© 2021. Published by The Company of Biologists Ltd | Disease Models & Mechanisms (2021) 14, dmm048884. doi:10.1242/dmm.048884 RESEARCH ARTICLE Synergic effect of atorvastatin and ambrisentan on sinusoidal and hemodynamic alterations in a rat model of NASH Miren Bravo1, Imma Raurell1,2, Aurora Barberá 1, Diana Hide1,2, Mar Gil1, Federico Estrella1, Marı́a Teresa Salcedo3, Salvador Augustin1,2, Joan Genescà1,2,* and Marı́a Martell1,2,* ABSTRACT INTRODUCTION In non-alcoholic steatohepatitis (NASH), decreased nitric oxide and The prevalence of non-alcoholic fatty liver disease (NAFLD) is increased endothelin-1 (ET-1, also known as EDN1) released by increasing worldwide in parallel with other metabolic epidemic sinusoidal endothelial cells (LSEC) induce hepatic stellate cell (HSC) disorders, including obesity and type 2 diabetes mellitus (Younossi, contraction and contribute to portal hypertension (PH). Statins 2019). Accordingly, there is an emerging consensus for NAFLD to improve LSEC function, and ambrisentan is a selective endothelin- be defined as MAFLD (Eslam et al., 2020). A subtype of MAFLD, receptor-A antagonist. We aimed to analyse the combined effects of characterized as non-alcoholic steatohepatitis (NASH), is a potentially atorvastatin and ambrisentan on liver histopathology and progressive liver disease that can lead to fibrosis and cirrhosis, hemodynamics, together with assessing the underlying mechanism with an increased risk of portal hypertension (PH), hepatocellular in a rat NASH model. Diet-induced NASH rats were treated with carcinoma and death (Lindenmeyer and McCullough, 2018). Still, the atorvastatin (10 mg/kg/day), ambrisentan (30 mg/kg/day or 2 mg/kg/ mechanisms regulating the development and progression of PH in day) or a combination of both for 2 weeks. Hemodynamic parameters the specific context of NASH are not completely understood. were registered and liver histology and serum biochemical In cirrhosis, sinusoidal architecture becomes grossly distorted, determinations analysed. Expression of proteins were studied by leading to increased intrahepatic vascular resistance (IHVR) (Bosch immunoblotting. Conditioned media experiments were performed et al., 2020). However, increasing clinical and experimental with LSEC. HSCs were characterized by RT-PCR, and a collagen evidence indicates that PH may develop in the early stages of lattice contraction assay was performed. Atorvastatin and NAFLD when fibrosis is far less advanced or absent (Francque ambrisentan act synergistically in combination to completely et al., 2010; García-Lezana et al., 2018; Mendes et al., 2012). normalize liver hemodynamics and reverse histological NASH by Non-parenchymal liver cells contribute to the disruption of 75%. Atorvastatin reversed the sinusoidal contractile phenotype, thus sinusoidal homeostasis, increasing IHVR in NASH. Among improving endothelial function, whereas ambrisentan prevented the these cells, liver sinusoidal endothelial cells (LSECs) become contractile response in HSCs by blocking ET-1 response. dysfunctional and acquire a vasoconstrictor phenotype, releasing Additionally, ambrisentan also increased eNOS (also known as decreased levels of vasodilators, such as nitric oxide (NO), and Nos3) phosphorylation levels in LSEC, via facilitating the stimulation increased levels of vasoconstrictors, such as endothelin-1 (ET-1, of endothelin-receptor-B in these cells. Furthermore, the serum also known as EDN-1) (Bravo et al., 2019; Pasarín et al., 2012). alanine aminotransferase of the combined treatment group Diminished NO production allows hepatic stellate cell (HSC) decreased to normal levels, and this group exhibited a restoration activation (DeLeve et al., 2008). In this transdifferentiation, HSCs of the HSC quiescent phenotype. The combination of atorvastatin and acquire a myofibroblast-like phenotype, with acquisition of alpha- ambrisentan remarkably improves liver histology and PH in a diet- smooth muscle actin increasing their contractility, which may induced NASH model. By recovering LSEC function, together with impede sinusoidal flow, augmenting intrahepatic resistance and inhibiting the activation and contraction of HSC, this combined portal pressure (PP) (Rockey et al., 1992, 1993). treatment may be an effective treatment for NASH patients. Statins have been shown to improve hepatic endothelial dysfunction (Bosch et al., 2020; Pose et al., 2019). Specifically, KEY WORDS: Ambrisentan, Atorvastatin, Endothelin-1, Hepatic chronic treatment with atorvastatin lowered PP by decreasing stellate cells, Liver sinusoidal endothelial cells, NAFLD-NASH intrahepatic resistance via the activation of eNOS/NO signaling in different experimental models of cirrhosis (Bravo et al., 2019; 1 Rodríguez et al., 2017; Trebicka et al., 2007). Disease Models & Mechanisms Liver Unit, Department of Internal Medicine, Hospital Universitari Vall d’Hebron, Vall d’Hebron Institut de Recerca (VHIR), Vall d’Hebron Barcelona Hospital Another key pathway regulating the HSC contractile phenotype Campus, Universitat Autònoma de Barcelona, Barcelona 08035, Spain. 2Centro de is endothelin signaling. Activated HSCs markedly upregulate Investigació n Biomé dica en Red de Enfermedades Hepá ticas y Digestivas endothelin receptors, suggesting an increased sensitivity to ET-1 (CIBEREHD), Instituto de Salud Carlos III, Madrid 28029, Spain. 3Department of Pathology, Hospital Universitari Vall d’Hebron, Vall d’Hebron Barcelona Hospital signal (Yokomori et al., 2001). ET-1 receptor A (ETA) stimulation Campus, Barcelona 08035, Spain. enhances HSC contraction and proliferation, whereas ET-1 receptor B (ETB) elicits anti-proliferative activity in HSCs, and induces eNOS *Authors for correspondence ( jgenesca@vhebron.net; maria.martell@vhir.org) activation in LSECs (Liu et al., 2003; Mallat et al., 1996). Furthermore, M.B., 0000-0001-5889-9125; S.A., 0000-0002-3515-9033; M.M., 0000-0002- HSCs experience a significant increase in their sensitivity to ET-1- 0935-0029 induced ETA stimulation during transdifferentiation (Reinehr, 2002). This is an Open Access article distributed under the terms of the Creative Commons Attribution This indicates that ETA antagonism would be more advantageous than License (https://creativecommons.org/licenses/by/4.0), which permits unrestricted use, ETB blocking when trying to reduce PP (Feng et al., 2009). distribution and reproduction in any medium provided that the original work is properly attributed. Ambrisentan, an ETA selective antagonist, has been previously Handling Editor: Monica J. Justice shown to moderately improve PH (Zipprich et al., 2016). However, Received 16 December 2020; Accepted 23 March 2021 there is some discrepancy between studies, which probably lies in 1 RESEARCH ARTICLE Disease Models & Mechanisms (2021) 14, dmm048884. doi:10.1242/dmm.048884 the different models, doses or routes of administration used (Pitts, ambrisentan groups also showed a significant increase in PBF 2009). compared to the HFGFD-Veh group. In the present study, we investigated whether the co-administration The reduction in IHVR of the HFGFD-Amhi group was of both drugs, atorvastatin and ambrisentan, improves NASH by significantly greater than that of the HFGFD-At group. However, restoring sinusoidal microcirculation. Specifically, we tested the in this did not translate into a greater effect on PP, probably due to the vivo effect of atorvastatin and ambrisentan on a rat model of NASH increase in PBF produced by ambrisentan