The cell biology of mammalian fertilization

Our individual body has a limited lifetime. However, through fertilization we are able to continue life as a species. The role of spermatozoa is to fertilize eggs. However, mammalian spermatozoa cannot accomplish this task when ejaculated. They must first undergo a physiological change called capacitation and a subsequent morphological change known as the acrosome reaction in the female reproductive tract. Spermatozoa also harbor the ability to migrate into the oviduct, where they interact with and subsequently fuse with the egg. A number of factors that contribute to sperm-egg interactions have been identified, based on observations using enzyme inhibitors and antibodies in in vitro fertilization systems (Box 1). This research led to the conclusion that various sperm enzymes within the acrosome dissolved the egg components and that various membrane proteins were used for binding with eggs. However, recent experiments using gene disruption of these factors did not result in an infertile phenotype, suggesting that they are not essential for fertilization, although they may indeed play a role during the fertilization event. By contrast, using in vivo gene-targeting experiments, a number of proteins have unexpectedly emerged as being essential factors for fertilization. In this Primer, I discuss the factors that have been implicated in the various stages of fertilization, ranging from sperm capacitation and migration to sperm-egg fusion. Newly arising views of mammalian fertilization are reviewed and compared with previously postulated models.

Ovaries are endowed at birth with a fixed number of oocytes enclosed in primordial follicles. This number declines as a result of ovulation and atresia during the reproductive life of the female (Faddy, 2000). Oocytes are arrested in the dictyate stage of first meiotic division. Over time, cohorts of oocytes enter into a growth phase and become among the largest cells in the body, their diameters reaching about 80 μm and 100 μm in mouse and human, respectively. During growth, eggs form an extracellular matrix called the zona pellucida (ZP) by secreting glycoproteins (Fig. 1A). The ZP of human eggs consists of four ZP glycoproteins (ZP1 to ZP4), whereas that of the mouse egg consists of three ZP proteins (ZP1 to ZP3; mouse Zp4 is a pseudogene).

Full oocyte growth must be supported by surrounding granulosa cells, which proliferate and form multiple layers of cumulus cells that surround the ovulated egg (Fig. 1A). Cumulus cells support fertilization, and in vitro fertilization can be achieved more efficiently with them than without them (Jin et al., 2011; Tokuhiro et al., 2012). The molecular basis of this observation remains obscure; however, a few studies have addressed the function of cumulus cells (Oren-Benaroya et al., 2008; Shimada et al., 2008).

Following meiotic maturation, eggs are ovulated in preparation for fertilization. After ovulation, the eggs are picked up by adhesion between the extracellular matrix of the cumulus cells and oviductal cells of the oviductal fimbriae and are subsequently transferred into the oviduct (Talbot et al., 2003). The eggs are then transported to the ampulla portion of the oviduct, where they await spermatozoa for fertilization. However, there is only a short ‘fertile window’, which is less than a day after ovulation in humans (Wilcox et al., 1995) and a few hours in the mouse, during which fertilization can successfully occur.

In human testes, ∼1000 spermatozoa are produced per second (Amann and Howards, 1980), although the reason why mammalian males produce so many spermatozoa to fertilize so few eggs is not understood. Spermatozoa produced in the testes are transferred to epididymides, where they receive various proteins (Busso et al., 2007), probably in part through a structure called the epididymosome (Frenette et al., 2010). The spermatozoa are

mature (Fig. 1B), yet they remain unable to fertilize eggs. According to Yanagimachi, the sperm plasma membrane remains ‘biologically frozen’ until spermatozoa leave the male’s body and begin the ‘defrosting’ process, known as capacitation, which is necessary for spermatozoa to be fertilization competent (Yanagimachi, 1994).

