Heterogeneous Processing and Zona Pellucida Binding Activity of Pig Zonadhesin*

Zonadhesin is a mosaic protein in sperm membrane fractions that binds directly and in a species-specific manner to the extracellular matrix (zona pellucida) of the oocyte. The active form of pig zonadhesin from ca-pacitated, epididymal spermatozoa comprises two covalently associated polypeptide chains of M r 105,000 (p105) and M r 45,000 (p45). Here we report detection and characterization of multiple zonadhesin isoforms in freshly ejaculated cells. Antibodies to the predicted von Willebrand D0-D1, D1, and D3 domains of pig zonadhesin recognized p105, p45, and additional M r 60,000– 90,000 polypeptides in particulate fractions of uncapacitated cells. Although the p105/45 form constituted a minority of all zonadhesin forms in sperm membrane fractions, it was the predominant form capable of binding to the pig zona pellucida. Zonadhesin-binding sites were distributed over the entire zona pellucida. Anion exchange chromatography resolved active, p105/45 zonadhesin from the p60–90 inactive forms. Without disulfide bond reduction some zonadhesin was M r > 300,000, including M r 300,000 and 900,000 proteins comprising in part multimers of p105/45. The multimeric forms did not bind the zona pellucida as avidly as did the p105/45 monomer. Expressed D1 and D3 domain fragments containing the CG(L/V)CG in HNE were stored at (cid:2) 70 °C. ZP Binding Assays— Detergent-solubilized proteins from sperm membrane fractions were mixed with isolated ZP, and zonadhesin that bound directly to the particulate, native ZP was detected either by Western blotting (4, 5) or by epifluorescence. For localization of binding sites, sperm proteins were biotinylated (4) prior to solubilization and incubation with ZP. The ZP with bound sperm proteins were washed extensively with 20 m M NaHEPES, 0.5 M NaCl, 1 m M EDTA, 1% (v/v) Triton X-100, 0.5% (w/v) sodium deoxycholate, 0.1% SDS, pH 7.5 (mRIPA) (4), and then bound, biotinylated proteins were detected by incubating for 15 min at 22 °C with Texas Red-labeled streptavidin (Molecular Probes, Eugene, OR) diluted 10,000-fold in 10 m M Tris-HCl, 150 m M NaCl, 0.1% (v/v) Tween 20, pH 7.5 (TBST). After washing three times for 5 min with TBST, ZP were dropped on coverslips, air dried, mounted with Fluoromount G (Electron Microscopy Sciences, Fort Washington, PA), and viewed by epifluorescence. ZP-bound forms of zonadhesin were characterized by Western blotting. Biotinylated zonadhesin polypeptides that remained bound to ZP after washing with mRIPA were detected by probing blots with horseradish peroxidase- streptavidin (4). Alternatively, zonadhesin polypeptides that remained bound after washing with 1% CHAPS/HNE were detected by probing blots with specific antisera as described below.

teins to complementary ligands in the ZP (1,2). The complexity of this process derives partly from cellular changes that occur during gamete interactions. Spermatozoa undergo physiological changes in the female reproductive tract that are required for fertilization and are collectively called capacitation (1,3). Although the molecular basis of capacitation is only partly understood, in some if not all species avidity of sperm-ZP adhesion increases as capacitation progresses. After capacitation is completed, the membranes involved in initial adhesion events are lost from the sperm surface during the acrosome reaction, but adhesion is sustained by interaction of newly exposed structures with the ZP (1,2). Unique adhesion molecule pairs likely function at different times during fertilization, and the activities of these molecules may change as fertilization progresses (2). It is therefore important to assess the biochemical and functional properties of sperm adhesion molecules at each stage in the fertilization process.
Several sperm proteins that may mediate adhesion to the ZP have been identified and characterized (2). Among these molecules zonadhesin is unique in its ability to bind directly and in a species-specific manner to native, particulate ZP (4,5). Zonadhesin from pig (5), mouse (6), rabbit, 2 and human (7) 3 spermatozoa is a mosaic protein with a predicted Type I integral membrane topology. In each of these species, the large extracellular region of the protein comprises primarily three domain types (meprin/A5 antigen/mu receptor tyrosine phosphatase, mucin, and von Willebrand D (VWD)) that are present in other adhesion molecules (8 -10). Although the domain structures of zonadhesin from these four mammals have been predicted from cDNA sequences, relatively little is known about the biochemical and functional properties of the proteins.
The active form of pig zonadhesin in membrane fractions of capacitated, epididymal spermatozoa is a two-chain molecule with disulfide-bonded M r 105,000 and 45,000 polypeptides, both of which are derived from a predicted 2467-amino acid nascent precursor (4,5). High M r forms of zonadhesin have also been observed, suggesting the possible formation of covalent oligomers (4). This possibility was further implied by the presence in the pig zonadhesin D1, D2, and D3 domains of a conserved CG(L/V)CG sequence motif (5) that is important for the oligomerization and proper function of von Willebrand factor (11) and for the oligomerization of porcine submaxillary mucin (12)(13)(14). These observations suggested that the protein at a minimum undergoes limited proteolysis and possibly also oligomerization as occurs in the functional maturation of vWF and other D-domain proteins (10). However, it is unclear when during sperm maturation such post-translational processing occurs or whether it is important for the ZP binding activity of zonadhesin.
