Garlic (Allium sativum) Lectins Bind to High Mannose Oligosaccharide Chains*

Two mannose-binding lectins, Allium sativum agglutinin (ASA) I (25 kDa) and ASAIII (48 kDa), from garlic bulbs have been purified by affinity chromatography followed by gel filtration. The subunit structures of these lectins are different, but they display similar sugar specificities. Both ASAI and ASAIII are made up of 12.5- and 11.5-kDa subunits. In addition, a complex (136 kDa) comprising a polypeptide chain of 54 ± 4 kDa and the subunits of ASAI and ASAIII elutes earlier than these lectins on gel filtration. The 54-kDa subunit is proven to be alliinase, which is known to form a complex with garlic lectins. Constituent subunits of ASAI and ASAIII exhibit the same sequence at their amino termini. ASAI and ASAIII recognize monosaccharides in mannosyl configuration. The potencies of the ligands for ASAs increase in the following order: mannobiose (Manα1–3Man) < mannotriose (Manα1–6Manα1–3Man) ≈ mannopentaose ≪ Man9-oligosaccharide. The addition of two GlcNAc residues at the reducing end of mannotriose or mannopentaose enhances their potencies significantly, whereas substitution of both α1–3- and α1–6-mannosyl residues of mannotriose with GlcNAc at the nonreducing end increases their activity only marginally. The best manno-oligosaccharide ligand is Man9GlcNAc2Asn, which bears several α1–2-linked mannose residues. Interaction with glycoproteins suggests that these lectins recognize internal mannose as well as bind to the core pentasaccharide of N-linked glycans even when it is sialylated. The strongest inhibitors are the high mannose-containing glycoproteins, which carry larger glycan chains. Indeed, invertase, which contains 85% of its mannose residues in species larger than Man20GlcNAc, exhibited the highest binding affinity. No other mannose- or mannose/glucose-binding lectin has been shown to display such a specificity.

The majority of the well characterized plant lectins have been isolated from the seeds of dicotyledonous species. But lectins of non-seed origin from other species are also emerging as promising tools chiefly because of two reasons: (i) a good number of them might contain novel sugar-binding sites; and (ii) they can provide valuable information regarding the biological roles of plant lectins, which to a large extent still remain elusive. In the recent past, there have been several reports of non-seed lectins from monocotyledonous families (1)(2)(3), especially Amaryllidaceae. The most remarkable property of these lectins is that they show strict specificity for mannose (2,4,5), unlike other mannose/glucose-binding plant lectins. Hence, they are being used extensively as affinity ligands for the purification of glycoproteins, viz. IgM, ␣ 2 -macroglobulin, haptoglobin, and ␤-lipoprotein (3,6).
Van Damme et al.
(3) examined a number of species (including Allium sativum) from the family Alliaceae (which is taxonomically close to the family Amaryllidaceae) and found them to accumulate mannose-binding lectins. They observed that lectins from both families share many common properties like their state of oligomerization, sugar specificity, amino acid composition, and serological interaction. We note that the bulbs of the species A. sativum contain an additional lectin (other than the one(s) described by them) that differs in its quarternary structure.
We found that the garlic lectins bind most avidly to invertase (which contains high mannose residues) among the glycoproteins tested. By exploiting this property, we developed a simple method to study their sugar specificities. This study reveals that the binding sites of these lectins accommodate a number of ␣1-2-linked mannose residues. None of the other mannose-binding lectins have been shown to exhibit this kind of specificity.
Preparation of Mannose-Sepharose Affinity Matrix-Mannose was coupled to epichlorohydrin-activated Sepharose 6B following the procedure of Sundberg and Porath (10).
Purification of A. sativum Agglutinins (ASAs) 1 -Healthy dry bulbs of A. sativum were purchased from the local market. The bulbs were homogenized with a blender using 20 mM phosphate buffer (pH 7.4) containing 150 mM NaCl (PBS). The extract was filtered and centrifuged at 10,000 rpm. The supernatant was subjected to (NH 4 ) 2 SO 4 cut * This work was supported by a grant from the Departments of Science and Technology and Biotechnology, Government of India. The costs of publication of this article were defrayed in part by the payment of page charges. This article must therefore be hereby marked "advertisement" in accordance with 18 U.S.C. Section 1734 solely to indicate this fact.
