Abstract
Targeting PARP1 for synthetic lethality is a new strategy for breast cancers harboring germline mutations in BRCA. However, these mutations are rare, and reactivation of BRCA-mediated pathways may result in eventual resistance to PARP1 inhibitor therapy. Alternative synthetic lethality approaches targeting more common sporadic breast cancers and preinvasive ductal carcinoma in situ (DCIS) are desirable. Here we show that downregulation of XRCC1, which interacts with PARP1 and coordinates base excision repair, is an early event in human breast cancer pathogenesis. XRCC1-deficient DCIS were aggressive and associated with increased risk of local recurrence. Human invasive breast cancers deficient in XRCC1 and expressing high PARP1 levels also manifested aggressive features and poor outcome. The PARP1 inhibitor olaparib was synthetically lethal in XRCC1-deficient DCIS and invasive breast cancer cells. We conclude that targeting PARP1 is an attractive strategy for synthetic lethality and chemoprevention in XRCC1-deficient breast cancers, including preinvasive DCIS.
These findings show that loss of XRCC1, which is associated with more malignant DCIS, can be exploited by PARP inhibition, suggesting its application as a promising therapeutic and chemoprevention strategy in XRCC1-deficient tumor cells.
Introduction
Targeting PARP for synthetic lethality is an exciting new strategy in BRCA germline-mutated breast cancers (1). However, BRCA germline-mutated cancers are rare. In addition, reactivation of BRCA-mediated pathways may result in eventual resistance to PARP inhibitor therapy (1). Therefore, alternative synthetic lethality targets enabling the extension of this approach including in sporadic DNA repair–deficient triple-negative breast cancers (TNBC) and preinvasive ductal carcinoma in situ (DCIS; ref. 2) is highly desirable.
XRCC1 (X-ray repair cross-complementing gene 1) is a critical factor in DNA base excision repair (BER) and single-strand break repair (SSBR) (3, 4). XRCC1 interacts with PARP1 and coordinates BER/SSBR (5, 6). In addition, XRCC1 is also involved in alternative nonhomologous end joining (alt-NHEJ) pathway for double-strand breaks (DSB; ref. 7) and nucleotide excision repair (8). XRCC1 deficiency delays SSB rejoining, thereby leading onto SSBs, which, if unrepaired, eventually to DSBs (3, 4). In addition, XRCC1 deficiency/mutation is associated with hyperactivation of PARP1 and ataxia but no cancer predisposition was reported in these individuals (9, 10). Whether PARP1 targeting will have translational application in XRCC1 deficient breast cancers is currently unknown.
In the current study, we show that XRCC1-deficient human invasive breast cancers with high PARP1 levels have aggressive pathology and worse survival. Preclinically, olaparib is synthetically lethal in XRCC1-deficient cells compared with XRCC1-proficient cell lines and 3D spheroids. In patients with preinvasive DCIS, XRCC1 deficiency is also linked to aggressive phenotype and recurrence. XRCC1 depletion promotes invasion, epithelial–mesenchymal transitions and increased sensitivity to olaparib monotherapy in MCF10DCIS cells. We conclude that PARP1 targeting in XRCC1-deficient–invasive cancers or DCIS is an attractive synthetic lethality and chemoprevention strategy.
Materials and Methods
Compounds, reagents, clonogenic assays, cell proliferation assays, confocal microscopy, functional assays (FACS, cell-cycle progression, apoptosis assays), invasion assay, migration assay, and 3D-spheroid assays are described in detail in Supplementary Materials and Methods.
Cell lines and culture
MDA-MB-231, MDA-MB-157 cells were purchased from ATCC. Cell line authentication was performed by AuthentiFiler PCR Amplification Kit. MDA-MB-231 was cultured in minimum essential amino acids medium supplemented with 10% FBS, 1% penicillin–streptomycin, 1% l-glutamine, and 1% nonessential amino acids. MDA-MB-157 was grown in IMDM medium supplemented with 15% FBS and 1% penicillin/ streptomycin). Previously well-characterized Chinese hamster (CH) ovary cells; CHO9 (wild-type), EM-C11 (XRCC1-mutant), EM-C12 (XRCC1-mutant) were a gift from the late Prof. M.Z. Zdzienicka. Cells were grown in Ham F-10 media (supplemented with 10% FBS and 1% penicillin/streptomycin). MCF10DCIS cells were cultured in DMEM-F12 supplemented with 10% horse serum, 5 mg/mL insulin, 1 mg/mL cholera toxin, and 100 μg/mL EGFR, 5 mg/mL hydrocortisone and 1% penicillin–streptomycin. XRCC1-deficient HeLa SilenciX cells and controls XRCC1-proficient HeLa SilenciX cells were purchased from Tebu-Bio (www.tebu-bio.com). SilenciX cells were grown in DMEM supplemented with 10% FBS, 1% penicillin/streptomycin, and 125 μg/mL hygromycin B. All cell lines were tested for Mycoplasma routinely every 3 months using MycoProbe Mycoplasma Detection Kit (R&D Systems). All cell lines were used between 15 passages window.
