Anumber of proteins are recruited to nuclear foci upon exposure to double-strand DNA damage, including 53BP1 and Rad51, but the precise role of these DNA damage–induced foci remain unclear. Here we show in a variety of human cell lines that histone deacetylase (HDAC) 4 is recruited to foci with kinetics similar to, and colocalizes with, 53BP1 after exposure to agents causing double-stranded DNA breaks. HDAC4 foci gradually disappeared in repair-proficient cells but persisted in repair-deficient cell lines or cells irradiated with a lethal dose, suggesting that resolution of HDAC4 foci is linked to repair. Silencing of HDAC4 via RNA interference surprisingly also decreased levels of 53BP1 protein, abrogated the DNA damage–induced G2 delay, and radiosensitized HeLa cells. Our combined results suggest that HDAC4 is a critical component of the DNA damage response pathway that acts through 53BP1 and perhaps contributes in maintaining the G2 cell cycle checkpoint.
A number of proteins, including 53BP1 (Schultz et al., 2000; Anderson et al., 2001; Rappold et al., 2001; Xia et al., 2001), Rad51, γH2AX, NBS1/Mre11/Rad50 complex (Maser et al., 1997; Paull et al., 2000), and BRCA1 (Scully et al., 1997; Zhong et al., 1999) have been observed to accumulate at multiple foci in the nucleus in response to DNA damage. 53BP1 was first identified in a yeast two-hybrid screen as one of two proteins that interacted with the transactivation domain of p53. Recently it was shown that 53BP1 participates in the phosphorylation of p53 and Chk2 and in the maintenance of S and G2 cell cycle checkpoints after DNA damage (DiTullio et al., 2002; Fernandez-Capetillo et al., 2002; Wang et al., 2002). 53BP1 foci are detectable within minutes after exposure to ionizing radiation, the number of foci increases with increasing dose, and the foci colocalize with that of γH2AX, suggesting an early role in the DNA damage response for 53BP1, as perhaps a marker of unrepaired double-strand breaks. Recent studies of cell lines derived from H2AX knockout mice show that this gene mediates radiation resistance and is essential for various components of the DNA damage response pathway to form foci (Bassing et al., 2002; Celeste et al., 2002). Although these data point to a role of chromatin in the DNA damage response, it remains unclear to what extent these foci represent alterations of the underlying chromatin.
Chromatin undergoes expansion and compaction in the course of many fundamental cellular processes, including gene expression, differentiation, and cell cycle progression. These alterations of the chromatin are largely mediated by histone acetylases and histone deacetylases (HDACs).*HDACs act on key acetylated lysine residues of core histones to induce chromatin compaction, which in turn results in gene silencing and heterochromatin formation (Taunton et al., 1996; Yang et al., 1996; Fischle et al., 1999; Grozinger et al., 1999). HDACs have been categorized into three classes that are based on sequence homology and domain organization. Class I HDACs (HDAC1, HDAC2, HDAC3, and HDAC8) are similar to the yeast deacetylase Rpd3 (Yang et al., 1996; Dangond et al., 1998; Emiliani et al., 1998; Buggy et al., 2000; Hu et al., 2000). Class II HDACs (HDAC4, HDAC5, HDAC6, and HDAC7) possess catalytic domains homologous to that of yeast Hda1 (Rundlett et al., 1996; Fischle et al., 1999; Grozinger et al., 1999; Miska et al., 1999; Verdel and Khochbin, 1999; Wang et al., 1999; Kao et al., 2000). Proteins similar to the yeast NAD1-dependent deacetylase Sir2 (Frye, 2000; Imai et al., 2000; Landry et al., 2000; Smith et al., 2000) compose the third class of HDACs. Class I and II HDACs have been found to function as corepressors recruited for transcriptional repression, whereas the class III HDACs are important for gene silencing at telomeres and HM (mating type) loci in yeast (Sherman et al., 1999).
