Rictor/mTORC2 Pathway in Oocytes Regulates Folliculogenesis, and Its Inactivation Causes Premature Ovarian Failure*

Background: The roles of Rictor/mTORC2 in folliculogenesis and follicle survival are unknown. Results: Loss of Rictor in oocytes causes excessive follicular atresia and the mutant mice demonstrate progressive POF phenotype. Conclusion: Rictor/mTORC2 plays key roles in folliculogenesis, follicle survival, and female fertility. Significance: This study establishes a novel function of mTORC2 in folliculogenesis and a potential link between mTORC2 and POF. Molecular basis of ovarian folliculogenesis and etiopathogenesis of premature ovarian failure (POF), a common cause of infertility in women, are not fully understood. Mechanistic target of rapamycin complex 2 (mTORC2) is emerging as a central regulator of cell metabolism, proliferation, and survival. However, its role in folliculogenesis and POF has not been reported. Here, we showed that the signaling activity of mTORC2 is inhibited in a 4-vinylcyclohexene diepoxide (VCD)-induced POF mouse model. Notably, mice with oocyte-specific ablation of Rictor, a key component of mTORC2, demonstrated POF phenotypes, including massive follicular death, excessive loss of functional ovarian follicles, abnormal gonadal hormone secretion, and consequently, secondary subfertility in conditional knock-out (cKO) mice. Furthermore, reduced levels of Ser-473-phosphorylated Akt and Ser-253-phosphorylated Foxo3a and elevated pro-apoptotic proteins, Bad, Bax, and cleaved poly ADP-ribose polymerase (PARP), were observed in cKO mice, replicating the signaling alterations in 4-VCD-treated ovaries. These results indicate a critical role of the Rictor/mTORC2/Akt/Foxo3a pro-survival signaling axis in folliculogenesis. Interestingly, loss of maternal Rictor did not cause obvious developmental defects in embryos or placentas from cKO mice, suggesting that maternal Rictor is dispensable for preimplantation embryonic development. Our results collectively indicate key roles of Rictor/mTORC2 in folliculogenesis, follicle survival, and female fertility and support the utility of oocyte-specific Rictor knock-out mice as a novel model for POF.

The reproductive lifespan of women and most other mammals is ultimately dependent on the size of the ovarian primordial follicle pool, which is generally established during embryonic development or after birth (1,2). To produce mature oocytes for fertilization, primordial follicles are recruited from the reservoir of dormant follicles into the growing follicle pool through a process termed follicular activation, and subsequently undergo a series of development steps. During this entire process, a large number of follicles undergo apoptotic death (atresia) if not selected for further growth (3)(4)(5). Menopause or ovarian failure occurs when the primordial follicle pool is exhausted. Premature ovarian failure (POF), 4 also known as premature menopause, is characterized by cessation of ovarian function, amenorrhea, hypoestrogenism, and elevated gonadotropin levels in women under 40 years of age, affecting 1-2 and 0.1% of women under the ages of 40 and 30, respectively (6). A disastrous consequence of POF is loss of fertility, which is due to the absence of follicles in most cases, and in others, the inability of the remaining follicles to respond to hormonal stimulation (7). Nevertheless, the detailed etiology of POF remains to be established, and the factors and mechanisms involved in regulating the processes of primordial follicle * This study was supported by The State Key Development Program for Basic activation, follicle growth, and follicle survival or atresia are not completely understood at present.
PI3K signaling is significant in folliculogenesis processes, including follicle activation, development, and survival (16 -19). As a downstream target of PI3K, mTORC1 is involved in the regulation of folliculogenesis. Stimulation of mTORC1 leads to overactivation of the entire pool of primordial follicles, causing POF and infertility (20 -22). However, the roles of Rictor/mTORC2 in folliculogenesis are currently unknown. In this study, we demonstrated that Rictor activity and downstream mTORC2 signaling are suppressed in ovaries treated with the ovotoxicant, 4-vinylcyclohexene diepoxide (VCD), an occupational chemical shown to cause POF in human and animal models by accelerating the apoptotic process of atresia (23)(24)(25)(26)(27). Additionally, induction of oocyte-specific ablation of Rictor with the Cre-loxp system using Zp3-Cre and Rictor loxp/loxp mice resulted in POF phenotypes. Our findings support a critical role of Rictor/mTORC2 in follicle survival and oogenesis.

