Inhibition of type I histone deacetylase increases resistance of checkpoint-deficient cells to genotoxic agents through mitotic delay

A functional DNA damage checkpoint pathway is essential for maintaining genomic integrity by facilitating the repair of damaged DNA. In S. pombe, the DNA damage checkpoint is regulated in a rad3⁺-dependent (ATM/ATR homologue) manner by a signaling cascade that involves Rad26, Rad17, Rad9, Rad1, Hus1, Crb2, Mrc1, Chk1, and Cds1 (1). Activation of this pathway in response to DNA damage ultimately leads to inactivation of Cdc2 activity and cell cycle arrest (the DNA damage checkpoint). At the same time, the repair of DNA lesions is also activated and cell cycle arrest allows time for DNA damage repair to take place. The importance of this pathway is underscored by the extreme sensitivity of DNA damage checkpoint and repair mutants to DNA-damaging agents. It has also become clear that the regulation of chromatin architecture is important for the detection/recognition and repair of DNA damage and the recovery/resetting of higher order chromatin structure following repair (2).

Histone acetyl transferases and histone deacetylases (HDAC) regulate gene expression and chromatin structure by modulating the acetylation status of lysine residues in core histones within chromatin (3). Early studies reported that UV-induced DNA damage resulted in an increase in global histone acetylation levels (4). Genetic studies have also shown that both histone acetyl transferase and HDAC mutants are sensitive to DNA-damaging agents and impaired in their ability to repair damaged DNA (5, 6). It is likely that the global changes in acetylation following DNA damage modulate gene expression and allow DNA repair proteins to access the damaged DNA within chromatin. HDAC activity is commonly deregulated in cancer cells. Small molecule inhibitors of HDAC activity have proven effective at inhibiting cancer cell proliferation in vitro and in vivo. As a consequence, several HDAC inhibitors (HDACi) have entered clinical trials with encouraging results (7, 8). Of particular interest has been the observation that HDACis can sensitize cancer cells to the effects of conventional DNA-damaging agents used in the treatment of cancer (9–11). Similarly, S. pombe mutants that have mutations within the clr6⁺ HDAC gene, or fail to express its associated cofactors, are sensitive to DNA-damaging agents (6). Interestingly, inhibition of HDAC activity has been shown to suppress the sensitivity of Saccharomyces cerevisiae checkpoint mutants to DNA damage and replication stress. Scott and Plon (12) observed that deletion of the RPD3 HDAC gene or its cofactor SIN3 enhanced the viability of the mec1Δ and rad9Δ checkpoint mutants following exposure to hydroxyurea (HU) or UV radiation. The suppression of sensitivity was shown to involve a RAD53-independent checkpoint and require a functional spindle checkpoint pathway. In many cancer cells, the ability to effectively detect and repair DNA damage is impaired. Indeed, mutations within genes that regulate DNA damage repair have been causally liked to the development and progression of cancer and the response to therapy (13, 14). The use of HDACis as hemosensitizers and radiosensitizers may thus be contraindicated in cancer cells in which the DNA damage response pathway is impaired.

In the current study, we set out to (a) determine if inhibition of HDAC activity suppresses the sensitivity of S. pombe DNA damage checkpoint and repair mutants to DNA damage, (b) to identify the specific HDAC proteins involved in the suppression of sensitivity, and (c) to characterize the molecular mechanisms that underlie the suppression sensitivity to DNA damage. We present evidence that the HDACi trichostatin A (TSA) suppresses the sensitivity of S. pombe checkpoint mutants to UV-induced DNA damage. A clr6-1 rad1Δ double mutant displayed significantly reduced sensitivity to UV radiation compared with rad1Δ single mutants. In contrast, deletion of the clr3⁺ and hos2⁺/hda1⁺ genes did not suppress sensitivity to DNA damage. In rad1Δ mutants, TSA attenuated UV-induced mitotic stimulation (15) but did not restore a normal DNA damage checkpoint. Furthermore, TSA did not induce Chk1 or Cdc2 phosphorylation in UV-exposed rad3Δ mutants and suppressed the UV sensitivity of cds1Δ chk1Δ double mutants. Similarly, clr6-1 rad1Δ double mutants underwent a reduced rate of cell division compared with rad1Δ mutants following UV exposure. HDAC inhibition did not, however, suppress the UV sensitivity of rad13Δ and rhp51Δ mutants that are defective in nucleotide excision repair (NER) and recombination, respectively. Our studies indicate that suppression of Clr6 activity modulates global histone acetylation and delays cell cycle progression. Together, these effects significantly suppress the UV sensitivity of S. pombe DNA damage checkpoint mutants.

