Structural Biology of CD4 and CDS Involvement 
in the Cellular Immune Response 
Wayne A. Hendrickson, Ph.D. — Investigator 
Dr. Hendrickson is also Professor of Biochemistry and Molecular Biophysics at Columbia University 
College of Physicians and Surgeons. He did his doctoral studies in biophysics at the Johns Hopkins 
University and remained for a year of postdoctoral research with Warner Love before going to the Naval 
Research Laboratory for continued postdoctoral study with Jerome Karle. He stayed on at NRL until 
he joined the faculty of Columbia University. 
THYMUS-derived lymphocytes, or T cells, dif- 
ferentiate upon maturation into two major 
cell types. These are principally distinguished by 
the exclusive occurrence of either the CD4 or 
CDS glycoproteins on their surfaces. The CD4- 
bearing cells are known as helper T cells, and the 
CD8-bearing cells as cytotoxic or killer T cells. T 
cells of both types are stimulated into action 
through the interplay of surface receptor mole- 
cules with peptide antigens, which must be pre- 
sented on target cell surfaces as complexes with 
molecules of the major histocompatibility com- 
plex (MHC). The helper T cells (CD4+ CDS") 
can only interact with class II MHC molecules, 
which occur on certain immune system cells 
such as macrophages and B cells, whereas the 
killer T cells (CD4~ CDS+) interact with the class 
I MHC molecules, found on all cells. 
The involvement of CD4 or CDS is essential for 
the efficient stimulation of the respective re- 
sponses. Stimulation of helper T cells by peptides 
from an invading pathogen leads to the produc- 
tion of cytokines, which in turn stimulate the pro- 
liferation of selected lymphocytes for antibody 
production or for cellular defense. Subsequent 
stimulation of appropriate killer T cells by pep- 
tide antigens from infected cells then leads to the 
production of lytic factors that can eliminate the 
infected cell. Thus the molecules CD4 or CDS 
can properly be described as co-receptors in the 
cellular immune response. 
We have undertaken a series of structural stud- 
ies on CD4 and CDS. This work is collaborative 
with Richard Axel (HHMI, Columbia University) 
and with Ray Sweet and his co-workers at Smith- 
Kline Beecham. Results from these studies, when 
taken in conjunction with mutational studies by 
others, are beginning to provide a picture of the 
molecular interactions implied by the specific 
stimulatory processes of the immune response. 
Moreover, the structural results also relate to the 
involvement of CD4 as a receptor for infection by 
the human immunodeficiency virus (HIV). 
Crystals of Soluble CD4 
Both CD4 and CDS are single-pass transmem- 
brane proteins. These can be made tractable for 
crystallographic study by the expression of solu- 
ble extracellular fragments of the protein. A solu- 
ble human CD4 fragment containing the entire 
extracellular portion of the protein was ex- 
pressed in mammalian cells, and Peter Kwong has 
crystallized this protein into several different 
crystal lattices. Unfortunately, none of these dif- 
fracts very well. From a characterization of these 
crystals, it is possible, nevertheless, to deduce 
some interesting features of the CD4 molecule. 
It appears that CD4 oligomerizes at the high 
concentrations needed for crystallization, and the 
lattice dimensions are consistent with a tetra- 
meric model of 1 25 A in length along a diad axis 
of the molecule. This propensity to associate may 
be relevant to signal transduction during T cell 
stimulation. Thus we are pursuing a crystallo- 
graphic analysis of one of these crystals, despite 
the prospect of resolution limited to approxi- 
mately 4 A. 
Structure of a CD4 Fragment 
The consistently poor diffraction from several 
polymorphs of a chemically homogeneous pro- 
tein preparation suggested to us that CD4 may be 
somewhat flexible. This possibility would be 
compatible with the four-domain structure de- 
duced from sequence comparisons and intron 
positions, and limited proteolysis experiments 
produced stable fragments corresponding to the 
first two domains (D1D2) and the second two 
domains (D3D4). A recombinant construct ex- 
pressing the D 1 D2 domain also produced a stable 
molecule, and Seong-Eon Ryu obtained diffrac- 
tion-quality crystals of this fragment. 
The resulting structure analysis revealed a Dl 
domain folded as in the variable domains of im- 
munoglobulins and a D2 domain in a variation of 
the immunoglobulin constant-domain topology. 
These two domains are intimately connected, 
with the last strand of Dl running directly into 
the first strand of D2. Despite the similarity of Dl 
to immunoglobulin variable domains, there are 
appreciable differences in the loop structures. 
181 
