Molecular constraints on CDR3 for thymic selection of MHC-restricted TCRs from a random pre-selection repertoire

Abstract

The αβ T cell receptor (TCR) repertoire on mature T cells is selected in the thymus, but the basis for thymic selection of MHC-restricted TCRs from a randomly generated pre-selection repertoire is not known. Here we perform comparative repertoire sequence analyses of pre-selection and post-selection TCR from multiple MHC-sufficient and MHC-deficient mouse strains, and find that MHC-restricted and MHC-independent TCRs are primarily distinguished by features in their non-germline CDR3 regions, with many pre-selection CDR3 sequences not compatible with MHC-binding. Thymic selection of MHC-independent TCR is largely unconstrained, but the selection of MHC-specific TCR is restricted by both CDR3 length and specific amino acid usage. MHC-restriction disfavors TCR with CDR3 longer than 13 amino acids, limits positively charged and hydrophobic amino acids in CDR3β, and clonally deletes TCRs with cysteines in their CDR3 peptide-binding regions. Together, these MHC-imposed structural constraints form the basis to shape VDJ recombination sequences into MHC-restricted repertoires.

Fig. 1

Overall αβTCR sequence overlap comparisons among various MHCr and MHCi repertoires. a, b Percentage of TCRβ (a) and TCRα (b) sequence overlaps observed within each strain of B6, B10.BR, BALB/c, Quad^ko (Q), or Quad^koBcl-2^tg (QB) animals. c, d Percentage of TCRβ (c) and TCRα (d) sequence overlaps observed between different strains of animals. The pair-wise sequence overlapping percentages are calculated using top 100, 200, 500, 1000, 2000, 10,000, 40,000 highest frequency and all sequences of each strain, respectively. e TCRβ and TCRα common sequences overlapping among the top 3000 sequences from B6, B10.BR, BALB/c, Quad^ko (Q), or Quad^koBcl-2^tg (QB). f, g The distribution of average length of CDR3β (f) and CDR3α (g) in MHCr and MHCi TCR repertoires with respect to their frequency intervals. Sequences from each strain were grouped by 5% intervals from the top 5% frequent sequences until the accumulation of all sequences (100%). The error bars indicate standard deviations from the means. The p-values all figures are calculated using unpaired student t-test with * = p ≤ 0.05, ** = p ≤ 0.01, *** ≤ 0.001, **** = p ≤ 0.0001. The results included in all panels are from at least two independent experiments with all data included

Fig. 2

CDR3 length dependent amino acid usage. a, b The frequencies of amino acids in FG-loops of CDR3β (a) and CDR3α (b) in B6, Q and QB repertoires with respect to their CDR3 length. c Cys usage of CDR3β (left panel) and CDR3α (right panel) is significantly lower in MHCr repertoires than MHCi repertoires. Our analysis excludes a germline-encoded CDR3 cysteine present in several mouse Vβ gene families because that cysteine is not in the FGβ-loop and does not contact peptide-MHC. d Usage of charged and hydrophobic amino acid usages of MHCr repertoires and MHCi repertoires

Fig. 3

Sequence diversity and CDR3 length distributions. a TCRβ species richness of various TCRβ repertoires based on Chaobc1 analysis. b The inverse simpson index (ISI) was calculated to estimate the clonal diversity of various TCRβ repertoires. c TCRβ sequence overlapping among the top 3000 common sequences to B6 DP1 from B6, B10.BR, BALB/c, Quad^ko (Q), or Quad^koBcl-2^tg (QB). d, e Heatmap showing CDR3 length distributions of TCRβ (d) and TCRα (e) for various MHCr, MHCi and pre-selection repertoires. The mice are identified with a number associated with their strain name (Table S1-S2). For strains with more than three sequencing attempts, only three repertoires are shown here

