Human Reproduction Update Advance Access originally published online on January 22, 2007
Human Reproduction Update 2007 13(3):289-312; doi:10.1093/humupd/dml062
| ||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||
Molecular mechanisms of ovulation: co-ordination through the cumulus complex
Research Centre for Reproductive Health, School of Paediatrics and Reproductive Health, The University of Adelaide, Adelaide, South Australia 5005, Australia
1 Correspondence address. School of Paediatrics and Reproductive Health, The University of Adelaide, Adelaide, South Australia 5005, Australia Tel: +61 8 8303 4096; Fax: +61 8 8303 4099; E-mail: darryl.russell{at}adelaide.edu.au
 |
Abstract |
|---|
 TOP Abstract IntroductionConclusionReferences |
|---|
Successful ovulation requires that developmentally competent oocytes are released with appropriate timing from the ovarian follicle. Somatic cells of the follicle sense the ovulatory stimulus and guide resumption of meiosis and release of the oocyte, as well as structural remodelling and luteinization of the follicle. Complex intercellular communication co-ordinates critical stages of oocyte maturation and links this process with release from the follicle. To achieve these outcomes, ovulation is controlled through multiple inputs, including endocrine hormones, immune and metabolic signals, as well as intrafollicular paracrine factors from the theca, mural and cumulus granulosa cells and the oocyte itself. This review focuses on the recent advances in understanding of molecular mechanisms that commence after the gonadotrophin surge and culminate with release of the oocyte. These mechanisms include intracellular signalling, gene regulation and remodelling of tissue structure in each of the distinct ovarian compartments. Most critical ovulatory mediators exert effects through the cumulus cell complex that surrounds and connects with the oocyte. The convergence of ovulatory signals through the cumulus complex co-ordinates the key mechanistic processes that mediate and control oocyte maturation and ovulation.
Key words: ovulation / fertility / ovarian follicle / oocyte / ovary
|
Introduction |
|---|
|
TOPAbstractIntroductionConclusionReferences |
|---|
Successful ovulation is a complex process whereby ovarian follicles reactivate oocyte meiosis, create a rupture pore in the apical follicle wall and initiate tissue restructuring and differentiation to form the corpus luteum. These processes are fundamental to successful establishment of pregnancy, but importantly also impact on the developmental potential of resultant embryos. A cascade of events drive ovulation, initiated upon receipt by the follicle of a single trigger, the surge of gonadotrophins: luteinizing hormone (LH) and follicle-stimulating hormone (FSH) from the pituitary (Figure 1). In response, detailed changes in gene expression and follicular structure occur with overlapping control and interdependent consequences in the theca, granulosa, cumulus and oocyte compartments of the ovarian follicle. Systemic and local inputs co-ordinate with signals from the oocyte; thus ovulation is under multipartite control facilitating synchronization of oocyte maturation with release and permitting the selection of oocytes with full developmental competence for succession to the reproductive pool.
|
This review focuses on the dynamic tissue remodelling, cell–cell communication, intracellular signalling and transcriptional changes that occur in each compartment of the ovulating follicle. Most of the recent advances in understanding of the molecular mechanisms of ovulation have come from rodent models, with the most complete knowledge of genes and molecular pathways in ovulation derived from studies with mice and these are a focus of this review. However, a considerable contribution from studies in domestic animal and primate species is also important. Which model species is most appropriate for understanding critical genes and processes in human ovulation is currently unclear, making translational research in humans very important, and available information from such research is highlighted here.
Mural granulosa cells: targets and transducers of ovulatory signals
Pre-ovulatory follicles contain two distinct sublineages of granulosa cells that arise during folliculogenesis as the cell populations segregate upon formation of the fluid-filled follicular antrum. The mural granulosa cells line the follicle wall and reside very close to the basement membrane; and the cumulus cells are those directly adjacent to the oocyte (Figure 1). These two cell populations exhibit highly divergent responses during ovulation. Indeed, the direct response to the ovulatory LH surge is predominant in mural granulosa cells due to greater receptor levels than in cumulus cells. Mural granulosa layers express LH-receptors (LH-R), at levels typically an order of magnitude higher than those in cumulus cells. Binding of human chorionic gonadotrophin (hCG) within intact follicles is consistently reported to be at least 9-fold higher in mural granulosa cells than in cumulus of rat (Amsterdam et al., 1975
; Bortolussi et al., 1979; Lawrence et al., 1980), pig (Channing et al., 1981), hamster (Oxberry and Greenwald, 1982) and mouse (Wang and Greenwald, 1993b). The earliest studies utilized in vivo and in vitro binding of radiolabelled hCG to detect sites of action within the ovary and found that cumulus cells bound little or no hCG in contrast to mural granulosa cells within the same proestrus follicles (Amsterdam et al., 1975; Bortolussi et al., 1979; Channing et al., 1981; Oxberry and Greenwald, 1982; Wang and Greenwald, 1993a). Isolated cumulus–oocyte complex (COC) were similarly reported to have a 10-fold less hCG-binding capacity than isolated mural granulosa cells (Lawrence et al., 1980; Channing et al., 1981). Subsequent studies using in-situ hybridization showed LH-R mRNA expression is stratified within the rat pre-ovulatory follicle, with mural granulosa cells expressing the highest, levels and cumulus cells and oocytes having low to undetectable levels (Peng et al., 1991). An immunohistochemical study in rats supported mRNA analyses, demonstrating far greater LH/hCG receptor protein in theca, stroma and luteal tissue than mural or cumulus, but low levels of cell surface receptor were detected on cumulus cells after ovulation (Bukovsky et al., 1993). Confirming these in vivo observations of follicular expression, LH-R mRNA was low or undetectable in freshly isolated murine, bovine and equine cumulus cells but highly expressed in mural cells (Eppig et al., 1997; Goudet et al., 1999; Robert et al., 2003).
Postovulatory human cumulus cells undergo structural changes consistent with luteinization (Motta et al., 1995; Familiari et al., 2006), and this process includes the induction of LH-R. In a recent study, the relative levels of LH-R mRNA 36 h after hCG treatment in human granulosa versus cumulus cells were reported to be similar (Foong et al., 2006). However, LH-R mRNA levels prior to LH treatment or the responsiveness of human COC to FSH versus LH was not reported. This is essential information for understanding whether human, similar to animal models, have higher LH-R in the mural granulosa of pre-ovulatory follicles.
