Estrogen Therapy Induces Receptor-Dependent DNA Damage Enhanced by PARP Inhibition in ER+ Breast Cancer

Abstract

Purpose: Clinical evidence indicates that treatment with estrogens elicits anticancer effects in ∼30% of patients with advanced endocrine-resistant estrogen receptor α (ER)-positive breast cancer. Despite the proven efficacy of estrogen therapy, its mechanism of action is unclear and this treatment remains underused. Mechanistic understanding may offer strategies to enhance therapeutic efficacy.

Experimental design: We performed genome-wide CRISPR/Cas9 screening and transcriptomic profiling in long-term estrogen-deprived ER+ breast cancer cells to identify pathways required for therapeutic response to the estrogen 17β-estradiol (E2). We validated findings in cell lines, patient-derived xenografts (PDX), and patient samples, and developed a novel combination treatment through testing in cell lines and PDX models.

Results: Cells treated with E2 exhibited replication-dependent markers of DNA damage and the DNA damage response prior to apoptosis. Such DNA damage was partially driven by the formation of DNA:RNA hybrids (R-loops). Pharmacologic suppression of the DNA damage response via PARP inhibition with olaparib enhanced E2-induced DNA damage. PARP inhibition synergized with E2 to suppress growth and prevent tumor recurrence in BRCA1/2-mutant and BRCA1/2-wild-type cell line and PDX models.

Conclusions: E2-induced ER activity drives DNA damage and growth inhibition in endocrine-resistant breast cancer cells. Inhibition of the DNA damage response using drugs such as PARP inhibitors can enhance therapeutic response to E2. These findings warrant clinical exploration of the combination of E2 with DNA damage response inhibitors in advanced ER+ breast cancer, and suggest that PARP inhibitors may synergize with therapeutics that exacerbate transcriptional stress.

Figure 1.. 17β-estradiol induces DNA damage in long-term estrogen-deprived cells that therapeutically respond to E2.

(A) Parental cells were HD for 3 d prior to seeding. Cells were treated as indicated for 4 wk and relative growth was measured. (B) 1428/LTED cells stably expressing Cas9 were transduced with a sgRNA library. Cells were treated ± 1 nM E2 for 3 wk, and beta-scores were calculated (18). Each point represents one gene. Genes with a differential beta-score ≥0.5 (n=1194, gray box) were analyzed for enrichment with Hallmarks gene sets (right). Cell cycle and DNA repair pathways are highlighted in red font. (C) HCC-1428 (HD for 3 d) and 1428/LTED cells were treated ± 1 nM E2 for 24 h, and RNA was harvested for sequencing. Single-sample gene set enrichment analysis (ssGSEA) for Hallmarks pathways was performed for each replicate sample, and normalized gene set enrichment scores (NES) were compared between treatment groups. Gene sets significantly (p<0.05) altered in 1428/LTED cells by E2 treatment are shown, and p-values and differences in mean NES induced by E2 treatment are indicated for each cell line. (D) Cells treated ± 1 nM E2 x 24 h were fixed and stained with propidium iodide (PI). DNA content was measured by flow cytometry. Sub-G1 cells were excluded from plots. Proportions of cells in each phase were compared. (E) Cells were treated ± 1 nM E2 as indicated. Three days after last medium change, cells were harvested, stained with FITC-tagged Annexin V (AnnV), and analyzed by flow cytometry. In (A/D/E), data are shown as mean of triplicates ± SD. *p<0.05, **p<0.005, ***p<0.0005, n.s. = not significant compared to control unless otherwise indicated.

Figure 2.. 17β-estradiol-induced DNA damage is dependent upon overexpression of ER.

(A/B) HCC-1428 (HD x 3 d) and 1428/LTED cells were treated ± 1 nM E2. Cells were fixed and stained for γH2AX (green) and DAPI (blue). γH2AX foci were counted in ≥100 nuclei/group. (C) HCC-1428 (HD x 3 d) and 1428/LTED cells were treated ± E2 for 21 h, and labeled with BrdU for another 3 h. Cells were stained for γH2AX, BrdU, and cleaved PARP for flow cytometry analysis. Cleaved PARP-positive cells were excluded from analysis. Proportions of cells with DNA breaks (γH2AX-positive) that were or were not in S-phase (i.e., did or did not incorporate BrdU) were plotted. Proportions of γH2AX+/BrdU+ cells were statistically compared. Data are shown as mean of triplicates + SD. (D) Cells were treated ± E2 ± abemaciclib x 24 h and analyzed as in (B). (E/F) T47D/pInd20-ESR1 cells were pretreated with HD x 7 d, and then treated with HD ± dox x 14 d prior to seeding. All cells were then treated as indicated x 24 h and analyzed as in (B). (G) 1428/LTED cell lines expressing dox-inducible shRNA targeting ESR1 (two independent constructs) or non-silencing control were treated ± dox for 2 d, and then treated ± E2 x 24 h and analyzed as in (B). In (A/E), representative images are shown. *p<0.05, **p<0.005, ***p<0.0005, ****p<0.0001, n.s. = not significant.

