Life and death of female gametes during oogenesis and folliculogenesis
The vertebrate ovary is an extremely dynamic organ in which excessive or defective follicles are rapidly and effectively eliminated early in ontogeny and thereafter continuously throughout reproductive life. More than 99% of follicles disappear, primarily due to apoptosis of granulosa cells, and only a minute fraction of the surviving follicles successfully complete the path to ovulation. The balance between signals for cell death and survival determines the destiny of the follicles. An abnormally high rate of cell death followed by atresia can negatively affect fertility and eventually lead irreversibly to premature ovarian failure. In this review we provide a short overview of the role of programmed cell death in prenatal differentiation of the primordial germ cells and in postnatal folliculogenesis. We also discuss the issue of neo-oogenesis. Next, we highlight molecules involved in regulation of granulosa cell apoptosis. We further discuss the potential use of scores for apoptosis in granulosa cells and characteristics of follicular fluid as prognostic markers for predicting the outcome of assisted reproduction. Potential therapeutic strategies for combating premature ovarian failure are also addressed.
KeywordsGranulosa cell Follicular fluid Follicular atresia Apoptosis Caspases Premature ovarian failure Autophagy Neo-oogenesis Gap junctions
Oocyte cell death and its regulation has always been one of the most active research areas in vertebrate reproductive physiology and developmental biology, and they have been inspiring scientists for over a century. The occurrence of cell death in the ovary is by no means a new discovery. Walther Flemming in 1885 made one of the first observations. He described a process he named “chromatolysis” as being responsible for the degeneration of inner epithelial cells (granulosa cells, GCs) in rabbit ovarian follicles. The cells dying by chromatolysis were characterized by cellular and nuclear condensation followed by fragmentation. Now it is well-known that these morphological characteristics perfectly match the typical features of apoptosis, which were described almost nine decades later by Kerr et al. . Two decades after Kerr, Gavrieli et al.  identified the oligonucleosomal DNA fragmentation that is pathognomonic for apoptosis in GCs of atretic mouse follicles. Since then, the instrumental role of granulosa cell apoptosis in follicular atresia was confirmed for many species, including humans. In parallel with the unraveling of the signaling pathways and the contributing gene products involved in other model systems, researchers analyzed the involvement of caspases, Bcl-2 family members, and death receptors and ligands in the development of the fetal gonad and in postnatal folliculogenesis. Independently of cell death research, more than 50 years ago investigators identified mouse genetic mutations affecting follicular endowment and thus fertility . Genes with well-known roles in PCD are now included in the list of genes affecting fetal endowment. In 1951 Sir Solomon Zuckermann  launched the central dogma in reproductive medicine on the finite stock of oocytes in mammalian ovaries and in doing so closed the discussion on the possibility of neo-oogenesis, which started at the beginning of the century .
Zuckermann’s dogma stimulated researchers to study the signaling pathways of programmed cell death in the ovary, with prospects of developing clinical applications, such as the development of new contraceptives, the prevention of spontaneous or chemotherapy-induced premature ovarian failure, and the optimization of in vitro fertilization (IVF) protocols. This dogma was recently questioned [6, 7], directing several research groups towards studying the possibility of oocyte neo-oogenesis in the adult ovary. In this way the story of the ovary is not only one of understanding cell death mechanisms, but also comprehending how new life can be generated. This review will discuss recent findings on the following subjects: (1) the role of programmed cell death in prenatal differentiation of the primordial germ cells and in postnatal folliculogenesis, including a discussion of the functional homology between invertebrate nurse cells and the vertebrate GCs; (2) the controversial issue of neo-oogenesis; (3) the contribution of pro-apoptotic and anti-apoptotic molecules to the regulation of granulosa cell apoptosis as the main driving force of follicular atresia; (4) the contribution of cell death associated with autophagy to follicular atresia; (5) the detection of apoptosis and the analysis of follicular fluid as tools for predicting the outcome of assisted reproduction; (6) the therapeutic strategies for combating premature ovarian failure.
Programmed cell death is instrumental in prenatal differentiation of primordial germ cells
The process of oocyte formation and follicular endowment includes seven prenatal steps that start with (i) the generation of primordial germ cells (PGCs), (ii) their migration to the future gonads and their concurrent proliferation, (iii) colonization of the gonads, (iv) differentiation of PGCs into oogonia, (v) proliferation of the oogonia, (vi) initiation of meiosis and, finally, (vii) arrest at the diplotene stage of prophase I of meiosis. A phenomenon that is common to vertebrates and invertebrates is that PGCs often arise in one portion of the embryo and migrate relatively long distances to where the gonad will ultimately be formed . During this migration, the founder PGC population (originally 45 in mice) proliferates. A growing number of spontaneous and transgene mouse models of ovarian failure with defects in one or more of the seven developmental stages serve as important tools in the analysis of programmed cell death mechanisms in the fetal ovary (for detailed review see [9, 10] and the databases presented elsewhere [11, 12]). The interaction of c-kit with its ligand (stem cell factor) has been well-documented as a prerequisite for migration, survival and proliferation of PGCs. The tyrosine kinase receptor, c-kit, is expressed at the germ cell surface and binds to the somatic cells along the migratory pathway, where its ligand is expressed. This explains the directed homing of germ cells . Mutations in either the receptor (KIT) or the ligand (KL) cause a dramatic reduction in fetal germ cells, discovered 50 years ago , besides mislocalization of germ cells. The process of prenatal germ cell loss, which basically occurs through apoptosis [14, 15], is designated oocyte attrition and occurs during each phase of oogenesis, in both mitotic and postmitotic germ cells. From a peak number of 6.8 × 106 germ cells about the time of the mitotic to meiotic transition (fifth month of fetal development in human) , the number of germ cells sharply decreases, particularly during two main periods, the pachytene stage of meiosis in oocytes and the formation of primordial follicles . In agreement with these pioneer studies, it was reported that during early meiosis in humans, the number of viable oocytes declined from seven million on week 20 of gestation to less than one million at birth . Thus, the characteristic normal fate of female germ cells during oogenesis is to commit suicide, but the rationale for this remains a mystery. The initiation of meiosis (step vi), whereupon the germ cells become oocytes, coincides with the formation of primordial follicles and is controlled by retinoic acid . Follicle formation occurs during the second trimester of human fetal development, whereas in the mouse ovary it occurs immediately after birth . The somatic cells involved in human folliculogenesis are derived from both the surface epithelium and the mesonephros . As for the molecular mechanisms underlying oocyte PCD, there is no clear evidence that a death receptor or extrinsic pathway is involved during the fetal period [21, 22]. The intrinsic pathway of apoptosis is triggered by extracellular and intracellular stresses, such as growth factor withdrawal, the presence of genotoxicants, and DNA damage.
Two hypotheses have been advanced to explain the basis of the universal prenatal germ cell death [23, 24]. The first hypothesis is death by neglect, caused by shortage of survival factors. Indeed, germ cells depend, like other cell types, on the availability of certain growth factors, such as kit ligand (KL) and leukemia inhibiting factor (LIF) . Dependence on growth factors could be a way of matching the number of oocytes to the number of somatic cells involved in folliculogenesis in the same way that the number of neurons in the central nervous system has to match the number of target cells. Though in vitro experimental approaches to study fetal oogenesis indicate that programmed cell death is due to lack of survival factors , it has not been determined whether this is a normal aspect of fetal oogenesis in vivo. Of course, the infertility of mice mutant for KL  due to apoptosis of primordial germ cells is highly suggestive of the death by neglect hypothesis. Interestingly, it was shown that KL might prevent germ cell apoptosis by maintaining weak expression of the pro-apoptotic gene bax . Although the extrinsic pathway of PCD induction is reported not to be active during the fetal period [21, 22], the phenotypic appearance of homozygous KIT-deficient mice (i.e. like KL mutants, lacking oocytes at birth) can be partly rescued by simultaneous presence of homozygous Fas deficiency. This indicates that Fas-mediated apoptotic signals in the fetal ovary crosstalk with KIT-mediated survival signals . Another hypothesis proposed to explain the engagement of oocytes in an intrinsic pathway leading to PCD is the death by defect hypothesis [23, 24], which means the elimination of oocytes with chromosomal abnormalities or defective mitochondrial genomes . Indeed, most sex chromosome aneuploidies are associated with gonadal dysgenesis owing to absence or near absence of germ cells in the gonads. Although the idea of a program of oocyte quality control seems appealing, it has not been ascertained whether the extensive apoptosis of oocytes in the ovaries of patients with Turner syndrome (XO) is a cause or a consequence of the meiotic defect . Difficult to reconcile with this hypothesis is that XO mice are fertile and have normal ovaries at birth, even though they suffer from premature ovarian failure . A third hypothesis explaining germ cell loss in vertebrates is the death by self-sacrifice hypothesis, which is comparable to what is seen during invertebrate oogenesis. In lagomorphs, rodents and humans, the PGCs are initially very closely clustered together. These groups of cells are comparable to Drosophila germ cell cysts and consist of interconnected, synchronously dividing germ cells surrounded by a layer of somatic cells. These cysts begin to break down due to germline cell death, allowing somatic cells to cocoon individual surviving oocytes and thereby giving rise to primordial follicles . This type of altruistic cell death would permit oocytes destined to form primordial follicles to acquire mitochondria from neighboring dying germline cells, thus behaving functionally like invertebrate nurse cells [33, 34]. Due to the different timing of follicle formation in mice and humans, this altruistic cell death accounts for fetal germ cell loss in humans but not in mice. To the best of our knowledge this third hypothesis has not been linked to a specific intra or extracellular stress initiating the intrinsic pathway.
