The establishment of the germ line is of fundamental importance to animal reproduction. In mice, the extraembryonic ectoderm induces germ cell determination [
50]. During early embryonic development in mice, uncommitted epiblast cells in the extra-embryonic mesoderm involute through the primitive streak to the yolk sac endoderm and become committed as primordial germ cells (PGCs). These cells proliferate and migrate via the yolk sac into the hind gut endoderm and dorsal mesentery, and finally to the genital ridges during gastrulation. Upon reaching their destination, PGCs lose their motility, become encapsulated by the primary sex cords and differentiate depending on the sex chromosome set up into oogonia or spermatogonia. The cortical sex cords give rise to the female ovaries, whereas the medulla slowly deteriorates [
51]. The process of gametogenesis starts as the PGCs leave the dorsal mesentery and continues as they enter and colonize the genital ridges to establish the prospective gonad. The role of the TGF-β superfamily ligands in ovarian organogenesis as well as folliculogenesis has been studied extensively in animals. In particular, the BMPs together with their antagonists have been shown to be prominent throughout embryonic development and organogenesis. Gene ablation studies in mice have identified BMP4, -8b and -2 as regulators of primordial germ cell (PGC) formation from epiblast cells, BMP-2 deriving from the embryonic endoderm and BMP-4 and -8b from the extra-embryonal ectoderm [
3]. Targeted mutations of either BMP-4 or BMP-8b lead to severe defects in PGC formation in the embryos that survive gastrulation [
3]. Also, altered germ cell migration in the absence of TGF-β signalling via ALK-5 has been reported [
52].
Within the established ovary, the progress of folliculogenesis is in part regulated by peripheral endocrine factors, the pituitary gonadotropins FSH and LH as well as growth hormone (GH) and prolactin in some species. In addition, intraovarian factors, such as steroids, cytokines and other growth factors act in a paracrine/autocrine manner and co-ordinately contribute to the processes of recruitment, development, atresia, selection and ovulation of follicles [
53]. The growth of the follicle is considered gonadotropin-independent to the small antral stage and during these early phases folliculogenesis appears to be driven by the local autocrine and paracrine signals from the oocyte and the surrounding somatic cells. A complex bi-directional communication between the oocyte and granulosa cells as well as between the granulosa and thecal cells drives the progression of follicular development through successive stages [
54]. Various TGF-β superfamily ligands expressed by the different ovarian cell types are important in this interaction and their expression is regulated in a developmental-stage related manner. Among the local factors at least activins, inhibins, TGF-β s, BMP-6, GDF-9 and its homologue GDF-9B (also known as BMP-15) as well as anti-Müllerian hormone (AMH, also known as Müllerian inhibiting substance, MIS) are implicated in having a role during the development of follicles (for review, see [
51]). The developing oocyte has been shown to express GDF-9, GDF-9B, BMP-6 and TGF-β 2, although the TGF-β protein may not be secreted [
55‐
59]. Granulosa cells produce activins, inhibins, TGF-β s, BMP-2, BMP-3 and BMP-6 as well as AMH at different stages of folliculogenesis, while the theca cells have been reported to produce all the isoforms of TGF-β, BMP-3b, BMP-4 and BMP-7 [
60,
61]. TGF-β s, activins and GDF-9 signal through the Smad2/3 pathway whereas GDF-9B, BMP-2, -4, -6 and -7 utilize the Smad1/5/8 pathway (see Table
1). Table
1 summarizes the different superfamily ligands expressed by the oocyte, granulosa cells and thecal cells as well as their receptors and the Smad pathways they activate when known.
Table 1
Signalling pathways of TGF-β superfamily ligands expressed in the ovary.
