The Mammalian Zona Pellucida: A Matrix That Mediates Both Gamete Binding and Immune Recognition?

Abstract

The crucial cell adhesion events required for mammalian fertilization commence when spermatozoa bind to the specialized extracellular matrix of the oocyte, known as the zona pellucida (ZP). Bound gametes then undergo a signal transduction cascade known as acrosomal exocytosis that enables them to penetrate this matrix and fuse with the oocyte to create a new individual. The ZP is therefore the target of intense investigation in the mouse, pig, bovine, and human models. Major goals in such studies are to define the adhesion molecules, signal transduction pathways, and the molecular basis for the species-restricted binding of gametes. Evidence exists indicating that protein-carbohydrate and to a lesser extent protein-protein interactions play a role in the initial gamete binding. More recent findings in an unusual sperm-somatic cell adhesion system indicate that tri- and tetraantennary N-glycans mediate initial sperm-oocyte binding in both the murine and porcine models, but conflicting data exist. A novel paradigm designated the “domain specific model” will be presented that could explain these inconsistencies. Another potential functional role of the ZP is immune recognition. Both spermatozoa and oocytes lack major histocompatibility (MHC) class I molecules that mediate the recognition of self in the immune system. This absence makes gametes less susceptible to class I restricted cytotoxic T lymphocytes, but more vulnerable to natural killer (NK) cells. Therefore a “fail safe” system for NK cell recognition should exist on both types of gametes. Another issue is that oocytes could begin to express paternal major histocompatibility antigens during the blastocyst stage prior to hatching, and thus mechanisms could also be in place to block the development of maternal adaptive immune responses. An enhanced understanding of these issues could facilitate the development of superior infertility treatments and contraceptive strategies, and define central operating principles of immune recognition in the female reproductive system.

KEYWORDS:

• egg binding proteins
• fertilization
• glycosylation
• immune response
• oligosaccharides.

Abbreviations
bZP:	=	bovine zona pellucida
EBP:	=	egg binding protein
huZP:	=	human zona pellucida
MHC:	=	major histocompatibility
mZP:	=	mouse ZP
NK:	=	natural killer
PTMs:	=	post-translational modifications
pZP:	=	porcine zona pellucida
ZP:	=	zona pellucida

INTRODUCTION

This review is focused on summarizing some of the more current data related to the structure and function of the mammalian zona pellucida (ZP) in four major models (mouse, pig, bovine, and human). In addition, a new paradigm for initial mammalian gamete binding is also presented in response to some contradictory data regarding the roles of protein-carbohydrate versus protein-protein interactions during murine fertilization. Several other excellent reviews focused on the structure and functions of ZP glycoproteins are available [Clark and Dell 2006; Gupta et al. 2007; Hedrick 2008; Shur 2008; Topfer-Petersen et al. 2008; Wassarman and Litscher 2008; Yonezawa et al. 2007]. First rate evaluations of the literature focused primarily on the induction of acrosomal exocytosis in the mouse model are also available [Buffone et al. 2008a; Florman et al. 2008].

ISOLATION OF MAMMALIAN ZP

The ZP is a specialized extracellular matrix covering the mammalian egg that is usually composed of 3–4 major glycoproteins. Unlike the egg coats found in many lower species, the ZP is generally less flexible and more difficult to penetrate, enabling it to serve as a substantial physical barrier for the protection of the oocyte. Analysis of ZP in metatherian (marsupial) species indicates that this matrix is thinner and more flexible than ZP found in eutherian species [Breed et al. 2002]. Why this transition occurred is unknown, but this shift happened at the same time that more sophisticated immune systems and the requirement for an extended gestational period developed in eutherians.

Obtaining sufficient highly purified ZP was one of the first obstacles that scientists working in this area of research had to overcome to pursue both biological and structural analyses. Large scale isolation of murine (mZP) and porcine ZP (pZP) from ovarian oocytes was initially reported over thirty years ago [Bleil and Wassarman 1978; Dunbar et al. 1978]. However, the unequivocal separation of ZP glycoproteins by gel electrophoresis was initially achieved in the mouse, revealing the presence of three major glycoproteins designated mZP1, mZP2, and mZP3 [Bleil and Wassarman 1980a]. The pZP glycoproteins resisted such simple separations, because of the heterogeneity in mass caused by the presence of polylactosamine sequences. This overlap was eliminated without compromising biological activity following the digestion of pZP with endo-β-galactosidase [Yurewicz et al. 1987], thus enabling the unambiguous separation of purified pZP glycoproteins that are now designated ZPA, ZPB, and ZPC [Harris et al. 1994; Hedrick 2008].

The bovine is also another useful mammalian model for investigating gamete binding. Definitive isolation of bovine ZP (bZP) was first reported in 1988, and subsequent analysis revealed that this matrix is also composed of three major glycoproteins [Florman and First 1988]. The bZP glycoproteins were designated ZP2, ZP3, and ZP4 that also correspond to ZPA, ZPB, and ZPC, respectively [Harris et al. 1994; Noguchi et al. 1994]. However, the amount of bZP that can be obtained is about ten fold less than pZP using the same amount of effort and ovaries, making it a more difficult model to study than the pig [Noguchi et al. 1994].

The advent of human in vitro fertilization in the early 1980s facilitated unprecedented access to human oocytes and therefore human ZP (huZP). Nonviable human oocytes are obtained under informed consent either after they have failed to fertilize during in vitro fertilization procedures or from ovaries removed from cadavers [Burkman et al. 1988; Overstreet and Hembree 1976]. The amount of huZP that can be obtained from this very limited number of oocytes is far less than what can be isolated from other mammalian models. Nonetheless, a proteomic study has revealed that the huZP consists of four distinct glycoproteins designated ZP1, ZP2, ZP3, and ZP4 [Lefievre et al. 2004]. Very small amounts of each of these precious glycoproteins was recently immunopurified to homogeneity for functional studies [Chiu et al. 2008a].

