Poly(ADP-Ribose) Polymerase Inhibitor Induces Accelerated Senescence in Irradiated Breast Cancer Cells and Tumors

Small molecules targeting cellular responses to DNA damage have long been considered an attractive strategy to improve the effectiveness of genotoxic cancer therapy (1). An early event in the double-strand break (DSB) response is rapid recruitment and activation of poly(ADP-ribose) polymerase 1 (PARP1), resulting in polymerization of poly(ADP ribose) into PARP1 itself, histones, and other proteins at DSBs and in the recruitment of macroH2AX to sites of DNA damage to stimulate chromatin remodeling and DNA repair (2–4). PARP activity is required for normal DNA damage tolerance. Whereas most attention has been paid to their potential in targeting malignancies defective in homologous recombination (5), PARP inhibitors are also promising as sensitizers for genotoxic agents and ionizing radiation (IR; refs. 6, 7).

Coincident with PARP1 recruitment, ATM-dependent phosphorylation of histone H2AX to form γH2AX at DSBs promotes further chromatin modifications and assembly of proteins at IR-induced foci (IRIF), such as MRE11/RAD50/NBS1, MDC1, 53BP1, and BRCA1 (8, 9). Tracking the accumulation and dispersal of IRIF proteins offers complementary reporters for checkpoint signaling and repair.

Herein, by exploiting green fluorescent protein (GFP) fused to the chromatin-binding domain of 53BP1 as a live-cell imaging reporter for DSB repair, we monitored the effects of PARP inhibition on irradiated breast cancer cells both in vitro and in vivo. ABT-888 blocked IRIF resolution and cell proliferation, driving tumor cells toward accelerated senescence and suppressing tumor regrowth compared with IR alone.

GFP fused to the human 53BP1 IRIF binding domain (10) was cloned into the pLVX-Tight-Puro lentiviral vector (Clontech), transduced into the MCF7 Tet-On Advanced cell line (Clontech), and cultured in high-glucose DMEM (Invitrogen) with 10% Tet system-approved fetal bovine serum (Clontech). MCF7 Tet-On Advanced is certified by Clontech as derived from MCF7 (American Type Culture Collection) by viral transduction and was used without further authentication. After induction for 48 hours with 1 μg/mL doxycycline (Sigma), GFP-positive cells were sorted to establish a stable MCF7^Tet-On GFP-IBD cell line.

17β-Estradiol pellets (1.7 mg; Innovative Research of America) were implanted in female athymic nude mice (Harlan) 7 days before s.c. injection of 1 × 10⁷ MCF7^Tet-On GFP-IBD cells in 100 μL of PBS. Once tumors grew to 300 mm³, 2 mg/mL doxycycline with 1% sucrose was added to the drinking water for 72 hours before IR. Mice received 0.5 mg of ABT-888 in water twice daily by oral gavage 48 hours before IR and thereafter as indicated.

Live-cell images were captured on an Olympus DSU spinning disk confocal microscope and a back-thinned EMCCD camera controlled by Slidebook v4.2 software or Zeiss Axiovert 200M and a Hammatsu Orca ER FireWire digital monochrome camera controlled by OpenLab software. For IRIF imaging in tumors, we used a Leica SP5 Tandem Scanner Two-Photon Spectral Confocal System controlled by LAS-AF 2.0 software.

Detailed methods on cell lines, shRNA knockdowns, quantitative PCR gene expression analyses, bromodeoxyuridine (BrdUrd) incorporation, clonogenic assays, PI staining, in vitro PARP activity assays, quantification of foci number and size, immunofluorescence, and SA-βGal staining are reported in Supplementary Data.

γH2AX foci and 53BP1 localization to IRIF can serve as proxies for unrepaired DSBs and the DNA damage response (8). The functional elements of the 53BP1 IRIF binding domain are a dimerizing domain, paired Tudor domains that recognize the stable histone marks H4-diMeK20 and/or H3-diMeK79, and a nuclear localization signal (10, 11). Cells lacking PARP activity display a delay in H2AX phosphorylation and persistence of γH2AX foci (12). 53BP1 binding at IRIF is partly dependent on H2AX phosphorylation and chromatin remodeling, also influenced by PARP activity. Thus, to examine PARP inhibitor effects on IRIF kinetics in living cells, we placed GFP fused to the 53BP1 IRIF binding domain (10) under tetracycline-inducible control (GFP-IBD; Supplementary Fig. S1) in a lentiviral vector. We transduced MCF7 Tet-On Advanced (MCF7^Tet-On, Clontech), which is a cell line derived from MCF7, a p53-positive, caspase-3–negative, and apoptosis-resistant human breast cancer–derived cell line that stably expresses the Tet-On Advanced transactivator.

