TGR(mREN2)27 rats develop non-alcoholic fatty liver disease-associated portal hypertension responsive to modulations of Janus-kinase 2 and Mas receptor

Prevalence of non-alcoholic fatty liver disease (NAFLD) is increasing. Resulting fibrosis and portal hypertension, as a possible secondary event, may necessitate treatment. Overexpression of mouse renin in the transgenic rat model, TGR(mREN2)27, leads to spontaneous development of NAFLD. Therefore, we used TGR(mREN2)27 rats as a model of NAFLD where we hypothesized increased susceptibility and investigated fibrosis and portal hypertension and associated pathways. 12-week old TGR(mREN2)27 rats received either cholestatic (BDL) or toxic injury (CCl4 inhalation). Portal and systemic hemodynamic assessments were performed using microsphere technique with and without injection of the Janus-Kinase 2 (JAK2) inhibitor AG490 or the non-peptidic Ang(1-7) agonist, AVE0991. The extent of liver fibrosis was assessed in TGR(mREN2)27 and wild-type rats using standard techniques. Protein and mRNA levels of profibrotic, renin-angiotensin system components were assessed in liver and primary hepatic stellate cells (HSC) and hepatocytes. TGR(mREN2)27 rats developed spontaneous, but mild fibrosis and portal hypertension due to the activation of the JAK2/Arhgef1/ROCK pathway. AG490 decreased migration of HSC and portal pressure in isolated liver perfusions and in vivo. Fibrosis or portal hypertension after cholestatic (BDL) or toxic injury (CCl4) was not aggravated in TGR(mREN2)27 rats, probably due to decreased mouse renin expression in hepatocytes. Interestingly, portal hypertension was even blunted in TGR(mREN2)27 rats (with or without additional injury) by AVE0991. TGR(mREN2)27 rats are a suitable model of spontaneous liver fibrosis and portal hypertension but not with increased susceptibility to liver damage. After additional injury, the animals can be used to evaluate novel therapeutic strategies targeting Mas.


Results
TGR(mREN2)27 rats develop spontaneous fibrosis and portal hypertension, but are not more prone to cholestatic or toxic liver damage. The extent of fibrosis was assessed in TGR(mREN2)27 and wild type (WT) rats, uninjured or subjected to BDL and CCl 4 (Fig. 1). Healthy uninjured TGR(mREN2)27 rats show more positive hepatic Sirius red staining compared to respective WT rats. There was no significant difference between the extent of fibrosis in Sirius red staining in WT and TGR(mREN2)27 rats after injury (Fig. 1A, Suppl. Fig. 1A). Interestingly, in uninjured TGR(mREN2)27 rats, periportal fibrosis was observed when compared to WT rats (Fig. 1A). Levels of hydroxyproline were significantly higher in TGR(mREN2)27 rats than in WT rats without injury (Fig. 1B). Control TGR(mREN2)27 and WT animals showed similar collagen I A I (Col1a1) mRNA levels. While there was only a trend towards decreased collagen transcription in TGR(mREN2)27 rat livers two weeks after BDL, the difference to wildtype rats became significant after four weeks of BDL. Congruent to this finding, ten weeks of CCl 4 intoxication, TGR(mREN2)27 rats showed less hepatic Col1a1 mRNA expression than the respective WT rats (Fig. 1C). While healthy TGR(mREN2)27 rats had more positive alpha smooth muscle actin (aSMA) areas, marker of activation of hepatic stellate cells, in livers compared to respective WT livers, the hepatic aSMA increased much less after CCl 4 intoxication in TGR(mREN2)27 rats compared to CCl 4 intoxicated WT livers (Fig. 1D, Suppl. Fig. 1B). This was also confirmed by the total protein levels of aSMA (Suppl. Fig. 1C), but mRNA levels showed no differences between WT and TGR(mREN2)27 livers at baseline and after pathological stimuli (Fig. 1E). Already two weeks after BDL, hepatic platelet derived growth factor (PDGF)-beta receptor transcription showed a trend towards lower mRNA levels in TGR(mREN2)27 rat livers, reaching significance four weeks after BDL and in the CCl 4 model (Fig. 1F).
TGR(mREN2)27 rats showed spontaneous portal hypertension, due to increased portal pressure and hepatic vascular resistance (Table 1). However, after four weeks of BDL injury, the degree of portal hypertension in TGR(mREN2)27 rats was similar to that found in WT rats, while CCl 4 -injury produced even less portal hypertension in TGR(mREN2)27 rats due to decreased hepatic vascular resistance compared to WT rats (Table 1). While mean arterial pressure (MAP) was higher in CCl 4 intoxicated TGR(mREN2)27 rats compared to WT rats, systemic vascular resistances were not significantly different ( Table 1).