BIOLOGICAL ACTIVITIES OF MAMMALIAN ZP GLYCOPROTEINS

Once isolated, solubilized ZP preparations or individual ZP glycoproteins are usually tested to determine if they can mediate specific biological effects in vitro. The major functions that were usually analyzed include: (i) the ability of heat or acid solubilized ZP or purified ZP glycoproteins to block sperm-oocyte or sperm-ZP binding in an in vitro competitive binding assay; and (ii) the capacity of such fractions to induce acrosomal exocytosis in mammalian spermatozoa.

In the mouse, acid solubilized mZP inhibits sperm-oocyte binding by about 85% at a final concentration of 15 μg/ml [Bleil and Wassarman 1980b]. Maximal induction of acrosomal exocytosis (equivalent to stimulation with the calcium ionophore A23187) was achieved in the presence of about 8 μg/ml of solubilized mZP. Purified mZP3 is the component glycoprotein of the mZP that inhibits murine sperm-oocyte binding and induces acrosomal exocytosis at relatively low concentrations (4–8 μg/ml) [Bleil and Wassarman 1983; Florman et al. 1984].

Heat solubilized pZP maximally inhibits the binding of boar spermatozoa to oocytes by 54% at a final protein concentration of 50 μg/ml [Berger et al. 1989a]. Sperm-oocyte binding in the pig is mediated by a heterocomplex of ZPB and ZPC that are orthologues of mZP1 and mZP3, respectively [Yurewicz et al. 1998]. By contrast, acid solubilized pZP is a relatively poor inducer of acrosomal exocytosis. The maximal proportion of acrosome reacted spermatozoa increased from 6% in controls to only 18% in the presence of 125 μg/ml of this matrix [Berger et al. 1989b]. The exact pZP glycoprotein(s) that trigger acrosomal exocytosis in pig spermatozoa remains to be determined.

The binding of bull spermatozoa to bovine eggs was decreased by 60% in the presence of 50 μg/ml of bZP and by 85% at 300 μg/ml [Amari et al. 2001]. Similar to the porcine model, a heterocomplex of ZPB and ZPC is required to mediate the binding of bull spermatozoa [Kanai et al. 2007]. Solubilized bZP is also a much more highly effective stimulator of acrosomal exocytosis compared to pZP, inducing maximal response at a final concentration of 25 μg/ml [Florman and First 1988]. However, the exact bZP glycoprotein or glycoproteins that are required to manifest this activity have not been identified.

Solubilized huZP is a very potent inhibitor of human spermatozoa-ZP binding in the hemizona assay, blocking binding by 78% at a final concentration of about 5 μg/ml [Franken et al. 1996]. The huZP is also a very potent stimulator of acrosomal exocytosis, increasing the percentage of acrosome reacted spermatozoa from 8% in controls to 44% in the presence of 10 μg/ml of this matrix [Chiu et al. 2008b]. Individually, purified huZP3 and huZP4 maximally increase the percentage of acrosome-reacted spermatozoa up to 30–40% at a final concentration of only 25–50 nM. However, when these two glycoproteins are mixed together at the same concentration, the percentage of acrosome reacted spermatozoa is increased to about 50%, thus recapitulating the inductive effect of solubilized huZP [Chiu et al. 2008b]. In summary, huZP proteins are very active mediators of these key biological activities in the human.

STRUCTURAL ANALYSIS OF MAMMALIAN ZP GLYCOPROTEINS

An understanding of molecular structure is necessary to determine how specific biological activities of ZP glycoproteins are manifested. Therefore the focus of many studies in this area of research has been to determine the protein sequence and post-translational modifications (PTMs) in the mouse, pig, bovine, and human zonae.

The mZP is composed of three major glycoproteins designated mZP1, mZP2, and mZP3 that have average molecular weights of 200 kDa (dimer), 120 kDa, and 83 kDa, respectively [Bleil and Wassarman 1980a]. The protein sequence of all three mZP glycoproteins was originally determined by genomic cloning methods [Chamberlin and Dean 1989; Epifano et al. 1995; Kinloch et al. 1988; Liang et al. 1990]. mZP1 apparently acts as an organizing molecule in the mZP [Rankin et al. 1999]. mZP2 has been proposed to mediate the secondary binding between acrosome-reacted spermatozoa and the mZP [Bleil et al. 1988]. mZP3 is the primary sperm receptor molecule and is responsible for the induction of acrosomal exocytosis [Wassarman 1990]. An extensive review of this molecule and the relationship of its structure to function is available for the early studies on this glycoprotein [Wassarman 1990]. Sequencing of both the N- and O-glycans linked to mZP3 has more recently been performed [Dell et al. 2003; Easton et al. 2000; Noguchi and Nakano 1993]. A reliable and highly detailed proteomic map of both mZP2 and mZP3 is also available [Boja et al. 2003]. The crystal structure of the N-domain of ZP3 has recently been reported [Monne et al. 2008]. Therefore mZP3 is now by far the best structurally characterized mammalian ZP glycoprotein.

The amino acid sequence of the pZP glycoproteins (ZPA, ZPB, and ZPC) were also determined by cDNA cloning and sequencing analysis [Harris et al. 1994; Taya et al. 1995; Yurewicz et al. 1993]. Several groups have analyzed the N- and O-glycans linked to pZP glycoproteins [Hirano et al. 1993; Hokke et al. 1993; 1994; Mori et al. 1998; Mori et al. 1991; Nakano et al. 1996, ; Noguchi et al. 1992; Noguchi and Nakano 1992; von Witzendorff et al. 2005]. Therefore pZP glycoproteins are also very well characterized.

The bZP consists of three major glycoproteins designated ZPA, ZPB, and ZPC [Florman and First 1988; Noguchi et al. 1994]. The amino acid sequences of these glycoproteins were determined by expression cloning [Yonezawa et al. 2001]. The sequences are 77%, 85%, and 75% identical to the corresponding ZPA, ZPB, and ZPC glycoproteins in pZP. The N-glycans associated with the bZP glycoproteins have also been sequenced [Katsumata et al. 1996].

