Targeting Stress Sensor Kinases in Hepatocellular Carcinoma-Infiltrating Human NK Cells as a Novel Immunotherapeutic Strategy for Liver Cancer

Abstract

Natural killer (NK) cells may become functionally exhausted entering hepatocellular carcinoma (HCC), and this has been associated with tumor progression and poor clinical outcome. Hypoxia, low nutrients, immunosuppressive cells, and soluble mediators characterize the intratumor microenvironment responsible for the metabolic deregulation of infiltrating immune cells such as NK cells. HCC-infiltrating NK cells from patients undergoing liver resection for HCC were sorted, and genome-wide transcriptome profiling was performed. We have identified a marked general upregulation of gene expression profile along with metabolic impairment of glycolysis, OXPHOS, and autophagy as well as functional defects of NK cells. Targeting p38 kinase, a stress-responsive mitogen-activated protein kinase, we could positively modify the metabolic profile of NK cells with functional restoration in terms of TNF-α production and cytotoxicity. We found a metabolic and functional derangement of HCC-infiltrating NK cells that is part of the immune defects associated with tumor progression and recurrence. NK cell exhaustion due to the hostile tumor microenvironment may be restored with p38 inhibitors with a selective mechanism that is specific for tumor-infiltrating-not affecting liver-infiltrating-NK cells. These results may represent the basis for the development of a new immunotherapeutic strategy to integrate and improve the available treatments for HCC.

Keywords: NK-cell; hepatocellular carcinoma; immunometabolism; immunotherapy; oncoimmunology.

Figure 1

Gene expression pattern of infiltrating NK cells. (A) Hierarchical clustering representation of the 181 genes identified as differentially expressed in tumor-infiltrating NK cells (TINK, n = 11), surrounding liver-infiltrating NK cells (LINK, n = 11), and normal liver-infiltrating NK cells (NLINK, n = 7) by ANOVA with Benjamini–Hochberg correction (p ≤ 0.05). Data were median-normalized before clustering; upregulated and downregulated genes are shown in red and green, respectively. (B) Principal component analysis of ANOVA-filtered data. (C) Enriched gene sets in TINK and LINK identified by Gene Set Enrichment Analysis (GSEA; MSigDB, C2 canonical pathways). NES, normalized enrichment score; FDR, false discovery rate. (D) Heat map of differentially expressed genes derived from GSEA in TINK and LINK related to mitochondrial and glycolytic function, cell cycle, and DNA damage/DNA repair. The upregulated genes are presented in red and the downregulated genes in green.

Figure 2

Metabolism assessment of infiltrating NK cells. (A) Mitochondrial membrane potential measured by the potentiometric probe JC-1. After anti-CD3 and anti-CD56 staining, JC-1 was added to NK cells from the three compartments: NLINK (n = 7), LINK (n= 12), and TINK (n = 12). Afterwards, the cells were stained with 7-AAD for viability and then analyzed on a flow cytometer. Depolarized NK cells (upper left panel) were quantified by the percentage of FL1high/FL2low cells (JC-1 staining) detected in the different samples. Median fluorescent intensity (MFI) of JC-1 in the FITC channel were analyzed in all study groups in the NK cells subsets (TOTAL, CD56^DIM, and CD56^BRIGHT). In the lower panels, representative examples of the two analyses. (B) Glucose uptake assay was performed on LINK (n = 12), NLINK (n = 7), and TINK (n = 12). The cells were stained with the glucose analog 2-NBDG. The frequency of 2-NBDG-positive NK cells was evaluated (upper left panel) as well as the MFI of the probe (upper right panel). Representative dot plots and histograms show the 2-NBDG uptake in NK cells. Statistical analysis was performed by Wilcoxon matched pairs test (LINK vs. TINK) and Mann Whitney test (NLINK vs. TINK and NLINK vs. LINK). Horizontal lines represent median values.

Figure 3

Phosphorylation status of p38 protein in NK cells infiltrating hepatocellular carcinoma (TINK), HCC-surrounding liver (LINK), and normal liver tissue (NLINK). (A) On the left, median fluorescence intensity (MFI) of phospho-p38 protein in total, CD56^DIM, and CD56^BRIGHT NK-cell subpopulations from TINK (n = 12), LINK (n = 12), and NLINK (n = 7). Right panel: detail of phospho-p38 MFI values of total NK cells from different study groups. Statistical analysis was performed by Wilcoxon matched pairs test (LINK vs. TINK) and Mann-Whitney test (NLINK vs. TINK and NLINK vs. LINK). (B) To show the segregation of NK subsets, we generated a two-dimensional map of NK cells from paired TINK and LINK from all the experimental samples. tSNE was applied to flow cytometry data (single-cell expression values). (C) tSNE colored by the expression intensity of phospho-p38 protein in TINK and LINK samples. Red indicates upregulation, while blue indicates downregulation. Different color shades represent intermediate levels.

