HIF-2α drives hepatic Kupffer cell death and proinflammatory recruited macrophage activation in nonalcoholic steatohepatitis

Abstract

Proinflammatory hepatic macrophage activation plays a key role in the development of nonalcoholic steatohepatitis (NASH). This involves increased embryonic hepatic Kupffer cell (KC) death, facilitating the replacement of KCs with bone marrow-derived recruited hepatic macrophages (RHMs) that highly express proinflammatory genes. Moreover, phago/efferocytic activity of KCs is diminished in NASH, enhancing liver inflammation. However, the molecular mechanisms underlying these changes in KCs are not known. Here, we show that hypoxia-inducible factor 2α (HIF-2α) mediates NASH-associated decreased KC growth and efferocytosis by enhancing lysosomal stress. At the molecular level, HIF-2α stimulated mammalian target of rapamycin (mTOR)- and extracellular signal-regulated kinase-dependent inhibitory transcription factor EB (TFEB) phosphorylation, leading to decreased lysosomal and phagocytic gene expression. With increased metabolic stress and phago/efferocytic burden in NASH, these changes were sufficient to increase lysosomal stress, causing decreased efferocytosis and lysosomal cell death. Of interest, HIF-2α-dependent TFEB regulation only occurred in KCs but not RHMs. Instead, in RHMs, HIF-2α promoted mitochondrial reactive oxygen species production and proinflammatory activation by increasing ANT2 expression and mitochondrial permeability transition. Consequently, myeloid lineage-specific or KC-specific HIF-2α depletion or the inhibition of mTOR-dependent TFEB inhibition using antisense oligonucleotide treatment protected against the development of NASH in mice. Moreover, treatment with an HIF-2α-specific inhibitor reduced inflammatory and fibrogenic gene expression in human liver spheroids cultured under a NASH-like condition. Together, our results suggest that macrophage subtype-specific effects of HIF-2α collectively contribute to the proinflammatory activation of liver macrophages, leading to the development of NASH.

Fig. 1.. Hepatic macrophages express HIF-2α.

(A) IHC analyses of HIF-2α expression in human liver biopsy samples from patients with different stages of NAFLD (scale bar = 10 μm). NAS, NAFLD activity score; Lob. Inflamm., lobular inflammation. (B and C) IHC of HIF-2α expression in the liver of NCD healthy (B) and (C) NASH diet mice. (D) Western blot of HIF-2α in KCs and BMDMs cultured at different oxygen (O₂) concentrations for 5h. (E) Western blot of HIF-2α in KCs and BMDMs incubated with different combinations of TGFβ, leptin, PA, or insulin treatment in low or high glucose medium. (F) scRNA-seq analysis of Hif2a expression in hepatic macrophages of healthy and NASH mice (right) (n=5 mouse/group). UMAP representation of five different hepatic macrophage clusters shown on the left. (G) Hif2a mRNA expression in KCs and BMDMs from NCD healthy mice. Hif2a mRNA expression was measured after incubating in low glucose (0.5mM) or high glucose (16.7mM)+PA (100 μM) medium for 48h (n=6 mice or wells/group). ***P < 0.001, ****P < 0.0001. All data are mean +/− SEM. Statistical analysis was performed by Kruskal-Wallis with Dunn’s multiple comparison tests (F) or 2-way ANOVA with Tukey’s multiple comparison tests (G).

Fig. 2.. Macrophage HIF-2α enhances NASH pathology.

