"r8 T C E L L S 671

destruction of stressed epithelial cells. Although expression of V8I is dominant in this ~,~ TCR + population, various -fB TCR phenotypes can be detected in the isolated and cultured ~,~ T cells. It is thus possible and perhaps even plausible that these intraepithelial ~,~ TCR+ cells are reactive with a number of (self and foreign) antigens. Until now, however, the antigen specificity of these ,8 T cells remains unknown.

Although the presence of ~,~ TCR +

cells in the epithelium of coeliac disease patients is reminiscent of the presence of "t'~ T cells in mouse epithelia, there is no evidence for a very restricted and specific usage of V~ and V8 region gene segments by these ~.~ T cells in man. This is clearly different f rom the situa- tion in the mouse, and indicates that the dominant presence of -f~ T cells in mouse epithelia may not be relevant for a general understanding of "t'~ T-cell function in man.

References.

ASARNOW, D.M., KUZIEL, W.A., BONYHADI, M., TIGELAAR, R.E., TUCKER, P.W. & ALLISON, J.P. (1988), Limited diversity of y~ antigen receptqr genes of Thy-1 ÷ dendritic epider- mal cells. Cell, 55, 837-847.

BONNEVILLE, M., JANEWAY, C.A. Jr, ITO, K., HASER, W., ISHIDA, I., NAKAN15HI, N. & TONEGA- WA, S. (1988), Intestinal intraepithelial lymphocytes are a distinct set of -~8 T cells. Nature (Lond.), 336, 479-481.

BRANDTZAEG, P., BOSNES, V., HALSTENSEN, T.S., SCOTT, H., lOLLeD, L.M. & VALNES, K.N. (1989), T lymphocytes in human gut epithelium preferentially express the 0J[3 anti- gen receptor and are often CD45/UCHLl-positive. Scand. J. ImmunoL, 30, 123-128.

GOODMAN, T. & LEFRANCOIS, L. (1989), Intraepithelial lymphocytes. Anatomical site, not T-cell receptor form, dictates phenotype and function. J. exp. Med., 170, 1569-1581.

HALSTENSEN, T.S., SCOTT, H. ~ BRANDTZAEG, P. (1989), Intraepithelial T cells of the TCR~8+CD8 - and VSI/J~I + phenotypes are increased in coeliac disease. Scand. J. Immunol., 30, 665-672.

JARRY, A., CERF-BEN'SUSSAN, N., BROUSSE, N., SELZ, F. & GuY-GRAND, D. (1990), Subsets of CD3 ÷ (T-cell receptor 0t[3 or ~,8) and CD3- lymphocytes isolated from normal hu- man gute_pithelium_, dis pl_a_y_pheno_t_yp_!c_a! featuresdefferent from their counterparts 111 pl~Ilpll¢lill OlOOLI. r.urop, j. l m m u e l O l . , &u, iU~l-i lO..5.

RusT, C.J.J., VERRECX, F., VIETOR, H. & KONINC, F. (1990), Specific recognition of staphylococcal enterotoxin A by human T cells bearing receptors with the V'l'9 region. Nature (Lond.), 346, 572-574.

SPENCER, J., ISAACSON, P.G., DIss, T.C. & MACDONALD, T.T. (1989), Expression of disulfide- linked and non-disulfide-linked forms of the T-cell receptor ~,/8 heterodimer iv hu- man intestinal intraepithelial lymphocytes, Europ. J. lmmuno!., 19, 1335-1338.

DISCUSSION

A. Hayday and S. Kyes:

As Bluestone and Matis discuss, what are the arguments that there is really much of a difference between 0t[3 and ~,~ T cells ? They are right in point- ing to a population of ~ T cells that can be found in lymphoid tissue, and that possesses reasonable diversity. We con-

sidered in our paper that these cells can infiltrate epithelia on a case-by-case ba- sis, and they possibly do so in response to a variety of antigenic stimuli. Born and O'Brien consider a similar point. However , one should also bear in mind Mackay and Hein 's data that although "r~ cells are most common in ruminants, there are very few such cells in lymph

672 33 rd F O R U M I N I M M U N O L O G Y

nodes. One wonders, did rearranging and [3 and "t" and ~ gene families evolve in cells that already had different ~ and [~ homing/c i rcula t ion propert ies? Perhaps cells in which ~'~ genes evolved were already more epithelial-tropic. Despite the claims of Rust e! al., there are increasing data that even in human, "r~ cells are epithelial-tropic (Tim Mac- Donald, Hermann Wagner, pets. comms).

To what degree are presenting ele- ments shared between 0t[3 and ~'~ T cells ? The many arguments for class-I-related molecules derive some support from the CDl-restriction described by Brenner and his colleagues, and from TL and Qa reactivity (Hedrick and Dent; Dembi6 and Vidovi~). However, both Triebel's and Bluestone and Matis's papers indi- cate that such reactivity may be rare. The latter paper leaves open the possi- bility of conventional class I/II presen- tation - - perhaps we have been confused where genetic restriction has been vainly sought in superantigen responses. The data of Spits et al., that "r~ responses are vulnerable to the same mutations in a presenting element as are ~[3 cells, strengthen this argument. Trie- bel, by contrast, offers an exciting although related alternative with the TCT molecule; but again, how general is this? Class II restriction may have been unjustifiably neglected (Holo- shitz.). In our original allo-modulation in mice, class II was a powerful stimu- lus, and class-II-reactive cells have been cloned out by Bluestone and Matis. As the latter workers point out, until we are confident of cell growth conditions, we are desperately short of a reliable ex vivo assay for "t'~ function.

