Potential conflict of interest: Nothing to report.
This study was supported by the Association pour la Recherche contre le Cancer (3435) and by the Agence Nationale de la Recherche (PHYSIO 2007). T.T. is supported by Assistance Publique–Hôpitaux de Paris. The authors thank Galya Vassileva and the Merck Research Laboratories (Kenilworth, NJ) for providing us with the C57Bl/6 Gpbar1−/− mice.
Address reprint requests to: Thierry Tordjmann, M.D., Ph.D., INSERM U.757, Université Paris Sud, Building 443, 91405 Orsay, France. E-mail: email@example.com; fax: +33 1 69 15 58 93.
Many regulatory pathways are involved in liver regeneration after partial hepatectomy (PH) to initiate growth, protect liver cells, and sustain functions of the remnant liver. Bile acids (BAs), whose levels rise in the blood early after PH, stimulate both hepatocyte proliferation and protection, in part through their binding to the nuclear farnesoid X receptor (FXR). However, the effect of the BA receptor, TGR5 (G-protein-coupled BA receptor 1) after PH remains to be studied. Liver histology, hepatocyte proliferation, BA concentrations (plasma, bile, liver, urine, and feces), bile flow and composition, and cytokine production were studied in wild-type (WT) and TGR5 KO (knockout) mice before and after PH. BA composition (plasma, bile, liver, urine, and feces) was more hydrophobic in TGR5 KO than in WT mice. After PH, severe hepatocyte necrosis, prolonged cholestasis, exacerbated inflammatory response, and delayed regeneration were observed in TGR5 KO mice. Although hepatocyte adaptive response to post-PH BA overload was similar in WT and TGR5 KO mice, kidney and biliary adaptive responses were strongly impaired in TGR5 KO mice. Cholestyramine treatment, as well as Kupffer cell depletion, significantly improved the post-PH TGR5 KO mice phenotype. After bile duct ligation or upon a cholic acid–enriched diet, TGR5 KO mice exhibited more severe liver injury than WT as well as impaired BA elimination in urine. Conclusion: TGR5 is crucial for liver protection against BA overload after PH, primarily through the control of bile hydrophobicity and cytokine secretion. In the absence of TGR5, intrahepatic stasis of abnormally hydrophobic bile and excessive inflammation, in association with impaired bile flow adaptation and deficient urinary BA efflux, lead to BA overload-induced liver injury and delayed regeneration. (Hepatology 2013;58:1451–1460)
After a two-thirds partial hepatectomy (PH), the rodent liver is restored to its initial mass within a few days, during which time a complex array of proliferative and hepatoprotective signaling cascades operates, involving cytokines, growth factors, and other paracrine and endocrine agonists.[1, 2] One of the first physiological consequences of PH is an immediate increase in blood bile acid (BA) concentration, reported to signal, by farnesoid X receptor (FXR) stimulation, for the need of hepatocyte division and protection. After PH, the remnant liver adapts to this immediate BA overload by down-regulating BA synthesis and uptake through FXR-dependent pathways.[5, 6] BAs also signal through the membrane-bound receptor, G-protein-coupled BA receptor 1 (GPBAR1, or TGR5) expressed in numerous cell types and organs[7, 8], whose functions in the liver in general, and during regeneration in particular, remain to be defined.
TGR5 is a G-protein-coupled receptor, from which activation by BA induces cyclic adenosine monophosphate (cAMP) synthesis. It is considered as a crucial regulator of energy homeostasis, as well as as a potential target for the treatment of metabolic syndrome and its complications, including nonalcoholic steatohepatitis (NASH), in the context of diabetes and obesity.[10, 11] TGR5 has not been significantly detected in rodent hepatocytes, whereas its activation by BA stimulates nitric oxide (NO) production by rat liver endothelial cells and decreases lipopolysaccharide (LPS)-induced cytokine gene induction in rat Kupffer cells (KCs). These anti-inflammatory properties have been reported to be the result of an inhibition of nuclear factor kappa B (NF-κB) signaling.[10, 14] TGR5 has also been proposed to play a role in the control of cholangiocyte chloride (Cl−) secretion in human gallbladder and in gallbladder-filling regulation.[16, 17] Because liver regeneration is associated with finely tuned inflammatory pathways and biliary homeostasis adaptive responses, we hypothesized that TGR5 might play a regulatory role after PH.
In this study, we provide evidence that, in TGR5 knockout (KO) mice, PH is followed by massive cholestasis and hepatocyte necrosis, and that liver regeneration is markedly delayed, as compared to wild-type (WT) mice. Based on data from several in vivo models of BA overload, our study suggests that TGR5 after PH may protect the BA-overloaded remnant liver primarily through control of BA hydrophobicity and through a fine-tuning of inflammatory processes; we also suggest that TGR5 regulates ion exchange in bile and BA efflux in urine, providing further protection against BA overload.
