Inhibition of Succinate Dehydrogenase Activity As a Mode of Action for Papyriferic Acid

Molecular Docking Overview

Molecular docking is a key tool in structural molecular biology and computer-assisted drug design. The goal of ligand–protein docking is to predict the predominant binding mode(s) of a ligand with a protein of known three-dimensional structure. Successful docking methods search high-dimensional spaces effectively and use a scoring function that correctly ranks candidate dockings. Docking can be used to perform virtual screening on large libraries of compounds, rank the results, and propose structural hypotheses of how the ligands inhibit the target, which is invaluable in lead optimization. A typical docking experiment involves the following steps:

Protein Receptor Preparation The receptor structure is usually obtained experimentally, by either x-ray crystallography or nuclear magnetic resonance. In this project, the mitochondrial respiratory membrane protein complex II (C-II) structure was obtained from available porcine x- ray structure in the Protein Data Bank (PDB code 1ZP0, http://www.rcsb.org/pdb), which was shown to have high amino acid sequence identity (99%) to rodent C-II through BLAST sequence alignment at NCBI (http:// www.ncbi.nlm.nih.gov/BLAST/). The BLAST search showed that the amino acid sequences between porcine, rats and mice complex II are nearly identical, 99% (NCBI Accessions: NP_001094009 and EDL13335) for the iron-sulphur protein subunit, 100% (NCBI Accessions: NP_001005534 and NP_079597) for the large trans-membrane subunit and 100% (NCBI Accessions: NP_942083 and NP_080124) for the small trans-membrane subunit.

PDB x-ray structures are not protonated. Therefore, all hydrogen atoms are added to the receptor. The partial charges are required by AutoDock program. They are calculated using MGLTools, a companion package for AutoDock program. MGLTools also merges nonpolar hydrogens as AutoDock uses a united-atom representation.

In this study, C-II was protonated at physiological pH 7.3 using the program PDB2PQR (Dolinsky and others 2004), which also determined the protonation states for histidine residues and optimized the internal hydrogen-bonding network of the protein. Amber Atom types and partial charges (Case and others 2005) for the protein, including the cysteines involved in the iron-sulphur clusters and the heme group, were assigned by the PDB2PQR program. MGLTools version 1.5.6 (Sanner 2005) was then used to prepare the protein for docking by merging nonpolar hydrogens.

1 Ligand Preparation Ligand preparation steps are similar to receptor preparation. The differences are: (a) Ligand structures can be built manually or obtained from PDB (or other small-molecule databases); (b) Ligand protonation is processed using PRODRG program instead of PDB2PQR because they are designed for small-molecules and proteins respectively; (c) Besides partial charges, MGLTools also adds a torsion tree for bond rotation because ligands are flexible during docking.

In this study, 2-Thenoyltrifluoroacetone (TTFA) coordinates were built from the crystal structure (PDB code 1ZP0); Ubiquinone-5 (UbQ) coordinates were built from the crystal structure (PDB code 1ZOY); α-Tocopheryl succinate (TOS) coordinates were built from the Vitamin E 3D structure which was retrieved from the PDB Ligand Expo Database (http://ligand- depot.rcsb.org/reports/V/VIT); PA coordinates were obtained from NCBI PubChem chemical structure database (CID 441683). All of the ligands were prepared for docking by MGLTools version 1.5.6, which included merging non-polar hydrogens, assigning Gasteiger charges and defining bond torsions.

AutoDock uses grid maps that must be calculated using AutoGrid. Each map describes a 3D grid of interaction energies with the receptor, one for each atom type in the ligand. We used AutoGrid version 4.2 (Morris et al., 2009) to create affinity grids centered on the protein ubiquinone binding site, specifically the proximal (Qp) and distal (Qd) sites. Each grid enclosed an area of 128 X 128 X 128 grids with 0.375 Å grid spacing. The affinity maps were calculated for the receptor atom types (A C Fe HD N NA OA SA), the ligand atom types (C OA HD A SA), electrostatic interactions and desolvation.

Docking The search space must first be defined. If there is previous information, such as ligands with known binding modes, active site residues, or mutagenesis data, then the search space can be reduced to focus on the region of interest, thus, simplifying the search problem. In this project, TTFA coordinates were used to define the docking search space. Then, the default Lamarckian genetic algorithm in (LGA) AutoDock was used to carry out the docking experiments.

In this study, docking parameters were optimized for the positive control docking of TTFA to the Qp and Qd sites of C-II (PDB code 1ZP0) and were as follows: trials of 100 dockings,

2 population size of 200, random starting position and conformation, translation step ranges of 2.0 Å, rotation step ranges of 50°, elitism of 1, mutation rate of 0.02, crossover rate of 0.8, local search rate of 0.06, 5 million energy evaluations and 27000 generations. Each docking calculation took about 4 hours on a 3.2 GHz AMD Phenom II X4 based workstation.

Evaluating Docking Results When evaluating the results of dockings, the main criterion to consider is how well the binding mode predicted by the docking matchs known structural data, where available. To do this, a crystal structure of the complex of the ligand bound to the receptor must be known, and then the root mean square deviation (RMSD) between the docked and the “reference” crystallographic binding mode of the ligand can be calculated. Success is usually counted as RMSD within a reasonable threshold comparable to ligand size.

When the search method used is stochastic such as LGA, it is important to consider how often a given binding mode was predicted across all the dockings that were run. This is usually achieved using conformational clustering, building families of related conformations using RMSD tolerances to decide whether two conformations belong in the same cluster. In the end, the predicted binding free energy is taken from the lowest energy conformation within the largest cluster.

In this study, docked conformations were clustered using a tolerance of 2.5 Å RMSD using MGLTools version 1.5.6 clustering script. Docking results were obtained from the lowest binding energy conformations of the most populated cluster.

The predicted binding mode(s) can be loaded into molecular visualization programs such as VMD (Humphrey and others 1996) for further analysis.

References

Case DA, Cheatham TE, Darden T, Gohlke H, Luo R, Merz KM, Onufriev A, Simmerling C, Wang B, Woods RJ. 2005. The Amber biomolecular simulation programs. J. Comput. Chem. 26:1668-1688.

3 Dolinsky TJ, Nielsen JE, McCammon JA, Baker NA. 2004. PDB2PQR: an automated pipeline for the setup of Poisson–Boltzmann electrostatics calculations. Nuc. Acids Res. 32(suppl 2):W665-W667. Humphrey W, Dalke A, Schulten K. 1996. VMD: Visual molecular dynamics. J. Mol. Graph. 14:33-38. Sanner MF. 2005. A Component-Based Software Environment for Visualizing Large Macromolecular Assemblies. Structure 13:447-462.