All strains (Table 1) were cultured on YES agar or in YES broth (0.5% yeast extract, 3% glucose, and 225 mg/liter each of adenine, histidine, leucine, lysine, and uracil) at 30°C except where indicated. DNA constructs and chromosomal disruptions were done as previously described (16, 17). Stock solutions of TSA (10 mg/mL; Errant Gene Therapeutics) in ethanol were stored at −20°C. Bleomycin stock solution (Sigma Aldrich) was made up according to the manufacturer's instructions and stored at −20°C. Stock solutions of HU (1 mol/L; Sigma-Aldrich) were stored at −20°C.

Cells were grown overnight in liquid medium until early or mid-log phase was reached. Cells were pelleted by brief centrifugation and resuspended in 25 mL of sterile water. A UVGL-58 short wave UV lamp with an emission peak of 254 nm was used as a UV source under conditions of continuous agitation. Following irradiation, cells were collected by centrifugation and resuspended in liquid medium with or without 20 μg/mL of TSA for 2 or 4 h. The cells were then equilibrated to an OD_{600 nm} of 0.2, serially diluted, spotted on plates, and incubated for 2 to 3 d at the indicated temperature. Alternatively, cells were grown to stationary phase and then diluted into fresh medium at an OD_{600 nm} of 0.2. The cells were then cultured at 30°C for 4 h to allow reentry into the cell cycle, serially diluted, and spotted on YES agar plates supplemented with the indicated compounds. The spotted cultures were allowed to dry in air and were then exposed to the appropriate dose of UV.

Early- to mid-log phase cultures were treated with 20 mmol/L HU alone or with 20 μg/mL TSA and incubated for 4 h with shaking at 30°C. Alternatively, cells were incubated with 1.5 mU/mL bleomycin alone or with 20 μg/mL TSA for 2 h. Following incubation, the cultures were equilibrated to an OD_{600 nm} of 0.2 to 0.3, serially diluted, spotted, and incubated for 2 to 3 d at the indicated temperature.

Cells were harvested by centrifugation and processed immediately or snap frozen in an ethanol bath and stored at −80°C. The cells were lysed in buffer A [50 mmol/L Tris-HCl (pH 8.0), 50 mmol/L NaCl, 0.2% Triton X-100, and 1% NP40] supplemented with Complete protease inhibitor and phosphatase inhibitor cocktails (Roche), using a FastPrep SP120 apparatus (Savant, Inc.) with a speed setting of 5.0 for 20 s. Lysates were resolved on 7%, 10%, or 15% SDS-polyacrylamide gels and transferred to nitrocellulose membranes using a semidry blotting apparatus. Proteins were detected using an anti-HA or anti-Myc probe (Santa Cruz Biotechnology), anti–phospho-Cdc2, anti–α tubulin (Sigma), anti-Cdc2 (Abcam), anti-acetyl H3 (K9, K14), and anti-acetyl H4 (K5, K8, K12; Upstate Biotechnology, Millipore AB). Horseradish peroxidase–conjugated secondary antibodies and the enhanced chemiluminescence Western blotting system (General Electric Bio-Sciences) were used for detection.

Cells were washed, fixed in 70% ethanol, and mounted in Vectashield mounting medium containing 4′,6-diamidino-2-phenylindole (DAPI; Vector Laboratories Ltd.) to counterstain DNA. Calcofluor staining of septa was carried out as described (18). Briefly, washed and fixed cells were mounted in 50 μg/mL calcofluor (Sigma-Aldrich). Stained cells were examined by fluorescence microscopy and at least 300 cells were scored for septation assays.

