Elevated Cyclin G2 Expression Intersects with DNA Damage Checkpoint Signaling and Is Required for a Potent G2/M Checkpoint Arrest Response to Doxorubicin*

Background: DNA damage triggers cell cycle checkpoints to halt cell division ahead of DNA repair. Results: Ectopic cyclin G2 (CycG2) induces a Chk2-dependent cell cycle arrest, and depletion of endogenous CycG2 attenuates doxorubicin-induced G2/M-phase cell cycle arrest. Conclusion: CycG2 influences checkpoint signaling and is required for G2/M arrest responses to genotoxic stress. Significance: Proper checkpoint function is important for genomic integrity and tumor suppression. To maintain genomic integrity DNA damage response (DDR), signaling pathways have evolved that restrict cellular replication and allow time for DNA repair. CCNG2 encodes an unconventional cyclin homolog, cyclin G2 (CycG2), linked to growth inhibition. Its expression is repressed by mitogens but up-regulated during cell cycle arrest responses to anti-proliferative signals. Here we investigate the potential link between elevated CycG2 expression and DDR signaling pathways. Expanding our previous finding that CycG2 overexpression induces a p53-dependent G1/S phase cell cycle arrest in HCT116 cells, we now demonstrate that this arrest response also requires the DDR checkpoint protein kinase Chk2. In accord with this finding we establish that ectopic CycG2 expression increases phosphorylation of Chk2 on threonine 68. We show that DNA double strand break-inducing chemotherapeutics stimulate CycG2 expression and correlate its up-regulation with checkpoint-induced cell cycle arrest and phospho-modification of proteins in the ataxia telangiectasia mutated (ATM) and ATM and Rad3-related (ATR) signaling pathways. Using pharmacological inhibitors and ATM-deficient cell lines, we delineate the DDR kinase pathway promoting CycG2 up-regulation in response to doxorubicin. Importantly, RNAi-mediated blunting of CycG2 attenuates doxorubicin-induced cell cycle checkpoint responses in multiple cell lines. Employing stable clones, we test the effect that CycG2 depletion has on DDR proteins and signals that enforce cell cycle checkpoint arrest. Our results suggest that CycG2 contributes to DNA damage-induced G2/M checkpoint by enforcing checkpoint inhibition of CycB1-Cdc2 complexes.

Genomic DNA is continually subject to lesions induced by environmental radiation, chemical carcinogens, and reactive oxygen species generated by cellular metabolism (1). If damage to chromosomal DNA is not corrected, these insults will lead to genomic instability and cancer. The presence of a lesion is relayed within minutes of the genomic insult through DNA damage response (DDR) 5 signal-transduction pathways. Signaling cascades including sensor, transducer, and effector proteins carry out a particular response (e.g. induction of cell-cycle arrest, DNA repair or apoptosis) dependent on the type and extent of the damage. Damage sensors initiate distinct DDR signaling pathways to coordinate activation of one of the phosphoinositide 3-kinase-related kinases that plays central roles in maintenance of organismal longevity, ataxia telangiectasia mutated (ATM), ATM and Rad3-related (ATR), and DNA-dependent protein kinase (DNA-PK) (1,2). ATM kinase activation is primarily stimulated by blunt double-stranded DNA (dsDNA) ends such as the DNA double-strand breaks (DSBs) incurred through ␥-irradiation (3), whereas ATR activation is most responsive to single-stranded DNA (ssDNA) like that presented by stalled DNA replication intermediates or resected DSB ends (4). DNA-PK is a critical participant in the non-homologous end-joining pathway for repair of V(D)J recombination-induced DSBs but is also thought to serve a vital DNA repair function during genotoxic stress DDRs (5). However, growing evidence suggests that extensive cross-talk between the DNA damage-responsive phosphoinositide 3-kinase-related kinases exists, the summation of which determines cell fate (4 -6).
DNA DSBs pose the most significant problem for maintenance of genomic stability. ATM is critical for the initial response to DSBs (3). The Mre11-Rad50-Nbs1 sensor complex (MRN (Mre11-Rad50-Nbs1) complex) promotes ATM activation and recognition of DSBs (3). It facilitates trans-autophosphorylation of inactive ATM dimers on Ser-1981 and thereby ATM dissociation into catalytically active monomers (3). Activated ATM interacts with and phosphorylates numerous proteins to amplify and propagate the signal. Studies indicate ATR is also activated by DSBs and plays a role in the later phase of the response, the progressive resection of blunt end DSB junctions to single strand ends ultimately triggering ATR activation (4,7). Once activated, ATM and ATR phospho-activate their respective target checkpoint kinases, Chk2 and Chk1. Chk1 and Chk2 in turn phosphorylate and modulate the activity of downstream effectors (Cdc25s A, B, and C; p53) to ultimately halt progression of cells through G 1 -and G 2 -phase checkpoints (3,4,6,8). This blockade of cellular proliferation allows DNA repair to proceed, but if the DNA damage is irreparable, cell death via apoptosis will ensue.
