The G Protein-Coupled Bile Acid Receptor TGR5 (Gpbar1) Modulates Endothelin-1 Signaling in Liver

TGR5 (Gpbar1) is a G protein-coupled receptor responsive to bile acids (BAs), which is expressed in different non-parenchymal cells of the liver, including biliary epithelial cells, liver-resident macrophages, sinusoidal endothelial cells (LSECs), and activated hepatic stellate cells (HSCs). Mice with targeted deletion of TGR5 are more susceptible towards cholestatic liver injury induced by cholic acid-feeding and bile duct ligation, resulting in a reduced proliferative response and increased liver injury. Conjugated lithocholic acid (LCA) represents the most potent TGR5 BA ligand and LCA-feeding has been used as a model to rapidly induce severe cholestatic liver injury in mice. Thus, TGR5 knockout (KO) mice and wildtype (WT) littermates were fed a diet supplemented with 1% LCA for 84 h. Liver injury and gene expression changes induced by the LCA diet revealed an enrichment of pathways associated with inflammation, proliferation, and matrix remodeling. Knockout of TGR5 in mice caused upregulation of endothelin-1 (ET-1) expression in the livers. Analysis of TGR5-dependent ET-1 signaling in isolated LSECs and HSCs demonstrated that TGR5 activation reduces ET-1 expression and secretion from LSECs and triggers internalization of the ET-1 receptor in HSCs, dampening ET-1 responsiveness. Thus, we identified two independent mechanisms by which TGR5 inhibits ET-1 signaling and modulates portal pressure.


Isolation of Hepatic Stellate Cells
HSCs were isolated from 4-6 month-old female WT C57BL/6 mice. Mice were kept under the same conditions as described above. Mice were anesthetized and placed under a heating lamp. After cannulation of the portal vein, the liver was perfused using HBSS buffer (37 • C). Following a perfusion with pronase E (0.2% in HBSS buffer) and collagenase type I (0.025% in HBSS buffer), that led to digestion of collagen and lysis of hepatocytes, the liver was removed and mechanically diced. The liver cell solution was filtered and resuspended in a pronase/Dnase solution (0.125%) for further digestion. After filtration through a 70 µm cell strainer, the solution was diluted in HBSS and centrifuged at 500× g for 7 min. The supernatant containing cell debris was removed. The cell pellet was resuspended in HBSS containing 0.25% BSA and then mixed with 28.7% Nycodenz gradient solution (Cosmo bio USA, Carlsbad, CA, USA) to obtain a final concentration of 18% Nycodenz (w/v). This preparation was covered with 8 mL 0.25% BSA/HBSS before centrifugation for 30 min at 1400× g without brake. After centrifugation, the layer between the two buffers was harvested and centrifuged at 450× g for 10 min. The supernatant was discarded and the pellet was resuspended in DMEM supplemented with 10% fetal calf serum (FCS) and 1% penicillin/streptomycin/amphotericin B. The medium was exchanged every other day. Cells were kept on cell culture plates covered with collagen type 1 in a density of 2.5-3 × 10 6 cells/well in a 6-well plate. Cells were used for experiments on the next day or after seven days for experiments with activated HSCs (Supplemental Figure S1). During experiments, cells were kept in DMEM without FCS or antibiotics.

Isolation and Cultivation of LSECs
The protocol for magnetic-activated cell isolation of LSECs was adapted from [31]. After induction of anesthesia, a cannula was inserted into the portal vein for perfusion with HBSS containing collagenase type 1, leading to removal of blood and dissociation of liver cells from connective tissue. Afterwards, livers were removed, diced, and incubated in the collagenase solution for 20 min for further digestion. The cell suspension was filtered through 100 µm cell strainers and centrifuged several times for removal of collagen and cell debris. To clear the cell suspension of CD45-positive immune cells, the cell suspension was preincubated with anti-CD45 magnetic beads for 15 min at 4 • C and transferred onto LD columns in the QuadroMACS separator (Miltenyi Biotec GmbH, Bergisch Gladbach, Germany). The flow-through was centrifuged for 10 min at 500× g and 4 • C. The supernatant was discarded and the pellet was resuspended in ice-cold MACS-buffer, according to the manufacturer's protocol. For positive selection of LSECs the cell suspension was incubated with anti-CD146 MicroBeads (Miltenyi Biotec). Then the cell suspension was put on cell columns in the OctoMACS separator (Miltenyi Biotec), according to protocol provided by the company. The flow-through was discarded and the columns were flushed with MACS-buffer containing 0.05% BSA after removal of columns from the magnetic field. The cell suspension was centrifuged and the pellet was resuspended in cell culture medium (DMEM, supplemented with 10% FCS and 1% antibiotics). Cells were cultured in cell culture dishes covered with collagen type I with the following densities: 24-well: 800,000-900,000 cells/well; and 6-well: 3-3.5 × 10 6 cells/well. Culture medium was removed and cells were washed in PBS after 24 h in culture. Cells were treated with serum-free medium containing DMSO, 10 µM TGR5 agonist, 25 µM taurolithocholic acid, or 10 µM forskolin for 24 h.