Approximately 10⁷ cells were harvested at the desired time points, resuspended in 70% ethanol, and stored at 4°C until use. Fluorescence-activated cell sorting (FACS) analyses were done according to the protocol of Sazer and Sherwood (19), using propidium iodide (32 μg mL⁻¹) as outlined on the Forsburg lab page.¹

Deletion of the class I HDAC gene, RPD3, has been shown to suppress DNA damage sensitivity in S. cerevisiae checkpoint mutants (12). To determine if the inhibition of HDAC activity in S. pombe checkpoint mutants would also suppress UV sensitivity, early-log phase cultures of wild-type (wt) and checkpoint mutant strains were UV irradiated and cultured with or without TSA (Fig. 1A). Wt cells exhibited little sensitivity to UV and this was not affected by TSA. As expected, the rad3-136, rad3Δ, rad1Δ, chk1Δ, and wee1Δ mutants were extremely sensitive to UV. When these mutants were cultured in YES medium containing TSA however, a significant increase in the survival was observed following UV exposure. The survival of rad3-136, hus1Δ, and rad1Δ mutants was also significantly enhanced when plated on YES agar containing TSA followed by exposure to UV (Supplementary Fig. S1A). Clr6, Hos2, and Clr3 are S. pombe class I/II HDACs, and of these, Clr6 has been shown to be sensitive to TSA (6). We generated clr6-1 rad1Δ, hos2Δ rad1Δ, and clr3Δ rad1Δ double mutants to compare their relative sensitivity to UV. Culture of the clr6-1 rad1Δ mutant at the semipermissive temperature (30°C) significantly suppressed UV sensitivity, an effect not observed at the permissive temperature, 25°C (Fig. 1B). In contrast, the hos2Δ rad1Δ and clr3Δ rad1Δ double mutants failed to grow following exposure to UV (Fig. 1B). Inhibition of HDAC activity by TSA also suppressed the sensitivity of the rad3Δ, rad1Δ, hus1Δ, and cdc2-3w mutants to HU (Fig. 1C). Similarly, clr6-1 rad1Δ double mutants grown at the semirestrictive temperature (30°C) were considerably less sensitive to HU than rad1Δ single mutants (Fig. 1D). FACS analyses further showed that progression through mitosis is delayed in clr6-1 and clr6-1 rad1Δ mutants at 30°C (Fig. 1D). However, TSA did not suppress the sensitivity of the rad3-136 mutant to bleomycin (Supplementary Fig. S1B). cds1 mutants undergo a Chk1-dependent cell cycle arrest when exposed to HU, but nevertheless lose viability due to their inability to resume DNA replication (20). In contrast, cdc2-3w mutants lose viability because they undergo mitosis in the absence of DNA replication when exposed to HU (21, 22). TSA suppressed the HU sensitivity of cdc2-3w but not cds1Δ mutants suggesting that the HDACi delays cell cycle progression (Fig. 1C). Accordingly, TSA also suppresses the UV sensitivity of chk1Δ and wee1Δ mutants (Fig. 1A), which are specifically defective in enforcing the G₂ DNA damage checkpoint (1).

S. pombe DNA damage checkpoint mutants fail to undergo cell cycle arrest when exposed to DNA-damaging agents and accumulate as septated cells (23, 24). Septation indices were monitored to determine cell cycle progression following UV exposure (120 Jm⁻²) with or without 20 μg/mL TSA (Fig. 2A). Mitotic division decreased rapidly in wt cells as monitored by the decrease in the number of septating cells and was not affected by TSA (Fig. 2A, I). In contrast, exposure of the rad3-136, rad1Δ, and chk1Δ mutants to UV did not result in cell cycle arrest (Fig. 2A, II–IV). As previously reported, exposure of rad3-136, rad1Δ, and chk1Δ mutants to UV resulted in an increase in the number of septated cells (Fig. 2A, II–IV). Addition of TSA to the culture medium prevented the accumulation of septated cells but did not restore a wt checkpoint in rad3-136, rad1Δ, and chk1Δ mutants; instead, the septation profiles were suggestive of a mitotic delay as the peak of septated cells was delayed or absent (Fig. 2A, II–IV). Similarly, attenuation of Clr6 activity also prevented the accumulation of septated cells when clr6-1 rad1Δ double mutants were exposed to UV (Fig. 2B). TSA did not restore Chk1 phosphorylation in a rad3Δ mutant or influence Cdc2 phosphorylation in rad1Δ mutants following UV exposure (Fig. 2C and D). Rather, the Chk1 level was lowered in rad3Δ mutants exposed to TSA. Together, these observations suggest that the suppression of Clr6 HDAC activity delays cell cycle progression independently of Cdc2 phosphorylation.

