Molecular dissection of IZUMO1, a sperm protein essential for sperm-egg fusion

The membrane fusion of spermatozoon and oocyte is the central event of fertilization, where the genetic materials of two gametes merge. Despite the biological importance of this event, targeted deletion studies have revealed only two proteins, CD9 and IZUMO1, to be necessary in sperm-egg fusion (Primakoff and Myles, 2007). Although CD9 is a ubiquitously expressed protein, gene disruption resulted in infertile females but otherwise healthy mice. Subsequent experiments revealed that CD9 functions as an essential factor for sperm-egg fusion in eggs (Kaji et al., 2000; Le Naour et al., 2000; Miyado et al., 2000). We have previously identified a sperm-specific protein, IZUMO1, which is indispensable for mouse sperm-egg fusion (Inoue et al., 2005). In mice lacking either Cd9 on eggs or Izumo1 on spermatozoa, membrane fusion was almost completely impaired. To date, how these two proteins mediate membrane fusion is unclear and therefore the molecular mechanism behind of sperm-egg fusion remains virtually unknown.

IZUMO1 is a 377 amino acid type I transmembrane protein with a large extracellular region, a single transmembrane region and a short cytoplasmic tail. In this study, we focused on the 319 amino acid extracellular region of murine IZUMO1. Protein domain search programs (Zdobnov and Apweiler, 2001; McGuffin and Jones, 2003; Letunic et al., 2006) have identified only one immunoglobulin-like domain in the extracellular region. In addition, Ellerman et al. have found that three novel proteins (IZUMO2, IZUMO3 and IZUMO4) share a local sequence similarity with IZUMO1 (Ellerman et al., 2009). These proteins were characterized by the presence of a new sequence motif, called the Izumo domain, with a conserved cluster of eight cysteines: C-X(2)-C-X(106,108)-C-X(3,4)-C-X(9)-C-X(2)-C-X(6)-C-X(4,5)-C. However, it is unclear how these motifs are related to functions of IZUMO1 in sperm-egg fusion.

Here, we have first tried to clarify the functional site in IZUMO1 for sperm-egg fusion by using fusion-inhibiting monoclonal antibodies and a series of recombinant IZUMO1 fragments (Fig. 1). A fluoresceinated fragment that includes the potential functional site was incubated with zona-free eggs and was used to probe its affinity to eggs. In addition, recombinant IZUMO1 fragments were added to the in vitro fertilization system and were examined for their fusion inhibitory activities. The biophysical properties of the putative functional site of IZUMO1 were subsequently analyzed using circular dichroism (CD), sedimentation equilibrium and small angle X-ray scattering (SAXS). Combining the above functional and structural data, we discussed the putative molecular function of IZUMO1 in gamete fusion.

All animal experiments were approved by the Animal Care and Use Committee of the Research Institute for Microbial Diseases, Osaka University, Japan.

Female rats were immunized with mouse sperm, and spleen cells were subjected to cell fusion with P3U1 cells. Indirect immunofluorescence staining as the first screening was employed (Okabe et al., 1987). Hybridomas expressing anti-IZUMO1 antibodies were then further selected by western blotting using a hybridoma culture as a source of primary antibodies. Each hybridoma culture was compared in terms of reactivity against proteins with a molecular weight of 56 kDa in the sperm extract.

DNA coding for the extracellular region of mouse IZUMO1 was constructed using synthetic DNA with codon use that was optimized for expression in Escherichia coli. IZUMO1 fragments were amplified by polymerase chain reaction (PCR) with a 3′-primer with and without sequence encoding Ala-Gly-Gly-His-His-His-His-His-His, a linker plus a hexahistidine tag. In addition, IZUMO1 fragments with the sequence encoding Gly-Gly-Gly-Gly-Ser-Cys were prepared for fluorescent labeling by thiol-maleimide coupling. IZUMO1_234-298 was designed to start from Pro233 with a His-tag and linker; however, the purified fragment lacked the first proline, as judged by matrix-assisted laser desorption ionization time-of-flight (MALDI-TOF) mass spectrometry. Amplified PCR products were cloned back into pAED4 (Doering and Matsudaira, 1996). In IZUMO1_PFF-Pro, Leu70, Leu81, Leu94 and Leu104 of IZUMO1_PFF were mutated to prolines using PCR-based mutagenesis.

