Poly (ADP-ribose) polymerase inhibitor exposure reduces ovarian reserve followed by dysfunction in granulosa cells

The use of poly (ADP-ribose) polymerase (PARP) inhibitors is expected to increase, but their effect on fertility is still unclear. The aim of this study was to investigate the effect of PARP inhibitors on ovarian function. In an in vitro study, cultures of ovaries and granulosa cells (GCs) exposed to the PARP inhibitor olaparib were evaluated by real-time RT-PCR, histological study, and hormone assays. In an in vivo study, mice were administered olaparib orally and evaluated via in vitro fertilization (IVF), follicle count, immunohistochemical staining, and real-time RT-PCR. In vitro, the gene expression of GC markers decreased in the olaparib-treated group. Olaparib also negatively affected estradiol production and the expression of GC markers in cultured GCs, with abnormal morphology of GCs observed in the treated group. The follicle number indicated depletion of follicles due to atretic changes in the treatment group, both in vitro and in vivo. Also, olaparib reduced the number of retrieved oocytes and the fertilization rate of IVF, but they recovered after 3 weeks of cessation. Our results indicate that olaparib is toxic to ovaries.

The number of cancer survivors has increased due to advances in treatment. The effects of cancer treatment on fertility are well known, and fertility preservation treatment and social awareness about fertility preservation are also increasing. The American Society of Clinical Oncology (ASCO) guidelines in 2013 categorize chemotherapy and radiation as treatments that may cause gonadal damage 1,2 . However, as a matter of course, ASCO guidelines cannot explain the gonadal damage caused by all cancer drugs, because new drugs are always under investigation. One class of cancer drugs which has unknown effects on fertility is poly (ADP-ribose) polymerase (PARP) inhibitors, which are presently used for breast cancer and ovarian cancer treatment.
Olaparib is a first-in-class oral PARP inhibitor used to treat patients with BRCA mutations that often cause breast cancer in young women, known as hereditary breast and ovarian cancer (HBOC) 3 , and patients with HER2-negative inoperable breast cancer who were previously treated with chemotherapy or who have recurrent breast cancer. Cancers with DNA repair defects, such as those with a BRCA1 or BRCA2 (BRCA1/2) mutation, can be treated with PARP inhibitors, which render them deficient in homologous recombination repair [4][5][6] . In tumors that lack this type of repair, PARP inhibitors block an alternative DNA repair pathway that is required for viability, resulting in their death 7 . As mentioned by the National Comprehensive Cancer Network, olaparib is also a standard maintenance therapy for patients with recurrent ovarian cancer after treatment with platinumbased chemotherapy, and for patients with or without BRCA mutations [8][9][10][11][12][13][14][15] . Moreover, in patients with advanced ovarian cancer with a BRCA mutation, olaparib is approved as a fourth-line or later treatment whether or not the patient is sensitive to platinum-based drugs 16 . Several studies have shown that olaparib as a maintenance therapy is beneficial in terms of progression-free survival in patients with newly diagnosed advanced ovarian cancer and a BRCA1/2 mutation; such treatment provides a 70% lower risk of disease progression or death compared to placebo 16 . Other studies have also suggested that PARP inhibitors may be effective for various other cancers [17][18][19][20] . Hence, PARP inhibitors are promising drugs that are likely to be used for more patients in the future.
In addition to their effects on DNA repair, PARP inhibitors also have anti-angiogenesis effects due to their inhibition of vascular endothelial growth factor (VEGF) action [21][22][23] . VEGF has effects on ovarian follicular development such as primordial follicle survival, mitogenic effects on granulosa cells (GCs), and the transition from primary to secondary follicles 24 . A follicle is a dynamic unit consisting of GCs, theca cells, and an
Histological evaluation of the effects of the PARP inhibitor on ovaries. As shown in Fig. 3, the shape of the GCs was not maintained in the olaparib group, and obvious abnormalities were observed via HE staining. VEGFR-2 (which is moderately present on oocytes 30 ) and Ki-67 (a cell proliferation marker) were detected by immunohistochemical staining. In the control group, VEGFR-2 was stained in blood vessel-like structures throughout the ovaries, whereas in the olaparib groups, VEGFR-2 was not present in the blood vessellike structures. Also, oocytes had a weak reaction for VEGFR-2 in the control group. In the control condition, Ki-67 was strongly stained, mainly in the GCs, whereas in the olaparib groups, it was not stained.
