Chemical Details of Cellular Regulation 
pected. Most of the direct interactions are hydro- 
gen bonds that appear to lash the protein to the 
sugar phosphate backbone on both sides of the 
DNA's major groove. In some instances a single 
phosphate accepts four hydrogen bonds. An inti- 
mate specific fit is partly accomplished by a de- 
formation of the DNA, which we believe to be 
allowed specifically by the target's sequence at 
little or no cost in internal energy. We refer to this 
sequence-dependent deformability as "indirect 
read-out" of the sequence. 
Most surprising is the array of fixed water mole- 
cules (20 per complex) that help mediate the 
specific interaction. Ten of these are also found 
in the uncomplexed repressor protein and there- 
fore represent noncovalent extensions of the pro- 
tein's "recognizable" surface. Four of the ten 
fixed water molecules provide six highly specific 
hydrogen bonds to bases that define the trp opera- 
tor as the target for the trp repressor. Recent mu- 
tational studies designed to test the functional 
significance of these specific water-mediated 
contacts support the idea that water molecules 
can mediate molecular recognition. The obliga- 
tory involvement of water in molecular recogni- 
tion has also been reported recently for highly 
specific interactions of proteins with sugars and 
RNA molecules. 
Our most recent work has focused on under- 
standing DNA recognition in higher cells. We are 
trying to determine the crystal structures of a vari- 
ety of regulatory protein-DNA complexes that 
control metabolism, grovvT:h, differentiation, and 
the expression of viral-specific genes. Some of 
these complexes closely resemble the products 
of oncogenes. 
We have just determined the crystal structure 
of the DNA-binding domain of the glucocorticoid 
(Cortisol) receptor in a complex with its DNA tar- 
get. The ability of this protein and each of its 
close homologues to bind to its specific target 
sequence lies at the heart of the selective expres- 
sion of genes in response to each of the different 
steroid hormones. Despite their discrimination in 
targeting the correct regulatory sequence, the 
DNA-binding domains of all the steroid receptor 
"superfamily" (thyroid hormone included) have 
a highly conserved sequence and presumably 
very similar structures. We found that these DNA- 
binding domains contain two intriguing sub- 
structures called zinc fingers, found in many 
genetic regulatory proteins. Mutagenesis experi- 
ments done elsewhere have indicated where to 
focus our attention, and we can now account for 
the mechanism by which the glucocorticoid re- 
ceptor can discriminate between its target DNA 
and those of the estrogen and thyroid receptors. 
Work on other systems is not quite so far along. 
We hope to grow better DNA-protein crystals 
containing the binding domain of the "leucine 
zipper proteins," another important type of tran- 
scription factor first characterized by Steven 
McKnight (HHMI at the Carnegie Institution). 
Work also continues with Laimonis Laimins 
(HHMI at the University of Chicago) on strongly 
diffracting crystals of the specific DNA complex 
of a regulator protein, E2, which can trigger the 
transforming activity of the potentially onco- 
genic papilloma virus. 
410 
