Specific CD4+ T cell phenotypes associate with bacterial control in people who 'resist' infection with Mycobacterium tuberculosis

Abstract

A subset of individuals exposed to Mycobacterium tuberculosis (Mtb) that we refer to as 'resisters' (RSTR) show evidence of IFN-γ^- T cell responses to Mtb-specific antigens despite serially negative results on clinical testing. Here we found that Mtb-specific T cells in RSTR were clonally expanded, confirming the priming of adaptive immune responses following Mtb exposure. RSTR CD4⁺ T cells showed enrichment of T_H17 and regulatory T cell-like functional programs compared to Mtb-specific T cells from individuals with latent Mtb infection. Using public datasets, we showed that these T_H17 cell-like functional programs were associated with lack of progression to active tuberculosis among South African adolescents with latent Mtb infection and with bacterial control in nonhuman primates. Our findings suggested that RSTR may successfully control Mtb following exposure and immune priming and established a set of T cell biomarkers to facilitate further study of this clinical phenotype.

Fig. 1. IFN-γ-independent T cell responses are not detected in low-exposure controls.

a, Representative flow cytometry showing the expression of IFN-γ and CD154 in CD4⁺ T cells from TST⁻IGRA⁻ or LTBI in the low-exposure cohort in response to ESAT6/CFP10 stimulation. b, COMPASS-generated probability heat map of CD4⁺ T cell subsets including IFN-γ⁻IL-2⁺CD154⁺, IFN-γ⁺IL-2⁻CD154⁺, IFN-γ⁺IL-2⁺CD154⁻ and IFN-γ⁺IL-2⁺CD154⁺ cells from TST⁻IGRA⁻ or LTBI in the low-exposure cohort in response to ESAT6/CFP10. The depth of purple shading correlates to the probability of a participant-specific response above background for a given cell subset. White, no function; black/gray, presence of a function. IFN-γ⁺ cell subsets are noted in gray. c, The frequencies of COMPASS-identified CD4⁺ T cell subsets are shown, background corrected by subtracting the frequency of each subset after DMSO stimulation from the frequency after ESAT6/CFP10 stimulation as in b. d, The frequencies of polyfunctional CD4⁺ T cells that expressed two or more cytokines in response to ESAT6/CFP10 in CD4⁺ T cells as in c. e, Polyfunctionality score of CD4⁺ T cells specific to ESAT6/CFP10. In c−e, the statistical significance was calculated using the Wilcoxon rank-sum test. Two-sided P values are shown, while in c the reported P values were adjusted for multiple hypothesis testing using the Bonferroni method.

Fig. 2. Mtb-specific T cells exhibit unique phenotypes and are clonally expanded in RSTR.

a, The gating strategy is shown for the sorting of ESAT6/CFP10-specific T cells from RSTR and LTBI in the discovery household contact cohort, based first on the expression of the activation marker CD69 and then CD154 or CD137. b, A heat map showing the median marker expression of ESAT6/CFP10-specific CD4⁺ T cell subsets from three RSTR and four LTBI participants in the discovery household contact cohort. Clustering was performed on flow cytometry mean fluorescence intensities and binarized read counts of profiled genes. c, Dimensionality reduction (t-SNE) projection of ESAT6/CFP10-specific CD4⁺ T cell subsets, as in b. The arrows highlight cluster 6 and cluster 11, which were preferentially detected in RSTR or LTBI participants, respectively. d, Distribution of clonal expansion based on TCRβ chain from single-cell targeted transcriptomics in ESAT6/CFP10-specific T cells, as in b. Each dot represents a clone as defined by the TCRβ chain CDR3 sequence. The size of the dot is proportional to the frequency, which is also depicted as log₂(counts) on the y axis. e, A box plot showing the median and interquartile range of frequency of TCRβ clonal expansion in ESAT6/CFP10-specific CD4⁺ T cells with whiskers representing minima and maxima. No statistical test was performed due to the small sample sizes. f, A histogram indicating the proportion of clonally expanded cells in ESAT6/CFP10-specific CD4⁺ T cell clusters (clusters 1–19), as in b, which were detected more than once in participants among the household contact cohort.

