Cell Cycle-dependent Phosphorylation of the Large Subunit of Replication Factor C (RF-C) Leads to Its Dissociation from the RF-C Complex*

The five subunit replication factor C (RF-C) complex plays a critical role in DNA elongation. We find that the large subunit of RF-C (RF-Cp145) is phosphorylated in vivo whereas the smaller RF-C subunits are not phosphorylated. The phosphorylation of endogenous RFCp145 is modulated in a cell cycle-dependent manner. Phosphorylation is maximal in G2/M and is inhibited by an inhibitor of cyclin-dependent kinases. Phosphorylation of purified recombinant RF-C complex in vitro reveals that RF-Cp145 is preferentially phosphorylated by cdc2-cyclin B but not by cdk2-cyclin A or cdk2-cyclin E. In vitro phosphorylation of RF-C complex by cdc2-cyclin B kinases leads to dissociation of phosphorylated RFCp145 from the RF-C complex. Using different approaches we demonstrate that phosphorylated RFCp145 is indeed dissociated from RF-Cp40 and RF-Cp37 in vivo. These results suggest that destabilization of the RF-C complex by CDKs may inactivate the RF-C complex at the end of S phase.

Cell cycle progression is regulated by distinct cyclin-dependent kinases (CDKs), 1 which activate at different times in the cell cycle. In mammalian cells, cyclin E-dependent kinase is activated at the G 1 to S phase transition, after the D-type cyclins but prior to A-type cyclins. The timing of activation of cyclin A kinase activity coincides with the onset of DNA synthesis (1). Cdc2-cyclin B kinase activation at the G 2 /M transition is required for cells to enter mitosis. Replication of DNA is a highly regulated process, which occurs during the S phase of the cell cycle. The DNA polymerase ␣-primase complex allows the generation of RNA-DNA primers for initiation of leading strand synthesis and synthesis of each Okazaki fragment during lagging strand replication. The polymerase switch from DNA polymerase ␣ to DNA polymerase ␦ then catalyzes the replication of the leading strand and for completion of the lagging strand. The replication factor C (RF-C) complex and PCNA are essential for processive DNA synthesis. RF-C loads PCNA onto DNA in an ATP-dependent process (2,3). The PCNA clamp then recruits DNA polymerase ␦ for processive DNA synthesis (4).
The Escherichia coli clamp loader, the ␥ complex, like the yeast and human RF-C complexes is comprised of five polypeptides. The DNA-dependent ATPase activity of the clamp in the ␥ complex is provided by three copies of the ␥ polypeptide whereas human (and yeast) utilize three different polypeptide chains hRF-Cp37 (yRF-C2), hRF-Cp36 (yRF-C3), and hRF-Cp40 (yRF-C4). The ␦ subunit (homologous to yRF-C1 in yeast and hRF-Cp145 in man) binds the clamp and has conserved hydrophobic residues required for clamp loading. Structural studies of the ␥ complex: ␦Ј-␥1-␥2-␥3-␦ reveals that the Cterminal domains of ␦Ј, ␥, and ␦ form a helical scaffold (circular collar) and the N-terminal ends appear to dangle under the C-terminal pentamer umbrella (17).
In this study we show that the RF-Cp145 is a target of CDKs in vivo. The other subunits in the RF-C complex are not targets of CDKs. Phosphorylation of RF-Cp145 increases as cells traverse S phase, peaking in G 2 /M. Treatment of cells with roscovitine, a specific inhibitor of cdk-cyclin kinases, inhibits RF-Cp145 phosphorylation in G 2 /M. This report demonstrates that RF-C complex is phosphorylated during cell cycle by cdk-cyclin kinases and reveals a unique function of RF-Cp145 phosphorylation in human cells.