Fig. 4

Normalized CDR3 length distributions. a–c Comparisons of normalized CDR3 length distributions of TCRβ (a–b) and TCRα (c) among MHCr, MHCi and pre-selection repertoires. The repertoire-specific preference of CDR3 length distributions is indicated by respective fold changes when normalized to B6 DN for TCRβ (a, b) or the average of MHCKO DP for TCRα (c) sequences. Unless otherwise noted, the CDR3 length and all other analyses reported here were carried out using all repertoire data listed in Supplementary Tables 1 & 2 without rejection and regardless of the sequencing depth variation. d MHCr repertoires contain fewer TCRβ sequences with CDR3 length of 15 or longer than MHCi and pre-selection repertoires. e–j The MHC contact area from CDR3 and germline CDR1&2 displayed as percentage of total TCR-MHC interface area in each MHC-TCR complex structures from Protein Data Bank(PDB). Panels e–j are fitted with linear regression using GraphPad Prism with dotted lines indicating 95% confidence

Fig. 5

Comparison of Amino acid usage in TCR CDR3 FG-loop from MHCr, MHCi and pre-selection repertoires. a, b Pearson correlations calculated using normalized amino acid usage among MHCi, MHCr and pre-selection TCR repertoires. The FGβ- (a) and FGα-loop (b) amino acid usage frequencies of individual repertoires are normalized against those of B6 DP_1 and MHCKO DP, respectively. The pair-wise Pearson correlation coefficients were displayed as heatmaps to indicate the distinct amino acid usages of MHCi, MHCr and pre-selection TCR repertoires. c, d The frequency of Cys-containing TCRβ (c) and TCRα (d) sequences in MHCr, MHCi and the pre-selection repertoires. e, f Fold changes of charged and hydrophobic amino acid usages observed in various repertoires. Amino acid frequencies are normalized to those in B6_DP1 repertoire

Fig. 6

V-gene and J-gene usage in various repertoires. a, b Representative Vβ gene (a) and Jβ gene (b) usage of MHCr, MHCi and pre-selection repertoires. c, d Representative Vα gene (c) and Jα gene (d) usage of MHCr and pre-selection repertoires

Fig. 7

V-gene usage comparison and the usage of common sequences. a, b Pair-wise Pearson correlation coefficients were calculated using Vβ-gene (a) and Vα-gene (b) frequencies and displayed as heatmaps for MHCi, MHCr and pre-selection TCR repertoires. The sequences of Vβ- and Vα-genes were from Adaptive Biotechnology and Irepertoire, respectively. c, d Cys and other amino acid usages of CDR3β FG-loop among public sequences from MHCr or MHCi repertoires. The fold change is calculated with respect to those present in B6_DN repertoire. The public sequences of Q and QB are analyzed together

Fig. 8

Frequency scatter plots. a, c Frequency scatter plots among MHC-restricted TCRβ (a) and TCRα (c) repertoires from individual B6 mice. b, d Frequency scatter plots among MHCi TCRβ (b) and TCRα (d) repertoires from individual Quad^ko sequences. Each panel contains the comparison of three independent sequence repertoires from each strain of mice with the horizontal axis representing the frequencies of the first repertoire sequences, vertical axis representing the frequencies of the second (red circle) and third (green circle) repertoires. Common sequences between repertoire 1 and 2 are shown as red circles in the first quadrant with horizontal and vertical coordinates representing their respective frequencies in repertoire 1 and 2. Common sequences between repertoire 1 and 3 are shown as green circles with horizontal and vertical coordinates representing their respective frequencies in repertoire 1 and 3. Non-overlapping sequences appear on their respective axes. Top 200 common sequences present in all three repertoires are shown as filled yellow circles. Linear regressions of the frequency distributions between B6 repertoires 1 and 2 are shown in red lines. e CDR3β length-dependent expansion of MHCr and MHCi sequences from pre-selection repertoires. The percentage of expanded TCRβ sequences during thymic selection were plotted according to their CDR3 length for B6, Quad^ko and Quad^koBcl-2^tg. The analysis included common sequences between pre-selection (B6_DN) and MHCr (B6) or MHCi (Q and QB mice) repertoires. Expansion is defined as sequences whose mature frequencies (normalized for all comparing repertoires) were higher than their pre-selection ones. f Charged amino acid usages of CDR3β FG-loop in common sequences between pre-selection and MHCr or MHCi repertoires. B6 (+ 5%), top 5% TCRβ sequences of B6 repertoires; B6⁺, B6⁻, Q⁺, and Q⁻ represent mature sequences either expanded (+) or not expanded (−) in frequency compared to their pre-selection repertoire