Although HDACs are most prominently linked with transcriptional repression, there are indications that HDACs may play broader roles in regulating cellular processes that affect survival after exposure to DNA-damaging agents. For example, p53 has been found to be deacetylated and inactivated by human Sir2α in mammalian cells, leading to reduced apoptosis and increased survival after exposure to ionizing radiation or etoposide. Conversely, expression of catalytically inactive Sir2α leads to increased apoptosis and radiosensitization in mammalian cells (Luo et al., 2001; Vaziri et al., 2001). Effects of the class III HDACs may extend beyond effects on p53. In budding yeast, members of the SIR2 family of HDACs appear to be important components of the DNA damage pathway. Whereas SIR2 appears to be static, SIR3 and SIR4 are mobilized to sites of DNA breaks (Gottschling et al., 1990; Tsukamoto et al., 1997; Martin et al., 1999; Moazed, 2001). SIR2, SIR3, and SIR4 are required for the efficient recircularization of linear plasmids by nonhomologous end joining, and mutations in these genes resulted in increased sensitivity to γ-radiation in fission yeast (Boulton and Jackson, 1998; Tsukamoto et al., 1997). A role in the DNA damage response has not yet been described for the class I and II HDACs, but is suggested by the observation that treatment with the class I and II HDAC inhibitor trichostatin A (TSA) leads to significant radiosensitization of human cells (Biade et al., 2001) (the class III Sir2 family deacetylases are resistant to TSA [Bernstein et al., 2000]). The mechanism of how TSA might lead to radiosensitization has likewise been unclear.
To assess for a role for class I and II HDACs in the DNA damage response in human cells, we examined several members for their ability to form foci in response to DNA damage. After exposure to double-strand DNA damage in a variety of human cells, we found that human HDAC4 is recruited to nuclear foci with kinetics similar to 53BP1. We found that HDAC4 and 53BP1 colocalize at foci and could be coimmunoprecipitated. HDAC4 foci formation was seen in a wide variety of cell lines, which included radiosensitive cell lines lacking the gene products of ATM (ataxia telangiectasia mutated), Nibrin, and DNA protein kinase (DNA-PK). HDAC4 foci disappeared after several hours in repair-proficient cells that were exposed to nonlethal doses of radiation. However, HDAC4 foci failed to resolve in radiosensitive cell lines or in normal cells that were exposed to a lethal dose of radiation. Silencing of HDAC4 resulted in decreased 53BP1 protein levels and did not markedly affect cell cycle distribution in unirradiated cells, but markedly abrogated the G2 checkpoint in radiosensitized cells after DNA damage. These results together suggest that HDAC4 is a critical component of the DNA damage response pathway and suggests an additional role for this protein beyond transcriptional silencing.
HDAC4 is recruited to nuclear foci after exposure to double-strand DNA damage
To investigate a potential role of HDACs in the DNA damage response in human cells, we studied the response of HDAC2, 4, and 6 to DNA damage. HeLa cells exposed to γ-irradiation (IR) or etoposide exhibited distinct foci of HDAC4 in their nucleus (Fig. 1, B and D), whereas unirradiated (Fig. 1 A) or UV exposure did not have this effect (Fig. 1 E). In contrast, DNA damage did not induce foci formation by HDAC2 (Fig. 1 I) or HDAC6 (Fig. 1 J) or noticeably alter their intracellular localization. We observed IR-induced HDAC4 foci in both transformed and untransformed cell lines, including the breast cancer cell lines MCF7, PA1, SKBR, and MO59J and K (both of gliomatous origin), the sarcoma cell lines U20S, HT29, and HT1080 (latter two are colon cancer lines), and Wi38 and WSC (normal human fibroblasts) (unpublished data). Thus, foci formation by HDAC4 in response to DNA damage appears to be a general cellular response that does not appear to be dependent on p53.
The formation of HDAC4 foci after exposure to DNA damage was not accompanied by significant changes in HDAC4 deacetylase activity or protein levels (Fig. 2, A and B). We noted however a dose-dependent increase in the average number of HDAC4 foci per cell (Fig. 2 C). IR resulted in approximately 30 foci per Gy up to 3 Gy, beyond which the increase in the number of foci with increasing dose was difficult to discern. Kinetics of foci formation was the same between cells exposed to nonlethal (1 Gy) and lethal (8 Gy) doses (Fig. 2 D). HDAC4 foci were evident within 5 min after IR and reached a maximum by 1 h. Interestingly, the number of foci in cells irradiated with 1 Gy progressively decreased to background levels by 24 h after IR. In contrast, the majority (87 ± 7%) of HeLa cells irradiated with 8 Gy retained their HDAC4 foci as long as 48 h after IR. Foci were evident beyond 48 h when there was a high level of cell death (unpublished data). The persistence of HDAC4 foci therefore appears to correlate with lethal doses of IR, which presumably result in accumulation of unrepaired DNA.