MATERIALS AND METHODS
Cell Cycle Synchronization-HeLa cells were synchronized by a double thymidine block. HeLa cells (1 ϫ 10 6 cells/10 cm 2 dish) were grown overnight in DMEM plus 10% bovine serum (Invitrogen). Thymidine (Sigma, final concentration 2 mM) was added to the medium and culture continued for 18 h. The cells were washed with phosphate-buffered saline, and transferred to regular medium for 8 h, followed by readdition of 2 mM thymidine to the medium for 18 h. Finally cells were washed and released from the block and synchronous cell cycle progression monitored by analyzing DNA content (propidium iodide) and MPM-2 positivity by flow cytometry.
In Vivo Labeling-HeLa cells collected at indicated times after release from a double thymidine block were washed with Hepes-buffered saline (20 mM Hepes, 150 mM sodium chloride) and grown in phosphatefree Dulbecco's modified Eagle's medium (DMEM) (Invitrogen) supplemented with 5% dialyzed bovine serum for 60 min, to deplete intracellular phosphate. The medium was aspirated and replaced with phosphate-free DMEM containing 0.5-1.0 mCi/ml inorganic [ 32 P]orthophosphoric acid (PerkinElmer Life Sciences, cat. no. NEX053) for 2 h. Finally, the cells were harvested by scraping and sequential extraction on ice in 1 ml each of low-salt (10 mM Hepes, 0.1% Triton X-100, 0.5 mM DTT, 1.5 mM MgCl 2 , 10 mM KCl, 5 mM EDTA) and high salt (50 mM Hepes, 0.1% Triton X-100, 0.5 mM DTT, 500 mM NaCl) lysis buffers to obtain chromatin-bound fractions of RF-C. Each lysis buffer was supplemented with protease (soybean trypsin inhibitor, 2 g/ml aprotinin, 2 g/ml leupeptin, 0.5 mM phenylmethylsulfonyl fluoride) and phosphatase (5 mM sodium fluoride, 1 mM sodium orthovanadate) inhibitors. In some experiments cells were lyzed directly in high salt buffer. Cell lysates were clarified by centrifugation (10,000 ϫ g for 10 min at 4°C), and the supernatants of the high salt extractions were used for immunoprecipitation.
Immunoprecipitation of Endogenous RF-C Subunits and Immunoblots-Polyclonal rabbit antisera specific for RF-Cp145 (large subunit of RF-C) were generated by immunizing rabbits with recombinant protein spanning amino acid residues 369 -480 of RF-Cp145. The sera were preadsorbed to GST beads and then affinity-purified on agarose affinity matrices covalently linked with RF-Cp145 domain A (369 -480) recombinant protein. The anti-RF-Cp37 and anti-RF-Cp40 antibodies were kindly provided by Dr. J. Hurwitz, the anti-RF-Cp145 specific monoclonal antibodies were provided by Dr. B. Stillman and the PCNAspecific antibodies (PC10) were from Santa Cruz Biotechnology. The rabbit polyclonal antibodies used for immunoprecipitation included anti-RF-Cp145, 4 l; anti-RF-Cp37, 1 l, and anti-RF-Cp40, 1 l. Immunoprecipitation was performed by mixing lysate from 1-2 ϫ 10 6 cell equivalents (in ϳ1-ml volume) with the appropriate antibody overnight at 4°C on a nutator. The immune complexes were isolated by incubation with 30 -40 l of protein G-Sepharose beads (Amersham Biosciences) for 1 h at 4°C, followed by five washes in cold TNE buffer (20 mM Tris-HCl, pH 7.5, 1% Nonidet P-40, 150 mM NaCl, 10 mM EDTA). Finally, the immune complexes were boiled in sample buffer, resolved by SDS-PAGE on 8.5% gels, and transferred to nitrocellulose, followed by autoradiography or immunoblot analysis. For immunoblotting, proteins were transferred to nitrocellulose, blocked with 5% nonfat milk/TBS containing 0.1% Tween 20, and incubated with the appropriate primary antibody using the following dilutions: anti-RF-Cp145, 1:3000; anti-RF-Cp37, 1:2000; anti-RF-Cp40, 1:2000; anti-PCNA, 1 g/ml. After overnight at 4°C, membrane was washed and incubated with the appropriate secondary antibody conjugated to horseradish peroxidase (Amersham Biosciences) followed by ECL (Amersham Biosciences).
