Transcriptional regulation of early oogenesis: in search of masters

Abstract

Transcription factors in the germline play important roles in ovary formation and folliculogenesis, and control both oocyte development and somatic cell function. Factor in the germline (Figla) and newborn ovary homeobox gene (Nobox) represent a growing number of oocyte-specific transcription factors that regulate genes unique to oocytes. Studies on oocyte-specific transcription factors are important in understanding the genetic pathways essential for oogenesis, pluripotency, and embryonic development. Likely, these genes regulate reproductive life span and represent candidate genes for reproductive disorders, such as premature ovarian failure, and infertility. Therefore, oocyte-specific transcription factors, and oocyte-specific genes regulated by such factors, are attractive tissue-specific pharmacological targets to regulate human fertility.

Submitted on May 3, 2005; accepted on June 27, 2005.

Introduction

The vertebrate egg is a fascinating cell. It is the only known cell that can reprogram somatic nuclei and initiate embryonic development via parthenogenesis, a process where an egg can develop into an embryo in the absence of sperm (Kono et al., 2004; Loebel and Tam, 2004). However, the molecular mechanisms necessary for the development of the oocyte are largely unknown. Cells differentiate through the activation and repression of gene transcription and it is has been hypothesized that master control genes regulate transcriptional cascades that generate cell and tissue-specific fates (Gehring, 1996). For example, the homeobox gene Pax6 (and its Drosophila homologue, eyeless) controls eye development (Halder et al., 1995; Gehring, 1996). In Drosophila, ectopic expression of eyeless or its vertebrate orthologs induces functional eye development on legs, wings, and antenna of fruit flies. Transcriptional cascades that activate an oocyte-specific developmental program may be used in the future to produce de novo oocytes from somatic cells. Although this idea might seem far away, it is currently possible to generate ‘oocyte-like’ cells from pluripotent ES cells (Hubner et al., 2003). Therefore, to understand how to ‘make’ an oocyte, the molecular program that specifies a female germ cell will have to be uncovered. Transcription factors, particularly oocyte-specific ones, are likely to be the critical switches. Moreover, oocyte-specific transcription factors are likely to control reproductive life span, success in fertilization, early embryo development and formation of ovarian tumors.

Oocyte transcriptional control must orchestrate expression of oocyte-specific genes necessary for oocyte growth and early embryonic development. Oocyte-specific genes are some of the most abundant transcripts in the ooplasm and include zona pellucida genes, Zp1, Zp2 and Zp3 (Rankin et al., 1996, 1999, 2001), growth differentiation factor 9, Gdf9 (McGrath et al., 1995), maternal antigen that embryos require, Mater (Tong et al., 2000), and zygote arrest 1, Zar1 (Wu et al., 2003). Multiple approaches have been undertaken to identify genes preferentially expressed in the oocyte and many new genes, including transcription factors, such as factor in the germline (Figla), Gpbox, newborn ovary homeobox gene (Nobox) and Obox have been discovered because of these efforts (Liang et al., 1997; Takasaki et al., 2000; Rajkovic et al., 2001, 2002; Agoulnik et al., 2002). We will review mouse models of transcription factors that are important in oogenesis with special emphasis on genes preferentially expressed in the germline and their potential roles in human reproduction (Table I).

The origin of germ cells

The molecular pathway(s) for the origin of germ cells in mammalian organisms are not well understood. More is known in invertebrate species such as Drosophila and Caenorhabiditis elegans. Drosophila germ cells arise from the posterior end of the fertilized egg where germ cell determinants are passed to the pole cells (Illmensee and Mahowald, 1974; Kobayashi et al., 1996; Rongo et al., 1997). Polar granules in the unfertilized eggs of C. elegans are asymmetrically distributed to daughter cells to be concentrated finally in P4 cells that give rise to the entire C. elegans complement of germ cells (Strome and Wood, 1982). Germplasm is not obvious in mice or humans and therefore, other mechanisms of germ cell segregation must exist (McLaren, 2003). The germ cell lineage in mice is induced from the proximal epiblast (embryonic ectoderm) of the egg cylinder at embryonic day 6.5 (E6.5) in response to signalling by Bmp4 and Bmp8b (Zhao et al., 1996; Lawson et al., 1999; Ying et al., 2001). A founder population of approximately 45 primordial germ cells (PGCs) at E7 gives rise to the germ cell lineage (Lawson and Hage, 1994). The earliest known molecular marker for mammalian germ cells is tissue non-specific alkaline phosphatase (TNAP) detected at E8.5 (Chiquoine, 1954) although transgenic studies, using β-galactosidase expressed under the control of TNAP promoter, detect TNAP in PGCs beginning at E7 (MacGregor et al., 1995). Although TNAP has been an invaluable marker to study PGCs, ablation of TNAP does not affect PGC migration and no excess embryonic loss occurs in mice lacking TNAP (MacGregor et al., 1995).

A variety of techniques can be used to identify genes specifically expressed in a particular tissue. Two groups have used differential screening and cDNA subtraction to identify Ifitm3 (interferon induced transmembrane protein 3, fragilis, mil-1) and Dppa3 (developmental pluripotency-associated 3, stella) from founder PGC cells against surrounding somatic cells (Saitou et al., 2002; Tanaka and Matsui, 2002). Ifitm3 is a member of the interferon inducible transmembrane protein family, expressed strongly at E7 in the founder PGC cluster. Ifitm3 expression fades by E8.5 when PGCs migrate. However, Ifitm3 expression is not limited to PGCs and multiple adult tissues express Ifitm3 (Su et al., 2002). A mouse knockout model for Ifitm3 has not been reported, and functional evidence for Ifitm3 in PGC formation is lacking. Even less is known about human IFITM3 in the human PGCs.

