Professor Robert Woods' research examines the relationships between the conformations of carbohydrate molecules and biological recognition and activity.Using computational chemistry, analytical chemistry and chemical biology methodologies the research team is pursuing: 1) characterization of carbohydrate-protein interactions, 2) the directed evolution of affinity/specificity of carbohydrate-binding biomolecules and 3) the development of computational biomolecular structure tools.
|Figure1: An antibody carbohydrate complex that is being studied via molecular dynamics simulation, in silico docking and hydroxyl radical footprinting.|
Areas of particular interest include carbohydrate antigenicity in immunological events, carbohydrate- processing enzymes, and the development of computational tools to explore protein-carbohydrate interactions. Such tools include our molecular dynamics force field GLYCAM, which is developed using quantum mechanical methods. Using GLYCAM and molecular dynamics simulations we can explore the structural properties of highly flexible biomolecules such as carbohydrates. To explore the relationship between carbohydrate sequence/structure and the affinity of carbohydrate-protein interactions we employ free energy perturbation and direct DG calculations, as well as in silico docking. These computational methods are complimented by experimental techniques, such as NMR spectroscopy and mass spectrometry based footprinting.
Furthermore, the design of Lectenz®, enzymes engineered with novel carbohydrate recognition activity, is initiated in silico to determine optimal carbohydrate-enzyme interactions (Figure 2). Simulated structure/function relationships are validated by generating biocombinatorial libraries for selection and downstream characterization of Lectenz® candidates.
|Figure2: We are engineering protein carbohydrate interfaces through in silico mutagenesis and in vitro selection using biocombinatorial libraries.|
NMR techniques exist that enable protein–ligand complexes to be probed from the point of view of the ligand alone. For example, the transferred NOE (tr–NOE) and saturation transfer difference (STD) experiments facilitate the determination of the bound carbohydrate conformation, and may be used to derive structure–activity relationships. In these experiments, spin magnetization is transferred from the saturated protein to the ligand, which is then observed. The simplicity of this approach facilitates its application of a broad range of interactions, the limitation being that the interaction must have a rapid kinetic off-rate. Since the protein itself is not being examined in this experiment STD NMR provides complementary information to surface footprinting experiments (described below). There is at present no single experimental method that can be relied on to determine the 3D structures of Fab–oligosaccharide complexes. The Woods group is examining the use of saturation transfer difference (STD) NMR experiments to provide experimental confirmation for our putative immune complexes.
Mass spectrometry applied to surface footprinting analysis of protein-ligand complexes. We employ protein surface contact analysis methods to determine which amino acids in the antibody CDRs are in contact with the polysaccharide antigens. These methods are all based on the concept that in a complex, the protein surface in the interaction region is shielded from exposure to solvent, and is therefore much slower to react with derivatizing reagents. After reaction, MS proteomics techniques may be used to identify which residues were protected from reaction. After transferring that information to a 3D model for the protein, it is possible to visualize the footprint left by the intermolecular contact. The experimental MS work will proceed locally via collaboration with Dr. Ron Orlando at the Complex Carbohydrate Research Center (CCRC, UGA) and with the group of Dr. Mark Chance (Case Western Reserve University).
The prediction of a protein-carbohydrate complex is an important step in understanding protein specificity in antibody-antigen recognition, cell-cell adhesion, enzyme-substrate specificity, molecular transport, etc. The carbohydrate recognition mechanism depends on 1.) the sequence of the monosaccharides in the glycan (i.e. glucose vs. mannose), 2.) the anomeric centers (i.e. α or β), 3.) the linkage positions (i.e. 1,3 vs. 1,4), 4.) the chemical modifications to the core glycan (i.e. sulfation, phosphorylation, methylation, acetylation, etc.). The strength of this interaction is also determined by the carbohydrate conformation and orientation with respect to the binding site. Our research uses two techniques to probe protein-carbohydrate interactions, Molecular Dynamics (MD) and Docking simulations. MD can be used to probe this interaction, but this depends on having a well-resolved model of the protein-carbohydrate complex before starting the simulation. Additionally, MD is computationally expensive, limiting the number of structures that can be examined. The Woods group uses high-throughput screening to dock a large number of carbohydrate ligands to many proteins using significantly less computational resources per complex. Computational docking also gives us a way to arrive at a well-resolved protein-carbohydrate structure without having known the binding conformation in advance. Once a docked model is obtained, it can then be used as a starting point for further refinement and exploration using MD simulations.
