The molecular mechanisms mediating mammalian fertilization

Fertilization is an essential step in sexual reproduction and consists of a carefully orchestrated series of events that culminate with the generation of a genetically unique zygote. Upon ejaculation in the female genital tract, millions of sperm ascend the uterus. However, only a few pass through the uterotubal junctions to migrate towards the ampulla of the oviduct. During this transit, sperm acquire the ability to fertilize eggs through a process defined as capacitation (Austin, 1951, 1960; Chang, 1951), which consists of a series of physiological and molecular changes that sperm acquire in the female reproductive tract. These changes pertain to the sperm motility pattern (Chung et al., 2014) and to their ability to undergo acrosome exocytosis (the fusion of the sperm plasma membrane with the outer acrosomal membrane), which is essential for fertilization (Puga Molina et al., 2018). Capacitated sperm transverse the cumulus oophorous (or cumulus mass), a mass of follicle cells kept together by hyaluronic acid and, within hours of ovulation, only one single spermatozoon successfully fuses with the egg. Effective mechanisms ensure monospermic fertilization, which is essential for successful embryonic development and healthy pregnancy. The first of these is a species-specific gamete recognition process at the zona pellucida (zona; ZP), the envelope surrounding mammalian oocytes, which mediates sperm binding and subsequent penetration through the zona. In this Review, we use ‘sperm binding’ to refer to the interaction between the sperm plasma membrane and the egg zona (Fig. 1, 1), and the term ‘sperm penetration’ to refer to the passage of the sperm through the zona (Fig. 1, 2). After penetration, the sperm reach a space enclosed between the inner aspect of the zona and the egg plasma membrane (the oolemma) known as the perivitelline space. Here, a second species-specific gamete recognition process mediates ‘gamete adhesion’; the interaction between the acrosome-reacted sperm plasma membrane and the oolemma (Fig. 1, 3). Gamete recognition at the zona and gamete adhesion prior to fusion occur in a species-specific manner. Finally, after gamete adhesion, the sperm membrane and the oolemma fuse to generate the zygote, which we refer to as ‘gamete fusion’ (Fig. 1, 4). Following gamete fusion is the exocytosis of cortical granules: membrane-bound vesicles derived from the Golgi apparatus that contain biochemical matrix-remodelling apparatus (Ducibella et al., 1988) (Fig. 1, 5). The resulting molecular changes in the zona and the oolemma establish effective blocks to polyspermy to prevent polyploidy, which is embryonic lethal in mammals (Fig. 1, 6).

Fertilization provides an excellent physiological platform for the study of ligand-receptor interactions. Moreover, understanding the molecular mechanisms that mediate fertilization will help to resolve the idiopathic causes of human infertility, establish novel treatments for providing superior fertility care and develop effective contraceptive agents. Here, we provide a comprehensive review of the fertilization process in mammals. We describe the most recent findings on the mechanisms mediating each of the steps described above. For each step, we report the most recent advances from studies performed on genome-edited mouse lines that describe the function and interactions of proteins during fertilization, focusing specifically on the proteins that are reported to be essential for fertilization.

A structurally intact zona pellucida plays major roles in mammalian fertilization: it mediates species-specific gamete recognition, it prevents polyspermy and it protects the preimplantation embryo from being resorbed into the oviduct epithelial lining (Bronson and McLaren, 1970; Modliński, 1970). The zona of most mammalian species contains either three or four glycoproteins: ZP1, ZP2/ZPA, ZP3/ZPC and ZP4/ZPB (Boja et al., 2003; Harris et al., 1994; Lefièvre et al., 2004) (Box 1, Fig. 2A). Extensive studies in mice have shed light on the structure of the zona, which is composed of ZP1, ZP2 and ZP3 (Bleil and Wassarman, 1980a).

Mutations of the genes encoding each zona protein have been shown to affect the zona structure either partially or severely, which leads to female subfertility or infertility (Box 2). Gene deletions in transgenic mice have shown that the absence of ZP1 results in a misshapen, thinner and more fragile zona structure compared with the wild type. Although still capable of supporting sperm binding and fertilization, ZP1 null mutations are associated with female subfertility (Rankin et al., 1999). A more severe phenotype is observed in mutant mice lacking either ZP2 or ZP3: homozygous null females form a very thin zona (Zp2^Null) or do not form a zona at all (Zp3^Null), which results in defective follicle formation, scarcity of ovulated eggs (that also lack the zona) and ultimately in female infertility (Rankin et al., 1996, 2001). To date, a complete crystal structure of a ZP protein has only been reported for chicken ZP3 (Han et al., 2010). Further characterization of the molecular architecture of the zona matrix will allow a better understanding of the putative interactions between zona proteins in establishing and maintaining a three-dimensional structure that guarantees successful sperm-egg recognition.

Uncovering the zona protein that mediates sperm binding to the egg has been a compelling and debated biological question over the past few decades. Results have varied based on the assays employed or the model systems adopted. Previous studies have reported each individual zona protein as the ligand for gamete recognition (Bleil and Wassarman, 1980b; Ganguly et al., 2010; Yonezawa et al., 2012). Controversy still persists on whether a single zona protein may be sufficient to support sperm binding and which of the four zona proteins mediate gamete recognition.

