Abstract

Multivalency in lectins is a phenomenon that has been discussed at considerable length. The structural basis for the role of multivalency in garlic lectin has been investigated here through computational studies. Biochemical studies have shown that the binding affinity of garlic lectin for high mannose oligosaccharides is orders of magnitude greater than that for mannose. Modeling and energy calculations clearly indicate that such increase in affinity cannot be accounted for by binding of these oligosaccharides at any of the six sites of a garlic lectin dimer. These studies also indicate that a given oligosaccharide cannot bind simultaneously to more than one binding site on a lectin dimer. The possibility of a given oligosaccharide simultaneously binding to and hence linking two or more lectin molecules was therefore explored. This study showed that trimannosides and higher oligomers can cross-link lectin dimers, amplifying the protein–oligosaccharide interactions severalfold, thus explaining the role of multivalency in enhancing affinity. A comprehensive exploration of all possible cross-links posed a formidable computational problem. Even a partial exploration involving a carefully chosen region of the conformational space clearly showed that a given dimer pair can be cross-linked not only by a single oligosaccharide molecule but also simultaneously by two oligosaccharides. The number of such possible double cross-links, including those forming interesting tetrameric structures, generally increases with the size of the oligosaccharide, correlating with the biochemical data. In addition to their immediate relevance to garlic lectin, these studies are of general interest in relation to lectin–oligosaccharide interactions.

Introduction

Lectins are carbohydrate-binding proteins that specifically recognize diverse sugar structures and mediate a variety of biological processes, such as cell–cell and host–pathogen interactions, serum glycoprotein turnover, and trafficking as well as in immune responses (Loris, 2002; Vijayan and Chandra, 1999). The exact nature and mechanisms of the function of lectins in these cellular processes still remains sketchy. Several lectins have been identified and characterized from plants and animals (Lis and Sharon, 1998; Bettler et al., 2003). Almost all our understanding of these molecules at the structural level have stemmed from X-ray crystallography studies of apo proteins and their complexes with monosaccharides or other simple sugars. Their larger function, however, is due to their ability to bind to complex sugars and glycoproteins, present often at the cell surfaces. Binding to the larger oligosaccharides, such as those that form part of the glycoproteins, have been determined biochemically by inhibition, surface plasmon resonance, and thermodynamic studies in many lectins (Barre et al., 1996; Dam et al., 1998; Bachhawat et al., 2001). Although the importance of multivalency in the function of lectins is well recognized (Weis and Drickamer, 1996), its structural basis is only beginning to be unraveled (Brewer, 2002; Sachhetini et al., 2001; Lee and Lee 2000). In this study we have used molecular modeling approaches to explore multivalency, employing garlic lectin.

Garlic lectin belongs to a well-conserved family of bulb lectins that are found in monocotyledonous families, such as Amaryllidaceae, Alliaceae, Araceae, Liliaceae, and Orchidaceae (Barre et al., 1996). The structures of several members of this group are now available, all belonging to the β-prism-II fold comprising three antiparallel four-stranded β-sheets arranged as a 12-stranded β-barrel. The bulb lectins, owing to their high mannose specificity, exhibit unique biological properties, such as selective inhibitory activity against HIV and other retroviruses and selective agglutination of rabbit but not human erythrocytes (Balzarini et al., 1991; Pusztai et al., 1993). These properties, which may lead to several useful applications, are presumably due to their ability to bind glycoproteins containing oligomannosides (Hester and Wright, 1996; Kaku et al., 1991). Although all bulb lectins have exclusive specificity toward mannose, their agglutination specificities and hence their biological roles vary. For example, lectins from Amaryllidaceae and Orchidaceae families (such as snowdrop lectin) are potent inhibitors of HIV due to their ability to bind to glyco protein 120, whereas garlic lectin, which has no detectable antiretroviral activity, can bind to high-mannose glycoproteins, such as invertase and alliinase with very high affinity (Barre et al., 1996; Dam et al., 1998).

The structural basis for these differences in binding specificities will provide a framework to study the molecular mechanisms of the larger functional roles of lectins. Complexes of lectins with large oligosaccharides are not easily amenable to X-ray crystallography, due to the inherent difficulties in crystallizing them, thus making it important to study them by computational methods. Here we seek to study the structures of complexes of garlic lectin with di-, tri- penta-, hepta-, octa-, and nonamannosides by a molecular modeling approach to explain its very high affinity to oligosaccharides exhibiting high mannose structures, specificities for the linkages in these complex sugars as well as the role of multiple binding sites in the lectin.

Results

Overview of the binding sites

Each subunit in the dimeric garlic lectin has three mannose-binding sites at locations similar to those observed in other bulb lectins (Chandra et al., 1999; Ramachandriah et al., 2002). These locations—E, F, G and I, J, K—are illustrated in Figure 1. The seventh, asymmetric location observed only in garlic lectin was ignored. Lectin–mannose interactions are similar at the six sites and involve a highly conserved sequence motif QXDXNXVXY (Ramachandraiah and Chandra, 2000). The O2 hydroxyl, which differentiates mannose from galactose and glucose is involved in strong hydrogen bonds with the second and the third consensus residues of the motif. Biochemical data from a systematic study, carried out to determine the types of oligosaccharides that would be recognized by the lectin, are available. This amounts to quantitative data on 17 sugars, as assayed by their ability to inhibit binding of Allium sativum agglutinin invertase (Dam et al., 1998). Data from an independent study using surface plasmon resonance is also available for many of these oligosaccharides. (Bachhawat et al., 2001). Of these, 9 representative sugars, which among them cover the mannosyl component of the 17, were used in the present analysis. The models of these nine sugars obtained using the program SWEET as described in the Materials and methods section, are shown in Figure 2 along with their schematic representations. The biochemical data show that the high-mannose oligosaccharides have a much greater affinity for the lectin, reaching up to a 14,000-fold increase for a nonamannoside over that of mannose.

Fig. 1.

A ribbon diagram of the crystal structure of garlic lectin (PDB: 1KJ1) showing the six mannose-binding sites in a dimer. E, F, and G correspond to the first, second, and third sites in the first subunit; I, J, and K refer to the three binding sites in the second subunit. The mannose molecules are shown in a CPK representation. This figure and Figures 4 and 5 are prepared using MOLSCRIPT (Kraulis, 1991).

Fig. 1.

A ribbon diagram of the crystal structure of garlic lectin (PDB: 1KJ1) showing the six mannose-binding sites in a dimer. E, F, and G correspond to the first, second, and third sites in the first subunit; I, J, and K refer to the three binding sites in the second subunit. The mannose molecules are shown in a CPK representation. This figure and Figures 4 and 5 are prepared using MOLSCRIPT (Kraulis, 1991).