GDF-9 | oocyte | BMPRII | ALK5 | Smad2/3 | [86-88] |
GDF-9B/BMP-15 | oocyte | BMPRII | ALK6 | Smad1/5/8 | [91] |
BMP-6 | oocyte, granulosa cell | BMPRII/ActRIIA/B | ALK2/ALK3/ALK6 | Smad1/5/8 | [94-96] |
TGF-β 1, -2, -3 | granulosa and theca cell | TβRII | ALK5 | Smad2/3 | [132, 133] |
Activin A/B | granulosa cell | ActRIIB | ALK4 | Smad2/3 | [134, 135] |
Inhibin α | granulosa cell | ActRIIA/ActRIIB | ? | ? | [136] |
BMP-2 | granulosa cell | BMPRII/ActRIIA | ALK3/ALK6 | Smad1/5/8 | [137, 138] |
BMP-3 | granulosa cell | ActRIIB | ? | ? | [106] |
AMH/MIS | granulosa cell | AMHRII | ALK2/ALK3/ALK6 | Smad1/5/8 | [111, 113-115] |
BMP-3b | theca cell | ? | ? | ? | [139] |
BMP-4 | theca cell | BMPRII/ActRIIA | ALK3/ALK6 | Smad1/5/8 | [137, 138, 140] |
BMP-7 | theca cell | BMPRII/ActRIIA | ALK2/ALK3/ALK6 | Smad1/5/8 | [140, 141] |
For appropriate signalling, an intact signalling cascade from ligands and receptors to intracellular effectors and accessory proteins has to be present and functional. The temporal and spatial regulation of these signalling cascade molecules determines the responsiveness of the cell type to each stimulus, and the direction of signalling within the follicle is dependent on cellular distribution of the whole signalling pathway. Reproductive defects can be found in knockout mice at all levels of the signalling cascade; at the ligand level activin β B, inhibin α, GDF-9, GDF-9B and AMH are known to cause fertility defects, at the receptor level, ALK6 (also known as BMP type IB receptor) and AMH type II receptor (AMHRII) affect female fertility, and finally at the intracellular effector level, Smad3 knockout mice exhibit reduced fertility [
1,
2,
62]. Ingman et al. very recently reported that TGF-β 1 null mice show severely impaired reproductive capacity and almost complete infertility [
63]. In the following section we discuss the main TGF-β superfamily ligands expressed by the different follicle cell types; the oocyte, granulosa and thecal cells as well as their target cells and signalling pathways.
The oocyte
In most mammalian species, both GDF-9 and GDF-9B mRNA and protein expression begin at the primary follicle stage and continue throughout the development of the maturing follicle [
55,
56,
64]. Depending on the species, both GDF-9 and GDF-9B are indispensable for normal progression of ovarian folliculogenesis as shown by the mouse and sheep animal models [
65‐
67]. GDF-9 deficient mice display arrested follicular development at the primary follicle stage, the theca cell layer is absent and in addition their oocyte development is compromised. Therefore, the homozygous female mice are infertile while heterozygous females and male mice are not affected [
65]. A naturally occurring mutation in the GDF-9 gene has been discovered recently in sheep that causes sterility in homozygous ewes due to abnormal follicle development, but surprisingly, increased ovulation rate and fertility in heterozygotes [
67].
In contrast to the mouse GDF-9 knockout, female mice completely lacking GDF-9B are fertile but exhibit reduced fertility due to defects in the ovulation process and the ability of oocytes to develop into normal embryos, whereas heterozygous females exhibit normal fertility [
68]. In sheep, four different mutations in the GDF-9B gene have been identified that affect fertility and ovulation rate, introducing premature stop codons (Belclare and Cambridge sheep FecX
G or Hanna mutation FecX
H) or non-conservative amino acid substitutions within the mature protein (Inverdale FecX
I or Belclare FecX
B) (reviewed in [
69]). All these mutations cause an arrest in folliculogenesis at the primary follicle stage in homozygous animals similar to the phenotype of the GDF-9 knockout mouse. In contrast, heterozygous ewes exhibit increased ovulation rates and fertility. The reason for these difference in phenotypes between mice and sheep is not fully understood but it has been suggested, however, that the differences may derive from the different ovulatory nature of these species, the sheep being a low ovulation rate species and the mice a poly-ovulatory species, or the different relative importance of these growth factors in sheep and mice [
68,
70].