Here we report that heterogeneous post-translational processing gives rise to multiple isoforms of pig zonadhesin in freshly ejaculated spermatozoa. Among these, only forms comprising the p105 and p45 polypeptides possess ZP binding activity, and the monomeric p105/45 form binds more avidly than do higher order covalent oligomers. Furthermore, we find that zonadhesin binds uniformly to homologous ZP and localizes to the apical head of pig spermatozoa. These properties further support a function for zonadhesin in sperm adhesion to the extracellular matrix of the egg.

EXPERIMENTAL PROCEDURES
Isolation of Sperm Membrane Fraction-Boar spermatozoa in extended, freshly ejaculated semen were washed and immediately disrupted by N 2 cavitation at 650 p.s.i. (15). Particulate fractions enriched in sperm plasma membranes were isolated from suspensions of disrupted cells by differential centrifugation (15) as for previous studies with cauda epididymal spermatozoa (4,5). Solutions for sperm fractionations were buffered at pH 7.5 and contained EDTA (1 mM) and diisopropyl fluorophosphate (1 mM) to prevent proteolysis by acidic proteases, Ca 2ϩ -dependent metalloproteinases, or serine proteases, respectively. In some experiments, solutions also contained 1 mM iodoacetamide to inhibit thiol proteases and to prevent thiol oxidation. Isolated membrane fractions in 20 mM NaHEPES, 130 mM NaCl, 1 mM EDTA, pH 7.5 (HNE), were stored at Ϫ70°C.
Isolation of Zona Pellucida-Porcine ZP were isolated from sliced ovaries by stepwise sieving through screens (16) and then further purified by ultracentrifugation through Percoll (Amersham Pharmacia Biotech) gradients (4). Isolated ZP in HNE were stored at Ϫ70°C.
ZP Binding Assays-Detergent-solubilized proteins from sperm membrane fractions were mixed with isolated ZP, and zonadhesin that bound directly to the particulate, native ZP was detected either by Western blotting (4,5) or by epifluorescence. For localization of binding sites, sperm proteins were biotinylated (4) prior to solubilization and incubation with ZP. The ZP with bound sperm proteins were washed extensively with 20 mM NaHEPES, 0.5 M NaCl, 1 mM EDTA, 1% (v/v) Triton X-100, 0.5% (w/v) sodium deoxycholate, 0.1% SDS, pH 7.5 (mRIPA) (4), and then bound, biotinylated proteins were detected by incubating for 15 min at 22°C with Texas Red-labeled streptavidin (Molecular Probes, Eugene, OR) diluted 10,000-fold in 10 mM Tris-HCl, 150 mM NaCl, 0.1% (v/v) Tween 20, pH 7.5 (TBST). After washing three times for 5 min with TBST, ZP were dropped on coverslips, air dried, mounted with Fluoromount G (Electron Microscopy Sciences, Fort Washington, PA), and viewed by epifluorescence. ZP-bound forms of zonadhesin were characterized by Western blotting. Biotinylated zonadhesin polypeptides that remained bound to ZP after washing with mRIPA were detected by probing blots with horseradish peroxidasestreptavidin (4). Alternatively, zonadhesin polypeptides that remained bound after washing with 1% CHAPS/HNE were detected by probing blots with specific antisera as described below.
Expression and Purification of D0-D1 Fusion Protein-The 1.7-kilobase EcoRI fragment of pig zonadhesin cDNA clone M2 (in pBluescript) was subcloned into the EcoRI site of pET-23d. The 5Ј sticky end of this fragment came from the EcoRI site in the adapter used to construct the cDNA library (5), and its 3Ј sticky end came from the EcoRI site at nucleotides 4063-4068 of the zonadhesin composite cDNA (GenBank TM accession number U40024). This construct specified an M r 64,000 fusion protein comprising 20 amino acids of N-terminal vector-encoded protein, 19 amino acids of C-terminal vector-encoded protein (including a hexahistidine tag), and zonadhesin amino acids Pro 683 -Ser 1224 . The fusion protein was expressed in Escherichia coli strain BL21/DE3 by induction with 0.5 mM isopropylthiogalactoside for 2 h at 37°C and isolated from inclusion bodies by preparative SDS-PAGE and electroelution.
Preparation of D0-D1 Antisera-Asp-Pro bonds of the purified D0-D1 fusion protein were hydrolyzed for 36 h with 70% formic acid at 37°C. The final hydrolysates contained a mixture of proteins with M r values corresponding to partial hydrolysis products predicted from the deduced amino acid sequence, including an M r 33,000 core polypeptide. Hydrolysates were lyophilized to remove formic acid prior to injection. Two female New Zealand White rabbits were immunized (intramuscular) with 0.2-0.5 mg of protein each in 1 ml of Freund's complete adjuvant (day 0). Booster injections (intramuscular) on day 45 consisted of 0.2-0.5 mg of protein each in 1 ml of Freund's incomplete adjuvant. Antisera were recovered from blood obtained by terminal exsanguinations on day 58.
