Prediction and Comparison of Coiled-Coil Proteins in Multiple Genomes

The coiled-coil structure is a typical hyper-secondary structure formed by the mutual intertwining of multiple alpha-helices, and is predicted as the structure of keratin of the intermediate filament by Crick in 1952 [2]. In 1988, the leucine zipper in the structure of transcription factors was found to be a form of the coiled-coil, and in succeeding years, a variety of biological applications of coiled-coils has been discovered: For example, structural proteins such as desmosomal proteins, motor proteins such as myosin, and some proteins related to the dynamic structure of membranes such as SNAREs. In ordinary case, coiled-coil molecules assemble with other coiled-coil molecules, so this may be the simple model of the general case of the receptor-ligand binding. Before now, several algorithms for prediction of coiled-coils were proposed: The first algorithm was proposed by Parry (1982)[6] and by Lupas (1991)[4], using the periodic occurrence of hydrophobic amino acids in the sequence of a coiled-coil. After that, pairwise correlation algorithm, which is the improvement of the scoring method, was proposed by Berger and Kim (1995)[1], and an another prediction method using hidden Markov model was proposed by Delorenzi and Speed (2002)[3]. Such a structure as to contribute to the broad range of biological phenomena and as to be predicted with fair accuracy is suitable for the comparative genomic analysis, though only few studies were published in which the coiled-coil prediction and the analysis was performed against the whole genome [5]. Hence, we collected complete genomic sequences of 18 eukaryotic and 147 prokaryotic organisms, and partial genomic sequences of 3 eukaryotic organisms, and predicted the coiled-coil molecules.