The biological functions of the oocyte secreted GDF-9 have been studied extensively over the past five years, and an essential role for GDF-9 in the early stages of folliculogenesis as well as during ovulation is emerging. Recombinant GDF-9 functions as a granulosa cell mitogen and has been shown to modulate granulosa cell steroidogenesis. In addition, it has been shown to stimulate the growth of preantral rat follicles and the proliferation of rat and mouse granulosa cells [
71‐
74] as well as to induce the cumulus cell phenotype during ovulation [
72]. Only a few target genes for GDF-9 have been identified so far. GDF-9 has been shown to induce granulosa cell hyalurononan synthase 2 (Has2), cyclo-oxygenase 2 (Cox2), and steroidogenic acute regulatory protein (StAR) mRNA expression and to suppress the protease urokinase plasminogen activator (uPa) and luteinizing hormone receptor (LHR) as detected with semi-quantitative RT-PCR [
72,
75]. With a microarray approach, Varani et al. found that GDF-9 induces pentraxin 3 expression in mural granulosa cells from preovulatory mouse follicles [
76] and recently, Pangas et al. identified gremlin, a BMP antagonist, as a gene regulated by GDF-9 in mouse granulosa cells from large antral follicles [
77]. GDF-9 may also modulate theca cell function as it is known to stimulate the expression of CYP17, a theca cell marker [
78], and androgen biosynthesis in rat theca-interstitial cells [
79] as well as to inhibit 3'5'-adenosine monophosphate-stimulated steroidogenesis in human theca cells [
80].
Few targets for GDF-9B have been identified, however, it is known to suppress FSH receptor mRNA expression [
81], to stimulate Kit ligand expression in rat granulosa cells [
82] and to simultaneously promote expression of anti-apoptotic Bcl-2 and suppress pro-apoptotic Bax [
83]. Recombinant GDF-9B functions as a granulosa and cumulus cell growth factor by actively preventing cell death and promoting DNA synthesis and proliferation
in vitro [
81,
83]. In the light of recent data it is becoming clear that GDF-9 and GDF-9B can co-operate to regulate granulosa cell functions e.g. proliferation and gonadotropin-induced differentiation [
84,
85].
GDF-9 has been shown to mediate its signal through cell surface receptors BMPRII, that normally functions as a BMP type II receptor, and ALK5, the type I receptor of TGF-β, and activates Smad2/3 pathway [
86‐
90]. The GDF-9 receptor combination is interesting since this is the first reported physical interaction between ALK5 and a BMP type II receptor. GDF-9B interacts with BMPRII and has also been shown to interact with ALK6 (or BMP receptor type IB) and causes the activation of Smad1/5/8 pathway [
91]. The receptor complex binding a GDF-9-GDF-9B heterodimer has not been reported yet, but it could be predicted to consist of two BMPRII molecules in complex with one ALK5 and one ALK6 molecule. Interestingly, a naturally occurring point mutation was found in sheep in the gene coding for ALK6 (Booroola gene FecB), the type I receptor for e.g. GDF-9B, which causes increased ovulation rates in heterozygous sheep compared to wild type sheep, and even higher ovulation rates in homozygotes [
92]. Although GDF-9B, BMP-2, -4, -6 and -7 are expressed in the mammalian ovary and can signal through ALK6, it is not known to what degree each of these is involved in the Booroola phenotype. However, based on the similarity of the phenotype with the heterozygous Inverdale ewe with a point mutation in the GDF-9B protein coding gene, the involvement of altered BMP-15 signalling is strongly suspected [
92], GDF-9B showing the highest affinity to ALK6 of the type I receptors [
91]. The follicles of these Booroola ewes mature and ovulate at smaller sizes with fewer granulosa cells than in wild-type ewes. The ALK6 knockout mouse phenotype differs from the Booroola sheep phenotype. The knockout mice appear to have normal ovarian follicular development and ovulation rates, but display reduced fertility which may be caused by the failure of normal cumulus expansion [
2].
