CD8+ tissue-resident memory T cells promote liver fibrosis resolution by inducing apoptosis of hepatic stellate cells

Abstract

Non-alcoholic steatohepatitis (NASH) is a leading cause of chronic liver disease that can progress to liver fibrosis. Recent clinical advance suggests a reversibility of liver fibrosis, but the cellular and molecular mechanisms underlying NASH resolution remain unclarified. Here, using a murine diet-induced NASH and the subsequent resolution model, we demonstrate direct roles of CD8⁺ tissue-resident memory CD8⁺ T (CD8⁺ Trm) cells in resolving liver fibrosis. Single-cell transcriptome analysis and FACS analysis revealed CD69⁺CD103^-CD8⁺ Trm cell enrichment in NASH resolution livers. The reduction of liver CD8⁺ Trm cells, maintained by tissue IL-15, significantly delayed fibrosis resolution, while adoptive transfer of these cells protected mice from fibrosis progression. During resolution, CD8⁺ Trm cells attracted hepatic stellate cells (HSCs) in a CCR5-dependent manner, and predisposed activated HSCs to FasL-Fas-mediated apoptosis. Histological assessment of patients with NASH revealed CD69⁺CD8⁺ Trm abundance in fibrotic areas, further supporting their roles in humans. These results highlight the undefined role of liver CD8⁺ Trm in fibrosis resolution.

Fig. 1. Abundance of CD8⁺ T cells in NASH resolution livers.

a Study design: male C57BL/6 mice were fed a ND or HFHC diet for 24 weeks. After 24 weeks, HFHC diet-fed mice were either continued on a HFHC diet (HFHC group) or switched to ND (RES group) for 2, 5, or 8 weeks. b Body weight change following the diet switch (n = 7 mice for each group). c Representative photomicrographs of Masson trichrome staining of liver sections. Scale bars: 200 µm. d Col1a1 mRNA levels in the whole liver. e Hydroxyproline levels in the liver (d, e: n = 3–6 mice for ND group, n = 6 mice for HFHC group, and n = 8 mice for RES group). f Frequency (left) and absolute numbers (right) of the indicated immune cells in CD45⁺ liver MNCs (n = 3 mice for ND, and HFHC groups, and n = 4 mice for RES groups). g Frequency of CD8⁺ T cells in CD45⁺ liver MNCs at week 8 following the diet switch (n = 6 mice for ND, and HFHC groups, and n = 8 mice for RES group). Data are presented as mean ± SEM. One-way ANOVA with Tukey’s multiple comparisons post-hoc test was applied. Source data are provided as a Source data file.

Fig. 2. CD8⁺ T cells play a direct role in NASH resolution.

a Study design: resolution induced mice were treated with anti-CD8α antibody (RES + CD8T depletion group) or the isotype control (RES group) once every 3 days for 3 weeks starting from week 2 after the diet switch, and compared to ND-fed group (n = 3 mice for ND group, n = 6 mice for RES, and RES + CD8T depletion groups). b Representative CD4/CD8α staining of CD45⁺TCRβ⁺ NK1.1⁻ gated liver MNCs. c Body weight. d Wet liver weight. e Serum ALT levels. f Representative photomicrographs of Masson trichrome staining of liver sections. Scale bars: 200 µm. g Representative photomicrographs (left) and positive area (right) of Sirius Red staining of liver sections. Scale bars: 200 µm. h Hydroxyproline levels. i Fibrosis associated genes (Col1a1, Col1a2, Acta2, Timp1, Desmin, Spp1) mRNA levels. j (left) Representative fluorescent photomicrographs of liver sections stained with 4′,6-diamidino-2-phenylindole (DAPI: blue) and anti-desmin Ab (red). Scale bars: 200 µm. (right) HSCs count in unit area of liver section. k Number of CD45⁺CD11b⁺CD11c⁻ macrophages in CD45⁺ liver MNCs. Data are presented as mean ± SEM. One-way ANOVA with Tukey’s multiple comparisons post-hoc test was applied. Source data are provided as a Source data file.