Numerous studies indicate that pre-ovulatory cumulus cells respond poorly, if at all, to direct LH exposure. Highly purified LH alone could not cause cumulus expansion in isolated mouse COCs (Eppig, 1979a, 1979b) in contrast to highly purified FSH. Furthermore, although cumulus expansion in vivo is induced by pure hCG treatment, this effect required the presence of other follicular compartments indicating the response in cumulus is indirect (Eppig, 1980). In rats, Hillensjo et al. (1981) reported that COC mucification is an FSH-specific effect and that FSH was 10-fold more potent than LH in stimulating progesterone (P) secretion by cumulus cells (Hillensjo et al., 1981). Subsequent experiments in mouse showed that FSH and LH can sequentially stimulate cumulus matrix synthesis, with FSH required prior to LH responsiveness, and that FSH or FSH plus estradiol induced LH-binding activity on cumulus cells, peaking after 8 h when cumulus expansion is well advanced in the mouse (Chen et al., 1994). Indeed most studies demonstrating LH action on COCs or maturation of oocytes in vitro include prior treatment or co-culture in the presence of FSH (Shimada et al., 2003; Foong et al., 2006). The preponderance of current evidence indicates that the ovulatory LH signal is received and responded to in the mural granulosa and theca cell layers and these may transmit a secondary signal to cumulus cells inducing myriad changes including the up-regulation of LH-R expression.
LH-induced intracellular signals
The LH surge triggers the activity of multiple intracellular signalling pathways in granulosa cells that culminate in altered transcriptional complexes mediating expression of ovulatory genes. The LH-R, a classical G
s G-protein-coupled receptor (GPCR) activates adenylate cyclase, resulting in a large intracellular cAMP increase that activates the cAMP-dependent serine kinase protein kinase A (PKA), a reaction which has been intensively studied (Marsh, 1976; Richards, 1994). Downstream of PKA, the cAMP regulatory element-binding protein (CREB) is phosphorylated at serine 133 (Mukherjee et al., 1996; Gonzalez-Robayna et al., 1999; Salvador et al., 2002; Russell et al., 2003b) and then recruits the CBP/p300 transcriptional coactivator (Arias et al., 1994). Phosphodiesterases are required to maintain tonic cAMP levels in responsive cells and PDE4D, in particular, regulates cAMP levels in granulosa cells (Tsafriri et al., 1996).
The LH surge is thought to stimulate additional signalling pathways, since LH treatment increased inositol triphosphate production in rat granulosa cells (Davis et al., 1986) as well as increased phospholipase C activity and intracellular calcium in non-ovarian cells (Gudermann et al., 1992; Zhu et al., 1994). Furthermore, cAMP and PKC activators, such as phorbol ester, are both required to emulate LH responses in cultured granulosa cells (Morris and Richards, 1995; Sriraman et al., 2003). Paradoxically studies have shown that LH-R activation (by hCG) does not activate PKC nor is a PKC inhibitor able to block hCG-mediated effects on rat granulosa cells (Salvador et al., 2002). Thus, although LH signalling likely involves activation of multiple parallel signal transduction pathways, PKC may not be one.
It is agreed that LH action on granulosa cells stimulates the extracellular regulated kinase (Erk1/2 or MAPK) pathway (Das et al., 1996; Maizels et al., 2001; Seger et al., 2001; Salvador et al., 2002; Choi et al., 2005), with activation of Erk being very rapid (Sela-Abramovich et al., 2005) and dependent on PKA (Salvador et al., 2002; Russell et al., 2003b). It is not yet entirely clear whether Erks are activated via direct phosphorylation by PKA or through some intermediate step such as inactivation of Erk phosphatases (Cottom et al., 2003), nor is the downstream function of Erks in granulosa cells, particularly in relation to ovulation, well understood; reports to date have primarily focused on their role in progesterone synthesis as cells luteinize (Seger et al., 2001; Dewi et al., 2002; Tajima et al., 2003). Recently, the transcription factor Rhox5 was identified as a transcriptional target of LH via Erks in ovulating follicles (MacLean et al., 2005). Ovulatory hCG also activates AP1 transcription factors of the Jun, Fos and Fra family (Sharma and Richards, 2000), consistent with general observations that Erk1/2 typically phosphorylate Fos, Myc, Elk-1 and Stat3 (Roux and Blenis, 2004). The effectors of Erk-mediated transcriptional regulation of granulosa cell gene expression during ovulation remain to be determined.
Follicular cGMP levels are increased during the course of ovulation, and at least one cGMP effector, the cGMP-dependent kinase cGKII, is also up-regulated at the time of ovulation by cAMP and Erk1/2 via progesterone receptor (PR) and epidermal growth factor (Egf) mediated pathways (Sriraman et al., 2005). Guanylyl cyclase receptors which generate cGMP, have been localized within the ovary (LaPolt et al., 2003). Expression of guanylyl cyclase-A receptor, and its ligand atrial natriuretic peptide (ANP), is up-regulated in mouse granulosa cells in response to ovulatory hCG (Sriraman et al., 2005); whereas guanylyl cyclase isoforms GC-B, GC-C and sGC are expressed at low levels and not hormonally regulated. Elevated production and activity of cGMP during the course of ovulation may alternatively occur via guanylyl cyclase activation by nitric oxide (NO), since nitric oxide synthase (NOS) in the theca produces NO during ovulation (Zackrisson et al., 1996). Clear effects of cGMP on granulosa cell gene expression, steroidogenesis and oocyte maturation are reported, but how this cascade impacts ovulation in particular is not known (LaPolt et al., 2003).
Studies using LH- or hCG-stimulated granulosa-lutein cells obtained from IVF patients confirm that PKA and PKC pathways are active in human granulosa cells (Freimann et al., 2004, 2005). More work is needed, however, to understand, as in rodents, the complex array of intracellular signalling pathways activated in response to the simple LH surge.
