Protein Phosphatase 2A Is Targeted to Cell Division Control Protein 6 by a Calcium-binding Regulatory Subunit*

The cell division control protein 6 (Cdc6) is essential for formation of pre-replication complexes at origins of DNA replication. Phosphorylation of Cdc6 by cyclin-dependent kinases inhibits ubiquitination of Cdc6 by APC/Ccdh1 and degradation by the proteasome. Experiments described here show that the PR70 member of the PPP2R3 family of regulatory subunits targets protein phosphatase 2A (PP2A) to Cdc6. Interaction with Cdc6 is mediated by residues within the C terminus of PR70, whereas interaction with PP2A requires N-terminal sequences conserved within the PPP2R3 family. Two functional EF-hand calcium-binding motifs mediate a calcium-enhanced interaction of PR70 with PP2A. Calcium has no effect on the interaction of PR70 with Cdc6 but enhances the association of PP2A with Cdc6 through its effects on PR70. Knockdown of PR70 by RNA interference results in an accumulation of endogenous and expressed Cdc6 protein that is dependent on the cyclin-dependent protein kinase phosphorylation sites on Cdc6. Knockdown of PR70 also causes G1 arrest, suggesting that PR70 function is critical for progression into S phase. These observations indicate that PP2A can be targeted in a calcium-regulated manner to Cdc6 via the PR70 subunit, where it plays a role in regulating protein phosphorylation and stability.

Precise regulation of DNA replication is necessary to ensure that daughter cells receive a complete and intact genome during mitosis. A crucial step in regulating DNA replication is the assembly of pre-replicative complexes at origins of replication (1). Coordination of DNA replication with the cell cycle is achieved through a periodic accumulation and destruction of proteins involved in formation of pre-RCs 3 is mediated by cyclin-dependent kinases (CDKs) and the E3 ubiquitin ligase, anaphase promoting complex/cyclosome (2). The mammalian Cdc6 protein is required for DNA replication and acts in conjunction with the Cdt1 protein to recruit the mini-chromosome maintenance complex into pre-RCs (1,3,4). Mammalian cells have multiple mechanisms to ensure that pre-RCs only assemble during late M and G 1 , including regulation of the levels and function of Cdc6 (2). Mammalian Cdc6 is regulated by phosphorylation of multiple sites within its N-terminal domain by cyclin-dependent protein kinases. Cdc6 is phosphorylated at canonical CDK sites, including serines 54, 74, and 106 of human Cdc6 (5,6). Experiments with exogenously expressed protein have shown that phosphorylation can regulate the nuclear localization of Cdc6 (5,(7)(8)(9). However, other studies have shown that a subpopulation of endogenous Cdc6 remains in the nucleus, bound to chromatin, throughout the cell cycle (10 -12). Phosphorylation of Cdc6 also plays an important role in regulating the stability of Cdc6. The N-terminal domain of Cdc6 contains RXXL (D box) and KEN (KEN box) destruction motifs, which are binding sites for the form of the APC/C containing the cdh1-targeting subunit (13). Cdc6 is polyubiquitinated and targeted for degradation by APC/C cdh1 , which prevents formation of pre-RCs in quiescent cells and during early G 1 by maintaining low levels of Cdc6 (14). Phosphorylation of Cdc6 by CDKs protects the protein from degradation by blocking recognition by cdh1 resulting in stabilization of Cdc6 during a window of time that allows formation of pre-RCs during G 1 (15). The importance of CDKmediated stabilization of Cdc6 is also supported by evidence showing that the cell cycle arrest caused by DNA damage is due to dephosphorylation and degradation of Cdc6 (16).
Because the extent of Cdc6 phosphorylation is controlled by the opposing actions of cyclin-dependent kinases and protein phosphatases, dephosphorylation of Cdc6 can also control formation of pre-RCs. Much less is known about mechanisms that regulate Cdc6 dephosphorylation. A previous study identified a fragment of PR70 as a member of the PPP2R3 family of PP2A regulatory subunits that interacted with Cdc6 and implicated PP2A in regulating Cdc6 phosphorylation (17). The major forms of PP2A contain a dimeric core complex composed of a scaffold (A) and a catalytic subunit (C). The AC core dimer associates with regulatory subunits that form heterotrimeric holoenzymes and target the catalytic subunit to specific phosphoprotein substrates (18 -20). In this study, the mechanism and functional consequences of targeting of PP2A to Cdc6 by PR70 were investigated. The results show that PR70 interacts with PP2A and Cdc6 through distinct regions of the protein, that the association of PP2A to Cdc6 is enhanced by calcium binding to PR70, and that loss of PR70 causes increased levels of Cdc6 and G 1 arrest.