Fig. 2.

The structures of the nine oligosaccharides constructed using SWEET. Schematic representations of each of them are also shown next to the structure. The nomenclature used in the text for the mannose residues is indicated in the diagram. Values in mM indicate the affinity of the sugar to garlic lectin as determined by an Allium sativum agglutinin invertase binding inhibition assay (Dam et al., 1998). Values in parentheses refer to the enhancement in affinity as compared to mannose.

Fig. 2.

The structures of the nine oligosaccharides constructed using SWEET. Schematic representations of each of them are also shown next to the structure. The nomenclature used in the text for the mannose residues is indicated in the diagram. Values in mM indicate the affinity of the sugar to garlic lectin as determined by an Allium sativum agglutinin invertase binding inhibition assay (Dam et al., 1998). Values in parentheses refer to the enhancement in affinity as compared to mannose.

Linkages recognized by the lectin

The first question to be addressed was concerned with the possible linkages from the mannose at the primary sites. All the hydroxyls, except that involving O1, are involved in hydrogen bonds with the lectin. Each hydroxyl group of mannose was methylated and their interactions with the lectin assessed. These calculations showed that methylation of the hydroxyl, and therefore any substitution in general at positions 2 and 4, led to severe unacceptable steric contacts. O1, on the contrary, interacts only with water. Thus this anomeric oxygen is the only reasonable candidate for linkage from mannose at the six crystallographically identified binding sites. This finding implies that oligosaccharides are likely to bind at any of the primary binding sites of the lectin, through their nonreducing ends rather than their reducing ends.

A systematic study was carried out to demonstrate that the inference from the methylated mannose studies holds good for oligosaccharides as well. It was realized that the procedure used for this also helps in determining whether the sugars exhibit differences in affinity among the six binding sites on the garlic lectin dimer. Docking of the dimannosides Manα1-2Man, Manα1-3Man, Manα1-4Man, and Manα1-6Man into the six sites was therefore attempted. Starting with the nonreducing end, each of the two mannose residues was separately docked into each site by superposing the ring atoms of the model on to the ring atoms of the crystallographically observed mannose molecule. Interactions with the protein were assessed at each stage both by graphics inspection as well as through measurement of hydrogen bonds and van der Waals interactions between the protein and the sugar. In each case, docking of the second residue into the primary site led to unacceptable steric clashes. The nonreducing residues of the dimannosides could be successfully docked into all the six sites except in the case of Manα1-4Man, which could be docked into only four sites. Docking of even the first residue of this dimannoside into sites E and I led to unacceptable steric clashes. However, energy minimization (see later discussion) led to acceptable docking at these sites as well. Thus 1–2, 1–3, 1–4, and 1–6 linkages with the nonreducing ends overlaying with the crystallographically identified mannoses, in all the six sites, is possible, at least at the disaccharide level, although the docking of Manα1-4Man at two of the sites was somewhat contrived. Therefore all the terminal mannosyl residues in the nine sugars considered are candidates to be docked at the binding sites. Also, binding at any site appeared to be completely independent of binding at other sites. During the procedure, the conformation of the oligosaccharides and the influence of conformational changes on the binding strengths were also evaluated. A range of 40° around each dihedral angle specified by SWEET for each sugar was searched at intervals of 5°. The systematic conformational search varying the dihedral angles clearly indicated that such changes did not significantly alter binding in any of the cases.

Lectin–carbohydrate complexes

Next the construction of complexes with all the nine oligosaccharides was attempted. Altogether 18 models were constructed with the same sugar docked in the same way in all the six binding sites. In view of the calculations outlined, only the nonreducing ends were used for docking at the primary sites. Thus the first model involved Manα1-2Man attached to all the six binding sites with the first residue superposing on the crystallographically observed mannose molecule in each case. Three more lectin–disaccharide complexes were similarly constructed. Two models could be constructed using the trisaccharide, one with one of the terminal mannosyl residue at the six primary sites and the other with the other terminal mannosyl residue at the sites. Likewise, three models each could be constructed with the penta-, hepta-, octa-, and the nona-oligosaccharides, because each of these sugars has three nonreducing mannosyl residues.

Energy minimization was carried out on all the 18 models. The root mean square (RMS) deviations in Cα positions between the original and the minimized models varied between 0.36 Å in the case of a complex with a disaccharide and 0.99 Å in the case of a complex with an oligosaccharide involving nine mannose residues. These values compare well with the RMS deviations in Cα positions between the two monomers in the original crystallographic model (0.85 Å). The RMS deviations when the monomers in the refined models of the dimers were superposed were between 0.86 to 0.90 Å in most cases. The values exceeded 1 Å (1.03–1.41 Å) in the case of three complexes with oligosaccharides containing eight or nine mannose residues. Similar RMS deviations were exhibited by the atoms in binding site residues and the sugar, when the original and the minimized models were compared. RMS deviations involving these atoms in the initial models indicated sites E and I to be most symmetric (0.61–0.72 Å) and sites G and K the least symmetric (1.34–1.77 Å), with respect to the noncrystallographic dyad that relates the two monomers in the lectin dimer. The same trend, though with somewhat higher RMS deviations, is maintained in the energy minimized models as well. The values cited indicate that the minimization had proceeded sensibly.

The interactions at each of the six sites were separately examined in all the models. Energy was calculated individually for the interaction of each carbohydrate molecule. Also calculated in each case were the surface area buried on complexation and shape complementarity of their surfaces. Unacceptable steric clashes in a few instances expressed themselves as high interaction energies, and such instances could be easily identified. The interaction energy, hydrophobic surface area buried on complexation and shape complementarity for the complexation of each carbohydrate molecule with the different sites on the lectin are listed in Table I.

Table I.