Expression and Purification of D1 and D3 Fusion Proteins-Glutathione S-transferase (GST) fusion proteins comprising in part amino acids Ser 923 -Met 993 of the D1 domain or amino acids Ile 1684 -Pro 1788 of the D3 domain were expressed in E. coli strain BL21. Polymerase chain reaction products (D1 sense primer, 5Ј-AGTGGATCCAGCACCTTCTC-TGG-3Ј; D1 antisense primer, 5Ј-ATAGAATTCTGCTAGGCCGTGTTG-3Ј; D3 sense primer, 5Ј-CATCGGATCCCAGGTCAAGTTTGACGG-3Ј; and D3 antisense primer, 5Ј-GGGGAATTCTAGGCCGCCTG-3Ј; underlined bases denote mismatches introduced to create restriction sites and stop codons) encoding the D1 and D3 domain segments were directionally cloned into the BamHI and EcoRI sites of pGEX-2T, and fusion protein expression was induced with 0.1 mM isopropylthiogalactoside at 37°C for 2 h. After washing the bacteria with 10 mM NaPO 4 , 150 mM NaCl, pH 7.4 (PBS), soluble fusion proteins were extracted by sonicating cell pellets in PBS containing 0.5 mM diisopropyl fluorophosphate, 1.0 mM EDTA, 10 mM E64, and 0.2% Triton X-100. Cell lysates were applied to a glutathione (GSH)-Sepharose column (15 ml of bed volume) equilibrated at 22°C in PBS. Nonbinding proteins were washed through with PBS, and fusion proteins were eluted with 5 mM GSH in 50 mM Tris-HCl, pH 8.0. Eluted fusion proteins were present at concentrations of 5-8 mg/ml in the pooled, peak fractions, and with prior disulfide bond reduction migrated as single bands in SDS-PAGE (10% gels). Total yields of purified fusion protein were 40 -45 mg/500 ml of culture.
Preparation of Domain-specific Antisera-Four female New Zealand White rabbits were immunized (intramuscular) with 1 mg of purified fusion protein/animal (two with GST-D1 and two with GST-D3) emulsified in 0.5 ml of Freund's complete adjuvant (day 0). Booster injections consisted of 1 mg of purified fusion protein/animal emulsified in 0.5 ml of Freund's incomplete adjuvant (intramuscular) on day 42, and 1 mg of soluble protein/animal in PBS (subcutaneous) on days 49 and 70. Antisera were recovered from blood obtained by terminal exsanguinations on day 81.
Preparation of GST, GST-D1, and GST-D3 Affinity Columns-Purified GST (100 mg), GST-D1 (70 mg), and GST-D3 (70 mg) were dialyzed at 4°C for Ͼ16 h in 0.1 M NaHCO 3 , 0.5 M NaCl, pH 8.3, to remove GSH and exchange into conjugation buffer. Dialyzed proteins were each coupled at 10 mg/ml swelled gel to CNBr-activated Sepharose 4B (Amersham Pharmacia Biotech). After washing by suction on a glass filter to remove uncoupled proteins, the remaining activated groups were blocked with 1 M ethanolamine, and the conjugated resins were washed with three cycles of alternating pH (0.1 M acetate, 0.5 M NaCl, pH 4.0, and 0.1 M NaHCO 3 , 0.5 M NaCl, pH 8.3). The affinity matrices were poured into 1-cm-diameter glass columns, equilibrated in PBS containing 0.02% NaN 3 , and then stored at 4°C.
Affinity Purification of D1 and D3 Antibodies-Antibodies to GST were removed by passing 20 ml of antisera through a 10-ml bed volume GST-Sepharose column equilibrated at 22°C in PBS. Antibodies to zonadhesin D1 or D3 domains were then affinity purified from their anti-GST-depleted sera by chromatography on GST-D1 or GST-D3 columns, respectively (7 ml of bed volume each, equilibrated in PBS at 22°C). Elution of bound antibodies with 0.2 M sodium citrate, 0.15 M NaCl, pH 3.0, was monitored continuously by A 280 . Peak fractions were pooled and immediately adjusted to pH 7 by addition of 1 M Tris (unbuffered). Antibodies to GST that were removed in the initial depletion steps were similarly eluted from GST-Sepharose and recovered for use as affinity-purified control antibodies. All purified antibodies were stored at Ϫ70°C.
Preparation of D3 Immunoaffinity Column-20 mg of affinity-purified antibody to the D3 domain (11.4 mg from rabbit R128 and 8.6 mg from rabbit R129) were desalted into 0.1 M NaHCO 3 , 0.5 M NaCl, pH 8.3 (coupling buffer) in two runs on four tandem 5-ml HiTrap desalting columns (Amersham Pharmacia Biotech). Desalted protein (15 mg in 8.8 ml) was coupled to 0.43 g (dry weight) of freshly swollen CNBractivated Sepharose 4B. A protein assay of uncoupled protein confirmed that more than 95% of the antibody (Ͼ14.5 mg) was coupled to the affinity matrix (1.5 ml of packed volume), which after blocking and washing as for the fusion protein affinity matrices was equilibrated in PBS containing 0.02% NaN 3 and stored at 4°C.
