Structural analysis of the oligosaccharides derived from glycodelin, a human glycoprotein with potent immunosuppressive and contraceptive activities.

Glycodelin, also known as placental protein 14 (PP14) or progesterone-associated endometrial protein (PAEP), is a human glycoprotein with potent immunosuppressive and contraceptive activities. In this paper we report the first characterization of glycodelin-derived oligosaccharides. Using strategies based upon fast atom bombardment and electrospray mass spectrometry we have established that glycodelin is glycosylated at Asn-28 and Asn-63. The Asn-28 site carries high mannose, hybrid and complex-type structures, whereas the second site is exclusively occupied by complex-type glycans. The major non-reducing epitopes in the complex-type glycans are: Galβ1-4GlcNAc (lacNAc), GalNAcβ1-4GlcNAc (lacdiNAc), NeuAcα2-6Galβ1-4GlcNAc (sialylated lacNAc), NeuAcα2-6GalNAcβ1-4GlcNAc (sialylated lacdiNAc), Galβ1-4(Fucα1-3)GlcNAc (Lewis), and GalNAcβ1-4(Fucα1-3)GlcNAc (lacdiNAc analogue of Lewis). It is possible that the oligosaccharides bearing sialylated lacNAc or lacdiNAc antennae may manifest immunosuppressive effects by specifically blocking adhesive and activation-related events mediated by CD22, the human B cell associated receptor. Oligosaccharides with fucosylated lacdiNAc antennae have previously been shown to potently block selectin-mediated adhesions and may perform the same function in glycodelin. The potent inhibitory effect of glycodelin on initial human sperm-zona pellucida binding is consistent with our previous suggestion that this cell adhesion event requires a selectin-like adhesion process. This result also raises the possibility that a convergence between immune and gamete recognition processes may have occurred in the types of carbohydrate ligands recognized in the human.

Bohn and co-workers originally isolated a glycoprotein from human placenta that they designated placental protein 14 or PP14 1 (Bohn et al., 1982). PP14 was subsequently found to be synthesized not by the placenta but by the secretory and decidualized endometrium (Julkunen et al., 1986a(Julkunen et al., , 1988. PP14 was therefore also referred to as progesterone-associated endometrial protein or PAEP in accordance with its endometrial origin. More recent evidence indicates that PAEP is also synthesized by the hematopoietic tissues of the bone marrow (Kamarainen et al., 1994) and perhaps other tissues. Since the glycoprotein referred to as PP14 or PAEP is not of placental origin nor is it exclusively synthesized in the endometrium, previous designations may not truly reflect its diverse sites of synthesis or its function. Therefore in this paper, we have designated PP14/PAEP as "glycodelin" to eliminate confusion over these issues and to emphasize the unique nature of oligosaccharides in this glycoprotein.
The temporal and spatial expression of glycodelin in the reproductive organs of the human female is highly regulated. During the menstrual cycle, glycodelin is not expressed in the proliferative endometrium but increases significantly from the fourth postovulatory day, peaking around the 12th day (Julkunen et al., 1986a). Thus glycodelin expression is at a minimum during the peri-ovulatory period of the cycle. However, at the time of implantation of the embryo, glycodelin synthesis in the decidua is induced to very high levels (4 -10% of total protein) (Julkunen et al., 1985). Glycodelin is also secreted into the amniotic fluid in substantial amounts, reaching a peak in the 10th to 14th week of gestation (Julkunen et al., 1985). In addition, glycodelin is also found in the serum under normal conditions and during pregnancy, although at a much lower level than in amniotic fluid or decidual cells (Julkunen et al., 1986b).
Glycodelin manifests several significant biological activities when tested in immunological assay systems. Crude decidual extracts containing this glycoprotein were initially shown to suppress thymidine uptake in both normal and mitogen-stimulated human mixed lymphocyte culture (Bolton et al., 1987;Pockley et al., 1988). Decidual extracts containing glycodelin also decreased the synthesis of cytokines (interleukin-1 and interleukin-2) and interleukin-2 receptors by mitogen-stimulated cells Bolton, 1989, 1990). Purified glycodelin also suppresses the lysis of K562 cells by human natural killer cells at low concentrations (Okamoto et al., 1991). Therefore, glycodelin may be one of several factors that induce highly regiospecific immunosuppression of the maternal response to the human embryo/fetus. Another significant biological activity of glycodelin is its ability to inhibit human sperm-zona pellucida binding in the hemizona assay system (Oehninger et al., 1995). To date, glycodelin is the most potent glycoprotein inhibitor of human sperm-zona pellucida binding in this assay system.
