Data-driven supervised learning of a viral protease specificity landscape from deep sequencing and molecular simulations

Abstract

Biophysical interactions between proteins and peptides are key determinants of molecular recognition specificity landscapes. However, an understanding of how molecular structure and residue-level energetics at protein−peptide interfaces shape these landscapes remains elusive. We combine information from yeast-based library screening, next-generation sequencing, and structure-based modeling in a supervised machine learning approach to report the comprehensive sequence−energetics−function mapping of the specificity landscape of the hepatitis C virus (HCV) NS3/4A protease, whose function—site-specific cleavages of the viral polyprotein—is a key determinant of viral fitness. We screened a library of substrates in which five residue positions were randomized and measured cleavability of ∼30,000 substrates (∼1% of the library) using yeast display and fluorescence-activated cell sorting followed by deep sequencing. Structure-based models of a subset of experimentally derived sequences were used in a supervised learning procedure to train a support vector machine to predict the cleavability of 3.2 million substrate variants by the HCV protease. The resulting landscape allows identification of previously unidentified HCV protease substrates, and graph-theoretic analyses reveal extensive clustering of cleavable and uncleavable motifs in sequence space. Specificity landscapes of known drug-resistant variants are similarly clustered. The described approach should enable the elucidation and redesign of specificity landscapes of a wide variety of proteases, including human-origin enzymes. Our results also suggest a possible role for residue-level energetics in shaping plateau-like functional landscapes predicted from viral quasispecies theory.

Learning peptide recognition rules for a low‐specificity protein

“…We sought an alternate approach to simple sequencebased metrics. Inspired by the literature on characteristic biochemical properties of intrinsically-disordered proteins, 29 as well as previous machine-learning approaches applied to biochemical interactions, 7,30,31 we hypothesized that the biochemical features of peptide tar-gets could be used to construct a predictive model of of peptide enrichment. For each peptide sequence identified in the phage display dataset, we calculated a set of 57 features covering an array of biochemical properties.…”

“…One can predict protein targets by searching for matching sequences within the proteome 6 . Some proteins deviate from this highly specific paradigm, requiring more sophisticated approaches 7 . For example, many PDZ binding domains exhibit binding “multi‐specificity”, in which peptide preference must be represented as a handful of binding motifs 8,9 .…”

“…6 Some proteins deviate from this highly specific paradigm, requiring more sophisticated approaches. 7 For example, many PDZ binding domains exhibit binding "multi-specificity", in which peptide preference must be represented as a handful of binding motifs. 8,9 Predicting interaction targets for such proteins is more difficult than for proteins with single binding motifs, but the same basic logic applies: search the proteome for sequences that match the binding motifs.…”

An automated protocol for modelling peptide substrates to proteases

“…Researchers can derive models of protease-peptide complexes to analyse the binding spectrum [ 29 , 30 ]. One approach is to analyse protease-peptide structures to ascertain amino acid preferences at different positions of a peptide substrate sequence [ 31 ]. These computational analyses usually employ experimental catalytic information stored in repositories such as M-CSA [ 32 ], bringing a clearer perspective of the structural role of the catalytic residues, and other pocket residues, during binding.…”

Structural basis for peptide substrate specificities of glycosyltransferase GalNAc-T2

“…Similarly, Pethe et al 36 were able to obtain significantly improved prediction accuracies for proteases (including HIV−1 protease) compared to previous methods (MFPred, sequence tolerance and pepspec) by employing machine learning and a discriminatory score based on geometric features, interface score terms from Rosetta and electrostatic score terms from Amber. In another work, Pethe et al 37 used supervised learning on experimentally obtained deep-sequencing data and information from structure-based models to chart the specificity landscape of 3.2 million substrate variants of the viral protease HCV.…”