During capacitation, various physiological changes occur in the head, acrosome and tail of spermatozoa, which biologically ‘defrost’ spermatozoa and prepare them for fertilization (Fig. 1B). It is known that spermatozoa in suspension are heterogeneous and, as such, when a sperm suspension is judged to be capacitated, a certain percentage of spermatozoa in the same suspension are already degrading. Numerous papers discuss the phosphorylation of sperm proteins during capacitation, but no clear mechanisms leading to capacitation have been elucidated (Bailey, 2010; Visconti et al., 2011). This may be due, at least in part, to the heterogeneity of spermatozoa.

Glucose is known to be essential for successful capacitation. It not only serves as an energy source allowing spermatozoa to swim, but also functions to enable spermatozoa to fertilize eggs (Goodson et al., 2012; Okabe et al., 1986). Cholesterol, bicarbonate, intracellular Ca²⁺ and many other factors are also involved in capacitation (Bailey, 2010; Evans, 2012; Florman and Ducibella, 2006; Yanagimachi, 1994); however, the precise molecular mechanisms underlying their action are yet to be clarified.

Once capacitated, spermatozoa demonstrate a vigorous swimming pattern called hyperactivation, a movement considered to give spermatozoa the strong thrusting power that allows them to penetrate the ZP. It is characterized by asymmetrical flagellar beating consisting of pro- and anti-hook bends accompanied by an increase in cytoplasmic Ca²⁺. The sperm-specific CatSper ion channels have been suggested to control the intracellular Ca²⁺ concentration and, thereby, the swimming behavior of spermatozoa (Chang and Suarez, 2011). CatSper1 to 4 are essential for fertilization in mouse (Qi et al., 2007). Furthermore, the CatSper channel seems to be activated by progesterone in human spermatozoa (Brenker et al., 2012; Lishko et al., 2011), which is interesting from the sperm-egg interaction viewpoint because cumulus cells are known to produce progesterone. However, progesterone is not reported to activate mouse spermatozoa.

After capacitation, the final step in sperm development prior to fertilization is the acrosome reaction (Fig. 1B). The acrosome, which is a subcellular organelle found at the apical tip of the sperm head, is filled with a variety of lytic enzymes and ZP-binding proteins. Through poorly understood mechanisms, which may involve SNARE proteins (De Blas et al., 2005), the plasma membrane and outer acrosomal membrane fuse and acrosomal contents are exocytosed into the environment (Yanagimachi, 1994). This process, referred to as the acrosome reaction, renders spermatozoa able to penetrate the ZP.

Fertilization is a complicated multi-step event, involving maturation and development of the spermatozoa and eggs, followed by sperm migration into the oviduct, and ending with sperm-egg interaction and then fusion. As such, various factors that play a role in each of these events are emerging.

The classical models of fertilization were mainly focused on sperm-zona binding and the zona-induced acrosome reaction, based on in vitro experiments, and many factors were postulated to be required for fertilization. With time, however, the genes encoding these factors were cloned and their roles were investigated by gene disruption experiments in mice (Box 2). Surprisingly, almost all of them showed no or a very minor phenotype in fertilization (Ikawa et al., 2010); although they may play a role, the knockout of these individual factors did not affect fertilization. In addition, various genes essential for successful fertilization began emerging from such gene disruption experiments. Most of these gene-disrupted mouse lines shared common phenotypes, showing: (1) no migration into the oviduct in vivo with aberrant zona-binding ability in vitro; or (2) no fusing ability (Table 1, Class I and II). Interestingly, all the spermatozoa from class I mouse lines lacked ADAM3 (Shamsadin et al., 1999) without exception, suggesting that ADAM3 plays a key role in fertilization in the mouse.

Below, I highlight the various factors that have been implicated in the fertilization process, placing an emphasis on those for which an essential role has been identified based on gene disruption approaches.