Like rat ZP, human ZP is somewhat unusual in that it is composed of four different glycoproteins designated huZP1, huZP2, huZP3, and huZPB (huZP4) with average masses of 100, 75, 55, and 65 kDa, respectively [Bauskin et al. 1999; Hoodbhoy et al. 2005; Lefievre et al. 2004]. The protein sequence of the human glycoproteins has also been determined by expression cloning [Chamberlin and Dean 1990; Hughes and Barratt 1999; Liang and Dean 1993]. Unlike all the other models, the carbohydrate sequences linked to the huZP have not yet been identified due to a lack of available huZP preparations.

THE ROLE OF GLYCOSYLATION IN THE SPECIES RESTRICTED BINDING OF MAMMALIAN SPERMATOZOA TO OOCYTES

There are several lines of evidence indicating that the binding of mammalian spermatozoa to oocytes is primarily mediated via the interaction between lectin-like egg binding proteins (EBPs) on the plasma membrane of spermatozoa and carbohydrate sequences linked to ZP glycoproteins. This emphasis was initially based on observations made in many invertebrate species indicating that carbohydrate recognition plays a major role in binding [Mengerink and Vacquier 2001]. As a consequence of this existing precedent, the potential role of glycosylation has been intensely studied in many mammalian models.

In the mouse, structural analyses revealed substantial diversity of different carbohydrate sequences. High mannose and biantennary, triantennary, and tetraantennary complex type N-glycans were detected (Fig. 1) [Easton et al. 2000; Noguchi and Nakano 1993]. These complex type N-glycans are terminated with the following antennae: Galβ1–4GlcNAc (lacNAc), NeuAcα2–3Galβ1–4GlcNAc, NeuGcα2–3Galβ1–4GlcNAc, the Sd^a antigen (NeuAcα2–3[GalNAcβ1–4]Gal, NeuGcα2–3[GalNAcβ1–4]Gal), or terminal GlcNAc. Linear polylactosamine type sequences bearing these same terminal ends are also found on some antennae. The majority of the O-glycans linked to mZP3 are core 2 type sequences terminated with sialic acid, lacNAc, GalNAcβ1–4GlcNAc (lacdiNAc), Galα1–3Galβ1–4GlcNAc, and the Sd^a antigen [Dell et al. 2003].

Figure 1  Complex type N-glycan cores present in mammalian ZP. The difference between the two triantennary sequences is the presence of either β1–4 linked GlcNAc (a); or β1–6 linked GlcNAc (b) at the positions indicated.

The two predominant models for murine gamete adhesion proposed that oligosaccharides bearing terminal β-linked GlcNAc or Galα1–3Gal sequences mediate initial binding [Bleil and Wassarman 1988; Lopez et al. 1985]. However, results obtained in other studies confirmed that this interaction could also be inhibited by oligosaccharides terminated with lacNAc or Lewis^x (Galβ1–4[Fucα1–3]GlcNAc) sequences, but only at relatively high concentrations of ligand [Johnston et al. 1998; Kerr et al. 2004]. A notable exception to these results involved an investigation performed with a panel of artificial oligosaccharides (Fig. 2, constructs S1-S5).

Figure 2  Glycans referred to in the text. The artificial oligosaccharide constructs (S1–S5) employed in the competitive murine gamete binding assays are shown [Litscher et al. 1995]. One of the highly branched polylactosamine type N-glycans found in rabbit erythrocytes is also displayed (S6). In this cell type, branched polylactosamine type N-glycans ranging in size up to 9 kDa were detected [Sutton-Smith et al. 2007]. The dashed boxes enclose the terminal branched polylactosamine sequences that are common to rabbit erythrocytes and constructs S1 and S2.

As noted earlier, the carbohydrate sequences in the porcine ZP have also been analyzed using a combination of different biophysical methods. Unlike the mouse, no high mannose type structures were detected in pZP N-glycans. The same complex type N-glycans shown in Figure 1 are also present, except that 39% of the neutral subset lacked one or more terminal β1–4 linked Gal residues [Mori et al. 1991; Noguchi et al. 1992; Noguchi and Nakano 1992]. In acidic chains, sulfate groups were linked to: (i) the C-6 position of GlcNAc located in a terminal H type 2 or Lewis^x sequence; (ii) the C-3 position of terminal GlcNAc; or (iii) the C-6 position of GlcNAc residues located within polylactosamine extensions of N-glycans [Mori et al. 1998]. The neutral O-glycans are primarily core 1 type with small amounts of core 3 structures [Hokke et al. 1994]. These sequences are usually terminated with either β-linked Gal or GlcNAc. The acidic O-glycans are primarily core 1 structures bearing extended polylactosamine structure with sulfate groups located at the C-6 position of GlcNAc and sialic acid in α2–3 position on some of the oligosaccharides [Hokke et al. 1994].

Though a heterocomplex of native ZPB and ZPC is required to inhibit porcine spermatozoa binding to oocytes, the receptor activity is specifically associated with ZPB following endo-β-galactosidase digestion that enables its separation from ZPC [Sacco et al. 1989]. This glycosidase treated glycoprotein reduced binding by 85% at a concentration of 20 μg/ml [Sacco et al. 1989]. An early study indicated that highly sulfated O-glycans but not N-glycans associated with the pZP mediated initial gamete binding [Yurewicz et al. 1991]. However, more recent studies indicate that tri-and tetraantennary N-glycans terminated with β1–4 linked Gal are the carbohydrate ligands that are recognized during this interaction [Kudo et al. 1998; Yonezawa et al. 2005].

The role of carbohydrate recognition has also been investigated in the bovine model. The N-glycans associated with the bZP have also been very well characterized [Katsumata et al. 1996]. In all, 23% of the oligosaccharides are neutral, with the remainder representing sialylated glycans. Remarkably, only a single neutral N-glycan is detected within this matrix, a high mannose type N-glycan designated Man₅GlcNAc₂. This sequence is the only one that that inhibits boar spermatozoa binding to bZP [Amari et al. 2001]. The acidic chains consist of biantennary, triantennary, and tetraantennary complex type N-glycans with some antennae extended with linear polylactosamine type sequences (Fig. 1).