Generation of XRCC1 knockouts using CRISPR/Cas-9 system
MDA-MB-231 and MCF-10DCIS were transfected with oligonucleotides carrying gRNA silencing XRCC1 cloned in a Plv-U6g-EPCG plasmid (Sigma). Briefly, cells were seeded at 50%–60% confluency in 6-well plates overnight. Cells were transfected with 2–3 μg of DNA using Lipofectamine 3000 (Invitrogen) in an Opti-MEM medium. Puromycin was used as a selection marker for 14 days. MDA-MB-231 cells were selected in 10 μg/mL and MCF-DCIS cells were selected in 5 μg/mL puromycin.
qRT-PCR analysis of epithelial–mesenchymal transition gene expression
Real-time PCR was performed using RT2 Profiler EMT PCR Array for 86 epithelial–mesenchymal transition (EMT) genes. The data were analyzed as per the manufacturer's recommendations (https://www.qiagen.com/us/shop/genes-and-pathways/data-analysis-center-overview-page/). RPLP0 was used for normalization of the data. All experiments were performed in duplicate.
XRCC1 expression in human invasive breast cancers and preinvasive DCIS
The study was performed in a consecutive series of 1,650 patients with primary invasive breast carcinomas who were diagnosed between 1986 and 1999 and entered into the Nottingham Tenovus Primary Breast Carcinoma series. All patients were treated in a single institution and have been investigated in a wide range of biomarker studies (11–15). Supplementary Table S1 summarizes patient demographics. Supplementary Treatment Data S1 summarizes various adjuvant treatments received by patients in this cohort. We also evaluated an independent series of 281 ERα-negative invasive breast cancers diagnosed and managed at the Nottingham University Hospitals between 1999 and 2007. All patients were primarily treated with surgery, followed by radiotherapy and anthracycline chemotherapy. The characteristics of this cohort are summarized in Supplementary Table S2. Tumors were arrayed in tissue microarrays and IHC profiled for XRCC1 (15), PARP1 (14) and other biological antibodies (Supplementary-Table S3) as described previously (11–13).
A total of 776 patients with noninvasive pure DCIS diagnosed between 1987 and 2012 were identified from the National Health System (NHS) database of the Nottingham University Hospitals. A cohort of 239 DCIS that coexist with invasive breast cancer as well as 50 normal breast tissues were also identified. Patients’ demographics are summarized in supplementary materials.
IHC scoring and statistical analyses for patient cohorts are described in detail in Supplementary Methods.
Written informed consent from patients was obtained where applicable. The study was conducted in accordance with the Declaration of Helsinki and ethical approval was obtained from the Nottingham Research Ethics Committee (C202313).
Results
XRCC1-PARP1 coexpression and aggressive human breast cancers
We have previously shown that loss of XRCC1 expression (16% of tumors) was significantly associated with high grade, absence of hormonal receptors (ER−/PgR−/AR−), presence of basal-like phenotypes, triple-negative phenotypes and poor survival including in patients with TNBCs (15). We have also demonstrated that PARP1 expression is linked to aggressive breast cancers (14). XRCC1 is known to interact with PARP1 during BER (7). In addition, XRCC1 deficiency/mutation can also hyperactivate PARP1 (9). We, therefore, proceeded to investigate XRCC1-PARP1 coexpression in a large clinical breast cancer cohort.
A total of 1011 breast tumors were suitable for XRCC1-PARP1 coexpression analyses (Fig. 1A). A total of 451 of 1,011 (44.6%) were XRCC1+/PARP1+, 61 of 1,011 (6%) were XRCC1−/PARP1+, 396 of 1,011 (39.1%) were XRCC1+/PARP1− and 103 of 1,011 (10.1 %) were XRCC1−/PARP1−. As shown in Supplementary Table S4, tumors that are XRCC1−/PARP+ are likely to be high grade, high mitotic index, high pleomorphism, ER−/PR− and high-risk Nottingham Prognostic Index (NPI > 3.4; all Padj < 0.0001). In the whole cohort (n = 997), XRCC1−/PARP+ tumors had the worst breast cancer–specific survival (P < 0.0001; Supplementary Fig. S1A). In ER+ breast cancers, XRCC1−/PARP+ tumors had the worst survival outcome (P < 0.0001) (Supplementary Fig. S1B) including in patients who had no endocrine therapy (P = 0.015) (Supplementary Fig. S1C) or endocrine therapy (P < 0.0001; Supplementary Fig. S1D). In TNBCs (n = 356), similarly, XRCC1−/PARP+ tumors had the worst survival outcome in the whole cohort (P = 0.001; Fig. 1B) including in patients who had no chemotherapy (P = 0.003; Fig. 1C) or chemotherapy (P = 0.038; Fig. 1D).
A, IHC expression of XRCC1 and PARP1 in breast cancers. B-D, XRCC1 and PARP1 coexpression and Kaplan–Meier curve for breast cancer–specific survival in TNBC (whole cohort; B), TNBCs that received no chemotherapy (C), and TNBCs treated with chemotherapy (D). E, XRCC1 expression in various cell lines. F-I, Clonogenic cell survival in olaparib-treated cells [CHO9 and EMC12 cells (F), HeLa control and HeLa XRCC1–deficient cells (G), 231 control and 157 cells (H), and 231 and 231:XRCC1_KO cells (I)]. J and K, XRCC1 mRNA levels in 231 and 157 cells. L, XRCC1 expression in 157 cells treated with azacytidine (1.5 μmol/L). M, Cell proliferation assay in cells treated with olaparib and azacytidine compared with olaparib alone. All figures are representative of three or more independent experiments.
