First Principles Modeling
First principles modeling mathematically represents quantum mechanical, reactive force field, and molecular dynamics principles in computational systems. This approach allows precise in silico description of amino acid, protein, and ligand systems as well as precise dynamic positional mapping of a system’s molecular and atomic constituents as they interact with each other in the biological environment. Homology modeling, in contrast, does not offer the precision required for the protein structure and ligand binding analysis that are fundamental to drug discovery and design.
Computationally Capturing GPCR Active States in CB1 Cannabinoid Receptor
Please see the following papers (in pdf form) for further information:
Computationally-predicted CB1 cannabinoid receptor mutants show distinct patterns of salt-bridges that correlate with their level of constitutive activity reflected in G protein coupling levels, thermal stability, and ligand binding Proteins, 2013 Aug;81(8):1304-17.
Precision: BX471 Binding Differential for Human CCR1 v. Mouse CCR1
Analysis undertaken for a large drug company provides an example of the precision enabled by first principles modeling. The task was to determine the structural difference between two protein variations that had an unusually high degree of amino acid sequence similarity, particularly in the relevant binding regions, to account for a differential binding result. Specifically, while the human and mouse versions of the GPCR CCR1 have 87% sequence identity in the GPCR’s 7-helix transmembrane region, a Berlex drug, BX471, nevertheless, binds strongly to the human protein but not at all to the mouse protein. First principles computational analysis showed a structural change resulting from a single amino acid substitution between the human and mouse versions of CCR1. As if to emphasize the sensitivity of structural conformation to amino acid sequence, the single amino acid substitution responsible for switching binding to non-binding is not even in the binding pocket. Instead, a substitution outside the binding pocket causes transmembrane loop 3 to rotate 30 degrees, which changes the actual binding site, thus resulting in BX471 binding to the human protein CCR1 but not the mouse protein CCR1. To contrast first principles modeling to homology modeling, the latter would not be able to explain the binding differential between these two protein variations, despite their unusually high degree of sequence similarity, even though homology modeling is based on inferences drawn from similarities among proteins’ respective amino acid sequences.
Results: Sanofi & Comparison to Merck Drug
The effectiveness of QioMed’s technology is also shown by work regarding the prostaglandin DP receptor. Here, QioMed technology was used to determine (a) the structure of the DP receptor, (b) 20 ligands that would bind to that structure, and (c) the respective binding affinities of those ligands for the receptor. Synthesis and testing of the 20 ligands showed the predicted binding affinities to be accurate, thus validating the structures, including the receptor, binding site and ligand structures. The technology was then used to determine 20 improved derivatives of three additional scaffolds (i.e., improved derivatives of three weakly binding compounds identified through high throughput screening (HTS) against the DP receptor). Of these 20 derivatives, one was designed to bind 1000 times more tightly to the DP receptor than did the original HTS compound. Synthesis and testing of these 20 ligands again showed the predicted affinities to be correct, again validating both the technology’s ability to determine the DP receptor structure and the technology’s ability to determine and accurately predict ligands, binding sites, and binding affinities for ligands binding to that target. The 1000-fold improved derivative was then intended for clinical development by the drug company for whom the foregoing work was conducted, Sanofi-Aventis.
The initial work in this process included analysis of a Merck drug that was known to bind to the DP receptor, and the initial 20 ligands were designed to explore the relationship between this Merck drug and the predicted structure of the DP receptor. Merck later published actual binding affinities for eight derivatives it had made of its own drug. Merck’s eight derivatives, previously unknown outside Merck, were, in fact, among the initial 20 ligands — i.e., the initial 20 predicted ligands ((b), above) included the eight Merck subsequently published. Merck’s post-facto data showed near perfect agreement with the predicted binding energies ((c), above), supplementing Sanofi’s prior synthesis and testing validation.
The above-described work was conducted by Professor William A. Godard, III. A paper describing this work and Professor Goddard’s results appeared in the Journal of the American Chemical Society. Click the following link for an excerpt: