Rad9A Is Required for G₂ Decatenation Checkpoint and to Prevent Endoreduplication in Response to Topoisomerase II Inhibition

Abstract

The Rad9A checkpoint protein interacts with and is required for proper localization of topoisomerase II-binding protein 1 (TopBP1) in response to DNA damage. Topoisomerase II (Topo II), another binding partner of TopBP1, decatenates sister chromatids that become intertwined during replication. Inhibition of Topo II by ICRF-193 (meso-4,4′-(3,2-butanediyl)-bis-(2,6-piperazinedione)), a catalytic inhibitor that does not induce DNA double-strand breaks, causes a mitotic delay known as the G₂ decatenation checkpoint. Here, we demonstrate that this checkpoint, dependent on ATR and BRCA1, also requires Rad9A. Analysis of different Rad9A phosphorylation mutants suggests that these modifications are required to prevent endoreduplication and to maintain decatenation checkpoint arrest. Furthermore, Rad9A Ser²⁷² is phosphorylated in response to Topo II inhibition. ICRF-193 treatment also causes phosphorylation of an effector kinase downstream of Rad9A in the DNA damage checkpoint pathway, Chk2, at Thr⁶⁸. Both of these sites are major targets of phosphorylation by the ATM kinase, although it has previously been shown that ATM is not required for the decatenation checkpoint. Examination of ataxia telangectasia (A-T) cells demonstrates that ATR does not compensate for ATM loss, suggesting that phosphorylation of Rad9A and Chk2 by ATM plays an additional role in response to Topo II inhibition than checkpoint function alone. Finally, we have shown that murine embryonic stem cells deficient for Rad9A have higher levels of catenated mitotic spreads than wild-type counterparts. Together, these results emphasize the importance of Rad9A in preserving genomic integrity in the presence of catenated chromosomes and all types of DNA aberrations.

Introduction

G₂ checkpoints guard genomic integrity by delaying cellular processes in the event of DNA damage or other chromosomal aberrations (1). Checkpoint sensors impart meticulous inspection of DNA during replication through to mitotic division and signal to the downstream cascade, thereby strictly controlling the sequence and timing of each (2). Mutation or altered expression of checkpoint genes can result in genetic instability, leading to polyploidy, aneuploidy, and tumorgenicity.

The central proteins in the G₂ checkpoint response, ATM and ATR, receive and transduce signals transmitted by sensor proteins that survey damaged DNA (2). G₂ checkpoint protein Rad9A is one of three proteins that sense damaged DNA, forming a proliferating cell nuclear antigen-like heterotrimeric ring around DNA, known as the 911 complex (3–5). The C terminus of Rad9A extends beyond the sequence homology with proliferating cell nuclear antigen and is thought to regulate the 911 complex through its many phosphorylation sites (6–8). Rad9A is constitutively phosphorylated at five C-terminal sites, Ser²⁷⁷, Ser³²⁸, Ser³³⁶, Thr³⁵⁵, and Ser³⁸⁷, and ablation of these sites causes prolonged G₂/M arrest following ionizing radiation (IR),² consistent with an S phase checkpoint defect (6, 7). Cells expressing Rad9A mutants lacking all known C-terminal phosphorylation sites are also sensitive to hydroxyurea, ultraviolet radiation (UV), topoisomerase I (Topo I) inhibitors, and topoisomerase II (Topo II) poisons (8, 9). Rad9A is also hyperphosphorylated in response to DNA damage by inducible phosphorylation at Ser²⁷² and at a damage and cell cycle-dependent site that has not yet been defined (6, 10). Rad9A Ser²⁷² is one of many targets of damage-induced phosphorylation by ATM, although paradoxically, phosphorylation at this site is not required for checkpoint activation in response to IR (7, 8). Furthermore, localization of replication and checkpoint protein Topo-binding protein 1 to damage-induced foci depends on the C-terminal 17 amino acids of Rad9A and was shown to interact specifically with Rad9A at Ser³⁸⁷ (7, 11). Thus, the Rad9A constitutive phosphorylation sites are necessary for the G₂ checkpoint signaling cascade and are required to sense many forms of chromosomal aberrations.

Many damage-induced ATM substrates are, however, critical to the checkpoint, such as phosphorylation of Chk2 at Thr⁶⁸ (12). Chk2 is an effector kinase downstream of ATM that suppresses the activation of the mitotic switch Cdc2/Cyclin B1 (12, 13). Mutations in Chk2 are found in patients with a form of Li-Fraumeni syndrome, characterized by hereditary and sporadic cancers akin to those caused by p53 mutation, underscoring the importance of p53 targeting by Chk2 in G₁ checkpoint signaling pathways (14). Chk2 Thr⁶⁸ is the major site of IR-induced phosphorylation by ATM and when activated, targets Cdc25C, a major point of convergence of the G₂ checkpoint (12, 13). Chk2 also interacts with Polo-like kinase 1 (Plk1), which regulates chromosomal segregation, and mitotic entry and exit, whereas Chk2 itself is localized to centrosomes in undamaged cell (15). Recently it was shown that inhibition of Chk2 causes failure of mitotic arrest, resulting in mitotic catastrophe and cell death, linking Chk2 activity to prevention of abnormal mitoses (16). This was confirmed when Chabalier-Taste et al. (17) showed that BRCA1 phosphorylation is dependent on Chk2 during the spindle assembly checkpoint and that this is required for successful chromosomal separation.

