HLA-B*15:01-positive severe COVID-19 patients lack CD8+ T cell pools with highly expanded public clonotypes

Abstract

Understanding host factors driving asymptomatic versus severe disease outcomes is of key importance if we are to control emerging and re-emerging viral infections. HLA-B*15:01 has been associated with asymptomatic SARS-CoV-2 infection in nonhospitalized individuals of European ancestry, with protective immunity attributed to preexisting cross-reactive CD8⁺ T-cells directed against HLA-B*15:01-restricted Spike-derived S_919-927 peptide (B15/S₉₁₉⁺CD8⁺ T-cells). However, fundamental questions remained on the abundance and clonotypic nature of CD8⁺ T-cell responses in HLA-B*15:01-positive patients who succumbed to life-threatening COVID-19. Here, we analyzed B15/S₉₁₉⁺CD8⁺ T-cell responses in COVID-19 patients from independent HLA-typed COVID-19 patient cohorts across three continents, Australia, Asia and Europe. We assessed B15/S₉₁₉⁺CD8⁺ T-cells in COVID-19 patients across disease outcomes ranging from asymptomatic to hospitalized critical illness. We found that severe/critical COVID-19 patients mounted B15/S₉₁₉⁺CD8⁺ T-cell responses lacking a highly expanded key public B15/S₉₁₉⁺CD8⁺ T-cell receptor (TCR; TRAV9-2/TRBV7-2) which recurred across multiple individuals in COVID-19 patients with a mild disease. Instead, B15/S₉₁₉⁺CD8⁺ T-cell responses in life-threatening disease had a prevalence of an alternate TCR clonotypic motif (TRAV38-2/DV8/TRBV20-1), potentially contributing, at least in part, to why B15/S₉₁₉⁺CD8⁺ T-cells in severe COVID-19 patients were less protective. Interestingly, the frequency, memory phenotype, and activation profiles of circulating B15/S₉₁₉⁺CD8⁺ T-cells did not differ across disease severity. Moreover, B15/S₉₁₉⁺CD8⁺ T-cells were better maintained into convalescence compared to other SARS-CoV-2-specificities. Our study thus provides evidence on the differential nature of the TCR clonal repertoire in 22.37% of HLA-B*15:01-positive COVID-19 patients who developed severe or critical disease in our cohorts, comparing to HLA-B*15:01-expressing individuals with mild COVID-19.

Keywords: CD8+ T cells; COVID-19; HLA-B*15:01; T cell receptors; severe disease.

Fig. 1.

B15/S₉₁₉⁺CD8⁺ T cell frequency and phenotype are maintained across time and disease severity. (A) Representative flow cytometry plots of enriched B15/S₉₁₉-specific tetramer⁺ CD8⁺ T cells. (B) B15/S₉₁₉⁺CD8⁺ T cell frequencies across time following primary SARS-CoV-2 infection. (C) Percentage of B15/S₉₁₉⁺CD8⁺ T cells cross-reactive with hCoV B15/S₁₀₁₂ across disease severity groups, median with interquartile range (IQR). (D) Frequency of B15/S₉₁₉⁺CD8⁺ T cells prepandemic and across time post infection, median with IQR. (E) Memory phenotype profiles across time for B15/S₉₁₉⁺CD8⁺ T cells, mean with SD. (F) Frequency of B15/S₉₁₉⁺CD8⁺ T cells across disease severity groups, median with IQR. (G) Phenotype profiles across disease severity groups for B15/S₉₁₉⁺CD8⁺ T cells, mean with SD. (H) Heatmap showing activation (CD71, CD38, HLA-DR, PD-1, and TIM-3) and memory phenotype of SARS-CoV-2 epitope-specific CD8⁺ T cells. (I) Matched B15/S₉₁₉⁺ and other SARS-CoV-2⁺CD8⁺ T cell frequencies within individuals across time. Statistical significance determined by (C, D, and F) Dunn’s multiple comparisons test, (E and G) Tukey’s multiple comparisons test and (I) Wilcoxon matched-pairs signed rank test. *P ≤ 0.05, **P ≤ 0.01, ***P ≤ 0.001, ****P ≤ 0.0001. Frequency of tetramer⁺CD8⁺ T cells are shifted by 10⁻⁶ (i.e., no detected tetramer⁺ events displayed as 10⁻⁶) to allow for visibility on a logarithmic y axis. Any samples with <10 tetramer⁺ events are shown as open symbols and only samples with 10 or more tetramer⁺ events are included in the phenotypic analysis (E, G, and H). For samples run in duplicate, averaged tetramer⁺ T cell frequencies are plotted.