Gene Expression Assay
RNA from liver samples of the above-mentioned mice or of LSECs was extracted using the Maxwell 16 LEV simply RNA Tissue Kit (Promega, Madison, WI, USA) and the Maxwell 16 Instrument (Promega, Madison, WI, USA), according to the manufacturer's instructions.
A DNA microarray platform (Affymetrix) was used for global gene expression analysis. Total RNA preparations were checked for RNA integrity by Agilent 2100 Bioanalyzer quality control. Mean RNA integrity number (RIN) was 7.3 ± 0.4 (range 6.5 to 8.1) for liver tissue from chow-fed animals and 7.8 ± 0.3 (range 6.5 to 8.8) for liver tissue from LCA-fed animals. RNA was further analyzed by photometric Nanodrop measurement and quantified by fluorometric Qubit RNA assays (Life Technologies).
Synthesis of biotin-labeled cDNA was performed according to the manufacturers' protocol (WT Plus Reagent Kit; Affymetrix, Inc., Thermo Fisher Scientific, Waltham, MA, USA). Briefly, 100 ng of total RNA was converted to cDNA. After amplification by in vitro transcription and 2nd cycle synthesis, cDNA was fragmented and biotin labeled by terminal transferase. Finally, end-labeled cDNA was hybridized to Affymetrix Mouse Gene 2.0 ST Gene Expression Microarrays for 16 h at 45 • C, stained by strepatavidin/phycoerythrin conjugate and scanned as described in the manufacturersṕ rotocol. Data analyses on Affymetrix CEL files are described in Section 2.14-Statistical Analysis of Expression Data. The gene expression data can be downloaded from the NCBI GEO database (https://www.ncbi.nlm.nih.gov/geo/, accession number GSE139075).
For real-time PCR, 1 µg of this RNA was used to generate cDNA utilizing the QuantiTect Reverse Transcription Kit (Qiagen, Hilden, Germany). Gene expression was quantified using Taqman Gene Expression Assays (Thermo Fisher Scientific, Waltham, MA, USA, assay order information can be obtained upon request) and the Lightcycler 480 II (Roche Diagnostics, Rotkreuz, Switzerland). Data were produced in duplicates for each gene. Mean values of cycle numbers of the target gene were subtracted from the mean of cycle numbers of the house-keeping gene succinatdehydrogenase (SDHA) for the respective sample. These values taken to the power of 2 are the mRNA expression of the target genes in relation to SDHA expression. The number of independent experiments performed are given in the text/figure legends. At least three independent experiments were performed. In order to rule out a regulation of the chosen house-keeping gene SDHA in response to LCA feeding, we initially analyzed expression of selected genes in relation to SDHA, hypoxanthine phosphoriosyltransferase-1 (HPRT-1), and glyceraldehyde-3-phosphate dehydrogenase (GAPDH). Since no difference in regulation was observed between these house-keeping genes, we continued the experiments with SDHA, as shown in Table 1 and Figure 3.

ET-1 ELISA
For analysis of ET-1 secretion from LSECs, cells were kept in culture after isolation as described for 24 h. After washing of cells with PBS, they were treated with serum-free medium containing DMSO, 10 µM TGR5 agonist, 25 µM TLC, or 10 µM forskolin for 24 h. Then, the supernatant was collected and transferred onto amicon ultra-10 centrifugal filters (Merck Millipore, Burlington, NJ, USA) to concentrate the supernatant (1:10). The concentrated supernatant was analyzed for ET-1 levels using the Quantikine kit system for murine ET-1 (Research and Diagnostic Systems Inc, Minneapolis, MN, USA), according to the provided protocol. LSECs were washed and incubated with a fixed amount of 50 µL lysis buffer for protein isolation. Measurement of protein concentration in lysate was performed in the Multiskan Spectrum (Thermo Fisher) and ET-1 concentration in the supernatant was normalized relative to protein concentration.

HSC Contraction Assay
For quantification of activated HSC contraction, 6-well plates were covered with a thick layer of 1 mL/well rat tail collagen. Rat HSCs were isolated as described before [32]. Cells were put on 6-well plates at a density of 6 × 10 5 cells/well. Cells were kept in DMEM supplemented with 10% FCS and 1% antibiotics for 7 days. Medium was exchanged every 48 h starting the day after isolation. At day 7 after isolation, cells were treated with serum-free medium containing DMSO, 10 µM TGR5 agonist, or 10 µM forskolin for 30 min. After detachment of collagen lattices from the 6-well surface, cells were treated with ET-1 (Sigma-Aldrich) at a concentration of 1nM to induce contraction for 2 h and 24 h. Photos of cell/collagen lattice areas were taken at the time of ET-1 addition and after 2 h and 24 h (Supplementary Figure S2). Surface area was measured using the ImageJ distribution Fiji V.2.0.0 [33,34]. The lattice area used as control treated with DMSO and vehicle for the respective time points was set to 1.0.

Measurement of Portal Pressure
TGR5 KO and WT mice on chow or LCA diet (1%, 84 h) were anesthetized and the portal vein was cannulated. Invasive blood pressure measurement was performed using the pressure transducer P75 and amplifier from Hugo Sachs Electronics (Harvard Apparatus GmbH, March, USA). Additionally, 6-week-old WT mice were anesthetized and cannulation of portal vein for blood pressure measurement was done. Mice livers were then perfused with either control HBSS solution or a solution containing 10 µM TGR5 agonist for 15 min. Subsequently, increasing concentrations of ET-1 (5, 10, and 15 nM) were added to the perfusion media for 15 min each. Perfusion pressure in the portal vein was measured constantly over the whole time course. Mean portal pressure was taken for each condition of each mouse. Portal pressure under control conditions (t = 0) of HBSS buffer perfusion supplemented with DMSO (=vehicle) were set to 1.0.