HDAC4 interacts with 53BP1
DNA damage induces 53BP1 to form foci within minutes after DNA damage, suggesting that they participate in the early steps of the DNA damage response pathway. In contrast, Rad50 and Rad51 foci appear hours after DNA damage (Paull et al., 2000). We found that HDAC4 formed foci with kinetics very similar to 53BP1. The average number of foci containing either 53BP1 or HDAC4 reached maximal levels by 1 h after 3 Gy and then progressively decreased to near baseline by 36 h (Fig. 2 E). The similarity in the kinetics of HDAC4 and 53BP1 foci formation motivated us to look further for interactions between HDAC4 and 53BP1. Immunoprecipitation experiments showed that HDAC4 and 53BP1 formed a complex in HeLa cells, but the formation of this complex was unaffected by DNA damage (Fig. 2 F). However, immunocytochemical data clearly showed that HDAC4 and 53BP1 colocalized at DNA damage–induced foci (Fig. 2 G). At later times after radiation when Rad51 foci began to accumulate, the numbers of HDAC4 and 53BP1 foci were reduced. The majority of Rad51 foci did not colocalize with HDAC4, although it may be possible that Rad51 was localized to sites that had previously been occupied by HDAC4 (see Figs. S1 and S2).
HDAC4 foci induced by IR persist in radiosensitive cells
We next examined in a range of human cell lines the genetic determinants that specified HDAC4 foci formation. We found that HDAC4 foci formation did not depend on DNA damage response genes, i.e., ATM (ataxia telangiectasia mutated), Nibrin, or DNA-PK, as cell lines defective for these genes formed foci with similar kinetics as HeLa cells (Fig. 3). However, the repair-deficient cell lines differed from HeLa (and other repair-proficient) cells in the persistence of HDAC4 foci after exposure to low doses of IR. For example, HDAC4 foci readily formed 1 h after radiation in the ATM-deficient FT169 cell line, as well as in its isogenic derivatives Y25 (in which ATM is restored by expression of a full-length cDNA) and PEB (expressing empty vector and hence remaining ATM deficient). 24 h after IR, HDAC4 foci were significantly reduced only in the ATM-positive Y25 cells (Fig. 3, A–D). A similar difference in the resolution of HDAC4 foci was observed between DNA-PK–deficient MO59J cells and DNA-PK–positive MO59K cells. Although MO59K cells did not completely resolve their HDAC4 foci, the average number of foci was less than in the MO59J cells (Fig. 3, E–G). We believe that MO59K cells did not efficiently resolve HDAC4 foci because of their inherent radiosensitivity relative to HeLa cells (Wang et al., 1997; unpublished data), which efficiently resolves foci at low doses of IR (Fig. 3 J). Lastly, we examined HDAC4 foci formation in the radiosensitive Nijmegen breakage syndrome (NBS) mutant cell lines and found that they too retained high levels of foci 24 h after IR (Fig. 3, H and I). We found that foci formation by HDAC4 in HeLa cells was unimpeded by TSA. However, the resolution of HDAC4 foci in HeLa cells was partially inhibited by TSA (Fig. 3, H and K).
HDAC4 foci are detected in interphase cells
HDAC4 foci were induced by DNA damage in nearly all cells, suggesting it was not limited to a specific phase of the cell cycle. We sought to confirm this in both asynchronous and synchronized HeLa cells (Fig. 4 A). Using cyclin E expression to distinguish G1–S from G2 cells, we found IR-induced HDAC4 foci in both. Next, we irradiated synchronized cells in S phase and then harvested when cells were beginning to progress into G2 (Fig. 4 B) or mitosis (Fig. 4 C). The average number of foci per cell was not decreased in the G2 cells but, interestingly, was decreased in prophase cells and not evident in mitotic cells (Fig. 4 D). These observations, however, do not permit us to distinguish whether HDAC4 foci form less efficiently in mitotic cells or whether the foci resolve as cells enter mitosis.
