Expanding the space of protein geometries by computational design ofde novofold families

Abstract

Naturally occurring proteins use a limited set of fold topologies, but vary the precise geometries of structural elements to create distinct shapes optimal for function. Here we present a computational design method termed LUCS that mimics nature’s ability to create families of proteins with the same overall fold but precisely tunable geometries. Through near-exhaustive sampling of loop-helix-loop elements, LUCS generates highly diverse geometries encompassing those found in nature but also surpassing known structure space. Biophysical characterization shows that 17 (38%) out of 45 tested LUCS designs were well folded, including 16 with designed non-native geometries. Four experimentally solved structures closely match the designs. LUCS greatly expands the designable structure space and provides a new paradigm for designing proteins with tunable geometries customizable for novel functions.One Sentence SummaryA computational method to systematically sample loop-helix-loop geometries expands the structure space of designer proteins.

Improved protein structure prediction by deep learning irrespective of co-evolution information

“…Baker group has shown that deep ResNet can fold 16 of the 18 de novo proteins designed by the same group [12]. Here we conduct a more comprehensive study on 21 de novo proteins designed by two research groups in 2018-2020 [26][27][28], among which 11 proteins have been used to test trRosetta. None of these 21 proteins has evolutionarily related homologs in the 2018 Cath S35 training proteins (HHblits with E-value<0.1).…”

“…We collected 35 de novo proteins (including 7 membrane proteins) designed by two research groups in the past several years. Meanwhile, 4 of them are designed by Kortemme group [28] and the others by Baker group. Among these proteins, 21 of them have HHblits E-value>0.001 with our training proteins in the 2018 Cath S35 dataset, so we only use these 21 proteins as our test set.…”

Dual film-like organelles enable spatial separation of orthogonal eukaryotic translation

“…Computer-aided de novo protein design is particularly suitable for designing completely novel structures because it allows sampling of a vast sequence space in silico. The resulting designs are often expressed and tested in E. coli (Cao et al, 2020;Glasgow et al, 2019;Pan et al, 2020;Xu et al, 2020). More recently, strategies for evolving mammalian systems have emerged (Berman et al, 2018;English et al, 2019;Piatkevich et al, 2018), although E. coli and yeast remain the preferred hosts for handling large libraries (Almhjell et al, 2018;Branon et al, 2018;Bryson et al, 2017).…”

A Brief History of De Novo Protein Design: Minimal, Rational, and Computational

“…2N8I; 177 loop-helix-loop unit combinatorial sampling, or LUCS, from Kortemme's group, e.g. 6VGA; 178 and TopoBuilder from Bruno Correia and colleagues for incorporating functional elements into de novo protein frameworks. 103,105 Gaps and remaining challenges: These advances in computational de novo protein design are allowing increasingly challenging design targets to be addressed such as structures rich in b structure -2KL8, 179 3WW7, 180 5BVL, 172 and 5KPE 181 ; and membrane-spanning peptide and protein assemblies -2MUZ, 106 6B85, 107 6MCT, 108 6M6Z, 84 6X9Z, 109 and 6YB1.…”