ECM1 expression in chronic liver disease: Regulation by EGF/STAT1 and IFNγ/NRF2 signalling

Abstract

Background & aims: The extracellular matrix protein 1 (ECM1) is essential for liver homeostasis by keeping latent transforming growth factor-beta quiescent. Upon hepatocyte damage, ECM1 is significantly downregulated, facilitating fibrosis and chronic liver disease (CLD) progression. We investigated the mechanism of ECM1 regulation in hepatocytes under pathophysiological conditions.

Methods: We used promoter analysis to predict Ecm1 transcriptional regulators and assessed the expression of Ecm1-related genes by single-cell RNA sequencing (scRNA-seq) and bulk RNA-seq. Functional assays were performed with AML12 cells, mouse and human primary hepatocytes, and liver tissue from mice and patients.

Results: In healthy hepatocytes, epidermal growth factor (EGF)/EGF receptor (EGFR) signalling sustains ECM1 expression through phosphorylating signal transducer and activator of transcription 1 (STAT1) at serine727 (S727), thus enhancing its binding to the ECM1 promoter and boosting gene transcription. This process is disrupted during liver inflammation by interferon gamma (IFNγ), which downregulates EGFR and inhibits EGF/EGFR/STAT1-mediated ECM1 promoter binding. Mechanistically, IFNγ-induced STAT1 phosphorylation at tyrosine701 (Y701) impairs the binding of p-STAT1 S727 to the ECM1 promoter. Additionally, IFNγ induces nuclear factor erythroid 2-related factor 2 (NRF2) nuclear translocation, which repressively binds to the promoter of ECM1, further reducing its expression. These findings are confirmed in several CLD mouse models (n = 2-6). Moreover, AAV8-ECM1 attenuates liver fibrosis and injury in Western diet-fed mice (n = 8-10), counteracting the effects of EGF signalling inhibition and IFNγ/NRF2 activation. In CLD patients (n = 22), ECM1 levels correlate positively with EGFR expression (p <0.0001) and negatively with IFNγ/NRF2 activation (p <0.0001).

Conclusions: EGF/STAT1 signalling promotes whereas IFNγ/NRF2 inhibits ECM1 expression in hepatocytes in health or disease, respectively. ECM1 has the potential to be developed as an antifibrotic agent, particularly in inflammation- or reactive oxygen species-driven CLD.

Impact and implications: This study reveals the regulatory mechanism of ECM1 in hepatocytes, demonstrating that EGF/STAT1 maintains ECM1 expression to prevent fibrosis, whereas IFNγ/NRF2 signalling inhibits ECM1 during chronic liver inflammation, thereby accelerating disease progression. These findings are important for researchers and clinicians to understand the pathogenesis of liver fibrosis, especially in CLD driven by inflammation or oxidative stress. Clinically, ECM1 levels correlate positively with EGFR expression and negatively with IFNγ/NRF2 activation, providing potential antifibrotic targets for CLD patients.

Keywords: Extracellular matrix; Growth factor; IFNγ; Liver fibrosis; TGF-β.

