Consequences of Amyloid-β Deficiency for the Liver

Abstract

The hepatic content of amyloid beta (Aβ) decreases drastically in human and rodent cirrhosis highlighting the importance of understanding the consequences of Aβ deficiency in the liver. This is especially relevant in view of recent advances in anti-Aβ therapies for Alzheimer's disease (AD). Here, it is shown that partial hepatic loss of Aβ in transgenic AD mice immunized with Aβ antibody 3D6 and its absence in amyloid precursor protein (APP) knockout mice (APP-KO), as well as in human liver spheroids with APP knockdown upregulates classical hallmarks of fibrosis, smooth muscle alpha-actin, and collagen type I. Aβ absence in APP-KO and deficiency in immunized mice lead to strong activation of transforming growth factor-β (TGFβ), alpha secretases, NOTCH pathway, inflammation, decreased permeability of liver sinusoids, and epithelial-mesenchymal transition. Inversely, increased systemic and intrahepatic levels of Aβ42 in transgenic AD mice and neprilysin inhibitor LBQ657-treated wild-type mice protect the liver against carbon tetrachloride (CCl₄)-induced injury. Transcriptomic analysis of CCl₄-treated transgenic AD mouse livers uncovers the regulatory effects of Aβ42 on mitochondrial function, lipid metabolism, and its onco-suppressive effects accompanied by reduced synthesis of extracellular matrix proteins. Combined, these data reveal Aβ as an indispensable regulator of cell-cell interactions in healthy liver and a powerful protector against liver fibrosis.

Keywords: 5xFAD; TGFβ; VEGF; eNOS; neprilysin; presenilin; β‐secretase 1.

Figure 1

In vitro effects of Aβ42 on LSEC, HSC, and hepatocytes. A) Aβ42 utilization in human LSEC (hLSEC) line measured by a decrease of Aβ42 in cell culture supernatant normalized to culture medium without cells supplemented with Aβ42 (DMEM+ Aβ); B,C) live cell imaging and quantification of FITC‐Dextran 150 kDa uptake by hLSEC line incubated 24 h with and without Aβ42 (3000 pg mL⁻¹, n = 10 per group); D) laminin 1(LAMA1) mRNA qPCR in hLSEC line (n = 3 per group); E) collagen 4a (Col4a2) mRNA qPCR in hLSEC line (n = 3 per group); F) Assessment of NO by Difluorofluorescein Diacetate (DAF) in primary hLSEC (n = 4 per group) and VEGF in primary human HSC (hHSC) by immunofluorescence staining (n = 3 per group); G) Western Blot (WB) analysis of TGFβ in hLSEC line incubated with Aβ40 or Aβ42 (1000 pg mL⁻¹ each), GAPDH served as a loading control (n = 3 per group); H) αSMA mRNA qPCR in primary hHSC ±Aβ42 (n = 4 per group); I) WB of CD31 and NOTCH1 in hLSEC line +/‐ Aβ42 (n = 4 per group); J) Hes1 mRNA qPCR in hLSEC line ±Aβ42 (n = 3 per group); K) Immunofluorescence staining of PCNA in primary murine hepatocytes ±Aβ42 (1000 pg mL⁻¹) counterstained with AFP and DAPI; L) Quantification of PCNA⁺ hepatocytes ±Aβ42 (1000 pg mL⁻¹; n = 5 per group). The data are presented as means ± SEM. ^* p < 0.05, ^** p < 0.01, ^*** p < 0.005, and ^**** p < 0.001; two‐tailed Student's t‐test (A–F,H, J, L).