(A) Hif2a expression in KCs and BMDMs derived from WT and H2MKO mice (n=3 or 6 mice/group). (B to J) Measurements in NASH WT and H2MKO mice. (B) Body weight (n=14 and 19 mice). (C) Liver and epididymal fat (eWAT) mass (n=10 and 9 mice). (D) Liver lipid contents (n=9-12 mice/group) (E) Plasma lipid concentrations (n=10 mice/group). (F) H&E and Sirius red staining of liver sections (scale bar = 100 μm). (G) NASH CRN scores (n=7 and 9). (H) Serum aminotransferase concentrations (n=10 and 9 mice). ALT, alanine aminotransferase; AST, aspartate aminotransferase. (I) Lipogenic, fibrogenic, inflammatory, and fatty acid oxidation gene expression in liver (n=10 and 9 mice). (J) Fatty acid oxidation in the liver (n=8 and 6). (K) Lipogenic, fibrogenic, inflammatory, and fatty acid oxidation gene expression in WT and KC-specific HIF-2α KO mouse liver spheroids incubated in a NASH-like medium for 7 days (n=6-9 wells/group). (L) Hif1a and Hif2a expression in KCs from WT and KC-specific HIF-2α KO mice (n=4 mice/group). (M) Serum aminotransferase (ALT and AST) concentrations in NASH WT and KC-specific HIF-2α KO mice (n=10 mice/group). (N) Lipogenic, fibrogenic, inflammatory, and fatty acid oxidation gene expression in livers of NASH WT and KC-specific HIF-2α KO mice (n=10 mice/group). (O) Liver TG contents in NASH WT and KC-specific HIF-2α KO mice. (n=10 mice/group). (P) Inflammatory and fibrotic gene expression in human liver spheroids incubated in normal or a NASH-like condition in the presence or absence of 100 μM HIF-2α-specific inhibitor (PT-2385) for 7 days (n=2, 5 and 7 wells per group). Throughout, mRNA expression data represent expression relative to 36B4 expression (A, L), or otherwise as fold change from control. *P < 0.05, **P < 0.01, ***P < 0.001, ****P < 0.0001. All data are mean +/− SEM. Statistical analysis was performed by the 2-tailed unpaired Student’s or Welch’s t tests (A-E,G-J,L,N,O), Mann-Whitney tests (C,D,G,M,N), 1-way (P) or 2-way (K) ANOVA with Tukey’s multiple comparison tests, Brown-Forsythe and Welch ANOVA with Dunnett’s multiple comparison tests (P), or Kruskal-Wallis tests with Dunn’s multiple comparison tests (P).

Fig. 3.. HIF-2α induces KC death in NASH.

(A-C) scRNA-seq analysis of liver leukocytes in NASH WT and H2MKO mice. UMAP projection of WT and KO liver leukocytes are shown separately (A) or together (B). (C) Pathway analysis using DE genes between NASH WT and H2MKO mice. NES, normalized enrichment score. Adj. P value, adjusted P value. (D) IHC analysis of active/cleaved caspase-3⁺ apoptotic KCs in NASH H2MKO and WT mice. Apoptotic cells were counted and plotted on the right (n=5 and 6 mice/group). (E) IHC analysis of Ki67⁺ proliferating KCs in NASH H2MKO and WT mice. Proliferating cells were counted and plotted on the right (n=14 and 6 mice/group). (F) The proportion of mki67⁺ proliferating cells in three different hepatic macrophage subtypes. Data were acquired by performing focused analysis of proliferating macrophages in scRNA-seq data in panel A and B. UMAP projection of proliferating macrophages is shown on the left. (G) The proportion of each of the macrophage subtypes among all macrophages in NASH H2MKO and WT mice. (H) Flow cytometry analysis of the proportion of KCs and total hepatic macrophages among all NPCs in NASH H2MKO and WT mice (n=6 and 4 mice). (I) The ratio of between different hepatic macrophage subtypes assessed by flow cytometry (n=5 and 4 mice per group). (J-K) In silico cell lineage tracing using RNA Velocity analysis in hepatic macrophages from NASH H2MKO and WT mice. (J) Color-coded UMAP showing start and end points (Markov process) of hepatic macrophages. (K) PCA plots showing bi-directional cell movement trajectory of RHMs. Arrows indicate vectoral changes (size and direction) in transcriptomic signatures in nearby cells and the principal curve (black) indicates integrated directional changes in overall cell movement. Quantitative changes in transcriptomic movement in each RHM subtype were plotted on the right. (L) CytoTrace analysis of cell maturity in each of the macrophage subtypes. *P < 0.05, **P < 0.01, ***P < 0.001, ****P < 0.0001. All data are mean +/− SEM. Statistical analysis was performed by the 2-tailed unpaired Student’s (D,E,H,I) or Welch’s (H,M) t tests, Mann-Whitney tests (K,L), or 1-way ANOVA with Tukey’s multiple comparison tests (F, G).