Periovulatory regulation of gene transcription
Signalling pathways induced by LH rapidly modify the transcriptional machinery of mural granulosa cells, reprogramming cellular function towards ovulation concurrent with luteinization. Extensive reprogramming of gene expression after the LH surge is achieved through modulation of a suite of transcriptional regulators, which in turn mediate effector gene transcription. A cohort of periovulatory genes with key roles in ovulation exhibits a common pattern of transcriptional induction involving binding of Sp1/Sp3 transcription factors to GC-box promoter elements. Sp1 and Sp3 are constitutively present in granulosa cells but exhibit sequence-specific, LH-regulated binding to gene promoters (Russell et al., 2003b). The signal transduction mechanism that modifies Sp1 binding activity in granulosa cells is uncertain, but in most tissues involves post-translational phosphorylation or glycosylation (Li et al., 2004).
Activity of Sp1/Sp3 complexes induces expression of additional transcription factors that further contribute to a cascade of ovulatory gene expression (Figure 2). Induction of PR in mouse and rat mural granulosa cells occurs in response to the LH surge through a cAMP/PKA-dependent mechanism, whereas phorbol ester (PKC activation) has a synergistic effect (Clemens et al., 1998; Sriraman et al., 2003; Sriraman and Richards, 2004). The cAMP-mediated induction of PR requires Sp1/Sp3 binding to multiple GC-box elements in the PR promoter (Sriraman et al., 2003) and consequent recruitment of additional transcription factors. In the rhesus monkey, PR is similarly induced in the periovulatory period (Stouffer, 2002). PR is present in human pre-ovulatory granulosa cells (Suzuki et al., 1994; Revelli et al., 1996; Chang et al., 2005). Two isoforms of PR are produced from distinct promoters on the same gene; PR-A is predominant in rat granulosa cells (Natraj and Richards, 1993) and is the most critical for ovulation, since mice lacking only PR-A have severely reduced ovulation (Mulac-Jericevic et al., 2000). Human ovaries express PR-A and PR-B isoforms, with PR-A most consistently found to predominate (Stouffer, 2003), however, the regulation of each isoform in ovulating follicles is not yet determined.
|
Early growth response-1 (Egr-1) is a transcription factor induced through cAMP/PKA and Erk signalling. Transcriptional induction of Egr-1 requires Sp1/Sp3 and phosphorylated-CREB binding to the proximal promoter. This is thought to lead to recruitment of a combinatorial transcription initiator complex involving CBP and other cofactors (Russell et al., 2003b
). Another transcription regulator Rhox5 is also regulated by the combination of Ets/GABP, Sp1 and Sp3 (MacLean et al., 2005).
These LH-induced transcription factors in turn co-ordinately activate transcription of a further suite of ovulatory effector genes. For instance, the protease cathepsin L is induced through Sp1/Sp3, PR, Egr-1 and CREB-mediated transactivation (Jaffe et al., 1989; Sriraman and Richards, 2004), and another ovulatory protease ADAMTS-1 is induced through Sp1/Sp3 and PR in combination with C/EBPß, and NF-1-like factor (Doyle et al., 2004). The Egf-like factor epiregulin is also regulated by Sp1/Sp3 binding its promoter in rat granulosa cells (Sekiguchi et al., 2002).
Thus, the cascade of diverse periovulatory gene expression in mural granulosa cells involves the induction and recruitment of a cohort of transcription factors that subsequently induce effector gene products, including ADAMTS-1 and cathepsin L proteases, Egf-like ligands (Egf-L) and others. These ovulatory genes exhibit a signature pattern of mRNA and protein expression: a rapid, transient increase after the LH surge followed by down-regulation. Recurring modifications of the transcriptional machinery interacting with these gene promoters mediate this characteristic expression. The common denominator is Sp1/Sp3 transcription factor binding, suggesting that these are universal mediators of periovulatory gene expression. The Sp factors recruit chromatin-remodelling complexes that can switch in response to changing cell context to modify transcriptional complex assembly, or alter promoter occupancy at neighbouring enhancer sites (Li et al., 2004) (Figure 2). This process has been well described for LH-R transactivation by Sp and nuclear receptor transcription factors in granulosa cells (Zhang and Dufau, 2003). Histone deacetylases recruited to Sp1/Sp3-bound promoters silence transcription, but can be rapidly exchanged for histone-acetylating complexes that open local chromatin structure to RNA polymerase access where Sp1/Sp3 are present. More detail is needed to better understand how LH-induced intracellular signals impact specific transcriptional complexes that distinctly regulate ovulation versus luteinization events and how this model relates to human periovulatory gene regulation.
Essential effectors of ovulation in mural granulosa cells
A number of mural granulosa cell genes play essential roles in ovulation. Mouse models, in particular, have yielded a great deal of information about the role of key genes in specific aspects of the ovulatory process. Of those identified to date, a few are expressed prior to the LH surge, such as the transcription co-regulator, RIP140, and the phosphodiesterase, PDE4D; however, the majority are transiently expressed during the discrete ovulatory process. Many are transcriptional regulators, including PR, Egr-1 and CAAT enhancer binding protein (C/EBPß). These co-ordinately mediate expression of ovulatory effector genes.
Transcriptional regulators: PR, RIP140, Egr-1 and C/EBPß
Progesterone and PR are very specific key regulators of follicular rupture. Antagonists of PR, including RU486 (Tsafriri et al., 1987; Loutradis et al., 1991), or inhibitors of P synthesis each reduces or completely blocks ovulation in several species including rats (Espey, 1998), sheep (Murdoch et al., 1986) and humans (Baird et al., 2003). Pre-ovulatory granulosa cells begin expressing progesterone in the days prior to the LH surge (Gore-Langton and Armstrong, 1994; Richards, 1994), whereas PR is acutely induced after the LH surge and is immediately activated by the high local concentration of ligand, before its expression is down-regulated. Mice with a targeted deletion of the PR gene (PRKO) exhibit normal follicle growth and luteinization, but a complete and specific block in ovulation (Lydon et al., 1996; Robker et al., 2000). Key PR-dependent genes expressed in the periovulatory period, but absent in PRKO mice and most likely contributing to their anovulation, include proteases ADAMTS-1 and cathepsin L (Robker et al., 2000).
