Radiosensitization by the Histone Deacetylase Inhibitor PCI-24781

The rationale for the development of histone deacetylase (HDAC) inhibitors as anticancer drugs resides in their ability to inhibit tumor growth, induce differentiation, and lower apoptotic threshold in transformed cells (1). PCI-24781 (previously called CRA-024781) is a hydroxamic acid–based HDAC inhibitor that was developed based on in vivo efficacy and therapeutic index (2). It is currently undergoing testing for safety, tolerability, and pharmacokinetics in a phase I trial. Preclinical results have established growth inhibitory concentrations for several tumor cell lines, as well as tumor growth inhibition in three xenograft models using various treatment schedules (2). Obtaining a good therapeutic ratio for tumor versus normal tissue is essential, and hydroxamic acid–based drugs seem to inhibit tumor growth in animals at nontoxic doses (3, 4). Tumor cells are thought to be more sensitive than normal cells to both the growth inhibiting and apoptosis promoting effects of most HDAC inhibitors (1, 5).

HDAC inhibition by a variety of chemical inhibitors enhances cell killing by ionizing radiation (6). Proposed mechanisms involved in radiosensitization include changes to chromatin structure, as well as changes in gene expression that influence cell cycle progression, DNA repair, and/or stress signaling pathways. For the majority of the cell culture studies summarized in Table 1, HDAC inhibitors were given 24 h before irradiation, and clonogenicity was used as the end point for analysis. Mechanisms for radiation potentiation have not been fully defined for any of these drugs and are complicated by cell type–dependent differences in response. Redistribution of cells into more radiosensitive phases of the cell cycle is a potential explanation because both p21^WAF1 and p27^KIP1 accumulation have been reported after treatment with several HDAC inhibitors (7, 8). Microarray analysis with PCI-24781–treated cells has confirmed up-regulation of p21 and caspases and down-regulation of cyclins (2). Increases in DNA accessibility caused by changes in acetylation may also enhance DNA damage and repair more directly (9). Several recent articles summarized by Karagiannis and El-Osta (10) suggest that inhibition of DNA repair may underlie radiosensitization by many HDAC inhibitors. HDAC inhibition is also known to enhance the toxicity of several anticancer drugs that target DNA (e.g., etoposide, doxorubicin, cisplatin; ref. 9).

As a first step toward examining the ability of PCI-24781 to sensitize human tumor cells to ionizing radiation, we examined toxicity and radiosensitization of SiHa cervical carcinoma cells and WiDr colon carcinoma cells in vitro. Additional experiments with normal human fibroblasts were done to determine the potential of preferential tumor radiosensitization. As a dose-dependent increase in H2AX phosphorylation has been reported after PCI-24781 treatment (2), we also examined the effect of combining PCI-24781 and radiation on γH2AX expression. Results indicate that (a) the S-phase fraction decreases after treatment with high doses of PCI-24781, (b) cells in G₁ phase at the end of treatment are more resistant to both killing and radiosensitization, (c) a longer treatment duration rather than a higher treatment dose is critical for radiosensitization, and (d) the initial rate of rejoining of radiation-induced double-strand breaks is not inhibited by PCI-24781 but residual damage is increased.

Cell lines and drug treatment. Human SiHa cervical carcinoma cells (human papillomavirus–positive, normal p53) and WiDr colon carcinoma cells (mutant p53) were purchased from American Type Culture Collection and maintained in exponential growth by subcultivating twice weekly in MEM containing 10% fetal bovine serum (Hyclone standard). Human skin fibroblasts, passage 31, were established in our laboratory from a skin biopsy and grown in culture in MEM containing 10% fetal bovine serum.

PCI-24781, kindly supplied by Pharmacyclics, Inc., was dissolved in DMSO at a stock concentration of 10 mg/mL. Drug was diluted in MEM with 10% fetal bovine serum. Cells in exponential growth were seeded at a density of 5 × 10⁵ cells per 60 mm tissue culture dish and allowed to attach overnight. Drug was added, and after 4 to 24 h of drug treatment, drug containing any suspended cells was removed, and suspended cells were recovered by centrifugation. Attached cells were trypsinized and combined with suspended cells. Samples were centrifuged, and aliquots of single cells were plated for cell survival as measured using a standard clonogenic assay. Remaining cells were fixed in 70% alcohol.