BMP-6 is the third TGF-β superfamily growth factor secreted by the oocyte from the primary stage onwards [
58] but it lacks the mitogenic activity of GDF-9 and GDF-9B (Gilchrist et al., 2005 submitted) [
93]. BMP-6 modulates granulosa cell steroidogenesis by inhibiting FSH-induced progesterone synthesis, but has no effect on estradiol production. BMP-6 suppresses the FSH action at the level of adenylate cyclase downstream of the FSH receptor in contrast to GDF-9B which suppresses FSH receptor expression [
81,
93]. The preference of cell surface receptors for BMP-6 in the ovary has not been determined yet but BMPRII, ActRII as well as ActRIIB have been implicated as type II receptors for BMP-6 [
94,
95] and all BMP ALKs (ALK2, -3 and -6) have been identified as potential BMP-6 type I receptors, with ALK6 having the strongest binding affinity [
95,
96]. BMP-6 can activate the Smad1/5/8 pathway in a human granulosa tumour cell line [
91].
The granulosa cells
Activins and inhibins were first discovered as gonadal proteins that regulate pituitary FSH secretion [
97]. Three types of activin are produced in the ovary by the granulosa cells, each consisting of a dimer of two related subunits β A and β B i.e. homodimeric activin A (β Aβ A) and activin B (β Bβ B), and a heterodimeric activin AB (β Aβ B). Two types of inhibin are also expressed by granulosa cells. Inhibins consist of one inhibin α-subunit and one activin subunit forming either inhibin A (α-β A) or inhibin B (α-β B). Activin produced by the secretory gonadotrophs in the anterior pituitary stimulates FSH production in a paracrine manner, and within the ovary activin promotes granulosa cell proliferation [
98] as well as potentiates FSH actions by increasing FSH receptor expression [
99]. Activin also modulates granulosa and theca cell steroidogenesis. Activins are produced by the granulosa cells and the expression pattern of the different activin subunit mRNAs changes during folliculogenesis.
Inhibins are also produced by the granulosa cells, and they act as endocrine hormones that are released into the circulation to suppress pituitary FSH production. Locally, inhibins also act as potent regulators of activin signalling. Inhibins compete with activin signalling by blocking activin binding to type II activin receptors. β-glycan, an inhibin co-receptor, facilitates inhibin binding to the activin type II receptor [
100]. Follistatin (FS) is yet another granulosa cell produced inhibitor of activin function which also regulates the actions several BMPs, including GDF-9B/BMP-15 [
83]. Follistatin antagonises activin through forming biologically inactive complexes. Activin subunits bind ActRIIB and ALK4, and activate the Smad2/3 pathway whereas the inhibin α-subunit binds to ActRIIA or ActRIIB. Cripto, a prototypic member of the epidermal growth factor-Cripto protein family, antagonises activin signalling by binding to the activin type II receptor and blocking ALK4 recruitment [
101].
Also BMP-2, BMP-3 and BMP-6 are expressed by the granulosa cells [
102,
103]. Recombinant BMP-2 has been shown to amplify FSH-induced estradiol and inhibin A production in sheep granulosa cells [
104] and stimulate inhibin β B subunit mRNA expression as well as inhibin B protein production in cultured human granulosa luteal cells [
105]. BMP-2 can signal through either BMPRII or ActRIIA and ALK3 and -6 activating the Smad1/5/8 pathway. BMP-3 mRNA has been shown to be expressed in human granulosa luteal cells but the biological function of BMP-3 in the ovary is still unclear. However, it was reported that the expression level of BMP-3 is regulated by human chorionic gonadotropin (hCG) [
103]. Recently, it was discovered that BMP-3 binds ActRIIB and functions as a novel inhibitor for both activin and BMP-4 signalling in
Xenopus embryos [
106]. BMP-6 expression in granulosa cells is first detected at the early secondary stage in rat follicles and it is rapidly lost at the time of dominant follicle selection suggesting that inhibition of BMP-6 gene activity may be required for the formation of the dominant follicle. BMP-6 is again highly expressed in atretic follicles supporting this hypothesis [
102].