All constructs were expressed in E. coli strain BL21 (DE3) pLysS (Agilent Technologies, La Jolla, CA). IZUMO1_PFF and IZUMO1_{Ig domain} were strongly expressed and accumulated in inclusion bodies. Inclusion bodies from 1.6 l of culture medium were suspended in 3 ml of 10 mM Tris-HCl (pH 8.5) and were solubilized by the addition of 3 g of solid guanidine hydrochloride (GdnHCl). These samples were then purified over a Superdex 75 (GE Healthcare, Waukesha, WI) that was pre-equilibrated with 6 M urea in 10 mM Tris-HCl (pH 8.5). IZUMO1_234-298, IZUMO1_26-113 and IZUMO1_5-56 were recovered in the soluble fraction and were purified using a HiTrap metal-chelating column (GE Healthcare). Finally, all samples were purified using reverse-phase chromatography, except IZUMO1_234-298 and IZUMO1_PFF variants with two or more cysteines. IZUMO1_234-298 was further purified by ion exchange chromatography. IZUMO1_PFF-core was prepared from partially purified IZUMO1_PFF by limited proteolysis using 1/500 (w/w) proteinase K (Merck KGaA, Darmstadt, Germany) at 10°C for 1.5-2 hours. Proteolyzed fragments were then applied to reverse-phase chromatography for final purification of IZUMO1_PFF-core.

MALDI-TOF mass spectrometry confirmed that the molecular weights of purified IZUMO1 fragments were identical to the expected values that were calculated from their amino acid sequences (with an error of ±0.025%). Protein stock concentrations were determined by measuring the absorbance at 280 nm in 6 M GdnHCl, 20 mM sodium phosphate (pH 6.5). Molar extinction coefficients were calculated based on the number of Trp and Tyr residues using the Edelhoch spectral parameters (Edelhoch, 1967) (supplementary material Table S1). In this article, the molar concentrations of IZUMO1 fragments were estimated by assuming that they are monomers in a solution.

The antibodies were immobilized to the CM-5 sensor chip by the amine coupling protocol that was supplied by the manufacturer at 25°C using Biacore 2000 (GE Healthcare). The surface plasmon resonance (SPR) spectra of the IZUMO1 fragments were measured in 10 mM HEPES, 150 mM NaCl, 3 mM EDTA and 0.005% Tween-20 at 20°C, and the bound fragments were washed with 10 mM glycine and 0.5 M NaCl at pH 2.0. All experiments were duplicated.

Mouse spermatozoa were collected from the cauda epididymis and were capacitated in vitro for 2 hours in a 200 μl drop of TYH medium that was covered with paraffin oil. B6D2F1 female mice (>8 weeks old) were superovulated with an injection of 5 IU of human chorionic gonadotropin (hCG) 48 hours after a 5 IU injection of equine chorionic gonadotropin (eCG). The eggs were collected from the oviduct 14 hours after the hCG injection. Eggs were placed in a 200 μl drop of TYH medium. These eggs were incubated with 2×10⁵ spermatozoa/ml for 2 hours at 37°C in 5% CO₂ with anti-IZUMO1 monoclonal antibodies, and unbound spermatozoa were washed away. Eggs were observed 24 hours after insemination for the two-cell development under a Hoffman modulation contrast microscope.

B6D2F1 female mice (>8 weeks old) were superovulated with an injection of 5 IU of hCG 48 hours after a 5 IU injection of eCG. The eggs were collected from the oviduct 14 hours after the hCG injection. Eggs were placed in a 200 μl drop of TYH medium. After being freed from cumulus cells with 0.01% (w/v) hyaluronidase, the zona pellucida was removed from mouse eggs using a piezo-manipulator, as previously reported (Yamagata et al., 2002). Zona-free mouse eggs were preloaded with 1 μg/ml Hoechst 33342 (Life Technologies, CA, USA) in TYH medium for 10 minutes and were washed prior to the addition of the spermatozoon. Zona-free eggs were co-incubated with 2×10⁵ mouse spermatozoa/ml for 30 minutes at 37°C in 5% CO₂ with and without antibodies or IZUMO1 fragments. After 30 minutes of incubation, the eggs were observed under a fluorescence microscope (UV excitation light) after fixing with 0.25% glutaraldehyde. This procedure enabled staining of only fused sperm nuclei by transferring the dye into spermatozoa after membrane fusion.