Silencing of HDAC4 expression via RNA interference (RNAi) results in decreased 53BP1 protein
To assess further the functions of HDAC4, we silenced its expression by RNA interference (RNAi) (Elbashir et al., 2001). We found that short interfering RNA (siRNA) directed against two regions of HDAC4 efficiently repressed levels of HDAC4 protein at 48 h after transfection. A control siRNA that was unrelated to HDAC4 had no effect (Fig. 5 A). We conducted a time course experiment to determine the optimal time when HDAC4 expression was at its lowest after siRNA treatment. Considerable amounts of HDAC4 protein were detectable 12 h after transfection, but the level of protein progressively decreased such that very little HDAC4 protein was detectable by 36 h, indicating that it is optimal to assess silencing after 24 h (Fig. 5 B). We next examined the effect of the HDAC4 siRNA on foci formation after IR. As expected, HDAC4 foci formation after IR was substantially reduced with both HDAC4 siRNAs, whereas the control siRNA had no effect (Fig. 5, C and D).
We next explored the effect of silencing HDAC4 expression on 53BP1 and were surprised to find that 53BP1 levels were reduced in cells transfected with HDAC4 siRNA (Fig. 5 E). We conducted the converse experiment and found that silencing 53BP1 also led to a reduction of HDAC4. These effects were not seen with a control siRNA. We conducted a time course experiment with both HDAC4 and 53BP1 siRNA and found that levels of either target protein were substantially decreased 24 h after treatment with either siRNA, whereas levels of an unrelated protein (tubulin) were unaffected (Fig. 5 F). To exclude the possibility that the HDAC4 and 53BP1 siRNAs inadvertently precipitated the degradation of 53BP1 and HDAC4 mRNAs, respectively, we assessed the levels of endogenous HDAC4 and 53BP1 mRNA after siRNA treatment. We found that each siRNA reduced only its target mRNA (Fig. 5 G). Thus, the stabilities of HDAC4 and 53BP1 proteins are dependent on each other. Consistent with the observed reduction in protein expression, cells treated with 53BP1 or HDAC4 siRNA showed a dramatic reduction in 53BP1 foci in response to DNA damage (Fig. 5, H and I). Together, these experiments further suggest that HDAC4 interacts in the DNA damage response pathway that involves 53BP1, perhaps as components of the same protein complex.
Silencing of HDAC4 expression via RNAi results in decreased cell viability
We further studied the effects of silencing HDAC4 on cell cycle distribution and cellular viability after DNA damage. Treatment with HDAC4 siRNA did not appear to have a major effect in altering the cell cycle distribution of unirradiated HeLa cells, compared with cells treated with control siRNA (Fig. 6, A and B) or untreated cells (unpublished data). IR of HeLa cells treated with control siRNA resulted in the expected accumulation of cells in G2, characteristic of the IR-induced checkpoint. In contrast, treatment with HDAC4 siRNA markedly decreased the proportion of cells with G2/M DNA content but was accompanied by an increase in cells with DNA content less than 2N (sub-G1) (Fig. 6, A and B). This reduction in G2/M cells was not due to an S-phase delay because HDAC4 siRNA did not affect normal S-phase progression, as determined by FACS® analysis and BrdU incorporation (unpublished data). Flow cytometry based on DNA content alone does not distinguish between cells in G2 and mitosis. We therefore performed immunofluorescence on irradiated cells that were transfected with control and HDAC4 siRNA. Control cells were blocked in G2 after IR, as determined by high cyclin B1 levels in the cytoplasm, CENP-F distributed uniformly in the nucleus, and uncondensed chromatin (Fig. 6, C and D), as described previously (Liao et al., 1995; Kao et al., 2001). In contrast, few of the HDAC4 siRNA–treated cells appeared to be in G2. Although many of these cells still retained cyclin B1 and CENP-F staining, the cyclin B1 was often not uniformly cytoplasmic, the CENP-F staining of the nuclei was frequently uneven, and many of nuclei were condensed or fragmented. Many of these condensed cells (>70%), in fact, showed high levels of phosphohistone H3 staining, along with cyclin B1, suggesting progression into mitosis (see Fig. S3). These observations together suggest that cells silenced for HDAC4 do not maintain a uniform G2 delay after IR. One notes, however, that as 53BP1 has been established to mediate the damage-induced G2 checkpoint, the abrogation of this checkpoint by silencing HDAC4 may be mediated, at least in part, through decreasing 53BP1 protein (DiTullio et al., 2002; Fernandez-Capetillo et al., 2002; Wang et al., 2002).
