Conspectus Natural metalloenzymes are often the most proficient catalysts in terms of their activity, selectivity, and ability to operate at mild conditions. However, metalloenzymes are occasionally surprising in their selection of catalytic metals, and in their responses to metal substitution. Indeed, from the isolated standpoint of producing the best catalyst, a chemist designing from first-principles would likely choose a different metal. For example, some enzymes employ a redox active metal where a simple Lewis acid is needed. Such are several hydrolases. In other cases, substitution of a non-native metal leads to radical improvements in reactivity. For example, histone deacetylase 8 naturally operates with Zn2+ in the active site but becomes much more active with Fe2+. For β-lactamases, the replacement of the native Zn2+ with Ni2+ was suggested to lead to higher activity as predicted computationally. There are also intriguing cases, such as Fe2+- and Mn2+-dependent ribonucleotide reductases and W4+- and Mo4+-dependent DMSO reductases, where organisms manage to circumvent the scarcity of one metal (e.g., Fe2+) by creating protein structures that utilize another metal (e.g., Mn2+) for the catalysis of the same reaction. Naturally, even though both metal forms are active, one of the metals is preferred in every-day life, and the other metal variant remains dormant until an emergency strikes in the cell. These examples lead to certain questions. When are catalytic metals selected purely for electronic or structural reasons, implying that enzymatic catalysis is optimized to its maximum? When are metal selections a manifestation of competing evolutionary pressures, where choices are dictated not just by catalytic efficiency but also by other factors in the cell? In other words, how can enzymes be improved as catalysts merely through the use of common biological building blocks available to cells? Addressing these questions is highly relevant to the enzyme design community, where the goal is to prepare maximally efficient quasi-natural enzymes for the catalysis of reactions that interest humankind. Due to competing evolutionary pressures, many natural enzymes may not have evolved to be ideal catalysts and can be improved for the isolated purpose of catalysis in vitro when the competing factors are removed. The goal of this Account is not to cover all the possible stories but rather to highlight how variable enzymatic catalysis can be. We want to bring up possible factors affecting the evolution of enzyme structure, and the large- and intermediate-scale structural and electronic effects that metals can induce in the protein, and most importantly, the opportunities for optimization of these enzymes for catalysis in vitro.

Two virtually identical acireductone dioxygenases, ARD and ARD′, catalyze completely different oxidation reactions of the same substrate, 1,2-dihydroxy-3-keto-5-(methylthio)pentene, depending exclusively on the nature of the bound metal. Fe 2+ -dependent ARD′ produces the α-keto acid precursor of methionine and formate and allows for the recycling of methionine in cells. Ni 2+ -dependent ARD instead produces methylthiopropionate, CO, and formate, and exits the methionine salvage cycle. This mechanistic difference has not been understood to date but has been speculated to be due to the difference in coordination of the substrate to Fe 2+ versus Ni 2+ : forming a five-membered ring versus a six-membered ring, respectively, thus exposing different carbon atoms for the attack by O 2 . Here, using mixed quantum-classical molecular dynamics simulations followed by the density functional theory mechanistic investigation, we show that, contrary to the old hypothesis, both metals preferentially bind the substrate as a six-membered ring, exposing the exact same sites to the attack by O 2 . It is the electronic properties of the metals that are then responsible for the system following different reaction paths, to yield the respective products. We fully explain the puzzling metal-induced difference in functionality between ARD and ARD′ and, in particular, propose a new mechanism for ARD′. All results are in agreement with available isotopic substitution and other experimental data.

No abstract