Telomere Shortening Alters the Kinetics of the DNA Damage Response after Ionizing Radiation in Human Cells

There has been growing recognition of a close relationship between telomeres and the DNA damage response (1). Proteins involved in the DNA damage response localize to telomeres and are required for normal telomere maintenance yet are also involved in the cellular response to telomere shortening and dysfunction (2). Cells with short or dysfunctional telomeres activate a damage response similar to the response observed with DNA double-strand breaks, including activation of ATM, phosphorylation of ATM targets, and formation of nuclear foci containing protein complexes involved in DNA repair (3–5). The telomere-induced damage response arises, in part, from direct uncapping of telomeres and is not solely attributed to end-to-end fusions of dysfunctional telomeres leading to chromosome breakage during anaphase.

The link between telomeres and the DNA damage response has been furthered by several studies, both at cellular and organismal levels, showing that telomere shortening and dysfunction are determinants of radiation sensitivity (6–11). However, the molecular basis underlying this sensitivity remains unclear. Understanding how telomere integrity influences sensitivity to DNA-damaging agents is especially relevant, given the emergence of telomerase inhibition as a therapeutic modality for human cancers. In the present study, we compared the cellular response to ionizing radiation (IR) in isogenic human cells with short and long telomeres. Our data show that the kinetics of the DNA damage response is altered in cells with short telomeres and suggest that telomere shortening is associated with chromatin structure changes that limit access of activated ATM to its chromatin-bound downstream targets.

The ATCC CRL-2091 normal primary human foreskin fibroblast strain (HFF) and the HeLa human cervical carcinoma cell line were obtained from the American Type Culture Collection. 293T cells were a gift from Dr. David Baltimore; no authentication of 293T cells was done by the authors. Cells were propagated in Dulbecco's Modified Eagle's Media (Biowhittaker) supplemented with 10% FBS (HyClone), l-Glutamine, and penicillin/streptomycin (Gibco) at 37°C and 5% CO₂ in a humidified incubator. Fibroblasts were trypsinized and counted every 7 days and reseeded at a density of 5 × 10⁵ cells per 75 cm² flask. HeLa cells were counted every 3 to 4 days and replated at a density of 0.5 × 10⁶ to 1 × 10⁶ cells per 100 mm dish. The histone deacetylase (HDAC) inhibitor vorinostat (LC laboratories) was dissolved in dimethyl sulfoxide and applied to cells at a concentration of 2 μmol/L. Cells were treated with vorinostat for 24 hours prior to irradiation.

Plasmids containing human TERT cDNA (pCI-neo-hEST2) were a gift from Dr. Robert Weinberg (Whitehead Institute for Biomedical Research, Cambridge, MA). Catalytically inactive TERT-dominant negative (TERT-DN) was generated by substituting the aspartic acid residues with valine residues at positions 868 and 869 by site-directed mutagenesis (Quick Change Site-directed Mutagenesis kit; Stratagene). The TERT and TERT-DN mutants were sequenced completely and subcloned into the retroviral expression vector MSCV-IGFP, which expresses the GFP reporter from an internal ribosomal entry site (IRES). Vesicular stomatitis virus-G (VSV-G)-pseudotyped retroviruses were packaged in 293T cells transfected with the plasmids pEQ-PAM3-E (containing the helper proteins), VSV-G (containing the envelope), and MSCV-IGFP, MSCV-IGFP-TERT, or MSCV-IGFP-TERT-DN. Virus-containing medium was applied to the cells of interest in 3 separate aliquots with polybrene, as previously described (12). Forty-eight hours after transduction, GFP-containing cells were isolated by fluorescence-activated cell sorting (FACS). The experiments described were carried out on HFFs taken in the early passages after transduction with the hTERT retrovirus when a sufficient number of cells were available after FACS sorting (estimated to be several population doublings). HeLa cells transduced with the mutant TERT-DN constructs were cloned by the limiting dilution method. The process of generating visible HeLa cell colonies required 20 population doublings.

Cell extracts were prepared according to protocols provided by the manufacturers. Telomerase activity was measured using the TRAPeze Telomerase Detection Kit (Intergen). For telomere length analysis, 5 μg genomic DNA was digested with HinfI and RsaI and analyzed by southern blotting using ³²P-labeled (TTAGGG)₆ probe. Signals were visualized using a PhosphorImager (Storm 860, Molecular Dynamics).