Purified RF-C Complex and in Vitro Kinase Assays-Recombinant RF-C complex was purified from insect cells as described previously (14). All kinase assays were performed in 18 l of kinase reaction mixture containing 40 mM Hepes, 8 mM MgCl 2 , 166 mM ATP, 1 Ci [␥-32 P]ATP (3000 Ci/mmol, Amersham Biosciences) and pre-activated cdk-cyclin kinases as described earlier (18) for 20 min at 25°C unless otherwise indicated.
Sucrose Gradients-Asynchronous HeLa cell cultures were labeled in phosphate-free DMEM containing 0.5-1.0 mCi/ml inorganic [ 32 P]orthophosphoric acid for 2 h as described above. Cells were harvested by scraping and sequentially extracted on ice in 1 ml each of low salt (10 mM Hepes, 0.1% Triton X-100, 0.5 mM DTT, 1.5 mM MgCl 2 , 10 mM KCl, 5 mM EDTA) and high salt (50 mM Hepes, 0.1% Triton X-100, 0.5 mM DTT, 500 mM NaCl) lysis buffers to obtain chromatin-bound fractions of RF-C. The high salt supernatants were then resolved on a sucrose step gradient (5/10/15/20% sucrose in high salt lysis buffer) by centrifugation in P100TU rotor in a Sorvall ultracentrifuge for 7 h at 45,000 rpm. The nine fractions collected after ultracentrifugation were then analyzed by immunoprecipitation with anti-RF-Cp145 specific antibodies. The presence of 32 P-labeled RF-Cp145 was determined by autoradiography and the presence of RF-Cp145 and RF-Cp37 proteins was determined by immunoblotting.  (36) have been described previously. Total cell lysate is shown as control. C, asynchronously growing NIH3T3, HeLa and COS-1 cells were labeled with 32 P-inorganic phosphate for 2 h. Cell lysates were subject to immunoprecipitation using either polyclonal anti-RF-Cp145 antibodies or preimmune serum (Control IP). RF-C in immune complexes was resolved by SDS-PAGE, transferred to nitrocellulose and visualized by autoradiography. The phosphorylated RF-Cp145 in immunoprecipitates was identified by its comigration with endogenous RF-Cp145 identified by immunoblotting HeLa cell lysates with polyclonal RF-Cp145-specific antibodies. D, COS-1 cells transfected with either empty vector or HA epitope-tagged RF-Cp145 expression vector were labeled with 32 P-inorganic phosphate for 2 h and cell lysates subjected to immunoprecipitation with anti-HA antibody. Immunoprecipitates were visualized by autoradiography as in C. This nitrocellulose filter was subsequently probed with anti-HA antibodies to visualize HA-RF-Cp145.

RF-Cp145 Is a Phosphoprotein in Vivo-
We generated antibodies specific for RF-Cp145 (large subunit of RF-C) using a recombinant protein spanning amino acid residues 369 -480 of RF-Cp145. This antibody specifically detects transfected fulllength RF-Cp145 in immunoblots (Fig. 1A). This RF-Cp145specific antibody also efficiently immunoprecipitated endogenous RF-Cp145 protein and the associated smaller subunits of the RF-C complex such as RF-Cp37 (Fig. 1B, right panel).
