33 rd F O R U M
Differentiation and repertoire selection o f 78 T cells. The model of T-cell differentiation proposed by De Villartay and Cohen, in which the dichotomy between 78 and ~[3 T-cell development is controlled by the activity of the TCR 0~-enhancer is very attractive, since it is completely consistent with all published data including the studies in transgenic mice, that have not been dealt with in this report. Whether or not 78 T cells are subject to repertoire selection in the thymus in the same way as ~[3T cells needs to be explored further. The data accumulated by Hedrick et al. indicate that negative selection can apply to 78 T cells, but perhaps it is relevant only for certain 78 T-cell subsets. The suggestion of Kyes and Hayday to consider the two subsets of murine 78 T cells with homogenous V75/VS1 or VT6/V81 receptors together as a distinct 78 T-cell type is relevant. These two subsets are generated during foetal development and must express certain homing receptors that determine their specific epithelial localization, since the receptor itself is not responsible for this (Bonneville et al., 1990, J. exp. Med., 171, IVIJ].
111 t l l ~
lents of these subsets are lacking. As mentioned by Spits et al., there is also an ordering in the rearrangement events (Krangel et al., 1990, J. exp. Med., 172, 847) with the most J-proximal V-gene segments rearranging first, while the early 78 T cells in man also have more limited junctional diversity. However, the early foetal wave of 78 T cells in man does not home selectively to epithelia of skin or reproductive organs. The suggestion that these two subsets of murine 78 T cells are positively selected within the thymus on the basis of TCR T8 interactions with the thymic stroma is valid (Kyes and Hayday). In the human system also, the equivalent of such interactions would be lacking, since at no timepoint during human foetal development does the thymus harbour only TCR 78 + cells (Campana et al., Z Immunol., 1989, 142, 57), as is the case for the VT5/VSl+ cells at day 15 during
murine foetal development while, in addition, the human foetal 78 T-cell repertoire is not homogenized as is the case for the VT5/VSI and V76/VS1 + T cells in the mouse. It would be of interest to determine whether the element responsible for this positive selection is polymorphic. Perhaps mouse strains remain to be identified in which the foetal 78 T-cell waves cannot be discriminated. There is a suggestion that positive thymic selection on the basis of recognition of class-I(-like) MHC molecules is less important for 78 T cells, since 78 T-cell numbers were not affected in the [32m-negative mice constructed by Zijlstra et al. (Nature, 1990, 344, 742). However, these experiments are not easy to interpret given the potential of T8 T cells to selectively expand in the periphery (see below). Therefore, the question as to whether positive thymic selection applies to at least certain subsets of T8 T cells is still open. "Positive selection" of 78 T cells in the peripheral compartment plays an important role in repertoire formation in man. The term "positive selection" is formally not correct, since this would imply the involvement of genetic (MHC) determinants. The various authors that have dealt with the predominance of V79/V82 receptors in human peripheral blood seem to agree mat ..... it is the result of a selective expansion of 78 T cells due to antigenic challenge. Augustin and Sim have observed two examples of extrathymic expansion of murine y8 T cells that might be determined by genetic factors and/or foreign antigens. The peripheral compartment of the mouse contains relatively high numbers of y8 T cells with Vy2Cy2- or VT1CT4-encoded chains (cell type 2 of Kyes and Hayday). It is perhaps an interesting thought to look upon this peripheral repertoire bias in the mouse as a .pcssible result of extrathymic expansion equivalent to that found in man. Support for this idea comes from the observation that these receptor types also dominate in the nude mouse. Both in human and murine intestinal epithelium the y8 T-cell repertoire, albeit diverse, differs from that in the periphery, with V81 receptors being dominant in man and V77 receptors in mouse (dis-
"~ T C E L L S cussed by Rust et aL). Since there is evidence in both species that these ceils have been activated previously, peripheral expansion might play a role in establishing the repertoire bias a!so at this site. It remains to be de~er.'nined whether this would be driven t~y interaction with a self protein or a foreign antigen. The observation that V-f7 cells also predominate in germ-free mice (Bandeira et al., 1990, ./. exp. Med., 172, 239) may indicate that in this case antigens are not the driving force.
Antigen recognition by ~'~ T cell i. Although MHC restriction l'as not been demonstrable in many case~ where antigen-specific -f8 T cells were found, all authors start from the ass amption that ~'8 T cells will recognize fcreign antigens presented by certain mo],'cules. The contributions of Triebel and Bluestone and Matis make an important point in saying that too much attention might have been paid to the nonconventional MHC class I molecules, CD1 in man and TL/Q~ in mouse, as presenting elements for -f~ T cells, because of initial isolated successes with certain ~'~ T-cell clones. It would appear tnc many examples of ailonow from "-MHC-specific "r~ T cells generated in man and mouse, that conventional class I and class II molecules are certainly relevant for ~'~ T cells. In additic.u, the alternative restriction elements have also been shown to be relevant for a[3 T cells. Obviously, what needs to be done is to carefully determine T-cell precursor frequencies within a[3 and ~'~ T-cell populations for recognition of each e f these MHC(-like) molecules. In such assay, care should be taken to use a functional read-out system, such as induction of Ca 2÷ flux, that would not introduce a bias based on differences in functional programs, but would only reflect the recognition event. In addition, there seems room for other, not yet molecularly defined restriction elements such as the TCT.1 structure (Triebel). Two types of antigen recognition by ~'~ T cells have emerged: "superanti-
gen"(-like) recognition and conventional antigen recognition. Human V~,9+ cells can interact with a well defined superantigen, Staphylococcal enterotoxin A (Rust et al., 1990, Nature (Lond.), 346, 572). In all other cases where superantigen recognition is proposed to play a role, the antigen has not been defined. A role for a superantigen is invoked, based on the observed expansion of 78 T cells sharing V-r and/or V~ expression, independent ef junctional diversity. The V~,9/V82 repertoire bias in human peripheral blood would come about by stimulation by such superantigens. Since mycobacteria induce significant responses of ~ T cells in mice and man that seem not to b,~ MHCrestricted, they are good candidates for a source of superantigens. Band et al. indeed indicate that human V~,9/V82 * cells respond to Mycobacterium tuberculosis independent of junctional diversity. Gatrill et al. discuss the fact that such stimulation of ~ T cells is not specific to mycobacteria, but can also be induced by other bacteria. These authors keep the two options open and suggest that conventional antigen recognition as well as superantigen-like recognition may be part of the observed response. Although a lot of speculation has been brought forward about ~'~ T ceils and response to stress proteins, it is ~ot at all clear how important bacterial heat shock proteins are in inducing these dominant ~'8 T-cell responses. Gatrill et al. have clear evidence that HSP do not provide the relevant antigen. In all other cases where dominant ~,~ T-cell responses to mycobacteria were seen in vivo (in man: lepra, rheumatoid arthritis), there is no followup yet to document the nature of the antigen at the peptide level. Perhaps some of the many ~,~ T cells activated by bacterial preparations display (conventional) reactivity to HSP-derived peptides, such as documented by Born and O'nrien. Since the y~ T-cell response to bacterial heat-shock proteins has not yet been well defined, except in this one case, it would seem that further speculation about preferential involvement of ~,8 T cells in response to autologous stress proteins (autoimmune disease) is not justified at the moment.