T CELL SPECIFICITY 
John W. Kappler, Ph.D., Investigator 
Most T cells use clonally variable receptors made 
up of two chains, a and P, to recognize antigen. 
Functional genes for each of these chains are con- 
structed by rearrangements of one each of a num- 
ber of segments, by mechanisms that are similar to 
those used for immunoglobulin genes. In the 
mouse and human, chain genes can use 1 of —50 
Vas and 1 of —50 Jas. In the mouse, P-chain genes 
can be constructed from 1 of —20 vps, 1 of 2 Dps, 
and 1 of 12 JPs. The figures are similar in humans, 
with the exception that there are —50 vp genes per 
haploid genome in this species. Additional variabil- 
ity between receptors on different T cells occurs be- 
cause of the introduction or deletion of a number 
of bases into a and P genes at the points of V-J, D-J, 
or V-D joining. 
The work of a number of laboratories has sug- 
gested that binding of a particular T cell receptor to 
its ligand [a peptide fragment of antigen bound to a 
cell surface major histocompatibility complex 
(MHC) protein] requires a contribution to binding 
of all the variable elements of the receptor, includ- 
ing Va, Vp, and Ja. Because any given combination 
of variable elements will be rare, T cells responding 
to most antigen-MHC complexes in a naive animal 
are rare, and primary responses to most infections 
are consequently small. 
Some years ago, work from this laboratory dem- 
onstrated a surprising exception to this rule: in 
some cases vp alone could determine the specificity 
of T cells for certain antigens, almost regardless of 
the other variable elements of the receptors ex- 
pressed on these cells. The antigens in question 
were called superantigens, because they stimulate 
almost all T cells bearing a particular vp(s), activate 
a relatively large percentage of all T cells, and give 
rise to enormous responses, even in a naive animal. 
For example, some mice express the superantigen 
Mls-2^. This antigen stimulates nearly all T cells 
bearing Vp3, providing these cells are derived from 
a mouse that does not itself express Mls-2^. Conse- 
quently, T cells from Mls-2'^-negative animals, 
which are usually —5% VPS"*^, respond strongly to 
Mls-2^-bearing cells from other animals. T cells 
from mice that themselves express Mls-2^ do not re- 
spond in this fashion, because the potentially self- 
reactive Vp3-bearing cells are eliminated during de- 
velopment in the thymus. 
So far about eight different superantigens have 
been found encoded in the genomes of various lab- 
oratory mice. The most biologically significant con- 
sequence of these superantigens to the mice that 
express them is the fact that they cause clonal dele- 
tion of the T cells with which they can interact. In 
some cases this amounts to deletion of a sizable 
percentage of the total T cell repertoire. Mls-1^, for 
example, has been shown by this laboratory to de- 
lete VPS.I^ T cells and by others to delete vp6- and 
vp9-bearing cells, —22% of all T cells in the mouse. 
Why do mice express these deleting super- 
antigens? In an attempt to deal with this question. 
Dr. Kappler and his colleagues analyzed wild mice 
to find out if these animals, which presumably have 
to deal with the full gamut of mouse pathogens, ex- 
press T cell-deleting superantigens. In collabora- 
tion with Drs. Edward Wakeland and Wayne Potts 
(University of Florida) 42 wild mice trapped near 
Gainesville were examined, vp-reactive, T cell-de- 
leting superantigens were found in nearly all the 
animals. Some of the animals were also homozy- 
gous for a large deletion at their VP loci; only 10 of 
all mouse Vp genes were present in these animals. 
Thus T cell deletion via Vp is not a rare event in the 
wild. On the contrary, such deletion is almost uni- 
versal; the strategy adopted in wild mice seems de- 
signed to cause a limited expression of Vps in any 
given animal and to permit different animals to ex- 
press different spectra of vps. 
At an HHMI sponsored workshop in 1987 on the 
Mis locus, Dr. Steven Buxser presented data on the 
curious properties of a collection of toxins pro- 
duced by Staphylococcus aureus. At the time these 
toxins were known to bind MHC class II proteins, 
and as such a complex to be powerful T cell stimu- 
lants in mice and humans. They were also known 
to cause several diseases in humans, including food 
poisoning, toxic shock, and scalded skin syndrome. 
This laboratory tested these toxins for their ability 
to stimulate T cells bearing different vps in mice 
and humans and found that in both species each 
toxin acted in a vp-specific way. Staphylococcal en- 
terotoxin B, for example, stimulates mouse T cells 
bearing Vp3, -7, -8.1, -8.2, and -8.3. The toxin re- 
sponsible for toxic shock in humans stimulates 
mouse T cells bearing Vpl5, -10, and -3 and human 
T cells bearing Vp2. Thus in every respect the 
staphylococcal toxins appear to be bacterial super- 
antigens, analogous to the Mis-like superantigens 
made by mice themselves. 
In collaboration with Dr. Brian Kotzin at the Na- 
Continued 
405 
