ADAMTS18-fibronectin interaction regulates the morphology of liver sinusoidal endothelial cells

Abstract

Liver sinusoidal endothelial cells (LSECs) have a unique morphological structure known as "fenestra" that plays a crucial role in liver substance exchange and homeostasis maintenance. In this study, we demonstrate that ADAMTS18 protease is primarily secreted by fetal liver endothelial cells. ADAMTS18 deficiency leads to enlarged fenestrae and increased porosity of LSECs, microthrombus formation in liver vessels, and an imbalance of liver oxidative stress. These defects worsen carbon tetrachloride (CCl4)-induced liver fibrosis and diethylnitrosamine (DEN)/high-fat-induced hepatocellular carcinoma (HCC) in adult Adamts18-deficient mice. Mechanically, ADAMTS18 functions as a modifier of fibronectin (FN) to regulate the morphological acquisition of LSECs via the vascular endothelial growth factor A (VEGFA) signaling pathways. Collectively, a mechanism is proposed for LSEC morphogenesis and liver homeostasis maintenance via ADAMTS18-FN-VEGFA niches.

Keywords: Biochemistry; Biological sciences; Cell biology; Molecular biology.

Graphical abstract

Figure 1

Spatiotemporal expression of Adamts18 mRNA during mouse liver development (A) ISH of wild-type embryos and liver tissues for Adamts18. ISH-positive signals are shown as pink dots in cells (red arrows). Scale bar, 100 μm. (B) Uniform manifold approximation and projection (UMAP) visualization of Adamts18 expression in cells from embryonic (E)11.5-E14.5 fetal liver. (C) Violin plot of Adamts18 expression levels in different cells (stromal cell, megakaryocytes, and endothelial cell). y axis indicates log-normalized expression. (D) Real-time qPCR for Adamts18 expression. Data are expressed as mean ± SD (n = 3/time point). (E) Schematic representation of Adamts18 mRNA expression levels at different stages of liver development in mice. Specifically, hepatoblasts originate from the gastrula endoderm and form liver buds at E8.5-E10. At E12, the liver bud is infiltrated by the vitelline vein, umbilical vein, and sinus venosus. Then the endothelium differentiates into the portal vein and the sinusoids. At E14–E15, LSECs become fenestrated to accommodate the hematopoietic function of the fetal liver through structural differentiation. At E18, central vein formation, functional differentiation allows LSECs to adapt to the mature liver function and microenvironment.

Figure 2

Effects of ADAMTS18 deficiency on liver development in mice (A) Representative images of liver in 8-week-old adult mice. The upper parts are the dorsal and ventral views. The lower parts show the liver lobes of dorsal view. LLL, left lateral lobe; ML, middle lobe; CL, caudate lobe; RUL, right upper lobe; RLL, right lower lobe. Red arrows point out the edge of the left lateral lobe, blue arrows indicate cracks in the middle lobe. Scale bar, 5 mm. (B) Ratio of liver weight to body weight at 3 and 6 weeks (W) of age. Data are represented as mean ± SD (n = 9∼11/time point). (C) Hematoxylin and eosin (H&E)-stained of mouse liver sections (n = 3/time point). Blue arrows indicate monocytes. Red arrows indicate aggregated Kupffer cells. PV, portal vein; CV, central vein. Scale bar, 100 μm. (D) Quantification of aggregated Kupffer cell clumps from liver tissues of 2-week-old mice in panel C. Data are represented as mean ± SD (n = 3). (E) Vascular casting of the portal vein in the left lateral lobe of 8-week-old mice. Blue, green, yellow, and white marks primary, secondary, tertiary and fourth vessel, respectively. Scale bar, 5 mm. (F) Quantification of different vascular branches and branching points in panel E by ImageJ. Data are represented as mean ± SD (n ≥ 10). Each dot or square represents one individual. Ns, no significance; ∗p < 0.05; two-tailed Student’s t test. See also Figure S1.