Immunoaffinity Purification of Zonadhesin from Spermatozoa-Sperm membrane fractions (100 mg protein) were Dounce homogenized in 10 ml of 1% SDS/HNE and incubated for 30 min at 22°C. The homogenate was diluted to 100 ml with HNE containing 0.5 mM diisopropyl fluorophosphate, 0.56% (w/v) sodium deoxycholate, and 1.1% (v/v) hydrogenated Triton X-100 to produce the composition of mRIPA, a detergent solution in which zonadhesin retains its ZP binding activity (4). After ultracentrifugation for 1 h at 100,000 ϫ g at 2°C, up to 50 ml of the supernatant solution (mRIPA extract) was applied to a 1.5-ml anti-D3 column, and nonbound proteins were washed through with mRIPA until A 280 (monitored continuously) returned to base line. The column was further washed with 10 ml of HNE containing 1% (v/v) hydrogenated Triton X-100, and then the protein was eluted with 10 ml of 0.2 M sodium citrate, 0.15 M NaCl, 1% (v/v) hydrogenated Triton X-100, pH 3.0. Eluted protein fractions containing purified zonadhesin were pooled, adjusted to pH 7 with 1 M Tris (unbuffered), and stored at Ϫ70°C.
Preparation of Antisera to Zonadhesin Holoprotein-Two female New Zealand White rabbits were immunized (intradermal) with 100 g of purified zonadhesin holoprotein/animal emulsified in 1 ml of Freund's complete adjuvant (day 0). After boosting with 100 g of purified protein/animal emulsified in 1.2 ml of Freund's incomplete adjuvant (intramuscular) on day 45, antisera were recovered from blood obtained by terminal exsanguinations on day 60.
In Vitro Multimerization-Wild type and mutant fusion proteins were purified by GSH affinity chromatography in the presence of 1 mM DTT to stabilize the expressed proteins in their monomeric forms. The purified fusion proteins were then desalted on Sephadex G50 spin columns equilibrated in 50 mM Tris-HCl, pH 7.4, containing 1 mM DTT. For time course studies, oxidized glutathione (GSSG) was added to desalted fusion protein (1 g in Tris-DTT) to a final concentration of 25 mM, which upon reaction with DTT produced a 24 mM GSSG/2 mM GSH redox buffer with an effective potential in solution (E h ) at pH 7.4 of Ϫ246 mV. The reactions were incubated at 37°C, and disulfide bond formation was terminated at various times by adding iodoacetamide to 60 mM. To determine dependence of multimer formation on E h , 5-25 mM GSSG was added to reactions to produce redox buffers ranging from Ϫ269 to Ϫ246 mV E h . After incubating for 2 h at 37°C, the reactions were terminated with iodoacetamide as for time course studies. Multimers in the terminated reactions were separated by SDS-PAGE and detected by staining with Coomassie Blue.
Immunofluorescence-All steps for immunolocalization experiments were done at 22°C. Spermatozoa were recovered from pig epididymides and washed with HEPES-buffered medium as described previously (4). For immunofluorescence, the cells were smeared on coverslips, air dried, and then fixed in methanol for 30 min. After blocking 30 min with 10% (v/v) heat-inactivated goat serum in PBS, the coverslips were floated 1 h on D0-D1 antisera diluted 1:400 in PBS/heat-inactivated goat serum. After washing coverslips with PBS, bound antibody was detected by incubating for 1 h with Texas Red-conjugated antibody to rabbit immunoglobulin (BIOSOURCE International) diluted 1:400 in PBS/heat-inactivated goat serum. After a final wash with PBS, cover-slips were mounted with Fluoromount G and viewed by epifluorescence and phase contrast microscopy.

RESULTS AND DISCUSSION
Zonadhesin from membrane fractions of capacitated, epididymal spermatozoa bound directly and with high affinity to intact ZP (Fig. 1). The bound zonadhesin comprised p105 and p45 polypeptides (Fig. 1a) as previously observed (4). Although earlier work established the species specificity of this interaction, the distribution of zonadhesin-binding sites in the pig ZP has not been characterized. We therefore visualized ZP-bound zonadhesin in situ by affinity fluorescence (Fig. 1, b and c). Zonadhesin protein was detected on the entire ZP, indicating that its binding sites were not regionalized in the ZP structure. In addition, the relative evenness of the labeling suggested that the binding sites are intrinsic to the ZP and not associated with adherent materials from the cumulus cell matrix or other potential contaminants of the ZP preparation.