‡ Post-doctoral Fellow in the Core Support Program in Drug and Molecular Design supported by a grant from the Department of Biotechnology, Government of India (to the Indian Institute of Science).
(70%), and centrifuged, and the protein pellet was resuspended in PBS and dialyzed extensively against PBS. The crude protein was loaded on the mannose-Sepharose column at 4°C. The column was then washed extensively at the same temperature with PBS until A 280 nm was below 0.005. The bound protein was eluted with 0.2 M mannose in PBS at room temperature.
The affinity-purified protein was dialyzed extensively initially against 20 mM PBS and finally against 50 mM PBS and concentrated using a Centricon filtration unit. The concentrated protein was then applied to a Bio-Gel P-200 column (1.8 ϫ 110 cm) equilibrated and eluted with 50 mM PBS.
Electrophoretic Procedures-SDS-polyacrylamide gel electrophoresis under reducing conditions was carried out as described by Laemmli (11). Molecular masses of the lectins were calculated according to the method of Weber and Osborn (12) using lysozyme (14.2 kDa), ␤-lactoglobulin (18.4 kDa), trypsinogen (24 kDa), hen egg ovalbumin (45 kDa), and bovine serum albumin (68 kDa) as the standards. The proteins were visualized on the gel by Coomassie staining. A P/ACE system 2100 (Beckman Instruments) was used for capillary electrophoresis with P/ACE system software controlled by an IBM PS/2 Model 50-Hz computer. Post-run data analysis was performed on System Gold software (Beckman Instruments). A standard capillary of 27 cm length (20 cm to the detector window) ϫ 20 m inner diameter, designed for P/ACE cartridges, was obtained from Beckman Instruments. On-line detection was set at 280 nm with a 50 ϫ 200-m aperture in the P/ACE cartridge. Temperature of the capillary during electrophoresis was maintained at 25°C. Samples were introduced by pressure injection for 5 s. Electrophoresis was performed at a constant voltage of 8 kV.
Determination of the Amino-terminal Sequence-Proteins after separation by 15% SDS-PAGE were electroblotted onto polyvinylidene difluoride membrane following the procedure of Matsudaira (13) using a Multiphor II electrophoresis system (Pharmacia Biotech Inc.). Nterminal analysis of the native protein was carried out on a Shimadzu automated gas-phase sequencer (Model PSQ-1) equipped with an online C-R4A120A Chromatopac Shimadzu phenylthiohydantoin analyzer.
Determination of Native Molecular Mass-The native molecular masses of ASAs were determined by gel filtration on a Bio-Gel P-200 column (1.8 ϫ 110 cm) calibrated with rabbit IgG, Vicia villosa agglutinin, soybean agglutinin, bovine serum albumin, hen egg ovalbumin, and pepsin. Void and inner volumes of the column were determined with blue dextran and myoglobin, respectively. Molecular masses of ASAI and ASAIII were also determined on a Bio-Gel P-60 column (8 ϫ 90 cm) calibrated with hen egg ovalbumin, chymotrypsinogen, myoglobin, and bovine pancreas ribonuclease A.
Protein Estimation-Protein concentration was determined following the method of Lowry et al. (14) using bovine serum albumin as the standard.
Sugar Assay-Total neutral sugar content was determined by the phenol-sulfuric acid method of Dubois et al. (15) using mannose as the standard.
Hemagglutination Assay-Hemagglutination was carried out at room temperature using rabbit and human erythrocytes (16). Hemagglutination inhibition tests were done by preincubating lectin (10 hemagglutinating units) with serially diluted sugars or glycoproteins in microtiter plates. Rabbit erythrocyte suspension (25 l of 4% (v/v)) was then added to the solution, and the results were noted after 1 h. Estimation of the Activity of Invertase Bound to ASAs-An appropriate amount of lectin was coated on an enzyme-linked immunosorbent assay plate and left overnight at 4°C. The plate was washed with 20 mM PBS and then blocked with 3% bovine serum albumin in PBS. After washing with the blocking buffer, invertase was added to the wells and incubated for 1 h. The plate was then washed with PBS, and 100 l of 50 mM sucrose solution was added to each well. After incubation for 1 h, the solutions were transferred to separate test tubes. The extent of hydrolysis of sucrose was assayed by estimating the free glucose according to the method of Nelson (17).