Successful separation of mitotic chromosomes is dependent on Topo II, which decatenates sister chromatids that become tangled during replication. Inhibition of Topo II decatenation activity by bisdioxopiperazine ICRF-193, a catalytic inhibitor that stabilizes Topo II in the closed clamp conformation and initiates a checkpoint-mediated mitotic delay (18, 19). What is known about the decatenation checkpoint signaling pathway scaffold is reminiscent of the G₂ checkpoints, as it is activated by ATR and ultimately targets Cdc2/Cyclin B1 (20). Cells expressing ATR kinase mutants do not demonstrate mitotic arrest in response to ICRF-193 treatment, but rather proceed through aberrant mitosis resulting in chromosomal aberrations (20). Conversely, ATM is not required to initiate decatenation checkpoint signaling (20). Although ATM is not required for the decatenation checkpoint, ICRF-193 treatment leads to autophosphorylation at Ser¹⁹⁸¹, suggesting that ATM plays an overlapping, nonessential role in this pathway (20, 21). The main downstream effector of ATR activation, Chk1, is not phosphorylated at Ser³¹⁷ or Ser³⁴⁵, nor was Chk2 reported to become hyperphosphorylated (20). Instead, the decatenation checkpoint causes cytoplasmic accumulation of Cyclin B1, sequestering the functional mitotic Cdk complex away from the nucleus, thereby arresting cells in G₂; this arrest can be overridden by expression of an exclusively nuclear Cyclin B1 (20, 22). Decatenation and the decatenation checkpoint are also dependent on the BRCA1 and WRN proteins. BRCA1 is phosphorylated in an ATR-dependent manner when Topo II is inhibited, and both BRCA1 and WRN (the RecQ helicase that is mutated in Werner syndrome) are necessary for proper decatenation of chromosomes (20, 23, 24). Interestingly, fission yeast RecQ helicase homolog Rad12 has been illustrated to function upstream of Rad9 in regulating exit from the S phase checkpoint (25). This suggests that resolution of altered chromosome structure by the RecQ class of helicases and Topo II is linked to the function of genomic surveyors such as Rad9A. However, no mammalian decatenation checkpoint sensors have as yet been identified.

Here, we demonstrate that Rad9A is required for the decatenation checkpoint. Although Rad9A is inducibly phosphorylated at Ser²⁷² when Topo II activity is inhibited by ICRF-193, mutation of the constitutive phosphorylation sites of Rad9A abrogates decatenation checkpoint activity. Moreover, when these sites are mutated, cells demonstrate endoreduplication, which is associated with ploidy defects and severe genomic instability. Rad9A deficiency is also associated with high levels of catenation detected in mitotic spreads after ICRF-193 treatment compared with wild-type (WT) cells.

EXPERIMENTAL PROCEDURES

Cell Lines, Cell Culture, and DNA-damaging Agents

HeLa (CCL-2; American Type Culture Collection (ATCC), Manassas, VA), IMR-90 (CCL-186; ATCC), and ATM^−/− (CRL-7201; ATCC) cells were grown in Dulbecco's modified Eagle's medium (Invitrogen) supplemented with 10% (v/v) fetal bovine serum (Sigma) and antibiotic-antimycotic (Invitrogen) at 37 °C in a 5% (v/v) CO₂ atmosphere. Cells were exposed to the indicated dose of ICRF-193 (BIOMOL, Plymouth Meeting, PA) or γ-irradiation, performed with ¹³⁷Cs in a Victoreen electrometer (Atomic Energy of Canada, Mississauga, Canada), at a dose rate of 0.50 Gy/min. Rad9A WT, Ser²⁷², and P5A phosphorylation-mutant stable cells were generated by transfection of HeLa cells expressing the pTet-Off vector (Clontech) with pTRE/Rad9A-myc WT, Ser²⁷², and P5A (6), respectively, and pTK-Hyg (Clontech), and selecting clones resistant to hygromycin, according to the manufacturer's instructions. Murine embryonic stem (mES) cells, both WT and containing a genomic deletion of the rad9A gene, were a gift from Dr. Howard Lieberman of Columbia University of New York, NY.

mES cells were maintained in gelatinized culture dishes in KnockOut Dulbecco's modified Eagle's medium (Invitrogen) supplemented with the following: 15% (v/v) fetal bovine serum (ES qualified; Invitrogen), 0.1 mm β-mercaptoethanol (BioShop, Burlington, Canada), 0.1 mm nonessential amino acids (Invitrogen), 2 mm l-glutamine, 50 μg/ml penicillin/streptomycin (Sigma), and 10³ units/ml ESGRO (Millipore, Etobicoke, Canada). Differentiation of mES cells was accomplished by growth in medium supplemented with 10 μm retinoic acid for 48 h, as described previously (26).

