A Conserved Physical and Functional Interaction between the Cell Cycle Checkpoint Clamp Loader and DNA Ligase I of Eukaryotes*

DNA ligase I joins Okazaki fragments during DNA replication and completes certain excision repair pathways. The participation of DNA ligase I in these transactions is directed by physical and functional interactions with proliferating cell nuclear antigen, a DNA sliding clamp, and, replication factor C (RFC), the clamp loader. Here we show that DNA ligase I also interacts with the hRad17 subunit of the hRad17-RFC cell cycle checkpoint clamp loader, and with each of the subunits of its DNA sliding clamp, the heterotrimeric hRad9-hRad1-hHus1 complex. In contrast to the inhibitory effect of RFC, hRad17-RFC stimulates joining by DNA ligase I. Similar results were obtained with the homologous Saccharomyces cerevisiae proteins indicating that the interaction between the replicative DNA ligase and checkpoint clamp is conserved in eukaryotes. Notably, we show that hRad17 preferentially interacts with and specifically stimulates dephosphorylated DNA ligase I. Moreover, there is an increased association between DNA ligase I and hRad17 in S phase following DNA damage and replication blockage that occurs concomitantly with DNA damage-induced dephosphorylation of chromatin-associated DNA ligase I. Thus, our results suggest that the in vivo interaction between DNA ligase I and the checkpoint clamp loader is regulated by post-translational modification of DNA ligase I.

nodeficiency (3). At the cellular level, there was a marked defect in Okazaki fragment joining and hypersensitivity to DNA damaging agents, in particular the DNA alkylating agent, methyl methanesulfate (4 -7). More recently, the mutation in the human LIG1 gene responsible for the cellular phenotype has been reiterated in the mouse (8). Notably, this mouse model of DNA ligase I deficiency exhibits genomic instability and an increased frequency of cancer.
Genetic, biochemical, and cell biology studies have together provided compelling evidence that human DNA ligase I is a key participant in DNA replication and DNA repair (2,9,10). An interaction between DNA ligase I and proliferating cell nuclear antigen (PCNA) 4 is required for the targeting of DNA ligase I to DNA replication foci and also for the efficient joining of Okazaki fragments and completion of long patch base excision repair (11)(12)(13). In these DNA transactions, the homotrimeric PCNA ring acts as a DNA sliding clamp that interacts with and coordinates the sequential action of DNA processing enzymes, ending with ligation by DNA ligase I (14 -16). The PCNA ring is loaded onto DNA by replication factor C (RFC), a heteropentameric complex that can also unload PCNA (17,18). RFC interacts with and inhibits DNA ligase I (19) but this inhibition is alleviated by PCNA in a reaction that is dependent upon the physical interaction between PCNA and DNA ligase I (19). Recently, we have shown that the physical interactions and functional interplay also occur among the homologous proteins of Saccharomyces cerevisiae (20). Thus, it appears that RFC plays a critical role in the ligation step that joins adjacent Okazaki fragments and completes long patch base excision repair.
PCNA and RFC are the prototypic members of an emerging family of DNA sliding clamps and clamp loaders that are involved in cell cycle checkpoints, sister chromatid cohesion, and genome stability in addition to DNA replication (21). In the cell cycle checkpoint response activated by either replication blockage or DNA damage, the DNA sliding clamp is a heterotrimer composed of hRad9, hRad1, and hHus1 proteins (21). The hRad9-hRad1-hHus1 clamp is loaded onto DNA by the heteropentameric hRad17-RFC clamp loader (22,23). Both the checkpoint and replicative clamp loader complexes contain the same four small RFC subunits, p36, p37, p38, and p40, but are distinguished by their large subunits, hRad17 and RFCp140, respectively (21). Following replication block, hRad17-RFC interacts with RPA-coated single strand regions and loads the hRad9-hRad1-hHus1 clamp onto the DNA (24,25). In the case of DNA damage, the checkpoint clamp complex is recruited to DNA lesions to activate the DNA damage checkpoint. Recent studies in S. cerevisiae indicate that the checkpoint clamp directly activates the checkpoint kinase Mec1, the ortholog of human Atr (26).
However, there is emerging evidence that Rad9-Rad1-Hus1 may also be recruited to the site of a DNA lesion by interactions with the DNA damage recognition proteins for base excision (27,28) and nucleotide excision repair (29). Because both PCNA and hRad9-hRad1-hHus1 interact with DNA pol ␤, FEN-1, and DNA ligase I (30 -34), enzymes involved in the gap-filling, processing, and ligation steps that complete DNA repair, it has been suggested that hRad9-hRad1-hHus1 may be able to substitute for PCNA in long patch base excision repair (35).
