Defining the carbohydrate specificities of aplysia gonad lectin exhibiting a peculiar D-galacturonic acid affinity.

Aplysia gonad lectin (AGL), which has been shown to stimulate mitogenesis in human peripheral lymphocytes, to suppress tumor cells, and to induce neurite outgrowth and improve cell viability in cultured Aplysia neurons, exhibits a peculiar galacturonic acid/galactose specificity. The carbohydrate binding site of this lectin was characterized by enzyme-linked lectino-sorbent assay and by inhibition of AGL-glycan interactions. Examination of the lectin binding with 34 glycans revealed that it reacted strongly with the following glycoforms: most human blood group precursor (equivalent) glycoproteins (gps), two Galalpha1-->4Gal-containing gps, and two d-galacturonic acid (GalUA)-containing polysaccharides (pectins from apple and citrus fruits), but poorly with most human blood group A and H active and sialylated gps. Among the GalUA and mammalian saccharides tested for inhibition of AGL-glycan binding, GalUA mono- to trisaccharides were the most potent ones. They were 8.5 x 10(4) times more active than Gal and about 1.5 x 10(3) more active than the human blood group P(k) active disaccharide (E, Galalpha1-->4Gal). This disaccharide was 6, 28, and 120 times more efficient than Galbeta1-->3GlcNAc(I), Galbeta1-->3GalNAc(T), and Galbeta1--> 4GlcNAc (II), respectively, and 35 and 80 times more active than melibiose (Galalpha1-->6Glc) and human blood group B active disaccharide (Galalpha1-->3Gal), respectively, showing that the decreasing order of the lectin affinity toward alpha-anomers of Gal is alpha1-->4 > alpha1-->6 > alpha1-->3. From the data provided, the carbohydrate specificity of AGL can be defined as GalUAalpha1-->4 trisaccharides to mono GalUA > branched or cluster forms of E, I, and II monomeric E, I, and II, whereas GalNAc is inactive.

microorganisms (3) and strongly agglutinates human papaintreated erythrocytes regardless of ABO blood groups (2). It was purified by heating to 70°C, precipitated with ammonium sulfate, and affinity chromatographed on Sepharose 4B (2). The purified lectin is a glycoprotein of molecular mass of about 65 kDa, composed of two identical subunits. It was shown to induce mitogenic stimulation and interleukin-2 formation in human lymphocytes (4), to suppress tumorigenicity of Lewis lung carcinoma cells (5), to modulate neurite outgrowth in cultured Aplysia neurons, and to increase neurite viability in vitro (6). AGL was also shown to be useful for ultrastructural characterization of galacturonic acid in plants and fungi (7) and for differentiation between I and i type human erythrocytes (8). Moreover, it was recently found to be useful for typing of halopilic Archaea and for the study of their S-layer structure (9). Although AGL has been shown to be specific for GalUA and Gal, its detailed carbohydrate specificity has not been established. It is important to elucidate the detailed carbohydrate specificity of this lectin, because it may function as a signaling adhesion molecule and has potential as a tool in experimental glycobiology, biochemistry, and immunochemistry (4 -9). In the present study, we defined the glycan affinity of this lectin by both enzyme-linked biotin/avidin-mediated microtiter plate lectin assay (ELLSA) and also by examination of the inhibition of AGL-glycan interaction (10,11). The great advantage of this method is that the amount of lectin and glycoform required is about 1/10 to 1/1000 of that required for the quantitative precipitin assay (12,13). The results show that the carbohydrate affinity hierarchy of this lectin can be regarded as: tri-Ga-lUA␣134 to mono-GalUA Ͼ branched and/or clusters of E(Gal␣134Gal), I(Gal␤133GlcNAc), and/or II(Gal␤134 GlcNAc) Ͼ Ͼ monomeric E and I Ͼ II, although GalNAc is inactive.

EXPERIMENTAL PROCEDURES
Lectin-The AGL was purified from extracts of the gonads of Aplysia depilans as described previously (2).
Biotinylation of the Lectin-For AGL biotinylation by biotinamidocaproate-N-hydroxy-succinimide ester (biotin ester, purchased from Sigma), the purified lectin preparation (200 g/250 l of phosphatebuffered saline) was mixed with 400 l of the biotin ester solution (100 g of biotin ester/200 g of lectin) and left for 30 min at room temperature. The biotinylated lectin was dialyzed for 2-3 h against distilled H 2 O and overnight against TBS. After dialysis, the sample volume was adjusted to 1 ml with TBS, and 20 l of 5% sodium azide was added (200 g/ml AGL in 0.1% NaN 3 ) (10 -11).