Effects of the PARP inhibitor on granulosa cell culture. GCs were cultured for 6 h before the extraction of RNA for real-time RT-PCR analyses of transcript levels. As shown in Fig. 4, both CYP19a and FSHR showed significantly decreased gene expressions in the olaparib group (p = 0.009).
Transition in the E2 level in cultured ovary media with the PARP inhibitor. The temporal changes in the E2 concentration are shown in Fig. 5. The E2 concentration in the control medium increased over time, but the concentration in the olaparib groups did not change. The E2 concentrations in day 8 medium were: control 3919.3 ± 1694.5, Ola10 51.6 ± 16.9, and Ola100 81.5 ± 5.4 pg/ml. The differences between the control and olaparib groups were significant (p = 0.009).
Olaparib effect on in vitro fertilization. As shown in Table 1, in vitro fertilization (IVF) was compared between mice administered olaparib and vehicle to evaluate the effect of olaparib on IVF. The factors evaluated were the body weight, number of retrieved oocytes, rate of mature oocytes, fertilization rate, and arrival rate of blastocysts. The numbers of oocytes retrieved were 11.7 ± 2.2 for the control and 3.8 ± 3.0 for the olaparib group. The number of oocytes retrieved was significantly lower in the olaparib group compared to the control group (p = 0.0002). The fertilization rate was also lower in the olaparib group compared to the control group (p = 0.03). www.nature.com/scientificreports/ In addition, as a result of IVF after 3 weeks of olaparib cessation, the number of retrieved oocytes had a tendency to be lower in the olaparib group compared to the control group (p = 0.094). Also, after the cessation of administration, the number of oocytes retrieved was obviously recovered in the olaparib group (p = 0.007).
Gene expression related to ovarian reserve changes in olaparib-treated mice. The expressions of CYP19a, FSHR, GDF9, VEGF, VEGF receptor-1 and -2, CD31, and Ki-67 were evaluated after 2 weeks of treatment with olaparib in vivo. As shown in Fig. 6, the gene expressions of CYP19a (GC marker) (p = 0.013), Figure 3. Histology evaluation of in vitro cultured ovaries with olaparib treatment. In the control group, VEGFR-2 was stained in blood vessel-like structures throughout the ovary, including around the follicle (black arrows). Also oocytes (red arrows) were stained for VEGFR-2. In the control, granulosa cells were stained for Ki-67 (blue arrows). However, granulosa cells were not stained in the olaparib group for VEGFR or Ki-67. Also, VEGFR was not expressed on oocytes of the olaparib group. HE staining shows that granulosa cells grown in olaparib did not maintain their shape and had obvious abnormalities. (a-h), scale bar = 100 µm; (i, j), scale bar = 10 µm (Color figure online).
Change in follicle development in olaparib-treated mice. The follicle number after treatment with olaparib for 2 weeks is shown in Fig. 7. All types of developing follicles were significantly reduced in the olaparib group (total: p = 0.028, primordial: p = 0.047, primary: p = 0.009, early secondary: p = 0.047, late secondary: p = 0.028, preantral: p = 0.047, antral: p = 0.026). In addition, in the olaparib group, no corpus luteum was observed. Regarding the distribution of follicles, the percentage of atretic follicles was significantly higher in the olaparib group compared to the control group (p = 0.009).

