A single domain of the ZP2 zona pellucida protein mediates gamete recognition in mice and humans

Monospermic fertilization is required for successful development in mice and humans. However, the molecular basis of sperm–egg interaction remains incompletely understood despite decades of investigation. Ovulated eggs are surrounded by an extracellular zona pellucida, to which sperm bind and penetrate before gamete fusion. After fertilization, the zona matrix is modified so that sperm do not bind to the early embryo. Humans have four genetic loci encoding ZP1, ZP2, ZP3, and ZP4 (Spargo and Hope, 2003), but mouse Zp4 is a pseudogene (Lefièvre et al., 2004), and the mouse zona pellucida contains only three glycoproteins (Bleil and Wassarman, 1980). Given the simple structure of the zona pellucida, it has been surprisingly difficult to genetically define a zona protein that is essential for fertilization and the postfertilization block to polyspermy.

To identify the zona ligand for sperm recognition, we have exploited two physiological dichotomies. One is that sperm bind to the zona pellucida surrounding eggs before, but not after, fertilization. The only biochemically documented change in the zona matrix is the postfertilization cleavage of ZP2 (Bleil et al., 1981; Bauskin et al., 1999), and the site has been defined as ¹⁶⁶LA↓DE¹⁶⁹ in mice (Gahlay et al., 2010). If mutated in transgenic mice or if the cleaving enzyme, ovastacin, is genetically ablated, ZP2 remains intact and sperm bind de novo to the zona pellucida surrounding early embryos despite fertilization (Gahlay et al., 2010; Burkart et al., 2012). The second dichotomy is that human sperm bind to human, but not mouse, zonae pellucidae (Bedford, 1977).

To capitalize on these differences, we have established transgenic mouse lines lacking ZP2 and expressing truncated or human/mouse chimeric isoforms to define ZP2^51–149 as the zona ligand for human and mouse sperm recognition, and document that in its absence female mice are sterile.

Formation of the extracellular zona matrix is mediated by zona domains (260 aa motifs with conserved cysteine residues) present near the C terminus of each secreted zona protein (Bork and Sander, 1992). The human zona pellucida is composed of four (ZP1–4) and the mouse of three (ZP1–3) glycoproteins (Bleil and Wassarman, 1980; Bauskin et al., 1999). Mouse ZP1, the least abundant protein, is not required for formation of the zona pellucida or fertility (Rankin et al., 1999). In the absence of ZP2, a thin zona matrix is formed around growing oocytes that does not persist in ovulated eggs (Rankin et al., 2001), and no zona matrix is formed in the absence of ZP3 (Liu et al., 1996; Rankin et al., 1996). Based on the phylogeny (Larkin et al., 2007) of human and mouse zona proteins, ZP1/ZP2/ZP4 fall into one clade and ZP3 into another (Fig. 1 A). The zona domains of ZP2 and ZP3 have 10 and 8 conserved cysteine residues, respectively, the linkage of which differs in the two clades (Fig. 1 B; Boja et al., 2003). We reasoned that if members of the first clade could substitute for one another, the presence of ZP4 along with ZP1 might permit formation of a matrix in the absence of mouse ZP2 and provide a loss-of-function assay for sperm–egg recognition. Because mouse Zp4 is a pseudogene (Lefièvre et al., 2004), human ZP4 was expressed in transgenic mice (Yauger et al., 2011) to establish a mouse line designated moQuad (moZP1, moZP2, moZP3, and huZP4). After the appropriate crosses, mice lacking mouse ZP2 in the presence of ZP4 were established and designated moQuad-Zp2^Null (moZP1, moZP3, and huZP4; Table S1).