at high dose. with PH and explored their underlying mechanisms on hepatic Ambrisentan dose reduction, both in monotherapy and in sinusoidal cells. combination, maintained the beneficial effects, significantly decreasing PP (HFGFD Amlo, 10.9%, P=0.004; HFGFD-AtAmlo, RESULTS 16.1%, P<0.001), together with IHVR compared to the vehicle. But, Treatment effect on body weight gain in addition, it prevented portal flow increase, so that both HFGFD- High fat glucose-fructose diet (HFGFD)-fed groups showed Amlo and HFGFD-AtAmlo groups did not show differences in PBF significantly greater body weight gain compared to the control values compared to the HFGFD-Veh group. Therefore, we used the diet (CD)-fed group after 8 weeks of each diet. Weight gain during lower dose for the histological and molecular analysis. the next 2 weeks of treatment did not differ between any group of The combination of atorvastatin and low-dose ambrisentan treated and untreated rats, with the exception of the group treated achieved a greater general improvement in liver hemodynamics with high-dose ambrisentan combined with atorvastatin (HFGFD- than that obtained with each of the drugs alone. As a consequence, AtAmhi), in which body weight gain was significantly lower than the HFGFD-AtAmlo group showed PP, PBF and IHVR values that of the vehicle group (HFGFD-Veh, 74% lower increase, comparable to those of the CD group (Table 1). P=0.001), but also to that of the atorvastatin (HFGFD-At, 60% lower increase, P=0.047) and high-dose ambrisentan (HFGFD- Combination treatment reverses histological NASH Amhi, 60% lower increase, P=0.036)-treated groups. Furthermore, HFGFD caused histological NASH, defined as the concurrence of despite the hypercaloric diet, the HFGFD-AtAmhi group gained steatosis, hepatocellular ballooning and lobular inflammation 59% less body weight than the CD group during the treatment (Fig. 2; Fig. S1). HFGFD-fed animals did not develop fibrosis weeks (Fig. 1). Animals treated with low-dose ambrisentan either (Fig. S2). alone (HFGFD-Amlo) or in combination with atorvastatin (HFGFD- Hepatic steatosis, analysed by the CRN system, persisted in AtAmlo) gained weight similar to the other groups during the most groups after the different treatments, although the combination treatment weeks (Fig. 1). of atorvastatin and ambrisentan markedly reduced the percentage of individuals with this feature (Fig. 2A,F). Moreover, Oil Red O Effect of treatments on biochemical parameters staining showed a significant decrease in the percentage of the There was a complete absence of both liver and muscular toxicity in steatosis area with all treatments. In the case of the combined all studied groups, considering that no animal exceeded alanine treatment, the percentage of steatosis was similar to that of the aminotransferase (ALT) and creatine kinase (CK) values shown by control group (Fig. 2E,F). Hepatocellular ballooning improved the group that received the vehicle treatment. Even with the significantly in the HFGFD-AtAmlo group (Fig. 2B,F), and the administration of high-dose ambrisentan there were no signs of liver lobular inflammation score improved with atorvastatin treatment toxicity, and its combination with atorvastatin did not produce liver both in monotherapy and in combination with ambrisentan or muscle toxicity either (Fig. 1). (Fig. 2C,F). A significant increase in serum ALT levels was observed in All three parameters correlated with serum transaminase levels HFGFD-Veh rats compared to CD animals. This elevation was in individual samples. Although the correlation between the abolished with all treatment groups, as well as with a reduction in steatosis score and ALT levels was not significant, Oil Red O the dose of ambrisentan (Fig. 1). staining did significantly correlate in these same individuals Table S1 shows serum biochemical parameters analysed in all (Fig. S3). samples. The combination of ambrisentan and atorvastatin lowered Accordingly, all animal groups receiving treatment showed lower serum cholesterol levels compared to the CD group, but not in NAFLD activity score (NAS) values than those achieved by the comparison with the HFGFD-Veh group. vehicle group. NAS also significantly correlated with serum ALT levels (Fig. S3E). Furthermore, both atorvastatin and ambrisentan Combined treatment markedly improves liver reduced the percentage of rats with histological NASH, although hemodynamics in rats with NASH only the combination of both achieved a superior histopathological Disease Models & Mechanisms Systemic circulation was not affected by the diet and neither by improvement, reversing histological NASH in 75% of individuals different treatments at any doses. (Fig. 2D). As expected, the PP was higher in HFGFD-Veh rats than in CD rats (27.8% increase, P<0.001). This was accompanied by a Atorvastatin improves insulin sensitivity significant increase in IHVR and decreased portal blood flow Fasting insulin levels increased significantly in the HFGFD-fed (PBF) (Table 1). In these rats, atorvastatin monotherapy caused a group, as well as insulin resistance [homeostatic model assessment significant decrease in PP (HFGFD-At, 12.8%, P=0.001), associated of insulin resistance (HOMA-IR)]. Atorvastatin treatment alone or with a moderate reduction in IHVR, whereas PBF remained in combination (HFGFD-At and HFGFD-AtAmlo) significantly unchanged. reduced serum insulin levels, and both groups showed a remarkable Treatments with high-dose ambrisentan, alone or in combination, insulin sensitivity recovery, although only the group with the achieved a significant decrease in PP (HFGFD-Amhi, 9.5%, combination achieved a statistically significant decrease in the P=0.011; HFGFD-AtAmhi, 15.7%, P<0.001) compared to the HOMA-IR index. Otherwise, in the HFGFD-Amlo group, both HFGFD-Veh group. This reduction in PP was due to a marked insulin levels and the HOMA-IR index were similar to those of the decrease in IHVR produced by ambrisentan. However, high-dose vehicle group (Table 2). 2 RESEARCH ARTICLE Disease Models & Mechanisms (2021) 14, dmm048884. doi:10.1242/dmm.048884 Disease Models & Mechanisms Fig. 1. Body weight gain, ALT and CK values. (A) Body weight gain at indicated week relative to the starting weight. (B) Starting body weight. (C) Body weight gain during 8 weeks of diet. (D) Body weight gain during 2 weeks of treatment. (E) End body weight. (F) Serum ALT levels. (G) Serum CK levels. Data are mean±s.e.m. CD (n=7); HFGFD-Veh (vehicle; n=6); HFGFD-At (10 mg/kg/day atorvastatin; n=8); HFGFD-Amhi (30 mg/kg/day ambrisentan; n=7); HFGFD-AtAmhi (10 mg/kg/day atorvastatin and 30 mg/kg/day ambrisentan; n=8); HFGFD-Amlo (2 mg/kg/day ambrisentan; n=7); HFGFD-AtAmlo (10 mg/kg/ day atorvastatin and 2 mg/kg/day ambrisentan; n=8). *P<0.05, **P<0.01, ***P<0.001 versus CD; ##P<0.01 versus ###P<0.001 HFGFD-Veh; †P<0.05 versus HFGFD-At; §P<0.05 versus HFGFD-Amhi (one-way ANOVA). Recovery of the intrahepatic vasoactive mediators by Akt and eNOS (also known as Nos3) phosphorylation reduction in treatment combination HFGFD-Veh rats compared to the CD group (Fig. 3B,C). These Western blot analysis revealed a decreased intrahepatic protein results suggest microvascular dysfunction as the underlying expression of Kruppel-like factor 2 (KLF2) (Fig. 3A), together with mechanism of increased IHVR in the model. Atorvastatin 3 RESEARCH ARTICLE Disease Models & Mechanisms (2021) 14, dmm048884. doi:10.1242/dmm.048884 Table 1. Hemodynamic studies SMABF SMAR PBF IHVR MAP (mmHg) (ml/[min·100 g]) (mmHg/ml·min·100 g) PP (mmHg) (ml/[min·100 g]) (mmHg/ml·min·100 g) CD (n=7) 122±4.13 3.20±0.29 36.70±2.52 8.45±0.25 2.99±0.13 2.86±0.14 HFGFD-Veh (n=6) 125.67±9.06 3.02±0.42 40.51±4.17 10.8±0.12*** 2.3±0.18* 4.83±0.37*** HFGFD-At (n=8) 113.95±6.6 3.61±0.51 32.96±5.56 9.42±0.27*## 2.22±0.15** 4.37±0.31*** HFGFD-Amhi (n=7) 118.86±5.04 3.58±0.37 32.75±3.81 9.77±0.32**# 3.5±0.25###††† 2.86±0.18###††† HFGFD-AtAmhi (n=8) 128.57±6.11 4.15±0.4 31.58±4.57 9.11±0.25### 3.05±0.18##†† 3.04±0.16###††† HFGFD-Amlo (n=7) 118.43±2.67 2.97±0.4 40.81±5.81 9.62±0.33**## 2.72±0.22 3.69±0.36*## HFGFD-AtAmlo (n=8) 122.10±3.02 3.77±0.35 32.36±3.89 9.06±0.16### 2.82±0.17† 3.31±0.23###†† Data are mean±s.e.m. n, number of rats; MAP, mean arterial pressure; SMABF, superior mesenteric artery blood flow; SMAR, superior mesenteric artery resistance; HFGFD-Veh, vehicle; HFGFD-At, 10 mg/kg/day atorvastatin; HFGFD-Amhi, 30 mg/kg/day ambrisentan; HFGFD-AtAmhi, 10 mg/kg/day atorvastatin and 30 mg/kg/day ambrisentan; HFGFD-Amlo, 2 mg/kg/day ambrisentan; HFGFD-AtAmlo, 10 mg/kg/day atorvastatin and 2 mg/kg/day ambrisentan. *P<0.05, **P<0.01, ***P<0.001 versus CD; #P<0.05, ##P<0.01, ###P<0.001 versus HFGFD-Veh; †P<0.05, ††P<0.01, †††P<0.001 versus HFGFD-At (one-way ANOVA). treatment increased KLF2, phosphorylated endothelial nitric oxide in cells isolated from all treatment groups, but especially in those synthase (P-eNOS) and phosphorylated protein kinase B (P-Akt) from HFGFD-At and HFGFD-AtAmlo animals, the contraction liver content (Fig. 3A-C). Ambrisentan administration significantly percentage of which was comparable to that shown by CD group increased eNOS and Akt phosphorylation (Fig. 3A, Fig. 2B), but cells. did not affect the expression of KLF2 (Fig. 3C). As expected, the In vitro treatment with exogenous ET-1 generated a superior combination of both treatments increased KLF2, P-eNOS and P-Akt contraction of the collagen gel in HSCs of HFGFD-Veh rats, intrahepatic levels (Fig. 3A-C). whereas cells of CD rats maintained their low percentage of Moreover, intrahepatic ET-1 expression was significantly contraction. Results showed increased contraction of HSCs from elevated in HFGFD-Veh rats compared to the CD group. In HFGFD-At in response to ET-1, whereas cells from HFGFD-Amlo HFGFD-At and HFGFD-AtAmlo groups, the protein expression exhibited a completely blocked ET-1-mediated contractile response. levels of this vasoconstrictor decreased significantly, but not in HSCs from the HFGFD-AtAmlo group showed an almost identical HFGFD-Amlo group (Fig. 3D). percentage of collagen gel contraction to those of the control group, Fig. S4 shows the complete western blot images used for the both in standard conditions and after ET-1 treatment (Fig. 5). quantification of the protein immunoblots. Combination treatment reverses HSC pro-contractile and Ambrisentan increases eNOS activation in LSEC pro-fibrogenic profile As ambrisentan has not yet been reported to increase eNOS ET-1 mRNA levels remained unchanged in HSCs from HFGFD- phosphorylation levels, we decided to corroborate this result directly Veh rats compared to the control group. Ambrisentan treatment in LSECs isolated from ambrisentan-treated rats. These cells significantly decreased ET-1 gene expression in these cells showed a significant increase in the expression of P-eNOS (Fig. 6A). Concerning ET-1 receptors, HFGFD-Veh-isolated compared to LSEC isolated from the vehicle group (Fig. 4A). HSCs showed increased levels of both receptor types, ETA and To further understand the ambrisentan mechanism for inducing ETB, compared to HSCs from the CD group (Fig. 6B,C). All eNOS activation, we decided to analyse the ET-1 pathway in treatment groups showed a significant reduction in ETA mRNA LSECs. Exposure of freshly isolated LSECs to ET-1 led to a levels (Fig. 6B). However, ETB expression was only significantly significant increase in eNOS phosphorylation (Fig. 4B). Besides, downregulated in HSCs from the HFGFD-AtAmlo group compared blockade of the ETB receptor with the specific receptor antagonist to those of the HFGFD-Veh group (Fig. 6C). BQ-788 inhibited ET-1-mediated eNOS phosphorylation, verifying HSCs isolated from HFGFD-Veh exhibited increased expression that in LSECs, ET-1-mediated eNOS activation derives from ETB of αSMA (also known as Acta2), Col1a1 and desmin mRNA, signaling (Fig. 4B). suggesting an activated phenotype in these cells (Fig. 6D-F). HSCs ET-1-induced eNOS activation was abolished when LSECs were isolated from both atorvastatin- and ambrisentan-treated animals exposed to conditioned medium from HSCs previously treated with showed a significant reduction in αSMA and Col1a1 levels, and a ET-1, showing no increases in P-eNOS levels in LSECs (Fig. 4C). marked desmin downregulation. The combined treatment achieved a However, when HSCs were previously treated with the ETA specific superior amelioration of the HSC phenotype, reducing significantly Disease Models & Mechanisms receptor antagonist BQ-123, ET-1 remained capable of inducing the levels of these three activation-associated markers (Fig. 6D-F). eNOS phosphorylation in LSECs exposed to HSC-conditioned The expression levels of platelet-derived growth factor receptor β medium. As expected, LSEC pre-treatment with BQ-788 reverted (PDGFR-β) and matrix metalloproteinase 2 (MMP2) in HSCs were eNOS activation in this condition (Fig. 4C). not modified in the HFGFD-Veh group compared to the CD group. However, cells from the HFGFD-Amlo and HFGFD-AtAmlo groups Atorvastatin reduces HSC contractility and ambrisentan exhibited significantly lower levels of both genes compared with blocks ET-1 induced contraction HSCs from the HFGFD-Veh group (Fig. 6G,H). To investigate whether IHVR reduction obtained by oral treatments is due to HSC relaxation, we performed contraction assays in the DISCUSSION basal situation and in response to ET-1, with HSCs isolated from the In this study, we showed that the combination of atorvastatin and different groups (Fig. 5). ambrisentan normalizes liver hemodynamics, reducing IHVR and In basal conditions, HSCs isolated from the HFGFD-Veh group PP in rats with histologically diagnosed NASH. showed a significantly higher percentage of contraction than cells PH and its derived complications represent the main cause of liver from the CD group. The contractile capacity significantly decreased failure and transplantation in patients with advanced chronic liver 4 RESEARCH ARTICLE Disease Models & Mechanisms (2021) 14, dmm048884. doi:10.1242/dmm.048884 Disease Models & Mechanisms Fig. 2. Histological evaluation of NASH activity following the NASH-Clinical Research Network system. (A-C) Bar graphs represent the percentage of individuals presenting steatosis (A), ballooning (B) and inflammation (C) in CD (n=7; HFGFD-Veh, n=6; HFGFD-At, n=8; HFGFD-Amlo, n=7; and HFGFD- AtAmlo, n=8). Each colour represents the percentage of individuals with the corresponding score (score 1, gray; score 2, black). (D) Bar graphs show the percentage of individuals with histological NASH represented with different NAS scores (score 3, light gray; score 4, dark gray; or score 5, black). (E) Bar graph representing the quantification of liver lipid by Oil Red O staining in liver sections. (F) Representative images (20× magnification) of liver sections stained with hematoxylin-eosin and Oil Red O used to perform the histological evaluation. *P<0.05; **P<0.01; ***P<0.001 versus CD; #P<0.05; ##P<0.01; ### P<0.001 versus HFGFD-Veh (Mann–Whitney U-test). 