Origin of Hypofunctional CD103+ NK Cells in Cirrhosis-Associated Ascites

Abstract

The occurrence of ascites is a frequent complication associated with the decompensation of liver cirrhosis. While it is known that cirrhosis leads to altered immune responses in the periphery, the immunological milieu of ascites remains poorly understood. In this study, we investigate the role and origin of natural killer (NK) cells in cirrhosis-associated ascites. Using high-dimensional flow cytometry and cytokine analysis, we analyzed matched peripheral blood and ascites fluid alongside liver and duodenum samples to discern tissue-specific differences. Interestingly, a subset of peritoneal NK cells displayed high expression of the tissue-residency receptor CD103. This subset of CD103⁺ ascites NK cells was distinct from blood, liver, and intestinal NK cells and presented with a less activated phenotype coupled with reduced effector capacity. Investigating their origin, we could identify that cytokines present in ascites, here predominantly IL-15 in synergy with IL-21 and TGFβ, can induce CD103 expression and that ascites supernatant further facilitates this process. These results indicate that the ascites in patients with decompensated liver cirrhosis harbor a heterogenous subset of CD103⁺ NK cells that is likely induced by the cytokine milieu.

Keywords: gut leakage; liver cirrhosis; natural killer cells; peritoneal cavity.

FIGURE 1

Study outline and frequency of NK cells in patients with liver cirrhosis. (A) Study outline and workflow. (B) Frequency of total NK cells out of lymphocytes in the peripheral blood of patients with decompensated liver cirrhosis (n = 53) compared with patients with compensated liver cirrhosis (n = 11), and healthy controls (n = 33). (C) Frequency of total NK cells out of lymphocytes in blood compared with matched ascites samples (n = 53), and liver samples from unrelated controls (n = 8) as a reference. (D, E) Frequency of (D) CD56^bright and (E) CD56^dim NK cells out of total NK cells in the ascites compared with matched blood samples of patients with decompensated liver cirrhosis (n = 53), as well as liver samples (n = 8) displayed as a reference. Kruskal Wallis test was performed for multiple comparisons. A paired t‐test was used to identify significance for normally distributed values, whereas the Wilcoxon test was performed on nonparametric datasets. Liver samples were run in a separate experiment and displayed as a reference without performing statistical tests. *p < 0.05; **p < 0.01; ***p < 0.001. Schematic figure created with BioRender.com (Strunz (2025) https://BioRender.com/7qi2sp6).

FIGURE 2

Specific tissue‐residency marker imprint on ascites NK cells. (A) Representative concatenated FACS plots displaying tissue‐residency marker expression on NK cells from blood, ascites, and liver samples. (B) Frequencies of the indicated tissue‐residency markers on total NK cells in matched blood and ascites (n = 10–45) from patients with decompensated liver cirrhosis, as well as unmatched liver (n = 5–8) samples from unaffected liver tissue acquired during liver resection. For comparison of paired blood and ascites samples, the Wilcoxon test was performed on nonparametric datasets. Liver samples are displayed as a reference and were run in a separate experiment. *p < 0.05; **p < 0.01; ***p < 0.001; ****p < 0.0001.

FIGURE 3

High‐dimensional data analysis reveals distinct clusters in blood, ascites, and liver. (A, B) UMAP plots of total NK cells displaying the three different origins (blood, ascites, liver) of total NK cells, analyzed were five matched ascites and blood samples from patients with decompensated liver cirrhosis as well as five unmatched liver samples with 20.000 NK cells exported from each sample. (B) Expression levels of the indicated markers in the UMAP analysis. (C) Phenograph analysis identifying 30 different clusters within NK cells from blood, ascites, and liver. (D, E) Relative abundance of each of the identified Phenograph clusters in blood, ascites, and liver (D), and the respective phenotype (E).

FIGURE 4

CD103⁺ NK cells in the ascites represent a heterogeneous group of NK cells. (A) Expression of indicated markers on CD103⁺ NK cells compared with their CD103^‐ counterpart (n = 10–20) in ascites. (B) UMAP analysis depicting the phenotype of CD103⁺ ascites NK cells generated via concatenated (n = 10) CD103⁺ NK cells derived from the ascites. A paired t‐test was used to identify significance for normally distributed values, whereas the Wilcoxon test was performed on nonparametric datasets. *p < 0.05; **p < 0.01; ***p < 0.001; ****p < 0.0001.

FIGURE 5

CD103⁺ NK cells can be generated through cytokines present in the ascites. (A) UMAP analysis of the total NK cell population stratified for sample origin (blood, ascites, duodenum, liver) and single marker plots identifying the main phenotypic clusters. Matched blood and ascites are derived from patients with decompensated liver cirrhosis and duodenum and liver samples from unmatched controls. UMAP analysis was performed after downsampling and concatenation of files. (B) Representative FACS plots showing CD103 expression on NK cells from different origins (blood, ascites, duodenum, liver). (C) Frequency of CD103⁺ NK cells in matched blood and ascites from patients with decompensated liver cirrhosis (n = 45) compared with blood (n = 23), liver (n = 8), and duodenum (n = 9) from healthy controls. (D) Abundance of the indicated cytokines (pg/mL) in matched blood and ascites from patients with decompensated liver cirrhosis (n = 16–18). (E) Correlation between the relative abundance of CD103⁺ NK cells in matched blood and ascites (n = 45). (F) Overview and (G) representative frequency of CD103⁺ among total NK cells after stimulation of peripheral blood NK cells from healthy controls (n = 6) for 5 days with the indicated cytokine combinations. (H) Frequency of CD103⁺ NK cells after stimulation of peripheral blood NK cells from healthy controls for 5 days in culture media supplemented with indicated cytokines and, when indicated, 50% pooled ascites supernatant. Paired t‐tests were used to identify significances for normally distributed values, whereas the Wilcoxon test was performed on nonparametric datasets. For multiple comparisons of matched nonparametric data, the Friedman test was performed. Spearman tests were used for the correlation between blood and ascites CD103⁺ NK cell frequencies. *p < 0.05; **p < 0.01; ***p < 0.001; ****p < 0.0001.

FIGURE 6

Peritoneal NK cells display altered functionality compared with blood and liver NK cells. (A) Heatmap summarizing the median frequency of responding cells for the indicated functional readouts following stimulation with different reagents for matched blood and ascites (n = 10) from patients with decompensated liver cirrhosis and liver (n = 5) samples from unmatched controls. (B) Multifunctional analysis of CD107a, Granzyme B, IFNγ, TNF, and CD69 responses for the indicated stimulation in matched blood and ascites NK cells; color denotes the number of simultaneously exhibited functions. (C) NK cell response following stimulation with IL‐12 + IL‐18 in matched blood and ascites samples (n = 11). (D, E) The proliferation of purified autologous peripheral blood T cells was assessed after anti‐CD3/anti‐CD28 stimulation for 5 days in the presence or absence of enriched primary ascites NK cells (n = 6). (F) Functional readout following stimulation with IL‐12 + IL‐18 of ascites CD103⁺ NK cells compared with ascites CD103^‐ NK cells (n = 6). Functional readouts of paired samples were compared using paired t‐test or Wilcoxon test when appropriate. For multiple comparisons, the Kruskal–Wallis test was performed. *p < 0.05; **p < 0.01; ***p < 0.001; ****p < 0.0001.