Immunofluorescence of Cells in Culture
HSCs were isolated and kept in culture as described above. Seven days after isolation, cells were incubated with serum-free medium containing either DMSO, 10 µM TGR5 agonist, or 10 µM forskolin. After 24 h, cells were washed with PBS and incubated with 5 µg/mL wheat germ agglutinin (WGA) (Thermo Fisher) labeled with AlexaFluor-594 in PBS for 10 min at 37 • C to stain glycoproteins or glycolipids within the outer leaflet of the plasma membrane [35]. Cells were fixed with ice-cold methanol for 3 min after removal of the WGA-solution. After washing of cells, unspecific binding was blocked using a solution of 5% FCS in PBS for 30 min. Activated HSCs were treated with either primary antibodies for Endothelin-A receptor (Abcam, Cambridge, UK) (1:500) or just blocking solution as negative control for 1 h. A secondary antibody labeled with AlexaFluor-488 (Dianova, Hamburg, Germany) (1:100) was used for detection of the primary antibody and incubated for 1 h. LSECs were fixed with methanol and incubated with antibodies directed against vascular endothelial Cadherine (Santa Cruz Biotechnology, Dallas, USA) (1:100) and TGR5 (Gpbar1 8/50, Roche) (1: 20). Secondary antibodies labeled with Cyanine-3 (Dianova) (1:500) or fluorescein (Dianova) (1:100) were used at dilutions of 1:500 and 1:100, respectively. Intranuclear DNA was labeled by Hoechst 34580 (Thermo Fisher) (1:20,000). Pictures of LSECs were taken using the confocal microscope LSM 510 (Carl Zeiss, Jena, Germany). HSCs were imaged using a LSM 810 confocal laser microscope (Carl Zeiss, Jena, Germany). Colocalization analysis of murine hepatic stellate cells was carried out using Pearson's correlation coefficient calculated by the coloc2 plugin for ImageJ after selection of representative regions of interest ROIs in cell surface areas [33,34].

Statistical Evaluation
The data are presented as means and the standard error of the mean (SEM). Statistical significance was tested using the two-sided Student's t-test or the Wilcoxon signed-ranked test as appropriate. A p-value < 0.05 was considered a significant result and a p-value < 0.01 was considered a highly significant result.

Statistical Analysis of Expression Data
Data preprocessing and all subsequent analyses were performed using the statistical programming language R, version 3.5.0 (R Development Core Team 2018). Normalization of the raw microarray data (CEL files) was done using RMA as implemented in the R package oligo. Normalization was performed on a set of, in total, 18 CEL files. To determine differentially expressed genes, the R package limma was used [36]. Adjustment for multiple testing was conducted with the method of Benjamini and Hochberg (FDR, false discovery rate) [37]. A gene was called differentially expressed if the adjusted p-value was <0.05 and log2 fold change was <−1.5 (downregulated) or >1.5 (upregulated). A volcano plot was generated that plots log2 fold change on the x-axis and statistical significance on the y-axis (-log10 of the FDR-adjusted p-value). Heatmaps were used to visualize z-scores (expression values standardized per gene to mean 0 and standard deviation 1), ordered according to average linkage hierarchical clustering of genes and experiments, respectively. Gene ontology enrichment analysis was performed based on probe set IDs with the topGO package [38], using Fisher's exact test and the elim method. Only results from the biological process ontology were considered. The cutoff for the enrichment p-value was set to 0.05.

LCA Feeding Results in a Significant Upregulation of Genes Associated with Cholestasis, Inflammation, and Extracellular Matrix Remodeling
Feeding of a chow diet supplemented with 1% LCA over 3.5 days (84 h) resulted in a severe cholestatic liver injury in wildtype (WT) mice, as described previously [21]. Liver tissue of LCA-fed WT mice was compared to chow-fed WT littermates by gene array analysis. In total, 332 and 263 genes were significantly up-or downregulated, respectively (p < 0.05; FDR-adjusted, and log2 fold change <−1.5 or >1.5; Supplementary Tables S1 and S2). Overrepresented Gene Ontology (GO) groups among the upregulated genes were associated with proliferation, inflammation, and extracellular matrix organization (Supplementary Tables S1 and S3, Figure 1). The four upregulated genes with the lowest p-values were the acute phase proteins serum amyloid A3 (Saa3) (46.2-fold up) and orosomucoid 2 (Orm2) (42.4-fold up), the bacterial siderophore binding protein lipocalin 2 (Lcn2) (32.4-fold up) that is expressed in neutrophils, and the pattern recognition receptor MARCO (macrophage receptor with collagenous structure, 17.4-fold up), which is found on macrophages and dendritic cells (Supplementary Table S1). The most significantly overrepresented GO groups among the downregulated genes were associated with oxidation-reduction processes, sodium-independent organic anion transport, and bile acid biosynthesis. The rate-limiting enzyme of bile acid synthesis, Cyp7a1 (89.0-fold down), was among the strongest downregulated genes (Supplementary Tables S2 and S4). Therefore, gene expression profiling of the livers of mice under LCA diet revealed a profile of an inflamed, cholestatic organ undergoing extracellular matrix reorganization [21,22]. We have previously reported that cholestatic liver injury in response to bile duct ligation is more severe in TGR5 knockout (KO) mice as compared to WT littermates [19]. Therefore, we analyzed liver damage following LCA feeding in TGR5 WT and TGR5 KO littermates.  to mean 0 and standard deviation 1) of the strongest differentially-regulated genes (p < 0.05 and log2 fold change >3 (upregulated) or <−3 (downregulated)). Each row corresponds to a gene, each column to a mouse. Rows and columns were ordered according to average linkage hierarchical clustering. (B) Volcano plot: On the x-axis, the log 2 fold change is plotted, and on the y-axis, -log10 of the false discovery rate (FDR)-adjusted p-value is plotted. We have previously reported that cholestatic liver injury in response to bile duct ligation is more severe in TGR5 knockout (KO) mice as compared to WT littermates [19]. Therefore, we analyzed liver damage following LCA feeding in TGR5 WT and TGR5 KO littermates.

