Neuregulin 4 suppresses NASH-HCC development by restraining tumor-prone liver microenvironment

Abstract

The mammalian liver comprises heterogeneous cell types within its tissue microenvironment that undergo pathophysiological reprogramming in disease states, such as non-alcoholic steatohepatitis (NASH). Patients with NASH are at an increased risk for the development of hepatocellular carcinoma (HCC). However, the molecular and cellular nature of liver microenvironment remodeling that links NASH to liver carcinogenesis remains obscure. Here, we show that diet-induced NASH is characterized by the induction of tumor-associated macrophage (TAM)-like macrophages and exhaustion of cytotoxic CD8+ T cells in the liver. The adipocyte-derived endocrine factor Neuregulin 4 (NRG4) serves as a hormonal checkpoint that restrains this pathological reprogramming during NASH. NRG4 deficiency exacerbated the induction of tumor-prone liver immune microenvironment and NASH-related HCC, whereas transgenic NRG4 overexpression elicited protective effects in mice. In a therapeutic setting, recombinant NRG4-Fc fusion protein exhibited remarkable potency in suppressing HCC and prolonged survival in the treated mice. These findings pave the way for therapeutic intervention of liver cancer by targeting the NRG4 hormonal checkpoint.

Keywords: HCC; NASH; NRG4; T cell; adipokine; endocrine; fatty liver disease; liver cancer; macrophage; microenvironment; single-cell RNA-seq.

Figure 1. Induction and regulation of TAM-like macrophages in NASH liver.

(A) Heatmap of top 20 marker genes for the macrophage subclusters. (B) Feature plots illustrating macrophage gene expression. (C) Virtual flow analysis of intrahepatic macrophages by gating for Csf1r and Apoe, Trem2, or Tgfbr1 mRNA expression. (D) Fluorescence images of liver sections from Trem2-Cre/Rosa26-tdTomato mice under different treatment conditions. Scale bars represent 100μm. (E) RNA velocity analysis of macrophage gene expression. Arrows denote likely trajectory of cell states among different subpopulations. (F) Flow cytometry analysis of liver macrophages in mice fed NASH diet following bone marrow transplantation. (G) Bubble plot illustrating relative mRNA expression for genes encoding secreted factors (top) and membrane proteins (bottom) with enriched expression in NAMs. (H) Violin plot of gene expression among different macrophage subclusters. (I) qPCR analysis of Tgfb expression in chow and NASH liver. (J) qPCR analysis of gene expression in cultured BMDMs treated with vehicle or 2.5 ng/ml TGFβ for 24 hrs. Data in (F), (I), and (J) represent mean ± SEM; two-tailed unpaired Student’s t-test. **p<0.01, ***p<0.001, ****p<0.0001.

Figure 2. NASH pathogenesis triggers CD8+ T cell exhaustion in the liver.

(A) UMAP illustrating T cells among liver NPCs (top) and three T cell subclusters (bottom). (B) Volcano plot of gene expression using averaged values of normalized expression levels for CD8+ T cells from chow and NASH livers. X-axis indicates log-transformed fold change of gene expression between NASH and chow livers. (C) Heatmap of a subset of genes differentially expressed in CD8+ T cells from chow and NASH mouse livers. (D) Dot plot illustrating relative abundance of CD8+ T cells expressing Cd8a in combination with the indicated genes in chow and NASH livers. (E) qPCR analysis of hepatic gene expression in mice fed chow (n=7) or NASH (n=7) diet for 4 months. (F) Confocal images of anti-PD1 immunofluorescence staining on liver sections. Scale bars=10μm. (G) Flow cytometry analysis of PD1 expression in intrahepatic T cells from mice fed chow (n=5) and NASH (n=5) diet. (H) Flow cytometry analysis of intracellular IFNγ and IL-2. Liver NPCs from mice fed chow (n=8) or NASH (n=6) diet for 5 months were treated with PMA/ionomycin for 6 hrs. (I) CFSE proliferation assay of CD8+ T cells. NPCs from chow (n=3) or NASH (n=3) mouse livers were incubated with CD3/CD28 Dynabeads for 24 and 120 hrs. (J) qPCR analysis of gene expression in liver biopsies from non-NASH (n=7) and NASH patients (n=7). Data in (E), (G), (H), (I) and (J) represent mean ± SEM; two-tailed unpaired Student’s t-test. *p<0.05, **p<0.01, ***p<0.001, ****p<0.0001.

Figure 3. NRG4 serves as a hormonal checkpoint for NASH-associated HCC.

(A) Volcano plot illustrating differential gene expression. Bulk liver RNA sequencing was performed on three pairs of pooled NASH diet-fed WT and NRG4 KO mice. X-axis indicates log-transformed fold change of gene expression between KO and WT mice. (B) Gene ontology analysis of upregulated (red) and downregulated (blue) genes in (A). (C) qPCR analysis of hepatic gene expression in WT (n=10) and NRG4 KO (n=11) mice fed NASH diet for 6 months. (D) A schematic outline of DEN/NASH liver tumor study using male WT (n=22) and NRG4 KO (n=28) mice. NASH diet feeding was initiated at 5 weeks of age. (E) Metabolic parameters and tumor count in WT (n=22) and NRG4 KO (n=28) mice subjected to the DEN/NASH protocol. (F) A schematic outline of DEN/NASH liver tumor study using male WT (n=22) and NRG4 KO (n=28) mice. NASH diet feeding was initiated at 5 weeks of age. (G) Metabolic parameters and tumor count in the treated mice. (H) Liver appearance and histology. Scale bar=20μm. Data in (C), (D), (E) and (G) represent mean ± SEM; two-tailed unpaired Student’s t-test. *p<0.05, **p<0.01, ***p<0.001, ****p<0.0001.