A, IHC expression of XRCC1 and PARP1 in breast cancers. B-D, XRCC1 and PARP1 coexpression and Kaplan–Meier curve for breast cancer–specific survival in TNBC (whole cohort; B), TNBCs that received no chemotherapy (C), and TNBCs treated with chemotherapy (D). E, XRCC1 expression in various cell lines. F-I, Clonogenic cell survival in olaparib-treated cells [CHO9 and EMC12 cells (F), HeLa control and HeLa XRCC1–deficient cells (G), 231 control and 157 cells (H), and 231 and 231:XRCC1_KO cells (I)]. J and K, XRCC1 mRNA levels in 231 and 157 cells. L, XRCC1 expression in 157 cells treated with azacytidine (1.5 μmol/L). M, Cell proliferation assay in cells treated with olaparib and azacytidine compared with olaparib alone. All figures are representative of three or more independent experiments.
Taken together, the data provide evidence that high PARP1 levels in XRCC1-deficient tumors is not only associated with aggressive breast cancer but also suggests that targeting PARP1 could be a promising personalized treatment strategy. To explore this possibility further we conducted detailed preclinical studies.
Selective toxicity of olaparib in XRCC1-deficient cells
We evaluated a panel of XRCC1-proficient and -deficient Chinese hamster [CHO9 (wild-type), EM-C12 (XRCC1 deficient); Fig. 1E], HeLa (control and XRCC1_KD-silenced X cells; Fig. 1E) and triple-negative breast cancer cells [MDA-MB-231 (XRCC1 proficient) and MDA-MB-157 (XRCC1 deficient); Fig. 1E]. We also generated XRCC1 knockout (KO) MDA-MB-231 cells using a CRISPR/Cas-9 system (Fig. 1E). Olaparib sensitivity was investigated in clonogenic survival assays. XRCC1-deficient cells are highly sensitive to olaparib treatment compared with control XRCC1-proficient cells (Fig. 1F–H and I). Interestingly, we observed a range of sensitivities to olaparib across different cell lines. We quantified relative XRCC1 and PARP1 protein levels in matched cell lines (Supplementary Fig. S2A and S2B). In MDA-MB-231 control XRCC1-proficient cells, the basal level of PARP1 was low. In MDA-MB-231_XRCC1_KO cells, there was a higher basal level of PARP1. In MDA-MB-157 cells that have low expression of XRCC1, PARP1 level was high. In contrast, in HeLa control and XRCC1-depleted cells, we did not observe significant changes in PARP1 levels. Together, the data suggest although XRCC1 depletion may result in increased PARP1 levels; it does not however, predict sensitivity to PARP inhibitor therapy.
We then investigated the mechanism of XRCC1 downregulation in MDA-MB-157 breast cancer cells. As shown in Fig. 1J, the qRT-PCR analysis indicated that XRCC1 mRNA expression was significantly low in MDA-MB-157 cells compared with MDA-MB-231 cells. Upon treatment with azacytidine (a hypomethylating agent), XRCC1 mRNA expression increased in MDA-MB-157 cells and was comparable with XRCC1 mRNA expression in MDA-MB-231 cells (Fig. 1K). Increased XRCC1 mRNA expression was associated with increased XRCC1 protein level as shown by Western blotting (Fig. 1L). Interestingly, azacytidine induced XRCC1 reexpression was associated with the development of resistance to olaparib therapy in MDA-MB-157 cells (Fig. 1M). The data provide supportive evidence that XRCC1 expression can influence olaparib sensitivity in cancer cells. To confirm a synthetic lethality relationship, we evaluated functional consequence of PARP inhibition in XRCC1-deficient and -proficient cells.
Olaparib induces DSBs in XRCC1-deficient cells
XRCC1 deficiency leads to single-strand breaks that promote binding and activation of PARP1. Subsequent PARylation of PARP1 substrate proteins coordinates DNA repair. In addition, PARP1 autoPARylation results in its release from the DNA. PARP1 inhibitors such as olaparib not only inhibit the catalytic activity of PARP1 but also “trap” PARP1 at the sites of DNA damage that leads to persistent SSBs, which, if unrepaired, ultimately result in accumulation of lethal DSBs in cells.
We first conducted immunofluorescence studies to monitor PARP1 and γH2AX (a marker of DSBs) levels following olaparib treatment in XRCC1-deficient and -proficient cells (Fig. 2A). As expected, basal levels of PARP1 are high in XRCC1-deficient breast cancer cells (Fig. 2A). Upon 24 hours of olaparib treatment, PARP1 level is sustained in XRCC1-deficient cells compared with proficient cells (Fig. 2A). There was a significant increase in γH2AX levels at 8 hours, which dramatically increased at 24 hours in olaparib-treated XRCC1-deficient cells compared with proficient cells (Fig. 2A). We then validated by FACS analysis. Cells were exposed to 10 μmol/L of olaparib for 24 hours and compared with untreated control. As shown in Figs. 2B–E, olaparib treatment significantly increased accumulation of γH2AX-positive cells in XRCC1-deficient cells (EM-C12, HeLa XRCC1_KD-silenced X cells, MDA-MB-231: XRCC1_KO cells, and MDA-MB-157 cells) compared with olaparib-treated XRCC1-proficient cells and untreated controls.