Mouse zona-free eggs (wild type and Cd9^-/-) were prepared as above. Eggs were incubated with 3 μM IZUMO1 fragments that were fluoresceinated with Alexa Fluor 546 (Life Technologies) and 1 μg/ml of Hoechst 33342 for 2 hours at 37°C in TYH medium. After the eggs had been washed twice in TYH medium for 5 minutes each, the eggs were observed. For these experiments, especially careful preparation of zona-free eggs was required to avoid the occasional aberrant strong staining of eggs by fluoresceinated fragments. We used an FV-1000 microscope (Olympus, Tokyo, Japan) to obtain the confocal images.

CD spectroscopy was performed using a J-820 spectropolarimeter (Jasco, Tokyo, Japan) with a 1 mm cell at 20°C in 10 mM Tris-HCl (pH 8.5) buffer. CD spectra were measured at protein concentrations of 10 μM, except for IZUMO1_{Ig domain}, which was measured at 20 μM. Sedimentation equilibrium analysis was carried out with an XLA-90 analytical centrifuge (Beckman Coulter, Fullerton, CA) in 10 mM Tris-HCl (pH 8.5) and 50 mM NaCl. The isoelectric points of IZUMO1_PFF and IZUMO1_PFF-core are 5.4 and 7.8, respectively. To increase the solubility and avoid unintended aggregation, we carried out all biophysical experiments, including SAXS, at pH 8.5, which is slightly more alkaline than the reported pH of uterine fluid (pH 7.9±0.4) (Iritani et al., 1971). The helix formation of IZUMO1_PFF and IZUMO1_PFF-core at neutral pH was confirmed by CD measurements in 10 mM potassium phosphate (pH 7.1) and 150 mM NaCl (supplementary material Fig. S1).

IZUMO1 fragments (100 μM) were incubated with 25 times molar excess of bis(sulfosuccinimidyl) suberate (BS³) (Thermo Scientific, Waltham, MA) in 0.1 M HEPES buffer (pH 7.5) at room temperature for 30 minutes, and then the reaction was quenched by 25 mM Tris-HCl (pH 8.5). The samples (1 μg) were subjected to sodium dodecyl sulfate PAGE (SDS-PAGE) under reduced conditions with untreated samples as controls. Mouse spermatozoa from the cauda epididymis were solubilized with phosphate-buffered saline (PBS) with 1% n-dodecyl-β-D-maltoside (DDM) and 1% protease inhibitor cocktail (Nacalai Tesque, Kyoto, Japan). Solubilized proteins were centrifuged at 15,000 g for 30 minutes at 4°C, and the supernatants were treated with 0.1, 0.5 and 1 mM BS³ for 2 hours on ice. After incubation, the reaction was stopped with 50 mM Tris-HCl (pH 7.4) for 15 minutes at room temperature. The samples were subjected to non-reducing SDS-PAGE followed by western blotting. IZUMO1 protein was detected with IZUMO1 monoclonal antibody.

SAXS measurements were carried out at BL-10C, Photon Factory, Tsukuba, Japan (Ueki et al., 1985). The wavelength of X-rays was 1.488 Å, as selected by a Si monochromator. The scattering profiles were collected using an R-Axis VII (Rigaku, Tokyo, Japan) and the circularly averaged intensity was used for further analysis. The sample cell had a path length of 1 mm. SAXS data were measured at 15°C in 50 mM Tris-HCl (pH 8.5). SAXS data of IZUMO1_PFF-core were collected at the protein concentrations of 12.9, 10.4, 7.9, 5.1 and 2.5 mg/ml. As a control, scattering profiles from lysozyme that was dissolved in the same buffer at the protein concentrations of 15.7, 12.6, 9.5, 6.2 and 3.1 mg/ml were also measured.

The raw data of SAXS measurements were initially analyzed by Igor Pro (WaveMetrics, Lake Oswego, OR) to calculate the scattering intensities from protein by subtracting the intensity curve that was obtained for the buffer solution and to correct for the protein concentration dependency. The scattering curves after these corrections were analyzed to compute p(r) functions using the program GNOM (Svergun, 1992). The analysis was performed by changing the approximate d_max value, and the value with which the p(r) function fell to a natural minimum was determined as the real d_max. Scattering data in the Q range of 0.02 to 0.24 Å^-1 were used in this calculation.

