Structural Basis of Interactions Within and Between Macromolecules 
This suggests a way to reconcile the different 
values for the apparent strength of the hydropho- 
bic effect. One can imagine two extreme situa- 
tions. In one case a leucine alanine replace- 
ment is constructed, and the protein structure 
remains completely unchanged. In this situation 
the size of the created cavity is large, and the mu- 
tant protein is maximally destabilized. In the 
other extreme, the protein structure relaxes in 
response to the leucine -•- alanine substitution, 
fills the space occupied by the leucine side 
chain, and so avoids the formation of any cavity 
whatsoever. In this case the decrease in energy of 
the mutant protein relative to wild type drops to a 
constant energy term that is characteristic for a 
leucine alanine replacement. 
Ligand Binding Within Cavities 
We have shown by crystallographic and thermo- 
dynamic analysis that the cavity created by the 
replacement leucine 99 alanine in T4 lyso- 
zyme is large enough to bind benzene and that 
ligand binding increases the melting temperature 
of the protein by 5.7°C. This shows that cavities 
can be engineered within proteins and suggests 
that such cavities might be tailored to bind spe- 
cific ligands. The binding of benzene at an inter- 
nal site 7 A from the molecular surface also illus- 
trates the dynamic nature of proteins, even in 
crystals. 
Receptor-Ligand Interaction 
To develop an understanding of the mode of 
action of growth factors and their interactions 
with their receptors, we have crystallized and de- 
termined the high-resolution structure of human 
fibroblast growth factor. The structure is very sim- 
ilar to that of interleukin-I/3. Clearly, many 
growth factors have similar overall structures. 
but the exact relationship of these factors in the 
vicinity of their receptor-binding regions remains 
to be clarified. Structural studies of human nerve 
growth factor are in progress. 
Protein-DNA Interaction 
We have been interested for some time in the 
interaction between proteins and nucleic acids. 
In 1981 we determined the structure of the Cro 
repressor protein of X bacteriophage (bacteria- 
infecting virus), one of the prototypical exam- 
ples of a DNA-interacting protein. The structure 
of Cro, as determined crystallographically, sug- 
gested that a characteristic part of the protein, 
now known as the helix-tum-helix motif, is espe- 
cially important in DNA binding. The helix-turn- 
helix unit can be considered as a "reading head" 
that fits into the grooves of the DNA and matches 
the DNA structure at the specific recognition site. 
The helix-turn-helix motif is now known to occur 
in a large number of DNA-binding proteins, and 
its functional role has been confirmed by struc- 
tures of several DNA-protein complexes. 
We have subsequently determined the crystal 
structure of Cro protein in complex with a tight- 
binding 1 7-base pair DNA operator. In general 
terms the structure of the complex supports the 
model for Cro-DNA interaction that was proposed 
on the basis of the uncomplexed protein, al- 
though the Cro dimer undergoes a substantial 
conformational change relative to the uncom- 
plexed crystal structure. 
Recently we have determined the structure of 
the biotin repressor from Escherichia coli. This is 
a more complicated protein that requires the 
presence of an effector molecule to bind DNA. It 
also acts as an enzyme. 
Studies of protein stability and protein-DNA in- 
teraction were supported in part by grants from 
the National Institutes of Health. 
278 
