BRE modulates granulosa cell death to affect ovarian follicle development and atresia in the mouse

The BRE (brain and reproductive expression) gene, highly expressed in nervous and reproductive system organs, plays an important role in modulating DNA damage repair under stress response and pathological conditions. Folliculogenesis, a process that ovarian follicle develops into maturation, is closely associated with the interaction between somatic granulosa cell and oocyte. However, the regulatory role of BRE in follicular development remains undetermined. In this context, we found that BRE is normally expressed in the oocytes and granulosa cells from the primordial follicle stage. There was a reduction in follicles number of BRE mutant (BRE−/−) mice. It was attributed to increase the follicular atresia in ovaries, as a result of retarded follicular development. We established that cell proliferation was inhibited, while apoptosis was markedly increased in the granulosa cells in the absence of BRE. In addition, expressions of γ-H2AX (marker for showing DNA double-strand breaks) and DNA damage-relevant genes are both upregulated in BRE−/− mice. In sum, these results suggest that the absence of BRE, deficiency in DNA damage repair, causes increased apoptosis in granulosa cells, which in turn induces follicular atresia in BRE−/− mice.

BRE is expressed in a various tissues that include the adrenal glands, brain, heart, kidneys, testis and ovaries. 1,2 The highest level of expression is in the nervous and reproductive systems, hence the gene was named BRE (brain and reproductive expressed). BRE is now considered to be an adapter protein or a homeostatic protein, which plays a role in stress response and DNA damage repair by some yet unknown mechanisms. 3 It has been reported that homeostasis is archived by modulating the action of hormones and cytokines in physiological and various pathological conditions (including inflammation, infection and cancers). 4 In addition, BRE also acts as the TNFRSF1A modulator, as it can modulate TNF signaling by directly binding to TNFR-1. 5 We have recently reported that BRE plays a vital role in controlling stem cell differentiation by maintaining stemness and also neurogenesis and somitogenesis during embryo gastrulation. 6,7 However, little is known of the physiological and pathological function of BRE in the reproductive system, in which normally express high levels of BRE. We could cause the lack of obvious reproductive defects in BRE knockout mice for the delay of investigating BRE functions on reproductive biology. Apparently, more elaborative studies on reproductive system are indispensable to elucidate whether or not BRE gene have functions in reproductive activities. Hence, we generated BRE mutant mice and carefully examined the development of ovarian follicles in these mice to elucidate how the absence of BRE affects crucial events during folliculogenesis.
Ova are the female reproductive cells that reside and develop within the ovaries, a pair of ductless female reproductive glands. The ovarian follicle, as the functional unit of the ovary, is morphologically composed of an oocyte surrounded by granulosa and theca cells. These cells protect and support the development of the oocytes. Given the appropriate hormonal environment, primordial follicles are induced to develop through the primary, secondary and mature follicular stages. However, most of follicles normally degenerate to be atretic follicles, which could occur in every stages of follicular development. 8 At birth, the ovary contains approximately one million hibernating primordial follicles and some of them become activated to undergo folliculogenesis during puberty. The various developmental stages that the activated primordial follicles pass through during folliculogenesis are also shared by many animals.
Ovarian follicle development is precisely regulated by a sequence of autocrine and paracrine factors. In addition, with input from endocrine hormones that includes pituitary and ovarian hormones. It is especially the balance of these hormones that determines whether a developing follicle becomes maturated or undergoes atresia. [9][10][11] Among these hormones, follicle-stimulating hormone (FSH) is the most important, playing a role in the survival of early antral-staged follicles and the growth, activation and differentiation of prenatal follicles. 12,13 The cellular and molecular mechanisms that determine the developmental fate of ovarian follicles is still poorly understood. 14 There is now accumulating evidences that indicate the death of follicular granulosa cells is partly responsible for causing follicular atresia. 15,16 Granulosa cells could become apoptotic by interfering with steroidogenesis and the addition of dexamethasone, which in turn trigger follicular atresia. In contrast, insulin-like growth factor (IGF) could protect the granulosa cells from apoptosis, induced by dexamethasone, which in turn represses follicular atresia. 17 In this context, we have investigated whether BRE is involved in regulating follicular development and atresia-through its effect on granulosa cell survival. We systematically examined the development of the ovarian follicles in BRE mutant mice and specially focused on the correlation between follicular atresia and granulosa cell growth and death.