Discussion
The ovaries have two main reproductive functions: (1) provide oocytes for fertilization and (2) secrete sex steroid hormones (i.e., estrogen and progesterone). Therefore, female gonadotoxic agents can cause the depletion of oocytes and dysfunction in hormone production. According to previous reports, PARP inhibitors affect embryo development 31,32 . However, there is no report concerning the gonadotoxicity of PARP inhibitors in vivo. To the best of our knowledge, the present study is the first that indicates ovarian toxicity of a PARP inhibitor in vivo. PARP constitutes a large family of 18 proteins encoded by different genes and displaying a conserved catalytic domain in which PARP-1 (113 kDa), the founding member, and PARP-2 (62 kDa) are so far the sole enzymes whose catalytic activity has been shown to be immediately stimulated by DNA strand breaks 33 . They can be inhibited by olaparib, which was used in the present study. Female fertility can be affected by cancer therapy by reducing the number of primordial follicles, compromising the hormonal balance, or inducing dysfunction of the ovaries, fallopian tubes, uterus, or cervix. Anatomic or vascular changes to these organs following surgery or radiation may also inhibit natural conception and pregnancy, leading to the need for assisted reproductive technology or a surrogate. The type of drug, dose, and protocol; the size and location of the radiation field, radiation dose, and intensity; the route of administration (oral versus intravenous); disease type; age; sex; and fertility pretreatment all impact the effect of radiation and chemotherapy 34 .
Females with cancer treated with chemotherapy and radiotherapy have a risk of premature menopause and primary ovarian insufficiency due to depletion of ovarian follicles, stromal fibrosis, and vascular damage 35 . One chemotherapy example is cyclophosphamide, which is an alkylating drug with high gonadotoxicity to dividing www.nature.com/scientificreports/ cells. This drug kills actively growing follicles and also activates quiescent follicles. Upon the growth of these previously quiescent follicles, they also become sensitive to the effects of cyclophosphamide 36 . Some treatments have known effects on gonads, but the effects of other treatments are still unknown. For example, molecular targeted drugs are relatively new, and the ovarian toxicity of many of these drugs, such as imatinib, bevacizumab, and olaparib, is unknown. Although molecular targeted drugs are not considered cytotoxic, they can affect growth factors and may thus affect follicular development and ovulation. Olaparib not only has an antitumor effect, but also has anti-angiogenesis, anti-inflammatory, and immune tolerance effects 37 . Consequently, olaparib may possibly be used not only as an anticancer agent but also as a therapeutic agent for chronic inflammatory diseases and autoimmune diseases.
VEGF is important for maintaining the number of follicles 24 . The role of VEGF in ovarian folliculogenesis and steroidogenesis is consistent with the observation that follicular development and E2 synthesis are blocked by short-term treatment of monkeys with anti-VEGF antibody at the late follicular stage 38 . Furthermore, VEGFR is also required for the survival of primordial follicles 39 , and inhibition of VEGFR-2 blocks follicular development 29 . In accordance with these reports, we confirmed VEGFR expression on oocytes (Fig. 3). An increasing dose of olaparib evidently decreased the level of VEGFR. The decrease in VEGFR directly affected GC function and ovarian reserve, both in vivo and in vitro, as evidenced by decreased CYP19a (a GC marker), follicular count, and GDF-9 (an oocyte marker). In addition, olaparib decreased CD31 gene expression in vivo and was not observed in blood vessel-like structures via immunohistochemical staining. Therefore, it is possible that as an indirect effect, olaparib suppresses angiogenesis, reducing follicle survival and folliculogenesis. Hence, our findings are consistent with previous reports. However, the present experiments only showed a decrease in VEGFR, and downstream effects on the VEGFR pathway are still unclear. These questions should be answered by future studies.
Our in vivo experiments showed that the number of oocytes retrieved was strongly reduced. The decrease in the number of oocytes retrieved and the low E2 level could be explained by the decrease in the number of follicles. However, low gene expressions of aromatase and FSHR, as seen with the GC culture, indicated that olaparib impaired the quality and/or function of GCs. Our study demonstrated that olaparib affected both the quality and the amount of GCs. In addition, olaparib decreased the fertilization rate, but the blastocyst development rate was maintained in IVF. The factors that affect successful fertilization include oocyte nuclear and cytoplasmic maturation 40 . Furthermore, in the fertilization process, several recognizable events take place, including the sperm acrosome reaction, penetration of the egg zona pellucida by sperm, and the egg cortical reaction and zona reaction. When these processes fail, fertilization fails. VEGFR has also been identified tentatively in the zona pellucida 39 . We speculate that these reactions were interfered with by the effects on VEGFR. In addition, according to a previous report, GC promotes the capacity of oocytes to undergo fertilization 41 . Therefore, it is also possible that the fertilization rate decreased due to GC damage. However, our experiments could not determine the cause of fertilization failure. It is necessary to perform intracytoplasmic sperm injection to investigate the cause.
