Antiparallel and parallel β-strands differ in amino acid residue preferences

Abstract

A β-strand is a particular type of extended sequence of amino acid residues, an element of secondary structure of proteins. β-sheets are an assembly of strands, often bringing together parts of the protein which are separated along the backbone. As such, β-sheets are an element of tertiary structure. Parallel β_P) and antiparallel (β_A) arrangements of strands in a sheet differ in the hydrogen bond pattern between strands, as shown schematically in Fig. 1, and in the type of chain connectivity they allow: short reverse turn connections for β_A and longer crossover connections for β_P (refs 1–3). Most present secondary structure prediction methods (for reviews refs 4–6) use a four-state distinction of secondary structure: α-helix, β-strand or extended, reverse turn, and ‘random coil’ (everything else). With a data base of 30–40 different protein structures, the conformational preferences for all amino acid residues in these four states seem to have converged⁷. However, the steadily increasing data base of structurally known proteins makes a refinement of the four-state description feasible. Although more refined classifications of conformational states based on finer subdivisions of (φ,Ψ)-space have been made^8,9, we prefer making distinctions based on structural environment. Using a novel definition of β-sheet structure in terms of the tertiary structure juxtaposition of strands, we have analysed residue contacts in known β-sheets and report here secondary structure preferences for the 20 amino acids, separately for antiparallel and parallel arrangements of strands. The distinction between the two arrangements results in strikingly different and sharpened sets of preference parameters, including some of the largest values reported so far for any substructure. These results point the way towards a basic improvement of secondary structure predictions by further distinction of secondary structure elements according to tertiary structure environment. Beyond secondary structure prediction, the different preferences for β_A and β_P may aid in predicting the tertiary interaction between strands.