Data-Driven Directed Enzyme Evolution for Highly Strained Ring Synthesis
General concepts outlined in this proposal. We propose to develop enzyme-catalyzed carbene transfer routes to synthesis of highly strained and densely functionalized N-heterocycles.
Evolution has generated myriad precise and efficient enzymes to assemble the molecules and materials of life with a chemical prowess no human chemist can mimic or master. Nature’s catalysts are catalytically promiscuous, i.e., enzymes can catalyze fortuitous side reactions in addition to the functions for which they have been selected in evolution. These promiscuous activities fuel the evolution of new enzymes when altered circumstances provide a fitness advantage to such innovations. The Arnold lab pioneered tools of directed evolution to exploit this catalytic promiscuity and generate enzymes that address challenges in synthetic chemistry. Notably, we have evolved hemoproteins that catalyze new-to-nature carbene and nitrene transfer reactions, including functionalizations of C─H, Si─H, and B─H bonds as well as cyclopropanations and aziridinations of unsaturated C─C bonds.
We will use this discovery to develop a modular carbene transferase platform for synthesis of highly strained and densely functionalized N-heterocycles. We envision a diversified enzymatic approach to these compounds, including: (1) cyclopropanation of unsaturated N-heterocycles with diazo compounds as carbene precursors and (2) cyclopropanation of hydrocarbon-based alkenes and alkynes using N-heterocycle-derived spirocyclic diazirines as carbene precursors. The proven capability of engineered hemoproteins to form highly strained ring systems with extraordinary selectivity indicates that biocatalysts are uniquely equipped to address existing synthetic limitations in this field. Catalytic activation of this new class of spirocyclic diazirines is unknown and would represent fundamental progress in catalysis and provide precedent for a general strategy for N-heterocycle installation.
Throughout the enzyme development, we will implement state-of-the-art tools in data-driven protein engineering, including every variant sequencing (evSeq), an accessible protocol that enables unprecedented access to paired sequence-function data for every variant produced during a protein engineering campaign. This comprehensive information will advance machine-learning guided directed evolution and thereby streamline the evolution of new enzyme function, as well as provide important insight into biochemical underpinnings that accompany such evolution.
