Inferring protein 3D structure from deep mutation scans

Abstract

We describe an experimental method of three-dimensional (3D) structure determination that exploits the increasing ease of high-throughput mutational scans. Inspired by the success of using natural, evolutionary sequence co-variation to compute protein and RNA folds, we explored whether 'laboratory', synthetic sequence variation might also yield 3D structures. We analyzed five large-scale mutational scans and discovered that the pairs of residues with the largest positive epistasis in the experiments are sufficient to determine the 3D fold. We show that the strongest epistatic pairings from genetic screens of three proteins, a ribozyme, and a protein interaction reveal 3D contacts within and between macromolecules. Using these experimental epistatic pairs, we compute ab initio folds for a GB1 domain (within 1.8 Å of the crystal structure) and a WW domain (2.1 Å). We propose strategies that reduce the number of mutants needed for contact prediction, suggesting that genomics-based techniques can efficiently predict 3D structure. Users may view, print, copy, and download text and data-mine the content in such documents, for the purposes of academic research, subject always to the full Conditions of use:

The mutational landscape of a prion-like domain

“…We have shown recently that the pattern of genetic (epistatic) interactions between mutations in a protein can report on the secondary structure of that molecule when it is performing the function that is being selected for 51,60 . In particular, when a sequence forms an α-helix, the side chains of residues separated by 3-4 AA are close in space and similarly oriented so that mutations in these AA interact similarly with mutations in the rest of the protein.…”

“…Epistasis is the nonindependence of mutation effects, i.e., the toxicity of double mutants is different from that expected given the toxicity of their constituent single mutant variants. We have previously shown that epistasis between double mutants can result from structural interactions within proteins and therefore can be used to infer secondary and tertiary structural features 51,60 . In brief, double mutants were classified as epistatic if they had more extreme toxicity values (below 5th percentile or above 95th percentile) than other double mutants with similar single mutant toxicities, which was estimated from non-parametric surface fits of double mutant toxicity as a function of a two-dimensional single mutant toxicity space ( Fig.…”

Evolutionary context-integrated deep sequence modeling for protein engineering

“…Mutations within the protein sequence can affect fitness in a non-independent way, which is also known as genetic interactions or epistasis. It was found that epistasis interactions, quantified by deep mutagenesis scanning (DMS) of proteins, can be used to infer protein contacts and structures [33,34] . As structurally proximal protein residues are often inferred from co-variation pairs from sequence evolution historically [35,36] , we hypothesized that co-evolution information can also be used to infer epistasis or fitness of proteins.…”

“…To test the ability or our method to predict epistasis, we compiled multiple DMS studies that contain the fitness values of both single and double amino acid variants. We obtained the DMS data of the GB1 domain, WW domain, RRM domain, and FOS-JUN heterodimer from [34] , and the prion-like domain of TDP-43 from [53] . A set of fitness of TEM-1 consecutive variants was also obtained from [52] .…”

DiMSum: an error model and pipeline for analyzing deep mutational scanning data and diagnosing common experimental pathologies

“…The resulting fitness landscapes are informative of protein [4][5][6] , RNA [7][8][9] and regulatory element [10][11][12][13][14][15][16][17][18] function, and have provided mechanistic insight into biological processes including the regulation of gene expression [10,19] , protein-protein interactions [20] , alternative splicing [21,22] and molecular evolution [7] . Deep mutational scans have the potential to improve human variant annotation [23,24] and protein and RNA structure determination [25][26][27] . In recognition of the growing number and importance of DMS assays in biomedical research, a dedicated platform for sharing, accessing and visualizing these datasets has recently been launched [28] .…”