With regard to the intracellular regulators of germ cell apoptosis, a diverse spectrum of pro- and anti-apoptotic susceptibility genes, including the Bcl-2 and caspase families, have been reported to be involved in prenatal oocyte attrition (for detailed review see ). But it should be emphasized that, depending on the specific stage of prenatal oogenesis (i to vii), the various pro- and anti-apoptotic molecules may either affect or not affect apoptotic signaling. This issue was revealed by counting germ cells in the bax−/− mouse. The results illustrated that while bax contributes to the loss of primordial germ cells and oogonia, it is not involved in apoptosis of oocytes entering meiotic prophase. This conversion during fetal oognesis from a bax-dependent to a bax independent apoptotic signaling cascade explains why the bax−/− mouse is born with a number of germ cells comparable to the wild type . In conclusion, as long as the central dogma in reproductive biology remains valid (that most if not all female mammals are born with a finite stock of germ cells), the study of signaling factors that affect fetal follicular endowment and the rationale of the universal prenatal germ loss remain of utmost clinical relevance.
Programmed cell death in postnatal folliculogenesis
The quiescent ovarian germ stockpile is decreased from less than one million viable oocytes at birth to about 300,000 oocytes at puberty, of which only ~400 will ovulate over the fertile lifespan [36, 37, 38]. During postnatal degeneration of follicles, the loss of quiescent follicles can be distinguished from the loss of growing follicles. The Greek word “atresia” is used to describe the closure of a natural opening. In the strictest sense, follicular atresia refers to antral follicles undergoing degenerative changes before rupture during ovulation. This term is nowadays used in a broader sense to describe degenerative changes taking place during ovarian follicular development. Growth of the dormant follicles is initiated before and throughout the female’s reproductive life. In humans, a number of follicles are recruited for development during each reproductive cycle. A single dominant follicle is usually selected for ovulation, while the cohort of antral follicles are sacrificed . Ovarian follicular development and atresia are regulated by the interaction of pituitary hormones (gonadotropins) and intra-ovarian regulators, and this interaction promotes proliferation, growth, differentiation, and apoptosis. Follicular atresia is initiated within the granulosa layer and subsequently in the theca cells [40, 41, 42]. Widespread cell loss within the granulosa layer provokes the death of follicles. Only in primordial and primary follicles it is likely that oocyte loss is responsible for subsequent follicular degeneration [43, 44]. The basic mechanism of follicular atresia in mammalian and avian species is apoptosis. This conclusion is supported by the identification of apoptotic features, such as DNA fragmentation in atretic follicles and by comparison of expression levels of apoptosis-related genes in atretic versus healthy follicles; this will be discussed later in this review . Furthermore, programmed cell death is also instrumental during ovulation  as well as during regression of the mammalian corpus luteum  or its homolog in birds . Though the main cell death type in the ovary is apoptosis, autophagic cell death has also been evidenced during follicular atresia  and luteolysis . Thus, programmed cell death serves as the basic mechanism in the ovarian cycle [50, 51], and its instrumental role in determining postnatal follicular atresia has been clarified in several well-studied knockout and transgenic animals; this will be addressed at several places in this review.
Functional homology between the invertebrate nurse cell and the vertebrate GC
One of the conserved features of postnatal oogenesis is the intimate relationship between the oocyte and the surrounding cells, the latter of which have a somatic origin in vertebrates (the follicle cells or GCs) but a mixed origin in invertebrates. While nurse cells are of germline origin, follicle cells are of somatic origin, as observed for example in insects and crustaceans.
Follicular atresia in the mammalian ovary and the neo-oogenesis debate
A basic doctrine in reproductive biology is that mammalian ovaries are endowed with a fixed number of quiescent primordial follicles during early life , while male mammals can reproduce throughout most of their adult lives by continuously generating sperm precursors from germline stem cells maintained within the testis. Females of the few invertebrate species that remain fertile throughout life, such as the fruit fly Drosophila melanogaster or Artemia francicana, a brine shrimp, contain germline stem cells like those of males, and use them to replenish oocyte precursors [60, 61]. Mitotically active germ cells were documented in some species of prosimian primates. However, there was no evidence of folliculogenesis and ovulation from the proliferating germ cells .
Contrary to long-held views, arguments for the existence of proliferative germ cells that sustain oogenesis and folliculogenesis in postnatal mouse ovary have been reported [6, 7, 63]. Based on calculations of the quantity of atretic follicles at a given moment and the duration of the visible stages of atretic follicles, it was concluded that postnatal neo-oogenesis is actually necessary to compensate for the prepubertal loss. Another team’s calculations showed that follicle numbers in the primordial follicle pool of C57BL/6 mice remain stable from day 7 to 100. In addition, follicular recruitment into the population of growing follicles was not paralleled by a decrease in the total number of healthy follicles, indicating that follicles are renewed postnatally . Positive staining for germ cell markers and early meiosis markers was detected in adult mouse ovaries by two groups . The most convincing experiments in favor of postnatal neo-oogenesis was the restoration of oocyte production by bone marrow transplantation in wild-type mice sterilized by chemotherapy as well as in telelangiectasia-mutated gene-deficient mice, which are otherwise incapable of producing oocytes. Donor-derived oocytes have also been observed in female mice following peripheral blood transplantation . Based on these data, Spradling  suggested that menopause in humans is due to depletion of germline stem cells and to the age-dependent incidence of follicular atresia.
Three independent laboratories have shown that embryonic stem (ES) cells can differentiate in culture into primordial germ cells and thereafter into either oocyte-like or spermatid-like cells [65, 66, 67]. However, the full potential of these ES cell-derived gametes has not been demonstrated so far , and furthermore, an intense debate continues on whether this type of oogenesis from germ cells occurs in vivo in the adult ovary [69, 70, 71]. Shortly after the first paper from Tilly’s group on neo-oogenesis in the adult human ovary , another research team reported similar data . They stated that the only source of new primordial follicles would be the surface epithelium of the ovary rather than an extragonadal source. However, their immunohistochemically stained cryosections are far from convincing. Moreover, old experimental data do not support the hypothesis of Bukovsky et al. because destruction of surface epithelium did not cause any reduction in the number of follicles compared to control mice . Another research team  used a mathematical model for assessing follicle progression dynamics to calculate whether the initial follicle pool was sufficient for adult fertility. The results of this analysis support the old dogma that the initial endowment of ovarian follicles is not supplemented by a significant number of stem cells. A recent study by Eggan et al.  dealing with the issue of neo-oogenesis addressed the physiological relevance of circulating cells for female fertility. By using a parabiotic and transplantation mouse model, they assessed the capacity of circulating bone marrow cells to generate ovulated oocytes. Their findings do not support the existence of postnatal oogenesis from an extra-ovarian origin. Furthermore, they claim that cells of bone marrow origin traveling to the ovary via the bloodstream exhibit properties characteristic of committed blood leucocytes. It was recently shown that bone marrow transplantation in mouse, though it does not result in pregnancies from donor-derived oocytes, generates immature oocytes of recipient origin . However, it remains to be shown whether bone marrow transplantation indeed reactivates host oogenesis and does not simply rescue fertility by rescuing a sufficient number of existing oocytes. In this context, it was reported that a young hypofertile woman suffering from Fanconi anemia and treated by chemotherapy, irradiation and bone marrow transplantation gave birth to a child genetically related to her and not to the donor of the bone marrow . Although this case report illustrates that women are not equally sensitive to chemotherapy and irradiation, it does not answer the question whether bone marrow transplantation reactivates host oogenesis or whether a sufficient number of existing oocytes survive the treatment.