Cells were seeded onto 6-well plates in concentrations predicted to yield 100 to 200 colonies per well. Increasing cell concentrations were used for increasing doses of IR. For the IR experiments, cells were subjected to the indicated doses of IR 20 hours after seeding. Two weeks after irradiation, the cells were fixed and stained with 0.01% formaldehyde and 0.1% crystal violet. Colonies containing at least 50 cells were counted.

All procedures are described in the works of Bakkenist and colleagues (3) and Kitagawa and colleagues (13). Briefly, cells extracts were prepared in radioimmunoprecipitation assay (RIPA) buffer (200 mmol/L NaCl, 10 mmol/L Tris at pH 7.5, 0.1% SDS, 1% NP-40, 0.5% deoxycholate, 1× protease inhibitor cocktail tablet from Roche), anti-SMC1 (BL308; Bethyl Laboratories), anti-SMC1S957-P, anti-p53 (Ab-6; Oncogene Research Products), anti-p53S15-P (Cell Signaling Technology), anti-β-actin (Novus Biologicals), anti-acetyl-Histone H3K9 (Upstate), anti-dimethyl-Histone H3K9 (Upstate), anti-H3, CT, pan, clone A3S (Upstate), anti-γH2AX (Millipore), anti-NBS1 (Novus Biologicals), and anti-NBS1S343-P (Signalway Antibody).

Extracts containing soluble histone proteins were prepared as described previously (13). In brief, cell pellets were resuspended in hypotonic buffer (10 mmol/L HEPES at pH = 7.9, 1.5 mmol/L MgCl₂, and 10 mmol/L KCl). Cells were then lysed in a hypertonic lysis buffer (10 mmol/L HEPES at pH = 7.9, 1.5 mmol/L MgCl₂, 250 mmol/L NaCl, 10 mmol/L KCl, 0.2 mol/L HCl, 0.5 mmol/L DTT, 1.5 mmol/L phenylmethylsulfonylfluoride, and 1× protease inhibitor cocktail tablet from Roche). Cell extracts were dialyzed against 0.1 mol/L acetic acid twice for 1 hour and 3 times against H₂O for 1 hour, 3 hours, and overnight, respectively.

Cells were permeabilized for 90 seconds with 0.01% egg-Lysolecithin (Sigma) diluted in 150 mmol/L sucrose, 80 mmol/L KCl, 35 mmol/L HEPES, pH = 7.4, 5 mmol/L K₂HPO₄, 5 mmol/L MgCl₂, and 0.5 mmol/L CaCl₂. Cells were then digested with 2 U/mL micrococcal nuclease (Sigma) in 20 mmol/L sucrose, 50 mmol/L Tris-HCl, pH = 7.5, 50 mmol/L NaCl, and 2 mmol/L CaCl₂ at room temperature for various time points. DNA was isolated and subjected to 6% PAGE to analyze the digestion products.

To evaluate the sensitivity of cells with short telomeres to IR, we used clonogenic survival assays at different population doublings of primary HFFs. We previously had shown that telomeres shorten in HFFs at each population doubling (3). Late-passage fibroblasts with short telomeres had significantly enhanced sensitivity to IR compared with early-passage fibroblasts (Fig. 1). To confirm that IR sensitivity is associated with short telomeres, we also introduced a catalytically inactive dominant-negative hTERT construct (hTERT-DN) to inhibit telomerase activity in HeLa cells. Transduced cells were cloned by the limiting dilution method. Sensitivity to IR was assessed at different population doublings using clonogenic survival assays. The introduction of hTERT-DN into HeLa cells resulted in complete inhibition of detectable telomerase activity (Supplementary Fig. S1A) and telomere shortening with successive population doublings (PD; Supplementary Fig. S1B). Cells with short telomeres were markedly more sensitive to IR than their counterparts with longer telomeres (Supplementary Fig. S1C).

Cellular survival following DNA damage depends on the integrity of the signaling pathways to repair DNA double-strand breaks (DSB) and restore genomic integrity. H2AX phosphorylation is an early step in the response to DNA damage. Upon irradiation, H2AX is rapidly phosphorylated by activated ATM and DNA protein kinases (DNA-PK) on serine 139 to generate γH2AX, a robust marker of DSBs (14, 15) that can be used to monitor DSB repair (16). Both activated ATM and γH2AX form nuclear foci upon DSBs. The extent of DNA damage post-irradiation and the kinetics of γH2AX foci formation and resolution were evaluated in late- and early-passage HFFs by immunofluorescence using γH2AX-S139 phospho-specific antibodies. Late-passage cells were presenescent, as evidenced by continued growth and lack of senescence-associated β-galactosidase staining (Supplemental Fig. S2). Although the cells were presenescent, they were growing at a slower rate than early-passage cells. To control for differences in cell growth, most of the experiments outlined below were carried out on contact-inhibited cells that were 3 weeks past the point of confluence.