We have recently proposed that human sperm-zona pellucida binding requires a selectin-like interaction between human sperm and human zona pellucida (Patankar et al., 1993a(Patankar et al., , 1993b. Therefore, we hypothesized that glycodelin probably manifested its immunosuppressive and contraceptive activities via its oligosaccharide chains. Previous work has indicated that this glycoprotein contains 17.5% carbohydrate by weight (Bohn et al., 1982). However, no oligosaccharide structures have been reported. In this study, we have performed structural analysis of glycodelin-derived N-linked oligosaccharides and glycopeptides using mapping strategies , Dell et al., 1983 based upon fast atom bombardment (FAB) and electrospray (ES) mass spectrometry. Glycodelin has three consensus sites for N-linked glycosylation (Julkunen et al., 1988) (Fig. 1), and we have shown that the first two of these sites are glycosylated with defined and substantially different heterogeneous populations of glycans. Many of these glycans have antennae composed of sialylated or fucosylated GalNAc␤1-4GlcNAc(lacdiNAc) sequences, which are rare in higher animals. N-Linked oligosaccharides of this type have been shown to be potent inhibitors of selectin-mediated adhesions (Grinnell et al., 1994), consistent with our hypothesis that glycodelin blocks both human sperm-zona pellucida binding and immune cell function via its oligosaccharide chains.

EXPERIMENTAL PROCEDURES
Isolation of Glycodelin-Isolation and purification procedures were the same as those described elsewhere  using 140 ml of midtrimester amniotic fluid as starting material.
Preparation of CNBr Fragments-Glycodelin (160 g) was dialyzed against 2 liters of 50 mM ammonium bicarbonate buffer, pH 8.5 at 4°C for 12 h, after which time it was dialyzed against 2 liters of water for another 12 h at 4°C and then lyophilized. The lyophilized sample was dissolved in 100 l of a solution of CNBr in 70% formic acid and left in the dark for 12 h. The reaction was terminated by drying in vacuo. An additional 5 l of water was added and the sample dried in vacuo. The sample was then dissolved in 25 l of triethylamine, 2.5 l of water and reduced using a 4-fold molar excess of dithiothreitol over the number of S-S bridges. The reaction was allowed to proceed for 30 min at 37°C, after which time it was dried in vacuo. An additional 5 l of water was added and the sample dried in vacuo.
PNGase F Digestion-PNGase F (EC 3.2.2.18, Boehringer Mannheim) digestion was carried out on tryptic digests of glycodelin (250 g) in ammonium bicarbonate buffer (50 mM, pH 8.4) for 16 h at 37°C using 0.6 unit of the enzyme. The reaction was terminated by lyophilization and the products were purified on C 18 -Sep-Pak (Waters Ltd.) as described (Dell et al., 1994).
Methanolysis-The reagent was prepared by bubbling dry HCl gas into methanol as described (Dell et al., 1994). After cooling, a 20-l aliquot of this reagent was added to the permethylated sample, which was then heated for 2 min at 40°C. A 1-l aliquot was removed for FAB-MS analysis, and the remainder of the sample was dried under nitrogen.
Chemical Derivatization for FAB-MS and GC-MS Analysis-Permethylation using the sodium hydroxide procedure was performed as described (Dell et al., 1994). Partially methylated alditol acetates were prepared from permethylated samples for GC-MS linkage analysis as described (Albersheim et al., 1967). Trimethylsilyl ether methyl glycosides were prepared for sugar analysis as described (Merkle and Poppe, 1994).
GC-MS Analysis-GC-MS analysis was carried out on a Fisons Instruments MD800 machine fitted with a DB-5 fused silica capillary column (30 m ϫ 0.32 mm, internal diameter, J & W Scientific). The partially methylated alditol acetates were dissolved in hexanes prior to on-column injection at 65°C. The GC oven was held at 65°C for 1 min before being increased to 290°C at a rate of 8°C/min. FAB-MS Analysis-FAB-MS spectra were acquired using a ZAB-2SE 2FPD mass spectrometer fitted with a cesium ion gun operated at 30 kV. Data acquisition and processing were performed using the VG Analytical Opus software. Solvents and matrices were as described (Dell et al., 1994).
