Simulations nab protein-folding mistakes
By I. Peterson
The human body manufactures thousands of different proteins, which
act as enzymes, structural elements, or carriers. Each protein starts
out as a long strand that must fold itself into the proper shape to
perform its specific function.
To gain insights into the folding process, researchers have
developed mathematical models that attempt to capture how a single
strand rapidly collapses into its correct configuration (SN: 5/9/98, p.
296). New computer simulations now reveal how one protein strand can
interfere with the folding of another.
"This is a step toward successful protein engineering," says
computer scientist Sorin Istrail of the Sandia National Laboratories in
Albuquerque. "It provides our first clue in how to design sequences of
laboratory proteins that can survive the essential but complicated
folding process."
The results could also help laboratory scientists understand the
mechanisms underlying protein-folding mistakes, which are implicated in
Alzheimer's disease (SN: 7/4/98, p. 4) and other maladies.
Istrail and his collaborators, biologist Jonathan A. King and computer
scientist Russell Schwartz of the Massachusetts Institute of
Technology, report their findings in an upcoming Journal of Computational Biology.
Each protein is a string of amino acids spelling out a
characteristic sequence. Scientists divide the 20 distinct amino acids
into two groups according to whether they attract or repel water
molecules, the main constituent of cells. Those interactions drive
protein folding, creating globular structures in which water-repelling
amino acids end up on the inside and water-attracting ones on the
outside.
A simple model portrays a protein as a stiff but jointed structure,
made of two types of beads, that can snap into a few positions at each
joint. Because of the huge number of different folding possibilities
for even a short strand, however, simulating the process on a computer
has proved difficult.
Moreover, protein strands in cells have many jostling neighbors,
King notes. Two partially folded protein molecules may end up sticking
together.
To capture that additional complication, Istrail and his coworkers
started with a well-studied, jointed-structure model originally
developed by Ken Dill of the University of California, San Francisco.
Taking advantage of Sandia's powerful computers (SN: 7/5/97, p. 5), the
researchers examined what happens when strands move about and bump into
each other.
Although the specific amino acid sequence determines the protein's
function, the location of water-attracting components appears to be
random. By tracking two highly simplified protein chains interacting on
a grid, Istrail and his coworkers showed that this irregular
arrangement plays an important role in preventing water-repelling units
from binding to those of another strand.
A protein with an ordered distribution of water-attracting units
tends to aggregate with other proteins to form an inert lump, Istrail
says. Random positioning appears necessary to keep different strands
apart, allowing the formation of the correct folds.