Yukio Kobayashi1, Nobuhiko Saito2, Makoto Ohta3, Hiroyuki Sasabe4, and Shigeki Mitaku3
1 Department of Information Systems Science, Faculty of Engineering, Soka University, Hachioji, Tokyo 192, Japan.
2 Department of Applied Physics, Waseda University, Shinjuku-ku, Tokyo 169, Japan.
3 Faculty of Technology, Tokyo University of Agriculture and Technology, Koganei, Tokyo 184, Japan.
4 Frontier Research Program, The Institute of Physical and Chemical Research(RIKEN), Wako, Saitama 351-01, Japan.
Prediction of the Structures of GPCT and HPR by the Island Model
We predict the structures of GPCT and HPR with ab initio method based
on the mechanism of protein folding, which is referred to as the
"island model"(Saito et al., 1988). This method is formulated on the
physicochemical basis, and thus it is applicable to any type of
proteins with low homology. We have selected these proteins consisting
of a small number of residues in order that the structures of these
proteins can be predicted with our method in time for the deadline of
prediction.
We assume that folding starts first with the formation of the secondary
structures and then proceeds to assemble them into the tertiary
structure. The procedures of predicting a protein structure are
summerized as follows:
1. Determination of secondary structures. Perform the three-state
prediction with statistical mechanical method already formulated before
(Saito, 1987) to provide simultaneously the probabilities of each
residue in alpha-helix, beta-strand or coil. This formulation involves the
following assumptions:
- (1) At least one residue in coil exists between neighboring secondary structures.
- (2) Interactions are considered only among the residues within four
residues along the chain in the same secondary structure. The above
probabilities are calculated with statistical weights for amino acid
pairs in alpha-helix or in beta-strand. We have estimated the weights so as
to minimize an objective function by referring to 80 proteins for
optimization which lack homology among them as far as
possible.
2. Model of protein chain.
- (1) Build a polypeptide chain with QUANTA (software applications
for modeling the structures and behavior of molecular systems).
- (2) Replace side chains of amino acid residues by a sphere of van
der Waals radius.
- (3) Locate each sphere at the average distance of the center of
gravity on the direction of the beta-carbon from the alpha-carbon.
- (4) Fix bond lengths and bond angles at the standard values.
- (5) Construct the initial conformation with the predicted secondary
structures and the extended conformations (f=180!, y=180!) for
others.
3. Packing of secondary structures into tertiary structure.
- (1) Calculate energies by considering long-range hydrophobic
interactions and Lennard-Jones (12,6)-potentials between relevant atoms
from short distance pairs to longer ones step by step.
- (2) Pack the secondary structures through the local structure
formation by searching for the conformation of lower energy.
4. Refinement of the conformation with QUANTA. Generate the atoms in
the side chains and introduce the electrostatic potentials ignored in
the above step.
We represent the predicted conformations of GPCT and HPR with the
distance maps, where the conformations are indicated on the basis of
the distances between alpha-carbons. These maps will be shown in the
poster presentation.
Saito, N. et al. (1988) Proteins, 3, 199-207.
Saito, N. (1987) Cell Biophys., 11, 321-329.
Asilomar Conference home page