As expected, at baseline before subjecting cells to IR, late-passage (PD 63) HFF cultures had a higher percentage of cells with γH2AX foci than early-passage cultures (PD 11). However, upon irradiation, a higher peak percentage of early-passage HFFs manifested γH2AX foci than late-passage HFFs (Fig. 2A). Moreover, the kinetics of γH2AX foci appearance and disappearance was faster in early-passage than late-passage HFFs (Fig. 2A and B). Similar results were obtained using Western blot analysis to assess γH2AX protein levels (Fig. 2C). Thirty minutes after 1 Gy of IR, early-passage HFFs showed a robust induction of H2AX phosphorylation, whereas late-passage HFFs did not increase γH2AX protein levels until 60 minutes after IR (Fig. 2C, lanes 3 and 4). At 5 hours post-irradiation, while γH2AX protein levels were still elevated in late-passage HFFs, γH2AX protein levels returned to near baseline levels in early-passage HFFs (Fig. 2C, lanes 9 and 10). γH2AX foci disappeared at later time point in late-passage cells, 16 hours post-irradiation (Fig. 2C, lanes 15 and 16).

To confirm that the altered kinetics of DNA repair observed in late-passage HFF was telomere dependent, we transduced HFFs at late passage (PD 64) with a retroviral vector that expresses hTERT. Transduced cells exhibited telomerase activity and elongated telomeres in early passages after the cells were harvested after the transduction (Supplementary Fig. S3). Ectopic telomerase expression in late-passage cells completely restored the kinetics of γH2AX foci formation and resolution to the levels observed with early-passage cells (Fig. 3A), showing that telomere dysfunction, rather than other effects of prolonged cell culture such as oxidative damage, was responsible for the altered H2AX phosphorylation. Similar results were obtained using Western blot analysis to assess γH2AX protein levels (Fig. 3B).

To further characterize the delayed DNA repair kinetics observed in late-passage HFFs, we assessed other members of the DNA damage response pathway. ATM is rapidly activated in response to DSBs and phosphorylates key proteins leading to cell-cycle arrest, senescence, and apoptosis (17). ATM substrates include p53, NBS1, H2AX, 53BP1, BRCA1, SMC1, and Chk2 (18, 19). Because late-passage cells showed delayed H2AX phosphorylation and γH2AX foci formation, we surmised that ATM activation may also be delayed. To test this hypothesis, the kinetics of ATM activation (measured by ATM-S1981P) after irradiation was evaluated in early- and late-passage HFFs that were contact inhibited for several weeks. Similar to the observation with H2AX, late-passage HFFs had a higher baseline level of ATM activation than early-passage cells. However, in contrast to the situation with H2AX phosphorylation, there was no striking difference in the level of ATM activation between early- and late-passage HFFs in response to IR (Fig. 4A). Moreover, the kinetics of ATM-S1981P nuclear focus formation was similar in early- and late-passage cells (Fig. 4B). We also evaluated the phosphorylation of p53, SMC1, and NBS1 in early- and late-passage HFFs. The accumulation of p53 protein and its phosphorylation on S15 occurred with the same kinetics after irradiation in early- and late-passage cells (Fig. 4C). However, the phosphorylation of SMC1 on S957 and NBS1 on S343 was delayed in late-passage cells (Fig. 4C and D). Together these results suggest that late-passage cells exhibit a delay in phosphorylation of ATM target proteins bound to chromatin such as H2AX, SMC1, and NBS1. Proteins that are not chromatin bound, such as p53 and ATM itself, have similar levels of activation in response to IR in late- and early-passage cells.

The delay in phosphorylation of H2AX, SMC1, and NBS1 might be induced by chromatin structure change, triggered by short telomeres, resulting in limited access of activated ATM to its chromatin-bound downstream targets. To test this hypothesis, we used the micrococcal nuclease assay to analyze chromatin structure in late- and early-passage cells. Consistent with the resistance of compacted chromatin to limited nuclease digestion (20), at an early time point (1 minute), chromatin from late-passage presenescent cells was more resistant to micrococcal nuclease digestion than early-passage cells, suggesting that late-passage chromatin is compacted (Fig. 5A and B). The resistance to micrococcal nuclease was completely abolished in late-passage cells expressing hTERT, linking this resistance to telomere dysfunction.