All of the 11 Class I lines of gene-disrupted males shown in Table 1 exhibited the same defect of sperm migration into oviduct, a defect which would not have been identified by in vitro studies. In mice, the uterus and oviduct meet in a structure called the uterotubal junction (UTJ), which significantly reduces the number of spermatozoa reaching eggs. The flow in the UTJ is bidirectional. After coitus, spermatozoa must move up to reach the eggs and, after gamete fusion, embryos resulting from fertilized eggs must move down the UTJ for implantation into the uterus. How the UTJ regulates this bidirectional flow is not known. It is clear, however, that sperm migration is not regulated by simple opening and closing of the UTJ entrance in mouse. A testis-specific molecular chaperone, calmegin, is one of the essential genes required for spermatozoa to migrate into oviduct (Ikawa et al., 1997). Using chimeric mice that ejaculate both wild-type and GFP-tagged, calmegin-disrupted spermatozoa, it was observed that only wild-type spermatozoa migrated into the oviduct, whereas the equally motile calmegin-disrupted spermatozoa remained in the uterus (Nakanishi et al., 2004). These observations indicated that entry into the oviduct is regulated by a recognition system between an individual spermatozoon and the UTJ, but the molecular mechanism underpinning this system remains to be determined.

The sperm-zona interaction is regulated almost in an all or none manner, as illustrated in Fig. 2. In vitro, many spermatozoa can bind (or attach) to the ZP of unfertilized eggs but they cannot bind to two-cell-stage eggs, suggesting that sperm-zona binding might be a controlled event. The loss of zona-binding ability, which was commonly observed in Class I knockout mouse lines, could easily be assessed by mixing spermatozoa with cumulus-free eggs, and initially no one doubted that the infertility was caused by the loss of zona-binding ability. However, this assumption was not correct as initially thought.

Mice deficient for the testis-specific molecular chaperone PDILT produce spermatozoa that, based on in vitro experiments, exhibit impaired zona-binding ability (Tokuhiro et al., 2012). After crossing with a transgenic fluorescently tagged sperm line (Hasuwa et al., 2010), the failure of sperm migration into the oviduct was indicated by live imaging. However, when the spermatozoa were deposited directly into the ampulla by bypassing the UTJ, the eggs could be fertilized (Fig. 2). Likewise, spermatozoa from infertile Adam1a^-/- mice, which also show impaired zona-binding ability in vitro, could fertilize eggs that were covered by cumulus cells (Nishimura et al., 2004). It is not known what the true sperm-zona interaction entails but these studies suggest that the so-called ‘zona-binding ability’ as measured by binding to cumulus-free eggs in vitro appears to be dispensable for fertilization in vivo (Fig. 2).

Many studies have also postulated the involvement of carbohydrates in sperm-zona binding. It must be noted, however, that many of these postulations were based on in vitro studies, and an essential role for such carbohydrate residues in vivo has not yet been confirmed. The enzymatic removal of terminal Gal (Bleil and Wassarman, 1988) or GlcNAc (Shur and Hall, 1982) residues from eggs, for example, abolished the ability of ZP3 to inhibit sperm binding. However, gene-disrupted mice lacking these residues on their zonae were fertile (Asano et al., 1997; Thall et al., 1995), suggesting that these residues are not essential per se for fertilization. Fucose in Lewis X- and A-containing glycans was also indicated to play a role in sperm-zona binding (Kerr et al., 2004), but mice lacking the corresponding fucosyltransferase (Fut9) were fertile (Kudo et al., 2004). A role for mannose present on glycans has also been proposed (Cornwall et al., 1991) but has been negated by N-glycanase treatment (Florman and Wassarman, 1985). In vitro experiments have also ascribed sperm binding activity to O-linked oligosaccharide side chains attached to Ser332 and Ser334 of ZP3 (Chen et al., 1998). However, transgenic mice harboring mutations in Ser332 and 334 were fertile (Liu et al., 1995). Moreover, glycosylation was not observed on these residues on native ZP proteins (Boja et al., 2003). Furthermore, disruption of the ZP3-binding protein SP56 (ZP3R - Mouse Genome Informatics) on spermatozoa (Bookbinder et al., 1995) resulted in almost no phenotype in fertilization (Muro et al., 2012). Finally, enzymes that make core glycans have also been disrupted, and eggs lacking complex and hybrid N-glycans as well as core-1-derived O-glycans were fertilized normally (Shi et al., 2004; Williams et al., 2007). The hypothesis that carbohydrates play an important role in sperm-egg interaction thus requires further support.

