Cell type-specific genotoxicity in estrogen-exposed ovarian and fallopian epithelium

Cancer cell lines were grown in Dulbecco’s DMEM supplemented with 10% calf serum plus 100 U/ml Penicillin and 100 mg/ml streptomycin. SKOV-3 (HTB-77) and A549 (CCL-185™) were purchased from ATCC. HO8910 (3111C0001CCC000280) and HO8910-PM (3111C0001CCC000281) were purchased from National Infrastructural of Cell Line Resources, Beijing. OVCAR-8TR and L02 lines were gifted by collaborating laboratory. Cell lines were authenticated upon purchase and regularly tested for mycoplasma contamination.

Ovarian tissues diagnosed with OvCa were used in this study along with one ‘healthy’” ovary without detectable malignancy. Freshly excised tissue blocks were embedded in OCT, followed by snap frozen in liquid nitrogen and stored in -80 °C. Deep frozen tissue blocks were sectioned at 8 μm using microtome (LEICA, CM3050 S). For ex vivo culture of OvCa biopsies, freshly excised tissues were subjected to mincing and enzymatic digestion (Collagenase A, Roche, 10,103,578,001; Trypsin, GIBCO, 25200–056) for 2 h with occasional vortex. Isolated single cells were dispersed in RPMI-1640/10% FBS and grown in media supplemented with 10 mg/ml Matrigel (BD, 356234). After 48 h, experiments were performed when 70% of cells were attached.

Eight reproductively mature female mice (C57/B6, 6 weeks old) were divided into two groups: group A, no estrogen treatment; group B, intraperitoneal injection with estrogen (Selleck, S1709, 1 mg/kg) for 6 h. Dissected liver, ovary and fallopian tube were instantly frozen in liquid nitrogen and cryosectioned for indirect immunofluorescent staining. All animals were housed in standard SPF condition throughout the experiments. Animals were sacrificed with minimal pain by neck broken protocol following approved CO2 euthanasia procedure.

Cryopreserved sections or cells grown on coverslip were fix with 4% paraformaldehyde (PFA) and permeabilize with 0.3%TritonX-100, followed by blocking in PBS with 3%BSA, 3% donkey serum and 0.2% Triton X-100. Primary antibodies were diluted with antibody buffer (3% Triton/10% BSA in PBS) and incubated overnight: phoshpo-BRCA1 (Bethyl, A300-001A, 1:1000), phosphor-RPA32 (NOVUS, NB100–544,1:3000), RAD51 (Proteintech, 14,961–1-AP, 1:500), Cyclin A2 (Huabio, ET-1612-26,1:500), anti-53BP1 (Bethyl, A300-272A,1:1000), anti-γH2AX (Millipore, 05–636,1:500), anti-BRCA2 (Invitrogen, MA5–32986, 1:500). Fluorescent images were acquired using OLYMPAS (BX51) and images were processed analysed using Image-Pro Plus software.

Cells were treated with hormones at the following concentrations: tamoxifen (Sigma, T5648), 50 nM; estrogen (17β-estradiol, Wako), 50 nM. Intraperitoneal injection of estradiol (Selleck, S1709, 1 mg/Kg) was applied to animals. Time of treatment are described for each experiment.

Lipofectamine 3000 transfection kit (Invitrogen, 3,000,015) was used for siRNA transfection for cells grown to 70% confluency. Cells were collected 48 h after transfection for indicated experiments. Individual siRNA duplexes used were: BRCA1 (target sequence: UCUGCUGUAUUGGAACAAAUU); BRCA2 (target sequence: AAC AACAAUUACGAACCAAACUU). Knockdown efficiencies of these genes are shown in Supplementary Figure 2. To quantify gene expression levels, total cellular RNA was extracted by using TRIzol® Reagent (Invitrogen) and cDNA was synthesized using Eastep RT Master Mix Kit (Promega). RT-PCR products were visualized by agarose gel electrophoresis. mRNA levels were normalized using GAPDH or actin RNA as internal control.

The Student’s t-test was performed on all data analysis. Each experiment had at least three independent biological replicates. Unless otherwise specified, data are showed as mean ± s.e.m. p < 0.05 was considered to be statistically significant. Excel and GraphPad Prism was used to create the graphs and calculate the p value.

In advanced stages of cancer, uncontrolled DNA replication in fast-proliferating cancer cells can increase the chances of DNA damage. To evaluate the level of endogenous DNA damage in OvCa cells, we examined the activation of DSB repair factors in OVCAR-8TR cells. In comparison with hTERT-transformed human primary cells (RPE1, non-cancerous) without obvious DNA breaks, ~ 11–17% percentage of OVCAR-8TR cells were marked with foci of 53BP1, phosphorylated RPA32 and BRCA1 (Fig. 1a), indicating significant levels of endogenous DNA damage in OvCa cells.

