The process of gamete fusion has been largely studied in the mouse and has revealed the crucial role of the tetraspanin CD9. By contrast, human gamete fusion remains largely unknown. We now show that an anti-α6 integrin mAb (GoH3) strongly inhibited human sperm-egg fusion in human zona-free eggs. Furthermore, a mAb directed against CD151, a tetraspanin known to associate with α6β1, partially inhibited sperm-egg fusion. By contrast, the addition of an anti-CD9 mAb to zona free eggs had no effect. The integrin α6β1, CD151 and CD9 tetraspanins were evenly distributed on human zona-intact oocytes. On zona-free eggs, the integrin α6β1 and tetraspanin CD151 patched and co-localized but the tetraspanin CD9 remained unchanged. CD9 mAb prevented α6β1 integrin clustering and gamete fusion when added prior to, but not after, zona removal. Antibody-mediated aggregation of integrin α6β1 yielded patches that were bigger and more heterogeneous in mouse oocytes lacking CD9. Moreover, a strong labelling of α6β1 could be observed at the sperm entry point. Altogether, these data show that CD9 controls the redistribution of some membrane proteins including the α6β1 integrin into clusters that may be necessary for gamete fusion.
Mammalian fertilization depends upon successful binding and fusion between the spermatozoon and the oocyte plasma membrane. The fusion process is mediated by a series of molecular interactions in which sperm fertilin, oolemmal integrins and members of the tetraspanin family have been suggested to play a role (Primakoff and Myles, 2002; Wassarman et al., 2001).
Integrins constitute a family of transmembrane αβ adhesion proteins involved in cell-cell and cell-extracellular matrix interactions. The β1 subunit can associate with 12 different α subunits (α1-11 plus αv) forming the largest subfamily of integrins. Integrins are considered as two-way signalling molecules (Liddington and Ginsberg, 2002). Integrin-mediated cell adhesion induces regulatory signals, including protein phosphorylation, that control cell growth, death, migration, gene induction and differentiation. Reciprocally, the integrin efficiency in mediating cell adhesion or ligand binding is modified upon stimulation with a variety of stimuli. This may be achieved through an increase of integrin affinity for ligand, but more likely, by an increased avidity which may be regulated by modifications of integrin clustering at the cell surface and lateral associations with other transmembrane proteins (Sanchez-Mateos et al., 1996; Bazzoni and Hemler, 1998; Schwartz, 1992).
Tetraspanins are surface proteins containing four transmembrane domains and forming complexes with each other and with various membrane proteins within a network of molecular interactions called the `tetraspanin web' (Boucheix and Rubinstein, 2001; Berditchevski, 2001; Hemler, 2003). They contribute to the formation of cell surface multimolecular complexes and thereby may participate in the functional regulation of molecules they associate with. For instance, the presence of a subset of β1 integrins in the tetraspanin web may explain in part how tetraspanins interfere with cell adhesion and migration (Boucheix and Rubinstein, 2001; Berditchevski, 2001; Hemler, 2003). Within the tetraspanin web, smaller primary complexes, containing only one tetraspanin tightly associated with one partner molecule have been identified, based on solubility in various detergents and on cross-linking experiments (Boucheix and Rubinstein, 2001; Hemler, 2003). Thus, two integrins present on the oocyte, namely the α6β1 and α3β1 integrins, form, on somatic cells, highly stoichiometric and stable complexes with the tetraspanin CD151 (Fitter et al., 1999; Serru et al., 1999; Yauch et al., 1998). A CD151 molecule deleted from its C-ter cytoplasmic end was shown to act as a dominant-negative inhibitor of α6β1-dependent cord-like structure formation, possibly through the regulation of α6β1 integrin adhesion strengthening, showing that CD151 can modulate the function of this integrin (Lammerding et al., 2003; Zhang et al., 2002). Biochemical data suggest that within the tetraspanin web, CD151 links α3β1 and α6β1 integrins to other tetraspanins, including CD9 (Serru et al., 1999; Yauch et al., 1998; Ito et al., 2003; Charrin et al., 2003; Berditchevski et al., 2002).