The locations of p105 and p45 tryptic peptides in the sequence of the pig zonadhesin precursor indicated that p45 comprises in part the D1 domain and that p105 comprises in part the D2 and D3 domains (ref. (5) and Fig. 2). To detect zonadhesin isoforms in spermatozoa and to characterize their polypeptide compositions, we prepared various domain-directed antisera and affinity-purified antibodies. The deduced sequence of the precursor specified numerous potentially antigenic regions, including segments located in approximately the same positions within the D1 and D3 domains that exhibited FIG. 1. Direct binding of zonadhesin to intact ZP. a, biotinylated polypeptides in a membrane fraction of capacitated, epididymal spermatozoa that bound to the pig ZP. CHAPS-solubilized sperm proteins were incubated with particulate ZP. The pig ZP with bound proteins were washed sequentially with CHAPS/HNE and mRIPA, and bound, biotinylated proteins were detected on Western blots (10% SDS-PAGE, disulfides reduced) by probing with horseradish peroxidase-streptavidin as described previously (4,5). Note that the p105 and p45 major polypeptides of zonadhesin remained bound under these conditions. Peptides comprising a minor portion of the bound zonadhesin (4) migrated at the dye front (df). b, epifluorescence image of zonadhesin bound to the pig ZP. Bound, biotinylated zonadhesin on intact ZP from the experiment shown in a was detected in situ by probing with Texas Red-streptavidin. The labeled ZP were then viewed by fluorescence microscopy. Note that the zonadhesin-derived fluorescence is uniform, and associated with all regions of the ZP fragments. c, differential interference contrast image of the ZP shown in b.
high predicted hydrophilicity, surface probability, and flexibility (Fig. 2). Accordingly, we raised antisera to a Gene 10 fusion protein spanning the D0 and D1 domains (Pro 683 -Ser 1224 ), as well as to two GST fusion proteins comprising in part the short, antigenic segments identified in the D1 and D3 domains (Ser 923 -Met 993 and Ile 1684 -Pro 1788 , respectively; Figs. 2 and 3). We also raised antisera to whole zonadhesin isolated from membrane fractions of pig spermatozoa (Fig. 3).
The D0-D1 fusion protein was recovered from inclusion bodies and purified by preparative SDS-PAGE (Fig. 3a). Rabbit antisera to this recombinant protein recognized the vectorencoded amino acids at the N and C termini of the fusion protein but did not cross-react with zonadhesin (not shown). We therefore hydrolyzed the purified fusion protein at Asp-Pro bonds to remove the short, vector-encoded peptides. The Asp-Pro hydrolysis products corresponded to those predicted by the precursor sequence, thereby confirming that the hydrolysate contained the desired zonadhesin fragments (Fig. 3a). This hydrolysate was used to prepare antisera to the D0-D1 domains.
In contrast to the D0-D1 fusion protein, the GST-D1 and GST-D3 proteins were expressed in soluble form and purified by chromatography on GSH-Sepharose (Fig. 3a). Antisera to these expressed, recombinant proteins were fractionated by chromatography on D1 and D3 affinity columns to produce reagents for specific detection of these domains. The D0-D1 antisera and the D1 antibody each recognized both the D0-D1 and GST-D1 fusion proteins but not the GST-D3 protein (Fig.  3a). Similarly, the D3 antibody recognized the GST-D3 protein but not the other two fusion proteins. The absence of crossreactivity with other D-domains indicated that these reagents were suitable for domain-specific detection of zonadhesin polypeptides.
We used the D3 antibody to prepare an immunoaffinity column for purification of zonadhesin from membrane fractions of freshly ejaculated spermatozoa (Fig. 3b). Without prior reduction of disulfide bonds, the purified zonadhesin migrated in SDS-PAGE primarily as an M r 150,000 protein (p150) and a less prominent M r Ͼ300,000 protein. When protein disulfides were reduced, the same zonadhesin preparation migrated primarily as p45, p105, and an M r 300,000 polypeptide (p300). This purified zonadhesin preparation was used to raise antisera to the fully processed and disulfide-bonded holoprotein and to confirm the reactivity and specificity of antisera and affinity-purified antibodies (Fig. 3b). Fig. 4 shows the reactivity of the four immunoreagents with zonadhesin isoforms on Western blots of membrane fractions from pig ejaculated spermatozoa. Antisera to the hydrolyzed D0-D1 protein recognized primarily p150 on blots of nonreduced proteins (Fig. 4a). In contrast to this relatively weak interaction with a single, disulfide-bonded form of zonadhesin, the D0-D1 antisera detected several polypeptides of disulfidereduced zonadhesin, including the p105 and p45 components described previously, as well as an M r 300,000 protein (p300) and at least two other polypeptides of intermediate size (designated p60 -90). The reaction of the D0-D1 antisera with p45 was consistent with the presence of p45 tryptic peptides in the D1 domain (Fig. 2). The reaction of these sera also with p105 indicated that the N terminus of p105 is likely upstream of Ser 1224 (the C terminus of the expressed D0-D1 fragment). The absence of the p60 -90 polypeptides from the holoprotein purified by D3 immunoaffinity chromatography (compare with Fig.  3) indicates that these D0-D1-reactive polypeptides neither contain nor are covalently associated with a D3 polypeptide.