Liver Damage in Response to LCA Feeding is Aggravated in TGR5 KO Mice and Results in Increased Portal Pressure
Liver damage in response to LCA feeding was more pronounced in TGR5 KO mice as compared to WT mice, as demonstrated by significantly higher serum levels of AST (WT LCA (n = 12) 6586 ± 898 IU/L vs. KO LCA (n = 10) 10,629 ± 1422 IU/L (1.6-fold increase), p < 0.05) and more extensive bile infarcts on liver histology (Figure 2A-C). Furthermore, a significant increase in portal venous pressure was observed in LCA-fed TGR5 KO mice as compared to LCA-fed WT or chow-fed TGR5 KO mice, respectively (WT LCA (n = 7) 3.9 ± 0.4 cmH 2 0 vs. KO LCA (n = 6) 5.7 ± 0.6 cmH 2 0 (1.4-fold increase), p < 0.05) ( Figure 2D). There was no difference in portal pressure between chow-fed mice of both genotypes. fold change >3 (upregulated) or <−3 (downregulated)). Each row corresponds to a gene, each column to a mouse. Rows and columns were ordered according to average linkage hierarchical clustering. (B) Volcano plot: On the x-axis, the log2 fold change is plotted, and on the y-axis, -log10 of the false discovery rate (FDR)-adjusted p-value is plotted.

Liver Damage in Response to LCA Feeding is Aggravated in TGR5 KO Mice and Results in Increased Portal Pressure
Liver damage in response to LCA feeding was more pronounced in TGR5 KO mice as compared to WT mice, as demonstrated by significantly higher serum levels of AST (WT LCA (n = 12) 6586 ± 898 IU/L vs. KO LCA (n = 10) 10,629 ± 1422 IU/L (1.6-fold increase), p < 0.05) and more extensive bile infarcts on liver histology (Figure 2A-C). Furthermore, a significant increase in portal venous pressure was observed in LCA-fed TGR5 KO mice as compared to LCA-fed WT or chow-fed TGR5 KO mice, respectively (WT LCA (n = 7) 3.9 ± 0.4 cmH20 vs. KO LCA (n = 6) 5.7 ± 0.6 cmH20 (1.4-fold increase), p < 0.05) ( Figure 2D). There was no difference in portal pressure between chow-fed mice of both genotypes. . All data are shown as mean ± standard error of the mean (SEM). * Statistically significant difference between the LCA-and . All data are shown as mean ± standard error of the mean (SEM). * Statistically significant difference between the LCA-and chow-fed mice of the same genotype (p < 0.05); # statistically significant difference between WT and TGR5 KO mice fed the same diet (p < 0.05).