Immunoblotting

Cells grown in 10-cm tissue culture dishes were washed twice with PBS, lysed in NETN buffer (150 mm NaCl, 1 mm EDTA, 20 mm Tris-Cl, 0.5% (v/v) Nonidet P-40, pH 8.0) supplemented with 0.5 mm AEBSF (BioShop), 20 μg/ml aprotinin (BioShop), 10 μg/ml leupeptin (BioShop), 10 μg/ml pepstatin A (BioShop), 1 mm Na₃VO₄ (ICN), 50 mm β-glycerol phosphate (Sigma), and 10 mm NaF (BD Biosciences, Mississauga, Canada), and removed by scraping. Lysates were incubated on ice for 60 min and then centrifuged at 16,000 × g for 10 min at 4 °C. Samples were then resuspended in SDS-PAGE sample buffer and heated to 95°C for 5 min. Samples for immunoblotting were treated as described (11). Rabbit polyclonal phospho-specific antibodies directed toward Rad9A Ser²⁷² and Thr²⁹² are described by St. Onge et al. (7). Rabbit polyclonal phospho-specific antibodies directed toward Chk2 Thr⁶⁸ were obtained from Cell Signaling Technology (Beverly, MA), and a rabbit polyclonal antibody against the pluripotency marker Oct4 was obtained from Santa Cruz Biotechnology, Inc. (Santa Cruz, CA). Horseradish peroxidase-conjugated α-rabbit (Bio-Rad) secondary antibodies were used for Rad9A Ser²⁷², Thr²⁹², and Chk2 Thr⁶⁸ Western blots. Chicken polyclonal (IgY) antibodies directed against Rad9A have been described previously (6) and were detected through horseradish peroxidase-conjugated affiniPure rabbit α-chicken secondary antibodies (Jackson ImmunoResearch Laboratories, West Grove, PA). In Figs. 1, 2, and 5, images of immunoblots were captured using a Kodak Image Station 4000MM Pro. Two-tailed, paired t tests were performed on log-transformed control versus experimental samples.

FIGURE 1.

Inhibition of Topo II causes phosphorylation of Rad9A Ser²⁷². A, HeLa cells were treated with 0–50 μm ICRF-193 as indicated for 2 h. Cells were then harvested and prepared for flow cytometry using the mitosis marker phospho-histone H3. Percent of cells positive for phospho-histone H3 is plotted as a percent of untreated cells for each dose of ICRF-193. B and C, HeLa cells were left untreated or treated with 10 μm ICRF-193 for an increasing amount of time (0–6 h, B) or with an increasing concentration of ICRF-193 for 2 h (0–100 μm, C), and lysate was immunoblotted (IB) with phospho-specific antibodies directed toward Rad9A Ser²⁷². D, IMR-90 and A-T fibroblasts were left untreated (−), treated with 10 μm ICRF-193 (ICRF) for 2 h, or exposed to 10 Gy of IR and allowed to recover for 2 h. Cells were then lysed and immunoblotted with α-Rad9A Ser(P)²⁷² (S_p272) antibodies as in B. Rad9A Ser(P)²⁷² is denoted with an arrow.

FIGURE 2.

Chk2 Thr⁶⁸ is phosphorylated in response to ICRF-193. A, A-T cells were treated with ICRF-193 and IR for 2 h prior to harvesting and processing for flow cytometric analysis of phospho-histone H3. Phospho-histone H3-positive cells are plotted as a percent of undamaged control cells. B and C, HeLa cells were left untreated (−), exposed to 10 μm ICRF-193 for 2 h (+, C) or for an increasing amount of time (0–6 h, B) as indicated. Cell lysate was then immunoblotted (IB) with phospho-specific antibodies directed toward Chk2 Thr(P)⁶⁸ (T_p68), Chk1 Ser(P)³¹⁷ (S_p317), or Chk1 Ser(P)³⁴⁵ (S_p345). D, IMR-90 and A-T fibroblasts were left untreated, treated with 10 μm ICRF-193 (ICRF) for 2 h, or exposed to 10 Gy of IR and allowed to recover for 2 h. Cells were then lysed and immunoblotted with antibodies directed toward Chk2 Thr⁶⁸ antibodies as in B. Chk2 Thr(P)⁶⁸, Chk1 Ser(P)³¹⁷, and Chk1 Ser(P)³⁴⁵ are denoted with arrows.

FIGURE 5.

Phosphorylation of Rad9A Thr²⁹² is not required for the decatenation checkpoint, and phosphorylation of Rad9A Ser²⁷² and the Rad9A C terminus is not required for Chk2 Thr⁶⁸ phosphorylation induced by ICRF-193. A, IMR-90 fibroblasts (upper panel) and HeLa cells (lower panel) were grown in the presence (+) or absence (−) of 10 μm ICRF-193 and 0.1 μg/ml nocodazole (Noc) as indicated for 18 h. Cells were then harvested, and lysates were immunoblotted with phospho-specific antibodies directed toward Rad9A Thr(P)²⁹² (T_p292), denoted by an arrow. B, HeLa cells, Rad9A-S272A, and Rad9A-P5A cell lines were left untreated (−) or treated with 10 μm ICRF-193 (+) as indicated for 2 h. Cells were then harvested, and lysates were immunoblotted with phosphospecific antibodies directed toward Chk2 Thr(P)⁶⁸ (T_p68) (arrow).

