Free Energy Methods in Drug Discovery

The successful use of computational methods to enhance the drug discovery process relies on the ability to adequately predict binding affinity of compounds prior to synthesis. Over the past several years several studies have pointed to the limitations in using empirical or molecular mechanics based scoring functions to predict relative potencies of congeneric series of compounds. A number of approximate free-energy based computational methods have been developed to more closely predict ligand binding affinities. The Molecular Mechanics-Poisson-Boltzmann Surface Area (MM-PBSA) is one approximate method that is being successfully applied in structure-based drug discovery. Variations to the MM-PBSA methodology include explicit or implicit solvent and choice of solvation model, Poisson Boltzmann (PB) or Generalized Born (GB).

On 14 October 2008 we will hold an eCheminfo Community of Practice conference session at Bryn Mawr College, Philadelphia to address the various current uses of the MM-PBSA/MM-GBSA methods in evaluating protein-ligand interactions.  The first portion of the session focuses on the fundamentals and applications of implicit solvent models. Details of the Generalized Born (GB) implicit solvent model as applied to MM-GBSA will be discussed by Alexey Onufriev. David Case will describe uses of MM-GBSA in rescoring docked ligands. Rommie Amaro will compare and contrast the use of explicit and implicit solvent models in molecular dynamics simulations of Influenza Neuraminidases.  Peter Coveney will demonstrate the use of MM-PBSA method to HIV protease as implemented over a highly distributed computational infrastructure.  Anna Kohlmann will elaborate on the use of MM-GBSA in post scoring of kinase inhibitors and its use in developing local models.  Scott Brown will close the session by presenting an analysis of MM-PBSA performance across several protein targets.

A description of the session which will be chaired by chaired by Judith Lalonde (Bryn Mawr College) with presentation abstracts follows:

Application of MM-PBSA Free Energy Methods in Drug Discovery

http://echeminfo.com/COMTY_confprog08freeenergy

(Please follow continuation here to read abstracts. Comments can be made at the end of post.)

Abstracts

Comparative Studies using Explicit and Generalized Born Molecular Dynamics Simulations of Influenza Neuraminidases

Rommie E. Amaro, University of California, San Diego

Avian influenza virus type A, subtype H5N1, is becoming the world’s largest pandemic threat due to its high virulence and lethality in birds, quickly expanding host reservoir, and high rate of mutations. Antigenic drift has given rise to new strains that are resistant to existing drugs and antigenic shift is resulting in new virulent subtypes of the flu virus, underscoring the need to design novel therapeutics. The first crystal structures of a group-1 NA in apo form and in complex with currently available drugs revealed that although the binding pose of Tamiflu was similar to that seen in previous crystallographic complexes, the 150-loop adopted a distinct conformation, opening a new cavity adjacent to the active site. These structures also suggested a slow conformational change may occur upon inhibitor binding. Despite this detailed structural information, the interpretation of the loop dynamics based on crystal structures alone is a difficult task. As a complement to the crystallographic structures, all-atom explicit solvent and generalized Born molecular dynamics (MD) simulations of the apo and Tamiflu-bound systems were carried out. These extensive simulations suggested that the 150-loop and adjacent binding site loops may be even more flexible than observed in the crystal structures. In addition, comparison of the avian- and human-type NAs are carried out. The comparative dynamics of the NAs that we present here allow us to make insights into the flexibility of the 150- and 430-loops, which are important due to their proximity to the sialic-acid binding site. Furthermore, the dynamics of residues that have been shown by computational solvent mapping and ligand docking to be potentially important in the binding of ligands to this expanded area are investigated and discussed. Lastly, the relative positions of Tamiflu in the different systems are reported and interpreted in the context of developing more specific inhibitors against the N1 strain.

In what Contexts can we reliably use MM-PBSA in Industrial Drug Discovery?

Scott P. Brown, Dept. of Structural Biology, Abbott Laboratories

We present an analysis of MMPBSA performance across a number of protein targets, examining results in different contexts based on the amount (and type) of experimental information available to guide the calculation. We wish to address the question: under what conditions can we expect MMPBSA to perform reliably? That is to say, we wish to identify a "domain of applicability" for MMPBSA-based rank-ordering of a set of candidate molecules by potency against a given protein target.