CCNG2 encodes cyclin G2 (CycG2), an unconventional cyclin homolog linked to cell cycle inhibition (9 -17). CCNG2 mRNAs are moderately expressed in proliferating cells (peaking during the late S/early G 2 phase) (9,11,12) but significantly up-regulated as cells exit the cell cycle in response to receptormediated negative signaling in B-lymphocytes and ovarian cancer cells (12,17). Transcript data from a variety of studies indicate that CCNG2 expression is up-regulated during cell cycle arrest responses to diverse growth-inhibitory signals and strongly repressed by mitogens, suggesting a positive role for CycG2 in the promotion or maintenance of cell cycle arrest (12, 18 -24). CCNG2 transcripts are also increased in cells treated with the DNA damaging chemotherapeutics actinomycin D and ecteinascidin-743 (25,26). In contrast to CCNG1 (the gene encoding the CycG2 homolog CycG1 (27,28)), CCNG2 does not contain p53 binding sites (29), but recent work showed that CCNG2 is a transcriptional target of the p53 homolog, p63 (30). Importantly, suppressed CCNG2 mRNA expression has been linked to cancer, including thyroid, oral, and breast carcinomas (14,30,31).
In previous work we determined that ectopic CycG2 expression inhibits DNA synthesis and induces a G 1 /S-phase arrest in a variety of cell lines (13,15,16). We showed that overexpression of CycG2 inhibits CDK2 activity and that the CycG2-mediated G 1 /S-phase cell cycle arrest is p53-dependent (13,15). Subsequent studies determined that even moderate up-regulation of ectopic CycG2 expression inhibits cellular proliferation (10,16,24). We found that exogenous and endogenously expressed CycG2 is a CRM1-dependent nucleocytoplasmic shuttling protein that localizes to the cytoplasmic-cytoskeletal compartment of replicating cells where it associates with centrosomes via AKAP450 (15). Here we examine CycG2 expression during cellular responses to treatment with the chemotherapeutic DNA DSB-inducing topoisomerase II poisons (32), etoposide, and doxorubicin. We relate changes in CycG2 expression to the effects doxorubicin treatment has on cell cycle progression and induction of phospho-activated forms of ATM/ATR pathway DDR proteins. By using transient overexpression of recombinant CycG2 and shRNA-mediated RNAi to knockdown endogenous CycG2, we investigate the involvement of CycG2 with DDR signaling pathways and its contribution to DNA damage-induced cell cycle arrest.
Immunofluorescence Microscopy-MCF10a cells were seeded at 1.5 ϫ 10 5 cells/35-mm well onto a 22-mm-square glass coverslip coated with 10 mg/ml collagen and 1 g/ml poly-L-lysine (Sigma) 14 -18 h before treatment. Coverslips were removed 16 h after treatment, rinsed with PBS, and immediately fixed with ice-cold MeOH at Ϫ20°C for 5 min. Specimens were stained and mounted, and images were collected by confocal microscopy as described (13).
Cell Cycle Analysis by Flow Cytometry-DNA content in untransfected cell cultures and stable MCF7 clones was assessed after fixation of cells with Ϫ20°C 70% EtOH. Washed pellets of fixed cells were resuspended in PBS containing 0.25 mg/ml RNase A (Fermentas) and 50 g/ml propidium iodide (Sigma) for 30 min at room temperature before DNA flow cytometry using a FACScan (BD Biosciences) as described (11)(12)(13). For cell cycle analysis of propidium iodide-stained DNA in GFP-expressing populations of transiently transfected HCT116, GM05849, and GM00637 cells, the GFP signal was retained by an initial 10-min fixation in PBS containing 0.5% paraformaldehyde and 10 mM EDTA before permeabilization and final fixation with Ϫ20°C 100% methanol as described (13).
In some experiments total DNA in live cells transiently expressing fluorescent marker proteins (e.g. GFP or RFP) was stained with Hoechst 33342, and DNA content in the unfixed fluorescent and non-fluorescent cell populations was assessed via flow cytometry using a quadruple laser LSR II flow cytometer (BD Biosciences) as described (13,15). In all cases assessment of DNA content distribution and cell cycle analysis was done using FlowJo 8.5 software. For statistical analysis, t tests and one-way analysis of variance tests (one-way ANOVA with the Tukey and Bonferroni post hoc tests) were done using Prism 4.0 software (GraphPad Software, Inc.). For indicated experiments, TOPRO negative cells were sorted on the basis of GFP expression with a MoFlo cell sorter (Beckman Coulter).

RESULTS
Cyclin G2-induced Cell Cycle Arrest Requires p53 and Chk2 but Is Only Partially p21-dependent-We reported that CycG2 induces a p53-dependent cell cycle arrest in HCT116 cells (15).