Decatenation Checkpoint Experiments

Approximately 2–5 × 10⁶ HeLa cells and Rad9A phosphorylation mutant cells were grown asynchronously or were synchronized by double-thymidine block as described (27). Following a 4-h release into complete Dulbecco's modified Eagle's medium (before cells had cycled into G₂), synchronized cells were treated with 10 μm ICRF-193 and 0.1 μg/ml nocodazole (Sigma) as indicated for a further 4 h. Cells synchronized to mitosis and asynchronous cells and were then washed twice in PBS and harvested by trypsinization. Cells were washed in PBS twice more, and the cell pellet was resuspended in 200 μl of PBS. Ice-cold ethanol (5 ml) was added dropwise during gentle vortexing, and samples were incubated at 4 °C overnight. Cells were then washed twice with PBS and treated with 0.1 mg/ml RNaseA (BioShop) for 20 min. Approximately 1 × 10⁶ cells from each sample were incubated with 1 μg of α-phospho-histone H3 (Ser¹⁰, Upstate) diluted in 100 μl of block solution (0.5% (v/v) Tween 20/PBS containing 0.5% (v/v) normal goat serum) for 1 h at room temperature. Samples were then washed in 5 ml of PBS and resuspended in 0.1 μg/μl Alexa Fluor 488 goat α-rabbit diluted in 100 μl of block, for 30 min at room temperature. Samples were then washed in 3 ml of PBS and resuspended in 80 μg/ml propidium iodide (CedarLane, Hornby, Canada) diluted in PBS. This protocol was largely based on that described in Marples et al. (28). Samples were analyzed using a flow cytometer (Beckman/Coulter EPICS ALTRA, Mississauga, Canada).

Cell Cycle Analysis

HeLa cells and Rad9A phosphorylation mutant cell lines were grown in the presence or absence of 10 μg/ml ICRF-193 for 24 h, and 10 μl of Cell Proliferation Labeling reagent (Amersham Biosciences) was added for the final 30 min. Cells were then harvested, fixed in ethanol, counted, and treated with RNaseA as described above. Cells were incubated in DNA denaturation solution (0.5% (v/v) Triton X-100, 4 n HCl) at room temperature for 30 min. Cells were then resuspended in 0.1 m sodium borate, pH 8.5, to neutralize the acid and washed in wash solution (0.5% (v/v) Tween 20, PBS). Fluorescein isothiocyanate-conjugated anti-BrdUrd (20 μl; BD Biosciences) was added to the cell pellet for 30 min at room temperature. Cells were washed in 3 ml of PBS and resuspended in PBS containing 80 μg/ml propidium iodide. Samples were analyzed by flow cytometry (Beckman/Coulter EPICS ALTRA).

Mitotic Spreads

Undifferentiated and differentiated mES cells were treated with 40 μg/ml colecmid for 4 h prior to harvest by trypsinization. Where indicated, cells were also treated with various concentrations of ICRF-193 for the same length of time. Cells were washed once with PBS, resuspended in 1 ml of warmed 75 mm KCl, and incubated at 37 °C for 10 min. Cells were pelleted via centrifugation for 10 min and fixed dropwise in 1 ml of very cold 3:1 (v/v) methanol:acetic acid. Nine milliliters of fix was added, and cells were incubated for 20 min at room temperature. Cells were pelleted via centrifugation and washed twice more in 10 ml of very cold fix. Cells were resuspended in 0.5–1 ml of cold fix and dropped onto very cold, clean glass slides. Slides were incubated at 37 °C for 1–2 h and then mounted in Antifade Gold supplemented with 4′,6-diamidino-2-phenylindole (Invitrogen). Spreads were examined via 100× objective mounted on an Olympus IX51 inverted microscope and imaged with a Retiga 2000R digital capture system (Olympus, Markham, Canada). Catenated versus normal mitotic spreads were scored as described previously (29).

RESULTS

Inhibition of Topo II by ICRF-193 Causes ATM-dependent Phosphorylation of Rad9A and Chk2

Because Topo-binding protein 1 interacts with the regulatory region of Topo II, and Topo-binding protein 1 in turn binds Rad9A (11), we investigated whether Rad9A was monitoring this pathway. HeLa cells treated with ICRF-193 for 2 h demonstrate mitotic delay (Fig. 1A). To evaluate the phosphorylation status of Rad9A in decatenation checkpoint-arrested cells, cell lysates were immunoblotted with phospho-specific antibodies directed toward Rad9A Ser²⁷². Rad9A Ser²⁷² phosphorylation was evident in cells that were treated with ICRF-193, but not in undamaged cells (Fig. 1B, denoted by arrow). Phosphorylation at Rad9A Ser²⁷² was detectable after cells had been incubated with ICRF-193 for 1 h and persisted throughout 6 h of treatment. Maximal induction of Ser²⁷² phosphorylation was seen 4 h after treatment with ICRF-192 and represented an 8× increase over untreated cells (n = 3, p = 0.02; Fig. 1B). Similarly, a range of ICRF-193 concentrations from 5 to 100 μm resulted in the phosphorylation of Rad9A Ser²⁷² (Fig. 1C). Treatment with 10 μm ICRF-193 for 2 h led to a 6-fold increase in Ser²⁷² phosphorylation over untreated cells (n = 4, p = 0.01; Fig. 1C). Therefore, inhibition of Topo II by ICRF-193 results in Ser²⁷² phosphorylation of Rad9A, suggesting that Rad9A may be participating in the decatenation checkpoint.