SiHa multicell spheroids were initiated as a single-cell suspension in 200 mL spinner culture flasks containing 200 mL MEM plus 10% fetal bovine serum. Medium was replenished daily as spheroid grew to a diameter of ∼350 μm. Spheroids were incubated for 24 h with 0, 0.3, or 3 μmol/L PCI-24781. They were then incubated for 20 min with 1 μmol/L Hoechst 33342, disaggregated for 8 min with 0.25% trypsin, and sorted on the basis of Hoechst fluorescence concentration using a Becton Dickinson FACS 440 dual laser cell sorter (11). Sorted cells were exposed to 0 to 8 Gy X-rays and plated for survival.

Exposure to X-rays was conducted at ambient temperature with a 300-kVp X-ray unit at a dose rate of 4.9 Gy/min.

Flow cytometry analysis of γH2AX, DNA content, and acetylated histone H3. Cells fixed in 70% ethanol were centrifuged and rehydrated in TBS, rinsed, and incubated for 2 h with primary anti-γH2AX antibody (Upstate monoclonal; 1:500 dilution) or anti–acetylated histone H3 (Upstate polyclonal; 1:200 dilution in 4% fetal bovine serum and 0.2% Triton X-100 in TBS). After staining with primary antibody, cells were washed and resuspended in secondary Alexa-488 goat-anti-mouse or goat anti-rabbit IgG-conjugated secondary antibody (Molecular Probes; 1:200 and 1:400 dilution, respectively). After 1 h, cells were rinsed and resuspended in 4′,6-diamidino-2-phenylindole to stain DNA. Apoptotic cells were identified based on the 10-fold higher expression of γH2AX in these cells (12). Apoptotic cells and cell fragments with sub-G₁ DNA content were not included in the analysis of drug-induced changes in γH2AX. Analysis was conducted using a dual-laser Coulter Elite cell sorter; samples were gated based on forward and peripheral light scatter.

Cell sorting and elutriation. Cell sorting based on DNA content was used to enrich for cells in G₁ phase. SiHa cells that had been incubated for 24 h with 0 or 3 μmol/L PCI-24781 were incubated for an additional 30 min with 5 μmol/L Hoechst 33342 (Sigma). Hoechst 33342 fluorescence intensity was used to define gates for cell sorting using a Becton Dickinson FACS440 dual laser cell sorter. Sorted cells were analyzed for DNA content to determine sorting accuracy and plated for assay of clonogenic cell survival. Alternatively, sorted drug-treated cells were exposed to 0 to 8 Gy and analyzed for their response to radiation. Drug-treated SiHa cells were also enriched in various cell cycle phases using a Beckman centrifugal elutriator, which separates cells based on cell size. SiHa cells were separated at the speed of 1,800 rpm. Populations obtained at various flow rates were fixed and analyzed for DNA content or plated for cell survival. As cells in G₁ phase were much more resistant to killing by PCI-24781 than cells in other phases, results are presented as the percentage of cells in G₁ phase versus clonogenic surviving fraction.

Measurement of γH2AX foci size and number. Primary human fibroblasts were allowed to attach to sterile slides, then incubated for 24 h with 0 or 3 μmol/L PCI-24781, exposed to 4 Gy, and fixed for 0.5 h in 2% paraformaldehyde in PBS. Cells were rinsed in TBS, placed briefly in −20°C methanol, rinsed, then placed for 20 min in TBS containing 1% bovine serum albumin and 0.2% Tween 20 (TTN), and finally incubated for 2 h with mouse monoclonal anti-γH2AX diluted 1:500 in TTN. Slides were washed and incubated with Alexa 488–conjugated goat anti-mouse IgG (H + L)F(ab′)² fragment conjugate (Molecular Probes) diluted 1:200 in TTN for 1 h at room temperature. Slides were rinsed then immersed in 0.05 μg/mL 4′,6-diamidino-2-phenylindole for 5 min, rinsed and mounted with coverslips using 10 μL Fluorogard (Bio-Rad) as the antifade mounting medium. Slides were viewed using a Zeiss fluorescent microscope using a 100× Neofluor objective and a Q-Imaging 1350 EX digital camera. Images were analyzed using Northern Eclipse 6.0 software (Empix Imaging). Individual nuclei were analyzed for foci number and area using Scion image software.