EXPERIMENTAL PROCEDURES
Mice, Husbandry, and Genotyping-Zp3-Cre mice (Jax number 003650) were obtained from the Model Animal Research Center of Nanjing University (Nanjing, China). Rictor loxp mice were kindly provided by Professor Mark A. Magnuson (Vanderbilt University). For oocyte-specific knock-out experiments, males (but not females) carrying the Cre transgene were used for breeding to prevent universal knock-out of target genes in the whole body (28). Male Zp3-Cre mice were mated with female homozygous Rictor loxp (Rictor loxp/loxp ) mice to yield males heterozygous for loxp Rictor and Zp3-Cre, which were then bred with female mice homozygous for loxp Rictor to obtain female mice homozygous for loxp Rictor and heterozygous for Zp3-Cre. These mice displayed oocyte-specific deletion of Rictor (Zp3-Cre ϩ , Rictor loxp/loxp ) and were therefore designated conditional knock-out (cKO) mice. Female mice homozygous for loxp Rictor without Zp3-Cre (Zp3-Cre Ϫ , Rictor loxp/loxp ) from the same litter were taken as control mice. Genotyping involved the use of the following primers: Zp3-F, 5Ј-GCGGTCTGGCAGTAAAAACTATC-3Ј, Zp3-R, 5Ј-GTG-AAACAGCATTGCTGTCACTT-3Ј; Rictor-F, 5Ј-GAAGT-TATTCAGATGGCCCAGC-3Ј; Rictor-R, 5Ј-ACTGAATAT-GTTCATGGTTGTG-3Ј. DNA extraction, PCR amplification, and agarose electrophoresis were performed according to the instructions of The Jackson Laboratory. All animal experiments were approved by the Southern Medical University Committee on the Use and Care of Animals and performed in accordance with the Committee's guidelines and regulations.
Western Blot Analysis-After euthanasia, ovaries were immediately removed, subsequently triturated and lysed on ice, and finally boiled in SDS loading buffer. Protein extracts were subjected to 6 -12% SDS-PAGE and subsequent processes following the standard protocol. Apart from the primary antibodies specified for immunohistochemistry analysis, those used for Western blot were as follows: anti-p-S6K (Thr-389), anti-S6K, anti-cleaved PARP, and anti-p-NDRG1 (Thr-346) (Cell Signaling Technology); anti-Akt and anti-␣-Tubulin (Santa Cruz Biotechnology); and anti-Foxo3a (Proteintech) (Chicago, IL). An ECL kit (PerkinElmer) was used for visualization.
Serum Hormone Measurement-After anesthesia, mid-line dermatotomy of the chest was performed, and blood was drawn via cardiocentesis. After 10 min of centrifugation at 3,000 ϫ g, serum was collected and stored at Ϫ20°C until use. Plasma FSH, luteinizing hormone (LH), estrogen hormone (E 2 ), and progesterone levels were measured using ELISA (Nanjing Jiancheng Bioengineering, Nanjing, China).
Fertility Assay and Embryonic Development Assessment-To evaluate the fertility of female mice, 4-or 6-month-old control or cKO mice were mated with wild-type male C57 mice with proven fertility at a proportion of 1:1. Successful conception was defined by the presence of vaginal plug and subsequent visibly growing abdomen, and pregnant female mice were separated and monitored. The number of pups from each litter was counted, each pup was weighed, and physical status was examined. The duration from the day of mating to the day of delivery was defined as a childbirth cycle. Female mice that did not conceive within 2 months of mating were defined as infertile. For embryonic development assessment, female control or cKO mice with vaginal plug were designated E0.5 and killed at E5.0, E10, and E15. Embryos and placenta were removed, counted, and weighed. E15 placentas were fixed in 4% paraformaldehyde, Rictor/mTORC2 Regulates Folliculogenesis processed using paraffin wax, and sectioned for histological analysis.