Fig. 3. Mtb-specific T cells exhibit a T_SCM cell-like phenotype in RSTR and LTBI.

a, A volcano plot depicting DEGs of clonally expanded activated T cells between RSTR (cell number, 231) and LTBI (cell number, 293) from the discovery household contact cohort in response to ESAT6/CFP10 based on SELECT-seq. Red, genes with |log₂(FC)| >0.5 and P value <0.05. b, GO Biological Process terms related to metabolic and basic activities based on upregulated DEGs from RSTR, as in a. False discovery rate (FDR) was calculated by a modified Fisher’s exact test with FDR correction. c, A heat map displaying the mean expression of genes involved in migration, adhesion or cytokine production in each RSTR and LTBI participant in the discovery household contact cohort. The mean expression level was calculated as the mean of the scaled log-normalized gene counts. d, Violin plots depicting the expression of stem memory like T cell genes (CCR7, FOXP1 and TCF7) in RSTR and LTBI, as in c. Statistical significance was calculated using two-sided Wilcoxon rank-sum tests with the Bonferroni method. *Adjusted P value <0.05, **adjusted P value <0.005, ***adjusted P value <0.001. e, Gating strategy for naive-like CD45RA⁺CCR7⁺ T cells in ESAT6/CFP10-specific CD4⁺ T cells from RSTR (n = 17) and LTBI participants (n = 20) in the validation household contact cohort. f, A box plot showing the median and interquartile range of frequencies of naive-like CD45RA⁺CCR7⁺CD4⁺ T cells in ESAT6/CFP10-specific CD4⁺ T cells, as in e, with whiskers representing minima and maxima. Statistical significance was determined by the two-sided Wilcoxon rank-sum test, and the unadjusted P value is shown.

Fig. 4. Mtb-specific T cells exhibit distinct activation phenotypes in RSTR compared to LTBI.

a, GO terms related to T cell activation in clonally expanded ESAT6/CFP10-specific CD4⁺ T cells among RSTR (cell number, 231) compared to LTBI (cell number, 293) from the discovery household contact based on SELECT-seq. b, Expressions of activation and costimulation genes in RSTR and LTBI, as in a. Statistical significance was calculated by two-sided Wilcoxon rank-sum tests with the Bonferroni method. **Adjusted P value <0.005, ***adjusted P value <0.001. c, Interaction network of DEGs related to T cell activation upregulated in RSTR compared to LTBI, as in a. d, Mean expression of T_reg cell-associated genes within each RSTR and LTBI, as in a. e, A box plot showing the median and interquartile range of frequencies of FOXP3⁺CD25⁺ T_reg cells in ESAT6^/CFP10-specific CD4⁺ T cells from the same RSTR (n = 3) and LTBI (n = 4) participants as in a using index sorting and targeted transcriptomics. The whiskers represent minima and maxima. No statistical test was performed due to small sample sizes. f, Gating strategy for CD25⁺ T cells and Foxp3⁺CD25⁺ T_reg cells in ESAT6/CFP10-specific CD4⁺ T cells from RSTR (n = 17) and LTBI participants (n = 20) in the validation household contact cohort. g, Frequencies of CD25⁺CD4⁺ T cells in ESAT6/CFP10-specific CD4⁺ T cells are shown as in f. h, Frequencies of Foxp3⁺CD25⁺ T_reg and IL-10⁺ CD4⁺ T cells in ESAT6^/CFP10-specific CD4⁺ T cells are shown as in f. i, MFIs of IL-10 in supernatants after 48 h ESAT6/CFP10 or 12 h Mtb lysate stimulation based on multiplex cytokine analysis. Significance was determined by two-sided Student’s t-test. Significance in g and h was determined by two-sided Wilcoxon rank-sum tests.