5 RESEARCH ARTICLE Disease Models & Mechanisms (2021) 14, dmm048884. doi:10.1242/dmm.048884 Table 2. Insulin resistance studies that evidence a lowered PP mediated by a significant n Insulina ng/ml HOMA-IR upregulation of KLF2 protein expression and eNOS phosphorylation in livers from atorvastatin-treated animals (Bravo et al., 2019; CD 7 4,47±0,68 1,39±0,17 HFGFD-Veh 6 12,29±2,26** 4,48±0,69** Rodríguez et al., 2017; Trebicka et al., 2007). On the other hand, HFGFD-At 8 6,76±1,27# 2,56±0,71 insulin also plays a very important role in the regulation of HFGFD-Amlo 7 12,59±2,93** 4,51±1,12** intrahepatic eNOS activity through the IRS/PI3K/Akt pathway HFGFD-AtAmlo 8 6,74±0,57# 2,04±0,29# (García-Lezana et al., 2018; Pasarín et al., 2011). Our model Data are mean±s.e.m. n, number of rats. **P<0.01 versus CD; #P<0.05 versus evidenced insulin resistance improvement by atorvastatin treatment, HFGFD-Veh. HOMA-IR (one-way ANOVA). suggesting that atorvastatin-induced P-Akt increase is, at least partially, mediated through the insulin vasodilator response diseases (García-Pagán et al., 2012). However, PH is not solely the recovery. However, we cannot rule out other mechanisms, as consequence of cirrhosis, as sinusoidal microvascular dysfunction statins have been shown to promote insulin-independent Akt- contributes also to increase IHVR, and consequently, sinusoidal mediated eNOS phosphorylation (Trebicka and Schierwagen, 2015; portal pressure, which may impact disease progression in MAFLD Trebicka et al., 2007). (Ryou et al., 2020). Consistent with these data, our NASH model Atorvastatin-induced endothelial function restoration was presents PH, together with a marked microvascular dysfunction accompanied by a lower production of ET-1 at the intrahepatic characterized by decreased eNOS activation in LSECs and level in our model. Our data support the hypothesis that atorvastatin increased HSC contraction. decreases HSC contraction and consequently reduces IHVR Decreased eNOS activity is probably the result of reduced (Trebicka et al., 2007). In accordance, our NASH model showed expression of KLF2, a transcription factor that confers endothelial enhanced HSC contraction, correlated with increased αSMA ( pro- protection in response to shear stress, inducing NO production contractile marker) expression, and this was also reversed by oral (Marrone et al., 2015). Our results are in agreement with previous administration of atorvastatin. Disease Models & Mechanisms Fig. 3. Intrahepatic expression of vasoactive modulators. (A-D) Bar graphs show the quantification of KLF2 (A), P-Akt/Akt (B), P-eNOS/eNOS (C) and ET-1 (D) using GAPDH as a loading control in CD, n=7; HFGFD-Veh, n=6; HFGFD-At, n=8; HFGFD-Amlo, n=7; and HFGFD AtAmlo, n=8. Protein levels are normalized with the HFGFD-Veh group and expressed as mean±s.e.m. Repesentative western blot bands are shown below. *P<0.05 versus CD; #P<0.05, ## P<0.01, ###P<0.001 versus HFGFD-Veh (unpaired two-tailed Student’s t-test). 6 RESEARCH ARTICLE Disease Models & Mechanisms (2021) 14, dmm048884. doi:10.1242/dmm.048884 Fig. 4. Effect of ET-1 in eNOS activation in LSECs. (A-C) Bar graphs show quantification of P- eNOS/eNOS ratio from LSECs (A) isolated from HFGFD-Veh (n=3) and HFGFD-Amlo (n=3) normalized to the HFGFD-Veh group. (B) LSECs treated with BQ-788 (1 µM) or vehicle for 15 min and exposed to ET-1 (10 nM) for 30 min. (C) LSECs treated with BQ-788 (1 µM) or vehicle for 15 min, and exposed for 30 min to conditioned medium from HSCs treated with BQ-123 (10 µM) or vehicle for 10 min, and incubated with ET-1 (10 nM) for 10 min. The immunoblot shown is representative of three different experiments, each performed with cells from a different isolation. *P<0.05; **P<0.01 (unpaired two-tailed Student’s t-test). On the other hand, ambrisentan also showed to be effective in These results reinforce the idea that ambrisentan is safe and not reducing PP. Endothelin receptor antagonists (ERAs) have been hepatotoxic. Nonetheless, more studies are needed to validate its indicated for the treatment of pulmonary arterial hypertension. safety in more advanced liver disease. Currently, several studies have demonstrated the ability of these Animals treated with atorvastatin and ambrisentan combined drugs to modulate liver hemodynamics in cirrhotic models, showed decreased serum cholesterol levels compared to the CD although with discordant results due to the great variation in ERA group. This suggests that the combined treatment might be type and administered dose (Cavasin et al., 2010; Feng et al., 2009; beneficial in controlling dyslipidaemia in NASH patients. Disease Models & Mechanisms Kojima et al., 2000). Our results demonstrate clearly that However, we cannot fully affirm this effect as our model does not ambrisentan at 2 mg/kg/day shows significant intrahepatic show hypercholesterolemia. beneficial effects. Increasing the dose does not improve the effect, In the case of ambrisentan treatment, PP reduction was but leads to the appearance of unwanted side effects, such as an accompanied by the improvement of the HSC phenotype by increase of PBF. blocking the ETA receptor, which is exclusively found in these Long-term administration of some ERAs, such as bosentan, have cells. Ambrisentan also reduced the contraction of HSC but, in been repeatedly associated with acute and severe liver injury particular, it completely inhibited their exacerbated contractile (Humbert et al., 2007). Yet, ambrisentan has not been associated response to exogenous ET-1. Thus, ambrisentan might directly with hepatotoxicity (Galiè et al., 2008; Kenna et al., 2015), not even decrease intrahepatic vasoconstriction response and thus decrease in patients who had previously discontinued treatment with intrahepatic vascular tone. bosentan due to alterations in aminotransferase levels (McGoon As expected, there was no improvement in insulin sensitivity or et al., 2009). In accordance with these data, none of the animals in increased KLF2 expression in ambrisentan-treated animals. However, our study showed liver toxicity, not even those receiving the high we observed increased intrahepatic eNOS phosphorylation levels. dose of ambrisentan, and all treatments normalized ALT levels. We even corroborated this increased P-eNOS level in LSECs freshly 7 RESEARCH ARTICLE Disease Models & Mechanisms (2021) 14, dmm048884. doi:10.1242/dmm.048884 Fig. 5. HSC contraction on collagen gel lattices. Contraction is expressed as the percentage of initial gel area as mean±s.e.m. All data are from experiments using three collagen lattices for each condition with HSCs isolated from CD, n=3; HFGFD-Veh, n=4; HFGFD-At, n=5; HFGFD-Amlo, n=4; and HFGFD AtAmlo, n=4. *P<0.05; ***P<0.001 versus CD; #P<0.05; ## P<0.01; ###P<0.001 versus HFGFD-Veh (one-way ANOVA). isolated from in vivo ambrisentan-treated rats. In the present study, we response when incubated with exogenous ET-1. These results demonstrate for the first time that ambrisentan has an indirect effect suggest an important role for HSCs and their response to ET-1 in the on LSECs, increasing eNOS activation. ETA receptor blockade in pathophysiology of PH in NASH. HSCs by ambrisentan increases the bioavailability of ET-1, which The combination of atorvastatin and ambrisentan was the only binds to ETB receptors located in LSECs. As described by others, and treatment capable of significantly reducing the expression levels of as we demonstrate in this study, ET-1 stimulation of