As noted previously, no studies have been performed on the expression of N- or O-glycans in the human. The only carbohydrate that inhibits the initial binding of human spermatozoa to huZP in the human hemizona assay is fucoidan, an algal fucan sulfate polysaccharide [Oehninger et al. 1990; Patankar et al. 1993]. Glycodelin isoforms obtained from amniotic fluid (GdA) and from follicular fluid (GdF) are also potent inhibitors of the binding of human spermatozoa in the hemizona assay system [Chiu et al. 2002; Oehninger et al. 1995; Yao et al. 1998]. By contrast, a differentially glycosylated form of glycodelin expressed in the seminal plasma (GdS) promotes binding in the same assay system and inhibits the capacitation of human spermatozoa [Chiu et al. 2005; Morris et al. 1996].

RABBIT ERYTHROCYTE: AN UNCONVENTIONAL SOMATIC CELL MODEL FOR STUDYING SPERM-ZP BINDING

Strong evidence that initial murine gamete binding is carbohydrate dependent was also obtained in a very unusual cell adhesion model. This model was first developed in 1993, when one of the prevailing paradigms for binding was that murine sperm recognize terminal Galα1–3Gal sequences [Bleil and Wassarman 1988]. Rabbit erythrocytes were then known to express a series of highly branched polylactosamine type glycosphingolipids ranging in size from 5–40 monosaccharides terminated with up to eight Galα1–3Gal sequences [Hanfland et al. 1988]. Therefore murine spermatozoa were incubated with rabbit erythrocytes under optimal conditions for gamete binding based on this structural relationship. Remarkably murine spermatozoa bind with exceptional affinity to these somatic cells, but do not undergo acrosomal exocytosis [Clark et al. 1996; Sandow and Clark 1993]. This unusual interaction was observed in a previous study, but had not been very well characterized [Yamagata 1985]. Exhaustive digestion with highly purified α-galactosidase did not diminish this activity, indicating that terminal Galα1–3Gal sequences were not obligatory for binding [Sutton-Smith et al. 2007]. However, mild periodate oxidation of the erythrocytes that attacks vicinal hydroxyl groups on oligosaccharides completely eliminated this interaction, indicating that this binding was solely dependent on the recognition of carbohydrate sequences [Sandow and Clark 1993; Sutton-Smith et al. 2007]. However, neither small glycopeptides obtained after exhaustive pronase digestion or glycosphingolipid derived-oligosaccharides released by ozonolysis from this cell type inhibited binding, even at very high concentrations [Clark unpublished data]. This unusual model was referred to as the sperm-somatic cell adhesion system.

The probable reason for these puzzling results became clear from findings made in a study involving the artificial oligosaccharide constructs shown in Figure 2 [Litscher et al. 1995]. A construct bearing a single terminal branched polylactosamine sequence capped with the Galα1–3Gal disaccharide (construct S1, dashed box) had no effect on murine sperm-oocyte binding in vitro. However, an analogue of construct S1 bearing two terminal branched polylactosamine type sequences (construct S2, dashed boxes) inhibited binding by 80% at a final concentration of 4 μM. An analogue of construct S1 lacking terminal α1–3 linked Gal (construct S3) did not block binding. However, the corresponding divalent analogue of construct S3 (construct S4) inhibited binding as well as construct S2. Removal of β1–4 linked Gal from construct S4 generated a derivative with very low inhibitory activity (construct S5) [Litscher et al. 1995].

These results are consistent with the concept that the major murine EBP accommodates branched polylactosamine sequences terminated with Galα1–3Galβ1–4GlcNAc or Galβ1–4GlcNAc, but only when two branches are presented in a very close divalent arrangement (Fig. 2, dashed boxes). Analysis of both protein- and glycosphingolipid-derived glycans isolated from rabbit erythrocytes indicate that the oligosaccharides present in this cell type are terminated with only a single branched sequence (Fig. 2, S6, dashed box) [Sutton-Smith et al. 2007]. These oligosaccharides are also inactive in the in vitro competition assays. However, rabbit erythrocytes express very large amounts of these glycans, likely enabling the EBP on mouse spermatozoa to bind via multivalent presentation of these carbohydrate sequences expressed on the surface of these somatic cells.

Biophysical analyses confirmed that branched polylactosamine sequences are not associated with the mZP [Dell et al. 2003; Easton et al. 2000; Noguchi and Nakano 1993]. The closest analogue to bioactive construct S4 in the mZP is a tetraantennary N-glycan sequence bearing four terminal lacNAc sequences (Fig. 1). However, N-glycans released from mZP also do not inhibit murine spermatozoa-oocyte binding in vitro [Litscher and Wassarman 1996]. These results are consistent with the concept that an appropriate steric presentation of such oligosaccharide ligands must occur to enable mZP3 to bind murine spermatozoa in vivo.

It is very intriguing that tri- and tetraantennary N-glycans have now been proposed to mediate the initial binding of gametes in both the porcine and murine models [Clark and Dell 2006; Kudo et al. 1998]. Additional data has recently been obtained in a related somatic cell adhesion system that supports this very interesting overlap. In preliminary studies, boar spermatozoa were also recently shown to bind with high affinity to rabbit erythrocytes, exactly like murine spermatozoa do when incubated with these somatic cells [Clark et al. 2009]. This result implies that the lectin-like EBP in the boar spermatozoa can also recognize branched polylactosamine sequences when presented in a polyvalent array (Fig. 2, S6). However, unlike the murine interaction, boar spermatozoa also undergo acrosomal exocytosis immediately after binding. This novel artificial adhesion system is currently under investigation [Clark et al. 2009].

