Fertilization is the process of union of two gametes. It occurs in several steps and culminates by the encounter of highly differentiated gametes. This interaction follows specific steps including recognition, adhesion and fusion. On the mammalian oocyte, JUNO and CD9 are the only two molecules known to be essential for adhesion/fusion during fertilization (Kaji et al., 2000; Le Naour et al., 2000; Miyado et al., 2000; Bianchi et al., 2014). While JUNO and its sperm partner IZUMO1 are now clearly identified as adhesion molecules (Aydin et al., 2016), the mechanism by which CD9 functions is still not completely understood, with most evidence pointing to an architectural or organizational role with other proteins embedded within the membrane (Runge et al., 2007; Jegou et al., 2011; Chalbi et al., 2014; Benammar et al., 2017). Indeed, the tetraspanin family, including CD9, contributes to the structural organization and plasticity of the plasma membrane. Regarding the architectural aspect, the concentration of CD9 molecules on microvillar membrane was found to be greater than that on the planar membrane. In addition, the complete absence of CD9 results in altered microvillar morphology, particularly a change of the radius of curvature of microvilli that are two fold higher than on wild-type oocytes (Runge et al., 2007) while a small radius of curvature is thought to promote fusibility (Wilson and Snell, 1998; McMahon and Gallop, 2005). An isolated large extracellular loop of CD9 was shown to inhibit fusion when preincubated with eggs, but not with sperm (Zhu et al., 2002) suggesting that CD9 is not a receptor for a sperm molecule. Only one article suggested that CD9 may function in a trans manner during gamete fusion by binding to an immunoglobulin (Ig) super family member, namely PSG17 (Ellerman et al., 2003). But this protein has never been found on sperm. Conversely, and in the sense of its cis action, CD9 associates with other tetraspanins, integrins, Igs and other adhesion receptors and proteins (reviewed in ((Charrin et al., 2014)). There are other members of Ig superfamily expressed by the oocytes with which CD9 interacts. Oocyte CD9 co-immunoprecipitates with two Ig superfamily cis partners, EWI-2 and EWI-F, which could have a role in linking CD9 to the oocyte microvillar actin core. Recently, crystal and cryo-electron microscopic structure of human CD9 and CD9-EWI-2 complex showed the interaction between CD9 and EWI-2, mediated by small residues in the transmembrane region and protein/lipid interaction suggesting the possible role of EWI-2 as a bridge between tetraspanins and other proteins, required for remodeling of membrane resulting in the formation of complex protein network. Cryo-electron microscopic structure of the CD9-EWI-2 complex and the mutation analysis revealed the importance of the small residues in the transmembrane region for the complex formation between CD9 and EWI-2 and the partial involvement of the CD9 large extracellular loop in the complex formation with EWI-2 (Umeda et al., 2020).

EWI-2 and EWI-F are tetraspanin CD9 and CD81 partners. They are immunoglobulins that share a Glu-Trp-Ile (EWI) extracellular motif. EWI-2 is also called IGSF8, PGRL, CD316 or KCT-4 and EWI-F is also known as FPRP or CD9-P1. Both associate directly with tetraspanins CD9 and CD81 while they do not associate with other tetraspanins or integrins (Charrin et al., 2001; Stipp et al., 2001; Sala-Valdes et al., 2006). In somatic cells, they induce tetraspanin relocalization into filopodia and they interfere in cell motility and aggregation (Charrin et al., 2001; Sala-Valdes et al., 2006; He et al., 2009). EWI-2 association with α-actinin regulates T cell immune synapses and HIV viral infection (Gordon-Alonso et al., 2012). Moreover, the expression level of membrane EWI-2 in Cd9-null oocytes is drastically reduced (He et al., 2009).