Using these RF-Cp145-specific antibodies, we examined if endogenous RF-Cp145 is phosphorylated in vivo. Cell extracts from Hela cells labeled with inorganic 32 P-orthophosphate were immunoprecipitated with anti-RF-Cp145 antibodies and the immunoprecipitates resolved on denaturating polyacrylamide gels were analyzed by autoradiography. A single major phosphoprotein, which comigrated with immunoreactive RF-Cp145 was identified in the RF-Cp145 immunoprecipitate (Fig.  1C). RF-Cp145 was also phosphorylated in vivo in other cell lines like COS-1, NIH3T3 (Fig. 1C), and MCF7 (data not shown). As expected no phosphoprotein corresponding to RF-Cp145 was observed in control immunoprecipitates with preimmune sera. In vivo phosphorylation of RF-Cp145 was also detected with other RF-Cp145-specific antibodies. These included three anti-RF-Cp145 monoclonal antibodies (8), which we have found to recognize the 1-368 region of RF-Cp145 (data not shown) and with a rabbit antibody specific for amino acid residues 481-728 of RF-Cp145 (gift of U. Hubscher, Zurich). Transfected HA epitope-tagged RF-Cp145 is phosphorylated in vivo. This was shown by immunoprecipitation of cell lysates from transfected cells with anti-HA epitope-specific antibodies (Fig. 1D). These results taken together demonstrate that RF-Cp145 is a phosphoprotein in vivo. RF-Cp145 is the primary target of phosphorylation in the RF-C complex since no phosphoproteins co-migrating with immunoreactive RF-Cp37 or RF-Cp30 were observed in anti-RF-Cp145 immunoprecipitates ( Fig. 1C and Fig. 3B).
Phosphorylation of RF-C by cdc2-cyclin B Kinase in Vitro-Recombinant RF-C complex was generated by co-infection of insect cells with five baculoviruses encoding individual RF-C subunits (p145, p40, p38, p37, and p36) and purified on Ni-NTA and Mono Q columns as described (14). The RF-Cp40 in this complex is tagged with a His epitope. The presence of this tag does not affect the biological activity of the RF-C complex (14). This purified recombinant RF-C complex ( Fig. 2A) was a good in vitro substrate for cdc2-cyclin B kinase (Fig. 2B). In contrast, cdk2-cyclin A and cdk2-cyclin E kinases although equally active in phosphorylating histone H1 (Fig. 2C) were unable to phosphorylate the RF-C complex efficiently (Fig. 2B). The large subunit of RF-C is the primary phosphorylation substrate of cdk kinases in such in vitro kinase reactions. In order to determine if such in vitro phosphorylation events occur physiologically, we set out to determine if the RF-C complex (RF-Cp145) is phosphorylated in vivo in a cell cycle-dependent manner.
Phosphorylation of RF-Cp145 Is Modulated during Cell Cycle Progression-Since the best-studied activity of RF-C is restricted to S phase during DNA replication we wanted to determine the temporal regulation of RF-Cp145 phosphorylation during cell cycle progression. For these studies, HeLa cells were synchronized at the G 1 /S border using a double thymidine block. After release from the block, cells were collected at various time intervals and analyzed for DNA content to determine their position in cell cycle. In addition, the cells were also stained for a mitotic marker, MPM2, to identify cells in mitosis. Immediately after release from the block, cells entered S phase and by 7 h, had a 4N DNA content. Cells at 7 h after release represented G2 since the peak of MPM2 staining occurred at 8 h after release (Fig. 3A). By 10 h cells were in G 1 and shortly thereafter lost synchrony.
We next examined the status of RF-Cp145 phosphorylation in synchronized cells collected at various time points after release from a double thymidine block. Cells were labeled with 32 P-inorganic phosphate before extracts were prepared for immunoprecipitation. The phosphorylation of RF-Cp145 increased as cells traversed through S phase, reaching a peak in G 2 /M (Fig. 3B). When lysates from the same synchronized HeLa cell populations were analyzed by immunoblotting we found that RF-Cp145 protein levels are not altered significantly in the different phases of the cell cycle (Fig. 3B). The differences in phosphorylation of RF-Cp145 through the cell cycle are thus not due to differences in the levels of RF-Cp145 protein. We therefore conclude that phosphorylation of RF-Cp145 is cell cycle-regulated.
