A role for the phosphorylation of hRad9 in checkpoint signaling.

The integrity of the human genome is preserved by signal transduction pathways called checkpoints, which delay progression through the cell cycle when DNA damage is present. Three checkpoint proteins, hRad9, hRad1, and hHus1, form a proliferating cell nuclear antigen-like, heterotrimeric complex that has been proposed to function in the initial detection of DNA structural abnormalities. hRad9 is highly modified by phosphorylation, in a constitutive manner and in response to both DNA damage and cell cycle position. Here we present evidence that Thr292 of hRad9 is subject to Cdc2-dependent phosphorylation in mitosis. Furthermore, our data are also consistent with four other hRad9 phosphorylation sites (Ser277, Ser328, Ser336, and Thr355) being regulated in part by Cdc2. We also identify Ser387 as a novel site of hRad9 constitutive phosphorylation and show that phosphorylation at Ser387 is a prerequisite for one form of DNA damage-induced hyperphosphorylation of hRad9. Characterization of nonphosphorylatable mutants has revealed that hRad9 phosphorylation plays a critical role in checkpoint signaling. Overexpression of these mutants blocks the interaction between hRad9 and the DNA damage-responsive protein TopBP1 and impairs the cellular response to DNA damage during S phase.

The integrity of the human genome is preserved by signal transduction pathways called checkpoints, which delay progression through the cell cycle when DNA damage is present. Three checkpoint proteins, hRad9, hRad1, and hHus1, form a proliferating cell nuclear antigen-like, heterotrimeric complex that has been proposed to function in the initial detection of DNA structural abnormalities. hRad9 is highly modified by phosphorylation, in a constitutive manner and in response to both DNA damage and cell cycle position. Here we present evidence that Thr 292 of hRad9 is subject to Cdc2-dependent phosphorylation in mitosis. Furthermore, our data are also consistent with four other hRad9 phosphorylation sites (Ser 277 , Ser 328 , Ser 336 , and Thr 355 ) being regulated in part by Cdc2. We also identify Ser 387 as a novel site of hRad9 constitutive phosphorylation and show that phosphorylation at Ser 387 is a prerequisite for one form of DNA damage-induced hyperphosphorylation of hRad9. Characterization of nonphosphorylatable mutants has revealed that hRad9 phosphorylation plays a critical role in checkpoint signaling. Overexpression of these mutants blocks the interaction between hRad9 and the DNA damage-responsive protein TopBP1 and impairs the cellular response to DNA damage during S phase.
Cell cycle checkpoints are signal transduction pathways that maintain the proper order of cell cycle events (1). Several checkpoints preserve the integrity of DNA by sensing genetic anomalies and delaying progression through the cell cycle so that enough time is provided for these anomalies to be corrected. The contribution of checkpoints to human health is illustrated by a growing list of checkpoint genes that are mutated in cancer and cancer predisposition syndromes (2)(3)(4)(5)(6)(7)(8).
The hRad9 protein is the human homologue of Schizosaccharomyces pombe Rad9, a member of the checkpoint Rad family of proteins. In fission yeast, the checkpoint rad genes (rad1 ϩ , rad3 ϩ , rad9 ϩ , rad17 ϩ , rad26 ϩ , and hus1 ϩ ) are required for the S phase (DNA replication) and G 2 (DNA damage) checkpoints (9 -13). Yeasts lacking these genes fail to inactivate Cdc2 and enter premature, lethal mitosis when challenged with agents that inhibit DNA synthesis or damage DNA (14,15). Like its yeast counterpart, hRad9 forms a ring-shaped, heterotrimeric complex with the hRad1 and hHus1 proteins (16 -18). Each member of the hRad9-hRad1-hHus1 complex (also known as the 9-1-1 complex), shares sequence homology with PCNA, 1 a homotrimer that encircles DNA and tethers DNA polymerase ␦ during DNA synthesis (19). PCNA is loaded onto DNA by the pentameric protein complex replication factor C (RFC), which is composed of one large subunit and four smaller subunits (20). In a manner analogous to PCNA and RFC, 9-1-1 is loaded onto DNA by a complex between hRad17 and the four smallest subunits of RFC (21). Since DNA damage induces hRad17-dependent association of 9-1-1 with chromatin, the 9-1-1 complex is believed to be involved in the direct recognition of DNA lesions during the initial stages of the checkpoint response (22). Also involved in this recognition are two phosphatidylinositol 3-kinase-related kinases, ATM and ATR, that regulate several cell cycle transitions and are central components of the cell checkpoint machinery (23). Even though these kinases appear to respond to different types of DNA lesions, they share a long list of common checkpoint substrates, including hRad17 (24 -26) and hRad9 (27). In fission yeast, Rad3 (which shares homology with both ATR and ATM) requires Rad9, Rad1, Hus1, and Rad17 to phosphorylate certain substrates (28). Similarly, in human cells, phosphorylation of hRad17 by ATR requires hHus1 (22). These findings support a model in which the 9-1-1 complex recruits substrates for ATM or ATR to sites of DNA damage or stalled replication forks (29).