Double-strand break detection using the neutral comet assay. Cells attached to tissue culture dishes were exposed for 20 h to 0.3 or 3 μmol/L PCI-24781 and then exposed to 75 Gy. Drug remained in the plates until 4 h after irradiation. At various times after irradiation, cells were trypsinized, and single cells were mixed with agarose. The neutral comet assay was used as previously described (13). Briefly, samples were placed in lysis solution containing 2% sarkosyl, 0.5 mol/L Na₂EDTA, and 0.5 mg/mL proteinase K (pH 8.0). The samples were incubated overnight at 37°C after rinsing in Tris-borate EDTA buffer (pH 8.0). Samples were electrophoresed for 20 min at 0.6 V/cm in Tris-borate EDTA buffer and then stained with propidium iodide. Slides were viewed using a fluorescence microscope with a CCD camera, and 150 individual comet images were analyzed from each sample for tail moment, DNA content, and percentage of DNA in tail (14). Apoptotic comets (>90% DNA in the tail) were not included in the analysis. Results were normalized to fraction of initial damage remaining at various times after irradiation, and two independent experiments are reported.

PCI-24781 toxicity. Increases in histone acetylation have been shown to occur quickly after the start of treatment with PCI-24781 or its earlier congener CRA-026440 (2, 15). At the earliest time measured here, 4 h, maximum levels of acetylated histone H3 were observed. However, no significant cell killing or cell cycle redistribution were observed for SiHa cells exposed to 1 or 10 μmol/L PCI-24781 until the incubation time exceeded ∼8 h (Fig. 1A).

The responses of SiHa and WiDr cell lines to increasing concentrations of PCI-24781 were measured after 4 or 24 h (Fig. 1B and C). Near-maximum histone acetylation was observed at the lowest dose examined in both cell lines, 0.1 μmol/L. The basis for the differences in maximum levels of histone H3 acetylation in SiHa and WiDr cells is not known but could involve expression of the various HDAC isoforms which are differentially inhibited by PCI-24781 (2). There was also a small but significant increase in the level of the DNA damage response indicator, phosphorylated histone H2AX. Apoptotic cells that develop with longer incubation times also express γH2AX but were not included in this analysis. As seen for histone acetylation, near-maximum phosphorylation of H2AX was observed at the lowest dose examined.

There was no evidence of toxicity when SiHa or WiDr cells were incubated for only 4 h with PCI-24781. However, treatment for 24 h produced an atypical dose dependency with a threshold for toxicity between 0.1 and 0.3 μmol/L (Fig. 1B and C). Above 1 μmol/L, no additional cell killing was observed for either cell line. This pattern was also observed for the percentage of apoptotic cells, number of cells recovered (indicative of growth inhibition), and relative number of cells in S phase. All end points showed maximum effects at ∼1 μmol/L for both cell lines. SiHa cells, which are more sensitive to killing by ionizing radiation and several anticancer drugs than WiDr cells (16), also showed a lower clonogenic survival than WiDr cells after exposure to PCI-24781.

Significant depletion of S-phase cells occurred for both SiHa and WiDr cells exposed to doses of PCI-24781 >0.3 μmol/L (Figs. 1B, C and 2A, B). The fact that PCI-24781 toxicity did not increase above 1 μmol/L suggested the presence of a drug resistant subpopulation. Centrifugal elutriation and cell sorting experiments confirmed that cells in G₁ phase at the end of drug treatment were much more likely to survive than cells in other phases of the cell cycle (Fig. 2C). The structural formula of PCI-24781 is also shown in Fig. 2C.