In the presence of huZP4, moQuad^(huZP4) (ZP1,2,3,4) and moQuad-Zp2^Null (ZP1,3,4) transgenic lines form zonae pellucidae during oocyte growth that is similar to normal oocytes (Fig. 1 C and Table S1). The zona pellucida persists after eggs are ovulated into the oviduct, and the composition of the moQuad-Zp2^Null zona matrix was confirmed with monoclonal antibodies that documented the absence of mouse ZP2 (Fig. 1 D). Ovulated eggs in cumulus (hyaluron interspersed with follicular cells) from moQuad^(huZP4) and moQuad-Zp2^Null mice were inseminated with mouse sperm using Zp3^EGFP mouse eggs (green zona) and normal mouse two-cell embryos, respectively, as positive and negative controls. Mouse sperm bound to moQuad^(huZP4) (41.4 ± 2.5, n = 30), but not to moQuad-Zp2^Null, eggs (1.6 ± 0.3, n = 43; Fig. 1 E), and the latter, but not the former, mouse line was sterile (Table S2). Thus, mouse ZP2 is required for in vitro sperm binding and in vivo mouse fertility.

The secreted ZP2 ectodomain (35–633 aa) lacks the signal peptide (1–34) that directs ZP2 into the endosomal pathway and the C terminus (634–713) that includes a transmembrane domain. After fertilization, the ectodomain is cleaved near the N terminus (¹⁶⁶LA↓DE¹⁶⁹) by ovastacin, an egg cortical granule metalloendoprotease, after which sperm no longer bind to the zona pellucida (Gahlay et al., 2010; Burkart et al., 2012). To investigate the importance of the N terminus in mouse gamete recognition, DNA recombineering was used to construct a transgene lacking ZP2^51–149 (moZp2^Trunc; Fig. 2 A and Fig. S1 A). After crossing into the Zp2^Null background, female mice reconstituted a zona pellucida that was thinner than normal. Therefore, these lines were crossed with huZP4 transgenic mice to establish moQuad-Zp2^Trunc mice that developed a more robust zona matrix during oocyte growth (Fig. 2 B).

The zona pellucida surrounding ovulated eggs from moQuad and moQuad-Zp2^Trunc mice were analyzed by confocal microscopy using ZP2 domain-specific monoclonal antibodies. The zonae from both genotypes reacted with antibodies to ZP1, ZP2^C-term, ZP3, and huZP4. However, the monoclonal antibody to the N terminus of ZP2 did not react with the zona pellucidae surrounding moQuad-Zp2^Trunc eggs, although it recognized normal ZP2 in the moQuad (Fig. 2 C). On immunoblots, the monoclonal antibody to the ZP2^N-term detected a 120-kD band in the zona pellucida isolated from moQuad, but not moQuad-Zp2^Trunc eggs (Fig. 2 D, left). A monoclonal antibody to the ZP2^C-term detected a 120-kD band in the zona pellucida isolated from moQuad eggs and a 92-kD band in eggs from moQuad-Zp2^Trunc mice (Fig. 2 D, right). These observations are consistent with deletion of 99 amino acids (ZP2^51–149) in the N terminus of ZP2 of the zona pellucida surrounding moQuad-Zp2^Trunc eggs.

Mouse sperm did not bind to either Zp2^Trunc (0.6 ± 0.2, n = 22) or moQuad-Zp2^Trunc (2.3 ± 1.3, n = 30) eggs (Fig. 2 E), and, when mated with normal males, moQuad-Zp2^Trunc females were sterile (Table S2). Based on these observations in transgenic mice, we conclude that the N terminus of mouse ZP2 is necessary for gamete recognition in vitro and in vivo.

To extend these observations to human biology, huQuad^(huZP1–4) mice containing all four human proteins and none of the endogenous mouse proteins (Baibakov et al., 2012) were used to establish huQuad-ZP2^Null (huZP1,3,4) mouse lines (Table S1). Both lines formed a zona surrounding growing oocytes within the ovary (Fig. 3 A), and, using monoclonal antibodies specific to the human proteins, the absence of huZP2 in huQuad-ZP2^Null ovulated eggs was confirmed (Fig. 3 B). After insemination with human semen, sperm bound robustly to huQuad^(huZP1–4) eggs (59.8 ± 5.3, n = 37), but rarely to huQuad-ZP2^Null eggs (1.5 ± 0.3, n = 36; Fig. 3 C). To ascertain if mouse ZP2 exerted a non-taxon-specific effect on the zona structure important for gamete recognition, mouse Zp2 was expressed in huQuad-ZP2^Null eggs. However, human sperm did not bind to the zona pellucida surrounding huQuad-ZP2^Null; Zp2^Mo eggs (0.7 ± 0.2, n = 30; Fig. 3 C).