EXPERIMENTAL PROCEDURES
Cloning of Full-length PR70-A human expressed sequence tag encoding the PR70 start codon was identified in the human expressed sequence tag data base using the MegaBLAST tool (www.ncbi.nlm.nih.gov/BLAST/) with the assembled PR70 sequence (21). A PR70 cDNA was constructed using the PR48 cDNA and the IMAGE Human Clone ID 5728169 (GenBank TM accession number BM544432), purchased from Invitrogen, using an internal NcoI restriction site present in the common region of BM54432 and PR48. A PCR fragment containing the translational start codon, the 5Ј-end, and the 3Ј-NcoI site of BM54432 was generated using the BM54432 cDNA as template with the PCR primers: 5Ј-CGGGATCCATGCCGCCCGGCA-AAGT-3Ј (sense strand) and 5Ј-GCGCCTTGATCCGGC-3Ј (antisense strand). The PCR product was digested with the restriction enzymes BamHI and NcoI. The 3Ј portion of the PR48 cDNA was excised from the PR48 cDNA (17) using NcoI and HindIII, and the fragments were ligated and subcloned into the pCMV-Tag2B vector (Stratagene) digested with BamHI and HindIII. The resulting construct encoded a full-length PR70 cDNA fused to an N-terminal FLAG epitope tag. The sequence was verified by automated sequencing.
Cell Culture, Transfection, and RNA Interference-COS-7, HeLa, and U2OS cells were maintained at 37°C in Dulbecco's modified Eagle's medium containing 10% fetal bovine serum in an atmosphere of 5% CO 2 . U2OS, obtained from the ATCC, is a human osteosarcoma cell line that expresses wild-type p53. For transient expression of proteins, cells were transfected with expression plasmids using Lipofectamine 2000 (Invitrogen) according to the manufacturer's protocol. Cells were harvested either 24 or 48 h after transfection. Transfection with small interfering RNA to knock down PR70 was carried out using Oligofectamine (Invitrogen) following the manufacturer's protocol. Annealed duplex siRNAs were purchased from Dharmacon and had the following sequences: PR70-1, 5Ј-AGCCGG-UCCUGAAGAUGAAdTdT-3Ј (sense strand) and PR70-2, 5Ј-AAAGCAUUCCGACCUUCUAdTdT-3Ј (sense strand). Controls included an siRNA that knocks down the MEKK2 protein kinase (22), an siRNA that knocks down protein phosphatase 5 (23), and an siRNA corresponding to the sequence of firefly luciferase (5Ј-TCGAAGTATTCCGCGTACGdTdT-3Ј). Cells were lysed and analyzed by immunoblotting 48 h after transfection. In some experiments, cells were co-transfected with PR70 siRNA and expression plasmids encoding wild-type Cdc6 or Cdc6 mutants in which all three N-terminal phosphorylation sites were mutated to alanine (AAA-Cdc6) or aspartic acid (DDD-Cdc6) using Lipofectamine 2000 and harvested 48 h later. The cDNAs encoding phosphorylation site mutants of Cdc6 were prepared using a PCR-based site-directed mutagenesis kit (Invitrogen) according to the manufacturer's protocol. Wild-type and mutant Cdc6 were expressed as a fusion proteins fused to the N terminus of enhanced green fluorescent protein using the pEGFP-N expression vector (Clontech).
Immunoprecipitation and Immunoblotting-Rabbit antisera were raised against a synthetic peptide corresponding to the C terminus of human PR70 (CDLYEYACGDEDLEPL) conjugated to keyhole limpet hemocyanin. Anti-PR70 antibodies were affinity purified on a peptide column made with the same peptide using the MicroLink Peptide Coupling Kit (Pierce) following the manufacturer's protocol. Rabbit antiserum against human Cdc6 was generated against a full-length Cdc6 fusion protein as described previously (4).