Comparison of the experimental findings in human and non-human mammalian ovaries that support the old dogma on the finite stockpile of oocytes at birth or conversely support the concept of neo-oogenesis
Meiotic, mitotic and germline markers
Inferential evidence histomorphometric assessment of follicle numbers versus death rates
BMT and peripheral blood generate donor derived oocytes in the recipient ovary, but offspring is of recipient origin
Grafting labeled ovary into unlabeled host ovary creates chimaeric follicles
Clinical experiments: BMT after chemotherapy rescues fertility in cancer patients
Absence of meiotic markers, mitotic and germline markers
Inferential evidence histomorphometric assessment of follicle numbers versus death rates do not indicate neo-oogenesis
BMT in a parabiotic mouse model does not generate donor derived ovulating eggs
Clinical experiments: BMT after chemotherapy does not rescue fertility in cancer patients
Apoptosis of granulosa cells in regulation of follicular atresia
Development and atresia of vertebrate ovarian follicles are tightly regulated by crosstalk between signals of cell death and survival. Apoptosis of GCs represents one of the primary pathways by which defective or excessive follicles are rapidly and effectively eliminated. Many apoptosis-related factors have been implicated in follicular atresia, including death ligands and receptors, caspases, pro- and anti-apoptotic Bcl-2 family members, gonadotropins, calcium, and gap junctional intercellular communication. All these will be discussed in this section. In addition, we will describe several genetically null mouse lines that lack various apoptosis regulatory proteins. Evaluation of female reproductive function in these knockout mice yielded interesting phenotypes and provided a possible means for distinguishing gene products essential for the execution of apoptosis from correlated gene products, i.e. gene products expressed during programmed cell death.
Pathways for induction of apoptosis in mammalian cells are mediated by cell-surface receptors known as death domain-containing receptors, which are members of the tumor necrosis factor (TNF) receptor family. The members of the TNF receptor family, including Fas receptor, TNF receptor, and tumor necrosis factor-related apoptosis inducing ligand (TRAIL), have been implicated in follicular atresia in mammalian ovaries. These membrane-anchored receptors, which are activated following ligand binding, are coupled to caspase activation by adaptor proteins such as Fas-associated death domain (FADD) and TNF receptor-associated death domain (TRADD) . FasL and Fas are the best characterized apoptotic signaling machinery in the GCs of many species, including humans [87, 88], mice , rats , pigs , and cattle . One line of evidence for the role of Fas in follicular atresia is provided by transgenic mouse models, including the lpr mutant mouse (lymphoproliferation), which contains a mutation in the gene encoding the Fas antigen. Young lpr mutant mice have morphologically normal ovaries, while adult mice have a significantly larger number of growing follicles than wild-type mice . These adult mice have larger ovaries because they have more follicles, indicating a crucial role for Fas in follicular atresia in the ovarian physiology of adult mice.
Activation of the Fas-FasL system can initiate apoptosis in GCs of rat, mouse [93, 94] and humans . Fas and Fas ligand have been localized to the granulosa layer in rat ovarian follicles that have been induced to undergo atresia in vivo by gonadotropin withdrawal . In human females, Fas is expressed in GCs of atretic antral follicles, and its level increases as atresia progresses . Treatment of cultured bovine GCs with estradiol E2 decreased the susceptibility to FasL-induced apoptosis . GCs from rats treated with equine chorionic gonadotropin (eGC) show reduced Fas and FasL content , whereas Fas content is increased in GCs from atretic mouse follicles . Nitric oxide, by suppressing activation of the caspases, inhibits apoptosis induced by the Fas/FasL system in rat GCs, pointing to a cross-talk between the Fas/FasL system-induced apoptosis pathway and NO-mediated anti-apoptotic pathway in ovarian follicle atresia . These findings strongly indicate that Fas/FasL is a key regulator of granulosa cell apoptosis and follicle atresia.
TNF, which is produced by several ovarian cell types, including GCs and the oocyte, is another important regulator of follicular development and atresia . This factor can induce either cell death or proliferation. TNF exerts its effects by binding to TNF receptor-1 (TNFR) or TNFR-2. TNFR-1 stimulates apoptotic signaling via its DD; TNFR-2, which lacks a DD, acts as a survival factor . TNF induces apoptosis in GCs isolated from rat preantral follicles and hen large white follicles [101, 102]. In addition to its pro-apoptotic function, TNF can promote survival and maintenance of follicular growth by inducing intracellular survival factors. For example, TNF induces NF-κB activation in rat GCs, resulting in expression of flice-like inhibitory protein (FLIP) and X-linked inhibitor of apoptosis (XIAP) [103, 104], which leads to survival of GCs isolated from follicles in the antral stage of development. These data indicate that TNF is an intraovarian modulator of granulosa cell function. Its pro-survival or pro-apoptotic effects on GCs depend on the stage of follicular development, which in turn determines the relative abundance of TNFR subtypes and the expression of various intracellular death and survival factors, such as Fas/FasL, XIAP and FLIP .
TVB is an avian death domain-containing receptor belonging to the TNF receptor family and is proposed to be the ortholog of mammalian DR5. The TVB receptor was originally identified by its ability to bind envelope coat proteins from cytopathic avian leukosis-sarcoma viruses (ALV). Treatment of cultured chick embryo fibroblasts with a recombinant ALV envelope fusion protein induces apoptosis that requires the presence of a functional TVB death domain . TVB is more abundantly expressed in hen GCs of atretic follicles than in healthy follicles . However, increased TVB expression does not precede induced follicle death in vitro. In addition, its expression in GCs was strongest during the final stages of follicle development, when follicles are highly resistant to apoptosis. These observations provide evidence that TVB receptor signaling in the ovary may also have functions other than mediating granulosa cell death and follicle atresia.
Increased mRNA expression of TNFR-associated DD protein (TRADD), which transmits the death signal from death receptor-4 (DR-4) and/or DR-5 to intracellular apoptosis-signal transduction components in GCs was demonstrated only in atretic follicles, indicating that the TRAIL-receptor system induces apoptosis in GCs during atresia in porcrine ovaries . It is noteworthy that mRNA of TRAIL, another member of the TNF family, is elevated in hen prehierarchal follicles undergoing spontaneous or induced atresia , indicating that at least four members of the TNF family may be involved in the process of follicle atresia.
Caspases constitute a family of intracellular cysteine proteases involved in both the initial and final stages of apoptosis in almost all types of vertebrate cells. Unprocessed caspase-3 can be found in GCs from healthy follicles, whereas GCs from atretic follicles contain larger amounts of activated caspase-3 . An antibody raised against the activated form of caspase-3 reacted strongly with GCs of degenerating antral follicles in both mouse and human . Analysis of caspase-3-deficient mice revealed that caspase-3 is required for granulosa cell apoptosis and thereby instrumental in follicular atresia, but dispensable for oocyte apoptosis . Activation of caspase-3 has been associated with cleavage of Poly-(ADPiribose)-polymerase (PARP) and actin, and the formation of oligonucleosomes . Caspase-9-deficient mice contain numerous developing follicles that fail to complete the process of atresia, apparently due to failure of granulosa cell apoptosis . Proteolytic activity of caspase-9 in pig GCs increased during follicular atresia, while the inactive zymogen (procaspase-9 protein) decreased . Caspase-9 and Apaf1 in murine GCs have also been shown to contribute to follicular atresia . In that context we have given evidence that in avian granulosa cytochrome c release, which is necessary to activate caspase 3 via the apoptosome complex, is confined to a subpopulation of mitochondria, while other mitochondria continue to respire .
On the other hand, caspase-2-deficient mice have excess follicles in neonatal ovaries due to attenuation of fetal germ cell loss . In addition, their oocytes were found to be resistant to cell death following exposure to chemotherapeutic drugs . These observations were substantiated by Morita et al. , who showed that germ cells in caspase-2-deficient fetal ovaries were also resistant to death caused by complete cytokine starvation in vitro, supporting the hypothesis that caspase-2 is central to the execution of PCD in oocytes.
Bcl-2 family members
As expected, the Bcl-2 family proteins are also implicated in granulosa cell apoptosis. One of the most studied members of this family is Bax. The survival of rat GCs mediated by gonadotropin correlates with decreased levels of bax expression in the absence of any change in Bcl-2 or Bcl-xL levels . Moreover, in man and other species, increased bax expression at the mRNA and protein levels was associated with granulosa cell apoptosis and follicular atresia. Microinjection of recombinant Bax protein into isolated oocytes induced apoptosis, thereby supporting the notion that increased cytoplasmic Bax levels are sufficient to induce apoptosis in female germ lines . Bax protein was abundantly expressed in GCs of early atretic follicles but was scarce or undetectable in healthy follicles . The rate of primordial and primary follicle atresia is substantially reduced in Bax-deficient mice due to defective postnatal oocyte apoptosis . Greenfeld et al.  studied antral follicles and showed that Bax deletion did not affect the extent of antral follicle atresia in vivo. The authors demonstrated similar numbers of atretic antral follicles in Bax−/− ovaries and wild type ovaries. In addition, the extent of apoptosis in Bax−/− ovaries was not different from that in wild type ovaries . Since atresia of immature follicles is initiated by oocyte death, whereas degeneration of antral follicles is controlled by death of GCs , it is conceivable that apoptosis in oocytes and GCs is regulated by different mechanisms. Importantly, as atresia of immature follicles can be reduced by Bax depletion , it is possible that oocyte death proceeds via a Bax-dependent pathway, whereas GCs die via a Bax-independent pathway .