Dppa3 was first detected at the late streak stage, E7.0, and its expression coincides with the appearance of nascent PGCs (Saitou et al., 2002). Oocytes throughout murine folliculogenesis express Dppa3 as do preimplantation embryos (Sato et al., 2002). Functional studies in knockout mice show that a lack of Dppa3 disrupts embryo development beyond the four-cell stage (Payer et al., 2003; Bortvin et al., 2004). Dppa3 is therefore a maternal effect gene: maternal mRNAs that are used to direct embryonic development before zygotic genome activation. It is possible that in humans, a subset of infertile patients whose embryos arrest early in development may have a mutation in Dppa3 and therefore Dppa3 is a candidate gene for post-ovulatory infertility. However, the human homologue of Dppa3 is highly divergent from the mouse and shares only 32% identity with the mouse protein. Even so, its expression pattern is similar to the murine Dppa3, and it is detectable in human embryonic stem cells, fetal and adult ovaries as well as unfertilized eggs (Clark et al., 2004) and thus future research on this gene in human disease will be important.

It is unclear what regulates the expression of Ifitm3 and Dppa3 in germ cells, but other proteins probably exist that are important in specifying the founder population. The molecular characterization of the early PGC transcriptome is technically difficult because of the very small number of founder cells, but will lead to a discovery of many genes critical for the germline and open fruitful research avenues into this area. In addition, analysis of the early PGC transcriptome may lead to the discovery of important, yet uncharacterized classes of transcription factors, potentially the master genes necessary to initiate germ cell differentiation.

There is a set of transcription factors that are down-regulated during germ cell development. The homeobox genes Hoxa1, Hoxb1, Lim1 and Evx1 are repressed in cells destined to become germ cells (Saitou et al., 2002). It is unclear what role this down-regulation plays in the genesis of germ cells. Evx1 is particularly interesting because of its expression in oocytes and fertilized egg (Su et al., 2002). Evx1 may be a maternal effect gene, with important functions in the regulation of the embryonic genome, but functional studies are lacking. Mice without functional Evx1 die early in embryogenesis (Spyropoulos and Capecchi, 1994), thus the question of whether Evx1 is expressed in the PGCs and early oogenesis is also unanswered and worth investigating.

Another homeobox gene of interest is Nanog, discovered both by mining expressed sequence tags (EST) databases as well as in a functional assay (Chambers et al., 2003; Mitsui et al., 2003). Nanog is first detected at the morula stage and is highly expressed in the inner cell mass of the blastocyst. Subsequently, Nanog is expressed in female and male PGCs at E11.5 but its expression beyond E11.5 in germ cells is not clear (Chambers et al., 2003). A loss of Nanog through gene targeting (Mitsui et al., 2003) blocks epiblast formation and embryonic stem cells lose pluripotency. Therefore, Nanog is important for stem cell pluripotency and self-renewal. Since most of the studies have been directed towards its role in embryonic stem cells, and its knockout phenotype precedes PGCs, Nanog’s role in PGC proliferation, migration and/or differentiation is currently unclear. A conditional knockout for Nanog, during later stages of development, may yield more clues whether this important homeobox gene functions in PGCs. Additionally, the downstream targets of Nanog are not known and will be important to identify these targets both in the inner cell mass as well as PGCs. In humans, Nanog expression is similar to the murine expression profile, suggesting that there is a potentially conserved role between species (Clark et al., 2004).

Much research has focused on the role of Pou5f1 (POU domain, class 5, transcription factor 1, Oct4) in the germline (Pesce et al., 1998). Pou5f1 undergoes a sequentially restricted expression pattern during embryogenesis. Before fertilization, Pou5f1 mRNA and protein are abundantly expressed in unfertilized eggs. Following fertilization, maternally derived Pou5f1 persists until the two-cell stage embryo stage and new production of Pou5f1 RNA commences at the four- to eight-cell stage. In the blastocyst, Pou5f1 expression is confined to the inner cell mass (Rosner et al., 1990; Scholer et al., 1990; Palmieri et al., 1994; Yeom et al., 1996) and after E7.5–8.0, Pou5f1 expression is restricted to the germline. Pou5f1 has two well-defined enhancers, the proximal and distal enhancer and a switch occurs from the use of proximal to distal enhancer as Pou5f1 expression becomes confined to PGCs. The mechanism of this switch is unclear. Pou5f1 is expressed in migrating germ cells as well as the proliferating germ cells. At approximately E13.5, female germ cells begin entry into the prophase I of meiosis. Pou5f1 expression in the female germline ceases at approximately E14.5 and reappears after birth. Pou5f1 is detected in oocytes of all follicular stages as well as unfertilized eggs (Pesce et al., 1998). Although Pou5f1 expression is confined to the large portion of oogenesis, its downstream targets in oocytes and precise function in oogenesis is unknown. The conventional knockout of Pou5f1 showed a block in embryo development at the blastocyst stage. The inner cell mass loses pluripotency and embryos die at the time of implantation (Nichols et al., 1998). Clearly Pou5f1 is required very early in embryogenesis, and because of the early embryonic lethality, unfortunately, this knockout mouse model could not answer the question of what functional role Pou5f1 plays in PGCs. A conditional knockout for Pou5f1 has also been generated (Kehler et al., 2004) using the Cre-LoxP system. The TNAP promoter was used to express the Cre recombinase, leading to excision of the floxed Pou5f1 locus in PGCs. This resulted in female sterility. The phenotype in the conditional knockouts was not observed until E10 even though Pou5f1 and TNAP (therefore Cre recombinase) are coexpressed in PGCs during their initial formation at E7.5. The number of PGCs was markedly lower in homozygous mutants as compared to heterozygotes, presumably because of premature apoptosis in PGCs. At 3 weeks post-partum, the number of primordial follicles was 25–100 times less in animals with both alleles excised, as compared to animals with only one allele excised, perhaps as a result of the initial PGC depletion at E10. Because not every cell excised the Pou5f1 locus, it is likely that this phenotype is hypomorphic, and therefore, excision in 100% of the cells could lead to complete loss of PGCs. Since the targets of Pou5f1 in PGCs are unknown, the conditional knockouts will be a useful model for allowing comparison of gene expression in PGCs and a look into which genes are up- or down-regulated in the absence of Pou5f1.