We have previously shown that CD9 generates fusion competent sites on the egg membrane for mammalian fertilization. These CD9-induced adhesion sites, named pro-fusional (i.e., able to lead to fusion), with strong adhesion, would be the actual location where fusion occurs. Indeed, in the absence of CD9, the adhesion between the two gametes exists but remains abortive (Jegou et al., 2011). Regarding the adhesive strength that we measured experimentally, the crucial difference between wild type (WT) and Cd9-null eggs comes from the strong interaction events that are absent when CD9 is absent. These adhesions are stronger because of the strong connection between the membrane and the cytoskeleton, that is not realized directly but through other molecules suggesting an involvement of CD9 in this process. Under these conditions, the link would likely occur via CD9-direct partners such as EWI-2 and EWI-F present in both somatic cells and oocyte (Rubinstein et al., 2006; Sala-Valdes et al., 2006; Runge et al., 2007). In turn, these CD9 partners bind directly to Ezrin-Radixin-Moesin (ERM) that connect the actin cytoskeleton to the plasma membrane (Sala-Valdes et al., 2006). Actin filaments cross-linked with the plasma membrane are important for generating cell-surface specialization areas. This is achieved through the regulation of actin-binding proteins. ERM proteins are thought to belong to the molecules linking the actin cytoskeleton and the plasma membrane. ERM proteins are widely distributed proteins located in the cellular cortex, in microvilli and adherens junctions. They contain a membrane binding domain and an actin binding domain. They have been implicated in mediating actin-membrane linkage and in regulating signaling molecules. In T lymphocytes for example, phosphorylated ERM are located at the posterior part of the cell where they have a potential role in the induction of uropod formation (Serrador et al., 1997; Lee et al., 2004). Indeed, ERM can have two conformations: a non-phosphorylated inactive cytoplasmic form and a phosphorylated active form located beneath the plasma membrane where it interacts with the cytoplasmic domain of adhesion proteins (For review (Louvet-Vallee, 2000; Fehon et al., 2010)). Furthermore, the function of lymphocyte Moesin is required for efficient HIV-1 viral fusion and infection (Barrero-Villar et al., 2009). More specifically, Dmoesin, the only member of the ERM family present in Drosophila, is required during oogenesis to anchor microfilaments to the oocyte cortex (Polesello et al., 2002) and Ezrin-deficient mice show microvilli disorganization in the developing intestine (Saotome et al., 2004). Mouse oocytes express all three ERM family members but seem to be particularly enriched in Radixin (Louvet et al., 1996; Larson et al., 2010). Finally, it has been shown that the disruption of ERM function reduces cortical tension in metaphase II eggs (Larson et al., 2010). It is known that cortical tension decreases during meiotic divisions and that a soft cortex is essential for asymmetric spindle positioning in mouse (Chaigne et al., 2013) which is guided by a narrow window of cortical tension (Chaigne et al., 2015). Recently it has been shown that aneuploidy in mouse oocytes can be generated by artificially decreasing cortical tension (Bennabi et al., 2020).

The purpose of our work was to analyze the role of ERM and EWI-2 and EWI-F proteins in gamete fertilization. We injected small interfering RNA (siRNA) to inhibit ERM and/or EWI-2 and EWI-F gene expression into oocyte cytoplasm. To assess the functional and morphological effects of inhibition, we assessed the fertilization index and studied oocyte microvilli morphology and measured cortical tension.

Figure 1 illustrates the experimental set-up including the different steps, ranging from oocytes preparation, siRNA injection, in vitro maturation and zona removal to the various tests that have been applied, namely immunofluorescence staining, zona-free in vitro fertilization assay, electron microscopy analyzes and measurement of cortical tension.

The aim of our study was to assess the role of the oocyte EWI-2, EWI-F and ERM protein family as potential partners of CD9 in gamete adhesion/fusion processes.