High amounts of γH2AX are expressed when DNA fragmentation occurs during apoptosis (12, 17). Although cells in all phases of the cell cycle were capable of undergoing apoptosis, by 24 h after the start of drug treatment there was a higher percentage of apoptotic SiHa cells with G₁-S DNA content and a higher percentage of apoptotic WiDr cells with G₂-M DNA content (Fig. 2D). WiDr cells that had been exposed for 24 h to PCI-24781, then irradiated and examined 4 h later, exhibited a third population intermediate in γH2AX intensity between normal and apoptotic cells (arrow in Fig. 2D). γH2AX expression in this third population is consistent with the larger γH2AX focus size that develops in cells in mitosis at the time of irradiation (18). A similar population was not evident in drug-treated SiHa cells.

PCI-24781 radiosensitization. Preliminary experiments indicated that histone acetylation dropped to background levels 2 h after removal of 0.3 μmol/L PCI-24781 and 8 h after removal of 3 μmol/L PCI-24781. Therefore, cells were incubated with PCI-24781 both before and after irradiation to maintain high levels of histone acetylation during the repair period. No radiosensitization was observed for SiHa or WiDr cells incubated for 2 h before and 2 h after irradiation with 1 or 10 μmol/L PCI-24781 (Supplementary Fig. S1). However, radiosensitization was observed for both cell lines incubated for 20 h before and 4 h after irradiation with 0.3 or 3 μmol/L PCI-24781 (Fig. 3A). Doses as low as 0.1 μmol/L PCI-24781 were able to sensitize SiHa cells to killing by 6 Gy, thus clearly separating the PCI-24781 dose required for toxicity from the dose required for radiosensitization (Fig. 3B). Although the lower drug dose was less cytotoxic and showed minimal cell cycle perturbations, both doses produced a similar dose-modifying factor (DMF; Fig. 3C and D). The DMF for a surviving fraction of 0.10 was 1.2 for WiDr cells and 1.5 for SiHa cells. If radiosensitization by PCI-24781 can be explained by the decrease in the fraction of radioresistant S-phase cells, then cells in G₁ phase should show much less (if any) radiosensitization. Consistent with this idea, contact-inhibited normal human fibroblasts in G₀ phase showed less killing and only a small DMF of 1.1 (Fig. 3E).

The importance of cell cycle was examined directly by determining whether SiHa or WiDr cells in G₁ phase at the time of irradiation would be sensitized by exposure to PCI-24781 before irradiation. SiHa cells were incubated for 24 h with 0 or 3 μmol/L PCI-24781 and then incubated for a further 30 min with 5 μmol/L Hoechst 33342. Cells were then sorted on the basis of DNA content as indicated by Hoechst 33342 fluorescence intensity. Sorted populations were fixed and examined for DNA content (Fig. 4A), and results confirmed excellent enrichment for the G₁ population (∼94% purity). Cells from the same sort fractions were exposed to 0 to 8 Gy and plated for survival (Fig. 4B and C). Of some importance in terms of mechanism, there was no postirradiation drug incubation in these experiments, yet radiosensitization was still observed. High concentrations of Hoechst 33342 that are necessary to obtain stoichiometric binding to DNA are known to protect against radiation damage (19). In experiments by Denison and Martin (20), a DMF of 1.3 at 10% survival (protection in this case) was observed for cells treated with 20 μmol/L Hoechst 33342. Radioprotection was confirmed in this study because the dotted lines in Fig. 4B, C lie below the responses for Hoechst 33342-treated cells. Because the radioprotective effect of Hoechst 33342 on PCI-24781–treated cells was less than the effect on untreated cells, the DMF was greater in these experiments. The reason for the lack of radioprotection in the PCI-pretreated cells may relate to changes in chromatin structure, although Hoechst treatment did not affect survival in PCI-24781–treated cells. Regardless of mechanism, PCI-induced radiosensitization, although diminished relative to S-G₂ cells, was clearly evident in the G₁ sorted cells. Similar results were obtained for WiDr cells (Supplementary Fig. S2).

