Folding Proteins on a Computer
January 19, 2001
Proteins only function when properly folded
In order for enzymatic reactions to proceed correctly, the enzyme and substrate must fit
together as precisely as a lock and key. The function of the enzyme and protein is determined by
the structure of the latter. The chain of amino acids that makes up the protein thus has to fold in a
very precise fashion. Nothing had better go wrong: the cattle epidemic of BSE seems to have
stemmed from an incorrectly folded protein.
How the folding process of proteins proceeds on an atomic level has so far not been
experimentally determined. A team of chemists working with Wilfred van Gunsteren is now hot
on the heels of protein folding -- with the help of computer simulations.
The number of possible conformations that a protein can theoretically adopt rises
exponentially with the length of its chain of amino acids. This quickly reaches astronomical
orders of magnitude, making it impossible to test all of the possibilities on a computer. Thanks to
the enormous computing power of modern supercomputers, it is now at least possible to simulate
the folding of very short peptide chains. Repulsion and attraction between individual atoms
determine the balance between the folded (native) and unfolded (denatured) states of the protein.
Taken together, all of the spatial characteristics of these forces make up a force field. Van
Gunsteren recognized that it is necessary to precisely represent both the final folded state and the
unfolded starting state of the protein as force fields in order to simulate the folding process. A
hopeless task, it seemed – the number of theoretically possible unfolded states is simply too huge.
Precisely this assumption has now been revealed to be false. In fact, the number of states is
actually comparatively small, as revealed by simulations carried out for several small peptides.
„This discovery pushes the simulation of protein folding processes within reach,“ says van
Gunsteren optimistically.
But how does one come up with the necessary force field for the unfolded protein when there
is no experimental data available? Van Gunsteren and his colleagues made do with data on the
interactions within and between small molecules in solution. Starting with this carefully worked
out force field, the team was in a position to simulate the folding of a series of peptides.
If this method of simulating the folding process also works for larger proteins, one of the
fundamental challenges of molecular biology comes within reach: the prediction of the spatial
structures of unknown proteins.
WILEY-VCH Verlag GmbH