To identify the zona ligand that mediates sperm binding, pioneer studies have used soluble SDS-PAGE mouse ZP proteins in in vitro competitive sperm binding assays with mouse ovulated eggs. These studies showed significant sperm binding inhibitory effects only in the presence of ZP3. On the other hand, with ZP1, ZP2 or ZP3 that were isolated from the two-cell embryo zonae, to which sperm are unable to bind, no effect on sperm binding to ovulated eggs was recorded (Bleil and Wassarman, 1980b). These observations introduced a widely accepted model that indicate ZP3 as the ligand for gamete recognition at the zona. This model later implicated O-glycans attached to Ser³³² and Ser³³⁴ of ZP3 as the zona ligands for sperm binding (Chen et al., 1998; Florman and Wassarman, 1985). More specifically, the α1,3 galactose and the N-acetylglucosamine were proposed as the ligand for mouse sperm binding (Bleil and Wassarman, 1988; Miller et al., 1992), and the sialyl-LewisX antigen for human sperm binding (Pang et al., 2011). However, fertility is preserved in genome-edited mice either lacking the α1,3 galactose or the putative sperm receptor for N-acetylglucosamine (Thall et al., 1995). Indeed, no carbohydrates have been found on Ser³³² and Ser³³⁴ in normal mouse zonae (Boja et al., 2003), and mutation of the sites to prevent glycosylation in transgenic mice has not led to female infertility (Gahlay et al., 2010). Moreover, mice that lack the glycosyl transferases MGAT1 and T-synthase are still fertile (Shi et al., 2004; Williams et al., 2007).

Another thought-provoking model suggests that the process of sperm binding to the zona might not be necessary for penetration, and that no gamete recognition at the zona occurs before sperm crossing the zona. This model originates from observations in knockout mice lacking ADAM3 (a disintegrin and metalloprotease 3). Although sperm from Adam3^Null mice show impaired sperm binding to the zona, the sperm can fertilize the eggs in vitro (Yamaguchi et al., 2009). In addition, deletion of a number of genes that are necessary for normal expression of ADAM3 in sperm, such as Ace (Krege et al., 1995), Clgn (Ikawa et al., 1997), Tpst2 (Marcello et al., 2011), Calr3 (Ikawa et al., 2011) and Pdilt (Tokuhiro et al., 2012), leads to defective sperm binding to the zona (Yamaguchi et al., 2009). Of note, humans have two orthologues of Adam3 (ADAM3A and ADAM3B), and both are pseudogenes. However, male fertility is preserved in the absence of the ADAM3 orthologues (Grzmil et al., 2001). Moreover, the observation that mammalian sperm bind to the zona with species-specificity (Bedford, 1981) indicates a role of the zona in mediating gamete recognition. For example, human sperm bind to the zonae of human and hominoid primates (gibbon, gorillas), but do not bind to the zonae of other sub-hominoid primates (baboons, rhesus monkeys, squirrel monkeys) or mouse and other rodents (Baibakov et al., 2012; Bedford, 1977; Hoodbhoy et al., 2005; Lanzendorf et al., 1992). Likewise, horse sperm can bind to the zonae of horse, but not to the zonae of pigs (Mugnier et al., 2009). Conversely, mouse and rat sperm bind with similar efficiency to the zonae of either rodent (Hoodbhoy et al., 2005), and pig sperm bind to equine and bovine zonae as effectively as to pig zonae (Mugnier et al., 2009; Takahashi et al., 2013). Thus, two main observations prompt reconsideration for alternative models mediating sperm binding to the zona. First, mouse sperm cannot bind to the two-cell embryos and the only characterized biochemical modification in the zona is a ZP2 cleavage after fertilization (Bleil et al., 1981) – remarkably, mouse sperm can bind de novo two-cell embryos when ZP2 remains uncleaved after fertilization (Baibakov et al., 2007; Gahlay et al., 2010). Second, as reported above, human sperm are selective and only bind to human and hominoid primates eggs (gibbon, gorillas), but do not bind to mouse eggs (Bedford, 1977; Lanzendorf et al., 1992).

Keeping this species-specificity in mind, gain-of-function assays using mouse genetics have been established by individually expressing human ZP1, ZP2 or ZP3 genes in the appropriate mouse Zp1, Zp2 or Zp3 null background (Rankin et al., 1996, 1999, 2001) (Fig. 2). The three established transgenic mouse lines have been defined as human (hu)ZP1^Rescue, huZP2^Rescue and huZP3^Rescue. Moreover, a fourth line expressing mouse ZP1-3 and human ZP4, defined as huZP4^Transgenic, has also been generated (Baibakov et al., 2012; Rankin et al., 1998, 2003). To assess which glycoprotein is sufficient to support human sperm binding, eggs from each huZP1^Rescue, huZP2^Rescue, huZP3^Rescue and huZP4^Transgenic line have been inseminated with human sperm, revealing that human sperm bind avidly only to the huZP2^Rescue eggs (Baibakov et al., 2012) (Fig. 2B,C,E). In addition, human sperm injected in vivo into the oviduct ampulla of huZP2^Rescue and huZP3^Rescue female mice have been observed only in the perivitelline space of the huZP2^Rescue eggs (Avella et al., 2014). After penetration through the zona, however, human sperm cannot fuse with the mouse eggs, owing to species-specific mechanisms controlling gamete adhesion and fusion (Bianchi and Wright, 2015; Quinn, 1979; Yanagimachi, 1984). These observations demonstrate that ZP2 is sufficient to support human sperm binding and penetration of the zona pellucida in transgenic mice (Avella et al., 2014; Baibakov et al., 2012), in the absence of sialyl-LewisX antigen (Avella et al., 2014; Pang et al., 2011).

