An evolution-based model for designing chorismate mutase enzymes

By W. P. Russ, Matteo Figliuzzi, Christian Stocker, Pierre Barrat-Charlaix, M. Socolich, P. Kast, Donald Hilvert, Rémi Monasson, Simona Cocco, Martin Weigt and 1 more. 2020

297 citations recorded by the source corpus.

Learning from evolution Protein sequences contain information specifying their three-dimensional structure and function, and statistical analysis of families of sequences has been used to predict these properties. Building from sequence data, Russ et al. used statistical models that take into account conservation at amino acid positions and correlations in the evolution of pairs of amino acids to predict new artificial sequences that will have the properties of the protein family. For the chorismate mutase family of metabolic enzymes, the authors demonstrate experimentally that the artificial sequences display natural-like catalytic function. Because the models access an enormous space of diverse sequences, such evolution-based statistical approaches may guide the search for functional proteins with altered chemical activities. Science, this issue p. 440 An evolution-based, data-driven engineering process can build artificial functional enzymes. The rational design of enzymes is an important goal for both fundamental and practical reasons. Here, we describe a process to learn the constraints for specifying proteins purely from evolutionary sequence data, design and build libraries of synthetic genes, and test them for activity in vivo using a quantitative complementation assay. For chorismate mutase, a key enzyme in the biosynthesis of aromatic amino acids, we demonstrate the design of natural-like catalytic function with substantial sequence diversity. Further optimization focuses the generative model toward function in a specific genomic context. The data show that sequence-based statistical models suffice to specify proteins and provide access to an enormous space of functional sequences. This result provides a foundation for a general process for evolution-based design of artificial proteins.

Subfields: Cellular and Molecular Neuroscience, Molecular Biology

How researchers see this paper

The researchers whose trusted networks rank it highest, strongest advocate first. Every researcher gets a different order.

In Molecular Biology

Citations in the corpus

MSA Transformer (2021, neutral): “…ﬁt to MSAs are widely used in computational biology including in applications such as ﬁtness landscape prediction (Riesselman et al., 2018), pathogenicity prediction (Sundaram et al., 2018; Frazer et al., 2020), remote homology detection (Hou et al., 2018), and protein design (Russ et al., 2020).”
Generating functional protein variants with variational autoencoders (2020, methodology): “…advances in efficient inference methods for these models have been shown to permit the generation of sequences which accurately reproduce the low-order statistics from natural MSAs [52, 59], and a concurrent work has demonstrated the successful application of these methods to protein design [60].”
Transformer protein language models are unsupervised structure learners (2020, methodology): “Unsupervised modeling of protein contacts has an important role in computational protein design (Russ et al., 2020; Tian et al., 2018; Blazejewski et al., 2019) and is a central element of all current state-of-the-art structure prediction methods (Wang et al.”
ProGen2: Exploring the Boundaries of Protein Language Models (2022, background): “Most still train on a ﬁxed protein family to capture coevolutionary signals present within a set of homologous sequences – ranging from direct coupling analysis techniques (Russ et al., 2020) to generative adversarial networks (Repecka et al., 2021).”
Guided Generative Protein Design using Regularized Transformers (2022, background): “…couplings themselves are likely insufﬁcient: past work has shown that protein function is governed by higher-order interactions between residues Russ et al. [2020] (i.e., epistasis) such that the collective effects of mutations can be non-additivand as a result, difﬁcult to predict Pokusaeva…”