Figure 4. Effects of NRG4 deficiency on the liver immune microenvironment.

Single-cell RNAseq analysis of liver NPCs isolated from WT (n=4) and Nrg4 KO (n=4) mice following 7 months of NASH diet feeding. (A) Macrophage subclusters and feature plots. (B) Pie chart of macrophage cell count from WT (red) and NRG4 KO (blue) livers in each subcluster. (C) Confocal images of liver immunofluorescence staining. Scale bar=10μm. (D) Immunoblots of total liver lysates from WT and NRG4 KO mice fed NASH diet for 6 months. (E) T cell subclusters and feature plots. (F) Virtual flow analysis of intrahepatic CD8+ T cells by gating for Cd8a in combination with Pdcd1 or Lag3 mRNA levels. (G) Confocal images of liver immunofluorescence staining. Scale bar=10μm. (H) Flow cytometry analysis of PD1 expression in intrahepatic CD8+ T cells from NASH diet fed WT (n=8) and NRG4 KO (n=9) mice. (I) CFSE proliferation assay of CD8+ T cells from WT (n=8) and NRG4 KO (n=10) mice fed NASH diet for 5 months. (J) Anti-PDL1 treatment study. WT and NRG4 KO mice were subjected to the DEN/NASH tumor induction protocol followed by treatments twice a week: WT IgG2b (n=10), WT anti-PDL1 (n=15), Nrg4 KO IgG2b (n=11), Nrg4 KO anti-PDL1 (n=10). Data in (G), (I) and (J) represent mean ± SEM; two-tailed unpaired Student’s t-test. *p<0.05, **p<0.01, and ****p<0.0001.

Figure 5. Inhibition of tumor-prone liver microenvironment and NASH-HCC by NRG4.

(A) A schematic overview of DEN/NASH liver tumor study using male WT (n=14) and NRG4 TG (n=14) mice. NASH diet feeding was initiated at 5 weeks of age. (B) Metabolic parameters and tumor count in treated mice. (C) Liver appearance. (D) qPCR analysis of gene expression WT and NRG4 TG mouse livers. (E) A schematic diagram of hNRG4-Fc fusion protein design and study outline. (F) H&E histology and Sirius red staining of liver sections from transduced mice. Scale bar=20μm. (G) Liver hydroxyproline content in mice transduced with AAV-Fc (n=7) or AAV-hNRG4-Fc (n=8). (H) qPCR analysis of hepatic gene expression in transduced mice. Data in (B), (D), (G) and (H) represent mean ± SEM; two-tailed unpaired Student’s t-test. *p<0.05, **p<0.01, ***p<0.001.

Figure 6. Suppression of oncogene-induced HCC by recombinant hNRG4-Fc fusion protein.

(A) Plasma concentrations of hNRG4-Fc, as measured by hIgG1 Fc ELISA, at different time points following a single injection of the fusion protein (2 mg/kg, i.p.). (B) Immunoblots of total Min6 cell lysates treated with Fc, hNRG4-Fc, or hNRG4 peptide at 0.8, 4, 20, or 100 nM for 15 minutes. (C) A schematic outline of hNRG4-Fc fusion protein treatment study, and the metabolic parameters and tumor burden in the treated mice. Four-month-old male mice were transduced with AAV-cMYC/AAV-nRAS followed by weekly treatment with Fc (2.5 mg/kg, n=13) or hNRG4-Fc at 0.5 mg/kg (n=11) or 2.5 mg/kg (n=14). (D) qPCR analysis of hepatic gene expression in treated mice. (E) Liver appearance and histology. Scale bar=200μm. (F) Survival curves of transduced mice treated with 1.5 mg/kg of Fc (n=15) or hNRG4-Fc (n=15) fusion protein. Data in (C) and (D) represent mean ± SEM; two-tailed unpaired Student’s t-test. *p<0.05, **p<0.01, ***p<0.001.

Figure 7. Interaction between NRG4 and the liver immune microenvironment in NASH-HCC.

(A) Heatmap illustrating expression patterns of NRG4-regulated genes among different liver cell types. (B) hNRG4-Fc and anti-PDL1 co-treatment study. WT male mice transduced with AAV oncogenes were randomly divided into four treatment groups: IgG2b+Fc (n=15), IgG2b+hNRG4-Fc (n=15), anti-PDL1+Fc (n=15), anti-PDL1+hNRG4-Fc (n=15). (C) Tumor burden in transduced WT and Trem2 KO mice: WT Fc (n=12), Trem2KO Fc (n=15), WT hNRG4-Fc (n=16), Trem2KO hNRG4-Fc (n=11). (D) CFSE proliferation assay of splenic CD8+ T cells from OT-1 transgenic mice co-cultured with BMDMs from WT and Trem2 KO mice in the absence or presence of OVA. (E) A model depicting NRG4 as a hormonal checkpoint in NASH-associated HCC. Data in (B) and (C) represent mean ± SEM; two-tailed unpaired Student’s t-test. *p<0.05, **p<0.01, ***p<0.001, ****p<0.0001.