Here we examined the effect of ectopic GFP-tagged CycG2 versus GFP expression on cell cycle progression in HCT116 cells nullizygous for the DNA damage checkpoint protein Chk2 and compared the effects to those observed in similarly transfected p53 null, p21 null, and wild-type (WT) cells. As anticipated, ectopic expression of GFP alone had no discernable effect on cell cycle progression in any of the isogenic cell lines, the cell cycle profile of each transfected population being similar to the non-expressing controls (Fig. 1, A and C). Multiple experimental repeats indicated that the G 1 /S-phase cell cycle arrest induced by ectopic CycG2 expression requires both the presence of Chk2 and p53 (p values Ͻ0.001), whereas loss of the p53 target gene p21 had only a moderate effect on CycG2 inhibitory activity ( Fig. 1, C and D). As Chk2/p53 checkpoint signaling is triggered by DNA damage-activated ATM, we tested whether ATM is required for the G 1 /S-phase cell cycle arrest induced by ectopic CycG2 expression (Fig. 1, B and C). Incubation of CycG2GFP-transfected cells with 10 M of the ATM inhibitor KU55933 (34) did not block the CycG2-induced cell cycle arrest of WT HCT116 cells (Fig. 1, B and C). Moreover, in contrast to GFP alone, expression of GFP-tagged CycG2 in the ATM null cell line GM05849 triggered a similar decrease in the proportion of cells in S-phase and an increase in those in G 1 phase (p values Ͻ0.01 and 0.001, respectively, Fig. 1, B and C, and supplemental Fig. S3). Together these results suggest that ectopic CycG2 expression induces a Chk2-and p53-dependent but ATM-independent G 1 -phase cell cycle arrest.
Ectopic Cyclin G2 Expression Induces Activation of Checkpoint Kinase Chk2-We evaluated whether ectopic expression of CycG2 modulates Chk2 and other DDR signaling proteins. HCT116 wild-type and isogenic p53-null cells were transfected with either GFP or CycG2GFP expression vectors and assessed for expression of phospho-activated Chk2 and Chk1 ( Fig. 2A). HCT116 cells treated with doxorubicin served as positive controls for induction of pChk1(Ser-345) and pChk2(Thr-68). We found in reproducible experiments that pChk2(Thr-68) expression was elevated in CycG2GFP compared with GFP-transfected cell lysates ( Fig. 2A). Interestingly, pChk2(Thr-68) levels were most prominent in the p53 null cultures expressing CycG2GFP. In contrast to the pChk1(Ser-345) levels in doxorubicin treated cells, pChk1(Ser-345) expression was undetectable in transfected p53 null and WT HCT116 cell lysates. As p53 activation is downstream of Chk2 and promotes both cell cycle arrest and cell death, this finding suggests that CycG2GFP expression and the concomitant induction of Chk2 activation is better tolerated in p53 null HCT116 cells.
To further investigate this issue and control for differences in transfection efficiency, we repeated the immunoblot analysis on FACS-sorted cell populations of similarly transfected WT and p53 null HCT116 cultures (Fig. 2B). As before, the nonexpressing populations isolated from CycG2GFP-transfected cultures did not show a modulation of phospho-Chk2 or phospho-Chk1 levels. However, lysates isolated from the sorted CycG2GFP-positive populations of both WT and p53 null cultures contained strongly increased levels of pChk2(Thr-68) and moderately elevated pNbs1(Ser-343) expression (Fig. 2B). Analogous to results shown in 2A, lysates of the CycG2GFP populations did not contain elevated levels of phospho-activated Chk1 (here pChk1(Ser-296); Fig. 2B). Repeated sorting experiments showed similar results and verified the specificity of the pChk2(Thr-68) immunosignal ( Fig. 2C and supplemental Fig. S4). Again, pChk2(Thr-68) and pNbs1(Ser-343) levels were enhanced in CycG2GFP-transfected WT HCT116, but as expected pChk2(Thr-68) was absent in Chk2 null cell lysates (Fig. 2C). Moreover, as seen with lysates from unsorted GFPtransfected controls, GFP expression did not modulate pChk2(Thr-68) or pChk1(Ser-345) expression levels (supplemental Fig. S4). Similar results were found for lysates of unsorted U2OS cells transiently transfected with CycG2GFP expression constructs (supplemental Fig. S4). Together our results show that ectopic up-regulation of CycG2 levels triggers signals that induce expression of phospho-activated forms of Chk2 and Nbs1 but not phospho-Chk1.
Endogenous Cyclin G2 Is Up-regulated during DNA Damage Responses Induced by Topoisomerase II Inhibitors and Accumulates in Nuclei of Doxorubicin-treated Cells-Our observations prompted us to investigate CycG2 expression in cells initiating checkpoint signaling in response to chemotherapeutic agentinduced dsDNA breaks. Exposure of the immortalized nontransformed breast epithelial cell line MCF10a to either doxorubicin or etoposide up-regulated CycG2 expression up to 5-fold within the first 4 h of treatment and remained at elevated levels in cultures treated for 24 h (Fig. 3A). A comparable response was also observed in similarly treated MCF7 cells (Fig.  3B). A similar up-regulation in CycG2 expression (3-5-fold) was observed upon treatment of NIH3T3 and U2OS cells with doxorubicin (supplemental Fig. S5, A and B).
In unperturbed cells endogenous and exogenous CycG2 behave as centrosome-associated nucleocytoplasmic shuttling proteins (15). We examined the distribution of CycG2 in doxo- rubicin-treated cells. Confocal immunofluorescence microscopy of MCF10a cells showed that doxorubicin-induced upregulation of CycG2 led to an accumulation of small bright puncta within the nuclei (39% increase in nuclear signal, p Ͻ 0.0001) of treated cells (Fig. 3C and supplement supplemental Fig. S5C). Moreover, doxorubicin treatment resulted in a 63% increase in CycG2 abundance at centrosomes (p ϭ 0.0018) but as expected did not alter the signal intensity for the integral centrosomal protein ␥-tubulin ( Fig. 3C and supplemental Fig.  S5C).