Whereas DNA ligase I interacts with both PCNA and hRad9-hRad1-hHus1, there appear to be clear differences in the mechanism and functional consequences of these interactions. In this study, we have confirmed that DNA ligase I interacts with and is stimulated by hRad9-hRad1-hHus1. More importantly, we have identified a novel physical and functional interaction between DNA ligase I and hRad17-RFC that also occurs between the homologous S. cerevisiae proteins, Cdc9 and Rad24-RFC. Unlike the replicative clamp loader, which inhibits DNA joining (19,20), the checkpoint clamp loader stimulates DNA joining. Interestingly, the physical and functional interaction between DNA ligase I and hRad17-RFC is modulated by post-translational modification of DNA ligase I.

EXPERIMENTAL PROCEDURES
Proteins-Recombinant wild type human DNA ligase I was overexpressed in and purified from Escherichia coli (36). A mutant version of human DNA ligase I, in which the adjacent Phe residues in the conserved PCNA binding motif (PIP box) are replaced by Ala residues (12,13), was constructed by sitedirected mutagenesis. The mutant protein was expressed in E. coli and purified as described (36). Fragments of DNA ligase I, N-terminal (residues 1-232), DNA-binding domain (DBD, residues 233-534), and catalytic domain (residues 533-919) were expressed as His 6 -tagged proteins in E. coli and purified by affinity chromatography followed by gel filtration and ion exchange chromatography (37). Recombinant wild type human DNA ligase I was also overexpressed in and purified from insect Sf9 cells (Invitrogen) infected with a baculovirus expressing human DNA ligase I (38). Briefly, a suspension culture of Sf9 cells was infected with the DNA ligase I baculovirus at a multiplicity of infection of 0.1 and then harvested 72 h post-infection. After lysis, the cleared lysate was fractionated by ammonium sulfate precipitation (40 -65%) and then further purified by phosphocellulose, Source Q, Superdex 200, Blue Sepharose, and Mono Q chromatography (36). DNA ligase I purified from E. coli is about 1.5-fold more active than the enzyme purified from insect cells.
Recombinant hRad9-hRad1-hHus1 and hRad17-RFC complexes were purified from baculovirus-infected insect cell as described. The complexes were estimated to be Ͼ95% homogeneous by Coomassie Blue staining after separation by SDS-PAGE (39).
Antibodies-Antibodies against hRad9, hRad17, hRFC p140, and hPCNA, were purchased from Santa Cruz Biotechnology. The Mek2 antibody was from Cell Signaling Technology, the Orc2 antibody from Abcam, the FLAG antibody from Sigma, and the DNA ligase I monoclonal antibody from GeneTex. Antibody against hRFC p140 was a gift from Dr. Bruce Stillman. Polyclonal antibodies against hRad17, hRad1, hRFC p37, and DNA ligase I have been described previously (11,39,41).
DNA Ligase I Affinity Chromatography-Nuclear and cytoplasmic extracts were prepared from a frozen pellet of HeLa S3 cells (10 9 cells) (11) and the nuclear extract (10 mg) was fractionated by DNA ligase I affinity chromatography as described previously (11,19). Fractions were analyzed for protein by immunoblotting after separation by SDS-PAGE (42).
Purified hRad17-RFC (1 g) or hRad9-hRad1-hHus1 (1 g) was incubated with 20 l of DNA ligase I and bovine serum albumin (BSA) beads (11) in 250 l of buffer A (50 mM Tris-HCl, pH 7.5, 100 mM NaCl, 1 mM dithiothreitol, 0.1% Nonidet P-40, and 5 g of BSA) for 1 h at 4°C with constant agitation. After collection by centrifugation, the beads were washed extensively with buffer A, and bound proteins were eluted with SDS sample buffer (42). After separation by SDS-PAGE, proteins were detected by immunoblotting.
Pull-down Assays-The plasmids pGEX-hRad9, pGEX-hHus1, pGEX-hRad1, pGEX-hRad17, and pGEX-hRad17N (1-320) were gifts from Dr. Eva Lee. After expression in E. coli, GST fusion proteins were purified from cell extracts by affinity chromatography using glutathione-Sepharose beads. A GST-PCNA fusion protein was expressed and purified as described previously (11). To prepare beads for pull-down assays, GST and GST fusion proteins (5 g of each) were incubated with a 20-l slurry of glutathione-Sepharose beads (Amersham Biosciences) equilibrated in buffer A for 1 h at 4°C with constant agitation. After washing with buffer A, the beads were resuspended in 250 l of buffer A containing purified full-length DNA ligase I or various truncated derivatives. Incubation was at 4°C for 1 h with constant agitation. In certain assays, phosphorylated DNA ligase I from insect cells was preincubated for 30 min at 30°C with -phosphatase (New England Biolabs) according to the manufacturer's instructions. The glutathione beads were collected by centrifugation and washed extensively in buffer A prior to resuspension in 20 l of SDS-PAGE sample buffer. After separation by SDS-PAGE, proteins were detected by immunoblotting with the DNA ligase I antibody.