MAMMALIAN EBPS

The most significant problem in the murine model is that in spite of all of the accumulated knowledge about mZP3, the EBP, or EBPs that interact with this glycoprotein to mediate adhesion remain highly controversial. A specific EBP designated sp56 associated with murine sperm was hypothesized to bind to vicinal O-glycans positioned at Ser-332 and Ser-334 [Chen et al. 1998]. However, sp56 was later localized to the acrosome [Foster et al. 1997], and Ser-332 and Ser-334 are not O-glycosylated [Boja et al. 2003]. However, sp56 has recently been proposed to be exposed on the plasma membrane of mouse spermatozoa during capacitation, forming a higher order aggregate that mediates initial gamete binding in the mouse [Buffone et al. 2008b].

Another major model indicates that a specific β1–4 galactosyltransferase expressed on the murine sperm plasma membrane mediates binding by interacting with β-linked GlcNAc residues at the terminal ends of O-glycans attached to mZP3 [Lopez et al. 1985]. However, mZP glycans are rarely terminated with β-linked GlcNAc sequences [Easton et al. 2000]. Lectin binding studies of freshly ovulated mouse oocytes are consistent with these structural analyses [Mori et al. 1997].

There are several other EBPs that have been proposed to mediate binding in the mouse, including SED-1, ZPB1, ZPB2, or proacrosin, but the deletion of genes encoding these proteins does not result in a complete loss of murine gamete binding [Baba et al. 1994; Ensslin and Shur 2003; Lin et al. 2007]. Rather, the subfertility or infertility observed in these studies results from defects in acrosome biogenesis, sperm morphogenesis, or dysregulation of acrosomal exocytosis [Buffone et al. 2008]. The hypothesis that a single major murine EBP is responsible for initial binding has now been supplanted by the concept that multiple recognition molecules are assembled into a functional complex during capacitation that mediates this interaction [Nixon et al. 2005]. Recent proteomic analysis in mouse spermatozoa confirms that many of the prominent putative EBPs (β1–4 galactosyltransferase, SED-1, PH-20, sp56) are localized in the detergent resistant membranes known as lipid rafts [Nixon et al. 2009]. The β-galactosyltransferase was proposed to mediate initial gamete binding in the mouse by interacting with O-linked glycans expressed on mZP3 [Miller et al. 1992]. None of the other potential murine EBPs has been shown to have any affinity for the oligosaccharides linked to the mZP.

In the case of the porcine model, several solubilized sperm membrane proteins have been shown to interact with pZP ghosts, including AQN-3, SED-1 (P47), fertilin-β and peroxiredoxin-5 [van Gestel et al. 2007]. Other potential porcine EBPs have also been isolated by affinity approaches [Ensslin et al. 1995; van Gestel et al. 2007]. AQN-3 is an appealing EBP because it displays high affinity binding to tri- and tetraantennary N-glycans previously implicated in binding [Calvete et al. 1996; Ensslin et al. 1995; Kudo et al. 1998]. Based on this evidence, AQN-3 has been proposed to be the major EBP mediating this initial interaction in the porcine model [Topfer-Petersen et al. 2008]. Another protein designated IAM38 has been implicated in the secondary binding interaction involving the inner acrosomal membrane [Yu et al. 2006]. In summary, there is far less controversy in the pig model when compared to the mouse.

Currently no EBP has been proposed in the bovine, perhaps due to the limited amount of available glycoproteins. However, studies in the human gametes have generated a candidate for this function. An unusual human protein designated glycodelin-A was previously shown to potently inhibit human sperm-ZP binding in the HZA [Oehninger et al. 1995]. The N-glycans associated with this glycoprotein were sequenced [Dell et al. 1995]. Unlike conventional human glycoproteins, glycodelin-A bears highly unusual lacdiNAc type sequences. Human spermatozoa express large amounts of a sugar modifying enzyme known as a fucosyltransferase (FUT5) on their surface. This glycosyltransferase binds with very high affinity to glycodelin-A and has therefore been proposed to act as the major EBP that mediates the firm adhesion of human sperm to the ZP [Chiu et al. 2007]. FUT5 could mediate the binding of human spermatozoa to the ZP by interacting with unusual lacdiNAc type sequences presented on this matrix.

GENETIC MANIPULATION OF ZP GLYCOPROTEINS AND GLYCOSYLTRANSFERASES

The primary mammalian model for in vivo genetic studies is the mouse. The initial focus was on manipulating the expression of the major glycoproteins present in mZP to determine their effects on the synthesis, structure, and function of this matrix. Targeted disruption of either mZP2 or mZP3 genes results in complete infertility in female mice [Liu et al. 1996; Rankin et al. 1996]. Oocytes isolated from null mice completely lack an mZP. Heterozygotes with one functional allele for mZP3 produce mZP that are half as thick as native mZP, indicating that mZP3 is limiting during the production of this matrix [Wassarman et al. 1997]. Targeted deletion of the mZP1 gene in female mice generates oocytes with a loosely organized mZP composed of only mZP2 and mZP3 [Rankin et al. 1999]. These female mZP1 null mice are far less fertile than wild type mice.

Formidable molecular biological approaches have also been employed to investigate the species-specificity of murine gamete binding. Human spermatozoa do not bind to murine eggs [Bedford 1977]. Dean and his colleagues proposed that by using transgenesis to substitute huZP3 for mZP3 in female mice, the oocytes derived from such animals should be able to bind human spermatozoa. However, human sperm did not bind to oocytes isolated from huZP3 rescue mice, but murine sperm were readily bound [Rankin et al. 1998]. Therefore, the taxon specificity of binding was preserved even in the absence of mZP glycoproteins. Similar results were observed in mice expressing huZP2 or both huZP2 and huZP3 simultaneously in their oocytes [Rankin et al. 2003]. These findings are consistent with the model that murine gamete binding is primarily dependent on the recognition of carbohydrate ligands but not the protein backbone.

Other types of in vivo genetic manipulations have also been performed in mice. Another putative EBP is a specific β1–4 galactosyltransferase. Male mice lacking this specific enzyme were also created [Lu and Shur 1997]. However, murine eggs from wild type mice bind 3–4 times more mutant spermatozoa lacking this galactosyltransferase than spermatozoa from wild type mice. These mutant male mice are also completely fertile, though their spermatozoa are defective in their ability to undergo mZP3 induced acrosomal exocytosis in vitro [Lu and Shur 1997].