To define CycG2 up-regulation in relation to activation of DDR proteins and cell cycle checkpoints, we exposed cultures of murine and human cell lines to 345 nM doxorubicin for up to 24 h, sampling the different cultures over the time course of treatment for immunoblot and cell cycle analysis. HCT116 cells are known to exhibit a minimal S-phase delay in response to DNA DSBs but do undergo a clear and potent DSB-induced G 1and G 2 -phase checkpoint arrest upon exposure to doxorubicin (Fig. 4A). A combination of immunoblot and cell cycle analysis of HCT116 and MCF7 cultures sampled over a time course of treatment determined that the doxorubicin-invoked increase in CycG2 levels trailed phosphorylation of the ATM/ATR target proteins, Chk2 and Chk1, but led the accumulation of cells at the G 2 /M boundary (Fig. 4, A and B). Although doxorubicininduced DNA damage in U2OS and NIH3T3 cells initially results in an apparent delay in S-phase progression, the onset of a clear G 2 -phase checkpoint arrest was observed between 8 and 16 h of treatment. CycG2 expression was elevated in NIH3T3 cultures within 4 h of doxorubicin addition, about 2 h after the appearance of phospho-activated forms of ATM and its target SMC1, and remained at increased levels for at least 20 h (supplemental Fig. S6C). A similar analysis of CycG2, pNbs1(Ser-343) and pChk2(Thr-68) expression levels and cell cycle distribution in doxorubicin-treated U2OS cells was performed (supplemental Fig. S6D). Consistently CycG2 up-regulation followed the appearance of phospho-activated forms of early response DDR proteins by about 2 h but preceded the induction of G 2 -phase checkpoint arrest (Figs. 4 and supplemental Fig. S6).
Transient Transfection of CCNG2-targeting shRNAs Blunts Doxorubicin-induced G 2 -phase DNA Damage Checkpoint Arrest Response in NIH3T3 and HCT116 Cells-To test for the contribution of CycG2 to the doxorubicin-induced cell cycle checkpoint response, we generated and tested CCNG2-targeting shRNAs for their ability to knockdown (KD) CycG2 expression levels (supplemental Fig. S2). We transfected cultures with validated shRNA constructs (supplemental Fig. S2) and assayed the effect these shRNAs had on the cell cycle profile of asynchronous cultures grown in the presence or absence of doxorubicin ( Fig. 5 and supplemental Fig. S7). NIH3T3 cultures were cotransfected with tracer amounts of GFP expression plasmids and either empty vector (pSilencer) or the pSilencer-Ex4.2 shRNA expression construct. After 48 h of growth, the cultures were treated with doxorubicin for 24 h. Cell cycle analysis of the GFP-expressing populations indicated that cells transfected with the Ex4.2 shRNA plasmid, in contrast to the empty vector control, did not exhibit a potent G 2 -phase cell cycle arrest response to doxorubicin (supplemental Fig. S7). This suggested a surprising block of the G 2 /M rather than the G 1 /S-checkpoint arrest response by CycG2 knockdown. Repeated experiments in NIH3T3 yielded similar results.
To determine whether the diminished G 2 /M checkpoint response in cells expressing CycG2-targeting shRNA is reproducible in human cell lines, analogous experiments with HCT116 cells were performed. HCT116 cultures were transfected with control (NSC or empty vector) or CCNG2-targeting (sh 1-B, ID3) shRNA vectors containing expression cassettes for marker fluorescent protein (RFP or GFP) and incubated for 72 h before the addition of doxorubicin or vehicle. After an additional 24-h incubation period, cultures were harvested for DNA flow cytometry (Fig. 5). As expected, shRNA-mediated repression of CycG2 did not alter the cell cycle profile of untreated asynchronous HCT116 cultures but did significantly blunt the G 2 /M checkpoint accumulation of doxorubicintreated HCT116 cells (Fig. 5, A-C). In contrast to cells expressing sh 1-B, HCT116 cells transfected with expression vectors for the non-silencing control (NSC) shRNA did not exhibit an abrogation of the G 2 /M checkpoint (Fig. 5A). Cells transfected with empty vector also exhibited a potent G 2 /M checkpoint arrest response to doxorubicin (Fig. 5A). Statistical analyses indicated that KD of CycG2 in HCT116 cells results in a significant (p Ͻ 0.001) blunting of the drug-induced G 2 -phase arrest response (Fig. 5B). To determine whether the G 2 /M checkpoint attenuation by shRNAs targeting the conserved Ex4.2/1-B site could be reproduced with shRNAs targeting another CCNG2specific site, we tested the effect that doxorubicin treatment had on HCT116 cells expressing the ID3 shRNA construct (Fig.  5C). As with expression of sh 1-B shRNA, suppression of CCNG2 via transient transfection with the ID3 shRNA construct did not alter the cell cycle distribution of untreated HCT116 cells but did potently repress the doxorubicin-induced accumulation of cells at the G 2 /M checkpoint (Fig. 5C). Taken together our results strongly suggested that loss of CycG2 alters the G 2 /M checkpoint arrest response of cells to doxorubicin.