ETB receptors in both ETA and ETB receptors. This shows once again a close LSECs increases eNOS phosphorylation and, thus, NO synthesis, relationship between HSC activation and the ET-1 system. Both leading to vasodilatation (Liu et al., 2003). Therefore, ambrisentan atorvastatin and ambrisentan obtained similar effects with respect to might indirectly contribute to sinusoidal microvascular function improving HSC phenotype, as indicated by expression reduction of improvement. activation markers (αSMA, Col1a1 and PDGFR-β). However, only Consistent with the results obtained from the atorvastatin the combination of both significantly reduced desmin expression, or ambrisentan monotherapies, combination treatment showed demonstrating a synergistic effect of both drugs on HSCs. synergistic benefits. Thus, a clear restoration of sinusoidal ET-1 has a prominent contractile effect on HSCs, which can microvascular function, together with significant IHVR and PP contribute to PH, but also promotes their proliferation and migration reduction, was exhibited in the double-treated rats. in early stages of the disease (Friedman, 2008). Our results show HSC contractile response to ET-1 has been shown to increase that treatment with ambrisentan inhibits the expression of MMP-2 Disease Models & Mechanisms with the progression of liver injury, being proportional to its induced by ET-1. This effect was described previously by in vitro activation degree (Kawada et al., 1993; Rockey and Weisiger, treatment of activated HSCs with BQ-123 (Koda et al., 2006). Here, 1996). In accordance, our NASH model showed increased levels of we confirm that this also occurs with in vivo treatment, suggesting intrahepatic ET-1, and an aggravated HSC contraction. Moreover, that decreasing MMP2 expression by ETA antagonisms could HSCs from our model had increased gene expression levels of both prevent the degradation of the normal subendothelial matrix and its ET-1 receptors, which would explain the exacerbated contraction subsequent replacement by a non-functional extracellular interstitial when incubated with this vasoconstrictor. HSCs isolated from matrix. These findings show the need for additional research in more atorvastatin-treated rats also exhibited this increased contraction in advanced models of NASH to analyse the effect of the combination response to exogenous ET-1. However, the HSCs of rats treated with of these drugs on fibrogenesis. We clearly acknowledge that our ambrisentan did not increase their contraction when incubated with NASH model lacks fibrosis generation and this is a limitation of our ET-1, demonstrating that much of the contraction exerted by the study. HSCs in response to ET-1 is via ETA. The combination of both Finally, the combination treatment markedly improved liver drugs maintained HSC contractility indistinguishable from that histopathology. Both drugs decreased steatosis area, but especially, from healthy animals, and like these, they did not show a contractile individuals from the HFGFD-AtAmlo group showed an Oil Red 8 RESEARCH ARTICLE Disease Models & Mechanisms (2021) 14, dmm048884. doi:10.1242/dmm.048884 Fig. 6. Expression of pro-contractile and pro-fibrotic markers in HSCs. Relative quantification of mRNA expression of endothelin-1 (ET-1), endothelin receptor A (ETA), endothelin receptor B (ETB), alpha-smooth muscle actin (αSMA), collagen type I alpha 1 (Col1a1), desmin, platelet-derived growth factor receptor-beta (PDGFR-β) and matrix metalloproteinase-2 (MMP2) by qRT-PCR in HSCs isolated from CD (n=4), HFGFD-Veh (n=4), HFGFD-At (n=5), HFGFD-Amblo (n=4) and HFGFD-AtAmlo (n=5). GAPDH was used as an endogenous control and the results were normalized to HSCs from CD. Data are mean±s.e.m. *P<0.05; **P<0.01 versus CD; #P<0.05; ##P<0.01 versus HFGFD-Veh (one-way ANOVA). O-stained area similar to that of the CD group. Atorvastatin, alone histology and PH in the early stages of NASH, and consequently, it or in combination, significantly improved lobular inflammation. might prevent disease progression. These findings support the Disease Models & Mechanisms However, ballooning improved significantly only with the idea that this combination could be a safe and effective treatment combined treatment. So, despite the fact that all treatments for patients with NASH. The potential long-term use of the reduced the NAS score, the combination achieved a superior combination treatment explored here in patients with NAFLD/ improvement, reversing histological NASH in 75% of individuals. NASH is very attractive for slowing the progression of the disease, Although we hypothesize that atorvastatin and ambrisentan act considering the low cost of the drugs, a very low toxicity profile and directly in LSECs and HSCs, and that eventually these effects might possible effects in human NAFLD physiopathology. Before that, indirectly collaborate in improving liver histology, a direct effect on first, proof of a beneficial effect in fibrotic NASH models would be hepatocyte function cannot be ruled out. Furthermore, there may be required and, from there, a demonstrated clinical effect in a clinical off-target effects involving other cell types, which brings out the trial would also be needed. need for additional studies to determine these complex interactions. In conclusion, this study suggests that the combination of MATERIALS AND METHODS atorvastatin and ambrisentan normalizes intrahepatic vascular tone, Animal model recovering LSEC function, together with inhibiting the proliferation All procedures were conducted in accordance with European Union and contraction of HSCs. This turns into amelioration of liver Guidelines for Ethical Care of Experimental Animals (EC Directive 86/ 9 RESEARCH ARTICLE Disease Models & Mechanisms (2021) 14, dmm048884. doi:10.1242/dmm.048884 609/EEC for animal experiments) and approved (file number: 10,989) by NASH was made as per current standards, based on the concurrence of the Animal Care Committee of the Vall d’Hebron Institut de Recerca steatosis, ballooning and inflammation, and a NAS≥3. (VHIR, Barcelona, Spain) and conducted in the animal facilities of VHIR. Male Sprague-Dawley OFA rats (Charles River Laboratories, L’Arbresle, Oil red O staining France) weighing 200-220 g were used for our previously described diet- Liver samples sections were frozen and sectioned into 8 μm slices, fixed in induced NASH model (García-Lezana et al., 2018). Rats were housed under 4% formaldehyde and stained with Oil Red O for 10 min. Lipid droplets 12h light/dark cycle at constant temperature and humidity. They were fed ad were quantified in five images (magnification 20×) of each section using Fiji libitum for 8 weeks with a HFGFD or CD. The HFGFD consisted of 30% software (Schindelin et al., 2012). fat (butter, coconut oil, palm oil and beef tallow), with mainly saturated fatty acids (5.73 Kcal/g), supplemented with cholesterol (1g/Kg) (Ssniff LSEC and HSC isolation Spezialdiaten GmbH, Soest, Germany), and a beverage of glucose-fructose LSECs and HSCs were isolated from rat livers as described previously (42g/L, 45% glucose-55% fructose). The CD consisted of a grain-based (Marrone et al., 2013). chow that consisted of 4.8% fat (3.43 Kcal/g) (Safe-150, SAFE, Augy, France) and tap water. Body weight and food consumption were monitored weekly. LSEC isolation LSECs were isolated from HFGFD-Veh (n=3) and HFGFD-Amlo (n=3) rat livers. Livers were perfused with collagenase through the portal vein for Drug administration/treatments 