A recent view that has been verified by gene-manipulated animals is that the zona loses its affinity for spermatozoa when ovastacin (also known as astacin-like metalloendopeptidase), an enzyme contained in egg cortical granules (Burkart et al., 2012), cleaves ZP2 after fertilization (Bleil et al., 1981; Gahlay et al., 2010). In addition, it has been suggested that liver-originated circulating glycoprotein fetuin-B is essentially required to inhibit premature cleavage of ZP2 by inhibiting ovastacin (Dietzel et al., 2013).

Together, these studies have shown that various sperm-zona interactions can occur but that many of these may not be absolutely required for successful fertilization. We also must remember that the ‘sperm-zona binding’ described above may not reflect the real sperm-egg interaction required for fertilization in vivo, as shown in Fig. 2.

Anatomical or behavioral differences between species guarantee that a male and a female from two different species do not normally mate. Nevertheless, taxon-specific zona penetration has been demonstrated (Rankin et al., 2003). However, if spermatozoa do not require a zona-binding ability for fertilization, how can they distinguish the ZP of the same species? One can anticipate that there is an as yet unclarified affinity between the cumulus-covered eggs and spermatozoa. We can further speculate that the same molecule responsible for zona binding is utilized in sperm-UTJ recognition, which could explain why the defects in zona-binding and UTJ penetration take place inseparably (Table 1).

The sperm protein zonadhesin (ZAN) is reported to contribute to species-specific zona binding (Tardif et al., 2010), but this has not yet been proven by rescue of Zan^-/- by other species’ Zan. Furthermore, when mouse ZP proteins were replaced with human homologs by septuple gene manipulation (mZP1 to mZP3 triple gene disruption and hZP1 to hZP4 quadruple transgenic insertion), it was shown that human spermatozoa could bind to and penetrate the humanized ZP (Baibakov et al., 2012). These data may indicate taxon-specific zona penetration. However, Gahlay et al. also mentioned that mouse spermatozoa could also penetrate the humanized ZP, which, in a way, indicated that taxon non-specific zona penetration is possible for mouse spermatozoa. It should also be noted that field vole spermatozoa have been shown to penetrate mouse and hamster zonae pellucidaes without the need for an acrosome reaction (Wakayama et al., 1996).

Various enzymes considered to aid spermatozoa in penetrating the cumulus cell layers and the ZP are released by acrosomal exocytosis (Florman and Ducibella, 2006). The timing of the acrosome reaction was thus believed to be important for spermatozoa as they approach eggs. Therefore, many researchers believed that the acrosome reaction of fertilizing spermatozoa was induced upon contact with the ZP and that spermatozoa that have undergone the acrosome reaction before contact with the ZP have no fertilizing ability (Bleil and Wassarman, 1983). More recently, it was shown that the addition of solubilized ZP can induce the acrosome reaction (Florman and Ducibella, 2006), and a partial ZP3 sequence can accomplish the same effect (Hinsch et al., 2005). In this context, zona-binding proteins were assumed to induce a signaling cascade in spermatozoa (Gong et al., 1995). In fact, various zona-binding proteins were sought and purified from spermatozoa. However, the gene disruption of these factors resulted in no significant phenotype in fertilization [e.g. GalTase (Lu and Shur, 1997), Sp56 (Muro et al., 2012), zonadhesin (Tardif et al., 2010), Crisp1 (Da Ros et al., 2008), acrosin (Baba et al., 1994)]. Thus, although these factors may play a role in inducing the acrosome reaction, the importance of these factors, and a zona-induced acrosome reaction, during fertilization is unclear.