The Woods group uses a variety of methods to characterize and study carbohydrate-protein interactions. Automated molecular docking can be utilized to identify binding sites in carbohydrate-binding proteins, or predict the bound orientation and conformation of a carbohydrate or drug mimic in a known binding site. Molecular dynamics (MD) simulations are employed to refine docked models and to study the dynamic nature of binding given a 3D structure. In order to quantify the energy of protein-carbohydrate interactions, post-processing methods such as Molecular Mechanics Poisson Boltzmann Surface Area (MM-PBSA) or Molecular Mechanics Generalized Born Surface Area (MM-GBSA) calculations can be used. MM-PB/GBSA can also be used to predict which residues have the greatest contribution to binding energy and are therefore most critical to the binding interaction in a protein-carbohydrate complex. Alternatively, Thermodynamic Integration (TI) can be used to calculate free energies of binding between two molecular states via alchemical mutation of a molecule or residue of interest over the course of a MD simulation. MD simulation trajectories can be analyzed to track hydrogen-bonds in simulated protein-carbohydrate complexes in order to identify key binding residues and characterize the binding interaction. The Woods group is primarily interested in protein-carbohydrate interactions relevant to viruses, bacteria, inhibitory drugs, and immune response.
The Woods group is developing a novel class of carbohydrate binding reagents. We refer to these reagents, which have lectin-like properties and which are derived from carbohydrate processing enzymes, as Lectenz®. The design of Lectenz® is initiated in silico to determine optimal carbohydrate-enzyme interactions using molecular dynamics simulations. Simulated structure/function relationships are validated by generating biocombinatorial libraries for selection and downstream characterization of Lectenz® candidates.
In this project, experimental techniques will be used to study interactions between carbohydrates (CHOs) and recombinant antibody fragments (rAbFs). CHOs on the surface of many infectious pathogens and tumor cells are desirable targets for therapeutic agents, such as rAbF, because they are often abundant, unique structures and are the main virulence factor for several pathogens. Single-chain variable fragments (scFvs) are among one of the most popular types of rAbFs as they have been successfully modified into a number of different antibody formats and are easily expressed by several expression systems, e.g. in E. coli. scFvs contain the complete antigen binding site of an antibody and are comprised of the variable heavy (VH) and variable light (VL) domains linked together by a flexible polypeptide linker (e.g. (GGGGS) 3).
The Woods group is currently working to characterize scFvs that bind to different bacterial and yeast antigenic CHO targets, such as the Group B Streptococcus type III polysaccharide 1, and β-1-2-mannotriose on Candida albicans 2, respectively. Determining the specificity and affinity of scFv for their CHO targets is crucial to determining their effectiveness as potential therapeutic agents. Different anti-CHO binding scFv in E. coli are being expressed and purified by immobilized metal affinity chromatography and size exclusion chromatography to obtain quantities of high purity for characterisation by surface plasmon resonance. The effect of specific point mutations on the ability of the scFv to bind the target CHO ligand is being analysed. References  Weisser NE, Almquist, KC, Hall. JC. 2007 Vaccine. 25, 4611-4622. [ 2] Han Y, Riesselman MH, Cutler JE. Infect Immun. 2000 68, 1649-1654.
The computational methods we develop here often require the use of novel procedures or novel applications of existing procedures. To make it easier for researchers to rapidly utilize our developments, we are constantly generating computational tools. Several of these tools are already available online, for example, our oligosaccharide and glycoprotein structure predictors which provide a convenient web-interface for generating 3D structures. Other tools are still in development including a tool for virtual, high-throughput screening of glycans against putative binders and a library of functions in C designed to facilitate the generation of highly novel computational methods. Students involved in these projects come from a diverse array of backgrounds including biochemistry, chemistry, physics, computer science and bioinformatics.
The Woods group uses computational methods to predict the structure of biomolecules in solution, crystal and gas phases through the development of the GLYCAM biomolecular force field. GLYCAM was originally developed to simulate carbohydrates using the AMBER simulation package. Recent developments have expanded the GLYCAM force field for work on glycolipids and glycosamino glycans, a highly charged class of carbohydrates. Research is ongoing to make GLYCAM parameters for nucleic acid systems which have a carbohydrate moiety in ribose (RNA) and deoxyribose (DNA). Research is ongoing to develop methods for simulating structural and thermodynamic properties of oligosaccharides, glycoproteins, glycolipids, lipids, and protein-carbohydrate complexes using the GLYCAM force field. Data from these simulations are directly compared to mass spec, isothermal calorimetry, glycan array, NMR, and crystallographic data to provide a better understanding of the biological activity of these molecules.