Genetics, energetics, and allostery in proteins with randomized cores and surfaces

A lack of systematic experimental data limits our understanding of protein evolution. In this study, we experimentally characterized proteins with randomized sequences. Vast numbers of amino acid combinations constitute stable protein cores and surfaces. However, alternative cores frequently disrupt protein function by indirect allosteric effects. Both protein stability and binding can be predicted by using simple additive energy models with a small contribution from pairwise energetic couplings. Indeed, energy models trained on one protein can predict functional cores and surfaces across more than a billion years of evolution, with only rare energetic couplings that we experimentally identify limiting the transplantation of cores between highly diverged proteins. Our results reveal the simple energetic architecture of proteins and suggest that allostery is an important constraint on sequence evolution. Overall design: We built combinatorial libraries in the hydrophobic cores of three small protein domains (FYN-SH3, CI-2A and CspA) using a reduced alphabet consisting of the amino acids F, L, M, V, I encoded by the DTS degenerate codon. By bottlenecking and pooling the libraries, in the sparse_DTS_core_mutagenesis experiment we sparsely measured the intracellular abundance of protein variants in yeast cells using abundancePCA, a protein complementation assay that couples cell growth rate with query protein intracellular abundance under selection by methotrexate. For the SH3 domain of the human FYN kinase, we also randomized 15 buried and 21 exposed sites using reduced amino acid alphabets that were optimized ad-hoc for each site (FYN-SH3_buried_n_exposed_randomization experiment). Also for FYN-SH3, we selected a few query core amino acid combinations that are severely deleterious in abundance fitness and designed a suppressor "permissivity" library by introducing non-core mutations associated with SH3 domains naturally carrying such query core combinations that are deleterious in FYN (FYN-SH3_core_permissivity experiment). For the same protein (FYN-SH3), we assessed the impact of core reconfiguration in function by measuring the binding to its short linear motif ligand PRD1super using bindingPCA, a protein complementation assay that couples cell growth rate with query variant intracellular binding to an interacting partner under selection by methotrexate (FYN-SH3_core_DTS_binding experiment).