In addition to interacting with hRad1 and hHus1, hRad9 also physically interacts with TopBP1, the human orthologue of Saccharomyces cerevisiae Dpb11, and Schizosaccharomyces pombe Cut5 (30). Each of these proteins contains multiple BRCA1 carboxyl-terminal (BRCT) domains, a putative proteinprotein interaction motif common in cell cycle control and DNA repair. In addition to the checkpoint rad genes, cut5 ϩ is also required for slowing S phase and delaying mitosis when DNA replication is challenged by DNA damage (13,31). A similar requirement also exists for DPB11 in budding yeast (32), and a physical interaction between Dpb11 and Ddc1 (the budding yeast orthologue of hRad9) may play a role in this response (33).
hRad9 is a unique member of the 9-1-1 complex in that it contains a C-terminal region (of about 110 amino acids) that does not share homology with PCNA and is not believed to be directly involved in association with hRad1 or hHus1. This region of the protein is both constitutively and transiently phosphorylated at several amino acid residues (34), and hence represents a potential regulatory region for the effector functions of 9-1-1. We have characterized the extensive phospho-rylation of hRad9 in this region and show that it is partially regulated by Cdc2. We also demonstrate that the C-terminal phosphorylation of hRad9 has roles in regulating both hRad9 interaction with TopBP1 and the cellular response to DNA damage in S phase.

EXPERIMENTAL PROCEDURES
Cell Culture and Transfections-HeLa cells, fibroblasts derived from an ataxia telangiectasia patient (ATM Ϫ/Ϫ ) (CRL-7201; ATCC, Manassas, VA), and IMR90 fibroblasts were maintained in Dulbecco's modified Eagle's medium (Invitrogen) supplemented with 10% fetal bovine serum (Invitrogen) at 37°C in 5% CO 2 atmosphere. hTERT-RPE1 cells, a human retinal pigment epithelial cell line that stably expresses the human telomerase reverse transcriptase subunit (BD Biosciences, Mississauga, Canada), were maintained as above, only in Dulbecco's modified Eagle's medium/F-12 medium (Invitrogen). Transient transfections were performed as described previously (34). The development of the stable HeLa Tet-Off cell line was performed according to the manufacturer's instructions (BD Biosciences). This cell line was maintained in 200 g/ml of Geneticin (G418) (Invitrogen). Transgene expression was limited by the addition of 1 g/ml tetracycline to the culture medium. Small interfering RNA (siRNA) transfections were conducted using OligofectAMINE (Invitrogen), according to the manufacturer's protocol. Briefly, siRNA transfections were carried out in six-well plates using 4 l of OligofectAMINE reagent and 0.1 nmol of siRNA duplex per well. siRNAs targeting hRad9 and luciferase (GL2) were purchased commercially (Dharmacon Research, Lafayette, CO). The siRNA sequence targeting hRad9 was (5Ј-AACCCUUGGAGGACGGGCUCU-3Ј).
Plasmids and Site-directed Mutagenesis-All hRad9 point mutants were generated using the Transformer site-directed mutagenesis kit (BD Biosciences) according to the manufacturer's instructions. GST expression plasmids were generated by PCR subcloning segments of the hRad9 cDNA (wild-type and point mutants) into the BamHI and EcoRI restriction sites of the pGEX-2T vector.
Flow Cytometry-HeLa cells were fixed in 50% ethanol in phosphatebuffered saline for at least 30 min on ice. Cells were then collected by centrifugation, resuspended in phosphate-buffered saline with 50 g/ml PI and 0.1 mg/ml RNase A, and analyzed using a flow cytometer (Beckman/Coulter EPICS ALTRA, Mississauga, Canada).
Immunoprecipitations and Immunoblotting-For immunoprecipitation reactions, cells were lysed in 250 mM NaCl, 1 mM EDTA, 20 mM Tris (pH 8.0), 0.5% Nonidet P-40, and 10% glycerol, supplemented with 20 g/ml aprotinin, 4 g/ml leupeptin, 2 mM sodium orthovanadate, 20 mM ␤-glycerophosphate, and 0.2 mM sodium fluoride. Lysates were incubated on ice for 30 min and centrifuged at 13,000 ϫ g. Supernatants were precleared with ␣-chicken IgY-agarose (Promega, Madison WI) or with protein G Sepharose (Amersham Biosciences) for 30 min at 4°C prior to immunoprecipitation. The immunoprecipitation was performed with 1 g of antibodies directed against the Myc epitope and 20 l of protein G-Sepharose or antibodies directed against hRad9 and ␣-chicken IgY-agarose for 2 h at 4°C. Immune complexes were washed four times with 1 ml of lysis buffer and then resuspended in SDS-PAGE loading buffer. Immunoblotting was performed essentially as previously described (17).