LC-ES-MS Analysis-LC-ES-MS was performed on various digests of glycodelin using an on-line microbore reverse phase high performance liquid chromatography system (Brownlee C 18 Aquapore column) coupled to a VG BioQ triple quadrupole electrospray mass spectrome-ter. The sample was dissolved in 0.1% trifluoroacetic acid (buffer A) for injection on the column. The column was held for 5 min at 0% B (90% acetonitrile in 0.1% trifluoroacetic acid) followed by increase to 100% B over 90 min. The flow rate was 50 l/min. After passage through a UV spectrophotometer with a microflow cell, monitoring at 214 nm, the eluant was mixed with a 1:1 mixture of propan-1-ol and 2-methoxyethanol prior to stream splitting 1:9 for ES-MS analysis and collection, respectively.

RESULTS
A unique feature of mass spectrometry, namely the ability to derive definitive structural information from mixtures (in contrast to other spectroscopies normally requiring pure samples for study), was recognized and exploited by us in specifically designed strategies for "mixture analysis" some years ago (Geddes et al., 1969;Morris et al., 1971;Morris et al., 1978). The masses of component peaks alone (even in the absence of fragmentation) are diagnostic not only for the composition of biopolymers but also for sequence, by relating the masses observed to biosynthetic pathways (oligosaccharides) or amino acid/cDNA-derived sequences (oligopeptides), leading to the concept of mapping biopolymer structures by mass spectrometry (Lemaire et al., 1982;Dell et al., 1983;Morris et al., 1983). These strategies have been applied here to glycodelin to map and differentiate the glycopeptides from proteolytic digests and to define and locate glycosylation in the molecule. FAB Mapping of Total Glycan Population-Glycodelin was digested sequentially with trypsin and PNGase F. Released glycans were separated from peptides and were analyzed by FAB-MS after permethylation (Fig. 2, Table I). Notable features of these data are: (i) the major ion at m/z 1557 has the composition of a high mannose structure containing five mannoses; (ii) molecular ions corresponding to complex-and hybridtype structures occur in the mass range from m/z 1800 to m/z 4000 and, taken together, these are significantly more abundant than m/z 1557; (iii) the majority of the molecular ions have compositions consistent with biantennary structures; (iv) the A-type fragment ions in Fig. 2b indicate that both Gal-GlcNAc (lacNAc) and GalNAc-GlcNAc (lacdiNAc) antennae are present; (v) the lacNAc and lacdiNAc antennae may be substituted with either sialic acid or fucose but not both; (vi) minor fragment ions of composition Hex 2 HexNAc 2 and Hex 3 HexNAc 3 are indicative of low levels of poly-N-acetyllactosamine.
FAB Mapping of Tryptic Peptides-FAB-MS was carried out on the tryptic digest before and after PNGase F digestion (Table II). The following conclusions may be drawn from these data. (i) The molecular ion for the non-glycosylated peptide spanning the consensus site at Asn-85 was observed prior to PNGase F digestion and there was no evidence for the formation of its Asp-85 analogue after PNGase F digestion, indicating that Asn-85 is unlikely to be glycosylated; (ii) the tryptic peptide spanning Asn-28 was observed only after PNGase F digestion and its mass was consistent with conversion of Asn to Asp during the digestion, consistent with Asn-28 being glycosylated; (iii) the tryptic peptide corresponding to the third consensus site was not observed in these experiments. Linkage Analysis of Total Glycan Population-Linkage analysis on the PNGase F-released glycans and their desialylated counterparts gave the data shown in Table III. Key features of   these data are as follows. (i) The majority of the complex glycans are biantennary but low levels of 2,4-Man and 2,6-Man suggest that minor tri-and/or tetraantennary structures are present; (ii) GalNAc, Gal, and Man are the major non-reducing sugars; (iii) after desialylation, 6-linked GalNAc and 6-linked Gal disappear and there is a concomitant increase in terminal GalNAc and terminal Gal, indicating that sialic acid residues were attached to the 6-positions of Gal and GalNAc prior to desialylation; (iv) some terminal GlcNAc is present but most of the GlcNAc is 4-linked, 3,4-linked, or 4,6-linked; (v) a very minor amount of 3-linked Gal is present, the majority of which is retained after desialylation and is therefore likely to be derived from the minor poly-N-acetyllactosamine moieties suggested by the FAB data (m/z 913 and 1362 in Fig. 2, see above); (vi) the minor 3,4,6-linked Man is indicative of some bisected structures; (vii) 3-linked Man and 6-linked Man are present but only as very minor components; they are indicative of hybrid and/or high mannose structures.
