Simulation
of Protein-Membrane Interactions
by
Implicit Solvent Models
This
project was motivated by the need for studying the molecular
mechanisms
of membrane damage induced by a protein toxin, the
cytolytic
toxin Cyt1A. Cyt1A exhibits strong cytolytic activity
against a
wide range of insect and mammalian cells, and has been
used as an
environmentally safe insecticide. However, since the
structural
details, which crucially link to the biological
mechanisms,
are too difficult to be produced by chemical and
physical
experiments, protein structure modeling seems to be the
only tool to
provide insight in the mechanism of Cyt1A-mediated
damage to
the lipid membrane.
The
ultimate goal of protein modelers is
to develop an algorithm
that will
compute the protein structure from the amino-acid
sequence,
which leads to the protein folding problem---one of the
most
challenging research subjects in science and technology. In
this project
we have a more modest objective, which is still,
nevertheless,
quite ambitious: the search for a fast and reliable
algorithm
for simulation of a family of particular proteins that
can ably
preserve their secondary structures a-helix and
b-sheet
in the folding process. According to the
preliminary
experimental data provided by the CO-PI of this
project, Dr.
Butko (Professor of Biochemistry), the protein Cyt1A
possesses
this feature.
Using
Cyt1A as a model protein, we will develop a new implicit
solvent
model for predicting the structure of protein.
The new
model will
be tested and used by Dr. Butko in
his study of the
Cyt1A-membrane
system. The new model will be extremely valuable in
analyzing
the damage mechanism of Cyt1A to the membrane. In
addition, a
new function of biomolecular potential energy function
will be
designed based on the special properties of protein Cyt1A.
The new
potential energy function will use only a small subset of
the
conformation coordinates of protein so that the computing
complexity
of simulation is reduced sharply.
The
involved mathematical problems in this project include (1) a
theoretical
and numerical analysis of a nonlinear partial
differential
equation --- the Poisson-Boltzmann equation (PBE),
and (2) a
global minimization of a strongly nonlinear large scale
function
(such as the biomolecular potential energy function). In
our project,
PBE is used to approximate the solvent effects on the
surface of
the protein through modeling the solvent as a
dielectric
continuum material. Fast solvers of PBE will be
developed by
using multigrid and domain decomposition techniques
and
truncated-Newton minimization techniques. The new global
minimization
algorithm will be designed by a novel combination of
several
advanced numerical techniques such as homology modeling,
lattice
modeling, Monte Carlo methods, and the truncated-Newton
minimization
algorithm implemented in TNPACK.
This
project will be performed in close collaboration with the co-PI
Dr. Peter Butko , who is a professor of Biochemistry at the University of Southern Mississippi.