Protein Purification-GST and His-tagged protein expression was induced in logarithmically growing BL21 bacteria, with 0.1 mM isopropyl-1-thio-␤-D-galactopyranoside for 3 h at 37°C. GST fusion proteins were batch-purified from bacterial lysates with glutathione-Sepharose (Promega) using standard techniques. Bound proteins were eluted with 20 mM reduced glutathione in 50 mM HEPES, 10 mM MgCl 2 , and 1 mM dithiothreitol (final pH of 7.4). His-hHus1 was purified under denaturing conditions (6 M guanidine HCl) on a nickel column. Kinase Assays-HeLa cells, arrested in mitosis with nocodazole or in S phase with thymidine, were lysed in 1 ml each of kinase lysis buffer (50 mM Tris (pH 7.4), 1 mM EDTA (pH 8.0), 25 mM NaCl, and 0.1% Nonidet P-40). The Cdc2 kinase was immunoprecipitated from these extracts using 4 g of mouse monoclonal antibody directed against Cdc2. Immune complexes were washed three times with 1 ml of phosphate-buffered saline and then two times with 1 ml of kinase reaction buffer (50 mM HEPES (pH 7.4), 10 mM MgCl 2 , 1 mM dithiothreitol, 50 M ATP) Reactions were carried out in 60 l of kinase reaction buffer in the presence of 4 Ci of [␥-32 P]ATP, and 3 g of GST fusion substrate for 30 min at 30°C. Reactions were stopped by the addition of 30 l of 3ϫ SDS-PAGE loading buffer. 20 l from each reaction was subjected to SDS-PAGE (12%), and proteins were stained with Coomassie Brilliant Blue. Gels were then dried and exposed to x-ray film (Eastman Kodak Co.) for 24 h.

RESULTS
hRad9 Is Phosphorylated at Thr 292 during Mitosis-We had previously identified four amino acid residues in hRad9 that contribute to the constitutive phosphorylation of the protein (34). Phosphorylation at Ser 277 , Ser 328 , Ser 336 , and Thr 355 results in a significant reduction in the electrophoretic mobility of hRad9 derived from undamaged, asynchronous cells. We had also previously shown that a T292A mutation prevented hyperphosphorylation of hRad9 in mitosis (34). To further characterize the nature of Thr 292 phosphorylation, phosphospecific antibodies were raised against a phospho-Thr 292 hRad9 pep-FIG. 1. Phosphorylation of hRad9 at Thr 292 during mitosis. A, HeLa cells, transiently expressing wild-type hRad9 or five hRad9 point mutants (indicated above), were grown in the presence (ϩ) or absence (Ϫ) of 0.1 g/ml microtubule inhibitor nocodazole (NOC) for 18 h. Mitotic (ϩNOC) and asynchronous (ϪNOC) cells were then lysed and immunoblotted with ␣-hRad9 antibodies (upper panel) or phosphospecific antibodies directed against phosphorylated Thr 292 of hRad9 (␣-p292; lower panel). B, similar to A but involving transiently expressed hRad9 protein harboring at least three mutations (S328A, S336G, and T355A). The S277A and T292A mutations were introduced into this mutant background, as indicated above (ϩ, wild type; Ϫ, alanine substitution).
tide. As shown in Fig. 1A, these antibodies (␣-p292) recognize wild-type (but not T292A) hRad9 from cells that were arrested in mitosis by treatment with the microtubule inhibitor nocodazole (ϩNOC). The ␣-p292 antibodies did not recognize hRad9 from asynchronous cultures (ϪNOC), indicating that hRad9 is indeed hyperphosphorylated at Thr 292 in mitosis.
We had also previously reported that mutation of Ser 277 (a constitutively phosphorylated residue in hRad9) to an alanine reduced the efficiency of hRad9 hyperphosphorylation in mitosis (34). To determine whether Thr 292 phosphorylation required prior phosphorylation at Ser 277 or any of the other constitutively phosphorylated residues above, the S277A, S328A, S336G, and T355A mutants were also immunoblotted with the ␣-p292 antibodies. We found that whereas Thr 292 phosphorylation was readily detectable in S328A, S336G, and T355A mutants derived from mitotic extracts, it was severely reduced in the S277A mutant (Fig. 1A). Similar results were obtained when hRad9 proteins harboring multiple mutations were analyzed (Fig. 1B). Whereas hRad9 protein containing the S328A, S336G, and T355A mutations was still efficiently phosphorylated at Thr 292 in mitosis, the addition of one or both of the S277A and T292A mutations inhibited this phosphorylation. Hence, the mitotic phosphorylation of hRad9 at Thr 292 requires prior phosphorylation at Ser 277 .
hRad9 Phosphorylation Is Partially Dependent on Cdc2-To further characterize the mitotic phosphorylation of hRad9, we examined the mobility of endogenous hRad9 as cells entered and exited nocodazole-induced mitotic arrest. Consistent with our previous observations (34), the predominant phospho form of hRad9 in asynchronous cultures (hRad9␣) migrated at an apparent molecular mass of ϳ60 kDa. The discrepancy between the apparent molecular mass of hRad9 and its predicted molecular mass of 43 kDa is largely due to phosphorylation at Ser 277 , Ser 328 , Ser 336 , and Thr 355 . Several less abundant, faster migrating phosphorylation intermediates (hRad9␤ forms) represent hRad9 protein that is only partially phosphorylated at these sites. When cells were blocked in mitosis by treatment with nocodazole, the abundance of both hRad9␣ and hRad9␤ forms was reduced, and the majority of hRad9 existed as an even slower migrating species, hRad9 (see Fig. 2A). When the mobility pattern of hRad9 was examined at increasing time points following the removal of nocodazole from the culture medium, the emergence of normal hRad9 mobility coincided with the exit of cells from mitosis, as defined by flow cytometry ( Fig. 2A, 120 and 400 min).
