extrude messages from each of the dsRNA segments. 
They are both ~ 700 A in diameter, and their crystals 
have unit cell dimensions > 1 ,000 A. Data to ~ 1 5-A 
resolution have been collected from the crystals of 
rotavirus SSPs, and observed diffraction extends to at 
least 5 A. Efforts are under way to compute the low- 
resolution structures of these particles. 
The envelope protein of tick-borne encephalitis 
virus (TBE) is the principal viral antigen, and it is 
believed to mediate receptor binding and mem- 
brane fusion. A dimeric soluble fragment (polypep- 
tide ~ 45,000), comprising ~80% of the mole- 
cule, can be released from virions by proteolytic 
cleavage. Crystals of this soluble form diffract to 
2.6-A resolution. A low-resolution (5-A) electron 
density map shows the overall shape of the subunit. 
Work is in progress to extend the resolution of the 
phase determination, in order to build a complete 
model of the folded polypeptide chain. 
Receptors 
The endocytosis of receptors for proteins such as 
growth factors, transferrin, and LDL (low-density li- 
poprotein) involves directed uptake by coated vesi- 
cles, transport to internal compartments where li- 
gand dissociation and sorting can occur, recycling 
to the cell surface in some cases, and delivery to 
lysosomes in others. The transferrin receptor (tfR), 
one of the best characterized of the proteins that 
participates in this pathway, is a homodimer of 90- 
kDa subunits. 
Crystals of a 70-kDa tryptic fragment, termed 
tfR-t, which represents nearly all of the extracellular 
part of tfR, will allow molecular details of such a 
receptor to be visualized. Substantial progress has 
been made during the past year toward determining 
the tfR-t structure. Crystals grown from tfR-t derived 
from placental preparations diffract to at least 3- 5-A 
resolution, but radiation sensitivity and a large unit 
cell make data collection difficult. Accurate diffrac- 
tion data can now be measured, using "frozen" crys- 
tals at — 170°C and high-intensity synchrotron 
beams, and several excellent data sets have been 
recorded. 
Expression of a recombinant, soluble receptor, 
corresponding almost precisely to the tryptic frag- 
ment, has also just been accomplished. The prod- 
uct, secreted in good yield from CHO cells, is di- 
meric and binds transferrin. Its availability should 
substantially facilitate the crystallographic studies. 
Crystals of the complete "four-domain" extracel- 
lular segment of the T cell co-receptor CD4 are also 
difficult to study — probably because CD4, like tfR, 
is an elongated molecule that packs awkwardly into 
a very large crystalline unit cell. Again in this case. 
low-temperature data collection has proved to be 
crucial for accurate diffraction measurements. The 
structure of a two-domain fragment was reported by 
Dr. Harrison's group in 1990, and a picture of the 
complete extracellular segment will be important 
for understanding the binding of CD4 to class II 
MHC (major histocompatibility complex) mole- 
cules. (The CD4 project is supported by grants from 
the National Institutes of Health.) 
DNA-binding Proteins for Transcriptional 
Regulation 
The yeast transcriptional activator GCN4 has a 
segment at its carboxyl terminus that forms a so- 
called basic region/leucine zipper (bZip) element, 
which governs dimerization and DNA binding. It is 
unfolded at low concentration, but the presence in 
free solution of specific DNA induces an almost com- 
pletely a-helical structure. Work in other laborato- 
ries has shown that the carboxyl-terminal part of the 
element (the leucine zipper) forms a dimeric a- 
helical coiled coil when this transition occurs. 
A crystal structure of the GCN4 bZip fragment, 
specifically bound to DNA, shows that essentially 
the entire 58-residue fragment forms a continuous a 
helix. The part corresponding to the basic region 
diverges gently from the carboxyl-terminal coiled 
coil, and each basic region in the dimer traverses a 
DNA major groove on either side of the center of the 
binding site. The entire complex has the appearance 
of a molecular tweezers grasping the DNA. 
A set of five residues in the basic region, largely 
conserved across the family of bZip proteins, speci- 
fies the sequence of the binding site. Indeed, many 
of the related proteins (such as the fos/jun hetero- 
dimer) bind to similar DNA sequences, and work on 
crystals of a fragment from the fos/jun complex is in 
progress. 
The amino-terminal 225 residues of GCN4, 
which include an acidic activator region, do not ap- 
pear to affect the conformation of the bZip portion. 
The structural properties of the amino-terminal pan 
remain to be explored. (The GCN4 project has been 
supported in part by funds from the Lucille P. Mar- 
key Charitable Trust.) 
Another yeast transcriptional activator, GAL4, is a 
very large protein with a small amino-terminal re- 
gion that mediates its binding to upstream activating 
sites (UASq). a fragment containing residues 1-65 
has been crystallized in complex with DNA contain- 
ing a UASq sequence, and the structure has been ana- 
lyzed at 2.7-A resolution. The protein binds as a 
dimer to the symmetrical 1 7-base pair sequence of 
the UASg site. A Zn^^-containing domain (residues 
8-40), termed the recognition module, specifies a 
472 