VCD Treatment-Female wild-type C57 mice (4 weeks old) were administered daily intraperitoneal injections of either sesame oil (2.5 ml/kg, vehicle control) or VCD (Sigma) dissolved in sesame oil (80 mg/kg) for 15 days, as described previously (25). At 12 h following the final dose, mice were killed. One ovary of each mouse was triturated, lysed, and boiled in SDS loading buffer for Western blot analysis. The other ovary was fixed in 4% paraformaldehyde and processed using paraffin wax and standard methods. Histological analysis was performed as described above.
Statistical Analysis-All experiments were carried out in duplicate. Data were expressed as mean values Ϯ S.E., and differences between groups were analyzed using Student's t test (SPSS 13.0) if data violated normal distribution via nonparametric Mann-Whitney test. p Ͻ 0.05 was considered statistically significant. For Western blot analysis, one representative set of data is shown.

4-VCD Destroys Early-stage Follicles and Suppresses the Rictor/mTORC2
Signaling Pathway-To assess the role of mTORC2 in POF, the activity of Rictor, a central component of the mTORC2 signaling pathway, was finely monitored in 4-VCD-treated ovaries. As expected, 15 days of VCD treatment reduced the number of primordial follicles by 45% and primary follicles by 55% that of control mice, respectively, with a weak toxic effect on follicles of higher grade (Fig. 1, A and B). Interestingly, the level of phospho-Rictor (p-Rictor, Thr-1135), an inhibitory modifier of Rictor, was up-regulated in VCD-treated ovaries (Fig. 1C). Rictor is a core component of mTORC2 that promotes cell survival through phosphorylating and activating Akt in various contexts, in turn, phosphorylating and inhibiting Foxo3a, a pro-apoptotic transcription factor (9, 15). As expected, expression of p-Akt (Ser-473) and p-Foxo3a (Ser-253) was reduced in VCD-treated ovaries (Fig. 1C). Simultaneously, the levels of Bad, Bax, and cleaved PARP, three typical proapoptotic proteins, were elevated ( Fig. 1, C and D). Serum-and glucocorticoid-regulated kinase 1 (SGK1), another mTORC2 substrate, was not affected because the phosphorylation of N-Myc downstream-regulated 1 (NDRG1) at Thr-346, a readout for SGK1 activity, was comparable between the control and VCD-treated mice (Fig. 1D). In addition, we observed an increase in Thr-389-phosphorylated S6K1 (Fig. 1D). These results provide in vivo evidence supporting the previous finding that S6K1 mediates direct phosphorylation and inhibition Rictor at Thr-1135, acting as a novel mTORC1-dependent inhibitory feedback loop on mTORC2 (30 -32). Thus, phosphorylation and inactivation of Rictor appear to be significantly involved in VCD-induced follicular apoptosis (Fig. 1E).
Oocyte-specific Deletion of Rictor Replicates the Effect of VCD on Oocytes-To identify the role of Rictor/mTORC2 in follicle survival, we generated mice with oocyte-specific deletion of Rictor using the Cre-loxp system (Fig. 2). As expected, Rictor expression was observed in ovarian follicles of control littermates, but not those of cKO mice (Fig. 3A), suggesting complete Cre-mediated recombination of the floxed Rictor gene. A decrease in p-Akt (Ser-473) was evident, attributable to the loss of Rictor (Fig. 3B). Interestingly, we observed specific reduction of p-S6 (Ser-235/Ser-236), a downstream signal/target of mTORC1, particularly in the corpus luteum (Fig. 3B). This result was in line with the previous finding that mTORC1 is activated by mTORC2 via inhibition of its negative regulator, tuberous sclerosis complex 1/2 (TSC1/2), through Akt (9, 10).