Fig. 5. Enrichment of Mtb-specific T_H17-like cells among RSTR.

a, Box plots showing the median and interquartile range of frequencies of RORC⁺TBX21⁺, RORC⁺TBX21⁻, RORC⁻TBX21⁺ and RORC⁻TBX21⁻ cells in ESAT6/CFP10-specific CD4⁺ T cells in RSTR (n = 3) and LTBI (n = 4) in the discovery household contact cohort using targeted transcriptomics. The whiskers represent minima and maxima. Statistical testing was not performed due to small sample sizes. b, SELECT-seq showing the mean expression of genes associated with T_H1 or T_H17 phenotypes from GO:0072539 in the MSigDB database in clonally expanded ESAT6/CFP10-specific CD4⁺ T cells within both RSTR and LTBI in the household contact cohort as in a. c, Expressions of T_H1 or T_H17 cell-associated genes in RSTR and LTBI in the household contact cohort, as shown in a. Significance was determined using two-sided Wilcoxon rank-sum tests with the Bonferroni method. ***Adjusted P value <0.001. d, Representative flow cytometry showing the expression of RORγt and T-bet in ESAT6/CFP10-specific CD4⁺ T cells from RSTR (n = 17) and LTBI participants (n = 20) in the validation household contact cohort. e, Frequencies of RORγ⁺T-bet⁺, RORγt⁺T-bet⁻, RORγt⁻T-bet⁺ and RORγt⁻T-bet⁻cells in ESAT6/CFP10-specific CD4⁺ T cells or Mtb lysate-specific CD4⁺ T cells from RSTRs and LTBI, as in d. f, Frequencies of IL-17A⁺ cells in ESAT6/CFP10-specific CD4⁺ T cells or Mtb lysate-specific CD4⁺ T cells from RSTRs and LTBI as in d. g, MFIs of IL-17A or IL-23 in supernatants of PBMC are shown after 24 h or 6 h stimulation, respectively, with Mtb lysate. Significance was determined by two-sided Student’s t-test. Significance in e and f was determined by two-sided Wilcoxon rank-sum tests.

Fig. 6. RSTR CD4⁺ T cell phenotypes associate with lack of progression to active TB in the ACS cohort and bacterial control in NHPs.

a,b, Box plots showing the median and interquartile range of RSTR-associated T_N_T_SCM cell module (a) or T cell activation module scores (b) from whole-blood bulk transcriptomics at serial time points before TB diagnosis among nonprogressors (n = 101) and progressors (n = 43) in the ACS (GSE79362) (ref. ). The whiskers represent minima and maxima. Scores were computed as geometric mean of gene expressions. Significance was determined by two-sided Student’s t-test. c, Frequencies of Mtb-specific CD4⁺ T cells expressing RORC and/or TBX21 or FOXP3 among TB nonprogressors (n = 35) and progressors (n = 35) in ACS using single-cell targeted transcriptomics. Significance was determined by two-sided Wilcoxon rank-sum tests. d,e, Heat maps showing the mean expression of the top 15 enriched genes in a stem-like T cell subset (d) and a T1–T17 cell subset (e) identified in granulomas from cynomolgus macaques in clonally expanded ESAT6/CFP10-specific CD4⁺ T cells in RSTR (cell number, 231) and LTBI (cell number, 293) in the household contact cohort. f, Stem-like T cell- and T1–T17 T cell-associated gene module scores as in d. g, RSTR-associated T_H17 cell gene module score from whole-blood bulk transcriptomics in rhesus macaques at day 0 (W0), day 2 (D2), week 2 (W2), week 4 (W4) and week 12 (W12), postintravenous vaccination with BCG. h, Flow cytometry showing cell count and frequency of IFNγ⁻IL-2⁻IL-17⁺TNF⁻ T_H17 CD4⁺ cells in the bronchoalveolar lavage of BCG-vaccinated rhesus macaques as in g. For c and f–h, significance was calculated using the two-sided Wilcoxon rank-sum tests.

Extended Data Fig. 1. Enrollment of low exposure cohort, study schema and gating strategy.

(a) Screening and enrollment of low exposure cohort participants for health assessment and blood draw. Enrollment took place in Uganda from July 2017 to March 2018. All healthy non-pregnant participants were eligible. TB, tuberculosis; HIV, human immunodeficiency virus; TST, tuberculin skin test; IGRA, IFN-γ release assay. (b) Schematic of the low exposure cohort, containing 19 LTBI and 17 TST⁻IGRA⁻ participants, and the household contact cohort, containing 45 RSTR and 45 LTBI participants. T cell responses to ESAT6/CFP10 or whole Mtb lysate were assessed using intracellular cytokine staining, targeted transcriptomics and SELECT-seq, and validated using flow cytometry and multiplex cytokine analysis on independent samples. (c) Flow cytometry gating strategy for CD4⁺ T cell subsets with cytokine and activation marker expressions from PBMCs in the low exposure control cohort stimulated with ESAT6/CFP10 peptide pool for 6 hours.