The potential role of the integrin α6β1 and the tetraspanin CD9 in mouse fertilization has been extensively studied. The role of the integrin α6β1 is still controversial. Based on antibody-inhibition assays, it had been suggested that the integrin α6β1 could be a receptor for sperm binding to the egg (Almeida et al., 1995), but this conclusion was not reached in a later study using a different method of removing the zona pellucida surrounding the egg (Evans, 1999). Recently, the analysis of oocytes from mice lacking the integrin α6 or β1 genes showed that these integrins were not essential for spermoocyte binding and fusion in the mouse (Miller et al., 2000). These data could not, however, completely exclude a role for the integrin α6β1. This integrin has been suggested to mediate the binding of fertilin β (ADAM 2) and cyritestin (ADAM 3), two members of the ADAM family (A Disintegrin And Metalloprotease) expressed on sperm (Chen et al., 1999; Chen and Sampson, 1999; Takahashi et al., 2001; Tomczuk et al., 2003), but others have concluded that fertilin β interacts with an alternate, but as yet unidentified, β1 integrin on the egg (Evans et al., 1997; Zhu and Evans, 2002).
The crucial role of CD9 in sperm-egg fusion was demonstrated by the inability of CD9 knock-out mouse oocyte to fuse with sperm (Le Naour et al., 2000; Miyado et al., 2000; Kaji et al., 2000) and by the ability of anti-CD9 antibodies to inhibit fusion (Chen et al., 1999; Le Naour et al., 2000; Miller et al., 2000). How CD9 functions in sperm-egg fusion is largely unknown. In a recent study, recombinant proteins including the large extracellular loop of CD9 were shown to partially inhibit fusion when added to the oocyte prior to insemination, but not when preincubated with sperm. This indicated that CD9 was probably not a sperm receptor and drew the attention to molecules that associate with CD9 at the oocyte surface and thus to the tetraspanin web (Zhu et al., 2002).
In humans, the fertilization mechanism is rarely studied, mainly because oocytes are rare and precious. While the expression of tetraspanins on human oocytes has not been reported, they express several integrins, including the α6 and β1 subunit (Fusi et al., 1992; Ji et al., 1998; Campbell et al., 1995b). We have previously suggested the implication of β1 integrin in the fusion process, as fused spermatozoa co-localize with β1 integrin oolemmal patches and as a function-blocking anti-β1 integrin mAb significantly inhibited spermoocyte fusion (Ji et al., 1998). In this study, we now demonstrate that CD9 controls the formation of integrin α6β1-containing clusters on the oocyte plasma membrane, both in human and mouse. Moreover, we show a role for the α6β1 integrin and the tetraspanin CD151 in human spermoocyte fusion.
In human zona-pellucida-free oocytes, the integrin α6β1 and the tetraspanins CD151 and CD81 are present in the same patches
We first investigated the presence of the tetraspanins CD9, CD81 and CD151 and of α6 and β1 integrin subunits on the oolemma (the oocyte plasma membrane) of human oocytes. All five molecules were detected on human unfertilized zona pellucida (ZP) intact oocytes (Fig. 1A-E).
Owing to French bioethical laws, in vitro human gamete fusion assays can only be performed using ZP-free oocytes. The ZP is a thick extracellular coat surrounding the egg that is composed of three glycoproteins (ZP1, 2, 3). It selects the fertilizing spermatozoon, triggers its acrosomal reaction and prevents the fusion with any extra spermatozoon (blockade to polyspermy) (Wasserman, 2001). As chemical and enzymatic ZP removal methods significantly alter oolemmal proteins and may reduce fertilization (Santella et al., 1992; Kinn and Allen, 1981; Kellom et al., 1992), a mechanical technique for ZP removal was performed using a pair of microdissection scissors under a binocular microscope (Ji et al., 1998). This technique is particularly gentle as no recovery time is necessary before insemination (Ji et al., 1998). However, it resulted in a modification of the α6 and β1 integrin subunits distribution as well as that of CD81 and CD151, that all formed patches (Fig. 1F,G,J,M). By contrast, CD9 remained evenly distributed (Fig. 1O) showing that the patching of surface molecules induced by ZP removal was restricted to a subset of surface molecules. These patches were observed whether labelling was done at 4°C or at room temperature, and were similar whether fixation was performed after or before labelling, indicating that patching was not induced by mAb after ZP removal.
Merging of α6 and β1 subunit staining (Fig. 1H) showed a co-localization of both integrin subunits in patches, suggesting the existence of oolemma α6β1 integrin heterodimers. All the α6 patches contained the β1 integrin subunit. However, some β1 integrin patches were devoid of α6, probably because it is associated with other α integrin chains. Interestingly, as shown in Fig. 1K, all α6 patches contained CD151, indicating their association on oocyte surface, as previously shown in other cell types (Fitter et al., 1999; Serru et al., 1999; Yauch et al., 1998). Some CD151 patches did not contain α6 integrin. In these patches, CD151 may interact with α3β1 integrin, another known CD151 partner (Serru et al., 1999) expressed on the human oocytes (Capmany et al., 1998; Campbell et al., 1995b). As shown in Fig. 1N, α6 patches are also co-localized with tetraspanin CD81 patches. In contrast, CD9 was not concentrated in α6 patches, which is consistent with CD9 being only indirectly associated with the α6β1 integrin (Serru et al., 1999; Charrin et al., 2003; Berditchevski et al., 2002).