To further explore the effects on cellular viability of silencing HDAC4 and 53BP1, we performed clonogenic assays to assess the effect on the survival of HeLa cells after DNA damage. Silencing of HDAC4 and 53BP1 had a dramatic effect on the plating efficiency of the cells (Fig. 6 E), indicating that HDAC4 and 53BP1 are likely key determinants of cell proliferation. However, even after correcting for the reduced plating efficiency of these cells, silencing of HDAC4 and 53BP1 further radiosensitized these cells (Fig. 6 F), with 53BP1 having the larger effect of the two. Together, these results suggest that HDAC4 and 53BP1 are required for cells to maintain in G2 after ionizing radiation; when either protein expression is silenced, cells do not block in G2, but instead rapidly lose viability.
Lastly, we examined whether abrogation of the G2 delay occurs when HDACs other than HDAC4 are silenced. Silencing of HDAC2 and HDAC6 did not interfere with the cell's ability to arrest in G2 in response to DNA damage (Fig. 7, A and B). Only silencing of HDAC4 appreciably diminished the proportion of cells delayed in G2 after IR.
Our studies have revealed a new role for a HDAC in the DNA damage response pathway. In response to double-stranded DNA breaks, HDAC4, but not HDAC2 or 6, rapidly formed foci that are coincident with the DNA damage response protein 53BP1. Foci formation by HDAC4 occurred throughout interphase and was observed in a wide panel of tumor and normal cell lines. HDAC4 foci formation did not depend on DNA damage response genes ATM, DNA-PK, or NBS, but their resolution was greatly delayed or blocked in these mutant cell lines. This is similar to the finding that ATM is not required for foci formation by 53BP1. When HeLa cells were exposed to a nonlethal dose of IR, where cells are presumably able to repair damaged DNA, HDAC4 foci were efficiently resolved. However, at lethal doses, where cells are presumably unable to effectively repair all of the damaged DNA, the HDAC4 foci persisted. We believe that the efficiency of resolution of HDAC4 foci correlates with cellular radiosensitivity. DNA-PK–defective MO59J cells are highly radiosensitive and were much less efficient at resolving their HDAC4 foci than DNA-PK–positive MO59K cells. MO59K cells, more radiosensitive than HeLa cells, were in turn less efficient at resolving foci than HeLa cells. The failure to resolve HDAC4 foci may reflect the radiosensitizing properties of TSA. These observations together suggest that HDAC4 foci formation is a general cellular response to DNA damage, and conversely, impaired resolution of HDAC4 foci may be correlated with radiosensitivity.
These data suggest that HDAC4 foci may facilitate recruitment of, or stabilize, repair factors such as 53BP1, which in turn may help maintain damage-induced cell cycle checkpoints. The interaction between HDAC4 and 53BP1 is established here by the findings that the two proteins colocalize and coimmunoprecipitate, as well as the unexpected findings that siRNA-mediated silencing of HDAC4 leads to reduced levels of 53BP1 protein and vice-versa. Thus, the stability of each protein appears to depend on the stability of the other, or on being associated in a common complex. Our findings that silencing of HDAC4 expression abrogates the damage-induced G2 delay and increases radiosensitivity support the notion that the checkpoint determines the ability of the cell to survive radiation damage. Along with 53BP1, localization of HDAC4 to unrepaired damage may signal the cell to stall at a checkpoint to allow the repair to be completed, while possibly serving as a marker to facilitate repair. These results, therefore, may be consistent with the model proposed by Fernandez-Capetillo et al. (2002) in which 53BP1 and other factors capable of signal transduction may accumulate into a chromatin microenvironment within megabases of an actual DNA double-strand break. This localized concentration of protein complexes may then generate the amplification of signal sufficient to invoke the G2 checkpoint.
These findings do not exclude roles for HDAC4 in mediating other cellular functions. HDAC4 is an established component of complexes mediating transcriptional repression (Guenther et al., 2001; Fischle et al., 2002). Although transcription is globally decreased during mitosis (Spencer et al., 2000; Pflumm, 2002), it appears unlikely that the role of HDAC4 in the DNA damage response is restricted to transcriptional repression. Radiation, at the doses used in this study, does not globally repress transcription (Maity et al., 1995), and drugs that globally inhibit transcription, such as α-amanitin, do not reverse the effects of silencing HDAC4 (unpublished data). The radiation dose dependency of the number of HDAC4 and 53BP1 foci does not exclude, but does reduce, the likelihood that foci are mere storage sites for these proteins.
