We show that the peptide backbone of an oc-helix places a severe thermodyna
mic constraint on transmembrane (TM) stability. Neglect of this constraint
by commonly used hydrophobicity scales underlies the notorious uncertainty
of TM helix prediction by sliding-window hydropathy plots of membrane prote
in (MP) amino acid sequences. We find that an experiment-based whole-residu
e hydropathy scale (WW scale), which includes the backbone constraint, iden
tifies TM helices of membrane proteins with an accuracy greater than 99%. F
urthermore, it correctly predicts the minimum hydrophobicity required for s
table single-helix TM insertion observed in Escherichia coli. In order to i
mprove membrane protein topology prediction further, we introduce the augme
nted WW (aWW) scale, which accounts for the energetics of salt-bridge forma
tion. An important issue for genomic analysis is the ability of the hydropa
thy plot method to distinguish membrane from soluble proteins. We find that
the method falsely predicts 17 to 43% of a set of soluble proteins to be M
Ps, depending upon the hydropathy scale used. (C) 2001 Academic Press.