α6β1 integrin and CD151 control human gamete fusion
In vitro fertilization of ZP-free human eggs was performed to test for a role of the α6β1 integrin and tetraspanins in the fusion process in human. As shown in Fig. 2A, in control experiments, the average number of sperms fusing with an oocyte was about 25 after 18 hours incubation. Except for ZP-removal, these experiments were done under the standard routine conditions used for in vitro fertilization of human intact oocytes before transfer to the uterus for clinical purposes. Because only one sperm fuses with zona-intact eggs under these conditions, these data further show the lack of blockade to polyspermy at the level of the plasma membrane in human. The same conclusion was reached using the SUZI technique, in which the number of sperms injected directly into the perivitelline space (the space between the ZP and the plasma membrane) correlated linearly to the number of fused sperm (Wolf et al., 1997; Tesarik and Kopecny, 1989). This is in contrast to what was reported in mice (Horvath et al., 1993).
As seen in Fig. 2A, pre-incubation of eggs with the anti-α6 blocking mAb GoH3 inhibited fusion in a dose-dependent fashion. GoH3 at a concentration of 10 or 50 μg/ml inhibited sperm-egg fusion by 71% and 96%, respectively (P<0.001). By contrast, an anti-α3β1 blocking mAb (P1B5) (range 10 to 200 μg/ml) had no effect on sperm-egg fusion (not shown). The anti-CD151 mAb also inhibited fusion by up to 50% (P<0.001) (Fig. 2B). This suggests that CD151 participates with the α6β1 integrin in the formation of a functional complex involved in human sperm-egg fusion. Surprisingly, although anti-mouse CD9 mAb inhibits fusion of ZP-free mouse oocytes with sperm (Chen et al., 1999), two anti-human CD9 mAb did not affect the fusion of sperm with human ZP-free oocytes (Fig. 2C). Anti-human CD81 mAb (Z81; range 10 to 200 μg/ml) had no effect either on gamete fusion (data not shown).
The tetraspanin CD9 controls α6 integrin patch formation and human gamete fusion
The lack of effect of the anti-CD9 mAb on human ZP-free oocytes was in apparent contrast to the mouse system in which CD9 mAb efficiently inhibits the fusion of ZP-free oocytes with sperm (Chen et al., 1999; Le Naour et al., 2000; Miller et al., 2000). We therefore hypothesized that in human oocytes, the formation of α6 integrin patches, induced by mechanical ZP removal, put the oocyte at a stage in which CD9 is no longer required for fusion. To challenge this hypothesis, oocytes were preincubated with anti-CD9 mAb, and the mAb was kept continuously present during ZP removal and the fusion assay. As shown in Fig. 3A, the presence of anti-CD9 mAb throughout the procedure inhibited α6 integrin patch formation upon ZP removal in a dose-dependent manner, and importantly, as seen in Fig. 3B, decreased fusion by 78% at 50 μg/ml [the number of fused sperm were 4±2 (mean ± s.d.) in treated oocytes versus 18±8 in control eggs; P<0.001]. The same experiment performed with the anti-human CD81 mAb (Z81, 100 μg/ml) did not inhibit sperm-egg fusion (data not shown). The CD9 mAb did not affect patching of CD151 and CD81 tetraspanins (Fig. 3C) and the CD81 mAb did not affect patching of the α6 integrin subunit (Fig. 3D).
CD9 controls antibody-induced α6β1 integrin redistribution on the mouse oocyte surface
The above data showed that in human oocytes, CD9 controls the formation of α6β1 integrin-containing patches. To further provide evidence that CD9 controls the lateral mobility of the integrin α6β1, integrins were aggregated at the surface of wild-type and CD9–/– ZP intact eggs using a combination of anti-integrin α6 and goat anti-mouse antibodies. The distribution of the α6β1 integrin was then studied by confocal microscopy. As a control, wild-type eggs that were fixed before labelling were analysed. A finely punctate distribution of this integrin in the microvillar region was observed both in control and CD9–/– oocytes (Fig. 4A). This is similar to the previously reported repartition of the integrin α6β1 on ZP-free eggs (Tarone et al., 1993; Takahashi et al., 2000). In wild-type oocytes, antibody-mediated aggregation of α6 integrins resulted in the formation of small, regularly arranged patches. By contrast, the patches were bigger and more dispersed in CD9-null oocytes (Fig. 4A). There were ∼40% less patches in CD9–/– oocytes after antibody-mediated aggregation of α6β1 integrins (Fig. 4B). Interestingly, aggregation of the integrin also resulted in a higher level of internalization in CD9–/– oocytes (data not shown). These results indicate that CD9 restricts the mobility of the integrin on the cell surface.