The nuclear receptor co-regulator RIP140 is most highly expressed in pre-ovulatory granulosa cells and is rapidly down-regulated after the LH surge in the periovulatory period (White et al., 2000). RIP140 plays a broad role in the regulation of genes that are involved in follicle rupture, suggesting that this transcription cofactor is part of a cell-specific transcriptional complex formed in the granulosa compartment of ovulating follicles. Mice null for RIP140 exhibit apparently normal follicular growth and luteinization, but a block in ovulation due to the failed induction of numerous critical mural and cumulus cell genes, such as ADAMTS-1, epiregulin, amphiregulin, betacellulin, hyaluronan synthase-2(HAS-2), cyclooxygenase-2(COX-2), pentraxin-3(PTX-3) and tumour necrosis factor stimulator gene-6(TSG-6) (Tullet et al., 2005). Thus, RIP140 may regulate differentiation of follicles to pre-ovulatory maturity or mediate a single master event in the periovulatory response to LH. The close similarity of the PRKO phenotype to that of RIP140 null mice suggests that these two transcriptional regulators are in the same essential pathway.
The LH surge also initiates rapid and transient induction of transcription factors Egr-1 (Russell et al., 2003b) and C/EBPß (Sirois and Richards, 1993) in mural granulosa cells. Egr-1 null mice are infertile, primarily due to loss of LH secretion due to its role in LHß gene synthesis in the pituitary (Lee et al., 1996; Topilko et al., 1998). Thus, fertility may be controlled at multiple levels by Egr-1, including both pituitary and ovarian roles in the ovulatory process. It has been reported that ovarian LH-R expression is activated by Egr-1 and is disrupted in Egr-1 null mouse ovaries (Topilko et al., 1998; Yoshino et al., 2002). However, closer analysis of these mice indicates that after exogenous gonadotrophin replacement, ovulation remains impaired but that ovarian LH-R is normally expressed (Russell 2004). Null mutants for C/EBPß show oocytes entrapped in CL; however, evidence suggests these are among the few models with affected luteinization in addition to ovulation (Sterneck et al., 1997). Consistent with this it was shown that, C/EBPß may regulate steroidogenesis through the StAR gene (Christenson et al., 1999).
cAMP metabolism
The phosphodiesterase PDE4D, an important regulator of LH-mediated cAMP accumulation in granulosa cells (Tsafriri et al., 1996), is required for normal ovulation. PDE4D null mice exhibit dramatically reduced ovulation rates and litter size due to altered responsiveness of granulosa cells to the LH surge and reduced expression of ovulatory genes, including PR, cathepsin L and COX-2 (Jin et al., 1999; Park et al., 2003). Many follicles fail to reach pre-ovulatory stage due to premature luteinization. However, by in-situ hybridization, it was shown that the morphologically normal pre-ovulatory follicles, although reduced in number, still express ovulatory genes such as the Egf-L (Park et al., 2004). Thus, rather than rendering the granulosa cells unable to respond to ovulatory LH, the lack of PDE4D primarily permits accumulation of cAMP such that follicles exhibit premature luteinization and entrapment of oocytes in immature luteinized follicles.
Egf-L: amphiregulin, betacellulin and epiregulin
The Egf-L, epiregulin, amphiregulin and betacellulin, are very rapidly induced specifically in mural granulosa cells within 1–3 h after the LH surge and act as major secondary signals that transmit the ovulatory signal to the cumulus complex (Park et al., 2004) (Figure 1). Epiregulin and amphiregulin are also LH inducible in human granulosa cells (Freimann et al., 2004). Owing presumably to functional redundancy between epiregulin, amphiregulin and betacellulin, null mutant mice for each of these genes are fertile (Luetteke et al., 1999; Lee et al., 2004). Each of the Egf-L factors are synthesized as integral membrane proteins, and cleavage of the precursor forms is required for their interaction with cumulus-expressed receptors since protease inhibitors can block transmission of the LH signal to cumulus cells, whereas cotreatment with epiregulin reverses this block (Ashkenazi et al., 2005). Identification of the processing enzyme will reveal a key regulatory event in intrafollicular co-ordination of the ovulatory processes.
Extracellular matrix protease ADAMTS-1
The extracellular protease ADAMTS-1 was identified as LH- and PR-induced gene in mural granulosa cells in the rat (Espey et al., 2000) and mouse (Robker et al., 2000). This gene has been identified in bovine (Madan et al., 2003), equine (Boerboom et al., 2003), primate (Young et al., 2004) and human ovaries (Freimann et al., 2005) and may be associated with the aetiology or symptoms of anovulatory human polycystic ovarian syndrome (Jansen et al., 2004), but has yet to be studied in human normal ovulation. The latent pro-form of ADAMTS-1 is synthesized by the mural granulosa cells, but the secreted, mature form is selectively localized to the extracelullar matrix (ECM) and cell surfaces of the expanded cumulus complex (Russell et al., 2003a), suggesting this is where it mediates its principal effects. Two lines of mice with null mutation of the ADAMTS-1 gene have 70–90% reduced ovulation rates and 75% reduction in litter size (Shindo et al., 2000; Mittaz et al., 2004). From these studies, it is clear that ADAMTS-1 is an important mediator of LH and progesterone effects in ovulation across species. Substrates for ADAMTS-1 include versican, which is also produced by rodent mural granulosa cells and stimulated by the LH surge (Russell et al., 2003a) and has been identified in human ovaries (Eriksen et al., 1999; Relucenti et al., 2005). As with ADAMTS-1, versican is produced in mural granulosa cells, but the majority of protein in the periovulatory period is incorporated into the cumulus cell matrix.
Clearly, many mural granulosa cell gene products are required for successful and optimal ovulation. Many of these genes encode proteins that sense and respond to systemic inputs, including the LH surge. A cohort of essential ovulatory mediators are transcription factors that are up- and down-regulated by the LH surge and function to translate the ovulatory trigger into global reprogramming of gene transcription and ultimately cell function. Interestingly, a new group of effector gene products is emerging, including the Egf-L, versican and ADAMTS-1, which are produced by mural granulosa cells, but specifically translocate to and act within the COC. This combined endocrine and paracrine intrafollicular communication enables the co-ordination of maternal receptivity with oocyte maturation and ovulation.