We next examined the BRE express pattern in the ovaries using immunofluorescent staining. The result revealed that BRE was expressed in the oocytes of primordial and primary BRE +/+ follicles (Figures 1d and d1), and the granulosa cells of secondary and antral follicles (Figures 1d2 and d3). The expression pattern in BRE +/ − follicles was similar to BRE +/+ follicles (Figures 1e and e1). Not unexpectedly, BRE was not expressed in BRE − / − ovaries, apart from autofluorescence from red blood cells (Figure 1f). The BRE expression pattern suggests that the gene might play a role in ovarian follicle development.
BRE mutation reduces number of follicles in the ovaries. The numbers and types of growing follicles were estimated  (Figures 2a-f). The results revealed that the numbers of primordial, primary, secondary and antral follicles were significantly decreased in 17-week BRE +/ − and BRE − / − ovaries as compared with BRE +/+ ovaries. However, there was no significant difference in the number of corpus luteum found between the three groups. It was estimated that in 17-week-old ovaries, the average number of follicles were significantly reduced in BRE − / − mice compared with BRE +/+ mice (Po0.05; for primordial-secondary follicles: in BRE +/+ = 6.4 ± 0.7, BRE +/ − = 5.5 ± 0.  Figure 2h).
Seventeen-and forty-week-old ovarian sections produced from BRE +/+ , BRE +/ − and BRE − / − mice were also immunofluorescently stained for FasL (Figures 6a-f) and Fas  Figures 6i-n). These two proteins play important roles in regulating apoptosis. The stained sections were analyzed to determine the ratio of FasL + and Fas + area (arrows) over the total area of granulosa cells. Statistical analysis revealed that the ratios for both FasL + and Fas + were significantly increased in comparison with BRE +/+ mice (FasL + 17 weeks: BRE +/+ = 0.60 ± 0. 16 Figure 7i). These findings suggest that, in the absence of BRE, there was significantly more DNA damage in the granulosa cells.

Discussion
In this study, we first examined and compared the morphology of wide-type and BRE knockout ovaries-as BRE is highly expressed in the reproductive system. We determined that the average weight and size of the BRE mutant ovaries were significantly smaller than wild-type ovaries-even though there was no difference between the overall weights of these mice. This suggests that BRE might play a specific role in ovarian follicle development. BRE immunofluorescent staining revealed that BRE was strongly expressed in the oocyte at primordial follicle stage and then it was reverted to the granulosa cells that surround the oocytes at later stages. The granulosa cells of BRE mutants did not express BRE and suggests that it maybe the potential cause of the smaller ovarian size and weight. Moreover, we carefully assessed the distribution of follicles, at various stages of development, in BRE +/+ , BRE +/ − and BRE − / − ovaries, and determined that there were significantly fewer follicles (at all developmental stages) in the BRE mutant ovaries. Hence, we investigated whether the reduction in the number of follicles was attributed to reduced cell proliferation, enhanced apoptosis or a combination of both.