Graphical abstract

Fig. 1

EGF-EGFR signalling maintains ECM1 expression in hepatocytes of quiescent liver. (A) mRNA expression of hepatic ECM1 in CLD patients, extracted from the GEO dataset GSE126848 (MASLD, n = 15; MASH, n = 16; healthy individuals with normal weight n = 14 and obesity n = 12), GSE103580 (alcoholic cirrhosis, n = 67; non-severe alcoholic hepatitis, n = 13; alcoholic steatosis, n = 6), and GSE94660 (non-tumour and tumour of HBV-HCC, n = 21), respectively. UMAP visualization of cell populations and hepatocytes, and heatmap of ECM1 mRNA expression in (B) DDC mouse model (GSE193850), (C) ChD-HFD fed mice (GSE232182), and (D) patients with CLD (GSE202379). (E) qRT-PCR for ECM1 mRNA expression in MPHs and HPHs treated with EGF. (F) Immunoblotting of Ecm1 in MPHs treated with EGF. (G) qRT-PCR and immunoblotting showing the effect of erlotinib on Ecm1 and p-Egfr Y1068 in EGF-treated MPHs. (H) Effects of Egfr knockdown on Ecm1 and Egfr expression in EGF-treated MPHs by qRT-PCR and immunoblotting. (I) qRT-PCR and immunoblotting for Ecm1 in liver tissues from erlotinib-treated mice (40 mg/kg body weight). DMSO was administered as a placebo. (J) Representative IF staining for Ecm1 in liver tissues from erlotinib-treated mice. DRAQ5 fluorescent probe stains DNA. (K) Representative tdTomato fluorescence of liver tissues from Ecm1-tdTomato mice treated with erlotinib. (L) Representative IF staining for LAP-D R58 in liver tissues from erlotinib-treated mice. Scale bar, 25 μm. The results of qRT-PCR were normalized to PPIA. In immunoblotting, β-actin, GAPDH, and α-tubulin are loading controls. Quantification of protein expression was performed by ImageJ (National Institutes of Health, Bethesda, MD, USA). p-values were calculated by unpaired Student’s t test. Bars represent the mean ± SD. ∗p <0.05, ∗∗p <0.01, ∗∗∗p <0.001.

Fig. 2

EGF regulates homeostatic ECM1 expression in hepatocytes through STAT1. (A) qRT-PCR and immunoblotting of Ecm1 and Stat1 in EGF-treated MPHs with or without Stat1 knockdown. (B) Immunoblotting showing the effect of EGF on target proteins in MPHs at different time points. (C) Immunoblotting for p-Stat1 S727 and Stat1 in EGF-treated MPHs with or without knockdown Egfr. (D) Immunoblotting for p-Stat1 S727 and Stat1 in liver tissue from erlotinib-treated mice. DMSO was administered as a placebo. (E) Immunoblotting for Ecm1, p-Stat1 S727 and total Stat1 in AML12 cells transfected with eGFP or eGFP-Stat1 plasmids. (F) Predicted binding motif of Stat1 to Ecm1 promoter by Jaspar (ATGGCAGGAAA at -59 to -49 bp to the Ecm1 TSS). (G) Luciferase reporter assay analysing the activity of Ecm1 promoter (-339 to +161 bp) in HEK293T cells transfected with or without Stat1 plasmid. (H) ChIP qRT-PCR showing the binding of S727-phosphorylated Stat1 to the Ecm1 promoter in MPHs, with or without EGF treatment. Fragment ‘-155 to +120 bp’ is calculated relative to the Ecm1 TSS. Rabbit IgG-bound chromatin served as negative control. (I) The amplified products of ChIP PCR are shown as bands on 2% agarose gel. Ppia represents non-specific binding. The results of qRT-PCR were normalized to Ppia. For immunoblotting, GAPDH, α-tubulin, and β-actin are loading controls. Fig. 1C and D shared the same loading controls as Fig. 1H and I. Quantification of protein expression was done by ImageJ. p-values were calculated by unpaired Student’s t test. Bars represent the mean ± SD. ∗p <0.05, ∗∗p <0.01, ∗∗∗p <0.001.

Fig. 3

IFNγ abrogates EGF-EGFR-maintained ECM1 expression. (A) qRT-PCR and immunoblotting showing the impact of IFNγ on Ecm1 and Egfr in EGF-treated MPHs. (B) Immunoblotting showing the effect of EGF or IFNγ on target proteins in MPHs at different time points. (C) ChIP qRT-PCR displaying the effect of IFNγ on the binding of p-Stat1 Y701 to the Ecm1 promoter in MPHs. ChIP PCR amplified products were separated on 2% agarose gel. The fragment positions ‘-155 to +120 bp’ are calculated relative to the Ecm1 TSS. Rabbit IgG-bound chromatin served as negative control. Ppia represents non-specific binding. (D) Fold enrichment from ChIP qRT-PCR showing the effect of IFNγ on EGF-induced Stat1 binding to the Ecm1 promoter in MPHs. (E) qRT-PCR and immunoblotting showing target genes or proteins in liver tissue from IFNγ-treated mice (400 μg/kg body weight, for 4 days). PBS was administered as placebo. (F) Representative IF staining for Ecm1 in the liver tissue from IFNγ-treated mice. DRAQ5 fluorescent probe stains DNA. (G) Representative tdTomato fluorescence of liver tissue from Ecm1-tdTomato mice treated with IFNγ. (H) Representative IF staining for LAP-D R58 in liver tissues from IFNγ-treated mice. (I) qRT-PCR showing the impact of IFNγ on fibrosis-related genes in liver tissues from IFNγ-treated mice. Scale bar, 25 μm. The results of qRT-PCR are normalized to Ppia. GAPDH was used as loading control. Quantification of protein expression was done by ImageJ. p-values were calculated by unpaired Student’s t test. Bars represent the mean ± SD. ∗p <0.05, ∗∗p <0.01, ∗∗∗p <0.001.