To determine if these observations pertain in vivo, an artificial insemination assay was established using mouse sperm as a positive control for fertility and for accumulation of sperm in the perivitelline space. After transcervical insemination (Fig. S2 A), control (Cd9^+/−) mice were fertile, albeit with smaller litters (2.0 ± 0.54 vs. 10.4 ± 0.81 for natural mating). Cd9^Null eggs to which sperm will not fuse (Le Naour et al., 2000) accumulated 1–6 mouse sperm in their perivitelline spaces (between the inner aspects of the zona matrix and the plasma membrane) in 9 of 113 eggs (Fig. 3 D). Subsequently, huZP2^Rescue (huZP2 replaces moZP2) and huZP3^Rescue (huZP3 replaces moZP3) mice (Table S1) were inseminated with human sperm (3 × 10⁷) and eggs recovered from the oviduct were examined by confocal microscopy to detect sperm in the perivitelline space where they accumulated, unable to fuse with mouse eggs. Of 29 eggs recovered from four huZP2^Rescue mice, human sperm were detected in the perivitelline space of four eggs (one egg per female), which is consistent with the small litters observed in control (Cd9^+/−) female mice. No sperm were observed in the perivitelline space of 28 huZP3^Rescue eggs (Fig. 3 E). To facilitate access to ovulated eggs, human sperm also were transferred directly to the oviduct (Fig. 3 F). After 2 h of incubation in vivo, 1–5 human sperm were observed in the perivitelline space of 22–77 eggs (29%) from huZP2^Rescue mice (Fig. 1 G). No sperm were observed in the perivitelline space of 85 control huZP3^Rescue eggs, which is consistent with their inability to support human sperm binding in vitro (Baibakov et al., 2012). Thus, human sperm bind and penetrate the zona pellucida of huZP2^Rescue, but not huZP3^Rescue, eggs in vitro and in vivo.

Mouse sperm, lacking taxon-specific gamete recognition (Bedford, 1977), also bound to huQuad^(huZP1–4) expressing human ZP2 (35.3 ± 1.9, n = 52) and the huQuad-ZP2^Null; Zp2^Mo (36.3 ± 2.2, n = 36) expressing mouse ZP2, but not the huQuad-ZP2^Null eggs missing both mouse and human ZP2 (2.4 ± 0.3, n = 44; Fig. S2 B). After mating, huQuad^(huZP1–4) and huQuad-ZP2^Null; Zp2^Mo female mice containing human or mouse ZP2 in their zonae pellucidae were fertile, but huQuad-ZP2^Null female mice lacking mouse and human ZP2 were sterile (Table S2). Collectively, these results indicate that ZP2 is necessary for human and mouse sperm binding and penetration through the zona pellucida, and for mouse fertility.

Earlier models of gamete recognition had focused on the role of specific glycans as ligands for sperm binding (Florman and Wassarman, 1985; Bleil and Wassarman, 1988; Miller et al., 1992; Chen et al., 1998). However, continued fertility after mutation of attachment sites (Liu et al., 1995; Gahlay et al., 2010) and genetic ablation of specific galactosyltransferases (Thall et al., 1995; Lowe and Marth, 2003; Shi et al., 2004; Williams et al., 2007) have not supported the candidacy of any of the proposed glycans. More recently, the sialyl-Lewis^X antigen has been reported to mediate human sperm binding to human zonae pellucidae (Pang et al., 2011). Although the sialyl-Lewis^X antigen was detected in the zona pellucida surrounding control human oocytes, it was not present in the zona matrix formed by normal, huZP2^Rescue, or huQuad^(huZP1–4) eggs (Fig. 3 H) and cannot account for the observed binding of human sperm under these experimental conditions.

