Incomplete immune response to coxsackie B viruses associates with early autoimmunity against insulin

Abstract

Viral infections are associated with autoimmunity in type 1 diabetes. Here, we asked whether this association could be explained by variations in host immune response to a putative type 1 etiological factor, namely coxsackie B viruses (CVB). Heterogeneous antibody responses were observed against CVB capsid proteins. Heterogeneity was largely defined by different binding to VP1 or VP2. Antibody responses that were anti-VP2 competent but anti-VP1 deficient were unable to neutralize CVB, and were characteristic of children who developed early insulin-targeting autoimmunity, suggesting an impaired ability to clear CVB in early childhood. In contrast, children who developed a GAD-targeting autoimmunity had robust VP1 and VP2 antibody responses to CVB. We further found that 20% of memory CD4(+) T cells responding to the GAD65247-266 peptide share identical T cell receptors to T cells responding to the CVB4 p2C30-51 peptide, thereby providing direct evidence for the potential of molecular mimicry as a mechanism for GAD autoimmunity. Here, we highlight functional immune response differences between children who develop insulin-targeting and GAD-targeting autoimmunity, and suggest that children who lose B cell tolerance to insulin within the first years of life have a paradoxical impaired ability to mount humoral immune responses to coxsackie viruses.

Figure 1. Antibody responses against CVB capsid proteins.

Antibody titers against VPs of CVB1-6 are represented as a heat map where each vertical line is a sample (n = 440), and the intensity of the color corresponds to the relative light units (RLU) immunoprecipitated in the LIPS assay. Four clusters were identified by hierarchical clustering of Euclidean distances calculated from anti-VP1 and anti-VP2 antibody titers. The clusters represent samples with weak responses to VP1 and/or VP2 (cluster 1, weak), samples with strong and dominant VP2 antibody responses (cluster 2, VP2), samples with strong and dominant VP1 antibody responses (cluster 3, VP1), and samples with strong VP1 and VP2 antibody responses (cluster 4, VP1/VP2). Within each cluster, there is a small subset of samples with strong antibodies to VP4.

Figure 2. Anti-VP1 antibodies inhibit CVB plaque formation.

Plaque neutralization tests using CVB3 were performed with 8 representative sera that were selected based on their anti-CVB profile (Supplementary Fig. 3) as strong anti-VP2 and VP1 antibodies (filled symbols), or strong anti-VP2 antibodies only (open symbols). Despite similar anti-VP2 titers in each serum, the anti-VP1 high titer sera had >100-fold stronger neutralization capacity than the anti-VP1 deficient sera. ****p-value < 0.0001 based on repeated measures ANOVA with Bonferroni post-test comparisons.

Figure 3. Early insulin autoimmunity is associated with a reduced anti-VP1 CVB antibody response.

Anti-CVB antibodies in sera from 89 children genetically at risk for type 1 diabetes and followed for beta cell autoantibody development in the BABYDIET study. Anti-VP1 (a), anti-VP2 (b) and anti-VP4 (c) antibodies at 3 years of age (y axis) are shown in children stratified as beta cell autoantibody negative throughout follow-up (no autoimmunity), beta cell autoantibody seroconversion before age 3 years with IAA as the first detected autoantibodies (early insulin), and beta cell autoantibody seroconversion with GAD as the first detected autoantibody target (GAD). Each point is an individual and the horizontal bar indicates the median. For the early insulin autoimmunity children, the unique symbols correspond to data points of an individual child. *p-value < 0.05 or **p-value < 0.01 based on comparison with early insulin autoimmunity group for the corresponding CVB serotype using the Mann Whitney U test.

Figure 4. Maternal anti-CVB antibodies are not associated with beta cell autoimmunity.

Anti-VP1 antibodies in cord blood serum of BABYDIAB study participants stratified as beta cell autoantibody negative throughout follow-up (no autoimmunity), beta cell autoantibody seroconversion before age 3 years with IAA as the first detected autoantibodies (early insulin), and beta cell autoantibody seroconversion with GAD as the first detected autoantibody target (GAD). Each point is an individual and the horizontal bar indicates the median. No significant differences based on comparison with early insulin autoimmunity group for the corresponding CVB serotype using the Mann Whitney U test.