Enzyme-linked Lectin Absorbent Assay-The sugar binding properties of ASAs were studied in detail using a modified enzyme-linked immunosorbent assay technique (Scheme 1). An optimum concentration of the lectin was coated on enzyme-linked immunosorbent assay plates and left overnight at 4°C. After washing three times with 20 mM PBS, the wells were blocked with 3% bovine serum albumin for 1 h at room temperature. Subsequent to washing thrice with the blocking buffer, different sugars were added at varying concentrations and allowed to interact with the lectin for 1 h. A fixed concentration of invertase was added to each well and incubated for 30 min. The wells were then washed thrice with blocking buffer and once with PBS. An equal volume of 50 mM sucrose solution in 100 mM acetate buffer (pH 5.0) was added to each well. After incubation (for 1 h), glucose oxidase in the same buffer was then added to the solution to produce H 2 O 2 from the liberated glucose. H 2 O 2 thus generated was assayed using horseradish peroxidase and o-phenylenediamine (0.05 mg/well in 100 mM citrate buffer). The reaction was stopped by 2 N HCl, and the absorbance was recorded on an enzyme-linked immunosorbent assay reader at 490 nm.

Bulbs of A. sativum Contain a Group of Mannose-binding
Lectins-Earlier reports (3) have shown that dimeric proteins of 25 kDa occur in the bulbs of A. sativum. By using a modified purification procedure, we have identified an additional lectin designated as ASAIII. The affinity-purified preparation revealed three peaks upon gel filtration on a Bio-Gel P-200 column with molecular masses of 136, 48, and 25 kDa, respectively ( Fig. 1). Taken together, the data from SDS-PAGE ( Fig.  2) and gel filtration show that peak 3 is a heterodimer of 12.5and 11.5-kDa subunits, whereas peak 2 is most likely a heterotetramer made up of two pairs of 12.5-and 11.5-kDa polypeptide chains, although occurrence of these subunits in other proportions cannot be ruled out. SCHEME 1. Enzyme adsorption assay for the binding of sugars to the garlic lectins. OPD, orthophenylenediamine; HRP, horseradish peroxidase.
The SDS-PAGE profile (Fig. 2) of peak 1 indicates that it is composed of 54-, 12.5-, and 11.5-kDa subunits, whereas peaks 2 and 3 are made up of only 12.5-and 11.5-kDa subunits. All three peaks show distinct patterns of migration when subjected to free flow capillary electrophoresis (Fig. 3), suggesting that they are distinct proteins. Peak 1 contains 6% neutral sugar, whereas peaks 2 and 3 are devoid of neutral sugars. Sequence analysis of the individual bands of SDS-PAGE showed that the 54-kDa subunit has the amino-terminal sequence KMTWT-MKADEEA, which is different from the sequence of the 12.5and 11.5-kDa subunits (RNILTNDEGLYAGQSLD), common to all three peaks. Van Damme et al. (18) reported the same amino-terminal sequence for the 12.5-and 11.5-kDa polypeptide chains (RNLLTNGEGLYAGQS), whereas Smeets et al. (19), from the deduced amino acid sequences of cDNA clones, showed slightly different amino-terminal sequences for the 12.5-kDa (RNLLTNGEGLYAGQS) and 11.5-kDa (RNILRN-DEGLYAGQS) polypeptide chains. Altogether, these results show that the 54-kDa band corresponds to the enzyme alliinase (cysteine-sulfoxide lyase, alliin lyase, EC 4.4.1.4), which is known to form