Scoring and Re-Scoring Ligand Binding Energies using Implicit Solvent Models

David A. Case, Dept. of Chemistry and Chemical Biology, and BioMaPS Institute for Quantitative Biology, Rutgers University

Molecular docking computationally screens thousands to millions of organic molecules against protein structures, looking for those with complementary fits. Many approximations are made, often resulting in low hit rates. A strategy to overcome these approximations is to rescore top-ranked docked molecules using a better but slower method. One such class of methods include mechanics generalized Born surface area (MM-GBSA) techniques. These more physically realistic methods have improved models for solvation and electrostatic interactions and conformational change compared to most docking programs. In a collaboration with Matt Jacobson and Brian Shoichet at UCSF, we recently re-ranked docking hit lists in three small buried sites: a hydrophobic cavity that binds apolar ligands, a slightly polar cavity that binds aryl and hydrogen-bonding ligands, and an anionic cavity that binds cationic ligands. In retrospective calculations, MM-GBSA techniques with binding-site minimization better distinguished the known ligands for each cavity from the known decoys compared to the docking calculation alone. A total of 33 molecules highly ranked by MM-GBSA for the three cavities were tested experimentally. Of these, 23 were observed to bind: these are docking false negatives rescued by rescoring. X-ray crystal structures were determined for 21 of these 23 molecules. In many cases, the geometry prediction by MM-GBSA improved the initial docking pose and more closely resembled the crystallographic result; yet in several cases, the rescored geometry failed to capture large conformational changes in the protein. I will discuss what we know about the origins of these successes and failures, and prospects for rescoring in biologically relevant targets.

Rapid and Accurate Determination of Binding Free Energies in Protein-Drug Systems using Automated Workflows across Federated Intercontinental Supercomputing Grids

Peter V. Coveney, Univeristy College London, UK

Medical practitioners have limited ways of matching a drug to the unique genetic profile of a virus population as it mutates within a patient under drug-related selective pressure. Currently, knowledge based decision support software, making use of existing clinical records and associated viral genotypic data, is used to aid inhibitor selection [1].

Our aim is to explain and accurately quantify the effects of resistance mutations on drug binding using fully-atomistic molecular dynamics (MD) simulations. Furthermore, we aim to demonstrate how the development of automating software which utilises suitable High Performance Computing (HPC) and grid distributed computational infrastructure can be employed to rapidly turn around large numbers of molecular dynamics-based binding affinity calculations. This makes the potential for implementing patient-specific decision support realizable [2] as well as relating molecular level insight directly to the clinical domain. In such a scenario the exact viral-genotypic sequence of a patient is used to deduce inhibitor efficacy across an array of inhibitors, using multiple binding affinity calculations in MD simulations, which return in clinically relevant timescales to confer decision support.

We have recently shown that it is possible, in principle, to quantitatively predict the differences in strength of inhibitors binding to wildtype and mutant HIV-1 protease (PR) enzymes using the single-trajectory MMPBSA and configurational entropy methods [3]. This allows the resistance conferred by an array of mutations to be ranked with respect to a given inhibitor. Computational models of the PR enzyme were fully atomistic, including solvent atoms with production runs of 10 ns duration. Absolute and relative binding affinities for three resistant HIV-1 protease mutants (L90M, G48V, and G48V/L90M) in complex with the inhibitor saquinavir were in excellent agreement to those obtained experimentally. A cross correlation coefficient of 0.96 was obtained for relative ranking of all variants and absolute affinities were all within 0.5 kcal/mol of experimental values.

In general, the approach I describe requires a highly distributed computational infrastructure as well as substantial automation in order to make such studies feasible. To this end, we have developed a tool, the Binding Affinity Calculator (BAC), for the rapid and automated construction, deployment, implementation and post-processing stages of the molecular dynamics simulations [4]. BAC makes use of the Application Hosting Environment (AHE) [5] to execute its constituent components and thus implement the various stages of the workflow involved in the calculation, in general, across multiple HPC and grid-based resources. In particular, the study presented here made use of the compute nodes of the UK National Grid Service (NGS) as well as the US TeraGrid, including the petascale machine, Ranger, at the Texas Advanced Compute Center (TACC).