The ectodomains of human (602 aa) and mouse (599 aa) ZP2 share 62% amino acid identity, but the N termini of the two ZP2 proteins are only 48% identical. Thus, we questioned whether the taxon specificity is mediated by the gamete recognition domain found to be essential for normal mouse sperm binding and fertility in Zp2^Trunc mice. Using DNA recombineering, genomic regions encoding the N termini of moZp2 and huZP2 were replaced with the corresponding human exons encoding huZP2^22–164 and mouse exons encoding moZP2^18–156, respectively (Fig. S1 A). After establishment by pronuclear injection, each transgenic line was crossed into the mouse Zp2^Null line to eliminate endogenous ZP2 protein and designated hu/moZp2^Rescue and mo/huZP2^Rescue mice, respectively (Fig. 4 A). The mice appeared normal and a robust zona pellucida was observed in ovarian sections from hu/moZP2^Rescue and mo/huZP2^Rescue female mice (Fig. 4 B).

Mouse ZP1 and ZP3 were present in the zona pellucida surrounding ovulated eggs from each of the three transgenic mouse lines (Fig. 4 C). The zona pellucida from huZP2 rescue mice (Table S1) reacted with monoclonal antibodies to huZP2^N-Term and not with either mouse-specific monoclonal antibody. The zona pellucida of the chimeric hu/moZP2 eggs reacted with huZP2^N-term and moZP2^C-term, but not moZP2^N-term monoclonal antibodies, and the zona pellucida surrounding the chimeric mo/huZP2 rescue eggs reacted with antibodies to the N terminus, but not the C terminus, of mouse ZP2 (Fig. 4 C). Sperm binding was assessed in vitro using noninseminated human oocytes as a positive control and normal mouse eggs with a green zona (Zp3^EGFP) as a negative control. Human sperm bound to the zona pellucida surrounding huZP2 (30.7 ± 1.3, n = 52) and hu/moZp2 (22.5 ± 1.7, n = 69), but not to the zona surrounding mo/huZP2 rescue eggs (2.9 ± 0.5, n = 49; Fig. 4 D). Based on these observations, we conclude that a domain near the N terminus of ZP2 mediates the taxon specificity of human sperm binding to the zona pellucida.

Using ZP2^39–154 as a backbone, a recombinant baculovirus encoding N-terminal chimeric human and mouse ZP2 peptides was constructed in which mouse replaced human sequence between cysteine residues to systematically express moZP2^52–83, moZP2^85–101, and moZP2^103–133 in place of the corresponding human sequence (Fig. 4 E and Fig. S2 C). Peptide beads were incubated with human sperm and washed to remove loosely adherent sperm, and binding was quantified from z projections of confocal images. HuZP2^39–154 (15.0 ± 0.9, n = 43) and moZP2^35–149 (1.2 ± 0.2, n = 43) peptide beads were used as positive and negative controls, respectively. Replacement of human sequence with moZP2^52–83 (3.2 ± 0.2, n = 16) dramatically decreased human sperm binding, whereas chimeric peptides with moZP2^85–101 (10.5 ± 1.0, n = 14) or moZP2^103–133 (11.7 ± 0.8, n = 14) had a more minimal effect (Fig. 4 F). These observations were quantified with box plots (Fig. 4 G), which indicate that the region between the first two cysteine residues (huZP2^55–88; moZP2^52–83) near the N terminus of ZP2 plays a significant role in human gamete recognition.