To determine where in the cell cycle the TSA-induced delay takes place, we next undertook analyses of cell cycle progression by microscopy, septation assays, and flow sorting. To determine if HDAC inhibition delays progression through G₂, we released cdc25-22 mutants, synchronized in early G₂ by incubating for 4 h at 36°C, into the cell cycle in the presence or absence of 20 μg/mL TSA. In the absence of TSA, cdc25-22 cells rapidly proceeded through mitosis into the subsequent G₁-S phase as indicated by the large number of binuclear and/or septated cells detectable between 80 and 140 minutes after release. In the presence of TSA, progression into mitosis was clearly delayed and the mitotic index did not peak until 140 to 160 minutes after release (Fig. 3A and B). FACS analyses further showed a marked difference in cell cycle kinetics between TSA-treated and untreated cells from Fig. 3A and B (Fig. 3C). In contrast, TSA did not affect progression into S phase when cdc10-M17 mutants, synchronized in G₁ by incubation at the restrictive temperature, were downshifted to 25°C (Fig. 3D). In S. pombe, S phase, septum formation, and cytokinesis are dependent on the completion of the preceding mitosis (25, 26). Our results thus strongly suggest that TSA delays entry into and/or progression through mitosis. Because cytokinesis occurs after the completion of S phase in S. pombe however, we could not rule out the possibility that TSA also delays cell cycle progression between the completion of mitosis, the subsequent G₁-S phase and/or cytokinesis. When asynchronous populations of rad3Δ mutants were exposed to HU and monitored by FACS (Fig. 4A), TSA delayed progression into S phase. In the presence of 20 mmol/L HU alone, within 2 hours, a majority of the cell population had proceeded through one round of mitosis and daughter cells accumulated with a predominantly <2C DNA content. In the presence of TSA however, a significant proportion of the cell population remained in G₂ at 2 hours after drug exposure, indicating that these cells had not yet progressed through mitosis and/or cytokinesis. Coculture with TSA also reduced the number of cells with chromosome segregation defects (cut phenotype) when rad3-136 or cdc2-3w mutants were exposed to HU (Fig. 3B and D). We noted that rad3-136 mutants cocultured with HU and TSA were longer at cell division when compared with cells cultured in the presence of HU alone (Fig. 3C). In cells with a defective G₂ checkpoint attempting mitosis in the presence of DNA damage, chromosomal missegregation will occur. Taken together, our findings suggest that a mitotic delay is the main mechanism by which HDAC inhibition suppresses the DNA damage sensitivity of checkpoint mutants. This delay provides additional time for DNA repair, thus reducing the number of cells with chromosomal segregation defects.

HDAC inhibition has been shown to delay mitosis in a manner dependent on the p38 mitogen-activated protein kinase (27). Deletion of sty1⁺ (the p38 homologue in S. pombe) in a rad1Δ background did not prevent the suppressive effect of TSA following exposure to UV (Fig. 4A). Interestingly, TSA also suppressed the UV sensitivity of sty1Δ mutants (Fig. 5A). Srk1, a downstream target of Sty1, has also been shown to delay G₂ progression in S. pombe (28). Deletion of srk1⁺ in a rad1Δ background did not suppress the protective effect of TSA, however (Fig. 5A). In S. cerevisiae, HDAC inhibition induces an alternative Mad1-dependent cell cycle delay when checkpoint mutants are exposed to agents that damage DNA (12). Codeletion of the S. pombe spindle checkpoint regulator gene mad2⁺ did not affect the ability of TSA to suppress UV sensitivity in rad1Δ mutants (Fig. 5A). Our findings indicate that HDAC inhibition delays cell cycle progression in S. pombe independently of Sty1, Srk1, and the spindle checkpoint pathway.

We investigated the ability of TSA to suppress the UV sensitivity of S. pombe mutants that were checkpoint competent but deficient in nucleotide excision or recombination repair. For these experiments, we used the rad13 mutant that is defective in NER and rhp51 mutants defective in recombination repair. Exposure of both mutants to 120 Jm⁻² UV resulted in a significant loss of viability. Culture of these mutants with TSA following UV exposure failed to suppress UV sensitivity (Fig. 5A). TSA nevertheless did suppress the DNA damage sensitivity of a rad13Δ rad1Δ double mutant indicating that the NER pathway is not required for the protective effects observed with HDAC inhibition. These observations suggest that TSA suppresses the UV sensitivity checkpoint mutants but not that of mutants defective in the NER and recombination repair pathways.