a complex with garlic lectins (20). Since the polypeptide chains corresponding to peaks 2 and 3 give the FIG. 1. Separation of ASAs from alliinase by gel filtration. a, the concentrated affinity-purified lectin preparation was applied to a Bio-Gel P-200 column (1.8 ϫ 110 cm) and eluted with 50 mM PBS. The Bio-Gel P-200 column was calibrated using rabbit IgG, which elutes at a molecular mass of 200 kDa on gel filtration (peak 1), V. villosa agglutinin (139 kDa; peak 2), soybean agglutinin (110 kDa; peak 3), bovine serum agglutinin (68 kDa; peak 4), hen egg ovalbumin (45 kDa; peak 5), and pepsin (34.7 kDa; peak 6). Myoglobin (17 kDa) was used for the determination of the inner volume of the column, and therefore, it was not used for constructing the calibration curve for estimating the molecular masses. b, shown is a calibration curve for the determination of the molecular masses of ASAI (peak 3) and ASAIII (peak 2) evaluated on a Bio-Gel P-60 column (2.8 ϫ 90 cm) calibrated with hen egg ovalbumin (45 kDa; ), chymotrypsinogen (25 kDa; ƒ), myoglobin (17 kDa; q) and bovine pancreas ribonuclease A (13.7 kDa; E). same sequence (RNILTNDEGLYAGQSLD), the lectins ASAI and ASAIII are made up of identical polypeptide chains that apparently differ at their carboxyl termini.
Hemagglutination Properties-The agglutinins ASAI and ASAIII interact strongly with rabbit erythrocytes, but considerably weakly with human erythrocytes, irrespective of their blood groupings.
ASAs Bind Most Avidly to Invertase-Hemagglutination inhibition studies of garlic agglutinins were carried out using a series of simple sugars and several glycoproteins. Among the monosaccharides tested, only methyl-␣-D-mannopyranoside was found to be inhibitory besides mannose, although the latter was less potent as an inhibitor (Table I). Glucose, a C-2 epimer of mannose, was inactive. Methyl-␣-D-glucopyranoside also did not interact. Replacement of the C-2 hydroxyl group of glucose with other groups did not alter its inhibitory property as both N-acetyl-D-glucosamine and glucosamine were inactive. Of all the glycoproteins used in this assay, invertase was the strongest inhibitor, and the minimum amount of this enzyme needed for complete inhibition was 0.7 nM (Table I). The invertase bound to both ASAI and ASAIII was found to retain its catalytic activity.
The strong affinity of these lectins for invertase, containing high mannose-type oligosaccharides, was instrumental in designing a sensitive enzyme-based assay to study the interaction of these lectins with various sugars (as shown in Scheme 1) and glycoproteins. The binding of invertase to varying amounts of garlic agglutinins coated on the wells of microtiter plates was checked (Fig. 4a). In other set of experiments, the amounts of horseradish peroxidase and glucose oxidase were varied, respectively, keeping the amounts of other ingredients fixed (Fig.  4, b and c). Based on these studies, the following concentrations were considered optimal for all subsequent sugar inhibition studies: 0.35 g of lectin, 20 ng of invertase, and 3 units of glucose oxidase for ASAI and 0.2 g of lectin, 20 ng of invertase, and 3 units of glucose oxidase for ASAIII. The amount of horseradish peroxidase was 10 ng and the concentration of sucrose was 50 mM in all experiments.
