Supplementary Figure 9: Sequence determinants of β-arch formation.

For the energy landscapes of different types of mutations, we calculated the frequency of formation of β-hairpins between every two consecutive β-strands. Increases of β-hairpin formation were correlated with a decrease of near-native sampling. a,d,g,i, Calculated energy landscapes for mutants assessing different types of interactions (black) are compared with the landscape of BH_10 (red). b,e,h, Effect of mutations on β-strand pairing (red, original design; black, mutant). Mutated loop connections are labeled with the corresponding amino acid substitution. Most of the mutations increase sampling of more local β-hairpin connections. Connection S4-S5 corresponds to the central β-hairpin of the β-helix. c, Sidechain packing interactions stabilizing β-arch loop connections that when mutated to alanine decrease β-arch stability and favor β-hairpin sampling. Mutant V18A favors hairpin sampling in the neighboring β-arch of the same β-arcade. f, Sidechain-backbone hydrogen-bonding interactions stabilizing β-arch loop geometry; upon removal by alanine substitutions, β-hairpin sampling increases. e, Mutations in the S6-S7 and S7-S8 connections favors sampling of β-hairpins between S6 and S8