10 min at a flow rate of 20 ml/min at 37°C with Hanks’ balanced salt Eight-week HFGFD-fed rats received daily oral doses of the corresponding solution (HBSS) without calcium and magnesium, containing 12 mM drug or vehicle for 2 weeks. Atorvastatin (10 mg/kg/day, Almirall, HEPES ( pH 7.4), 0.6 mM EGTA and 1.6% bovine serum albumin. The Barcelona, Spain) (HFGFD-At), 30 mg/kg/day (HFGFD-Amhi) or 2 mg/ livers were then perfused through the portal vein for 30 min at a flow rate of kg/day (HFGFD-Amlo) ambrisentan (GlaxoSmithKline, Dublin, Ireland), a 5 ml/min at 37°C with 0.01% collagenase A, HBSS containing 12 mM combination of the same dose of atorvastatin with both doses of ambrisentan HEPES ( pH 7.4) and 4 mM CaCl2, and then excised and in vitro digested (HFGFD-AtAmhi and HFGFD-AtAmlo) or equivalent volume of water for 10 min at 37°C, also with the same buffer. Resulting cells were filtered (HFGFD-Veh) were administered by gastric gavage. Each group continued through a 100 µm nylon filter, collected in cold Krebs buffer and centrifuged to have access to the original diet during the entire treatment period. at 50 g for 5 min to eliminate hepatocytes. The supernatant was then centrifuged at 800 g for 10 min, and the pellet was resuspended in ice-cold Biochemical parameters PBS and centrifuged in a two-phase Percoll gradient (25%/50%). The Blood samples were collected from the cava vein after completing the central fraction containing LSECs and Kupffer cells (KCs) was collected, hemodynamic study. Glucose, creatinine, bilirubin, aspartate washed with PBS, resuspended in LSEC medium [RPMI with 10% fetal aminotransferase, ALT, alkaline phosphatase (ALP), CK, total cholesterol, bovine serum (FBS), 2 mM L-glutamine, 1% penicillin/streptomycin, 1% high-density lipoprotein (HDL), low density lipoprotein (LDL), amphotericin B, 0.1 mg/ml heparin and 0.05 mg/ml endothelial cell growth triglycerides and albumin were measured using standard methods at the supplement (ECGS)] and seeded in a non-coated plate for 30 min at 37°C Hospital Vall d’Hebron CORE lab. Insulin was measured using an ELISA (5% CO2). KCs attached to the plate were discarded and the non-adherent kit (EMD Millipore, Billerica, MA, USA). Insulin resistance was estimated LSECs were seeded in collagen-coated culture plates, incubated for 45 min by applying the homeostasis model of insulin resistance index (HOMA-IR): (37°C, 5% CO2) and washed afterwards. The resulting cells were incubated at 37°C in a humid atmosphere with 5% CO2 in LSEC medium (RPMI Fasting insulinðng=mlÞ  Fasting glucoseðmg=dlÞ HOMA  IR ¼ : with 10% FBS, 2 mM L-glutamine, 1% penicillin/streptomycin, 1% 405 amphotericin B, 1% heparin and 1% ECGS). Hepatic and muscular toxicity due to statin treatment was defined based on ALT and CK levels in vehicle rats. HSC isolation HSCs were isolated from CD-Veh (n=4), HFGFD-Veh (n=4), HFGFD-At Hemodynamic measurements (n=5), HFGFD-Amlo (n=4) and HFGFD-AtAmlo (n=5) rat livers. Briefly, Ninety minutes after the last treatment administration, fasted rats were livers were perfused with pronase, collagenase and DNase through the portal intraperitoneally anesthetized with ketamine hydrochloride (100 mg/kg) vein for 10 min at a flow rate of 20 ml/min at 37°C with Gey’s balanced salt plus midazolam (5 mg/kg), and body temperature was maintained at 37°C solution (GBSS). The livers were then perfused through the portal vein for for continuous recording of hemodynamic parameters. Mean arterial 30 min at a flow rate of 5 ml/min at 37°C, excised and in vitro digested for pressure (MAP, mmHg) was measured by catheterization ( polyethylene 10 min at 37°C, also with pronase, collagenase and Dnase. The resulting PE-catheter, PE50) of the femoral artery, and PP (mmHg) was assessed by cells were filtered and centrifuged at 50 g for 5 min to eliminate the ileocolic vein catheterization using highly sensitive pressure transducers hepatocytes. The supernatant was then centrifuged at 800 g for 10 min and (Harvard Apparatus, Holliston, MA, USA). Superior mesenteric artery the pellet was resuspended in ice-cold GBSS and centrifuged in Optiprep Disease Models & Mechanisms blood flow (SMABF, ml/[min×100 g]) and PBF (ml/[min×100 g]) were gradient (11%). The fraction containing HSCs was collected, washed with measured using a perivascular ultrasonic transit-time flow probe (1 mm GBSS, resuspended in HSC medium (Iscove’s modified Dulbecco’s diameter, Transonic Systems Inc, Ithaca, NY, USA). Superior mesenteric medium with 10% FBS, 2 mM L-glutamine, 1% penicillin/streptomycin artery resistance (SMAR, mmHg/ml×min×100 g) and IHVR (mmHg/ and 1% amphotericin B) and seeded in a non-coated plate at 37°C (5% CO2) ml×min×100 g) were calculated as [(MAP-PP)/SMABF] and (PP/PBF), overnight and then washed. respectively. Western blot analysis Histological analysis Livers were perfused with saline for exsanguination and samples were Hematoxylin-eosin staining directly frozen in liquid nitrogen, crushed to powder and homogenized in Liver samples were fixed in 4% formaldehyde for 24 h, embedded in Triton X-100-lysis buffer [25 mM Tris/HCl ( pH 7.6), 137 mM NaCl, paraffin and sectioned in 4 μm thick slides. Hematoxylin-eosin was used to 2.7 mM KCl, 20 mM NaF, 10 mM Na4P2O7, 10 nM okadaic acid, 2 mM assess liver. Samples were evaluated by an expert liver pathologist blinded Na3VO4, 2 µg/ml antipain, 2 µg/ml aprotinin, 2 µg/ml chymostatin, 2 µg/ml to animal interventions. leupeptin, 2 µg/ml pepstatin A, 2 µg/ml trypsin inhibitor, 40 µg/ml NAS was used to quantify NAFLD activity, obtained from the unweighted phenylmethylsulfonylfluoride, and 10% (v/v) Triton X-100]. LSECs were sum of the histological components: steatosis (0-3), lobular inflammation washed with PBS and lysed using Triton X-100-lysis buffer. Homogenized (0-3) and hepatocellular ballooning (0-2) (Kleiner et al., 2005). Table S2 livers and cell lysates were sonicated and centrifuged at 14,000 g at 4°C for shows CRN quantification system definitions. The diagnosis of histological 10 min. Supernatant protein concentration was assessed using a BCA 10 RESEARCH ARTICLE Disease Models & Mechanisms (2021) 14, dmm048884. doi:10.1242/dmm.048884 Protein Assay Kit (Thermo Fisher Scientific, Rockford, IL, USA). Equal non-parametric post-hoc test according to variance homogeneity amounts of protein were run on a 10% SDS-PAGE. Proteins were blotted determined by Levene’s test. P<0.05 was considered statistically significant. onto a polyvinylidene difluoride membrane (Thermo Fisher Scientific, Waltham, MA, USA). Membranes were blocked in 5% phosphoblocker Competing interests (Cell Biolabs, San Diego, CA, USA) and incubated with the relevant Salvador Augustin has received consulting fees from Gilead, IQVIA, Intercept, primary antibody (Table S3) overnight at 4°C. Then, membranes were Novartis and Pfizer; speaking fees from Allergan, Gilead, MSD, Menarini and Novartis; and research grants from Gilead. incubated with the corresponding secondary peroxidase-coupled antibody for 1 h at room temperature, developed using an ECL kit (GE Healthcare) Author contributions and quantified by Image Studio Lite (Lincoln, NE, USA). Glyceraldehyde- Conceptualization: J.G., M.M.; Methodology: M.B.; Validation; J.G., M.M., S.A.; 3-phosphate dehydrogenase (GAPDH, 1/5000, Ambion, Austin, TX, USA) Investigation: M.B., I.R., A.B., D.H., M.G., F.E., M.T.S.; Writing - original draft: M.B.; was used as loading control. Writing - review & editing: M.M., J.G.; Supervision: J.G., M.M.; Project When detecting the phosphorylated and the total protein in the same administration: I.R.; Funding acquisition: J.G., M.M., S.A. membrane (for Akt and eNOS), the detection of the phosphorylated protein was performed first. Membranes were then stripped by incubating them with Funding Restore WB Stripping Buffer (Thermo Fisher Scientific, Rockford, IL, This work was supported by the Instituto de Salud Carlos III (ISCIII) and co-funded USA) for 30 min at 55°C. Thereafter, they were blocked again and incubated by the European Union (European Regional Development Fund/European Social with the primary and secondary antibody as described above to detect total Fund, ‘Investing in your future’) [PI17/00310 to S.A., PI17/00754 to M.M., and PI18/ protein levels. In this case, the ratio between the intensities of the 00947 and AC18/00033 (ENM3 2018) to J.G.]