Sensitivity in Binding Free Energies Due to Protein Reorganization

Abstract

Tremendous recent improvements in computer hardware, coupled with advances in sampling techniques and force fields, are now allowing protein-ligand binding free energy calculations to be routinely used to aid pharmaceutical drug discovery projects. However, despite these recent innovations, there are still needs for further improvement in sampling algorithms to more adequately sample protein motion relevant to protein-ligand binding. Here, we report our work identifying and studying such clear and remaining needs in the apolar cavity of T4 Lysozyme L99A. In this study, we model recent experimental results that show the progressive opening of the binding pocket in response to a series of homologous ligands.1 Even while using enhanced sampling techniques, we demonstrate that the predicted relative binding free energies (RBFE) are sensitive to the initial protein conformational state. Particularly, we highlight the importance of sufficient sampling of protein conformational changes and demonstrate how inclusion of three key protein residues in the ‘hot’ region of the FEP/REST simulation improves the sampling and resolves this sensitivity.

Predictions of Ligand Selectivity from Absolute Binding Free Energy Calculations

“…However, severe sampling issues, such as the failure to access relevant states (e.g., side-chain rotamers or loop motions), might contribute toward the errors. 25,72 Identifying specific sampling deficiencies on a case-by-case basis can, however, be both challenging and impractical, in particular with the prospect of being able to apply these calculations to large numbers of protein–ligand pairs. The combination of free energy calculations with enhanced sampling techniques might thus be beneficial, as it does not require the identification of specific degrees of freedom that have been under-sampled.…”

Towards full Quantum‐Mechanics‐based Protein–Ligand Binding Affinities

“…The approach that is mostly used is the calculation of relative binding free energies between ligands, making de novo design of new ligands without an experimentally determined starting point somewhat challenging. Another potential source of problems is the alchemical perturbation itself: transformations between chemically very different ligands can make the simulations hard to converge, leading to longer, and thus more costly, simulations . This can be especially problematic in scaffold‐hopping modifications, when rings are closed or their size is changed in single‐topology approaches .…”

“…Another potential source of problems is the alchemicalp erturbation itself:t ransformationsb etween chemically very different ligands can make the simulations hard to converge, leadingt ol onger,a nd thus more costly,s imulations. [5] This can be especially problematic in scaffold-hopping modifications, [6][7][8][9] when rings are closed or their size is changed in single-topology approaches. [10,11] However,t his is not the case for dual-topology approaches, [12] and there has been significant progress to alleviate that problem in single-topology approaches recently.…”

Impact of domain knowledge on blinded predictions of binding energies by alchemical free energy calculations

“…However, for a given set of protein-ligand complexes it remains difficult to anticipate the predictive power of AFE calculations. Uncertainties in binding modes [12][13][14] protonation/tautomeric states [15,16], binding site water content [17][18][19], and choice of potential energy functions [20,21], can profoundly influence the outcome of such calculations. Accordingly, there is much interest in defining as much as possible a domain of applicability for the technology [22].…”