Early studies using transgenic spermatozoa with GFP in their acrosomes were unsuccessful in observing an acrosome reaction on the ZP, and addition of the calcium ionophore A23187 to zona-bound spermatozoa was required to induce an acrosome reaction (Nakanishi et al., 1999). Another group suggested that a penetrating action through the mesh structure rather than by surface binding could induce acrosome exocytosis, because they too could not observe an acrosome reaction on ZP-binding spermatozoa (Baibakov et al., 2007). Recent in vitro studies also suggested that spermatozoa did not require contact with the ZP to cause the acrosome reaction. Rather, most of the fertilizing spermatozoa were in fact acrosome-reacted before reaching the ZP in vitro (Jin et al., 2011). This, together with the report that spermatozoa can fertilize eggs without zona-binding ability under defined experimental conditions (Nishimura et al., 2004; Tokuhiro et al., 2012), questions the importance of the zona-induced acrosome reaction and whether it is physiologically relevant.

Interestingly, hamster spermatozoa complete the acrosome reaction around the time they pass through the cumulus, or shortly before, or after contacting the surface of the ZP in vivo (Cummins and Yanagimachi, 1982). Likewise, guinea pig spermatozoa participating in fertilization appear to undergo the acrosome reaction after reaching the proximal part of the oviduct or when they are very near the eggs (Yanagimachi and Mahi, 1976). In addition, rabbit spermatozoa recovered from the perivitelline space of fertilized eggs can still penetrate and fertilize fresh oocytes (Kuzan et al., 1984). These results suggest that the fertilizing ability of spermatozoa is maintained for a while after the acrosome reaction. Consistent with this notion, a compartmentalized structure within the acrosome (Kim et al., 2001) and asynchronous release of acrosomal proteins during acrosomal exocytosis have been reported (Hardy et al., 1991; Kim and Gerton, 2003; Nakanishi et al., 2001).

More recently, acrosome-reacted mouse spermatozoa were shown to penetrate fresh, cumulus-enclosed, zona-intact eggs, and the resultant fertilized eggs developed to term after uterine transplantation (Inoue et al., 2011). These results suggest that the timing of acrosome reaction can be flexible. It also suggests that the crucial sperm-zona interaction occurs between the ZP and acrosome-reacted, rather than acrosome-intact, spermatozoa.

Upon fusion, spermatozoa activate the egg thereby inducing calcium oscillations and completion of the second meiotic cell division (Miyazaki and Ito, 2006) PLCζ from spermatozoa is considered to be the responsible activator during this process (Nomikos et al., 2012), but other factors may also be involved (Harada et al., 2011). Activation of the egg leads to exocytosis from peripherally located cortical granules. Ovastacin is an enzyme that accumulates in cortical granules as described earlier, and is reported to result in cleavage of ZP2, which is considered to decrease the affinity to spermatozoa (Burkart et al., 2012). This phenomenon, which prevents polyspermy, is called the zona reaction.

To identify factors involved in sperm-egg fusion, the antigens of monoclonal antibodies that inhibit fusion in vitro (Blobel et al., 1992; Okabe et al., 1987; Okabe et al., 1992) were characterized, and this led to the identification of molecules such as CD46 (Taylor et al., 1994), IZUMO1 (Inoue et al., 2005) and fertilin (ADAM1b/ADAM2 heterodimer) (Blobel et al., 1992). To assess the roles of these molecules in vivo, researchers turned to gene-disruption experiments in mice. Initially, fertilin gathered attention but turned out not to be required for fertilization in vivo (Box 2). A Cd46-disrupted mouse line was also produced but, even though the testis is the only place in which CD46 is expressed in mouse, no effects on the fertilizing ability of spermatozoa from these mice was observed (Inoue et al., 2003). It is possible that these proteins may play redundant roles and, when knocked out, their function can be compensated for by other proteins. Alternatively, it is possible that these factors interact with or rely on other factors, and knockout of additional interacting factors is required to observe an effect in vivo (see Box 2 for a further discussion of the limitations of such knockout approaches).