AMH is expressed by the Sertoli cells in the testis and granulosa cells in the ovary [
107]. In the male AMH causes the regression of Müllerian ducts that in the female differentiate into the oviducts, the uterus and the upper part of the vagina. In the granulosa cells AMH is first expressed postnatally in primordial follicles recruited to growth and continues to be expressed until the growing follicles are selected for dominance by the action of FSH. AMH deficient mice are fertile but their pool of primordial follicles is depleted earlier than wild type mice [
108]. AMH has been shown to inhibit the initiation of primordial follicle growth in the neonatal mouse ovaries as well as inhibit the stimulatory effect of FSH on the growth of preantral and small antral follicles [
107]. Instead, in the rat, AMH promotes preantral follicle growth in the presence of FSH but not preantral follicle cell differentiation and apoptosis [
109] and in human, it was recently found that AMH induces growth of primordial follicles from ovarian cortical tissue [
110]. AMH binds to the AMH type II receptor [
111] and causes the activation of Smad1/5/8 in a granulosa tumour cell line [
112], but the type I receptor has yet to be conclusively confirmed (ALK2,-3 and -6 are implicated) [
113‐
115].
The theca cells
A theca cell layer forms to surround the developing follicle outside the basal lamina at the primary/secondary transition. Theca cells from rat follicles have been shown to produce BMP-3b, BMP-4 and BMP-7, as well as all the isoforms of TGF-β [
102,
116,
117]. Recombinant BMP-4 and -7 have been found to modulate FSH signalling by promoting FSH induced estradiol production and inhibiting progesterone biosynthesis [
116]. Granulosa cells in growing follicles produce estradiol but no progesterone
in vivo in response to FSH stimulation until the periovulatory period, whereas
in vitro cultured granulosa cells produce progesterone as well as estradiol in response to FSH stimulation. Therefore, it has been suggested that the biological function of theca cell derived BMPs might be to function as selective inhibitors of progesterone synthesis (luteinization inhibitors) as neither BMP-4 or BMP-7 affect granulosa cell steroidogenesis in the absence of FSH in the rat [
116]. Both BMP-4 and -7 bind to the BMPRII/ActRIIA receptors and ALK3 and -6 which are predominantly expressed in granulosa cells [
116], BMP-7 may also signal through ALK2 [
118]. Both activate the Smad1/5/8 pathway in their target cells.
In conclusion, the oocyte secreted factors GDF-9, GDF-9B and BMP-6 can activate either Smad2/3 or Smad1/5/8 pathways in the granulosa cells and modulate their differentiation and proliferation. However, Smad2/3 signalling is by far the predominate pathway used by oocyte secreted factors to promote granulosa and cumulus cell growth and cumulus cell expansion (Gilchrist et al., submitted). It is not clear, however, to what extent Smad mediated signalling is involved in the oocyte maturation process during folliculogenesis or whether the oocyte secreted factors have autocrine effects on the oocyte itself. The human oocyte shows immunostaining for Smad2 and Smad4 at primordial and primary stages of folliculogenesis, having therefore the capacity to respond to TGF-β-like ligands but little is known of the significance of the Smad1/5/8 pathway in the oocyte development [
119]. The granulosa cell expressed ligands may act locally in a paracrine or autocrine fashion affecting the granulosa cells themselves or the thecal cells and possibly even the oocyte, or they may have systemic effects acting on for example the pituitary gonadotropin expression of activins and inhibins.