Stable Expression of CycG2-targeting shRNAs Attenuates Doxorubicin-induced G 2 -phase Checkpoint Responses in MCF7
Cells-As CycG2 has been implicated as an important biomarker for breast cancers (16,20,30), we sought to determine whether CycG2 contributes to the DNA damage checkpoint response of MCF7 breast cancer cells to doxorubicin. Puromy-cin-resistant clones of MCF7 cells stably harboring vectors encoding sh 1-B, NSC, and ID3 shRNAs were established and characterized ( Fig. 6 and supplemental Fig. S8). Immunoblot analysis determined that both exogenous (supplemental Fig.  S8A) and endogenous human (Fig. 6, A and B) CycG2 expression were strongly repressed in clones containing expression cassettes for the CycG2-targeting shRNAs sh 1-B and ID3 but not in clones harboring the NSC control shRNA vector (Fig. 6,  A and B). Importantly, in contrast to NSC and wild-type MCF7 controls, those shRNA-expressing clones exhibiting significant (p Ͻ 0.001) repression of doxorubicin-induced CycG2 levels also showed an altered G 2 /M checkpoint arrest response to doxorubicin (Fig. 6, C and D, and supplemental Fig. S8B). This response was reproducible with multiple doxorubicin-treated CycG2 KD clones displaying a statistically significant (p values Ͻ0.01-0.001) reduction in the percentage of G 2 /M-arrested cells (Fig. 6D).
Because the closest homolog of CycG2, CycG1, is a DNA damage response protein linked to regulation of G 2 /M transition (35)(36)(37), we assessed whether KD of CycG2 affected CycG1 expression (Fig. 7A). As predicted doxorubicin-induced DNA damage triggered up-regulation of CycG1 in MCF7 WT and NSC cells. Importantly, doxorubicin induction of CycG1 expression was maintained in all of the CycG2 KD clones (Fig.  7A and supplemental Fig. S9A). Given that ectopic CycG2-induced cell cycle arrest requires expression of Chk2 and p53 and promotes expression of the phospho-activated forms of Chk2 and Nbs1, we examined the effect CycG2 KD has on the expression of phospho-activated forms of these proteins (Fig. 7B). Notably, results indicated that, compared with the response in MCF7 WT and NSC control cells, depletion of CycG2 did not appreciably effect the DNA damage response induction of phospho-Nbs1 or -Chk2 in the KD clones (Fig. 7B). Passage from G 2 phase into mitosis requires active CycB1-Cdc2 complexes, but once in mitosis cyclin B1 (CycB1) is targeted for proteasomal-mediated degradation (8). Importantly, DNA damage-induced accumulation of CycB1 observed in the doxorubicin-treated WT and NSC control cells was much reduced in the CycG2-KD clones (Fig. 7C), consistent with the relative reduction in the amount of cells arrested at the G 2 /M boundary (Fig. 6, C and D). DNA damage signaling is known to inhibit CycB1-Cdc2 activation through maintenance of the Wee1-and Myt1 kinase-mediated inhibitory phosphorylation of Cdc2 on Thr-14 and Thr-15 (8). Immunoblot analysis indicated that, in contrast to doxorubicin-treated cultures of control WT and NSC cells, Thr-15-phosphorylated Cdc2 levels were not strongly increased in drug-treated CycG2 KD clones (Fig. 7C  and supplemental Fig. S9C).
Activation of Cdc2 is largely promoted through dephosphorylation of its inhibitory sites by the dual specificity phosphatases Cdc25B and Cdc25C. Consistent with the known effects of genotoxic stress on Cdc25B expression levels (38 -40), immunoblot analysis revealed reduced expression of Cdc25B in doxorubicin-treated, relative to untreated, MCF7 WT and NSC cells (Fig. 7D). Interestingly, extracts from doxorubicin-challenged CycG2 KD clones did not show a noticeable decrease in Cdc25B expression levels relative to the basal level in the respective undosed clone control (Fig. 7D). Rather, Cdc25B abundance in the doxorubicin-treated CycG2 KD clones appeared to be similar to or even increased above the level of its respective nontreated control. Although the basal level of Cdc25B in untreated cultures of CycG2 KD clones appeared lower than that in unperturbed MCF7 WT and NSC populations (Fig. 7D), the fact that this difference was not reflected by a respective increase in the percentage of CycG2 KD cells in G 2 /M (Fig. 6, C and D and supplemental Fig. S8B) suggests that the CycG2 KD clones have adapted to this lower basal level. Contrasting the difference in modulation of Cdc25B by doxorubicin that was observed between the CycG2-expressing and KD populations (the latter showing no decrease in Cdc25B levels), Wee1 abundance, induction of phospho-inhibited Cdc25C and modulation of Cdc25C levels were comparable among all drug-treated populations; the doxorubicin-dosed MCF7 WT, NSC, and CycG2 KD clones all exhibited a similar expression pattern (data not shown). Taken together these results suggest that blocking CycG2 up-regulation during a DNA damage response to DSBs promotes activation of CycB1-Cdc2 complexes by hindering the DDR-induced down-regulation of Cdc25B expression.

