Formation of ovarian follicles during fetal development
It is well established that in human and nonhuman primates, formation of the pool of primordial follicles available for reproductive function in adulthood occurs
in utero (
Rabinovici and Jaffe 1990;
Pepe et al., 2006). Thus, following differentiation of the bipotential gonad into an ovary very early in gestation (
Simpson and Rajkovic, 1999), presumptive germ cells undergo mitosis (now termed oogonia) and become colonized by proliferating pregranulosa cells (
Byskov, 1986;
Sawyer et al., 2002). Several oogonia then initiate meiosis to form oocytes which arrest in the diplotene/dictyate phase of prophase I of meiosis. At about midgestation, the pregranulosa cells and oocytes begin to contact each other to form somatic cell-germ cell complexes which progressively fuse to form a structure termed a germ cell nest which is encased by a basal lamina. During this period of development, follicles form within the nest and it is presumed that emergence of primordial follicles in the ovarian cortex involves breakdown of the basal lamina of the nest, a process termed germ cell nest breakdown (
Sawyer et al., 2002). Finally, primordial follicles formed
in utero must develop mechanisms/structures to support their survival for up to 40 years (
Pepe et al., 2006). Thus, because the foundation for normal adult reproductive function is established during fetal life (
Rabinovici and Jaffe, 1990;
Pepe et al., 2006), inadequate ovarian development
in utero may impair reproductive potential in the adult. Hence, a more comprehensive understanding of the factors underpinning intrauterine development of the pool of follicles available in adulthood is a prerequisite to managing adult ovarian dysfunction. However, elucidation of fetal ovarian development and its impact on reproductive function and fertility (i.e. programming) requires studies in which invasive experimental paradigms can be performed
in utero and ovarian function assessed postnatally, studies which for ethical reasons cannot be performed in humans. Therefore, although the cellular development of the human fetal ovary has been well-characterized (
Baker, 1963;
van Wagenen and Simpson, 1965), our understanding of the regulation of human fetal ovarian development is limited and based primarily on isolated case reports of genomic defects, morphological characterization of abortuses, serum hormone levels, and
in vitro approaches. Consequently, despite significant advances in adult reproductive medicine, infertility/ovarian dysfunction continues to adversely affect a large segment of the female population (
Crisponi et al., 2001). For example, a significant number of infertile women exhibit ovarian failure manifest as a deficit in ovarian follicles ranging from complete absence and thus lack of pubertal development, to a reduction in the number of germ cells which collectively comprise the clinical syndrome of premature ovarian failure (
Crispoini et al., 2001;
Simpson and Rajkovic, 1999;
Layman et al., 1998;
Beck-Peccoz and Persani, 2006). Therefore, a nonhuman primate model is needed to study fetal ovarian development and to translate this information to the human.
Estrogen receptor expression in fetal ovary
Our laboratories have been instrumental in establishing the baboon as a model for the study of human reproductive endocrinology and as discussed above, showed that estrogen, the maternal and fetal levels of which increase with advancing gestation as in women (
Albrecht and Pepe, 1988), plays a central integrative role in regulating fetal-placental development (
Albrecht and Pepe, 1999). Like the human, the baboon fetal ovary at midgestation is comprised of numerous germ cell nests containing oocytes and pregranulosa cells but very few primordial follicles and by late gestation is comprised almost exclusively of primordial follicles and very few germ cell nests. Moreover, our laboratories were among the first to show that fetal ovarian follicular development in the baboon was associated with expression of the mRNAs and proteins for estrogen receptor (ER) α and ERβ (
Pepe et al., 2002). Thus, ERα protein was localized to nuclei of surface epithelium and pregranulosa cells but not oocytes on day 100 of gestation and in the surface epithelium and in several but not all granulosa cells of primordial follicles on days 165-180 of gestation (term = day 184). ERβ was also localized and abundantly expressed in the nuclei of pregranulosa cells on day 100 and extensively expressed in granulosa cells of follicles on day 165 of gestation. Moreover, oocytes obtained from near term fetal ovaries by laser capture microdissection expressed the mRNA for ERβ but not ERα (
Bocca et al., 2008).
