NEWS AND VIEWS IMMUNOLOGY----------------------------------- An affinity for learning David H. Margulies CHEMISTS and physicists have been known to remark that biology - even molecular biology, which underpins most of the sub- ject as a whole - remains a 'soft' science. The reason is that so much work measures a distant result of many complex processes. More proximal measurements can of course be technically very difficult to make, but at least in one area of immunology progress is being made. On page 616 of this Agonist ligand Antagonist ligand Negative selecti on Positive selection APC r T cell -l --- 'High ' affinity Most stable complex k 0 rt-0 .0 2 s- 1 r M Lower affinity Less stable complex k0 ff-0 ,039- 0.146 s- 1 Mechanistic models to explain the devel- opmental fate of thymic T cells are based on the phenomena of positive and negative selection, events that are dependent on the interaction of developing thymocytes with thymic APC that either rescue them from programmed cell death or promote it. These models fall into two major cate- gories: those governed by qualitative changes in the signal delivered by the TCR Null ligand No selection r I M on binding one or another MHC/peptide ligand; and those determined by quanti- tative differences in the strength of the interaction between the TCR and differ- ent MHC/peptide ligands. Over the past few years, strategies for producing the appropriate reagents and for measuring the appropriate Lowest affinity parameters have been de- No complexes veloped. Building on this detected work, Alam et al now report Correlation of activity of MHC/peptide complexes as ligands for peripheral T-cell activation with the outcome of thymic edu- cation. Examples of agonist, antagonist or null MHC/peptide ligand displayed on the surface of the antigen-presenting cell (APC) are illustrated. The MHC class I/peptide complex, con- sisting of a heavy chain (red) , a light chain !32-microglobulin (pal e blue) and a peptide (dark blue ), interacts with the ai3 T- eel! receptor (TCR , green and yellow) to initiate positive or neg- ative selection in the thymus, or agonist or antagonist effects on peripheral mature T cells. TCR- MHC/ peptide interactions tend to be of moderate affinity. But the highest ones seem to be those in which the agonist peptide parti ci pates, which in the case studied by Alam et al. '- are charact eri z ed by a disso- ciation rat e constant ( kotf) of - 0.02 s- 1 . Antagonist peptides seem to produce complexes of weaker affin i ty for the TCR (indicated here by greater distance between the APC and T-cell membrane) and null peptides produc e MHC/peptide complexes that do not detectably interact with t he TCR. the strongest evidence to date directly correlating the affinity of the interaction between a TCR and MHC/ peptide ligands with the out- come of thymic selection. For the time being, quantita- tive models are attractive, and they promise to tell us a great deal. Alam et al. have used sol- uble, purified preparations of the main components of the TCR-MHC/peptide interaction to explore the kinetics and affinity of bind- ing of a well-characterized TCR to MHC/peptide com- plexes in which the bound issue 1, Alam et al. provide quantitative data on a central issue for immunologists: how the fate of developing T lymphocytes is decided. The background to the problem is as follows. T lymphocytes undergo a complex 'educational' programme of cellular and molecular development as they pass through the thymus, before being dis- patched to peripheral tissues where they sit poised to initiate and control the immune response to infection or to dysregulated protein expression. Both the developmen- tal programme of the T cell and the speci- ficity of its action in recognizing aberrant molecular structures rest on the interaction of clonally distinct T-cell receptors (TCR) with cell-surface molecular complexes composed of major histocompatibility complex (MHC) molecules bound to anti- genic or to self peptides on 'antigen- presenting cells' (APC). 558 peptide was varied. The TCR was cloned from a mature T cell that is activated by an MHC class I molecule (H-2Kh) bound to an octamer peptide derived from chicken ovalbumin. Variants of the ovalbumin peptide have been previously characterized not only for their effects as agonist and antagonist in in vitro T-cell activation assays, but also for their effects on positive selection of thymic cells from animals transgenic for the same receptor 2 " 1 • Agonist peptides bind H-2Kb and deliver a stimulatory signal to the T cell resulting in sensitization for cytolysis; antagonists deliver an inhibitory signal that counters agonist-mediated activation. Pos- itive selection is assessed by exposure of fetal thymic cells in organ culture to syn- thetic peptides; and by measurement, in an appropriate genetic background, of the development of T cells bearing the trans- genic receptor as well as the CDS marker of mature MHC class I restricted cells. Using an evanescent wave biosensor to monitor the binding of the various MHC/peptide complexes to the immobi- lized specific TCR, Alam et al. show that the interaction of the MHC/peptide com- plex (consisting of the wild-type ovalbu- min-derived peptide bound to a purified soluble version of H-2Kb) with the TCR is weak. It is characterized by a rapid dissoci- ation rate (k 0 ff - 0.02 s- 1 , corresponding to a tin of -24.5 s), and relatively slow associ- ation rate (k 00 - 3,100 M- 1 s- 1 ), giving a calculated equilibrium constant, Kd, of 6.5 x 10 - 6 M. But because such measure- ments are subject to a variety of artefacts and potential misinterpretations4, it is im- portant that internal tests of consistency, as well as measurement of parameters in the opposite orientation (MHC/peptide on the solid phase, TCR in solution) lead to essentially the same results. The authors then evaluated a set of four synthetic variant peptides that have been previously characterized with respect to their agonist or antagonist activity with mature peripheral T cells, as well as in the fetal-thymic-organ culture assay. Three peptides, known to form positively select- ing ligands for T-cell precursors bearing the same TCR, form MHC/peptide complexes that bind the TCR less well than those formed with the full agonist. Variant pep- tides that on binding the MHC fail either to activate mature T cells, or to exert positive selection in the transgenic fetal- thymic-organ culture assay, have k0 ff values that exceed the discriminatory capacity of the method. Surprisingly, the apparent window of affinity that distinguishes the ag- onist peptide that causes negative selection in the thymic assay from the antagonist peptides that induce both positive selection and act as antagonists for the T-cell clone in vitro is quite narrow - the antagonist peptides generate complexes that have roughly two to ten times less affinity for the TCR than does the agonist. Alam and colleagues' results comple- ment those previously published evaluating a set of agonists in a different TCR- MHC/peptide system 5 • They lead to the neat conclusion, summarized in the figure, that correlates affinity with degree of agonism as well as with the outcome of thymic selection. These data are consistent with kinetic models that view the result of TCR-mediated signals as resulting from differential occupancy 6 • 7 ; and they tend to count against mechanisms that depend on structural changes, either in conformation or topology8, of the TCR that transmit differences in the ligand. But we should not settle too com- placently for this direct affinity/outcome model. In isolation, the new measurements have not integrated the contribution of the CDS co-receptor, a T-cell surface molecule that not only provides a signature for MHC class I restricted T cells, but also provides NATURE · VOL 381 · 13 JUNE 1996