Cumulus cells: conduit of ovulatory signals
The pre-ovulatory oocyte is surrounded by several cell layers, known as the cumulus granulosa cells, which are tightly connected to each other and the oocyte through intercellular membrane processes and gap junctions that facilitate exchange of glucose metabolites, small signalling molecules and ions. Oocytes depend on the cumulus cells for metabolism of glucose and supply of pyruvate for energy production (Biggers et al., 1967; Gardner et al., 1996; Preis et al., 2005). The cumulus cells exhibit hormone responsiveness and gene expression profiles distinct from those of the mural granulosa layers. In preovulatory follicles, cumulus and mural cells present abundant cell surface FSH receptors, but the cumulus mass and granulosa cells very close to the antrum show the greatest proliferative response to FSH during follicle growth (Robker and Richards, 1998). The cumulus are the cells most directly exposed to mitogenic factors secreted by the oocyte (Erickson and Shimasaki, 2000). Oocyte-secreted factors (OSF) belonging to the Tgfß superfamily establish a morphogen gradient that promotes the cumulus cell characteristics including enhanced cell proliferation, and repression of LH-R and PR gene transcription, and progesterone biosynthesis (Eppig et al., 1998; Elvin et al., 1999a; Gilchrist et al., 2004). The rate of proliferation of these cells has been suggested to correlate with implantation potential in human assisted reproduction (Gregory, 1998). Following the ovulatory LH surge, cumulus cells respond with a unique pattern of gene induction leading to production and stabilization of a local ECM that envelopes the COC. This process, known as cumulus expansion or mucification, is dependent on a specific cascade of intracellular signals and ECM gene expression within the COC and plays a key role in ovulation (Chen et al., 1993; Hess et al., 1999; Fulop et al., 2003; Salustri et al., 2004) and subsequent fertility success (Tanghe et al., 2003; Somfai et al., 2004; Yang et al., 2005).
Intracellular signalling in periovulatory cumulus cells
Cumulus cells respond rapidly and change dramatically after the gonadotrophin surge. The program of gene induction in the cumulus cells responding to the gonadotrophin surge is potentially mediated, at least in part, by high FSH that accompanies the LH surge. In women, as in other mammalian species, FSH levels surge with similar timing kinetics to LH at the midcycle (Hoff et al., 1983). Local paracrine effectors including Egf-L from granulosa cells, interleukin-1 from the thecal compartment and oocyte-derived signals also contribute to cumulus cell responses (Figure 1).
Cumulus response to endocrine and paracrine stimuli
Follicle stimulating hormone
In isolated COCs in vitro, FSH induces periovulatory changes in cumulus cells and oocytes (Su et al., 2002) and in vivo very high concentrations of pure FSH have even been found to initiate oocyte maturation and ovulation in rodents (Tsafriri et al., 1976b; Galway et al., 1990; Wang and Greenwald, 1993a) and primates (Zelinski-Wooten et al., 1998). However, the levels of FSH used to activate COC expansion in vitro are supraphysiological, and in animal models or human IVF cycles, levels of FSH, which reflect physiological circulating levels during the ovulatory surge, do not enhance molecular events associated with ovulation or oocyte development (Tsafriri et al., 1976b; Vermeiden et al., 1997). Although periovulatory gene expression is induced by FSH in the cumulus, it has been widely shown that ovulation and oocyte maturation are efficiently induced in vivo by pure exogenous LH preparations that mimic the levels of an endogenous LH surge. These observations indicate that FSH may play a minor role, but is not essential, whereas intrafollicular responses to the LH surge mediate ovulation.
Since FSH is effective in stimulating ovulatory cumulus gene expression, it is widely used to study molecular aspects of this process. The FSH receptor is a GPCR closely related to LH-R that activates adenylate cyclase mediating cAMP synthesis and resulting in PKA-activated intracellular signalling (Figure 3). In addition, FSH independently, activates the PI3K/Akt pathway (Gonzalez-Robayna et al., 2000; Zeleznik et al., 2003) and recruits the 14-3-3 adaptor protein (Cohen et al., 2004) to mediate cell survival and proliferation in granulosa cells. Through the Akt pathway, FSH also increases transcription of the hypoxia inducible factor-1 (Alam et al., 2004) leading to the induction of expression from genes such as vascular endothelial growth factor (VEGF), a critical mediator of angiogenesis during ovulation/luteinization (see below). Most importantly, FSH activates ERK signalling in granulosa (Das et al., 1996) and in cumulus cells (Su et al., 2002) and Erk activity is requisite in cumulus cells but not in oocytes for COC expansion and oocyte meiotic maturation (Su et al., 2002). Inhibitors of Erk, p38 kinase or Akt block COC expansion induced by FSH or cAMP treatment (Ochsner et al., 2003a), indicating that each signalling kinase pathway converges to mediate important periovulatory cumulus gene expression.
|
Prostaglandins
Cumulus cells in ovulating COC produce prostaglandins via induction of COX-2 and other prostaglandin synthetic enzymes. Induction of COX-2 in COC can be stimulated by Egf in cows (Nuttinck et al., 2002
) and mice (Park et al., 2004) or by FSH through cAMP/PKA activation (Joyce et al., 2001). Pharmacological activators of cAMP synthesis induce COX-2, but this is sensitive to inhibition of p38-MAPK or Erk1/2 (Ochsner et al., 2003a), indicating that Erk-mediated transcriptional induction, distinct from or in conjunction with the classical cAMP pathway, is required for COX-2 expression. The COX-2 gene promoter has been studied in detail and shown in mouse granulosa cells to be dependent on oocyte-derived Gdf-9 (Elvin et al., 2000) and induced through an E-box element that interacts with the upstream stimulatory factor (USF) transcription factor in a manner regulated by PKA (Morris and Richards, 1996; Sayasith et al., 2004; Sayasith et al., 2005). The predominant prostaglandin produced is prostaglandin (PGE2) which interacts with the PGE2-receptor (EP-2) to induce changes in cumulus gene expression (Segi et al., 2003) critical for COC expansion and ovulation (Hizaki et al., 1999). Prostaglandins made by the cumulus cells potentially mediate important signalling effects through other receptors also, since disruption to fertility in EP-2 null mice is far less severe than in COX-2 nulls that produce no prostaglandins (Table I). Very recently, the PGE2 receptor, EP-4, was also shown to be induced in periovulatory cumulus cells and to participate with EP-2 in oocyte maturation, cumulus cell survival and matrix production (Takahashi et al., 2006). Both PGE2 receptors are GPCR that stimulate cAMP production; thus in cumulus cells, prostaglandin synthesis establishes an autocrine signalling loop synergizing with FSH and reinforcing cAMP/PKA and Erk-dependent gene induction in cumulus cells (Figure 3). This loop of cAMP-mediated PKA activation in cumulus cells may be controlled by the regulator of G-protein signalling (RGS-2), which has a temporal and spatial pattern of expression very similar to COX-2 and EP-2 in periovulatory COCs (Ujioka et al., 2000). Adenylate cyclase activity can be either augmented or inhibited by RGS-2 (Kehrl and Sinnarajah, 2002), and its role in COC has yet to be more fully investigated. Human granulosa-lutein cells express EP-2 (Harris et al., 2001)and secrete PGE2 (Chandras et al., 2004), indicating that a similar signalling loop likely exists in human ovulation.