Determination of Fucosyl Linkages-The attachment sites of the fucosyl residues were established by linkage analysis after removal of the fucoses by mild methanolysis and remethylation of the newly formed hydroxyl groups (see footnotes to Table  III). Comparison of linkage data before and after mild methanolysis indicates that loss of fucosyl residues from the antennae is accompanied by loss of the 3,4-linked GlcNAc and a concomitant increase in 4-linked GlcNAc. Importantly no 3-linked GlcNAc was observed after methanolysis. These data establish that fucose is attached to the 3-position of 3,4-linked GlcNAc. Exoglycosidase Digestions-The glycan mixture was treated sequentially with ␣-sialidase, ␣-fucosidase, ␤-hexosaminidase, and ␤-galactosidase, and the reactions were monitored by FAB-MS after permethylation. The FAB spectrum of the fully digested sample was dominated by an A-type ion at m/z 872 (Hex 3 HexNAc ϩ ) and an [M ϩ H] ϩ ion at m/z 1149 corresponding to Hex 3 HexNAc 2 (data not shown), confirming that the majority of the complex structures can be degraded to the trimannosyl core by this series of exoglycosidases. Thus the NeuAc and Fuc residues are in normal ␣ linkages and the Gal, GalNAc, and GlcNAc residues are all ␤ linked. The signal at m/z 1557 (see Fig. 2) was unaffected by the above exoglycosidase digestions, a result that is consistent with the assignment of a high mannose structure to this ion. This was corroborated in a separate experiment in which the intact glycans were subjected to ␣-mannosidase digestion. The resulting FAB spectrum was very similar to Fig. 2 except that m/z 1557 had disappeared and a new signal was present at m/z 1149 corresponding to Man 3 GlcNAc 2 (data not shown). In addition the signal at m/z 2368 was no longer present, consistent with the proposed composition of NeuAcHex 6 HexNAc 3 (Table I), which corresponds to a hybrid structure.
LC-ES Mapping of Glycodelin-In the analysis of tryptic digests of glycodelin, a well resolved intense late-eluting peak in the UV chromatogram gave electrospray data transformed to a mass of 4749, which maps onto the non-glycosylated disulfide bridged peptide Ile-84 to Arg-124. Summation of scans surrounding this peak and in earlier eluting fractions produced no evidence for glycosylated versions of this peptide (data not shown). Thus it is clear that site Asn-85 is not glycosylated at the level of detection of the method. In a separate experiment, glycodelin was digested with cyanogen bromide and following a disulfide reduction step the products were analyzed by on-line microbore LC-ES-MS (Table IV). Fig. 3 shows the UV (analog) trace above the total ion current (TIC) trace produced on ionization of samples entering the mass spectrometer ion source. The short time delay indicates the UV cell to source flow time at 100 l min Ϫ1 . Scans corresponding to UV and TIC peaks were summed, and the data on multiply charged ions thus created were transformed to give masses shown in Table IV. From these LC-ES-MS digest data, the whole of the molecule is effectively mapped, including the glycosylation sites discussed below. Scans 53-60 (analogue peak 7.10) gave the transformed ES mass spectrum shown in Fig. 4. The cluster of molecular ions observed is indicative of a heterogeneous mixture of glycopeptides. Taking into account the location of methionine in the sequence (Fig. 1), it was considered likely that this early eluting glycopeptide peak comprised glycoforms of the peptide spanning residues 25-32, which includes the consensus site at Asn-28. Subtraction of the mass of this peptide from each observed signal in the ES spectrum yields the tentative glycan    Table V. In order to confirm these assignments, including resolving the NeuAc/Fuc 2 ambiguity (see legend to Table V), and to check for additional minor components, the collected fractions 41-43 were pooled and digested with PNGase F. The released glycans were permethylated and analyzed by FAB-MS (Fig. 5, Table VI). The FAB data show that, with the exception of Hex 4 HexNAc 5 Fuc 2 , all the fucosylated glycans contain only a single fucose residue ruling out the other tentative Fuc 2 assignments in Table V. The FAB spectrum contains one minor molecular ion not observed in the ES spectrum (corresponding to NeuAcHex 5 HexNAc 4 Fuc), but otherwise all molecular ion signals observed in the FAB spectrum (Fig. 5, Table VI) have their counterparts in the ES spectrum (Fig. 4, Table V). The peptide released in the glycanase experiment was found to produce a quasimolecular ion at m/z 834, corresponding to peptide residues 25-32. Glycosylation site 28 is thus proven to carry the glycans in Table VI, and the LC-ES-MS data show it to be well separated from other molecular species.
