One possible explanation for the association between viral infections and autoimmune disease is molecular mimicry, in which foreign antigens activate T cells that crossreact with self-antigens. But how similar do the foreign and self-antigens have to be for the crossreaction to lead to clinical disease? Two papers by Ohashi and colleagues and von Herrath and colleagues show that the affinity of the self-reactive T-cell receptor (TCR) for the foreign antigen is an important factor in regulating the initiation of autoimmunity.

Gronski et al. used a mouse model of diabetes in which the lymphocytic choriomeningitis virus glycoprotein (LCMV-gp) is expressed as a 'self-antigen' under the control of the rat insulin promoter (RIP–gp) by pancreatic β-cells; these mice were crossed with P14 mice, which express a transgenic TCR specific for the gp33 epitope of LCMV-gp presented by H–2Db. Immunization of P14 × RIP–gp mice with wild-type LCMV results in β-cell destruction and diabetes. They looked at the ability of two LCMV variants (crossreacting 'foreign antigens') — LCMV-L6F and LCMV-C4Y, which each have an amino-acid substitution in the most important LCMV-gp epitope — to induce disease in this system.

Both variants bound H–2Db; however, LCMV-L6F had 5-fold lower affinity for the P14 TCR than did wild-type LCMV-gp, and LCMV-C4Y had 20-fold lower affinity. After immunization of P14 mice, wild-type LCMV and LCMV-L6F led to similar changes in activation-marker expression and cytotoxic activity by P14 T cells, although LCMV-L6F led to less T-cell proliferation. The lower-affinity LCMV-C4Y resulted in efficient but reduced cytotoxic activity. These differences in T-cell activation and proliferation properties were reflected in differences in disease induction in P14 × RIP–gp mice infected with the LCMV variants. All mice infected with wild-type LCMV develop diabetes, but only half of the mice infected with the medium-affinity variant, LCMV-L6F, developed disease. None of the mice immunized with the low-affinity variant, LCMV-C4Y, became diabetic.

Christen et al. provide similar evidence that, unless the crossreactive antigen is of sufficient affinity for the self-reactive TCR, disease is not initiated. They used a RIP–NP H–2b mouse model, in which LCMV nucleoprotein is expressed in the pancreas and diabetes is induced in 95% of mice by infection with LCMV. However, infection with Pichinde virus (PV) — which has a crossreactive NP205 epitope that shares six out of eight amino acids with LCMV-NP205 and binds H–2Kb with similar affinity — failed to induce diabetes. This is consistent with the 100-fold lower avidity of PV-NP205 compared with LCMV-NP205 for the crossreactive TCR. But, although low-affinity mimics could not initiate disease in either study, this paper showed that they can accelerate an ongoing autoimmune process when PV infection occurs 1 month after LCMV infection in RIP–NP mice, by selectively expanding the LCMV-NP205-specific T-cell population in the pancreas.

Both studies therefore indicate that molecular mimicry might be unlikely to initiate autoimmunity, except in the case of rare high-affinity mimics. The second study shows that mimicry might have a more important role in accelerating disease in susceptible individuals with pre-existing inflammation of the target organ, and the first study also showed that defects in negative regulation, such as through CBL-B, could contribute to susceptibility.