Importantly, targeted expression of Bcl-2 in mouse oocytes during either fetal development  or postnatal life  suppresses apoptotic cell death. Moreover, mice overexpressing Bcl-2 show decreased apoptosis of ovarian somatic cells, enhanced folliculogenesis, and increased susceptibility to germ cell tumorigenesis . In contrast, ablation of Bcl-2 gene expression is accompanied by a decrease in the numbers of oocytes and primordial follicles in the postnatal ovary . These data provide in vivo proof that Bcl-2 family members regulate oocyte fate during prenatal development and during adult life. In addition to Bax and Bcl-2, other pro-apoptotic (e.g. Bcl-xs, Mtd/Bok, Diva/Boo, Bad, Bim and Bod) and anti-apoptotic (e.g. Mcl-1) proteins of the Bcl-2 family are known to be expressed in ovarian germ cells and/or GCs of various species [117, 124]. For example, ectopic overexpression of Mtd/Bok or Bad can induce granulosa cell apoptosis [125, 126].
These data indicate that the decision of any given follicle to either undergo atresia or to survive and subsequently ovulate is controlled by a balance between multiple pro- and anti-apoptotic Bcl-2 family members. Further elucidation of the role of Bcl-2 members in the tissue-specific regulation of apoptosis could facilitate an understanding of normal physiology and allow development of new therapeutic approaches for pathological states (discussed in more detail in section “Therapeutic options for combating premature ovarian failure”). For example, oocytes obtained from Bcl-2 transgenic mice and cultured in vitro were found to be resistant to spontaneous and anticancer drug-induced apoptosis providing a possible means to obtain resistance of the female germ line to naturally occurring and chemotherapy-induced apoptosis . These observations suggest a possible way to acquire resistance of the female germ line to naturally occurring and chemotherapy-induced apoptosis by targeted expression of Bcl-2 only in oocytes.
Gonadotropins and intraovarian regulators
The decision of the developing follicles to continue to grow and eventually to ovulate or to undergo atresia depends mainly on the coordinated action and interaction of cell survival and cell death factors within the GCs. It is likely that multiple molecules are involved in regulation of granulosa cell apoptosis, including follicle-stimulating hormone (FSH), GDF-9, Nodal, prohibin, TNF, IGF-I and p53. The response of follicles to these survival factors depends on the growth stage. For example, follicular development of primordial to secondary follicles does not require gonadotropin support. Intraovarian factors, such as GDF-9, control transition of the follicle from the preantral to the early antral stage. FSH is required for follicle growth from the time of past antrum formation until ovulation . In this section we will briefly review the role of these factors in follicle atresia and granulosa cell apoptosis. For a more detailed picture we recommend consulting other published reviews [39, 127, 128].
Gonadotrophin-mediated inhibition of apoptosis in ovarian GCs is partially related to changes in the expression of several cell death-related factors (see section “Death receptors”). Gonadotrophin withdrawal by antibody neutralization [129, 130], hypophysectomy on the day of proestrus, or metabolic clearance after a single hormonal injection induces granulosa cell apoptosis and follicular atresia [131, 132, 133]. Gonadotrophins can induce the expression of pro-survival molecules, including bcl-2, GATA-4, FLIP and XIAP, and decrease the expression of pro-apoptotic molecules, such Bax, Apaf1, Fas/FasL and p53 [112, 132, 134]. It has been shown that FSH can mediate both progesterone-dependent and progesterone-independent survival pathways in pre-ovulatory avian GCs .
Besides gonadotropins, other factors that are synthesized and secreted within the follicle have a direct impact on granulosa cell apoptosis. Down regulation of growth differentiation factor 9 (GDF-9) by intraoocyte injection of a GDF-9 antisense morpholino increased caspase-3 activation and granulosa cell apoptosis, but this response was attenuated by exogenous GDF-9 . In addition, GDF-9 suppressed ceramide-induced apoptosis in primary GCs from early antral, but not from large/preovulatory follicles . The phosphatidylinositol 3-kinase inhibitor, LY294002, and a dominant negative form of Akt prevented the protective effect of GDF-9. These data suggest that GDF-9 is antiapoptotic in preantral follicles and protects GCs from undergoing apoptosis by activating the phosphatidylinositol 3-kinase/Akt pathway.
Nodal, a member of the TGF-β family, exerts its biological effects by signaling through a cell surface serine/threonine kinase receptor complex composed of types I and II receptors and intracellular Smad proteins. Nodal can induce apoptosis and inhibit cell growth. Nodal and its type I receptor, Alk7, are expressed in the theca layer and in GCs, respectively. However, they are co-localized in the GCs when follicular atresia is induced by gonadotropin withdrawal . Addition of recombinant Nodal to GCs from large antral follicles, forced overexpression of Nodal, and expression of a constitutive active form of Alk7 (Alk7-ca) result in the induction of apoptosis [127, 137]. Over-expression of either Nodal- or Alk7-ca-activated caspase-3 and -9 increased apoptosis . Moreover, addition of recombinant Nodal or forced expression of Nodal or Alk7-ca in primary GCs induces phosphorylation and nuclear accumulation of Smad2, as well as down-regulation of phospho-Akt and Xiap content . All these data support the role of the Nodal/Alk7 signaling pathway in promoting follicular atresia.
The intracellular protein prohibin also contributes to granulosa cell apoptosis, and its effect is stage-dependent. Overexpression of prohibin in undifferentiated GCs from preantral follicles markedly attenuated apoptosis induced by ceramide, staurosporine, or serum withdrawal . However, over-expression of prohibin in differentiated GCs from antral follicles induces apoptosis . Further studies are required to identify factors regulating its expression and function in the follicle.
The p53 protein is an anti-proliferative transcription factor that regulates the rate of transcription of various genes involved in mitosis and apoptosis. The expression of p53 is increased during gonadotropin withdrawal, and overexpression of p53 resulted in extensive granulosa cell apoptosis [132, 138]. These data suggest that induction of atresia is a p53-dependent process.
IGFs and IGF-binding proteins (IGFBPs) are thought to play a critical role in ovarian follicle selection and follicular growth. Indeed, knockout of IGF-I in mice results in arrested follicular development at the preantral and early antral stages, leading to ovulation failure [139, 140]. In addition, IGF-I seems to play a crucial role in the responsiveness of the ovary to FSH action, because FSH receptor expression was severely reduced in preantral IGF-I null follicles, and restored to wild type levels after two weeks of exogenous IGF-I supplementation . Treatment of the rat pre-ovulatory follicles with IGF-I prevents spontaneous onset of apoptosis . Proliferation is promoted and apoptosis is suppressed in primary cultured porcine GCs by IGF-I and in avian GC explants by a combination of IGF-I and LH [142, 143]. In rat and bovine GCs, IGF-I activates PI3-K and Akt, with IGF-I driving phosphorylation, indicating that IGF-I plays an anti-apoptotic role in GCs by sustaining PI3-K-Akt signaling . Although IGF-I has an essential role in the development of ovarian follicles in many species, IGF-II is more abundant and likely more important in the human and non-human primate ovary, where it appears to act similarly to IGF-I [144, 145, 146]. All these observations led to the suggestion that the ultimate fate of the follicle depends on intrafollicular synthesis and bioavailability of IGF, and ultimately to its interaction with its cognate receptor . Actually, IGF’s bioavailability rather than its concentration dramatically changes during growth and atresia of ovarian follicles. IGFBPs have been proposed to play an essential role in IGF bioavailability by sequestering IGFs. In particular, transgenic female mice overexpressing IGFBP-1 were shown to present a decrease in serum IGF-I levels and bioavailability, and a reduction in natural and PMSG-induced ovulation rate [148, 149]. Large differences between the IGF/IGFBP systems of different mammalian species have been described. One exception to this rule is the degradation of IGFBP-2 and -4 that can be found in follicles undergoing terminal follicular growth, while these IGFBPs have been described to increase in atretic follicles . In response to the gonadotropin surge, the compact cumulus-oocyte complex undergoes expansion by synthesis and deposit of an intercellular matrix enriched in the mucopolysaccharide, hyaluronan . Hyaluronan in granulosa cell layers may be involved in cell locomotion , and in the prevention of fragmentation or segmentation of oocytes in vitro . In addition, hyaluronan decreases the occurrence of degenerated oocytes . Although little is known about regulation of the hyaluronan receptors in reproductive tissue, it has been suggested that the most common hyaluronan receptor, i.e. CD44, plays an important role during human oocyte maturation  and prevents apoptosis in human granulosa cells in a hyaluronan-dependent way .