Extended Data Fig. 2. The composition of Mtb-specific T cell subsets.

(a) Gating strategy of ESAT6/CFP10-specific CD4⁺ T cells among PBMCs from the household contact cohort using index sort. T cells were identified as CD3⁺CD14⁻/CD19⁻. T cells positive for the TCRɑβ marker were gated on and separated into CD4⁺ or CD8⁺ subsets. Activated CD3⁺TCRαβ⁺ T cells were gated on using CD69, CD137, and CD154 markers. (b) Box plots showing the median and interquartile range of frequencies of ESAT6/CFP10-specific CD4⁺ T cell clusters defined in Fig. 2b among RSTR (n = 3) and LTBI (n = 4) donors in the household contact cohort using index sort and targeted transcriptomics. Whiskers represent minima and maxima. Statistical testing was not performed due to the small sample sizes. (c) Dimensionality reduction (t-SNE) projection of activated CD4⁺ T cell subsets sorted from PBMCs among the household contact cohort after stimulation with Mtb lysate for 12 hours.

Extended Data Fig. 3. Quality control filtering and key T cell transcriptional factor gene visualization of the SELECT-Seq data.

(a) Filtered cell number by quality control parameters, including the number of unique expressed genes, total gene counts, and percentage of mitochondrial (MT) gene expression, from the whole transcriptomic data of clonally expanded ESAT6/CFP10-specific CD4⁺ T cells in the household contact cohort. (b) Dimensionality reduction (UMAP) projection of these clonally expanded CD4⁺ T cells showing group, subject, sex, sequencing batches, and subset clusters. (c) Expression of transcriptional factor genes essential for T cell polarization on these clonally expanded CD4⁺ T cells. The expression levels were calculated as the log-normalized gene counts from the whole transcriptomic data.

Extended Data Fig. 4. Gating strategy for T_H panel and naïve-like Mtb lysate-specific cell frequency in the validation cohort.

(a) Flow cytometry gating strategy for cells expressing memory markers, transcriptional factors and intracellular cytokines associated with T_H1 and T_H17 among the validation household contact cohort in response to stimulation with ESAT6/CFP10. A time gate was applied to exclude events affected by sample acquisition aberrations, followed by T cell identification by lymphocyte size, CD3, CD14, and CD19 markers. A singlet gate was then applied, followed by the identification of viable cells and a second lymphocyte gate. A secondary gate was applied to identify IL-17A⁺ cells due to spillover spreading. The same gating strategy was applied to samples stimulated with whole Mtb lysate (not shown). (b) Flow cytometry showing the median and interquartile range of frequencies of CD45RA⁺CCR7⁺ naïve-like cells in Mtb lysate-specific CD4⁺ T cells among validation household contact cohort RSTR (n = 17) and LTBI (n = 20). Whiskers represent minima and maxima. Statistical significance was determined by the two-sided Wilcoxon rank-sum test.

Extended Data Fig. 5. RSTR T_SCM cells proliferate similarly to LTBI T_SCM cells in response to Mtb antigen.

(a) Schematic of T_SCM study where RSTR (n = 12) and LTBI (n = 12) PBMC from the household contact cohort were thawed, rested for 2 hours, stained with 5 µM CFSE, and stimulated with ESAT6/CFP10 or Mtb lysate overnight. (b) Gating strategy of activated naive-like CD69⁺CD154⁺CCR7⁺CD45RA⁺CD45RO⁻lymphocytes. (c) Representative flow cytometry showing sorted T_SCM cells from RSTR and LTBI expressing the proliferation marker CFSE after 7-day culture. Representive CD45RA and CCR7 staining of the CFSE^low population in response to ESAT6/CFP10 and Mtb lysate 7-day post-stimulation (d) Box plots showing the median and interquartile range of frequencies of T_SCM cells (CD95⁺CCR7⁺CD45RA⁺CD45RO⁻) at the time of sorting from PBMC stimulated with Mtb lysate or ESAT6/CFP10. Whiskers represent minima and maxima. (e) Frequencies of CFSE^lowCD95⁺ CD4⁺ T_SCM cells after 7-day culture. (f-g) Frequencies of proliferated cells in response to (f) Mtb lysate and (g) ESAT6/CFP10 stimulation. CM, central memory; EM, effector memory. Significance in d-g was calculated using the two-sided Wilcoxon rank-sum tests.