A, Immunofluorescence staining for PARP1 and γH2AX in 231control, 231:XRCC1_KO, and 157 cells treated with olaparib (10 μmol/L) is shown here. B-E, Quantification of γH2AX-positive cells by flow cytometry in CHO9 & EMC12 cells (B), HeLa control and HeLa XRCC1–deficient cells (C), 231 control and 157 cells (D), and 231 control and 231:XRCC1_KO cells (E). F-I, Cell-cycle analysis by flow cytometry in XRCC1-proficient and -deficient cells treated with olaparib (10 μmol/L) in CHO9 and EMC12 cells (F), HeLa control and HeLa XRCC1–deficient cells (G), 231 control and 157 cells (H), and 231 control and 231:XRCC1_KO cells (I). All cell lines were plated overnight and treated with olaparib (10 μmol/L) for 24 hours before harvesting for flow cytometry experiments as described in Materials and Methods. All figures are representative of three or more independent experiments.
A, Immunofluorescence staining for PARP1 and γH2AX in 231control, 231:XRCC1_KO, and 157 cells treated with olaparib (10 μmol/L) is shown here. B-E, Quantification of γH2AX-positive cells by flow cytometry in CHO9 & EMC12 cells (B), HeLa control and HeLa XRCC1–deficient cells (C), 231 control and 157 cells (D), and 231 control and 231:XRCC1_KO cells (E). F-I, Cell-cycle analysis by flow cytometry in XRCC1-proficient and -deficient cells treated with olaparib (10 μmol/L) in CHO9 and EMC12 cells (F), HeLa control and HeLa XRCC1–deficient cells (G), 231 control and 157 cells (H), and 231 control and 231:XRCC1_KO cells (I). All cell lines were plated overnight and treated with olaparib (10 μmol/L) for 24 hours before harvesting for flow cytometry experiments as described in Materials and Methods. All figures are representative of three or more independent experiments.
Olaparib and cell-cycle arrest in XRCC1-deficient cells
Accumulation of DSBs, if unrepaired, will result in cell-cycle arrest. We initially investigated cell-cycle progression upon olaparib treatment in an asynchronous population of cells. In EM-C12, olaparib treatment induced G1 phase arrest (Fig. 2F; see Supplementary Table S5 for quantification). In HeLa XRCC1_KD-silenced X cells, and MDA-MB-157 cells, olaparib treatment induced S-phase arrest (Fig. 2G–H; see Supplementary Table S5 for quantification). However, in MDA-MB-231: XRCC1_KO cells, olaparib treatment induced G2–M-phase arrest compared with control cells (Fig. 2I; see Supplementary Table S5 for quantification). We then synchronized cells by serum starvation for 16 hours. As expected, all cell lines arrested at G1 phase (Fig. 3A). The cell lines were then released in serum-containing media with or without olaparib treatment. At 24 hours, similar to asynchronous population, we observed substantial accumulation of cells in S-phase upon olaparib treatment in HeLa XRCC1_KD-silenced X cells (Fig. 3A and B) and MDA-MB-157 cells (Fig. 3A and C). In MDA-MB-231: XRCC1_KO cells, G2–M arrest was evident and substantial compared with olaparib-treated control and untreated cells (Fig. 3A and C).
A, Cell-cycle synchronization in various cell lines. B and C, FACS analysis in synchronized HeLa control and HeLa XRCC1–deficient cells treated with olaparib (10 μmol/L; B), 231 control and 231:XRCC1_KO and 157 cells (C). D, Cell-cycle analysis in synchronized 231 control, 231:XRCC1_KO and 157 treated with Nutlin 3a (10 μmol/L), and/or olaparib (10 μmol/L). E-H, Annexin V analysis by flow cytometry in CHO9 and EMC12 cells (E), HeLa control and HeLa XRCC1–deficient cells (F), 231 control and 157 cells (G), and 231 control and 231:XRCC1_KO cells (H). All cell lines were plated overnight and treated with olaparib (10 μmol/L) for 24 hours before harvesting for flow cytometry experiments as described in Materials and Methods. I, Representative photo micrographic images of 3D spheres of 231 control, 231:XRCC1_KO, and 157 cells untreated or treated with olaparib (10 μmol/L). Measurement of spheres surface area in square pixels for 231 control untreated or treated with olaparib (10 μmol/L). J and K, 231:XRCC1_KO as well as 157 untreated or treated with olaparib (10 μmol/L; K). L, Quantification of viable and dead cells is shown here. All figures are representative of three or more independent experiments.
A, Cell-cycle synchronization in various cell lines. B and C, FACS analysis in synchronized HeLa control and HeLa XRCC1–deficient cells treated with olaparib (10 μmol/L; B), 231 control and 231:XRCC1_KO and 157 cells (C). D, Cell-cycle analysis in synchronized 231 control, 231:XRCC1_KO and 157 treated with Nutlin 3a (10 μmol/L), and/or olaparib (10 μmol/L). E-H, Annexin V analysis by flow cytometry in CHO9 and EMC12 cells (E), HeLa control and HeLa XRCC1–deficient cells (F), 231 control and 157 cells (G), and 231 control and 231:XRCC1_KO cells (H). All cell lines were plated overnight and treated with olaparib (10 μmol/L) for 24 hours before harvesting for flow cytometry experiments as described in Materials and Methods. I, Representative photo micrographic images of 3D spheres of 231 control, 231:XRCC1_KO, and 157 cells untreated or treated with olaparib (10 μmol/L). Measurement of spheres surface area in square pixels for 231 control untreated or treated with olaparib (10 μmol/L). J and K, 231:XRCC1_KO as well as 157 untreated or treated with olaparib (10 μmol/L; K). L, Quantification of viable and dead cells is shown here. All figures are representative of three or more independent experiments.