Figure 3.. 17β-estradiol therapy induces DNA damage in endocrine-resistant tumors.

(A) Biopsy samples of advanced ER+ breast tumors were obtained from 2 patients before and 2 wk after treatment with E2. FFPE tumor sections were stained for ER (red), γH2AX (green), and DAPI (blue). γH2AX intensity was quantified in ≥100 nuclei/specimen. Representative exposure-matched image pairs are shown. (B) Ovx mice bearing tumors ~200 mm³ were randomized to treatment ± E2. Tumor volumes were serially measured. Data are shown as mean + SEM, and were analyzed by nonlinear mixed modeling. (C) Tumors (n=3/group) were harvested from ovx mice treated ± E2 for 24 h. FFPE sections were analyzed by IHC. Data are shown as mean ± SD. *p<0.05, ****p<0.0001, n.s. = not significant.

Figure 4.. 17β-estradiol-induced R-loop formation drives DNA damage.

(A/B) Cells were treated ± 1 nM E2 x 24 h, fixed, and stained for DNA/RNA hybrids (S9.6 antibody) and with DAPI. Fluorescence intensity was quantified in ≥100 nuclei/group. (C) Tumors were harvested from ovx mice treated ± E2 (n=3/group). FFPE sections were stained for DNA/RNA hybrids. Proportions of positively staining nuclei were calculated. Data are shown as mean + SD. (D/E) Cells were transiently transfected with plasmids encoding RNase H1 or vector control. Two days later, cells were treated ± 1 nM E2 x 24 h, then fixed and stained for γH2AX and DAPI. γH2AX foci were counted in ≥100 nuclei/group. In (A/D), representative images are shown. *p<0.05, **p<0.005, ***p<0.0005, ****p<0.0001, n.s. = not significant.

Figure 5.. PARP inhibition synergizes with 17β-estradiol to induce DNA damage and inhibit growth in vitro.

(A) T47D/pInd20-ESR1 cells were HD x 7 d, and pretreated with HD + dox x 14 d prior to seeding. Both lines were then treated ± olaparib for 2 d, followed by treatment ± olaparib ± E2 for 4 wk, and relative growth was measured. Data are shown as mean of triplicates + SD. (B/C) T47D/pInd20-ESR1 cells were pretreated as in (A) before seeding. All lines were then treated ± olaparib for 2 d, followed by treatment ± E2 ± olaparib for 24 h. Cells were fixed and stained for γH2AX and DAPI. Representative images are shown. γH2AX foci were counted in ≥100 nuclei/group. *p<0.05, **p<0.005, ***p<0.0005, ****p<0.0001, n.s. = not significant.

Figure 6.. PARP inhibition synergizes with 17β-estradiol against tumors in vivo and prevents recurrence.

(A/B) Ovx mice bearing tumors ~200 mm³ were randomized to treatment as indicated. E2 was delivered continuously, and olaparib was administered daily for 28 d (gray shading). Data are shown as mean + SEM. (C) After 10 wk of treatment in (A), mice without palpable tumors were monitored for recurrence. Time to recurrence was calculated as time from treatment start until tumors re-grew to baseline volume. Proportions of mice that were recurrence-free over time are shown. (D) Ovx mice bearing WHIM16 tumors were treated as in (A), and tumors were harvested 4 h after drug treatment on Day 2. Tumor lysates were analyzed by immunoblot. vinc.- vinculin loading control. (E) Mice bearing tumors that recurred during E2 monotherapy in (C) were treated with estrogen deprivation starting on Day 0, which stunted tumor growth. Each line represents one mouse. Tumors eventually resumed growth, which was defined as two consecutive biweekly volume measurements above baseline. (F) Mice with tumors that resumed estrogen-independent growth in (D) were randomized to treatment with a second cycle of E2 ± olaparib. Data are shown as mean + SEM. ***p<0.0005, n.s. = not significant.