. J.G. is a recipient of a clinical intensification award and D.H. is a recipient of a Sara Borrell grant (both from ISCIII). phosphorylated and the total protein (P-Akt/Akt or p-eNOS/eNOS) were M.G. and A.B. have predoctoral fellowships from the Agència de Gestió d’Ajuts calculated without requiring GAPDH normalization. Universitaris i de Recerca and ISCIII, respectively. Centro de Investigació n Biomé dica en Red de Enfermedades Hepá ticas y Digestivas (CIBERehd) is Primary LSEC culture with conditioned medium from HSCs supported by ISCIII. Endothelin-1-related paracrine effects between LSECs and HSCs were determined through a conditioned medium study (Dirscherl et al., 2020). References HSCs were isolated from healthy rats and allowed to attach overnight in 12- Bosch, J., Gracia-Sancho, J. and Abraldes, J. G. (2020). Cirrhosis as new indication for statins. Gut 69, 953-962. doi:10.1136/gutjnl-2019-318237 well plates at confluency. Culture medium was replaced by fresh medium Bravo, M., Raurell, I., Hide, D., Ferná ndez-Iglesias, A., Gil, M., Barberá , A., (serum free RPMI), and incubated for 10 min with the endothelin receptor A Salcedo, M. T., Augustin, S., Genescà, J. and Martell, M. (2019). Restoration of specific antagonist BQ-123 (10 µM) or vehicle. Then, ET-1 (10 nM) was liver sinusoidal cell phenotypes by statins improves portal hypertension and added to the medium and incubated for 10 min more. After incubation, the histology in rats with NASH. Sci. Rep. 9, 20183. doi:10.1038/s41598-019-56366-2 conditioned medium was transferred to overnight serum-starved LSECs Cavasin, M. A., Semus, H., Pitts, K., Peng, Y., Sandoval, J., Chapo, J. and Plato, (isolated from healthy rats and seeded in 12-well plates at confluency) C. F. (2010). Acute effects of endothelin receptor antagonists on hepatic previously incubated with endothelin receptor B-specific antagonist BQ- hemodynamics of cirrhotic and noncirrhotic rats. Can. J. Physiol. Pharmacol. 88, 636-643. doi:10.1139/Y10-038 788 (1 µM) or vehicle for 15 min, and incubated for 30 min. DeLeve, L. D., Wang, X. and Guo, Y. (2008). Sinusoidal endothelial cells prevent rat stellate cell activation and promote reversion to quiescence. Hepatology 48, HSC collagen gel contraction assay 920-930. doi:10.1002/hep.22351 Twenty four-well plates were used to examine the contractile capacity of Dirscherl, K., Schlä pfer, M., Roth Z’graggen, B., Wenger, R. H., Booy, C., Flury- Frei, R., Fatzer, R., Aloman, C., Bartosch, B., Parent, R. et al. (2020). Hypoxia HSCs as described previously with slight modifications (Rockey and sensing by hepatic stellate cells leads to VEGF-dependent angiogenesis and may Weisiger, 1996). In brief, 1.5 mg/ml collagen (Collagen R Solution 0.2% contribute to accelerated liver regeneration. Sci. Rep. 10, 4392. doi:10.1038/ SERVA, Heidelberg, Germany) gels were prepared using 1 N NaOH for pH s41598-020-60709-9 adjustment. An aliquot of 500 µl of the solution was added to each well and Eslam, M., Newsome, P. N., Sarin, S. K., Anstee, Q. M., Targher, G., Romero- incubated at 37°C for at least 1 h to allow gelatinization. Freshly isolated Gomez, M., Zelber-Sagi, S., Wai-Sun Wong, V., Dufour, J. F., Schattenberg, HSCs (2×106) from the study groups were seeded on each gel in 1 ml of J. M. et al. (2020). A new definition for metabolic dysfunction-associated fatty liver disease: an international expert consensus statement. J. Hepatol. 73, 202-209. medium and incubated overnight. Cells were starved for 1 h, and the doi:10.1016/j.jhep.2020.03.039 medium was then replaced by medium with 10% FBS, or medium with 10% Feng, H.-Q. Q., Weymouth, N. D. and Rockey, D. C. (2009). Endothelin FBS and 100 nM ET-1. The tip of a 200 μl pipette was used to gently detach antagonism in portal hypertensive mice: Implications for endothelin receptor- the gel from the plates. After incubation for 24 h, the areas of the gels were specific signaling in liver disease. Am. J. Physiol. Gastrointest. Liver Physiol. 297, measured using Fiji software (Schindelin et al., 2012). Triplicates of each G27-G33. doi:10.1152/ajpgi.90405.2008 condition were made, with cells isolated from at least three rats from each Francque, S., Verrijken, A., Mertens, I., Hubens, G., Van Marck, E., Pelckmans, P., Van Gaal, L. and Michielsen, P. (2010). Noncirrhotic human nonalcoholic study group. fatty liver disease induces portal hypertension in relation to the histological degree of steatosis. Eur. J. Gastroenterol. Hepatol. 22, 1449-1457. doi:10.1097/MEG. Disease Models & Mechanisms RNA extraction and gene expression 0b013e32833f14a1 RNA extraction was performed with HSCs isolated from the study groups Friedman, S. L. (2008). Hepatic stellate cells: protean, multifunctional, and enigmatic cells of the liver. Physiol. Rev. 88, 125-172. doi:10.1152/physrev. using an RNeasy Mini Kit (Qiagen, Venlo, The Netherlands) following the 00013.2007 manufacturer’s instructions and reverse transcribed to cDNA (High Galiè, N., Olschewski, H., Oudiz, R. J., Torres, F., Frost, A., Ghofrani, H. A., Capacity cDNA Reverse Transcription Kit, Thermo Fisher Scientific). Badesch, D. B., McGoon, M. D., McLaughlin, V. V., Roecker, E. B. et al. (2008). cDNA was added to Taqman universal PCR master mix plus the specific Ambrisentan for the treatment of pulmonary arterial hypertension. Circulation 117, probe (Table S4) and loaded into 384-well plates (Thermo Fisher Scientific). 3010-3019. doi:10.1161/CIRCULATIONAHA.107.742510 qRT-PCR was performed using a 7900HT Fast Real-Time PCR system Garcı́a-Lezana, T., Raurell, I., Bravo, M., Torres-Arauz, M., Salcedo, M. T., (Thermo Fisher Scientific). The relative gene expression was normalized to Santiago, A., Schoenenberger, A., Manichanh, C., Genescà, J., Martell, M. et al. (2018). Restoration of a healthy intestinal microbiota normalizes portal GAPDH. Data were analysed using the Relative Quantification qPCR hypertension in a rat model of nonalcoholic steatohepatitis. Hepatology 67, Application in Thermo Fisher Cloud. 1485-1498. doi:10.1002/hep.29646 Garcı́a-Pagá n, J.-C. C., Gracia-Sancho, J. and Bosch, J. (2012). Functional Statistical analysis aspects on the pathophysiology of portal hypertension in cirrhosis. J. Hepatol. 