Our data showed that fertilization was not affected by the inhibition of one of these proteins, the fertilization index being not decreased after inhibition of either EWI-2, EWI-F, Ezrin, Radixin or Moesin, separately. Conversely, injection of ERM siRNA or combined EWI-2/EWI-F siRNA inhibited fertilization, decreasing the FI by 59 and 43% respectively. The lack of single siRNA effect did not appear to be due to a lack of siRNA efficiency since the injection of EWI-F or EWI-2 siRNA induced significant decreases in protein expression (55 and 68% respectively), as well as Radixin siRNA alone caused a decrease in Radixin protein expression (56%) similar to that caused by the combined injection of ERM siRNA (62%). Thus, the lack of effect on fertilization could be explained by a redundancy mechanism, like already highlighted in the context of knockout gene experiments. Indeed, EWI-2 and EWI-F are both immunoglobulins known to be associated to tetraspanins in somatic cells as well as in oocytes (Charrin et al., 2001; Stipp et al., 2001; Charrin et al., 2003; Runge et al., 2007; Glazar and Evans, 2009; He et al., 2009). EWI-2-deficient mice have been demonstrated to be fertile. Their eggs retain the normal level and localization of CD9, resulting in normal microvilli formation (Inoue et al., 2012). While CD9 regulates the expression and localization of EWI-2, these results and ours indicate that EWI-2 and EWI-F do not regulate the expression and localization of CD9. These results also suggest a possible compensation by EWI-F. In addition, unlike the anti-CD9 antibody, an anti-EWI-2 antibody has moderate inhibitory effects on sperm-egg binding and fusion (Glazar and Evans, 2009). Similarly, in our previous experiments, a monoclonal antibody directed against the EWI-F (CD9P-1) molecule failed to inhibit fusion (Rubinstein et al., 2006). Otherwise, EWI-2 and EWI-F have been shown to associate directly to ERM in somatic cells (Sala-Valdes et al., 2006). ERM are thought to constitute a bridge between the actin cytoskeleton and the plasma membrane. Ezrin-deficient mice, that die by 3 weeks of age (explaining that their fertility cannot be explored), showed abnormalities in their intestinal epithelium morphogenesis with microvilli disorganization. It should be noted that Ezrin is the only ERM protein detected in the developing intestinal epithelium, and this is likely why no redundancy mechanism may occur in this case (Saotome et al., 2004). Radixin-deficient mice are viable and fertile (Kikuchi et al., 2002). Moesin-deficient mice are normal and fertile without any Ezrin or Radixin hyper-regulation (Doi et al., 1999). However, the mutations of the Dmoesin gene in drosophila have dramatic consequences on embryo development. Dmoesin is required during oogenesis for anchoring microfilaments to the oocyte cortex (Polesello et al., 2002), since in mutant oocytes, the subcortical actin network is detached from the cell membrane (Jankovics et al., 2002). This is consistent with the fact that no redundancy is possible in this non-mammalian species that lacks the ezrin and radixin genes. Indeed, ERM are encoded by three highly homologous genes that have ∼85% similarity at the amino acid level (For review (Louvet-Vallee, 2000; Fehon et al., 2010)).

Given that ERM proteins promote the CD9-cytoskeleton interaction (Sala-Valdes et al., 2006), we can hypothesize that ERM are also critical in gametes adhesion/fusion mechanism via this function. The fertilization inhibition obtained by siRNA injection was however less dramatic than the one we have reported in Cd9-knockout eggs (Jegou et al., 2011). This discrepancy could be due to the fact that although almost no pERM expression was detected by immunofluorescence after siRNA injection, it cannot be ruled out that few ERM proteins remained or that further ERM phosphorylation may take place. In this case, our results would underestimate their role. A significant decrease of protein expression necessitated 48 h after the injection of EWI-2/EWI-F siRNA, while 24 h post injection of ERM siRNA were sufficient to detect an important decrease suggesting that the EWI-2 and EWI-F proteins would be more stable than ERM proteins.

It has been shown that overexpression of ERMBMPs (ERM-binding membrane proteins) in collaboration with activated ERM proteins in several cell lines, induced formation and elongation of microvilli (Yonemura et al., 1999). A mixture of ezrin/radixin/moesin antisense oligonucleotides could induce the destruction of both cell-cell and cell-substrate adhesion, as well as the disappearance of microvilli (Takeuchi et al., 1994).

In addition, it has been shown that the mechanical properties of the cell cortex can predict the developmental potential of oocytes after fertilization in mice in humans (Yanez et al., 2016). Cortical tension is highly regulated during oocyte maturation and could also play a role during gamete adhesion/fusion.