Like the D0-D1 antisera, the D1-and D3-specific antibodies also detected p150 zonadhesin in nonreduced sperm proteins (Fig. 4a). However, in contrast to the complex pattern of disulfide-reduced proteins the D0-D1 antisera recognized, the D1 antibody recognized only p45. Similarly, the D3 antibody bound primarily to p105 and more weakly to an M r 60,000 polypeptide. Affinity-purified control antibodies to GST did not bind significantly to sperm proteins (not shown). The antiserum to the zonadhesin holoprotein detected p150 in nonreduced sperm proteins but primarily recognized proteins with M r Ͼ Ͼ300,000. Overall, the D0-D1 antisera reacted much more weakly with nonreduced forms of zonadhesin than it did with the reduced, constituent polypeptides. In contrast, the D1 and D3 antibodies bound similarly to both nonreduced and reduced zonadhesin, and the antiserum to the zonadhesin holoprotein reacted strongly with nonreduced zonadhesin (Figs. 3b and 4a) but did not recognize the protein's separated, disulfide-reduced polypeptides. The differential binding of our antibodies to re-FIG. 2. Domain structure, polypeptide composition, and predicted characteristics of the pig zonadhesin precursor. The N-terminal domain (designated N) is composed of one partial and one full meprin/A5 antigen/mu receptor tyrosine phosphatase domain, whereas this region of mouse and human zonadhesins comprises three tandem meprin/A5 antigen/mu receptor tyrosine phosphatase domains. Vertical bars mark the locations in the precursor of tryptic peptides that were previously isolated from p45 and p105 and sequenced (5). Note that p45 must include much of the D1 domain and that p105 includes most or all of the D2 and D3 domains. Horizontal bars denote segments in the tandem VWD domains that were expressed as fusion proteins for production of antisera. Two potentially antigenic segments in the D1 and D3 domains (short horizontal bars) exhibited high predicted hydrophilicity, flexibility, and surface probability. duced and nonreduced zonadhesin likely reflects structural properties determined in part by the many intramolecular disulfide bonds present in von Willebrand D-domains (21). The cross-reaction of the antibodies to the GST-D1 and GST-D3 proteins with nonreduced zonadhesin extracted from spermatozoa, both on blots and in the isolation of the holoprotein by immunoaffinity chromatography, further suggested that the tertiary structures of these soluble fusion proteins are similar to that of the native protein.
Two-dimensional SDS-PAGE (first dimension, disulfides not reduced; second dimension, disulfides reduced) revealed that the p60 -90 zonadhesin polypeptides migrated with the same mobility in each dimension (i.e. on the gel diagonal) and are therefore not covalently bound to other polypeptides. In con-trast, p105 and p45 were components of M r 150,000, 300,000, and Ն900,000 complexes stabilized by intermolecular disulfide bonds (Fig. 4b). The presence of both p105 and p45 in the M r 150,000 nonreduced protein (p150) indicated that this zonadhesin form comprised primarily one each of the two polypeptides, consistent with both polypeptides being derived by proteolysis from a single precursor molecule. Similar compositions of the M r 300,000 and Ն900,000 nonreduced proteins, in particular the presence of mostly p105 and p45 in a ratio similar to that of p150, suggested that these large complexes are covalent dimer and hexamer respectively of p150 (Fig. 4b). The relative amounts of the M r 150,000, 300,000, and Ն900,000 zonadhesins did not change substantially when 1 mM iodoacetamide was included in membrane isolation buffers to inhibit poten- Similarly, the anti-D3 antibody recognized GST-D3 but not the other fusion protein. Note also that antisera to the purified, disulfide-bonded zonadhesin holoprotein did not recognize the disulfide-reduced fusion proteins under these conditions. All developed blots for a were exposed to film for 40 s. b, composition and immunoreactivity of processed zonadhesin holoprotein purified from membrane fractions of pig ejaculated spermatozoa. Shown are protein stains and Western blots of SDS-PAGE (4 -12% linear gradient gels, protein disulfides either reduced (lanes labeled R) or not reduced (lanes labeled NR) as indicated). Note the different mobilities of the constituent polypeptides of the purified protein when separated without and with prior reduction of disulfide bonds (lanes overlined with Silver stains). Note also that the different antisera and antibodies each recognized a unique subset of the disulfide reduced and nonreduced polypeptides (lanes overlined with Western blots, containing 220 ng zonadhesin/lane). The developed anti-D0-D1, anti-D1, and anti-D3 blots of b were exposed to film for 40 s, whereas the anti-holoprotein blot was exposed for 5 s. tially artifactual thiol-disulfide exchange (data not shown). Thus the M r 300,000 and 900,000 zonadhesins likely exist as true multimers in the sperm cell. Other VWD domain proteins form oligomers that are functionally important. For example, the absence of the largest vWF multimers causes a mild clotting deficiency distinct from other forms of von Willebrand disease (22), presumably because the higher inherent avidity of multivalent interactions is necessary for proper clot formation. Although vWF multimerizes extensively (23,24), porcine submaxillary mucin preferentially forms trimers in transfected cells (12,25). The apparent absence in spermatozoa of zonadhesin multimers other than dimer and hexamer indicates that its oligomerization, like that of porcine submaxillary mucin, is more limited and specific than the multimerization of vWF.