The cloned mouse Izumo1 cDNA was ligated into the mammalian expression vector pCXN-2 and Cos-7 cells were transiently transfected with this vector using lipofectamine LTX (Life Technologies). After 2 days, transfected Cos-7 cells were collected with 10 mM EDTA-PBS, washed three times with PBS, and suspended in TYH medium. Mouse eggs expressing green fluorescent protein (GFP) in the cytosol were obtained from transgenic mice with an enhanced GFP cDNA under the control of a chicken β-actin promoter and cytomegalovirus enhancer (Okabe et al., 1997). Wild-type and GFP-expressing zona-free oocytes were prepared as above and incubated with transfected Cos-7 cells at 37°C in TYH medium. IZUMO1 was stained by Alexa Fluor 546-fluoresceinated anti-IZUMO1 monoclonal antibodies.

Our previous report indicated that glycosylation was not essential for IZUMO1 function (Inoue et al., 2008). Therefore, we speculated that there must be primary amino acid sequences in IZUMO1 that are important for fusion. We tried to identify the functional site using three new anti-IZUMO1 monoclonal antibodies (Mabs), Mab34, Mab120 and Mab125, which were prepared by immunizing female rats with mouse spermatozoa. In western blot analysis, all of the antibodies detected 56 kDa proteins in wild-type spermatozoa but not in Izumo1-null spermatozoa (Fig. 2A), indicating that all of the Mabs are specific for the detection of IZUMO1 in western blotting. We added these Mabs to an in vitro fertilization system and found that only Mab34 and Mab120 significantly inhibited fertilization (Fig. 2B). When the spermatozoa were incubated with zona-free eggs in the presence of these antibodies, the sperm-egg binding was not affected by either of these inhibitory antibodies, but the fusion step was significantly inhibited at a concentration of 10 μg/ml (Mab34 and Mab120) by the sperm-egg fusion assay (Fig. 2C). We measured the concentration dependence of the inhibitory activity of these antibodies, and the inhibition of in vitro fertilization was observed even at 0.2 μg/ml of Mab34 and Mab120 (supplementary material Fig. S2).

To search for the functional site of IZUMO1, we separated the sequence of the extracellular region of IZUMO1 into three sections, taking into account the immunoglobulin-like domain. These regions correspond to Asp5-Leu113, Glu146-Leu232 and Pro234-Arg298 (Fig. 1). Of these fragments of IZUMO1, the fusion inhibitory Mabs, Mab34 and Mab120, bound only to the fragment from Asp5 to Leu113, as judged by SPR spectroscopy (Fig. 3A,B). This result indicated that the epitopes of these antibodies are located at the N-terminal region of IZUMO1 (from Asp5 to Leu113) and constitute a putative functional site. We therefore called this fragment ‘IZUMO1 putative functional fragment’ (IZUMO1_PFF) (Fig. 1). The dissociation constants of Mab34 and Mab120 were calculated as 1.4 nM (supplementary material Fig. S3A) and 36 nM (supplementary material Fig. S3B), respectively, based on the monomer concentration of IZUMO1_PFF. Noninhibitory Mab125 recognized IZUMO1_234-298, which is a proline-rich juxtamembrane region corresponding to the area between the immunoglobulin-like loop and the transmembrane domain (Fig. 3C). OBF13, which is the original anti-IZUMO1 Mab of the IgM class that was reported to inhibit sperm-egg fusion (Okabe et al., 1987; Inoue et al., 2005), also recognized the IZUMO1_PFF fragment, albeit weakly (Fig. 3D).

To further scrutinize the antigenic and biophysical properties of IZUMO1_PFF, we pursued the identification of the structural core of this fragment. Partially purified IZUMO1_PFF was incubated with proteinase K at 10°C for 1.5-2 hours, and a major protease-resistant fragment corresponding to residues Val57-Leu113 was recovered by reverse-phase chromatography, suggesting that this fragment folds into a stable structural core. Therefore, we named this fragment IZUMO1_PFF-core (Fig. 1). In addition to this fragment, an N-terminal region of IZUMO1_PFF [named IZUMO1_5-56 (Asp5-Ala56)] and a short version of IZUMO1_PFF lacking the N-terminal 21 amino acid residues [IZUMO1_26-113 (His26-Leu113)] were prepared (Fig. 1). None of these short fragments interacted with Mab34 and Mab120 (supplementary material Fig. S4), suggesting that the antibody recognition depends on the tertiary structure of the antigen.