Next, we monitored the expression of downstream signals/ targets of Rictor/mTORC2. Loss of Rictor led to decreased p-Foxo3a (Ser-253) and increased Bad, Bax, and cleaved PARP levels (Fig. 3, C and D). These changes in signaling molecules were confirmed via Western blot analysis (Fig. 3, D and E). Thus, oocyte-specific ablation of Rictor duplicates the effects of VCD on the underlying signaling pathway controlling follicular survival in the ovary. Our results collectively reveal a central role of Rictor in the signaling axis of mTORC2/Akt/Foxo3a/ pro-apoptotic proteins for regulation of follicular survival.
Loss of Rictor in Oocytes Causes Progressively Extensive Cell Apoptosis and Follicle Loss-Next, we focused on the follicle number in cKO mice. cKO mice developed normally but had smaller ovaries than control mice, with a ovary weight decreased to 45% of the control mice at the age of 8 months (Fig.  4A). Consistent with enhanced expression of pro-apoptotic proteins, significantly fewer healthy follicles and less corpus luteum and more atretic follicles were morphologically observed in the ovaries of cKO mice. Notably, follicles were rarely found in cKO mice at the age of 8 months (Fig. 4B). To precisely assess follicle loss caused by Rictor deletion, the number of follicles at each stage of five sections from each ovary was classified and counted. When compared with control mice, we observed 40% loss of healthy follicles and 50% decrease in corpus luteum in 6-month-old cKO mice. In 8-month-old cKO mice, a 67% loss of healthy follicles and a 71% decrease in corpus luteum were observed (Fig. 4C), suggesting a progressive follicle loss after Rictor deletion in oocytes. Corpus luteum develops from follicular granulosa cells after ovulation (33). Less corpus luteum indicates lower levels of ovulation or availability of a smaller dominant follicle reserve for ovulation, which may be attributed to excessive follicle loss or arrest of development from small to mature follicles. In this regard, it is reasonable to believe that excessive follicle loss is one of the critical causes because more atretic follicles were detected in cKO mice (Fig.  4C). To further determine at which follicular stage follicle loss occurs, the number of follicles at each stage was analyzed. As shown in Fig. 4D, although cKO mice had fewer small follicles (including primordial and primary follicles) than control mice, this difference was not statistically significant. It has been shown that ZP3-Cre-mediated gene deletion occurs at the primary and/or later follicular stages (34), and thus deprivation of Rictor in oocytes did not cause significant primordial follicle loss. Indeed, massive follicle loss in cKO mice occurred at large and mature follicles (Fig. 4D). Consistently, acceleration of cell apoptosis was observed in large and mature follicles in 8-month-old cKO mice (Fig. 4, E and F), in line with the result of elevated pro-apoptotic protein levels in cKO mice. These results indicate that deletion of Rictor in oocytes enhanced cell apoptosis and follicle loss, and thus decreased follicular reserve available for ovulation. MARCH

Depletion of Rictor in Oocytes Causes Subfertility in cKO
Mice-In general, diminished follicle reserve routinely results in reduced female fertility. To further evaluate the fertility of cKO mice, 4-, 6-, and 8-month-old female control and cKO mice were mated with wild-type C57BL/6J males with proven fertility. As expected, progressive loss of fertility was observed in cKO mice. All control mice successfully conceived and delivered their pups, whereas one cKO mouse (11%) from both the 4-month-old and the 6-month-old groups did not conceive within the 2-month observation period. Of note, all 8-monthold cKO mice were sterile (Table 1). Interestingly, one mouse (11%) from the 4-month-old group and four (44%) from the 6-month-old group died of dystocia (Table 1), characterized by aberrantly long gestation duration and delivery of dead pups.
Consequently, the cKO group had a significantly lower number of live pups than the control group (Table 1). Even for the cKO mice displaying successful parturition, the frequency of delivery was decreased (Table 1). In summary, cKO mice exhibited a typical age-dependent decline in reproductive fitness and became sterile at the age of 8 months.