Interaction indices depicting the maximum and the minimum values in the range, observed for binding of the different mannosides to each of the six binding sites of the garlic lectin dimer

No. of Man units Residues in the binding pocket Interaction energy (k/Cal/mole) Buried surface area (Å2)
 
 Shape complementarity 

 

 

 
Total
 
Nonpolar
 

 
M1 −13……−24 358.…….267 145……103 0.84.…….0.73 
M1(1–2 linkage) −15……−24 410.…….286 159……..98 0.80.…….0.73 
M1(1–3 linkage) −13……−28 426.…….288 152……..82 0.81.…….0.73 
M1(1–4 linkage) −16……−22 398.…….303 142……..93 0.80.…….0.71 
M1(1–6 linkage) −13……−23 399.…….276 147……..91 0.82.…….0.73 
M1 −13……−23 397.…….274 146……..89 0.83.…….0.72 
M3 −15……−29 463.…….296 173……..93 0.83.…….0.72 
M1 −14……−24 427.…….278 146……..95 0.85.…….0.75 
M3 −16……−31 593.…….414 219…….119 0.84.…….0.64 
M5 −16……−2 555.…….374 222…….150 0.81.…….0.72 
M1 −16……−28 660.…….395 265…….151 0.76.…….0.70 
M4 −20……−38 778.…….561 240…….188 0.78.…….0.66 
M7 757……−37 698.…….545 282…….198 0.76.…….0.54 
M1 −16……−28 659.…….395 263…….151 0.76.…….0.70 
M4 1255……−9 943.…….571 366…….180 0.76.…….0.47 
M8 −33……−42 932.…….749 448…….307 0.78.…….0.61 
M1 −16……−30 678.…….418 246…….147 0.85.…….0.72 
M5 −23……−38 727.…….586 253…….215 0.79.…….0.65 
M9 2602……−43 1025.…….927 450…….400 0.73.…….0.43 
No. of Man units Residues in the binding pocket Interaction energy (k/Cal/mole) Buried surface area (Å2)
 
 Shape complementarity 

 

 

 
Total
 
Nonpolar
 

 
M1 −13……−24 358.…….267 145……103 0.84.…….0.73 
M1(1–2 linkage) −15……−24 410.…….286 159……..98 0.80.…….0.73 
M1(1–3 linkage) −13……−28 426.…….288 152……..82 0.81.…….0.73 
M1(1–4 linkage) −16……−22 398.…….303 142……..93 0.80.…….0.71 
M1(1–6 linkage) −13……−23 399.…….276 147……..91 0.82.…….0.73 
M1 −13……−23 397.…….274 146……..89 0.83.…….0.72 
M3 −15……−29 463.…….296 173……..93 0.83.…….0.72 
M1 −14……−24 427.…….278 146……..95 0.85.…….0.75 
M3 −16……−31 593.…….414 219…….119 0.84.…….0.64 
M5 −16……−2 555.…….374 222…….150 0.81.…….0.72 
M1 −16……−28 660.…….395 265…….151 0.76.…….0.70 
M4 −20……−38 778.…….561 240…….188 0.78.…….0.66 
M7 757……−37 698.…….545 282…….198 0.76.…….0.54 
M1 −16……−28 659.…….395 263…….151 0.76.…….0.70 
M4 1255……−9 943.…….571 366…….180 0.76.…….0.47 
M8 −33……−42 932.…….749 448…….307 0.78.…….0.61 
M1 −16……−30 678.…….418 246…….147 0.85.…….0.72 
M5 −23……−38 727.…….586 253…….215 0.79.…….0.65 
M9 2602……−43 1025.…….927 450…….400 0.73.…….0.43 

Binding of each sugar through each of its nonreducing ends has been explored. The first column indicates the number of mannose residues in the sugar, thus referring to the type of oligosaccharide. The second column indicates the particular nonreducing mannose through which the oligosaccharide is bound to the lectin.

The possibility of a given oligosaccharide simultaneously binding to more than one site on the same lectin dimer was also explored. The distances involving pairs of the six binding sites in the lectin varied between about 22 and 52 Å. However, the maximum distance between any two terminal mannose residues in the oligosaccharides considered was about 17 Å. Therefore this possibility could be easily ruled out. This also suggests that binding of the sugars at one site is completely independent of the binding at other sites.

Linked lectin molecules: exploratory effort

Singly linked pairs of lectin molecules

The next obvious step was to explore the possibility of a given oligosaccharide simultaneously binding to and hence linking two or more lectin molecules. Considering that each dimer has six binding sites and each carbohydrate molecule has one to three terminal mannosyl residues capable of binding to these sites, the number of possible distinct pairs or groups of linked lectin molecules becomes large. The possibility, even in principle, of linking three molecules exists only in the case of the pentasaccharide and the higher oligomers. Careful visual examination showed that the docking of any of the oligosaccharides simultaneously to three lectin molecules in any manner would lead to severe steric clashes, which could not be expected to be relieved through energy minimization. Therefore only linked lectin pairs were considered.

The disaccharides have only one nonreducing end each. Therefore they cannot link two lectin molecules. In the trisaccharide–lectin complex, in which a trisaccharide is bound to each of the binding sites, there are six free terminal residues, one belonging to each of the six trisaccharides bound to the lectin, which are capable of docking into one of the six binding sites in a neighboring lectin dimer. In each of the complexes involving higher oligosaccharides, there are 12 such free terminal residues. Each of them can in principle dock into any one of the six sites in another dimer. In practice, all the possibilities are not realized owing to steric clashes. All the possible linked dimers were generated by systematically superposing, one by one, each of the terminal residues in each energy-minimized complex onto the bound mannose in each of the six binding sites in another lectin dimer. Of the 72 (or 36 in the case of the trisaccharide) linked dimers—those involving more than 20 steric clashes with an interatomic distance of 2.3 Å or less—were rejected as sterically unacceptable. This criterion is arbitrary, but appeared reasonable. This resulted in the singly linked dimers listed in Table II. In Table II, each pair is identified in terms of the terminal residues involved and the binding sites used. For example, among the pairs linked by the pentasaccharide, M1F–M3G refers to the linked pair in which the terminal mannose M1 occupies the binding site F in one dimer and the terminal mannose M3 occupies binding site G in other.

Table II.

Dimer pairs constructed using one energy-minimized conformation for each oligosaccharide

Trimannoside
 
Pentamannoside
 
Heptamannoside
 
Octamannoside
 
Nonamannoside
 
Singly linked pairsa     
M1F–M3* :6 M1F–M3* :6 M1E–M4* :6 M1F–M7I :1 Nil M1F–M5* :4 
M1G–M3* :6 M1G–M3* :6 M1F–M4* :3 M1G–M7I :1  M1G–M5* :5 
M1 J–M3* :6 M1 J–M3* :6 M1G–M4* :6 M1 J–M7I :1  M1 J–M5* :3 
M1K–M3* :6 M1K–M3* :6 M1I–M4* :5 M1K–M7I :1  M1K–M5* :4 
  M1 J–M4* :4   
  M1K–M4* :6   
Doubly linked pairsb     
M1E–M3 J + M1 J–M3E M1E–M5E + M1F–M5G M1E–M7E + M1 J–M4 J M1F–M8F + M1 J–M8 J M1E–M9E + M1 J–M5 J 
M1F–M3G + M1I–M3E M1E–M5G + M1F–M5F M1E–M7I + M1 J–M4F M1F–M8 J + M1 J–M8F M1E–M9I + M1 J–M5F 
M1F–M3K + M1I–M3I M1E–M5K + M1F–M5 J M1F–M4F + M1I–M7I  M1F–M5F + M1I–M9I 
 M1F–M3G + M1I–M3E M1F–M4 J + M1I–M7E  M1F–M5 J + M1I–M9E 
 M1F–M3K + M1I–M3I    
 M1I–M5E + M1 J–M5G    
 M1I–M5G + M1 J–M5F    
Trimannoside
 
