Cytokine production by stimulated T cells. (A) The number of cytokine-producing T cells per 1000 total CD45+ cells for each stimulation at 24 h with the background (unstimulated) cytokine-production removed. (B) Cluster analysis of IFN-γ secreting cells with clusters 2, 3, 5, 8, 10, 11, 12, 14, 15, 16, and 24 corresponding to T cells. (C) Clusters significantly affected by stimulation with (D) cluster histograms indicating CD4, CD8, CD45RA, and CD27. (E) Cluster analysis of IL-2 secreting cells with clusters 5, 6, 8, 12, 14, and 15 corresponding to T cells. (F) Clusters significantly affected by stimulation with (G) cluster histograms indicating IFN-γ and TNF-ɑ secretion. (H) Cluster analysis of IL-6 secreting cells with cluster 5 corresponding to T cells. (I) Cluster 5 is significantly affected by stimulation (left) and comparison of PPD stimulation on donors with active TB (ATB), latent TB (LTB) and healthy controls (HC) in cluster 5 (right). (J) Cluster histogram indicating GrzB secretion. (K) Cluster analysis of GM-CSF secreting cells with cluster 2, 8, 12, and 13 corresponding to T cells. (L) Cluster 13 is significantly affected by stimulation (left) and comparison of PIM stimulation on donors with ATB, LTB and HC in cluster 13 (right). (M) Cluster dot plots indicating CD4, CD8, CD45RA, and CD27 expression. (N) Cluster analysis of IL-17A secreting cells with cluster 1, 2, 5, 8, 9, 10, 11, 14, 15, 16, 17, and 18 corresponding to T cells. (O) Clusters significantly affected by stimulation (left) and comparison of PIM and LAM stimulation on donors with active ATB, LTB and HC in clusters 1 and 16, respectively (right) (P) Cluster histograms indicating IFN-γ, TNF-ɑ, and IL-2. Statistical differences between stimulations in (A, I, L) were evaluated by Friedman’s test with Dunnet’s posttest, while comparisons within multiple clusters (C, F, O left panels) were evaluated by a matched-pair two-way ANOVA with Geissner-Greenhouse correction followed by Tukey’s posttest (n=15/stimulation). Groups (ATB/LTB/HC) (I, L, O right panels) were compared using Kruskal-Wallis with Dunn’s posttest (n=5/group) *p<0.05, **p<0.01, ***p<0.001, ****p<0.0001.