Because IZUMO1_PFF was a candidate for the functional site of sperm-egg fusion, we asked whether it bound to eggs. We produced Alexa Fluor 546-conjugated IZUMO1_PFF, IZUMO1_PFF-core and IZUMO1_5-56 via an additional C-terminal cysteine by thiol-maleimide coupling. These fluoresceinated fragments were incubated with zona-free unfertilized eggs (Fig. 4A-C). The fragments bound to eggs, and the eggs became fluorescent. The binding of IZUMO1_PFF and IZUMO1_PFF-core to eggs was significantly stronger than that of IZUMO1_5-56. To investigate precisely the localization of IZUMO1_PFF bound on eggs, we acquired differential interference contrast and confocal images of eggs that were incubated with fluoresceinated IZUMO1_PFF (Fig. 4D,E). IZUMO1_PFF was observed at microvilli-rich regions but not at microvilli-negative regions on the surface of metaphase II chromosomes. The nonuniform distribution of bound IZUMO1_PFF suggested that the association between IZUMO1_PFF and the egg surface that was observed here was caused by a specific interaction.

Next, we measured sperm-egg fusion in the presence of IZUMO1_PFF to ask whether IZUMO1_PFF directly affected fertilization (Fig. 5A-C). No significant inhibition was demonstrated by the addition of IZUMO1_5-56, IZUMO1_PFF-core and IZUMO1_{Ig domain}. However, the addition of more than 8 μM IZUMO1_PFF significantly inhibited the sperm-egg fusion, and the fusion index (number of spermatozoa that fused to each egg) significantly decreased (Fig. 5C; supplementary material Fig. S5) without affecting sperm motility and egg-binding ability, indicating that the inhibitory effect on fusion occurs during the sperm-egg fusion step and is not based on the sperm-egg binding. These results show that the IZUMO1_PFF region is sufficient for both the egg-binding ability and sperm-egg fusion inhibitory activity, and they suggest that this fragment includes the functional site of IZUMO1. In addition, the inhibitory activity of IZUMO1_PFF might not come from the saturation of an unknown egg surface receptor for IZUMO1 because IZUMO1_PFF-core bound to eggs but did not prevent the sperm-egg fusion.

IZUMO1_PFF bound to the egg surface (Fig. 4) and inhibited sperm-egg fusion in vitro (Fig. 5). Therefore, it is plausible that binding partners of IZUMO1 such as CD9 exist on the egg surface. To our surprise, although CD9 is an essential factor for eggs to fuse with spermatozoa, IZUMO1_PFF could bind to Cd9-null eggs (Fig. 6). Fluoresceinated IZUMO1_PFF staining exhibited larger and denser spots than CD9 staining (Fig. 4E) (Inoue et al., 2012). These indicate that there is no physical binding between IZUMO1_PFF and CD9. The lack of interaction between IZUMO1_PFF and CD9 led us to hypothesize the existence of unknown factors that are involved in gamete fusion. Although CD9-related microvillus formation is considered crucial for fusion, our results suggest that the binding of IZUMO1 to the egg surface may not require a CD9-related strong interaction and/or microvillous formation.

A variety of biophysical analyses of protein solution gives us important information about the structure and size of the target molecule. The CD spectroscopy, which monitors the differences between the absorptions of left-handed and right-handed circularly polarized light, is often used to estimate the contents of secondary structures, as well as to examine whether the target protein is denatured, in combination with equilibrium denaturation experiments. A single measurement of a CD spectrum typically can be carried out with less than 25 μg of protein, which is less protein than required by other biophysical methods. The sedimentation equilibrium is one of the most reliable methods with which to determine the molecular weight and, thus, the oligomeric state of proteins in solution. The method is almost free from nonspecific interactions with carriers and effects of charged residues of proteins that often cause ambiguous interpretations of data from other methods such as gel filtration or electrophoresis without SDS or denaturants. By SAXS measurement, we can obtain information about the distance distribution between each pair of X-ray diffractive atoms, such as carbon and nitrogen, in a protein. Using the SAXS data, the size and low-resolution shape of the protein in solution can be calculated. We believe that the molecular information from such biophysical analysis as mentioned above supplements the results from physiological experiments, which will be helpful in understanding the precise mechanism of sperm-egg fusion.