The data suggest that the cell-cycle response in various cell lines may be influenced by genetic backgrounds including p53 status. We profiled p53 (Supplementary Fig. S2C) for mutant p53, phosphorylated forms of p53 at Ser 46, Ser 20, and Ser 392 and acetylated form of p53 at Lys 382. As shown in Supplementary Fig. S2C, HeLa cells are wild-type for p53. MDA-MB-231 and MDA-MB-157 cells harbor activating p53 mutation. Nutlin-3, a potent inhibitor of MDM2–p53 interaction can activate p53-mediated pathway in cells (16). To determine whether the cell-cycle arrest observed in olaparib-treated XRCC1-deficient cells may be influenced by p53, we monitored cell-cycle progression in cells treated either with Nutlin alone or with Nutlin/olaparib combination. Olaparib alone resulted in S-phase arrest in MDA-MB-157 cells (Fig. 3A) but olaparib/Nutlin combination lead to G2–M arrest in MDA-MB-157 cells (Fig. 3D). Interestingly, in MDA-MB-231_XRCC1 KO cells, olaparib alone resulted in G2–M arrest (Fig. 3C) but olaparib/Nutlin combination led to S-phase arrest in MDA-MB-231_XRCC1 KO cells. (Fig. 3D). In HeLa cells, olaparib treatment resulted in S-phase arrest (Fig. 2G) and with olaparib/Nutlin combination, the cells remain arrested in S-phase (Supplementary Fig. S2D). When p53 was depleted using siRNA in HeLa cells, olaparib treatment resulted in G2–M arrest (Supplementary Fig. S2D). Taken together, the data provide evidence that cell-cycle response to olaparib treatment in XRCC1-deficient cells is complex and p53-mediated pathway may not be significantly involved in the cell-cycle response. We then evaluated regulators of cell-cycle progression by Western blotting in olaparib-treated and untreated cells (Supplementary Fig. S2E). In S-phase–arrested HeLa XRCC1_KD-silenced X cells, we observed an increase in the level of cyclin E1 and a reduction in the level of p21 after olaparib treatment. In addition, olaparib treatment also reduced the level of p21 in MDA-MB-157 cells, leading onto S-phase arrest. In G2–M–arrested MDA-MB-231: XRCC1_KO cells, we observed an increase in the level of p-cyclin B1 upon olaparib treatment. Together, the data illustrate a complex cell-cycle regulator response following olaparib treatment in various XRCC1-deficient cells.
Olaparib and apoptosis in XRCC1-deficient cells
SSBs/DSBs, if unrepaired, will lead to cell-cycle arrest and eventually induce apoptosis in cells. Accordingly, we observed significant induction of apoptosis at 48 hours in olaparib-treated EM-C12, HeLa XRCC1_KD-silenced X cells, MDA-MB-157 and MDA-MB-231: XRCC1_KO cells compared with control and untreated cells (Fig. 3E–H). To recapitulate an in vivo system, we then generated 3D spheroids of MDA-MB-231 control, MDA-MB-231 XRCC1 KO, and MDA-MB-157 cells. Similar to control cells, untreated XRCC1 KO cells retain spheroid-forming capacity. However, upon olaparib treatment in XRCC1 KO cells, there was not only a reduction in spheroid size (Fig. 3I, J, and K) but also an accumulation of apoptotic cells (Fig. 3L).
Olaparib as a chemoprevention strategy in XRCC1-deficient breast DCIS
DCIS, a preinvasive breast cancer, continues to increase in incidence (17). The main aim of treatment is to prevent DCIS from progressing to invasive cancer. Accordingly, surgery (mastectomy or wide local excision), with or without adjuvant radiotherapy are the main treatment modalities. However, personalization of DCIS therapy is an area of unmet need. For example, it is likely that a subset of low-grade DCIS may never progress to invasive cancer. In a subset of high-grade disease, despite surgery and adjuvant radiotherapy may still recur. Therefore, development of biomarkers of aggressive phenotype is highly desirable. Emerging data suggest that aggressive DCIS may result from the accumulation of somatic mutations. We hypothesized that impaired DNA repair capacity due to XRCC1 deficiency may be a key contributor to the development of high-grade DCIS. We evaluated XRCC1 expression (Fig. 4A) in a cohort of 776 patients with DCIS diagnosed between 1987 and 2012 at Nottingham University Hospitals (demographics summarized in Supplementary Table S6). Whereas normal breast tissues have high levels of XRCC1 expression, the XRCC1 level substantially reduced in DCIS and invasive breast cancers (Fig. 4B). The data provide clinical evidence that loss of XRCC1 expression (observed in 54.3% of DCIS) is an early event during breast cancer pathogenesis. In addition, high-grade DCIS (P = 0.05), ER negativity (P = 0.015), and older age (P = 0.048) were also more common in patients with XRCC1-deficient DCIS (Supplementary Table S7). Importantly, the local recurrence-free interval was significantly reduced in patients with XRCC1-deficient DCIS compared with XRCC1-proficient DCIS (P = 0.001; Fig. 4C). The clinical data, therefore, provide evidence that low XRCC1 expression is a biomarker of aggressive DCIS. To evaluate the potential of olaparib as a chemoprevention strategy in XRCC1-deficient DCIS we proceeded to preclinical investigations.
