Self-Determination in the T Cell Repertoire

Abstract

The number of T cells specific for various antigens can vary dramatically. In this issue of Immunity, Nelson et al. (2015) report that these differences might be, at least in part, set by the number of cross-reactive self peptides encountered by T cells during development.

How many T cells recognize a given peptide antigen presented by a major histocompatibility complex (MHC)? Generally speaking, the more naive T cells that recognize a peptide antigen bound to MHC, the stronger the resulting immune response (Jenkins and Moon, 2012). But this conceptually simple question has proven difficult to answer in practice.

Several years ago, Jenkins and colleagues devised an elegant method to determine the number of T cells that recognize an antigen: by enriching T cells stained with peptide-MHC (pMHC) tetramers of interest, they can directly count naive, antigen-specific T cells (Moon et al., 2007). Using this technique, they have found that the number of naive T cells that recognize a given pMHC is consistent from mouse to mouse, but can vary by orders of magnitude between different pMHCs (Moon et al., 2007). But what causes these differences? In this issue of Immunity, Nelson et al. showed that T cells recognize many (but not all) MHCs displaying peptides that share T cell receptor (TCR) contact epitopes. They also found the number of T cells that recognize a given pMHC in the mature T cell repertoire is negatively correlated with the number of self antigens that share that pMHC’s TCR contact epitope, likely due to editing in the thymus during T cell development (Nelson et al., 2015) (Figure 1).

Figure 1. A Model of How the T Cell Repertoire Is Shaped.

In the thymus (left), self peptides with different TCR contact epitopes (orange and purple) bound to MHC are present in different amounts. T cells that recognize higher-abundance peptides are more likely to be deleted. In the periphery, when mature T cells encounter antigen (right), there are fewer T cells, with lower affinity TCRs, recognizing non-self antigens that share TCR contact epitopes with highly prevalent self antigens. For T cells that recognize fewer self-antigens during selection (bottom right), the higher abundance T cells might also be more likely to initiate autoimmunity upon encountering non-self peptides with a shared TCR recognition epitope.

Nelson et al. took advantage of two properties of pMHC-TCR interactions to find potentially cross-reactive peptides with an array of model antigens bound to the mouse class II MHC I-A^b. First, by knowing the binding register of peptides for class II MHC, they predicted which peptide side chains would be seen by the TCR (position or “P”2, 3, 5, and 8) and which would primarily bind the MHC (P1, 4, 6, and 9) (Jones et al., 2006). Second, because TCR cross reactivity can occur through limited changes to TCR contact residues and a larger degree of changes to MHC contact residues, they identified peptides predicted to cross react with their previously characterized model pMHCs (Birnbaum et al., 2014).

The authors experimentally tested their predictions by immunizing mice with microbial peptides predicted to cross react with their model system antigens and then looked for expansion of T cells reactive with the model system pMHCs. Many, but not all, of the microbial peptides with shared TCR contacts increased the prevalence of T cells recognizing the model system antigens. Why do some peptides with a shared TCR contact epitope cross react while others do not? A panel of single alanine mutants of the model system peptides showed that while all TCR contact epitopes residues were crucial for recognition, MHC contact substitutions also significantly reduced the T cell response for a subset of the peptides.

If unrelated peptides that share TCR contacts are cross reactive, then a peptide expressed during thymic selection should be able to delete T cells that are responsive to an antigen that shares TCR contacts. To test this, Nelson et al. used a mouse transgenic for the 2W peptide to look for T cells able to recognize 2W109, a peptide that shares every TCR contact, but no MHC contacts, with 2W. In these mice, they found the number of 2W109-reactive T cells cut in half. Even though 2W and 2W109 share a TCR contact epitope, there were T cells able to distinguish between the two. The reduction of 2W109-reactive T cells was caused by the loss of T cells also able to bind 2W-tetramers, while the number of T cells that bound to only 2W109 tetramers remained unchanged. 2W109-reactive T cells that emerge from thymic selection in the presence of the overexpressed 2W peptide showed differences when compared to their control group comparators. The remaining T cells were stained by 2W109 tetramer with lower intensity than 2W109 cells from a wild-type (WT) mouse (indicative of a lower affinity interaction), and were more sensitive to mutations of MHC contact residues.