Fig. 4.. HIF-2α suppresses lysosomal gene expression, causing decreased efferocytosis and increased death in KCs.

(A-B) Venn diagrams showing the number of genes decreased (A) or increased (B) in NASH WT KCs compared with healthy WT KCs, of which expression was reversed by HIF-2α KO commonly or specifically in KCs or KLRMs in mice. (C) Heatmap showing HIF-2α target genes specifically expressed in KCs or KLRMs. (D) Pathway analysis in genes that were commonly up or down-regulated in KCs and KLRMs of NASH H2MKO compared with NASH WT mice. (E) Relative lysosomal mass in KCs from NCD WT and H2MKO mice after incubation in low glucose (control) or high glucose+PA medium (M/S) (n=6 or 8 mice/group). (F) Relative lysosomal activity in KCs from NCD WT and H2MKO mice after incubation in control or M/S medium (n=5-7 mice/group). (G) Efferocytic activity in KCs from NCD WT and H2MKO mice after incubation in control or M/S medium (n=6-12 wells/group). (H) mRNA expression of phagocytic or lysosomal gene expression in KCs isolated from NCD WT or H2MKO mice and incubated in control or M/S medium (n=6 mice/group). Data were normalized by 36B4 expression. (I to K) Efferocytic activity (n=9 wells/group) (I), lysosomal cathepsin D (J), and cytosolic cathepsin D abundance (K) in KCs from NASH WT or HIF-2α MKO. (L-N) mtROS (L), cellular MDA (M), and caspase-3/7 activity (N) in WT or HIF-2α KO KCs after incubation in control or M/S medium (n=6 or 10 wells/group). (O) Caspase-3/7 activity. WT and HIF-2α KO KCs were transfected with mock or Ctsb- or Ctsd-specific siRNAs. 24h later, cells were incubated in control or M/S medium in the presence or absence of 10 μM CA-74-ME for 24 h (n=6 wells/group). Scrb, scrambled. ns, not significant. *P < 0.05, **P < 0.01, ***P < 0.001, ****P < 0.0001. All data are presented as mean +/− SEM. Statistical analysis by the 2-tailed unpaired Welch’s t test (I) or 1- or 2-way ANOVA with Tukey’s multiple comparison tests (E-H, L-O).

Fig. 5.. HIF-2α induces mTOR and ERK-dependent TFEB phosphorylation/inhibition in KCs.

(A) Volcano plot showing phosphoproteins changed in KCs isolated from NASH H2MKO mice compared with NASH WT mice. (B) Volcano plot showing pathways decreased in global (left) and phospho-proteome (right) of HIF-2α KO KCs (terms from the MSigDB_Hallmark_2020 gene set). (C) Volcano plot showing differential kinomes in HIF-2α KO vs WT KCs, predicted from the phosphoproteomics data in (A). (D) Total (t-) and/or phosphorylated (p-) mTOR and TFEB in KCs from NASH WT and H2MKO mice (n=2 mice/group). (E) Total (t-TFEB) and phosphorylated TFEB (p-TFEB) in WT and HIF-2α KO KCs after 24h incubation in low glucose (control) or high glucose+PA medium (metabolic stress, M/S). (F) Total (t-) and phosphorylated (p-) mTOR and ERK abundance in WT and HIF-2α KO KCs after 24h incubation in control or MS/MS medium. (G) Phosphorylated (p-) TFEB, S6K, ERK, and mTOR in WT and HIF-2α KO KCs after 24h incubation in control or M/S medium in the presence or absence of U0126 or rapamycin (Rapa). (H) Marco, Ctsb, Slc7a5, and Slc7a11 expression in WT KCs transfected with empty or CA-HIF-2α overexpressing plasmid (n=3 wells/group). (I) Slc7a5 and Slc7a11 expression in WT and HIF-2α KO KCs after 48h incubation in control or M/S medium (n=6 wells/group). (J) Caspase-3/7 activity in KCs transfected with scrambled ASOs, Flcn-ASO, and/or CA-HIF-2α expressing vector. Caspase-3/7 activity was measured after incubating in control or M/S medium for 48h (n=7 wells/group). (K) mRNA expression of NASH-associated genes in WT or KC-specific CA-HIF-2α-overexpressing mouse liver spheroids transfected with Flcn-specific or scrambled control ASOs. Gene expression was measured 7 days after incubation in control or NASH medium (n=12 or 20 wells/group). (L-O) 4-month-old NASH WT mice were treated with scrambled or Flcn-specific ASOs (Flcn-ASO) for 4 weeks (n=10 mice/group). (L) Serum ALT and AST concentrations. (M) Liver TG contents. (N) H&E and Sirius Red staining of liver sections (scale bar = 100 μm). (O) Lipogenic, fibrogenic, and pro-inflammatory gene expression in liver. Data were normalized by 36B4 expression. *P < 0.05, **P < 0.01, ***P < 0.001, ****P < 0.0001. All data are mean +/− SEM. Statistical analysis was performed by the 2-tailed unpaired Student’s (H,L,M,O) or Welch’s (H,O) t tests, Mann-Whitney tests (O), 1-(J,K) or 2-way (I) ANOVA with Tukey’s multiple comparison tests, or Kruskal-Wallis tests with Dunn’s multiple comparison tests (K).