A, XRCC1 expression in DCIS. IBC, invasive breast cancer. B, Box plot of XRCC1 expression in normal, DCIS, and invasive tumors. C, Kaplan–Meier curve showing association between XRCC1 and local recurrence. D, XRCC1 expression in 231, MCF-7, MCF10DCIS, and MCF10A cell lines. E, Photo micrographic images of cell invasion in MCF10A, MCF10DCIS, and 231 cells as well as quantification data. F, XRCC1 expression in MCF10DCIS_KO cells or controls. G, PARP1 expression in MCF10DCIS_KO cells or controls. H, Photomicrographic images of cell migration in MCF10DCIS control and MCF10DCIS_XRCC1_KO as well as quantification data. I, Cell morphology of MCF10DCIS_XRCC1_KO & MCF10DCIS control cells as well as EMT gene expression in MCF10DCIS control and XRCC1_KO cells using RT2-PCR profiler is shown here. J, Olaparib sensitivity in MCF10DCIS control and MCF10DCIS_XRCC1_KO cells. K, Quantification of γH2AX-positive cells in MCF10DCIS control and XRCC1_KO cells treated with olaparib (10 μmol/L) for 24 hours. L, Cell-cycle analysis in MCF10DCIS control and MCF10DCIS_XRCC1_KO cells. M, Annexin V analysis by flow cytometry in MCF10DCIS control and MCF10DCIS_XRCC1_KO cells. All cells were plated overnight and treated with olaparib (10 μmol/L) for 24 hours before harvesting for flow cytometry experiments as described in Materials and Methods. N, Photomicrographic images of olaparib (10 μmol/L)-treated MCF10 DCIS 3D spheres (see Materials and Methods for details). All figures are representative of three or more independent experiments.
A, XRCC1 expression in DCIS. IBC, invasive breast cancer. B, Box plot of XRCC1 expression in normal, DCIS, and invasive tumors. C, Kaplan–Meier curve showing association between XRCC1 and local recurrence. D, XRCC1 expression in 231, MCF-7, MCF10DCIS, and MCF10A cell lines. E, Photo micrographic images of cell invasion in MCF10A, MCF10DCIS, and 231 cells as well as quantification data. F, XRCC1 expression in MCF10DCIS_KO cells or controls. G, PARP1 expression in MCF10DCIS_KO cells or controls. H, Photomicrographic images of cell migration in MCF10DCIS control and MCF10DCIS_XRCC1_KO as well as quantification data. I, Cell morphology of MCF10DCIS_XRCC1_KO & MCF10DCIS control cells as well as EMT gene expression in MCF10DCIS control and XRCC1_KO cells using RT2-PCR profiler is shown here. J, Olaparib sensitivity in MCF10DCIS control and MCF10DCIS_XRCC1_KO cells. K, Quantification of γH2AX-positive cells in MCF10DCIS control and XRCC1_KO cells treated with olaparib (10 μmol/L) for 24 hours. L, Cell-cycle analysis in MCF10DCIS control and MCF10DCIS_XRCC1_KO cells. M, Annexin V analysis by flow cytometry in MCF10DCIS control and MCF10DCIS_XRCC1_KO cells. All cells were plated overnight and treated with olaparib (10 μmol/L) for 24 hours before harvesting for flow cytometry experiments as described in Materials and Methods. N, Photomicrographic images of olaparib (10 μmol/L)-treated MCF10 DCIS 3D spheres (see Materials and Methods for details). All figures are representative of three or more independent experiments.
MCF10DCIS breast cancer cell line was previously derived from a xenograft originating from premalignant MCF10AT cells injected into SCID mice (18). Injection of the MCF10DCIS cells into SCID mice results in a predominantly comedo DCIS phenotype (18, 19). We show robust levels of XRCC1 expression in MCF10DCIS cells (Fig. 4D). In contrast to MDA-MB-231, MCF10DCIS cells are as expected noninvasive, a phenotype similar to MCF10A noncancerous epithelial cells (Fig. 4E). We then proceeded to generate MCF10DCIS_XRCC1_KO using CRISPR/Cas-9 system (Fig. 4F). XRCC1 _KO was associated with increased PARP1 levels compared with control cells (Fig. 4G). In contrast to control cells, MCF10DCIS_XRCC1_KO cells not only acquired an invasive phenotype (Fig. 4H and I), but was also associated with upregulation of several genes involved in EMT (Fig. 4I; Supplementary Table S8). In addition, spindle-shaped cells with elongated cellular processes and diminished cell-to-cell contacts were more evident in MCF10DCIS_XRCC1_KO cells (Fig. 4I). Importantly, MCF10DCIS _XRCC1 _KO cells were very sensitive to olaparib treatment (Fig. 4J) compared with control cells. Increased olaparib sensitivity was associated with accumulation of γH2AX-positive cells (Fig. 4K), G2–M cell-cycle arrest (Fig. 4L) and apoptosis (Fig. 4M). To recapitulate an in vivo system, we then generated 3D spheroids of MCF10DCIS _XRCC1 _KO cells and MCF10DCIS control cells. Similar to control cells, untreated MCF10DCIS _XRCC1_KO cells retain spheroid-forming capacity (Fig. 4N). However, olaparib treatment strikingly reduced viability and spheroid-forming ability (Fig. 4N). Taken together, the data provide evidence that XRCC1-deficient human DCIS are aggressive and olaparib is synthetically lethal in XRCC1-deficient MCF10DCIS cells.