Glycoproteins and Polysaccharides-The blood group substances were purified from human ovarian cyst fluid by the procedures as described previously (14 -19). Regardless of their A, B, H, Le a , or Le b activity, the purified water-soluble blood group substances have a similar overall structure. They are polydispersed macromolecules (M r 2.0 ϫ 10 5 to 1.0 ϫ 10 6 ) of similar composition (75-85% carbohydrates, 15-20% protein). They bear multiple heterosaccharide branches attached by glycosidic linkages at their internal reducing ends to serine or threonine of the polypeptide backbone (14 -22).
The P-1 fractions of cyst glycoproteins represent the nondialyzable portion of the blood group substances after mild hydrolysis at pH 1.5-2.0, 100°C for 2 h, which removed most of the L-fucopyranosyl end groups, as well as some blood group A and B active oligosaccharide side-chains (14,23,24). The 1st Smith-degraded products of blood group A active substances (MSS 10% 2X, see Table I), in which almost all of the sugar groups at the nonreducing ends were removed, were prepared as described earlier (18,20). Both P-1 fractions and 1st Smith degradation products, prepared from human ovarian cyst glycoprotein, are defined as precursor equivalent glycoproteins (Structure I, [25][26][27][28]. The human blood group P 1 -active substance, purified from sheep hydatid cyst glycoprotein (29,30), was kindly provided by Dr. W. M. Watkins (University of London, Royal Postgraduate Medical School, Hammersmith Hospital, London, UK). The mucus glycoprotein (native bird nest glycoprotein), the so-called nest-cementing substance (Structure II), from the salivary gland of Chinese swiftlets (genus Collocalia), was extracted with distilled H 2 O at 60°C for 20 min from the commercial bird nest substance (Kim Hing Co., Singapore) (31,32).
The Pneumococcus type XIV polysaccharide was prepared as described previously (33,34). Fetuin (Life Technologies, Inc.), which is the major glycoprotein in fetal calf serum (35), has a molecular mass of 48,400 with the following composition: 78% amino acids, 8.7% sialic acid, 6.3% hexosamines, and 8.3% neutral sugars (36). It bears six oligosaccharide side chains/molecule, three of them (of two types) are O-glycosyl-linked to Ser or Thr residues of the protein core, and the other three are N-glycosyl-linked to asparagine (37)(38)(39).
The rat sublingual glycoprotein (RSL) was prepared by the method of Moschera and Pigman (40). Its molecular mass is 2.2 ϫ 10 6 and it contains 81% carbohydrates (40). The carbohydrate side-chains (Structure III) are O-glycosyl-linked to Ser or Thr residues of the protein core. The established structure has 9, 10, 12, 13, and 15 sugar residues with NeuNAc␣2,6 linked to Gal, GalNAc␣13 Ser/Thr (Tn), and GlcNAc groups at the nonreducing ends, as well as a repeating unit, Gal␤134GlcNAc␤13, in the carbohydrate core structure (Structure III) (41). It was found that its carbohydrate chains also contain the Tn active determinant (42).
Porcine salivary mucin (PSM), bovine submandibular glycoproteinmajor, and armadillo salivary glycoprotein were purified according to the method of Tettamanti and Pigman (43) with some modifications (44 -47). Sialic acids were removed from sialylated glycoproteins by mild acid hydrolysis with 0.01 N HCl at 80°C for 90 min and dialyzed against distilled water for 2 days to remove small fragments (47).
The Microtiter Plate Lectin-Enzyme Binding Assay (ELLSA)-ELLSA was performed according to the procedures of Duk et al. (10) or as described previously (12,13). The volume of each reagent applied to the plate was 50 l/well, and all incubations, except for coating, were performed at 20°C. The reagents, if not otherwise indicated, were diluted with TBS containing 0.05% Tween 20. The TBS buffer or 0.15 M NaCl containing 0.05% Tween 20 was used for washing the plates between incubations.
For inhibition studies, the serially diluted inhibitor samples were mixed with an equal volume of lectin solution containing a fixed amount of lectin. The control lectin sample was diluted 2-fold with TBS containing 0.05% Tween 20. After 30 min at 20°C, the samples were tested in the binding assay, as described above. The inhibitory activity was estimated from the inhibition curve and expressed as the amount of inhibitor (nmol/well) giving 50% inhibition of the control lectin binding.
