Statistical analysis of predicted transmembrane K-helices

Statistical analyses were undertaken for putative transmembrane K-helices obtained from a database representing the subset of membrane proteins available in Swiss-Prot. The average length of a transmembrane K-helix was found to be 22^21 amino acids with a large variation around the mean. The transfer free energy from water to oil of a transmembrane K-helix in bitopic proteins, 348 kcal/mol, is higher than that in polytopic proteins, 339 kcal/mol, and is nearly identical to that obtained by assuming a random distribution of solely hydrophobic amino acids in the K-helix. The amino acid composition of hydrophobic residues is similar in bitopic and polytopic proteins. In contrast, the more polar the amino acids are, the less likely they are to be found in bitopic proteins compared to polytopic ones. This most likely reflects the ability of K-helical bundles to shield the polarity of residues from the hydrophobic bilayer. One half of all amino acids were distributed nonrandomly in both bitopic and polytopic proteins. A preference was found for tyrosine and tryptophan residues to be at the ends of transmembrane K-helices. Correlated distribution analysis of amino acid pairs indicated that most amino acids are independently distributed in each helix. Exceptions are cysteine, tyrosine, and tryptophan which appear to cluster closely to one another and glycines which are preferentially found on the same side of K-helices. ß 1998 Elsevier Science B.V. All rights reserved.