Inhibition of ASA-Invertase Binding by Mono-and Disaccharides-Sugar inhibition assays were carried out in triplicate, and each value is an average of three experiments. The amount of sugar required for 50% inhibition was calculated from complete inhibition curves (Fig. 5), and values are listed in Tables  I-III. Relative affinities of the lectins for different sugars were determined from the concentration of sugars required for 50% inhibition of the binding of invertase. Consistent with hemagglutination inhibition studies, methyl-␣-D-mannopyranoside was better as an inhibitor than mannose (Tables I and II). The presence of nonpolar p-nitrophenyl aglycon in D-mannose did not improve the binding affinity, but the introduction of a nonpolar 4-methylumbelliferyl aglycon at the anomeric position in ␣-linkage slightly enhanced its inhibitory potency. The lectins did not bind to 4-methylumbelliferyl-␤-mannopyrano-side, indicating that ␤-linked mannose was not conducive for binding. N-Acetyl-D-mannosamine was inactive, suggesting that the axially oriented hydroxyl group of mannose cannot be substituted with a bulky acetamido group. The monosaccharide binding propensities of ASAI and ASAIII are broadly similar to other well studied mannose-binding lectins from snowdrop (4) and daffodil and amaryllis (21). Of all the mannobioses tested, Man␣1-3Man was the most potent inhibitor. Its potency was 12 times greater than that of mannose. Man␣1-2Man and Man␣1-6Man were almost eight and six times more active, respectively, over mannose. But the other mannobioses, including Man␣1-4Man, were poor ligands. Among the other mannose-binding lectins, GNA recognizes only terminal Man␣1-3Man, whereas NPA (daffodil) and HHA (amaryllis) prefer ␣1-6-linked mannose. On the other hand, ConA is known to  4. Optimization of the components of the enzyme-based lectin absorbent assay. a, except for the lectin, all other parameters, viz. 20 ng of invertase, 3 units of glucose oxidase, 10 ng of horseradish peroxidase, and 50 mM sucrose, were constant. In b and c, the concentrations of horseradish peroxidase and glucose oxidase were varied, whereas the same amounts of invertase, sucrose, and lectin (0.35 g of ASAI/0.2 g of ASAIII) were used. The profiles presented in a-c were produced by ASAIII. ASAI also showed identical patterns. The experimental protocol is described under "Material and Methods." exhibit greater affinity for Man␣1-2Man (22). Artocarpin (A. integrifolia mannose-binding lectin) shows higher specificity for Man␣1-3Man (7). The ␤-linked disaccharides like Man␤1-4GlcNAc and GlcNAc␤1-2Man were poor inhibitors, whereas Man␤1-6GlcNAc and GlcNAc␤1-6Man were inactive toward ASAs. The lectins did not interact with disaccharides like lactose (Gal␤1-4Glc), N-acetyllactosamine (Gal␤1-4GlcNAc), and melibiose (Gal␣1-6Glc).
Binding Specificities of ASAI and ASAIII for Manno-oligosaccharides-To understand the carbohydrate specificities of ASAI and ASAIII in greater detail, their binding to a carefully chosen panel of manno-oligosaccharides was then undertaken (Table II). The core structure of N-linked oligosaccharides, Man␣1-3(Man␣1-6)Man, was 30 times stronger an inhibitor than mannose and showed four, two, and five times more potency than Man␣1-2Man, Man␣1-3Man, and Man␣1-6Man, respectively. The relative inhibitory potencies of mannopentaose and mannotriose were identical. Man 5 GlcNAc and Man␣1-3Man␤1-4GlcNAc were marginally better ligands than mannopentaose and Man␣1-3Man, respectively, unlike artocarpin, in which the addition of GlcNAc at the reducing end of Man␣1-3Man caused a dramatic enhancement of its binding ability (7).
When the ␣1-3and ␣1-6-linked mannose residues of mannotriose were substituted with GlcNAc as in GlcNAc 2 Man 3 , no remarkable change in activity was noted. In this regard, ASAI and ASAIII bear similarity to ConA, which displays equal affinities for mannotriose and GlcNAc 2 Man 3 (23-25), but for artocarpin, the inhibitory activity is reduced by 25-fold (7). Man 3 GlcNAc 2 has a 5-fold higher potency relative to mannotriose. Among the manno-oligosaccharides tested, Man 9 GlcNAc 2 -Asn is the best ligand, followed by Man 8 GlcNAc 2 Asn and Man 7 GlcNAc 2 Asn, indicating that the ␣1-2-linked mannosyl residues at the nonreducing end are highly preferred by the binding sites of these lectins. The preference of ASAI and ASAIII for the cluster of ␣1-2-linked mannosyl residues as in Man 9 GlcNAc 2 Asn is unique among the lectins described so far.
Interaction of ASAI and ASAIII with Glycoproteins-The ability of several mannose-containing glycoproteins to inhibit the binding of ASAI or ASAIII to invertase was checked to confirm their specificities. The extent of binding of these glycoproteins was qualitatively consistent with the relative affinities exhibited by manno-oligosaccharides.
