Is inhibition process better described with MD(QM/MM) simulations? The case of urokinase type plasminogen activator inhibitors

Abstract

Urokinase plasminogen activator (uPA) is an enzyme involved in cancer growth and metastasis. Therefore, the design of inhibitors of uPA is of high therapeutic value, and several chemical families have been explored, even if none has still emerged, emphasizing the need of a rationalized approach. This work represents a complete computational study of uPA complexed with five inhibitors, which present weak similarities. Molecular dynamics simulations in explicit solvent were conducted, and structural analyses, along with molecular mechanics (MM)/Poisson-Boltzmann surface area free energies estimations, yield precious structure-activity relationships of these inhibitors. Besides, we realized supplemental QM/MM computations that improved drastically the quality of our models providing original information on the hydrogen bonds and charge transfer effects, which are, most often, neglected in other studies. We suggest that these simulations and analyses could be reproduced for other systems involving protein/ligand molecular recognitions.

Simulation with quantum mechanics/molecular mechanics for drug discovery

“…Implication of these effects in a drug design process is generally hard to evaluate. However, in a comparative study of 5 urokinase inhibitors, it has been demonstrated that charge transfer process corrects the overestimation of free energy values determined through molecular dynamics and (QM/MM)PBSA analyses [31]. This correction mostly concerns the ionic inhibitors [31,32] which realize a salt-bridge with the Asp192 of urokinase.…”

“…For nucleic acids, where the macromolecule and generally the ligand are charged, this effect is really Reproduced from [31] with permission of John Wiley and Sons. important [36].…”

“…Indeed, biomacromolecules are plastic objects and their required conformational sampling is generally made by molecular dynamics for several nanoseconds. It is particularly the case for the determination of the free energy binding of inhibitors with the QM/MM-PBSA or QM/ MM-GBSA methods [31,38,39,75]. In this case, the accuracy of the free energy value depends more on the number of considered structures than the level of theory in the QM part [31].…”

“…It is particularly the case for the determination of the free energy binding of inhibitors with the QM/MM-PBSA or QM/ MM-GBSA methods [31,38,39,75]. In this case, the accuracy of the free energy value depends more on the number of considered structures than the level of theory in the QM part [31]. When QM/MM calculations are realized for deciphering a chemical reaction pathway, such as an enzymatic mechanism, the conformational exploration is dreadfully increased by the use of supplemental trajectories to obtain a PMF made by techniques such as umbrella sampling, metadynamics, or steered molecular dynamics [50].…”

“…important [36]. Besides, QM computation can also lead to a better consideration of hydrogen-bonding [31,[37][38][39]. Indeed, in MM simulations, this interaction is not taken into account explicitly but rather calculated within the force-field equation as a mix between electrostatic and Van der Waals effects [3,6].…”