This study was approved by Institutional Review Board of the Catholic University of Korea (MC21SESI0005). All healthy donors provided written informed consent before participation in this study. Healthy donors who did not have chronic diseases such as diabetes, thyroid, and hypertension and were not taking medications. A complete blood cell count test and additional tests for HBV, HCV, HIV-1, HIV-2, and syphilis were performed on all donors. The absence of active tuberculosis was confirmed by regular medical check-up. The participants are ranged from 26.4 ± 4.0 years of age and composed of 6 females and 27 males. PBMCs were isolated by density gradient centrifugation using Ficoll–Hypaque (GE Healthcare). CD4⁺ T cells were isolated and CD8⁺ T cells were depleted using magnetic microbeads (AutoMacs Pro separator; Miltenyi Biotec), and the purity of the CD4⁺ T cells and CD8^- PBMCs was confirmed with flow cytometry (CD8⁺ T cells in CD8^- PBMC: 0.7% ± 0.7%, Supplementary Figure S1). CD8⁺ T cells were depleted to precisely measure the response by CD4⁺ T cells. Suspended CD4⁺ T cells and CD8^- PBMCs in fetal bovine serum (Gibco) containing 10% dimethyl sulfoxide (Mylan) and 50% RPMI 1640 medium (Lonza) were cryopreserved in liquid nitrogen until ELISPOT assay or cultured. The HLA was then genotyped using sequencing-based typing and next-generation sequencing at the Catholic Hematopoietic Stem Cell Bank (Table 1) (NCBI BioProject Accession: PRJNA721949) (MiSeqDx, Illumina), as previously described (32, 33).

The panels of aAPCs expressing a single HLA class II allotype were established in the previous study (32, 33). Briefly, cDNAs of HLA-DRA allele, 21 HLA-DRB1 alleles, 10 HLADQA1 alleles, 11 HLA-DQB1 alleles, 3 HLA-DPA1 alleles and 11 HLA-DPB1 alleles were obtained from lymphoblastoid cell lines of donors (740902.50; Macherey Nagel, RT300M; Enzynomics). PCR products including cDNA sequences of HLA class II alleles were cloned into the pCDH lentivector (#CD523A-1; System Biosciences) using In-Fusion Cloning (EZ015TL; Enzynomics). The sequences were verified by Sanger sequencing. 293TN cells (System Biosciences) were transfected (Lipofectamine 2000; Invitrogen) with psPAX2 (RRID: Addgene_12260), pMD.2G (RRID: Addgene_12259), and pCDH encoding a single HLA class II allele. After 48h the supernatant was harvested. The K562-based aAPCs, which do not express HLA molecules, were transduced with lentiviruses encoding an alpha and a beta chain at a multiplicity of infection (MOI) of 20. The transduced aAPCs were isolated by fluorescence-activated cell sorting (FACSAria Fusion, BD Biosciences) using the following monoclonal antibodies; HLA-DR (clone G46-6; RRID: AB_1727527), HLA-DQ (clone Tü169; RRID: AB_2738963, ü39; RRID: AB_395940), and HLA-DP (clone B7/21).The aAPCs expressing a single HLA class II allotype were then cultured in RPMI 1640 supplemented with 2 mM L-glutamine, 100 U/ml penicillin–streptomycin–amphotericin B mixture (Lonza), and 10% fetal bovine serum (Gibco). The expression of HLA class II on aAPCs was confirmed using flow cytometry (FACSCanto, BD Biosciences). All HLA-DR alpha chain were omitted because DRA1*01:01 alone was used.

5×10⁴ or 5×10³ aAPCs were loaded with 60 nM of 15 -mer amino acid spanning the entire M. tuberculosis TB10.4, Ag85b, ESAT-6 and CFP-10 protein with 11 -mer amino acid overlap (JPT Peptide Technologies) for 3h in serum-free medium in a 96-well plate. The peptide loaded aAPCs were washed three times with serum-free medium using centrifuge and microplate washer (405LSR; BioTek).