Estrogen regulation of fetal ovarian folliculogenesis
In the adult rat, estrogen regulates several aspects of ovarian function (
Richards, 1980; Billig
et al., 1983) and male and female αβ ER knock-out mice are infertile (
Couse and Korach, 1999). Therefore, we originally proposed (
Pepe et al., 2002) that estrogen regulates fetal ovarian development
in utero. To test this hypothesis, follicle development was examined in baboons treated daily on days 100-165 of gestation with the highly specific aromatase inhibitor, letrozole (CGS 20267; Norvartis Pharm AG, Basel Switzerland), which suppressed maternal and fetal estradiol levels by >95% or with letrozole and estradiol benzoate which restored estrogen levels in the mother and fetus to 90-120% and 10%-30% of normal, respectively (
Albrecht et al., 2000;
Zachos et al., 2002). The incomplete restoration of fetal serum estrogen levels reflects transplacental metabolism of maternally administered estradiol (
Albrecht and Pepe, 1990). In estrogen-suppressed baboons, although fetal ovarian weight and expression of ERα or ERβ were not significantly changed, follicle numbers were reduced by 50% (
Zachos et al., 2002), whereas the number of germ cell nests containing pregranulosa cells/oocytes which failed to form follicles as well as follicles which were not released into the ovarian stroma was increased 3-fold. Moreover, whereas most of the follicles in untreated baboons contained oocytes with an intact cytoplasm and appeared healthy, in estrogen depleted fetuses, the number of intact oocytes was markedly reduced. Thus the majority (>70%) of follicles that formed in estrogen-depleted fetuses contained oocytes in which the cytoplasm exhibited vacuolization (i.e. not intact) and were comprised of swollen and less electron-dense mitochondria.
Exposure of the fetus particularly early in gestation to excess androgen has also been shown to disturb fetal ovarian development (
Steckler et al., 2005;
Manikkam et al., 2006;
Abbott et al., 2006). However, because aromatization is blocked and androgen (i.e. androstenedione and testosterone) levels comparably increased in baboons treated with letrozole alone or with estradiol (
Zachos et al., 2002) and androgen receptor protein is not detected in the near term baboon fetal ovary (Pepe and Albrecht, unpublished), the suppression of follicle development in estrogen deficient baboons reflects the absence of estrogen and not increased androgen (
Pepe et al., 2006). Consistent with our original observations, in aromatase knock-out mice totally devoid of estrogen synthesis, the number of primordial follicles formed was also reduced due to their lack of formation from germ cell nests (
Britt et al., 2004).
It also is unlikely that fetal pituitary gonadotrophins regulate primordial follicle development in the primate. Thus, in the baboon (
Zachos et al., 2003), as in humans (
Reyes et al., 1981; Kaplan and Grumbach, 1976), fetal serum FSH and LH levels increase to a maximum at midgestation (i.e. before follicle formation) and decline steadily thereafter (i.e. during the period of folliculogenesis). Moreover, ovarian folliculogenesis apparently proceeded normally up to week 32 of gestation in human anencephalic fetuses (
Baker and Scrimegour, 1980). Although FSH receptor (FSHR) binding and mRNA are detected in the near-term fetal rhesus monkey (
Huhtaniemi et al., 1987) and baboon (
Zachos et al., 2003), we showed that the reduction in the number of primordial follicles formed in near-term baboon fetuses deprived of estrogen between mid and late gestation was associated with an increase in fetal serum FSH and no change in fetal ovarian FSHR mRNA (
Zachos et al., 2003). Finally, in two women with hypogonadism due to a mutation in the FSH β-subunit gene and thus no detectable FSH, one conceived after induction of ovulation (
Matthews et al., 1993), while the other exhibited increased follicle growth (
Barnes et al., 2002) after exogenous FSH, indicating that follicles had developed
in utero in the absence of FSH (
Layman et al., 1998;
Themmen and Huhtaniemi, 2000). Collectively, our studies provided the first definitive evidence that development of a stockpile of healthy follicles available for adult ovarian function in the primate is an intrauterine estrogen-dependent event.
