The G-protein coupled bile salt receptor TGR5 is expressed in liver sinusoidal endothelial cells


  • Potential conflict of interest: Nothing to report.


Sinusoidal endothelial cells (SEC) constitute a permeable barrier between hepatocytes and blood. SEC are exposed to high concentrations of bile salts from the enterohepatic circulation. Whether SEC are responsive to bile salts is unknown. TGR5, a G-protein–coupled bile acid receptor, which triggers cAMP formation, has been discovered recently in macrophages. In this study, rat TGR5 was cloned and antibodies directed against the C-terminus of rat TGR5 were developed, which detected TGR5 as a glycoprotein in transfected HepG2-cells. Apart from Kupffer cells, TGR5 was detected in SEC of rat liver. SEC expressed TGR5 over the entire acinus, whereas endothelial cells of the portal or central veins were not immunoreactive toward TGR5 antibodies. In isolated SEC, TGR5 mRNA and protein were detected by reverse transcription (RT) PCR, immunofluorescence microscopy, and Western blot analysis. Bile salts increased cAMP in isolated SEC and induced mRNA expression of endothelial NO synthase (eNOS), a known cAMP-dependent gene. In addition, bile acids activated eNOS by phosphorylation of eNOS at amino acid position 1177. In line with eNOS activation, bile acids induced NO production in liver slices. This is the first report on the expression of TGR5 in SEC. Conclusion: The data suggest that SEC are directly responsive toward specific bile salts. Regulation of eNOS in SEC by TGR5 connects bile salts with hepatic hemodynamics. This is of particular importance in cholestatic livers when bile salt concentrations are increased. (HEPATOLOGY 2007;45:695–704.)

Bile salts are required for cholesterol excretion and lipid absorption.1 However, high concentrations of lipophilic bile salts have toxic effects. Bile salts alter membrane fluidity2 and can act pro- or anti-apoptotic.3–6 Many liver diseases are aggravated by the cholestatic potential of lipophilic bile salts. Therefore, several mechanisms exist to maintain bile salt homeostasis. These include the coordinated expression and action of bile salt transporters at the sinusoidal and canalicular membrane of liver parenchymal cells,7 alternative pathways for bile salt synthesis and metabolism,8, 9 and the involvement of extrahepatic tissues (such as the gut and the kidneys) in bile salt excretion.10–12 These mechanisms are closely regulated by nuclear receptors sensitive for bile salts, which control the expression of transporter proteins and enzymes. They comprise the farnesoid X receptor,13–15 the pregnane X receptor,16, 17 and the vitamin D receptor.18 Recently, a G-protein–coupled plasma membrane receptor responsive to bile salts has been discovered by high-throughput screening. This receptor, named TGR5,19 M-BAR, or BG37,20 stimulates adenylate cyclase on activation and increases the production of cyclic adenosine monophosphate (cAMP). Thereby, bile salts not only may be involved in the regulation of transcription but also may influence rapid, cAMP-dependent mechanisms in TGR5 expressing cells. So far, TGR5 expression has been demonstrated in enteroendocrine cells,21 where bile salts stimulate the secretion of glucagon-like peptide-1 via TGR5 and in alveolar macrophages,19 which secrete smaller amounts of cytokines in response to endotoxin, when bile salts are present. Recently, bile salts were shown to influence energy consumption in brown adipose tissue involving the TGR5–cAMP pathway.22 Which liver cells express TGR5 remains elusive. In the current study a TGR5-specific antibody was developed and was used to identify the expression pattern of TGR5 in the liver. TGR5 was found not only in liver macrophages (Kupffer cells) but also in the plasma membrane of sinusoidal endothelial cells, where bile salts led to an increase in intracellular cAMP and to the activation and enhanced expression of the endothelial nitric oxide (NO) synthase.


cAMP, cyclic adenosine monophosphate; DAF-FM-DA, 4-amino- 5-methylamino-2′,7′-difluorofluorescein diacetate; eNOS, endothelial NO synthase; KC, Kupffer cells; NO, nitric oxide; SEC, sinusoidal endothelial cells; TC, taurocholate; TCDC, taurochenodeoxycholate; TLC, taurolithocholate.