In S. cerevisiae, exposure to UV results in the rapid increase of global histone H3 and H4 acetylation levels (29, 30). In contrast to S. cerevisiae, exposure of wt S. pombe cells to 120 Jm⁻² UV resulted in a rapid decline of global histone H3 and H4 acetylation levels and persisted for up to 2 hours (Fig. 5B and C). The global levels of histone H3 acetylation declined similarly in rad3Δ mutants following exposure to UV (Fig. 5D). TSA induced acetylation of H3 in both wt and rad3Δ mutants following exposure to UV (Fig. 5D). Our observations suggest that global increases in histone acetylation (at least of histones H3 and H4) are not part of the normal response to UV in S. pombe.

In the current study, we have extended previous findings from S. cerevisiae (12), demonstrating that inhibition of HDAC activity suppresses the sensitivity of checkpoint deficient strains to DNA damage. Our findings suggest that a major mechanism underlying this is that HDAC inhibition delays progression into mitosis in S. pombe cells, which in a cycling population are largely in G₂, and can partially suppress the need for a functional DNA damage checkpoint pathway. In contrast to S. cerevisiae, however, this mitotic delay occurs independently of a functional spindle checkpoint pathway. Furthermore, we have shown that chemical inhibition of HDACs by TSA similarly suppresses the sensitivity of checkpoint mutants to DNA damage, and that the HDAC Clr6 is the TSA target responsible for this effect. HDACis are currently in clinical trials to determine their efficacy as chemosensitizers and radiosensitizers (31). A proper understanding of how HDACis influence the response of cancer cells to DNA damage is required to facilitate effective use of these agents as modulators of drug sensitivity.

Numerous studies have shown that HDACis sensitize cancer cells to the cytotoxic and genotoxic effects of conventional cancer agents (31, 32). The precise mechanisms whereby HDACis sensitize cancer cells to these treatments remain unclear but may involve prevention of effective DNA repair (31). Cancer cells often harbor defects in DNA damage checkpoint pathways and thus have a limited repertoire of responses to deal with genotoxic insults (14). Conceptually, chemosensitization and radiosensitization stratagems seek to attenuate or inhibit the activity of the remaining checkpoints in cancer cells, thus enhancing their sensitivity to genotoxins. Accordingly, agents that abolish DNA damage checkpoints sensitize cancer cells to genotoxic agents (33). HDACis induce cell cycle arrest in both normal and cancer cells (34). This fact raises the possibility that the cell cycle effects of HDACis counteract their chemosensitizing activity. Disruption of an HDAC gene in S. cerevisiae has previously been shown to suppress the sensitivity of checkpoint mutants to UV and HU (12). It is currently unclear if HDAC inhibition will similarly compensate for the loss of checkpoint function in cancer cells. Given that cancer cell progression often selects for the loss of checkpoint pathways (35, 36), a better understanding of the interactions between HDACs and DNA damage response pathways is desirable.

The effects of HDAC inhibition on S. pombe DNA damage checkpoint and repair mutants have not been reported earlier. In the current study, we have characterized the effect of HDAC inhibition with TSA on the sensitivity of S. pombe checkpoint mutants to UV. TSA effectively suppressed the UV sensitivity of rad3, rad1, hus1, chk1, and cds1 chk1 mutants to UV and HU. Deletion of the clr6⁺ HDAC gene, but not of other HDAC genes, similarly suppressed the HU and UV sensitivity of rad1Δ mutants.