Another important transcription factor is Nr6a1 (nuclear receptor subfamily 6, group A, member 1, also known as Nr6a1), a member of the nuclear receptor superfamily (Chen et al., 1994; Hirose et al., 1995; Bauer et al., 1998; Lan et al., 2003a). Nr6a1 is found in many species, including humans, frogs and zebrafish and binds to a direct repeat of the estrogen receptor half site (AGGTCA) (Yan et al., 1997). Nr6a1 functions as a transcriptional repressor (Bauer et al., 1997; Cooney et al., 1998) and a conventional knockout of Nr6a1 shows that mice deleted in both loci cannot survive beyond E10.5. Nr6a1 knockout mice suffer multiple defects, presumably because of disrupted somitogenesis and posterior development of the embryo. In the adult, Nr6a1 is expressed in the oocytes of primary follicles and all subsequent stages of folliculogenesis. In the germline, Nr6a1 is expressed in unfertilized eggs and preimplantation embryos and may regulate zygotic gene expression. The role of Nr6a1 in the genesis of germ cells is not understood. However, Nr6a1 has been shown to play an important role in repressing Pou5f1 expression during early embryogenesis (Fuhrmann et al., 2001). Nr6a1 expression during PGC migration, during proliferation and in the gonads of mouse embryos has not been characterized.

PGC migration and proliferation

PGC migration commences at E9.5 until E11.5 (Clark and Eddy, 1975; Donovan et al., 1986; Anderson et al., 2000) and progresses from the hindgut epithelium to developing genital ridge. Very little is understood with regards to germline specific transcription factors necessary to maintain migration and proliferation and initiate sexual differentiation. The migration of PGCs was elegantly studied by utilizing Pou5f1-driven green fluorescent protein and time-lapse confocal microscopy (Anderson et al., 2000). This study showed active migration of PGCs from the time they exited the primitive streak. It is possible that continuing expression of Pou5f1 during germ cell migration and proliferation plays an essential role in maintaining transcription of yet unknown subset of germ cell specific genes but these targets of Pou5f1 have yet to be identified.

Psx2 (placenta specific homeobox gene 2, Gpbox) is another homeobox gene discovered by sequencing cDNA libraries derived from E12 and E13 urogenital ridges in an effort to identify genes essential for sexual differentiation (Takasaki et al., 2000). Psx2 is expressed in the male and female germline at E12 and persists until E15.5 with no apparent expression in the adult tissues including the ovary. Psx2 shares 82% amino acid identity with Psx1 (placental specific homeobox gene 1), and RNA for both genes is present from E11.5 until E15.5 female genital ridge. Psx1 and Psx2 have previously been described in the placenta (Han et al., 1998, 2000). Both Psx2 and Psx1 are located on the X chromosome. The Psx2 knockout has no obvious pathology, perhaps because Psx1 compensates for the loss of Psx2. Generating a Psx1 knockout and Psx2/Psx1 double knockouts will help us to understand the role of this family of homeobox genes in female gonadal development. Redundancy within gene families, while clearly desirable for the organism, confounds experimental studies.

Interactions between germ cells and surrounding somatic cells are essential for the formation of the ovary and subsequent folliculogenesis. Transcription factors in somatic cells are known to play important functions in the gonad formation and germline survival. These include but are not limited to LIM homeodomain gene 9 (Lhx9), empty spiracles homologue 2 (Emx2), Gata4, Steroidogenic factor 1 (Sf1) and Wilms tumour 1 (Wt1). Lhx9, Emx2, Gata4, Sf1 and Wt1 are expressed in multiple adult tissues as well as somatic cells of the urogenital ridge (MacLaughlin and Donahoe, 2004). Lhx9 encodes a protein with a homeodomain and two cysteine-rich LIM domains. Mouse Lhx9 expression can be detected as early as E9.5 in the presumptive gonadal region of the urogenital ridge, and in the adult ovary, its expression appears confined to the surface epithelium (Birk et al., 2000; Mazaud et al., 2002). Although Lhx9 is expressed in the embryonic central nervous system, pancreas and limbs, loss of Lhx9 in knockout mice produced viable mice with gonadal agenesis and genetic males that are phenotypically female (XY sex reversal). PGCs migrated on time to the urogenital ridge of Lhx9 knockout mice, but subsequent proliferation of somatic cells in the urogenital ridge at E11.5 did not occur. There are currently no published conditional knockouts for Lhx9 and its role in ovarian development and/or folliculogenesis is not known. Mice lacking Emx2, die soon after birth and lack kidneys, ureters and gonads (Miyamoto et al., 1997). Interestingly, PGC migration also occurs normally in the Emx2 knockout. Gata4 (GATA binding protein 4) is expressed in the soma of the urogenital ridge as early as E11.5 and in the adult ovary its expression is confined to granulosa, theca and interstitial cells of primary through antral follicles (Heikinheimo et al., 1997; Viger et al., 1998; Tevosian et al., 2000; Vaskivuo et al., 2001; Salmon et al., 2004). Gata4 is a zinc finger transcription factor that binds a consensus target sequence (T/A)GATA(A/G) and requires interaction with Zfpm2 (zinc finger protein multi type 2, also known as Fog2) for in vivo function. Zfpm2 is also expressed in the soma of the urogenital ridge at E11.5 and granulosa cells of adult ovaries (Tevosian et al., 2000; Salmon et al., 2004). Deletion of Gata4 in mice leads to early death at E9.5 due to defects in ventral morphogenesis (Kuo et al., 1997; Molkentin et al., 1997) and deletion of Zfpm2 leads to embryonic death by E14.5 (Tevosian et al., 2000). The embryonic gonads in mice that lack Zfpm2 and mice that carry a Gata4 knock-in allele (Gata4^ki, a V217G amino acid substitution that abrogates the interaction between Gata4 and Fog2) are abnormal, although exact defects and molecular abnormalities in the XX gonad are unclear (Tevosian et al., 2002). The role of Gata4 and Zfpm2 in adult ovary and folliculogenesis is unclear and awaits generation of conditional knockouts that specifically ablate Gata4 in granulosa cells of the ovary. Sf1, also known as a nuclear receptor subfamily 5, group A, member 1 (Nr5a1), is a transcription factor without a known ligand, hence known as orphan nuclear hormone receptor. Sf1 binds a canonical (T/C)CAAGG(T/C)C(A/G) motif and promotes transcription of numerous genes including anti-Mullerian hormone (Amh) (Shen et al., 1994), all the cytochrome P-450 steroid hydroxylase enzymes (Parker and Schimmer, 1995), 3β-hydroxysteroid dehydrogenase (Leers-Sucheta et al., 1997), sex determining sex reversal Y (Sry), Sox9 and Dax. Sf1 are expressed in the murine gonadal ridge as early as E9, before sexual determination (Ikeda et al., 1993). Migration of PGCs to the urogenital ridge is unaffected in female mice with disrupted Sf1, but subsequent degeneration of the embryonic female gonad leads to no ovaries at the time of birth (Luo et al., 1994). Wt1 is a zinc finger DNA-binding protein that associates with Sf1 to promote anti-Mullerian hormone (Amh) expression (Nachtigal et al., 1998). Similar to Sf1, the homozygous knockout for Wt1 does not develop gonads before sexual differentiation (Kreidberg et al., 1993). Wt1 is expressed in humans in the soma of the gonadal ridge (Hanley et al., 1999) and murine granulosa cells of adult ovaries (Hsu et al., 1995; Salmon et al., 2004). The above transcription factors expressed in the somatic cells are clearly important in the formation of gonads but these genes are not thought to play a significant role in female sexual differentiation (MacLaughlin and Donahoe, 2004).

