Structural Studies of Macromolecular Assemblies 
Stephen C. Harrison, Ph.D. — Investigator 
Dr. Harrison is also Professor of Biochemistry and Molecular Biology at Harvard University and Research 
Associate in Medicine at the Children's Hospital, Boston. He received his A.B. degree in chemistry 
and physics from Harvard College and his Ph.D. degree in biophysics from Harvard University. 
Dr. Harrison is a member of the National Academy of Sciences. 
HOW do regulatory proteins activate or in- 
hibit transcription of particular genes? How 
do viruses leave and enter cells? How do recep- 
tors and their ligands cycle from cell surface to 
ceil interior and back? These questions deal with 
molecular recognition in the determination of 
cell organization. They represent groups of proj- 
ects in our laboratory, all of which involve eluci- 
dation of detailed atomic structures as a prerequi- 
site for tackling functional problems. 
Transcriptional Regulatory Complexes 
A common characteristic of eukaryotic tran- 
scriptional regulatory elements is the presence of 
sites in multiple copies varying slightly in se- 
quence, often with two or more related proteins 
that can bind to them. The best-understood pro- 
karyotic paradigm is in the immunity region of 
temperate bacteriophages, where two proteins, 
repressor and Cro, bind two sets of three sites, 
with appropriately graded affinities. We have 
made an effort to understand the mechanism of 
this regulatory switch, by determining the struc- 
tures of a series of specific protein/DNA com- 
plexes containing the Cro protein of phage 434 
or the DNA-binding domain of its repressor. We 
are using computational approaches to link ob- 
served structural differences among these various 
complexes with the corresponding free energies 
of binding. 
We are studying structural aspects of eukaryo- 
tic transcriptional regulation, initially by examin- 
ing the DNA-binding domains of regulatory pro- 
teins in complex with synthetic binding sites. 
GAL4, the prototype of a class of such proteins, is 
a regulator of galactose metabolism in yeast. It 
binds as a dimer to the upstream activity se- 
quences of several genes involved in galactose 
and melibiose catabolism. The DNA-binding re- 
gion is at the amino terminus of the 881 -residue 
polypeptide chain. We have determined the 
structure of a fragment containing residues 1-65, 
in a specific complex with DNA. A small domain 
containing zinc ions (residues 8-40) recognizes 
a conserved CCG triplet at each end of the 17- 
base pair binding site, through direct, major- 
groove contacts. A short a-helical coiled-coil ele- 
ment imposes twofold symmetry. A segment of 
extended polypeptide chain, linking the metal- 
binding module to the dimerization element, 
specifies the length of the site. The complex has a 
relatively open structure, which would allow an- 
other protein to bind coordinately with GAL4 . Co- 
ordinate binding of two or more proteins has 
been shown to be an important feature of many 
eukaryotic control elements. 
GCN4, also a yeast transcriptional regulator, 
represents yet another class of DNA-binding pro- 
teins. It contains a dimerization element, gener- 
ally called a leucine zipper, which forms an 
a-helical coiled coil about 30 residues in length. 
This segment is preceded in the protein sequence 
by a positively charged region, which has little 
ordered structure in the free protein but which 
also acquires a-helical structure when it binds 
to DNA. We have prepared crystals of the basic 
region/leucine zipper fragment of GCN4, in 
complex with a synthetic binding site (having a 
so-called AP- 1 sequence) . The structure determi- 
nation is nearly complete. Preliminary analysis 
shows that each chain is a continuous a-helix. 
The part corresponding to the basic region lies 
along the major groove, and contacts can be made 
by suitable side chains to four base pairs on either 
side of a central GC. 
TflllA, which controls 5S RNA transcription in 
Xenopus, represents the so-called zinc finger 
class of regulatory proteins. The finger is a small, 
30-residue domain, stabilized by a tightly bound 
zinc ion. A recombinant fragment comprising 
seven of the nine fingers from TflllA binds to a 
30-base pair DNA containing an appropriate part 
of the total binding site, and we have crystallized 
this complex. The fragment also binds specifi- 
cally to 5S RNA. 
Understanding how these various structures 
recognize their DNA sites is only a beginning. The 
specificities of interactions between other do- 
mains of these proteins and additional compo- 
177 