Evaluation of niraparib and talazoparib sensitivity in XRCC1-deficient cells
The data presented thus far suggest that the hypersensitivity observed in XRCC1-deficient cells may be influenced by the effectiveness of the olaparib to "trap" PARP proteins. However, to validate this hypothesis we evaluated other clinically relevant PARP inhibitors including niraparib and talazoparib that have increased ability to “trap” PARP. Talazoparib is about 100 times more potent than niraparib for PARP trapping. Niraparib, in turn, traps PARP more potently than olaparib (1). As expected, XRCC1-deficient cells are sensitive to niraparib (Supplementary Fig. S2F) and to talazoparib (Fig. 5A) compared with control cells. LD50 analysis (Supplementary Table S9) revealed that at doses tested, niraparib was five times more potent than olaparib. Talazoparib was 20 times more potent than olaparib (Supplementary Table S9). Given the impressive hypersensitivity to talazoparib, we then conducted detailed functional studies. Confocal studies revealed that nuclear PARP accumulation was evident within 4 hours of talazoparib treatment, which substantially increased at 16 hours and persisted at 24 hours in XRCC1-deficient cells (Fig. 5B and C). γH2AX accumulation was also evident at 8, 16, and 24 hours. We confirmed DSB accumulation using γH2AX FACS (Fig. 5D). Furthermore, increased sensitivity to talazoparib was associated with G2–M cell-cycle arrest (Fig. 5E) and increased apoptosis (Fig. 5F). Importantly, in 3D spheroids experiments, we observed that XRCC1-deficient 3D spheroids were sensitivity to talazoparib, as evidenced by the substantial reduction in spheroid size as well as increased apoptotic cells (Fig. 5G–I). Taken together, the data provide additional validation that PARP “trapping” is likely to contribute to the observed PARP inhibitor sensitivity in XRCC1-deficient cells.
A, Clonogenic cell survival in talazoparib-treated cells. B, Immunofluorescence staining for PARP1 and γH2AX in 231control, 231:XRCC1_KO cells untreated or treated with talazoparib (5 μmol/L). C, Quantification of PARP and γH2AX nuclear fluorescence. D, Quantification of γH2AX-positive cells by flow cytometry in talazoparib (5 μmol/L)-treated cells. E, Cell-cycle analysis by flow cytometry is shown here. F, Quantification of apoptotic cells. All cell lines were plated overnight and treated with talazoparib (5 μmol/L) for 24 hours before harvesting for flow cytometry experiments as described in Materials and Methods. G, Photomicrographic images of talazoparib (5 μmol/L)-treated 3D spheroids. H, Quantification of spheres surface area by ImageJ software. I, Quantification of spheroids cell viability by flow cytometry.
A, Clonogenic cell survival in talazoparib-treated cells. B, Immunofluorescence staining for PARP1 and γH2AX in 231control, 231:XRCC1_KO cells untreated or treated with talazoparib (5 μmol/L). C, Quantification of PARP and γH2AX nuclear fluorescence. D, Quantification of γH2AX-positive cells by flow cytometry in talazoparib (5 μmol/L)-treated cells. E, Cell-cycle analysis by flow cytometry is shown here. F, Quantification of apoptotic cells. All cell lines were plated overnight and treated with talazoparib (5 μmol/L) for 24 hours before harvesting for flow cytometry experiments as described in Materials and Methods. G, Photomicrographic images of talazoparib (5 μmol/L)-treated 3D spheroids. H, Quantification of spheres surface area by ImageJ software. I, Quantification of spheroids cell viability by flow cytometry.
Discussion
XRCC1 is a key scaffolding protein intimately involved in BER, SSBR, and alt-NHEJ (3, 4, 7). XRCC1 loss promotes genomic instability (3, 4). XRCC1 interacts with PARP1 during DNA repair (5, 6). A previous high-throughput siRNA screen identified XRCC1 as a synthetic lethality partner for PARP inhibition (20). Preclinically, Xrcc1−/− mouse embryonic fibroblasts were shown to be hypersensitive to PARP inhibitor treatment compared with Xrcc1 +/+ mouse embryonic fibroblasts (21) suggesting that this approach could have clinical relevance in human tumors including breast cancers.