ZP2 was first proposed as a primary sperm-binding ligand in Xenopus laevis (Tian et al., 1999) and more recently in humans (Baibakov et al., 2012). However, like all other candidates, it had not been defined as essential by genetic ablation. We now report that transgenic mice expressing human ZP4 form a zona pellucida in the absence of ZP2. Using this model system, we have identified a domain within ZP2 that accounts for the taxon specificity of human sperm binding to the zona pellucida and, by genetic ablation, document that this domain is essential for female mouse fertility. Collectively, these results are consistent with a model (Fig. 5) in which mouse and human sperm bind to ZP2^51–149 and hyperactive sperm penetrate the zona matrix to fuse and fertilize ovulated eggs. This triggers egg cortical granule exocytosis in which ovastacin, a metalloendoprotease (Quesada et al., 2004) encoded by Astl, is released and cleaves upstream of a di-acidic motif (¹⁶⁶LA↓DE¹⁶⁹) in ZP2 (Burkart et al., 2012). Although a single cleavage site is reported in human ZP2 (Bauskin et al., 1999), multiple degradation products are observed in mouse ZP2 (Burkart et al., 2012). In each case, the postfertilization cleavage of ZP2 provides a definitive block to polyspermy. Sperm that do not bind to the zona pellucida cannot penetrate the zona matrix or fuse with the egg plasma membrane. The destruction of the N terminus of ZP2 provides an effective block to polyspermy and ensures the monospermic fertilization required for the successful onset of development.

Targeted mutagenesis with a PGK-neomycin cassette was used to genetically ablate Zp1 (replaced exon 1–3), Zp2 (replaced exon 1–2), Zp3 (disrupted exon 1), and Cd9 (replaced promoter and exon 1) in embryonic stem (ES) cells. Correctly targeted ES cells were identified for injection into blastocysts to establish mouse lines with null alleles (Rankin et al., 1996, 1999, 2001; Le Naour et al., 2000). Cd9 null mice were a gift of C. Boucheix (Institut National de la Sante et de la Recherche Medicale, Villejuif, France). Mouse lines expressing human zona proteins were established after pronuclear injection of genomic DNA encoding human ZP1 (11.9 kb including 2.2 promoter, 8.1 coding region, and 1.5 of 3′ flank), ZP2 (16.8 kb including 2.3 promoter, 14 kb coding region, and 0.5 kb of 3′ flank), ZP3 (16 kb including 6.0 kb promoter, exons 1–4 followed by cDNA encoding exons 5–8 and a BGH polyadenylation signal), and ZP4 (11.6 kb including 2.4 promoter, 8.2 coding region, and 1.0 of 3′ flank; Rankin et al., 1998, 2003; Yauger et al., 2011; Baibakov et al., 2012). In addition, three new mouse lines were established using chimeric hu/moZp2, chimeric mo/huZP2, and moZp2^Trunc transgenes (Fig. S2 A) constructed by GalK DNA recombineering (Warming et al., 2005).

To establish the new lines, bacterial artificial chromosome (BAC) DNA (Life Technologies) that include either mouse Zp2 (RP23-65I3) or human ZP2 (RP11-1023A8) were transformed into SW102 bacterial cells containing the λ prophage recombineering system (Liu et al., 2003). For hu/moZp2, human genomic DNA encoding huZP2^41–168 (bp 26 in exon 2 to bp 21 in exon 6) replaced mouse genomic DNA encoding moZP2^37–165 (bp 59 in exon 2 to bp 6 in exon 5). To construct this transgene, a PCR fragment (1,331 bp) containing the galK operon flanked by 50 bp homologous to moZp2 gene 5′ and 3′ of the sequence encoding moZP2^22–161 protein was amplified (huZP2-GalK primers; Table S3) using NEB Phusion (New England Biolabs, Inc.). After digestion with DpnI and overnight gel purification (0.7% agarose, 15 V, 16 h), the PCR fragment was electroporated into the BAC containing SW102 cells, and recombinants were selected by growth on minimal media with galactose. Using a clone from this first step, the galK cassette was replaced by recombineering with a second PCR fragment (5,783 bp) encoding huZP2^41–168 protein with 100-bp arms homologous to moZP2 on either side (moZp2-huZP2 primers; Table S3). Mutant clones were selected on minimal media with 2-deoxy-galactose and confirmed by DNA sequencing of PCR products using gene specific primers (Table S3) to isolate 5′ and 3′ junction fragments.