Figure 4

Autophagy capacity in tumor-infiltrating NK cells (TINK), non-tumorous liver-infiltrating NK cells (LINK), and normal liver-infiltrating NK cells (NLINK). Cyto-ID median fluorescence intensity (MFI) values in TINK (n = 12), LINK ( n = 12), and NLINK ( n = 7) from untreated (A) and chloroquine-treated (C) samples (total, CD56^DIM, and CD56^BRIGHT NK cells). Examples are shown on the right of each panel. Statistical analysis was performed by Wilcoxon matched pairs test (LINK vs TINK) and Mann–Whitney test (NLINK vs TINK and NLINK vs LINK). Horizontal lines represent median values. Correlation between the Cyto-ID MFI values of untreated (B) and chloroquine-treated (D) and phospho-p38 protein in total NK cells from TINK samples. Statistics by Pearson correlation. *p < 0.05.

Figure 5

Functional analysis of NLINK, LINK, and TINK. (A, B) IFN-γ and TNF-α production in infiltrating NK cells was evaluated with or without phorbol 12-myristate 13-acetate and ionomycin incubation for 4(h) (C) The cytotoxic potential of NK cells was measured by CD107a degranulation assay. (D) Representative dot plots showing IFN-γ, TNF-α, and CD107a expression in tumor- and liver-infiltrating NK cells from NLINK, LINK, and TINK upon PMA/ionomycin stimulation. (A–C) Data are presented as the delta between the frequency of cytokine⁺ or CD107a⁺ NK cells in unstimulated and stimulated samples. Horizontal lines indicate median values. (E) NK cell adhesion to target cells (K562) was evaluated by staining NK cells and K562 with two distinct fluorescent dyes (anti-CD3 and anti-CD56 and 0.5 μM CFSE, respectively). The effector-to-target ratio was 5:1. Cells were incubated at 37°C for 0, 15, and 30 min, followed by fixation. Median fluorescence intensity (MFI) ratio was measured at different time points (left panel). Cell conjugation is presented as MFI values of CD56⁺ cells in CFSE-FITC target cells. Comparison of TINK, NLINK, and LINK at 30 min (right panel). (A–E) NLINK (n = 7), LINK (n = 12), and TINK (n = 12) *p < 0.05.

Figure 6

Restoration assays of hepatocellular carcinoma-infiltrating NK cells. IL-12 + IL-18 O/n stimulation of NK cell stimulation with or without specific p38 inhibitors was followed by flow cytometry determination of IFN-γ (A), TNF-α (B) and CD107a (C) production in study groups and in all NK cell subsets (total, CD56^DIM, and CD56^BRIGHT). Data are presented as the ratio between the frequency of cytokine and CD107a-positive NK cells tested in the presence of inhibitor or untreated cultures (fold change). Horizontal lines represent median values. (D) Bar graph showing IFN-γ production upon PMA/ionomycin stimulation from single TINK samples (upper panel). On the middle and lower panels, corresponding fold change values in the two p38 inhibitor-treated TINK samples. Right panels: correlation between IFN-γ production and response to p38 blockade. (E) Conjugation assay performed on LINK and TINK samples with or without p38 inhibition. (A–C) Statistical analysis by Wilcoxon signed-rank tests.(D) Statistics by Pearson’s correlation. (A–E) LINK (n= 11) and TINK (n= 11). *p < 0.05, **p < 0.01.

Figure 7

Metabolic restoration of HCC (n = 8) and non-tumorous liver-infiltrating NK cells (n = 8). IL-12 + IL-18 overnight stimulation with or without specific p38 inhibitors was followed by flow cytometry determination of JC-1 (A) 2-NBDG (B) and Cyto-ID (with and without chloroquine) (C). Data are presented as the ratio between the metabolic values in inhibitor-treated vs. untreated NK cells from liver and tumor counterparts (fold change). Horizontal lines represent median values. Statistics by Wilcoxon signed-rank tests.