According to previous reports, PARP inhibition affects embryo development by clearly reducing the embryo development rate and the number of blastocysts, and by increasing the apoptotic index. The inhibition of PARylation impacts cumulus cell expansion because of abnormal changes in maturation-and expansion-related gene expression, and it decreases the embryo developmental rate and parthenogenetic activation of embryo quality 31 . Furthermore, PARylation is associated with pronuclear fusion in post-fertilization processes 32 . The www.nature.com/scientificreports/ result of embryo development is different between the present study and previous reports, which may be due to different study protocols. According to the recommendation of the United States Food and Drug Administration for genotoxic pharmaceuticals after the cessation of therapy, a contraception period of 6 months, 5 times the half-life period, covers the growth and maturation phase of folliculogenesis and is expected to allow for the elimination of most damaged follicles in humans 42 . Based on this, the contraception period is approximately 3 weeks in mice, and IVF was performed 3 weeks after cessation. Comparing the olaparib group and cessation group, the number of retrieved oocytes and the fertilization rate in cessation group were significantly higher than in the olaparib group. Therefore, the effect of olaparib on IVF may be temporary. Also, we speculate that the recovery of the number of retrieved oocytes after 3 weeks of cessation was due to the elimination of the olaparib effect on cumulus cell expansion 31 . However, the number of retrieved oocytes after 3 weeks of cessation tended to decrease compared to the control group, which also had 3 weeks of cessation (p = 0.094). We speculate that this result may depend on the decrease of ovarian reserve due to olaparib treatment.
At present, olaparib is used for breast cancer and ovarian cancer, but it is used in ovarian cancer patients with and without BRCA mutations. Also, olaparib has been further expanded in the European Union for the treatment of germline BRCA -mutated HER2-negative advanced breast cancer. Some young patients use olaparib because those with BRCA mutations develop cancer at young age 43 . Although it is still controversial, BRCA mutations may be associated with low ovarian reserve and occult primary ovarian insufficiency 44 . Therefore, more aggressive fertility preservation treatment for patients with BRCA mutations may be considered.
Clinical trials for pancreatic cancer, prostate cancer, and pediatric cancer with solid tumors are being conducted, and further expansion of olaparib use is expected 45,46 . Therefore, understanding the gonadotoxicity of olaparib is important. One reason for the anti-cancer effect of olaparib is its anti-angiogenic effect, and our results are consistent with the hypothesis that this indirect effect reduces follicle survival and folliculogenesis. Based on this, the use of olaparib treatment for young cancer patients who wish to have children should be carefully considered, because the present study provides preliminary evidence of a possible risk of infertility to such patients.
In summary, we investigated the ovarian toxicity of a PARP inhibitor, a chemotherapeutic drug that may be widely administered to younger female patients in the future. According to our investigation, olaparib has direct (without affecting angiogenesis) and indirect (affecting angiogenesis) effects on ovarian follicles that reduce follicle survival and folliculogenesis. Hence, we conclude that olaparib may lead to infertility based on low ovarian reserve and GCs dysfunction (Fig. 8).
We have shown that olaparib affects ovaries. These findings suggest the necessity for fertility preservation among young patients who will receive olaparib treatment. However, to clarify this, additional analyses using non-human primates or human clinical research are needed to extrapolate these findings, because the present study was performed with a rodent model. Also, the long-term effects of PARP inhibitors and methods to protect ovaries from damage should be studied. In addition, the effects of olaparib on the pregnancy course (pregnancy rate, live birth rate, etc.) should be examined.