The importance of proliferation during drug treatment was examined by exposing SiHa multicellular spheroids to PCI-24781 for 24 h followed by cell sorting, irradiation, and plating for cell survival. Cell proliferation was significantly reduced in the innermost cells of spheroids that are enriched for G₁ phase. In this experiment, the Hoechst 33342 fluorescence gradient was used to identify the position of individual cells on the outside versus inside of the spheroids (11). As a result of the steep Hoechst gradient, the innermost cells from spheroids were exposed to 50-fold to 100-fold lower doses of Hoechst 33342, which are too low to cause radioprotection (21). As shown in Supplementary Fig. S3, the largely noncycling (<3% S phase cells) inner cells of spheroids were still sensitized to PCI-24781, although both toxicity and the extent of sensitization were reduced relative to the external spheroid cells or to asynchronously growing monolayers (Fig. 3C). Drug penetration did not seem to be a consideration because the toxicity of PCI-24781 was the same for inner versus outer cells of spheroids exposed to 3 μmol/L PCI-24781 (dotted lines in Supplementary Fig. S3).

DNA double-strand break rejoining. In several studies, HDAC inhibitors have been shown to increase the retention of radiation-induced γH2AX foci, an indication of possible inhibition of double-strand break repair. PCI-24781 also increased γH2AX when analyzed by flow cytometry; however, this increase was relatively small (40%) and was enhanced to the same extent in untreated, as well as irradiated, cells (Fig. 5A). Rate of loss of γH2AX after irradiation can be delayed in repair deficient cells and by treatments that affect double-strand break rejoining (22, 23). However, after correcting for DNA content, the initial rate of loss of γH2AX after irradiation was not affected in cells exposed to PCI-24781 treatment for 24 h before irradiation (Fig. 5A). Similar results were obtained with another hydroxamic acid, TCA (Supplementary Fig. S4). The fact that γH2AX foci intensity increased in parallel with the increase in histone acetylation suggests that these two events are related. At 24 h after irradiation, the fraction of cells lacking γH2AX foci was significantly lower in PCI-24781–pretreated cells compared with controls and was consistent with the extent of radiosensitization (Fig. 5B). To determine whether foci number or foci size might be affected by PCI-24781 treatment, individual images of G₁ fibroblast nuclei were examined for γH2AX foci. PCI pretreatment did not significantly increase foci number, but foci size was increased when examined 4 h after 1 Gy in cells that had been pretreated for 24 h (Fig. 5C). However, at 24 h after irradiation and drug removal, there was no apparent difference in size of residual foci.

The neutral comet assay was used to determine whether physical rejoining of radiation-induced DNA double-strand breaks would be delayed in cells exposed to 0.3 μmol/L PCI-24781 for 20 h before and 4 h after irradiation. No differences in either initial rate of rejoining or amount of residual double-strand breaks measured 24 h after irradiation were observed for cells that had received radiation only or radiation after PCI-24781 treatment (Fig. 5D).

HDAC inhibitors have been shown to sensitize tumor cell lines and rodent tumors to ionizing radiation, so it is not surprising to find that PCI-24781 also acts as a radiosensitizer in two human tumor cell lines. Although several potential mechanisms of radiosensitization by HDAC inhibitors have been proposed, the mechanism for sensitization by PCI-24781 has not been examined and is relevant to its optimal combination with radiation in the clinic. Results shown here point to misrepair of radiation damage, but this will require further substantiation. Our results also rule out two potential mechanisms of sensitization: redistribution to a more radiosensitive cell cycle phase and inhibition of initial rejoining of radiation-induced double-strand breaks.

Initial experiments examined the toxicity of PCI-24781 in exponentially growing SiHa and WiDr cell lines. Histone acetylation was near maximum after the lowest dose (0.1 μmol/L) and treatment time (4 h; Fig. 1) in line with previous results (2). This contrasts to results using another hydroxamic acid, trichostatin A, that shows increases in histone acetylation with increasing drug dose (10). Microarray analysis has shown changes in expression of ∼15% of all expressed genes within 2 to 4 h after PCI-24781 treatment, including those involved in apoptosis and cell cycle regulation (2). However, we observed no loss of clonogenicity after a 4-h exposure of SiHa or WiDr cell lines to doses of PCI-24781 up to 10 μmol/L. Apparently, HDAC inhibition by PCI-24781 must be maintained for a longer period to effect changes in target proteins. This is consistent with a study that reported that, in spite of the fact that valproic acid could increase histone acetylation within 30 min, a 50% decrease in expression of chromatin maintenance proteins and dispersion of heterochromatin required exposure for 48 h (24).