Full-length DNA ligase I and truncated fragments were synthesized and labeled by coupled in vitro transcription and translation reactions using the TNT Quick Coupled Transcription/Translation system (Promega, Madison, WI) as described previously (19). After partial purification by ammonium sulfate precipitation, labeled polypeptides were incubated with the glutathione beads at 4°C for 1 h with constant agitation. Next, the beads were collected by centrifugation and washed extensively in buffer A prior to resuspension in 20 l of SDS-PAGE sample buffer. After separation by SDS-PAGE, labeled polypeptides were visualized by PhosphorImager analysis (Amersham Biosciences).
Purified His-tagged Cdc9 (5 g) was bound to nickel-agarose beads (20 l) (Qiagen Inc., Valencia, CA) equilibrated in buffer B (50 mM potassium phosphate, pH 7.5, 100 mM NaCl, 1 mM 2-mercaptoethanol, 0.1% Nonidet P-40, 20 mM imidazole, 10% glycerol, and 1 mg/ml BSA) for 30 min at 4°C with constant agitation and then resuspended in 250 l of buffer B. Purified Rad24-RFC (1 g) was incubated with either the nickel-agarose beads alone or the nickel-agarose beads liganded by His-tagged Cdc9 at 4°C for 1 h with constant agitation. Beads were collected by centrifugation, washed extensively with buffer B, and then resuspended in 20 l of SDS-PAGE sample buffer. After separation by SDS-PAGE, proteins were detected by silver staining (Bio-Rad).
DNA Joining Reaction with Biotin-labeled Linear Substrate-A 5Ј-biotinylated 90-mer oligonucleotide was annealed with two complementary 15-mers, one of which was 5Ј-end labeled with 150 Ci of [␥-32 P]ATP and T4 polynucleotide kinase (New England Biolabs), to generate a partial duplex of 30 bp containing a single ligatable nick in the middle flanked by singlestranded regions of 30 nucleotides (19). Streptavidin-agarose beads (10 l; Pierce) were incubated with 1 pmol of the biotinylated linear DNA substrate in phosphate-buffered saline for 30 min at room temperature. After washing three times in ligation buffer (50 mM Tris-HCl, pH 7.5, 10 mM MgCl 2 , 1 mM dithiothreitol, 0.25 mg/ml BSA, 100 M ATP, and 100 mM NaCl), the beads were incubated with 1 pmol of E. coli single-stranded binding protein/pmol of DNA in the same buffer for 15 min at room temperature. This substrate was then incubated with 50, 150, and 500 fmol of hRad17-RFC and hRad9-hRad1-hHus1 both individually and together at 30°C for 2 min. DNA ligase I (50 fmol) was added, and the reaction was incubated at room temperature for 5 min. The beads were then spun down, and the reaction was terminated by the addition of 10 l of stop mixture (50% glycerol, 1% SDS, 20 mM EDTA, and 0.05% bromphenol blue). Similar reactions were carried out with yeast Rad24-RFC, yRFC, and Cdc9 DNA ligase. After heating at 100°C for 3 min to denature DNA, a 2.5-l aliquot was mixed with 2.5 l of denaturing PAGE dye (80% formamide, 0.05% bromphenol, and 0.05% xylene cyanol) and then separated by electrophoresis through a 12% denaturing polyacrylamide gel. The gel was dried, exposed to a Storage Phosphor screen and subjected to PhosphorImager analysis (Amersham Biosciences).
Treatment of HeLa Cells and 46BR.1G1 Cells-To synchronize HeLa cells in the G 1 phase of the cell cycle, an asynchronous population was incubated with 500 M mimosine for 24 h (43). HeLa cells and derivatives of 46BR.1G1, one of which stably expresses FLAG-tagged wild type DNA ligase I (12), were synchronized in S phase by double thymidine arrest (43). An asynchronous population was incubated with 2 mM thymidine for 18 h and then released for 8 h. After incubation with 2 mM thymidine for another 18 h, the cells were released for 4 h. Cell cycle distributions were determined by fluorescence-activated cell sorter analysis in the Flow Cytometry Core of the University of Maryland Marlene and Stewart Greenebaum Cancer Center. To induce replication blockage, HeLa cells that had been synchronized in S phase were incubated for 1 h with medium containing 10 mM hydroxyurea. Where indicated, cell populations were irradiated with 12 gray of ionizing radiation using a ␥-ray irradiator (Mark I, model 68A, JL Shepherd and Associates).
Subcellular Fractionation-Chromatin fractionation was performed as described previously (44). Briefly, equal numbers of synchronized or treated cells were washed twice in ice-cold phosphate-buffered saline followed by incubation in buffer C (10 mM HEPES-KOH, pH 7.9, 10 mM KCl, 1.5 mM MgCl 2 , 0.34 M sucrose, 10% glycerol, 0.5% Triton X-100) supplemented with protease inhibitor mixture (Sigma) and phosphatase inhibitors mixtures I and II (Sigma) for 5 min. After centrifugation at 1,300 ϫ g for 4 min, the supernatant/cytoplasmic fraction (Cyto) was removed. The pellet containing nuclei was washed in buffer C and then lysed in buffer D (3 mM EDTA, 0.2 mM EGTA, and protease and phosphatase inhibitors as above). Soluble nuclear proteins (Nuc Sol) were separated from the chromatin fraction (Chr) by centrifugation at 1,700 ϫ g for 4 min. The chromatin-enriched pellet was washed in buffer D, resuspended in SDS sample buffer, and then sonicated three times for 15 s.