Redistribution of CD9 and α6β1 integrin during fertilization
The dual inhibition by anti-CD9 mAb of both patch formation and human gamete fusion when added before ZP removal strongly suggested that these integrin α6β1-containing patches were involved in fusion. Thus we examined the distribution of CD9 and integrin α6β1 during fertilization. The distributions of CD9 and α6β1 integrin were subjected to dramatic changes. Four major trends were observed. (1) A labelling of CD9 and α6β1 was observed at the site of interaction of the sperm and oocyte (Fig. 5D,K). (2) A large area surrounding the sperm entry point was devoid of CD9 and of integrin α6β1. This area became larger as meiosis 2 division proceeded (Fig. 5F,H). No particular staining of the sperm entry point could be observed except when the flagellum remained partially outside (Fig. 5J,M). (3) These molecules gathered into clusters that were heterogeneous in size and shape (Fig. 5F,H,J). These patches occurred rapidly as they were observed in eggs before completion of sperm nucleus decondensation and entry into telophase (Fig. 5F). (4) The α6β1 integrin and to a lesser extent CD9, accumulated in the mitotic cleavage furrow (Fig. 5J). Since this reorganization was not observed in CD9–/– oocytes (which do not fuse), we suggest that it was consecutive to fusion (data not shown).
CD9 is strongly expressed on the oocyte surface where it is essential for sperm-egg fusion (Le Naour et al., 2000; Miyado et al., 2000; Kaji et al., 2000; Chen et al., 1999; Miller et al., 2000). So far, the mechanisms by which it controls this process are unknown. An isolated large extracellular loop of CD9 was recently shown to inhibit fusion when preincubated with eggs, but not with sperm (Zhu et al., 2002). This indicated that CD9 was not a receptor for a sperm molecule. Because of the exceptional ability of CD9 to associate with other transmembrane proteins including the α6β1 integrin (Boucheix and Rubinstein, 2001), these data suggested that CD9 could in some way modulate the function of a sperm receptor on the oocyte surface.
Our data show that the fusion of human gametes is more dependent on α6β1 integrin than mouse gamete fusion. Indeed, the mAb GoH3 strongly inhibited the fusion (up to 96% inhibition), and this was achieved using concentrations that were similar to the concentrations necessary to block the binding of somatic cells to laminin, a major α6β1 integrin ligand (Sonnenberg et al., 1990). Consistent with this role of α6β1 integrin, we also found that a mAb directed to CD151, a tetraspanin that intimately associates with α6β1 (Serru et al., 1999), could partially inhibit the fusion. This is the first demonstration of a role for this tetraspanin in this process. This result contrasts with the finding that CD151 knockout mice are fertile (Wright et al., 2004). However, mice lacking CD151 appear to be grossly normal and healthy while human patients with mutations in this molecule have severe disorders of skin and kidney (Karamatic Crew et al., 2004). This discrepancy was suggested to be linked by the expression in the mouse, but not in humans, of a molecule closely related to CD151 that may compensate for CD151 deficiency in knockout animals. The strong inhibitory effect of the anti-α6 integrin mAb on human sperm-egg fusion contrasts with the controversial role of this integrin in the mouse. The notion that α6β1 integrin in mouse could play a role in gamete fusion came from a study showing that the anti-α6 integrin-blocking mAb GoH3 diminished in a dose-dependent fashion sperm binding to mouse eggs whose ZP had been previously removed by protease digestion. In this study, fusion was only inhibited at 400 μg/ml, concentration that almost completely blocked sperm binding to eggs (Almeida et al., 1995). More recently, these data were challenged by the finding that GoH3 did not inhibit binding of sperm to mouse oocytes after ZP removal using an acidic treatment (Evans et al., 1997), and that it did not reduce the fertilization rate of ZP-intact mouse oocytes at 500 μg/ml (Miller et al., 2000). Moreover, oocytes lacking the α6 integrin subunit could be fertilized (Miller et al., 2000). Thus the integrin α6β1 does not play an essential role in sperm-egg fusion in the mouse, although its participation cannot be excluded. This apparent discrepancy may be due to the different techniques used to remove the ZP. Protease treatment may modify some proteins resulting in a loss of function of the main receptor, making the egg susceptible to GoH3 inhibition (Miller et al., 2000). This is the reason why we used a totally chemical- and enzyme-free mechanical technique to remove the ZP in human eggs in which α6β1 appears crucial for fusion. This technique of zona removal appears less deleterious since in this condition the oocyte is immediately competent for fertilization and does not need further incubation time to recover its ability to be fertilized. Use of ZP-free eggs for sperm-fusion assays is common because its removes any possible influence of the ZP and cumulus cells. Moreover, for ethical reasons, removal of ZP is mandatory to study the process of fusion in human.