The observations made in Figs. 1 and 2A, are consistent with each of Ser 277 , Ser 328 , Ser 336 , Thr 355 , and Thr 292 being quantitatively phosphorylated during mitosis, although the vast majority of hRad9 is already phosphorylated at the four former residues prior to mitosis. Each of these five amino acids share the common consensus sequence ((S/T)PX(R/P)) and are thus potential targets for cyclin-dependent kinases. For these reasons, we sought to determine whether Cdc2, the cyclin-dependent kinase that controls the G 2 /M transition, could phosphorylate Ser 277 , Thr 292 , Ser 328 , Ser 336 , and Thr 355 in vitro. To this end, we purified a series of hRad9 C-terminal peptides as GST fusion proteins that encompassed the amino acids indicated above, either left intact or mutated to alanine or glycine residues. Peptides were used as substrates for Cdc2, which was purified by immunoprecipitation from HeLa cells arrested in mitosis or S phase.
These analyses revealed that Cdc2 could phosphorylate each of the five (S/T)PX(R/P) sites in vitro, with the exception of Thr 292 (Fig. 2B). Phosphorylation of hRad9 and the histonepositive control was enhanced if Cdc2 was derived from mitotic rather than S phase extracts, illustrating the specificity of the ␣-Cdc2 antibody. Phosphorylation of Ser 277 was demonstrated by introduction of the S277A mutation in GST-hRad9 266 -301 , which abolished phosphorylation of this protein by Cdc2. Nei-

FIG. 2. hRad9 is phosphorylated by Cdc2 in vivo and in vitro.
A, hyperphosphorylation of hRad9 in mitosis. Soluble cell extracts from HeLa cells, grown in the absence of nocodazole (Asyn), grown in the presence of 0.1 g/ml nocodazole for 18 h (Block), or harvested the indicated number of minutes after release from nocodazole block, were immunoblotted with antibodies directed against hRad9. In asynchronous cells, the slowest migrating hRad9 form is designated hRad9␣, and faster migrating partially phosphorylated intermediates are designated hRad9␤. The predominant mitotic form of hRad9 is designated hRad9. The percentage of cells in G 2 /M was determined by flow cytometry and is indicated below. B, phosphorylation of hRad9 by Cdc2 in vitro. Three hRad9 peptides (amino acids 266 -301, 314 -344, and 348 -391; indicated below) were purified as GST fusion proteins. Peptides either contained wild-type (WT) hRad9 amino acid sequence or contained one or multiple mutations at Ser 277 , Thr 292 , Ser 328 , Ser 336 , and Thr 355 , as indicated above. The P3A 348 -391 peptide contains the T355A mutation as well as the S375A and S380G mutations (two other C-terminal S-P sites. Purified peptides, as well as histone H1 and GST, were used as in vitro substrates for Cdc2 immunoprecipitated from mitotic or S phase HeLa cell extracts, as indicated on the right. Reactions were subjected to SDS-PAGE (12% acrylamide), and proteins were stained with Coomassie Brilliant Blue (a representative gel is shown in the top panel). Each gel was dried and exposed to x-ray film for 24 h. C, phosphorylation of hRad9 by Cdc2 in vivo. Asynchronous HeLa cells were left untreated (Asyn) or were treated with 0.1 g/ml nocodazole (NOC) for 18 h, as indicated. These cells were grown in the presence (ϩ) or absence (Ϫ) of a 100 M concentration of the CDK inhibitor roscovitine (Ros) from hour 16 to 18, as indicated. Soluble cell lysates were size-fractionated by SDS-PAGE (8%) and immunoblotted with antibodies directed against hRad9.
hRad9 Phosphorylation and Checkpoint Control ther the S277A mutant nor the S277A/T292A double mutant were phosphorylated at levels significantly above background, indicating that Thr 292 was not phosphorylated by Cdc2 in this assay. Phosphorylation of Ser 328 and Ser 336 was shown using GST-hRad9 314 -344 . Whereas the S328A and S336G single mutants of GST-hRad9 314 -344 were still modestly phosphorylated, the S328A/S336G double mutant showed only background levels of 32 P incorporation. Thr 355 was more efficiently phosphorylated than any of the other hRad9 sites (see GST-hRad9 348 -391 ). There was no appreciable difference in phosphorylation of T355A and P3A (T355A/S375A/S380G), indicating that Ser 375 and Ser 380 are not phosphorylated by Cdc2 in vitro. This is consistent with our previous work in which we found no evidence for phosphorylation of these residues in vivo (34). Furthermore, there did not appear to be any significant phosphorylation of non-Cdk consensus sites in hRad9, thus underscoring the specificity of Cdc2 toward Ser 277 , Ser 328 , Ser 336 , and Thr 355 of hRad9 in this assay.
To determine whether Cdc2 phosphorylates hRad9 in vivo, we employed the Cdk inhibitor roscovitine. Roscovitine is a selective inhibitor of both Cdc2 and Cdk2 (35). When HeLa cells were arrested in mitosis with nocodazole and then treated with roscovitine (Ros) for 2 h, a drastic reduction in hRad9 phosphorylation was observed (Fig. 2C). Roscovitine had no effect however, on hRad9 phosphorylation in asynchronous cultures (Fig. 2C).