Primordial follicle formation

At birth, the mouse ovary is filled with oocytes, most of which are present in clusters with no evidence of surrounding granulosa cells (Figure 1). These collections of oocytes have various terms including ‘naked’ oocytes, germ cell cysts, clusters, nests or syncitia. In mice, the vast majority of oocytes have entered meiosis during embryonic life, and at birth some oocytes are in the transitory stages of prophase (pachytene and early diplotene), while others have entered late diplotene and dictyate in which they apparently remain until meiosis resumes shortly before ovulation (Pedersen and Peters, 1968). More centrally located are oocytes that have already been enclosed by a layer of flat squamous pre-granulosa cells and these represent primordial follicles. Primordial follicles in the adult ovary are located in the periphery of the ovary underneath the epithelial surface. The number of germ cell clusters declines rapidly after birth with few clusters remaining beyond postnatal day seven (Pepling and Spradling, 2001). By postnatal day 3, some primordial follicles become primary follicles as granulosa cells undergo a squamous to cuboidal transition and the oocytes grow beyond 20 µm. The transcription of numerous oocyte-specific genes is initiated during these early stages of the primordial to primary follicle transition. Genes that are critical both in folliculogenesis such as Gdf9, as well as embryogenesis, such as Mater and Zar1, are transcribed early in folliculogenesis. Two transcription factors, Figla (Liang et al., 1997; Soyal et al., 2000) and Nobox (Suzumori et al., 2001; Rajkovic et al., 2004) are oocyte specific and play a critical function during early folliculogenesis.

Figla is a basic helix-loop-helix transcription factor that is expressed exclusively in germ cells and regulates transcription of zona pellucida (Zp) genes through an E-box motif (CANNTG) (Liang et al., 1997). Figla binds the E-box motif that is located approximately 200 base pairs upstream of the zona pellucida transcription start sites. Although Figla is expressed as early as mouse embryonic day 13, embryonic gonads in Figla knockouts appear grossly normal until birth, after which formation of primordial follicles is blocked and oocytes are rapidly lost in a matter of a few days (Soyal et al., 2000). Figla is therefore a critical transcription factor for early steps in folliculogenesis. Zp1, 2 and 3 gene expression is absent in Figla knockouts as predicted, but other important genes analysed such as Gdf9, Bmp15, Kit (encoding a transmembrane receptor with tyrosine kinase activity), Kitl (kit ligand), Cx43 (connexin 43) and Fgf8 (fibroblast growth factor 8) are present in both knockouts and wild types. Because the individual zona pellucida knockouts are unaffected in early folliculogenesis, it is likely that there are additional genes that Figla regulates at this stage. A comparison of oocyte transcriptomes between wild type and Figla knockout ovaries will help to identify genes that Figla regulates. In humans, FIGLA is expressed as early as 14 weeks gestational age with a dramatic increase in transcripts by midgestation (19 weeks of gestational age)–the time of primordial follicle formation in humans. Human FIGLA also binds an E-box in the human ZP2 promoter, suggesting a similar conserved function of the human and mouse FIGLA proteins (Bayne et al., 2004).

Nobox was isolated using an in silico screen to detect transcripts specifically expressed in newborn ovaries (Rajkovic et al., 2001; Suzumori et al., 2001). Nobox mRNA is preferentially expressed in germ cells, as early as E15, and throughout folliculogenesis, including germ cell cysts, primordial, growing and antral follicles. The Nobox pattern of RNA expression is very similar to that of Figla, and like Figla knockouts, Nobox deficiency similarly disrupts early folliculogenesis. Ovaries lacking Nobox formed what appeared to be histologically primordial follicles, but the oocytes rarely grow beyond 20 µm and the number of somatic cells surrounding oocytes rarely exceeds seven (Rajkovic et al., 2004). Also similar to Figla knockouts, the Nobox knockout ovaries rapidly lose oocytes, and by 2 weeks after birth there are only few degenerating oocytes left in the Nobox knockout ovaries. Well characterized genes such as Kitl and its receptor Kit, important in survival, migration and proliferation of PGCs as well as primordial follicle formation (Matsui et al., 1990; Manova and Bachvarova, 1991; Driancourt et al., 2000) are expressed in Nobox knockout ovaries. Moreover, genes important in apoptosis and meiosis such as Bax, Bcl2, Bcl2l2, Casp2, Mlh1 and Msh5 are also expressed in Nobox knockout ovaries. However, transcripts corresponding to numerous genes preferentially expressed in oocytes, such as Pou5f1, Gdf9, Bmp15, Rfpl4, H1oo, Zar1 and Mos are drastically reduced in newborn ovaries that lack Nobox (Rajkovic et al., 2004). Future studies are necessary to determine whether Nobox affects the amount of oocyte-specific transcripts by directly or indirectly regulating the transcription of such genes. Since Nobox knockouts lose oocytes rapidly, these mice represent good models to identify genes responsible for oocyte loss in humans. Even though there is such a complete loss of oocytes after birth, the genes usually implicated in apoptosis, such as Bax, Bcl2, Casp2 and Bcl2l2, are expressed equivalently in Nobox knockout ovaries as wild type ovaries, and this may suggest that other (potentially novel) oocyte-specific genes may be necessary for the oocyte survival. In the adult ovary, Nobox expression persists throughout folliculogenesis, likely maintaining transcription of numerous oocyte-specific genes. The expression pattern of the human NOBOX in ovaries has not been documented, but the lack of human ESTs corresponding to human NOBOX is most likely because of the restriction of human NOBOX expression to oocytes. NOBOX is likely not the only oocyte-specific factor not represented in the public databases, and there is a real need for sequenced human oocyte cDNA libraries in public EST databases, such as Unigene, to further the cross-species comparison of genes preferentially expressed in oocytes.