Further data will be necessary to determine the precise role of the α6β1 expressed by human oocytes in gamete fusion. By analogy with the mouse system, it is possible that in human, the integrin α6β1 functions as a sperm receptor through its ability to bind sperm ADAM proteins. Indeed in the mouse, based on the capacity of specific mAb to inhibit the binding of fertilin β recombinant proteins to the oocyte, and the direct binding to this integrin of fertilin β peptides, the integrin α6β1 has been suggested to be a receptor for fertilin β (ADAM2) (Chen and Sampson, 1999; Chen et al., 1999; Takahashi et al., 2001), but these data have not been confirmed by others (Evans et al., 1997; Zhu and Evans, 2002). More recently, this α6β1 integrin has been shown to mediate the binding of cyritestin (ADAM3), another member of the ADAM family expressed on sperm (Takahashi et al., 2001). The role of fertilin β in the fusion process is highlighted by the fact that sperm lacking this protein have a 50% decrease ability to fuse with the oocyte (Cho et al., 1998). In human, a peptide containing the integrin recognition sequence of human fertilin β (Phe-Glu-Glu) was shown to also inhibit sperm adhesion to oocytes and their penetration (Bronson et al., 1999).
The essential role of CD9 in sperm-egg fusion in the mouse has now been largely documented using knockout mice (Le Naour et al., 2000; Miyado et al., 2000; Kaji et al., 2000) and mAbs that have been shown to block sperm-egg fusion (Chen et al., 1999; Le Naour et al., 2000; Miller et al., 2000). Surprisingly, anti-human CD9 mAb failed to inhibit the fusion of human ZP-free oocytes with sperm, when added at the time of insemination, under conditions in which anti-mouse CD9 mAb inhibit the fusion in mouse. However, anti-CD9 mAb strongly inhibited the fusion when added to human intact oocytes and kept continuously present during ZP removal and fusion, indicating that mechanical ZP removal put the oocyte at a stage where CD9 is no longer required for fusion. A major effect of ZP removal in human was to induce the formation of patches containing the integrin α6β1 and CD151. Actually, CD9 mAb added before ZP removal inhibited the formation of these patches along with fusion, showing that these patches play a key role in human gamete fusion and that the role of CD9 in sperm-egg fusion (in human) is to control the formation of these patches. Study of CD9–/– mouse oocytes provided additional evidence for a role of CD9 in the control of α6β1 integrin lateral mobility. Indeed, antibody-mediated aggregation of α6β1 integrin yielded patches that were more dispersed and larger in the absence of CD9. We hypothesize that CD9 is necessary for the maintenance of a tetraspanin web to which the integrin α6β1 is linked, thus controlling its lateral mobility. Surprisingly, the tetraspanin CD151 still formed patches in the presence of the CD9 mAb. This suggests that this treatment may dissociate the interaction of the integrin α6β1 with CD151. Alternatively, as a fraction of CD151 formed patches not containing this integrin (Fig. 1), it is possible that only the patches containing CD151 but not the α6β1 integrin are still observed after CD9 mAb treatment.
The fact that a CD9 mAb inhibits the formation of the patches formed after ZP removal of human oocytes strongly suggests that it uncovers a physiological mechanism of patch formation. The perivitelline space of unfertilized oocytes (the space between the oolemna and ZP) contains a hyaluronan-containing matrix, well developed in humans but scant in rodents including the mouse (Talbot and Dandekar, 2003). The hyaluronan-rich matrix attaches to the plasma membrane of the oocytes. CD44, a major hyaluronan receptor is actually present on human oocytes (Campbell et al., 1995a) and has been shown to associate with tetraspanins (Jones et al., 1996). Moreover, during fertilization, acrosome-reacted sperm carry substantial amount of hyaluronidase localized at the inner acrosomal membrane that may facilitate fusion (Cowan et al., 1991). The hyaluronan-rich matrix attached to the membrane may have a stabilizing effect that is abolished upon ZP removal, which removes proteins and vesicles present in the perivitelline space. This role is further supported by the fact that adding exogenous hyaluronan decreases polyspermy during conventional porcine IVF (Suzuki et al., 2000). Therefore, although this remains speculative, it is possible that hyaluronan participates in maintaining unclustered the tetraspanin/integrin α6β1-containing membrane structures. The fact that hyaluronan-containing matrix is well developed in human and less in mice could explain the differences observed in integrin relocation upon ZP removal between these two species (Talbot and Dandekar, 2003).