Complementary loss-of-function assays have been performed by crossing mouse lines to obtain zonae lacking either mouse or human ZP2. Transgenic zonae containing mouse ZP1, ZP3 and human ZP4 cannot support mouse sperm binding. Likewise, transgenic zonae containing human ZP1, ZP3 and ZP4 cannot support mouse and human sperm binding, and female mice are infertile (Avella et al., 2014). These observations further support the model that ZP2 is necessary and sufficient for mouse and human sperm binding in transgenic mice (Avella et al., 2014) (Fig. 2D). Moreover, the gamete recognition domain has been refined to the N terminus of ZP2 by deleting a region coding for the ZP2 N terminus (ZP2^51-149). These mutant zonae cannot support mouse sperm binding and female mice are infertile (Avella et al., 2014) (Fig. 2D,E). In addition, a recombinant fusion mouse Zp2^35-149/huZP4 protein-encoding gene expressed in the Zp2^Null background has been shown to re-establish formation of a zona matrix, which could support sperm binding in vitro, resulting in fertile female mice (Tokuhiro and Dean, 2018) (Fig. 2D). Genetic ablation of the proposed glycan attachment site on the N terminus of ZP2 does not affect sperm binding or female fertility in transgenic mice (Tokuhiro and Dean, 2018), which demonstrates that gamete recognition mediated by the N terminus of ZP2 is glycan independent (Boja et al., 2003; Avella et al., 2014; Tokuhiro and Dean, 2018) (Fig. 2A). The N terminus of ZP2 accounts for the species-specificity observed for human sperm binding. Chimeric human/mouse ZP2 proteins, with the human N-terminal domain in place of the mouse N terminus, support human sperm binding analogous to the huZP2^Rescue zonae matrices (Avella et al., 2014) (Fig. 2C,E). All these results are consistent with the N-terminal region of ZP2 being necessary and sufficient for sperm binding to the zona in mice and humans, along with female fertility in mice (Avella et al., 2014; Tokuhiro and Dean, 2018). These recent results support a model of gamete recognition in which mouse and human sperm bind to the zona via the N terminus of ZP2 and the binding appears to be independent of O- or N-linked zona protein glycans (Tokuhiro and Dean, 2018).

A plausible alternative interpretation for these observations is that sperm may not directly interact with ZP2 and that the ZP2 N terminus is necessary to preserve a zona structure that guarantees normal sperm binding. Indeed, the absence of the ZP2 N terminus (Avella et al., 2014) may mimic a possibly modified zona structure occurring after fertilization, which would impede sperm binding and lead to infertility. Because sperm binding is necessary for penetration through the zona (Avella et al., 2014; Baibakov et al., 2012; Tokuhiro and Dean, 2018) and occurs with species-specificity (Bedford, 1977), one would expect that sperm recognize a putative zona sperm binding site, the structure of which is preserved by an intact ZP2 N terminus. After fertilization, the cleavage of ZP2 would disrupt the native conformation of this putative sperm binding site, impeding supernumerary sperm binding to the zona. If this model is indeed correct, the data obtained from genetically edited mouse lines may help to localize the putative sperm binding site on the zona: Zp1^Null female mice still form a zona that is capable of supporting sperm binding, so the sperm binding site should reside on ZP2 and/or ZP3. Also, we may exclude the ZP2 C terminal region, which is not required for sperm binding, because recent studies in transgenic mice show that ZP2^51-149 is necessary (Avella et al., 2014) and sufficient (Tokuhiro and Dean, 2018) for sperm binding. In addition, human sperm, which are normally unable to bind to mouse zonae (Bedford, 1977), can bind and penetrate huZP2^Rescue zonae, but not huZP3^Rescue zonae, in vitro and in vivo (Avella et al., 2014; Baibakov et al., 2012). Together, this evidence indicates that the putative sperm binding site should not reside on human ZP3. Moreover, recombinant mouse or human ZP2 N termini directly interact with mouse or human sperm in peptide-bead binding assays (Avella et al., 2014, 2016; Baibakov et al., 2012). From all these data, we hypothesize a direct interaction between sperm and the N terminus of ZP2 and we conclude that, for successful fertilization, sperm bind to the zona via the N terminus of ZP2.