Gap junctional intercellular communication
GCs can communicate either through the local production of intraovarian factors such as cytokines  and growth factors  that act as paracrine and/or autocrine modulators, or through gap junctions . Gap junctions between mammalian oocytes in follicles and GCs serve to transfer nutrients from the GCs to the oocyte, sustaining its growth, and to regulate oocyte meiosis. The gap junctions connecting mouse oocytes and GCs during oocyte growth consist of homomeric, homotypic channels composed of Cx37 . The cumulus and mural GCs are themselves connected by gap junctions composed of Cx43 [161, 162]. Mouse GCs express both Cx37 and Cx43 but target them to different cell-surface-membrane domains: Cx37 to contacts with the oocyte and Cx43 to contacts with each other [160, 163]. It has been shown that gap junctions play an important role in granulosa cell development, differentiation, and luteinization . The developmental importance of the gap junctions that couple growing oocytes with GCs was clearly demonstrated when the gene encoding Cx37 was knocked out . Although viable and without overt abnormalities, female mice lacking Cx37 are sterile due to disruption of folliculogenesis in the ovaries. In the absence of the coupling between oocyte and granulosa cell provided by Cx37, null mutant oocytes suffer growth retardation and do not survive to become meiotically competent . The growth of follicles in these mice is also interrupted. The mutant GCs form structures resembling corpora lutea, which would normally develop only after the mature oocyte has been expelled from the follicle during ovulation. Therefore, Cx37 gap junctions are essential for maintenance of oocyte growth and survival, which in turn is necessary for maintaining proper granulosa cell function [163, 167].
However, whether the modulation of gap junction formation and permeability is a primary or secondary event in controlling apoptosis is not yet clear. It was found that glucocorticoids enhanced Cx-43 expression, formation of gap junctions, and appearance of intact gap junctions in a rat pre-ovulatory granulosa cell line (RGSP53-10), indicating that gap junctional intracellular communication is an important mediator in glucocorticoid protection against apoptosis in GCs . In contrast, the same group reported earlier that induction of apoptosis by LH and forskolin was accompanied by increased expression of Cx43 in human luteinized GCs . This discrepancy might be related to the use of different cell models and apoptosis-inducing stimuli. We have established an avian model (Japanese quail, Coturnix coturnix japonica) of granulosa cell apoptosis in which granulosa explants are isolated from the largest ovarian follicle belonging to the follicle hierarchy . Since these granulosa cell explants consist of single-layered GCs sandwiched between vitelline and basement membranes, this model system is suitable for studying cell-cell contacts in vitro. We reported that induction of apoptosis was accompanied by decreased Cx43 immunoreactivity in immunocytochemical and immunoblotting procedures, suggesting that Cx43 expression per se may play a role in the survival process . A similar observation was made by Cheng et al. , who observed in porcine tertiary follicles that Cx43 was expressed most strongly in GCs of healthy follicles, with only trace levels in cells of early atretic and progressed atretic follicles, an indication that the expression levels of Cx43 protein decrease during follicular atresia . However, we also observed that initiation of apoptosis was accompanied by an overall increase in the level of gap junctional coupling, and apoptosis was dose-dependently inhibited by the gap junction blocker α-glycyrrhetinic acid . Therefore, these studies indicate that the functional state of gap junctional communication rather than their physical integrity may contribute to the resistance of GCs to apoptotic signals.
Role of calcium as a first and second messenger in survival and apoptosis of GCs
Calcium, one of the most versatile and universal signaling agents involved in cell growth and differentiation, has attracted interest as a potential second messenger in apoptosis since Kaiser and Edelman  demonstrated in 1977 that glucocorticoid-stimulated thymocyte apoptosis is associated with enhanced Ca2+ influx. Calcium coupling to the apoptotic effector pathway has been documented for a myriad of targets, including activation of proteases, transglutaminases, endonuclease(s), and Ca-dependent protein kinases and phosphatases, leading to alterations in gene transcription and affecting the enzymes involved in the maintenance of phospholipid asymmetry in the plasma membrane . Furthermore, it was shown that bcl-2 suppresses apoptosis by a mechanism that is linked to intracellular Ca2+ compartmentalization, and it appears that Ca2+ alterations in some cases of apoptosis occur as the result of changes within the mitochondria . We have shown for quail granulosa explants that a transient intracellular calcium rise due to influx of calcium is causally related to apoptosis induction [170, 176]. Similar data supporting the notion that calcium acts as a second messenger were reported for rat GCs in serum-free cultures.
Moreover, we suspected that Ca2+ also acts as a first messenger, because elevation of the extracellular calcium load in the absence of an intracellular calcium rise prevented apoptosis from occurring . The role of the extracellular calcium-sensing receptor (CaR) in maintaining Ca2+ homeostasis has been extensively studied, particularly in tissues contributing to calcium homeostasis, such as the parathyroids. We have shown for the quail ovary that CaR is expressed in GCs of the follicles selected for ovulation, but not in follicles that are lost during follicular atresia, which indicates that CaR might play a regulatory role in follicle selection. When granulosa explants are cultured in serum-free media for 24 h and then the CaR is challenged with different agonists, apoptosis is inhibited (unpublished results, D’Herde K). As far as we know, the CaR has not been linked to a survival pathway in the ovary.
Follicular atresia is not under exclusive control of apoptosis: autophagic cell death in GCs
Apoptosis of GCs and follicular fluid (FF) features as predictive markers for the outcome of assisted reproduction
Traditional methods for evaluating oocyte quality rely on the morphological assessment of the follicle, cumulus-oocyte complex, polar body and/or meiotic spindle . These methods are undoubtedly inexpensive; however, the use of morphological characteristics to predict oocyte developmental competence cannot be supported . Therefore, the need for more objective and accurate predictors has prompted the study of additional forecasters of oocyte quality associated with the oocyte itself, i.e. the surrounding follicular cells and the FF.
It is now widely recognized that oocyte-follicle communication is bidirectional and essential for their correct function and development. GCs are essential in ovarian folliculogenesis, because they produce factors that are necessary in normal follicular maturation, such as steroids and growth factors . Moreover, for an adequate follicular response to gonadotropins, it is necessary to have a suitable number of GCs, which is determined by the rates of proliferation and of apoptotic cell death . Taking the latter observations into consideration, it seems logical to propose GC apoptosis and FF characteristics as predictive markers for the developmental potential of the oocyte. Factors in GCs and those secreted in FF would, indeed, be an ideal source of non-invasive predictors of oocyte competence. Studies have been conducted in diverse species, physiopathological conditions, and methodologies, and variable results have been reported. The current section exclusively covers apoptosis-related prognostic markers for the outcome of assisted reproduction. Additional information on oocyte and embryo viability markers falls outside the scope of this review.
Apoptosis in follicular aspirates of GCs
It has been claimed that a key aspect in achieving high rates of pregnancy after IVF is selection of oocytes derived from women with low incidence of apoptotic cells on mural and/or cumulus GC masses derived from follicular aspirates during conventional IVF [184, 185, 186, 187]. Conversely, various studies reported that GC apoptosis cannot be used to predict the result of assisted reproduction techniques (ARTs) [188, 189]. Some major variations in the experimental setup of research trials that could lead to seemingly inconsistent results are discussed thoroughly later in this review.
Different techniques used to detect apoptosis
In the pursuit of markers of oocyte quality, several techniques have been employed to assess apoptosis of GCs. These include TUNEL staining, nuclear chromatin staining with fluorescent probes, and annexin V and/or PI labeling. Evaluation has been carried out mainly by microscopy, immunoblot and flow cytometry.
A pioneer study  examined the correlation between the occurrence of GC apoptosis, detected by Southern blotting following the TUNEL assay, and compromised oocyte quality, in terms of fertilization rates or in vitro development of embryos to the four-cell stage, on a per follicle basis. No correlation could be found based on apoptosis as a categorical variable, namely whether or not apoptosis could be detected. Later studies considered apoptosis as a continuous variable, and thus assessed the incidence or extent of apoptotic bodies in GCs in follicular aspirates. These studies found an association between apoptosis and IVF outcome when apoptosis was assessed by fluorescence microscopy upon chromatin staining  or by flow cytometry, and using the TUNEL assay in both instances [185, 190]. Flow cytometry has the extremely important advantage of counting large numbers of cells in a short time . However, flow cytometry may not be suitable for determining oocyte quality by evaluating individual follicles and measuring apoptosis because of the large number of GCs needed. Alternatively, apoptosis may be estimated by fluorescence microscopy. Although many studies morphologically evaluated the percentage of apoptotic cells upon nuclear staining, Lee et al.  reported that the fragmented, condensed nuclei of cumulus GCs could not be observed clearly upon staining with H33258. Hence, these authors recommended use of the TUNEL assay. As each type of technique has inherent advantages and disadvantages, it is recommended that more than one technique be used in order to achieve reliable results and to characterize a specific form of cell death [192, 193].