CD and sedimentation equilibrium analysis revealed that IZUMO1_PFF forms a helix dimer (Fig. 7). The limited proteolysis of IZUMO1_PFF suggested that IZUMO1_PFF-core is a structural core of the IZUMO1_PFF protein. Indeed, IZUMO1_PFF-core exhibited greater negative ellipticity at 222 nm (Fig. 7A). However, the structure of IZUMO1_5-56, which is the residual N-terminal region of IZUMO1_PFF after IZUMO1_PFF-core is removed, was unfolded, as judged by CD spectroscopy. If the N-terminal region corresponding to IZUMO1_5-56 is tightly folded in the context of the whole IZUMO1_PFF, then IZUMO1_PFF should either be more stable than IZUMO1_PFF-core or show a biphasic unfolding transition curve. To examine the stability of both IZUMO1_PFF and IZUMO1_PFF-core, we tracked ellipticity over a range of GdnHCl-induced unfolding conditions. The midpoint concentrations of unfolding and slopes of transition for both IZUMO1 fragments were identical (Fig. 7B). Combining these data, IZUMO1_PFF had a helical structure in its C-terminal region and a mostly unstructured N-terminal region. To assess whether these IZUMO1 fragments formed oligomers, we measured the sedimentation equilibrium profiles at pH 8.5. IZUMO1_PFF (Fig. 7C) and IZUMO1_PFF-core (supplementary material Fig. S6A) each exhibited an apparent molecular weight that was 1.9 times its calculated molecular weight, indicating the formation of a dimer in solution, which contrasted with the monomeric characteristic of the IZUMO1_{Ig domain} (supplementary material Fig. S6B). The dimerization of IZUMO1_PFF and IZUMO1_PFF-core was confirmed by crosslink experiments (supplementary material Fig. S6C). After the reaction of the fragments with the homobifunctional amine-to-amine crosslinker BS³, we observed decreased monomer fractions and increased dimer fractions by SDS-PAGE of IZUMO1_PFF and IZUMO1_PFF-core. However, crosslinking is not significant in other fragments. The increased dimeric fraction in the presence of BS³ was also observed in the intact IZUMO1 in sperm extract (supplementary material Fig. S6D). It is interesting that a small amount of dimeric IZUMO1 can be detected by SDS-PAGE in the absence of BS³, suggesting that the dimeric structure of intact IZUMO1 is highly stable and that a certain amount of the dimeric form can survive the heat treatment with SDS.

To investigate the molecular shape of IZUMO1_PFF-core, we measured the SAXS of both IZUMO1_PFF-core and hen egg white lysozyme as a control because lysozyme has a similar molecular weight to dimeric IZUMO1_PFF-core. The p(r) function, which is the probable frequency distribution of all vector lengths between atoms within the scattering particle, can provide information about the shape of the protein molecule. The p(r) function of IZUMO1_PFF-core showed a larger maximum diameter d_max (∼60 Å) than that of lysozyme (∼46 Å) (Fig. 7D). This difference in d_max indicates that the structure of IZUMO1_PFF-core is considerably more elongated than that of a typical globular protein such as lysozyme, suggesting that IZUMO1_PFF-core forms an elongated ellipsoidal shape. The presence of a shoulder in the right side of the p(r) function of IZUMO1_PFF-core also supports this idea.

Proline is known as a strong helix breaker in aqueous environments (Levitt, 1978). To investigate the functional role of the helix that was identified in the C-terminal region of IZUMO1_PFF, we prepared a proline mutant of this fragment (IZUMO1_PFF-Pro; Fig. 1). As expected, IZUMO1_PFF-Pro lost most of its secondary structure, as shown by CD spectrum (Fig. 8A). The interaction of IZUMO1_PFF-Pro with egg was much weaker than that of IZUMO1_PFF, but it was similar to that of IZUMO1_5-56 (Fig. 8B-D). In addition, no significant fusion inhibitory activity was observed in IZUMO1_PFF-Pro (Fig. 8E). These findings suggested that the helix structure in IZUMO_PFF is related to the egg binding and fusion inhibitory effects of IZUMO1_PFF.

To examine whether the IZUMO1 alone can induce the egg binding, which is anticipated from the experiments using IZUMO1_PFF and IZUMO1_PFF-core, we explored the interaction between egg and recombinant Cos-7 cells that were transfected with the expression vector, including mouse intact Izumo1 cDNA. After 3 hours of incubation of IZUMO1-expressing Cos-7 cells with wild-type eggs, most of the cultured cells were bound to eggs (Fig. 9). This phenomenon was not observed in Cos-7 cells without IZUMO1 expression. In addition, IZUMO1-expressing cells did not aggregate, indicating that the binding of IZUMO1-expressing Cos-7 cells with eggs is a specific interaction that is mediated by IZUMO1. The fusion of two cells, which was judged by the movement of stained nuclei, did not occur even after 24 hour of incubation. We also confirmed that the cytosol fraction of eggs did not disperse within Cos-7 cells after 24 hours of incubation using the eggs that were obtained from transgenic mice expressing GFP. Fluorescence staining by antibody revealed that IZUMO1 accumulated at the interface between Cos-7 and egg cells (Fig. 9B-D). In addition, Cos-7 cells that were bound to eggs were deformed along the egg surface.

