concanavalin A- Carbohydrate interactions: a computer modelling approach

Abstract

Lectins are cell-agglutinating, carbohydrate-binding proteins. They began to generate interest when it was discovered in the 1940s that some of them were capable of specifically agglutinating human erythrocytes. Most lectins studied to date have been from seed sources. Concanavalin A (ConA), a plant lectin from jack beans (Canavalia ensiformis), binds specifically to -D-mannose and -D-glucose containing carbohydrates. Hence, ConA has been used to study cell-surface carbohydrates and the changes they undergo during cell growth, malignancy, and differentiation. ConA preferentially agglutinates transformed cells against their normal counterparts. In addition, the use of ConA in cell fractionation and in the purification of carbohydrates and glycoconjugates is becoming increasingly important.

ConA was the first lectin for which the amino acid sequence and the three-dimensional structure of the native lectin were determined. Hapten inhibition studies on ConA revealed its specificity to carbohydrates. A variety of studies, such as hapten inhibition, NMR, stopped-flow, fluorescence, calorimetric, and computer modelling, have been carried out to understand the preferences of different carbohydrates and glycoproteins containing mannose and glucose residues towards ConA. In spite of these studies, the specific interactions between the sugar ligands and the lectin are not clearly understood.

Three-dimensional (3D) structures of protein-carbohydrate complexes are required to understand the interactions between them at an atomic level. X-ray crystallography is the method of choice to elucidate 3D structures of macromolecules at atomic resolution. Even with the advent of sophisticated facilities like synchrotron, area detectors, and supercomputers, this method often takes a lot of time to solve the 3D structures at a reasonably high resolution. Moreover, obtaining stable crystals of a required protein-ligand complex has always been the rate-limiting step of this technique. Recently, molecular modelling techniques using computers are being used to generate possible 3D structures of unknown protein-ligand complexes using the available X-ray structural data even on the native protein. Earlier studies from this laboratory using contact criteria have led to significant results in the area of carbohydrate-protein interactions. Although methods which use contact criteria alone can give information about the range of possible conformations, it is not possible to discriminate between them. Hence, in the present work, the CCEM (Contact Criteria and Energy Minimization) method has been developed to generate the most probable conformations of the lectin-carbohydrate complexes using the available X-ray crystallographic data of native ConA.

This thesis contains six chapters. Chapter 1 briefly describes details of the structure, specificity, and biophysical properties of ConA relevant to the present study. In Chapter 2, the conformational parameters which define the sugars and the protein are given. This chapter also includes a description of the CCEM method, used in the docking of the ligands to the protein in the most probable orientation. The potential functions and the parameters used have also been discussed.

Chapter 3 discusses the binding of methyl--D-mannopyranoside to ConA. Different possible complexes of ConA and MeMan have been described. One of the predicted complexes agrees well with a recent X-ray crystallographic study. This has marginally higher energy than the lowest energy complex arrived at using the CCEM method.

Chapter 4 deals with the binding of (1-2) linked mannobiose, mannotriose, and mannotetraose to ConA. Studies on the (1-2) linked manno-oligosaccharides have indicated that the nonreducing terminal mannose residue of the mannobiose preferentially occupies the primary sugar binding site compared to that at the reducing end. This is in agreement with the experimental results obtained from kinetic studies. For the mannotriose and mannotetraose, also the nonreducing and middle mannose residues occupy the primary binding site in preference to the reducing terminal residue. In fact, the middle mannose residues are the most favored in the primary sugar binding site as the adjacent residues of these oligosaccharides effectively interact with the protein. Based on these results, an extended binding site, which can accommodate at least three mannose residues of the (1-2) linked manno-oligosaccharides, has been suggested. As the preference of the mannose residues of the oligosaccharides to occupy the primary binding site increases from the nonreducing end towards the reducing end (with the exception of the reducing end residue), a sliding mechanism of binding is proposed for a tetrasaccharide binding.

In Chapter 5, the binding of (1-3) and (1-6) linked mannobioses and of the trimannosyl oligosaccharide, Man (1-6)Man (1-3)Man, a recognition element for ConA on the cell surface, are discussed. The trimannosyl oligosaccharide is connected to the protein on the cell surface through a core fragment comprising two N-acetyl-D-glucosamine (GlcNAc) residues and an L-fucose residue. The effect of the core on the binding of this trimannosyl oligosaccharide has also been examined. These studies suggest that in the case of trimannosyl oligosaccharides without the core sugar residues, the (1-6) linked terminal mannose residue has preference over the (1-3) end mannose residue to reach the binding site of ConA. However, this trend is reversed when the trimannosyl fragment is substituted with the core sugar residues.

Chapter 6 deals with the binding of some biantennary, complex-type oligosaccharides, with 3 (1-2) linked GlcNAc residues or bigger fragments substituted on the terminal (1-3) and (1-6) linked mannose residues. This substitution affects the affinity of ConA for the (1-3) linked mannose residue to a greater extent than that for the (1-6) linked mannose residue, thus making the latter the most favored residue at the primary sugar binding site.

Thus, the present work provides information about the binding of different carbohydrate ligands to ConA. Energy calculations have revealed:

Two most probable conformations for the methyl--D-mannopyranose–ConA complex, one of which agrees with recent X-ray crystallographic studies.

Calculated energies of the (1-2) linked manno-oligosaccharides explain the experimental binding affinities of these saccharides to ConA and suggest the existence of an extended binding site which can accommodate at least three mannose residues.

In trimannosyl oligosaccharides, both the (1-3) and (1-6) linked mannose residues can reach the binding site; however, their relative preference depends on the presence of other substituents.

These studies thus show that computer modelling methods, along with available X-ray crystallographic data, can be successfully used to understand protein-carbohydrate interactions.