In addition to the dual inhibition by CD9 mAb of fusion and patch formation, several lines of evidence suggest that α6β1 integrin-containing patches play a role in sperm binding and fusion. Concerning humans, we have previously shown that fused spermatozoa co-localized with β1 integrin patches (Ji et al., 1998). In the mouse (using ZP-free eggs), clustering of integrins at sperm contact and fusion occurred concomitantly, and the percentage of eggs in which the integrin α6 and β1 subunits appeared clustered at a sperm-binding site was closely correlated to the percentage of eggs with fused sperm. This led Takahashi et al. to suggest that sperm-egg fusion occurs at sites where the α6β1 integrin is clustered (Takahashi et al., 2000). In our study, we observed, before fusion, a strong labelling of CD9 and to a lower extent of integrin α6β1 at the site of interaction of the sperm with the oocyte. After fusion occurred, we detected a local high concentration of the integrin α6β1 (and to a lower extent CD9) when the flagella had remained partially outside the ZP. At latter time, the sperm entry point could not be detected based on CD9 and integrin labelling, suggesting that these molecules diffuse rapidly in the plasma membrane after fusion. This result is consistent with the study of Takahashi et al. who observed that the α6β1 integrin disappeared from the site of sperm penetration (Takahashi et al., 2000). These data further suggest a role for integrin α6β1 patches during sperm-egg fusion.
The non-essential role of the integrin α6β1 in the mouse (Miller et al., 2000) seems to conflict with a model in which α6β1 integrin-containing clusters play a major role in fusion. However, because other molecules (integrins and other types) associate with tetraspanins within the tetraspanin web (Boucheix and Rubinstein, 2001; Berditchevski, 2001; Hemler, 2003), it is possible that the actual sperm receptor(s) in the mouse oocyte, that remain(s) to be identified, also cluster(s) in a CD9-dependent mechanism. The recent demonstration that none of the integrins known to be present on the mouse egg are essential for sperm-egg binding and fusion suggests that the actual sperm receptor in mice is not an integrin (He et al., 2003). We propose that the α6β1-containing patches constitute a marker for the presence of multimolecular complexes assembling in large clusters necessary for fusion. Inside these clusters, the avidity for the ligand(s) may be increased. In this regard, it has recently been demonstrated that an anti-CD9 mAb inhibited the binding to eggs of fertilin β disintegrin domains immobilized onto beads, but not the binding of the soluble protein (Zhu and Evans, 2002). In contrast to soluble proteins, beads are subjected to strong detaching shear forces during washing steps, and thus their binding to oocytes depends not only on the affinity of the receptor to fertilin, but importantly on avidity. Our results suggest that this effect of the CD9 mAb on the binding of fertilin β-coated beads reflects a decreased avidity of the receptor linked to an inefficient clustering. This hypothesis is strengthened by the recent finding that the tetraspanins CD151 and CD81 regulate adhesion strengthening of the integrins α6β1 and α4β1, respectively (Lammerding et al., 2003; Feigelson et al., 2003).
We have also shown that fertilization induces a major redistribution of CD9 and of the α6β1 integrin. In particular, a large area surrounding the sperm lacked these two molecules after fusion. This is probably a consequence of the formation of a fertilization cone, that does not discernibly bind conA, in relation with the absence of microvilli in this region (Davies and Gardner, 2002; Maro et al., 1984). The intense co-localization of CD9 and integrins into heterogeneous patches, together with the control of integrin α6β1 diffusion by CD9, suggest that CD9 may play a role in this reorganization. However, CD9 does not play an essential post-fusion role, as intracytoplasmic sperm injection into CD9–/– eggs results in normal stage 2 embryo formation and implantation efficiency. Moreover, we have observed that the α6β1 integrin strongly accumulates in the mitotic cleavage furrow, raising the possibility that it may play a role in cytokinesis. In this regard, integrins are linked to the actin cytoskeleton, which reorganizes at the growing end of the cleavage furrow of Xenopus egg during cytokinesis (Noguchi and Mabuchi, 2001). Thus, the integrin α6β1 may also play a role in cytokinesis.