In addition to Nobox and Figla, Pou5f1 and Nr6a1 transcripts are detectable during early folliculogenesis (see section The origin of germ cells). The role of Pou5f1 during early folliculogenesis is unknown and hampered by early embryonic death in corresponding knockout animals. Conditional knockouts of Pou5f1 during early folliculogenesis will be necessary to verify whether it is important during this stage of follicular development. Nr6a1 conditional knockout experiments indicate that Nr6a1 is not essential in early folliculogenesis or follicular growth (Lan et al., 2003b).

Transcription factors expressed in the somatic cells of the ovary are also important in the formation of primordial and primary follicles. Foxl2 is one such transcription factor associated with human disease and expressed highly in the pre-granulosa cells of primordial follicles, granulosa cells of pre-antral and antral follicles and eyelids (Crisponi et al., 2001). Foxl2 is a member of the forkhead/hepatocyte nuclear factor 3 gene family of transcription factors (Crisponi et al., 2001). In humans, nonsense mutations and duplications in the Foxl2 gene cause blepharophimosis ptosis and epicanthus inversus syndrome with (BPES I) or without ovarian failure (BPES II) (Crisponi et al., 2001; Prueitt and Zinn, 2001). Ovaries from women with BPES II may contain primordial follicles and may appear as a ribbon of white connective tissue devoid of germ cells as detected at the time of laparoscopy and histology on ovarian biopsies (Fraser et al., 1988; Nicolino et al., 1995). By multitissue Northern blot analysis, human FOXL2 is expressed preferentially in the ovary. Two research groups have successfully disrupted the Foxl2 gene in mice (Schmidt et al., 2004; Uda et al., 2004). Ovaries from Foxl2 knockout mice contain oocytes surrounded mostly by flat granulosa cells that grow slowly compared to wild type ovaries (Schmidt et al., 2004; Uda et al., 2004). The follicular block appears at the primordial follicle stage. Expression of oocyte-expressed genes Figla, Gdf9 and Kit are not affected by the lack of Foxl2, and the ovarian morphology, as well as oocyte count, are also unaffected at birth. However, the granulosa cell markers, activin βA and Amh are diminished in Foxl2 mutant ovaries, although the significance of these findings is unclear. Since Foxl2 is a transcription factor, the identification of direct downstream target genes would be useful for understanding genetic pathways that Foxl2 controls. One such candidate gene is the steroidogenic acute regulatory (Star) gene that contains multiple putative forkhead DNA consensus sites. Foxl2 binds these sites directly (Pisarska et al., 2004). Foxl2 most likely acts as a repressor of Star, and dominant negative mutations within Foxl2 likely interfere with repressor activity of the wild type protein with subsequent derepression of Star (Pisarska et al., 2004). STAR is expressed in steroidogenic tissues including adrenal glands, granulosa and theca cells of the ovary and is responsible for the increased mobilization and delivery of cholesterol precursors to the inner mitochondrial membrane. However, derepression of Star transcription in mice that lack Foxl2 is probably not the cause of the early follicular block at the primordial/primary follicular stage.

Follicular growth

During follicular growth a remarkable increase in the size of the murine oocyte occurs–from <20 µm in the primordial follicle to greater than 70 µm in the antral follicle–with accompanying proliferation of somatic granulosa and differentiation of thecal cells. Several transcription factors expressed both in the somatic cells and oocytes are important for follicles to grow. Interestingly, many genes expressed early in the embryonic gonad formation and germline establishment are also expressed during follicular growth either in the oocyte or surrounding somatic cells. Oocyte-expressed transcription factors, Nr6a1, Pou5f1, Figla and Nobox are expressed throughout folliculogenesis and most likely play a critical function in accumulation of transcripts necessary for follicular growth (such as Gdf9, Bmp15) as well as transition from fertilization to embryonic development (Zar1, Mater).

There are many question with regards to the role of Pou5f1 in follicular growth and oocyte maturation. Does down-regulation of Pou5f1 at E14.5, at the time when female germ cells enter meiosis, lead to the observed apoptosis from E14.5–2 days after birth? What is the role of Pou5f1 throughout folliculogenesis? Pou5f1 expression throughout most of the germ cell life span may signify that Pou5f1 plays different roles at different stages of oocyte development, and targets of Pou5f1 regulation at different stages of germ cell development need to be determined. Availability of mice with a Gdf9 promoter driven Cre recombinase (Lan et al., 2004) may help to define the role that Pou5f1 plays in murine postnatal oogenesis, since Gdf9 is expressed postnatally in primordial follicles and through all stages of folliculogenesis including unfertilized eggs (Rajkovic et al., 2004).