The recruitment of HDACs to sites of DNA damage may be at least partially evolutionarily conserved. Members of the SIR2 family of HDACs have been implicated in the DNA response in budding yeast, contributing either directly or indirectly to the repair of double-strand breaks (Astrom et al., 1999; Bennett et al., 2001). Whereas SIR2 has been most prominently implicated in the silencing of chromatin at the mating type loci and at telomeres, SIR3 is recruited to sites of DNA damage to form foci that are visible at the cytological level. It has been proposed that the accumulation of some of these deacetylases may induce a repressed chromatin state to facilitate repair or protect unrepaired broken DNA ends (Tsukamoto et al., 1997; Martin et al., 1999). Although HDAC4 does not belong to the same family of deacetylases, we propose that HDAC4 may perform an analogous function in mammalian cells. As a component of protein complexes that mediate transcriptional repression by inducing heterochromatin formation, HDAC4 likely serves a similar role in processing chromatin in the DNA damage response. Recruitment of HDAC4 to foci after DNA damage might reflect its role in silencing the chromatin near the site of damage. The silenced chromatin would prevent processes such as DNA replication and transcription from passing through damaged DNA. In addition, the silenced chromatin might also reduce nonspecific end joining of broken DNA ends or unwanted recombination events.
In linking a HDAC with the DNA damage response to agents inducing double-strand breaks, the results presented here may also have clinical implications for treating patients with cancer. Inhibitors of HDAC inhibitors have entered clinical trials with reports of efficacy in certain tumors (Piekarz et al., 2001; Sandor et al., 2002). These protocols have largely entailed administering the HDAC inhibitor as a single agent. With further development and confirmation, the observations reported here suggest that it may be fruitful to pursue strategies to block HDAC function in combination with standard treatments, such as radiation and chemotherapy, to maximize the killing of cancer cells.
Materials And Methods
Polyclonal antibodies were generated by using full-length human HDAC2, 4, and 6 expressed in bacteria to immunize rabbits. For each antigen, the antibody was purified from at least two rabbits and confirmed to have equivalent specificity. The serum was affinity purified using protein immobilized on Affigel beads (Bio-Rad Laboratories). Specificity of the antibody was further confirmed by preabsorbing purified antibody with the immunizing antigen (Fig. S4). Anti-53BP1 antibody was generated against the corresponding domain in human 53BP1 encoded by the 1.5-kb HindIII–EcoRI fragment of XL53BP1 (the 53BP1 cDNA was a gift from Y. Adachi, University of Edinburgh, Edinburgh, UK). Anti-HDAC4, anti-53BP1, anti-Mad1 (gift from M. Campbell, Yen lab), anti–CENP-F (Yen lab), anti–α tubulin (Sigma-Aldrich), and anti–cyclin B1 (BD Biosciences) were used at 1:1,000 for staining and/or immunoblotting. Anti–cyclin E (HE12, sc-247; Santa Cruz Biotechnology, Inc.) and anti-Rad51 antibody (rabbit polyclonal, catalog no. PC130, Oncogene Research Products; mouse monoclonal, RAD51–14B4, GeneTex) was used at 1:500. The anti–phosphohistone H3 antibody recognizes histone H3 phosphorylated on serine 10 (no. 06–570; Upstate Biotechnology) and was used at 1:1,000. Cells were fixed with methanol/acetone 50:50 at the indicated times after IR before staining with the respective antibodies.
All cells were grown in DME medium supplemented with 20% FBS at 37°C with 5% CO2. The ATM-deficient FT169A (ATM−), ATM-restored YZ5 cells (ATM+, consisting of FT169A transfected with and stably expressing full-length ATM), and ATM-deficient vector-control PEB cells (ATM control, consisting of FT169A cells transfected with the parental vector only) were provided by Y. Shiloh (Tel Aviv University, Ramat Aviv, Israel). Nibrin-deficient human cells were obtained from the American Type Culture Collection. Cells grown on coverslips were irradiated with cesium-137 γ rays from a JL Shepherd and Associates 81–14R panoramic irradiator at a dose rate of 1.35 Gy/min. UV IR was delivered in a single pulse (50 J/m2) using a Stratalinker UV source (Stratagene). Before UV IR, the culture medium was removed, and the medium was replaced immediately after IR. All cells were returned to the incubator for recovery and harvested at the indicated times. Etoposide was used at 20 μg/ml for 20 min. To block cells in mitosis, cells were exposed to 0.04 μg/ml of nocodazole for 15 h.