hRad9 Is Constitutively Phosphorylated at Ser 387 -Site-directed mutagenesis of potentially phosphorylated residues in the C terminus of hRad9 identified a mutant, S387A, which was not normally hyperphosphorylated in response to the DNA synthesis inhibitor hydroxyurea (HU). To characterize this residue, phosphospecific antibodies were raised against a phospho-Ser 387 hRad9 peptide. The quality of these antibodies was determined by testing their reactivity toward phosphorylated and dephosphorylated wild-type and S387A hRad9. Although transiently expressed wild-type and S387A hRad9 were indistinguishable in terms of their SDS-PAGE mobility, they were clearly distinct with regard to their detection when immunoblotted with ␣-phospho-Ser 387 (␣-p387) antibodies (Fig. 3A). Whereas the ␣-p387 antibodies recognized all differentially migrating forms of wild-type hRad9, they were completely nonreactive toward the S387A mutant (Fig. 3A, right blot). In addition, the ␣-p387 antibodies did not detect wild type hRad9 that had been dephosphorylated with calf intestinal phosphatase. Thus, hRad9 is phosphorylated at Ser 387 , and the ␣-p387 antibodies are effective in specifically recognizing Ser 387phosphorylated hRad9.
To determine whether phosphorylation at Ser 387 of hRad9 was regulated in a cell cycle-or DNA damage-dependent manner, we collected HeLa cells that were in G 1 , S, G 2 , and M phases of the cell cycle, as we have previously described (34). HeLa cells were synchronized with a double thymidine block and then released and collected 2 h (S), 7 h (G 2 ), and 11 h (G 1 ) later. One hour prior to harvest, these cells were treated with 20 Gy of ionizing radiation (IR), as indicated. Cells were arrested in mitosis (M) by treating asynchronous cells with nocodazole. We also collected HeLa cells that were grown in the presence of HU for 18 h or were harvested 18 h after a 20-Gy dose of IR. At the time of harvest, these cells were arrested in early S phase and G 2 , respectively. In all cases, cell cycle position was confirmed using flow cytometry of propidium iodide-stained nuclei. Soluble cell lysates from these cells were subjected to immunoblotting with ␣-hRad9, ␣-p387, and phosphospecific antibodies directed against phosphorylated Ser 272 of hRad9 (␣-p272), a site of DNA damage-dependent phosphorylation (27).
In agreement with several previous reports (16,22,27,34,36,37), the constitutively phosphorylated hRad9 protein underwent hyperphosphorylation under various conditions (Fig.  3B, top panel). Ionizing radiation induced rapid phosphorylation of hRad9 at Ser 272 regardless of cell cycle position (27). Whereas Ser 272 phosphorylation did produce a subtle mobility shift, it was more easily visualized by immunoblotting with FIG. 3. hRad9 is constitutively phosphorylated at Ser 387 . A, phosphospecific antibodies against Ser 387 of hRad9. Wild-type (WT) or S387A hRad9 proteins were transiently expressed in HeLa cells and immunoprecipitated using antibodies directed against hRad9. hRad9 protein was then incubated at 37°C for 30 min in the presence (ϩ) or absence (Ϫ) of calf intestinal phosphatase (CIP). Proteins were subjected to immunoblotting with antibodies directed against hRad9 (␣-hRad9) or phosphospecific antibodies directed against phosphorylated Ser 387 of hRad9 (␣-p387), as indicated below. The slowest migrating form of hRad9 (which co-migrates with the majority of endogenous hRad9) is designated hRad9␣. Faster migrating hRad9 forms, which represent partially phosphorylated intermediates, are designated hRad9␤. De-phosphorylated hRad9 is designated hRad9␦. B, Ser 387 phosphorylation is not regulated in a cell cycle-or DNA damage-dependent manner. HeLa cells were synchronized in S phase, G 2 , or G 1 by double thymidine block and release and harvested 1 h after mock irradiation or irradiation with 20 Gy of IR. HeLa cells were also arrested in G 2 phase, early S phase, or mitosis by administration of 20 Gy of IR (IR(18h)), 5 mM HU (HU(18h)), or 0.1 g/ml nocodazole (M), 18 h prior to harvest, respectively. Soluble cell lysates were immunoblotted with ␣-hRad9, ␣-p272, or ␣-p387 antibodies, as indicated. C, Ser 387 is phosphorylated in normal and ATM Ϫ/Ϫ cell lines. Soluble cell lysates, derived from the indicated cell types, were immunoblotted with ␣-hRad9, ␣-p272, or ␣-p387 antibodies. Cells were harvested 1 h following mock irradiation (Ϫ) or irradiation (ϩ) with 20 Gy of IR. x, unknown proteins that react with the ␣-p387 antibodies.

hRad9 Phosphorylation and Checkpoint Control
␣-p272 antibodies (Fig. 3B, middle panel). The ␣-p272 blot revealed that IR-induced phosphorylation at Ser 272 was not only rapid but also transient and dissipated by 18 h following irradiation. The hRad9 protein from cells harvested 18 h after IR exhibited a second IR-induced mobility shift. This IR-induced modification, however, which persisted after phosphorylation at Ser 272 had dissipated, was less rapidly induced and was not readily detectable 1 h after IR (Fig. 3B, top panel). The hRad9 protein also underwent similar mobility shifts in G 2 , in response to HU, and in mitosis.
