MULTIMAP TARGETED FREE ENERGY ESTIMATION: A MECHANICAL EMBEDDING QM/MM SCHEME FOR PROTEIN-LIGAND AFFINITIES

ChemRxiv

Abstract

Accurate yet rapid prediction of protein-ligand binding affinities remains one of the central challenges of computationally aided drug design. State-of-the-art approaches rely largely on free-energy perturbation (FEP) techniques, whose accuracy is limited by the underlying classical force fields (FFs). These limitations become particularly severe for pharmaceutically relevant systems that are difficult to describe classically, such as proteins containing transition-metal centers. Quantum mechanics/molecular mechanics (QM/MM) molecular dynamics simulations could, in principle, overcome these limitations, but their computational cost has so far kept them beyond the reach of routine binding free-energy calculations. The recently introduced multimap targeted free-energy perturbation (TFEP) approach, which is available on Github at github.com/andrrizzi/tfep under an MIT license, offers a promising route to bridge classical and quantum descriptions by substantially increasing the overlap between the FF and QM/MM potential-energy distributions, thereby dramatically accelerating the calculation of free-energy corrections from FF to QM. While the method has so far shown excellent performance for small molecules in vacuum, its applicability to realistic biomolecular systems, such as protein–ligand complexes in solution, remains to be established. Here, we first implement multimap TFEP for the calculation of protein–ligand binding affinities at the semiempirical QM/MM level, using mechanical embedding between the QM and MM regions. In this scheme, the QM region, comprising the ligand, is treated quantum mechanically, while its interactions with the surrounding MM environment are described at the classical force-field level. Next, using a standard benchmark consisting of the p38 kinase in complex with a series of inhibitors, we show that multimap TFEP substantially increases the overlap between the FF and QM/MM potentials in both the protein-bound and solvated states, enabling efficient estimation of the corresponding quantum free-energy corrections. Finally, we identify the main challenges that must be addressed to extend TFEP efficiently to QM/MM electrostatic embedding schemes. Such an extension would allow the electronic polarization of the QM region by its biomolecular environment to be explicitly accounted for, paving the way toward practical calculations of protein–ligand binding affinities with quantum-mechanical accuracy.

Authors 5

1. Jacques Gay

   Georgia Institute of Technology

   Affiliation as printed

   Georgia Institute of Technology

   School of Chemistry and Biochemistry

2. Andrea Rizzi

   Forschungszentrum Jülich · Italian Institute of Technology

   Affiliation as printed

   Atomistic Simulations, Italian Institute of Technology, Via Morego 30, Genova, Italy, 16163

   Computational Biomedicine INM-9

   Forschungszentrum Jülich GmbH

3. Shina Caroline Lynn Kamerlin

   Georgia Institute of Technology · Lund University

   Affiliation as printed

   Department of Chemistry

   Georgia Institute of Technology

   Lund University

   School of Chemical and Biomolecular Engineering

   School of Chemistry and Biochemistry

4. Davide Mandelli

   Forschungszentrum Jülich

   Affiliation as printed

   Computational Biomedicine INM-9

   Forschungszentrum Jülich GmbH

5. Paolo Carloni Aachen

   Italian Institute of Technology · RWTH Aachen University

   Affiliation as printed

   Atomistic Simulations, Italian Institute of Technology, Via Morego 30, Genova, Italy, 16163

   Department of Physics and Universitätsklinikum

   RWTH Aachen University