Materials and Methods

Cell culture media and geneticin were from Gibco Life Technologies (Gaithersburg, MD). Penicillin/streptomycin and trypsin were from Biochrom (Berlin, Germany). Fetal calf serum was from PAA Laboratories (Pasching, Austria). Bile salts were from Sigma-Aldrich (Deisenhofen, Germany). Forskolin was from Calbiochem (Schwalbach, Germany). AccuPrime Taq DNA Polymerase and Lipofectamine 2000 transfection reagent were from Invitrogen (Carlsbad, CA). BglII, KpnI were from New England BioLabs (Beverly, MA). All other chemicals were of analytical grade and purchased from either Sigma or Merck (Darmstadt, Germany).


Polyclonal antibodies were raised in a guinea pig (M38) and a rabbit (K36) against a 24–amino acid oligopeptide containing amino acids 306 through 329 from the C-terminus of rat TGR5 (KRANPGPSTAYHSSSQCSTDLDLN). The peptide was synthesized automatically, coupled via the N-terminal lysine to ovalbumin and used for immunization of a guinea pig and a rabbit, respectively. The mouse monoclonal anti-Reca-1 antibody, which recognizes rat endothelial cells, and the anti-rat CD163 (ED2) were purchased from Serotec (Oxford, UK).23 The monoclonal anti–GFP antibody (sc-9996), known to be reactive for YFP,24 was from Santa Cruz (Santa Cruz, CA). The anti-nucleoporin antibody was from BD Biosciences (Paolo Alto, CA). The anti-phospho- endothelial NO synthase (eNOS) (Ser1177) antibody was from Cell Signaling (Beverly, MA), the monoclonal anti-eNOS (Clone 3) was from BD Transduction Laboratories (San Jose, CA), the rabbit anti-CD95 antibody (C20) and mouse anti-CD95 (M20) were from Santa Cruz, and the mouse anti-phosphoserine was from Biomol (Hamburg, Germany).

Isolation and Cultivation of Kupffer Cells and Sinusoidal Endothelial Cells.

Cells were isolated from 1-year-old male Wistar rats by collagenase-pronase perfusion. Experiments were approved by the local ethical authority. Cells were separated by a single Nycodenz gradient and centrifugal elutriation as described previously.25, 26 Kupffer cells (KC) were seeded onto culture plates, and sinusoidal endothelial cells (SEC) were plated onto collagen-1–coated (Sigma) culture dishes. Both KC and SEC were maintained in RPMI 1640 medium supplemented with 10% fetal calf serum and 1 % penicillin/streptomycin.

Cloning of Rat TGR5 cDNA.

Total RNA was prepared from isolated rat Kupffer cells using the RNA extraction kit (Qiagen, Hilden, Germany). Complementary DNA was obtained with the first-strand cDNA synthesis kit (Roche, Mannheim, Germany) and served as template for the subsequent PCR. The forward primer was 5′-gaagatctatgatgtcacacaacaccactgagct-3′ and contained a BglII restriction site upstream of the start codon. The stop codon in the reverse primer was replaced by a KpnI restriction site, the primer was 5′-ccggtaccgaattcaagtccaagtcagtgctgc-3′. The resulting PCR product was cloned into the pCR2.1Topo-Vector (Invitrogen). After propagation in Escherichia coli, the insert was excised with BglII and KpnI and ligated into the pYFPN1 vector (Clontech, Palo Alto, CA). The sequence of the resulting plasmid TGR5-YFP was confirmed by sequencing and alignment with the respective reference sequence (accession number NM_177936).

Cell Culture and Stable Expression in HepG2 Cells.

HepG2- and Hek293-cells were cultured in Dulbecco's modified Eagle's medium-Nutrimix F12 with 10% fetal calf serum and were kept at 37°C and 5% CO2. Rat TGR5-YFP was transfected into HepG2 cells with LipofectAMINE 2000 according to the manufacturer's guidelines. After 48 hours, the cells were split, and stable transfectants were selected using medium containing 300 μg/mL geneticin. Resistant clones were screened by immunofluorescence microscopy and immunoblot analysis for TGR5-YFP expression.