Proteins were immunoprecipitated following the protocol described previously (17). Briefly, the media was aspirated and the cells were washed with cold PBS. The cells were incubated on ice for 20 min in 300 l of IP lysis buffer containing 20 mM Tris-HCl (pH 7.5), 0.2% Nonidet P-40, 20% glycerol, 200 mM NaCl, 1 mM EDTA, and protease inhibitor mixture (Roche Applied Science). Lysates were centrifuged at 14,000 ϫ g for 10 min, and protein complexes were immunoprecipitated from the supernatant. Endogenous PR70 and Cdc6 were immunoprecipitated from 1.2 ϫ 10 6 HeLa cells lysed in 300 l of IP lysis buffer as described above. PR70 was immunoprecipitated using a rabbit antiserum generated against the peptide CDLYEY-ACGDEDLEPL conjugated to hemocyanin. Cdc6 was immunoprecipitated using a rabbit polyclonal antibody generated against a full-length Cdc6-GST fusion protein described previously. As a negative control, immunoprecipitations were performed using pre-immune serum collected from the rabbits immunized against PR70 or Cdc6. 10 l of antiserum and 40 l of protein A-Sepharose (Sigma-Aldrich) were added to 300 l of lysate, and the mixture was incubated for 2 h at 4°C. The protein-A beads were washed three times with IP lysis buffer, and protein was solubilized in 60 l of 2ϫ SDS-PAGE loading buffer. Thirty microliters of solubilized material was resolved on a 10% SDS-PAGE gel and transferred to a nitrocellulose membrane. The membrane was cut into pieces, which were probed with anti-PP2A C-subunit monoclonal antibody 1F6 (24), anti-PP2A A-subunit antiserum (C-20, Santa Cruz Biotechnology), anti-Cdc6 monoclonal antibody (clone DCS-180, Upstate), or anti-PR70 antiserum. Following incubation with horseradish peroxidase-conjugated secondary antibodies, the blots were developed using the enhanced chemiluminescence detection system (Amersham Biosciences).
Transiently expressed FLAG-tagged proteins were immunoprecipitated from 1.5 ϫ 10 6 cells using 7 g of anti-FLAG polyclonal antibody (Sigma-Aldrich) or 7 g of non-immune rabbit IgG (Sigma-Aldrich) and 40 l of protein A-Sepharose (Sigma-Aldrich) for 2 h at 4°C. The immunoprecipitates were washed three times with lysis buffer and solubilized in 60 l of 2ϫ SDS-PAGE loading buffer. 30 l of solubilized protein was resolved on a 10% SDS-PAGE gel and transferred to a nitrocellulose membrane. The membrane was probed with anti-FLAG M2 monoclonal (Stratagene), anti-PP2A C-subunit 1F6, and anti-PP2A A-subunit (C-20, Santa Cruz Biotechnology) antibodies and developed as described above.
Calcium Overlay Assay-In vitro 45 Ca 2ϩ overlay assays were carried out using a protocol described previously (25). Purified GST fusion proteins were resolved by SDS-PAGE and transferred to a nitrocellulose membrane. The membrane was washed three times in IMK buffer (10 mM imidazole-HCl, pH 6.8, 5 mM MgCl 2 , and 60 mM KCl) for 1 h at room temperature. The membrane was then incubated in IMK buffer containing 5 Ci/ml 45 Ca 2ϩ for 10 min. The membrane was washed three times in 30% ethanol for 5 min, dried, and exposed to x-ray film for 12 h.

Cloning of GST-tagged PR70 and EF-hand Mutants-PR70
and PR70 EF-hand mutant cDNA were cloned into the pGEX-4T-1 vector (Amersham Biosciences). The cDNAs were amplified by PCR using the pCMV-Tag2B vector containing the fulllength PR70 or EF-hand mutant cDNA as template with the following primers: 5Ј-CGGGATCCATGCCGCCCGGCAA-AGT-3 (sense strand) and 5Ј-ATTTGCGGCCGCTCACAG-CGGCTCCAGGTC-3Ј (antisense strand). The products were digested with BamHI and NotI and ligated into pGEX 4T-1, which had been cut with the same restriction enzymes. The resulting constructs encode the GST protein fused to the N terminus of full-length PR70 or EF-hand mutant proteins. The sequences were verified by automated sequencing.