In this paper, we have determined that a putative functional site exists in the N-terminal region of IZUMO1. The latter half of this region is important for the binding of IZUMO1 to the egg surface because the fragment corresponding to this part, IZUMO1_PFF-core, bound to the egg surface. Although CD9 is an essential factor for sperm-egg fusion in eggs, IZUMO1_PPF maintains the binding ability on the egg surface of Cd9^-/- eggs, indicating that the presence of unknown factors that are related to the gamete fusion. A series of CD, sedimentation equilibrium and SAXS experiments showed that IZUMO1_PFF was composed of an N-terminal unfolded region and a C-terminal helix dimer with an ellipsoidal shape. In addition, the IZUMO1_PFF mutant, IZUMO1_PFF-Pro, which has a defective helical formation, lost egg-binding and fusion inhibitory activities. These results suggest that the formation of a helical dimer at the N-terminal region is essential for the function of IZUMO1.

Recombinant Cos-7 cells expressing whole IZUMO1 did not fuse to eggs. However, we found that these cultured cells bound to eggs and that IZUMO1 accumulated at the interaction site of the two cells. These observations suggest that IZUMO1 alone cannot cause sperm-egg membrane fusion; rather, it induces cellular surface interactions such as membrane tethering. We speculate that the distance between membranes of the spermatozoon and the egg may be not short enough to trigger membrane fusion when IZUMO1_PFF binds to the egg surface. Instead, the helical region of IZUMO1 corresponding to IZUMO1_PFF-core may keep the membranes close together. The separation of IZUMO1_PFF-core from the transmembrane region by an immunoglobulin-like domain does not contradict the above notion. There are two examples of essential proteins for fusion, which are supposed to tether membranes. Mitofusin is essential for mitochondrial fusion and forms trans complexes between mitochondria. Mfn1, mammalian mitofusin, forms an antiparallel helix dimer. Because the arrangement of helices is antiparallel and the linker between transmembrane and helical regions is relatively long (>40 amino acids), two membranes that are connected by dimeric mitofusin are distant from each other. Thus, Mfn1 is thought to tether mitochondrial membranes during fusion (Koshiba et al., 2004; Detmer and Chan, 2007). EEA1 is essential for early endosome fusion, and it is proposed to tether endosomes by binding phosphatidylinositol at its C terminus and Rab5 at its N terminus (Christoforidis et al., 1999). The crystal structure of a C-terminal fragment of EEA1 indicates that a parallel helix bundle is formed next to the phosphatidylinositol-binding FYVE domain. The presence of a strong heptad repeat throughout the sequence suggests that EEA1 forms long, continuous coiled coils and holds two membranes together at a significant distance (Dumas et al., 2001). It should be noted that possible membrane interactions that are induced by IZUMO1 should be different from the binding of spermatozoa around eggs because Izumo1-deficient spermatozoa seem to have normal attachments to eggs, but this binding may not facilitate sperm-egg fusion (Inoue et al., 2005). CD9 is also suggested to be involved in membrane adhesion to trigger sperm-egg fusion (Jégou et al., 2011), although spermatozoa bound to Cd9-null eggs (Miyado et al., 2000).

Chen and Olson referred to IZUMO1 as a typical non-helix protein that was involved in biological membrane fusion because only the immunoglobulin-like domain can be predicted from the primary sequence as a known structural motif (Chen and Olson, 2005). The β-sheet-rich C2 domain is proposed to induce membrane curvature and to be involved directly in the fusion step (Martens and McMahon, 2008). Because of the similarity between the immunoglobulin-like domain and the C2 domain, fusion-related proteins with immunoglobulin-like domains, including IZUMO1, were hypothesized to work in a similar manner to proteins containing C2 domains, such as synaptotagmin. Our finding requires reconsidering such a view about IZUMO1 and also encourages a search for helix motifs in other proteins that are related to cell-cell membrane fusions, such as myoblast formation, multinucleation of macrophages and epithelial fusion in Caenorhabditis elegans (Podbilewicz et al., 2006; Sapir et al., 2007; Helming and Gordon, 2009; Abmayr and Pavlath, 2012).