Quantitative Reverse Transcription Polymerase Chain Reaction.

Total RNA was isolated using the RNA extraction kit (Qiagen), and cDNA was obtained with the QuantiTect Reverse Transcription Kit (Qiagen), which includes a DNAse I digestion. The level of gene expression in SEC and KC was measured by real-time SYBR Green PCR with the 7500 Real-time PCR System (Applied Biosystems, Foster City, CA) as described previously.27 In brief, data were produced in triplicates for each gene. Mean values of cycle numbers of the target genes (TGR5, eNOS, ICAM) were subtracted from the mean of cycle numbers of the housekeeping gene beta-glucuronidase (GUS), for the respective sample. This value taken to the power of 2 is the mRNA expression of the target gene in relation to GUS expression, termed “relative mRNA unit.” Real-time primers were as follows: TGR5-forward: 5′-aaaggtggctacaagtgcttc-3′; TGR5-reverse: 5′-ttcaagtccaagtcagtgctg-3′; GUS-forward: 5′-ggatccacctcgcatgttc-3′; GUS-reverse: 5′-ttctccaggagagctgtctgg-3′, eNOS-forward: 5′-gtgctggcatacagaaccca-3′, eNOS-reverse: 5′-gaccccatagtgcagagggc-3′, ICAM-forward: 5′-atcgggatggtgaagtctgtc-3′, ICAM-reverse: 5′-gctaaaggcacggcacttgt-3′.

Measurement of cAMP in Isolated SEC.

Intracellular cAMP was measured by the cyclic AMP(3H) assay system from Amersham Biosciences (Freiburg, Germany) according the manufacturer's instructions. SEC were cultured at a density of 6 × 106 cells/6-cm dish. After stimulation for 4 minutes, SEC were washed with ice-cold phosphate-buffered saline (PBS), harvested with 120 μL Tris-EDTA buffer (0.05 M Tris pH 7.5; 4 mM EDTA), incubated at 80°C for 2 minutes, and centrifuged at 12,000g for 1 minute. Cyclic AMP was analyzed from the supernatants. Aliquots from each sample were taken for protein determination using the Advanced Protein Assay (Cytoskeleton, Denver, CO) and the amount of cAMP (pmol/mg protein) was calculated. Relative increase in intracellular cAMP concentration per sample was determined in relation to untreated control SEC.

Immunofluorescence and Confocal Laser Scanning Microscopy.

For immunofluorescence studies, cells were grown on glass coverslips (48 hours for HepG2, 24 hours for SEC) and fixed in 100% methanol (5 minutes, −20°C). Cryosections (5 μm) of perfused rat livers were prepared with a Leica cryotome, air-dried for 1 hour, and fixed in 100% methanol at −20°C for 5 minutes. Immunofluorescence was performed as described previously.27 The primary antibody dilutions were as follows: M38 1:50, K36 1:150, Reca-1 1:20, and CD163 (ED2) 1:50. Fluorescein or cyanine-3–conjugated secondary antibodies (Jackson Immuno Research Laboratories, West Grove, PA) were diluted 1:100 and 1:500, respectively. Immunostained samples were analyzed with a Zeiss LSM 510 META confocal microscope (Zeiss, Oberkochen, Germany).

Immunoprecipitation and Immunoblotting.

For the detection of TGR5, cells were homogenized with a Labsonic 2000 sonifier (B. Braun Biotech, Melsungen, Germany) in hypotonic buffer (0.1 mM EDTA, 0.5 mM sodium phosphate, pH 7.0) containing complete protease inhibitor cocktail (Roche). Crude membrane fractions were collected by centrifugation (20,000g, 4°C, 1 hour) and resuspended in hypotonic buffer. For detection of phosphorylated and total eNOS SEC were lysed in a buffer consisting of 20 mM Tris-HCl (pH 7.4), 1% Triton X-100, 140 mM NaCl, 10 mM Na4P2O2, 1 mM EDTA, 1 mM EGTA, 10 mM NaF, 1 mM sodium vanadate, 20 mM β-glycerophosphate, and complete protease inhibitor cocktail. Immunoprecipitation of CD95 was performed as described recently in detail.28 Equal protein amounts were separated by SDS-PAGE and blotted onto PVDF membranes. Proteins were visualized by enhanced chemiluminescence. Primary antibodies were diluted as follows: M38 1:5000, anti-GFP 1:1,000, anti-eNOS 1:5000, anti-phospho-eNOS 1:1000, anti-CD95 (M20) 1:2500, anti-phosphoserine 1:2500. Densitometry was performed using the Totallab100 software (Nonlinear Dynamics, Durham, NC).