Immunoprecipitation-HeLa cells and 46BR.1G1 cells (10 7 ) were lysed in IP buffer (50 mM Tris-HCl, pH 7.5, 150 mM NaCl, 10 mM MgCl 2 , 0.1% Triton X-100, 1 mM dithiothreitol, 50 g/ml ethidium bromide, 1 g/ml leupeptin, pepstatin, and chymostatin, 0.1 mM phenylmethanesulfonyl fluoride, 50 mM sodium fluoride, 1 mM sodium vanadate). Approximately 3.5 mg of the clarified extract was used for each immunoprecipitation. Extracts were precleared by incubation for 1 h at 4°C with 50 l of Protein A-Sepharose and Protein G-Sepharose beads (Amersham Biosciences) equilibrated with IP buffer, prior to the addition of antibodies against DNA ligase I and hRad17. After incubation at 4°C overnight, 50 l of Protein A-Sepharose and Protein G-Sepharose beads were added, and the incubation continued for 1 h. The beads were collected by centrifugation, washed three times with IP buffer lacking ethidium bromide, and then resuspended in SDS sample buffer. After separation by SDS-PAGE, proteins were detected by immunoblotting.
Immunofluorescence and Microscopy-HeLa cells were cultured on cover glasses, fixed in 4% formaldehyde, and permeabilized in 0.25% Triton X-100 before immunostaining with antibodies as recommended by the antibody manufacturer (Abcam). Images were captured using a Nikon E400 fluorescence microscope.

RESULTS
Physical Interactions between DNA Ligase I and Both the hRad17-RFC Clamp Loader and hRad9-hRad1-hHus1 DNA Sliding Clamp-Previously we fractionated a HeLa nuclear extract by DNA ligase I affinity chromatography to provide initial evidence for the association of DNA ligase I with the replicative DNA sliding clamp PCNA and its cognate loader RFC (11,19). Using the same strategy, we observed the specific retention of the hRad17 subunit of the hRad17-RFC cell cycle checkpoint clamp loader complex and the hRad9 subunit of the cell cycle checkpoint DNA sliding clamp, hRad9-hRad1-hHus1 (Fig. 1A). We estimate that at least 50% of the loaded hRad17 and hRad9 was retained by the DNA ligase I affinity beads and eluted in the 150 and 300 mM NaCl eluates. Moreover, these proteins co-eluted with RFCp140, the unique large subunit of RFC (Fig. 1A). To demonstrate that DNA ligase I interacts directly with the hRad17-RFC and hRad9-hRad1-hHus1 complexes, we performed pull-down assays with DNA ligase I beads and recombinant hRad17-RFC and hRad9-hRad1-hHus1 complexes purified from insect cells (39). As shown in Fig. 1B, the hRad17 and p37 subunits were specifically retained by the DNA ligase I beads. Similarly, the hRad9 and hRad1 subunits of the purified hRad9-hRad1-hHus1 complex also bound specifically to the DNA ligase I beads (Fig. 1C).
Mapping of the Interaction between DNA Ligase I and the hRad9-hRad1-hHus1 DNA Sliding Clamp-Unlike homotrimeric PCNA, the presence of three different subunits in the heterotrimeric hRad9-hRad1-hHus1 complex raises the possibility that each subunit interacts specifically with different protein partners. In support of this idea, it has been shown recently that a DNA polymerase, nuclease, and DNA ligase interact specifically with different subunits of the heterotrimeric Sulfolobus solfatricus DNA sliding clamp (45). However, DNA ligase I does not exhibit an obvious specificity for any one of the subunits of the hRad9-hRad1-hHus1 complex, interacting with each of the subunits to a similar extent ( Fig. 2A).
Although there are additional albeit much weaker interactions with the DBD of DNA ligase I (46), the N-terminal PIP box of DNA ligase I plays a major role in complex formation with homotrimeric PCNA (Fig. 2B, lower panel) as expected (12). The interaction with the subunits of the hRad9-hRad1-hHus1 complex also involves both the non-catalytic N-terminal region and the DBD domain of DNA ligase I (data not shown) but, in this case, amino acid substitutions that disrupt the DNA ligase I PIP box have no significant effect on the interaction with the hRad9-hRad1-hHus1 complex (Fig. 2B, upper panel).