Fig. 6.. HIF-2α drives proinflammatory RHM activation by increasing ANT2-dependent opening of the mtPTP in NASH.

(A and B) Volcano plots showing changes in global protein expression (A) and pathways increased (right) or decreased (left) in the global proteome (B) in HIF-2α MKO and WT Ly6C⁺ RMs isolated from NASH mice. (C) scRNA-seq analysis of Ant1/Slc25a4 and Ant2/Slc25a5 expression in hepatic macrophages in healthy and NASH mice. Color-coded UMAP shows 5 different hepatic macrophage clusters. Heatmaps (upper right) and violin plots (bottom) show Ant1 and Ant2 expression in each of the hepatic macrophage clusters. (D) Ant2 expression in RHMs from NASH WT and HIF-2α MKO mice. (E and F) ANT1 and ANT2 protein (E) and mRNA (F) expression in WT and HIF-2α KO BMDMs after 24h incubation in low glucose (control) or high glucose+PA medium (metabolic stress, M/S). (n=6 wells/group). (G-H) Opening of the mtPTP (G) and mtROS (H) in WT and HIF-2α KO BMDMs incubated in control or M/S medium for 24h in the presence or absence of cyclosporin A (CsA) or Tro19622 (TRO) (n=5-9 wells/group). (I) Inflammatory gene expression in WT and HIF-2α KO BMDMs incubated in M/S medium for 24h in the presence or absence of CsA (n=4 wells/group). (J) mtROS in ANT2-overexpressing WT and HIF-2α KO BMDMs. 24h after empty (mock) or ANT2-overexpressing plasmid vector transfection, cells were incubated in control or M/S medium for 24h. mtROS were measured using MitoSOX (n=8 or 9 wells/group). (K) Inflammatory gene expression in Ly6C⁺ RMs from NASH WT and H2MKO mice after in vitro incubation in the presence or absence of CsA for 24h (n=6 mice/group). (L) Cytokine secretion from Ly6C⁺ RMs from NASH WT and H2MKO mice (n=8 mice/group). (M) ANT1 and ANT2 expression in Ly6C⁺ RMs from NASH WT and H2MKO mice (n=2 mice/group). (N-O) Fatty acid oxidation (N) and Cpt1 expression (O) in WT hepatocytes incubated in NASH diet WT or H2MKO mouse KC-CM for 24h (n=8-11 wells/group). Data in panels F, I, K, and O were normalized by 36B4 expression. *P < 0.05, **P < 0.01, ***P < 0.001, ****P < 0.0001. All data are mean +/− SEM. Statistical analysis was performed by the 2-tailed unpaired Student’s (L) or Welch’s (L,O) t tests, Mann-Whitney tests (D,L,N), 2-way (F-K) ANOVA with Tukey’s multiple comparison tests, or Kruskal-Wallis tests with Dunn’s multiple comparison tests (C).