Also, this model raises the prediction of the existence of a putative sperm receptor for the ZP2 N terminus, which would mediate sperm binding to the zona. In the past decades, numerous studies have focused on ZP3 as a ligand to identify the sperm receptor (Bleil and Wassarman, 1990; Cheng et al., 1994; Ensslin and Shur, 2003; Hanayama et al., 2004; Muro et al., 2012; Neutzner et al., 2007; Silvestre et al., 2005; Tardif et al., 2010). However, ablation of each novel candidate in transgenic mice has not resulted in male infertility and the identity of this receptor is still unknown. Here, we report a number of examples including SED1 [secreted protein that contains Notch-like epidermal growth factor (EGF) repeats and discoidin/F5/8 type C domains] (also known as MFGE8), the ZP3 receptor (ZP3R; also known as sp56), and zonadhesin. The sperm surface protein SED1 was originally identified using porcine zona proteins from the boar orthologue, p47. Western blot analyses have shown that SED1 interacts with ZP3 (Ensslin and Shur, 2003). However, deletion of the SED1-encoding gene, either shows no impact on male fertility (Hanayama et al., 2004; Neutzner et al., 2007) or leads to male subfertility (Ensslin and Shur, 2003; Silvestre et al., 2005). Thus, SED1 is not necessary for sperm binding. A second candidate, ZP3R, localizes on the sperm plasma membrane and shows binding affinity for the glycans on ZP3 (Bleil and Wassarman, 1990; Cheng et al., 1994). In an in vitro competitive binding assay, ZP3R (recombinant or native) can inhibit sperm binding to ovulated mouse eggs, but not to embryos. Nonetheless, genetic ablation of the gene encoding ZP3R does not affect male fertility in transgenic mice (Muro et al., 2012). Finally, zonadhesin has been also an appealing candidate; it is a testis-enriched protein that presents several cell-adhesion domains and becomes exposed in the sperm plasma membrane during capacitation. Recombinant zonadhesin can bind to the zona, and antibodies against zonadhesin significantly inhibit sperm binding and in vitro fertilization (Tardif et al., 2010). Nevertheless, mutant null males are fertile, which leads to the conclusion that zonadhesin is not necessary for sperm binding (Tardif et al., 2010).

Today, the definition that the ZP2 N terminus is the ligand necessary and sufficient for gamete recognition at the zona in mice and humans may offer a new path toward the identification of novel sperm receptor candidates. Moreover, recent studies reported that the folding of the N terminus of ZP2 shows structural similarities with a specific region of the egg coat protein VITELLINE ENVELOPE RECEPTOR for LYSIN (VERL) from the marine mollusc, abalone. During abalone fertilization, the egg VERL directly interacts with the sperm lysin to guarantee species-specific gamete recognition (Raj et al., 2017). Even though it appears plausible that ZP2 may not interact with a mammalian homolog of lysin, the functional (Avella et al., 2014; Baibakov et al., 2012; Tokuhiro and Dean, 2018) and structural (Raj et al., 2017) definition of the oocyte ligand open new perspectives for the identification of the putative receptor for mammalian sperm binding to the zona.

The acrosome is a Golgi-derived subcellular organelle that underlies the anterior plasma membrane of mammalian sperm heads. Inner and outer acrosomal membranes delimit this vesicle (Fig. 3A), which includes a mixture of digestive enzymes – although none of the digestive enzymes is necessary for binding or penetration through the zona pellucida (Buffone et al., 2014). To gain fusion competence with the egg, capacitated mammalian sperm must undergo the acrosome reaction, an exocytotic event, which involves the fusion of the sperm plasma membrane with the outer acrosomal membrane. This membrane fusion event lets the acrosome-reacted sperm expose the inner acrosomal membrane, together with the equatorial segment. The sperm equatorial segment is defined by two regions, one in which the inner and outer acrosomal membrane are tightly apposed, and another one in which the inner and outer acrosomal membranes separate to include the acrosomal matrix (Yanagimachi and Noda, 1970). Acrosome-reacted sperm can fuse with the oolemma, the egg plasma membrane, via a remnant of the sperm plasma membrane, overlying the equatorial segment (Fig. 3A).

Only acrosome-reacted sperm are found in the perivitelline space of eggs from guinea-pig, mouse and rabbit (Fleming and Yanagimachi, 1982; Inoue et al., 2011; Kuzan et al., 1984), which indicates that, at some point before fusion, sperm must undergo acrosome exocytosis. The site of acrosome reaction has been an intriguing matter of debate for decades and remains highly controversial (Fig. 3B). Based on functional studies reporting the ability of solubilized mouse zonae or mouse ZP3 to trigger acrosome exocytosis in vitro (Bleil and Wassarman, 1990), acrosome reaction has been thought to be induced during gamete recognition at the zona pellucida (Saling et al., 1979). However, binding to the zona has been reported to be insufficient to induce acrosome exocytosis of Acr^GFP sperm, even several hours after binding (Baibakov et al., 2007). These observations prompted consideration of alternative models of acrosome exocytosis induction.

Early studies in rabbit show that acrosome-reacted sperm are able to bind and fertilize de novo rabbit ovulated eggs (Kuzan et al., 1984). This observation could be repeated in mouse using supernumerary perivitelline sperm recovered from inseminated Cd9^Null eggs, which are unable to fuse with mouse sperm (discussed below) (Inoue et al., 2011). In addition, time-lapse imaging of Acr^GFP mouse sperm shows that acrosome-reacted sperm penetrate the zona with higher efficiency than acrosome-intact sperm (Jin et al., 2011). These observations are consistent with recent findings showing that both mouse and human sperm can remain bound to the N terminus of ZP2 after acrosome exocytosis (Avella et al., 2016). However, this model does not fully explain the acrosomal shrouds (residues of the vesiculated acrosomal caps) that are found attached to the zonae surrounding fertilized eggs in different mammalian species (VandeVoort et al., 1997; Wakayama et al., 1996; Yanagimachi and Phillips, 1984). Indeed, if sperm undergo acrosome exocytosis before binding to the egg, no shroud should be observed on the zona. Moreover, species-specificity has been observed in the induction of acrosome exocytosis during zona pellucida penetration; field vole sperm penetrate mouse and hamster zonae, and the acrosome remains in the perivitelline space, but they are unable to fuse with the eggs (Wakayama et al., 1996).