Folliculogenesis involves the activation of a small number of primordial follicles, which then develop and pass through the primary, secondary, antral and follicle stages. Only a few of these mature follicles are ovulated, whereas the majority normally undergoes atresia in the mouse. For follicles to develop normally to maturity and not undergo atresia, it involves very precise cellular and molecular interactions. 18,19 It has been reported that BMP15 and GDF9 were important for inducing follicular cells to differentiate during follicle development and maturity. Our RT-PCR analysis revealed that both Bmp15 and GDF9 expression was significantly reduced in BRE mutant ovarian tissues. Furthermore, Kitl, Pgr and FGFR (follicles development-relevant genes) expression was also repressed. 20 We propose that the abnormal expression pattern of these genes maybe one of the reason why there were fewer developing follicles in the BRE mutant. However, it is generally recognized that the interaction of autocrine and paracrine effectors, including FSH and LH, ultimately determines the developmental fate of the developing follicles. 18,19 Our PAS and Masson's trichrome staining revealed that there were more atretic follicles in the BRE mutants, which may explain why there were fewer follicles in these mutant ovaries. We tried to establish why there were more atretic follicles by focusing on the granulosa cells. These cells are normally indispensable for inducing and supporting the development of the follicles. 21,22 In this context, we first examine how the absence of BRE affected granulosa cell proliferation. Immunofluorescent staining was performed on an ovarian section using PCNA and pHIS3 antibodies. We found that there were significantly fewer PCNA + granulosa cells in 17-and 40-week-old BRE − / − follicles than BRE +/+ follicles. We also found that BrdU + granulosa cells were markedly decreased in 17-week-old BRE − / − antral follicles compared with the BRE +/+ group. Nevertheless, there was no significant difference in the numbers of pHIS3 + cells between the two groups. Using flow cytometry of PI staining, the cell cycle profile of COV434 cells following BRE-siRNA transfection was revealed, which confirmed that the proportion of cells during S phase was reduced after 72-h transfection. The discordance between PCNA, BrdU, pHIS3 expressions and cell cycle analysis suggested that BRE mainly plays a role in the G1-S phase of the cell cycle, which was confirmed by the experimental results of Kim et al. 21 Cyclins are a family of proteins that control the cell progression through the cell cycle. 22 Transitions between the different phases of the cell cycle are governed by positive (cyclins and cyclin-dependent kinases (CDK)) and negative (CDK inhibitors) cell cycle regulatory proteins. 23 CyclinE1, a functional complex as a subunit of CDK2, is required for the G1/S cell cycle transition. Our experiments suggest that silencing BRE can activate the expression of CDK inhibitors p21 and p27, which can block the cell cycle through inhibiting of CyclinE1.
Undoubtedly, the extent of granulosa cell proliferation and apoptosis affects the number of viable follicles that develop to the antral stage. 21 Increased granulosa cell death is most likely to be the cellular mechanism that directly or indirectly induces follicle atresia. 24 Quirk et al. have reported that IGF-I and estradiol could promote bovine granulosa cell proliferation and survival because of their increased resistance to apoptosis. 25 It has been demonstrated that cell cycle arrest and excessive apoptosis of granulosa cells during follicle development could be induced by a high-fat diet. Similarly, we have demonstrated that granulosa cell death (indicated by c-Capase3 labeling) significantly increased in 40-week-old ovaries in the absence of BRE. This implies that BRE conferred an increase in resistance to apoptosis in granulosa cells under normal physiological conditions. Moreover, we have found high level of Fas and FasL expression in BRE − / − mouse ovaries indicating that Fas-FasL signaling has been activated to induce apoptosis in the granulosa cells. The significance of Fas-FasL signaling in granulosa cells has already been confirmed. Next, we asked the question why granulosa cell death was increased in the absence of BRE expression. It has been reported that BRE is mainly involved in DNA damage repair and stress response. Therefore, we decided to determine the presence of γ-H2AX foci, a marker of DNA double-strand breaks, 26 in BRE − / − and BRE +/+ ovaries. We established that there were significantly more γ-H2AX cells in BRE − / − than in BRE +/+ follicles. Moreover, we also revealed that ATM, PUMA, Fas and p53 (DNA damage repair-relevant genes) 27 expression was significantly upregulated in BRE − / − ovarian tissues. These findings implied that increased granulosa cell death in BRE mutant follicles maybe attributed to an excessive accumulation of DNA damage-as BRE is not there to assist in DNA repair.
We have schematically illustrated in Figure 8 how we believe BRE influences follicle development and atresia. Briefly, the illustration shows that BRE is mainly expressed in oocyte of primordial follicles, and then it is the specifically expressed granulosa cells that surround the primary, secondary and antral follicles. The expression pattern indicates that BRE might exert its role in folliculogenesis and follicular atresia via the granulosa cells. Moreover, when granulosa cells cannot express BRE, it would lead to growth arrest at G1 and G2/M phase, and enhances excess apoptosis (as cells are less efficient in conducting DNA repair). These events in turn induce the follicles to undergo atresia. Nevertheless, there are still more experiments that need to be conducted before we can completely understand how BRE functions in the female reproductive biology.