Abnormal Hormone Secretion in cKO Mice-As a reproductive organ, the ovary is largely regulated by gonadal hormones (FSH and LH), and in turn, produces other gonadal hormones (E 2 and P 4 ). Accordingly, we assessed the hormone parameters in cKO mice. Both FSH and LH levels in cKO mice were higher, although not to a significant extent, when compared with those in control mice (Fig. 5, A and B). Considering the existence of follicular development arrest and deficient ovulation, these changes may represent certain positive feedback responses because FSH is responsible for follicular development and LH is responsible for ovulation. In contrast, both estradiol and progesterone secretion were significantly decreased in cKO mice (Fig. 5C), especially progesterone with a reduction of almost 50% that of the control level (Fig. 5D). As an important secretory source for estradiol and the exclusive source for progesterone (33), less functional corpus luteum in cKO mice consequently produced insufficient hormones, partly leading to subfertility in cKO mice.
Maternal Rictor Is Dispensable for Embryonic Development-Because recombination mediated by Zp3-Cre specifically occurs in oocytes, it is not only an excellent tool to understand the physiological roles of target genes during folliculogenesis and oogenesis, but also widely used to explore maternal gene effects during embryonic development (34 -37). Notably, a recent study on multiallelic disruption of the Rictor gene revealed that mTORC2 is essential for fetal growth and viability (38). To further determine whether maternal Rictor is necessary for embryonic development, sexually mature female cKO (Zp3-Cre ϩ , Rictor loxp/loxp ; loss of Rictor in ovulated oocytes) and control (Zp3-Cre Ϫ , Rictor loxp/loxp ; normal ovulated oocytes) mice were crossed with wild-type C57BL/6J males with proven fertility. In this mating scheme, embryos derived from the oocytes of cKO mice were devoid of maternal Rictor, giving rise to embryos with the genotype Ϫ/Rictor. However, we failed to observe visible physical or histological differences between embryos of control and cKO mice (Fig. 5). Although cKO mice had lower numbers of embryos at each scheduled time point (E5, E10, or E15), the total number of embryos within either the control or the cKO group was relatively stable throughout the duration of observation (E5 to E15) (Fig. 6A), indicating that maternal Rictor deficiency is acceptable for normal embryonic development. This assumption was supported by the finding that the average mass of embryos at each time point is comparable between control and cKO mice (Fig. 6B). In addition, his-   tological data revealed no obvious morphological or structural abnormalities in E16 placentas from cKO mice (Fig. 6C), and placental lactogen 1 (PL-1), a marker for trophoblast giant cells, exhibited similar expression and distribution patterns to those of control mice (Fig. 6D). Although cKO mice were susceptible to delivery of dead pups, no physically visible developmental

Rictor/mTORC2 Regulates Folliculogenesis
defects were observed in the pups (Fig. 6E). However, specific histological differences were evident between the uteri of control and cKO mice, with smaller cavities and fewer secretory glands in cKO mice (Fig. 6F), which possibly led to abortion or dystocia. These results support the theory that paternal Rictor alone is sufficient for embryonic development.

DISCUSSION
In this study, using a POF mouse model induced by the ovotoxicant, 4-VCD, and a mutant mouse model with oocyte-specific deletion of Rictor, we showed that Rictor/mTORC2 functions in oocytes to protect follicles from apoptosis. Upon inhibition of Rictor in oocytes, follicular atresia was accelerated and follicle loss was enhanced. We additionally identified an underlying Rictor/mTORC2/Akt signaling axis controlling follicle survival during folliculogenesis, which inhibits Foxo3a and induces a decrease in the protein levels of pro-apoptotic Bad, Bax, and cleaved PARP. Interestingly, Rictor-cKO mice demonstrated a distinctive ovarian phenotype of POF, featuring a similar decrease in the follicular reservoir, fertility, and serum estrogen level, along with mild elevation of FSH and LH. Our findings have broad physiological and clinical implications and contribute to in-depth understanding of both normal ovarian physiology and development of ovarian diseases.