To test the ZP binding activity of the zonadhesin isoforms present in ejaculated cells, we solubilized membrane fractions of spermatozoa in 1% CHAPS/HNE, incubated the extract with particulate ZP, and then removed nonbinding proteins by sequential washing with CHAPS/HNE and mRIPA. Under these conditions, zonadhesin was the only protein from membrane fractions of capacitated, epididymal spermatozoa that re-mained bound to the ZP ( Fig. 1 and Ref. 4). In contrast to previous ZP binding studies that used biotinylated sperm proteins, however, in this experiment ZP-bound zonadhesin was detected on Western blots with the D0-D1 antisera. Only zonadhesin forms comprising the p105 and p45 polypeptides bound to the ZP (Fig. 5a), and these p105/45 zonadhesin molecules were separable by anion exchange chromatography from other forms that lacked ZP binding activity (Fig. 5b). In twodimensional SDS-PAGE (Fig. 6), the predominant ZP-bound form of p105/45 zonadhesin migrated with M r 150,000 in the first dimension (disulfides not reduced). The M r 300,000 (dimer) and Ն900,000 (hexamer) proteins also bound to the ZP, but a significant proportion of these zonadhesin forms remained in the nonbinding fraction (Fig. 6). Thus the p105/45 monomeric form of zonadhesin bound preferentially to the ZP in vitro, although p105/45 multimers also possessed ZP binding activity.
At high concentrations in storage, the purified D1 and D3 fusion proteins formed viscous gels that liquefied upon the addition of 10 mM DTT. This observation suggested that the fusion proteins, which each contained the CG(L/V)CG sequence motif, had spontaneously formed intermolecular disulfide bonds even though a mild reductant was present (the 5 mM GSH used to elute them from GSH-Sepharose). SDS-PAGE without prior reduction of disulfides revealed that covalent multimers were indeed present in stored preparations of both fusion proteins but not in identically stored GST (Fig. 7a). Including DTT in isolation buffers preserved the proteins in their monomeric forms (Fig. 7b). The addition of oxidized DTT at concentrations up to 100 mM, which raised the effective potentials in solution (E h ) at pH 7.4 as high as Ϫ292 mV, did not induce formation of covalent multimers in vitro (not shown). However, the addition of 25 mM GSSG to produce a Ϫ246 mV redox buffer induced rapid formation of disulfidebonded multimers of both proteins (D1 shown in Fig. 7b; D3 not shown). Most of the D1 fusion protein was converted to multimers within 30 min, and multimer formation continued until very little monomeric protein remained (Fig. 7b). To determine whether the vicinal cysteines in the CG(L/V)CG sequence motif FIG. 4. Identification of multiple zonadhesin forms in membrane fractions of freshly ejaculated spermatozoa. Proteins were separated by SDS-PAGE and Western blotted, and zonadhesin forms were detected with antisera or affinity-purified antibodies. a, reaction of D0-D1 antisera, D1 antibody, D3 antibody, and zonadhesin holoprotein antisera with zonadhesin separated on 4 -12% linear gradient gels. Note the differences in polypeptides recognized depending on whether protein disulfides were reduced (lanes labeled R) or not reduced (lanes labeled NR), and the M r Ն300,000 disulfide nonreduced protein (asterisks) detected by the D3 antibody and the holoprotein antisera. b, reaction of D0-D1 antisera with zonadhesin polypeptides separated by two-dimensional SDS-PAGE. Note the presence of p60 -90 on the diagonal and the migration in the first dimension of p105 and p45 as M r 150,000, 300,000, and 900,000 complexes. were important for multimer formation, we compared the multimerization kinetics of the Cys 3 Ser mutants with those of the wild type proteins. Single mutants of the D1 protein (C933S and C936S) formed multimers at the same apparent rates as the wild type proteins (not shown). The double mutant of the D1 protein (C933S,C936S) also formed multimers at Ϫ246 mV with the same kinetics as the wild type protein (Fig. 7b). Furthermore, the multimerization of the wild type and double mutant D1 fusion proteins were indistinguishable at various E h values ranging from Ϫ269 to Ϫ246 mV (Fig. 7c). Neither the wild type nor the mutant protein multimerized significantly at Ϫ269 mV, but multimer formation was clearly evident when E h was raised to Ϫ259 mV. Double mutation of the 1709 CGVCG 1713 motif in the D3 fusion protein to SGVSG also did not perturb multimer formation (not shown). Collectively, these results demonstrate that the expressed D1 and D3 fragments of pig zonadhesin spontaneously form intermolecular disulfide bonds and that the reaction is dependent on E h but not on the vicinal cysteines in the CG(L/V)CG sequence motif known to be important for vWF multimer formation. Our D1 and D3 fragments contained two and three additional cysteines, respectively, downstream of the two in the CG(L/V)CG motif, one or more of which must form cystine in the multimerization of these proteins. Studies on porcine submaxillary mucin have identified half-cystines that are not in the CG(L/V)CG motif but are nonetheless important for the dimerization of fragments (13) or multimerization of full D-domains (12). Nevertheless, when they are expressed in eukaryotic cells, Cys 3 Xaa mutants of the CG(L/V)CG motif in vWF (11) or in submaxillary mucin (12) generally exhibit defects in multimer formation. Our results show that cysteines other than those in the CG(L/V)CG can mediate spontaneous multimerization of purified zonadhesin fragments in vitro. Further studies, including testing whether mutants of downstream cysteines (Cys 975 and Cys 985 ) in GST-D1 form multimers, will be necessary to determine whether our results reflect differences in methods or true dif- Binding experiments were performed as for Fig. 5a. Proteins in the nonbinding and ZP-bound fractions were separated by two-dimensional electrophoresis, and zonadhesin forms were detected on Western blots with the D0-D1 antisera. a, nonbinding fraction. b, binding fraction. Note that only zonadhesin forms comprising the p105 and p45 polypeptides exhibited ZP binding activity. Note also that the proportion of p105/45 monomer (M r 150,000 when disulfides were not reduced) that bound to the ZP was higher than those of the other p105/45 forms (dimer and hexamer).