Expression and Purification of GST-Cdc6, GST-A, and GST-PR70 Fusion Proteins-A GST-Cdc6 fusion protein was prepared by a modification of a method previously described (6). Briefly, 1 liter of Sf9 cells (2 ϫ 10 6 cells/ml) was infected with recombinant GST-Cdc6 baculovirus (a gift of Dr. Ellen Fanning, Vanderbilt University) at a Sf9 culture:baculovirus ratio of 1:20 (v/v) for 60 h. The cells were collected by centrifugation and washed once with PBS. Cells were lysed on ice in 40 ml of buffer A (100 mM Tris-HCl, pH 7.4, 100 mM NaCl, 5 mM KCl, 0.5 mM MgCl 2 , 0.5% Nonidet P-40, 1 mM dithiothreitol, 10 mM NaF, 1 mM EGTA, 2 mM EDTA, and a protease inhibitor tablet (Roche Applied Science)) using a Dounce homogenizer. Lysates were centrifuged at 30,000 ϫ g for 30 min at 4°C to remove cellular debris, and the lysate was mixed with 2 ml of glutathione-agarose (Sigma-Aldrich) for 2 h at 4°C. The resin was recovered by centrifugation and washed twice with PBS, once with PBS containing 1.5 M NaCl, and once with PBS containing 1.5 mM NaCl and 0.1% (v/v) Nonidet P-40, and then re-equilibrated in PBS. The GST-Cdc6 fusion protein immobilized on glutathione agarose beads was resuspended in buffer B (20 mM HEPES, pH 7.6, 100 mM KCl, 1 mM dithiothreitol, 1 mM EDTA, and 50% glycerol) and stored at Ϫ80°C.
GST-A fusion protein, GST-PR70, and GST-EF-hand mutants were expressed in bacteria and prepared as previously described (26). The GST fusion proteins immobilized on glutathione agarose beads were resuspended in buffer B and stored at Ϫ80°C until use.
GST Pulldown Assays-GST, GST-A, and GST-Cdc6 immobilized on glutathione-agarose beads were used to assess the binding of PR70. Wild-type or mutant FLAG-PR70 was expressed by transient transfection of COS-7 cells. The cells were lysed on ice in 300 l of IP lysis buffer or in 300 l of IP lysis buffer containing 10 mM EDTA or 10 mM CaCl 2 for 20 min. GST pulldown assays (26) were conducted by incubating 300 l of lysate with either GST, GST-A, or GST-Cdc6. The samples were incubated for 1 h at room temperature with agitation. The calpain inhibitor calpeptin (Calbiochem) was added at a concentration of 50 M in some experiments. Following incubation, the sample was washed three times with IP lysis buffer supplemented with EGTA, CaCl 2 , or CaCl 2 and calpeptin, and the beads were collected by centrifugation. After washing, the bound proteins were solubilized in 60 l of 2ϫ SDS-PAGE loading buffer. 30 l of solubilized protein was resolved on a 10% SDS-PAGE gel and transferred to a nitrocellulose membrane. The membrane was probed with anti-FLAG monoclonal (M2, Stratagene), anti-PP2A C-subunit 1F6, and anti-PP2A A-subunit (C-20, Santa Cruz Biotechnology) antibodies and developed as described above.
Flow Cytometry-U2OS cells (4 ϫ 10 5 ) were seeded into 60-mm dishes and transfected with PR70 or control siRNA 24 h later. The cells were then incubated for 48 h and harvested by trypsinization, washed once with PBS, and resuspended in 0.5 ml of PBS. The cell suspension was then added to 4.5 ml of 70% ethanol and incubated on ice for 2 h. Cells were collected by centrifugation, washed once with PBS, and suspended in 1 ml of propidium iodide/Triton X-100 staining solution with RNase (0.1% Triton X-100, 0.2 mg/ml DNase-free RNase, and 10 g/ml propidium iodide in PBS). The DNA content of 10,000 cells was determined using a BD Biosciences FACScan flow cytometer and FlowJo software. Single cells were gated away from clumped cells using an FL3 width versus FL3 height dot plot, and the DNA content of individual cells was plotted as FL3 area versus cell number.

RESULTS
Interaction of PR70 with Cdc6-The original cDNA for PR70, termed PR48, was identified in a yeast two-hybrid screen using the human Cdc6 protein as bait (17) and subsequently shown to be a fragment of a longer cDNA (27). A full-length human PR70 cDNA was constructed by ligating expressed sequence tag BM54432 to the PR48 cDNA using an internal NcoI restriction site. The predicted open reading frame of the PR70 cDNA encodes a protein with a calculated molecular mass of 65.1-kDa and corresponds to the longer transcript (variant 1) of the PPP2R3B gene (GeneID: 28227). The predicted amino acid sequence of PR70 is highly similar to the human PR72 and mouse PR59 members of the PPP2R3 gene family, but more distantly related to the G5PR protein (supplemental Table S1). An alignment of the PPP2R3 family (supplemental Fig. S1) revealed a highly conserved central domain, termed the R3 domain, that contains two conserved EF-hand calcium binding motifs previously identified in PR72 (27). Rabbit antisera were raised against a peptide corresponding to the unique C terminus of PR70 and affinity purified on a peptide column. The purified antibodies recognized a protein band of M r ϭ 70,000 in lysates of HeLa cells. The 70-kDa protein recognized by the antibody was greatly reduced in cells treated with two different siRNAs corresponding to sequences within PR70 but not with control siRNA (supplemental Fig. S2).