In order to check if the functional finding, the fertilization decrease, resulted from microvilli morphological alteration and/or cortical tension deregulation, we measured the radius of curvature of microvilli on oocyte membranes and cortical tension after injection of ERM and/or EWI-2/EWI-F siRNA compared to CTRL siRNA. Indeed, the inhibition of the expression of these proteins also caused an increase in the radius of curvature of the oocyte microvilli and a decrease in cortical tension.

Several works have shown the link between the radius of curvature and fusion. A small radius of curvature at the tip of microvilli maximizes the efficiency of membrane fusion (Wilson and Snell, 1998; McMahon and Gallop, 2005). In several cases of cell fusion, a small radius of curvature of the membrane protrusions promotes fusion (Wilson and Snell, 1998). Similarly, the fusion efficiency is higher when vesicles have a smaller radius of curvature (Lentz et al., 1992). Otherwise, several studies showed the involvement of Ezrin in microvilli morphology (Saotome et al., 2004; Bonilha et al., 2006). Moreover, in accordance with our results, in thymoma cells cultured in presence of antisense oligonucleotides complementary to ERM sequences, microvilli disappear, whereas when each one of the ERM sequences was inhibited separately, microvilli remain (Takeuchi et al., 1994). Finally, the involvement of oocyte microvilli in gamete interaction is strongly suggested by the fact that this interaction has been reported to be restricted to the microvilli-rich region (Yanagimachi, 1978; Runge et al., 2007). In addition, consistently with our results, the radius of curvature of microvillar tips on wild-type oocytes was found to be half that of the Cd9-null oocytes (Runge et al., 2007). But contrary to our results, Cd9-knockout oocytes were also shown to have denser arrays of microvilli than wild-type oocytes. Surprisingly, we did not observe any difference in terms of density of microvilli after injection of ERM or EWI-2/EWI-F siRNA whereas this decreased after the combined injection of the two (Mix siRNA). This result seems to indicate that ERM and EWI-2/EWI-F would be involved in the formation of microvilli while CD9 would negatively regulate this one, probably by acting indirectly on the organization of other proteins. The absence of this negative regulation in the context of Cd9-KO could explain microvilli density increase.

If a correlation between the increase in the radius of curvature of the microvilli and the decrease in the fertilizing ability of the oocytes seems to be rational and very likely (Runge et al., 2007), the link between the decrease in cortical tension and that in the fertilization index is more questionable. Cortical tension measures the mechanical characteristics of both the membrane and the subcortical cytoskeleton. But the major contribution of cortical tension comes from the cytoskeleton. Therefore, an inhibition of the links between the two may result in a slight drop in cortical tension. Indeed, the difference we obtained in terms of cortical tension between the oocytes microinjected by CTRL siRNAs and those microinjected by EWI-2/EWI-F and ERM siRNAs were too small to explain the important difference in fertilization index between these two groups. Especially since even more important differences of cortical tension, such as those observed between GV and metaphase I or metaphase II oocytes do not completely prevent these oocytes, whatever their stage of maturation, from fusing with sperm (Kryzak et al., 2013). However, it is possible that optimal adhesion/fusion during fertilization requires a narrow range of cortical tension, but additional investigations are necessary to prove it.

Since we cannot totally exclude the existence of other molecules able to overcome the lack of EWI-2/EWI-F or ERM proteins, the confirmation of redundancy mechanism on one hand between EWI-2 and EWI-F and on the other hand between Ezrin, Radixin and Moesin remains to be provided by experiments using EWI-2 and EWI-F double-knockout mice and ERM conditional oocyte specific triple-knockout mice respectively.

In conclusion, we highlighted here the role of oocyte EWI-2, EWI-F and ERM in fertilization, specifically in the step of gamete adhesion/fusion. The ERM and EWI-2/EWI-F proteins seem involved in gamete interaction through their role in the formation and maintenance of functional oocyte microvilli. These results confirm the role of microvilli in gamete interaction and support the hypothesis of the existence of a link between CD9 and the cytoskeleton through EWI-2/EWI-F and ERM proteins.