Following induction with doxycycline, unirradiated MCF7^Tet-On cells expressing inducible GFP-IBD (MCF7^Tet-On GFP-IBD) displayed pan-nuclear fluorescence, with only rare nuclear foci (mean, 0.4 ± 0.7 per cell). Consistent with previous reports, the GFP-IBD reporter relocalized within minutes after IR to form nuclear foci that colocalized with γH2AX, endogenous 53BP1, and MDC1 proteins (Supplementary Fig. S2). The GFP-IBD foci then slowly resolved over the next 24 hours. The ATM kinase inhibitors KU-55933 and CGK733 decreased GFP-IBD foci formation (data not shown). In turn, shRNA knockdown of proteins required for 53BP1 relocalization to IRIF, including ATM, MDC1, and RNF8, blocked the formation of GFP-IBD foci after IR (Fig. 1A). Significantly, knockdown of endogenous 53BP1 increased the number of GFP-IBD foci in unirradiated cells and slowed their resolution after IR, indicating that 53BP1 remains active in MCF7^Tet-On GFP-IBD cells.

We examined IRIF formation and resolution in relation to IR dose and time in MCF7^Tet-On GFP-IBD cells in vitro (Fig. 1B and C). Most GFP-IBD foci resolved by 24 hours at doses up to 8 Gy. For 2 Gy, the mean of 47 ± 13 IRIF at 3 hours decreased to 3.2 ± 1.7 at 24 hours, whereas for 8 Gy, the mean of 55 ± 15 IRIF at 3 hours decreased to 12 ± 6 at 24 hours. After 12 Gy, the mean of 53 ± 14 IRIF at 3 hours decreased only to 37 ± 18 at 24 hours. The increased IRIF persistence with higher IR dose suggests saturation of repair capacity or other damage responses. Indeed, doses above 6 Gy had greater effects on clonogenicity, a likely consequence of persistent DNA damage (Fig. 1D).

Treating MCF7^Tet-On GFP-IBD cells with IR in the presence of the PARP1 inhibitor ABT-888 (veliparib; 2-[(R)-2-methylpyrrolidin-2-yl]-1H-benzimidazole-4-carboxamide; ref. 13) prevented PARP activation (Supplementary Fig. S3) and markedly increased residual IRIF at 3 and 24 hours (Fig. 2A and B). Neither the GFP-IBD reporter nor ABT-888 seemed to alter γH2AX localization or the recruitment of MDC1 and endogenous 53BP1 to IRIF (Fig. 2B). Time-lapse live-cell imaging of GFP-IBD revealed that in cells treated with 6 Gy, IRIF appeared within 15 minutes and began to decrease noticeably by 60 minutes (Fig. 2C; Supplementary Fig. S4). However, after 6 Gy + ABT-888, IRIF continued to appear up to 60 minutes, perhaps via conversion of single-strand breaks to DSBs (7), but remained largely unchanged thereafter. Previous data suggest that the growth of IRIF might maintain DNA damage signaling from unrepaired DSBs (14). Whereas the mean size of IRIF formed in cells treated with ABT-888 was clearly smaller (Supplementary Fig. S5), the total volume of IRIF per cell at 24 hours with IR + ABT-888 (187 ± 11/μm³) was significantly greater than that with IR alone (87 ± 10/μm³; P = 0.005, t test).

ABT-888 alone slightly decreased colony formation at 10 μmol/L (100 ± 1% for control versus 88 ± 0.6% for ABT-888), but significantly reduced colony formation following 2 Gy (29.7 ± 1.5% for IR alone versus 11.3 ± 0.6% for IR + ABT-888; P < 0.001, t test), with similar fold reductions at each IR dose up to 6 Gy. We next examined the potential mechanisms of growth suppression after IR + ABT-888. PARP inhibition did not dramatically affect MCF7^Tet-On GFP-IBD cell death after IR. Even 7 days later, few cells exhibited propidium iodide permeability, suggesting that PARP inhibition might induce MCF7^Tet-On GFP-IBD cell cycle arrest rather than apoptosis or necrosis. This is consistent with the previous observation that inhibition of ADP-ribosylation could block apoptosis, and a transient burst of PARP activity was required for apoptosis (15). Indeed, whereas ABT-888 alone did not appreciably decrease proliferating cells at 24 hours (58 ± 1% BrdUrd⁺ for control versus 56 ± 1% for ABT-888; not shown), the antiproliferative effects of 3 Gy (41 ± 1%) were enhanced by ABT-888 (27 ± 1%). In turn, MCF7^Tet-On GFP-IBD cells treated with 3 or 6 Gy alone showed a higher recovery of proliferative capacity compared with IR + ABT-888 (Fig. 2D).