To verify that PR70 associates with PP2A and Cdc6, HeLa lysates were immunoprecipitated with PR70 and Cdc6 antibodies. Immunoprecipitation of Cdc6 co-precipitated a diffuse protein band that migrated at the position of PR70 that was not present in immunoprecipitates obtained with the preimmune serum (Fig. 1A). The anti-Cdc6 serum also co-precipitated the A-and C-subunits of PP2A. Although the anti-PR70 antibodies co-precipitated the A-and C-subunits of PP2A (Fig. 1B), a complex between PR70 and Cdc6 could not be detected. The inability to detect Cdc6 may be due to steric hindrance by the anti-C-terminal antibody, because the C terminus of PR70 is required for interaction with Cdc6 (see below). The association of PR70 with PP2A and Cdc6 was also tested in co-immunoprecipitation experiments using exogenously expressed proteins. FLAG-tagged PR70 was expressed in HeLa cells and immunoprecipitated with anti-FLAG antibodies. Analysis of the immunoprecipitates by immunoblotting showed that the endogenous A-and C-subunits of PP2A and HA-tagged Cdc6 co-precipitated with FLAG-PR70 (Fig. 1C). These results indicate that PR70 can interact with both the PP2A core dimer and Cdc6 in intact cells.
Calcium Enhances the Interaction of PR70 with the AC Core Dimer and Recruits PP2A to Cdc6-Analysis of the amino acid sequence of PR70 identified two EF-hand calcium binding motifs that are conserved within the PPP2R3 family (supplemental Fig. S1). The roles of these motifs were tested in a gel overlay assay with wild-type PR70 and PR70 containing inactivating mutations of the EF-hand motifs. Point mutants were constructed that had substitutions of amino acids involved in calcium binding (28), including alanine substitutions at both the x and y coordinates and a conservative change at the -z coordinate ( Fig. 2A). Wild-type PR70 bound calcium in the in vitro 45 Ca 2ϩ overlay assay (Fig. 2B). Mutation of the first EFhand (EF1) resulted in reduced binding of calcium compared with wild-type PR70. Mutation of the second EF-hand (EF2) severely reduced calcium binding, whereas the double mutation of EF1 and EF2 nearly abolished the ability of PR70 to bind calcium.
Calcium binding causes a conformational change in the PR72 member of the PPP2R3 family that is associated with enhanced interaction with the A-subunit of PP2A (27). Therefore, the effects of calcium on the interaction of PR70 with PP2A and Cdc6 were determined. FLAG-tagged PR70 was expressed in COS-7 cells, which were lysed in buffer containing EGTA or calcium. The lysates were incubated with either GST-A or GST-Cdc6 and bound proteins detected by immunoblotting. FLAG-PR70 interacted with both GST-A and GST-Cdc6 but not GST alone (Fig. 2C, lanes 1-2 and 6 -7). Compared with lysates prepared with standard buffer or EGTA, the addition of calcium enhanced the binding of PR70 to GST-A, but not to GST-Cdc6 (Fig. 2C, lanes 4 and 9). Although calcium did not enhance the binding of PR70 to GST-Cdc6, it did cause a significant increase in the amount of A-and C-subunits associated with GST-Cdc6 (Fig. 2C, lanes 9 and 10). Although an excess of calcium was used in the experiments shown in Fig. 2C, other experiments showed that enhanced binding of the AC core dimer was also observed at calcium concentrations of 100 M (not shown).