Fig. 4

IFNγ inhibits ECM1 expression through NRF2. (A) qRT-PCR for mRNA expression of relevant genes in liver tissues from IFNγ-treated mice or (B) in IFNγ-treated MPHs. (C) ROS generation fluorescence detection in the IFNγ-treated MPHs. (D) Representative IF staining for Nrf2 in liver tissues from IFNγ-treated mice. Scale bar, 12.5 μm. (E) Immunoblotting for cytoplasmic and nuclear localization of Nrf2 in MPHs treated with EGF and/or IFNγ. (F) Representative IF staining showing localization of Nrf2 in IFNγ-treated MPHs. Scale bar, 25 μm. (G) Immunoblotting showing the impact of Nrf2 knockdown on Ecm1 in IFNγ-treated MPHs. (H) qRT-PCR and immunoblotting showing the impact of H₂O₂ on Ecm1 and Nrf2 in MPHs. (I) Representative IF staining showing Nrf2 in H₂O₂-treated MPHs. Scale bar, 25 μm. (J) Immunoblotting showing the impact of knockdown Nrf2 on Ecm1 in H₂O₂-treated MPHs. (K) qRT-PCR and immunoblotting showing the effect of H₂O₂ on target genes or proteins in EGF-treated MPHs. The results of qRT-PCR are normalized to Ppia. GAPDH/α-tubulin and histone H3 are loading controls for cytoplasmic and nuclear proteins, respectively. DRAQ5 fluorescent probe stains DNA. Quantification of protein expression was done by ImageJ. p-values were calculated by unpaired Student’s t test. Bars represent the mean ± SD. ∗p <0.05, ∗∗p <0.01, ∗∗∗p <0.001.

Fig. 5

NRF2 inhibits ECM1 expression through negative regulatory binding to its promoter. (A) qRT-PCR and immunoblotting showing the effect of OPZ on Nrf2 in MPHs. (B) Representative IF staining for Nrf2 in OPZ-treated MPHs. Scale bar, 25 μm. (C) qRT-PCR and immunoblotting displaying the effect of OPZ on Ecm1 in MPHs at different time points. (D) qRT-PCR and immunoblotting showing the effect of Nrf2 knockdown on Ecm1 in OPZ-treated MPHs. (E) qRT-PCR and immunoblotting showing the effect of Nrf2 overexpression on Ecm1 and Nrf2 in EGF-treated MPHs. (F) Immunoblotting displaying the effect of OPZ on Ecm1 in EGF-treated MPHs. (G) Predicted binding motif (antisense GGACATGACTCAGAA) of Nrf2 to the Ecm1 promoter by Jaspar. (H) Luciferase reporter assay analysing the activity of Ecm1 promoter (-953 to -454 bp) in HEK293T cells transfected with or without Nrf2 plasmid. (I) ChIP qRT-PCR showing the effect of OPZ on binding of Nrf2 to the Ecm1 promoter in MPHs. The fragment positions ‘-865 to -586 bp’ were calculated relative to the Ecm1 TSS. Rabbit IgG-bound chromatin served as negative control. (J) The amplified products of ChIP PCR were separated on 2% agarose gel. Ppia represents non-specific binding. (K) qRT-PCR and immunoblotting showing Ecm1 expression in OPZ-treated mice (150 mg/kg body weight). Olive oil was administered as placebo. (L, N) Representative IF staining showing Nrf2, Ecm1, and LAP-D R58 in liver tissues from OPZ-treated mice. (O) mRNA expression of target genes in liver tissue from p62-KO mice injected with/without adenovirus p62, extracted from the GEO dataset GSE134188. The results of qRT-PCR are normalized to Ppia. GAPDH and β-actin are loading controls. DRAQ5 fluorescent probe stains DNA. Scale bar, 25 μm. Quantification of protein expression was done by ImageJ. p-values were calculated by unpaired Student’s t test. Bars represent the mean ± SD. ∗p <0.05, ∗∗p <0.01, ∗∗∗p <0.001.