Inhibitor of cdk-cyclin Kinases Abrogates RF-Cp145 Phosphorylation-The cell cycle-dependent in vivo RF-C phosphorylation suggests that phosphorylation of RF-Cp145 may be regulated in vivo by cdk-cyclin kinases. Consistent with such an idea phosphoamino acid analysis of RF-Cp145 shows that it is phosphorylated on serine/threonine residues (data not shown). We therefore examined the effect of cdk-cyclin kinase inhibitor on RF-Cp145 phosphorylation in synchronously cycling cells. Synchronously cycling Hela cells in G 2 /M, released from a double thymidine block, were cultured either in the presence of 32 P-inorganic phosphate alone or in combination with 80 M roscovitine. Roscovitine has been used previously as a specific inhibitor of cdk2/cdc2-cyclin kinases (19,20). Roscovitine dramatically inhibits the phosphorylation of RF-Cp145 during cell cycle progression through G 2 /M (Fig. 4).
Phosphorylation by cdc2-cyclinB Promotes Dissociation of RF-Cp145 from the RF-C Complex-We next determined if phosphorylation with cdc2-cyclin B leads to a dissociation of RF-Cp145 from the RF-C complex. For these studies we used recombinant RF-C complex, which was purified as described (14).
First, the RF-C complex phosphorylated in vitro with cdc2cyclin B in the presence of [␥-32 P]ATP was immunoprecipitated with either anti-RF-Cp145 or anti-His (RF-Cp40)-specific antibodies. Immunoprecipitation with anti-His (RF-Cp40)-specific antibodies revealed that more phosphorylated RF-Cp145 was present in the supernatant as compared with the supernatant from anti-RF-Cp145 immunoprecipitations (Fig. 5A, left panel). In contrast when non-phosphorylated RF-C complex is immunoprecipitated with anti-His (RF-Cp40) antibodies (Fig. 5A, middle panel) almost all of the RF-Cp145 and RF-Cp40 is in the pellet. These results suggest that phosphorylation by cdc2cyclin B kinase in vitro leads to dissociation of RF-Cp145 from the RF-C complex. We then confirmed whether the phosphorylated RF-Cp145, which remains in the supernatant after immunoprecipitation with anti-His-specific antibodies represents RF-Cp145 not associated with RF-Cp40. For this the 32 P-phosphorylated RF-C complex was immunoprecipitated with anti-His antibodies as in Fig. 5A. The supernatant was then subjected to immunoprecipitation with either normal rabbit serum, or anti-RF-Cp145-, or anti-RF-Cp40-specific antibodies. The majority of the phosphorylated RF-Cp145 came down with immunoprecipitated RF-Cp145 but not with either normal rabbit serum or with immunoprecipitated RF-Cp40 (Fig. 5B).
Finally, the RF-C complex was first immunoprecipitated with anti-His (RF-Cp40) antibodies and the immunoprecipitate was then either mock-phosphorylated or phosphorylated in vitro with cdc2-cyclin B kinase in the presence of cold ATP. The release of RF-Cp145 from the RF-Cp40 immunoprecipitate after phosphorylation was then determined by immunoblotting the immunoprecipitates with anti-RF-Cp40 and anti-RF-Cp145 antibodies. Less RF-Cp145 was found in the pellet after phosphorylation (Fig. 5C) whereas RF-Cp40 remained unchanged. These results together support the conclusion that in vitro phosphorylation of the RF-C complex with cdc2-cyclin B kinase leads to a dissociation of the large subunit from the smaller RF-C subunits like RF-Cp40 of the RF-C complex.