To clarify these diverse sets of observations, it is conceivable that different mechanisms could induce acrosome exocytosis, which would depend on the status of the egg investments encountered by the fertilizing sperm. A tight cumulus mass surrounding freshly ovulated eggs could induce acrosome exocytosis before sperm penetration through the zona matrix. Conversely, a more dispersed or absent cumulus mass, as it is the case for some marsupials (Rodger and Bedford, 1982; Talbot and DiCarlantonio, 1984), would fail to induce acrosome exocytosis and would therefore be dependent on zona penetration (Baibakov et al., 2007; Wakayama et al., 1996). Adding more complexity to the model, recent studies have indicated that the majority of sperm have undergone acrosome exocytosis before encountering the cumulus mass (Hino et al., 2016; Muro et al., 2016), and only ∼5% of the sperm traversing the ampulla are acrosome-intact (La Spina et al., 2016). Therefore acrosome exocytosis of the fertilizing sperm might be occurring during sperm migration through the female oviduct and the acrosome-reacted sperm cross the cumulus before binding to the zona (Fig. 3B). This model raises some intriguing predictions to be tested in future assays: as sperm migrate through the female oviduct or as they cross the cumulus mass, what are the molecular mechanisms that mediate the induction of acrosome exocytosis? Also, both acrosome-intact and acrosome-reacted sperm have the ability to bind to the zona, so is there only one putative sperm plasma membrane receptor for binding to the zona pellucida, which is relocated to the equatorial segment upon acrosome reaction? Or are there two or more different sperm receptors, located in different sites of the sperm head? A clear understanding of the mechanisms and sites of the induction of acrosome exocytosis are necessary for a better understanding of the processes mediating gamete recognition and fusion in mammals.

Gamete adhesion between the oolemma and the sperm plasma membrane overlying the equatorial segment is a precursor step necessary for gamete fusion and shows extensive species-specificity in mammals (Bedford, 1977; Bianchi and Wright, 2015; Quinn, 1979; Yanagimachi, 1984). Gamete adhesion begins with known or putative molecular interactions between proteins of sperm and the oolemma. Upon gamete adhesion, a closer apposition of the membranes establishes mixing of lipid bilayers, followed by the formation of fusion pores, which generates cytoplasmic continuity and gamete fusion (Yanagimachi, 1984). Gamete adhesion and fusion have been intensively investigated over the past decades. Pioneering in vitro fertilization (IVF) studies have reported how defective or uncontrolled mammalian gamete adhesion and fusion causes failed fertilization, including eggs remaining unfertilized (Chang, 1959; Edwards et al., 1970) or polyspermy (Wentz et al., 1983), both of which lead to infertility.

Despite decades of investigation, only a few proteins have been found to be necessary for the adhesion of the sperm plasma membrane to the oolemma. Using loss-of-function assays in transgenic or mutant mice, two sperm proteins have been found to mediate the sperm adhesion to the oolemma that is necessary for fertilization: IZUMO1 (Inoue et al., 2005) and SPACA6 (Lorenzetti et al., 2014). IZUMO1 is a testis-specific cell-surface protein and part of the immunoglobulin type-I cell superfamily, characterized by a cytoplasmic C-terminal tail, a transmembrane region and a conserved ‘Izumo domain’, which is linked to an extracellular immunoglobulin-like (Ig-like) C2-type domain by a central β-hairpin region. Each of these domains plays an important role in mediating gamete adhesion (Aydin et al., 2016; Inoue et al., 2005; Ohto et al., 2016). Upon acrosome exocytosis, IZUMO1, which is not detectable on the plasma membrane of acrosome-intact sperm, localizes to the equatorial segment (Satouh et al., 2012) to mediate gamete adhesion with the egg plasma membrane. Absence of IZUMO1 leads to impairment of gamete adhesion and to an accumulation of sperm in the perivitelline space (Inoue et al., 2005). SPACA6 is a testis-specific protein that includes an Ig-like domain. Random integration of a transgene has been reported to disrupt the open reading frame of Spaca6 (intron 2-exon 6), which results in loss of SPACA6 expression and sperm-egg adhesion impairment, with an accumulation of supernumerary sperm in the perivitelline space (Lorenzetti et al., 2014).

On the oolemma, two proteins are necessary for gamete adhesion. CD9 was the first protein identified to mediate gamete adhesion and belongs to the family of tetraspanins. Females lacking CD9 show severely reduced fertility, or infertility, due to the inability of the oolemma to support sperm adhesion (Le Naour et al., 2000). A second tetraspanin, CD81, has also been found to play a role in mediating gamete adhesion. Cd81^Null female mice accumulate supernumerary sperm in the perivitelline space, although they are still fertile – albeit with reduced fertility (Rubinstein et al., 2006).

More recently, an oocyte glycosylphosphatidylinositol (GPI)-anchored protein, folate receptor 4 (FOLR4, also known as IZUMO1R), has been identified as the oolemma-bound receptor for IZUMO1 (Bianchi et al., 2014). Although this protein presents structural similarity with other folate receptors, it is unable to bind to folates and therefore it is also called JUNO, after the Roman goddess of fertility and marriage. Female mice lacking JUNO are infertile, owing to the inability of the egg to adhere to sperm (Bianchi et al., 2014). Structural studies using X-ray crystallography have confirmed direct interaction between human IZUMO1^22-254 and JUNO^20-228, and show that the two proteins form a stable complex that is necessary for gamete adhesion (Aydin et al., 2016; Ohto et al., 2016). These structural studies have identified 19 residues in JUNO that interact with 20 amino acids in IZUMO1 (Aydin et al., 2016; Ohto et al., 2016). In addition, this IZUMO1-JUNO interaction (Jean et al., 2019) regulates the species-specific gamete adhesion observed between human sperm and zona-free hamster oocytes (discussed in detail below) (Bianchi and Wright, 2015; Han et al., 2016; Yanagimachi, 1984). Additional evidence suggests that after gamete adhesion, a transmembrane region of IZUMO1 is involved in protein dimerization, which allows dimeric IZUMO1 to interact with a putative oocyte receptor (Inoue et al., 2015). To date, JUNO and IZUMO1 are the only egg receptor and sperm ligand known to interact directly to guarantee sperm-egg adhesion and this binding ability is conserved in several mammalian species (Bianchi and Wright, 2015; Grayson, 2015).

The mechanisms mediating the transition between gamete adhesion and fusion are still uncertain. Nevertheless, it is conceivable that during gamete adhesion, certain proteins may act in cis to recruit other molecular players which would mediate gamete fusion.

Fusogen proteins mediate fusion of membranes during cell interactions. Examples of fusogens include the mammalian syncytins, which are necessary for cytotrophoblast fusion during placentation, the Caenorhabditis elegans Epithelial Fusion Failure-1 (EFF-1) and Anchor-cell Fusion Failure-1 (AFF-1) proteins, which mediate cell-fusion mechanisms during embryo/larval development (reviewed by Hernández and Podbilewicz, 2017). Despite intensive investigation, the existence and identity of gamete fusogens remain uncertain and our understanding of the mechanisms mediating gamete fusion are incomplete.

One model for cell-cell fusion that might be applied to fertilization arises from studies on myoblast fusion. Upon cell adhesion, myoblasts generate actin-propelled membrane protrusions that invade the other cell, which responds in turn through a myosin II-mediated increase of cortical tension to the myoblast invasion. The contrast between invasive and resistance forces disrupt the lipid bilayers of each cell, leading to the formation of fusion pores (Kim et al., 2015). This compelling model is reminiscent of early observations documenting species-specificity in gamete fusion between human sperm and hamster oocytes (Yanagimachi and Noda, 1970). In these IVF assays, the zona was removed from ovulated hamster oocytes and denuded eggs were inseminated with human sperm. Soon after gamete adhesion, protruding microvilli from the egg surrounded the human sperm, leading to gamete fusion (Yanagimachi, 1984). As mentioned above, this cross-species gamete fusion event is dependent on the IZUMO1-JUNO interaction (Bianchi and Wright, 2015; Han et al., 2016). The ‘myoblast fusion’ model raises the prediction that interaction between JUNO and IZUMO1 should be sufficient to induce gamete fusion. To test this hypothesis, IZUMO1 and JUNO were ectopically expressed in eukaryotic COS-7 cells, and despite the occurrence of cell-adhesion, no cell fusion was recorded. These data indicate that the IZUMO1-JUNO interaction is not sufficient to induce cell fusion (Inoue et al., 2015), which is consistent with the fact that neither IZUMO1 nor JUNO show sequence similarities to currently identified membrane fusogens (Bianchi and Wright, 2015).

One final model envisions a pre-fusion step (in between gamete adhesion and fusion), during which the direct bond between IZUMO1 and JUNO generates an accumulation of CD9 in the adhesion area (Chalbi et al., 2014). By acting independently (Ohnami et al., 2012), both CD9 and CD81 directly interact with three oocyte tetraspanin partners, CD9P-1 (PTGFRN), α6β1 integrin and IGSF8 (also known as EWI-2). In addition, CD9 and CD81 may recruit in cis other proteins (although still undefined) to mediate gamete fusion (Jegou et al., 2011; Ohnami et al., 2012; Rubinstein et al., 2006). In particular, studies in vitro have shown that gamete fusion might be mediated by a three-residue epitope (SFQ) in the EC2 domain of CD9, which organizes a tetraspanin protein network. More specifically, wild-type CD9 mRNA partially rescues the ability of Cd9^Null eggs to adhere with sperm, whereas mutant CD9 mRNA that lacks the SFQ epitope fails to rescue the fusion competence in Cd9^Null eggs (Zhu et al., 2002). Moreover, incubation of wild-type eggs with monoclonal antibodies against the SFQ epitope can inhibit sperm-egg fusion in mice, sheep and goats (Xing et al., 2010), but not gamete adhesion in mice (Miller et al., 2000; Zhu et al., 2002). Incubation of the recombinant EC2 domain with sperm does not preclude gamete adhesion and fusion, which implicates that sperm may lack a receptor acting as a trans partner for CD9 (Miller et al., 2000; Zhu et al., 2002). Thus, upon IZUMO1-JUNO interaction, it is conceivable that the SFQ epitope in the EC2 domain may play a part in mediating gamete fusion by recruiting still-unidentified fusogens. Then, the sperm and egg fusogen interactions would mediate fertilization (Fig. 3C).