Fig. 3. Landscape of CD8⁺ T cells in NASH and NASH resolution livers.

a–d ScRNA-seq analysis of liver CD8 T cells. a UMAP projection of combined liver CD8⁺ T cells (n = 25,851) derived from ND, HFHC, and RES (5 weeks) mice, showing the formation of 17 main clusters represented by different colors. The functional description of each cluster is determined by the gene expression characteristics of each. b Heat map of 17 CD8⁺ T cell clusters with top 10 unique signature genes. c UMAP projection of liver CD8⁺ T cells derived from ND (left), HFHC (middle), and RES (right) mice. d Fractions of five clusters (naïve, Tcm, Tem, Trm, and others) in CD8⁺ T cells. e Dot plot visualizing the expression of representative genes in the Trm clusters (cluster 2, 3, 6–9). The color represents the average expression level, and the circle size represents the proportion of cells expressing each gene. Source data are provided as a Source data file.

Fig. 4. Characteristics of CD8⁺ Trm cells in NASH and NASH resolution livers.

a Representative CD44/CD62L staining of CD45⁺TCRβ⁺NK1.1⁻CD8α⁺ gated liver MNCs (upper) and KLRG1/CD69 staining of CD45⁺TCRβ⁺NK1.1⁻CD8α⁺CD44⁺CD62L⁻ gated (Tem + Trm) liver MNCs (lower) of RES (5 weeks) mice. b Frequency of each CD8⁺ T cell subset in CD45⁺ liver MNCs derived from the indicated groups of mice (n = 6 mice for ND and HFHC groups, and n = 8 mice for RES group). Data are presented as mean ± SEM. One-way ANOVA with Tukey’s multiple comparisons post-hoc test was applied. c Relative abundance of partner or parent cells in the blood and liver MNCs of mice two weeks following the parabiosis surgery between Ly5.2⁺ RES (5 weeks) mice and Ly5.1⁺ RES (5 weeks) mice (n = 6 pairs). Data are presented as mean ± SEM. d–f Bulk RNA-seq analysis of isolated liver naïve CD8⁺ T cells from ND mice (ND naïve; n = 3), CD8⁺ Tem cells from HFHC mice (HFHC Tem; n = 3), CD8⁺ Trm cells from HFHC mice (HFHC Trm; n = 3), and CD8⁺ Trm cells from RES (5 weeks) mice (RES Trm; n = 3). d Heat map of key upregulated (upper) or down-regulated (lower) genes characteristic of Trm. e K-means clustering analysis. Most variable 1200 genes were clustered to six groups. Featured genes determined by scRNA-seq were shown. f PCA analysis plot of top 1000 genes. g–i ScTCR-seq analysis of combined liver or spleen CD8 T cells derived from ND and RES (5 weeks) mice. g UMAP projection of combined liver and spleen CD8⁺ T cells (n = 53,659) derived from ND(Liver), ND (Spleen), RES (Liver), and RES (Spleen) mice. Green circle highlights clusters of Trm cells. h UMAP projection and i percentage of top 10 and other TCR clonotypes in each cell subset. Source data are provided as a Source data file.

Fig. 5. Direct antifibrotic effect of CD8⁺ Trm cells in NASH progression.

a Study design: male C57BL/6 (Ly5.2) mice were fed a MCD diet for 8 weeks. Mice were intravenously injected with either PBS, liver CD8T cells isolated from Ly5.1 HFHC mice or Ly5.1RES (5 weeks) mice, or liver non-CD8T cells isolated from RES mice (2 × 10⁶ cells each) on day 1, 6, and 11 starting from week six (n = 6 mice for MCD group, and n = 7 mice for the other groups). b Representative CD45.2/CD45.1 staining of CD45⁺TCRβ⁺NK1.1⁻CD8α⁺ gated liver MNCs (left). Representative CD44/CD62L (center) and KLRG1/CD69 (right) staining of CD45⁺TCRβ⁺NK1.1⁻CD8α⁺ CD45.1⁺ (Transferred CD8T) gated liver MNCs. c Frequency of the transferred CD8⁺ T cell subset in CD45⁺ liver MNCs of the indicated groups of mice. d Serum ALT levels. e Representative photomicrographs of Masson trichrome staining of liver sections. Scale bars: 200 µm. f Representative photomicrographs (left) and positive area (right) of Sirius Red staining of liver sections. Scale bars: 200 µm. g Hydroxyproline levels. h Fibrosis associated genes (Col1a1, Col1a2, Acta2, Timp1, Desmin, Spp1) mRNA levels. i HSCs count in unit area of liver section. j Number of CD45⁺CD11b⁺ CD11c⁻ macrophages in CD45⁺ liver MNCs. k (left) Representative fluorescent photomicrographs of liver sections stained with DAPI (blue), anti-desmin Ab (green), and TUNEL (pink). Scale bars: 50 µm. (right) Count of TUNEL positive HSCs in unit area of liver section. Data are presented as mean ± SEM. One-way ANOVA with Tukey’s multiple comparisons post-hoc test was applied. Source data are provided as a Source data file.