Inhibition of HDAC activity in S. pombe mutants may suppress UV sensitivity through the following mechanisms: (a) activation of an alternative checkpoint, (b) expression of DNA repair proteins not expressed in mutants exposed to UV alone, or (c) enhanced repair resulting from changes in chromatin structure due to increased histone acetylation. Our observations indicate that the suppression of UV sensitivity by TSA neither requires the damage recognition nor the cell cycle checkpoint activities of checkpoint proteins, because the sensitivity of rad3 mutants is also suppressed (Fig. 1). We noted that TSA also suppressed the characteristic increase in septating cells following UV exposure that occurs in rad3, rad1Δ, and chk1Δ mutants (Fig. 2A; refs. 23, 24). The mitogen-activated protein kinase–activated protein kinase Srk1 has been shown to be capable of inducing cell cycle arrest in G₂ (37). Srk1 is activated by the Sty1 stress-activated protein kinase, a homologue of mammalian p38^SAPK2, following cellular exposure to stress (37, 38). Sty1 has itself been shown to be involved in the response of S. pombe cells to UV exposure and sty1 mutants are moderately sensitive to this agent (39). Deletion of sty1⁺ in a rad1Δ mutant to UV did not prevent the suppressive effect of TSA following exposure. Similarly, deletion of srk1⁺ in a rad1Δ background did not suppress the protective effect of TSA or result in increased sensitivity to UV (Fig. 5). In S. cerevisiae, the suppression of UV sensitivity induced by the inhibition of HDAC activity requires an intact spindle checkpoint (12). Codeletion of the spindle checkpoint regulator gene mad2⁺ did not affect the ability of TSA to suppress UV sensitivity in S. pombe rad1Δ mutants (Fig. 5). We conclude therefore that in fission yeast, the suppression of UV sensitivity by TSA does not involve activation of either of these alternative cell cycle checkpoints. TSA failed to suppress the sensitivity of rad13Δ mutants, defective in NER, and of an rhp51Δ mutant, defective in recombination repair (Fig. 5). These mutants arrest normally following exposure to UV but fail to repair UV-induced DNA damage (40). TSA, thus specifically suppresses the DNA damage sensitivity of checkpoint mutants, but not of DNA repair mutants.

TSA suppressed the UV sensitivity of chk1Δ mutants (Fig. 1). According to current models, the UV sensitivity of chk1 mutants results from their inability to undergo G₂ arrest following DNA damage. These mutants thus undergo mitosis in the presence of damaged DNA (41). We postulate that TSA suppresses UV sensitivity by delaying progression into mitosis, thus abrogating the need for checkpoint induction. The precise mechanisms by which TSA delays mitotic progression remain unclear but are clearly linked to the increased global acetylation induced by this agent. Studies in mammalian cells suggest, however, that the prevention of entry into mitosis as well as its deregulation is indeed a common consequence of HDAC inhibition (42). TSA has been shown to suppress the expression of mitotic regulators such as cyclin B1 and Plk1 in human cancer cells (43). In S. pombe, osmotic stress has been shown to delay mitosis independently of mad2⁺ and rad3⁺ by enhancing the affinity of Cut2 (securin) for Cut1 (separase; ref. 44). We are currently investigating if HDACis prevent the initiation of mitosis by similar mechanisms in S. pombe.

It is increasingly evident that the modulation of histone acetylation by histone acetyl transferases and HDACs is important for the repair of DNA damage (45). Histone acetylation enhances DNA repair, and S. cerevisiae gcn5Δ mutants are mildly sensitive to UV (46). In S. pombe, loss of Clr6 HDAC activity or deletion of its associated cofactors results in increased sensitivity to higher doses of UV (6). Histone acetylation is believed to allow efficient access of repair proteins via its effects on transcription and chromatin architecture (45). Interestingly, we observed that exposure to UV does not induce increased, but rather a moderately decreased global histone acetylation in S. pombe. Indeed, global deacetylation seems to accompany cell cycle arrest in S. pombe irrespective of cell cycle phase (Supplementary Fig. S1C; ref. 47). Our studies show that HDAC inhibition does not suppress the sensitivity of mutants with DNA repair defects. They do not, however, exclude the possibility that TSA enhances the rate of DNA repair as a consequence of global histone acetylation. We are currently investigating this possibility.

In mammals, a defect within the ATM-regulated checkpoint pathway sensitizes cells to DNA damage but also predisposes toward the development of cancer. It is possible that the modulation of HDAC activity in cells with defective checkpoint pathways may contribute toward genetic instability by facilitating their survival. Our findings also provide further support for the notion that the use of HDACis as chemosensitizers or radiosensitizers may be contraindicated in tumors with defective DNA damage checkpoints (12).

No potential conflicts of interest were disclosed.

We thank Beáta Grallert, Erik Boye, Karl Ekwall, Shiv Grewal, Nancy Walworth, and the Yeast Genetic Resource Centre for the S. pombe strains.