Fig. 6

Activation of IFNγ/NRF2 signalling and downregulation of EGFR/ECM1 in various mouse models of hepatic injury. (A–E) Heatmaps of relative mRNA expression for target genes in CCl₄-injured (twice per week for 6 weeks), BDL surgical (7 days after ligation), Fxr-KO, DDC diet (for 1 week) and ChD-HFD fed (for 3 months) mouse models, extracted from GEO datasets GSE222576, GSE166867, GSE76163, GSE193850, and GSE232182, respectively. (F) Immunoblotting for Ecm1, pY-Egfr, and Egfr in liver tissues from oil/CCl₄-treated mice. (G) Representative IF staining for Ecm1 and Nrf2 in liver tissues from oil/CCl₄-treated mice. (H) Immunoblotting for Ecm1, pY-Egfr, and Egfr in liver tissues from CD/WD-fed mice. (I) Representative IHC staining for Nrf2 in liver tissues from CD/WD-fed mice. (J) Schematic illustration of in vivo experiment with AAV8-ECM1 (i.v.) and WD-fed WT mice. (K–M) Representative IF staining for LAP-D R58, Ecm1, a-SMA, and Col1A1 in liver tissues from CD/WD-fed mice treated with control or AAV8-ECM1. DRAQ5 fluorescent probe stains DNA. Scale bar, 25 μm. (N) Representative H&E and PSR staining for liver tissues from CD/WD-fed mice treated with control or AAV8-ECM1. Scale bar, 70 μm. (O) Quantification of staining for LAP-D R58, a-SMA, Col1A1, and PSR, and for lipid droplets according to the white bubbles in H&E staining. GAPDH, α-tubulin are loading controls. p-values were calculated by unpaired Student’s t test. Bars represent the mean ± SD. ∗p <0.05, ∗∗p <0.01, ∗∗∗p <0.001.

Fig. 7

Activated IFNγ/NRF2 and decreased ECM1 expression aggravate the progression of fibrotic liver diseases. (A) Representative IF or IHC staining for ECM1, EGFR, NRF2, and TGF-β1 LAP-D R58 in patients with F1–F2 and F3–F4 fibrosis. DRAQ5 fluorescent probe stains DNA. Scale bar, IF 25 μm, IHC 87 μm. (B, C) Correlation analysis of EGFR/NRF2 and ECM1 expression levels in the liver tissue of patients with F1–F2 or F3–F4 fibrosis. (D) Differences in pathway activity were evaluated using GSVA scores for each patient with advanced (n = 32) and mild liver fibrosis (n = 40). The analysis reports t-values derived from linear models. (E) mRNA expression of target genes in liver tissue from patients with F0–F1 or F3–F4 fibrosis, extracted from GEO dataset GSE49541. (F, G) UMAP visualization of cell clusters. (H) Analysis of target genes in the hepatocytes from human normal liver (n = 2) and MASLD cirrhotic liver (n = 2), extracted from scRNA-seq dataset GSE174748. (I) Heatmap of target genes in MASLD/MASH patients (GSE202379). (J) Scheme depicting the regulation of ECM1 expression in hepatocytes in healthy and diseased liver (figure created with BioRender.com). p-values were calculated by unpaired Student’s t test. Bars represent the mean ± SD. ∗p <0.05, ∗∗p <0.01, ∗∗∗p <0.001.