Combining scans 139 -145 produces the raw data (multiply charged) shown in Fig. 6. The complex appearance of this spectrum, which contrasts with the clean single signals of different charge states expected for a peptide, is immediately indicative of the expected heterogeneity seen in a glycopeptide. Computing the charge states shown, and transformation of the data produces component masses of 11,835.5, 11,879.3, and 12,126.8 for the most abundant peaks. These masses are approximately 2000 Da higher than the anticipated mass of peptide Ala-33 to Met-117, indicating glycosylation of the peptide. The peptide contains potential glycosylation sites Asn-63 and Asn-85, but since we have already proven that Asn-85 is not glycosylated (see earlier FAB and ES mapping experiments), it follows that the glycans on peptide Ala-33 to Met-117 are attached to Asn-63. Their identities were studied in detail by FAB-MS analysis after their release by PNGase F from the glycopeptides in collected fractions 56 -57 (see Fig. 7 and Table  VII). It is noteworthy that the majority of glycans in this sample are different from those attached at Asn-28 (see below).
Occupancy of Consensus Sites-Unequivocal evidence for full glycan occupancy of Asn-28 was provided by a combination of the FAB Mapping experiments on tryptic and PNGase F digests of glycodelin (Table IV) and the on-line LC-ES-mapping experiments on tryptic and CNBr digests (Fig. 3). The latter experiments also established that only one (Asn-63) of the remaining two consensus sites is glycosylated at observable levels (Fig. 6). The separation of the occupied sites in the LC-ES mapping experiments on reduced CNBr digests of glycodelin allowed identification of the differing oligosaccharide structures at Asn-28 and Asn-63 (see below). Assignment of Oligosaccharide Structures-The proposed structures for the major oligosaccharides are shown in Fig. 8. The glycans fall into three classes, namely high mannose (i), hybrid (ii and iii) and complex (iv-xx), of which the first two classes are only found at Asn-28, whereas complex structures occur at both glycosylation sites. Among the complex structures only (ix and xiii) are common to both sites. The most notable differences between the complex structures at Asn-28 and Asn-63 are the increased levels of sialylation and fucosylation of the glycans at the latter site. The high sensitivity achieved in the FAB-MS analyses of the total glycan population allowed the detection of very minor components giving molecular ions at masses above m/z 3000 (see Fig. 2 and Table I). These correspond to tri-and tetraantennary structures and/or bi-and tri-antennary structures with N-acetyllactosamine repeats in their antennae. The FAB fragmentation data and the linkage analysis results suggest that both types of structure are present (see above). The very low abundance of these components has to date precluded precise structural analysis or determination of attachment sites.

DISCUSSION
The majority of the glycodelin N-linked oligosaccharides characterized in this study are not typically found in mammalian glycoproteins. In particular, the presence of lacdiNAccontaining antennae is unusual because, with the exception of the pituitary glycohormones, this sequence has been rarely observed in the glycoproteins of higher animals (see Dell and Khoo (1993), van den Eijnden et al. (1995), and references cited therein).
The best characterized family of mammalian lacdiNAc glycoproteins are the pituitary glycohormones, which contain sulfated lacdiNAc structures (Baenziger and Green, 1988). The GalNAc transferase, which adds GalNAc to these glycoproteins, recognizes the tripeptide motif Pro-Xaa-Arg/Lys (PXR/K) located 6 -9 residues NH 2 -terminal to an Asn glycosylation site (Smith and Baenziger, 1992). This GalNAc transferase, together with the sulfotransferase responsible for synthesizing the unique sulfated epitope on the pituitary glycohormones, is present in a number of tissues other than the pituitary, and the two enzymes appear to be co-ordinately expressed (Dharmesh et al., 1993). However, we consider it unlikely that glycodelin is a substrate for the PXR/K-specific GalNAc transferase because it does not contain a recognition motif 6 -9 residues upstream of either glycosylation site. Furthermore, we were not able to detect sulfated structures in glycodelin using acetylation/ FAB-MS strategies, which are optimized for the detection of sulfated oligosaccharides (data not shown) .