Mammalian eggs cannot tolerate the physiological polyspermy observed in other vertebrate species (such as urodeles, elasmobranchs, reptiles and birds) owing to the absence of a mechanism that selects one single male pronucleus for interdigitating with the female pronucleus. Hence, polyploidy in mammals is embryonic-lethal: early studies in humans show ∼20% of spontaneously aborted concepti to be polyploid (Hassold et al., 1980), of which 66-69% are diandric – the result of the fertilization by one diploid sperm or by two sperm (Jacobs et al., 1978; Zaragoza et al., 2000). Therefore, promptly after fertilization cortical granule exocytosis releases zinc (Box 3) and a cortical granule-specific protease that biochemically modifies the zona to prevent supernumerary sperm binding with and penetrating the zona matrix (known as the ‘zona block’) (Stewart-Savage and Bavister, 1988). Moreover, molecular changes at the oolemma prevent other sperm adhering to and fusing with the oolemma (referred to as the ‘membrane block to polyspermy’) (Sato, 1979). These events together are defined as the block to polyspermy, which ensures monospermic fertilization (Fig. 4). These distinct biological processes are independent of each other and work with different efficiencies in different mammalian models.

In mammals, the membrane block does not depend on the rapid membrane depolarization (Jaffe et al., 1983) that occurs rapidly (a few seconds) after fertilization in species such as the starfish (Tyler et al., 1956) and the sea urchin (Jaffe, 1976). Studies in mice have shown that zona-free, unfertilized oocytes remain competent for fusion for several minutes before a membrane block is established (Wolf and Hamada, 1977). One important aspect to note is that this membrane block is independent of sperm entry into the cytoplasm (Wolf and Hamada, 1977): when one-cell zygotes or two-cell embryos were generated by intracytoplasmic sperm injection (ICSI) (Wolf and Hamada, 1977), which consists of the injection of a single sperm in the oocyte cytoplasm, bypassing gamete fusion (Horvath et al., 1993; Maleszewski et al., 1996), their cell plasma membranes still presented the ability to fuse de novo with other sperm.

The molecular mechanism that regulates the membrane block in mammals remains incompletely understood. In mice, a twofold decrease in CD9 protein content and a reorganization of microvilli distribution have been reported to play a putative role in the membrane block (Żyłkiewicz et al., 2010). Moreover, shedding of JUNO upon fertilization impedes further sperm adhesion to the fertilized egg oolemma (Bianchi et al., 2014) (Fig. 4, 2). However, as JUNO appears to persist for ∼40 min after fertilization, other mechanisms may account for the effective membrane block to polyspermy at the oolemma. Unfortunately, to date, no fusion molecule(s) have been identified on either the egg or sperm plasma membrane, and what the exact mechanism mediating the membrane block is remains a compelling, but still unanswered, biological question.

A better understanding of the molecular basis of the membrane block will be beneficial to elucidate the complexity of the different effective membrane blocks observed among mammalian species. In fact, in vitro and in vivo observations appear to indicate that this membrane block works with different efficiencies in different mammalian species. For example, tens to hundreds of sperm are found in the perivitelline space of one-cell zygotes recovered from rabbit, pocket gopher and mole females, whereas in other species such as mice, rats, cats, pigs, dogs, sheep and humans perivitelline sperm are rarely observed (Gardner and Evans, 2006). Such observations raise the prediction that the former group may present a more effective membrane block, whereas the latter possess a more effective zona block.

In the last few years, the molecular mechanisms mediating the zona block have been extensively characterized in transgenic mice. Early studies report that, within hours of fertilization, the zona from fertilized eggs loses the ability to support sperm binding (Bleil et al., 1981). Thirty minutes after fertilization, the initiation of a post-fertilization cleavage of ZP2 at a diacidic residue in the N-terminal region is the only biochemical modification documented (Bleil et al., 1981) (Fig. 2A), and mutation of this ZP2 cleavage site permits sperm to bind to the zona surrounding fertilized eggs (Gahlay et al., 2010). This post-fertilization cleavage of ZP2 occurs following cortical granules exocytosis and a precise translocation of the cortical granules to the cortex of the oocytes is necessary to preserve the zona block. This translocation is guaranteed by a RAB27a-regulated cytoplasmic actin network (Cheeseman et al., 2016; Bhuin and Roy, 2014). In ashen mice, a splice site mutation in Rab27a generates a nonfunctional RAB27a protein (Wilson et al., 2000), which is associated with the disruption of cortical granules translocation and results in the accumulation of supernumerary perivitelline sperm in fertilized homozygous Rab27a mutant eggs (Cheeseman et al., 2016). In addition, recent studies used high- and super-resolution imaging to document cortical granule dynamics at single granule resolution in transgenic mice (Vogt et al., 2019). In these studies, MATER, a component of the subcortical maternal complex (SCMC) of the oocyte has been shown to be necessary in anchoring the cortical granules at the oocyte cortex. Conditional genetic ablation of MATER in the oocyte has been associated with a spread distribution of the cortical granules throughout the cytoplasm, which delays cortical granule exocytosis after fertilization. This prevents the post-fertilization cleavage of ZP2, which results in supernumerary sperm in the perivitelline space (Vogt et al., 2019). Upon gamete fusion, the oocyte internal calcium concentration increases, which prompts the cortical granules to fuse with the oolemma and exocytose their contents of trypsin-like proteases, ovoperoxidase, N-acetylglycosaminidase and zinc (Ducibella et al., 2002) (Box 3) (Fig. 1, 5).