In general, mAb directed to various tetraspanins trigger the same functional effects, and this is thought to be related to their presence in the tetraspanin web (Boucheix and Rubinstein, 2001). In this study, we found that the three tetraspanins examined behave differently with respect to sperm-egg fusion in human. Thus, no effect of CD81 on fusion could be observed, although in previous studies anti-mouse CD81 mAb was shown to partially inhibit sperm binding to the oocyte and fusion (Takahashi et al., 2001), and CD81 was shown to compensate for the lack of CD9 when exogenously expressed in the oocyte (Kaji et al., 2002). These discrepancies may relate to differences in the human and mouse processes. Two other tetraspanins were shown to be involved at different stages, with CD9 controlling the formation of clusters that contain the tetraspanin CD151, the α6β1 integrin and possibly a sperm receptor at the surface of human oocyte. We have demonstrated for the first time that CD151 plays a role in sperm-fusion, since mAb directed to CD151 could partially inhibit this process. CD151 plays its role downstream of CD9 since this inhibition was achieved after the formation of patches induced by ZP removal. This is consistent with the tight association of CD151 with the integrin α6β1 (Serru et al., 1999) and with a role of this integrin in the fusion process in human. This is the first demonstration that two tetraspanins are involved sequentially in a physiological process and this shows the dynamic behaviour and plasticity of the tetraspanin web.
Materials and Methods
The function blocking mAb against human β1 integrin (DE9), and against α6 integrin (GoH3) were purchased from Chemicon International (London, UK). The mouse anti-human CD151 mAb TS151, the two mouse anti-human CD9 antibodies (ALB-6, SYB-1) and the rat anti-mouse CD9 antibody 4.1F12, coupled or not with Alexa Fluor 488, were described previously (Charrin et al., 2001; Le Naour et al., 2000; Rubinstein et al., 1996). 11B1G4 (anti-CD151) was provided by L. K. Ashman (Fitter et al., 1999), Z81 (CD81) by F. Lanza (Azorsa et al., 1999) and 1D6 (CD81) by S. Levy (Higginbottom et al., 2000). Fluorescein isothiocyanate-conjugated donkey anti-mouse immunoglobulin and rhodamine-conjugated donkey anti-rat immunoglobulin (Jackson Laboratories) were used for second antibodies in immunocytochemistry experiments.
Preparation of human gametes and fusion assays
Immature human oocytes and unfertilized human oocytes were donated by patients undergoing intracytoplasmic sperm injection and in vitro fertilization respectively, with the approval of local Ethical Committee. Exceptionally, fresh cohorts of mature oocytes were used when no sperm were available. Similar results were obtained with these fresh oocytes and unfertilized or in vitro matured oocytes for immunostaining and functional test of fusion. Unfertilized normal metaphase II (MII) stage human oocytes were collected 48 hours after insemination. In vitro matured metaphase II were obtained by incubating germinal vesicle or metaphase I stage oocytes under a 5% CO2 atmosphere at 37°C for 24 hours. ZP was removed with a chemical- and enzymatic-free technique, using a pair of microdissection scissors under a binocular microscope, in order to preserve the integrity of the surface proteins (Kellom et al., 1992).
Semen from donors of proven fertility were collected after 3 days of abstinence. Sperm samples were kept at 37°C until liquefaction was completed. A two-step 90/45% Puresperm® gradient (Pharmacia), was used to select motile spermatozoa. The sperm were then kept under capacitating conditions for 2 hours before use.
Sperm-egg fusion assay
ZP-free mature eggs were inseminated with 4000 capacitated motile human spermatozoa in 20 μl Ferticult Medium (FertiPro, Belgium) for 18 hours under paraffin oil. These are the standard conditions (except that ZP is not removed) for in vitro fertilization of human oocytes before implantation in women. In some experiments, the eggs were preincubated with different concentrations of mAb for 30 minutes before insemination. For analysis, oocytes were washed and loaded with DNA-specific fluorochrome Hoechst 33342 (Sigma) at 5 μg/ml for 20 minutes. After washing they were fixed in 4% paraformaldehyde in 1% PBS-BSA for 30 minutes at room temperature. Fusion was considered to have occurred when the nucleus of the spermatozoa stained with Hoechst and was decondensed. The samples were analysed using a Zeiss Axiophot microscope equipped with a camera and connected to an Imaging System Package (Applied Imaging, Newcastle-upon-Tyne, UK).