Unrepaired DNA damage can promote accelerated or premature senescence, even in cells with otherwise unlimited proliferative capacity (16–18). Accelerated senescence following IR has been observed in MCF7^Tet-On GFP-IBD cells both in vitro and in vivo (19, 20). At 4 days after IR + ABT-888, cells displaying persistent GFP-IBD foci began to exhibit a morphology characteristic of senescence. At 7 days, surviving cells remained adherent, became enlarged with a flat morphology, and displayed multiple nuclear GFP-IBD foci (Fig. 3A). We investigated other hallmarks of accelerated senescence (16, 17), including SA-βGal staining (Fig. 3B) and increased expression of the cyclin-dependent kinase inhibitor p21^Cip1/WAF1 (Fig. 3C). After 6 Gy + ABT-888, 76 ± 4% of surviving cells showed SA-βGal staining compared with 1.2 ± 1.0% for ABT-888 and 2.5 ± 2.0% for 6 Gy (P < 0.001), and p21^Cip1/WAF1 gene expression was significantly upregulated following IR + ABT-888 compared with IR alone (P < 0.02, t test). Immunocytochemistry suggested that the accumulation of p21^Cip1/WAF1 was greatest in cells with persistent IRIF (Fig. 3D). Accelerated senescence following IR and ABT-888 treatment was not limited to cells with wild-type p53, as we observed the same phenotype in MCF7^Tet-On GFP-IBD cells treated with p53 inhibitor pifithrin (data not shown), as well as in breast and other cancer cell lines with mutations in p53 (Supplementary Fig. S6).

To visualize IRIF in vivo, MCF7^Tet-On GFP-IBD cells were injected into nude mice to form xenograft tumors. Imaging of GFP-IBD by two-photon microscopy revealed that the kinetics of IRIF formation and resolution in tumors were comparable to that observed in MCF7^Tet-On GFP-IBD cells in vitro (Fig. 4A). When mice were treated with ABT-888 twice daily for 2 days before IR and then twice daily thereafter, we observed no increase in IRIF number at early time points but the number of cells with residual IRIF increased at 24 hours. Twenty-two percent of tumor cells treated with 3 Gy exhibited ≥4 IRIF/cell whereas 42% of tumor cells with 3 Gy + ABT-888 had ≥4 IRIF/cell (P < 0.001, t test; Fig. 4B). To evaluate DNA damage–induced senescence in vivo, we examined SA-βGal staining in frozen tumor sections at 7 days (Fig. 4C). ABT-888 alone slightly enhanced SA-βGal staining above background but markedly increased staining when combined with IR. To compare in vivo growth delay with that observed in vitro, we performed a tumor regrowth experiment. Mice bearing MCF7^Tet-On GFP-IBD tumors were treated with ABT-888 for 2 days before a single 6-Gy dose and then only for 2 days after IR. This short course of PARP inhibition significantly slowed MCF7^Tet-On GFP-IBD tumor regrowth compared with 6 Gy alone [day 9, P = 0.021; day 12, P = 0.013; day 14, P = 0.001 (t test); Fig. 4D].

Our data confirm previously reported enhancement of IR effects by PARP inhibition (6, 11) and implicate IRIF persistence as a potential mechanism of accelerated tumor cell senescence. Persistent cell cycle arrest and accelerated senescence are ascribed to accumulation of unrepaired DNA damage and chromatin perturbation, among other inducers (17, 18). We speculate that the efficacy of PARP inhibitors toward homologous recombination–deficient BRCA1-, BRCA2-, or PTEN- negative cancer cells (21) may similarly reflect a cellular response to accumulation of unrepaired endogenous DNA damage. Indeed, preliminary analysis of the PTEN mutant cell line PC-3 suggests that ABT-888 accelerates senescence, particularly in combination with radiation. Whereas it remains a dogma that IR and genotoxic agents mediate their lethal effects via enhanced apoptosis, necrosis, or mitotic catastrophe, senescence is an alternative terminal phenotype that may be highly relevant as a determinant of outcomes for cancer treatment (16, 22, 23). Alone or in combination with other epigenetic drugs such as histone deacetylase inhibitors that promote IRIF persistence and accelerated senescence (24), PARP inhibitors may have a significant effect by inducing senescence as a novel mechanism for sensitization to radiation and chemotherapy.

No potential conflicts of interest were disclosed.

We thank T. Halazonetis for generously providing reagents, and M. Pejovic, R. Torres, and S. Bond for assistance.

Grant Support: The University of Chicago Ludwig Center for Metastasis Research, the Foglia Family Foundation, The University of Chicago Comprehensive Cancer Center, and NIH grants CA138365 and GM60443.

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.