Genes Associated with HSC Activation are Significantly Upregulated in TGR5 KO Mice as Compared to WT Littermates
Gene expression analysis by semi-quantitative real-time PCR revealed a similar pattern of gene regulation in both genotypes after LCA feeding. As expected, induction of cholestasis by LCA feeding resulted in a significant downregulation of Cyp7a1 and Cyp8b1, the rate limiting enzymes of bile acid synthesis. In line with the increase in liver damage, a significant upregulation of the genes encoding receptor-interacting serine/threonine-protein kinases 1 and 3 (RIPK1 and RIPK3), as well as caspase-3, was observed (Table 1). Furthermore, we found a significant induction of pro-inflammatory genes such as tumor necrosis factor-α (TNF-α), CCL2, interleukin-1β (IL-1β) and pro-fibrogenic chemokine receptors CCR1, CCR5 (Table 1) [21,22,39]. Interestingly, genes related to sinusoidal endothelial dysfunction, activation of HSCs, and fibrosis development, such as platelet-derived growth factor receptor-α (PDGFRα), PDGFRβ, transforming growth factor-β1 (TGF-β1), and endothelin-1 (ET-1), were upregulated in both WT and TGR5 KO mice following LCA feeding. However, the observed increase was significantly higher in TGR5 KO mice (PDGFRα: WT LCA (n = 7) 3.52 ± 0.8 vs. KO LCA (n = 10) 7.30 ± 1.4, 2.1-fold increase in TGR5 KO, p < 0.05; PDGFRβ: WT LCA (n = 12) 4.0 ± 0.6 vs. KO LCA (n = 12) 6.23 ± 0.75, 1.6-fold increase, p < 0.05; TGF-β1: WT LCA (n = 7) 7.81 ± 0.8 vs. KO LCA (n = 9) 11.8 ± 1.4, 1.5-fold increase, p < 0.05; ET-1: WT LCA (n = 14) 8.58 ± 1.1 vs. KO LCA (n = 11) 13.9 ± 2.4, 1.6-fold increase, p < 0.05) ( Figure 3A-D). Sirius red staining of liver tissue after 3.5 days of LCA feeding revealed a slight increase in periportal collagen deposition, but no overt hepatic fibrosis in both genotypes, which was in line with measurements of hydroxyproline content in liver tissue ( Figure 4A,B). Expression of genes encoding collagen-1α1, collagen-1α2 were upregulated in both genotypes following LCA feeding for 3.5 days, indicating initiation of extracellular matrix synthesis, as described earlier ( Figure 4B) [21,22]. Table 1. Selection of genes significantly regulated in livers of TGR5 wildtype (WT) and TGR5 knockout (KO) after LCA feeding for 84 h as compared to chow-fed animals of the same genotypes. Data were generated by real-time PCR, as described in Section 2-Materials and Methods. Gene expression is presented in relation to the house-keeping gene SDHA. Values for WT mice on chow diet were set to 1.0. * Significantly different as compared to chow-fed animals of the same genotype (p < 0.05); ** significantly different as compared to chow-fed animals of the same genotype (p < 0.01); # significantly different as compared to WT animals of the same treatment group (p < 0.05). All data are expressed as mean ± SEM. Green and red boxes indicate a statistically significant downregulation or upregulation, respectively, of a gene in TGR5 KO as compared to WT littermates receiving the same treatment.  Figure 3. Expression of genes related to portal hypertension and fibrosis development are significantly higher in liver tissue of TGR5 KO mice as compared to WT littermates in response to LCA feeding. (A-D) Semi-quantitative real-time PCR analysis of liver tissue revealed an upregulation of mRNA expression PDGFRα, PDGRFβ, endothelin-1, and TGF-β1 following LCA feeding. The increase was more pronounced in mice deficient for TGR5. Data are presented as mean ± SEM (n = 4-12). * Statistically significant difference between the LCA-and chow-fed mice of the same genotype (p < 0.05); # statistically significant difference between WT and TGR5 KO mice fed the same diet (p < 0.05; using the two-sided Student's t-test).

Figure 3.
Expression of genes related to portal hypertension and fibrosis development are significantly higher in liver tissue of TGR5 KO mice as compared to WT littermates in response to LCA feeding. (A-D) Semi-quantitative real-time PCR analysis of liver tissue revealed an upregulation of mRNA expression PDGFRα, PDGRFβ, endothelin-1, and TGF-β1 following LCA feeding. The increase was more pronounced in mice deficient for TGR5. Data are presented as mean ± SEM (n = 4-12). * Statistically significant difference between the LCA-and chow-fed mice of the same genotype (p < 0.05); # statistically significant difference between WT and TGR5 KO mice fed the same diet (p < 0.05; using the two-sided Student's t-test). (B) Measurement of hydroxyproline content in liver tissue showed a slight increase following LCA feeding, which did not reach significance (n = 8-10 animals per genotype LCA group, n = 3 animals per genotype for chow group). However, expression of collagen-1α1 and collagen-1α2 increased significantly after LCA feeding in both genotypes (qPCR: n = 12-14 LCA-fed animals per genotype, n = 4 chow-fed animals per genotype). Liver tissue from WT animals is shown in grey, while tissue from TGR5 KO mice is represented by black bars. All data are shown as mean ± SEM. * Statistically significant difference between the LCA-and chow-fed mice of the same genotype (p < 0.05) (n = 8-10).

Activation of TGR5 in LSECs Lowers ET-1 Expression and Secretion
ET-1 in liver is mainly secreted by LSECs under physiological conditions [24]. We have previously demonstrated that rat and human LSECs express TGR5 [1,4]. To determine whether secretion of ET-1 from LSECs was modulated by TGR5, LSECs were isolated from WT mice. Expression and localization of TGR5 in these primary cells was confirmed by immunofluorescence staining with an antibody against TGR5 and an antibody against vascular endothelial cadherin (Ve-Cad). TGR5 was localized in the plasma membrane, as well as in intracellular vesicular structures ( Figure 5A). Treatment of primary murine LSECs with vehicle (DMSO), taurolithocholic acid (TLC, 25 μM), a non-bile acid TGR5 agonist (RO5527239, 10 μM) [27], and forskolin (10 μM) for 24 h resulted in a significant decrease in ET-1 mRNA expression (DMSO vs. TLC (n = 10) 1.0 vs. 0.76 ± 0.06 (1.3- liver tissue showed a slight increase following LCA feeding, which did not reach significance (n = 8-10 animals per genotype LCA group, n = 3 animals per genotype for chow group). However, expression of collagen-1α1 and collagen-1α2 increased significantly after LCA feeding in both genotypes (qPCR: n = 12-14 LCA-fed animals per genotype, n = 4 chow-fed animals per genotype). Liver tissue from WT animals is shown in grey, while tissue from TGR5 KO mice is represented by black bars. All data are shown as mean ± SEM. * Statistically significant difference between the LCA-and chow-fed mice of the same genotype (p < 0.05) (n = 8-10).