Pentamannoside
 
Heptamannoside
 
Octamannoside
 
Nonamannoside
 
Singly linked pairsa     
M1F–M3* :6 M1F–M3* :6 M1E–M4* :6 M1F–M7I :1 Nil M1F–M5* :4 
M1G–M3* :6 M1G–M3* :6 M1F–M4* :3 M1G–M7I :1  M1G–M5* :5 
M1 J–M3* :6 M1 J–M3* :6 M1G–M4* :6 M1 J–M7I :1  M1 J–M5* :3 
M1K–M3* :6 M1K–M3* :6 M1I–M4* :5 M1K–M7I :1  M1K–M5* :4 
  M1 J–M4* :4   
  M1K–M4* :6   
Doubly linked pairsb     
M1E–M3 J + M1 J–M3E M1E–M5E + M1F–M5G M1E–M7E + M1 J–M4 J M1F–M8F + M1 J–M8 J M1E–M9E + M1 J–M5 J 
M1F–M3G + M1I–M3E M1E–M5G + M1F–M5F M1E–M7I + M1 J–M4F M1F–M8 J + M1 J–M8F M1E–M9I + M1 J–M5F 
M1F–M3K + M1I–M3I M1E–M5K + M1F–M5 J M1F–M4F + M1I–M7I  M1F–M5F + M1I–M9I 
 M1F–M3G + M1I–M3E M1F–M4 J + M1I–M7E  M1F–M5 J + M1I–M9E 
 M1F–M3K + M1I–M3I    
 M1I–M5E + M1 J–M5G    
 M1I–M5G + M1 J–M5F    
a

See text for nomenclature. Different types of cross-links are shown for each oligosaccharide. Also shown is the number of each type of cross-link that has been found possible. When an anchor residue can interact with more than one binding site on a lectin dimer, the fact is indicated by an asterisk. The number of possible links is also indicated in each case. For example, M1F–M4* :3 refers to links with M1 sugar bound to the first dimer at the F site and M4 sugar bound to any one of the three possible sites on the second dimer.

b

For example, M1E–M3 J + M1 J–M3E denotes M1 sugar of a trisaccharide bound to site E of the first dimer bound through its M3 residue to the second dimer at the J site, while a second tri-saccharide simultaneously binds through its M1 sugar to the first dimer at the J site and through its M3 residue to the E site of the second dimer.

Doubly linked pairs of lectin molecules

In the course of the modeling, it became apparent that there could be doubly linked pairs as well. Each of the bound oligosaccharides with five or more residues, presents two terminal residues capable of binding other lectin molecules. Thus the number of possible pairs for double linkage is 12C2. As there are six binding sites in the second dimer also, the number of ways these sites can be used for forming two links at a time is 6P2. Thus 12C2×6P2 doubly linked pairs each are in principle possible in the case of the penta-, hepta-, octa-, and nonasaccharides. This number is 450 in the case of the trisaccharide. The number in each case is therefore very large.

The intersite distances (Sij), measured as the distance between the centers of the bound mannose rings, were calculated. Only those pairs (Figure 3) of terminal residues in the complex in which the distance between centers of the two rings (Tij) lies within ± 3 Å of any of the Sij's were retained for further calculations. This distance filter reduced the number of pairs of terminal residues to be considered for double linkages. For example, in the case of the nonasaccharide, the number came down from 1980 to 188. To each possible pair, a lectin dimer was docked such that the least squares deviation between the 12 ring atoms in the pair of terminal residues and those in the bound mannoses in the 2 concerned sites in the incoming lectin dimer is a minimum. Among the double-linked dimers thus constructed, those with more than 40 unacceptable contacts (interatomic distance 2.3 Å or less) were rejected. The remaining possible double linked dimers are listed in Table II.

Fig. 3.

Schematic representation of interterminal (Tij) and intersite (Sij) distances employed in exploring double linkages between pairs of lectin dimers (shown as shaded triangles). E, F, G, I, J, and K refer to the six binding sites on the garlic lectin dimer, shown in detail in Figure 1.

Fig. 3.

Schematic representation of interterminal (Tij) and intersite (Sij) distances employed in exploring double linkages between pairs of lectin dimers (shown as shaded triangles). E, F, G, I, J, and K refer to the six binding sites on the garlic lectin dimer, shown in detail in Figure 1.

It may be mentioned that somewhat different strategies have been used in the construction of singly linked pairs and doubly linked pairs. Therefore, contrary to what one might expect, the individual links in the double links are not always present in the list of single links, in the case of every oligosaccharide. In the case of singly linked pairs, exact superposition of the ring in the terminal mannose of the oligosaccharide complex and the bound mannose in the second lectin dimer was used; it was hoped that the rather liberal criterion employed for steric acceptability would take care of the intrinsic flexibility of the system. In the case of double linkages, where the conformational flexibility of two oligosaccharides is simultaneously involved, it was felt that the requirement of exact superposition of sugar rings would be too restrictive. Therefore, double-linked lectin pairs in which the concerned Tij is within ± 3 Å of an Sij were accepted as possible, if the criterion based on steric contacts was also satisfied.

Further calculations on linked lectin molecules

One anomaly in the calculations using a single, energy-minimized conformation for each oligosaccharde was immediately obvious. Several possible singly linked lectin pairs involving the heptasaccharide could be identified (Table II). The same was true with those involving the nonasaccharide, too. However, none of them could be identified when the octasaccharide was used as the link. This appeared strange as the heptasaccharide could be obtained by the deletion of a mannosyl residue at one end of the octasaccharide, whereas the nonasaccharide resulted when a mannosyl residue was added at another end of the octasaccharide. Careful examination of interactions involving the three oligosaccharides in their minimum energy conformations showed that this anomaly was due to a steric effect. Several linkages involving M1 and M4 of the heptasaccharides occur. However, when a mannosyl residue is added to M7, the added residue, in the minimum energy conformation, prevents the interaction of M4 with the second lectin in the pair. The nonasaccharide results from the addition of a mannosyl residue (M5) to M4. Now, an M1–M5 linkage is possible in spite of the presence of M9. It could also be seen that an M1–M4 linkage involving the octasaccharide becomes possible when the conformation of the M6-M7-M8 arm is varied.