We have previously demonstrated that XRCC1 deficiency is linked to aggressive breast tumors including in TNBCs (15). In this study, we not only show that high PARP1 levels in XRCC1-deficient tumors is associated with aggressive breast cancer but also provide compelling evidence that PARP1 targeting is suitable for synthetic lethality application. A model for XRCC1-directed synthetic lethality has been proposed previously (21). PARP1 binds to DNA repair intermediates, such as single-strand breaks, and gets activated, which, in turn, leads to the synthesis of PAR polymers. PARP1 auto-PARylation recruits other BER factors (including XRCC1) at the sites of DNA damage, resulting in efficient DNA repair. Inhibition of PARP1 catalytic activity (by inhibitor) prevents auto-PARylation, impairs BER recruitment, and stabilizes binding of PARP1 to DNA intermediate. DNA-bound immobilized PARP-1 disrupts replication fork progression, leads to DSB accumulation and DSB-mediated apoptosis. In XRCC1-deficient cells with increased SSB accumulation, PARP inhibition–mediated accumulation of DSB is more pronounced compared with XRCC1-proficient cells, leading to synthetic lethality (21). Accordingly, in XRCC1-deficient breast cancer cells treated with olaparib, we observed nuclear PARP accumulation, increased γH2AX foci in the nucleus, cell-cycle arrest and induction of apoptosis. Interestingly, we also observed cytoplasmic PARP and cytoplasmic γH2AX in olaparib-treated XRCC1-deficient cells. Although majority of cellular PARP activity is localized to the nucleus, the cytoplasm has been shown previously to have PARP activity (22) including in the mitochondria (23). Besides DNA repair, PARP has recognized roles in mitochondrial homeostasis, oxidative stress, and cell death (23). Similarly, cytoplasmic γH2AX has also been reported (24). Jung and colleagues have shown that Tropomyosin-related kinase A–mediated cytoplasmic γH2AX via JNK signaling may have a role in apoptosis (24). We therefore speculate that the observed cytoplasmic PARP and γH2AX may reflect their roles in apoptosis in olaparib-treated XRCC1-deficient cells. Taken together, the data provide the first translational evidence that PARP1 targeting (e.g., olaparib) may have a wider clinical application in XRCC1-deficient sporadic breast cancers. Interestingly, the cell-cycle pattern observed in MDA-MB-231 control cells treated with olaparib was unexpected. A previous study (25) showed that PARP-1 promoted cell proliferation by inhibiting Sp1 signaling pathway. PARP inhibitors significantly inhibited proliferation of hepatoma cells and induced G0–G1 cell-cycle arrest in hepatoma cells. Inhibition of PARP-1 enhanced the expression of Sp1-mediated checkpoint proteins including p21, resulting in G0–G1 cell-cycle arrest in that study. We observed a similar phenotype in MDA-MB-231 control cells where olaparib treatment induced G1 cell-cycle arrest (Fig. 3C) and induction of p21 expression (Supplementary Fig. S2E).
XRCC1 deficiency promotes aggressive invasive cancerous phenotypes. To test whether XRCC1 also influences clinical outcomes in preinvasive DCIS, we investigated XRCC1 expression in a large clinical cohort of human breast DCIS. In contrast to normal breast tissue that had a high XRCC1 expression, there was a dramatic reduction in XRCC1 levels in DCIS particularly those with high grade or ER-negative. Although the mechanism of XRCC1 downregulation in DCIS is currently unknown, the data suggest that loss of DNA repair could be an early event in human breast cancer pathogenesis. More importantly, we also observed that XRCC1 deficiency was also linked to increased risk of local recurrence. In a preclinical MCF10DCIS model, we show that XRCC1 KO dramatically increased invasion associated with upregulation of markers of EMT. Whether XRCC1 is directly or indirectly involved in upregulation of several genes involved in EMT is unknown but the data shown here would concur with previous studies in melanoma (26) and clear cell renal cancer (27) cells where XRCC1 depletion was also shown to promote invasion. A novel observation in this study is that MCF10DCIS_ XRCC1_ KO cells are extremely sensitive to Olaparib therapy. Interestingly, olaparib maintenance almost completely abolished the 3D spheroid–forming ability of MCF10DCIS_XRCC1_KO cells compared with controls. The data provide the first promising evidence that olaparib may have a role in chemoprevention of XRCC1-deficient DCIS.
Impaired DNA repair drives mutagenicity, which can increase neoantigen load and immunogenicity. Whether PARP targeting in combination with immune checkpoint inhibitor can improve therapeutic efficacy is currently an area of intense clinical investigation. However, biomarkers that could predict such an approach are currently unknown. We recently investigated XRCC1 and T-cell infiltration in invasive breast cancers (13). Tumors that expressed low XRCC1 were associated with high CD8+ tumor-infiltrating lymphocyte (TIL) counts, aggressive phenotype and reduced survival. Importantly, PD-1+ or PD-L1+ breast cancers with low XRCC1 were linked to aggressive cancers and reduced survival including in ER− breast cancers (13). As immune microenvironment (including PD-L1+ TILs infiltration) in DCIS can influence aggressive phenotypes (28, 29), we propose that PARP1 targeting may be a promising personalized approach either alone or in combination with immune checkpoint inhibitors in XRCC1-deficient invasive cancers and in preinvasive DCIS.
Disclosure of Potential Conflicts of Interest
No potential conflicts of interest were disclosed.
Authors' Contributions
Conception and design: R. Ali, V. Band, E.A. Rakha, S. Madhusudan
Development of methodology: R. Ali, C. Seedhouse, E.A. Rakha, S. Madhusudan
Acquisition of data (provided animals, acquired and managed patients, provided facilities, etc.): R. Ali, M.S. Toss, A.R. Green, I.M. Miligy, C. Seedhouse, S. Mirza, E.A. Rakha, S. Madhusudan
Analysis and interpretation of data (e.g., statistical analysis, biostatistics, computational analysis): R. Ali, A. Al-Kawaz, M.S. Toss, A.R. Green, E.A. Rakha, S. Madhusudan
Writing, review, and/or revision of the manuscript: R. Ali, A. Al-Kawaz, M.S. Toss, A.R. Green, I.M. Miligy, K.A. Mesquita, V. Band, E.A. Rakha, S. Madhusudan
Administrative, technical, or material support (i.e., reporting or organizing data, constructing databases): R. Ali, M.S. Toss, I.M. Miligy, K.A. Mesquita, E.A. Rakha
Study supervision: I.M. Miligy, V. Band, E.A. Rakha, S. Madhusudan
Other (construction of the TMA for invasive breast cancer and preinvasive DCIS): I.M. Miligy
Acknowledgments
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.