In a similar recombineering strategy, DNA encoding huZP2^22–161 protein was replaced with sequence encoding moZP2^18–156 using huZP2-GalK and huZP2-moZp2 primers (Table S3), and DNA encoding moZP2^51–149 protein was removed using moZp2-GalK and bridging moZp2 primers (Table S3) to establish the mo/huZP2 and moZp2^Trunc transgenes, respectively. For the mo/huZP2 transgene, mouse genomic DNA encoding moZP2^18–156 (bp 1 in exon 2 to bp 150 in exon 5) replaced human genomic DNA encoding huZP2^22–161 (bp 1 in exon 2 to bp 152 in exon 5), and for the moZp2^Trnuc transgene, DNA encoding moZP2^51–149 (bp 59 in exon 2 to bp 27 in exon 6) was deleted (Fig. S1 A).

NotI fragments containing the hu/moZp2 (17.9 kb including 2.3 and 0.9 kb of the 5′ and 3′ flanking regions, respectively), mo/huZP2 (16.8 kb including 1.5 and 1.7 kb of the 5′ and 3′ flanking regions, respectively), and truncated moZp2^Trunc (13.9 kb including 2.3 and 0.9 kb of the 5′ and 3′ flanking regions, respectively) transgenes were retrieved from the BAC with pl253 (Gahlay et al., 2010), and the fidelity of coding regions was confirmed by DNA sequencing. After gel purification, the transgenes were injected into the male pronucleus of fertilized FVB/N eggs by the Taconic Transgenic Mouse Facility. At least two founders were established for each transgene and crossed into mouse Zp1-3 null and human ZP1-4 transgenic mouse lines.

Hu/moZp2 mice were genotyped by PCR ([95°C for 30 s, 58°C for 30 s, 72°C for 1 min] × 30 cycles, 72°C for 7 min, and 4°C for >30 min) using primers in intron 5 of human ZP2 and intron 6 of mouse Zp2, mo/huZP2 mice were genotyped using primers in intron 4 of mouse Zp2 and intron 5 of human ZP2, and moZp2^Trunc mice were genotyped by PCR using primers in intron 2 and intron 5 of mouse Zp2 (Table S4). Each transgenic line was crossed into the Zp2^Null background (Rankin et al., 2001) to establish hu/moZp2, mo/huZP2, and moZp2^Trunc rescue lines. To avoid ambiguity of PCR product sizes, the normal Zp2 allele was detected by three distinct primer sets for hu/moZp2 (Set 1), mo/huZP2 (Set 2), and moZp2^Trunc (Set 3) rescue lines (Fig. S1 A; Table S4).

Expression of normal Zp2, as well as hu/moZp2, mo/huZP2, and moZp2^Trunc transgenes, was detected by RT-PCR using as gene-specific primer sets (Table S5) and total RNA isolated from 8–12-wk-old mouse tissue (Fig. S2 B). Gapdh transcripts served as a control for RNA integrity (Baibakov et al., 2012).

Samples were mounted in PBS, and images of eggs, embryos, and beads were obtained with a confocal microscope (LSM 510; Carl Zeiss) using a 63×/1.2 NA water immersion objective lens at room temperature (Baibakov et al., 2007; Yauger et al., 2011). LSM 510 images were exported as full-resolution TIF files and processed in Photoshop CS5.5 (Adobe) to adjust brightness and contrast. Alternatively, confocal optical sections were projected to a single plane with maximum intensity and combined with DIC images of eggs or peptide beads using LSM image software.

