X-ray Diffraction and Computer Simulation Studies 
of Protein Function 
John Kuriyan, Ph.D. — Assistant Investigator 
Dr. Kuriyan is also Assistant Professor of Molecular Biophysics at the Rockefeller University. He graduated 
from Juniata College in Pennsylvania, with a B.S. degree in chemistry. He received his Ph.D. degree from 
the Massachusetts Institute of Technology, working jointly with Gregory Petsko and Martin Karplus at 
Harvard University on the dynamics of proteins. He continued in the Karplus laboratory as a postdoctoral 
fellow before moving to Rockefeller as a University Fellow. He is also a Pew Scholar in the Biomedical 
Sciences. 
OUR interests are in characterizing the struc- 
tures of proteins, using crystallography — 
namely, x-ray diffraction experiments — and 
computer simulations. Crystallography and 
computer simulations are complementary ap- 
proaches to understanding protein function. The 
x-ray experiments provide three-dimensional 
structures, which are currently impossible to gen- 
erate from theoretical considerations alone, and 
the simulations allow us to visualize the effects of 
extrapolations that are not experimentally feasi- 
ble. We apply the knowledge gained from these 
studies toward the design of mutations and inhibi- 
tors to modify protein activity. 
Part of our work involves applications of the 
molecular dynamics (MD) method to problems 
in protein crystallography. This powerful tool for 
simulating protein structure and structural rela- 
tionships involves computing the forces between 
all the atoms in a protein and generating trajec- 
tories of the atomic motion in response to these 
forces. Reasonably realistic MD simulations of 
solvated enzymes or proteins in crystal lattices 
are now possible. The simulations, however, rely 
on rather approximate interatomic potential 
functions and, even with the fastest computers 
available today, are limited to very short time- 
scales. We are carefully testing and evaluating the 
results of MD calculations through simulation of 
small proteins, such as crambin and ribonucle- 
ase, in the crystal environment, thus permitting 
detailed comparison with high-resolution x-ray 
diffraction data. 
In related work, we are developing methods 
for the better treatment of dynamics in refining 
crystallographic structure. 
Our projects in the determination of crystallo- 
graphic structure encompass two major areas-, 
one involves redox proteins and the transcrip- 
tional response to oxidative stress (in collabora- 
tion with Peter Model and Anthony Cerami at the 
Rockefeller University and with Gisela Storz at 
the National Institutes of Health), and the other is 
concerned with transcription factors in Drosoph- 
ila development (in collaboration with Claude 
Desplan, HHMI, the Rockefeller University). We 
focus here on just the first topic. 
One of the efforts in this area is aimed at devel- 
oping an understanding of the catalytic and sub- 
strate recognition mechanisms of two related 
redox enzymes, thioredoxin reductase and try- 
panothione reductase. These are members of a 
widely distributed family that channel the reduc- 
tive power of NAD(P)H via a protein-bound fla- 
vin (FAD), to an active-site disulfide bond. Other 
well-known members include glutathione reduc- 
tase (for which a high-resolution x-ray structure 
was first determined), lipoamide dehydrogenase, 
and mercuric ion reductase. 
Thioredoxin is a small redox-active protein 
that is the reductant of ribonucleotide reductase 
in the DNA synthesis pathway. It has diverse other 
functions, ranging from the light-activated regula- 
tion of enzyme pathways in plants to the catalysis 
of protein disulfide isomerization. In Esche- 
richia coli, thioredoxin is maintained in the re- 
duced state by the action of NADPH-dependent 
thioredoxin reductase (TR). Previous x-ray dif- 
fraction studies on the related enzyme glutathi- 
one reductase (GR) revealed a dimeric structure 
(Af, = 2 X 52,400) with four domains within 
each molecule: the FAD- and NADPH-binding do- 
mains, the "central" domain that is also responsi- 
ble for binding FAD, and the carboxyl-terminal 
domain that provides the dimer interface as well 
as residues that are critical for substrate binding 
and catalytic activity. 
Although TR and GR both catalyze the same 
chemical reaction, that of disulfide reduction in 
their substrates, TR is significantly smaller (M^ = 
2 X 34,500) and lacks the interface domain that 
in GR forms part of the active site. Another funda- 
mental difference in their sequences involves the 
redox-active disulfide, which in GR is part of a 
highly conserved hexapeptide (Cys-Val-Asn-Val- 
Gly-Cys) in the FAD domain, and in TR is a shorter 
segment (Cys-Ala-Thr-Cys) found instead in the 
NADPH domain. 
The three-dimensional structure of TR, deter- 
mined at 3.0 A resolution by multiple isomor- 
phous replacement (in collaboration with Peter 
Model and with Charles Williams at the Univer- 
sity of Michigan) and initially refined at a resolu- 
253 