Ser²⁷² is the major site of Rad9A phosphorylation by ATM in response to IR (6, 10). To determine whether ATM also phosphorylates Rad9A at Ser²⁷² when stimulated by Topo II inhibition, we evaluated the status of Rad9A Ser²⁷² phosphorylation in A-T cells, which lack functional ATM. Normal diploid fibroblast (IMR-90) and A-T fibroblast cells were exposed to Topo II inhibition by ICRF-193 or, as a control, IR. Relative to untreated IMR-90 cells, there was a increase in Rad9A Ser²⁷² phosphorylation following either ICRF-193 treatment (14-fold increase, n = 2, p = 0.05) or exposure to 10 Gy of IR (20-fold increase, n = 2, p = 0.01). By contrast, the same treatments do not lead to any detectable Rad9A Ser²⁷² phosphorylation in A-T fibroblast cells (Fig. 1D). Thus, activation of the decatenation checkpoint causes phosphorylation of Rad9A at Ser²⁷² is that is dependent on ATM.

ATM is not strictly required for the decatenation checkpoint because A-T cells demonstrate an effective mitotic block when treated with ICRF-193 (20). We have also confirmed this finding (Fig. 2A). More recently, however, cancer cell lines with defective ATM expression were shown to be hypersensitive to ICRF-193, suggesting that ATM may be involved in sensing this type of damage (21). Therefore, these results suggest that ATM and Rad9A may be playing a role distinct from checkpoint-mediated cell cycle arrest in response to Topo II inhibition.

Although the decatenation checkpoint was shown to be dependent on ATR, it was hypothesized that this pathway did not signal through the usual checkpoint effector kinases because neither Chk1 nor Chk2 was observed to be phosphorylated in response to ICRF-193 (20). However, because phosphorylation of Rad9A Ser²⁷² is dependent on ATM in response to DNA damage and ICRF-193 treatment, we decided to examine Chk2 Thr⁶⁸. Chk2 Thr⁶⁸ is a key ATM target that, like Rad9A, is also phosphorylated by ATM in response to DNA damage (12, 30, 31). Fig. 2B clearly demonstrates that cell lysates exposed to ICRF-193 do indeed contain phosphorylated Chk2 Thr⁶⁸ and that this phosphorylation is persistent after 6 h of ICRF-193 treatment with a 4-fold increase in band intensity compared with untreated samples (n = 3, p = 0.02). These results differ from previous observations in which Chk2 was not seen to be phosphorylated in response to ICRF-193 treatment (20). It is likely that our observations regarding Chk2 Thr⁶⁸ phosphorylation differ from that of earlier conclusions because of our use of a phospho-specific antibody that detects a Chk2 Thr⁶⁸ phospho-specific band that is difficult to detect by mobility shift analysis alone. In agreement with earlier results, neither Chk1 Ser³¹⁶ (1.5-fold increase, n = 2, p = 0.5) nor Ser³⁴⁵ (1.5-fold increase, n = 2, p = 0.5) showed significantly increased phosphorylation following ICRF-193 treatment (Fig. 2C). ICRF-193 treatment also stimulated a modest Chk2 Thr⁶⁸ phosphorylation in normal human fibroblasts compared with untreated controls (4-fold increase, n = 3, p = 0.1), but no phosphorylation was detectable in A-T fibroblasts (Fig. 2D), suggesting that ATM is also required for Chk2 phosphorylation under these conditions. It should be noted that despite the lack of a Chk2 mobility shift in cells treated with ICRF-193, there is an easily detectable Chk2 mobility shift observed when IMR-90 cells were treated with IR, suggesting the additional posttranslational modifications of Chk2 occur only after IR.

Constitutive Phosphorylation of Rad9A Is Required for the Decatenation Checkpoint

To confirm that ICRF-193 treatment arrests cells in G₂, cells were synchronized to late S (4-h release from double-thymidine block) and treated with ICRF-193, and/or nocodazole, which blocks cells in mitosis, for a further 4 h. Phospho-histone H3 cell cycle analysis of synchronized cells illustrated that cells were arrested in mitosis following nocodazole treatment but were arrested in G₂ following incubation with ICRF-193 alone or in combination with nocodazole (Fig. 3A). This confirms that ICRF-193 arrests cells in G₂ as they progress through the cell cycle, prior to mitotic entry defined by phospho-histone H3 and prior to mitotic checkpoint arrest. We then went on to analyze the effect of ICRF-193 treatment in cell lines stably transfected with Myc-tagged constructs of Rad9A-WT, the Rad9A-S272A mutant, and the Rad9A-constitutive phosphorylation site mutant P5A (S277A, T292A, S328A, S336A, T355A) (6). We analyzed cells during M phase, when the highest numbers of cells are positive for phospho-histone H3 staining. Analysis of phospho-histone H3 by flow cytometry demonstrated that ICRF-193 induced mitotic delay in synchronized Rad9A-S272A mutant cells which was not significantly different from that of untransfected cells or Rad9A-WT cells (Fig. 3B). These results are consistent with our above conclusions stating that phosphorylation of Ser²⁷² is not required for the decatenation checkpoint. However, cells expressing the Rad9A-constitutive phosphorylation site mutant P5A abrogated the decatenation checkpoint response. Untransfected cells and Rad9A-P5A mutant cells treated with ICRF-193 were inhibited to 64 ± 16% and 108 ± 16% of untreated cells, respectively (Fig. 3B, p < 0.05). The increased number of Rad9A-P5A mitotic cells in the ICRF-193-treated samples is likely because the cells proceed through mitosis at a reduced rate when chromosomes are tangled (32). These results suggest that the constitutive phosphorylation sites of Rad9A are required for cell cycle arrest mediated by the decatenation checkpoint.