Doxorubicin-induced Up-regulation of Cyclin G2 Is ATMindependent-To investigate the influence of ATM activity on
CycG2 expression, we first tested whether pharmacological inhibition of ATM kinase activity blunts DNA damage-induced up-regulation of CycG2 levels (Fig. 8A). Consistent with previously reported effects of caffeine on ATM/ATR activity (41) and their target checkpoint kinases (42), treatment of MCF7 cultures with doxorubicin in the continual presence of 3 mM caffeine blunted the expression of pChk1(Ser-345) but not pChk2(Thr-68) (Fig. 8A). The later may be due to the ability of DNA-PK to phosphorylate Chk2 in the absence of ATM and ATR activity (43,44). Importantly, we found that 3 mM caffeine also dampened the doxorubicin-induced elevation of CycG2 expression (Fig. 8A). Cotreatment of MCF7 cells with 10 M of the more specific ATM inhibitor KU55933 (34) had no effect on doxorubicin-induced elevation of CycG2 expression but as expected (45,46) did reduce the expression of Thr-68-phosphorylated Chk2 (Fig. 8A). These results suggest that DNA damage-mediated up-regulation of CycG2 expression does not require ATM activity.
To further investigate the relationship of CycG2 to ATM signaling, we tested the effects of doxorubicin on CycG2 expression and cell cycle progression in cells devoid of ATM function. Importantly, although ATM-deficient cells do not maintain a G 1 -phase cell cycle arrest upon induction of DSBs, ATM-independent G 2 /M checkpoint arrest responses to genotoxic stressors (including doxorubicin) do occur (47)(48)(49)(50)(51). As expected, culture of the ATM-deficient (AT) human fibroblast line GM05849 with doxorubicin for 24 h did not, in contrast to WT cells, arrest them in G 1 phase (Fig. 8B and supplemental Fig. S10) but did provoke a potent G 2 /M checkpoint arrest response (Fig. 8B). To verify that doxorubicin-induced up-regulation of CycG2 is ATM-independent, WT and AT cells were cultured in the presence of 345 nM doxorubicin or vehicle control for 2, 4, or 6 h and assessed for activation of ATM pathway signaling and CycG2 expression (Fig. 8C). Immunoblot analysis of lysates from WT cultures indicated that, as expected, levels of phospho-activated forms of Nbs1 and Chk2 increased markedly within 2 h of doxorubicin treatment. As seen in other ATM competent cells (Figs. 3 and 4 and supplemental Figs. S5 and S6), up-regulation of CycG2 expression in AT cells was detectable within 4 h of exposure to doxorubicin, increasing 2-fold by 6 h of treatment (Fig. 8, C and D). Doxorubicin induction of pChk2(Thr-68) and pNbs1(Ser-343) expression in AT cultures was, as expected, nearly undetectable during the first 6 h of treatment. However, in contrast to the obvious deficiencies in the AT cell DDR, CycG2 expression was still up-regulated within 4 h of treatment, increasing nearly 2-fold by 6 h of treatment (Fig. 8, C and D). Notably, the basal level of CycG2 in AT  68), and the loading control proteins ␣-actinin (␣-Act) and PP2A/C in WT and AT fibroblasts cultured over a time course ϩ or Ϫ doxorubicin (C). D, shown is CyG2 expression relative to GAPDH loading control in WT and AT cells cultured for 6 h in the presence (ϩ) or absence (Ϫ) of doxorubicin. Note that although there is a reduced phospho-modification of Chk2 and Nbs1 in doxorubicin-treated AT compared with WT cells, doxorubicin treatment increased CycG2 expression ϳ2-fold over the respective basal level for each culture. cells was higher than the basal level in the WT control (Fig. 8, C  and D), indicating that increased basal expression of CycG2 is better tolerated in the absence of ATM. Nevertheless, doxorubicin did up-regulate CycG2 expression over basal levels to a similar degree in both WT and AT cells (Fig. 8, C and D). Collectively our results show that DDR induction of CycG2 expression is ATM-independent.

DISCUSSION
CCNG2 expression is up-regulated as cells undergo cell cycle arrest in response to a variety of growth inhibitory signals (9, 10, 12, 16 -19, 22, 24 -26). We previously showed that unscheduled CycG2 expression inhibits DNA synthesis, blunts CDK2 (but not CDK4) activity (13), and induces a p53-dependent G 1 -phase cell cycle arrest (15). Similar results have since been replicated by others using various epitope-tagged forms of CycG2 in several cell lines (10,13,14,16,24). Here we show that the potent cell cycle arrest response of HCT116 cells to exogenous CycG2 requires intact alleles for Chk2 and p53 (Fig. 1). That loss of p21 only partially reduces CycG2-medited cell cycle inhibition suggests that additional effectors downstream of Chk2 or p53 are involved. Thr-68 phosphorylation of Chk2 by activated ATM triggers Chk2-mediated G 1 checkpoint arrest responses to DNA DSBs (6,52). However, the G 1 -phase cell cycle arrest response induced by ectopic CycG2 does not require ATM (Fig. 1). Consistent with the Chk2-dependent arrest, ectopic elevation of CycG2 also promoted the expression of the Thr-68 phospho-activated form of Chk2 (Fig. 2). Because phosphorylation of p53 by Chk2 is known to promote G 1 -phase checkpoint arrest (53), and pChk2(Thr-68), but not pChk1(Ser-345) levels, were robustly elevated in CycG2-overexpressing WT and p53-deficient HCT116 cells (Fig. 2), the p53-dependent G 1 -phase arrest induced by ectopic CycG2 is likely downstream of activated Chk2.
