Deciphering The Underlying Resistance-Conferring Mechanisms Of Dihydrofolate Reductase Using Energy Landscape Theory

Official URL: https://risc01.sabanciuniv.edu/record=b3205842

How can short timescale molecular dynamics simulations can be of use? In this work, we tried to address the antibiotic resistance problem with 210 ns-long simulations utilizing smart constructs. Biologically, dihydrofolate reductase (DHFR) is an enzyme working on folate synthesis network producing purine/pyrimidine precursors need for DNA replication. Its substrate is dihydrofolate (DHF) which is converted into tetrahydrofolate (THF) by a hydride transfer reaction. E. coli DHFR is inhibited by the inhibitor trimethoprim. This work focuses on the evolutionary trajectory of E. coli DHFR under TMP resistance and explains the differences in working dynamics of different mutations. The work consists of four pillars, in one pillar we delve into energy landscape of DHFR side chains and understand the characteristics of the dynamics, the tiered nature of the landscape, discuss ergodic behavior of mutations, binding dynamics and an interesting phenomenon - synchronization of the enzyme to the substrate-. Then, we utilize short time hydrogen bonding dynamics to explain fitness. With the acquired know-how, we propose two applications of hydrogen bond dynamics as repurposing a drug as a binder to DHFR to the newly discovered cryptic site by our work and explain working mechanism of an inhibitor proposed by our group. Finally, we exploit DHF and TMP-based trajectories using a newly thermodynamic cycle constructed for alchemical free energy perturbation calculations, for a sample size of 13 mutations, we were able to predict reliable free energy estimates with a root mean squared (RMS) error of 1.16 kcal/mol. We claim that for understanding ligand-protein interactions and the effects rooting to ligand-protein interactions can well be explained by short-range simulations where conformational changes requires more sampling.