There are a few possibilities explaining how MHC contacts can affect a peptide’s T cell recognition. Most prosaically, while alanine is reported to be an accepted substitution at each MHC contact position for I-A^b, it is possible that some sets of MHC contacts are less amenable to alanine substitution, so the mutations disrupt peptide binding to the MHC. Second, even though the peptide’s position in the class II MHC groove is largely fixed (Jones et al., 2006), mutations to the peptide can sometimes result in minute changes to peptide conformation that affect TCR affinity and T cell activation (Kersh et al., 2001). Correspondingly, MHC contact residues that permit TCR recognition of any given peptide are often a subset of those permissive for MHC binding and might vary for different TCRs (Birnbaum et al., 2014).

What do these results tell us about how the T cell repertoire is shaped in a WT mouse (or person)? First, the authors hypothesized that pMHC with TCR contact epitopes more heavily represented in the “self” proteome should be recognized by a smaller pool of naive T cells and that these interactions should be lower affinity. Indeed, the data shows a weak negative correlation between the number of naive T cells able to recognize a given peptide and the number of potentially homologous self-peptides. A weak correlation is not surprising, given that the number of possible homologous peptides in the mouse proteome is likely an imperfect proxy for what epitopes are in high abundance in the thymus. The correlation between number of T cells recognizing a peptide of interest and the intensity of staining was significantly stronger.

The second hypothesis offered is that autoimmunity can arise from incomplete deletion of clones that are high affinity for self antigens, perhaps due to those antigens being underrepresented or absent in the thymus. To this end, they found a relatively large pool of naive T cells able to recognize I-A^b bound to myelin oligodendrocyte glycoprotein (MOG) peptide and that these T cells were activated when immunized with bacterially derived peptides with a shared TCR contact epitope.

The data presented by Nelson et al. corroborate our recent findings that TCR cross reactivity can enable recognition of seemingly unrelated peptides that share TCR contact epitopes (Birnbaum et al., 2014). In C57BL/6 mice, which encode a single class II MHC, the authors estimated 160,000 (or 20⁴) potential epitopes for the CD4⁺ T cell repertoire to patrol (Nelson et al., 2015). The subtle effects of MHC contact residues on TCR recognition might increase the number of potential epitopes, but this is balanced out by many TCRs that permit broader recognition of pMHC than a single residue for each TCR-facing position along the peptide (Birnbaum et al., 2014). Perhaps, then, the surprise is not that cross reactivity between self and foreign antigens occurs, but that it is not more rampant. With a limited set of pMHC epitopes, it seems plausible that for nearly any T cell of interest, one would be able to find cross-reactive and perhaps activating self peptides. Conversely, strict negative selection based on all possible self peptides seemingly would not leave any T cells to fight infection. Negative selection in the thymus might therefore be more likely to weed out the worst-acting T cell clones: those that are very high affinity to a high abundance self antigen or excessively cross reactive (Huseby et al., 2005). It also speaks to the likely importance of factors such as the prevalence of antigen and an inflammatory local environment to activate T cells in the periphery.

The work here presents an interesting contrast to recent studies by Germain and colleagues, who find that T cells that received a stronger TCR signal during selection are more active in the mature immune response (Mandl et al., 2013), and Allen and colleagues, who show that thymic expression of a single self pMHC known to induce positive selection of moth cytochrome C (MCC)-specific TCRs skews the TCR repertoire to contain more clones that recognize MCC (Lo et al., 2014). Allen and colleagues have also recently shown that T cell clones expressing TCRs with similar affinity for a given non-self pMHC can show significant signaling differences, which are set during positive selection in the thymus (Persaud et al., 2014).

Determining how these mechanisms of self determination fit together will be important to more fully understand the shaping of the T cell repertoire and might provide insight into autoimmunity and developing better vaccines.