Mapping of the Interaction between DNA Ligase I and the hRad17-RFC Clamp Loader-Previously we have shown that the interaction between DNA ligase I and RFC involves the Nand C-terminal regions of DNA ligase I and the non-catalytic N-terminal region of RFC p140 and the p36 and p38 subunits of RFC (19). Because hRad17 replaces p140 in the hRad17-RFC complex, we examined whether hRad17 interacts with DNA ligase I. As shown in Fig. 3A, DNA ligase I bound specifically to glutathione beads liganded by either full-length hRad17 fused to GST or an N-terminal fragment of hRad17 (residues 1-320) fused to GST.
Next we mapped the interacting regions of DNA ligase I using fragments synthesized and labeled by in vitro transcription and translation. As noted above, an N-terminal fragment (residues 1-118) and a C-terminal fragment (residues 479 -919) of DNA ligase I interacted with the N-terminal region of RFC p140. By contrast, only an internal fragment of DNA ligase I (residues 109 -479) containing most of the DBD of DNA ligase I (37) was specifically retained by GST beads liganded by full-length hRad17 fused to GST (Fig. 3B). The role of the DBD in the interaction with hRad17 was confirmed in similar experiments with purified fragments of DNA ligase I (Fig. 3C). Thus, DNA ligase I interacts with the N-terminal domains of the large subunits of the hRad17-RFC and RFC clamp loaders but there are differences in the regions of DNA ligase I that are involved in these interactions.
Effect of DNA Ligase I Phosphorylation on the Interaction with hRad17-RFC-Although DNA ligase I becomes increasingly phosphorylated during cell cycle progression (43), DNA ligase I phosphorylation status does not significantly affect its interac-  tion with PCNA (43). Similarly, we did not observe any significant difference in the binding of hRad9-hRad1-hHus1 with phosphorylated DNA ligase I purified from insect cells (47) compared with non-phosphorylated DNA ligase I purified from E. coli (data not shown). In contrast to the studies with the DNA sliding clamps, RFCp140, the unique large subunit of the replicative clamp loader complex, preferentially binds to a specific phosphorylated species of DNA ligase I. 5 Notably, the binding of hRad17 is also influenced by DNA ligase I phosphorylation, but, in this case, hRad17 preferentially interacts with unmodified DNA ligase I purified from E. coli (Fig. 4A). To demonstrate that phosphorylation reduces the binding of DNA ligase I to hRad17, we incubated phosphorylated DNA ligase I purified from insect cells with -phosphatase. As expected, the phosphatase treatment significantly increased the amount of DNA ligase I retained by GST beads liganded by GST fusions of full-length hRad17 and the N-terminal region of hRad17 (Fig. 4B).
hRad17-RFC Stimulates DNA Joining by Non-phosphorylated DNA Ligase I-Previously we have shown that RFC inhibits DNA joining by human DNA ligase I and yeast Cdc9 DNA ligase, whereas PCNA has no effect (19,20). Notably, the inhibition by RFC was alleviated by co-incubation with PCNA in a reaction that was dependent upon the physical interaction between the DNA ligase and PCNA (19,20). Therefore, we examined the functional interplay between human DNA ligase I and the checkpoint clamp loader and clamp, hRad17-RFC and hRad-hRad1-hHus1, respectively, using a linear partial duplex with a central nick (Fig. 5A). In accord with published reports (31,34), hRad9-hRad1-hHus1 enhanced DNA joining by unmodified DNA ligase I (Fig. 5B,  lanes 5-7). Under these reaction conditions, DNA joining was increased about 5-fold at a 10:1 ratio of hRad9-hRad1-hHus1 to DNA ligase I (Fig. 5C). In similar assays, hRad17-RFC also stimulated DNA joining by unmodified DNA ligase I (Fig. 5B, lanes 2-4). With a 10:1 ratio of hRad17-RFC to DNA ligase I, there was about a 5-fold increase in the number of ligation events (Fig. 5C). Although individually hRad17-RFC and hRad9-hRad1-hHus1 have a similar stimulatory effect on DNA joining, co-incubation of these complexes with DNA ligase I did not significantly increase the amount of ligation over that observed with either factor alone (Fig. 5C). Because hRad17 preferentially interacts with unmodified DNA ligase I (Fig. 4), we compared the effect of hRad17-RFC on DNA joining by unmodified and phosphorylated DNA ligase I. Strikingly, in contrast to unmodified DNA ligase I, phosphorylated DNA ligase I was not stimulated by hRad17-RFC (Fig. 5D).
Yeast Rad24-RFC Stimulates DNA Joining by Cdc9 DNA Ligase-Because the physical and functional interactions among the replicative DNA ligase and the replicative clamp loader and clamp are conserved in eukaryotes (19,20), we asked whether this is also the case for the checkpoint clamp loader and clamp. Rad24-RFC, the yeast homolog of hRad17-RFC, was  specifically retained by the Cdc9 beads (Fig. 6A). However, under similar conditions, we did not detect a specific interaction between Cdc9 and Rad17-Mec3-Ddc1, the yeast homolog of hRad-hRad1-hHus1 (data not shown). Next we compared the effect of the yeast clamp loader on Cdc9 DNA ligase. Notably, Rad24-RFC enhanced DNA joining by Cdc9 ligase (Fig. 6B,  compare lanes 1 and 2), whereas, as expected (20), yRFC inhibited DNA joining (Fig. 6B, compare lanes 1 and 3). Thus, we conclude that the physical and functional interaction between the replicative DNA ligase and cell cycle checkpoint clamp loader is conserved in eukaryotes.