Materials and Methods
Mice. BRE-wild-type (Bre +/+ ), -heterozygote (Bre +/ − ) and -knockout (Bre − / − ) mice were obtained from the Chinese University of Hong Kong Animal Centre and maintained at 25°C on a 12 h light/dark cycle. Bre − / − mutant mice were generated based on the Cre/LoxP recombination against a C57/BL/6J background. The BRE targeted strain (B6Dnk; B6N-Bre tm1a (EUCOMM)Wtsi/H ), in which the exon 3 of BRE gene is flanked by two loxP sites, was purchased from European Conditional Mouse Mutagenesis Programme (EUCOMM). TNAP-Cre mice (129-Alpl tm1 (cre) Nagy /J, stock number: 008569, Jackson Laboratory, Bar Harbor, ME, USA), which are primordial germ cell-specific transgenic mice, were used to cross with female BRE fx/fx mice to generate BRE − / − mice. All of the mice were maintained under a 12 light/12 dark cycle at a constant temperature of~23°C and humidity between 35 and 75%. All animal procedures were approved by AEEC (Animal Experimentation Ethics Committee) of Chinese University of Hong Kong and Hong Kong Government Department of Health. The animal experiments were conducted in accordance with the approved guidelines. 28 Cell culture and gene transfection. COV434 (human ovarian granulosa cells) was attained from GuangZhou Jennio Biotech Co., Ltd, China (Guangzhou, Guangdong, China). The cells were cultured in a humidified incubator with 5% CO 2 at 37°C in six-well plates (1 × 10 5 cells per ml) containing HAM'S/F-12 (Myclone, Logan, UT, USA) supplemented with 10% fetal bovine serum (Gibco, Gaithersburg, MD, USA). For the gene transfection, the COV434 cells were transfected by Control-siRNA (5′-AAGCCUCGAAAUAUCUCCU-3′) or BRE-siRNA (5′-CTGGACTGGTG AATTTTCA-3′), with the help of lipofectamin 3000 (Invitrogen, Carlsbad, CA, USA). 4 Cells were plated to 50-70% confluence at the time of transfection and the preparation of siRNA-lipid complexes, which were subsequently added to the cells. Histology. Briefly, 17-and 40-week-old Bre +/+ , Bre +/ − and Bre − / − ovaries were fixed in 4% paraformaldehyde at 4°C for 24 h. The specimens were then dehydrated, cleared in xylene and embedded in paraffin wax. The embedded specimens were serially sectioned at 5 μm using a rotary microtome (Leica, Frankfurt, Germany). The sections were either stained with hematoxylin and eosin, PAS reaction or Masson's trichrome dyes. 29 The sections were also immunohistochemically stained. The PAS and Masson staining were used to reveal the presence of atretic follicle in the ovarian sections. The stained histological sections were photographed using an epifluorescence microscope and an attached camera (Olympus IX51, Tokyo, Japan; Leica DM 4000B, Frankfurt, Germany) at × 200 magnification.
Classification of developing follicles in ovarian sections. The follicles in the ovarian histological sections were developmentally staged according to their morphology as: primordial, primary, secondary, antral or atretic follicles. Briefly, an oocyte surrounded by a single layer of squamous granulosa cells was classified as a primordial follicle. Oocyte surrounded by a single or several layer/s of cuboidal granulosa cells was classified as a primary or secondary follicle, respectively. When an antrum is present, it was described as an antral follicle. The presence of zona pellucida remnants was classified as an end-stage atretic follicle. 30 Every fifth and sixth histological sections were selected for comparison and evaluation. Follicles were only counted if they appeared in one histological section but not in the other. 30 Immunohistological staining. Sections of mouse ovaries were dewaxed, hydrated, incubated in citrate buffer (pH 6.0) and then heated in a microwave for antigen retrieval. Immunofluorescent staining was conducted on these treated sections using various antibodies. Briefly, the sections were incubated in the following primary antibodies diluted using PBT-NGS: BRE  show γ-H2AX staining. Daminobenzidine tetrahydrochloride substrate (DAB kit, MXB, Fuzhou, China) was used to visualize the immunostaining. Photographs were taken of the stained histological sections using an epifluorescence microscope (Olympus IX51, Leica DM 4000B) at × 200 magnification.