References
1. Kantor, R., R. Machekano, M. J. Gonzales, K. Dupnik, J. M. Schapiro and R. W. Shafer, 2001. Human immunodeficiency virus reverse transcriptase and protease sequence database: an expanded data model integrating natural language text and sequence analysis programs. Nucleic Acids Research 29(1):296–299.
2. Sadiq, S. K., Mazzeo, M. D., Zasada, S. J., Manos, S., Stoica, I., Gale, C. V., Watson, S. J., Kellam, P., Brew, S. and P. V. Coveney. Patient-specific simulation as a basis for clinical decision making. Preprint.
3. Stoica, I., Sadiq, S. K. and P. V. Coveney (2008). Rapid and accurate prediction of binding free energies for saquinavir-bound HIV-1 proteases. Journal of the American Chemical Society, 130, 2639-2648.
4. Sadiq, S. K., Zasada, S. J., Wright, D., Stoica, I. and P. V. Coveney. An automated molecular simulation-based binding affinity calculator for ligand-bound HIV-1 proteases. Preprint.
5. Coveney, P. V., R. S. Saksena, S. J. Zasada, M. McKeown and S. Pickles, 2007. The application hosting environment: Lightweight middleware for grid-based computational science. Computer Physics Communications 176:406–418.

Deploying Prime MM-GBSA in Kinase Inhibitor Lead Optimization: from Docking to Local QSAR Models

Anna Kohlmann, Ariad Pharmaceuticals

Scoring functions based on MM-GBSA have been successfully applied to predict potency in kinase inhibitors. In this work, we explore the possibilities of MM-GBSA, in the context of docking and scoring compounds during the lead optimization process. Two series of small-molecule inhibitors of Src and Abl kinases were docked, then post-scored with MM-GBSA to compute the free energies of binding and compare them to the experimental data. Then, in an effort to create a scoring function tailored to our series of ligands, we extracted energy terms produced by the MM-GBSA calculation to use as descriptors in a local PLS model. The predictivity of this model is compared to that of Glide and LIE scoring functions as well as other local models based on various sets of descriptors. We also address the feasibility of MM-GBSA calculations at different stages of drug discovery, from virtual screening to library design.

Implicit Solvent Models: Practical Uses, Advantages and Limitations

Alexey Ornufriev, Virginia Tech

The implicit solvent framework has become popular in many biomolecular applications due to its computational efficiency. Accurate representation of electrostatic interactions is key, which are often hardest to compute. Here I will introduce the fundamentals of the implicit solvent methodology, and the latest developments in the area of the generalized Born models. The trade-offs between accuracy and speed will be discussed. An example of application of the implicit solvent methodology to protein-ligand docking will be presented.

Barry Hardy

eCheminfo Community of Practice

Structural Biology eCheminfo cheminformatics chemoinformatics bioinformatics Medicinal Chemistry Computational Chemistry Virtual Screening Poisson Boltzmann Free Energy Molecular Modelling Molecular Modeling pharmaceutical pharma meeting workshop training Oxford Critical Path toxicology Bursary Life Sciences Pharma Drug Discovery Research and Development Drug Development Healthcare Innovation Knowledge Management events

Comments

[Oprah]

Good words.

[Tamiflu Side Effects]

My name is Maura Logan and i would like to show you my personal experience with Tamiflu.

I am 63 years old. Have been on Tamiflu for 5 days now. Formerly had no negative feelings about "Big Pharmaceutical" but this medication has changed my mind. Definitely needed more extensive testing by the FDA.

I have experienced some of these side effects-
Horrible itching started after 8 pills (fourth day) and has lasted for six more days--and counting. Also suffering insomnia, and mood swings--crying and sour temper. Dr. wasn't even sure I had the flu (headache, severe body ache, exhaustion but no cold symptoms). Med seemed to help, but the after-effects are totally miserable. Hugely expensive med and not worth the risk. Absolutely HATE, HATE, HATE this medication!

I hope this information will be useful to others,
Maura Logan