Mouse ovaries were fixed in glutaraldehyde and embedded in glycol methacrylate before staining with periodic Schiff’s acid and hematoxylin (Yauger et al., 2011). Ovulated eggs were fixed in 2% paraformaldehyde before staining with rat or mouse monoclonal antibodies (1:50) specific to: moZP1 (Rankin et al., 1999), N terminus of moZP2 (East and Dean, 1984), C terminus of moZP2 (Rankin et al., 2003), moZP3 (East et al., 1985), huZP1 (Ganguly et al., 2010), huZP2 (Rankin et al., 2003), huZP3 (Rankin et al., 1998), and huZP4 (Bukovsky et al., 2008). Monoclonal antibodies to human ZP1 and ZP4 were a gift from S. Gupta (National Institute of Immunology, New Delhi, India). Primary antibody binding was detected with goat anti–mouse Alexa Fluor 568 (1:100) or goat anti–rat Cy3 (1:100) antibodies (Life Technologies). Monoclonal antibody CSLEX1 (BD) that binds to the sialyl-Lewis^X antigen was diluted 1:50 and detected with goat anti–mouse IgM-FITC (1:20; Life Technologies). Images were obtained at room temperature on a microscope (Carl Zeiss; Axioplan 2) equipped with a 40×/0.75 NA lens using a camera (AxioCam 1Cc1) and AxioVision software (all from Carl Zeiss).

To assay mouse sperm binding to the zona pellucida surrounding normal and transgenic mouse eggs, sperm were released from cauda epididymides, capacitated in human tubal fluid (HTF; EMD Millipore) supplemented with 0.4% BSA (Sigma-Aldrich) for 40 min (37°C, 90% N₂, 5% O₂, and 5% CO₂), and added to eggs in cumulus and embryos in 100 µl of HTF, 0.4% BSA at a final concentration of 10⁵ ml⁻¹ progressive motile sperm as determined by a HTM-IVOS (Version 12.3) motility analyzer (Hamilton Thorne; Gahlay et al., 2010). Zp3^EGFP mouse eggs (green zona) and two-cell embryos served as positive and negative wash controls, respectively. Samples were fixed in 2% paraformaldehyde and stained with Hoechst to identify nuclei. Bound sperm were quantified from z projections obtained by confocal microscopy (Baibakov et al., 2007), and results reflect the mean ± SEM from at least three independently obtained samples each containing 10–50 mouse eggs/embryos.

To assay human sperm binding, individual aliquots (0.5 ml) of liquefied human semen (Genetics & IVF Institute Fairfax Cryobank) were added to an Eppendorf tube (2.0 ml) containing 0.5 ml of 40% of PureSperm (Nidacon) layered over 0.5 ml of 80% PureSperm. After centrifugation (swinging bucket, 20 min × 300 g, 20°C) and removal of the supernatant, sperm were resuspended in the residual buffer and transferred into 1.0 ml HTF. After a second centrifugation (5 min × 300 g), sperm were resuspended in 0.2 ml of HTF/BSA, and 2–4 aliquots were mixed before evaluation by HTM-IVOS. Sperm were then diluted in HTF, 0.5% BSA to 10⁷ ml⁻¹ progressive motile sperm. Transgenic eggs in cumulus, normal eggs, Zp3^EGFP eggs (Zhao et al., 2002), and noninseminated, immature human oocytes were incubated in droplets (100 µl) of sperm (10⁷) in HTF, 0.5% BSA under mineral oil for 4 h (37°C, 90% N₂, 5% O₂, and 5% CO₂) in a BT37GP incubator (Planer). Eggs/embryos/oocytes were washed by serial transfer through 500 µl of HTF/BSA and fixed for imaging (Baibakov et al., 2012). For each experiment using human sperm, initial evaluations were conducted using 10 huZP2^Rescue eggs as positive controls to test batches of HTF media and quality of human sperm preparations, which varied.

To assess fertility, females (≥5) from each mouse line were singly co-caged with a fertile female (control, NIH Swiss) and mated (2:1) with a male (NIH Swiss) proven to be fertile. Litters were recorded until the NIH Swiss fertile female gave birth to at least three litters or after 5 mo of mating.

