β-Sheet twisting is thought to be mainly determined by interstrand hydrogen bonds with little contribution from side chains, but some proteins have large, flat β-sheets, suggesting that side chains influence β-structures. We therefore investigated the relationship between amino acid composition and twists or bends of β-strands. We calculated and statistically analyzed the twist and bend angles of short frames of β-strands in known protein structures. The most frequent twist angles were strongly negatively correlated with the proportion of hydrophilic amino acid residues. The majority of hydrophilic residues (except serine and threonine) were found in the edge regions of β-strands, suggesting that the side chains of these residues likely do not affect β-strand structure. In contrast, the majority of serine, threonine, and asparagine side-chains in β-strands made contacts with a nitrogen atom of the main chain, suggesting that these residues suppress β-strand twisting.

BackgroundMany β-strands are not flat but bend and/or twist. However, although almost all β-strands have a twist, not all have a bend, suggesting that the underlying force(s) driving β-strand bending is distinct from that for the twist. We, therefore, investigated the physical origin(s) of β-strand bends.MethodsWe calculated rotation, twist and bend angles for a four-residue short frame. Fixed-length fragments consisting of six residues found in three consecutive short frames were used to evaluate the twist and bend angles of full-length β-strands.ResultsWe calculated and statistically analyzed the twist and bend angles of β-strands found in globular proteins with known three-dimensional structures. The results show that full-length β-strand bend angles are related to the nearby aromatic residue content, whereas local bend angles are related to the nearby aliphatic residue content. Furthermore, it appears that β-strands bend to maximize their hydrophobic contacts with an abutting hydrophobic surface or to form a hydrophobic side-chain cluster when an abutting hydrophobic surface is absent.ConclusionsWe conclude that the dominant driving force for full-length β-strand bends is the hydrophobic interaction involving aromatic residues, whereas that for local β-strand bends is the hydrophobic interaction involving aliphatic residues.Electronic supplementary materialThe online version of this article (doi:10.1186/s12900-015-0048-y) contains supplementary material, which is available to authorized users.

No abstract