All experiments were done in duplicate or triplicate, and the data presented are mean values of the results. The standard deviation did STRUCTURE I. Proposed carbohydrate side chains of blood group active glycoproteins prepared from human ovarian cyst fluid. The blood group substance was purified from human ovarian cyst fluid by digestion with pepsin and precipitation with ethanol; the dried ethanol precipitates were extracted with 90% phenol, the insoluble fraction being given after the name of the blood group substance (e.g. Cyst Beach phenol insoluble). The supernatant was fractionally precipitated by the addition of 50% ethanol in 95% phenol to the indicated concentrations. The designation 10 or 20%(ppt) denotes a fraction precipitated from phenol at an ethanol concentration of 10 or 20%; 2X signifies that a second phenol extraction from ethanol precipitation was carried (e.g. Cyst OG 20% 2ϫ). The 4-branched structure (I-IV) shown above represents the internal portion of the carbohydrate moiety of blood group substances to which the residues responsible for A, B, H, Le a , and Le b activities are attached. This structure represents precursor blood group active glycoproteins (25) and the precursor equivalent gp can be prepared by Smith degradation of cyst A, B, and H active glycoproteins (4,19,20,22,25) or mild acid hydrolysis (P-1 gp) (4,12,23,24). Numbers in parentheses indicate the site of attachment for the human blood group A, B, H, Le a , and Le b determinants. These determinants as well as the structural units at the nonreducing end are the sources of lectin A/A h , B, I, II, T, and Tn determinants. A megalosaccharide of 24 sugars has not been isolated. However, most of the carbohydrate chains isolated are parts of this structure. For this structure, various human blood group determinants are attached and illustrated in Table I.
not exceed 10% and in most experiments was less than 5% of the mean value. The control wells, where either coating or addition of biotinylated lectin was omitted, gave low absorbance values (below 0.1, read against the well filled with buffer) and were used as blank. It has been shown that blocking the wells before lectin addition was not necessary when Tween 20 was used in TBS.

reacted with AGL.
Inhibition of AGL-Glycoform Interaction by Various Glycans-The abilities of various glycans to inhibit the binding of AGL with Cyst Beach P-1 glycoprotein by ELLSA were analyzed and are shown in Fig. 2 and Table III. Among the glycans tested for inhibition of that interaction, six human blood group precusor equivalent gps (curves 1, 5, 6, 9, 10, and 11 in Fig. 2 and Table III); two Gal␣134Gal containing glycoproteins (sheep hydatid cyst gp and asialo bird nest, curves 3 and 4 of Fig. 2 and Table III), and two GalUA-containing polysaccharides (pectin-A and pectin-C, curves 7 and 8 of Fig. 2 and Table  III) were the best inhibitors, requiring less than 12 ng to inhibit 50% of the interaction. They were much more active than monomeric Gal␣134Gal and Gal but much weaker than Ga-lUA ( Fig. 2 and Table III). The precursor equivalent and asialo glycoproteins were much more active than their further glycosylated, native or sialylated compounds (curve 4 versus curve 19, curves 5 and 6 versus corresponding native compounds in FIG. 2. Inhibition of AGL binding to Cyst Beach P-1-coated enzymelinked immunosorbent assay plates with various glycoproteins and Tncontaining glycopeptides. The quantity of Cyst Beach P-1 in the coating solution was 10 ng/well. The quantity of lectin used for inhibition assay was 5 ng/well. Total volume, 50 l. A 405 was recorded after a 2-h incubation. When 277.8 ng of glycoprotein were used to inhibit AGL-Beach P-1 glycoprotein binding, the concentration required to induce 50% inhibition was determined. In this assay, Hog gastric mucin 4; Hog gastric mucin 9; Cyst MSS 10% 2x; Cyst Mcdon; PSM; Cyst JS phenol insoluble; Cyst Tighe phenol insoluble; RSL; fetuin; asialo fetuin; bovine submandibular gp-major; asialo bovine submandibular gp-major; OSM; asialo OSM (curves 22 to 35) did not reach 50% inhibition. Fig. 2 and Table III, etc.). The decreasing order of the reactivity of these glycoforms is Cyst OG 10% 2x PPT (one of the human blood group precursor gps, Structure I, curve 1 in Fig. 2) and sheep hydatid cyst gp (blood group P 1 active gp) (curve 3 in Fig.  2) Ͼ asialo bird nest gp (curve 4), five human blood group precursor gps (curves 5, 6, 9, 10, and 11) and two GalUAcontaining gps; pectin-A and pectin-C (curves 7 and 8) Ͼ a blood group B active gp (curve 12); mild acid-hydrolyzed hog gastric mucin 14 (II), hog mucin 21 (II), and asialo rat sublingual gp (II, structure III) Ͼ Ͼ blood group A and H active glycoproteins and sialylated glycoproteins (Figs. 2 and 4). With several exceptions, the inhibitory reactivities of glycoforms toward AGL agree, in general, with the maximum absorbance values recorded in the binding assay ( Fig. 1 and Table II).
