Large-scale discovery, analysis and design of protein energy landscapes

Library design The initial set of 15,715 domain sequences was organized into five batches and further divided into 18 libraries (mix 1–4, libraries 1 and 4; libraries 7–15; and mutants 2–4): (1) mix 1–4: de novo designed ααα, βαββ and ββαββ sequences 21 ; (2) libraries 1 and 4: de novo designed αββα proteins 11 ; (3) libraries 7–14: natural domains from the Pfam database, including LysM, PASTA, WW, SH3, pyrin and cold-shock; (4) library 15: PDB-derived monomeric proteins devoid of cysteine residues and metal cofactors; (5) mutant libraries containing single and double mutants from EEHEE_rd4_0871 and HHH_rd4_0518 low-cooperativity proteins. Sequences were randomly assigned to libraries within each batch, ensuring a minimum mass difference of 50 ppm between nearest-neighbour sequences for mass spectrometry compatibility (except library 15 where two sequences are 36 ppm apart). After SUMO cleavage (see below), all proteins begin with the dipeptide HM (the scar from the NdeI ligation). Some sequences were modified with C-terminal padding (G, S, GG or GS) to optimize mass spacing. All sequences were reverse-translated and codon-optimized for E. coli using DNAworks (v.2.0) 68 . To standardize amplification efficiency, a ‘GGS’ sequence was appended after the stop codon. Oligo libraries encoding the original 15,715 sequences were purchased from Agilent Technologies, while the 280 designed mutations were sourced from Twist Bioscience. …