Chromatin-associated DNA Ligase I Is Dephosphorylated in Response to Ionizing Radiation-Although DNA ligase I becomes increasingly phosphorylated during cell cycle progression (43), treatment of cells with etoposide, a specific inhibitor of topoisomerase II that causes DNA double strand breaks, induces dephosphorylation of DNA ligase I (49,50). Because the hRad17 subunit of the checkpoint clamp loader complex preferentially interacts with unmodified DNA ligase I, we asked whether ionizing radiation, another agent that causes DNA double strand breaks, induced dephosphorylation of DNA ligase I in chromatin-associated and soluble fractions isolated from cell populations enriched for the G 1 and S phases of the cell cycle. An increased amount of phosphorylated DNA ligase I was observed in S phase compared with G 1 but this difference was only evident in the chromatin fraction (Fig.  7, compare left and right panels). In both G 1 and S phase cells, ionizing radiation induced dephosphorylation of DNA ligase I, resulting in about a 2-fold increase in the ratio of dephosphorylated versus phosphorylated DNA ligase I. Once again this effect was only observed with chromatin-associated DNA ligase I. Treatment of S phase cells with hydroxyurea, which activates the hRad17-dependent checkpoint by blocking DNA replication also caused DNA ligase I dephosphorylation of chromatin bound-DNA ligase I albeit to a lesser extent than IR (data not shown).
Because the amount of DNA ligase I associated with chromatin in the G 1 and S phases of the cell cycle and following DNA damage did not change significantly, it is likely that it is the chromatin-bound pool of DNA ligase I that is being phosphorylated during S phase and dephosphorylated in response to DNA damage and replication blockage. In etoposide-treated cells, the dephosphorylation of DNA ligase I occurred during the execution phase of apoptosis (49,50). By contrast, the dephosphorylation of DNA ligase I observed 1 h after ionizing radiation is not a consequence of apoptosis as there is no significant increase in the percentage of apoptotic cells at this time and only a small increase 24 h after radiation (data not shown). Thus, it appears that DNA ligase I dephosphorylation in response to ionizing radiation reflects an attempt by the cell to survive DNA damage rather than activation of apoptosis.
Association between hRad17 and DNA Ligase I Is Enhanced by DNA Damage: Co-localization of hRad17 and DNA Ligase I in Response to DNA Damage-Our in vitro binding studies show that hRad17 preferentially interacts with non-phosphorylated DNA ligase I. Because this form of DNA ligase I is most abundant in the G 1 phase of the cell cycle, our results predict that hRad17 and DNA ligase I will associate in extracts of G 1 phase cells. In accord with this idea, DNA ligase I and hRad17 were co-immunoprecipitated from extracts of HeLa cells enriched for the G 1 phase of the cell cycle (Fig. 8A). Treatment of G 1 phase HeLa cells with ionizing radiation did not significantly increase the amount of hRad17 co-immunoprecipitated by the DNA ligase I antibody (Fig. 8A). In similar experiments with an S phase-enriched cell population, hRad17 was co-immunoprecipitated by the DNA ligase I antibody (Fig. 8B). FIGURE 5. Effect of hRad17-RFC and hRad9-hRad1-hHus1 on DNA joining by unmodified and phosphorylated DNA ligase I. A, diagram of the labeled nicked linear DNA substrate. Two 15-mer oligonucleotides were annealed to a biotin-labeled 90-mer as described under "Experimental Procedures" to generate a partial duplex with a single ligatable nick. The asterisk indicates the labeled phosphate group. Unmodified DNA ligase I (50 fmol) purified from E. coli (panel B) was incubated with the DNA substrate (1 pmol) and the indicated combinations of hRad17-RFC (Rad17RFC: 50, 150, and 500 fmol) and hRad9-hRad1-hHus1 (911: 50, 150, and 500 fmol). After separation by denaturing gel electrophoresis, the labeled substrate (15-mer) and product (30-mer) were detected by phosphorimaging analysis. The results of three independent experiments are shown graphically in C. DNA joining is expressed as -fold stimulation compared with joining by DNA ligase I alone. hRad17-RFC alone (gray bars), hRad9-hRad1-hHus1 alone (black bars), and co-incubation of hRad17-RFC and hRad9-hRad1-hHus1 (white bars) are shown. D, unmodified DNA ligase I (LigI, 50 fmol) and phosphorylated DNA ligase I (P-LigI, 50 fmol) were incubated with the DNA substrate (1 pmol) and hRad17-RFC (Rad17RFC, 500 fmol). The results of three independent experiments are shown graphically. DNA joining is expressed as -fold stimulation compared with joining by DNA ligase I alone.