To determine whether late-passage compacted chromatin is related to heterochromatin, we examined H3K9 acetylation and methylation levels. Consistent with the features seen with compact chromatin (21), our results indicated reduced levels of H3 acetyl K9 and increased levels of H3 dimethyl K9 in late-passage cells (Fig. 6, lanes 1 and 3). Ectopic expression of hTERT in late-passage cells reversed H3K9 acetylation and methylation profiles, suggesting that these histone modifications are caused by short telomeres (Fig. 5, lane 4). Moreover, this reversal required an active hTERT on telomeres, as the ectopic expression of hTERT-DN or hTERT-HA (a form of hTERT that is catalytically active but unable to elongate telomeres and prevent senescence; ref. 22) did not reverse these profiles (Fig. 6, lanes 5 and 6).

To further assess whether chromatin state was responsible for the diminished DNA damage response in late-passage cells, we treated late-passage HFFs with the HDAC inhibitor vorinostat, which promotes histone acetylation and an open chromatin state. Treatment with the HDAC inhibitor restored the intensity and rapidity of H2AX phosphorylation in late-passage cells (Fig. 6B). Together, these results suggest a model in which short telomeres induce chromatin structure change consistent with heterochromatin, causing chromatin compaction and limited access of ATM kinase to its chromatin targets, thereby potentially contributing to enhanced radiosensitivity in late-passage cells.

In the present study, we evaluated the mechanism of enhanced radiosensitivity in cells with short telomeres. As expected, primary and transformed cells with short telomeres had markedly enhanced sensitivity to IR compared with cells with longer telomeres. These results mirror the results observed in Terc knockout mice, which showed that progressive telomere shortening was associated with enhanced cellular and organismal sensitivity to IR (7, 11). Similar results have been observed in human cells (23–26). Although some studies have suggested that telomerase expression per se influences radiosensitivity (8, 27), our results and the results of other groups suggest that telomere length is the primary determinant of radiosensitivity (7, 11, 25, 28, 29, 30). In these studies, ectopic telomerase expression or telomerase inhibition had no effect on sensitivity to IR until a measurable change in telomere length was observed.

The biological mechanism underlying the relationship between telomere integrity and radiation sensitivity remains to be determined. A simple model to explain this correlation is that cells with short telomeres possess a baseline level of DNA damage from uncapped telomeres even before they are subjected to IR. In this model, the preexisting damage would be additive with the extra damage induced by IR, thereby enhancing radiation sensitivity. Arguing against this model is our observation that the baseline number of γH2AX foci in late-passage fibroblasts is equivalent to an IR dose of only 0.1 Gy (data not shown). Adding 0.1 Gy to the levels of IR that were used in the clonogenic survival assays would be insufficient to explain the enhanced sensitivity to IR that we observed, though the protracted 0.1-Gy equivalent exposure in late-passage cells may be biologically different from the short-lived exposure associated with a one-time dose of IR. Although the baseline number of γH2AX foci in late-passage cells was elevated compared with early-passage HFFs, the late-passage cells had diminished H2AX phosphorylation in response to IR at early time points (Fig. 2). If the enhanced radiation sensitivity in late-passage cells was merely the result of an additive effect of radiation damage to telomere damage, we would have expected the peak levels of H2AX phosphorylation to be higher in the late-passage cells than the early-passage cells. A second model that has been proposed is that telomeres serve as repositories of proteins that participate in the DNA damage response, which can readily translocate to DNA breaks (11). According to this premise, once telomeres become short, repair proteins can no longer bind telomeres and are no longer immediately available to sites of DNA damage. However, the DSB repair proteins that associate with telomeres, such as the MRE11/RAD50/NBS1 (MRN) complex, are uniformly distributed in the nuclei of unirradiated cells (31). Because the telomere-bound component comprises just a small fraction of the abundant pool of repair proteins, it seems unlikely that a decrease in telomere length would have a major effect on the availability of repair proteins to be recruited to DNA breaks.

