PREFERENCE FUNCTIONS FOR PREDICTION OF MEMBRANE-BURIED HELICES IN INTEGRAL MEMBRANE-PROTEINS

The preference functions method is described for prediction of membran e-buried helices in membrane proteins. Preference for the a-helix conf ormation of amino acid residue in a sequence is a non-linear function of average hydrophobicity of its sequence neighbors. Kyte-Doolittle hy dropathy values are used to extract preference functions from a traini ng data set of integral membrane proteins of partially known secondary structure. Preference functions for beta-sheet, turn and undefined co nformation are also extracted by including beta- class soluble protein s of known structure in the training data set. Conformational preferen ces are compared in tested sequence for each residue and predicted sec ondary structure is associated with the highest preference. This proce dure is incorporated in an algorithm that performs accurate prediction of transmembrane helical segments. Correct sequence location and seco ndary structure of transmembrane segments is predicted for 20 of 21 re ference membrane polypeptides with known crystal structure that were n ot included in the training data set. Comparison with hydrophobicity p lots revealed that our preference profiles are more accurate and exhib it higher resolution and less noise. Shorter unstable or movable membr ane-buried alpha-helices are also predicted to exist in different memb rane proteins with transport function. For instance, in the sequence o f voltage-gated ion channels and glutamale receptors, N-terminal parts of known P-segments can be located as characteristic alpha-helix pref erence peaks. Our e-mail server: predict@drava.etfos.hr, returns a pre ference profile and secondary structure prediction for a suspected or known membrane protein when its sequence is submitted. (C) 1998 Elsevi er Science Ltd. All rights reserved.