Apoptosis at the mural region and at the cumulus region
A space or antrum containing FF physically separates mural and cumulus granulosa components in a preovulatory follicle [194, 195, 196]. When a woman ovulates, virtually all of the granulosa components are obtained by aspiration. The mural GCs can be found at the FF while the cumulus GCs are those attached to the oocyte. Differences in the apoptotic percentage have been reported between mural and cumulus compartments. Indeed, the mural granulosa cell region had a higher incidence of apoptotic bodies than did the cumulus cell region in each follicle  and on a per patient basis [185, 197, 198]. These observations imply that, although both components seem to be predictive of oocyte quality and pregnancy outcome, the uniformity of the sample being evaluated should be taken into consideration to increase the accuracy and validity of the result.
Apoptosis in individual follicles or per patient
When the general capacity of folliculogenesis of each patient is to be assessed, it is important to analyze the incidence of apoptotic bodies on a per patient basis. However, if multiple follicles are aspirated per patient per cycle, no direct conclusions can be drawn concerning the relationship between the incidence of GC apoptosis and oocyte competence on a per follicle basis .
The intensity of apoptosis on a per follicle basis during the same superovulatory period clearly indicates that a higher incidence of apoptotic bodies is associated with empty follicles, poor oocyte fertilization, and poor embryo quality . Therefore, the occurrence of high-order pregnancies may be avoided by selectively transferring considerably fewer high quality embryos. Although investigation of apoptosis on the follicular basis seems to be the proper way, it is difficult to perform in the clinical setting. Moreover, even after observation on the follicular basis, it is still not certain which embryo is implanted after transferring two or three embryos. Therefore, more refined studies are required through investigation of apoptosis on a follicular basis and with single embryo transfer.
The assisted reproduction technique employed
The most frequently used assisted reproduction techniques are IVF and intracytoplasmic sperm injection (ICSI). While there is a fairly broad consensus that the incidence of apoptosis in GCs is indicative of ovarian function and is a prognostic factor in IVF program [184, 185, 200], it remains controversial for ICSI. Indeed, two research groups could not relate the percentage of apoptosis in GCs to oocyte maturity and fertilizability by ICSI or to follicular quality in stimulated cycles of normal women [189, 200]. The authors suggested that ICSI does not need high quality oocytes, defined as low apoptosis of GCs, because this technique bypasses the natural barriers by its invasive nature. Oocytes of relatively low quality, including those of older women and those with poor ovarian fecundity, have been proposed to have a greater chance of being fertilized with ICSI than with IVF. However, in another study, maturity and fertilization of the oocyte were found to be correlated with apoptosis in the cumulus cells during ICSI . In addition, apoptosis of GCs induced by oxidative stress, measured by the 8-hydroxy-2′-deoxyguanosine index, was correlated with the fertilization rate and embryo rate independently of the insemination method, i.e. IVF or ICSI . Therefore, the latter research groups [185, 201, 202] pointed to the fact that the quality of the oocyte is more relevant than the insemination method in determining the outcome.
Oocyte acquisition upon ovarian stimulation
Fresh, mature oocytes in metaphase II are usually obtained after pituitary down-regulation using gonadotropin releasing hormone agonists (GnRHa) followed by FSH administration. Ovulation is then stimulated by human chorionic gonadotropin, and GnRH antagonists may be simultaneously administered to avoid early ovulation [180, 203]. It was shown that GC apoptosis tends to be low in follicles stimulated with the survival factor FSH [204, 205]. Therefore, it is understandable not to find consistent results between GC apoptosis and parameters of oocyte quality for oocytes obtained after ovarian stimulation with FSH [206, 207].
Superovulation with GnRHa may raise the probability of fertilization and pregnancy by increasing the number of fertilizable oocytes ovulated per cycle, the chances of ovum pickup, and the density of gametes in the female reproductive tract [208, 209]. However, GnRHa superovulation may result in oocytes of poor quality. Indeed, GnRHa potentially increase the incidence of human GC apoptosis in vitro as well as in vivo [210, 211]. Hence, protocols excluding GnRHa may increase fecundity by improving oocyte quality. Due to the decrease in oocyte quality observed upon ovarian stimulation, the trend is now to harvest a single oocyte in the natural cycle with minimal stimulation.
Studies during IVF under various pathophysiological conditions
Endometriosis patients have a higher incidence of apoptotic GCs compared to normal women [184, 212]. Moreover, the incidence of GC apoptotic bodies increased as the severity of the endometriosis increased . Patients with deficit of the ovarian reserve also have a high percentage of apoptosis compared with patients with other causes of infertility and with normal women . Accelerated GC apoptosis has been proposed to play a pivotal role in the etiology of unexplained infertility . Patients from whom a large number of oocytes were retrieved had a lower incidence of apoptosis in GCs compared to those from whom fewer oocytes were retrieved [184, 197]. No difference was found in the first study  on the incidence of apoptotic cells among different age groups in which endometriosis patients were included. However, when evaluating normo-ovulatory women undergoing IVF due to infertility of their partners, those who were over 40 years old had significantly more apoptosis, lower fertilization rate, and fewer oocytes retrieved and mature than younger patients [191, 214]. Thus in order to obtain conclusive results, care should be taken when selecting the population to be included in an experiment.
Basal vs. induced GC apoptosis
Apoptosis of the cells retrieved from the stimulated follicle is a rare event during ART. Indeed, when evaluating apoptosis in unstimulated GCs, the average apoptotic index is low, ranging from 0.43% in a group with a good IVF-embryo transfer (IVF-ET) prognosis up to 1.81% in women with bad IVF-ET prognosis . On the other hand, upon stimulation with IFN-γ and triggering apoptosis with anti-Fas antibody, the percentage of apoptotic GCs drastically increased . In the latter study, mean apoptotic indices for good and bad IVF-ET outcome varied between 11.6% and 59.5%, respectively. The subtle differences obtained in unstimulated GCs could easily be missed compared with stimulated samples when counting less than ~1,000 cells per sample. Therefore, induction of apoptosis in GCs may more precisely predict oocyte quality and subsequently outcome of IVF-ET.
Biochemical features of apoptotic GCs
As previously discussed in this review, many pro- and anti-apoptotic proteins are involved in GC regulation. However, few have been correlated with the oocyte developmental potential in ART. Expression of the anti-apoptotic protein Bcl-2 was found to be significantly higher in the pregnant group upon IVF . However, the test was not sensitive enough to enable the use of Bcl-2 to predict pregnancy outcome . It is known that apoptosis is induced by a variety of stimuli, including oxidative stress. Eight-hydroxy-2′-deoxyguanosine (8-OHdG) is a sensitive indicator of DNA damage due to oxidative stress. A negative correlation was found between 8-OHdG and the fertilization rate or production of good embryos . Further studies regarding the relative expression of biochemical markers of apoptosis are required before they can be used as a screening method. The use of novel molecular biology techniques, such as cDNA expression array, will help to elucidate the activation of clusters of genes involved in the regulation of follicular atresia, and especially the regulation of apoptosis in GC.
Apoptosis-related metabolites present in FF
The composition of the FF could create a suitable microenvironment for oocyte development and maturation . Interestingly, the concentration of factors potentially influencing apoptosis in FF have been demonstrated to correlate with the developmental potential of the human oocyte [216, 217].
Intrafollicular steroid concentrations have been analyzed to detect the oocytes or embryos that have the highest potential to proceed to each stage of development. The results are often inconsistent because various stimulation protocols have been employed in the different IVF institutes, and IVF protocols have not been standardized yet [218, 219]. Estradiol and progesterone concentrations in large follicles were significantly higher while the percentage of GCs with nuclear fragmentation was significantly lower compared to those in medium and small follicles. On the other hand, testosterone concentration in small follicles containing the highest percentage of apoptotic bodies was significantly higher than in large follicles . Another study reported that the incidence of apoptotic bodies and the free testosterone level were significantly higher in the group of follicles from which no oocytes were retrieved, while estradiol and progesterone levels showed no difference between the two groups. Each of the hormones could be linked with oocyte maturity, quality or high IVF rate . In two other reports, no correlations were found between levels of steroid hormones (E2, progesterone and testosterone) in FF and the number and proportion of GCs undergoing apoptosis derived from healthy women [189, 214]. These authors argued that the results could be explained by the maintenance of steroidogenesis observed in apoptotic GCs , due to the fact that steroidogenic organelles, i.e. mitochondria and smooth endoplasmic reticulum, remain intact during the initial stages of apoptosis.
The expression of the insulin growth factor binding proteins (IGFBP) has been linked to follicular atresia in humans . Accordingly, IGF ligands serve to regulate GCs and theca cell growth and differentiation, and they are also thought to act as antiapoptotic factors [223, 224, 225]. The IGFBP expression profile of FF can be used to better predict oocyte developmental competence in bovines . Therefore, the evaluation of IGFBP in human FF could also be predictive of IVF-ET outcome.