Activation of TGR5 in HSCs Attenuates ET-1-Mediated Contractility
Activated HSCs are able to contract, leading to an increase in portal pressure due to their perivascular location in the space of Disse [40]. ET-1 has been identified as one of the key mediators regulating HSC contractility [24,41]. Since we observed an increase in portal pressure solely in TGR5 KO mice after LCA feeding, we hypothesized that activation and contraction of HSCs may contribute to this phenotype. To determine whether ET-1-dependent HSC contraction is modulated by TGR5, HSC contraction in response to ET-1 was analyzed. Rat HSCs were cultivated on collagen lattices for 7 days [40]. After this interval, DMSO, a TGR5 agonist (10 µM), or forskolin (10 µM) were added to the culture medium 30 min prior to stimulation with ET-1 (1 nM). Measurement of the collagen lattice surface area 2 h and 24 h after addition of ET-1 revealed a significantly smaller reduction in relative surface area in TGR5 agonist-or forskolin-treated cells, indicating an attenuation of HSC contractility ( Figure 6A); 2 h: DMSO/ET-1 vs. TGR5 Ago/ET-1 1.0 vs. 1.64 ± 0.11 pixels 2 ; DMSO/ET-1 vs. forskolin/ET-1 1.0 vs. 2.85 ± 0.36 pixels 2 , n = 12, p < 0.01; 24 h: DMSO/ET-1 vs. TGR5 Ago/ET-1 1.0 vs. 1.89 ± 0.27 pixels 2 ; DMSO/ET-1 vs. forskolin/ET-1 1.0 vs. 3.96 ± 0.51, n = 12, p < 0.01. It was previously shown that cAMP may desensitize the ET-1 receptor-A (ET A R) towards its ligand on activated HSCs through internalization. Thus, the ET A R's responsiveness is being shifted from picomolar to nanomolar concentrations [42]. Since TGR5 is coupled to a stimulatory G protein [6,8], we hypothesized that ligand binding to TGR5 could result in increased intracellular cAMP levels and internalization of the ET A R. Immunofluorescence staining localized the ET A R within the plasma membrane of HSCs, as demonstrated by colocalization with the plasma membrane marker wheat germ agglutinin (WGA) ( Figure 6B). Treatment with the TGR5 agonist reduced the amount of ET A R within the plasma membrane, indicating internalization of the receptor ( Figure 6C) (DMSO vs. TGR5 Ago (n = 8-9) 0.37 ± 0.06 Pearson's correlation coefficient vs. 0.11 ± 0.04, 3.4-fold decrease, p < 0.01).

Activation of TGR5 Inhibits the ET-1-Mediated Increase in Portal Pressure
Perfusion of mouse livers with buffer containing only DMSO as vehicle did not cause any significant changes in portal perfusion pressure. A challenge of these mice with increasing concentrations of ET-1 led to a dose-dependent increase in portal perfusion pressure. Addition of a TGR5 agonist to the ET-1-enriched perfusion buffer significantly reduced the ET-1-mediated rise in portal pressure as compared to vehicle containing perfusion buffer, indicating that TGR5 activation rapidly reduces ET-1-mediated portal hypertension (DMSO vs. TGR5 Ago/+1 nM ET-1 2.41 ± 0.26 cmH 2 0 vs. 1.42 ± 0.18 cmH 2 0 (1.7-fold decrease); DMSO vs. TGR5 Ago/+5 nM ET-1 2.9 ± 0.18 cmH 2 0 vs. 1.89 ± 0.28 cmH 2 0 (1.5-fold decrease), DMSO vs. TGR5 Ago/+15 nM ET-1 3.14 ± 0.15 cmH 2 0 vs. 2.20 ± 0.3 cmH 2 0 (1.4-fold decrease), (n = 5-6), p < 0.05) ( Figure 7A). Figure 6. Stimulation of TGR5 in isolated HSCs triggers internalization of the endothelin A receptor (ETAR), thus reducing the contractile response towards ET-1. (A) Contraction of rat HSCs on collagen lattices was measured in response to ET-1 after pre-incubation with DMSO, a TGR5 agonist (10 μM), or forskolin (10 μM), which was used as TGR5-independent positive control. Surface area of collagen lattices served as indirect measure of contractile activity 2 h and 24 h after ET-1 addition. Surface area Figure 6. Stimulation of TGR5 in isolated HSCs triggers internalization of the endothelin A receptor (ET A R), thus reducing the contractile response towards ET-1. (A) Contraction of rat HSCs on collagen lattices was measured in response to ET-1 after pre-incubation with DMSO, a TGR5 agonist (10 µM), or forskolin (10 µM), which was used as TGR5-independent positive control. Surface area of collagen lattices served as indirect measure of contractile activity 2 h and 24 h after ET-1 addition. Surface area of cells/collagen lattices treated with DMSO and ET-1 was set to 1.0. Data are shown as mean ± SEM (n = 12); * statistically significant from DMSO/ET-1-treated cells (p < 0.01). (B) Immunofluorescence staining of murine HSCs using an antibody directed against the ET A R (in green) and Alexa-543 coupled wheat germ agglutinin (WGA, shown in red) to stain the cell surface. HSCs were incubated with either DMSO or a TGR5 agonist (10 µM) for 24 h prior to fixation and immunofluorescence staining. Colocalization of ET A R and WGA was determined using the coloc2 plugin for ImageJ. Colocalized pixels are shown in white in the merge images. (C) Colocalization was quantified using Pearson's correlation coefficient for colocalization, as determined by the coloc2 plugin for ImageJ for each selected region of interest (ROI). Data are shown as mean ± SEM; * statistically significant from DMSO-treated cells (n = 8-9; 2-3 different ROIs per condition of three different cell isolations) (p < 0.05).
with either DMSO or a TGR5 agonist (10 μM) for 24 h prior to fixation and immunofluorescence staining. Colocalization of ETAR and WGA was determined using the coloc2 plugin for ImageJ. Colocalized pixels are shown in white in the merge images. (C) Colocalization was quantified using Pearson's correlation coefficient for colocalization, as determined by the coloc2 plugin for ImageJ for each selected region of interest (ROI). Data are shown as mean ± SEM; * statistically significant from DMSO-treated cells (n = 8-9; 2-3 different ROIs per condition of three different cell isolations) (p < 0.05). Portal perfusion pressure of TGR5 transgenic mice (TGR5tg) was continuously measured in the presence of DMSO or a TGR5 agonist and increasing concentrations of ET-1. Perfusion was performed with either DMSO or a TGR5 agonist (10 µM) and increasing concentrations of endothelin-1 (ET-1, 5-15nM). Data are shown relative to controls, where the perfusion pressure in the presence of DMSO-containing perfusion buffer following an equilibrium phase of 15 min at the starting point was set to 1.0. All data are shown as mean ± SEM; * statistically significant from DMSO-treated cells (p < 0.05) (n = 6-9). (B) Cartoon depicting the mechanisms of TGR5-dependent inhibition of ET-1 signaling within the sinusoid. Activation of TGR5 through cAMP inhibits transcription and subsequent secretion of ET-1 from LSECs. ET-1, produced by LSECs, is known to induce contraction of activated HSCs cells by binding to the ET A R, leading to an increase of intrahepatic vascular resistance and, hence, the development of portal hypertension. Activation of TGR5 in HSCs promotes internalization of the ET A R, reducing ligand binding at the cell surface and thus contractile activity. In conclusion, ligand binding to TGR5 activates two mechanisms that act synergistically to reduce intrahepatic vascular resistance and, thus, portal hypertension.