Briefly, an ELISPOT plates (Human IFN-γ ELISPOT set, BD, USA) were coated with IFN-γ capture antibody and incubated overnight at 4°C. The plates were washed 3 times with phosphate-buffered saline (PBS) and then blocked with RPMI 1640. After 2h the blocking, the plates were washed with PBS. The 5×10⁵ CD4⁺ T cells were co-incubated with peptide-pulsed or not 5×10⁴ aAPCs for 20h. The cells were removed and the plates were washed 3 times with 0.05% Tween 20/PBS using microplate washer (405LSR; BioTek). Biotinylated detection antibody for IFN-γ was added and incubated for 2h at 37°C. After 4 times washes with 0.05% Tween 20/PBS, avidin-horseradish peroxidase was added and plates were incubated for 1h at 37°C. The plates were washed 4 times with 0.05% Tween 20/PBS, followed by 2 washes with PBS and the addition of AEC substrate (BD) per well. The reaction was stopped by washing with deionized water, and the plates were dried overnight. The spot forming units were counted using an AID ELISPOT Reader System (AID Diagnostika GmbH). The frequencies of an HLA allotype-restricted CD4⁺ T cell response to antigens were calculated as [(response to aAPCs expressing HLA pulsed with peptide pools) − (response to aAPCs expressing HLA)] − [(response to aAPCs pulsed with peptide pools) − (response to aAPCs)], as previously described (32, 33).

On Day 0, isolated CD4⁺ T cells and CD8^- PBMCs were mixed 2:1 (8-12×10⁶) at 4×10⁶/ml. Each M. tuberculosis antigen peptide pools 0.1ug/ml were added to RPMI 1640 supplemented with 2 mM L-glutamine, 100 U/ml Penicillin-Streptomycin-Amphotericin B Mixture (Lonza), 10% fetal bovine serum (Gibco), 500U/ml IFN-γ (LG Chemical) and 200U/ml hIL-2 (JW CreaGene) and incubated in 5% CO₂, 37°C. On Day 1, 4, and 7, hIL-2 was added to medium 200U/ml. On Day 11, cells were sub-cultured to replenish the medium at 2×10⁶/ml with hIL-2 200U/ml. On Day 14, cells were harvested and measured IFN-γ secretion by ELISPOT assay. The 5×10⁴ cultured CD4⁺ T cells and 5×10³ antigen-pulsed or not aAPC were co-incubated for 20h at 37°C. M. tuberculosis antigens-specific responses of cultured CD4⁺ T cell responses were measured by IFN-γ ELISPOT assay as described in cultured ELISPOT assay. To compare ex vivo and cultured ELISPT, the frequencies of each ELISPOT were adjusted to account for differences in cell numbers and proliferation folds among the donors. It was calculated as (SFCs/5×10⁴ CD4⁺ T cells) × (1×10⁶) × proliferation fold. These were determined as adjusted Spots-Forming Cells (aSFCs) by cultured ELISPOT.

To analyze and visualize the data, Microsoft Excel, GraphPad Prism 7, and FlowJo v10 (BD) were used. Statistical significance was determined by Wilcoxon test, Pearson’s correlation analysis, Welch’s t-test (with a two-tailed test of significance), one-way analysis of variance [ANOVA]. Values of P <0.05 were considered significant. The data are expressed as means ± standard deviation or standard error of the mean, and the sample sizes are presented in the figures.