Non-sulfated lacdiNAc structures of the type present in glycodelin have previously been found in a three categories of mammalian glycoproteins. The first comprises glycoproteins produced by bovine mammary glands, including lactotransferrin (Coddeville et al., 1992), CD36 (Nakata et al., 1993), and butyrophilin (Sato et al., 1995). The second contains three human glycoproteins, all of which are serine proteases with important physiological functions, namely Bowes melanoma tissue plasminogen activator (Chan et al., 1991) and urinary type plasminogen activator (urokinase) (Bergwerff et al., 1992), both of which convert plasminogen to plasmin, and urinary kallidinogenase (Tomiya et al., 1993), which cleaves kininogens to liberate lysyl-bradykinin, a vasoactive peptide. The third category contains only a single glycoprotein at present, namely human recombinant Protein C (rHPC) expressed in human kidney 293 cells (Yan et al., 1993), but we anticipate the discovery of many more examples with the increasing use of this human cell line for the expression of recombinant glycoproteins. Due to its availability in large quantities, rHPC is among the best characterized of the mammalian lacdiNAc-containing glycoproteins. Like glycodelin, rHPC carries a heterogeneous population of complex-type oligosaccharides composed of lac-NAc and lacdiNAc building blocks, which are substituted with either sialic acid or fucose (Yan et al., 1993). Interestingly, the sialylated lacNAc antennae in rHPC have both ␣2-3 and ␣2-6 linked sialic acid, but the the former linkage was not observed in the lacdiNAc antennae. It is notable that NeuAc␣2-  3GalNAc␤1-4GlcNAc has not, to our knowledge, been found in mammalian glycoproteins, although this structure has been identified in serine proteases derived from snake venoms (Pfeiffer et al., 1992;Lochnit and Geyer, 1995). It is possible that, by analogy with PXR/K-specific GalNAc transferase and sulfotransferase (Dharmesh et al., 1993), there could be coordinate expression of GalNAc transferase and ␣2-6-sialyltransferase in mammalian cell lines that synthesize sialylated lacdiNAc structures. Several lines of evidence indicate that oligosaccharides are essential recognition sequences in cell-mediated adhesions in both inflammatory and immune responses (Phillips et al., 1990;Springer, 1990;Lasky, 1992;Bevilacqua, 1993). Oligosaccharides terminated with sialylated or sulfated Lewis x/a type sequences have been shown to act as specific ligands for selectinmediated adhesions (Berg et al., 1991;Yuen et al., 1992). Other oligosaccharide sequences can, however, act as selectin ligands (Varki, 1994). Importantly, in the rHPC study (see above) it was shown that a biantennary N-linked oligosaccharide bearing GalNAc␤1-4(Fuc␣1-3)GlcNAc antennae is a potent inhibitor of E-selectin-mediated adhesion (Grinnell et al., 1994). Since the same fucosylated epitope is also expressed on glycodelin, it is possible that a component of the immunosuppressive effect exhibited by glycodelin is mediated via blocking of the selectin-like binding sites by this carbohydrate sequence.
Other specific antennae associated with glycodelin may also interact with alternative bioactive receptor proteins of the human immune system. CD22 is a B cell-associated receptor of the immunoglobulin superfamily that acts as both an adhesion molecule and an activation molecule (Clark and Lane, 1991;Clark, 1993;Ledbetter et al., 1993;Peaker, 1994). Transfected cells that stably express CD22 on their surfaces show greatly FIG. 8. Structures of the major Nglycans present at Asn-28 (a) and Asn-63 (b) of glycodelin. Panel a, superscript a indicates that minor forms may exist with different arm structures as indicated by presence of 3-and 6-linked mannose; superscript b indicates that fucose residue may be 3-linked to the Glc-NAc on either arm. Panel b, superscript c indicates that the fucose residue may be 3-linked to the GlcNAc on either arm, but is not on the arm bearing the sialic acid. enhanced binding to T and B lymphocytes (Wilson et al., 1991). CD22 is closely associated with the subset of responsive B lymphocytes as defined by stimulation with anti- (Pezzutto et al., 1988). CD22 also binds to CD45, the leukocyte-specific receptor-linked phosphotyrosine phosphatase involved in T-cell activation (Stamenkovic et al., 1991). Previous studies have revealed that CD22 binds to NeuAc␣2-6Gal␤1-4GlcNAc sequences (Powell and Varki, 1994). More recent studies indicate that CD22 also binds the NeuAc␣2-6GalNAc disaccharide with approximately equal affinity as it does the NeuAc␣2-6Gal␤1-4GlcNAc sequences (Powell et al., 1995). Therefore we believe that glycodelin may bind to CD22 via its NeuAc␣2-6Gal␤1-4GlcNAc and/or NeuAc␣2-6GalNAc␤1-4GlcNAc antennae and may inhibit specific immune cell adhesion and activation events mediated via this receptor protein.