FIG. 7. Multimerization of D1 and D3 fusion proteins in vitro.
a, spontaneous multimerization of purified GST-D1 and GST-D3 in storage. Proteins were separated by SDS-PAGE (10% gel, silver stained) with (left gel) or without (right gel) prior reduction of disulfide bonds. Note the high M r multimers present in the stored D1 and D3 preparations but not in identically stored preparations of GST. b, time course of D1 wild type and double mutant (C933S,C936S) multimerization at Ϫ246 mV E h . The fusion proteins were purified in the presence of 1 mM DTT to inhibit premature multimer formation. GSSG (25 mM) was then added to the purified proteins (upper gel, wild type; lower gel, double mutant) to initiate multimer formation. Reactions were terminated with iodoacetamide at intervals ranging from 15 min to 18 h as shown, and multimers were detected by SDS-PAGE (10% gels). Note that GSSG induced significant multimer formation within 30 min, whereas in the absence of GSSG neither protein formed multimers even after prolonged incubation (compare GSSG at time 0 with GSSG at 18 h). c, dependence of multimer formation on E h . Purified GST-D1 wild type and double mutant proteins were incubated with various concentrations of GSSG as indicated to promote multimer formation. E h ranged from Ϫ269 mV (at 5 mM added GSSG) to Ϫ246 mV (at 25 mM added GSSG). The reactions were terminated as for panel b, and multimers were detected by SDS-PAGE on a 4 -12% linear gradient gel. Note the presence of multimers in all reactions containing Ն10 mM added GSSG (E h to ՆϪ259 mV). ferences in multimer formation between zonadhesin and other D-domain proteins.
The D0-D1 antisera detected pig zonadhesin on the apical heads of methanol-fixed, epididymal spermatozoa (Fig. 8). All of the cells were strongly labeled when they were prepared by methanol fixation, which both denatures proteins and disrupts membranes. Thus pig zonadhesin is present in the anterior head of pig spermatozoa, which is a part of the cell that interacts with the ZP. However, because methanol fixation disrupts cell membranes, we cannot discern from this experiment whether the pig protein is present on the sperm cell surface. These results are consistent with those reported by Gao and Garbers (6), who used antisera to the unique, partial D domain repeats of mouse zonadhesin to detect the protein on the apical heads of paraformaldehyde-fixed spermatozoa.
These studies are relevant both to the potential function of zonadhesin in mammalian fertilization and to the functions of VWD domains in diverse proteins. To our knowledge no other sperm protein exhibits the physicochemical heterogeneity of zonadhesin. In ovine spermatozoa, hyaluronidase is present as a remarkably heterogeneous mixture of disulfide-bonded multimers (26 -28). Nevertheless, even this protein does not undergo the combined proteolytic removal of protein domains, generation of specific constituent polypeptides, and multimerization that occur in the processing of zonadhesin. In addition, hyaluronidase does not exhibit such marked variation in all species (29), whereas we find that zonadhesin is heterogeneous in spermatozoa from all mammals examined (eight species). 4 Our detection of differences in ZP binding activity among the various zonadhesin forms suggests that the heterogeneous processing of this protein is functionally important, just as proteolytic activation and heterogeneous multimerization are important for the proper function of vWF (10). Unlike vWF multimers, however, zonadhesin multimers appear to bind less avidly than the monomer. This observation, together with our detection of potential differences in the way zonadhesin and vWF form multimers, indicates that VWD domains are versatile structures that share certain properties but nonetheless have unique functions in different proteins. Multimerization of zonadhesin could represent a mechanism for storing the protein in a latent form that can be activated when it is required for interaction with the ZP, or it could reflect an additional function of the protein as a scaffold or other structural element of the sperm head. Further studies will be required to determine whether the ZP binding activity of zonadhesin is dynamically regulated during fertilization, for example, as a component of sperm capacitation.