Endogenous DNA breaks in OvCa cells were further solidified by assays in ex vivo culture from fresh surgical biopsies. We found 63% primary OvCa cells displayed endogenous 53BP1 foci (> 3 per cell) (Fig. 1b). However, it seems that the basal level of 53BP1 was not correlated with the genomic replication in ex vivo culture, as manifested by the low rate of BrdU-positive OvCa cells (2%). Thus, we conclude that OvCa cells suffer from strong genotoxic burden that is not accompanied by genomic replication.

Ovarian surface epithelium is periodically exposed to high level of steroid hormones (estrogen and progesterone), triggering cycles of proliferation, damage and repair known as ovulation trauma [23]. Besides potent proliferation stimulation, estrogen also has a direct genotoxic impact on estrogen-positive cells (ie. breast), albeit the precise mechanism remains elusive [24‐26]. We hypothesized that the high level of endogenous DNA lesions observed in above OvCa cells, including a significant fraction of from replication-independent damage, may largely arise from exposure to estrogen containing in culturing media.

To determine the impact of steroid hormones on the origin of DNA lesions in ovarian cells, estrogen was added to proliferating OVCAR-8TR cells, followed by the monitoring of 53BP1 and γH2AX-represented DSBs. Clearly, addition of 50 nM estrogen for 12 h elevated the endogenous level of 53BP1 and γH2AX foci, indicating a robust generation of DNA breaks upon steroid stimulation (Fig. 2a-b). These hormone-induced DNA lesions were correlated to the expression levels of estrogen nuclear receptor (ER), as cells originated from non-female reproductive tissues including L02 (liver) and A549 (lung cancer) displayed minimal basal level of γH2AX (< 1 per cell), and did not respond to estrogen treatment (Fig. 2b). Coherently, expression levels of ER in OvCa cells (OVCAR-8TR, HO-8910 and HO-8910 PM) as measured by RT-PCR were appreciably higher than L02 and A549 cells (Fig. 2c). Furthermore, addition the same concentration of tamoxifen, which is the chemical mimics of estrogen but biologically inactive, did not induce equivalent number of 53BP1 foci in OVCAR-8TR cells (Fig. 2d). Thus, we conclude that exposure of ER-positive OvCa cells to estrogen triggers genotoxicity.

Sasanmura et al. detected DSBs in G1 phase upon estrogen induction20, in which stage HR is not active due to the absence of homologous sister chromatids. We investigated if the estrogen-dependent elevation of 53BP1 occurs in cell cycle phases beyond G1. Although previous work described foci formation after treatment with 10 nM estrogen for 2 h in MCF-7 cells, DNA lesions were not detectable shorter than 6 h in OVCAR-8TR cells. In asynchronized OVCAR-8TR culture, γH2AX increase (> 10 foci per cell) was observed in 40% of cyclin A-positive cells, which represents G1 phase cells (Fig. 2e). This is consistent to the estrogen-induced accumulation of 53BP1 foci in serum starved OVCAR-8TR cells (Fig. 2f), confirming the impact of estrogen on G1 cells. In parallel, we also detected comparable elevation of 53BP1 in a subpopulation of cyclin A-negative cells (29%), indicating that non-G1 phase cells (S/G2 and mitosis) are also liable to the genotoxicity of estrogen (Fig. 2e). Consistently, arresting cell cycle in S phase by pre-treating OVCAR-8TR with hydroxyurea produced a remarkable additive elevation of 53BP1 foci on top of estrogen exposure, indicating that S-phase cells are sensitive to estrogen (Fig. 2g). Thus, estrogen-induced genotoxicity can occur in both G1 and S phases of OvCa cells, implying the involvement of HR-dependent and independent mechanisms to eliminate these damages.

In the light of the specific impact of estrogen on ER-positive OvCa cells, we predict ovary-derived epithelial cells are more dependent on HR genes (ie. BRCA1/BRCA2) than ER-negative cell types. We first examined the activation of BRCA1 by its phosphorylated form (pS1524), and noticed a ubiquitous phosphorylation signals in OvCa biopsies of both borderline and higher malignant stages (moderately differentiated) (Fig. 3a). A complete negativity in normal ovarian tissues was observed in parallel. Reminiscent of the 53BP1 foci formation in ex vivo culture (Fig. 1b), the ratio of phosphorylated BRCA1 positivity is as high as 85% in cancerous epithelial cells, while the corresponding proliferation rate was only 11% as indicated by Ki67 positivity (Fig. 3b). These observations imply that BRCA1 can be activated in both resting and proliferative OvCa cells, consistent to its dual function in G1 phase (cooperating with MRE11 nuclease) and S/G2 (facilitating resection and committing HR).