IZUMO1, IZUMO2, IZUMO3 and IZUMO4 share a novel sequence motif called the Izumo domain, which is characterized by the presence of a conserved cluster of eight cysteines: C-X(2)-C-X(106,108)-C-X(3,4)-C-X(9)-C-X(2)-C-X(6)-C-X(4,5)-C (Ellerman et al., 2009). IZUMO1_PFF corresponds to the region between the second and third cysteines. The sequence similarity between IZUMO1 and IZUMO2, IZUMO3 and IZUMO4 is not obvious at this region, where the CLUSTAL W scores were 12.2-15.7 (Thompson et al., 1994). Ellerman et al. observed the dimeric form, in addition to monomeric and trimeric forms, of the Izumo domain of IZUMO1 by SDS-PAGE with mildly denaturing conditions. This is consistent with the data that are presented here. However, they could observe neither the binding of recombinant IZUMO1 fragments, which included the Izumo domain, to the egg nor the inhibitory activity of IZUMO1 fragments on sperm-egg fusion in vitro. The discrepancy between our data and those of Ellerman et al. could be explained by the presence of cysteine clusters close to the IZUMO1_PFF sequences. These cysteines may affect the structure and negatively regulate the function of IZUMO1. Another possibility is that the cysteines prohibit the correct folding of their recombinant fragments by the aberrant formation of disulfide bonds. Recombinant fragments with those cysteines that were expressed in E. coli had intermolecular disulfide bonds under oxidized conditions and thus formed oligomers and aggregates (supplementary material Fig. S7). Although the functional and structural roles of the cysteine cluster are unclear, the IZUMO1_PFF fragment was mono-dispersed in solution, had a significant amount of secondary structure, was unfolded cooperatively by a denaturant, exhibited egg-binding activity and inhibited sperm-egg fusion. Furthermore, inhibitory antibodies, which were obtained by the immunization of mouse spermatozoa, recognized the tertiary structure of IZUMO1_PFF, indicating that our recombinant IZUMO1_PFF shares the structure of native IZUMO1. These results strongly suggested the correct and functional folding of our fragment. To uncover the physiological functions of the novel IZUMOs, it will be interesting to study the egg-binding and fusion inhibition activities and structural analysis of fragments of IZUMO2, IZUMO3 and IZUMO4 that correspond to IZUMO1_PFF.

IZUMO1_PFF-core formed an ellipsoidal dimer, and its helical structure is required for the fusion inhibition and egg-binding activities of this fragment. These features are reminiscent of the known fusion-related proteins such as class 1 viral envelope proteins, SNAREs, mitofusin and EEA1, which consist of a bundle of coiled-coil structures. However, we could not identify an unequivocal coiled-coil region by the prediction programs MultiCoil, LearnCoil and Paircoil2 (Wolf et al., 1997; Singh et al., 1999; McDonnell et al., 2006). The mean residue ellipticity at 222 nm of IZUMO1_PFF-core is larger than the ellipticity that is expected in a typical coiled-coil region (O’Shea et al., 1989), but it is similar to that of the SNARE complex (Fasshauer et al., 1997; Rice et al., 1997). It is fascinating that the machinery in sperm-egg fusion may share the structural motif of a helix bundle with other biological viral and vesicle fusions. A high-resolution structural analysis is required to confirm this hypothesis. We found that IZUMO1 fragments can be crystallized with polyethylene glycol as a precipitant; however, the quality of X-ray diffractions of these crystals was not sufficient to solve the three-dimensional structure. We believe that co-crystallization with an antibody fragment may help improve crystal packing (Koide, 2009).

A function-inhibiting and egg-binding IZUMO1_PFF, which was identified in this report, will be useful in the search for other components of the egg-fusion machinery. Identifying the essential components of the fusion machinery will greatly assist in the elucidation of the sperm-egg fusion mechanism.

We are very grateful to Dr Chris C. Liu at University of Toronto and Dr Adam M. Benham at Durham University for critically analyzing the manuscript and providing valuable advice. We thank Dr Eisuke Mekada and Mr Takao Nishikawa at Osaka University for kindly giving us the Cd9 disrupted mouse line and for technical assistance, respectively. We also thank Professor Amy E. Keating at Massachusetts Institute of Technology and Professor David C. Chan at California Institute of Technology for their helpful comments on the manuscript.