Human Hepatic CD56bright NK Cells Display a Tissue-Resident Transcriptional Profile and Enhanced Ability to Kill Allogenic CD8+ T Cells

Abstract

Liver-resident CD56^brightCD16^- natural killer (NK) cells are enriched in the human liver and are phenotypically distinct from their blood counterparts. Although these cells are capable of rapid cytotoxic effector activity, their functional role remains unclear. We hypothesise that they may contribute to immune tolerance in the liver during transplantation. RNA sequencing was carried out on FACS sorted NK cell subpopulations from liver perfusates (n=5) and healthy blood controls (n=5). Liver-resident CD56^brightCD16^+/- NK cells upregulate genes associated with tissue residency. They also upregulate expression of CD160 and LY9, both of which encode immune receptors capable of activating NK cells. Co-expression of CD160 and Ly9 on liver-resident NK cells was validated using flow cytometry. Hepatic NK cell cytotoxicity against allogenic T cells was tested using an in vitro co-culture system of liver perfusate-derived NK cells and blood T cells (n=10-13). In co-culture experiments, hepatic NK cells but not blood NK cells induced significant allogenic T cell death (p=0.0306). Allogenic CD8⁺ T cells were more susceptible to hepatic NK cytotoxicity than CD4⁺ T cells (p<0.0001). Stimulation of hepatic CD56^bright NK cells with an anti-CD160 agonist mAb enhanced this cytotoxic response (p=0.0382). Our results highlight a role for donor liver NK cells in regulating allogenic CD8⁺ T cell activation, which may be important in controlling recipient CD8⁺ T cell-mediated rejection post liver-transplant.

Keywords: CD8+ T cell; NK cell; liver; resident; tolerance; transplant.

Figure 1

Human hepatic CD56^bright NK cell populations express a distinct transcriptional profile association with tissue residency. RNA-sequencing analysis of flow cytometry sorted NK cell subpopulations from healthy human liver perfusate (n = 5) and healthy human peripheral blood (n = 5). (A), PCA analysis of 405 differentially expressed genes (ANOVA q-value < 0.05) and (B), K-means clustering of PCA. (C), Heatmap of average gene expression of selected NK cell-related genes. (D), Gene-set enrichment analysis comparing 769 genes with >2 fold change between liver and blood CD56^brightCD16^- NK cells, against up-regulated genes in tissue-resident lymphocytes from Kumar et al., 2017 (40). (E), Gene-set enrichment analysis comparing 769 genes with >2 fold change between liver and blood CD56^brightCD16^- NK cells, against up-regulated genes in tissue-resident lymphocytes from Mackay et al., 2016 (41). (F), Representative histograms of protein validation of differentially regulated genes via flow cytometry on CD56^brightCD16^- NK cells (n = 5-11).

Figure 2

Co-expression of CD69, CXCR6, TIGIT, CD160, Ly9 and NKG2D defines a liver-resident CD56^bright NK cell subset. Flow cytometry analysis liver-residency associated markers on NK cell subsets in liver perfusate and tissue samples (n = 5-25). (A), Frequency of NK cell subsets in liver perfusate and tissue samples (n = 5-25, matched samples indicated with a line). (B, D), Percent frequency of CD69⁺, CXCR6⁺, TIGIT⁺, CD160⁺, Ly9⁺, NKG2D⁺, and CD49a⁺ CD56^brightCD16^- (B), CD56^brightCD16⁺ (C), and CD56^dimCD16^- (D) NK cells in perfusate and tissue samples. (E), tSNE analysis of NK cells from 5 matched liver perfusate and tissue samples visualising CD56^bright (CD16^-), CD56^brightCD16⁺, and CD56^dim (CD16⁺) NK cell subsets’ distinct cell surface receptor expression patterns. (F, G), SPICE analysis visualising co-expression of cell surface receptors on CD56^brightCD16^- NK cells in matched liver perfusate and tissue samples represented as a scatter plot (F) and a pie chart (G). Data analysed using an unpaired Mann-Whitney test (B–D) or a paired t test (F). * p< 0.05, **p < 0.01, ***p < 0.001.