Figure 2

Phenotypic changes in APP knock‐down human liver spheroids and expression of APP processing enzymes and NOTCH activation genes in human fibrotic liver. A–C), Expression of APP, αSMA, and COL1A1 are shown in control (Ctrl.) and APP knock‐down (APP‐KD) human spheroids consisting of primary hepatocytes and HSC (n≥3); D) representative immunofluorescent images of CYP3A4 and αSMA in APP‐KD and control (Ctrl.) spheroids. Note that the overlap of αSMA and CYP3A4 drastically increases in APP‐KD, indicative of epithelial‐mesenchymal transition; E) quantification of the overlap of αSMA and CYP3A4 signals using Mander‘s Overlap Coefficient; F) qPCR of ADAM10, ADAM17, ADAM9, NOTCH3, JAG1, MMP2, UCHL, BACE2, and LRP‐1 mRNA in normal (Ctrl.) and fibrotic (fibrosis) human liver tissue. ^* p < 0.05, ^** p < 0.01, and ^**** p < 0.001 using a two tailed t‐test (A–F), ns: not significant. For NOTCH3 and ADAM9 (in F) Mann–Whitney test was used.

Figure 3

Assessment of fibrotic, inflammatory, and liver sinusoidal permeability markers in APP‐KO and 3D6‐immunized 5xFAD mice. A–J) Representative images of APP‐KO versus WT and 5xFAD mouse liver sections after 8‐month immunization with 3D6 versus IgG2a control antibodies. Immunofluorescence staining of (A–D) αSMA /Col1 /DAPI; E–H) TGFβ /αSMA /DAPI; I,J) GFAP /αSMA /DAPI (n = 4 per group); K–P) Multiplex analysis of TNFα, IL‐6, IL‐13, VEGF‐A, IL‐10, and IFNγ in liver homogenates of APP‐KO versus WT (n = 5 per group) and 5xFAD mice after 8‐month immunization with 3D6 versus IgG2a control antibodies (n = 9 per group); Q) WB of eNOS and GFAP in liver homogenates of 5xFAD mice after 8‐month immunization with 3D6 versus IgG2 control antibodies (n = 3 per group); R) Quantification of eNOS+ cells in liver sections of APP‐KO versus WT mice in 10 liver slices from of n = 3 mice per group. The data are presented as means ± SEM. ^* p < 0.05, ^** p < 0.01, ^*** p < 0.005, and ^**** p < 0.001; two‐tailed Student's t‐test (K–P, R).

Figure 4

NOTCH pathway in APP‐KO and 3D6‐immunized 5xFAD mice. A–H) Representative images of APP‐KO versus WT and 5xFAD mouse livers after 8‐month immunization with 3D6 versus IgG2a control antibodies. Immunofluorescence staining of liver sections for NICD/αSMA/DAPI, (n = 4 per group); Arrowheads in G and D indicate αSMA/NICD positive binucleated cells putatively reflecting the ongoing EMT in hepatocytes; Arrows in H indicate microvessels surrounded by multiple layers of αSMA+/NICD+ HSC; I) Quantification of Hes1+ cells in liver sections of APP‐KO versus WT and 3D6 versus IgG2a treated 5xFAD mice from, n = 9 per group. J–M) qPCR of Notch 1, Adam‐10, Adam‐17, and Psen1 in liver samples of APP‐KO versus WT (n = 12 per group) and 5xFAD mice after 8‐month immunization with 3D6 versus IgG2a control antibodies (n = 6 per group); N) BACE1 protein assessed by ELISA in liver homogenates of 5xFAD mice after 8‐month immunization with 3D6 versus IgG2a control antibodies (n = 7 per group). The data are presented as means ± SEM. ^* p < 0.05, ^** p < 0.01, and ^**** p < 0.001; two‐tailed Student's t‐test (I–N).

Figure 5

Aβ protects 3xTg‐AD mice from CCl₄‐induced fibrosis. Data sets acquired from liver samples of CCl₄‐ treated 3xTg‐AD (3xTg‐CCl₄) versus BL/6 (BL/6‐CCl₄) and Corn oil treated BL/6 controls (BL/6‐Ctrl.) after 5 weeks of CCl₄ versus corn oil treatment; A–C) Representative images of Sirius Red staining (n = 4 per group); D) plasma liver enzymes (AST, ALT, and AP), n = 8 per group; E) qPCR of αSMA (acta2) mRNA (n = 8 per group); F) Multiplex Analysis of Osteopontin (OPN) and TGFβ (n = 8 per group); G–L) Immunofluorescence staining of liver sections for (G–I) NICD/ αSMA/DAPI, (n = 4 per group); J–L) glutamine synthetase (GS)/ αSMA/ DAPI, (n = 4 per group); M) Western Blot of CD31, Notch1, Hes1 (n = 3 per group); N) TNFα, IL‐6, IL‐13, VEGF‐A, IFNγ, MMP‐12, and MMP‐9 multiplex analysis of liver homogenates (n = 8 per group). The data are presented as means ± SEM. ^* p < 0.05, ^** p < 0.01, ^*** p < 0.005, and ^**** p < 0.001; One‐way ANOVA with Bonferroni”s post hoc test (D–F, N) and Kruskal–Wallis for AP analysis (in D).