Despite their potential limitations, gene disruptions sometimes bring us serendipitous findings. For example, the gene encoding the tetraspanin CD9 was initially disrupted in order to clarify its role in immunology, but it became the first known essential factor for sperm-egg fusion (Kaji et al., 2000; Le Naour et al., 2000; Miyado et al., 2000). More recently, we disrupted yet another gene, Izumo1, that was shown to be essential for fertilization; Izumo1^-/- mice produce normal-looking spermatozoa but are completely infertile. The Izumo1-disrupted spermatozoa are able to penetrate cumulus layers and the ZP normally but they fail to fuse with eggs (Inoue et al., 2005), thus highlighting the importance of IZUMO1 in the sperm-egg fusion event.

Based on such gene-disruption studies, CD9 on the egg and IZUMO1 on spermatozoa are the only two essential factors for fusion described to date. However, interaction between the two factors has not been observed and, structurally, neither factor has a fusogenic domain, which suggests that they might act via additional interacting proteins. However, the recent identification and disruption of an IZUMO1-associating protein (angiotensin converting enzyme 3; ACE3) resulted in fertile males (Inoue et al., 2010), implying that there must be other additional factors that are involved in fusion.

To assess the role of IZUMO1 during fertilization, the localization of IZUMO1 at the time of fusion was investigated using mCherry-tagged IZUMO1. IZUMO1 is initially hidden under the plasma membrane before the acrosome reaction but then moves out to the plasma membrane from the outer acrosomal membrane during the acrosome reaction in live spermatozoa (Satouh et al., 2012). This suggests that an additional function of the acrosome reaction is to render spermatozoa capable of fusing with the egg by transferring IZUMO1, and possibly other proteins, onto the sperm surface (Fig. 3A). Of note, three IZUMO1 localization patterns after acrosome reaction were observed: (1) an equatorial (EQ) fluorescence pattern; (2) a head (H) fluorescence pattern; or (3) a dim acrosomal cap (AC) fluorescence pattern (AC_dim) (Fig. 3C-E). The formation of the AC_dim pattern was noteworthy, because this pattern was considered to be produced by the phagocytosis of the inner acrosomal membrane after the first (marginal) fusion event is accomplished, with part of the sperm membrane thereby detaching from the fusion plane (dividing fusion) (Fig. 3D,E). The formation of an AC_dim pattern is the product of fusion step 2 (Fig. 3F). As each fusion event has a characteristic nature, each might require individual fusion factors. In any case, we need to identify more factors involved in sperm-egg fusion.

A new view (Fig. 4) on the mechanism of fertilization is emerging. This view includes diminished importance of the so-called ‘zona-binding ability’ of sperm, a zona-independent acrosome reaction, and the zona-penetrating ability of spermatozoa recovered from the perivitelline space. This updated view is largely based on in vivo observations of gene-manipulated animals, whereas classical models were mainly centered on observations of in vitro fertilization (IVF). In IVF conditions, eggs must be exposed to 10⁵ spermatozoa per ml (comprising a mixture of acrosome-intact, acrosome-reacted, and degrading spermatozoa). However, only a few (presumably acrosome-reacted) spermatozoa approach eggs for fertilization in vivo. Furthermore, recent studies have shown that many factors reported to be important for fertilization in vitro can be genetically disrupted with little or no perturbation of fertility (although see Box 2 for further discussion of potential reasons for this). These studies emphasize that, to avoid misplaced enthusiasm that could mislead the field, it will be important to validate candidate molecules in gene-manipulated animals in order to focus on those essential for fertility. Fertilization might be one of the most suitable research fields to require the use of gene-manipulated animals to perform in vivo studies, as the mysterious behavior of gametes is seemingly difficult to reproduce in vitro.