To test the function of the individual EF-hand motifs in the calcium-enhanced interaction with PP2A, GST pulldown experiments were performed with the calcium-binding mutants. Compared with assays in the presence of EGTA, the addition of calcium resulted in enhanced binding of wild-type PR70 and the EF1 mutant to GST-A, but not to GST-Cdc6 ( Fig.  3A and B, lanes 2-5). Although it did not increase the amount of PR70 or the EF1 mutant associated with Cdc6, calcium did increase the association of the A-and C-subunits with GST-Cdc6 (Fig. 3B, lanes 2-5). Mutation of EF2 or mutation of both EF1 and EF2 resulted in loss of the calcium-enhanced binding of PR70 to GST-A (Fig. 3A, lanes 6 -9) and the calcium-dependent association of the A-and C-subunits with GST-Cdc6 (Fig.  3B, lanes 6 -9).
The effect of calcium on the interaction of PR70 with PP2A was also assessed by expression and immunoprecipitation in COS-7 cells. Both wild-type PR70 and the EF1 mutants interacted with endogenous PP2A (Fig. 3C, lanes 2-4). The EF1(-z) mutant interacted as well as wild-type PR70, but interaction of the EF1(x,y) mutant was reduced suggesting that mutation of the x and y residues causes a structural defect in PR70. Mutation of EF2, or both EF1 and EF2, resulted in a nearly complete loss of interaction with PP2A (Fig. 3C, lanes 5-8). A longer exposure of the blot showed that a weak interaction of the Aand C-subunits with the EF2 and EF1/EF2 double mutants could still be detected (Fig. 3C, OE). The combination of intact cell data and in vitro binding assays provide evidence that PR70 is a calcium binding protein and that interaction with the core dimer of PP2A is enhanced by binding of calcium to the second EF-hand motif. Calcium does not affect interaction of PR70 with Cdc6 but increases the association of the PP2A core dimer with Cdc6 in a manner dependent upon binding of calcium to the second EF-hand of PR70. PP2A and Cdc6 Bind to Distinct Regions of PR70-Comparison of the amino acid sequences of the PPP2R3 regulatory subunits identified a conserved domain in the central region of PR70 (supplemental Figs. S1 and S4A). The R3 domain is 66% identical and 82% conserved between human PR70 (PPP2R3B) and PR72 (PPP2R3A). A series of truncation mutants were constructed to identify regions within PR70 that were important for interaction with PP2A and Cdc6. FLAG-tagged mutants were expressed in COS-7 cells and immunoprecipitated with anti-FLAG antibody. The ability of the mutants to incorporate into endogenous PP2A heterotrimers was determined by immunoblotting for associated A-and C-subunits. The ⌬N1 mutant contains a deletion of the entire N-terminal PR70unique region and interacted with endogenous PP2A subunits to the same extent as full-length PR70 (Fig. 4B). Deletion of a C-terminal segment that included the PR70-unique region (⌬C) had little, if any, effect on the interaction with PP2A. However, deletions of N-terminal regions of the conserved R3 domain, ⌬N2 and ⌬N3, resulted in proteins that failed to interact with PP2A. These data indicated that the region between amino acids 125 and 136 of PR70 were necessary for interaction with the PP2A core dimer.
The sequence between residues 125 and 136 of PR70 contains a hydrophobic motif (FYF) that was conserved in PPP2R3 proteins from humans to flies (Fig. 5A). The role of the FYF motif was tested by mutating these residues to alanines (Fig. 5B) and determining the effects on interaction with the AC core dimer. Mutation of any one of these residues resulted in a significant loss of interaction with endogenous PP2A (Fig. 5C). A longer exposure of the immunoblot showed that small amounts of the A-and C-subunits could be detected in immunoprecipitates of each of the mutants (not shown). Although the interaction of the FYF mutants was severely compromised in intact cells, these mutants still bound to PP2A when assayed in vitro by GST pulldown experiments (not shown). These results indicate that the FYF motif contributes to the interaction of PR70 with the A-subunit.
The N-and C-terminal truncation mutants of PR70 were also tested for their ability to interact with Cdc6. Pulldown experiments with GST-Cdc6 were performed with full-length PR70, the ⌬N3, and the ⌬C mutants in the presence and absence of calcium. As expected, full-length PR70 and the ⌬C mutant interacted with the A-subunit of PP2A, whereas the ⌬N3    mutant did not (Fig. 6A). Furthermore, there was an enhanced interaction of PR70 and the ⌬C mutant in the presence of calcium. As shown previously with full-length PR70 (Fig. 2C), the enhanced binding of PR70 and the ⌬C mutant was accompanied by a decrease in the amount of C-subunit associated with GST-A (Fig. 6A, lanes 3 and 9). This decrease in associated C-subunit was not observed with the ⌬N3 mutant, which did not bind to GST-A. These observations suggest that the decrease in C-subunit in the presence of calcium is due to displacement of endogenous regulatory and catalytic subunits from GST-A by excess free PR70.
