Immunogenic epitope panel for accurate detection of non-cross-reactive T cell response to SARS-CoV-2

Abstract

The ongoing COVID-19 pandemic calls for more effective diagnostic tools. T cell response assessment serves as an independent indicator of prior COVID-19 exposure while also contributing to a more comprehensive characterization of SARS-CoV-2 immunity. In this study, we systematically assessed the immunogenicity of 118 epitopes with immune cells collected from multiple cohorts of vaccinated, convalescent, healthy unexposed, and SARS-CoV-2-exposed donors. We identified 75 immunogenic epitopes, 24 of which were immunodominant. We further confirmed HLA restriction for 49 epitopes and described association with more than 1 HLA allele for 14 of these. Exclusion of 2 cross-reactive epitopes that generated a response in prepandemic samples left us with a 73-epitope set that offered excellent diagnostic specificity without losing sensitivity compared with full-length antigens, and this evoked a robust cross-reactive response. We subsequently incorporated this set of epitopes into an in vitro diagnostic Corona-T-test, which achieved a diagnostic accuracy of 95% in a clinical trial. In a cohort of asymptomatic seronegative individuals with a history of prolonged SARS-CoV-2 exposure, we observed a complete absence of T cell response to our epitope panel. In combination with strong reactivity to full-length antigens, this suggests that a cross-reactive response might protect these individuals.

Keywords: Antigen; COVID-19; Clinical Trials; Peptides; T cells.

Figure 1. Characteristics of the peptide set.

(A) Number of epitopes selected from each indicated publication (detailed in Supplemental Table 1) for the MHC-I (left) and -II (right) sets. The distribution of the peptides according to the number of HLA that they bind is shown at top. The x axis displays the number of predicted binding alleles per peptide. The y axis shows the percentage of peptides that bind to a given number of alleles. Numbers below the SARS-CoV-2 genome schematic indicate the number of peptides derived from each gene. (B) The number of HLA class I (left) and -II (middle) alleles alone or in combination (right) that are predicted to bind at least 1 peptide from the set per individual among 2210 donors from the BM registry. (C) Antigen response among the healthy (HD-2019) cohort (n = 52). The normalized mean of 2 duplicate wells and the median and interquartile range. Cross-reactive MHC-II peptides are marked with red arrows. The positive threshold is indicated by the dotted line.

Figure 2. Response to MHC-II peptides differs significantly in Vac and CP donors.

(A) Response to the indicated antigens as measured by ELISpot for Vac (n = 43) and CP (n = 51). A normalized mean of 2 duplicate wells and a median with interquartile range. Mann-Whitney U test (S peptides, P = 0.013; MHC-II peptides, recombinant S protein, P < 0.0001). (B) Volcano plot shows the effect of a particular HLA allele on response to the same peptide sets and antigens. The x axis denotes the decimal logarithm of the ratio of the median response among HLA carriers to that of individuals without the HLA. The y axis denotes the negative decimal logarithm of the P value. The 3 most significant associations are annotated. P = 0.05 is depicted by the dotted line (Fisher’s exact test). (C–E) Receiver operating characteristic (ROC) curves for MHC I + II peptides versus S peptides in Vac versus HD-2019 samples (C); MHC I + II peptides versus S, N, or M peptides in CP versus HD-2019 samples (D); and MHC I + II peptides versus the sum of S, N, and M peptides and versus multiple regression model, incorporating S, N, and M peptides in CP versus HD-2019 samples (E). Three HD-2019 donors with cross-reactive responses were excluded from the ROC analysis for MHC I + II peptides.

Figure 3. T cell response to MHC-I and MHC-II epitopes in CP and Vac.

(A and B) Immunogenicity in carriers of various HLA-I (A) or HLA-II (B) alleles. For the right sides of the plots, gray bars indicate the number of tested carriers; colored bars indicate the number of responses. Colors indicate source protein, with 3-letter amino acid codes at left. Superscript numbers indicate peptide length, and the asterisk denotes a peptide with an ambiguous HLA association. The left sides of the plots show the number of responses in donors without the indicated HLA. For HLA-I, the association between the response and a single HLA allele is shown; for HLA-II, the best associations (including associations with several HLAs) are shown. Fisher’s exact test; P < 0.05 were considered significant (exact P values are specified in Supplemental Table 3). (C) Number of S protein–derived MHC-I and MHC-II epitopes recognized per individual for the CP and Vac cohorts; median and interquartile range are shown

Figure 4. Clinical trial confirmed the high accuracy of the Corona-T-test in discriminating healthy donors from vaccinated and convalescents.

(A) Scatter plot shows Corona-T-test result. A normalized mean of 2 duplicate wells for the Vac_trial (n = 69) and CP_trial (n = 50) and HD-2021 (n = 95) cohort with median and interquartile range. Results in the gray zone (n = 6) were excluded from the ROC-analysis. Kruskal-Wallis with Dunn’s post hoc test; ****P < 0.0001. (B) ROC curve for Vac_trial (AUC = 0.98) an CP_trial (AUC = 0.97) versus HD-2021_trial. (C and D) Number of confirmed HLA-I and -II alleles per individual binding at least 1 peptide from any protein (C) or S protein (D) (n = 2210).

Figure 5. Evidence for a cross-reactive rather than SARS-CoV-2–specific response in the healthy exposed (HE) cohort.

(A) Response to the antigens (normalized mean of 2 wells after subtracting negative control) in HE (n = 37), CP (n = 51), and HD-2019 (n = 52) cohorts. Each bar represents an individual donor; colors represent particular antigens. (B) Comparison of the response to antigens between cohorts as measured with ELISpot. A normalized mean of 2 duplicate well, and the median response and interquartile range. Kruskal-Wallis with Dunn’s post hoc test; ****P < 0.0001, **P < 0.01. (C) Difference in responses to MHC-II peptides before and after exclusion of 2 cross-reactive peptides in the HE cohort (Wilcoxon test; P = 0.0045).**P < 0.01.