The discovery by Rust et al., that hu- man VT9 is superantigen-reactive fur- thers the parallels between ~.8 and 0c[3 cells. Is such reactivity responsible for general bacterial and eukaryotic (e.g. Daudi cell) reactivity ? To accommodate this, we may need to revise our defini- tion of superantigen reactivity. We pro- pose that it should include all responses that are not dependent on a specific an- tigenic peptide in the presenting element groove. Hence, junctional diversity is likely to be reasonably unlimited in such

responses (as Band et al., discussL but there may still be a requirement for a specific "t' and ~ chain pairing (Bolhuis et al. ).

Superantigen responses may reflect internal changes in celle that modulate the appearance/conformation of cell surface molecules. We first presented data at Segovia (June 1990) that the lev- el of expression of [32-microglobulin and associated molecules may be upregulat- ed by treatment of cells with PPD or mycobacterial extract. These changes may simply be a marker for altered ki- netics of the antigen presentation machinery. Although some responses to PPD seem to be dependent on peptide- TCR junction interactions (Born and O'Brien), other "r~ cells may respond to induced cell surface changes, largely ir- respective of TCR junctional sequence. The case for self reactivity as a fun- damental property of at least one sub- set of y~ T cells is considered again in our own paper. It is most appealing in the case of murine epithelial cell that have almost homogeneous receptors. These cells bear V~l chains. Is this the only true murine V~ gene, and V82 the only true human V8 gene ? Casorati and Migone, and Triebel discuss the in- creasing cases of 0c-~ crossover of the other V segments. Murine V~ 1 and hu- man V~2-bearing cells are generated ear- ly in thymic ontogeny (discussed by ourselves and by Spits et al.). While the products of these rearrangements are readily traced in epithelia of adult mice, their fate in postnatal humans is less clear. We should not like to exclude completely the possibility that the foe- tal receptors may be used in situ in foe- tal tissue. Mackay and Hein draw attention to the clustering of thymic ~.~ cells around the keratinized Hassall's corpuscules, and Mak and his col- leagues have argued for some time for a role of "r~ cells in the regulation of lymphopoiesis.

Finally, the 0c-8 cross-over in V-gene usage reminds us that these families may presently be in dynamic evolutionary flux. Much as we ourselves have consi- dered that autoreactive ~,~ cells may be the more primitive TCR1 (echoed by Holoshitz), the opposite m a more re-

"(~ T C E L L S 673

cent selected limitation in the de facto diversity of a rearranging gene family - - could equally be true.

Z. Dembi~ and D. Vidovi~:

Most participants in the "Forum" present many fi.cts and hypotheses which fit into the mosaic aimed at an- swering the question of whether the "r~ T-cell population is (and to what extent) different from u[3 T cells. Analysing them (see contributions to this issue) we obtained the following hypothetical "picture": there seems to be an over- lapping equivalence between 0t~ and ~,~ cells in the general structure of their TCR (they belong to the Ig supergene family), recognition specificity (both have the ability to recognize MHC molecules and some other antigens), repertoire and its selection (both have a potential to generate diversity in the repertoire which seems to be positively and negatively selected), development (both can develop in the thymus), hom- ing (both can circulate and settle in peripheral lymphatic organs and epithe- lia), function and host defence (both seem to show cytolytic or regulatory ac- tivity, and perhaps both play a role in protecting the organism against bacter- ia and viruses). But, according to the available data, each population shows a characteristic bias in every aspect men- tioned: (1) structure of TCR: they use different set of genes; (2) recognition specificity: ~8 show preference towards certain antigens like hsp and relatively invariant antigens structurally similar to class I MHC molecules; (3) repertoire: due to usage of the limited set of V ~/8 region gene segments, the repertoire seems to be skewed towards the recog- nit ion of yet unknown ligands; (4) repertoire selection: the ~,8 repertoire seems to be opportunistically selected

(positively and negatively) both in the (foetal) thymus (Lafaille et aL, 1990; Hedrick and Dent, this issue) and in the periphery (Augustin and Sim; Bolhu;s et al.; Borst and van Dongen, this is- sue); (5) development: in -/'8 cells, not all V~. are rearranged at the same time during ontogeny, and a large proportion seem not to be thymus-dependent (Havran and All ison, 1990) ; (6) homing: a larger proportion of the total "t'8 cell population home to epitheli- al tissues (skin, lung, intestinal mucosa and reproductive organs); (7) function: "t'~ so far seem to have a bias towards cytolytic activity; (8) host defence: there is a possibility that "r~ can be em- ployed in certain specific host defences against intracellular microbes (Gatrill el a!., this issue). Interesting is the proposal that a discrimination, analo- gous to "self-nonself" 0y 0t[3 cells, might, in the ~,~ T-cell pool, rely oa he;- real versus the stressed state of the tar- get cell (Born and O'Brien, this issue). What has not been reported so far is the relationship with other tissues and or- gans in the body (respoosiveness to cytokines, stress and other hormones, releasing factors, rnodulat ars secreted by neural tissue and others). Perhaps there lies the missing link between an or- ganism and its immune system.

What emerges from this multitude of information is net easy to reconcile with the notion that "rB cells are a unique and functionally uniform T-cell population. Rather, it is more compatible with the notion that this population consists of more distinct subsets each performing a different function (Kyes and Hayday; Bluestone and Matis; Dembi6 and Vi- dovi6, this issue). Each of them might have been expanded to a different ex- tent during phylogeny among species, implying heterogeneity in ~B-cell fea- tures and function (Mackay and Hein, this issue).

References.

HAVRAN, W.L. & ALLISON, J.P. (1990), Origin of Thy-1 + dendritic epidermal cells of adult mice from fetal thymic precursors. Nature (Lond.), 344, 68-70.

LAFAILLE, J.J., HAAS, W., COUTINHO, A. • TONEGAWA, S. (1990), Positive selection of gamma delta T cells. ImmunoL Today, II , 75-78.