The cortical granules also release an oocyte-specific astacin-like metallo-endoprotease encoded by the Astl gene, named ovastacin, which recognizes the cleavage site on ZP2 (Burkart et al., 2012) (Fig. 4, 3). Ablating the gene encoding ovastacin prevents cleavage of ZP2, and as a consequence, zona from these mutant eggs still support sperm binding after fertilization (Burkart et al., 2012). Ovastacin cleaves ZP2 with species-specificity; mouse ovastacin does not cleave human ZP2 in transgenic mice, which allows transgenic zonae surrounding two-cell embryos to support de novo mouse sperm binding (Baibakov et al., 2007). To ensure monospermic fertilization, ovastacin must act at the right time and in the right place. A precocious activity of ovastacin on the zona before fertilization would be detrimental, as it would induce a premature block to sperm binding that would, in turn, cause female infertility. The ovastacin signal peptide localizes the protein ectodomain into the endomembrane system, in which it is then sequestered into the cortical granules of the oocyte cortical area. During oocyte maturation, the precise localization of ovastacin in the cortical granules is ensured by a seven amino acid motif (Xiong et al., 2017). Indeed, deletion of this motif affects ovastacin localization because of the mutant protein being retained in the endomembrane system and leads to a premature and partial cleavage of ZP2. With the zona of MII eggs unable to support normal sperm binding, female fertility is severely affected (Xiong et al., 2017). These observations are puzzling; it is conceivable to expect that the retention of mutant ovastacin in the endomembrane system could inhibit ZP2 cleavage, thus leading to a faulty polyspermy block. However, it could be hypothesized that the mutant ovastacin is released at higher concentration compared with the normal physiological release of native ovastacin observed shortly before ovulation (Ducibella et al., 1988). This continuous release would be the reason for the partial premature ZP2 cleavage observed (Xiong et al., 2017) and would be consistent with previously reported phenotypes associated with premature cleavage of ZP2. FETUB, a cystatin superfamily protein (thiol protease inhibitors), plays an important role by inhibiting the proteolytic action of the premature physiological release of ovastacin that occurs before ovulation (Ducibella et al., 1988). The effects of FETUB last until cortical granule exocytosis when the overcoming amount of ovastacin acts on the zona by cleaving ZP2 (Dietzel et al., 2013). Female mice lacking FETUB are infertile, owing to a premature cleavage of ZP2 by an early release of ovastacin from the cortical granules (Dietzel et al., 2013). In addition, a block to sperm penetration through the zona, dependent on cortical granule exocytosis (Inoue and Wolf, 1975), acts transiently. This has been recently shown using eggs expressing a mutant ZP2 that encodes a protein that remains uncleaved after fertilization and cortical granule exocytosis (Gahlay et al., 2010). The study showed that after artificial induction of cortical granule exocytosis by exposure to strontium chloride, the zona of eggs inseminated de novo with mouse sperm are capable of supporting sperm binding, although it remains refractory to sperm penetration (Tokuhiro and Dean, 2018). Nine hours post-cortical granule exocytosis this block disappears, and the mutant zona becomes permissive for penetration (Tokuhiro and Dean, 2018). This transient block appears to be mediated by zinc released upon fertilization (Que et al., 2017) (Box 3) that, together with a block to fusion and block to zona binding, provides an effective block to polyspermy in mammals that is imperative for the successful onset of development.

In conclusion, a number of important recent discoveries using genetically modified mice have led to a novel model for fertilization, but fundamental questions remain answered. The current model envisions sperm binding to the zona via the N terminus of ZP2, which predicts the presence of a putative sperm receptor yet to be discovered (Avella et al., 2014; Raj et al., 2017; Tokuhiro and Dean, 2018) (Fig. 2). Indeed, this putative sperm receptor might also be involved in the induction of sperm acrosome exocytosis during sperm penetration through the zona, however, the location and regulation of sperm acrosome exocytosis remains poorly understood (Inoue et al., 2011; Jin et al., 2011). After zona penetration, direct interaction between IZUMO1 and JUNO (Aydin et al., 2016; Kato et al., 2016; Ohto et al., 2016) mediates gamete adhesion (Bianchi et al., 2014) and drives accumulation of CD9 in the adhesion area (Chalbi et al., 2014). CD9 recruits in cis other oocyte proteins, which need to be identified (Fig. 3C). Similarly, whether other undefined sperm proteins form complexes upon sperm adhesion with the oolemma to mediate gamete fusion remains to be shown (Ellerman et al., 2009). Following fertilization, the cortical granules at the periphery of the oocyte undergo exocytosis to release zinc and ovastacin that cleaves ZP2 at the N terminus. Zinc release affects the motility of supernumerary sperm penetrating the zona, whereas cleavage of ZP2 and shedding of JUNO from the oolemma impede further sperm binding and fusing with the egg (Bianchi et al., 2014; Burkart et al., 2012; Tokuhiro and Dean, 2018; Xiong et al., 2017) (Fig. 4).

It is remarkable that, despite decades of intensive investigation, we still know little about the molecular interactions mediating fertilization, potentially because of the nature of the extracellular interactions that regulate gamete recognition and fusion. Interactions that involve transmembrane proteins can be challenging to identify, because some receptor-ligand interactions present low affinity, are transient or require local protein clustering to increase binding efficiency. This can be circumvented by using recombinant oligomerized ectodomains as probes for the identification of cis/trans-acting factors controlling gamete recognition and fertilization (Bianchi et al., 2014). Finally, recent advances in chemoproteomic and genome editing technologies currently offer outstanding tools with which to address these and other long-standing issues in reproductive biology.

We apologize to colleagues whose work could not be cited owing to space limitations. We thank Drs Bonett and Toomey for critical comments on the manuscript.