Liver Slice Preparation and NO Measurements.

Livers were explanted from anesthetized male Wistar rats (180–200 g) and slices of 400–500 μm thickness were cut from a single liver lobule. Slices were immediately placed in ice-cold Dulbecco's modified eagle medium containing 1 g/L D-glucose without phenol red (GibcoBRL Life-Sciences, Gaithersburg, MD). For detection of nitric oxide, liver slices were incubated with 5 μM of 4-amino-5-methylamino-2′,7′-difluorofluorescein diacetate (DAF-FM-DA) for 60 minutes at 37°C and 5% CO2. DAF-FM-DA is taken up by cells, deacetylated by intracellular esterases to the nonfluorescent DAF-FM. Nitric oxide rapidly oxidizes DAF-FM to a fluorescent benzotriazole, which detects nitric oxide down to a concentration of 3 nM.29 After washing with Dulbecco's modified eagle medium, slices were examined on an inverted fluorescence microscope (Zeiss, Oberkochem, Germany) using a 100× lens. Excitation wavelength was 488 nm (generated by a monochromator), emission was measured at 515 to 565 nm wavelength using a CCD camera (QuantiCell 2000, VisiTech, Sunderland, UK). Emission intensity of unstimulated liver slices (control) was set to 1, and changes in emission intensity were expressed as relative changes as compared with control.


Membrane fractions (50 μg) were denatured by incubation with 1% SDS in a total volume of 10 μL at 37°C for 30 minutes. One hundred microliters digestion buffer (17 mM NaH2PO4; 33 mM Na2HPO4; 0.2 mM NaN3; 5 mM EDTA, 1% Triton X-100) were added to the denatured proteins in the presence or absence of 5 units of peptide N-glycosidase F (EC (Roche). After overnight incubation at 37°C, samples were examined by immunoblot analysis.

Reporter Gene Assay.

HepG2- or Hek293-cells were transfected with a cAMP-sensitive reporter gene construct (Plasmuc, Bayer AG, Leverkusen, Germany), containing five cAMP-responsive elements in front of the luciferase gene. This plasmid was co-transfected with a Renilla expression vector (Promega, Madison, WI) and rTGR5-YFP or with pYFP-N1 (control) using LipofectAMINE (Invitrogen) according to the manufacturer's instructions. Forty-eight hours after transfection, cells were incubated with or without taurolithocholate or forskolin for another 16 hours. Cell lysis and luciferase assays were performed using the dual-luciferase kit (Promega) according to the manufacturer's guidelines. Transfection efficiency was monitored by the co-transfection of a Renilla expression vector (Promega), and luciferase activity was normalized to Renilla luminescence.


Values from RT-PCR and densitometric analysis are given as means ± standard errors of mean. The one-sided Student t test was used for statistical analysis, with a P < 0.05 considered to be statistically significant.


Expression of Rat TGR5 in HepG2 Cells.