Mild acid hydrolysis (pH 1.5, 100°C for 2 h), which removes the terminal L-Fuc␣13 linked and some blood group A and B active oligosaccharide side chains (14 -17), and Smith degradation, which removes almost all nonreducing terminal sugars, should significantly increase their interactions with this lectin (Tables II and III and Figs. 1 and 2) (18). The Gal␤13 determinant of human blood group precursor or precusor equivalent gp (Structure I) is similar to Bombay type erythrocytes (O h ) that are more strongly agglutinated by this lectin than the O(H) blood type (8).
Inhibition of Lectin-Glycan Interaction by Mono-and Oligosaccharides-The ability of various sugars to inhibit the binding of AGL to cyst Beach P-1 gp (human blood group precursor equivalent gp purified from human ovarian cyst fluid) is shown in Fig. 3, and the amounts of ligand required for 50% inhibition of the lectin-glycan interaction are listed in Table IV. Among the oligo-and monosaccharides tested, di Ͼ tri-GalUA␣134 to mono-GalUA were the most active, up to 8.5 ϫ 10 4 times more active than Gal, indicating that COOH at carbon-6 is the most important factor for binding. GalUA was about 1.5 ϫ 10 3 times more active than human blood group P k active disaccharides (E, Gal␣134Gal), which was 6, 28, and 120 times more active than Gal␤133GlcNAc(I), Gal␤133GalNAc(T), and Gal␤134GlcNAc (II), respectively. These results, show that each lectin has its own binding characteristics (26 -28) and that the carbohydrate specificity of AGL can be defined as GalUA␣134 di Ͼ trisaccharides to mono GalUA Ͼ branched or cluster forms of E, I, and II Ͼ Ͼ monomeric E(Gal␣134Gal) Ͼ I(Gal␤133GlcNAc) Ͼ T(Gal␤133GalNAc) Ͼ B(Gal␣133Gal) Ͼ II(Gal␤134GlcNAc), and L(lactose).
Gal␤134Man was about 2.5 times less active than Gal␤13 3GlcNAc (I), but 6.7 and 8 times more active than Gal␤134Glc-(L) and Gal␤134GlcNAc (II), respectively. These results show that the configuration at C-2 in the subterminal hexopyranose is also important for the binding and that substitution with ϪNHCOCH 3 at C-2 reduces the inhibitory power. Melibiose and raffinose (Gal␣136Gal␤132DFructo-furanoside) were almost equally active and 4.0 times more active than stachyose (Gal␣136Gal␣136Glc␤132DFructofuranoside), suggesting that the combining size for ␣136 oligosaccharides is probably most accessible with less than a trisaccharide structure.
Of the monosaccharide derivatives studied, phenyl␤Gal (Fig.  3a, curve 9) was the best inhibitor 2 ϫ 10 4 times less active than GalUA (Fig. 3b, curve 3), but 1.8 times more active than Gal and 1.2 times more than p-NO 2 -phenyl␤Gal and p-NO 2 -phenyl␣Gal. As shown in Table IV, the phenyl␣-derivative of Gal was about 2 times better than the methyl-␣-derivative (Fig.  3, curves 13 versus 20), whereas no significant difference was observed between the methyl-and p-NO 2 phenyl derivatives of ␤-Gal (Fig. 3a, curves 12 and 14).
GalNAc, which was tested up to a concentration exceeding that of Gal inducing 50% inhibition by 2.4-fold ( Fig. 3 and Table  IV), was inactive indicating that the N-acetamido group at C-2 of the Gal pyranose ring strongly interferes with its interaction with AGL. D-Fuc and L-Ara showed no inhibition up to three times the amount of Gal giving 50% inhibition, suggesting that the OH group at C-6 or the CH 2 OH at C-6 of Gal is essential for lectin binding. GlcNAc, Glc, methyl-␣-Glc, and methyl-␤-Glc were tested at concentrations from 380 to 463 nmol, but no inhibition of lectin binding was observed.
From these assays, we conclude that the major contributions of this study are: 1) establishment of the binding relationship between AGL and mammalian carbohydrate structural units; 2) illustration of the high specificity of AGL for GalUA-related glycotopes, which is rare and is an important consideration in animal lectins; 3) demonstration that the COO Ϫ group rather than OH Ϫ at carbon-6 dramatically enhances binding reactivity; and 4) presentation that the presence of carboxylated (ϪCOOH) or hydroxylated (ϪOH) carbon-6 and the configuration of Gal at carbon-4 are essential for binding (Gal versus Glc). This information should be useful for elucidating the mechanism of adhesion in the life process of Aplysia and other aspects of glycobiology.  (5 ng/50 l) to Cyst Beach P-1 gp (10 ng/50 l) The inhibitory activity was estimated from the inhibition curve in Fig. 3 and is expressed as the amount of inhibitor giving 50% inhibition. Total volume 50 l.