In humans, POF is an early ovarian dysfunction clinically defined as the cessation of ovarian function with hypoestrogenism and elevated gonadotropin before or at 40 years of age. This condition is characterized by the presence of primary or secondary amenorrhea for at least 4 months, low serum estrogen, and elevated serum gonadotropin concentrations (e.g. FSHϾ40 IU/liter), and above all, loss of fertility (6,7). A wide spectrum of pathogenic mechanisms lead to the development of POF, including genetic (e.g. X-chromosome abnormalities), autoimmune, metabolic (galactosaemia), infectious (mumps), and iatrogenic (anticancer treatment) causes (39). However, the majority of cases of POF are idiopathic. Studies on POF have been hindered by inaccessibility of the ovary and occurrence of follicular depletion before the onset of symptoms. Thus, several POF models have been generated with mice via genetic manipulation. For example, deletion of Pten from mouse oocytes causing overactivation of PI3K signaling leads to premature activation of the entire pool of primordial follicles (16). Deletion of Tsc1 (20) or Tsc2 (21) in oocytes induces overactivation of mTORC1 signaling and results in global activation of all primordial follicles around the time of puberty, ending with follicular depletion in early adulthood. In contrast, POF of Rictor-cKO mice was caused by excessive follicular atresia, which occurred at every developmental stage, rather than entire overactivation of the primordial follicle pool in the three mutant mice. Furthermore, the cKO mice exhibited progressive loss of follicles and fertility and became infertile at the age of 8 months. The phenotypic features are similar to the natural and time point pathological changes in humans. Our data suggest that increased follicle apoptosis resulting from genetic variations or mutations also results in POF in mice, advancing our understanding of the pathological processes of POF. Indeed, chemo-or radiotherapy-induced follicular apoptosis contributes to many cases of POF (40).
It is hypothesized that resting primordial follicles are under constant inhibitory local influence to remain dormant (41). PTEN (phosphatase and tensin homolog), TSC1, and TSC2 in oocytes are part of the inhibitory mechanisms that maintain quiescence of primordial follicles. Other similar inhibitory molecules include the cyclin-dependent kinase (Cdk) inhibitor, p27 kip1 (p27 or Cdkn1b) (42). In this sense, mTORC1 appears constantly suppressed in oocytes, and selective activation of primordial follicles occurs as a result of selective elevation of mTORC1 activity. In contrast, mTORC2 is believed to be persistently active in oocytes to maintain follicular survival, and apoptosis is enhanced when mTORC2 is inhibited under antagonistic conditions, such as ovotoxicant-induced injury or deficit in survival factors. Taken together, the data indicate that mTORC1 signaling and mTORC2 signaling in oocytes act synergistically to regulate follicular activation, development, and atresia. Further clarification of the mTOR pathway may facilitate the development of improved contraceptives for adjusting follicular activation and survival to an optimal level, thus preserving the follicular reserve pool until fertility is desired.
Rictor/mTORC2 was recently shown to be essential for fetal growth and viability in mice using a multiallelic gene-targeting strategy. Rictor-null mice exhibited placental defects and embryonic lethality (38). However, neither embryonic nor placental defects were observed in embryos from Rictor-cKO mice. This discrepancy led to the assumption that maternal Rictor is not essential for preimplantation embryonic development. To date, it is widely accepted that the first phase of embryonic development is dependent on maternal transcripts and proteins accumulated during oogenesis until broad embryonic genome activation after combination and reprogramming of maternal and paternal genomes (43,44). The finding that embryos from cKO mice develop normally suggests that mater-

Rictor/mTORC2 Regulates Folliculogenesis
nal Rictor is dispensable for oocyte-to-embryo transition or that transcripts controlled by Rictor/mTORC2 accumulated before the primary follicle are sufficient for preimplantation embryonic development because Zp3-Cre has been shown to express Cre recombinase in oocytes at the primary and/or later follicular stages (34).
In summary, our results indicate that maternal Rictor is not required for preimplantation embryonic development. However, disruption of Rictor in oocytes causes early depletion of functional ovarian follicles, aberrant gonadal hormone secretion, and secondary subfertility in cKO mice, reminiscent of POF phenotypes. Based on the collective findings, we conclude that Rictor/mTORC2 plays a critical role in folliculogenesis, follicle survival, and female fertility and that its inactivation in oocytes causes POF.