Gene deletion studies targeting glycosyltransferases have also been employed to determine the functional role of glycosylation in mediating murine spermatozoa-oocyte binding. In one model, vicinal O-glycans positioned at Ser-332 and Ser-334 of mZP3 terminated with Galα1–3Gal sequences expressed on mZP3 were proposed to be the receptor site for the binding of spermatozoa [Bleil and Wassarman 1988; Chen et al. 1998]. However, female mice deficient in terminal Galα1–3Gal sequences retain their fertility and sperm binding capacity [Liu et al. 1997; Thall et al. 1995]. Oocyte specific deletion of the β1–3 galactosyltransferase designated T-syn necessary for core 1/core 2 O-glycan synthesis in fact results in increased fertility, indicating that O-glycans are not essential for binding or may in fact be negative regulators of sperm ZP interactions [Williams and Stanley 2008]. Therefore the proposal that O-glycans mediate binding in the murine model is unsupported by genetic manipulation studies.

Conditional deletion of the glycosyltransferases involved in N-glycan synthesis in female mice has generated far more substantial biological effects in mice. Inactivation of N-acetylglucosaminyltransferase V, an enzyme essential for the synthesis of triantennary and tetraantennary N-glycans, yields mutant mice that display about 50% of the fecundity of control mice [Clark and Dell 2006]. Conditional deletion of N-acetylglucosaminyltransferase I (GnT I), an enzyme essential for complex type N-glycan synthesis in oocytes, reduces their spermatozoa binding capacity by 81% compared to ova from wild type mice [Hoodbhoy et al. 2005] and their fertility by about 50% [Shi et al. 2004]. However, the projected lifetime fecundity of female mice that are conditionally deficient in both GnT I and T-syn in their oocytes is reduced by more than 97% compared to controls, indicating an absolute necessity for complex type glycosylation to support viable fertility in this species [Williams et al. 2007].

Recombinant ZP glycoproteins and proteins have also been expressed in cell lines and bacteria, respectively, for subsequent testing in in vitro systems. The value of this approach was initially demonstrated in F9 embryonal carcinoma cells. Recombinant mZP3 synthesized in this cell line is functionally equivalent to native mZP3 [Kinloch et al. 1991]. Further analysis employing an exon swapping strategy indicated that a specific region of exon 7 of the mZP3 gene between Ser-329 and Ser-334 (SNSSSS) is required to inhibit murine sperm-egg binding in vitro. The conversion of either Ser-332 or Ser-334 (SNSSSS) to Ala results in the synthesis of mutant recombinant mZP3 that lacks the capacity to inhibit murine sperm-ZP binding [Chen et al. 1998]. Therefore vicinal O-glycans at Ser-332 and Ser-334 were proposed to represent the carbohydrate ligands essential for binding. However, as outlined previously, subsequent proteomic analysis of mZP3 indicated that Ser-332 and Ser-334 are not O-glycosylated [Boja et al. 2003]. In addition, mutant mice expressing ZP glycoproteins deficient in core1/core 2 O-glycan synthesis due to the deletion of T-syn actually display enhanced fertility [Williams and Stanley 2008].

Nakano and colleagues expressed pZP glycoproteins in Sf9 insect baculovirus cells, generating recombinant glycoproteins (rZPA, rZPB, and rZPC) [Yonezawa et al. 2005]. Boar spermatozoa did not bind to any of the beads coated with recombinant pZP glycoproteins either singly or in combinations. However, bovine spermatozoa were bound to beads coated with rZPB and rZPB/rZPC, but not to beads coated with rZPA or rZPC. Structural analyses indicate these glycoproteins expressed pauci and high mannose type chains [Yonezawa et al. 2005]. In another study, a heterocomplex of bovine recombinant ZPB and rZPC coexpressed in Sf9 cells exhibited inhibitory activity for bovine spermatozoa-ZP binding similar to that of a native bovine ZPG mixture, whereas neither bovine rZPB nor rZPC alone inhibited binding [Kanai et al. 2007]. Therefore the ability of the recombinant pZP/bZP glycoproteins to interact with bovine spermatozoa is dependent on the expression of high mannose type N-glycans rather than the protein backbone.

Recombinant huZP3 has been produced in CHO cells and E. coli that induces acrosomal exocytosis in human spermatozoa [Chapman et al. 1998; van Duin et al. 1994]. Recently, Gupta and coworkers demonstrated that soluble recombinant huZP3 and huZP4 synthesized in both E. coli and insect baculovirus systems bind to human spermatozoa in vitro, though they have distinct binding sites [Chakravarty et al. 2008]. However, the glycosylated forms produced in the insect baculovirus cells induced acrosomal exocytosis, but the nonglycosylated bacterial forms did not [Chakravarty et al. 2005]. These investigators have localized both human spermatozoa binding activity to the C-terminal region of huZP3 by making peptide constructs from different regions of this glycoprotein in insect baculovirus and bacterial systems [Bansal et al. 2009]. The glycosylated C-terminal region peptides (aa 214–305 and aa 214–348) produced in the baculovirus system could both bind to human spermatozoa and induce acrosomal exocytosis. However, the same huZP3 peptides made in E. coli bind to human spermatozoa, but cannot induce acrosomal exocytosis. The results of these studies imply that glycosylation of huZP3 and huZP4 is not required for the binding of these glycoproteins to human spermatozoa, but is essential to induce acrosomal exocytosis.

THE DOMAIN SPECIFIC MODEL FOR MAMMALIAN SPERM-OOCYTE BINDING

As noted earlier, the oocyte specific loss of core 1/core 2 O-glycans increased fertility in mutant female mice [Williams and Stanley 2008]. These findings clearly indicate that the O-glycans associated with mZP3 are not essential for fertilization. By contrast, the conditional oocyte specific inactivation of GnT I leading to the loss of complex type N-glycosylation in mZP is accompanied by an 81% decrease in murine gamete binding and a 50% loss in fertility [Hoodbhoy et al. 2005; Shi et al. 2004]. Clearly, these results support the concept that N-glycans play a crucial role in the binding of spermatozoa in the murine model [Clark and Dell 2006].