When the treatment time was extended beyond 4 to 8 h, toxicity gradually increased. However, there was a switch from no toxicity to maximum toxicity over a relatively narrow drug concentration range (0.1-1 μmol/L). All measures of toxicity, including loss of clonogenicity, apoptosis, growth inhibition, and cell cycle redistribution, showed a similar pattern in terms of dose and time responses. This contrasts to reported dose-dependent toxicities for other hydroxamates, such as SAHA (25, 26), and suggests that the targets for these inhibitors may differ. About 20% to 40% of the cells seemed resistant to cell killing; although with longer exposure times, this population would be expected to decline. Cell sorting confirmed that surviving SiHa cells were likely to be in G₁ phase at the end of the 24-h treatment. It is possible that blockage of cells in G₁ phase may protect cells from permanent damage. In line with this idea, primary skin fibroblasts maintained in G₀ phase showed only 30% cell killing by a dose of PCI-24781 that caused 60% to 80% cell killing in the two asynchronously growing tumor cell lines.

Cell cycle variations caused by PCI-24781 and other HDAC inhibitors are cell type specific. A higher proportion of WiDr cells were blocked in G₂-M at the end of the 24-h treatment, in spite of the fact that both cell lines show a similar cell doubling time (20-22 h). Moreover, treatment with 0.3 μmol/L PCI-24781 for 24 h reduced the S-phase content of WiDr cells to very low levels but caused a minimal reduction in SiHa cells (Fig. 1). The greater apoptosis and cell loss in WiDr cells may explain this difference. WiDr cells, which express mutant p53, showed a greater accumulation of cells in M phase than SiHa cells which have normal p53 but express human papillomavirus E6 protein. The arrest of WiDr cells in mitosis, as indicated by the γH2AX staining pattern after irradiation, is consistent with drug-induced blockage in mitosis rather than the conventional G₂ block typically associated with HDAC inhibitors (1). The role of p53 in the response to HDAC inhibition has been discussed in several studies (27–29). Flatmark et al. (27) reported that cells with functional p53 were depleted of G₁-phase cells after exposure to trichostatin A, whereas cells with mutant p53 accumulated in both G₁ and G₂-M, as was observed here.

There is conflicting evidence concerning the importance of cell proliferation for toxicity by HDAC inhibitors. Burgess et al. reported that the hydroxamic acid SBHA was equally effective in inducing apoptosis in proliferating and nonproliferating tumors cells, with no effect on normal cells (30). However, a leukemic cell line that was induced to express the cyclin-dependent kinase inhibitor p16 was resistant to the toxic effect of HDAC inhibitors (31). Our results indicate that noncycling fibroblasts are more resistant to killing, SiHa cells in G₁ phase at the end of a 24-h treatment are more likely to survive than cells in other phases, and SiHa cells in spheroids (both outer and inner cells) showed a higher survival after PCI treatment than monolayers exposed to the same dose. Cells in G₁ phase would be more dependent on the nonhomologous end-joining pathway for repair of radiation-induced double-strand breaks and, therefore, more prone to errors in rejoining. In line with this idea, greater sensitivity of nonhomologous end-joining–deficient cells to trichostatin A toxicity has previously been reported (32).

Hydroxamic acid–based drugs are reversible in their ability to inhibit HDACs. The lack of radiosensitization for the 4-h exposure to PCI-24781 is consistent with a requirement for a longer exposure to allow for the effects of changes in gene transcription to become manifested as changes in proteins. The degree of radiosensitization in SiHa and WiDr cells after 24 h is similar to that reported for several other HDAC inhibitors (Table 1). Radiosensitization by PCI-24781 was similar for the high and low drug doses tested in spite of the fact that 0.3 μmol/L produced only modest effects in terms of cell killing and cell cycle redistribution; even 0.1 μmol/L produced similar radiosensitization in SiHa cells (Fig. 3H). Zhang et al. (33) also reported that radiosensitization by trichostatin A was achieved at a dose that was lower than that required for other antiproliferative responses. For PCI-24781, loss of radioresistant S-phase cells could explain part of the radiosensitization, an idea that was supported by the inability to sensitize G₁ fibroblasts. However, it is apparent from the fact that low drug doses are as effective at sensitizing cells as high doses and, from the results with sorted populations in Fig. 4, that mechanisms other than cell cycle redistribution must be involved in sensitization. Experiments with multicell spheroids confirm that noncycling tumor cells are more resistant to radiosensitization than proliferating cells (Supplementary Fig. S3).