Rat TGR5 was cloned and fused to the enhanced yellow fluorescent protein (YFP). In stably transfected HepG2-cells, most of TGR5-YFP was localized to the cell membrane, where it was clearly detected by the guinea pig antibody directed against rat TGR5 (M38) (Fig. 1A–C). The pre-immune serum from the same animal showed no specific immunoreactivity (pre-M38, Fig. 1D–F). Similarly, preincubation of the antibody M38 with the peptide, which was used for immunization of the guinea pig, resulted in a complete disappearance of the immunoreactivity toward M38, demonstrating the specificity of this antibody (Fig. 1G–I). There was no cross-reactivity toward endogenous TGR5 of HepG2-cells, because M38 showed specific staining only in transfected, but not in untransfected, HepG2-cells (Fig. 1J–L, arrows). Hek293-cells were co-transfected with TGR5-YFP and a cAMP reporter plasmid as described in Materials and Methods. Vector-transfected (pYFP) Hek293-cells were used as controls. Luciferase activity served as a measure of intracellular cAMP elevation. Forskolin, a direct activator of adenylate cyclase, stimulated luciferase activity independent of TGR5-expression. In contrast, taurolithocholate increased luciferase activity only in TGR5-expressing Hek293-cells by 3.6 ± 0.7-fold but did not change luciferase activity in normal Hek293-cells (Fig. 2A). Similar results were obtained in HepG2-cells, although triple transfection was less effective in these cells. The expected molecular mass of the TGR5-YFP fusion protein is approximately 64 kDa. In whole cells lysates of HepG2-cells stably expressing TGR5-YFP, a broad band of approximately 105 kDa was detected by an anti-GFP/YFP antibody that was not visible in untransfected cells (Fig. 2B). When membrane fractions were treated with N-glycosidase F (PNGase F) to remove all N-linked complex sugar chains, a shift of this band from 105 to approximately 64 kDa was seen in Western blot analysis using either the anti-GFP/YFP or the anti-TGR5 antibody (Fig. 2C). Sensitivity toward PNGase F indicates that rat TGR5-YFP is heavily glycosylated in HepG2-cells.

Figure 1.

Characterization of the polyclonal anti-rat TGR5 antiserum (M38) by immunofluorescence microscopy. Recombinant rat TGR5-YFP was localized in the plasma membrane of stably transfected HepG2-cells (green in B, E, H and K). The fluorescence obtained with the M38 antiserum (red) co-localized with the TGR5-YFP fluorescence (C, L). The M38 preimmune serum showed no specific staining (D, F). Preincubation of the M38 antiserum with the synthetic M38 peptide completely abolished the membrane staining of transfected HepG2-cells (G, I). In transiently transfected HepG2-cells, M38 only stained TGR5-YFP expressing cells (J-L), whereas untransfected cells (arrows) did not show specific staining. Bars = 10 μm.

Figure 2.

Functional and immunoblot analysis of recombinant TGR5-YFP in transfected cells. (A) Taurolithocholate increased luciferase activity from a 5× cAMP-responsive elements reporter plasmid in TGR5 expressing, but not in Hek293-cells transfected with the control plasmid (pYFP). Forskolin, which was used as a positive control, increased cAMP independent of TGR5 expression. (B) The anti-GFP/YFP antibody (GFP-ab) detected rat TGR5-YFP in transfected but not in un-transfected HepG2-cells. (C) Both the anti-GFP/YFP antibody as well as the M38 antiserum (M38-ab) detected a broad band of approximately 105 kDa and a minor band of approximately 64 kDa. Incubation of the crude membrane fractions with peptide-N-glycosidase F (PNGase F) reduced the molecular mass of TGR5-YFP to approximately 64 kDa.

Localization of TGR5 in Rat Liver.

Rat liver slices were stained with the M38 antibody. The strongest immunoreactivity was detected in extended structures radiating toward the central vein. Co-staining with an anti-Reca-1 antibody,23 which is directed against an epitope present in all endothelial cells, uncovered a co-localization between Reca-1 and TGR5 within the endothelium of the acinus but not in vessels of the portal field (Fig. 3A–C) or of the central veins (Fig. 3D–F). These findings suggest that TGR5 is specifically expressed in SEC but not in other liver endothelium. In addition some immunoreactivity was found distant from SEC (Fig. 3G–I, arrows). This immunoreactivity of minor intensity was attributed to Kupffer cells (liver macrophages), because in these areas, TGR5-immunoreactivity co-localized with ED2, a marker of macrophages30 (Fig. 3J–L). Kupffer cells belong to CD14-positive cells, which were shown to have the highest TGR5-mRNA expression level compared with other white blood cells.19

Figure 3.