Phosphorylated RF-Cp145 Is Not Associated with RF-Cp37 or RF-Cp40 in Vivo-We next examined if phosphorylated RF-Cp145 is associated with RF-Cp40 and RF-Cp37 during cell cycle progression. Hela cells at different times after release from a double thymidine block were labeled with 32 P-inorganic phosphate, and the 32 P-labeled cell extracts were immunoprecipitated with anti-RF-Cp37-and anti-RF-Cp40-specific antibodies. The associated phosphorylated RF-Cp145 in the immunoprecipitates was determined by autoradiography. Phosphorylated RF-Cp145 was not associated in a complex with either RF-Cp37 or RF-Cp40 (Fig. 6A). In contrast, non-phosphorylated RF-Cp145 was associated with both RF-Cp37 and RF-Cp40 (Fig. 6A). Under conditions when anti-RF-Cp37 and anti RF-Cp145 antibodies immunoprecipitate similar amounts of RF-Cp145, less than 5% of phosphorylated large subunit was associated with RF-Cp37 immunoprecipitates as compared with RF-Cp145 immunoprecipitates (Fig. 6A).
In order to further demonstrate that phosphorylated RF-Cp145 was not associated with smaller RF-C subunits, we resolved 32 P-labeled HeLa cell extracts on sucrose gradients and immunoprecipitated each fraction with anti-RF-Cp145specific antibodies. We reproducibly find that phosphorylated RF-Cp145 is found in fractions that do not contain RF-Cp37 or RF-Cp40 (Fig. 6B). As expected, RF-Cp145 immunoblots revealed that RF-Cp145 is present in two types of complexes in mammalian cells. A non-phosphorylated RF-Cp145 complex, which contains associated RF-Cp37 and a second, phosphosphorylated RF-Cp145 complex that is not associated with smaller subunits of RF-C like RF-Cp37 or RF-Cp40 (Fig. 6B). Taken together, the results in Figs. 5 and 6 suggest that phosphorylated RF-Cp145 is not associated with smaller subunits of the RF-C complex in vivo.

DISCUSSION
In this study we show that the large subunit of RF-C is a phosphoprotein and is preferentially phosphorylated as compared with the smaller subunits of RF-C. RF-Cp145 is phosphorylated in vivo in a cell cycle-dependent manner. Our conclusion that the cdc2-cyclin B kinase phosphorylates RF-Cp145 is based on three sets of observations. First, in vitro cdc2-cyclin B is much better at phosphorylating RF-Cp145 as part of an RF-C complex as compared with cdk2-cyclin A or cdk2-cyclin E. Second, phosphorylation of endogenous RF-Cp145 is maximal in the G 2 /M phase of the cell cycle. Third, roscovitine, a specific inhibitor of cdc2/cdk2 kinases inhibits phosphorylation of endogenous RF-Cp145 during G 2 /M. One important conclusion emerging from our studies is that The supernatant from the first RF-Cp40 immunoprecipitation was then immunoprecipitated with either anti-RF-Cp145 or anti-RF-Cp40 (anti-His)-specific antibodies or preimmune antibodies. The beads were washed with TNNE buffer and the presence of phosphorylated RF-Cp145 in the immunoprecipitate was examined by autoradiography. The majority of the phosphorylated RF-Cp145 came down with anti-RF-Cp145 immunoprecipitates but not with anti-RF-Cp40 immunoprecipitates. C, phosphorylation of immunoprecipitated RF-C complex results in release of RF-Cp145 from the complex. Purified RF-C complex (1 l) in 20 l of TNNE buffer was immunoprecipitated with 2 l of anti-His (RF-Cp40) antibodies preincubated with protein G-Sepharose. After 2 h at 4°C, the beads were washed with TNNE buffer. The RF-C complex on beads was phosphorylated in a standard kinase assay in the presence of cold ATP. After the kinase assay, the beads were washed with TNNE buffer and the remaining RF-Cp145 and RF-Cp40 in the pellet was examined by immunoblotting with anti-RF-Cp145 and anti-His (RF-Cp40)-specific antibodies. Less RF-Cp145 was found in the pellet after phosphorylation whereas the amount of RF-Cp40 remained unchanged. The amount of RF-C complex loaded in the input lane is equivalent to the amount used for immunoprecipitation. phosphorylation of RF-Cp145 leads to its dissociation from the RF-C complex. First, in vitro phosphorylation of RF-C complex leads to dissociation of RF-Cp145 from RF-Cp40. Second, when immobilized RF-C complex is phosphorylated in vitro, less RF-Cp145 remains associated with RF-Cp40 after phosphorylation as compared with mock phosphorylated immunoprecipitates. Third, in vivo phosphorylated RF-Cp145 is not associated with endogenous RF-Cp40 or RF-Cp37.