Unscheduled enforced expression of CycG2 in the absence of coordinated dsDNA DDR signaling ultimately has more profound effects on G 1 /S compared to G 2 /M transition. It is however unclear how CycG2 overexpression promotes expression of Thr-68 phospho-activated Chk2. As the CycG2-induced cell cycle arrest was independent of ATM, the effects of ectopic CycG2 on Chk2 are likely ATM-independent. That phosphorylation of the ATR target Chk1 was not promoted by ectopic CycG2 expression suggests that CycG2 overexpression did not activate ATR. Recent work indicates that pChk2(Thr-68) serves a DNA damage-independent function during mitosis to ensure proper spindle assembly and maintain chromosomal stability (54,55). The kinases PLK1, TTK/hMps1, and DNA-PK can each interact with and phosphorylate Chk2 on Thr-68 and play DDR-independent roles in regulating mitosis and spindle assembly checkpoints (43, 52, 56 -58). Because ectopic CycG2 expression promotes formation of nocodazole-resistant microtubules and aberrant nuclei (13), its overexpression may trigger a defective mitosis that provokes Chk2 activation through one of these kinases. As cells exit mitosis this stress response could ultimately result in a G 1 -phase arrest. Alternatively, as CycG2 can interact and form complexes with PP2A B56 and C subunits (13,59) and phosphorylation of Chk2 on Thr-68 is negatively regulated by B56 isoforms of PP2A (60, 61), it is possible that in otherwise unperturbed cells, overexpressed CycG2 acts as a PP2A sink and in so doing inhibits PP2A-mediated dephosphorylation of Chk2. In this context it is notable that CycG2, PP2A, and Chk2 all associate with centrosomes (15,62,63).
We explored the possibility that CycG2 contributes to cell cycle control during genotoxic stress-induced DDRs. Genotoxic topoisomerase II poisons (32) play a central role in cancer chemotherapies. Although doxorubicin and etoposide belong to distinctly different classes of chemotherapeutics (1,32), both induce DNA DSBs by trapping topoisomerase II-DNA intermediates and provoke a potent G 2 -phase checkpoint arrest in treated cells (1,32). We found that endogenous CycG2 expression is increased up to 8-fold when cancer cells are cultured with doxorubicin or etoposide (Figs. 3 and 4 and supplemental Figs. S5 and S6). Concordant with the idea that CycG2 is upregulated in response to activation of DNA DSB signaling pathways, previous studies indicated significant elevation of CCNG2 mRNAs upon ␥-irradiation-induced DNA damage (64). Because the doxorubicin-stimulated increase in CycG2 levels is clearly detectable within 4 h of dosing, well before an obvious arrest at the G 2 /M boundary ( Fig. 4  Doxorubicin-induced DNA damage triggers ATM and ATR activation (1,32). Although ATR and ATM both enforce the DSB DDR delay in M-phase entry, ATR activity is thought to regulate the majority of the late (2-9 h post ␥-irradiation) phase of the checkpoint response (4,46,65). In the absence of ATM, both ATR and DNA-PK regulate DNA DSB G 2 /M checkpoint responses (50,51). We showed (Fig. 8A) that doxorubicin induction of CycG2 (and Chk1 phosphorylation) in MCF7 cells is not blocked by 10 M of the ATM inhibitor KU55933 (ATM IC 50 ϭ 13 nM, ATR IC 50 Ͼ 100 M (34)) but is blunted by 3 mM caffeine (ATM IC 50 ϭ 0.2 mM, ATR IC 50 ϭ 1.1 mM). Moreover, we determined that this genotoxic stress elevates CycG2 expression to a similar degree in WT and ATM-deficient fibroblasts (Fig. 8, B and C). Thus, doxorubicin-triggered CycG2 up-regulation does not require early ATM-initiated signaling.
The concentration of caffeine that repressed CycG2 expression is within the IC 50 range for ATR (1.1 mM) and mTOR (0.4 mM) but severalfold lower than the IC 50 for DNA-PK (10 mM) (41). Because direct inhibition of mTOR (via rapamycin) promotes rather than represses CycG2 expression (16,19,66), the ability of 3 mM caffeine to repress doxorubicin-induced up-regulation of CycG2 is most likely independent of its effects on mTOR. In contrast to KU55933, caffeine did not diminish pChk2(Thr-68) levels in doxorubicin-treated MCF7 cells. However, caffeine-insensitive DNA damage induction of Chk2 phosphorylation has been reported by others (47,(67)(68)(69). DNA-PK activity is sensitive to KU55933 (IC 50 ϭ 2.5 M), but 10 M KU55933 did not blunt doxorubicin-induced CycG2 expression, suggesting that this response is DNA-PK-independent. The late-phase DDR increase in CycG2 levels coupled with its ATM independence and caffeine sensitivity, suggests that doxorubicin-induced CycG2 up-regulation is ATR-dependent; however, we cannot exclude DNA-PK involvement.