Immunofluorescence localization of TGR5 in rat liver. Cryosections from perfused normal rat liver were incubated with the M38 antiserum to detect TGR5 (red). TGR5 was localized to the endothelial cells of the sinusoids as demonstrated by co-localization with the immunofluorescence pattern of the Reca-1 antibody, a marker of endothelial cells (green). The endothelium of vessels of the portal field (PF) (delineated by a white line) was Reca-1 positive but was not reactive to the M38 antiserum (A-C). Similarly, Reca-1 stained the central veins (CV, double arrow), which were not reactive to M38 (D-F). SEC in the immediate vicinity of the central vein were reactive to RECA-1 and the TGR5 antibody (simple arrows). Some M38-immunoreactivity was observed in some distance to Reca-1–positive endothelial cells (G-I, arrows), which could be attributed to KC because of the colocalization with ED2 (J-L). Bars = 10 μm.

Expression of Functional TGR5 in Sinusoidal Endothelial Cells.

Crude membrane fractions of isolated SEC were separated by sodium dodecyl sulfate polyacrylamide gel electrophoresis. The guinea pig antibody M38 and the rabbit antibody K36, both directed against the same epitope of rat TGR5, detected an equivalent band of approximately 65 kDa (Fig. 4A). Treatment with PNGase F shifted this band to 45 kDa (Fig. 4B).

Figure 4.

Expression of TGR5 in isolated sinusoidal endothelial cells (SEC). (A) M38 and K36 antisera detected a broad band of approximately 70 kDa in crude membrane fractions of isolated SEC. (B) In crude membrane fractions of SEC, a band of lower molecular mass appeared after treatment with PNGase F. (C) Immunofluorescence localization of TGR5 in SEC. Nuclei were stained with nucleoporin (green). (D) Relative TGR5-mRNA expression is stronger in SEC as compared with Kupffer cells (KC).

In SEC cultured for 24 hours, M38 led to a reticular staining pattern (Fig. 4C). Because SEC possess a discontinuous surface due to their fenestrae, this pattern might well reflect a membranous distribution of TGR5. Relative mRNA levels of TGR5 were determined by RT-PCR, in relation to the expression level of the beta-glucuronidase as a housekeeping gene. Relative mRNA was higher in SEC as compared with Kupffer cells (Fig. 4D), which is in accordance with the stronger immunoreactivity in SEC compared with Kupffer cells in liver slices. Physiologically relevant bile acids and their taurine-conjugates stimulate TGR5 in the order taurolithocholate (TLC) ≥ taurodeoxycholate > taurochenodeoxycholate (TCDC) > taurocholate (TC).19, 20 Here we tested TLC, TCDC, and TC in their potential to stimulate cAMP production in SEC. In unstimulated SEC, cAMP concentration was 2.1 ± 0.7 pmol/mg protein (n = 4). Incubation of SEC with TC, TCDC, or TLC (25 μM each, n = 4) for 4 minutes induced an increase of intracellular cAMP by 3.2 ± 0.9-fold; 2.4 ± 0.9-fold, and 2.9 ± 0.8- fold, respectively, compared with control (Fig. 5A). Forskolin (10 μM) increased cAMP 39.6 ± 3.7-fold (n = 4). In line with the known potency to stimulate cAMP production,19, 20 TLC induced a maximum of cAMP already at a concentration of 10 μM, whereas higher concentrations of TC compared with TLC were needed for a similar effect on cAMP production (Fig. 5B). When mRNA of SEC was isolated 12 hours after stimulation with TLC (100 μM), a 1.6 ± 0.1-fold (n = 3) increase of eNOS-mRNA was detected in TLC-treated SEC compared with unstimulated SEC. Forskolin (10 μM) increased eNOS-mRNA 1.9 ± 0.3-fold. In contrast, ICAM-mRNA was not significantly changed by TLC or forskolin (Fig. 5C). Production of nitric oxide (NO) by eNOS depends on the amount but also on the activity of eNOS. Because phosphorylation of eNOS at a serine at amino acid position 1177 is critical for its activity, the phosphorylation state of eNOS was measured in SEC, which were treated with TLC or forskolin for 30 minutes. TLC (25 μM) induced a 2.4 ± 0.1-fold increase (n = 5) of P-Ser-eNOS compared with control, similar to forskolin (10 μM), which increased P-Ser-eNOS by 3.3 ± 0.6-fold (n = 5) (Fig. 6A,B). The data suggest that bile acids instantaneously activate eNOS in SEC involving the TGR5-cAMP pathway. When rat liver slices were incubated with TLC (25 μM), NO production increased within minutes as measured by DAF-FM fluorescence (Fig. 6C). As a positive control, peroxynitrite (ONOO, 5 mM) was added at the end of the experiment, leading to a strong increase in DAF-FM fluorescence. The spatial distribution of bile salt–induced changes of DAF-FM fluorescence was compatible with a sinusoidal pattern (Fig. 6D). Because SEC are the main source for NO production in the liver, the results confirm the functional activity of TGR5 in SEC. The CD95 receptor (FAS-R), which is expressed in SEC,31 is another potential target of cAMP action. CD95 is known to be involved in bile acid–induced apoptosis5; however, cAMP-dependent phosphorylation of CD95 at serine residues was recently shown to prevent apoptosis in liver parenchymal cells.28 Treatment of SEC with TLC resulted in serine phosphorylation, as shown by immunoprecipitation of CD95 and subsequent sodium dodecyl sulfate polyacrylamide gel electrophoresis and probing with a phosphoserine-specific antibody (Fig. 7). The relative increase of phosphorylated CD95 was 1.5 ± 0.5-fold in TLC-treated and 1.7 ± 0.5-fold in forskolin-treated SEC (n = 4), when control was set to 1.