Figure 6

Transcriptome analysis of CCl₄‐treated BL/6 and 3xTg‐AD mouse livers. A–C) Liver samples from CCl₄‐treated BL/6 and 3xTg and corn oil‐treated BL/6 control mice were analyzed using Clariom S arrays. A) Number of differentially expressed genes (DEGs) using RMA and limma analysis. Results were filtered for FDR p‐value <0.05 and absolute logFC >1.5. B) Heatmap of selected DEGs categorized into lipid and glucose homeostasis, hepatocellular carcinoma (HCC), inflammation, fibrosis, Aβ generation and degradation, and drug transport and metabolism; C) Microarray gene expression of fibrosis markers; D) Enriched Gene Ontology (GO) molecular function and KEGG pathways. The color legend indicates the degree of normalized enrichment score (NES). Significance is reflected by p ^**** = 10⁻⁴, ^*** = 10⁻³, ^**10⁻², ^* = 0.05, and ns = not significant.

Figure 7

Inhibition of neprilysin protects BL/6 mice from CCl₄‐induced fibrosis. A) Schematic presentation of treatment timeline: 5‐week CCl₄‐ treatment (2x/week) of BL/6 mice ± two different dosages of neprilysin inhibitor sacubitrilat (LBQ657), 5 mg or 30 mg kg⁻¹ body weight versus corn oil‐treated BL/6 controls (BL/6‐Ctrl.); B) plasma liver enzymes (AST, ALT, and AP), n = 8 per group; C) Sirius Red staining (n = 4 per group); D) Immunofluorescence staining of liver sections for Col1/ αSMA/DAPI, (n = 4 per group); E) Western Blot of αSMA and CD31 (n = 3 per group); F–I) Multiplex analysis of Osteopontin (OPN), TGFβ, TNFα, IL‐6, IL‐13, VEGF‐A, IFNγ, MMP‐12, and MMP‐9 in liver homogenates (n = 8 per group); The data are presented as means ± SEM. ^* p < 0.05, ^** p < 0.01, ^*** p < 0.005, and ^**** p < 0.001; One‐way ANOVA with Bonferroni‘s post hoc test (B, F–I) and Kruskal–Wallis test for AP analysis (in B).

Figure 8

Schematic presentation of processes regulated by Aβ in the liver. APP‐ and NOTCH‐cleaving enzyme PSEN1 favors NOTCH processing in the situation of decreased hepatic levels of Aβ during fibrosis. Left panel: neutralization of Aβ by 3D6‐antibody treatment (anti‐Aβ) or induction of liver fibrosis in wild type (WT) mice by CCl₄ or amyloid precursor protein (APP) knockout (APP‐KO) results in downregulation of Aβ in the liver leading to decreased ammonia detoxification by glutamine synthetase (GS), HSC activation, extracellular matrix (ECM) deposition, decreased liver endothelial cell (LSEC) permeability reflected by downregulation of eNOS and VEGF and an increased expression of CD31 in transgenic 5xFAD, CCl₄‐treated wild type (WT) and APP‐KO mice; knockdown of APP in human liver spheroids induces epithelial‐mesenchymal transition (EMT) of hepatocytes. Right panel: high systemic and hepatic levels of Aβ in 3xTg‐AD mice and treatment of BL/6 mice with the neprilysin inhibitor sacubitrilat provide protection against liver fibrosis, normalizing the processes presented in the right panel. High hepatic Aβ level is accompanied by an overweight of PSEN1‐mediated APP cleavage over its NOTCH‐cleaving function. The figure was created with BioRender.com.