Determinants governing T cell receptor α/β-chain pairing in repertoire formation of identical twins

Abstract

The T cell repertoire in each individual includes T cell receptors (TCRs) of enormous sequence diversity through the pairing of diverse TCR α- and β-chains, each generated by somatic recombination of paralogous gene segments. Whether the TCR repertoire contributes to susceptibility to infectious or autoimmune diseases in concert with disease-associated major histocompatibility complex (MHC) polymorphisms is unknown. Due to a lack in high-throughput technologies to sequence TCR α-β pairs, current studies on whether the TCR repertoire is shaped by host genetics have so far relied only on single-chain analysis. Using a high-throughput single T cell sequencing technology, we obtained the largest paired TCRαβ dataset so far, comprising 965,523 clonotypes from 15 healthy individuals including 6 monozygotic twin pairs. Public TCR α- and, to a lesser extent, TCR β-chain sequences were common in all individuals. In contrast, sharing of entirely identical TCRαβ amino acid sequences was very infrequent in unrelated individuals, but highly increased in twins, in particular in CD4 memory T cells. Based on nucleotide sequence identity, a subset of these shared clonotypes appeared to be the progeny of T cells that had been generated during fetal development and had persisted for more than 50 y. Additional shared TCRαβ in twins were encoded by different nucleotide sequences, implying that genetic determinants impose structural constraints on thymic selection that favor the selection of TCR α-β pairs with entire sequence identities.

Keywords: T cell repertoire; major histocompatibility complex; monozygotic twins; single-cell sequencing.

Fig. 1.

Pairing of V and J elements of TCR α- and β-chains is largely stochastic. (A) Heatmaps show the number of TRAV–TRBV sequence combinations detected in naive (Top) and memory CD4 (Bottom) T cells from one representative individual before (Left) and after (Right) normalization for gene segment frequencies. (B) Observed frequencies of TCR α–β pairs were correlated to those predicted based on the product of the respective gene segment frequencies. Scatter plots show V (blue) and J (red) segment frequencies for naive (Left) and memory (Right) CD4 subsets from one representative individual. Correlation coefficients are from fitting to an identity function without linear regression. (C) Summary box plots of Pearson correlation coefficients for all individuals as shown in Fig. 1B. Different cell populations are indicated as circles (total T), squares (naive CD4), or diamonds (memory CD4), with variable (blue) and joining (red) regions shown separately. Two-sided P values were determined by one-sample Wilcoxon test.

Fig. 2.

Genetic influence on frequencies of TCRαβ pairs is due to biased gene segment usage. (A) Gene segment usage of TRAV, TRAJ, TRBV, and TRBJ elements were compared in twin pairs (red) and unrelated individuals (black). Pearson correlation coefficients for all possible comparisons are shown as box plots for total T (Left), naive CD4 (Middle), and memory CD4 (Right). One-sided P values were determined by permutation test (*P = 0.067; **P = 0.002). (B) Frequencies of TRAV–TRBV (Top) and TRAJ–TRBJ (Bottom) gene segment combinations were compared between twins and unrelated individuals. Scatter plots show correlations in total T (Left), naive CD4 (Middle), and memory CD4 (Right) cells. For each cell population, all possible combinations between 2 individuals are shown comparing twin pairs (red dots) or unrelated pairs (black dots). (C) Summary box plots show Pearson r coefficients calculated for each possible pair of individuals comparing twin pairs (red) vs. unrelated pairs (black). Different cell populations are indicated as circles (total T), squares (naive CD4), or diamonds (memory CD4); Top shows correlation coefficients for variable regions comparisons, and Bottom for joining regions. One-sided P values were determined by permutation test (*P = 0.067; **P = 0.002).

Fig. 3.

Increased sharing of identical TCRαβ pairs in twins. Jaccard indices were computed for sharing of identical TCR α- and TCR β-chain sequences between twin pairs (red) or unrelated individuals (black) as well as for TCRαβ pairs. As lower boundary of detecting identical sequences, sharing in replicate samples (gray) is shown. Sharing across entire variable and joining regions (V-CDR3-J, Left) and sharing within isolated CDR3 regions (Right) are shown. Box plots show results from total T (A), naive CD4 (B), and memory CD4 (C) cells; if no identical sequences were found (TCRαβ pairs in B), the detection limit is shown. Fold differences between twins and unrelated individuals average sharing of identical sequences are indicated. One-sided P values were determined by permutation test (*P = 0.067).

Fig. 4.

Mechanisms driving sharing of identical TCRαβ sequences. (A) TCR α- and β-chain amino acid CDR3 lengths were analyzed independently comparing TCRαβ sequences shared vs. those not shared between twin–twin pairs. Data are shown as Gaussian distributed histograms. Top shows the distribution of α-chain CDR3 lengths in total T (Left), naive CD4 (Middle), and memory CD4 (Right) cells; Bottom shows those for β-chains. Two-sided P values were determined by unpaired t test. (B) Box plots show percentage of identical nucleotide TCRαβ sequences in the shared TCRαβ amino acid clonotypes among twins and unrelated individuals across all cell types; P value was determined by Mann–Whitney U test. (C) Jaccard indices were computed for the sharing of identical sequences in combined total T and memory CD4 samples between any 2 unrelated individuals. Results are plotted vs. the number of shared MHC class I (Left) and class II (Right) alleles; P values were determined by trend test.