Secondly, recruitment of HR factors to DNA lesions in response to estrogen was examined. Higher level of RPA phosphorylation was detected in estrogen-treated OVCAR-8TR cells, indicating a robust generation of single-strand DNA that is required for recombination (Fig. 3c). In consistence, foci formation of RAD51 recombinase and phosphorylated BRCA1 was also increased upon estrogen exposure (Fig. 3d-e), consistent to the S phase response under the same condition (Fig. 2g). Therefore, we conclude that estrogen exposure stimulates homologous recombination in OvCa cells, and that BRCA1 can function beyond HR in both resting and replicating cells.

To establish the role of HR genes in estrogen-induced genotoxicity in ovarian cancer cells, we examined DSB formation upon combined treatment of estrogen and BRCA1/BRCA2 ablation in OvCa cell lines from different cancer origins (HO-8910, HO-8910 PM, OVCAR8-8TR and SKOV3), as well as L02 and A549 from non-reproductive tissue. Although depletion of BRCA1/siBRCA2 caused moderate but insignificant increase of 53BP1 and γH2AX in L02 and A549 cells, no additive impact was observed upon estrogen addition (Fig. 4a-b), indicating the irresponsiveness of these non-reproductive cell types due to their low ER expression. In sharp contrast, 53BP1 focal numbers in all OvCa cells but SKOV3 were significantly induced upon estrogen exposure (Fig. 4c-f). When BRCA1/BRCA2 were depleted by RNA interference, all cell lines exhibited 53BP1 elevation relative to scramble transfection.

Intriguingly, BRCA1/BRCA2 silencing caused additive impact on 53BP1 accumulation in sibling HO-8910 and HO-8910 PM cell lines on top of estrogen addition, as shown by the comparison of Scramble/E2 versus siBRCA/E2 (Fig. 4c-d). This indicates that HO-8910/HO-8910 PM require the function of BRCA genes for efficient elimination of estrogen-induced DNA breaks. Unlike its sibling line, HO-8910 PM cells did not display excessive increase of 53BP1 foci upon double treatment relative to siBRCA1/siBRCA2 alone. This could be attributable to its higher ER expression level than HO-8910 (Fig. 2c), conferring a more rigorous response to the basal level of estrogen in culture medium in the absence of BRCA1/BRCA2.

For OVCAR-8TR cells, estrogen induced more 53BP1 foci upon ablating BRCA1/BRCA2, but did not exacerbate the overall number of breaks (Fig. 4e), suggesting that BRCA genes do not contribute to repair estrogen-induced DNA damage in OVCAR-8TR. Alternatively, the breaks have reached plateau upon double treatment, considering the median number in OVCAR-8TR is ~ 10 foci per cell compared to 4–5 in other three cells. Taken together, these data consolidate our speculation that HR are required by OvCa cells to remove DNA lesions upon estrogen exposure, though differential responses were observed for individual cell lines.

It is postulated that OvCa arise from epithelial cells in epithelial compartments of ovary and fallopian tubes [27, 28]. Based on the results described above, we hypothesis that storming of steroid hormones may impair the genomic integrity of non-cancerous cells locating at these pro-carcinogenic sites. To evaluate the in vivo response of ovarian cells to estrogen exposure, we examined the damage responses of epithelial compartments after intraperitoneal injection of estrogen (1 mg/Kg) to mature female mice for 6 h. Indirect immunofluorescent staining of dissected ovarian cryosections showed that a vast number of 53BP1 and γH2AX foci emerged in estrogen-exposed epithelial cells (Fig. 5a). About 86% of ovarian epithelial cells were positive for 53BP1 foci, that is in sharp contrast to the mock-treated ovary and control tissues such as liver (Fig. 5b).

The dramatic DDR response is also observed in fallopian tubes, where similar levels of foci induction were monitored for 53BP1 and γH2AX (Fig. 5c). Intriguingly, we noticed a pronounced cell type specificity in response to estrogen treatment, in that the estrogen-induced foci are nearly exclusively detected in fallopian epithelial monolayer instead of smooth muscle (Fig. 5c-d). This is consistent to the previous detection of BRCA1 mutation and precancerous lesions in fallopian epithelium [27, 29]. Taken together, we conclude that exposure to estrogen specifically challenges the genome stability of ovarian as well as fallopian epithelium, making HR mechanisms obligatory for eliminating the genotoxicity of these cells.