Other direct observations made in the human and murine model are seemingly in direct conflict with the hypothesis that N-glycans are the major ligands recognized during binding. Nonglycosylated recombinant huZP3 and huZP4 but not huZP2 synthesized in E. coli bind to human spermatozoa [Chakravarty et al. 2008]. The digestion of mZP3 with papain or V8 protease generates a 55 kDa C-terminal glycopeptide fragment (gp55) that blocks mouse sperm-oocyte binding and induces acrosomal exocytosis as well as intact mZP3 [Litscher and Wassarman 1996]. Digestion of this 55 kDa glycopeptide with peptide N-glycosidase F removes its N-glycans, generating a much smaller 21 kDa product that retains its ability to inhibit murine sperm-oocyte binding in vitro. The quite logical conclusion that the investigators reached from this experiment is that O-glycans mediate binding [Litscher and Wassarman 1996]. However, this conclusion is not supported by a more recent study demonstrating that the inactivation of core1/core 2 O-glycan synthesis increases fertility [Williams and Stanley 2008].

Data obtained in another recent study also calls into question the role of N-glycans in mediating murine gamete binding. Li and coworkers synthesized three small recombinant proteins in E. coli comprising different domains of mZP3, designating them P1, P2, and P3 [Li et al. 2007]. These investigators were able to completely renature these small recombinant proteins lacking glycosylation, and verified this accomplishment by analyzing their products using both circular dichroism and mass spectrometric methods. However only P3 comprising the C-terminal domain of mZP3 (aa185–354) displayed any inhibitory activity, blocking murine gamete binding by 60% at a final concentration of 2 μM [Li et al. 2007]. However, these authors reported that P3 does not induce acrosomal exocytosis.

Therefore a crucial question arises from these observations: how can N-glycans be essential for initial sperm-ZP binding when ZP derivatives lacking these oligosaccharides inhibit the binding of murine spermatozoa to mouse oocytes? Similarly, how can recombinant forms of huZP3 and huZP4 also bind to human spermatozoa? Though seemingly incompatible, these observations can be reconciled. Glycosylation confers many advantages on a protein, including the stabilization of its structure, resistance to proteolysis, and optimization of folding [Varki 1993]. Such properties are certainly under positive selection, since the scope and diversity of glycosylation increases in tandem with organismal complexity. However, the major disadvantage of glycans is that they have substantial spheres of hydration, and thus impede access to the peptide backbone. It is evident that the domain of the mZP3 that mediates binding is located at its C-terminus, not its N-terminus, based on the ability of the gp55 fragment to both inhibit initial gamete binding and induce acrosomal exocytosis. It is also quite clear that gp55 is very heavily glycosylated, because 60% of its mass can be removed by N-glycanase digestion [Litscher and Wassarman 1996]. Therefore the logical conclusion is that the majority of the amino acid sequence in this ZP3 domain is sterically hindered by carbohydrate chains when presented on the surface of the oocyte. It is only reasonable to conclude that the carbohydrate sequences attached to mZP3 must have been selected to act as the major ligands that mediate binding based on the available evidence. Once binding is achieved and acrosomal exocytosis occurs, the acrosomal glycosidases released during this reaction could remove monosaccharides from ZP glycoproteins near the binding region thus exposing sites for the secondary binding interaction.

Therefore it is extremely likely that an EBP (or several EBPs forming an egg binding complex) surveys this entire C-terminal domain during initial binding, including the carbohydrate sequences and accessible amino acid sequences. Results obtained in both the mouse and human models indicate that about 80% of the binding sites for spermatozoa are carbohydrate dependent, whereas the remaining 20% are due to protein-protein interactions [Clark and Dell 2006; Ozgur et al. 1998]. This synergy could exist because glycosylation is a stochastic process where even preferred glycosylation sites are not always occupied (Fig. 3). If precise glycosylation was always necessary to mediate binding, then the absence of a key carbohydrate ligand would completely inactivate a subset of mZP3 molecules. However, in this domain specific model, binding could still occur even if a carbohydrate chain was missing, because the egg binding complex can also bind to amino acid sequences in the domain via protein-protein interactions that would become more accessible when N-glycans are absent (Fig. 3). This type of synergism could enable virtually all mZP3 molecules to participate in binding, though the affinity of each molecule would depend on its glycosylation state. Binding analyses indicate that both high and low affinity binding sites for mZP3 exist on mouse spermatozoa, consistent with this hypothesis [Thaler and Cardullo 1996]. This type of synergistic binding would certainly be under positive selection, because individuals that could engage both protein-carbohydrate and protein-protein interactions to bind spermatozoa would likely be more fertile than those that could not. It is evident that even a 1% edge in fertility over many generations could confer a major benefit in the ecological setting and thus a major evolutionary advantage.

Figure 3  The domain specific model for mammalian gamete binding. In the majority of cases, there is access to the key carbohydrate ligand recognized by the EBP (shaded green). Other glycans (shaded blue) that are not ligands for the EBP are nearby, and may also partially obscure regions that can mediate binding via protein-protein interactions (shaded red). In scenario 1, all glycans are in place, and binding is mediated primarily by protein-carbohydrate interactions. In scenario 2, the carbohydrate ligand recognized by the EBP is absent enabling more access to the sites for protein-protein interactions (arrows). Carbohydrate dependent binding is minimal. In scenario 3, one of the other nonbinding glycans is absent, providing more access for protein-protein interactions in addition to the protein-carbohydrate interaction (arrow). In scenario 4, all the glycans are absent. This situation is completely unnatural in mammals. However, if the protein is correctly folded, then the sites for protein-protein interactions now become both accessible and functional, enabling binding to occur in the complete absence of any carbohydrate sequences.

