Integral membrane proteins (of the alpha-helical class) are of central
importance in a wide variety of vital cellular functions. Despite con
siderable effort on methods to predict the location of the helices, li
ttle attention has been directed toward developing an automatic method
to pack the helices together. In principle, the prediction of membran
e proteins should be easier than the prediction of globular proteins:
there is only one type of secondary structure and all helices pack wit
h a common alignment across the membrane. This allows all possible str
uctures to be represented on a simple lattice and exhaustively enumera
ted. Prediction success lies not in generating many possible folds but
in recognizing which corresponds to the native. Our evaluation of eac
h fold is based on how well the exposed surface predicted from a multi
ple sequence alignment fits its allocated position. Just as exposure t
o solvent in globular proteins can be predicted from sequence variatio
n, so exposure to lipid can be recognized by variable-hydrophobic (var
iphobic) positions. Application to both bacteriorhodopsin and the euka
ryotic rhodopsin/opsin families revealed that the angular size of the
lipid-exposed faces must be predicted accurately to allow selection of
the correct fold. With the inherent uncertainties in helix prediction
and parameter choice, this accuracy could not be guaranteed but the c
orrect fold was typically found in the top six candidates. Our method
provides the first completely automatic method that can proceed from a
scan of the protein sequence databanks to a predicted three-dimension
al structure with no intervention required from the investigator. With
in the limited domain of the seven helix bundle proteins, a good chanc
e can be given of selecting the correct structure. However, the limite
d number of sequences available with a corresponding known structure m
akes further characterization of the method difficult. (C) 1994 Wiley-
Liss, Inc.