The present data support a third model for increased radiation sensitivity in cells with short telomeres. One of the earliest events in the DNA damage response is ATM phosphorylation/activation and swift phosphorylation of H2AX (γH2AX) on serine 139, primarily by activated ATM and DNA-PK (15, 18). Upon DSB, phosphorylated H2AX (γH2AX) rapidly forms DNA repair foci that include the MRN complex, 53BP1, and BRCA1 (32). Our results showed that late-passage cells had attenuated peak levels of H2AX phosphorylation and that the kinetics of H2AX phosphorylation and dephosphorylation were delayed. Because γH2AX plays a crucial role in the recruitment of proteins to the repair focus and in its stabilization to recruit late factors, such as cohesions, to tether sister chromatids for homologous recombination repair pathway (33–36), the altered H2AX phosphorylation kinetics may account for the enhanced radiosensitivity in late-passage cells. Recently, it was shown that telomerase-null and telomerase-inhibited mouse embryo fibroblasts (MEF) formed fewer γH2AX foci and had delayed appearance and disappearance of such foci following heat shock and IR exposure compared with telomerase-positive MEFs (6). Although these results focused on telomerase expression and not telomere length, they are concordant with our results. Furthermore, delayed kinetics of DNA DSB processing, including the rates of recruitment of DSB repair proteins to γH2AX foci, has been observed in normal and pathologic aging (9).

Our data suggest a mechanism for the altered kinetics of γH2AX foci formation and disappearance in late-passage cells. Late-passage chromatin was resistant to micrococcal nuclease digestion and had reduced levels of H3K9 acetylation and higher levels H3K9 dimethylation, indicative of heterochromatinization. These results are consistent with the observation that senescent cells accumulate foci of heterochromatin (37), though the late-passage fibroblasts used in our experiments were presenescent (Supplementary Fig. S2). The compacted chromatin could explain why phosphorylation and activation of chromatin-bound DNA damage proteins (H2AX, SMC1, and NBS1) were delayed in late-passage cells, whereas activation of non–chromatin-bound proteins (p53 and ATM) was unchanged. Moreover, we observed that dephosphorylation of γH2AX was delayed in late-passage cells. Dephosphorylation of γH2AX is achieved by protein phosphatase 2A (38), by the HTP-C phosphatase complex in budding yeast (39) and by Wip1 phosphatase (40–42). Like delayed phosphorylation of chromatin-bound proteins, delayed dephosphorylation may be caused by reduced access of phosphatases to late-passage chromatin. Furthermore, we and others have shown that H3K9Ac is a DNA damage–responsive modification [Zhang and Drissi, unpublished data; ref. (43)], suggesting that this histone modification is a mark of heterochromatin and DNA damage.

Telomerase protein has also been described to regulate chromatin state and the DNA damage response (27). Suppression of hTERT in human fibroblasts led to enhanced radiosensitivity and delayed kinetics of γH2AX foci formation, without measurable changes in telomere length. Abrogation of hTERT was associated with a pattern of euchromatin (enhanced sensitivity to micrococcal nuclease–induced DNA digestion, increased H3K9 acetylation, and decreased H3K9 dimethylation). It is interesting that although the end result of telomerase suppression (enhanced radiosensitivity and delayed DNA repair kinetics) was similar to the effect we observed with telomere shortening, telomerase suppression increased euchromatin content whereas telomere shortening increased heterochromatin content. These results indicate that although telomerase repression and telomere shortening are closely intertwined, the two processes do not have identical consequences for the cell.

It remains to be determined how telomere length influences chromatin structure and, in turn, how chromatin structure influences the DNA repair process. It has recently been shown that ATM-mediated phosphorylation of heterochromatin-promoting proteins, such as KAP-1, increases the repair of DNA breaks within heterochromatin (44). Our data showed that ATM activation kinetics in late-passage cells was not affected by short telomeres but that phosphorylation of H2AX, SMC1, and NBS1 was delayed. It is possible that telomere shortening–induced heterochromatin introduces an impediment to ATM-dependent chromatin modifications that facilitate DNA DSB repair.

Finally, our results support several predictions about the use of telomerase inhibitors as therapeutic agents for human cancer. First, telomerase inhibition in itself can induce apoptosis and growth arrest in cancer cells, though this response may be delayed, as observed in our HeLa cells. Second, the concurrent administration of telomerase inhibitors and IR may have a synergistic effect once sufficient telomere shortening has occurred. The concurrent application of DNA-damaging agents and telomerase inhibitors may provide a therapeutic benefit.

No potential conflicts of interest were disclosed.

The research was supported by grants CA87903 and CA111347 from the NIH to J.S. Dome, the American Lebanese Syrian Associated Charities (ALSAC) of St. Jude Children's Research Hospital, and the Children's Cancer Foundation.

The costs of publication of this article were defrayed in part by the payment of page charges. This article must therefore be hereby marked advertisement in accordance with 18 U.S.C. Section 1734 solely to indicate this fact.