Both full-length PR70 and the ⌬N3 mutant bound to GST-Cdc6 (Fig. 6B). However, only full-length PR70 was able to recruit additional A-and C-subunits in the presence of calcium. In contrast, the ⌬C mutant bound very poorly to Cdc6 in the presence or absence of calcium. A low level of the A-and C-subunits was pulled down with GST-Cdc6 from lysates expressing the ⌬N3 or ⌬C mutants (Fig. 6B, lanes 5-6 and 8 -9). Similar amounts of these subunits were also bound to GST-Cdc6 in lysates from non-transfected cells (not shown) suggesting that GST-Cdc6 can interact with endogenous A-and C-subunits in the absence of expressed PR70 (presumably by binding to endogenous PR70). The amounts of A-and C-subunits bound to GST-Cdc6 in experiments with the ⌬N3 or ⌬C mutants were increased in the presence of calcium. This observation provides additional support for the conclusion that calcium can regulate the association of PP2A with Cdc6 and shows that a C-terminal region of PR70, which includes the PR70-unique domain, is necessary for interaction with Cdc6.
PR70 Regulates Cdc6 Levels-Because phosphorylation of the N-terminal regulatory sites of Cdc6 inhibits degradation, loss of the phosphatase that dephosphorylates these sites should promote accumulation of Cdc6. Therefore, RNA interference was used to determine if knockdown of PR70 affected the levels of Cdc6. Knocking down the catalytic subunit of PP2A increased the levels of endogenous Cdc6 in HeLa cells (Fig. 7A). Treatment of cells with either a control siRNA or an siRNA that knocks down protein phosphatase 5 had no effect on Cdc6 levels. Knocking down the PR70 subunit also caused a substantial increase in the levels of Cdc6 (Fig. 7B). The increase in Cdc6 levels occurred with two PR70 siRNAs targeted to distinct regions of the mRNA. The accumulation of Cdc6 following knockdown with PR70-1 siRNA appeared to be greater than that with PR70-2 siRNA, which is consistent with the greater efficiency of the PR70-1 siRNA in reducing PR70 levels (supplemental Fig. S2). Phosphorylation site mutants of Cdc6 were then used to test the role of phosphorylation in the accumulation of Cdc6 caused by knockdown of PR70. Knockdown of PR70 caused an increase in the levels of expressed wild-type GFP-Cdc6 compared with transfection with a control siRNA (Fig. 7B). Transfection with a mutant of Cdc6 in which all three N-terminal phosphorylation sites had been mutated to phospho-mimicking aspartic acid residues (DDD-Cdc6) resulted in substantially higher levels of expression than those observed with the wild-type protein as previously reported (15). PR70-1 siRNA had little or no effect on the levels of DDD-Cdc6. The ability of PR70 knockdown to cause accumulation of Cdc6 was also greatly diminished when the phosphorylation sites were    1, 4, and 7). mutated to non-phosphorylatable alanine residues (AAA-Cdc6). Similar results were seen in U2OS cells. These data indicate that knockdown of PR70 results in an increase in the levels of endogenous and exogenous Cdc6 that is dependent on the presence of phosphorylatable residues at the N-terminal phosphorylation sites.
The effects of overexpressing PR70 on Cdc6 levels were also determined. When HeLa cells were transfected with expression plasmids containing the CMV promoter, FLAG-tagged PR70 was expressed at levels 5-to 10-fold higher than the endogenous protein (not shown). Co-expression of CDK2 and Cdc6 caused a substantial increase in Cdc6 levels as reported previously (15). Expression of wild-type PR70 caused an increase in the levels of both co-transfected and endogenous Cdc6 (Fig.  8A). Expression of the ⌬N3 or EF1/2 mutants, which cannot interact with the AC core dimer but bind to Cdc6, also increased the levels of Cdc6. In contrast, expression of the ⌬C mutant, which binds to the AC core dimer but not to Cdc6, had  relatively little effect on the levels of exogenous or endogenous Cdc6. The potential role of phosphorylation in the effects of overexpressed PR70 was tested using the phosphorylation site mutants of Cdc6. Although co-expression of PR70 caused some increase in the levels of the AAA mutant of Cdc6, the effect was much less than its effect on wild-type Cdc6 (Fig. 8B). Similarly, co-expression of PR70 had little effect on the levels of the DDD mutant of Cdc6 even though endogenous Cdc6 was increased. Thus, the ability of overexpressed PR70 to cause accumulation of the protein was inhibited when the phosphorylation sites in Cdc6 were mutated. The effects of PR70 overexpression to cause accumulation of Cdc6 suggest it acts in a dominant-negative manner to block Cdc6 dephosphorylation (see "Discussion").