Treatment of the S phase cells with either ionizing radiation or hydroxyurea did result in a small but reproducible increase in the amount of co-immunoprecipitated hRad17 (1.4-and 1.8fold, respectively). In reciprocal co-immunoprecipitation experiments, the co-immunoprecipitation of DNA ligase I by hRad17 antibody occurred with extracts from S phase cells pretreated with either ionizing radiation or hydroxyurea but was barely detectable in similar experiments with extracts from untreated S phase cells (Fig. 8B). To confirm that DNA ligase I and hRad17 associate in cell extracts, we used a derivative of the DNA ligase I-deficient human cell line 46BR.1G1 (12) that stably expresses a FLAG-tagged version of wild type DNA ligase I at about the same level as endogenous DNA ligase I in GM00847 cells, an SV40-immortalized fibroblast cell line established from a normal individual (Fig. 8C, left panel). FLAG-tagged DNA ligase I was specifically co-immunoprecipitated from extracts of S phase-enriched cell populations by antibody against hRad17 and, once again, DNA damage treatment resulted in about a 1.7-fold increase in association between hRad17 and DNA ligase I (Fig. 8C, middle and right  panels).
To provide further evidence for the DNA damage-dependent association of DNA ligase I and hRad17, we examined the subcellular localization of these proteins by immunofluorescence. In undamaged cells, both hRad17 and DNA ligase I are distributed diffusely throughout the nucleus (Fig. 8D). Following DNA damage, both hRad17 and DNA ligase I redistribute to form distinct foci. Moreover, the patterns of hRad17 and DNA ligase I foci exhibit substantial overlap, indicating that a significant fraction of these proteins co-localize following DNA damage.

DISCUSSION
During DNA replication, the generation of an intact lagging strand requires the action of a DNA ligase to join adjacent Okazaki fragments. The specific participation of the replicative DNA ligase in eukaryotic DNA replication is determined by conserved interactions with PCNA (11,13) and RFC (19,20), the replicative DNA sliding clamp and clamp loader, respectively. These complexes are the prototypes of an emerging family of clamp loaders and clamps involved in different aspects of DNA metabolism (21). In accord with recent studies (31,34), we have detected a physical and functional interaction between DNA ligase I and the hRad9-hRad1-hHus1 heterotrimeric DNA sliding clamp that functions in cell cycle checkpoint pathways activated by DNA damage and replication block (21). More importantly, we have identified and characterized a novel conserved physical and functional interaction between the replicative DNA ligase and checkpoint clamp loader of eukaryotes.
Although human DNA ligase I interacts with the DNA sliding clamps involved in DNA replication (11-13, 51, 52), and cell cycle checkpoints (31,34), there are differences both in the interaction interfaces and the functional consequences of the interactions. Specifically, amino acid changes that disrupt the DNA ligase I PIP box, the primary binding site for PCNA (12), have no effect on the interaction with hRad9-hRad1-hHus1. In addition, we and others have found that hRad9-hRad1-hHus1 increased DNA joining by DNA ligase I (31,34), whereas PCNA did not (11,19,51). It should be noted that Bambara and colleagues (34,52) have reported that, like hRad9-hRad1-hHus1, PCNA also stimulates DNA ligase I. A likely explanation for the discrepancy between the studies on PCNA and DNA ligase I is that the modest stimulation of DNA ligase I activity was only observed when PCNA was present in at least 100-fold molar excess (52). Interestingly, the stimulation of DNA ligase I by PCNA was dependent upon the PCNA ring encircling the nicked DNA, whereas the stimulatory effect of the hRad9-hRad1-hHus1 ring was not (34,52).
Recently, we demonstrated that human DNA ligase I and yeast Cdc9 DNA ligase interact with and are inhibited by RFC, the clamp loader for PCNA (19,20). Notably, the inhibitory effect of RFC is alleviated by PCNA in a reaction that is dependent upon the physical interaction between PCNA and the DNA ligase (19,20), suggesting that there are functional interactions among DNA ligase I, PCNA, and RFC. In this study we have shown that there is a specific physical interaction between the replicative DNA ligase and checkpoint clamp loader and that, unlike RFC, the checkpoint clamp loader does not inhibit but instead stimulates DNA ligase activity.
