Three-Dimensional Macromolecular and Cellular Structure 
analysis, x-ray crystallographic structural analy- 
sis, and site-directed mutagenesis. Of key impor- 
tance has been the availability of tight-binding 
peptide boronic acids, which provide an excel- 
lent model for the reaction transition state or 
nearby intermediates. Approximately 30 high- 
resolution, extremely well-refined crystal struc- 
tures have now been determined and analyzed. 
These have provided significant new insights into 
the structure of the transition state and the im- 
portance of substrate hydrogen bonding for its 
stabilization, as well as fundamental information 
on steric exclusion, substrate specificity, and en- 
zyme flexibility. 
By mutation we have been able to alter dramati- 
cally the pattern of specificity while maintaining 
or even increasing enzyme activity. Detailed 
structural analysis of two mutants as free enzymes 
and as complexes have provided surprising in- 
sights into the mechanism and complexity of 
specificity and have indicated the crucial role of 
protein flexibility in selectivity. During the past 
year we have made numerous other mutations 
and examined their kinetic and structural 
properties. 
In the past year we began a collaboration with 
Vladimir Basus (University of California, San Fran- 
cisco) to perform a complete two-dimensional 
nuclear magnetic resonance (NMR) analysis on 
the structure of a-lytic protease. We hope this 
will provide new insights to correlated motions 
within the enzyme and be useful as well for the 
folding studies mentioned below. 
Recently we have developed a new method for 
predicting the energetics of protein-substrate in- 
teractions. This approach, based on Ponder- 
Richards rotamers, combined with energetics and 
solvation terms, can predict k^^JK^ with stun- 
ning accuracy. We have used this method to de- 
sign a new enzyme with particular properties, 
and so far the predictions have been remarkably 
accurate. 
Structural and Biochemical Probes of 
Folding of a-Lytic Protease 
a-Lytic protease is synthesized as a prepro- 
enzyme. Experiments in Escherichia coli have 
demonstrated that the 1 66-amino acid precursor 
domain is absolutely required for the proper fold- 
ing of the 198-amino acid protease domain. Fur- 
thermore, we have shown that proper folding can 
be accomplished either in vivo or in vitro with 
the pro region covalently attached or in trans. 
Our current thinking is that a high-energy barrier 
exists between the folded and unfolded states 
that the mature protease cannot cross by itself. 
The precursor acts as a "foldase" to stabilize the 
transition state for folding, essentially serving as a 
template on which the mature enzyme finds its 
active conformation. 
Current efforts are focused on analyzing this 
folding reaction in vitro. Amazingly, we have 
been able to trap and purify a stable folding inter- 
mediate that is rapidly refolded upon addition of 
the pro region. We plan to use a combination of 
physical approaches including two-dimensional 
NMR to probe the structure of the intermediate 
and the role of the pro region in the final stage of 
folding. 
Structure of Apolipoprotein E 
Apolipoprotein E is an important protein in 
cholesterol metabolism in mammals. It is one of 
two proteins that can bind to the low-density li- 
poprotein (LDL) receptor (the other is apolipo- 
protein B) and thus has a major role in triglycer- 
ide and cholesterol metabolism. The protein 
itself has two distinct structural and functional 
domains. The amino-terminal 22-kDa domain 
contains the receptor binding functionality, 
whereas lipid binding resides primarily with the 
10-kDa carboxyl-terminal domain. In collabora- 
tion with the Mahley group at the Gladstone 
Foundation Laboratories for Cardiovascular Dis- 
ease, we have obtained crystals of the 22-kDa re- 
ceptor-binding domain and recently finished the 
high-resolution structure determination. The 
protein is an unusually elongated four-helix bun- 
dle. Although the surface is exceptionally 
charged, positive and negative charged groups 
are precisely balanced, except in what we be- 
lieve is the receptor-binding region. Currently, 
we are examining the structures of two human 
mutants that disrupt receptor binding. 
6 