FIGURE 3.

Constitutive Rad9A phosphorylation sites are required for the decatenation checkpoint. A, HeLa cells were synchronized by double-thymidine block and released for 4 h. 10 μm ICRF-193, 0.1 μg/ml nocodazole (Noc), or both, were added as indicated, and cells were allowed to grow for a further 4 h until mitosis. Cells were then harvested and processed for flow cytometry of phospho-histone H3, as described under “Experimental Procedures.” Shown here is the cellular PI content versus phospho-histone H3. Percent refers to number of phospho-histone H3-positive cells gated in ellipse. B, untransfected cells (HeLa) and stable cell lines expressing Rad9A-WT (WT), Rad9A-S272A (S272A), and Rad9A-P5A (P5A) mutants were synchronized by double-thymidine block and released into late S/G₂ phase (6-h release from double-thymidine block). Cells were then left untreated (gray bars) or treated with 10 μm ICRF-193 for 2 h (black bars) prior to harvesting and processing for flow cytometry of phospho-histone H3. Phospho-histone H3-positive cells are plotted as a percent of undamaged cells. Error bars represent the S.D. from four independent experiments (HeLa), two independent experiments (WT and S272A), and three independent experiments (P5A). Asterisk indicates significant difference (p < 0.05) compared with untransfected cell controls.

Constitutive Phosphorylation of Rad9A Is Required to Prevent Endoreduplication in the Presence of ICRF-193

We next examined the cell cycle distribution of Rad9A cell lines that had been exposed to prolonged treatments of ICRF-193. HeLa, Rad9A-P5A, and Rad9A-S272A cells were exposed to ICRF-193 for 24 h and labeled with BrdUrd and propidium iodide for analysis by flow cytometry. In all cases, untreated cells displayed a normal cell cycle distribution, with the majority of cells in G₁ phase (Fig. 4). Treatment with ICRF-193 caused essentially all of the cells to accumulate in G₂/M, suggesting that inhibition of Topo II causes cells to arrest in G₂, as expected. Strikingly, 17.5% of Rad9A-P5A cells stained positively for BrdUrd with a >4 n DNA content, indicated by shifted PI staining. This suggests that these cells advanced into S phase without completing mitosis because sister chromatids remain catenated in the presence of ICRF-193. This phenotype is characteristic of cells exhibiting endoreduplication. Although there were a few HeLa and Rad9A-S272A cells that also stained positively for BrdUrd with a >4 n DNA content, there was a striking increase in the extent of the endoreduplication-like phenotype in Rad9A-P5A cells.

FIGURE 4.

Mutation of Rad9A constitutive phosphorylation sites induces an endoreduplication-like phenotype in prolonged ICRF-193 treatment. Untransfected, Rad9A-P5A, and Rad9A-S272A cells were untreated or treated with 10 μm ICRF-193 for 24 h. BrdUrd labeling reagent was added for the final 30 min, and cells were fixed and stained with α-BrdUrd fluorescein isothiocyanate and propidium iodide. Gated populations represent cells that stain positively for BrdUrd with a 4 n or greater DNA content. This experiment is representative of three independent experiments.

The Rad9A-P5A mutant cell line expresses Rad9A constructs that harbor mutations at Thr²⁹². We have previously demonstrated that this residue is phosphorylated during mitosis in a Cdc2-dependent manner (7). We wondered whether the phenotype observed in the Rad9A-P5A mutant cell lines could be attributed to the mutation at Thr²⁹² alone because the cells demonstrated attenuated G₂ arrest and entered into S phase without progressing through mitosis. To address this question, IMR-90 and HeLa cells were exposed to ICRF-193 and/or nocodazole as indicated for 18 h (Fig. 5A). Immunoblotting with phospho-specific antibodies raised against Rad9A Thr²⁹² revealed that phosphorylation of Thr²⁹² is apparent only after nocodazole treatment (7-fold increase, n = 2, p = 0.05) and not after ICRF-193 alone (0.7-fold increase, n = 2, p = 0.7) or in combination with both drugs (0.9-fold increase, n = 2, p = 0.7). This indicates that Thr²⁹² is not phosphorylated at, and therefore not required for, the decatenation checkpoint.