|
Epidermal growth factor-like ligands
Egf-L (amphiregulin, betacellulin and epiregulin) from LH-stimulated mural granulosa cells also activate periovulatory COC gene expression (Espey and Richards, 2002
; Park et al., 2004; Rimon et al., 2004). These ligands activate the Egf-receptor tyrosine kinase and subsequently the Erk activity (Kansra et al., 2004) (Figure 3). Cumulus cell expression of ECM genes, COC expansion and ovulation are blocked by pharmacological inhibition of the Egf-receptor tyrosine kinase or downstream Erk function (Lorenzo et al., 2001; Su et al., 2002; Park et al., 2004; Tsafriri et al., 2005). Ovulation induction by hCG is also blocked by inhibition of Egf-receptor activity; oocytes are entrapped in follicles surrounded by compact cumulus cells and cumulus matrix gene induction in cultured follicles is inhibited (Ashkenazi et al., 2005). Thus, both FSH and Egf-L signals promote cumulus periovulatory gene expression through activation of Erk signalling.
Androgen
Cumulus cells of several species, including rat (Pelletier et al., 2000; Szoltys et al., 2003) and human (Hickey et al., 2005), express the androgen receptor and thus are thought to respond directly to androgens produced by theca cells. There is limited evidence for the consequences of androgen action in ovulation; however, in female mice lacking androgen receptor a minor, albeit significant reduction in ovulation rate was reported (Hu et al., 2004). Furthermore, ovulatory genes, including PR, HAS2 and TSG-6, were not induced, and the cumulus complexes ovulated were reported to have a disorganized structure, suggesting a role for androgen receptor in COC expansion and the ovulatory process (Hu et al., 2004).
Interleukin-1ß
Other cells within the theca interna, possibly leukocytes surrounding periovulatory follicles, may also contribute to cumulus expansion in vivo (Figure 1). These cells produce IL-1ß (Jasper and Norman, 1995; Wang et al., 1997; Minge et al., 2005) in response to LH (Kol et al., 1999). The IL-1ß receptor activates a receptor-associated kinase, again resulting in Erk activation (Eder, 1997). Thus, in vivo, paracrine signals from the theca also transmit signals to the cumulus and co-ordinate the cumulus response to the LH surge. Hyaluronan (HA) synthesis and cumulus expansion are induced by IL-1ß (Kokia et al., 1993; Singh and Armstrong, 1997), and cumulus expansion and ovulation can be blocked by IL-1 antagonism in rodents (Simon et al., 1994). In humans, IL-1ß and IL-1 have been found to be induced by gonadotrophins and serum levels correlate with successful fertilization (Mendoza et al., 1999) or implantation in assisted reproductive cycles (Karagouni et al., 1998). However, subsequent studies evidenced either no correlation (Barrionuevo et al., 2000) or reverse correlation (Mendoza et al., 2002).
Extracellular matrix
The COC matrix itself may have important effects in mediating or modulating cumulus intracellular signalling. The HA receptor, CD44 is expressed on cumulus cells, induced after the LH surge in many species including humans (Ohta et al., 1999; Yokoo et al., 2002; Schoenfelder and Einspanier, 2003) and can transduce multiple intracellular signals (Bourguignon, 2001) via cytoskeletal changes, which transactivate the Egf-R family member ErbB2 and activate Erk. Proteoglycan components of the COC matrix, such as CD44 and versican, likely bind growth factors and modulate signalling. However, CD44 null mice have normal fertility (Schmits et al., 1997; Protin et al., 1999), suggesting the involvement of redundant HA receptors such as RHAMM, which is present in bovine COC (Turley et al., 2002). Actions of the proteoglycan inter-alpha trypsin inhibitor (II) found in the COC matrix include modulating kinases Erk and Akt to mediate gene expression changes (Kobayashi et al., 2001). Microarray analysis identified 29 genes up-regulated and 5 genes down-regulated 12 h after hCG in COC of mice that lack II relative to wild-type controls (Suzuki et al., 2004), suggesting that it regulates gene expression changes in addition to being a key matrix component.
To date, transcription regulators modulating periovulatory cumulus cell gene expression have not been studied in detail due to lack of an appropriate in vitro model for such studies. Repeatedly it has been reported that signals that activate periovulatory cumulus gene expression act through the Erk kinase family. The transcription factors most commonly phosphorylated by Erk kinases include AP-1 factors, signal transducer and activator of transcription (STAT) and serum-responsive transcription factors. However, in addition, cumulus-cell-specific gene expression requires oocyte-derived signals.
Cumulus cell response to signals from the oocyte
Mature oocytes, although transcriptionally inactive, secrete signals that play a key permissive role in periovulatory cumulus gene expression in mice (Eppig et al., 2002). This oocyte-derived activity, originally referred to as cumulus-expansion-enabling factor (CEEF), is secreted by mouse, pig and cow oocytes (Eppig et al., 1993; Singh et al., 1993; Vanderhyden et al., 1993; Gilchrist et al., 2004). This ability is linked to the oocytes' developmental competence (Eppig et al., 1993). A range of experiments seeking to identify the CEEF in mice suggests that Gdf-9 as well as other Tgfß superfamily members can support cumulus ECM gene expression (Vanderhyden et al., 2003; Dragovic et al., 2005). However, the identity of the true oocyte-secreted factor(s) facilitating COC expansion remains controversial, since antibody neutralization of Gdf-9 secreted by oocytes only partially blocks COC expansion (Dragovic et al., 2005), whereas suppression of Gdf9 mRNA in mouse oocytes was more highly effective (Gui and Joyce, 2005).