Figure 3

Enlarged LSEC fenestrae in Adamts18^-/- mice (A) TEM of LSEC morphology in 2-week-old Adamts18^+/+ and Adamts18^−/− mice. Green arrows indicate fenestrae. Red square marks lymphocytes infiltration. Yellow arrows indicate vacuolization of hepatocytes. Hep, hepatocyte; L(Sinusoid), lumen of the sinusoid; Lym, lymphocyte. Scale bar, 1 μm. (B) SEM of LSEC fenestrae morphology in 2-week (W)-old and 3-month (M)-old mice. Red arrows: fenestra. Yellow circles: sieve plate. Scale bar, 1 μm. (C) Quantification of mean fenestra diameter, porosity and frequency of LSECs in panel C by ImageJ. Fenestra porosity is the percentage of the cell surface area covered by fenestra. Fenestra frequency is the number of fenestrae per area. Each dot or square represents one block. Each liver section was analyzed for 3 fields. Data are represented as mean ± SD (n = 3/time point). (D and E) Left panel: oil red O-stained frozen mouse liver sections at 2-week (W)-old (D) and 3-month (M)-old (E). Scale bar, 100 μm. Right panel: Quantification of relative area of oil red O staining by ImageJ. Each dot or square represents one individual. Each liver section was analyzed for 5 fields. Data are represented as mean ± SD (n = 3/time point). Ns, no significance, ∗p < 0.05, ∗∗p < 0. 01; ∗∗∗p < 0. 001; ∗∗∗∗p < 0.0001; two-tailed Student’s t test. See also Figure S2.

Figure 4

Tortuous liver sinusoidal circulation and liver micro-thrombosis in Adamts18^−/− mice (A) Multiphoton microscopy of liver sinusoids in 2-week-old mice. The green luminescent area represents the liver sinusoids, and the darker dots represent the red blood cells (white arrows). The ratio of the actual walking distance of the liver sinusoid to the straight walking distance is the tortuosity of the liver sinusoid (solid white lines). Scale bar, 50 μm. (B) Quantification of diameter (left panel), tortuosity (middle panel) and relative density (right panel) of liver sinusoids in panel A by ImageJ. Data are represented as mean ± SD (n = 3). (C) Representative images of Evans blue stained mouse liver sections at two weeks of age. Scale bar, 50 μm. (D) Immunofluorescence of CD11b (red) and CD41 (green) in the liver vessels of 2-week-old mice, DAPI (blue). Scale bar, 50 μm. (E and F) The level of lipid peroxidase (LPO) (E) and superoxide dismutase (SOD) (F) in the liver of 2-week-old mice determined by colorimetric method. Data are represented as mean ± SD (n = 8). Each dot or square represents one individual. Ns, no significance, ∗p < 0.05, ∗∗p < 0. 01; two-tailed Student’s t test.

Figure 5

Altered liver ECM compositions in Adamts18^-/- livers (A) Upper panel: representative immunohistochemistry images of fibronectin (FN) in 2-week-old mouse livers. PV, portal vein; CV, central vein. Lower panel: Quantification of FN density in different hepatic vascular areas by ImageJ. Each liver section was analyzed for 5 fields. Data are represented as mean ± SD (n = 3). Scale bar, 100 μm. (B) Upper panel: representative images of Masson’s trichrome staining of collagen (Col) in 2-week-old mouse livers. Lower panel: immunohistochemistry images of Laminin in 2-week-old mouse livers. Scale bar, 100 μm. (C) Quantification of collagen area and laminin density in panel B by ImageJ. Each liver section was analyzed for 5 fields. Data are represented as mean ± SD (n = 3). (D) The levels of liver FN, Col I, III, IV, and laminin determined by sandwich ELISA. Data are represented as mean ± SD (n = 5∼7). (E) Real-time qPCR results of liver ECM molecules (Fn1, Fbn1, Fbn2, Col1a1, Col1a2, Col3a1, Col4a1, Lama1, Lama5, Lamb1, and Lamc1). Data are represented as mean ± SD (n = 3). (F) Representative immunofluorescence images of colocalization of FN (green) and VEGFA (red) in the liver of 2-week-old mice. Yellow arrows indicate that FN and VEGFA are co-located around blood vessels. Scale bar, 100 μm. Each dot or square represents one individual. Ns, no significance, ∗p < 0.05, ∗∗p < 0.01; two-tailed Student’s t test.

Figure 6

Enhanced VEGFA and Rho signaling pathway in LSECs from Adamts18^−/− mice (A) Immunofluorescence staining for CD14 in primary LSECs from 2-week-old Adamts18^+/+ and Adamts18^−/− mice. Scale bar, 50 μm. (B) Representative western blots for VEGFA, pSrc, Src, RhoA-GTP, RhoA, pYAP, YAP, pMLC, and MLC in primary LSECs. (C) The relative quantity of VEGFA is normalized to that of GAPDH and phosphorylated proteins is normalized to that of corresponding total proteins. Data are expressed as mean ± SD (n = 3). (D) Representative immunofluorescence staining for RhoA-GTP, pYAP, and pMLC in primary LSECs. Scale bar, 10 μm. (E) Quantification of the fluorescence intensity in panel D by ImageJ. Each LSEC slide was analyzed for 5 fields. Data are represented as mean ± SD (n = 3). Each dot or square represents one individual. Ns, no significance, ∗p < 0.05, ∗∗p < 0.01; two-tailed Student’s t test.

