ondary structure, providing an equilibrium model 
for the early "hydrophobic collapse" phase of pro- 
tein folding. Under acidic conditions, certain pro- 
teins, such as apomyoglobin, enter a molten globule 
state with some secondary structure but without a 
defined tertiary structure. The molten globule is 
thought to represent a later step in the protein- 
folding pathway. These nonnative states are often 
difficult to study by current biophysical methods. 
To explore the structure of the model protein- 
folding intermediates described above, a chemical 
approach has been developed to map close contacts 
between a variable reporter residue site and all 
other residues in the protein. A polar iron chelator 
has been developed that can be specifically attached 
to cysteine residues engineered into the protein 
chain. In the presence of ascorbate, the protein che- 
lator adduct generates hydroxyl radicals, which in 
turn cleave accessible peptide bonds in close prox- 
imity to the chelator. The cleavage pattern has been 
examined for a number of nuclease and apomyoglo- 
bin variants. The approach provides structural infor- 
mation for nonnative protein conformations, such as 
the apomyoglobin molten globule. The cleavage of 
a small peptide-reagent adduct has been examined 
in great detail, leading to the proposal of a new 
cleavage mechanism involving diffusible hydroxyl 
radicals. 
Characterization of Folding Intermediates 
in Staphylococcal Nuclease 
The protection of backbone amide hydrogens 
from exchange with solvent protons during protein- 
refolding kinetic experiments has been used to ex- 
amine the onset of secondary structure formation for 
several protein systems in a number of other labora- 
tories. In each case, nuclear magnetic resonance 
(NMR) spectroscopy has been used to follow the 
extent of protection of many assigned amide pro- 
tons, providing detailed structural information for 
folding intermediates. 
This year such an experimental approach was ap- 
plied to staphylococcal nuclease to complement 
the cleavage-mapping approach described above. 
In these experiments, amide-deuterated nuclease 
P117G was rapidly diluted into an H20-based re- 
folding buff^er at low pH, where the intrinsic amide 
exchange rate is slow. After a variable folding time 
(5 ms to several seconds), the protein was "pulsed" 
in a high-pH buff'er in water. Under this condition, 
backbone amide hydrogens exposed to solvent ex- 
change rapidly, while hydrogen-bonded amides 
were protected from exchange. The protein fin- 
ished folding at low pH, where exchange was again 
slow. 
Suitable NMR samples were prepared in a D2O 
buff'er, two-dimensional proton COSY (correlated 
spectroscopy) spectra collected, and amide cross- 
peak volumes integrated to assess the extent of 
amide protection. The results indicate that a subset 
of amide protons are partially protected early in the 
folding pathway, while others are protected only 
gradually with time. The COSY cross-peaks are 
currently being assigned for the nuclease P117G 
variant to the amino acid sequence and structure in 
order to identify the early-folding domain. 
Genetic Analysis of a /3-Turn 
A genetic system has been developed in the labora- 
tory to explore the amino acid sequence require- 
ments for the formation of a jS-turn distant from the 
active site in staphylococcal nuclease (residues 27- 
31) . A combinatorial library over those positions 
has been constructed and screened with the aid of a 
DNase plate assay. A statistical analysis of the result- 
ing sequence data indicates that there are strong and 
different biases for and against many residues at 
each of the positions of the type I' 18-turn. 
Sufficient sequence and activity data have now 
been collected to allow a linear model for /S-turn 
formation to be developed. The resulting model 
predicts turn formation in the nuclease experiment 
and is also successful in predicting type I' |8-turns in 
other proteins of known crystal structure. Crystal 
structures of three nuclease /3-turn variants suggest 
the type I' |8-turn present in the wild-type protein 
should be preserved in a large majority of the vari- 
ants examined. A thermodynamic analysis of this li- 
brary of (8-turn variants is in progress. (This work is 
supported by a grant from the National Institutes of 
Health.) 
Role of Amino Acid Sequence in Determining 
/?-Turn Type 
Staphylococcal nuclease occurs in at least two 
folded conformations that are in slow exchange on 
the NMR time scale. The interconversion of these 
folded conformers and the thermal unfolding ki- 
netics have been investigated by magnetization- 
transfer NMR experiments. Previous work indicates 
that this conformational heterogeneity is due to cis- 
trans isomerization about the peptide bond preced- 
ing proline- 1 17, which is predominantly in the cis 
configuration in the crystal structure. The nuclease 
conformational equilibrium provides an opportu- 
nity to address the physical basis of the relationship 
between the amino acid sequence and protein con- 
formation. The coexistence of two native conforma- 
tions in slow exchange on the NMR time scale al- 
lows a systematic experimental investigation into 
470 