Human DNA ligase I is phosphorylated in a cell cycle-dependent manner by casein kinase II and cyclin-dependent kinase, Cdk2 (43,53). Here we show that it is the chromatinassociated pool of DNA ligase I that is phosphorylated in S phase. This is consistent with the co-localization of DNA ligase I, Cdk2, and cyclin A within replication foci (54) and suggests that only the DNA ligase I molecules actively engaged at the replication fork are phosphorylated. In support of this idea, we have found that RFC specifically associates with a specific phosphorylated species of DNA ligase I present in S phase cells. 5 Previous studies have shown that treatment of cells with the topoisomerase II inhibitor etoposide results in DNA ligase I dephosphorylation and its release from chromatin and replication foci (49,55). Here we show that treatment with ionizing radiation also induces dephosphorylation of DNA ligase I but the dephosphorylated protein remains associated with chromatin. This suggests that the cellular response to replication-dependent DNA double strand breaks caused by etoposide is different from the response to DNA damage generated by ionizing radiation. Indeed, etoposide-induced dephosphorylation of DNA ligase I occurs during the execution of apoptosis (50), whereas the dephosphorylation induced by ionizing radiation was evident prior to the onset of apoptosis. Interestingly, the dephosphorylation of DNA ligase I correlates with the enhanced association of DNA ligase I with hRad17 following DNA damage and replication blockage. Taken together with the in vitro binding studies showing that hRad17 preferentially interacts with unmodified DNA ligase I, our studies suggest that, hRad17, which is chromatin associated throughout the cell cycle (44), forms a specific chromatin-bound complex with dephosphorylated DNA ligase I that is a component of the cellular response to survive after DNA damage.
Following replication blockage, hRad17-RFC is recruited to the resultant single strand gaps coated with RPA and loads the hRad9-hRad1-hHus1 ring onto the DNA (24,25,56,57). Studies with purified proteins have shown that the heterotrimeric clamp is loaded at the 5Ј junction of a gap coated with RPA, whereas RFC loads PCNA onto a 3Ј primer-template junction during DNA replication (23,58). Although the loading of the Rad9-Rad1-Hus1 complex is necessary for activation of the checkpoint, it is not known whether Rad9-Rad1-Hus1 and associated proteins, such as DNA ligase I participate in processing of the abnormal DNA intermediate. At some stage it will be necessary to turn off the checkpoint and restart DNA replication. We have suggested that FIGURE 8. Effect of DNA damage and replication blockage on the association between DNA ligase I and hRad17. HeLa cells and derivatives of 46BR.1G1 cells were synchronized in either the G 1 or S phase of the cell cycle and then treated with hydoxyurea or ionizing radiation as described under "Experimental Procedures." A, cell extract from a G 1 phase-enriched population of HeLa cells that either had (ϩ) or had not (Ϫ) been irradiated with 12 gray of ionizing radiation. DNA ligase I and hRad17 in the whole cell extracts (WCE, 20 g) were detected by direct immunoblotting. B, cell extracts from an S phase-enriched population of HeLa cells that were untreated, treated with hydroxyurea, or irradiated with 12 gray of ionizing radiation as indicated. Proteins were immunoprecipitated from the extracts in the presence of ethidium bromide with preimmune serum, hRad17 antibody, and DNA ligase I antibody and then detected by immunoblotting with the antibodies indicated on the right. DNA ligase I and hRad17 in whole cell extracts (20 g) were detected by direct immunoblotting. C, derivatives of the DNA ligase I-deficient cell line 46BR.1G1 that either stably express FLAGtagged wild type DNA ligase I (wt) or were stably transfected with empty expression vector (V/O) were enriched for S phase cells as described under "Experimental Procedures." Extracts were prepared from cells that either had (ϩ) or had not (Ϫ) been irradiated with 12 gray of ionizing radiation. Left and middle panels, FLAG-tagged DNA ligase I, DNA ligase I, hRad17, and ␤-actin in the whole cell extracts (20 g) were detected by direct immunoblotting with the indicated antibodies. Right panel, proteins were immunoprecipitated from extracts in the presence of ethidium bromide with hRad17 antibody and then detected by immunoblotting (IB) with the indicated antibodies. D, subcellular localization of hRad17 and DNA ligase I. HeLa cells, either mock treated (Mock) or treated with 12 gray of ionizing radiation (IR) were fixed as described under "Experimental Procedures." hRad17 and DNA ligase I were detected by indirect immunofluorescence using anti-hRad17 (red) and anti-DNA ligase I antibodies (green). the interaction between RFC and DNA ligase I may be involved in the unloading of PCNA after the joining of adjacent Okazaki fragments (19,20). It is possible that DNA ligase I and Rad17-RFC may play a similar role in unloading the Rad9-Rad1-Hus1 complex, thereby switching off the checkpoint.
Based on interactions between the Rad9-Rad1-Hus1 complex and excision repair proteins that also interact with PCNA, it has been suggested that the heterotrimeric clamp and presumably Rad17-RFC may be able to substitute for PCNA and RFC in certain DNA excision repair pathways (30 -35). An attractive feature of this idea is that it provides a potential mechanism by which DNA repair pathways can operate when DNA replication is inhibited and PCNA is bound by p21 (48). However, at the present time, it is difficult to reconcile this model with the biochemical properties of the Rad17 clamp loader and the structure of the DNA repair intermediates, in particular in base excision repair.