In the adult, ovarian function and follicle recruitment are regulated by an interplay between pituitary gonadotrophins, estrogen and various intragonadal autocrine/paracrine factors most notably inhibin and activin (
Findlay, 1993;
Mather et al., 1997;
Cook et al., 2004). Inhibins are composed of an α subunit and one of two β subunits, whereas activins are homo/heterodimers of either of the two β subunits (
Matzuk, 2000 a,
b). The action of activins is mediated by interaction at the cell membrane with type I and/or type II receptors (
Pangas and Woodruff, 2000) and phosphorylation of a class of substrates collectively known as Smads (
Massague, 1998;
Zhang and Derynck, 1999), e.g., Smads 2 and 3 which bind to a common mediator Smad 4 that translocates to the nucleus to activate target genes (
Lagna et al., 1996). Because a specific receptor for inhibin remains to be characterized (
Bernard et al., 2002), inhibins act either by binding β subunits to produce an inactive dimmer (i.e. inhibin) and/or binding to the activin receptor to prevent the action of activin (
Matzuk, 2000). Therefore, we determined whether the baboon fetal ovary expressed these important functional antagonists and whether expression was altered in association with the decrease in follicle formation in estrogen-deprived fetuses. Immunocytochemical and Western blot showed that proteins for activin βA and βB, the activin receptors (
Billiar et al., 2003), as well as Smad 2, 3 and 4 (
Billiar et al., 2004), were abundantly expressed and localized in oocytes and pregranulosa cells at midgestation and in granulosa cells at term. Moreover, the site of expression and levels of the activin subunits, the activin receptors and Smads 2, 3 and 4 were not altered by development or estrogen-deprivation. In contrast, expression of α-inhibin which was minimal in fetal ovaries of untreated baboons at both mid and late gestation, was markedly up-regulated in estrogen-suppressed fetuses in late gestation and detected primarily in pregranulosa cells of germ cell nests and granulosa cells of primordial follicles, an effect prevented when estrogen was restored (
Billiar et al., 2003). Therefore, we have proposed that estrogen regulates encapsulation of oocytes by granulosa cells, i.e. folliculogenesis, by controlling the intraovarian activin: inhibin ratio. Consistent with this suggestion, Mayo and colleagues have shown that over-expression of the α-inhibin subunit gene in mice in the presence or absence of FSH caused several ovarian pathologies including development of ovarian cysts (
McMullen et al., 2001;
Cho et al., 2001).
Estrogen regulation of fetal ovary microvilli formation
Elegant studies of others have shown that oocytes are dependent upon surrounding granulosa cells for uptake and exchange of metabolic products and thus survival (
Eppig, 1979;
Heller et al., 1981;
Brower and Schultz, 1982). In epithelial cells of the kidney and small intestine, nutrient/substrate exchange is accomplished primarily by microvilli developed and present in large numbers on the cell surface (
Lange 2002). Microvilli are present on oocytes of primordial follicles in the human fetal (
Gondos et al., 1971;
Motta and Makabe, 1982) and adult (
Familiari et al., 1993) ovary and in the ovary of adult rats oocytes exhibiting cytoplasmic vacuolization show a marked retraction of their microvilli (
Devine et al., 2000). Estrogen has previously been shown to increase development of microvilli in rat pituitary cells (
Antalky et al., 1980) and microvilli are disrupted/depleted in ovarian primordial follicles of women undergoing chemotherapy, an effect which is partially prevented by pretreatment with estrogen containing oral contraceptives (
Familiari et al., 1993). Therefore, we determined whether estrogen regulates the formation and/or maintenance of oocyte microvilli in the primate fetal ovary and whether development of microvilli by the fetal oocyte is associated with oocyte structural integrity (
Zachos et al., 2004).
Oocyte microvilli, examined by transmission electron microscopy, were not detected in the fetal ovary at midgestation but became abundant in primordial follicles by late gestation and were maintained in primordial follicles of the adult ovary. Importantly, oocyte microvilli were minimal in fetal ovaries of baboons depleted of estrogen, but well developed in fetal baboons treated with letrozole and estrogen. Thus, as confirmed by image analysis, the number of microvilli/intact oocyte in estrogen-suppressed fetuses was 56% lower (P < 0.01) than normal and restored by letrozole and estrogen. Similar results were noted when the number of microvilli was expressed per square micrometer oocyte surface area. Moreover, average height of microvilli on intact oocytes of letrozole-treated baboons was 54% lower (P < 0.01) than in baboons untreated or treated with letrozole plus estrogen. Interestingly, no microvilli were detected on the surface of non-intact oocytes in estrogen-suppressed baboons. These results are the first to demonstrate that development of a microvillus brush border by the oocyte of primordial follicles is regulated in utero in the primate fetal ovary by estrogen and support our hypothesis that estrogen regulates fetal ovarian folliculogenesis and development of follicles with healthy oocytes composed of microvilli critical for nutrient uptake and presumably long-term survival.