The specific roles for Gdf-9, BMP-15 and perhaps other Tgfß family members such as BMP-6, in ovulation have also been difficult to fully interpret because of species variations and their critical role early in folliculogenesis. In pigs and cows, the oocyte does not need to be present at the time of cumulus expansion, but in cows, BMP-15 is a necessary cumulus cell survival factor during oocyte maturation and COC expansion (Hussein et al., 2005). The impact of growth factor signals from the human oocyte has not been directly tested due to obvious constraints on available COCs for analysis. Thus, the model system that best represents the human is unknown.
In all species, the initiator of ovulatory cumulus gene expression is clearly not Gdf-9 and/or BMP-15, since these factors are constitutively synthesized by oocytes and affect follicular somatic cell development throughout folliculogenesis (Elvin et al., 1999b; Juengel et al., 2002), including in humans (Di Pasquale et al., 2004), without activating matrix production. It is thought that in all species, oocyte signals specify the cumulus cell lineage which orchestrates the cumulus-specific responses after the LH surge (Erickson and Shimasaki, 2000; McNatty et al., 2005). Therefore, by producing Gdf-9 and BMP-15, the oocyte controls the differentiation of its immediate environment and hence its ovulation.
Signalling by Gdf-9 in cumulus cells is achieved by binding to the BMP type-II receptor (BMPRII), followed by recruitment and activation of the type-I receptor, Alk5 (Mazerbourg et al., 2004) (Figure 3). Thus Gdf-9 uses similar intracellular signalling to Tgfß. Activated Alk5 in human granulosa-lutein cells phosphorylates transcription factors SMAD2 and SMAD3, which dimerize with SMAD4, translocate to the nucleus and bind to SMAD-response elements in gene promoters containing the core nucleotide sequence CAGA, such as the inhibin-B promoter (Kaivo-Oja et al., 2005). Not surprisingly, factors that share the capacity to activate SMAD2/3 in cumulus cells are those previously shown to share the CEEF activity.
The mRNA and protein for SMAD2 and 4 are induced in whole mouse ovary by FSH (Gueripel et al., 2004), indicating that the pre-ovulatory follicle is primed for a robust response to Gdf-9 and other Tgfß-related ligands in the periovulatory phase. Knockout of the SMAD3 gene results in anovulatory infertility primarily due to defects in follicle growth (Tomic et al., 2004), a similar but not identical phenotype to Gdf-9–/–. This may indicate that the Gdf-9 functions partially but not exclusively through SMAD3. SMAD2 and SMAD4 knockout mice die in embryonic development. Whether SMADs bind directly to promoters of each cumulus-expressed periovulatory gene or regulate these genes through indirect means is not known. It has been suggested that Gdf-9 also activates Erk signalling in cumulus (Su et al., 2003), but since other Erk activators such as Egf cannot substitute for the signal from oocytes, SMAD 2/3 activation is more likely to be the requisite mechanism of Gdf-9-mediated cumulus gene expression.
Essential cumulus gene expression and regulation
The set of genes expressed in periovulatory cumulus cells are critical for normal rates of ovulation and fertility (Table I). Among the earliest induced genes is COX-2, the rate-limiting enzyme in prostaglandin production (Sirois and Richards, 1992; Wong and Richards, 1992). COX-2 production is rapidly induced within 2 h of an hCG stimulus in rodent mural and cumulus cells (Sirois et al., 1992), but expression is higher and persists longer in the COC (Elvin et al., 1999a; Joyce et al., 2001; Segi et al., 2003). In larger species with longer periovulatory periods, this induction happens later and COX-2 has been proposed to be part of a molecular clock that sets the species-specific timing of ovulation (Sirois et al., 2004). This suggestion is supported by studies in the macaque monkey, but in this model, the timing of prostaglandin secretion is dependent on induction of COX-2, microsomal prostaglandin-E synthase and most particularly on expression and post-translational activation of phospholipase A2 in granulosa cells (Duffy et al., 2005). This mechanism places more stringent control on the process required for prostaglandin synthesis and the timing of ovulation. It is not known whether a similar complex regulatory mechanism operates in human ovulation.
Null mutation of the mouse COX-2 gene results in defective cumulus expansion, reduced ovulation rate and infertility (Dinchuk et al., 1995; Lim et al., 1997; Davis et al., 1999; Ochsner et al., 2003b). This is explained by the reduced expression of PGE2/EP-2-dependent genes such as TSG-6 in COX-2–/– as well as EP-2–/– mice (Ochsner et al., 2003a). Likewise, treatment of mice and rats with COX-2 inhibitors indomethacin or NS398 also blocks ovulation and TSG-6 expression (Mikuni et al., 1998; Yoshioka et al., 2000; Espey and Richards, 2002). Indomethacin treatment also caused delayed ovulation by 2–12 days in women, (Athanasiou et al., 1996) and another COX-2 inhibitor, rofecoxib, had a similar effect (Pall et al., 2001). These studies collectively demonstrate that prostaglandin production via COX-2 enzyme is a key event in the cumulus gene expression cascade and vital to ovulation across mammalian species.