We also find it significant that glycodelin has glycoforms carrying NeuAc␣2-6GalNAc␤1-4GlcNAc and GalNAc␤1-4(Fuc␣1-3)GlcNAc antennae on a single biantennary oligosaccharide (structure xviii; Fig. 8). Although this structure has been previously observed in rHPC (Yan et al., 1993), we are now the first to demonstrate its expression in a naturally occurring glycoprotein. The biological activities expressed by this glycan remain to be determined. It is possible that this oligosaccharide could interact with the selectins or other adhesion molecules with selectin-like specificity via the fucosylated antenna whereas its sialylated antenna could bind to CD22. Such an oligosaccharide could manifest multiple biological effects, including blocking inflammatory responses, attenuating CD22-dependent immune responses or perhaps inhibiting other selectin-like adhesion processes. It is also possible that certain carbohydrate-binding proteins associated with either the immune or reproductive systems may require the precise spatial arrangement of fucose and sialic acid provided by the antennae for optimal binding.
Evidence collected from diverse species in both the plant and animal kingdoms indicates that the appropriate recognition of surface carbohydrates is a crucial event in the binding of sperm to the eggs during fertilization (Macek and Shur, 1988;Miller and Ax, 1990;Wassarman, 1990). In the mouse, oligosaccharides associated with the zona pellucida glycoprotein ZP3 have been shown to be recognized by specific egg-binding proteins located on the sperm plasma membrane (Wassarman, 1990). It is probable that a similar paradigm is utilized in the human system.
We have previously suggested that initial human spermzona pellucida binding involves a selectin-like adhesion (Patankar et al., 1993a(Patankar et al., , 1993b. This proposed specificity was initially based upon our observation that fucoidan blocked initial human sperm-zona pellucida binding (Oehninger et al., 1990) and a selectin-mediated adhesion process (lymphocyte homing) in the same concentration range (Yednock and Rosen, 1989). Fucoidan also blocked induction of the sperm's acrosome reaction by solubilized human zona pellucida, consistent with its ability to block sperm-zona pellucida binding (Mahony et al., 1991). We recently reported in a preliminary study that sialyl-Lewis x oligosaccharide and human orosomucoid also inhibit initial human sperm-zona pellucida binding in the same concentration-dependent manner as is observed for E-selectin-mediated adhesion (Clark et al., 1995a). Although these studies suggest that the egg-binding protein is a selectin, our preliminary studies using specific anti-selectin monoclonal antibodies indicate that these adhesion proteins are not expressed on human sperm (Clark et al., 1995b). Therefore, we have hypothesized that the human egg-binding protein, though not itself a selectin, may have converged with the selectins in its carbohydrate binding specificity. Glycodelin also inhibits initial human sperm-zona pellucida binding in a potent concentration-dependent manner (Oehninger et al., 1995). The expression of putative selectin ligands on this glycoprotein provides further evidence supporting our hypothesis that initial human spermzona pellucida binding is dependent upon a selectin-like adhesion process.
The structural studies reported in this paper provide the necessary foundation for effectively addressing the above issues. We are now investigating the potential contribution of the variety of glycans attached to glycodelin to its immunosuppressive and contraceptive activities. Finally, since glycodelin is expressed in bone marrow (Kamarainen et al., 1994;Morrow et al., 1994) and perhaps other tissues, it will be of great interest to see whether their glycosylation and function are the same. Until that information is available we propose to designate this glycoprotein isolated from amniotic fluid "glycodelin-A."