Evolution and Models of Sequence Covariation for Protein Design

Abstract

Enzymes, as genetically encoded biocatalysts, possess immense potential for various applications in medicine and industry due to their ability to facilitate diverse chemical reactions. However, designing task-specific enzymes with optimal performance remains a formidable challenge, primarily due to the intricate relationship between enzyme amino acid sequence and function, as well as the colossal combinatorial search space one must traverse to design optimized variants. For instance, the median length of an E. coli protein (277 amino acids) has 20^277 possible sequence variants, far exceeding the number of atoms in the known observable universe estimated at 10^78 to 10^82. In this work, we aim to develop evolution-inspired computational and experimental methodologies to efficiently navigate the vast protein sequence space and design optimized enzymes for industrial-scale plastic degradation, thereby enabling closed-loop plastics recycling. In chapter 1, we develop a method using evolutionary models of sequence covariation to design variants of the model enzyme TEM-1 β-Lactamase, demonstrating their effectiveness in creating optimized enzyme variants. In chapter 2, we apply this knowledge to the design of plastic-degrading enzymes, showcasing the potential of these models to modulate a wide array of industrially relevant enzyme properties. In chapter 3, we bring evolution into the laboratory and explore the development of ultra-high-throughput screening technologies for evolving plastic-degrading enzymes by directed evolution. We hope that our findings will significantly contribute to the field of protein design and offer valuable insights for future research endeavors.