As previously reported for sodium butyrate (34), vorinostat (26), and valproic acid (35), we found that phosphorylation of histone H2AX was enhanced by treatment with PCI-24781. However, once correction was made for apoptosis and cell cycle redistribution, the increase in γH2AX was relatively small (∼40% increase above endogenous levels). The fact that high and low drug doses produced similar increases in γH2AX indicates that H2AX phosphorylation can be dissociated from toxicity (Fig. 1). Moreover, PCI-24781–induced increases in γH2AX were observed for endogenous foci, as well as radiation-induced foci (Fig. 5A), also found to be the case for cells exposed to sodium butyrate (34). In human fibroblasts that were not significantly sensitized by PCI-24781, we saw an even larger increase in γH2AX. The fact that exposure for 4 h produced a similar increase in γH2AX intensity as exposure for 24 h indicates that foci size can also be dissociated from radiosensitization.

Although changes in histone γH2AX foci intensity could be relevant to DNA repair, it is typically the kinetics of foci loss or degree of retention, not the intensity of the foci themselves, that is correlated with radiosensitivity (22, 23). The lack of differences in rate of rejoining was measured using the comet assay, and the lack of an effect on the rate of loss of γH2AX after irradiation (Fig. 5) do not support a role for PCI-24781 in direct inhibition of double-strand break rejoining after irradiation. Moreover, in spite of the fact that the 4-h postirradiation drug treatment was omitted from the cell sorting experiments, radiosensitzation was still observed, so the drug need not be present during rejoining to cause radiosensitization. It is more likely that accuracy of repair of radiation-induced DNA damage is compromised by changes in gene expression or chromatin structure. This is consistent with the observation that a higher fraction of cells contained γH2AX foci 24 h after irradiation when they had been pretreated with PCI-24781 (Fig. 5B). We and others have noted that a deficiency in ATM does not affect initial DNA rejoining kinetics in the comet assay (36, 37), although lack of ATM can cause misrejoining and an increase in residual damage (38). Cells deficient in other genes involved in homologous recombination, like RAD51, Bloom syndrome, and BRCA1, do not seem to rejoin breaks more slowly than repair-proficient cells (39–41). Previous reports that HDAC inhibitors may decrease the expression of genes involved in homologous recombination and nonhomologous end-joining (Table 1) and a recent abstract indicating that PCI-24781 reduces RAD51 expression and homologous recombination (42) support the idea that fidelity of repair may be compromised.

In summary, our results indicate that PCI-24781 sensitizes tumor cells to ionizing radiation. Although inhibition of histone deacetylation occurs rapidly, toxicity and radiosensitization require exposures longer than 4 h. Radiation resistant S-phase cells are depleted during a 24-h pretreatment with high dose of PCI-24781, and cells may also block and die in G₂-M during this period. Cells that survive the toxic effects of PCI-24781 are more likely to be in G₁ phase, thus more sensitive than asychronous cells to killing by ionizing radiation. However, as G₁ cells can also be sensitized by PCI-24781, and minimally toxic doses are effective in sensitization, cell cycle redistribution is not a major mechanism for radiosensitization. There is, however, some evidence that cell proliferation during drug exposure may affect the degree of sensitization, and DNA repair fidelity may be compromised by PCI-24781 treatment. Noncycling human fibroblasts are more resistant to the toxic and radiosensitizing effects of PCI-24781, supporting the potential for an improvement in therapeutic ratio for the combination of PCI-24781 and ionizing radiation.