Left-handed conformations of glycyl residues may confer protection against protein aggregation.

Abstract

The lack of Cβ atom allows glycyl to adopt left-handed Ramachandran conformations, typically disallowed for l-amino acids. The underlying significance remains under-appreciated. Through conformational analysis of glycyls at 1104 disease and 343 benign variant sites, we show that the left-handed glycyls are over-represented (odds ratio > 1.3) at disease variant sites and are evolutionarily conserved. Mutations involving l-disallowed glycyls destabilize native folding by altering free energies (P = 2.4 × 10^-4). The l-disallowed glycyls are enriched at the aggregation gatekeepers, more profoundly so in thermophiles (P = 2.0 × 10^-6), implying heightened selection to impede aggregation. Mutations of l-disallowed glycyls also reduce the protein solubility (P = 0.001). Due to mostly positive Φ dihedral-angle, Cα atom of l-disallowed glycyl flips to conform a crescent that likely disrupts β-strand alignment, discouraging the intermolecular aggregation of β-strands. Deep learning confirms the predictive value of l-disallowed glycyls in identifying pathogenic variants (accuracy = 0.81 vs. 0.69, area under the curve = 0.88 vs. 0.79). The findings underscore the evolutionary selection of l-disallowed conformations of glycyls to maintain proteostasis by modulating protein stability and aggregation, and suggest applications for disease-associated genetic prioritization and soluble protein design.