Expanding Enzyme Function Through Data-Guided Evolution

Author: Lal, Ravi Goel

Year: 2026

Degree: Dissertation (Ph.D.)

Advisor: Arnold, Frances Hamilton

Committee Members: Shapiro, Mikhail G.; Mayo, Stephen L.; Parker, Joseph; Arnold, Frances Hamilton; Stoltz, Brian M.

Option: Chemical Engineering

DOI: 10.7907/5zaf-8w29

Abstract

Through the process of evolution, Nature has optimized enzymes for the chemical transformations which drive biology. The use of enzymes for chemical synthesis is enticing as these privileged catalysts can facilitate reactions in a highly sustainable and selective manner. Directed evolution (DE) is a strategy for engineering proteins to improve a desired function. This approach has been demonstrated to be of great effect for the development of enzymatic activities which have never been naturally observed. These ‘new-to-nature’ chemistries not only showcase the power of DE to unlock novel functions, but also the potential of biocatalysis to deliver high-value chemical compounds. This thesis details the exploration of new-to-nature biocatalytic reactions of hemoproteins using traditional DE and novel machine learning-assisted directed evolution (MLDE) methods. Chapter I provides an overview of how DE has enabled new-to-nature-biocatalysis, and the challenges associated with developing this novel biocatalytic reactions. In Chapter II, a cytochrome P411 is found to catalyze a carbene-mediated [1,2]-Stevens rearrangement to furnish azetidines via ring expansion, and is engineered to deliver these products with high yield and enantioselectivity. The evolution of this activity is a demonstration of how established DE techniques can be utilized to arrive at enzymes capable of unprecedented chemistry. Chapter III describes efforts towards advancing this ring-expansion chemistry for the synthesis of proline analogs. Though this activity could not be found with existing sequence diversity, several engineering insights were made about protoglobins (small, thermostable hemoproteins). Chapters II and III both highlight challenges associated with DE. In Chapter IV, active-learning assisted directed evolution (ALDE) is introduced as a workflow which leverages MLDE in an iterative fashion to greatly accelerate DE efforts. Alongside simulations on combinatorially complete datasets, ALDE was validated in the wet lab by simultaneously evolving a protoglobin-based cyclopropanation for improved yield and stereoselectivity. Finally, Chapter V describes the use of ALDE to engineer protoglobins with active sites which have been optimized to catalyze a broad scope of nitrene and carbene transfer reactions. This demonstration of enzyme ensemble engineering displays the power of diversity-oriented evolution to provide broad solutions in biocatalysis.