Computational Structural Biology 
of proteins based on their sequence remains im- 
possible. This fundamental problem of structural 
biology (the "folding problem") is still un- 
solved, despite improvements in computational 
techniques for macromolecular simulation and 
CQjnputer hardware. Nevertheless, macromolecu- 
lar simulation has been successful in predicting 
localized conformations if sufficient experimen- 
tal constraints or restraints (e.g., in the form of an 
x-ray structure) are available. It is therefore con- 
ceivable that other more global predictions are 
possible if appropriate experimental information 
is available. We have embarked on trying to pre- 
dict the association and stability of helices that 
form coiled coils. Conformational search strate- 
gies are being employed with empirical energy 
functions, using molecular dynamics and energy 
minimization. 
Presently we shall apply this approach to the 
family of leucine zipper proteins, which are se- 
quence-specific DNA-binding proteins that regu- 
late gene expression in certain mammalian cells. 
We have successfully predicted the structure of 
the dimerization domain of GCN4, for which a 
high-resolution x-ray has become available 
(Thomas Alber, University of Utah). 
Macromolecular Simulation of Free-Energy 
Differences 
We are involved in a number of projects 
aimed at simulating free-energy differences be- 
tween two states of a biological system, using 
the so-called free-energy perturbation tech- 
nique. The goal is to investigate microscopi- 
cally the structure and stability of protein sec- 
ondary structural elements and protein-peptide 
complexes. Furthermore, we would like to eval- 
uate the reliability of free-energy calculations 
and molecular dynamics simulations as tools 
for this purpose. 
One project concerns the complexes of bovine 
pancreatic ribonuclease S and a number of mu- 
tants of the S-peptide for which x-ray crystal 
structures, binding free energies, and enthalpies 
have been obtained by Frederic Richards and Ju- 
lian Sturtevant (Yale University) . Another project 
involves the study of a number of site-directed 
mutants of |8-turns in staphylococcal nuclease in a 
joint project with Robert Fox (HHMI , Yale Univer- 
sity) for which x-ray crystal structures and infor- 
mation about the cis to trans equilibria of a pro- 
line side chain in the turn have been obtained in 
Dr. Fox's laboratory. 
Illustration of hulk solvent regions in protein crystal struc- 
tures. The regions are indicated by blue dots in this unit cell of a 
crystal of penicillopepsin from Penicillium janthinellum, whose 
structure was solved in Michael fames' laboratory (Alberta, 
Canada) at 1.8- A resolution. The protein backbone atoms ap- 
pear as yellow lines; ordered water molecules, as pink dots. 
Bulk solvent typically constitutes 40-60 percent of the unit cell 
contents, yet its physical characteristics and biological func- 
tion are poorly understood. 
Research and photograph by Axel Brunger, using a graphics 
interface developed by Warren Delano. 
58 