Peptide pools of TB10.4, Ag85b, ESAT-6, and CFP-10 were used to assess CD4⁺ responses to M. tuberculosis according to a single HLA class II allotypes in healthy donors. HLA-DRB1, -DQA1, -DQB1, -DPA1, and -DPB1 alleles were genotyped in 33 healthy donors (Table 1). The single HLA class II allotype-expressing aAPCs were established for those whose alleles were typed. The CD4⁺ T cells from peripheral blood were stimulated by aAPCs pulsed with the M. tuberculosis antigens. M. tuberculosis antigen-specific CD4⁺ T cell responses were measured by IFN-γ ELISPOT assay and were designated to ex vivo ELISPOT (Figure 1). Cultured ELISPOT was conducted to measure the responses of memorial CD4 T cells by culturing with M. tuberculosis antigens for 14 days (Figure 1A). Individual donors showed various values of proliferation fold. The average of cell proliferation fold during the culture was 0.8 (min 0.04, max 3.8) (Figure 1B). Using aAPCs expressing a single HLA class II allotype, CD4⁺ T cell responses by HLA class II allotypes possessed by each donor were measured with ex vivo ELISPOT and cultured ELISPOT (Figure 1C). Cultured CD4⁺ T cells were analyzed expression of CCR7 and CD45RO to define T cell subsets (Supplementary Figure S3). The cultured CD4⁺ T cells were CD45RO⁺CCR7^- which indicates cultured CD4⁺ T cells are effector memory.

CD4⁺ T cell responses specific to M. tuberculosis antigens were analyzed by HLA-DR, -DQ, and -DP loci to compare the responses by each locus (Figure 2). The response by a locus was calculated by summing the responses by two alleles of a locus. The response by HLA-DR was higher than by HLA-DQ and -DP within donors in ex vivo ELISPOT (Figure 2A, Wilcoxon test, DR vs. DQ p = 0.0051; DR vs. DP p = 0.0136). There was no significant difference in responses by HLA-DQ and -DP (Wilcoxon test, p = 0.6939).

Next, the responses of memory T cells were measured by cultured ELISPOT. The response in cultured ELISPOT was increased 39-fold compared to ex vivo ELISPOT (3791 SFCs per a million in cultured ELISPOT; 98 SFCs per a million in ex vivo ELISPOT). In cultured ELISPOT, the response by HLA-DR was higher than by HLA-DQ and -DP, consistent with ex vivo ELISPOT (Figure 2B, Wilcoxon test, DR vs. DQ p = 0.0002; DR vs. DP p = 0.0002). The response by HLA-DQ was similar to the response by HLA-DP, as in ex vivo ELISPOT (Wilcoxon test, p = 0.7259). In the sum of the responses by loci, there was no pattern in the responses by HLA class II in cultured and ex vivo ELISPOT (Figure 2C, Pearson’s correlation, p = 0.0986). According to the analysis of the response by each locus, the response by HLA-DR was correlated in cultured and ex vivo ELISPOT (Figure 2D, Pearson’s correlation, p < 0.0001). On the contrary, the response by HLA-DQ and -DP was not correlated in cultured ELISPOT and ex vivo ELISPOT, and it was detected only in cultured ELISPOT or only in ex vivo ELISPOT (Figure 2E , F, HLA-DQ p = 0.441; HLA-DP p = 0.2544).

Using aAPCs expressing a single HLA class II allotype, the CD4⁺ T cell responses by HLA class II allotypes possessed by each individual were measured with ex vivo ELISPOT and cultured ELISPOT, and the distribution of responses to each allotype was investigated. In cultured ELISPOT, DRB1*01:01, DRB1*14:06, DRB1*15:01, DRB1*15:02, DRB1*04:03, DRB1*09:01, DRB1*13:02, DRB1*04:04, DRB1*08:02, DRB1*07:01, DRB1*04:06, DRB1*11:01 and DRB1*03:01 among 20 HLA-DR allotypes (Figure 3A), DQA1*01:02/DQB1*06:09, DQA1*05:03/DQB1*03:01, DQA1*02:01/DQB1*02:02, DQA1*01:03/DQB1*06:01 and DQA1*01:01/DQB1*05:01 among 16 HLA-DQ allotypes (Figure 3B), and, DPA1*01:03/DPB1*03:01, DPA1*02:02/DPB1*05:01 and DPA1*01:03/DPB1*04:01 among 13 HLA-DP allotypes (Figure 3C) showed an average of more than 200 aSFCs per a million and they showed a high response in the order listed. Most of allotypes in ex vivo ELISPOT showed less than 200 SFCs per a million.