Although Nr6a1 (Gcnf) knockouts indicate that it has an important role in early embryogenesis, its continual expression throughout folliculogenesis suggests that Nr6a1 may play important roles postnatally as well. Excision of the floxed Nr6a1 with Cre recombinase under the control of the zona pellucida 3 promoter causes female subfertility (Lan et al., 2003b). The zona pellucida 3 promoter is oocyte specific and expresses as early as primary follicles (Lan et al., 2003b, 2004). Although there is no difference in the size and weight of ovaries from conditional knockouts and controls, a disrupted estrous cycle in the knockout mice may account for the reduced number of litters per month. NR6A1 binds multiple DR0 elements in the Bmp15 and Gdf9 promoters, and Nr6a1 ablation causes overexpression of Bmp15 and Gdf9. It is thought that this overexpression of Bmp15 and Gdf9 accounts for reduced steroid hormone levels as well as the misexpression of Star, 3β-hydroxysteroid dehydrogenase I and 17-hydroxylase with subsequent prolongation of diestrus in the Nr6a1 conditional knockout. Although knockouts of Gdf9 and Bmp15 affect folliculogenesis and granulosa cell function, currently there are no transgenic overexpression models for Gdf9 or Bmp15. As such, it is hypothetical whether overexpression of these genes in conditional Nr6a1 knockouts leads to subfertility via this mechanism. It is clear, however, that Nr6a1 does affect the postnatal ovary and may have other target genes.

An oocyte-specific family of transcription factors, called Obox, is greatly up-regulated from primordial to primary follicles and beyond (Rajkovic et al., 2002; Yeh et al., 2002). Unlike Figla,Nobox, Pou5f1 and Nr6a1, Obox transcripts are not present in primordial follicles, but are first detected in growing primary follicles. Obox1, Obox5 and Obox6 are only expressed in the ovary and are abundantly present in unfertilized eggs. Homeodomain alignment shows that the OBOX1 homeodomain shares 92 and 56% similarity with OBOX5 and OBOX6, respectively. The role of these maternally expressed transcription factors is unknown, and future studies are needed to understand their function in folliculogenesis as well as potential functions in early embryogenesis.

Taf4b is another transcription factor preferentially expressed in germ cells (Su et al., 2002; Falender et al., 2005). Taf4b was initially isolated from a differentiated human B cell line and thought to play an important function in the immune response (Dikstein et al., 1996). Transgenic mice that lack Taf4b apparently have normal immune function, but abnormal postnatal development of the ovary (Freiman et al., 2001). Females that lack Taf4b are infertile. Folliculogenesis is arrested at the pre-antral stage, although it is unclear whether the initial pool of primordial follicles is different in the knockouts, or if Taf4b has an embryonic function in PGCs/oocytes. Microarray experiments indicate that many genes in the knockout ovary are down-regulated including inhibin βB, aromatase, inhibin βA, cyclin D2, follistatin, inhibin alpha and 17-β hydroxysteroid dehydrogenase. The down-regulation of these genes in the Taf4b knockout ovary most likely reflects smaller number of granulosa and theca cells in the knockouts and not necessarily direct targets of Taf4b. Careful investigations during ovarian development are necessary to evaluate when the knockout phenotype is first observed during ovarian development. These studies will be important to identify the stage in ovarian development that will be useful for future studies to identify targets of Taf4b. Studies on the expression of human TAF4B are lacking and the role of TAF4B in human reproductive pathology is unknown.

Many important transcription factors are expressed in the somatic cells that surround the oocyte. These include Foxo3a, Sf1, liver receptor homologue 1 (Lrh-1), Wt1 and Sox3. Some of these transcription factors such as Sf1, Lrh-1, Sox3 and Wt1 are also expressed early in embryogenesis as well as folliculogenesis. Foxo3a belongs to the family of forkhead transcription factors (Foxo) important in cell cycle arrest, apoptosis and embryogenesis (Hosaka et al., 2004). In contrast to the embryonic lethal Foxo1 knockout, the mouse knockout of Foxo3a is viable but anaemic (Castrillon et al., 2003; Hosaka et al., 2004). Fascinatingly, Foxo3a knockout females are initially fertile but by 15 weeks of age are infertile. The infertility in Foxo3a knockout mice appears to be due to widespread and premature activation of folliculogenesis before sexual maturity, followed by large scale atresia of the oocytes. Interestingly, loss of Foxo3a in mice also leads to spontaneous lymphoproliferation in lymphoid organs presumably by inhibiting NF-kappaB activation (Lin et al., 2004). It is unclear whether Foxo3a inhibits NF-kappaB activation in the ovary as well. In humans, FOXO3A transcripts exist in multiple tissues including brain, spleen, thymus and prostate as well as newborn ovaries (Castrillon et al., 2003). It is unknown when Foxo3a is first expressed in human or murine ovarian development. The accelerated loss of the ovarian reserve as well as the hormonal profile (elevated levels of serum FSH and LH) is reminiscent of premature ovarian failure in women (Castrillon et al., 2003). A subset of women with anaemia and ovarian failure may carry mutations in FOXO3A, although there are currently no known mutations in the FOXO3A gene associated with human disease.

In humans, SF1 is expressed when the indifferent gonadal ridge forms at 32 days post-ovulation and beyond (Hanley et al., 1999) including granulosa cells in the adult ovaries. Interestingly, conditional knockouts of Sf1, whereby Sf1 is presumably excised as early as E11 in the somatic cells of embryonic ovaries, did not affect embryonic ovarian development (Zhao et al., 2001, 2004). Sf1 excision was accomplished by driving Cre recombinase expression under the promoter of the anti-Mullerian hormone type 2 receptor gene (Zhao et al., 2004). The AMH receptor 2 promoter drives expression of Cre recombinase in somatic cells of the embryonic gonad as well as granulosa cells in the ovary and parallels Sf1 expression in the gonads. Female mice with a conditional excision of Sf1 contain antral follicles and haemorrhagic cysts but lack corpora lutea. Therefore, Sf1 conditional knockouts do not ovulate. The haemorrhagic cysts may result from disruption of estrogen production since estrogen receptor alpha and aromatase knockouts display similar haemorrhagic cysts (Lubahn et al., 1993; Britt et al., 2000). It is unclear, whether the number of primordial and primary follicles is the same between the wild type and conditional knockouts. It is also possible that Sf1 is expressed in cell types other than Amhr2 expression, and that the conditional knockout only represents partial Sf1 loss. This possibility is real given that Sf1 is also expressed in the theca cells of rodent ovaries (Ikeda et al., 1993; Falender et al., 2003). In addition, it is unclear how efficient Sf1 excision was in the somatic cells. Inefficient excision of the Sf1 allele will lead to mosaicism in Sf1 expression and result in a variable phenotype.