Fig. 6. Roles for IL-15 signaling in the maintenance of CD8⁺ Trm cells in NASH resolution.

a Study design: resolution induced mice were intraperitoneally injected with anti-IL-15 antibody (RES + IL-15 neutralization group) or the isotype control (RES group) once every 3 days starting from 2 weeks following the diet switch to 5 weeks, and compared to the ND-fed mice (n = 4 mice for ND group, and n = 6 mice for RES and RES + IL-15 neutralization groups). b Proportion and number of CD8⁺ Trm cells and number of KLRG1⁻ CD8⁺ Tem and KLRG1⁺ CD8⁺ Tem in the liver CD45⁺ MNCs. c Proportion of CD8⁺ Trm cells (left) and Tem cells (left) in the combined population. d Il15 mRNA levels in the whole liver (n = 6 mice for ND and HFHC group, and n = 8 mice for RES groups). e Serum ALT levels. f Representative photomicrographs of Masson trichrome staining of liver sections. Scale bars: 200 µm. g Representative photomicrographs (left) and positive area (right) of Sirius Red staining of liver sections. Scale bars: 200 µm. h Hydroxyproline levels. i Fibrosis associated genes (Col1a1, Col1a2, Acta2, Timp1, Desmin, Spp1) mRNA levels. j HSCs count in unit area of liver section. k Number of CD45⁺CD11b⁺CD11c⁻ macrophages in CD45⁺ liver MNCs. l (left) Representative fluorescent photomicrographs of liver sections stained with DAPI (blue), anti-desmin Ab (green), and TUNEL (pink). Scale bars: 50 µm. (right) Count of TUNEL positive HSCs in unit area of liver section. Data are presented as mean ± SEM. One-way ANOVA with Tukey’s multiple comparisons post-hoc test was applied. Source data are provided as a Source data file.

Fig. 7. CD8⁺ Trm cells target HSCs in a CCR5-dependent manner in NASH resolution.

a (Left) Representative fluorescent photomicrographs of liver sections of ND, HFHC, and RES (3 weeks) mice stained with 4′,6-diamidino-2-phenylindole (DAPI: blue), anti-CD8 Ab (green), and anti-desmin Ab (red). Insets show higher magnification. Scale bars: 50 µm. (Right) Count of CD8⁺ T cells co-localized with HSCs in unit area of liver section　(n = 4 mice per group). b, c Ccl3, Ccl4, and Ccl5 mRNA levels in sorted CD8 T cell subsets (b) (n = 3 mice per group), and isolated hepatocytes (c) (n = 4 mice per group). d Frequency of CCR5⁺ cells in CD45⁺ liver MNCs and HSCs of the indicated groups (n = 3 mice per group). e Cell numbers of migrated HSCs in lower compartment containing supernatants of CD3/28-stimulated CD8⁺ T cell subset with or without CCR5 inhibitor, Marovenic (MVC) (left; HSCs and CD8⁺ T cells isolated from RES mice, right; isolated from HFHC mice, n = 3 biologically independent samples per group). f Study design: RES mice were intravenously treated with vitamin A-control siRNA liposomes (RES group) or vitamin A-Ccr5 siRNA liposomes (RES + HSC specific CCR5KD group) twice per week starting from 2 weeks following the diet switch to 5 weeks (n = 6 mice per group). g Ccr5 mRNA levels in isolated HSCs, Kupffer cells, and hepatocytes of the indicated groups. h (left) Representative fluorescent photomicrographs of liver sections stained with DAPI (blue), anti-desmin Ab (red), and anti-CD8 Ab (green). Scale bars: 50 µm. (right) Count of CD8⁺ T cells co-localized with HSCs in unit area of liver section. i Serum ALT levels. j Hydroxyproline levels. k Representative photomicrographs (left) and positive area (right) of Sirius Red staining of liver sections. Scale bars: 200 µm. l Fibrosis associated genes (Col1a1, Col1a2, Acta2, Timp1, Desmin, Spp1) mRNA levels. m HSCs count in unit area of liver section. n Number of CD45⁺CD11b⁺CD11c⁻ macrophages in CD45⁺ liver MNCs. o (left) Representative fluorescent photomicrographs of liver sections stained with DAPI (blue), anti-desmin Ab (green), and TUNEL (pink). Scale bars: 50 µm. (right) Count of TUNEL positive HSCs in unit area of liver section. Data are presented as mean ± SEM. One-way ANOVA with Tukey’s multiple comparisons post-hoc test (a–c, e) or two-sided unpaired Student’s t test (g–o) was applied. Source data are provided as a Source data file.