Expression of RAS components in TGR(mREN2)27 rats. Interestingly, hepatic angiotensinogen mRNA levels decreased after BDL, which was more pronounced in TGR(mREN2)27 rats ( Fig. 2A). ACE and ACE2 mRNA increased after liver injury, but to a lesser extent in TGR(mREN2)27 rat livers than in WT livers (Fig. 2B,C). Moreover, hepatic rat renin mRNA expression was down-regulated after liver injury, with down-regulation most pronounced in TGR(mREN2)27 rats after four weeks of BDL. This direction of regulation was apposed in CCl 4 TGR(mREN2)27 rat livers compared to the respective WT livers (Fig. 2D). In contrast, expression of hepatic transgene (mouse renin) was down-regulated in TGR(mREN2)27 rats (Fig. 2E), which was most likely due to the down-regulation in the hepatocytes, as shown in primary isolated hepatocytes from BDL TGR(mREN2)27 rats compared to control TGR(mREN2)27 rats (Fig. 2F).
AT1R-dependent pathways in TGR(mREN2)27 rats. Since TGR(mREN2)27 rats showed no major differences compared to WT rats in the in vivo disease models, we next analyzed HSC as the major cellular contributor of liver fibrosis. Compared to WT HSC, HSC isolated from TGR(mREN2)27 showed higher aSMA mRNA expression without changes in collagen mRNA (Fig. 3A).
To evaluate the motility of the cells, we investigated migration using the scratch assay, in which HSC isolated from TGR(mREN2)27 showed higher migration capacity than WT HSC (Fig. 3B).  Interestingly, the mRNA levels of ACE, AT2R and Mas suggest an elevated cellular alternative RAS pathway in activated HSC of TGR(mREN2)27 rats gained from livers of uninjured TGR(mREN2)27 rats (Fig. 3C).
Although most downstream pathways activated by AT1R were similarly regulated in WT and TGR(mREN2)27 HSC, congruent to the unaltered mRNA expression of the receptor, HSC from TGR(mREN2)27 HSC showed significantly elevated ROCK2 expression (Fig. 3D). Interestingly, the total hepatic AT1R protein expression was elevated, but only on basal condition TGR(mREN2)27 rats and this difference disappeared after additional liver injury (Fig. 3E, Suppl. Fig. 1D). In parallel to this finding, JAK2 phosphorylation and its downstream component Arhgef1 were overexpressed in the control TGR(mREN2)27 rat livers when compared to WT (Fig. 3F, Suppl. Fig. 1C), a difference that could not be observed under liver injury conditions.
In order to test whether this up-regulation is relevant for the spontaneous portal hypertension observed in TGR(mREN2)27 rats (Table 1), AG490 compound was used to inhibit JAK2. Indeed, AG490 decreased portal pressure and hepatic vascular resistance in TGR(mREN2)27 rats (Fig. 4A). This effect was also confirmed by in situ liver perfusion experiments in TGR(mREN2)27 livers, showing a dose-dependent relaxation of the intrahepatic vascular system (Fig. 4B). Furthermore, migration of TGR(mREN2)27 HSC was blunted by AG490 (Fig. 4C).

Role of Mas in TGR(mREN2)27 rats.
Interestingly, hepatic Mas expression tended to be higher in TGR(mREN2)27 rats than in WT rats at mRNA and protein levels, reaching significance in CCl 4 induced fibrosis ( Fig. 5A,B, Suppl. Fig. 1D). To test whether this up-regulation of Mas is relevant in vivo, its agonist AVE0991 was injected in control TGR(mREN2)27 animals, as well as after four weeks of BDL and ten weeks of CCl 4 . A significant decrease of portal pressure due to a drop in hepatic vascular resistance was observed in TGR(mREN2)27 rats in all models (Fig. 5C,D, Table 2). Secondary changes in the systemic circulation were also observed. However, in splanchnic vascular resistance, they were not significant (Table 2).
Interestingly, AVE0991 decreased Mas expression but led to a relative increase in ACE2 mRNA levels ( Fig. 5E,F), which might be a feedback regulation to the reduced Mas mRNA expression in response to AVE0991. Notably, ACE and AT1R were not substantially influenced (Suppl. Fig. 1E,F).