Wt1 is a zinc finger DNA-binding protein that associates with Sf1 to promote Amh expression (Nachtigal et al., 1998). Similar to Sf1, homozygous knockouts for Wt1 do not develop gonads before sexual differentiation (Kreidberg et al., 1993). WT1 is expressed in humans in the soma of the gonadal ridge (Hanley et al., 1999) and in mice, granulosa cells of adult ovaries (Hsu et al., 1995; Salmon et al., 2004). In addition to gonads and kidneys, Wt1 is important in the formation of mesothelial tissues such as diaphragm, heart, lung and spleen. It will be interesting to see whether a conditional knockout that inactivates Wt1 expression specifically in the gonad disrupts folliculogenesis.

Lrh-1, officially known as nuclear receptor subfamily 5, group A, member 2 (Nr5a2), is another orphan nuclear receptor expressed in multiple tissues including liver, pancreas, spleen and uterus. Lrh-1 is abundantly expressed in granulosa cells and corpora lutea of the rodent ovary (Falender et al., 2003; Liu et al., 2003). Lrh-1 shares 60% amino acid identity with Sf1, binds similar DNA sequences, and promotes transcription of steroidogenic enzymes Cyp11a1, Cyp17 and Star (Sirianni et al., 2002; Kim et al., 2004, 2005). Mice that lack Lrh-1 die at E6.5–7.5 and this phenotype precludes studying the effects of Lrh-1 deficiency on ovarian development and folliculogenesis (Botrugno et al., 2004; Pare et al., 2004). It is likely that Lrh-1 plays important function in folliculogenesis by exerting control over steroidogenesis and conditional knockouts will be necessary to ascertain this hypothesis in vivo. Cotransfection experiments in human granulosa cell tumors indicate that LRH-1 can bind the promoter of the gene encoding cholesterol side-chain cleavage cytochrome P450 enzyme (.). In addition, adenovirus express Lrh-1 mRNA appears to induce progesterone biosynthetic pathways during granulosa cell differentiation (Saxena et al., 2004). The above experiments indicate that Lrh-1 most likely plays important functions in the adult ovary that are likely to be distinct from Sf1.

Sox3 is a member of the HMG family of transcription factors that includes the testis determining gene Sry. Sox3 is presumed to function as a transcription factor by binding DNA through the HMG box. Sox3 is expressed in the brain, pituitary, as well as gonads. In the gonads, Sox3 is expressed in the somatic cells of the urogenital ridge, and RNA can be detected by PCR in the granulosa cells as well as oocytes of the adult ovaries, but no protein can be detected by immunohistochemistry (presumably due to low protein levels) (Weiss et al., 2003). Researchers previously hypothesized that a lack of Sox3 will reverse genotypic XX gonads to male gonads and that Sox3 is therefore critical in female gonadal differentiation (McElreavey et al., 1993). On the contrary, Sox3 knockout mice did not show sex reversal and developed anatomically and morphologically normal ovaries (Weiss et al., 2003). Histology revealed all stages of folliculogenesis with numerous small atretic follicles observed in prepubertal mice aged 4.5–5.5 weeks. Sox3 knockout females were subfertile and superovulation of Sox3 knockouts yielded fewer oocytes that were almost three times more likely to be deformed as compared to wild type controls. Fertilization of Sox3 knockout oocytes was normal and reproductive tract anatomy appeared normal. Sox3 is also transcribed in the pituitary, and although FSH and LH levels in the knockout mice appear unaffected, it is possible that pituitary dysfunction may play a role in the subfertile phenotype of Sox3 knockouts. The mechanism of substantial decrease in fertility in the Sox3 knockout females is not clear, and more work will be necessary to determine exact functions of Sox3 in folliculogenesis.

Relevance to human reproduction

Non-syndromic ovarian failure affects 1–2% of women. Causes of premature ovarian failure include autoimmune oophoritis, chemotherapy and radiation treatment but for most of the cases, the etiology is idiopathic and probably genetic. Premature ovarian failure can result from several different genetic mechanisms including X chromosomal abnormalities, autosomal recessive genes causing various types of XX gonadal dysgenesis and autosomal dominant genes with action restricted to premature ovarian failure. Autosomal recessive inheritance appears to be the major mode of transmission. Follicle stimulating hormone receptor (FSHR) is the only well characterized autosomal recessive gene that causes non-syndromic premature ovarian failure (Aittomaki et al., 1995). A mutation in the oocyte-specific growth factor, BMP15, was recently discovered that causes ovarian failure (Di Pasquale et al., 2004). The above mutations, however, account for a minority of cases with ovarian failure.

A growing list of human diseases is due to mutations affecting transcription factors (Online Mendelian Inheritance in Man, http://www.ncbi.nlm.nih.gov/omim/). The vast majority of these mutations affects multiple organs and cause syndromes. FOXL2 is an example of a transcription factor involved in syndromic ovarian failure. Mutations in the FOXL2 gene cause eyelid defects and ovarian failure. Possibly there may be additional mutations in the FOXL2 gene that cause only ovarian failure without eyelid defects. For example, a 30 base pair deletion that removes 10 of 14 alanines downstream of the winged helix/forkhead domain of the FOXL2 gene may be the cause of ovarian failure in a 17-year-old woman with small uterus and small ovaries without visible follicles (Harris et al., 2002) though she does not appear to have eyelid defects. Functional studies in vitro or in mouse models would be necessary to prove whether this deletion is the cause of her ovarian failure and a larger pedigree analysis is necessary to ascertain whether the penetrance of this mutation is variable.