Hyaluronic acid is an antiapoptotic component of the extracellular matrix and is produced by cumulus GCs after induction of ovulation by luteinizing hormone . The hyaluronic acid concentration in FF is affected by the regimens of assisted reproduction stimulation . A significantly lower content of hyaluronan was found in FF harvested from patients treated with GnRHa in combination with human menopausal gonadotropin (hMG) in comparison to those receiving clomiphene citrate in combination with hMG or with hMG alone. The hyaluronan level in FF containing subsequently normally fertilized oocytes was significantly lower than that in FF with unfertilized oocytes . Hence, the concentration of hyaluronan in FF has been pointed to as an indicator for the estimation of oocyte viability for fertilization.
Further, hyaluronic acid inhibits apoptosis of granulosa cells via CD44 . This molecule is a ubiquitous multistructural and multifunctional cell surface adhesion molecule involved in cell-to-matrix interactions. Soluble CD44 (sCD44) was expressed more strongly in follicles containing oocytes that were not fertilized than in those containing fertilized oocytes. On the other hand, shedding of CD44 from the fertilized oocytes resulted in their development into good-quality embryos, while the oocytes in the follicles containing a small amount of sCD44 turned into poor-quality embryos. A balance of CD44 on GCs and sCD44 in FF appears, therefore, to influence the quality of the embryos . These authors suggested that levels of sCD44 in human FF may be useful in assessing the prognosis of different IVF programs.
Soluble components of the Fas-FasL system are abundantly expressed in FF. Levels of the anti-apoptotic molecule, soluble Fas, have been demonstrated to be significantly higher in FF samples containing immature oocytes compared with those containing atretic oocytes. However, the similarity of the levels of soluble Fas and FasL detected in FF among patients with or without clinical pregnancy indicates that these soluble apoptotic factors may not be predictive of the success of IVF .
Many members of the transforming growth factor (TGF-β) superfamily are involved in controlling cellular growth and differentiation . Different research groups have evaluated the predictive value of the inhibin subfamily in pregnancy outcome. While one study has shown that inhibin levels in FF are positively associated with fertilization outcome and embryo quality , other reports have evidenced that this compound could not predict oocyte quality [233, 234].
In summary, oocyte quality assessment is important for successful pregnancy during assisted reproduction procedures. For now, the most convenient evaluation system for oocyte quality is based on oocyte morphology and status of oocytes-cumulus complexes. However, morphological assessment could be imprecise and subjective because there is no clear correlation between oocyte morphology and rate of fertilization or pregnancy. Several studies have linked the use of oocyte quality markers to the clinical outcome of patients undergoing ART. Among these investigational prognostic markers is the direct or indirect measurement of apoptosis in GCs. This appears to be a valuable tool that can be applied to more precisely predict the success of IVF/ICSI-ET. However, results are not definitive yet and much remains to be understood before a standard method can be introduced in assisted reproduction.
Therapeutic options for combating premature ovarian failure
Premature ovarian failure (POF; also known as premature menopause) is a well-known adverse effect of cancer treatment. It involves an accelerated decline in the number of oocyte-granulosa cell units to a critical threshold. Additional consequences associated with premature menopause are vasomotor, psychosocial, skeletal and cardiovascular problems [235, 236]. Ovarian damage induced by cancer treatment depends on the patient’s age, specific chemotherapeutic agents used, irradiation field, and total doses administered. Older women have a higher incidence of complete ovarian failure and permanent infertility in comparison with younger women [237, 238, 239]. This can be explained by the age-related decline of primordial follicle reserve . However, POF is mainly a long-term consequence of successful treatment of young cancer patients . Rapidly dividing cells, such as bone marrow, gastrointestinal tract and thymus, do recover, but it seems that damage to the postnatal oocyte pool, a cell-lineage arrested in meiosis-I and thus postmitotic, is irreversible. Alkylating agents were found to cause the highest risk of ovarian failure [242, 243]. Patients who received both alkylating agents and abdominal-pelvic radiation were more likely to suffer from POF than those who did not receive combined therapy . In addition, the risk of POF increases with increasing dose of abdominal-pelvic radiation and alkylating agent’s dose [244, 245].
Therapeutic options for combating POF include assisted-reproduction techniques, such as ovarian transposition (oophoropexy) and embryo, oocyte or ovarian tissue cryopreservation [246, 247]. Additionally, recent evidence indicates that a major mechanism of genotoxicity induced by cancer therapy is mediated by triggering apoptosis of the germ stockpile [248, 249]. Therefore, one of the main avenues of research to protect the germline from anticancer treatments is controlled manipulation of oocyte apoptosis . Indeed, ovarian tissue can potentially be protected in situ by administration of GnRHa in parallel with chemotherapy. It has been shown that GnRHa administration before cyclophosphamide treatment in rats significantly increased the pregnancy/mating rate and the number of implantations/mated animals . Moreover, 65% of the primordial follicle population in Rhesus monkeys was lost following cyclophosphamide treatment, in contrast to 29% loss in GnRHa-treated monkeys . This preliminary experience in rats and Rhesus monkeys is in line with recent clinical results. Only 7% of patients developed POF after co-treatment with GnRH agonist and chemotherapy, compared to more than half of the patients in the chemotherapy control group . Ongoing clinical trials further reinforce the potential use of GnRHa in minimizing chemotherapy-associated gonadotoxicity . The protective effect of GnRHa to minimize chemotherapy-associated gonadotoxicity could be explained by their ability to upregulate intragonadal anti-apoptotic molecules such as sphingosine-1-phosphate (S1P) . Because disruption of the gene encoding acid sphingomyelinase in female mice results in birth of female mice with twice as many oocytes as wild type animals, it was hypothesized that ceramide functions as a critical second messenger in female germ cell apoptosis . This hypothesis could be confirmed by experiments in which the ceramide antagonist S1P could enhance germ cell survival in wild type animals .
Notwithstanding all the above, several studies have reported that disruption of the Bax gene or enhanced expression of the Bax antagonist Bcl-2 in mice protects oocytes against the effects of doxorubicin [40, 248] and environmental toxicants [109, 256]. Additionally, oocytes from caspase-2 and caspase-3 double knockout female mice were more susceptible to apoptosis induced by DNA damaging agents and, conversely, more resistant to methotrexate-induced apoptosis compared to wild type oocytes . A recent study demonstrated that inactivation of the pro-apoptotic Bax gene in mice sustains ovarian lifespan into advanced age, extends fertile potential, and minimizes postmenopausal related diseases . Importantly, and contrary to popular belief, in this study prolongation of ovarian function into advanced age by Bax deficiency did not lead to an increase in tumor incidence.
Such etiological strategies for restoring apoptotic balance might provide an elegant approach to improving oocyte quality when undergoing cancer therapy. A better understanding of the genes responsible for regulating apoptosis is crucial for elucidating further the mechanisms involved and the reasons for its occurrence. Additional studies are needed to demonstrate the safety and efficacy in controlled manipulation of oocyte apoptosis.
Precise knowledge of the signals, receptors and intracellular signaling pathways leading to apoptosis of GCs is limited. It is likely that multiple molecules are involved, and here we sought to briefly overview these potential actors. Many more interesting and challenging findings concerning the molecular basis of ovarian life and death are expected to emerge, with consequent resolution of the controversial issue of neo-oogenesis. Further, this knowledge will stimulate the development of new treatment strategies for infertility, the improvement of strategies for preventing chemotherapy-induced infertility, and the understanding of normal and pathological processes leading to reproductive senescence.
Dmitri V. Krysko is supported by a postdoctoral fellowship from the BOF (Bijzonder Onderzoeksfonds 01P05807), Ghent University. We thank Dr. Amin Bredan for editing the manuscript.