Fig. 2.

Altered B15/S₉₁₉-specific TCR repertoire in severe/critical COVID-19 patients. (A) Circos plots depicting TRA and TRB gene usage for TCRαβ clonotypes specific to B15/S₉₁₉. Left outer arch color indicates TRAV usage, right outer arch and segment color depicts TRBV usage (n = sequences). (B and C) Alluvial plots showing frequency of TRA (B) and TRB (C) gene usage in B15/S₉₁₉-specific TCR repertoires (n = sequences). Connections between bars represent shared CDR3 usage between individuals of different COVID-19 severity. (A–C) Colors represent variable gene segment usage, while divisions represent TCRαβ clonotypes with the same CDR3 sequence. (D) Overall proportion of singleton and expanded clonotypes (≥2 clonotypes in one donor) from total TCRαβ sequences per disease group and prepandemic. Statistical significance determined by (B and C) Simpson’s Diversity Index and (D) Fisher’s exact test for comparison between disease groups. *P ≤ 0.05, **P ≤ 0.01, ***P ≤ 0.001, ****P ≤ 0.0001.

Fig. 3.

B15/S₉₁₉-specific TCR repertoire in severe/critical patients dominated by an alternate motif. (A) Network analysis of paired B15/S₉₁₉-specific TCRαβ clonotypes. Network nodes (circles) represent clonotypes of paired TCRαβ, and are connected by edges (lines) to other clonotypes that are ≤120 TCRdist units (22). Size of the nodes represents the size of the clonotypes. (B) Logo plots for networked clonotype clusters with ≥5 clonotypes detected in B15/S₉₁₉-specific TCR repertoire. Cluster numbers indicated in (A). Sequence logo shows the probability of each amino acid residue at each CDR3 position. Colors represent amino acid chemistry: red, acid; blue, basic; black, hydrophobic; purple, neutral; green, polar. (C and D) Probability of generation (P_gen; log10 transformed) for TCRα and TCRβ chains were generated with tcrdist3 (23) and plotted for (C) prepandemic and disease severity groups and (D) clusters. (C) Clonally expanded TCRs within an individual were reduced to a single data point for this analysis. (D) Subclusters represent clusters consisting of 3 or 4 clonotypes, pairs represent clusters consisting of two clonotypes and clones are singleton TCRs; clonally expanded TCRs were reduced to a single data point for this analysis. Plots represent the median with 95% CI. Statistical significance determined by (C and D) Dunn’s multiple comparisons test. *P ≤ 0.05, **P ≤ 0.01, ***P ≤ 0.001, ****P ≤ 0.0001.

Fig. 4.

No HLA associations with asymptomatic or symptomatic SARS-CoV-2 infection across four independent Asian ancestry cohorts. (A) Asian ancestry cohorts for HLA association analysis. (B) Forest-plot depicting odds ratios (OR) and 95% CI with no HLA-B*15:01 association with asymptomatic SARS-CoV-2 infection across four independent Asian ancestry cohorts. (C) OR and 95% CI for HLA associations with asymptomatic or symptomatic SARS-CoV-2 infection in four independent Asian ancestry cohorts with Bonferroni correction across alleles. (A–C) Additional information in SI Appendix, Tables S3–S6. ctrl, control; cf, confer; OR, odds ratio; CI, confidence interval; P_adj, adjusted P value.