In contrast to these dynamic phosphorylation changes, the extent of Ser 387 phosphorylation did not change significantly in response to cell cycle perturbation or DNA damage. In fact, the reactivity of the ␣-p387 antibodies toward hRad9 seemed to mirror that of the ␣-hRad9 antibodies (Fig. 3B, bottom panel) in that they recognized all differentially migrating forms of hRad9. We went on to show that Ser 387 of hRad9 is phosphorylated constitutively in normal (noncancerous) cell lines and those with homozygous mutations in ATM (Fig. 3C). Thus, Ser 387 represents a novel, ATM-independent, constitutive phosphorylation site in hRad9. Ser 387 is distinct from previously identified constitutive sites in hRad9, however, in that it does not fit the (S/T)PX(R/P) consensus sequence seen in these other sites (Table I), and it is not phosphorylated by Cdc2 in vitro (Fig. 2B, peptide 348 -391).
Ser 387 Phosphorylation Regulates DNA Damage-dependent Hyperphosphorylation of hRad9 -We next examined the dependence of DNA damage-induced hRad9 hyperphosphorylation events on hRad9 constitutive phosphorylation. As described previously (27) and as illustrated in Fig. 3B, hRad9 becomes hyperphosphorylated in a DNA damage-dependent manner at Ser 272 . We had previously shown that this process occurs independently of constitutive phosphorylation at (S/T)P sites (34). Using ␣-p272 antibodies, this observation was confirmed in Fig. 4A, since a P4A mutant (harboring the S277A, S328A, S336G, and T355A mutations) was still inducibly phosphorylated at Ser 272 1 h after a 20-Gy dose of IR. Similarly, the S387A mutant was also inducibly phosphorylated at Ser 272 at levels comparable with that of wild-type hRad9 (Fig. 4A), indicating that Ser 387 phosphorylation is also not a prerequisite for DNA damage-induced phosphorylation at Ser 272 .
We had previously reported that hRad9 undergoes DNA damage-dependent hyperphosphorylation that is distinct from phosphorylation at Ser 272 (34). This second hyperphosphorylation event was illustrated in Fig. 3B. When cells were subjected to DNA damage in late S phase/G 2 , treated with hydroxyurea, or harvested after extended periods of time following IR, hRad9 underwent a mobility shift that was similar to that observed in mitotic cells (Fig. 3B). This observation had initially led us to believe that this DNA damage-induced phosphorylation of hRad9 was occurring at Thr 292 (the site of mitotic phosphorylation; Fig. 1). To address this directly, hRad9 protein was harvested from cells grown under each of the above conditions and immunoblotted with ␣-p292 antibodies. As shown in Fig.  4B, only hRad9 derived from mitotic cells was phosphorylated at Thr 292 . Thus, the location of Ser 272 -independent, DNA damage-induced hRad9 phosphorylation remains unclear.
As mentioned earlier, our initial interest in Ser 387 was sparked from the observation that a S387A mutant was not efficiently hyperphosphorylated in response to HU. Whereas this had initially led us to believe that Ser 387 phosphorylation was induced by HU, our analysis with ␣-p387 antibodies (Fig.  3) has indicated this to be incorrect. We therefore reasoned that Ser 387 phosphorylation was a prerequisite for HU-induced hyperphosphorylation and may also be a prerequisite for hyperphosphorylation of hRad9 under other conditions. To address this possibility, we examined the response of wild-type, S272A, and S387A hRad9 protein to growth conditions known to induce DNA damage-dependent/Ser 272 -independent hyperphosphorylation. These proteins were transiently expressed at low levels using the Tet-Off expression system and were Myctagged to distinguish them from endogenous (wild-type) hRad9. In these experiments, transient expression levels near that of endogenous hRad9 were necessary, since overexpressed hRad9 is not efficiently hyperphosphorylated in response to HU. 2 Unlike wild-type and S272A hRad9, the S387A mutant was not efficiently hyperphosphorylated when HeLa cells were synchronized in G 2 , harvested 18 h after 20 Gy of IR, or treated with HU (Fig. 4C). We refer to this hyperphosphorylated form of hRad9 as hRad9 to distinguish it from mitotic hRad9 (hRad9), which is readily observed in the S387A mutant. As shown in Fig. 4D, neither the S272A nor the S387A mutation prevented mitotic phosphorylation of hRad9 at Thr 292 . A summary of the hRad9 phosphorylation sites described in this paper is given in Table I. hRad9 Constitutive Phosphorylation Sites Regulate Association with TopBP1-hRad9 forms a ring-shaped, PCNA-like heterotrimeric complex with hHus1 and hRad1 (16 -19). The C terminus of hRad9, which does not exhibit sequence or structural similarity to PCNA, has been shown to interact with the DNA damage-responsive protein TopBP1 (30). Using hRad9 C-terminal deletion mutants, we refined the region of hRad9 required for the interaction with TopBP1 to 17 amino acids at the extreme C terminus of hRad9. 3 Because this region encompasses Ser 387 , we hypothesized that phosphorylation at Ser 387 could be regulating the interaction between hRad9 and TopBP1. To test this, HeLa cells were transiently transfected with plasmids that overexpress wild-type, S272A, S387A, or P4A hRad9. hRad9-containing complexes were immunoprecipi- hRad9 Phosphorylation and Checkpoint Control tated from transfected cell lysates with ␣-hRad9 antibodies. Immunoprecipitated proteins were then size-fractionated by SDS-PAGE and immunoblotted with ␣-TopBP1, ␣-hRad9, ␣-p387, ␣-hHus1, and ␣-hRad1 antibodies. Due to high levels of transiently expressed hRad9 relative to endogenous hRad9, the vast majority of the immunoprecipitated hRad9 protein was plasmid-derived. Accordingly, the ␣-p387 immunoblot (Fig. 5A, third panel) does not exhibit any endogenous, Ser 387 -phosphorylated hRad9. TopBP1 co-immunoprecipitated with wild-type, S272A, and to a lesser extent P4A hRad9, but did not coimmunoprecipitate with the S387A mutant (Fig. 5A, top panel). This was confirmed in a similar experiment employing epitopetagged hRad9 proteins. Whereas TopBP1 co-immunoprecipitated with Myc-tagged wild-type and S272A hRad9, it did not co-immunoprecipitate with Myc-tagged S387A (Fig. 5B, top  panel). Immunoblotting hRad9-containing immune complexes with ␣-hRad1 and ␣-hHus1 antibodies revealed that none of the mutations tested prevented hRad9 association with either hRad1 or hHus1 (Fig. 5A, bottom two panels).