This model should now be considered very carefully in the context of how complex multicellular organisms manifest their biological activities. There could be other biological systems where both oligosaccharides and accessible amino acid sequences within a specific protein domain work synergistically to mediate a very specific biological activity. The surprising inference of this system is that even if a non-glycosylated recombinant protein can mediate a biological activity associated with a mammalian glycoprotein, such a demonstration does not preclude the possibility that carbohydrate sequences could be the primary functional groups in the physiological setting because of their potential for sterically hindering access to protein domains. The increase in the scope, diversity, and complexity of post-translational modifications in higher organisms indicates that they are likely being employed to both optimize the activities of existing proteins and create new functions. Logically, this pathway rather than the generation of unique proteins could very well be the primary driving force behind the development of increased complexity in metazoans that also drives the speciation process. Therefore it is important to consider the “big picture” involving every aspect of a protein in order to understand how specific functions are manifested.

THE POTENTIAL ROLE OF THE MAMMALIAN ZP IN MEDIATING IMMUNE RECOGNITION

The possibility also exists that the mammalian ZP acts as an immune barrier for the oocyte and the developing blastocyst. Murine and human gametes lack major histocompatibility markers for “self” and thus can avoid certain types of deleterious cell mediated responses (reviewed in [Clark et al. 2006]). However, such oocytes likely require recognition signals for other immune effector cells, especially natural killer (NK) cells that target cells lacking histocompatibility markers, a concept known as “missing self” [Karre 2002]. HLA class Ia glycoproteins (HLA-A, HLA-B, and HLA-C) possess a single N-glycosylation site that projects the carbohydrate sequence well away from the binding cleft [Barber et al. 1996]. Another unusual aspect of these glycoproteins is that their N-glycans display unusual uniformity, with bisecting biantennary type N-glycans expressed on 35–92% of these molecules [Barber et al. 1996]. It is therefore significant that the expression of bisecting type N-glycans has been associated with the protection of MHC class I negative K562 erythroleukemia cells [el Ouagari et al. 1995; Yoshimura et al. 1996]. Such carbohydrate sequences are also expressed at significant levels on the surface of human oocytes and spermatozoa [Pang et al. 2007; Patankar et al. 1997].

The mammalian zygote could also become susceptible to immune destruction after fertilization when it begins to express foreign paternal histocompatibility markers. Carbohydrate structural analyses of mZP glycans have confirmed that many of the sequences expressed on this matrix are also highly elevated on many different lymphocyte populations following their activation [Clark et al. 2001]. These carbohydrate sequences could be employed in the immune system as protective markers to prevent the accidental destruction of lymphocytes during their activation (reviewed by [Clark et al. 2006]). The incorporation of such carbohydrate markers into the surface coating of eggs would certainly have facilitated immune recognition in ancient sexually reproducing metazoans. This “handshaking” could be the shielding that protects histoincompatible mammalian blastocysts prior to hatching.

There is evidence for this type of system on the murine oocyte. The β1–6-linked N-acetyllactosamine sequence attached to the core α1–6-linked mannose residue of tri- and tetraantennary N-glycans is often extended with linear polylactosamine chains [Demetriou et al. 2001]. The same types of carbohydrate modifications are associated with mZP3 based on the results of previous studies [Easton et al. 2000; Noguchi and Nakano 1993]. Galectins are postulated to bind to these polylactosamine sequences on T cells, forming lattices that block signal transduction involving T cell receptors [Demetriou et al. 2001]. A major question that arises is whether these glycans expressed on other cell types outside of the immune system provide protection from immune responses. Besides a decrease in fertility, Mgat V knockout mice develop very severe froms of autoimmune disorders that affect some but not all tissues [Clark and Dell 2006; Demetriou et al. 2001]. There is compelling evidence that extended linear polylactosamine type sequences are also expressed on pZP, bZP, and huZP, but their enzymatic removal does not adversely affect the binding of spermatozoa to these gametes [Katsumata et al. 1996; Ozgur et al. 1998; Yurewicz et al. 1987]. An interesting question for future investigation is whether such polylactosamine type sequences affect the adaptive immune response directed against ZP glycoproteins.

CONCLUSIONS

Considerable progress has been made in our understanding of mammalian gamete binding since 1989, due to the development of very powerful genetic, biochemical, and mass spectrometric methods. The murine model is the most accessible to genetic manipulation, but thus far this major advantage has not been translated into a consensus about how the binding of spermatozoa to the mZP is mediated. The greatest progress has been made in the porcine model, where an adhesion molecule (AQN-3) that recognizes native pZP and pZP-associated glycans has been identified. An inner acrosomal membrane protein (IAM38) has also been defined that mediates the secondary binding interaction that occurs after acrosomal exocytosis. Significant information has also been collected in the bovine model, though an EBP has not yet been proposed. The greatest clinical benefit will be obtained by investigating the functional roles of the human ZP. It is very likely that soon the biophysical methods will become so sensitive that the glycans linked to huZP glycoproteins will also be sequenced. Whether this knowledge will quickly lead to any useful clinical applications cannot be predicted. The oligosaccharides linked to mZP3 were also characterized in great detail six years ago. However, no consensus has developed yet about the carbohydrate ligands or EBPs that mediate the binding of murine spermatozoa to the mZP. Data indicating that both protein-carbohydrate and protein-protein interactions play a role in binding has greatly complicated the situation. How the mammalian ZP mediates immune recognition is not yet known, but is a fertile area for further inquiry. Nonetheless, it is very likely that many of the functional roles of the mammalian ZP will finally be revealed over the next two decades.

ACKNOWLEDGMENTS

Studies cited in this review that were performed by the author were supported by National Institutes of Health Grant HD35652, the Breeden-Adams Foundation, the Elsa U. Pardee Foundation, the Jeffress Trust, and the Mission Enhancement Program in Reproductive Biology and Medicine (C8783) funded by the state of Missouri.

Declaration of Interest: The authors report no conflicts of interest. The authors alone are responsible for the content and writing of the paper.