Replication protein A (RPA) is another example in which subunit interactions is destabilized by phosphorylation. RPA the major eukaryotic single strand-specific DNA-binding protein, consists of three subunits, RPA70, RPA32, and RPA14. The middle subunit, RPA32, is phosphorylated in a cell cycledependent manner. The phosphorylated RPA32 subunit is not associated with RPA70, whereas unmodified RPA32 remains associated with RPA70 (21). Cdk-cyclins regulate the activity of other replication proteins in the cell cycle, by either influencing subcellular localization (22,23) or altering the rates of protein degradation (24,25) or perturbing biological activity (26,27).
We show that RF-Cp145 phosphorylation increases late in S phase and peaks in G 2 /M phase. Further phosphorylated RF-Cp145 is not associated with smaller subunits. These results suggest that phosphorylation of RF-Cp145 by cdc2-cyclin B may be part of the general mechanism which coordinates entry into M phase. The studies of Rao and Johnson (28) established that cells in G 1 but not G 2 are competent to undergo DNA replication when fused to S phase cells. The role of cdk-cyclin activity in preventing replication in G 2 /M is supported by the generation of a conditional human cdc2 mutant cell line (29) and earlier observations in yeast that mitotic cyclin-dependent kinases inhibit replication (30,31). The mutually exclusive presence of cyclin B-cdc2 and MCMs at origins of replication suggests that cyclin B-cdc2 inhibits replication probably by phosphorylating origin-bound proteins (32). Our results suggest that in addition to inactivating proteins in the preinitiation complex cdc2-cyclin B could also down-regulate replication proteins involved in DNA elongation.
We have earlier reported that cdc2 is associated with replicating DNA (33). Consistent with the observations reported here, CDKs have been reported to be associated with the replicative complex, which includes RF-C (34). CDKs associate in a cell cycle-dependent manner with the chromatin-associated complex (34). In this study we extend the scope of the earlier observations by showing that RF-C proteins are targets of CDK activity. What is not clear is if other kinases besides CDKs play additional roles in regulating RF-C activity. In this direction phosphorylation of RF-C by Ca 2ϩ /calmodulin dependent protein kinase II has been reported to inactivate RF-C activity in vitro (35). Our demonstration in this study that RF-C phosphorylation can be inhibited by roscovitine suggests that in vivo, kinases like Ca 2ϩ /calmodulin-dependent protein kinase II target RF-C subsequent to CDK phosphorylation.
In transfection experiments with full-length and RF-Cp145 domains we find that full-length RF-Cp145 The immunoprecipitates were resolved on denaturing polyacrylamide gels, transferred to nitrocellulose followed by autoradiography (top panels). Phosphorylated RF-Cp145 is immunoprecipitated only by anti-RF-Cp145 antibody but not by antibodies specific for RF-Cp37 or RF-Cp40 subunits. The membranes were subsequently used for immunoblotting with anti-RF-Cp145-, anti-RF-Cp37-, and anti-RF-Cp40-specific antibodies. The RF-Cp40 and RF-Cp37 antibodies are as effective as RF-Cp145-specific antibodies in their ability to immunoprecipitate non-phosphorylated RF-Cp145. B, RF-Cp145 is present in at least two distinct complexes. Cell lysates were prepared from asynchronous HeLa cells labeled for 2 h with 32 P-inorganic phosphate. These labeled cell lysates were fractionated on sucrose gradients, and the fractions immunoprecipitated with anti-RF-Cp145-specific antibodies. The immunoprecipitates were resolved on denaturing polyacrylamide gels, transferred to nitrocellulose followed by autoradiography (top panel). The membranes were subsequently used for immunoblotting with anti-RF-Cp145-and anti-RF-Cp37-specific antibodies. Phosphorylated RF-Cp145 is not associated with RF-Cp37 whereas non-phosphorylated RF-Cp145 is associated with RF-Cp37. tides present in full-length RF-Cp145 are the sum of the phosphopeptides in domain N and domain B (data not shown). The C-terminal end (729 -1148 region) of RF-Cp145, which is both required and sufficient to form the RF-C complex in vitro (36) and in vivo (37) is unlikely to be phosphorylated in vivo.
