During immune-cell development, potentially self-reactive T cells are eliminated. It emerges that recruitment of a co-receptor bound to the T-cell receptor by the enzyme Lck is the rate-limiting step in this negative selection.
Immunological tolerance is a developmental process that enables the immune system to be poised to respond to potential pathogens without inappropriately responding to the body's own molecules. As T cells of the immune system develop in the thymus, those whose T-cell receptor (TCR) binds to a ligand with high strength — an indication that the ligand belongs to a self-molecule — are induced to die. Writing in Cell, Stepanek et al.1 identify a molecular mechanism to explain the relationship between TCR recognition strength and this negative selection. The mechanism they propose is based on the 'kinetic proofreading' model originally put forward nearly 20 years ago for T-cell activation2, which suggests that the induction of an activating signal requires the TCR to bind to its ligand for long enough that the resulting downstream signalling cannot be stopped by the eventual TCR–ligand dissociation. The novelty in Stepanek and colleagues' work is that they have linked the half-life of this interaction to the recruitment of the signalling molecule Lck to the interaction complex, which turns out to be a rather rare event.
Each mature T cell expresses a slightly different TCR that will bind to a complex formed of a specific short peptide (the antigen) and a major histocompatibility complex (MHC) protein on the surface of antigen-presenting cells. The strength of this binding determines the strength of the signal transduced in the T cell. If the antigen is from a foreign organism, this signal needs to be strong enough to activate the cell to respond appropriately to the potential pathogen. But because the TCR repertoire is generated in immature T cells (thymocytes) in a fairly random way, some TCRs will recognize self-antigens, and these cells need to be deleted by negative selection during T-cell development. The challenge in understanding this process has been to determine how a continuous variable — the strength of MHC–peptide binding to the TCR — can be translated into a digital response, in which too-strong signalling leads to death, whereas signalling below this threshold induces positive selection, leading to thymocyte survival and differentiation into mature naive T cells.
In addition to the TCR, developing thymocytes express cell-surface co-receptor proteins, called CD4 and CD8, which bind to MHC class II (MHCII) and class I (MHCI) proteins, respectively. Thymocytes bearing TCRs that recognize MHCII lose CD8 co-receptor expression and become CD4+ T cells, whereas MHCI-restricted thymocytes develop into CD8+ T cells. Expression of these co-receptors enhances cellular sensitivity to antigen, mainly through their recruitment of Lck, a kinase enzyme required for triggering TCR signalling, into the vicinity of the ligated TCR.
The same research group had previously identified ligands that define the threshold between positive and negative selection for MHCI-restricted TCRs3,4. Now, they have extended this to MHCII-restricted TCRs, and measured the binding affinity of various MHCII–peptide complexes to the TCR at the negative-selection threshold. They find that the affinity thresholds are similar — around 450 micromolar (μM) for MHCI and around 300 μM for MHCII. However, when they used direct imaging to measure the half-life, or 'dwell time', of these threshold ligands binding to live thymocytes, they found larger differences.
These on-cell binding measurements include the effects of binding of the MHC–peptide complex to both the TCR and the co-receptor; CD8 binds to MHCI more strongly than CD4 binds to MHCII. As a result, the dwell time for the interaction of the MHCI threshold ligands with immature pre-selection thymocytes is about 0.9 seconds, whereas that for MHCII threshold ligands was calculated to be about 0.2 s (unfortunately, it was too short to be measured directly). But a 0.9-s dwell time in the MHCI system corresponds to the dwell time for strongly negative-selecting ligands in the MHCII system, making it difficult to see how the same TCR signal strength can be translated into different functional outcomes in the two systems.
To solve this conundrum, the researchers considered the different roles of the co-receptors in recruiting Lck to the TCR signalling complex. Both CD4 and CD8 bind Lck in a similar way, but CD4 binds rather better than CD8 (ref. 5). Stepanek and co-workers measured the amount of Lck that was bound to the two co-receptors in unstimulated thymocytes, and found that only around 7% of CD4 molecules and around 0.6% of CD8 were bound by Lck. When they factored in the proportion of Lck molecules that were active (on the basis of phosphorylation at a specific site), they were left with a paltry 1.8% and 0.16%, respectively. The authors also showed that a CD8 molecule engineered to use the Lck-binding site from CD4 lowered the negative-selection threshold, such that ligands that were normally at the threshold became strong negative selectors and ligands that were just within the positive-selection range were tipped over the threshold into the negative-selecting realm.
The authors then mathematically modelled the effect of differential co-receptor–Lck coupling on T-cell activation. The model that best fits the data proposes that the probability of a Lck-bound co-receptor being recruited to the TCR–MHC–peptide complex is the crucial factor in kinetic proofreading, because only Lck-bound co-receptors stay bound to the signalling complex long enough to transmit a negative-selection signal (Fig. 1). The probability of CD8–Lck being recruited is lower than that of CD4–Lck, so the TCR complex will need to 'examine' more CD8 molecules than CD4 molecules before it finds one that bears active Lck. Thus, a longer TCR–MHC–peptide dwell time is required when CD8 is involved.
Although this model provides a molecular mechanism for how developing thymocytes translate TCR dwell time into distinct signalling and functional outcomes, and how this varies between MHCI- and MHCII-restricted thymocytes, there are some points that it does not resolve. For example, antigen-independent co-receptor interaction with MHC molecules can concentrate MHC at the interface between an antigen-presenting cell and a thymocyte, with the effect of speeding the rate at which the TCR can bind to MHC–peptide6. Because of the higher affinity of CD8 for MHCI than of CD4 for MHCII, this effect will be more marked for MHCI-restricted TCRs and might lower the threshold affinity, but not dwell time, for signalling. Moreover, concentration at this interface also applies to the co-receptor and its associated Lck, so CD8 should be more concentrated than CD4, and the density of CD8-associated Lck at the interface could be higher than estimates obtained from whole-cell analyses.
Another potentially confounding point is that formation of the TCR signalling complex has been identified as a two-step interaction, in which the co-receptor binds to MHC to stabilize the complex only after TCR binding and early signalling events lead to Lck interaction with the TCR complex7,8,9. According to this model, Lck-bound co-receptors preferentially associate with TCRs that have just bound MHC–peptide, which would make the proportion of Lck molecules that are associated with co-receptors less important than in Stepanek and colleagues' model.
Despite such unresolved details, the new model is an attractive variant of the kinetic-proofreading model for T-cell activation, taking into account features of Lck and co-receptor interactions that were not previously accommodated. In particular, co-receptor–Lck interactions change during T-cell differentiation10, in parallel with changes in antigen sensitivity, and the co-receptor-scanning model provides a simple mechanistic explanation for this phenomenon.