Discussion
Liver injury of various etiologies triggers morphological and functional changes within hepatic sinusoids, resulting in development of fibrosis and portal hypertension [24,43,44]. Disturbances of the intercellular crosstalk within the sinusoids, especially between LSECs and HSCs, contribute to fibrosis development and microvascular dysfunction [24,43,44]. Multiple paracrine signals, such as NO, ET-1, TGF-β, PDGF, and VEGF, as well as various cytokines and chemokines that are being secreted not only from LSECs, hepatocytes, cholangiocytes, immune cells, but also from platelets, modulate LSEC and HSC phenotype under physiological and pathophysiological conditions [44,45]. Gene expression analysis of liver tissue following LCA feeding for 84 h revealed an enrichment in signaling pathways related to inflammation, proliferation, and matrix remodeling, which was in line with histological analysis of the livers. Interestingly, mice deficient for the G protein-coupled BA receptor TGR5 were more susceptible towards LCA-induced liver injury, resulting in elevated AST serum levels, more pronounced bile infarcts, and an elevated portal perfusion pressure. TGR5 is highly expressed in cholangiocytes, where activation of the receptor triggers formation of the bicarbonate umbrella, promotes tight junction integrity, and inhibits apoptosis, thus protecting cholangiocytes from bile acid toxicity [9,11,46,47]. Lack of TGR5 therefore renders mice more susceptible towards bile acid-induced biliary injury, as observed previously, in response to common bile duct ligation (CBDL) [19] and now in the LCA model. Furthermore, TGR5 exerts anti-inflammatory functions in monocytes and macrophages [4,6,13,14,18]. Absence of TGR5 resulted in elevated expression and secretion of chemokines and cytokines in response to lipopolysaccharide injection or common bile duct ligation [18,19]. Following LCA feeding, we observed a significant induction of hepatic mRNA expression of various cytokines, chemokines, and pro-fibrogenic chemokine receptors such as CCR1 and CCR5 [39]. Interestingly, expression of the chemokine receptor CCR5 was already induced 3-fold in TGR5 KO mice under chow-fed conditions, which has been previously reported for white adipose tissue in TGR5-deficient animals [13]. CXCL10 and its receptor CXCR3 have been implicated in fibrosis development in models of congenital hepatic fibrosis and carbon tetrachloride (CCl 4 )-mediated fibrosis [48,49]. While upregulation in comparison to chow-fed WT animals of CXCL10 was 9-fold in WT and 5-fold in TGR5 KO, respectively, CXCR3 mRNA expression was induced 3-fold in WT and 5-fold in TGR5 KO following LCA feeding. Interestingly, in patients with chronic liver disease, serum CXCL10 levels, but not hepatic CXCL10 mRNA levels, were positively correlated with portal hypertension, fibrosis stage, and disease progression [50,51]. To which extent the CXCL10-CXCR3 signaling pathway contributes to the phenotypes of our LCA-fed mice is unclear. CXCL1, which is secreted from LSECs following mechanical stretching, triggers formation of sinusoidal microthrombi, thereby increasing portal pressure independent of cirrhosis [52]. Hepatic CXCL1 mRNA expression was significantly induced in LCA-fed mice. In comparison to chow-fed WT mice, CXCL1 levels were 14.5-and 9.4-fold higher in LCA-fed WT and TGR5 KO mice, respectively. However, in TGR5 KO mice, CXCL1 levels were low under chow-fed conditions and increased 27-fold in response to LCA. Thus, microthrombosis may also contribute to portal hypertension in LCA-fed TGR5 KO animals.
Moreover, expression of PDGFRα/β, TGF-β1 and ET-1, which are related to HSC activation, development of fibrosis, and portal hypertension, were upregulated in both genotypes in response to LCA. However, mRNA levels were significantly higher in livers of TGR5 KO mice as compared to liver tissue from WT littermates. While periportal fibrosis, as measured by Sirius red staining and hydroxyproline content, was not significantly increased in either genotype as early as 3.5 days after starting the LCA feeding, expression of collagen was significantly upregulated in both genotypes, suggesting initiation of fibrogenesis in line with previous data [21]. Since TGR5 is expressed both in LSECs and