Figure 5.

Bile salts stimulate cAMP production and eNOS-mRNA expression in sinusoidal endothelial cells (SEC). (A) TLC, TC, and TCDC (all 25 μM, 4 minutes) stimulated cAMP production in SEC compared with untreated SEC. (B) TC increased cAMP production in a dose-dependent manner with a maximum at a concentration of 50 μM, whereas TLC at a concentration of 10 μM already induced maximal cAMP production. (C) TLC significantly induced an increase in eNOS-mRNA expression similar to forskolin, whereas ICAM-mRNA expression was not affected in SEC.

Figure 6.

eNOS is activated by bile salts. (A) TLC (25 μM) induced the phosphorylation of eNOS as detected by an antibody specific for 1177P-Ser-eNOS, which is essential for the activity of eNOS. As a positive control, stimulation of the adenylatecyclase by forskolin (10 μM) had a similar effect (representative immunoblot). Total eNOS was not different in control SEC, TLC, or forskolin-treated SEC. (B) Densitometric analysis of five immunoblots from five independent cell preparations (Ser1177 phosphorylated eNOS normalized to total eNOS). The changes were statistically significant with a P < .05. (C) Liver slices were loaded with the NO-sensitive fluorescent dye DAF-FM (see Materials and Methods for details). After addition of TLC to the liver slice, fluorescence increased, indicating the reaction of DAF-FM with NO. Peroxynitrite, which induced a strong increase in fluorescence, served as a positive control (representative trace from four independent experiments). (D) Fluorescence images from a DAF-FM loaded liver slice before (control), 2, 4, and 6 minutes after addition of TLC (25 μM). The spatial distribution of DAF-FM fluorescence, which was induced by TLC, showed a sinusoidal pattern. Bar = 10 μm.

Figure 7.

TLC-induced serine-phosphorylation of CD95. SEC were treated with 25 μM TLC. After cell lysis CD95 was immunoprecipitated and serine-phosphorylated versus total CD95 was determined. TLC induced a significant increase in serine phosphorylation of CD95 as compared with control, similar to forskolin as a positive control.