The biochemical mechanisms by which estrogen regulates oocyte microvilli development in the baboon fetal ovary remain to be elucidated (
Pepe et al., 2006). It is well established that ezrin plays an important role in the formation and maintenance of microvilli (
Bretscher et al., 1997;
Hanzel et al., 1991;
Berryman et al., 1993) and thus microvillus structures are completely lost after treatment of cultured epithelial cells with antisense nucleotides that block ezrin formation (
Takeuchi et al., 1994;
Yonemura et al., 1999). A preponderance of evidence shows that microvillus formation requires movement of ezrin from the cytoplasm to the plasma membrane and binding to membrane phosphatidylinositol-4,5-bis-phosphate (PIP2) and phosphorylation of ezrin by intracellular kinases (
Fievet et al., 2004 for review). Ezrin phosphate then links to the membrane spanning protein ezrin-binding protein 50 (EBP50), which maintains ezrin phosphate at the membrane (Reczek and Bretscher, 1998) and facilitates binding to f-actin to complete the formation of the microvilli (
Fievet et al., 2004). Thus, in association with microvilli breakdown in epithelial cells, ezrin is not phosphorylated and ezrin and EBP50 relocate from the surface of the cell to the cytoplasm (
Morales et al., 2004). Ongoing studies in our laboratories suggest that comparable mechanisms are operative in the fetal oocyte and regulated by estrogen. Thus, recently completed studies indicate that ezrin and EBP50 are expressed in the oocyte of primordial follicles in baboon fetal ovary and that cellular localization is modified and thus becomes localized away from the membrane in oocytes of fetal ovaries from estrogen-suppressed baboons (
Zachos et al., 2008).
In summary, the baboon fetal ovary expressed ERα and ERβ and fetal ovarian development was significantly altered in animals in which estrogen had been depleted by administration of an aromatase inhibitor during the second half of gestation and restored by concomitant administration of estradiol. Thus, in the absence of estrogen, follicle numbers were decreased by approximately 50%, whereas the number of germ cell nests comprised of pregranulosa cells and oocytes was increased and associated with a marked upregulation of expression of α-inhibin but not activins or the activin receptors and signalling molecules. We also showed that the majority of the follicles that did form in ovaries of estrogen-deprived fetuses appeared unhealthy and contained oocytes with a marked depletion/reduction in microvilli, structures essential for uptake of substrates from surrounding granulosa cells. We propose therefore, as illustrated in , that estrogen regulates formation of the pool of primordial follicles comprised of healthy oocytes essential for adult reproductive function by controlling the formation of microvilli on the oocyte membrane.
Significance
Finally, the significance of our studies is heightened by studies showing that endocrine disruptors that create either an excess (e.g. diethylstilbestrol) or deficit (e.g. deletion of ER genes; increased exposure to environmental factors that bind but do not activate ER) in estrogen availability/action have profound effects on fetal development (
McLachlan, 2001;
Guillette and Moore, 2006;
Abbott et al., 2006 for review). Moreover, because estrogen receptors are required for maintenance of germ and somatic cells in the postnatal rodent ovary (
Couse and Korach, 1999;
Couse et al., 1999), it is possible that in cases of estrogen deficiency/inhibition of estrogen action in human pregnancy there is a risk of impairment of fetal and consequently adult ovarian development either directly and/or secondary to enhanced sensitivity to environmental endocrine disruptors (
Vandenbergh, 2003) and/or biohazards which have been shown to accelerate follicular atresia and decrease primordial follicle development in estrogen-deficient but not estrogen-treated prepubertal rats (
Thompson et al., 2002). Thus, our studies in the baboon coupled with the corollary work of others demonstrate the need for translational research studies using nonhuman primate models to address impact of estrogen deficiency/impairment of estrogen action and exposure to environmental disruptors
in utero on ovarian maturation and fertility in adulthood.