Crystallographic and modelling studies on a mannose-specific lectin from garlic bulbs

Abstract

Lectins are multivalent carbohydrate-binding proteins which exert their biological effects through their ability to specifically bind different sugars. They occur in plants, animals, bacteria, and viruses. They exhibit widely different folds, the only common property they share being the ability to specifically bind carbohydrates. Among the plant lectins, indeed among lectins as a whole, legume lectins are the best-studied family. The agglutinin from garlic belongs to the bulb lectin family. Snowdrop lectin was the only bulb lectin whose structure was known when the present study was initiated.

Chemical studies indicated garlic lectin to be either a heterodimer of two subunits of approximately 11.5 and 12.5 kDa or a homodimer of 12 kDa subunits. This mannose-specific lectin was crystallised in the presence of a large excess of -D-mannose in space group C2 with cell dimensions a = 203.24 Å, b = 43.78 Å, c = 79.27 Å and = 112.4°, with two dimers in the asymmetric unit. X-ray diffraction data were collected to a nominal resolution of 2.4 Å (effective resolution 2.8 Å) on a MAR imaging plate. Data were processed using XDS and the structure was solved by molecular replacement. It was at this stage that the author joined the project. The structure, refined to an R-factor of 22.6% and an R-free of 27.8% using X-PLOR, reveals a -prism II fold, similar to that in snowdrop lectin, comprising three antiparallel four-stranded -sheets arranged as a 12-stranded -barrel, with an approximate internal threefold symmetry.

This agglutinin is, however, a dimer unlike snowdrop lectin, which exists as a tetramer, despite a high degree of sequence similarity between them. A comparison of the two structures reveals a few substitutions in the garlic lectin which stabilise it into a dimer and prevent tetramer formation. Three mannose molecules have been identified on each subunit. In addition, electron density is observed for another possible mannose molecule per dimer, resulting in a total of seven molecules in each dimer. Although the mannose-binding sites and overall structure are similar in the subunits of snowdrop and garlic lectin, their specificities to glycoproteins such as gp120 vary considerably. These differences appear, in part, to be a direct consequence of the differences in oligomerisation, implying that variation in quaternary association may be a mode of achieving oligosaccharide specificity in bulb lectins.

Repeated attempts to grow better crystals with a view to improving the definition of the structure did not succeed. The available raw data were then reprocessed using DENZO. The structure was re-refined with both X-PLOR and CNS separately using the reprocessed data, which extended to a resolution of 2.2 Å. These two sets of refinements, along with the two sets using the XDS-processed data, afforded an opportunity to compare the performance of different data-processing and refinement packages when dealing with data from weakly diffracting crystals. The best results were obtained when CNS was employed for refinement using data processed by DENZO. The quality and resolution of the map and the definition of the structure improved substantially. In particular, the amino acid residues at the variable locations in the sequence, and hence the isolectins, could be identified with a high degree of confidence. It could be established that the crystal asymmetric unit contains two identical heterodimers. The newly refined structure also provided a better definition of other finer structural details, such as the network of water molecules interacting with the tryptophans present close to the internal threefold axis of the molecule.

The detailed study of mannose binding by garlic lectin led to an exploration of the sequence and structural determinants of mannose recognition. Mannose, an abundant cell surface monosaccharide, binds to a diverse set of receptors that are involved in a variety of important cellular processes. Structural analysis has been carried out on all the proteins containing non-covalently bound mannose as a monosaccharide in the Protein Data Bank to identify common recognition principles. Proteins highly specific to mannose, belonging to the superfamily of bulb lectins, are found to contain a consensus sequence motif QXDXNXVXY, which has been identified as essential for mannose binding. Analysis of this motif in the crystal structures of bulb lectins has led to an understanding of the contribution of individual residues to mannose recognition. Comparison with other mannose-binding proteins reveals common hydrogen-bonding patterns in all of them, despite differences in sequence, overall fold, and the substructures at the binding sites of individual proteins. A database analysis also suggests that although the topology of the backbone, as at the binding site in bulb lectins, can generate mannose-binding capability in a few other proteins, the sequence and disposition of not only the residues in the motif but also the residues in the neighbourhood play a crucial role in allowing that property to be retained.

Multivalency in lectins is a phenomenon that has been discussed at considerable length. The very substantial enhancement, as in garlic lectin, of the binding affinity of lectins for oligosaccharides compared to that for the appropriate monosaccharide is related to this phenomenon. Inhibition studies have given a measure of the binding affinities of garlic lectin for mono-, di-, tri-, penta-, hexa-, octa-, and nona-mannosides. The affinity for the higher oligosaccharides is orders of magnitude higher than that for the monomer. Modelling and energy calculations clearly indicated that this cannot be accounted for by the enhanced interaction energy between the lectin dimer and the oligosaccharides. The possibility of each oligosaccharide binding simultaneously to more than one binding site in the lectin dimer could also be ruled out on the basis of straightforward geometrical considerations. It was then realised that trimannosides and higher oligomers can crosslink lectin dimers.

A comprehensive exploration of all possible crosslinks posed a formidable computational problem. Even a partial exploration involving a carefully chosen region of conformational space clearly showed that the dimers can be crosslinked not only by a single oligosaccharide molecule but also simultaneously by two oligosaccharides involving two distinct pairs of binding sites. The number of such possible “double-crosslinks,” which lead to interesting tetrameric structures, generally increases with the size of the oligosaccharide. This observation is in consonance with the results of experimental studies on sugar binding by garlic lectin. It was also realised that the crosslinks can lead to different types of lectin aggregates. In addition to their immediate relevance to garlic lectin, these modelling studies are of general interest in relation to lectin–oligosaccharide interactions.