Designed active-site library reveals thousands of functional GFP variants

Abstract

Mutations in a protein active site can lead to dramatic and useful changes in protein activity. The active site, however, is extremely sensitive to mutations due to a high density of molecular interactions, drastically reducing the likelihood of obtaining functional multipoint mutations. We introduce an atomistic and machine-learning-based approach called htFuncLib to design a sequence space in which mutations form low-energy combinations that mitigate the risk from incompatible interactions. We screened 11 million htFuncLib designs that targeted the GFP chromophore-binding pocket and isolated >16,000 unique fluorescent designs encoding as many as eight active-site mutations. Many designs exhibit substantial and useful diversity in stability and activity. By eliminating incompatible active-site mutations, htFuncLib generates a greater diversity of functional sequences than evolutionary or mutational scanning approaches for optimizing enzymes, binders, and other functional proteins.

Engineering of highly active and diverse nuclease enzymes by combining machine learning and ultra-high-throughput screening

“…However, it may be fruitful to more carefully balance the cost of DNA synthesis with the level of control it provides. In early rounds of experiments, for example, it would have been interesting to employ ML to design random combinatorial libraries, which are significantly cheaper to assemble 38,[87][88][89][90][91][92][93][94] . For a given budget, improvements to assays 95 and enrichment factor estimation 96 can also help improve the quality of datasets produced by sorting-based experiments.…”

Combinatorial assembly and design of enzymes

“…CADENZ addresses this conflict by designing a spanning set of low-energy and mutually compatible protein fragments that can be assembled into thousands of diverse and functional proteins. We have also begun to investigate how EpiNNet can be implemented to design large repertoires of active site sequence variants ( 49 ). Our approach increases the number and diversity of functional enzymes that can be interrogated relative to the natural diversity, providing an alternative to metagenomic libraries ( 12 ).…”

Combinatorial assembly and design of enzymes

“…CADENZ addresses this conflict by designing a spanning set of low-energy and mutually compatible protein fragments that can be assembled into thousands of diverse and functional proteins. In a companion paper( 49 ), we demonstrate that EpiNNet can also be implemented to design large repertoires of active-site sequence variants. This approach therefore increases the number and diversity of functional enzymes that can be interrogated relative to the natural diversity, providing an alternative to metagenomic libraries( 12 ).…”