Genes encoding COC matrix components are also induced in cumulus cells by FSH or Egf in conjunction with oocyte signals including HAS-2, the enzyme that synthesizes the HA backbone of the cumulus matrix, (Eppig, 1979a, 1981; Tirone et al., 1997; Elvin et al., 1999a; Dragovic et al., 2005) and HA binding proteins TSG-6 (Fulop et al., 1997; Yoshioka et al., 2000) and pentraxin-3 (PTX-3) (Varani et al., 2002; Salustri et al., 2004). The inhibitor of BMP signalling Gremlin is expressed in pre-ovulatory mural and cumulus cells, but maintained only in cumulus cells after ovulation induction through Gdf-9 action (Pangas et al., 2004). The need for oocyte factors as well as FSH or Egf for induction of these cumulus specific genes has been characterized, but the full regulatory mechanism in cumulus cells has not been studied in detail. Interestingly, induction of TSG-6 in cumulus cells is dependent on autocrine/paracrine prostaglandin action, yet HAS-2 expression is not (Ochsner et al., 2003b). However, PGE2 treatment of isolated COC can induce HAS-2 mRNA (Eppig, 1981). These observations indicate that HAS-2 induction requires cAMP-PKA and/or Erk signalling in addition to Gdf-9/BMP-15 and that diverse regulatory mechanisms control different periovulatory genes in cumulus cells in vivo.
Cumulus matrix function
Appropriate formation of the expanded cumulus matrix is critical for ovulation. In fact, the success of follicular rupture as well as fertilization is exquisitely sensitive to perturbations in the composition and hence functional capacity of the cumulus matrix. This has been clearly established in mice by gene ablation of COC matrix proteins (Sato et al., 2001; Zhuo et al., 2001; Varani et al., 2002; Fulop et al., 2003; Salustri et al., 2004) (Table I). Other approaches that inhibit the expansion, organization or function of cumulus matrix proteins support the observations from knockout studies (Chen et al., 1993; Hess et al., 1999; Ochsner et al., 2003a).
Composition of expanded cumulus matrix
The backbone of the expanded cumulus matrix is HA, a large disaccharide chain common to many extracellular matrices including cartilage where it provides strong osmotic force and viscoelastic properties. Synthesis of HA requires glucose as a substrate for the hexosamine pathway, which generates the disaccharide subunits of HA. Glucose uptake and glycolytic activity in cumulus cells is markedly stimulated by the LH surge in rodents, cows and humans (Tsafriri et al., 1976a; Zuelke and Brackett, 1992; Roy and Terada, 1999; Sutton et al., 2003). Glucose flux through the hexosamine pathway in bovine COC increases during oocyte maturation (Sutton-McDowall et al., 2004). Strands of HA matrix are stabilized and organized into a specific functional structure by cross-linking proteins that bind HA through canonical link-protein-like modules. Protein cross-linkers play a fundamental role of stabilization of the HA matrix, as demonstrated by its dissolution by treatment with protease (Cherr et al., 1990). Organizers of the COC matrix, which bind HA directly, include TSG-6 (Carrette et al., 2001; Mukhopadhyay et al., 2001; Ochsner et al., 2003a), synthesized by cumulus cells and versican synthesized by the mural granulosa cells (Russell et al., 2003c) and the heavy chains (HCs) of II, a serum protein secreted by hepatocytes (Steinbauch and Loeb 1961). Circulating II is a trimeric protein complex comprising two HC subunits joined through chondroitin sulphate bridges to bikunin (Bost et al., 1998). It circulates in high concentrations and enters the ovulating ovary in blood plasma as a result of increased follicular vascular permeability during ovulation (Castillo and Templeton, 1993). The HC binds HA in the expanding COC ECM, whereas the bikunin subunit remains in solution and takes no further part in the COC matrix formation. Cartilage link protein may also be a product of FSH or IGF-1-stimulated mural granulosa cells (Sun et al., 2003) and is found in pre-ovulatory follicles of rodents and humans (Kobayashi et al., 1999). Exogenous link protein enhances volumetric COC expansion in vitro, but is not an essential matrix component (Sun et al., 2002). PTX-3 is an immune protective and anti-inflammatory protein that does not possess an HA-binding motif, but localizes intensely to the cumulus ECM in rodents and human. In addition, fibronectin, tenascin-C and laminin have been reported in the human COC matrix (Familiari et al., 1996).
Function of HA cross-linking proteins in cumulus matrix
Evidence suggests that the function of the cumulus matrix is dependent on a specific structure produced through the binding of HC and HC-TSG-6 complexes to HA (Carrette et al., 2001; Mukhopadhyay et al., 2001; Fulop et al., 2003; Ochsner et al., 2003a) (Figure 4). Formation of the appropriate covalent complex of II-HC to HA requires an intermediate interaction with TSG-6 (Chen et al., 1996; Carrette et al., 2001; Jessen and Odum, 2003). Loss of either TSG-6 or II synthesis results in almost identical phenotypes: HA is produced by cumulus cells but not stabilized, resulting neither in matrix aggregation nor morphological COC expansion in vitro or in vivo. As a result, fewer (<50%) oocytes are ovulated (Sato et al., 2001; Zhuo et al., 2001; Fulop et al., 2003). Of the few oocytes that are ovulated in TSG-6–/– mice, no fertilization or implantation has been observed in natural matings, but fertilization under IVF conditions has not been assessed (Fulop et al., 2003). A TSG-6 neutralising antibody can prevent cumulus matrix assembly in vitro (Ochsner et al., 2003a). Treatment of mice with inhibitors of the HA synthetic pathway (Chen et al., 1993) or short HA oligosaccharides that adsorb TSG-6 (Hess et al., 1999) also block COC expansion and ovulation. These data combined with TSG-6–/– studies indicate that TSG-6 as well as II interactions with HA are required for the formation of the COC matrix.
|
The role of PTX-3 is different; mice with PTX-3 null mutation achieve COC matrix expansion in vivo, but the cumulus cells appear abnormally distributed. Oocytes are ovulated but fail to fertilize because of the rapid degradation and loss of the COC matrix during or after release from the follicle (Varani et al., 2002
; Salustri et al., 2004). Thus, PTX-3 appears to be required for matrix stability and retention. In vivo fertilization was <2% of the normal rate in one PTX-3–/– mouse line (Varani et al., 2002) and completely deficient in another (Salustri et al., 2004). Under IVF conditions, fertilization of PTX-3–/– oocytes occurred at the normal rate. This is expected since it is known that the need for the COC matrix can be circumvented in vitro by the use of high sperm concentrations; it also indicates that there is no profound oocyte deficiency in the absence of PTX-3 (Salustri et al., 2004). PTX-3 protein forms homo-decameric complexes that may interact with HA-bound TSG-6 molecules to form anchoring complexes