Antigen-MHC complexes are recognized by two distinct structures on the T cell: the antigen-specific T cell receptor (TCR) and the coreceptor molecules, CD4 or CD8. The antigen-specific receptors are, of necessity, variable and recognition of class I or class II MHC can be undertaken by the same TCR VP-encoded products, though there may be more restriction on the usage of Vu genes. On the other hand, the coreceptors are invariant and there is tight linkage between recognition of class I MHC and expression of CD8 and the recognition of class II MHC and expression of CD4 by the T cell (Parnes, 1989). Furthermore, during differentiation in the thymus the decision to express either CD4 or CD8, rather than the MHC specificity of the T cell receptor per se, has been correlated with commitment to a helper (Corbella et al., 1994) or cytotoxic (Davis et al., 1993) phenotype: effector functions which, in order to be delivered to the appropriate cell type, depend upon recognition of antigen on the correct class of MHC molecule.

How are the coreceptors, CD4 and CD8, involved in T cell differentiation and mature cell function? The coreceptors share two common features: the first is that the extracellular domains bind MHC and contribute to avidity, and the second is that the cytoplasmic domains bind the tyrosine kinase,~ 56~~, and contribute to signaling (Micelli and Parnes, 1993). It might be expected, therefore, that CD8 and CD4 would closely resemble each other structurally, but they do not. CD8 is a disulphide-linked heterodimer of two unrelated polypeptide chains a and 3, while CD4 is a monomer (Parnes, 1989). Of the CD8 polypeptides, a most resembles CD4 in that it has been shown to bind both MHC and Ick when expressed as homodimers, raising questions about whether the CD89 chain is redundant. Yet class l-restricted TCRaP+ T cells obsessively express CD8a3 heterodimers on their surface despite the fact that aa homodimers are present within the cell (Zamoyska and Parnes, 1988) although other types of cells, such as NK cells or extrathymic intraepithelial T cells, are quite content to express the homodimer form (Moebius et al., 1991). It is reassuring, therefore, to find that T cells know something we do not; as several recent papers show that disruption of the CD83 gene (Crooks and Littman, 1994 [this issue of/mmunify]; Fung-Leung et al., 1994; Nakayama et al., 1994) or even removal of the CD8f3 cytoplasmic domain (Itano et al., 1994 [this issue of hrmunify]), has significant consequences, particularly for the maturation of MHC class l-restricted T cells. In light of this, it is perhaps time to reevaluate what we think the coreceptor molecules are doing and how we think they achieve it.