The soybean lectin and ovalbumin, which bear several terminal Man␣1-2Man linkages in their high mannose-containing glycan chains, emerged as the best inhibitors (26). It has been reported that the sole glycan chain of the ovalbumin molecule is occupied by high mannose-or hybrid-type chains (27)(28)(29); as expected, it displayed a fairly good binding activity, unlike GNA, which did not interact with ovalbumin. 2 Some of the serum glycoproteins, viz. ␣ 1 -acid glycoprotein, fetuin, transferrin, and fibrinogen, were active, although the binding affinities were substantially weaker than that of soybean agglutinin. Binding of these glycoproteins in their native structure suggests that the lectins, like NPA and HHA, could recognize internal D-mannosyl residues. Alternatively, it can be said that these lectins can bind to the core pentasaccharide of N-linked glycans (Man␣1-3(Man␣1-6)Man␤1-4GlcNAc␤1-4GlcNAc) even when both of the ␣1-3and ␣1-6-mannosyl residues are substituted with NeuAc␣2-3/6Gal␤1-4GlcNAc. Removal of the terminal sialic acids improved the potency of these glycoproteins. The affinities of ASAI and ASAIII for jacalin and sheep IgG, which are substituted at their C-2 and C-4 hydroxyl groups of ␤-linked mannose by ␤-linked xylose and GlcNAc, respectively, are much weaker as compared with that of artocarpin. These results suggest that substitution with xylose in ␤1-2-linkage as in horseradish peroxidase (and jacalin) or bisection with GlcNAc in ␤1-4-linkage as in sheep IgG reduces their binding potencies for ASAI and ASAIII. Analyses of binding of saccharides derived from horseradish peroxidase indicate that the substitution of the ␤-linked mannose with xylose in ␤1-2-linkage compromises the binding potency to a greater extent than the substitution of reducing end GlcNAc with fucose in ␣1-3-linkage. Invertase carries Man 9 -20 GlcNAc in its seven accessible glycan chains (30). As a result, this glycoprotein binds to ASAI and ASAIII with such an avidity that its potency surpasses the activity of all the glycoproteins tested.

DISCUSSION
The high mannose-binding lectins (ASAI and ASAIII) from garlic bulbs were purified in two steps using affinity chromatography and gel filtration. Van Damme et al. (3) reported a mannose-binding lectin that resembles ASAI in its subunit composition. However, it differs from ASAI reported by them in displaying the same amino-terminal sequence for both of the polypeptide chains. ASAI and ASAIII are the heterodimer and heterotetramer, respectively, of similar polypeptide chains, present in equimolar proportion, that vary slightly in their molecular masses, viz. 12.5 and 11.5 kDa, yet the N-terminal sequences of their large (12.5 kDa) and smaller (11.5 kDa) subunits are identical. Despite the dissimilarities in their subunit compositions, the contours of the carbohydrate-binding sites of ASAI and ASAIII are identical. The biosynthetic and functional significance of the occurrence of lectins that differ in their subunit composition but display similar binding propensities within the same tissue is not presently understood. Our failure to isolate ASAII as a homodimer of 12-kDa subunits is perhaps related to differences in the purification methods used by us and by Van Damme et al. (3). It is also possible that our isolation procedure permits the purification of isolectins that have high propensities for interaction with alliinase (20). To ascertain their biological function and to develop them as potential tools for research, a knowledge of the carbohydrate specificities of these lectins becomes imperative. The sugar specificities of ASAI and ASAIII were elucidated in two steps: hemagglutination inhibition and a coupled enzyme-based assay. The hemagglutination inhibition study confirmed its exclusive specificity for D-mannose, like three other Amaryllidaceae lectins (GNA, NPA, and HHA) that do not recognize D-glucose (4,17). Our studies also show that these lectins display extraordinary avidity for invertase as compared with other glycoproteins tested. Based on this finding, a sensitive enzyme-based assay system was designed to investigate their detailed binding specificities.