Female mice (8–12 wk old) were hormonally stimulated at 12:00 midnight with 5 IU of pregnant mare serum gonadotrophin (PMSG), and hCG was injected intraperitoneally 48 h later. Epididymal mouse sperm from three males was released into 1.5 ml of HTF, 0.4% BSA, and equilibrated for 1–5 h (37°C, 90% N₂, 5% O₂, and 5% CO₂). 500 µl was used to inseminate each of two female mice. Alternately, commercially obtained human semen (proven to bind and penetrate human ZP2^Rescue eggs in vitro) was prepared as described in “Assessment of sperm binding and fertility“ and diluted in 500 µl HTF 0.5% BSA that had been equilibrated (≥24 h) at 37°C, 90% N₂, 5% O₂, and 5% CO₂. After capacitation (1.5 h), the contents of two vials of human semen were used to inseminate a single female mouse. At 11:00 p.m. (11 h after hCG injection), females were restrained and inseminated with mouse or human sperm using an NSET device (ParaTechs) connected to a 1-ml syringe, which reached a single uterine horn without anesthesia and surgery (Snell et al., 1944). Females were mated overnight with sterile, vasectomized males, and ∼10 h later, eggs were collected from the oviducts of females with copulatory plugs.

HuZP2^Rescue and huZP3^Rescue females were stimulated with gonadotrophins and anesthetized. Human sperm were injected into the oviduct, proximal to the swollen ampulla. 2 h later, the mice were euthanized to collect eggs that were fixed and imaged by confocal microscopy (Sato and Kimura, 2001; Tokuhiro et al., 2012).

Ovulated eggs with intact zonae pellucidae from moQuad^(huZP4) and moQuad-moZp2^Trunc transgenic mice were resolved by SDS-PAGE and probed with monoclonal antibodies to the N and C termini of mouse ZP2 (Burkart et al., 2012).

cDNA encoding human (39–154 aa) and mouse (35–149 aa) ZP2 were cloned into pFastBac-HBM TOPO (Invitrogen) downstream of a polyhedron promoter and a 23-aa honeybee melittin signal peptide. Each clone was tagged with 6-histidine at the C terminus to facilitate purification. Using the human cDNA as a backbone, chimeric mouse–human ZP2 clones were generated by DNA synthesis and used to make additional baculovirus expression constructs in which sequences encoding mouse ZP2^52–83, ZP2^85–101, and ZP2^103–133 replaced endogenous human sequence. Recombinant peptides were expressed in Hi-Five cells, purified on IMAC Sepharose High Performance beads (GE Healthcare), and assayed on SDS-PAGE (Fig. S2 C) as described previously (Baibakov et al., 2012). Results reflect the mean ± SEM from at least three independently obtained samples, each containing 20–25 beads. Cloning, expression, and attachment to IMAC beads were performed in the Protein Expression Laboratory of the Advanced Technology Program, SAIC-Frederick, Inc.

Experiments with normal and transgenic mice were conducted in compliance with the guidelines of the Animal Care and Use Committee of the National Institutes of Health under a Division of Intramural Research, National Institute of Diabetes and Digestive and Kidney Diseases–approved animal study protocol. Human sperm-binding assays were conducted in compliance with an Institutional Review Board approved protocol, and the Office of Human Subject Research determined that federal regulations did not apply to the anonymous human sperm and noninseminated, immature oocytes.

Fig. S1 shows the exon maps of transgenic mice and documents their tissue-specific expression. Fig. S2 provides a schematic of transcervical insemination, documents mouse sperm binding to huQuad, huQuad-ZP2^Null, and huQuad-ZP2^Null; Zp2^Mo eggs, and resolves the mouse–human chimeric peptides expressed in baculovirus on SDS-PAGE. Table S1 summarizes the protein composition of the zonae pellucidae of the transgenic mouse lines, and Table S2 summarizes their fertility. Tables S3–S5 list primers used for evaluating transgenes (Table S3), genotyping (Table S4), and RT-PCR (Table S5).

We thank Galina Baibakov for her excellent animal husbandry of the mouse colony necessary for these investigations, and our colleagues for their useful discussions. Noninseminated, immature human oocytes were obtained from Shady Grove Fertility Clinic (Rockville, MD).

This research was supported by the Intramural Research Program of the National Institutes of Health, National Institute of Diabetes and Digestive and Kidney Diseases (ZIA-DK015603).

The authors declare no competing financial interests.