There are currently no other transcription factors implicated in ovarian failure in 46,XX phenotypic females. The search for these genes in humans is hampered by a lack of patients with fortuitous autosomal translocations and small pedigrees to map the loci. In addition, ethical issues surrounding human germ cell research make it difficult to study genes important in ovarian differentiation and development. Ovarian pathologic conditions in transgenic mice closely resemble conditions observed in mutated human homologues, as exemplified in cases involving the FSHβ subunit and FSHR (Kumar et al., 1997; Dierich et al., 1998; Abel et al., 2000). Therefore, genes that cause ovarian failure in mice are candidate genes for ovarian failure in humans. Taf4b, Figla and Nobox, which were identified and first studied in mice, represent excellent candidate genes for non-syndromic primary ovarian failure because of their oocyte-specific expression pattern and restricted ovarian phenotype. In humans, hypomorphic mutations that partially affect Nobox and Figla function, may cause ovarian failure to occur later in life but before 40 years of age, (i.e. premature ovarian failure). It is also conceivable that ovaries in Turner syndrome lose oocytes because of perturbation in the expression of oocyte-specific transcription factors, and future experiments should examine expression of transcription factors such as Figla and Nobox in Turner syndrome ovaries. Another interesting group of genes expressed in the germline and encoding transcription factors include Nanog, Pou5f1 and Nr6a1. Nanog, Pou5f1 and Nr6a1 play important functions early in embryogenesis based on the respective mouse knockouts. Human null mutations in Nanog, Pou5f1 or Nr6a1 are likely to cause embryonic lethality, as observed in mouse knockouts. However, hypomorphic mutations in Nanog, Pou5f1 and Nr6a1 could possibly lead to selective loss of the germline and streak gonads.

Transcription factors expressed in the somatic cells of the gonads usually cause syndromes because of their expression and functional importance in multiple tissues. Mutations in the SF1 gene cause adrenal insufficiency and sex reversal, and it is unclear whether SF1 mutations in 46,XX females disrupt ovarian function. Mutations in the WT1 gene cause multiple congenital anomalies including streak gonads in 46,XX females. SOX3 resides on the X chromosome and in-frame duplication of 33 base pairs encoding for 11 alanines in a polyalanine tract of the SOX3 gene was associated with mental retardation and isolated growth hormone deficiency (Laumonnier et al., 2002). Duplication of SOX3 may also cause X-linked hypopituitarism (Solomon et al., 2004). It is not known whether phenotypic females with an XX karyotype and mutated SOX3 are infertile. Foxo3a is another candidate gene expressed in the somatic cells that may be responsible for premature ovarian failure in women with anaemia. It is also conceivable that specific mutations in transcription factors that are expressed in somatic cells causes only ovarian failure, without any other syndromic features. Transcription factors that are known to disrupt ovarian development in mice should be sequenced in patients with primary and premature ovarian failure.

The size of the germ cell pool, as well as the rate by which the germ cells are lost, determines the time of the menopause. We are far from being able to manipulate menopausal age. Identifying and studying the function and mechanisms of transcription factors that orchestrate germ cell formation and proliferation will be essential if we ever want to manipulate the germ cell reservoir. Oocyte-specific transcription factors and other oocyte-specific genes theoretically represent ideal pharmacological targets to regulate fertility, reproductive life-span and infertility in a tissue-specific manner without affecting other organ systems.

Conclusion

Germ cell biology is important in understanding pluripotency and will continue to make important contributions to the expanding field of stem cell biology. In much of the plant and animal kingdom, except for mammals, parthenogenesis can give rise to viable offspring. It is thought that the lack of viable offspring in mammals is due to imprinting, but recently, altering expression of two imprinted genes Igf2 and H19, enabled production of viable offspring through parthenogenesis in mice (Kono et al., 2004; Loebel and Tam, 2004). Clearly, even in mammals, oocytes may be the ultimate stem cells. Identification and functional characterization of transcription factors that regulate genesis of germ cells as well as growth and development of oocyte during folliculogenesis, will be essential not only to understand genetic pathways that give rise to oocytes, but also to understand mechanistically the origin of pluripotency. Oocytes, however, are limited in number in adult mammals. Postnatal oocyte stem cells may exist and replenish oocytes (Johnson et al., 2004), but such germline stem cells may be rare. In the future, efforts to derive eggs from somatic cells will benefit from understanding oocyte-specific transcription factors and their downstream target genes. Many transcription factors reviewed here are expressed in the somatic cells of the gonad. Genes such as Sf1, Wt1, Lhx9 and Emx2 are necessary for the survival of the gonad and testis differentiation but not critical for development of the female gonad. Only a few oocyte-specific transcription factors, such as Figla, Pou5f1, Nr6a1, Taf4b and Nobox, are known to play crucial roles at various stages during oocyte development. Other germ cell transcription factors await discovery; this will be aided by the availability of the sequenced mammalian genomes and ovarian transcriptomes. Identifying such germ cell specific transcription factors is necessary to understand the genetic cascades that lead to the formation of the female gonad and ovarian follicles. One place to start is by identifying the upstream regulators of Figla and Nobox, which may lead us to the molecular switches that control female sexual determination. The growing list of oocyte-specific genes indicates that non-syndromic ovarian failure is likely to be genetically highly heterogenous (Rajkovic and Matzuk, 2001; Dean, 2002). Ultimately, a better understanding of the genetics of oocyte development will help better control fertility, improve oocyte quality and perhaps generate functional eggs in vitro.

Currently there are no oocyte cell lines, and transgenic animals remain the mainstay of research into the mechanisms of germline origin and differentiation. Mouse models represent invaluable tools to study in vivo genetics of mammalian oogenesis and to understand the mechanisms that lead to infertility. Multiple mutagenesis approaches in mice will eventually give us a comprehensive list of genes with reproductive phenotypes including phenotypes that specifically affect early embryogenesis and may have relevance to cases of embryonic developmental arrests seen in IVF laboratories. Our understanding of the human ovarian and oocyte transcriptomes is still incomplete and efforts are necessary to generate good EST libraries from these difficult to obtain human tissues. These databases would be invaluable for comparing genes found to be part of genetic networks important in murine and human oogenesis. Ultimately, the better we understand transcriptional control in oogenesis, the more rational our approach will be to modulate human fertility.

Acknowledgements

Supported by NIH grants HD44858 and HD47514 to Dr. Rajkovic and a National Research Service Award (F32 HD46335–01A1) to Dr. Pangas.

References

Author notes

1Department of Pathology and 2Department of Obstetrics and Gynecology, Baylor College of Medicine, Houston, TX, USA