To address the significance of the Rad9A-P5A mutation on Chk2 Thr⁶⁸ phosphorylation induced by ICRF-193 treatment, HeLa, Rad9A-S272A, and Rad9A-P5A cells were treated with ICRF-193, and lysates were analyzed for Chk2 phosphorylation status by immunoblotting. Fig. 5B demonstrates that Rad9A-S272A and -P5A mutants have no effect on the ICRF-193-induced phosphorylation of Chk2 Thr⁶⁸ (n = 3, p = 0.01). This suggests that Rad9A is not required for phosphorylation of Chk2 Thr⁶⁸ in response to catenated chromosomes and is consistent with ATM phosphorylation of Chk2 being independent of the 911 complex (33). Because cell cycle arrest mediated by the decatenation checkpoint involves nuclear exclusion of Cyclin B1, Rad9A may be instead operating upstream of this event.

Deletion of Rad9A Results in Higher Levels of Catenated Mitotic Spreads

To examine DNA quality in the presence and absence of Rad9A, mES cells were prepared for mitotic spread. Due to the lack of decatenation checkpoint in undifferentiated mES cells (29), differentiation was induced via exposure to retinoic acid for 48 h. For each sample, more than 200 mitotic cells were scored as having either a normal spread (Fig. 6A) or an aberrant spread (Fig. 6B) and plotted in relation to the concentration of ICRF-193 for both undifferentiated and differentiated cells (Fig. 6C). Cells lacking Rad9A were sensitive to very small amounts of ICRF-193, producing a higher level of aberrant mitotic spreads compared with WT cells. Differentiation was confirmed via immunoblotting against Oct4, a well known marker for pluripotency (Fig. 6D) (34). Fig. 6C demonstrates the level of catenated chromosomes is higher in cells deficient in Rad9A and suggest a critical role for Rad9A in the decatenation checkpoint during resolution of catenated chromosomes (see Fig. 7).

FIGURE 6.

Rad9A depletion results in increased levels of aberrant mitotic spreads in differentiated murine embryonic stem cells after exposure to ICRF-193. Cells were exposed to the indicated concentration of ICRF-193 for 4 h and prepped for mitotic spreads as described under “Experimental Procedures.” A, normal metaphase spread. B, catenated metaphase spread. C, >200 metaphase spreads analyzed/sample and the number of catenated spreads plotted versus ICRF-193 concentration. Error bars represent S.D. for two independent trials. D, immunoblot (IB) against Oct4 of whole cells lysates from both WT and Rad9A-nulls cells confirming differentiation.

FIGURE 7.

Role of Rad9A in the G₂ decatenation checkpoint. Inhibition of Topo II by ICRF-193 prevents catenated chromosomes from being resolved following replication. Catenated chromosomes induce a G₂ checkpoint delay, mediated by ATR. ATR acts upstream of both BRCA1 and Plk1 (not illustrated) in a pathway that results in exclusion of Cyclin B1 from the nucleus and delay in mitotic progression. Rad9A may play a role in this pathway as a substrate of ATR or through interactions with Topo-binding protein 1 (TopBP1), which also binds Topo II. ATM is also activated, as indicated by autophosphorylation at Ser¹⁹⁸¹. ATM phosphorylates Rad9A Ser²⁷² and Chk2 Thr⁶⁸ in response to catenated chromosomes. Rad9A and Chk2 both function to prevent re-replication and cut-like phenotypes, and ATM may function in a redundant manner to prevent cell cycle progression in cancer lung cells.

DISCUSSION

We have shown here that Rad9A is part of the human decatenation checkpoint. The decatenation checkpoint is attenuated in differentiated Rad9A-null cells, as demonstrated by the high number of aberrant mitotic spreads (Fig. 6C). Furthermore, we show that the C terminus of Rad9A is required for this checkpoint because cell lines expressing Rad9A-P5A fail to show mitotic delay when Topo II is inhibited by ICRF-193 (Fig. 3B), and these cells exhibit endoreduplication in the prolonged presence of catenated chromosomes (Fig. 4). Our results suggest that Rad9A and the 911 complex function to preserve genomic integrity in the presence of catenated chromosomes.

We have demonstrated that Rad9A is phosphorylated in an ATM-dependent manner at Ser²⁷² in response to ICRF-193 treatment (Fig. 1D). Notably, ATR was not able to compensate for the lack of ATM-dependent phosphorylation at this site. Rad9A Ser²⁷² is located within an ATM consensus sequence (S/TQ) and is also phosphorylated by ATM in response to IR (6, 10). However, neither the DNA damage checkpoint nor the decatenation checkpoint appears to require phosphorylation at Rad9A Ser²⁷² despite the requirement of ATM in the former but not the latter. This may be because Rad9 Ser²⁷² is overexpressed in our cell lines. However, phosphorylation of Rad9A at Ser²⁷² is only required for activation of the G₁/S checkpoint (10), therefore this site may be involved in some other aspect of damage response in G₂, such as maintenance of the checkpoint.

Alternatively, phosphorylation at Rad9A Ser²⁷² may not be absolutely required for checkpoint signaling because there are redundant pathways to compensate for this loss. Recent evidence has suggested that in response to DNA cross-linking agents, the S phase checkpoint bifurcates into two arms stemming from ATR: ATR-Chk1 and ATR-NBS1-FANCD2 (35). Knockdown of Chk1 expression in WT cells partially arrested S phase cells in response to cross-linking agents, but abrogated the S phase checkpoint in NBS1-deficient cells (35). Therefore, cells overexpressing Rad9A-S272A mutations may not demonstrate a checkpoint-defective phenotype in the presence of an intact redundant signaling pathway.