Extended Data Fig. 6. Gating strategy of T_reg and its functional profiles.

(a) Flow cytometry gating strategy for cells expressing T_reg cell- and activation- associated markers, transcriptional factors and intracellular cytokines among the validation household contact cohort in response to stimulation with ESAT6/CFP10. A time gate was applied to exclude events affected by sample acquisition aberrations, followed by T cell identification by lymphocyte size, CD3, CD14, and CD19 markers. A singlet gate was then applied, followed by the identification of viable cells and a second lymphocyte gate. The same gating strategy was applied to samples stimulated with whole Mtb lysate (not shown). (b) ELISA showing the median and interquartile range of background-corrected absorbance level of TGF-β in conditioned supernatants from PBMCs in response to ESAT6/CFP10 stimulation among RSTR (n = 18) and LTBI (n = 20) in the validation household contact cohort. Whiskers represent minima and maxima. (c) Ratio of anti-inflammatory FoxP3⁺CD25⁺ T_reg cell frequency to pro-inflammatory CD4 T cell frequency, including RORγt^-T-bet⁺ T_H1, RORγt⁺T-bet⁻T_H17, and RORγt⁺T-bet⁺ T_H1* cells, among validation donors in response to ESAT6/CFP10 (left) or Mtb lysate (right) from flow cytometry data. Significance in b,c was determined by the two-sided Wilcoxon rank-sum tests.

Extended Data Fig. 7. Cytokine secretion from LTBI and RSTR PBMC following ESAT6/CFP10 stimulation.

(a) Heatmap showing paired t-statistic of cytokine expression levels between ESAT6/CFP10 and DMSO stimulations from supernatants among RSTR (n = 18) and LTBI (n = 20) donors at 6, 12, 24, and 48 hours after stimulation. (b) Box plots showing the median and interquartile range of cytokine mean fluorescence intensities at the timepoint that exhibited the highest signal identified as in a. Whiskers represent minima and maxima. Significance was calculated by two-sided student’s t-test. n.s. (not significant), p-value > 0.05; *, p-value < 0.05; **, p-value < 0.01; ***, p-value < 0.001.

Extended Data Fig. 8. Cytokine secretion from LTBI and RSTR PBMC following Mtb lysate stimulation.

(a) Heatmap showing paired t-statistic of cytokine expression levels between Mtb lysate and DMSO stimulations from supernatants among RSTR (n = 18) and LTBI (n = 20) donors at 6, 12, 24, and 48 hours after stimulation. (b) Box plots showing the median and interquartile range of cytokine mean fluorescence intensities at the timepoint that exhibited the highest signal identified as in a. Whiskers represent minima and maxima. Significance was calculated by two-sided student’s t-test. n.s. (not significant), p-value > 0.05; *, p-value < 0.05; **, p-value < 0.01; ***, p-value < 0.001.

Extended Data Fig. 9. Representative staining of CXCR3⁺CCR6⁺ TH1 cells and IFN-γ⁺ CD4⁺ T cell frequencies.

(a) Representative staining for CXCR3⁺CCR6⁺ T_H1 cells in ESAT6/CFP10-specific CD4⁺ T cells from a participant in the validation household contact cohort. (b) Median and interquartile range of frequencies of CXCR3⁺CCR6⁺ T_H1 and IFN-γ⁺ CD4⁺ T cells in ESAT6/CFP10-specific CD4⁺ T cells or Mtb lysate-specific CD4⁺ T cells among RSTR (n = 17) and LTBI (n = 20). Whiskers represent minima and maxima. Significance was determined by two-sided Wilcoxon rank-sum tests.