The CCNG1 gene encoding the closest CycG2 homolog, CycG1, is a direct transcriptional target of p53, and its transcript levels increase severalfold in response to DNA damage (25,27,28,70). Predictably, CycG1 expression was significantly up-regulated in doxorubicin-treated MCF7 cells (Figs. 7A and supplemental Fig. S9). Importantly we found that doxorubicininduced DDR elevation of endogenous CycG1 expression was unaffected by shRNA-mediated KD of CycG2, further indicating the specificity of our CCNG2-targeting shRNA constructs. CycG1 has been linked to G 2 /M checkpoint control (35)(36)(37); however, whether it promotes or inhibits either cell cycle arrest or cell death in response to DNA damage is controversial (35-37, 70 -72). Because CycG2 depleted cells exhibited a reduced G 2 /M checkpoint despite the rise in CycG1 levels suggests that CycG1 does not compensate for loss of CycG2 and that these two homologs do not serve fully redundant functions.
In variance with the effects of ectopic CycG2 expression on Chk2, shRNA-mediated blunting of CycG2 in MCF7 cells had no affect on the DDR-induced elevation of pChk2(Thr-68) (Fig.  7B). CycB1 expression levels are normally increased as cells enter G 2 phase and decreased as cells proceed through mitosis (8). As predicted, the doxorubicin-triggered G 2 /M checkpoint led to accumulation of CycB1 levels in WT and shRNA control cultures. In contrast, CycG2 KD clones did not show increased CycB1 expression under the same conditions (Fig. 7C). In accord with their blunted G 2 /M checkpoint arrest response, doxorubicin-treated CycG2 KD clones also exhibited diminished levels of phospho-inhibited Cdc2 when compared with treated WT and shRNA controls (Fig. 7C). Inhibitory phosphorylation of Cdc2 on Thr-14 and Thr-15 by the Myt1 and Wee1 kinases is counterbalanced by the Cdc25 phosphatases that dephosphorylate these sites (6,8). During DDR signaling the dual specificity phosphatases Cdc25B and Cdc25C are themselves subject to Chk1 and Chk2 inhibitory phosphorylation that promotes Cdc25 degradation and/or restricted subcellular localization (8). Although Cdc25B expression is not required for G 2 /M transition in otherwise unperturbed somatic cell populations, it is essential for resumption of cell cycle progression after DNA damage-induced checkpoint arrest (8). We found that Cdc25B levels were diminished in doxorubicintreated compared with mock-treated WT and NSC cells, but no such doxorubicin-induced decrease from base-line levels was apparent for the CycG2 KD clones. Given that increasing Cdc25B expression levels even moderately impairs G 2 /M checkpoint control (39,73,74), our results (Figs. 6 and 7) suggest that the weakened G 2 /M checkpoint arrest in CycG2 KD cells is due to a disruption of the regulatory circuit controlling Cdc25B expression. Thus, CycG2 may contribute to G 2 /M checkpoint enforcement by constraining the Cdc25B/ CycB1-Cdc2 axis.
CCNG2 transcripts are up-regulated during G 1 -phase cell cycle arrest responses to a variety of DDR-independent antimitogenic signaling cascades (9,12,16,17). RNAi KD of CCNG2 has been shown to blunt the G 1 -phase arrest response to some of these growth inhibitory signals (14,17). Given these observations and the effects that ectopic CycG2 expression has on G 1 /S phase transition, the diminished G 2 /M checkpoint arrest response of CycG2 KD cells to doxorubicin was somewhat surprising. However, such seemingly contradictory findings are not unprecedented for cell cycle inhibitors and have been described for both p53 and p21 (75)(76)(77)(78)(79). Although most of the evidence in the literature supports a role for CycG2 in limiting G 1 /S-phase transition, there are indications that CycG2 could participate in G 2 /M regulation (30, 80 -83). The idea that CycG2 has a regulatory function in G 2 /M-phase transition is also supported by the discovery that CycG2 is a substrate of the anaphase promoting complex (APC), being both ubiquitinated and degraded in mitotic cell extracts enriched with APC-Cdc20 complexes (84). Consistent with the notion that CycG2 helps restrict G 2 /M transition, we show for the first time that 1) a caffeine-sensitive but KU55933-insensitive and ATM-independent DDR pathway promotes CycG2 up-regulation during the late phase of doxorubicin-induced G 2 /M checkpoint and 2) that CycG2 depletion attenuates G 2 /M checkpoint-induced down-regulation of Cdc25B, inhibitory phosphorylation of Cdc2, and accumulation of CycB1. Given the report that elevated CycG1 expression promotes transcriptional activation of CycB1 and abrogation of G 2 /M checkpoint arrest (72), it is possible that there is a Yin and Yang relationship between these two G-type cyclins and that CycG2 acts to restrict CycG1-associated activity during DNA damage responses. The single CycG homolog in Drosophila is an essential protein for embryonic development that restricts cell proliferation and growth (85). Whether the two mammalian CycG paralogs evolved to serve opposing or complementary functions is an open question. Future studies will be needed to determine the exact mechanism by which CycG2 modulates the Cdc25B-Cdc2/CycB1 regulatory loop during G 2 /M checkpoint.