Figure 7

Interaction of ADAMTS18 and FN (A) Representative immunofluorescence images of colocalization of ADAMTS18 (green) and FN (red) in subcutaneous tumors. Scale bar, 100 μm. (B) The basic domain organization of the ADAMTS18 and its major functional groups. ∗ Indicates the putative region of ADAMTS18 binding FN. (C) The location and function of 3CAL (94-159aa), 2CG7(93-182aa), 2CG6 (93–182), 3ZRZ (94-182aa), 4PZ5 (94-182aa), 2RKZ (183-271aa), 2RKY (93-182aa), 2RL0 (184-272aa), 6MFA (906-1268aa), 1FNF (1173-1540aa), 4LXO (1448-1631aa), 2CK2 (1538-1633aa), and 1FNA (1543-1633aa) in FN. (D) The interface area and ΔiG (kcal/mol) of different protein fragments of FN and ADAMTS18 GRAMM-X docking models analyzed by PDBePISA. The red dashed boxes indicate the proper models. (E) Two proper docking models of ADAMTS18 (green) and FN fragments (blue) generated by GRAMM-X. Interaction analysis is visualized by PyMol tool. The potential hydrogen bonding interactions (purple dashed lines) are shown in the expanded view. The upper model reveals that residues Ser48, Ser51, Ser55, Ser102, Ser192, Ser1029, Gln1030, Cys39, Gly53, and Asp59 of ADAMTS18 form hydrogen bonds with Gly144, Lys143, Thr147, Trp146, Pro126, Thr122, Gly131, Thr74, Thr65, Arg101, Glu128, and Gln115 in FN fragment (2CG7). The following model reveals that residues Arg20, Gly21, Leu22, Gly24, Gly26, Arg27, Ala31, Cys35, Lys945, His989, and Ala1022 of ADAMTS18 form hydrogen bonds with Pro87, Ser89, Leu8, Val10, Leu62, Val29, Lys86, Ala13, Thr16, Ser53, and Thr49 in FN fragment (1FNF). (F) CoIP analysis using cellular protein extracts from HEK293T cells with co-expression and separate expression of ADAMTS18 and protein fragment of FN. Mouse immunoglobulin G (IgG) agarose gel and anti-Flag agarose gel were used for immunoprecipitation. Anti-Flag and anti-His antibody recognizing Flag-ADAMTS18, His-FN were used for immunoblotting. See also Figure S3.

Figure 8

Increased susceptibility of Adamts18^−/− mice to carbon tetrachloride (CCl4)-induced liver fibrosis and diethylnitrosamine (DEN)-induced hepatocellular carcinoma (HCC) (A) Mode diagram of CCl4-induced liver fibrosis in 8-week-old male mice. Oliver oil acts as control. (B) Upper panel: representative micrographs of α-smooth muscle actin (α-SMA) staining in control (a, b) and CCl4 treatment groups (c, d). Masson’s trichrome staining of mouse livers with CCl4 treatment (e, f). Scale bar, 100 μm. Lower panel: quantification of fibrosis in CCl4 treatment mouse livers. Data are represented as mean ± SD (n = 3). Each dot or square represents one individual. Each liver section was analyzed for 3 fields. (C) Induction of hepatocellular carcinoma (HCC) formation using DEN treatment in 2-week-old male mice. Mice were subsequently given high-fat diet at 5W and sacrificed at 9M. (D) Representative images of liver in 9-month-old DEN treatment mice (a, b). Red arrows point out surface nodules. Scale bar, 5 mm. H&E-stained of mouse liver sections (c, d). Black arrows indicate nodules in the deep liver tissues. Scale bar, 100 μm. (E) Analysis of tumor burden in DEN/high-fat-treated mice. Data are represented as mean ± SD (n = 8∼9). (F) Correlation of ADAMTS18, 13, 15, 16 levels with the overall survival rate of human liver cancer patients in the KMplot dataset. ∗p < 0.05, ∗∗p < 0.01, ∗∗∗p < 0.001; two-tailed Student’s t test. See also Figure S4.