The subunit molecular mass of one of the lectins isolated by Gupta and Sandhu (31) is similar to that of the larger subunit of peak 1 on the gel filtration column. However, unlike the alliinase-ASA complex or ASAI and ASAIII, the former is not retained on a mannose-Sepharose matrix. In the absence of the N-terminal sequence and comprehensive sugar binding properties of the high molecular mass lectin reported in Ref. 31, it is not possible to conclude that it is derived from another isolectin rather than being a glycoprotein contaminant, viz. alliinase. Garlic plant contains at least five different lectins and lectin genes (32). The processing and post-translational modifications of the primary translation products of monocot mannose-binding lectins are rather complex as evident from the report of several lectins with different molecular masses (33).  Compared with other mannose-binding lectins, ASAI and ASAIII bind to mannose very weakly. Methyl-␣-D-mannopyranoside is six times better an inhibitor than mannose. The relative potencies of D-mannose and its epimers suggest that the equatorial orientation of the hydroxyl groups at C-3, C-4, and C-6 and an axial hydroxyl group at C-2 as in mannose are necessary for interaction with these lectins. Compared with other mannose-binding lectins, some of the mannobioses show much higher potency than mannose. Of all the mannobioses tested, Man␣1-3Man exhibited the highest affinity as found by Kaku et al. (34), but unlike the present result, they recorded a lower potency of trimannoside than Man␣1-3Man. The difference might be attributed to the different techniques used. (Although assay-dependent differences in binding affinities are not common, they have, however, been seen in some instances. For example, GNA shows altered affinities for murine IgM under different assay conditions such as in a precipitation assay and in immobilized form (34).) The observation that ASAs are complementary to ␣1-3-mannosyl units merits reconsideration as (i) extension of ␣1-3Man as in Man␣1-3Man␣1-3Man␣1-3Man␣1-2Man-ol diminishes its binding ability (34); (ii) the affinity increases in the order of Man␣1-3Man Ͻ mannotriose Ϸ mannopentaose Ͻ Ͻ Man 9 -oligosaccharides; and (iii) it may not be an ␣1-3-linked mannopyranosyl residue, but the number of residues in ␣1-2-linkage at the nonreducing end as discussed subsequently that determine higher affinity of manno-oligosaccharides. This is evident by the dramatic increase in the potency of oligosaccharides that carry increasing numbers of ␣1-2-linked mannose residues. Unlike ASAI and ASAIII, mannotriose is the most complementary inhibitor for ConA, whereas artocarpin displays only slightly higher affinity for mannopentaose over mannotriose. When the reducing end of Man␣1-3Man and mannopentaose was substituted with a GlcNAc residue, no appreciable change in activity was observed. For artocarpin, the same substitution led to an enhancement in potency by severalfold (7). The reducing end GlcNAc in ASAs is probably accommodated adjacent to the primary binding site without involving any significant interactions with the lectin. When the terminal mannose residues of mannotriose are substituted with two GlcNAc residues as in GlcNAc 2 Man 3 , it leads to some improvement in activity, suggesting that the combining site of ASAs can access the masked/internal mannose residues of the oligosaccharides. On the other hand, the addition of two GlcNAc residues in ␤1-4-linkage at the reducing end of mannotriose (i.e. the core pentasaccharide of N-linked glycans) improves the affinity by five to six times. These observations are in accordance with the speculation, made by Barre et al. (33), that the mannose-bind-ing site of the monocot lectins is part of a more extended site, which explains the stronger binding of ASAI and ASAIII to complex glycans. Interestingly enough, the extension of the core trimannosidic structure by ␣1-2-linked mannosyl residues as in the glycopeptides of quail ovalbumin and soybean lectin leads to increased potencies. This enhancement also highlights the preference of garlic lectins for a cluster of ␣1-2-linked mannose residues.