Methods
Animals and experimental conditions. ICR mice (Japan SLC, Tokyo, Japan) were housed with a 12-h/12-h light/dark cycle in an approved animal facility, with food and water provided ad libitum. For in vitro studies, ovaries were collected from day 10-11 female mice and placed in L-15 medium (Lonza, Tokyo, Japan) www.nature.com/scientificreports/ supplemented with 3 mg/ml bovine serum albumin (Sigma-Aldrich, St. Louis, MO, USA). Also, ovaries were collected from mice at 12 weeks of age for the GCs culture assay. For the in vivo study (real-time PCR, IVF, and histological analysis), in brief, olaparib was mixed into 1% carboxymethyl cellulose (Wako, Tokyo, Japan) to make a 30 mg/ml suspension. Female mice were orally administered control fluid (1% carboxymethyl cellulose) or olaparib (300 mg/kg) for 2 weeks using a feeding needle, starting at postnatal day 21. The dose was determined regarding the minimum lethal dose described in the package insert. Also, as an additional study, IVF was performed to confirm the recovery from olaparib treatment after 3 weeks of cessation.
Ethics approval. All  Histological analysis. At the end of the culture, ovaries (n = 5) were fixed in 10% buffered formalin for 12 h and embedded in paraffin wax. Serial sections of 6 µm thickness were cut and stained with hematoxylin and eosin (HE) for histomorphological observation and follicle number counting. The stages of follicular development were defined as follows: primordial follicle (a single layer of flattened pre-GCs), primary follicle (one cuboidal GC layer), early secondary follicle (two cuboidal GC layers, no antrum), late secondary follicle (more than two cuboidal GC layers without an antrum cavity), preantral follicle (multiple layers of cuboidal GCs with scattered antrum cavities), and antral follicle (multiple layers of cuboidal GCs with a large single antrum cavity). Follicles with pyknotic nuclei in the GCs and/or degeneration of the nucleus of the oocyte were defined as atretic follicles. To prevent double-counting, only follicles with oocyte nuclei were counted 48-50 . Immunohistochemical analysis. Immunohistochemical staining was performed according to the manufacturer's instructions. Briefly, sections were blocked in 5% skim milk in phosphate-buffered saline (PBS) (Wako) at room temperature for 10 min after antigen retrieval with heating. Immunoreactions were performed with primary antibodies in PBS at 4 °C overnight. Primary antibodies for immunohistochemistry were rabbit monoclonal anti-VEGFR-2 (Cell Signaling Technology, Danvers MA, USA; 1:100) and rabbit monoclonal anti-Ki-67 (Cell Signaling Technology; 1:400). After washing with PBS, the sections were reacted with reagents from the Histostain Plus immunohistochemistry kit (Life Technology, Carlsbad, CA, USA). The peroxidase reaction was developed using a diaminobenzidine reaction kit (Abcam, Tokyo, Japan). The sections were observed with a microscope (Bio-Revo BZ-X700, KEYENCE, Tokyo, Japan). The relative abundance of specific transcripts was normalized using β-actin as a housekeeping gene 48 . The relative expression of the mRNAs of interest was calculated with the ΔΔCT method. The primers used are shown in Table 2.

Real-time RT-PCR analyses.
Hormone assays. 17β-Estradiol (E2) concentrations in medium (n = 5) were measured using an Estradiol EIA Kit (Cayman Chemical Co., Ltd., Ann Arbor, MI, USA) following the manufacturer's protocol. The plate was read at a wavelength of 412 nm.
Granulosa cell culture. Antral follicles were separated from 10 ovaries for the collection of GCs. The GCs were collected by gently scraping the follicle wall with 27G needles and were pooled. After washing with DMEM, the collected GCs were cultured in plastic 24-well plates in 400 µl Ham's F-12: Dulbecco's minimal essential medium (1:1, v/v) (Sigma-Aldrich) containing 10% fetal bovine serum (Cosmo Bio Co., Ltd., Tokyo, Japan) and 1% antibiotic-antimycotic (Nacalai, Kyoto, Japan) with/without olaparib at 37 °C in 5% CO 2 for 6 h. The GCs were washed with PBS after culturing for 6 h and were used for real-time RT-PCR analyses of transcript levels.
In vitro fertilization (IVF). Animals (n = 10) were administered olaparib or control fluid and were then treated i.p. with human chorionic gonadotropin (hCG) ( Statistics. The data are expressed as mean ± standard error (hormone assay) and as mean ± standard deviation (others). Statistical analyses were performed with the Steel-Dwass test or Mann-Whitney U test as appropriate, and a P value of less than 0.05 was considered statistically significant. The JMP Pro12 package program was used for statistical analyses.