This observation indicated that the conformational flexibility, although it did not matter much at the disaccharide level, needs to be taken into account to avoid misleading results on the higher oligosaccharides. It is reasonable to assign three possible values, one close to that in the energy-minimized conformation and two at 20° on either side of it, for each torsion angle. Two torsion angles are involved in defining the conformation about a 1–2 or 1–3 linkage. Three are involved when the linkage is 1–6. Then the number of conformers to be considered for the hepta-, the octa-, and the nonasaccharides work out to be 314 (∼5 million), 316 (∼43 million), and 318 (∼387 million), respectively. These are formidable numbers, and it is nearly impossible to explore all the conformers. A comparison of the results involving the trimer, the pentamer, and the heptamer indicated them to be apparently on expected lines. It was therefore decided to keep the conformation of the pentamer fixed as corresponding to the energy-minimized structure. The number of torsion angles to be varied in the heptamer, octamer, and nonamer then reduced to 4, 6, and 8, respectively, and the number of different conformers to be dealt with to 81, 729, and 6561, respectively.

The same method used in calculations with energy minimized oligosaccharide structures, was used for searching pairs of lectins linked by the different conformers. The single linkages found possible with the heptamer, octamer, and nonamer are given in Table III. Understandably many more linkages than those listed in Table II could be identified when the conformational flexibility of the oligosaccharides was also partially taken into account. The same is true about double linkages (Table III). In general, several pairs of conformers can lead to a given double linkage.

Table III.

Dimer pairs with different sugars when partial conformational flexibility was introduced

Heptamannoside
 
Octamannoside
 
Nonamannoside
 
Singly linked pairs   
M1E–M4* :6 M1E–M4* :5 M1F–M5* :5 
M1F–M4* :4 M1F–M4* :3 M1G–M5* :6 
M1G–4* :6 M1G–M4* :5 M1 J–M5* :6 
M1I–M4* :6 M1I–M4* :5 M1K–M5* :6 
M1 J–M4* :6 M1 J–M4* :2 M1E–M9* :6 
M1K–M4* :6 M1K–M4* :5 M1F–M9* :6 
M1E–M7* :6 M1E–M8* :6 M1G–M9* :6 
M1F–M7* :6 M1F–M8* :6 M1I–M9* :6 
M1G–M7* :6 M1G–M8* :6 M1 J–M9* :6 
M1I–M7* :6 M1I–M8* :6 M1K–M9* :6 
M1 J–M7* :6 M1 J–M8* :6 M5E–M9* :6 
M1K–M7* :6 M1K–M8* :6 M5F–M9* :5 
  M5G–M9* :6 
  M5I–M9* :6 
  M5 J–M9* :6 
  M5K–M9* :6 
Doubly linked pairs   
M1E–M7* + M1 J–M4* :14 M1E–M8* + M1 J–M4* :2 M1E–M5* + M1 J–M5* :2 
M1E–M7* – M1 J–M7* :4 M18–M8* + M1 J–M8* :8 M1E–M9* + M1 J–M5* :5 
M1F-M4* + M1I-M7* :12 M1F–M4* + M1I–M8* :2 M1E–M9* + M1 J–M9* :8 
M1F–M7* + M1I–M7* :4 M1F–M8* + M1I–M8* :6 M1F–M5* + M1I–M5* :2 
M1F–M7* + M1 J–M7* :2 M1F–M8* + M1 J–M8* :2 M1F–M5* + M1I–M9* :4 
  M1F–M5* + M1 J–M9* :3 
  M1F–M9* + M1I–M9* :10 
  M1F–M9* + M1 J–M9* :2 
  M1F–M9* + M1 J–M5* :2 
  M5E–M1* + M5G–M9* :14 
  M5F–M9* + M5G–M1* :4 
  M5I–M1* + M5K–M9* :6 
  M5 J–M9* + M5K–M1* :2 
Heptamannoside
 
Octamannoside
 
Nonamannoside
 
Singly linked pairs   
M1E–M4* :6 M1E–M4* :5 M1F–M5* :5 
M1F–M4* :4 M1F–M4* :3 M1G–M5* :6 
M1G–4* :6 M1G–M4* :5 M1 J–M5* :6 
M1I–M4* :6 M1I–M4* :5 M1K–M5* :6 
M1 J–M4* :6 M1 J–M4* :2 M1E–M9* :6 
M1K–M4* :6 M1K–M4* :5 M1F–M9* :6 
M1E–M7* :6 M1E–M8* :6 M1G–M9* :6 
M1F–M7* :6 M1F–M8* :6 M1I–M9* :6 
M1G–M7* :6 M1G–M8* :6 M1 J–M9* :6 
M1I–M7* :6 M1I–M8* :6 M1K–M9* :6 
M1 J–M7* :6 M1 J–M8* :6 M5E–M9* :6 
M1K–M7* :6 M1K–M8* :6 M5F–M9* :5 
  M5G–M9* :6 
  M5I–M9* :6 
  M5 J–M9* :6 
  M5K–M9* :6 
Doubly linked pairs   
M1E–M7* + M1 J–M4* :14 M1E–M8* + M1 J–M4* :2 M1E–M5* + M1 J–M5* :2 
M1E–M7* – M1 J–M7* :4 M18–M8* + M1 J–M8* :8 M1E–M9* + M1 J–M5* :5 
M1F-M4* + M1I-M7* :12 M1F–M4* + M1I–M8* :2 M1E–M9* + M1 J–M9* :8 
M1F–M7* + M1I–M7* :4 M1F–M8* + M1I–M8* :6 M1F–M5* + M1I–M5* :2 
M1F–M7* + M1 J–M7* :2 M1F–M8* + M1 J–M8* :2 M1F–M5* + M1I–M9* :4 
  M1F–M5* + M1 J–M9* :3 
  M1F–M9* + M1I–M9* :10 
  M1F–M9* + M1 J–M9* :2 
  M1F–M9* + M1 J–M5* :2 
  M5E–M1* + M5G–M9* :14 
  M5F–M9* + M5G–M1* :4 
  M5I–M1* + M5K–M9* :6 
  M5 J–M9* + M5K–M1* :2 

The nomenclature is similar to that in Table II.

Discussion

Thermodynamic and kinetic measurements in this laboratory have earlier demonstrated, among other things, a dramatic increase in the binding affinity of oligosaccharides to garlic lectin compared to that of the monomeric sugar (Dam et al., 1998). Selected dimannosides bind to the lectin with about eightfold affinity compared to that of mannose. The comparative affinity increases by about 30-fold in the case of tri- and pentamannosides. This enhancement of affinity becomes several thousandfold in the sugars containing seven or eight mannose residues and more than 10,000 in the oligosaccharides containing nine residues. The modeling studies described seek to provide a ready qualitative explanation for this nonlinear, dramatic enhancement of the affinity of the higher oligomers of mannose to the lectin.