RNAi was performed with siRNA that was commercially synthesized (Dharmacon) and used as described in protocols provided by the manufacturer. Cells were treated with siRNA to a final concentration of 10 μM. siRNA against HDAC6 was applied twice on consecutive days, whereas all other siRNAs were applied once and harvested as described for each experiment. Paired siRNA sequences targeting each protein were as follows: HDAC4, GACGGGCCAGUGGUCACUG and CAGUGACCACUGGCCCGUC; 53BP1, CACACAGAUUGAGGAUACG and CGUAUCCUCAAUCUGUGUG; HDAC2, GCCUCAUAGAAUCCGCAUG and CAUGCGGAUUCUAUGAGGC; HDAC6, CCAGCCAGCGAAGAAGUAG and CUACUUCUUCGCUGCCUGG. Control siRNA consisted of the unannealed single-strand RNA and siRNA targeted against luciferase (both of which did not affect levels of endogenous proteins).
HDAC assays were performed as previously described (Huang et al., 2000). In brief, [3H]acetylated histones purified from HeLa cells (25,000 cpm/10 μg) were incubated with enzymes at 37°C for 15 min. The reaction was stopped by the addition of concentrated HCl, extracted with 1 ml of ethylacetate, and the amount of radioactivity released into the organic layer was quantitated using a scintillation counter. In experiments in which HDAC4 deacetylase activity was inhibited, HeLa cells were pretreated with 1 μM TSA. For immunoprecipitations, for each sample, 0.1 μg of antibody or 10 μl of preimmune rabbit serum was incubated with cell lysate from 1.0 × 106 cells and incubated in the presence of ethidium bromide (10 μg/ml) to exclude nonspecific protein–DNA associations (Lai and Herr, 1992). Cell preparation, image acquisition and processing, and FACS® analysis were as previously described (Kao et al., 1997, 2001). Endogenous mRNA was isolated using Trizol reagent (GIBCO BRL), as per the manufacturer's instructions, and assessed via RT-PCR. The Titan One Tube RT-PCR System (Roche) was used with the following primers: HDAC4, CAAGAACAAGGAGAAGGGCAAAG and GGACTCTGGTCAAGGGAACTG; 53BP1, AGGTGGGTGTTCTTTGGCTTCC and TTGGTGTTGAGGCTTGTGGTGATAC; glyceraldehyde-3-phosphate dehydrogenase, CAACTTTGGTATCGTGGAAGGACTC and AGGGATGATGTT-CTGGAGAGCC (specific details regarding the PCR parameters used are available from the authors).
Plating efficiency was defined as the proportion of cells that remained viable 8 h after trypsinization and replating in fresh media. Cell viability was assessed by trypan exclusion. Clonogenic survival assays were performed as previously described (Biade et al., 2001), except that cells were counted and plated 48 h after treatment with siRNA and colonies of at least 50 cells were counted 14 d after plating. Statistical analyses were performed with SPSS for Windows Release 10 and Microsoft Excel (Office 2000).
Online supplemental material
Additional data (Figs. S1–S4) regarding antibody specificity, Rad51 localization after IR, and phosphorylated histone H3 (Pi-Histone H3) labeling of siRNA-treated cells are available as supplemental material.
We are grateful to Dr. T. Halazonetis for helpful comments. Drs. G.K.T. Chan and S.T. Liu, Mr. James Hittle, and Ms. Jennifer DeVirgilis in the Yen lab provided technical advice and assistance. Dr. Fang Liu and Ms. Kimberly Clay of the Kao lab assisted in the protein assays and RT-PCR.
G.D. Kao is a recipient of an award from the WW Smith Charitable Trust. T.J. Yen, R.J. Muschel, and W.G. McKenna are supported by PO1-CA75138-05. T.J. Yen was also supported by grants from the National Institutes of Health, PO1 core grant CA06927, and an Appropriation from the Commonwealth of Pennsylvania.
The online version of this article includes supplemental material.
Abbreviations used in this paper: ATM, ataxia telangiectasia mutated; DNA-PK, DNA protein kinase; HDAC, histone deacetylase; IR, irradiation; NBS, Nijmegen breakage syndrome; RNAi, RNA interference; siRNA, short interfering RNA; TSA, trichostatin A.