activated HSCs, we further explored TGR5-mediated effects in these cell types. Stimulation of TGR5 on LSECs reduced ET-1 expression, as well as its secretion, and reduced the responsiveness of activated HSCs towards ET-1 through internalization of the ET A R. These two mechanisms may act synergistically to reduce ET-1 signaling within the sinusoid ( Figure 7B). In rats, treatment with a TGR5 agonist (BAR501) for 6 days prior to cannulation of the portal vein lowered the rise in portal perfusion pressure in response to norepinephrine [53]. Furthermore, administration of the TGR5 agonist inhibited portal hypertension in mice treated for 9 weeks with CCl 4 , while it did not affect fibrosis development [53], suggesting an effect on the hepatic microvasculature independent of extracellular matrix deposition. As underlying molecular mechanisms, the authors demonstrated that TGR5 activation induces expression and also non-genomically promotes activation of cystathioneγ-lyase (CSE), resulting in increased production of hydrogen sulfide (H 2 S), a potent vasodilator [53,54]. Amongst others, we have previously shown that TGR5 may also trigger serine phosphorylation of eNOS, thereby promoting NO generation, which again leads to vasodilation of hepatic sinusoids [1,53]. Taken together, TGR5 agonists promote generation and secretion of vasodilatory agents (H 2 S and NO) and inhibit expression and secretion of the potent vasoconstrictor ET-1 from LSECs [1, 7,46,53,54]. While the function of TGR5 in LSECs has been previously studied, the role of TGR5 for HSC activation and function remained elusive. Interestingly, the rise in hepatic PDGFRα/β expression was more pronounced in TGR5 KO mice as compared to WT littermates after LCA-feeding. PDGF-β, which may be derived from either LSECs or platelets, can signal to activated HSCs through its receptor PDGFRβ and promote cell proliferation, migration, and development of a pro-angiogenic HSC phenotype [55,56]. Furthermore, it was recently demonstrated that PDGF-β plays a role in the trans-differentiation processes of HSCs into an activated phenotype, and that administration of an anti-PDGF-β antibody (MPR8457) inhibits development and progression of biliary fibrosis in Abcb4 KO mice, which serve as a chronic model for sclerosing cholangitis [56,57]. Moreover, cAMP may desensitize activated HSCs towards ET-1 through internalization of the ET A R [42]. Using a non-BA TGR5 ligand, we could demonstrate that activation of the receptor promotes retrieval of ET A R from the plasma membrane, explaining the reduced ET-1-dependent contractility of activated HSCs in the presence of a TGR5 agonist in vitro. In summary, TGR5 activation reduces hepatic vascular resistance through several mechanisms, both in LSECs and in HSCs, that act synergistically. Through modulation of PDGFRβ expression, it may also contribute to the response of HSCs to microvascular thrombosis and platelet-derived signals. A recent study explored the effect of the non-bile acid TGR5 ligand oleanolic acid in a rat model of liver fibrosis development [58]. Treatment with oleanolic acid significantly reduced fibrogenesis in vivo; however, a direct effect on HSCs in vitro was not observed, since the cell lines used did not express TGR5 [58]. It was previously reported that TGR5 expression, which is very low in quiescent HSCs, is upregulated during the activation of HSCs into a myofibroblast-like phenotype [4,26]. Thus, stimulation of TGR5 on activated HSCs may contribute to the beneficial effects of oleanolic acid on fibrosis development in the above model [58]. Furthermore, mice deficient in both TGR5 and the nuclear bile acid receptor FXR were generated and showed an enrichment in gene expression pathways associated with liver fibrosis and inflammation in line with our study [59]. Since FXR agonists convey beneficial effects in preclinical models of liver fibrosis and portal hypertension [60], it is highly anticipated that TGR5/FXR double-KO mice will also develop LSEC and HSC dysfunction and portal hypertension. Thus, targeting of BA signaling becomes an interesting strategy to reverse morphological changes and dysfunction of cell types residing in the hepatic sinusoids, thereby attenuating portal hypertension and fibrosis development.