Topo II inhibition by ICRF-193 also causes ATM-dependent phosphorylation of Chk2 Thr⁶⁸ (Fig. 2D). It is possible that ATM-dependent phosphorylation of Rad9A and Chk2 occurs simply because in addition to producing catenated chromosomes, ICRF-193 is also causing double-strand breaks. Although it has been reported that ICRF-193 treatment can cause double-strand breaks (36), numerous studies have demonstrated that the number of double-strand breaks caused by bisdioxopiperazines is significantly less than those generated by Topo II poisons, such as etoposide (37–39), and that ICRF-193 does not cause double-strand breaks in HeLa cells (32). Additionally, we (Fig. 2C) and others (20) have shown that Chk1 Ser³¹⁷ and Ser³⁴⁵ are not phosphorylated in cells treated with ICRF-193. ICRF-193 does cause ATM Ser¹⁹⁸¹ autophosphorylation and activation, and lung cancer cells lacking functional ATM are hypersensitive to ICRF-193 (21). This suggests that ATM is playing a redundant and cooperative role with ATR in the decatenation checkpoint (21). By extrapolating from these findings, a possible resolution of these data is that ATM activation of Chk2 is required for activation of p53, to preserve chromosomal stability and to prevent polyploidy. Rad9A-S272A and Rad9A-P5A cells showed normal induction of Chk2 Thr⁶⁸ in response to ICRF-193 treatment (Fig. 5B), therefore phosphorylation of this site does not likely require Rad9A regulation. Similarly, Rad9A is not required for Chk2 activation in response to IR, UV, and hydroxyurea (8).

The constitutive phosphorylation sites of Rad9A are, however, required for the decatenation checkpoint. Cells expressing Rad9A mutants lacking the constitutive phosphorylation sites abrogated the mitotic delay induced by Topo II inhibition (Fig. 3B). This result is consistent with the requirement of the Rad9A constitutive phosphorylation sites for IR-induced prolonged G₂/M arrest (7). Additionally, mutation of all known Rad9A C-terminal phosphorylation sites causes a defect in checkpoint response to UV and hydroxyurea treatment, as well as sensitivity to the Topo II poison etoposide (8, 9). Loegering et al. (9) demonstrated that cells expressing a Rad9A C-terminal nonphosphorylatable mutant ablated the S phase checkpoint to cytarabine, which is also a catalytic inhibitor of Topo II. Thus, the constitutive phosphorylation sites of Rad9A are indispensable for checkpoint activation when challenged by a variety of genotoxins causing DNA damage and Topo II decatenation inhibition and support a role for Rad9A as a general sensor of genomic instability.

Prolonged inhibition of Topo II decatenation activity causes a phenotype consistent with endoreduplication in cell lines expressing Rad9A-constitutive phosphorylation site mutants (Fig. 4). Although the majority of Rad9A-P5A cells arrest at G₂/M after 24 h of ICRF-193 treatment, reentry into replication is observed ∼17% of cells in the absence of chromosomal segregation, resulting in S phase cells with a >4 n DNA content. ICRF-193 does not affect exit from mitosis, nor does it interfere with S phase (40), although it does prolong the second S phase when mitosis has not intervened (41). Thus, some Rad9A-P5A cells continue through an aberrant mitosis and re-replicate chromosomes in the absence of mitotic segregation. Infrequently, we observed cells that had a full 8 n profile, but more often prolonged exposure to ICRF-193 resulted in cell death, as the number of cells surviving 48- and 72-h treatments was greatly reduced, independently of Rad9A status.

Originally we hypothesized that the Rad9A-P5A endoreduplication phenotype could be due to mutation at Thr²⁹² alone, as this site is specifically phosphorylated during mitosis (7). We hypothesized that if Thr²⁹² were required for mitotic progression, then mutation at Thr²⁹² would produce the same phenotype, regardless of the type of damage. However, as shown in Fig. 5A, Thr²⁹² is not required for the decatenation checkpoint, as it is not phosphorylated in response to ICRF-193 treatment. This is likely because these cells arrest in G₂ before mitosis is initiated (Fig. 3A).

Finally, analysis of metaphase spreads from differentiated mES cells treated with ICRF-193 reveals that Rad9A-deficient cells show higher levels of catenated mitotic spreads compared with WT cells (Fig. 6C). Differentiation was required to reinstate the decatenation checkpoint, which is suspended in embryonic stem cells, possibly hastening the cell cycle during embryonic development (29). The Rad9A-null cells also show greater sensitivity to ICRF-193 compared with WT cells as 0.05 μg/ml ICRF-193 caused an increase the number the of aberrant spreads (Fig. 6C). In contrast, 0.20 μg/ml ICRF-193 is required to cause similar levels of aberrant mitotic spreads in the WT cells (Fig. 6C). This indicates that in the presence of Topo II inhibition Rad9A-nulls cells are not able to resolve catenated chromosomes. Interestingly, mitotic arrest by colcemid also resulted in significant levels of catenated chromosomes in Rad9A-null cells, suggesting that progression through the cell cycle alone is sufficient to result in catenated DNA. Taken together, these data suggest a role for Rad9A and the 911 complex in the detection or resolution of chromosomal catenation.