CD4⁺ T cell responses restricted by an HLA class II allotype by cultured ELISPOT were analyzed in individual donors to investigate the allotype dominance to M. tuberculosis antigens. A positive response was defined in the case that the response by an allotype is higher than 100 SFCs per 5×10⁴. Of the 33 donors, ten showed a positive response in an allotype; seven showed a positive response by only one allotype; three showed a positive response by two allotypes (Figure 4). The remaining 23 donors did not show a positive response in any allotype. We classified the response by an HLA allotype in the order of responses’ magnitude to compare the response without an arbitrary threshold in a donor. The highest response by an allotype was higher than the second, third, and fourth highest responses by the other allotypes (Supplementary Figure S4, one-way ANOVA, 1^st vs. 2^nd p = 0.0033, 1^st vs. 3^rd and 1^st vs. 4^th p < 0.0001). And we also classified the response by an HLA allotype of individuals showed positive responses in cultured ELISPOT in the order of responses’ magnitude. The highest response by an allotype was higher than the second, third, and fourth highest responses by the other allotypes (Supplementary Figure S5, one-way ANOVA, 1^st vs. 2^nd p = 0.0429, 1^st vs. 3^rd p = 0.0144, and 1^st vs. 4^th p = 0.0127). Allotype dominance exists in ex vivo ELISPOT within individuals having allotypes showed positive responses in cultured ELISPOT. These results suggest that there is an allotype dominance even in the M. tuberculosis antigens.

The responses of CD4⁺ T cells to a single HLA class II allotypes have been assessed using a mixture of four M. tuberculosis antigens. We further defined which of the four M. tuberculosis antigens mainly used for the positive response by an allotype. (Figure 5A). In ex vivo ELISPOT, all HLA class II allotypes except 2 allotypes showed less than 100 SFCs per 5×10⁵, which was not suitable for subsequent analysis (Supplementary Figure S6). Even in the three donors who showed positive responses by two allotypes, the each allotype had responses predominantly to a single antigen (Figure 5B). The positive responses by two allotypes targeted to different proteins of M. tuberculosis. In HD02, DRB1*01:01 and DPA1*01:03/DPB1*03:01 showed antigen specificity for TB10.4 and for ESAT-6, respectively. In HD05, DQA1*01:02/DQB1*06:09 and DPA1*02:02/DPB1*05:01 showed antigen specificity for TB10.4 and for ESAT-6. In HD33, DRB1*09:01 and DPA1*01:03/DPB1*04:01 showed antigen specificity for Ag85b and for TB10.4.

The seven donors that showed a positive response by a single allotype showed a response to only one antigen (Figure 5C). The DRB1*15:01 showed an antigen specificity for TB10.4 in HD15, HD07 and HD14. The DRB1*15:01-restricted responses targeted for the same antigen, which is TB10.4. The DRB1*04:03 showed an antigen specificity for CFP-10 in HD04. The DRB1*14:06 showed an antigen specificity for Ag85b in HD23. The DQA1*02:01/DQB1*02:02 showed an antigen specificity for TB10.4 in HD25. The DRB1*04:06 showed an antigen specificity for CFP-10 in HD28. Comparing the immunogenicity of each antigen based on the positive responses by an allotype, the average frequencies of the CD4⁺ T cell responses to each antigen were 135 SFCs/5×10⁴ in TB10.4; 59 SFCs per 5×10⁴ in ESAT-6, 50 SFCs per 5×10⁴ in CFP-10; 26 SFCs per 5×10⁴ cells in Ag85b (Figure 5D). The CD4⁺ T cell responses were significantly higher to TB10.4 than Ag85b (Welch’s t-test. p = 0.0373). These results suggest that HLA class II allotype dominance targeted predominantly to a single M. tuberculosis antigen (Supplementary Figure S7).