In recent years direct effects of bile salts on several physiological processes have been described, which are mediated by a set of nuclear receptors.7 These nuclear receptors act as transcription factors, which regulate transporter proteins and enzymes on a long-term scale at the transcriptional level. TGR5, the first plasma membrane–bound bile salt–sensitive receptor, was discovered recently.19, 20 TGR5 is expressed in many organs such as liver, lung, spleen, kidney, uterus, placenta, heart, and brain but also in leukocytes.19, 20, 22 In leukocytes, it was especially expressed in CD14-positive cells, which represent monocytes/macrophages. Apart from CD14-positive cells, TGR5 has been detected in an enteroendocrine cell line21 and human skeletal muscle myofibroblasts22 at the cellular level. Our study demonstrates that TGR5 is expressed in liver macrophages (Kupffer cells) and to a higher extent in sinusoidal endothelial cells, whereas endothelial cells from the portal or central veins or from branches of the hepatic artery do not express TGR5. Sinusoidal endothelial cells (SEC) have an exceptional position within the enterohepatic circulation. They are exposed to highly variable concentrations of nutrients, including bile salts. The concentration of bile salts in portal blood increases in response to food intake after gallbladder contraction and re-uptake of bile salts from the intestine. Under pathological conditions such as chronic liver diseases, increased systemic bile salt concentrations due to defects in bile salt secretion may occur.27, 32 Expression of TGR5 by SEC provides the possibility to adapt to changing bile salt concentrations. TGR5 has been shown to signal via cAMP.19, 20, 22 In line with these results, primary and secondary bile salts induce an increase of cAMP in SEC. The effects of cAMP in SEC remain elusive. However, by immunohistochemistry, a high concentration of cAMP was detected in a sinusoidal pattern,33 which may be attributed to SEC. Cyclic AMP may protect SEC from injury induced by cold storage/reperfusion34 involving the adenosine A2 receptor.35 Cyclic AMP inhibits apoptosis in many cell types and was recently shown to induce serine-threonine-phosphorylation of the CD95 receptor in a PKA-dependent mechanism.28 CD95 phosphorylation at these residues acts anti-apoptotic by abrogating the signaling downstream of CD95.28 In SEC, treatment with TLC (as well as cAMP- elevation by forskolin) led to serine-threonine-phosphorylation of the CD95 receptor, which may be connected to an anti-apoptotic mechanism in SEC mediated by TGR5/cAMP. Such a proposed mechanisms may be especially relevant for SEC, which are exposed to temporarily high concentrations of bile salts and may protect from bile acid–induced SEC injury. SEC differ from other endothelial cells because of their fenestrae and lack of a basal membrane.36–39 Fenestrae are dynamic structures that allow the access of larger molecules (such as albumin) to liver parenchymal cells,40 thus enhancing the hepatic metabolic activity.39 Several vasoconstrictors were shown to induce contractions of fenestrae, which may represent an essential regulatory site for sinusoidal blood exchange with the space of Dissé. For example, endothelin-1 causes a contraction of the sinusoidal fenestrae via the endothelin B receptor,41 which can signal through cAMP.42 This pathway also may be triggered by bile acids via TGR5. Whereas the relevance of SEC contractions for vascular resistance and overall portal pressure under healthy conditions remains controversial,38, 43 SEC have a central role as a donor of nitric oxide (NO).44 NO production by SEC relates to the expression of eNOS rather than the inducible isoform of NO synthase.45 In the presence of eNOS, endothelial cells produce NO, which mediates vasodilation in a paracrine manner. Inadequate production of NO is an important factor in the development of portal hypertension.46 In line with this, a fourfold increase in eNOS expression was found in cirrhotic as compared with noncirrhotic human livers.47 Because bile salts are increased in cirrhosis and the promoter of the human eNOS contains a cAMP-responsive elements,48 the increase in eNOS expression in cirrhotic livers may be partly mediated by the “bile salt–TGR5–cAMP pathway” as supported by our data. Apart from effects of NO on vascular resistance, NO may protect from oxidative stress and lipid peroxidation.49, 50 Because bile salts can cause the production of reactive oxygen species,6 rapid NO release from SEC in response to bile salts as shown here may represent another protective mechanism. Taken together, the detection of the membrane-bound bile acid receptor in SEC points to another, not yet recognized role of bile salts in liver physiology, and a direct link between bile salt concentration and hepatic microcirculation can be postulated based on our results.


Expert technical assistance by Elisabeth Winands and Stefanie Winandy is acknowledged.