Structural studies with the E. coli clamp loader, ␥ complex: ␦Ј-␥1-␥2-␥3-␦ (17) demonstrate that the C-terminal domains of ␦Ј (hRF-Cp38 homolog), ␥, and ␦ (hRF-Cp145 homolog) form a helical scaffold (circular collar) and the N-terminal ends appear to dangle under the C-terminal pentamer umbrella. The presence of an extensive interface suggests that C-terminal regions of ␦Ј and ␥ are tightly linked and not easily perturbed. The C-terminal domain of ␦ probably swivel with respect to the rest of the helical scaffold as the ␥ complex undergoes conformational changes based on structural and modeling studies (38). Based on in vitro reconstitution studies with recombinant human RF-C complex, RF-Cp145 has been shown to have subunit contacts with RF-Cp40 and RF-Cp38 (14,15,39). The structural studies with the E. coli ␥ complex reveals that the hRF-Cp145 equivalent ␦ subunit contacts ␦Ј and ␥3 subunits (17,40). This demonstrates a remarkable conservation of intersubunit interactions from the ␥ complex to the human RF-C complex.
Does RF-Cp145 phosphorylation induce its dissociation from the RF-C 2-5 complex? Our in vitro experiments in Fig. 5C suggest that when the RF-C complex is immobilized, phosphorylation with cdc2-cyclin B leads to dissociation of the RF-Cp145 subunit from the rest of the RF-C complex. This result would be consistent with the conclusion from the ␥ complex structure that the large subunit is less tightly bound as compared with the smaller subunits and has the capacity to undergo major structural transitions. The subunit interactions of recombinant RF-C complex are consistent with this conclusion. RF-Cp145 is recruited to the RF-C complex after the smaller subunits p40/p38/p37/p36 (RF-C 2-5 ) form a core complex (14,15). It is therefore likely that RF-Cp145 dissociates from the RF-C 2-5 core, which remains intact.
Using different approaches we clearly demonstrate that in vivo phosphorylated RF-Cp145 is not associated with RF-Cp37 or RF-Cp40. The smaller RF-C 2-5 subunits associate with proteins, which substitute for the RF-Cp145 subunit to form three alternative RF-C complexes. First, Rad17 (Rad24) forms a complex with RF-C 2-5 , which plays a critical role in checkpoint signaling (41)(42)(43). Second, RF-C 2-5 associates with Cft18 to play a critical role in sister chromatin cohesin (44). Third, the Elg1-RF-C complex has been suggested to be required for maintaining genomic integrity (45). Our results raise the possibility that RF-Cp145 phosphorylation may play a critical regulatory role in destabilizing the replicative RF-C complex, which would thus allow the checkpoint or the cohesin RF-C complex to be formed with Rad17 or Cft18, respectively. In addition our results leave open the possibility that in mammalian cells the phosphorylated RF-Cp145 subunit associates with other unknown proteins after it dissociates from the pentameric RF-C complex. The size of the phosphorylated RF-Cp145 complex is consistent with such a possibility. Answering these questions is a challenge for future studies.