The mode of interaction of ASAs with the glycoproteins studied substantiates the above findings. The ability to recognize internal mannose and the affinity for core pentasaccharide were, once again, proved through the interaction of ASAs with some glycoproteins. It appears that the binding site can withstand the terminal sialic acids as well as the penultimate galactose residues, although the removal of sialic acids enhances potency. Sheep IgG and jacalin displayed moderate binding potencies compared with other glycoproteins. Since the substitution of reducing end GlcNAc with ␣1-6-linked fucose as in Man 3 GlcNAc 2 Fuc is impervious to binding, the relatively weak interaction of jacalin, horseradish peroxidase, and its oligosaccharide as well as goat IgG appears to be due to the substitution of the ␤-linked 3,6-disubstituted mannose with xylose in ␣1-2-linkage (as in jacalin and horseradish peroxidase) and its substitution with GlcNAc in ␤1-4-linkage (as in sheep IgG). Using a different protocol (surface plasmon resonance analysis), Barre et al. (33) found that immobilized fetuin and asialofetuin do not bind to garlic lectins (viz. ASAI and ASAII). But hemagglutination and the enzyme-based assay reported here show that garlic lectins (ASAI and ASAIII) can bind to both fetuin and asialofetuin. Surface plasmon resonance analysis in the concentration range used appears to have failed to detect this interaction because of their moderate affinities. Notwithstanding the identity of the polypeptide chain in the lectin preparation reported by Gupta and Sandhu (31), they had purified the garlic lectin(s) using an asialofetuin-silica affinity column, confirming that they indeed are able to bind glycoproteins such as asialofetuin.
Invertase was the strongest glycoprotein ligand. This is attributed to the presence of several ␣1-2-linked mannose residues. Invertase contains nine N-linked high mannose oligosaccharides, seven of which are accessible. About 85% of the mannose of the accessible oligosaccharides is in species larger than Man 20 GlcNAc (30). We believe that the availability of these high mannose oligosaccharides with several ␣1-2-linked mannose residues at their nonreducing ends is instrumental for the observed potency of ASAs. This is also borne out by the inhibitory activities of the glycopeptide from soybean agglutinin. From molecular modeling studies, it is suggested (33) that the garlic lectins, like their counterparts in the monocot mannose-binding lectin family, possess three identical mannosebinding sites per monomer and that the mannose-binding site is part of a more extended site. As a result, these lectins are expected to accommodate a larger number of mannose residues. This explains, at least in part, (i) the comparatively low affinity for the monosaccharide (mannose), (ii) the increased affinity with increasing numbers of mannose residues, and (iii) the enhanced avidity for high mannose-containing oligosaccharides/glycoproteins. Although there is no report on the interaction of high mannose-containing glycoproteins, such as invertase, with other members of this lectin family, at least one member (bulb lectin from Allium cepa) demonstrates identical affinity for invertase. 3 The interaction of high mannose-containing oligosaccharides/glycoproteins and other larger glycans 3 T. K. Dam and A. Surolia, unpublished observation. with the members of the monocot mannose-binding lectin family constitutes a novel specificity among lectins studied to date. If compared with the mannose-binding (GNA, NPA, and HHA) and mannose/glucose-binding (ConA and artocarpin) lectins, the topology of the binding site(s) of ASAs would appear quite distinct. GNA recognizes only terminal ␣1-3-linked mannose residues. NPA and HHA interact with both the terminal and internal mannosyl residues, but the best inhibitors of NPA and HHA are ␣1-6-linked mannotrioses and oligosaccharides with ␣1-3or ␣1-6-mannose residues, respectively. ConA displays high affinities for oligosaccharides containing ␣1-2-linked mannose, but its binding site is most complementary to mannotriose. Artocarpin does not recognize ␣1-2-linked high manno-oligosaccharides and is most complementary to Man 3 -GlcNAc 2 Fuc containing a xylose ␤1-2-linked to the 3,6-disubstituted core mannose. In conclusion, our studies illustrate that the exquisite specificity of lectin-glycoprotein enzyme interaction when coupled with the catalytic power of an enzyme provides a simple and sensitive method for elucidating the carbohydrate recognition propensities of lectins. The ability of ASAs to bind high mannose-containing oligosaccharides and glycoprotein with enhanced potencies places them in a unique position among the mannose-binding lectins reported so far. This specificity of ASAs can be utilized for several biochemical studies, viz. biosynthesis and functional aspects of high mannose oligosaccharides and purification of high mannose-containing glycoproteins like invertase and carboxypeptidase Y.