In vivo BrdU labeling. BRE-wild-type (Bre +/+ ) and -knockout (Bre − / − ) mice were injected with BrdU (100 mg/kg of body weight) at 10 mg/ml in PBS intraperitoneally for 2 h before killing. The ovaries were then extracted and fixed in 4% paraformaldehyde at 4°C for 16 h. For histology, the fixed ovaries were washed, dehydrated and embedded in paraffin wax. The paraffin-embedded ovaries were serially sectioned at 4 μm. The sections were immunohistologically stained using rat anti-BrdU antibody (1:200; Abcam) and counterstained with hematoxylin for morphological observation. The extent of follicular cell proliferation within follicles was determined by the presence of BrdU + in randomly selected ovarian sections.
RNA isolation and PCR analysis. Total RNA was isolated from 17-and 40-week-old Bre +/+ , Bre +/ − and Bre − / − ovaries or COV434 cells using Trizol (Invitrogen) according to the manufacturer's instructions. Three ovaries from each group were used. First-strand cDNA was synthesized at a final volume of 25 μl using a SuperScript III First-Strand kit (Invitrogen). Following reverse transcription, RT-PCR amplification of the cDNA was performed using specific primers (Supplementary Figure 1). PCR was performed in a Bio-Rad S1000 Thermal cycler (Bio-Rad, Richmond, CA, USA). The cDNAs were amplified for 30 cycles. One round of amplification was performed at 98°C for 10 s, at 60°C for 15 s and then at 72°C for 30 s (Takara, Tokyo, Japan). The PCR products (20 μl) were resolved on a 2% agarose gels (Biowest, Madrid, Spain) in 1 × TAE buffer (0.04 M trisacetate and 0.001 M EDTA) plus GeneGreen Nucleic Acid Dye (TIANGEN, Beijing, China). The reaction products were visualized using a transilluminator (Syngene, Cambridge, UK) and a computer-assisted gel documentation system (Syngene). qPCR analysis was also performed by SYBR Premix Ex Tag (Takara) using a 7900HT Fast Real-Time PCR system (Applied Biosystems, Foster City, CA, USA). The primer sets used for the qPCR are provided in Supplementary Figure 2. Each of these experiments was replicated at least three times. The RT-qPCR results were produced from four independent sets of experiments. The housekeeping gene, β-actin, was run in parallel to confirm that equal amounts of RNA were used in each reaction. The ratio intensity for the fluorescently stained bands of genes of interest, and β-actin was calculated and normalized to quantify the level of gene expression.
Image acquisition and analysis. Whole ovaries were photographed using a fluorescence stereomicroscope (Olympus MVX10) and analyzed imaging software (Image-Pro Plus 6.0 Media Cybernetics, Media Cybernetics, MD, USA). Sections of the stained ovaries were photographed using an epi-fluorescent microscope (Olympus IX51, Leica DM 4000B) at × 200 and × 400 magnification, and analyzed using an Olympus software (Leica CW4000 FISH).
For quantification of proliferation, apoptosis and DNA damage, we counted the number of PCNA + , pHIS3 + , C-Caspase3 + and γ-H2AX + granulosa cells versus total DAPI + granulosa cells for each antral follicle in 17-and 40-week-old mouse ovaries (Bre +/+ , Bre +/ − and Bre − / − ). The results were then compared between each group with the follicles only at the same developmental stage. For immunofluorescent staining of 17-40-week-old ovaries, total Fas + , FasL + , CD34 + and α-SMA + granulosa or theca cells in antral follicles were counted. 31 Statistical analysis was performed using a SPSS 19.0 software (SPSS software, Armonk, NY, USA), and the data were presented as mean ± S.E.M. Six ovaries of each group (Bre +/+ , Bre +/ − and Bre − / − ) were used.
Data analysis. Data analyses and construction of statistical charts were performed using GraphPad Prism 5 software (GraphPad Software, La Jolla, CA, USA). The results were presented as the mean value (x ̅ ± S.E.M.). Statistical analysis was performed using IBM SPSS Statistics 19.0 software (IBM SPSS Statistics software, Armonk, NY, USA). Statistical significance was determined using an independent sample's t-test, and non-parametric independent samples Kruskal-Wallis test. Po0.05 was considered to be statistically significant.

Conflict of Interest
The authors declare no conflict of interest.