57, 458-461. doi:10.1016/j.jhep.2012.03.007 Statistical analysis was performed using the IBM SPSS 20.0 statistical Humbert, M., Segal, E. S., Kiely, D. G., Carlsen, J., Schwierin, B. and Hoeper, package (IBM, New York, USA). Data are reported as mean±s.e.m. and M. M. (2007). Results of European post-marketing surveillance of bosentan in statistical significance was determined using Student’s t-test, U-Mann– pulmonary hypertension. Eur. Respir. J. 30, 338-344. doi:10.1183/09031936. Whitney test or one-way ANOVA, followed by a parametric or 00138706 11 RESEARCH ARTICLE Disease Models & Mechanisms (2021) 14, dmm048884. doi:10.1242/dmm.048884 Kawada, N., Tran–Thi, T., Klein, H. and Decker, K. (1993). The contraction of Pasarı́n, M., Abraldes, J. G., Rodrı́guez-Vilarrupla, A., La Mura, V., Garcı́a- hepatic stellate (Ito) cells stimulated with vasoactive substances: possible Pagá n, J. C. and Bosch, J. (2011). Insulin resistance and liver microcirculation in involvement of endothelin 1 and nitric oxide in the regulation of the sinusoidal a rat model of early NAFLD. J. Hepatol. 55, 1095-1102. doi:10.1016/j.jhep.2011. tonus. Eur. J. Biochem. 213, 815-823. doi:10.1111/j.1432-1033.1993.tb17824.x 01.053 Kenna, J. G., Stahl, S. H., Eakins, J. A., Foster, A. J., Andersson, L. C., Bergare, Pasarı́n, M., La Mura, V., Gracia-Sancho, J., Garcı́a-Calderó , H., Rodrı́guez- J., Billger, M., Elebring, M., Elmore, C. S. and Thompson, R. A. (2015). Multiple Vilarrupla, A., Garcı́a-Pagá n, J. C., Bosch, J. and Abraldes, J. G. (2012). compound-related adverse properties contribute to liver injury caused by Sinusoidal endothelial dysfunction precedes inflammation and fibrosis in a model endothelin receptor antagonists. J. Pharmacol. Exp. Ther. 352, 281-290. doi:10. of NAFLD. PLoS ONE 7, e32785. doi:10.1371/journal.pone.0032785 1124/jpet.114.220491 Pitts, K. R. (2009). Endothelin receptor antagonism in portal hypertension. Expert Kleiner, D., Brunt, E., Van Natta, M., Behling, C., Contos, M., Cummings, O., Opin. Investig. Drugs 18, 135-142. doi:10.1517/13543780802680186 Ferrell, L., Liu, Y., Torbenson, M., Unalp-Arida, A. et al. (2005). Design and Pose, E., Trebicka, J., Mookerjee, R. P., Angeli, P. and Ginès, P. (2019). Statins: validation of a histological scoring system for nonalcoholic fatty liver disease. Old drugs as new therapy for liver diseases? J. Hepatol. 70, 194-202. doi:10.1016/ Hepatology 41, 1313-1321. doi:10.1002/hep.20701 j.jhep.2018.07.019 Koda, M., Bauer, M., Krebs, A., Hahn, E. G., Schuppan, D. and Murawaki, Y. Reinehr, R. (2002). Regulation of endothelin-A receptor sensitivity by cyclic (2006). Endothelin-1 enhances fibrogenic gene expression, but does not promote adenosine monophosphate in rat hepatic stellate cells. Hepatology 36, 861-873. DNA synthesis or apoptosis in hepatic stellate cells. Comp. Hepatol. 5, 5. doi:10. doi:10.1016/S0270-9139(02)00092-7 1186/1476-5926-5-5 Rockey, D. C. and Weisiger, R. A. (1996). Endothelin induced contractility of Kojima, H., Yamao, J., Tsujimoto, T., Uemura, M., Takaya, A. and Fukui, H. stellate cells from normal and cirrhotic rat liver: implications for regulation of portal (2000). Mixed endothelin receptor antagonist, SB209670, decreases portal pressure and resistance. Hepatology 24, 233-240. doi:10.1002/hep.510240137 pressure in biliary cirrhotic rats in vivo by reducing portal venous system Rockey, D. C., Boyles, J. K., Gabbiani, G. and Friedman, S. L. (1992). Rat hepatic resistance. J. Hepatol. 32, 43-50. doi:10.1016/S0168-8278(00)80188-9 lipocytes express smooth muscle actin upon activation in vivo and in culture. Lindenmeyer, C. C. and McCullough, A. J. (2018). The natural history of J. Submicrosc. Cytol. Pathol. 24, 193-203 Rockey, D. C., Housset, C. N. and Friedman, S. L. (1993). Activation-dependent nonalcoholic fatty liver disease—an evolving view. Clin. Liver Dis. 22, 11-21. contractility of rat hepatic lipocytes in culture and in vivo. J. Clin. Invest. 92, doi:10.1016/j.cld.2017.08.003 1795-1804. doi:10.1172/JCI116769 Liu, S., Premont, R. T., Kontos, C. D., Huang, J. and Rockey, D. C. (2003). Rodrı́guez, S., Raurell, I., Torres-Arauz, M., Garcı́a-Lezana, T., Genescà, J. and Endothelin-1 activates endothelial cell nitric-oxide synthase via heterotrimeric G- Martell, M. (2017). A nitric oxide-donating statin decreases portal pressure with a protein betagamma subunit signaling to protein jinase B/Akt. J. Biol. Chem. 278, better toxicity profile than conventional statins in cirrhotic rats. Sci. Rep. 7, 40461. 49929-49935. doi:10.1074/jbc.M306930200 doi:10.1038/srep40461 Mallat, A., Pré aux, A. M., Serradeil-Le Gal, C., Raufaste, D., Gallois, C., Brenner, Ryou, M., Stylopoulos, N. and Baffy, G. (2020). Nonalcoholic fatty liver disease D. A., Bradham, C., Maclouf, J., Iourgenko, V., Fouassier, L. et al. (1996). and portal hypertension. Explor. Med. 1, 149-169. doi:10.37349/emed.2020. Growth inhibitory properties of endothelin-1 in activated human hepatic stellate 00011 cells: a cyclic adenosine monophosphate-mediated pathway: Inhibition of both Schindelin, J., Arganda-Carreras, I., Frise, E., Kaynig, V., Longair, M., Pietzsch, extracellular signal-regulated kinase and c-Jun kinase and upregulation of T., Preibisch, S., Rueden, C., Saalfeld, S., Schmid, B. et al. (2012). Fiji: an endothelin B receptors. J. Clin. Invest. 98, 2771-2778. doi:10.1172/JCI119103 open-source platform for biological-image analysis. Nat. Methods 9, 676-682. Marrone, G., Russo, L., Rosado, E., Hide, D., Garcı́a-Cardeña, G., Garcı́a- doi:10.1038/nmeth.2019 Pagá n, J. C., Bosch, J. and Gracia-Sancho, J. (2013). The transcription factor Trebicka, J. and Schierwagen, R. (2015). Statins, Rho GTPases and KLF2: new KLF2 mediates hepatic endothelial protection and paracrine endothelial-stellate mechanistic insight into liver fibrosis and portal hypertension. Gut 64, 1349-1350. cell deactivation induced by statins. J. Hepatol. 58, 98-103. doi:10.1016/j.jhep. doi:10.1136/gutjnl-2014-308800 2012.08.026 Trebicka, J., Hennenberg, M., Laleman, W., Shelest, N., Biecker, E., Schepke, Marrone, G., Maeso-Dı́az, R., Garcı́a-Cardena, G., Abraldes, J. G., Garcı́a- M., Nevens, F., Sauerbruch, T. and Heller, J. (2007). Atorvastatin lowers portal Pagá n, J. C., Bosch, J. and Gracia-Sancho, J. (2015). KLF2 exerts antifibrotic pressure in cirrhotic rats by inhibition of RhoA/Rho-kinase and activation of and vasoprotective effects in cirrhotic rat livers: Behind the molecular endothelial nitric oxide synthase. Hepatology 46, 242-253. doi:10.1002/hep. mechanisms of statins. Gut 64, 1434-1443. doi:10.1136/gutjnl-2014-308338 21673 McGoon, M. D., Frost, A. E., Oudiz, R. J., Badesch, D. B., Galie, N., Olschewski, Yokomori, H., Oda, M., Ogi, M., Kamegaya, Y., Tsukada, N., Nakamura, M. and H., McLaughlin, V. V., Gerber, M. J., Dufton, C., Despain, D. J. et al. (2009). Ishii, H. (2001). Enhanced expression of endothelin receptor subtypes in cirrhotic Ambrisentan therapy in patients with pulmonary arterial hypertension who rat liver. Liver 21, 114-122. doi:10.1034/j.1600-0676.2001.021002114.x discontinued bosentan or sitaxsentan due to liver function test abnormalities. Younossi, Z. M. (2019). Non-alcoholic fatty liver disease – A global public health Chest 135, 122-129. doi:10.1378/chest.08-1028 perspective. J. Hepatol. 70, 531-544. doi:10.1016/j.jhep.2018.10.033 Mendes, F. D., Suzuki, A., Sanderson, S. O., Lindor, K. D. and Angulo, P. (2012). Zipprich, A., Schenkel, E., Winkler, M. and Ripoll, C. (2016). Selective endothelin- Prevalence and indicators of portal hypertension in patients with nonalcoholic fatty a-blocker decrease portal pressure in patients with cirrhosis. A feasibility study liver disease. Clin. Gastroenterol. Hepatol. 10, 1028-1033.e2. doi:10.1016/j.cgh. combining the local intraarterial and systemic administration. Z. Gastroenterol. 54, 2012.05.008 1343-1404. doi:10.1055/s-0036-1597411 Disease Models & Mechanisms 12