Investigation of hydrophobic moment and hydrophobicity properties for transmembrane α-helices

Integral membrane proteins are the primary targets of novel drugs but are largely without solved structures. As a consequence, hydrophobic moment plot methodology is often used to identify putative transmembrane α-helices of integral membrane proteins, based on their local maximum mean hydrophobic moment (〈μH〉) and the corresponding mean hydrophobicity (〈H〉). To calculate these properties, the methodology identifies an optimal eleven residue window (L = 11), assuming an amino acid angular frequency, θ, fixed at 100°. Using a data set of 403 transmembrane α-helix forming sequences, the relationship between 〈μH〉 and 〈H〉, and the effect of varying of L and / or θ on this relationship, was investigated. Confidence intervals for correlations between 〈μH〉 and 〈H〉 are established. It is shown, using bootstrapping procedures that the strongest statistically significant correlations exist for small windows where 7 ≤ L ≤ 16. Monte Carlo analysis suggests that this correlation is dependent upon amino acid residue primary structure, implying biological function and indicating that smaller values of L give better characterisation of transmembrane sequences using 〈μH〉. However, varying window size can also lead to different regions within a given sequence being identified as the optimal window for structure / function predictions. Furthermore, it is shown that optimal periodicity varies with window size; the optimum, based on 〈μH〉 over the range of window sizes, (7 ≤ L ≤ 16), was at θ = 102° for the transmembrane α-helix data set. © 2004 Wallace et al; licensee BioMed Central Ltd.