Knockdown of PR70 Causes G 1 Arrest-The potential role of PR70 in progression through G 1 was determined by determining the cell cycle distribution of cells in which PR70 was depleted by RNA interference. Knockdown of PR70 caused accumulation of cells in G 0 /G 1 and depletion of cells in S and G 2 /M (Fig. 9). The apparent G 1 arrest occurred with either of two siRNAs that target distinct regions of PR70. The level of G 1 arrest correlated with the extent of PR70 knockdown. The lower levels of PR70 achieved with the PR70 siRNA-1 compared with PR70 siRNA-2 corresponded to a greater increase in the number of G 1 cells (76% versus 65%). The G 1 arrest caused by knockdown of PR70 supports a role for this PP2A regulatory subunit in progression through G 1 phase.

DISCUSSION
The formation of pre-replicative complexes during the initiation of DNA replication is regulated, in part, by the availability of Cdc6. Cyclin-dependent kinases phosphorylate regulatory sites within the N-terminal domain of Cdc6 and block ubiquitination by APC/C cdh1 and subsequent degradation by the proteasome (15). The results reported here help establish the form of PP2A complexed with the PR70 regulatory subunit as a physiological Cdc6 phosphatase and are consistent with a model in which PR70 targets PP2A to Cdc6 through direct protein-protein interactions. Knockdown of PR70 by RNA interference results in an increase in the levels of Cdc6 protein, consistent with a role for this subunit in regulating the stability of Cdc6. Overexpression of PR70 appeared to act in a dominant-negative manner to also increase the levels of Cdc6. The observations that increased protein levels did not occur with phosphorylation site mutants of Cdc6 are consistent with a role for PR70 in regulating Cdc6 phosphorylation and stability. A novel aspect of this model is the potential regulation of Cdc6 dephosphorylation by calcium. Calcium enhances the recruitment of the core dimer of PP2A to Cdc6 by binding to the EF-hand motifs of PR70, raising the possibility that changes in intracellular calcium can regulate the accumulation of Cdc6 and initiation of DNA replication. However, it remains to be determined if physiological changes in intracellular calcium ily. The A-subunit of PP2A is a HEAT repeat protein (36). The FYF motif in PR70 resembles the FG amino acid repeats (FXFG and GLFG) within the nucleoporin family of nuclear pore proteins. The nucleoporins interact with nuclear transport factors, including importin-␤, which are also HEAT repeat proteins. The FG repeats of nucleoporins bind to shallow hydrophobic pockets in importin-␤ (37,38). The A-subunit of PP2A contains exposed hydrophobic surfaces, predicted to play a role in interaction with the regulatory subunits (36), that are possible sites of interaction with the FYF motif of PR70.
The requirement for the N-terminal region of the R3 domain of PR70 for binding to the A-subunit is distinct from results observed with the PR72 protein. A fragment of PR72 consisting of amino acids 219 -473 (corresponding to residues 257-509 of PR70) interacts with the A-subunit in the yeast two-hybrid assay (27). This fragment of PR72 is missing the N-terminal region of the R3 domain. Two fragments of PR72 containing putative A-subunit binding domains prepared by in vitro translation (corresponding to residues 234 -339 and 378 -436 of PR70) interacted with the A-subunit in vitro using GST pulldown assays (39). These PR72 fragments also do not contain the conserved N-terminal region of the R3 domain that was necessary for interaction of PR70 with the AC core dimer in our assays. These observations indicate that additional regions of PR70, beyond those required in PR72, are required for binding to the A-subunit, or that the differences observed are due to different assays employed.
In summary, the present study shows that the PR70 regulatory subunit targets PP2A to Cdc6 and that PP2A is likely to be a physiological Cdc6 phosphatase. The targeting of PP2A to Cdc6 is enhanced by binding of calcium to PR70 raising the possibility that changes in intracellular calcium can influence formation of pre-replicative complexes through regulation of Cdc6 dephosphorylation.