminal NES required for nuclear export.
Phosphorylation of HsCDC6 was expected to be linked with origin firing. Therefore, the A4 mutant might have acted as a dominant negative protein, which causes a block to origin firing. It is surprising that A4 does not cause a block in G 1 /S. Although this result is consistent with the results of Petersen, et al. (21), it is possible that levels of exogenously expressed A4 is lower than endogenous HsCDC6, thus allowing origin firing by endogenous HsCDC6. Further, as in yeast, very few origins may need to fire per chromosome for complete and rapid replication of the genome (31). Therefore, we cannot yet state that the failure to phosphorylate HsCDC6 is without any effect on the replication machinery, particularly in view of the effects on replication seen in other experimental systems with similar mutants (22,32).