B7-1 and B7-2 are generally thought to have comparable structures and affinities for their receptors, CD28 and CTLA-4, each of which is assumed to be bivalent. We show instead (1) that B7-2 binds the two receptors more weakly than B7-1, (2) that, relative to its CTLA-4 binding affinity, B7-2 binds CD28 2- to 3-fold more effectively than B7-1, (3) that, unlike B7-1, B7-2 does not self-associate, and (4) that, in contrast to CTLA-4 homodimers, which are bivalent, CD28 homodimers are monovalent. Our results indicate that B7-1 markedly favors CTLA-4 over CD28 engagement, whereas B7-2 exhibits much less bias. We propose that the distinct structures and binding properties of B7-1 and B7-2 account for their overlapping but distinct effects on T cell responses.

How the T cell receptor engages antigen is known, but not how that 'triggers' intracellular signaling. The first direct support for a mechanism based on the spatial reorganization of signaling proteins, proposed 10 years ago and referred to as the 'kinetic-segregation' model, is now beginning to emerge, along with indications that it may also apply to the triggering of nonclonotypic receptors. We describe here the development of the model, review new data and suggest how the model fits a broader conceptual framework for receptor triggering. We also consider the capacity of the model, versus that of other proposals, to account for the established features of TCR triggering.