As indicated earlier, the first attempt was to dock all the nine sugars individually to each of the six binding sites on the lectin. This led to 18 possible lectin–sugar complexes in which all the sites are occupied by the same sugar. These calculations showed that oligosaccharides bind to the protein through their nonreducing ends only. Energy minimization indicated docking at all the sites to be possible in most instances. The interaction energies and surface area buried on complexation were not dramatically higher in the case of higher oligomers. In fact these indices could not explain even the higher affinities of the lectin toward dimannosides compared to that for mannose. It may, however, be noted that not unexpectedly, the surface area buried on complexation (including the nonpolar component) generally increased with increased size of the oligosaccharide. As already mentioned, it was also ascertained that no sugar can simultaneously bind to two or more sites, as the distances between binding sites are higher than those between the terminal sugars in the oligosaccharides. It is in this context that the possibility of each sugar binding to more than one lectin molecule was systematically explored.

The pentasaccharide and higher oligomers have three terminal residues each, which can interact with the sugar binding sites in the lectin. M1 (Figure 2), which is involved in almost all cross-links, may be referred as the primary anchor in all cases, including the trisaccharide. M3 in the pentamer, M4 in the heptamer and the octamer, and M5 in the nonamer could be appropriately called the middle anchor. Geometrically, M3 in the trisaccharide corresponds to the middle anchor in the pentamer. M5 in the pentamer, M7 in the heptamer, M8 in the octamer, and M9 in the nonamer may be called the terminal anchors. This nomenclature will be used in the following discussion wherever appropriate.

The M1-M2-M3 fragment is common to the tri- and pentasaccharides. The torsion angles that define the conformation of this fragment are also very similar in the two sugars; the difference between the corresponding angles in the two cases is in the range of 2–23°. No single linkages involving M5 of the pentamer exist. Therefore, not surprisingly, the trimer and the pentamer are involved in exactly the same type of single linkages (Table II). Because the calculations used to explore double linkages do not require rigorous superposition of the anchor residues and the mannose molecules bound to the lectin, M5 also figure in such linkages. Two of the three double linkages generated by the trimer are produced by the pentamer also. In the case of the third double linkage, the distance between M1 and M3 in the pentasaccharide just falls outside the appropriate range on account of the slight difference in the conformation of M1-M2-M3 fragment between the trimer and the pentamer. Both tri- and pentasaccharides led to similar type of single cross-linked aggregates, consistent with similarities in their binding affinities.

Interestingly, all the linkages produced by the heptamer and the octamer involve M1. In the case of pairs connected by the nonamer, linkages involving the middle anchor and the terminal anchor also occur. These are the linkages that primarily contribute to the additional numbers in the pairs linked by the nonamer. The numbers directly indicate the potential for the formation of linked lectin molecules. The higher potential of the high-mannose oligosaccharides to form linked molecules as compared to the tri- and the pentamanno-oligosaccharides is reflected in the number of different types of aggregations found possible with it, again correlating with the orders of magnitude higher affinity for the lectin. Admittedly more linkages involving the trimer and the pentamer would have resulted if their conformational flexibility was taken into consideration. However, for reasons already mentioned, it was computationally impossible to carry out calculations for all the possible conformers. Thus, as far as the heptamer and the higher oligomers are concerned, the attempt is to identify linkages additional to those made by the pentamer, as the object of the exercise is comparison of the potential of different oligosaccharides to form single and double linkages. Table III clearly shows that the linkages formed by the heptamer and the higher oligomers would be very substantially higher than those formed by the pentamer for a given conformation of the pentasaccharide. In general, on account of the presence of multiple binding sites on the lectin, linked pairs can be successively linked to other pairs to produce large aggregates. Although no rigorous proof exists for a direct relation between ability to form linkages and affinity, it appears reasonable to expect the ability to form more linkages to result in larger aggregates. The differences in affinity appears to correlate, admittedly qualitatively, with the ability to form single and double linkages.

Typical singly and doubly linked pairs of lectins are illustrated in Figures 4 and 5. Of particular interest are symmetrically doubly linked dimers (Figure 5d), which appear as tetramers with 222 symmetry. “Tetramers” of this type are generated by hepta-, octa-, and nonasaccharides. Indeed modeling involving lower oligomers did not lead to such tetramers. In this type of aggregation, in addition to cross-linking of the two dimers through their F and J sites, the hepta-, octa-, and nona-oligosaccharides are also capable of forming single cross-linked aggregates through their G and K sites with two other protein dimers, which in turn are free to repeat the cross-linking pattern through their F and J sites to more number of protein dimers, thus resulting in the formation of a lattice. It is interesting to note that the second of the three binding sites (F and J sites, see Figure 1) is involved in each subunit in this type of arrangement (Figure 5d) leaving the third site (G and K sites) free for forming an additional single linkage with yet another lectin molecule per subunit. The symmetry in the interactions in the aggregate lends itself to the formation of homogenous aggregates or lattices (Figure 6). As a result, such lattices can be expected to have more intrinsic stability as compared to a heterogeneous aggregate formed by a combination of other linkages demonstrated here.

Fig. 4.

Typical dimer pairs singly linked by (a) tri-, (b) penta-, (c) hepta-, and (d) nonasaccharides. The type of link is mentioned below each figure. Sugars involved in the links are shown in CPK representation. Other sugar molecules bound to the lectin are also indicated. One subunit in each dimer is shown in black and the other is in gray.

Fig. 4.

Typical dimer pairs singly linked by (a) tri-, (b) penta-, (c) hepta-, and (d) nonasaccharides. The type of link is mentioned below each figure. Sugars involved in the links are shown in CPK representation. Other sugar molecules bound to the lectin are also indicated. One subunit in each dimer is shown in black and the other is in gray.

Fig. 5.

Typical dimer pairs doubly linked by (a) tri-, (b) penta-, (c) hepta-, and (d) octasaccharides. Conventions used in the figure are the same as those in Figure 4.

Fig. 5.

Typical dimer pairs doubly linked by (a) tri-, (b) penta-, (c) hepta-, and (d) octasaccharides. Conventions used in the figure are the same as those in Figure 4.

Fig. 6.

A lattice-like aggregation starting from the tetramer-like doubly linked pair, shown in Figure 5d. The tetramers are linked to each other through M1G–M9G single linkages. This figure is prepared using BOBSCRIPT (Esnouf, 1999).

Fig. 6.

A lattice-like aggregation starting from the tetramer-like doubly linked pair, shown in Figure 5d. The tetramers are linked to each other through M1G–M9G single linkages. This figure is prepared using BOBSCRIPT (Esnouf, 1999).