Fig. 8. CD8⁺ Trm cells induce HSCs apoptosis in a FasL dependent manner in NASH resolution.

a (left) Representative fluorescent photomicrographs of liver sections from ND, HFHC, and RES (3 weeks) mice stained with anti-desmin Ab (green) and TUNEL (pink). Scale bars: 50 µm. (right) Count of TUNEL positive HSCs in unit area of liver section (n = 4 mice per group). b, c Percent cytotoxicity of HSCs by CD8 Trm (red) or CD8 T naive (blue) isolated from RES 5 weeks mice (b) or HFHC mice (c) (n = 3 biologically independent samples per group). d Fasl mRNA levels in sorted CD8⁺ T cell subsets (n = 3 mice per group). e Count of FasL positive CD8⁺ T cells in unit area of liver section (n = 4 mice per group). f Percent cytotoxicity of HSCs co-cultured with CD8⁺ Trm cells in the presence of the indicated apoptosis inhibitors for 8 h (n = 3 biologically independent samples per group). g Fas mRNA levels in isolated HSCs and hepatocytes of the indicated mice (n = 4 mice per group). h (left) Representative fluorescent photomicrographs of liver sections stained with DAPI (blue), anti-desmin Ab (green), and anti-Fas (red). Scale bars: 50 µm. (right upper) Count of Fas positive HSCs in unit area of liver section (n = 4 mice per group). (right lower) Positive area of Fas positive HSCs in unit area of liver section. i Study design: resolution induced mice were intraperitoneally injected with anti-FasL antibody (RES + FasL neutralization group) or the isotype control (RES group) once every 3 days for 3 weeks starting at week 2 following diet switch (n = 7 mice per group). j Serum ALT levels. k Hydroxyproline levels. l Representative photomicrographs (left) and positive area (right) of Sirius Red staining of liver sections. Scale bars: 200 µm. m Fibrosis associated genes (Col1a1, Col1a2, Acta2, Timp1, Desmin, Spp1) mRNA levels. n HSCs count in unit area of liver section. o Number of CD45⁺CD11b⁺ CD11c⁻ macrophages in CD45⁺ liver MNCs. p (left) Representative fluorescent photomicrographs of liver sections stained with DAPI (blue), anti-desmin Ab (green), and TUNEL (pink). Scale bars: 50 µm. (right) Count of TUNEL positive HSCs in unit area of liver section. Data are presented as mean ± SEM. One-way ANOVA with Tukey’s multiple comparisons post-hoc test (a, d–h) or two-sided unpaired Student’s t test (j–p) was applied. Source data are provided as a Source data file.

Fig. 9. Abundance of CD69⁺CD8⁺ Trm cells in human advanced NASH livers.

a, b Representative (left) H&E staining and (right) immunofluorescent staining with anti-CD8α Ab and anti-CD69 Abs of resected liver tissues from normal control (a) and NASH LC (F4) patient (b). Middle and lower panels show each upper image at a higher magnification. Scale bars: 500 µm (normal control) or 1 mm (NASH F4) in the top panel, 250 µm in the middle panel. c Number of CD69⁺ and CD8⁺ co-stained cells in the liver of indicated groups of patients (n = 5 for normal control, n = 4 for F1-2, and n = 9 for F3-4). Data are presented as mean ± SEM. One-way ANOVA with Tukey’s multiple comparisons post-hoc test was applied. Source data are provided as a Source data file.