Figure 3

Hepatic NK cells but not peripheral blood NK cells dose-dependently kill activated allogenic peripheral blood CD8⁺ T cells. PBMCs cultured alone or with magnetically sorted hepatic NK cells (H-NK) or peripheral blood NK cells (P-NK) cells for 24 hr with or without anti-CD3/28 activation stimulation (n = 6-13). (A), Schematic of experimental design. (B), Representative flow plots of % dead CD56^-CD16^-CD19^- cells at each indicated ratio for both H-NK and P-NK cocultures. (C, D), Percent dead CD56^-CD16^-CD19^- cells (C) and CD56^-CD16^-CD19^-CD8⁺ cells (D) after 24 hr co-culture with H-NK (clear circle) or P-NK (triangle). Grey dashed line represents PBMCs cultured alone. (E–G), Percent dead CD4⁺ T cells (E) and CD8⁺ T cells (F) post-PBMC culture alone or with H-NK cells and/or anti-CD3/28 activation (n = 12-13) (G), Data from (E, F) directly comparing CD4⁺ T cell and CD8⁺ T cell death post-coculture with or without anti-CD3/CD28 activation. Data analysed using a 2-Way ANOVA (C, D), a one-way ANOVA (E, F) or a paired t test (G). *p<0.05, ** p<0.01, and ***p<0.001.

Figure 4

CD160 engagement on hepatic NK cells enhances killing of allogenic CD8⁺ T cells. Activated CD3⁺ T cells co-cultured with H-NK cells with or without a mAb against CD160 for 24 hr (n = 13). (A), Schematic of experimental design. (B), Representative flow plots of the percent dead total T cells in each treatment group. ‘Ctrl’ indicates cocultures of CD3⁺ T cells with H-NK cells without the addition of any antibody. ‘aIgG1’ indicates cocultures of CD3⁺ T cells with H-NK cells with the addition of an IgG1 antibody. ‘aCD160’ indicates cocultures of CD3⁺ T cells with H-NK cells with the addition of an anti-CD160 mAb. (C–E), Percent dead total T cells (C), CD8⁺ T cells (D) and CD4⁺ T cells (E) following 24 hr co-culture of CD3⁺ T cells with hepatic NK cells. (F), Representative flow plots of HLA-A/B/C and HVEM MFI of CD8+ T cells at 0 hr and for each treatment at 24 hr. (G, H), MFI values for HLA-A/B/C (G) and HVEM (H) on CD8⁺ T cells and CD4⁺ T cells at 0 hr and for each treatment group at 24 hr. Data analysed using a Freidman test with Dunn’s multiple comparison (C–E) or a repeated measures one-way ANOVA (G–J). *p < 0.05, ** p < 0.01, and ***p < 0.001.

Figure 5

Agonistic anti-CD160 mAb acts on hepatic CD56^bright NK cells to enhance their killing of allogenic T cells. Activated CD3⁺ T cells co-cultured with H-NK cells or FACS-sorted hepatic CD56^brightCD16^+/- or CD56^dimCD16⁺ NK cells with or without a mAb against CD160 for 24 hr (n = 6-10). (A), Schematic of experimental design (B), Representative flow plot of CD160 expression on NK and T cell populations at 0 hr prior to coculture. ‘Ctrl’ indicates cocultures of CD3⁺ T cells with the specified H-NK cell subset without the addition of any antibody. ‘aIgG1’ indicates cocultures of CD3⁺ T cells with the specified H-NK cell subset with the addition of an IgG1 antibody. ‘aCD160’ indicates cocultures of CD3⁺ T cells with the specified H-NK cell subset with the addition of an anti-CD160 mAb. (C, D), Percent CD160 expression on NK cell (C) and T cell (D) subsets at 0 hr. (E–G), Percent dead total T cells (E), CD8⁺ T cells (F) and CD4⁺ T cells (G) following 24 hr co-culture of CD3⁺ T cells and CD56^brightCD16^+/- hepatic NK cells. (H–J), Percent dead CD56^-CD19^- cells (H), CD8⁺ T cells (I) and CD4⁺ T cells (J) following 24 hr co-culture of CD3⁺ T cells with hepatic CD56^dimCD16⁺ NK cells. Data analysed using a repeated measures One Way ANOVA test (C–J). * p< 0.05, **p < 0.01, and ***p < 0.001.