Overexpression of hRad9 Constitutive Phosphorylation Mutants Results in Prolonged G 2 /M Arrest following IR-We used siRNA directed against the hRad9 transcript to determine whether reduction of hRad9 protein levels would impair the response of cells to DNA damage. Immunoblotting revealed that the hRad9 siRNA duplex was successful in reducing steady state hRad9 protein levels, relative to mock-and luciferase (Luc) siRNA-transfected controls (Fig. 6A, top). Sixteen hours after exposure of cells to 20 Gy of IR, the majority of mock-, hRad9 siRNA-, and Luc siRNA-transfected cells were arrested with a G 2 /M DNA content (Fig. 6A, bottom). When these cells were monitored for exit from G 2 /M arrest, hRad9-siRNA-transfected cells exhibited prolonged G 2 /M arrest in relation to the control cells (Fig. 6A). No measurable changes in cell cycle distribution were observed in unperturbed hRad9 siRNA transfected cells relative to control cells. 2 We next wanted to determine if cells overexpressing the hRad9 constitutive phosphorylation mutants exhibited the prolonged G 2 /M arrest defect observed in hRad9 hypomorphic cells. To this end, HeLa cells were transiently transfected with expression plasmids encoding wild-type, S272A, S387A, or P4A hRad9. In all cases, the transfection efficiency was about 50%, based on indirect immunofluorescence/flow cytometry. In this experiment, cells were arrested in early S phase with thymidine, released from arrest, and then irradiated 15 min later with 10 Gy of IR. Cells were then collected at the indicated time points, fixed, and stained with propidium iodide (Fig. 6B). By 9 h after irradiation, a time at which unirradiated cells had surpassed the G 2 /M transition (34), most irradiated cells were arrested with a G 2 /M DNA content. Even though the effect was not as pronounced as that of the hRad9-siRNA-transfected cells ( Fig. 6A), cells overexpressing S387A and P4A exited G 2 /M at a reduced rate when compared with cells overexpressing wildtype or S272A hRad9 (Fig. 6B, 13-h time point). Thus, overexpression of hRad9 mutants that cannot be phosphorylated at Ser 387 or constitutive (S/T)P sites results in prolonged G 2 /M arrest following IR, a defect that is also exhibited by cells with reduced hRad9 protein levels. DISCUSSION We have characterized the extensive constitutive phosphorylation at the C terminus of the hRad9 checkpoint protein. We find that Thr 292 is phosphorylated exclusively in mitosis, in a manner that is consistent with it being an in vivo substrate of Cdc2. Four other amino acids in hRad9 (Ser 277 , Ser 328 , Ser 336 , and Thr 355 ) that contribute to the constitutive phosphorylation of the protein and share a common consensus sequence with Thr 292 ((S/T)PX(R/P)) may also be subject to phosphorylation by Cdc2. We have also identified a fifth constitutive site of hRad9 phosphorylation at Ser 387 . Analysis of hRad9 phosphorylation mutants has indicated that interdependencies exist between distinct hRad9 phosphorylation events. Perhaps more importantly, however, these phosphates appear to be of critical importance for hRad9 association with TopBP1. A P4A mutant, harboring the S277A, S328A, S336G, and T355A mutations, displays reduced affinity for TopBP1 relative to wild-type hRad9, and S387A displays no detectable interaction with TopBP1. Transient expression of the S387A and P4A mutants also disrupts cell cycle progression of cells irradiated during S phase.
