Tracking DNA-based antigen-specific T cell receptors during progression to type 1 diabetes

Abstract

T cells targeting self-proteins are important mediators in autoimmune diseases. T cells express unique cell-surface receptors (TCRs) that recognize peptides presented by major histocompatibility molecules. TCRs have been identified from blood and pancreatic islets of individuals with type 1 diabetes (T1D). Here, we tracked ~1700 known antigen-specific TCR sequences, islet antigen or viral reactive, in bulk TCRβ sequencing from longitudinal blood DNA samples in at-risk cases who progressed to T1D, age/sex/human leukocyte antigen-matched controls, and a new-onset T1D cohort. Shared and frequent antigen-specific TCRβ sequences were identified in all three cohorts, and viral sequences were present across all ages. Islet sequences had different patterns of accumulation based upon antigen specificity in the at-risk cases. Furthermore, 73 islet-antigen TCRβ sequences were present in higher frequencies and numbers in T1D samples relative to controls. The total number of these disease-associated TCRβ sequences inversely correlated with age at clinical diagnosis, indicating the potential to use disease-relevant TCR sequences as biomarkers in autoimmune disorders.

Fig. 1.. Antigen-specific TCR sequences within bulk TCRβ chain repertoires from individuals genetically at risk and with new-onset T1D.

(A) Study design showing the three patient cohorts and number of peripheral blood samples from each cohort used to perform bulk TCRβ chain sequencing (~50 million total sequences). This large database of TCRβ sequences was queried using a curated list of 1743 antigen-specific TCR sequences, islet antigen or viral protein reactive, that was compiled from the published literature. (B) Graphs displaying the samples obtained by age from cases during progression to clinical T1D (left) and age-matched controls (right). Each row is a participant, with open circles indicating a sample. In preclinical T1D cases, orange circles indicate the sample at which the participant became islet autoantibody positive (seroconversion), and red circles indicate the age of clinical T1D onset. (C) Shown are the results of the sample search using the curated list of known antigen-specific TCRβ sequences, grouped by T cell subset and antigen specificity. Public TCRβ sequences were found in at least one sample, and frequent TCRβ sequences were found in at least 1% of the samples (i.e., >4 samples of 357) from any cohort.

Fig. 2.. Viral antigen–specific CD8 TCRβs are present across ages and patient cohorts.

Scatterplots displaying viral antigen TCRβ sequence total template numbers for each sample relative to age in years for (A) controls, (B) cases that progressed to clinical T1D, and (C) patients with new-onset T1D. Each dot represents the sum of all templates for TCRβ sequences with a given antigen specificity in one sample. Plots for all viral antigen–specific TCRβ sequences include both CD4 sequences (n = 2) and CD8 sequences (n = 103), while the remaining plots display viral CD8 TCRβ sequences grouped by antigen specificity; influenza (n = 51), EBV (n = 43), CMV (n = 2), and ADV (n = 7). P values were calculated using mixed-effects models to account for multiple measurements in cases and controls; linear regression was used for new-onset T1D. *P < 0.05.

Fig. 3.. Differential patterns of islet antigen–specific TCRβ sequences during progression to T1D.

Bar graphs showing the total number of TCRβ sequence templates in controls (blue) and preclinical T1D cases (red) within defined age groups. (A) Results for all islet antigen–reactive CD4, (B) PPI-reactive CD4, and (C) GAD-reactive CD4. The bottom panel of graphs depicts (D) all islet antigen–reactive CD8, (E) PPI-reactive CD8, and (F) ZnT8-reactive CD8 TCRβ sequence templates. Sample numbers for each age bin: 0 to 3 (controls, n = 24; cases, n = 26), 3 to 6 (controls, n = 27; cases, n = 30), 6 to 9 (controls, n = 24; cases, n = 26), 9 to 12 (controls, n = 14; cases, n = 17), and 12+ (controls, n = 11; cases, n = 15). P values were calculated using mixed-effects models to compare cases to controls within each age bin and to test for an interaction between group (cases and control) and time. None of the trajectories between cases and controls over time were statistically significant. *P < 0.05.

Fig. 4.. Identifying T1D disease–associated TCRβ chain sequences.

Depicted are scatterplots with each TCRβ sequence plotted as the difference in sample frequency (%, x axis) versus the difference in templates (#, y axis) in preclinical T1D cases minus controls (left column) and in new-onset T1D minus controls (right column). Graphs are displayed by antigen specificity and T cell subset for (A) PPI CD4, (B) GAD CD4, and (C) all islet antigen–reactive CD8. Disease-associated TCRβs (n = 73 unique TCRβ sequences) are indicated in dark red (case-associated) or in light red (new onset–associated). Black circles represent those TCRβ sequences that are not case- or new onset–associated. Two data points are not displayed on the case plots to help visualize the data: [−12.23, 14.00 in (A)] and [−18.54, −34.00 in (B)].

Fig. 5.. Temporal changes within individual disease-associated islet antigen TCRβ sequences during T1D development.

Shown are multivariable plots for each disease-associated (D-A) TCRβ, grouped by antigen specificity and T cell subset for (A) PPI CD4 (n = 21 TCRβ sequences), (B) GAD CD4 (n = 40), and (C) PPI CD8 (n = 4), and ZnT8 CD8 (n = 8). Case samples are grouped in the first three rows by seroconversion status: before islet autoantibody seroconversion (Aab⁻), at islet autoantibody seroconversion, and after islet autoantibody positivity (Aab⁺). Results from the new-onset T1D cohort are shown in the bottom row. Dot size depicts the sample frequency within a patient cohort for each disease-associated TCRβ sequence, while a darker color depicts a higher number of TCRβ chain templates in case or new-onset T1D samples. *Disease-associated TCRβ sequence numbers 3, 4, 13, and 15 respond to HIPs.

Fig. 6.. Total numbers of disease-associated islet antigen TCRβ sequences correlates to an earlier age at T1D diagnosis.

(A) Scatterplots depicting the total number of unique disease-associated TCRβ sequences (left) or number of templates (right) versus age at T1D diagnosis for the new-onset cohort and (B) preclinical T1D cases for all samples after islet autoantibody seroconversion versus age at diagnosis. P values were calculated using linear regression. *P < 0.05.