CRYSTALLOGRAPHIC PHASING AND REFINEMENT 
Axel T. Brunger, Ph.D., Assistant Investigator 
Dr. Briinger's research is focused on the interface 
between theory and experiment in the area of 
structural biophysics. The research tools are simu- 
lation methods of computational chemistry adapted 
to the requirements of macromolecular systems. 
The current effort centers on development and ap- 
plications of macromolecular structure determina- 
tion and refinement based on x-ray crystallographic 
or nuclear magnetic resonance (NMR) spectro- 
scopic data. 
I. Refinement of the Influenza Virus Hemagglutinin 
by Simulated Annealing. 
The hemagglutinin (HA) glycoprotein of the in- 
fluenza virus membrane mediates the receptor- 
binding and membrane fusion activities required 
for entry of the virus into a host cell and is also the 
primary antigen of the virus. The HA is a homotri- 
mer consisting of a large ectodomain, a small trans- 
membrane region, and a small domain inside the 
virus. Each monomer consists of tw^o disulfide 
linked chains, HA^ and HA^, formed by post-transla- 
tional cleavage of a single polypeptide precursor; 
the carboxyl-terminal region of HA^ anchors the HA 
in the membrane. Treatment of the X:31 virus with 
the protease bromelain produces the trimeric 
water-soluble ectodomain BHA (molecular weight 
—210 kDa). Structures of mutant BHAs with altered 
receptor binding and complexes of BHA with cellu- 
lar receptor analogues have been obtained in 
Dr. Don C. Wiley's laboratory (HHMI, Harvard Col- 
o 
lege). Despite the 3 A resolution limit imposed 
by the BHA crystals, there is great interest in using 
these structures for modeling studies, as part of 
an effort to design anti-influenza drugs. In collab- 
oration with Dr. Wiley, these structures were re- 
fined, using Dr. Briinger's method of crystallo- 
graphic refinement by simulated annealing, with 
the goal of obtaining models as close to idealized as 
possible. 
The mutant structure G146D with the best dif- 
fraction data was refined first as a monomer, using 
a reciprocal space method for averaging the x-ray 
structure factor derivatives over the threefold non- 
crystallographic symmetry, and a nonbonded en- 
ergy term describing the interactions of the mono- 
mer with its trimer symmetry mates. Subsequently 
the entire trimer was refined to model lattice inter- 
actions properly, using positional and isotropic 
temperature factor noncrystallographic symmetry 
restraints in those portions of the molecule not in- 
volved in lattice contacts. This structure was then 
used as the basis for refinement of the other three 
crystallographically isomorphous HA mutant struc- 
tures, L226Q, L226QS, and D1112Gs, where the lat- 
ter two structures are complexed with influenza 
virus receptor, sialic acid. 
The refinements accomplished the goals of ob- 
taining HA models of low R factor and favorable 
empirical energies for modeling or analogue drug 
design purposes. The sialic acid complexes 
obtained in this work have significantly better ge- 
ometry than the original coordinates reported pre- 
viously. Refinement by simulated annealing re- 
quired little manual intervention. However, this 
refinement is not fully automatic; there were always 
some regions of the structures that required man- 
ual adjustment. Overall, however, the process 
worked well at idealizing a relatively low resolution 
crystal structure with a minimum of manual inter- 
vention. The refined coordinates of G146D, L226Q, 
L226Qs, and D1112Gs have been deposited in the 
Brookhaven data bank. 
II. Extension of Molecular Replacement: A New 
Search Strategy Based on Patterson Correlation 
Refinement. 
In macromolecular crystallography, the initial de- 
termination of phases by molecular replacement 
(MR) is often attempted if the structure of a similar 
or homologous macromolecule is known (search 
model). MR involves the placement (i.e., rotation 
and translation) of the search model in the unit cell 
of the target crystal to obtain the best agreement 
between calculated and observed diffraction data. 
The optimally placed search model is used to ob- 
tain initial phases for structure building and refine- 
ment. This approach may or may not succeed; 
many successful cases reported involve search mod- 
els with a backbone atomic root-mean-square dif- 
ference of < 1 A from the target structure. 
Recent progress in obtaining approximate three- 
dimensional models of macromolecules from infor- 
mation other than diffraction data suggests an in- 
creased use of MR to solve crystal structures. For 
instance, the database of known protein sequences 
and protein structures is growing rapidly. Tech- 
niques for aligning sequences, such as consensus 
Continued 
569 