ATM phosphorylates hRad9 at Ser 272 in response to ionizing radiation (27). hRad9 also undergoes a second DNA damageregulated hyperphosphorylation that occurs on an amino acid residue other than Ser 272 (Fig. 4C). Whereas both of these events are dependent on DNA damage, they differ significantly in their regulation in a number of ways. Ser 272 phosphorylation is response to IR occurs rapidly (27) (Fig. 3A), independently of cell cycle position (27) (Fig. 3A), independently of constitutive phosphorylation at Cdk sites and Ser 387 (Fig. 4A), and is not FIG. 5. The S387A and P4A mutants display reduced affinity for TopBP1. A, wild-type (WT), S272A, S387A, or P4A (S277A/S328A/ S336G/T355A) hRad9 proteins were transiently overexpressed in HeLa cells. hRad9 protein was immunoprecipitated from cell extracts using antibodies directed against hRad9. hRad9 containing immune complexes and an aliquot of the corresponding soluble transfected cell lysates were immunoblotted with ␣-TopBP1, ␣-hRad9, ␣-p387, ␣-hHus1, or ␣-hRad1 antibodies, as indicated on the left. B, similar to A except that proteins were expressed with an N-terminal Myc tag and immunoprecipitated with antibodies directed against the Myc epitope. Immune complexes were immunoblotted with ␣-TopBP1, ␣-Myc, or ␣-p387 antibodies, as indicated. hRad9 Phosphorylation and Checkpoint Control limited when hRad9 is overexpressed (Fig. 4A). Ser 272 -independent IR-induced phosphorylation, on the other hand occurs less rapidly (Fig. 3B), requires prior phosphorylation at Ser 387 (Fig. 4C), and does not occur efficiently when hRad9 is overexpressed. 2 The fact that a similar Ser 387 -dependent hRad9 mobility shift is observed in cells that are exposed to hydroxyurea or synchronized in G 2 (Fig. 4C) suggests that this event is also cell cycle-dependent and may be a response to DNA damage detected during S phase. Others have reported that hyperphosphorylation of hRad9 under similar conditions requires the activity of phosphatidylinositol 3-kinase-related kinases (37). Further experimentation will be required to determine whether this modification is the same as the one reported here.
Our results also demonstrate that the constitutive phosphorylation of hRad9 at (S/T)PX(R/P) sites and Ser 387 is dispensable for interaction with hHus1 and hRad1 (Fig. 5A). These results are in agreement with a previous report, which demonstrated that constitutive phosphorylation of hRad9 does not influence the stability of the 9-1-1 complex (38). In contrast, a recent report has suggested that hRad9 phosphorylation is important for 9-1-1 stability (39). Analysis of a hRad9 mutant, in which phosphorylation is completely ablated, will be required to fully reconcile these seemingly conflicting results. We have shown, however, that hRad9 interaction with TopBP1 is partially compromised in the P4A mutant and undetectable in the S387A mutant. Since the majority of cellular hRad9 exists as part of the 9-1-1 complex (40), our data are consistent with hRad9 phosphorylation regulating protein interactions with the 9-1-1 complex rather than interactions between 9-1-1 members.
Depletion of hRad9 protein using hRad9-directed siRNA results in a prolonged accumulation of cells in G 2 /M following irradiation (Fig. 6A). Furthermore, overexpressing hRad9 mutant proteins that cannot be phosphorylated at Ser 387 or (S/T)PX(R/P) constitutive sites (P4A) produces a similar, albeit less pronounced, effect (Fig. 6B). Prolonged G 2 /M accumulation following irradiation is a common phenotype of S phase checkpoint-defective cells, such as those lacking ATM, BRCA1, Nbs1, and Smc1 (41,42). It has been proposed to be a compensatory mechanism that provides cells that failed to appropriately arrest DNA synthesis extra time to deal with replicated damaged DNA (41). Therefore, S phase checkpoint dysfunction may be a plausible explanation for the cell cycle defects associated with hRad9 protein depletion and S387A or P4A mutant overexpression. The fact that these mutants are also defective in their ability to interact with TopBP1 further suggests that the hRad9/TopBP1 interaction plays a functionally significant role in the S phase checkpoint. This is supported by work in S. cerevisiae, which indicates an important role for the Dpb11/Ddc1 interaction in responding to genotoxins (33). The abrogation of DNA damage-dependent, Ser 272 -independent phosphorylation of hRad9 (Fig. 4C) may also be a contributing factor to the prolonged G 2 /M phenotype. Mapping the site(s) of this phosphorylation will be required to fully resolve this possibility.
Finally, the data reported here suggest that hRad9 is regulated in part by Cdc2, a kinase generally thought to represent the ultimate target of hRad9-associated signaling pathways. Whereas the maintenance of basal phosphorylation of the hRad9 C terminus appears to be essential for hRad9-mediated DNA damage response, the hyperphosphorylation of hRad9 at Thr 292 in mitosis is particularly intriguing and could potentially represent a means of resetting hRad9 activity for future cell cycles. The fact that this phosphorylation is exclusively a mitotic event is consistent with Thr 292 being an in vivo substrate of Cdc2. Whereas Thr 292 is the only amino acid in hRad9 with the (S/T)PX(R/P) consensus sequence that was not phosphorylated by Cdc2 in vitro (Fig. 2B), this is probably due to the dependence of Thr 292 phosphorylation on prior phosphorylation at Ser 277 (Fig. 1, A and B). The extent to which Ser 277 , Ser 328 , Ser 336 , and Thr 355 are phosphorylated outside of mitosis suggests that these sites may be regulated by other kinases in addition to Cdc2. In fact, we have observed that Cdk2 can also phosphorylate each of these four residues in vitro. 2 In addition, protein kinase C␦ has recently been shown to directly regulate hRad9 phosphorylation, although the location of protein kinase C␦-mediated phosphorylation in hRad9 remains unclear (39).
The non-PCNA-like, C-terminal region of hRad9 represents an ideal regulatory domain for the effector functions of the 9-1-1 complex. Its extensive phosphorylation may enable hRad9 and 9-1-1 to coordinate multiple functions during the DNA damage response. The complex nature of hRad9 phosphorylation, however, as illustrated by the large number of phosphorylation sites and the interdependencies between these sites, has made deciphering the precise role of individual phosphates a formidable task. The results presented here advance the current understanding of how hRad9 phosphorylation contributes to checkpoint signaling.