• No results found

Molecular modelling

Chapter 2 General Methods

2.10 Molecular modelling

The crystal structures of fibrinogen, C3 and C3b (Protein Data Bank [PDB] ID codes 3GHG, 2A73 and 2I07 respectively) were visualised using, Pymol (Schrödinger, Inc.), which is free for academic groups. There is little information currently available regarding the interaction sites of C3(b) and fibrinogen and

78

given that both are relatively large molecules, current molecular modelling software is not able to identify sites of protein-protein interaction due to the vast computational memory required. By concentrating on the C3 molecule it was possible to identify potential sites of protein interaction through the use of the molecular modelling software, AutoLigand (Scripps Research Institute). This is an automated predictor of ligand binding sites in proteins which works by analysing the geometry of a protein surface. The majority of binding sites on a protein are within pockets or clefts and this method is able to search the entire surface of a protein structure and identify areas of maximal affinity for ligand binding sites232. Once suitable areas of binding are identified, the electronic high throughput screening software, eHiTS (SimBioSys, Inc.) will be used to target these regions. eHiTS is an automated docking tool which allows the rapid evaluation of large compound libraries in order to identify ligands which can bind to the protein of interest. This algorithm utilises a database of small molecules, in this case, the small molecule screening library from Peakdale Molecular. It divides the 3-dimensional ligand structures into rigid fragments and flexible chains and each rigid fragment is docked into pockets, or receptor sites, identified on the surface of the protein. Different orientations or “poses” of these fragments that are compatible with chemistry and steric constraints are tested within the receptor site and a score is determined based on interacting surface points between the ligand and cavity. The flexible chains that connect the rigid fragments are fitted to reform the ligand using the lowest energy chain configuration possible233. Validation studies of this docking programme has demonstrated its ability to replicate ligand binding compared to known crystal

79

structures of bound ligands and it has been used previously as a screening tool to identify novel drug targets234-236.

Protein-peptide interactions are a subset of protein-protein interactions and are known to be able to regulate certain biological processes and may be manipulated therapeutically given that the interaction surface is small in relative terms237. This approach of studying protein-peptide interactions was a logical step in order to identify sites of interaction between C3 and fibrinogen through the use of peptide sequences, identified from techniques such as phage display and microarray screening that bind C3 and fibrinogen.

Modelling of the predicted binding sites of these peptides with C3 and fibrinogen was achieved through the use of freely available software, PepSite, (www.pepsite2.russelllab.org) that is capable of predicting the binding sites of peptides, up to 10 aa in length, to a known protein surface (using the PDB ID code). PepSite is available via a web server and the initial version has been superseded by PepSite2237. The original method studied 405 different protein- peptide interactions that had a known 3-dimensional structure. From this data every peptide residue was allocated a spatial position-specific scoring matrix (S- PSSM), which essentially encodes the calculated preferred binding environment for each peptide residue. PepSite2 utilises these S-PSSMs and scans them against the protein structure of interest and predicts preferred sites of amino acid binding within the protein, termed hotspots. These hotspots are subsequently matched against every residue within the peptide sequence to be analysed. The resulting pairs of matched residues are subjected to distance constraints so that they lie within a range typically seen in peptide structures237. The result is an approximation of the peptide structure bound to the protein

80

surface and is presented visually using a Java viewer, Jmol, and in a table in order of statistical significance. The results can also be downloaded as a PDB file and visualised using Pymol.

Molecular modelling technique Description

Autoligand Searches a known protein structure for potential ligand binding sites

eHiTS Uses a library of small molecules to identify ligands to a known protein structure

Pepsite2 Determines predicted binding sites of amino acids within a peptide sequence (up to 10 amino acids in length) to a known protein structure

AutoDock Predicts the docking poses of 3-

dimensional peptide structures to a known protein structure.

Table 2-1 Summary of molecular modelling techniques used

Brief description of the molecular modelling techniques utilised and the information generated as a result

Another method of identifying protein-peptide interaction sites is to use the docking program, AutoDock (Scripps Research Institute). AutoDock can be used when the structure of both the ligand and target molecule are known, but the site of interaction is not238. It attempts to dock the peptide or ligand to the protein of interest using a grid based method that permits identification of the binding energy between the two molecules. A grid is placed over the surface of the target protein and a probe atom from the ligand is placed at each grid point which then stores the calculated interaction energy between the protein and probe. The software also generates different “trial” conformations of the ligand based on the types of bonds present and thus generates a number of potential

81

docking sites based on the binding energies produced by the trial conformations239. This technique will be employed using peptide sequences from fibrinogen identified from the microarray and phage display screening and also using Adhirons of interest. A summary of the different molecular modelling techniques employed can be seen in Table1-1. The Autoligand and AutoDock screening were performed by Dr Katie Simmons, School of Chemistry, University of Leeds.

Related documents