Molecular modelling of the interactions of complex carbohydrates with proteins

Abstract

Glycosaminoglycans (GAGs) are ubiquitous complex carbohydrate molecules present on the cell surfaces and in extracellular matrices (ECM) of vertebrate and invertebrate tissues. The interactions of sulphated GAGs such as heparin and heparan sulphate (HS) with numerous immunologically-relevant proteins is now attracting considerable interest as a source of new therapeutics for the treatment of infectious diseases, inflammation and allergies, and cancers. Various computational molecular modelling methods are being employed to determine the nature of the interactions of heparin oligosaccharides with various proteins in order to establish the structural requirements that determine their binding specificity and selectivity.The first part of this research focused on understanding the inflammatory cytokine CXCL-8 (Interleukin-8 or IL-8) and its interactions with GAGs. A variety of molecular modelling methods were used to investigate the binding of complex carbohydrates to this protein. A number of consensus heparin/HS binding motifs were predicted to be required for the binding of monomeric or oligomeric structures of CXCL-8. Bioinformatics analyses showed that the basic residues in the heparin binding site are highly variable within the CXC family and amongst various CXCL-8 species. Drug-like carbohydrate mimetic molecules (cyclitols) that bind optimally to CXCL-8 were identified and characterised. It was established that both an optimum number of sulphates and a certain length of alkyl spacers are required for the interaction of cyclitol inhibitors with the dimeric form of CXCL-8. Furthermore, explicit solvent molecular dynamics simulations of dimeric CXCL-8 showed how its two anti-parallel helices exhibit domain movements that can bring them in closer proximity. In addition, these simulations revealed shearing movements in the C-terminal helices in the CXCL-8 dimer. This inherent flexibility of the CXCL-8 dimer can be exploited in drug design as it plays an important role in the understanding of the interactions of molecules such as sulphated cyclitols with the two monomers.Structural bioinformatics and molecular modelling methods were used to generate and analyse a three-dimensional model of heparanase, an enzyme involved in metastasis and angiogenesis in cancer, in order to gain insight into its protein sequence, and its structural and functional relationships. The interactions of heparanase with disaccharide substrates and GAG mimetics were modelled to investigate the structural determinants of their protein binding specificity and selectivity. The choice of structural template for modelling the binding site of heparanase is very critical. Analyses of active-site similarity across groups of homologous template structures revealed that these bound oligosaccharides can block ligand binding to the catalytic and heparin binding sites of heparanase. Ligand-protein docking simulations further revealed the existence of a large binding site extending at least two saccharide units beyond the cleavage site (towards the non-reducing end) and at least three saccharides towards the reducing end (towards heparin-binding site 2). Extensive modelling of substrate and inhibitor interactions with the catalytically-active glutamic acids and the two binding sites for heparan sulphate of heparanase provides information useful for future drug discovery efforts focused on the identification of novel inhibitors of this enzyme.Free energy calculations of the binding of sGAGs to GAG-binding proteins were pioneered with the proteins PECAM-1 (platelet endothelial cell adhesion molecule) and Annexin, using the molecular mechanics/Poisson Boltzmann surface area (MM/PBSA) method. Analysis of the free energy of binding components revealed that the major contributors to complex stability are electrostatic interactions, with equally important contributions from van der Waals interactions and vibrational entropy changes, against a large, unfavourable desolvation penalty due to the high charge density of sGAG. The calculated absolute free energies of binding of the molecules investigated were found to be inaccurate compared to experimental values, but the method performed well in discriminating weak and strong binding.A final focus of this research was to investigate the conformational properties of sulphated iduronic acid (IdoA2S), a hexopyranose present in heparin. IdoA2S adopts more than one conformation (skew boat and chair) when internally positioned within an oligosaccharide or when it interacts with proteins. The influence of the solvent on the flexibility and conformations of this saccharide is of significant interest given that its biomolecular interactions occur in an aqueous environment. Therefore, molecular dynamics simulations were used to investigate the ability of the GROMOS force field and the GLYCAM carbohydrate parameter set in the presence of explicit solvent to successfully predict rotamer populations for this ring system. Calculations of theoretical proton NMR coupling constants showed that the GROMOS96 force field can predict the skew-boat to chair conformational ratio in good agreement with experiment; however, the accuracy of the GLYCAM force field in representing ring conformer populations during unconstrained molecular dynamics simulations is still debatable.