Discussion
This study shows for the first time that TGR(mREN2)27 rats develop spontaneous fibrosis and portal hypertension, in addition to previously described steatosis and inflammation 12 . Renin-induced portal hypertension can be ameliorated either by JAK2 inhibitors or Mas agonists. Surprisingly, renin overexpression does not exacerbate cholestatic or toxic liver damage. www.nature.com/scientificreports www.nature.com/scientificreports/ Modulation of RAS might play an important role in the treatment of human NAFLD 27,28 , suggesting a pathogenic involvement of RAS in the development and progression of NAFLD. NAFLD is an emerging etiology of liver disease and almost 30% of the patients suffer from portal hypertension, which is associated with fibrosis in almost 90% of the patients 2 . The TGR(mREN2)27 rat model has been described to exhibit pronounced steatosis and inflammation already at an age of 12 weeks 12 and the present study demonstrates that these animals also develop fibrosis and portal hypertension. This further confirms TGR(mREN2)27 as a suitable model to investigate renin-induced liver injury and portal hypertension. Moreover, TGR(mREN2)27 rats represent a NAFLD model of rats with mild fibrosis and significant portal hypertension regardless of any specific diet. To date, all NAFLD rat models necessitate an extended diet treatment and, while the animals might develop portal hypertension, they nevertheless fail to develop inflammation or fibrosis [29][30][31] . Thus, the TGR(mREN2)27 rat model possibly offers an advantage over the diet models in wild type rats [29][30][31] .
The concept that co-factors of liver injury may maintain portal hypertension has already been adopted by the Baveno VI guidelines 32 , and was tested in TGR(mREN2)27 rats. However, this concept could not be implemented in the rats in our study, since the mouse renin expression -responsible for the development of liver fibrosis and portal hypertension -was reduced upon cholestatic or toxic liver injury. This reduction was probably caused by the injury inducing loss of hepatocytes, which are the majority cell type in the liver. During liver injury the number of hepatocytes diminishes, and as shown by our study the gene expression of mREN is decreased in the hepatocytes from fibrotic animals. This reduced mouse renin expression due to injury might explain the lacking difference between WT and TGR(mREN2)27 rats after liver injury.
However, the TGR(mREN2)27 rats without an additional liver injury are an interesting model of liver disease since these animals develop NAFLD with fibrosis and portal hypertension. Indeed, classic and alternative RAS play an important role in different models of experimental liver cirrhosis and in different etiologies of human liver cirrhosis 8,9,[14][15][16] . TGR(mREN2)27 rats exhibit these properties, which confirms their suitability as a model of liver disease. One of the pathways mediating portal hypertension in TGR(mREN2)27 rats is the overactivation of the JAK2-dependent pathway downstream of AT1R. Several lines of evidence confirmed this conclusion. First, the transcription levels of the main components of the pathway including AT1R were increased. Further, activation and protein expression of JAK2 and Arhgef-1, and, finally, inhibition of JAK2 using AG490 decreased hepatic resistance and thereby portal pressure in these animals. Previously, we demonstrated that downstream of AT1R, JAK2/Arhgef1 is activated and mediated through ROCK fibrosis and portal hypertension 8,9 . These findings were independently confirmed by others 33 . In uninjured TGR(mREN2)27 rats, JAK2 inhibition mediated relaxation of hepatic stellate cells, slowed their migration and thereby decreased portal pressure in vivo via a decrease in hepatic vascular resistance, which was shown in vivo and in isolated liver perfusions.
Similar to human liver cirrhosis with portal hypertension, TGR(mREN2)27 rats showed elevated Mas expression 15 . AVE0991, the non-peptidic agonist of Mas, could blunt portal hypertension due to massive hepatic vasodilation in TGR(mREN2)27 rats, in either uninjured or cirrhotic (BDL, CCl 4 ) models. Again, this indicates that Mas is an important effector of the vasculature in the presence of portal hypertension as described previously by others and our group [14][15][16] . The assessment of the loss-of-function of the masR would be extremely interesting in this setting and this is a limitation of the present work and can be further analyzed in future studies. Especially the hemodynamic results suggest that not only uninjured TGR(mREN2)27 rats, but also TGR(mREN2)27 rats after cholestatic or toxic injury are suitable for analyzing the effects of Mas. Interestingly, especially the incubation with AVE0991 led to a decrease of masR mRNA, significantly only in the CCl4-model. This might be due to an yet unknown feedback-loop, which might also explain why AVE0991 did not reduce fibrosis in the long-term treatment.