- 4.Zuckerman S (1951) The number of oocytes in the mature ovary. Recent Prog Horm Res 6:63–108Google Scholar
- 17.Gondos B (1978) Oogonia and oocytes in mammals. In: Jones RE (ed) The vertebrate ovary. Plenum Press, New York, pp 83–120Google Scholar
- 46.Dharmarajan AM, Goodman SB, Atiya N, Parkinson SP, Lareu RR, Tilly KI et al (2004) Role of apoptosis in functional luteolysis in the pregnant rabbit corpus luteum: evidence of a role for placental-derived factors in promoting luteal cell survival. Apoptosis 9:807–814. doi:10.1023/B:APPT.0000045783.71178.c4 PubMedGoogle Scholar
- 50.Tilly JL, Billig H, Kowalski KI, Hsueh AJ (1992) Epidermal growth factor and basic fibroblast growth factor suppress the spontaneous onset of apoptosis in cultured rat ovarian granulosa cells and follicles by a tyrosine kinase-dependent mechanism. Mol Endocrinol 6:1942–1950. doi:10.1210/me.6.11.1942 PubMedGoogle Scholar
- 76.Lee HJ, Selesniemi K, Niikura Y, Niikura T, Klein R, Dombkowski DM et al (2007) Bone marrow transplantation generates immature oocytes and rescues long-term fertility in a preclinical mouse model of chemotherapy-induced premature ovarian failure. J Clin Oncol 25:3198–3204. doi:10.1200/JCO.2006.10.3028 PubMedGoogle Scholar
- 77.Veitia RA, Gluckman E, Fellous M, Soulier J (2007) Recovery of female fertility after chemotherapy, irradiation, and bone marrow allograft: further evidence against massive oocyte regeneration by bone marrow-derived germline stem cells. Stem Cells 25:1334–1335. doi:10.1634/stemcells.2006-0770 PubMedGoogle Scholar
- 91.Peng X, Maruo T, Matsuo H, Takekida S, Deguchi J (1998) Serum deprivation-induced apoptosis in cultured porcine granulosa cells is characterized by increased expression of p53 protein, Fas antigen and Fas ligand and by decreased expression of PCNA. Endocr J 45:247–253. doi:10.1507/endocrj.45.247 PubMedGoogle Scholar
- 98.Chen Q, Yano T, Matsumi H, Osuga Y, Yano N, Xu J et al (2005) Cross-talk between Fas/Fas ligand system and nitric oxide in the pathway subserving granulosa cell apoptosis: a possible regulatory mechanism for ovarian follicle atresia. Endocrinology 146:808–815. doi:10.1210/en.2004-0579 PubMedGoogle Scholar
- 112.Robles R, Tao XJ, Trbovich AM, Maravel DV, Nahum R, Perez GI et al (1999) Localization, regulation and possible consequences of apoptotic protease-activating factor-1 (Apaf-1) expression in granulosa cells of the mouse ovary. Endocrinology 140:2641–2644. doi:10.1210/en.140.6.2641 PubMedGoogle Scholar
- 115.Morita Y, Maravei DV, Bergeron L, Wang S, Perez GI, Tsutsumi O et al (2001) Caspase-2 deficiency prevents programmed germ cell death resulting from cytokine insufficiency but not meiotic defects caused by loss of ataxia telangiectasia-mutated (Atm) gene function. Cell Death Differ 8:614–620. doi:10.1038/sj.cdd.4400845 PubMedGoogle Scholar
- 125.Hsu SY, Kaipia A, McGee E, Lomeli M, Hsueh AJ (1997) Bok is a pro-apoptotic Bcl-2 protein with restricted expression in reproductive tissues and heterodimerizes with selective anti-apoptotic Bcl-2 family members. Proc Natl Acad Sci USA 94:12401–12406. doi:10.1073/pnas.94.23.12401 PubMedGoogle Scholar
- 134.Wang Y, Asselin E, Tsang BK (2002) Involvement of transforming growth factor alpha in the regulation of rat ovarian X-linked inhibitor of apoptosis protein expression and follicular growth by follicle-stimulating hormone. Biol Reprod 66:1672–1680. doi:10.1095/biolreprod66.6.1672 PubMedGoogle Scholar
- 143.Mao J, Smith MF, Rucker EB, Wu GM, McCauley TC, Cantley TC et al (2004) Effect of epidermal growth factor and insulin-like growth factor I on porcine preantral follicular growth, antrum formation, and stimulation of granulosal cell proliferation and suppression of apoptosis in vitro. J Anim Sci 82:1967–1975PubMedGoogle Scholar
- 146.Thierry van Dessel HJ, Chandrasekher Y, Yap OW, Lee PD, Hintz RL, Faessen GH et al (1996) Serum and follicular fluid levels of insulin-like growth factor I (IGF-I), IGF-II, and IGF-binding protein-1 and -3 during the normal menstrual cycle. J Clin Endocrinol Metab 81:1224–1231. doi:10.1210/jc.81.3.1224 PubMedGoogle Scholar
- 147.Adashi EY (1994) Regulation of intrafollicular IGFBPs: possible relevance to ovarian follicular selection. In: Baxter RC, Gluckman PD, Rosenfeld RO (eds) The insulin-like growth factors and their regulatory proteins. Excerpta medica international congress series 1056. Elsevier Science BV, Amsterdam, pp 341–419Google Scholar
- 175.McConkey DJ (1996) The role of calcium in the regulation of apoptosis. Scanning Microsc 10:777–793; discussion 793–774Google Scholar
- 179.Duerrschmidt N, Zabirnyk O, Nowicki M, Ricken A, Hmeidan FA, Blumenauer V et al (2006) Lectin-like oxidized low-density lipoprotein receptor-1-mediated autophagy in human granulosa cells as an alternative of programmed cell death. Endocrinology 147:3851–3860. doi:10.1210/en.2006-0088 PubMedGoogle Scholar
- 180.Elder KCJ (2007) Human preimplantation embryo selection, 1st edn. Informa HealthcareGoogle Scholar
- 189.Clavero A, Castilla JA, Nunez AI, Garcia-Pena ML, Maldonado V, Fontes J et al (2003) Apoptosis in human granulosa cells after induction of ovulation in women participating in an intracytoplasmic sperm injection program. Eur J Obstet Gynecol Reprod Biol 110:181–185. doi:10.1016/S0301-2115(03)00243-4 PubMedGoogle Scholar
- 196.Dirnfeld M, Goldman S, Gonen Y, Koifman M, Lissak A, Kraiem Z et al (1993) Functional differentiation in progesterone secretion by granulosa versus cumulus cells in the human preovulatory follicle and the effect of different induction of ovulation protocols. Fertil Steril 60:1025–1030PubMedGoogle Scholar
- 201.Host E, Gabrielsen A, Lindenberg S, Smidt-Jensen S (2002) Apoptosis in human cumulus cells in relation to zona pellucida thickness variation, maturation stage, and cleavage of the corresponding oocyte after intracytoplasmic sperm injection. Fertil Steril 77:511–515. doi:10.1016/S0015-0282(01)03006-0 PubMedGoogle Scholar
- 202.Seino T, Saito H, Kaneko T, Takahashi T, Kawachiya S, Kurachi H (2002) Eight-hydroxy-2’-deoxyguanosine in granulosa cells is correlated with the quality of oocytes and embryos in an in vitro fertilization-embryo transfer program. Fertil Steril 77:1184–1190. doi:10.1016/S0015-0282(02)03103-5 PubMedGoogle Scholar
- 205.Ruvolo G, Bosco L, Pane A, Morici G, Cittadini E, Roccheri MC (2007) Lower apoptosis rate in human cumulus cells after administration of recombinant luteinizing hormone to women undergoing ovarian stimulation for in vitro fertilization procedures. Fertil Steril 87:542–546. doi:10.1016/j.fertnstert.2006.06.059 PubMedGoogle Scholar
- 212.Toya M, Saito H, Ohta N, Saito T, Kaneko T, Hiroi M (2000) Moderate and severe endometriosis is associated with alterations in the cell cycle of granulosa cells in patients undergoing in vitro fertilization and embryo transfer. Fertil Steril 73:344–350. doi:10.1016/S0015-0282(99)00507-5 PubMedGoogle Scholar
- 217.Fanchin R, Mendez Lozano DH, Frydman N, Gougeon A, di Clemente N, Frydman R et al (2007) Anti-Mullerian hormone concentrations in the follicular fluid of the preovulatory follicle are predictive of the implantation potential of the ensuing embryo obtained by in vitro fertilization. J Clin Endocrinol Metab 92:1796–1802. doi:10.1210/jc.2006-1053 PubMedGoogle Scholar
- 219.Kamel MA, Zabel G, Bernart W, Neulen J, Breckwoldt M (1994) Comparison between prolactin, gonadotrophins and steroid hormones in serum and follicular fluid after stimulation with gonadotrophin-releasing hormone agonists and human menopausal gonadotrophin for an in-vitro fertilization programme. Hum Reprod 9:1803–1806PubMedGoogle Scholar
- 224.Bencomo E, Perez R, Arteaga MF, Acosta E, Pena O, Lopez L et al (2006) Apoptosis of cultured granulosa-lutein cells is reduced by insulin-like growth factor I and may correlate with embryo fragmentation and pregnancy rate. Fertil Steril 85:474–480. doi:10.1016/j.fertnstert.2005.08.014 PubMedGoogle Scholar
- 233.Fried G, Remaeus K, Harlin J, Krog E, Csemiczky G, Aanesen A et al (2003) Inhibin B predicts oocyte number and the ratio IGF-I/IGFBP-1 may indicate oocyte quality during ovarian hyperstimulation for in vitro fertilization. J Assist Reprod Genet 20:167–176. doi:10.1023/A:1023656225053 PubMedGoogle Scholar