Identification of a particular type of lattice formation also forms a framework to design further experiments to determine the exact mechanism of achieving high affinities in garlic lectin. The oligosaccharides studied here all have α-1-2 glycosidic linkages at their anchor residues, as indicated in the schematic representation given in Figure 3. The particular type of tetrameric arrangement in the aggregate or the pattern observed in the lattice appears to be specific to 1–2 linkages at these positions. If the linkages were to be replaced by 1–3, 1–4, or 1–6 linkages, the conformations of the anchor sugars and their positions with respect to each other in an oligosaccharide molecule would vary significantly, thus making it impossible to bind to this type of protein aggregate.

The protein–carbohydrate interactions are relatively weak on the biological scale, with dissociation constants usually in the milli to micromolar range (Lee and Lee, 1995). To overcome the problem of weak affinity, nature appears to have evolved a mechanism of multivalency, in which multiple copies of both the oligosachhardies and the interacting segments of lectins are displayed on cell surfaces such that many weak interactions reinforce one another in a cumulative manner. Multivalency not only enhances the affinity or strength of binding but also amplifies binding selectivity or specificity, thus allowing nature to use subtle changes in oligosaccharide structures to create an impressive array of ligands for highly specific biological processes.

3D structures have been determined for more than 200 different lectins, which have provided detailed information about the structure, binding site and the mode of action of individual molecules of these lectins. Yet in most cases, they have not been sufficient by themselves to explain multivalency or generation of specificity in recognizing a wide variety of oligosaccharides. They do, however, provide excellent frameworks to integrate biochemical information through computational approaches and help in addressing some of the crucial issues, as illustrated in this article. Modeling of protein–carbohydrate cross-linked aggregates of garlic lectin reported here, clearly reveal the role of multivalency in increasing specificity to higher oligosaccharides. Formation of large aggregates through cross-linking has been studied earlier using physicochemical, electron microscopic, and X-ray techniques (Bhattacharyya et al., 1990; Lee et al., 1984, Lee and Lee, 2000; Brewer, 2002). The X-ray studies on soybean agglutinin complexed with bivalent pentasaccharides are particularly noteworthy (Olsen et al., 1997). The garlic lectin–oligosaccharide interactions, considered here, present a more complex case because the lectin molecule has six binding sites and the higher oligosaccharides are trivalent. It is perhaps for the first time such a complex set of lectin–carbohydrate interactions are sought to be studied through detailed rigorous modeling. Multivalency is a common and in fact characteristic feature of lectins. In addition to providing a qualitative explanation for the dramatic increase in the affinity of the lectin for higher oligosaccharides compared to that for lower ones, the study provides insights into the general problem of the multivalency of lectins.

Materials and methods

One of the two crystallographically independent dimers (A and D) in the crystal structure of garlic lectin–mannose complex (1KJ1) (Berman et al., 2000) determined by us recently (Chandra et al., 1999; Ramachandraiah et al., 2002) was used as the starting model to construct all oligosaccharide complexes. Structures of the oligosaccharides were obtained by using the program SWEET, available online at www.dkfz-heidelberg.de/spec (Bohne et al., 1998, 1999). This program builds carbohydrates based on the glycosidic linkages input by the user for each sugar residue and minimizes the whole structure, taking into account the range of flexibility allowed for each dihedral angle. Glycosidic torsion angles ϕ, ψ, and ω of disaccharides constructed in this manner were similar to the values reported in the crystal structures of disaccharides available in the Cambridge Structure Database (Allen and Kennard, 1993).

The carbohydrates were docked to the desired binding sites on the lectin molecule by superposing the ring of the appropriate residue on the ring of the mannose molecule bound to that site. A locally developed program using Kearsley's (1989) procedure based on Mackay's algorithm was used for this purpose. The model of methylated mannose was constructed using INSIGHT II (version 98) in calculations involving methylated mannose. The same program suite was used for search in dihedral space for visual examination or for ascertaining steric contacts.

Structures of lectin-oligosaccharide complexes were energy-minimized using X-PLOR (Brünger, 1992) with a distance-dependent dielectric constant. In each case 500 steps of minimization were carried out. The main chain atoms of the lectins were restrained with a force constant of 200 kCal/mol/Å2. One lectin–carbohydrate hydrogen bond each from the hydroxyl oxygens O2, O3, and O4 from the mannose residue at the primary site were restrained to lie between 2.4 and 3.6 Å with a force constant of 100 kCal/mol/Å2. When calculating interaction energies, amino acid residues with one or more atoms within a distance of 6 Å from any atom in the oligosaccharide were included. Buried surface areas were estimated using Lee and Richard's (1971) algorithm implemented in the NACCESS package (Hubbard and Thornton, 1993) with 1.4 Å as the radius of water molecule. The shape complementarity coefficient (Lawrence and Colman, 1993) was calculated using CCP4 (1994) with interface of 6Å. The coefficient can vary between 0 and 1, where 1 indicates ideal compatibility between two surfaces. Steric contacts were found using CCP4 (1994).

Different sugar linkages that would be accepted by the lectin for noncovalent binding, the residues within a oligosaccharide that would bind to the lectin, as well as aggregations involving single and double linkages between lectins and oligosaccharides, were explored using locally developed programs in a manner described at appropriate places.

1
To whom correspondence should be addressed; e-mail: mv@mbu.iisc.ernet.in

The computations and model building were carried out at the Super Computer Education and Research Centre at the Indian Institute of Science, the Bioinformatics Centre, and the Graphics facility supported by the Department of Biotechnology. G.R. is a CSIR senior research fellow. The work is supported by the Department of Science and Technology.