Preparation of mouse gametes
Preparation of oocytes
CD9–/– and inbred wild-type female mice (6- to 12-weeks old) were obtained as previously described (Le Naour et al., 2000). They were superovulated by intraperitoneal injections of 5 IU PMSG (Folligon, Intervet, France) and 5 IU hCG (Chorulon, Intervet, France) 48 hours apart. Fifteen to 16 hours after hCG injection, the animals were sacrificed by cervical dislocation. Cumulus-oocyte complexes were collected by tearing the oviductal ampulla, then inseminated in Whittingham medium with 3% BSA and 106 spermatozoa/ml.
Spermatozoa were collected from CD9+/+ male mice after squeezing two caudae epididymis in a 500 μl drop of Whittingham's medium supplemented with 30 mg/ml BSA under mineral oil (Whittingham, 1971). Spermatozoa were incubated at 37°C in 5% CO2 for 90 minutes for capacitation.
In vitro fertilization
Cumuli were inseminated with 106/ml capacitated spermatozoa for 3 hours. There were then washed and loaded with the DNA-specific fluorochrome DAPI at 1.5 μg/ml for 15 minutes. After washing they were fixed in 4% paraformaldehyde in 1% PBS-BSA for 30 minutes at room temperature. Fertilization was considered to have occurred when the oocyte nucleus and decondensed sperm head showed fluorescence under UV light and the second polar body had been extruded.
To analyse different fertilization steps, insemination was sequentially interrupted (at 30 minutes, 1 hour, 2 hours and 3 hours) by dropping them in 4% paraformaldehyde.
Fluorescent staining of eggs
Single or double staining of human eggs was performed by incubating intact or ZP-free eggs with primary antibodies for 1 hour at the following concentrations: anti-α6 (20 μg/ml) and/or anti-β1 (10 μg/ml) and/or anti-CD151 (10 μg/ml) or anti-CD9 antibodies (10 μg/ml). These incubations were followed by extensive washing with Ferticult® IVF culture medium prior to a staining procedure with a mixture of FITC-conjugated donkey anti-mouse IgG (10 μg/ml) and/or a rhodamine-conjugated donkey anti-rat IgG (10 μg/ml) antibody. Eggs were then fixed with 4% paraformaldehyde solution for 30 minutes. Oocytes were mounted in Immumount antifade solution and analysed under a fluorescence microscope. The primary antibodies were omitted for control staining. To verify that clustering was not mediated by the antibodies, labellings were also performed after paraformaldehyde fixation and identical results were obtained.
Fertilized oocytes were separated from unfertilized oocytes. Both were incubated with 20 μg/ml rat mAb GoH3 for 1 hour. Eggs were then treated with a donkey anti-rat IgG (10 μg/ml) antibody for 1 hour. For additional labelling of CD9, the eggs were treated with rat IgG and then with an Alexa Fluor 488-coupled CD9 mAb. The eggs were deposited in a small drop of medium on a Labteck coverslip and covered with mineral oil. Confocal analysis of α6 integrin and CD9 tetraspanin expression was performed with a TCS SP2 confocal microscope (Leica, Wetzlar, Germany), using a 63× objective. Images of about 100 sections, representing nearly half of the eggs were collected, and non-specific labelling in the ZP was removed in each using the program Paintshop Pro. The maximum projection function of the LCS software (Leica) was then used to superimpose the different sections.
Antibody-mediated aggregation of α6β1 integrin
Wild-type and CD9–/– intact eggs were successively incubated with anti-α6 GoH3 (20 μg/ml) and, after washing, with Alexa Fluor 488-labelled goat anti-rat antibodies (10 μg/ml) for 45 minutes at 37°C under 5% CO2. They were then fixed with 4% paraformaldehyde and analysed by confocal microscopy. There were two control groups of oocytes: one group was fixed prior to any antibody incubation and the other before the secondary antibody. The distribution of the integrin α6β1 was identical in both groups. The number of patches were quantified using the `particle analysis' menu of Image J.
We thank P. Fontanges (Hôpital Tenon, Paris, France), S. Chambris (Université Paris 13, Bobigny, France) for their technical assistance, L. Ashman, F. Lanza and S. Levy for the gift of monoclonal antibodies. We are grateful to A. Hazout, P. Cohen-Bacrie and A.-M. Junca from Eylau laboratories for providing us with human material. This work was supported by grants from Inserm, the Association pour la Recherche sur le Cancer, NRB-Vaincre le Cancer, and the Institut de Cancérologie et Immunogénétique.