Fig 2: Human CD69+CD16- lung NK cell subsets have unique characteristics. a Representative overlays and b summary of data of expression of NKG2A and CD57 on CD16- human lung CD69-CD49a-CD103- (n = 20 for both), CD69sp (n = 20 for both), CD69+CD49a+CD103- (n = 14 and n = 13, respectively), and CD69+CD49a+CD103+ (n = 17 and n = 15, respectively) NK cells. CD56dimCD16+ NK cells are shown for comparison (n = 20 and n = 19, respectively). Numbers in a indicate %NKG2A+ and %CD57+ NK cells, respectively. c Representative overlays and d summary of mean fluorescence intensity (MFI) of CD56 (n = 18 for CD16-CD69-CD49a-CD103-NKG2A+CD57-, n = 21 for CD69+CD49a-CD103-CD16-, n = 16 for CD69+CD49a+CD103-CD16-, n = 19 for CD69+CD49a+CD103+CD16-, and n = 20 for CD56dimCD16+), CXCR3 (n = 3) and CXCR6 (n = 7) on human lung CD69-CD49a-CD103-NKG2A+CD57-, CD69sp, CD69+CD49a+CD103-, and CD69+CD49a+CD103+ NK cells within the CD16- subset. e Representative overlay and f summary of MFI of perforin expression in CD16-CD69-CD49a-CD103-NKG2A+CD57-, CD69spCD16-, CD69+CD49a+CD103-CD16-, CD69+CD49a+CD103+ CD16-, and CD69-CD49a-CD103-CD16+ NK cells in human lung (n = 5). CD127+CD161+ cells from lung tissue are shown as a comparison (n = 5). g Representative overlays and h summary of MFI of Eomes (left panel) (CD16-CD69-CD49a-CD103-NKG2A+CD57- (n = 5, other subsets n = 7) and T-bet (right panel) (CD16-CD69-CD49a-CD103-NKG2A+CD57- (n = 4, other subsets n = 6) expression in CD16- human lung CD69-CD49a-CD103-NKG2A+CD57-, CD69sp, CD69+CD49a+CD103-, and CD69+CD49a+CD103+ NK cells. Expression in CD16+CD69-CD49a-CD103- NK cells and non-NK cell ILCs (CD56-CD16-NKG2A-KIR-CD161+) are shown as a reference for Eomes (n = 4), and T-bet (n = 6), respectively. i Principal component analysis based on 1,000 most variable genes across subsets. j Euler diagram based on the 230 differentially expressed genes (DEGs, p < 0.05, log2 fold change > 1) between CD69+CD49a+CD103+NKG2A+CD16-, CD69+CD49a+CD103-NKG2A+CD16-, CD69spNKG2A+CD16- and CD69-NKG2A+CD16- NK cells. DEGs among CD69+CD49a+CD103+NKG2A+CD16- and CD69+CD49a+CD103-NKG2A+CD16- were pooled in the analysis. k Heatmap of 142 DEGs (padj < 0.01, log2 fold change > 1) with genes ordered by hierarchical clustering. Genes involved in regulation of tissue-residency, homing, and/or NK cell function are highlighted. Two replicates per donor and subsets are shown. Significant DEGs identified using deSeq2. Colour scale depicts z-score. b, d, f, h Kruskal-Wallis test, Dunn’s multiple comparisons test, *p < 0.05, **p < 0.01, ***p < 0.001, ****p < 0.0001. Means ± SD are shown. Source data are provided as a Source Data file

Fig 3: CD69+CD16− NK cells in the human lung co-express CD49a and CD103. a Representative dot plots and b summary of data of CD49a and CD103 expression on CD69+CD16− and CD69+CD16+ NK cells (n = 64). Mean ± SD is shown. c Linear regression analysis of CD49a and CD103 co-expression on NK cells in human lung (n = 98). d Representative histogram and e summary of data of CD16 expression on CD69− (n = 20), CD69sp (n = 20), and CD69+CD49a+CD103− (n = 17), CD69+CD49a−CD103+ (n = 15) and CD69+CD49a+CD103+ (n = 19) NK cells in human lung. Mean ± SD is shown. f Summary of data for CD69 expression on human lung CD49a−CD103−CD16− (orange), CD49a+CD103−CD16−, CD49a−CD103+CD16−, and CD49a+CD103+CD16− (all blue) NK cells (n = 37). Mean ± SD is shown. g Frequency of NK cell subsets defined by the expression of CD16, CD69, CD49a, and CD103 out of total NK cells (n = 37). Mean ± SD is shown h Percentages of CD49a+ and i CD103+ cells among CD127+ ILCs (defined as live CD45+CD14−CD19−CD3−CD16−NKG2A−KIR−CD69+CD127+) (n = 19) and CD69+CD8+ T cells in the human lung (n = 23). Mean ± SD is shown. Statistical analysis in b, f RM one-way ANOVA, Tukey’s multiple comparison test, Geisser-Greenhouse’s correction. *p < 0.05, **p < 0.005, ***p < 0.001, ****p < 0.0001; ns, non-significant; e Kruskall-Wallis test, Dunn’s multiple comparisons test. **p < 0.005, ****p < 0.0001; h, i Two-tailed Wilcoxon matched-pairs signed rank test, ****p < 0.0001. Source data are provided as a Source Data file

Fig 4: Human lung trNK cells are functional upon stimulation. a Counts per million (CPM) reads for selected genes in human lung CD69-NKG2A+CD16-(dark grey), CD69spNKG2A+CD16- (orange), CD69+CD49a+CD103-NKG2A+CD16-(light blue), and CD69+CD49a+CD103+NKG2A+CD16-(dark blue) NK cells. CD56dimCD16+CD57-NKG2A+ NK cells (light grey) are shown for comparison. Two biological replicates from two donors are shown. Mean ± SEM is shown. b Representative histograms and c summary of data for degranulation and cytokine production of PMA/ionomycin-stimulated lung NK cells. CD49a- and CD49a+ NK cells were gated, and background responses in unstimulated controls were subtracted (CD107a, n = 4; GM-CSF, n = 6; IFN-?, n = 5; MIP-1ß, n = 6; TNF, n = 5). Box and Whiskers, min to max with central line indicating the median and individual data points shown, the mean is indicated as ‘ + ’. Wilcoxon matched-pairs signed rank test, *p < 0.05. Numbers in b indicate percentage of marker-positive NK cells. d Degranulation and cytokine production by sorted CD16- and CD16+ human lung NK cells after PMA/ionomycin stimulation. CD49a and CD103 were used to identify distinct NK cell subsets (n = 3) after stimulation. e Representative overlays and f summary of data for expression of Ki67, perforin, granzyme B, and CCL5 in sorted CD16+, CD16-CD49a-, and CD16-CD49a+ lung NK cells, respectively, following stimulation with IL-15 for 3 days. (n = 4). Mean ± SD is shown. n.a. = not analyzed. The majority of the sorted CD16-CD49a+ NK cells also co-expressed CD103. Numbers in e indicate percentage of marker-positive NK cells. g SPICE analysis for proliferation and effector molecule expression in sorted CD16+, CD16-CD49a-, and CD16-CD49a+ lung NK cell subsets following stimulation with IL-15 for 3 days (n = 3). Floating bars, min-max. Individual data points are shown. Source data are provided as a Source Data file