Hybrid computational approaches for the elucidation of complex biocatalytic mechanisms

We report two mechanistic investigations on recently discovered, industrially relevant biocatalytic systems conducted using a hybrid computational approach integrating Quantum Mechanics/Molecular Mechanics (QM/MM) with the Molecular Dynamics/Perturbed Matrix Method (MD-PMM). In the first study, concerning the reduction of 2-pentanone catalyzed by the Horse Liver Alcohol Dehydrogenase (HLADH) in the presence of the biomimetic cofactor, benzyl-1,4-dihydronicotinamide (BNAH), we propose a revised mechanism that follows a sequential pathway in which an endothermic hydride transfer (HT) precedes a strongly exothermic proton transfer (PT). The Gibbs free-energy barrier for HT, calculated using MD-PMM, is 14 kcal mol−1, in excellent agreement with the value derived from experimental data. The protein environment is found to be responsible for lowering the energy barrier compared to the gas phase. A plausible PT pathway is hypothesized in which His51 is first protonated from the bulk solvent and then relays the proton via a bridging water molecule to the oxygen atom of Ser48 and subsequently to the substrate carbonyl. One or more water molecules forming a hydrogen-bond network between His51 and Ser48 are essential to activate the proton transfer chain. The second study focuses on fatty acid photodecarboxylase from Chlorella variabilis (CvFAP), a recently discovered flavin adenine dinucleotide (FAD)-containing photoenzyme that catalyzes the light-driven decarboxylation of long-chain fatty acids into the corresponding alkanes, a reaction of considerable relevance for biofuel production. Despite its remarkable biotechnological potential, the mechanistic pathway of this transformation remains only partially understood, encompassing uncertainties about the protonation state of the substrate, the origin of the unusual features in the enzyme’s FAD absorption spectrum (which exhibits the main absorption peaks red-shifted by approximately 10–15 nm compared with those observed in other flavoproteins), and the nature of the red-shifted FAD species (FADRS) observed experimentally after decarboxylation. Our QM/MM calculations strongly indicate that the bound fatty acid exists predominantly in its deprotonated form. Building on this result, the FAD absorption spectrum in CvFAP was simulated using MD–PMM calculations, which show that the contribution of flavin conformational bending to the observed red shift is minor (4 nm), whereas the protein environment is the dominant factor (15 nm). Regarding the origin of the FADRS species, our calculations support the hypothesis that its formation is likely driven by a pronounced reorganization of the active site following product formation (alkane, CO2, and bicarbonate), with the bicarbonate interacting strongly with the FAD moiety. Our results provide a solid basis for understanding the reaction mechanisms of these enzymes, offering mechanistic insights that could inform future environmentally friendly industrial applications. Moreover, from a computational point of view, this thesis demonstrates the complementarity of the two methodologies (QM/MM and MD–PMM) used here highlighting that their integration is fundamental for achieving a detailed and accurate characterization of enzymatic reaction mechanisms.