References

Allen, F.H. and Kennard, O. (
1993
) 3D search and research using the Cambridge structural database.
Chem. Des. Auto. News
 ,
8
(1),
31
–37.
Bachhawat, K., Thomas, C.J., Amutha, B., Krishnasastry, M.V., Khan, M.I., and Surolia, A. (
2001
) On the stringent requirement of mannosyl substitution in mannooligosaccharides for the recognition by garlic (Allium sativum) lectin.
J. Biol. Chem
 .,
276
,
5541
–5546.
Balzarini, J., Schols, D., Neyts, J., Van Damme, E.J.M., Peumans, W., and De Clerq, E. (
1991
) α(1,3) And α(1,6)-mannose specific plant lectins are markedly inhibitory to human immunodeficiency virus and cytomegalovirus infections in vitro.
Antimicrob. Agents Chemother.
 ,
35
,
410
–416.
Barre, A., Van Damme, E.J.M., Peumans, W.J., and Rouge, P. (
1996
) Structure-function relationship of monocot mannose-binding lectins.
Plant Physiol.
 ,
112
,
1531
–1540.
Berman, H.M., Westbrook, J., Feng, Z., Gilliland, G., Bhat, T.N., Weissig, H., Shindyalov, I.N., and Bourne, P.E. (
2000
) The protein data bank.
Nucleic Acids Res.
 ,
28
,
235
–242.
Bettler, E., Loris, R., and Imberty, A. (
2003
) Lectins database. Available online at http://webenligne.cermav.cnrs.fr/lectines. Accessed 18 August 2003.
Bhattacharyya, L., Fant, J., Lonn, H., and Brewer, C.F. (
1990
) Binding and precipitating activities of Lotus tetragonolobus isolectins with L-fucosyl oligosaccharides. Formation of unique homogeneous cross-linked lattices observed by electron microscopy.
Biochemistry
 ,
29
,
7523
–7530.
Bohne, A., Lang, E., and von der Leith, C.W. (
1998
) W3-SWEET: carbohydrate modeling by internet.
J. Mol. Model.
 ,
4
,
33
–43.
Bohne, A., Lang, E., and von der Leith, C.W. (
1999
) SWEET-WWW-based rapid 3D construction of oligo-and polysaccharides.
Bioinformatics
 ,
9
,
767
–768.
Brewer, C.F. (
2002
) Binding and cross-linking properties of galectins.
BBA
 ,
1572
,
255
–262.
Brünger, A.T. (
1992
) X-PLOR. Version 3.1. A System for X-ray Crystallography and NMR. Yale University, Connecticut.
Chandra, N.R., Ramachandraiah, G., Bachhawat, K., Dam, T.K., Surolia, A., and Vijayan, M. (
1999
) Crystal structure of a dimeric mannose-specific agglutinin from garlic: quaternary association and carbohydrate specificity.
J. Mol. Biol.
 ,
285
,
1157
–1168.
CCP4 (Collaborative Computational Project, No. 4.) (
1994
) The CCP4 suite: programs for protein crystallography.
Acta Cryst.
 ,
D50
,
760
–763.
Dam, T. K., Bachhawat, K., Geetha Rani, P., and Surolia, A. (
1998
) Garlic (Allium sativum) lectins bind to high mannose oligosaccharide chains.
J. Biol. Chem.
 ,
273
,
5528
–5535.
Esnouf, R.M. (
1999
) Further additions to MolScript version 1.4, including reading and contouring electron density maps.
Acta Cryst.
 ,
D55
,
938
–940.
Hester, G. and Wright, C.S. (
1996
) The mannose-specific bulb lectin from Galanthus nivalis (snowdrop) binds mono- and dimannosides at distinct sites. Structure analysis of refined complexes at 2.3 Å and 3.0 Å resolution.
J. Mol. Biol.
 ,
262
,
516
–531.
Hubbard, S.J. and Thornton, J.M. (
1993
) NACCESS. Computer program, Department of Biochemistry and Molecular Biology, University College, London.
Kaku, H., Goldstein, I.J., and Oscarson, S. (
1991
) Interaction of five D-mannose-specific lectins with a series of synthetic branched trisaccharides.
Carbohydr. Res.
 ,
213
,
109
–116.
Kearsley, S.K. (
1989
) On the orthogonal transformation used for structural comparisons.
Acta Cryst.
 ,
A45
,
208
–210.
Kraulis P. (
1991
) Molscript: a program to produce both detailed and schematic plots of protein structures.
J. App. Crystallogr.
 ,
24
,
946
–950.
Lawrence, M.C. and Colman, P.M. (
1993
) Shape complementarity at protein/protein interfaces.
J. Mol. Biol.
 ,
234
,
946
–950.
Lee, B. and Richards, F.M. (
1971
) The interpretation of protein structures; estimation of static accessibility.
J. Mol. Biol.
 ,
55
,
379
–400.
Lee, R.T. and Lee, Y.C. (
2000
) Affinity enhancement by multivalent lectin-carbohydrate interaction.
Glycoconj. J.
 ,
17
,
543
–551.
Lee, R.T., Lin, P., and Lee, Y.C. (
1984
) New synthetic cluster ligands for galactose/N-acetylgalactosamine-specific lectin of mammalian liver.
Biochemistry
 ,
23
,
4255
–4261.
Lee, Y.C. and Lee, R.T. (
1995
) Carbohydrate-protein interactions: basis of glycobiology.
Acc. Chem. Res.
 ,
28
,
321
–327.
Lis, H. and Sharon, N. (
1998
) Lectins: carbohydrate-specific proteins that mediate cellular recognition.
Chem. Rev.
 ,
98
,
637
–674.
Loris, R. (
2002
) Principles of structures of animal and plant lectins.
Biochim. Biophys. Acta
 ,
1572
,
198
–208.
Olsen, L.R., Dessen, A., Gupta, D., Sabesan, S., Sacchettini, J.C., and Brewer, C.F. (
1997
) X-ray crystallographic studies of unique cross-linked lattices between four isomeric biantennary oligosaccharides and soybean agglutinin.
Biochemistry
 ,
36
,
15073
–15080.
Pusztai, A., Grant, G., Spencer, R.J., Duguid, T.J., Brown, D.S., Ewen, S.W., Peumans, W.J., Van Damme, E.J., and Bardocz, S. (
1993
) Kidney bean lectin-induced Escherichia coli overgrowth in the small intestine is blocked by GNA, a mannose-specific lectin.
J. Appl. Bacteriol.
 ,
75
,
360
–368.
Ramachandraiah, G. and Chandra, N.R. (
2000
), Sequence and structural determinants of mannose recognition.
Proteins Struc. Func. Genet.
 ,
39
,
358
–364.
Ramachandraiah, G., Chandra, N.R., Surolia, A., and Vijayan, M. (
2002
) Re-refinement using reprocessed data to improve the quality of the structure: a case study involving garlic lectin.
Acta Cryst.
 ,
D58
,
414
–420.
Sacchettini, J.C., Baum, L.G., and Brewer, C.F. (
2001
) Multivalent protein-carbohydrate interactions. A new paradigm for supermolecular assembly and signal trasduction.
Biochemistry
 ,
40
,
3009
–3015.
Vijayan, M. and Chandra, N. (
1999
) Lectins.
Curr. Opin. Struct. Biol.
 ,
9
,
707
–714.
Weis, W.I. and Drickamer, K. (
1996
) Structural basis of lectin-carbohydrate recognition.
Annu. Rev. Biochem.
 ,
65
,
441
–473.

Author notes

2Molecular Biophysics Unit, Molecular Biophysics Unit, Indian Institute of Science, Bangalore 560012, India; and 3Bioinformatics Centre, Indian Institute of Science, Bangalore 560012, India