In summary, TGR(mREN2)27 rats develop spontaneous liver fibrosis and portal hypertension and are a suitable NAFLD model with mild liver fibrosis and portal hypertension. In particular after toxic or cholestatic liver injury, may TGR(mREN2)27 rats be used to develop novel therapeutic strategies targeting Mas. www.nature.com/scientificreports www.nature.com/scientificreports/ Hepatic-vascular resistance was measured in control, BDL and CCl 4 intoxicated TGR(mREN2)27 rats before and after AVE injection. Results are shown as mmHg*min*100 g/ml. (E) Hepatic Mas mRNA expression with and without AVE injection in TGR(mREN2)27 rats. Hepatic Mas mRNA was measured in control, BDL and CCl 4 injected TGR(mREN2)27 rats without and with AVE injection. Data were normalized to control mRNA data without AVE treatment and represented as x-fold change. (F) Hepatic ACE2 mRNA expression with and without AVE injection in TGR(mREN2)27 rats. ACE2 mRNA expression was measured in livers of control, BDL and CCl 4 intoxicated TGR(mREN2)27 rats with and without AVE injection. All data were normalized to control data without AVE injection und shown as x-fold change. Error bars are means ± s.e.m. Statistical analyzes; Mann-Whitney t-test. */**/*** Indicates p < 0.05/p < 0.005/p < 0.0001 compared to respective WT data. #/### Indicates p < 0.05/p < 0.005/p < 0.0001 compared to TGR(mREN2) 27

Materials and Methods
Animals and models of liver disease. Animals. We used 79 Sprague-Dawley wild type (WT) and 90 TGR(mREN2)27 rats. Experimental procedures were approved by the Animal Ethics Committee of Austin Health and of North Rhine-Westphalia (LANUV 84-02.04.2014.A137). WT and TGR(mREN2)27 rats were housed in a controlled environment (12 hour light/dark, temperature 22 °C to 24 °C) and fed standard rat chow ad libitum (Norco, Lismore NSW, Australia; Ssniff, Soest, Germany) with free access to water.
Hemodynamic studies. In-vivo hemodynamic studies. Once rats had developed ascites as a definite sign for the presence of portal hypertension, the animals were used for hemodynamic studies as described previously 9,34 . To assess the acute effect of AVE0991 or AG490, invasive measurements of mean arterial pressure (MAP) and portal pressure (PP) were performed in cirrhotic rats. AVE0991 or AG490 was administered at a dose of 1 mg/kg in the femoral vein.
Microsphere technique. To investigate hemodynamics, the colored microsphere technique was performed as described previously 9,34 . Before and 1 h after injection of AVE0991 or AG490, 300.000 systemic (red/white) microspheres (15 µm diameter, Triton-Technologies, San Diego, USA) were injected in the left ventricle. Mesenteric portosystemic shunt volume was estimated before and after injection of 150.000 microspheres (yellow/blue) in the ileocecal vein.
In situ liver perfusion. In ten cirrhotic CCl 4 intoxicated TGR(mREN2)27 rats, in situ liver perfusion was performed in a recirculating system as described previously 9,34 . After a stabilization period of 30 minutes, pre-contraction of the liver was induced by adding 5 µM Ang II to the Krebs-Henseleit solution. Thereafter, AG490 was added to the Krebs-Henseleit solution (10 −6 and 10 −5 M).
Hepatic hydroxyproline content. The hepatic hydroxyproline content was determined photometrically in analogue segments (200 mg) of snap-frozen livers as described previously 8,35 . Tissue collection. Healthy control rats and animals after induction of liver fibrosis were anesthetized and laparotomy was performed for tissue collection. The livers were cut into fragments and stored at −80 °C until they were used for qRT-PCR and Western blot analysis as described previously 8,35 . Segments of each liver were fixed in formaldehyde (4%) for paraffin embedding as described previously 8,35 .
Cell culture. Rat hepatic stellate cells were incubated in cell culture medium (DMEM + 20% FBS + penicillin/ streptomycin) in 250 ml plastic flasks at 37 °C. After reaching 80% confluence, cells were passaged with a 1:3 split ratio. Detachment was achieved by incubating the cells with 0.05% Trypsin/EDTA solution (solved in PBS) for five minutes at 37 °C. Confluent hepatic stellate cells were incubated in media with 5 µM AG490 or with 10 −5 M AVE0991 and harvested three days later for qRT-PCR.
Wound healing assay. Cells were cultured in cell culture medium (DMEM + 20% FBS + penicillin/streptomycin) in 24-well plastic dishes at 37 °C. When cells were confluent the cell culture medium was withdrawn by suction and changed with the respective media. A scratch was made with a size of 1 mm. After one, two, and three days, the diameter of the scratch was measured and quantified using a Zeiss microscope (Primo Star, SF18). The results are shown as the percentage of the gap size.