Materials and Methods
Surgical Procedures Used on Animals
C57Bl/6 Gpbar1−/− mice (referred to in this study as TGR5 KO mice) and their C57Bl/6 WT littermates were provided by Merck Research Laboratories (Kenilworth, NJ) and used to found our colonies of TGR5 KO and control animals. TGR5-overexpressing transgenic mice were generated as previously described. The study was performed on male 10-16-week-old mice. Two-thirds PHs were performed as previously described. Bile duct ligation (BDL) and bile flow measurements were performed as previously described.
In some experiments, liposomal clodronate was injected (retro-orbital) 48 hours before inclusion in the experiments to eliminate KCs.
Tissue fragments were removed at various times after surgery and either frozen in nitrogen-cooled isopentane and stored at −80°C until use or fixed in 4% formaldehyde and embedded in paraffin.
Further Applied Methods
Additional animal treatments, immunohistochemistry, immunoblottings, biochemical assays, and reverse-transcriptase polymerase chain reaction (PCR) experiments followed standard procedures and are further described in the Supporting Materials.
The Student t test was used to compare sample means with paired controls. Results are expressed as means ± standard error of the mean. P values ≤0.05 were considered statistically significant.
Surgical Lack of TGR5 Impairs Liver Regeneration and Results in Cholestasis and Liver Injury After PH
WT and TGR5 KO mice had similar body weights; however, KO mice had a significantly smaller liver/body-weight ratio (Fig. 1A,B). Hepatocyte size, analyzed on phalloidin-stained liver sections, was similar in WT and TGR5 KO mice (Supporting Fig. 1A). Hematoxylin and eosin (H&E) staining revealed normal liver histology in the majority of WT and TGR5 KO mice, although approximately 20% of TGR5 KO mice exhibited mild portal inflammation and fibrosis (Supporting Fig. 1B). Basal biochemical blood parameters (alanine aminotransferase [ALT], alkaline phosphatase [ALP], bilirubin, total bile acids [TBA], glucose, and insulin concentrations) fell in the normal range in both genotypes (Supporting Table 2).
After PH, TGR5 KO mice exhibited a significantly slower liver mass restoration (Fig. 1C and Supporting Fig. 2A,B) and a reduced mitotic activity, as compared to WT mice, especially at 2 and 3 days, whereas at later time points (days 5 and 9), there was a significant trend to compensate this deficit in TGR5 KO mice (Fig. 1D-F).
A majority of TGR5 KO mice (60%-75%) exhibited jaundice as soon as 2-3 days after PH and recovered afterwards. H&E staining after PH showed, exclusively in TGR5 KO mice, periportal patchy hepatocyte necrosis (Fig. 2A), increasingly extensive up to 72 hours, closely mimicking clusters of injured hepatocytes (“bile infarcts”) observed after BDL in mice. At 5, 9, and 15 days afterwards PH, hepatocyte necrosis and inflammatory infiltrates progressively declined (data not shown), whereas periductular fibrosis appeared in a majority of TGR5 KO mice (day 15), but was lacking at day 21 (Supporting Fig. 2E). In WT mice, TBA raised immediately after PH in plasma, but also in liver during the first hours (Fig. 2B,C). Although this rise was transient in WT mice, massive and prolonged TBA accumulation in both plasma and liver was observed in TGR5 KO mice. No increase in post-PH mortality was noticed in TGR5 KO, as compared to WT mice (data not shown).
The TGR5 KO phenotype could not be explained by a deficient hepatic adaptive response to post-PH BA overload, because Na+ taurocholate cotransporting polypeptide (NTCP), cholesterol 7α-hydroxylase (CYP7a1), organic solute transporter beta (OST-β), and bile salt export pump (BSEP) messenger RNAs (mRNAs) were adequately regulated. This regulation was even stronger in TGR5 KO mice at days 3 and 5, when necroticoinflammatory injury and cholestasis were peaking, suggesting that FXR-dependent pathways were functional in those mice (Supporting Fig. 3E).
Lack of TGR5 Results in Excessive BA-Induced Liver Injury and Inflammation Upon BA Overload
In line with the fact that post-PH injury observed in TGR5 KO livers was suggestive of bile-induced toxicity, we first observed that liver necrosis occurred very early on (4 hours) after PH in TGR5 KO mice, at a time when BA—in particular, hydrophobic BA—had already accumulated in liver (Supporting Fig. 4A-D). Second, we found that a diet enriched with the bile acid sequestrant, cholestyramine (2% CT), rescued the post-PH phenotype in TGR5 KO mice, as shown by normal liver histology, reduced plasma and hepatic BA overload, and partial restoration of mitotic activity (Fig. 3A and Supporting Fig. 4E). Third, we found that TGR5 KO mice were significantly more sensitive to extrahepatic cholestasis induced by BDL, because they exhibited larger and more numerous necrotic areas on H&E-stained liver sections, higher ALT, bilirubin, and ALP elevations in plasma, and increased BA overload, as compared to WT mice (Fig. 3B,C and Supporting Fig. 4F). Finally, we found that TGR5 KO mice fed with a 1% cholic acid (CA)-enriched diet exhibited an increase in plasma ALT, ALP, and bilirubin (Supporting Fig. 4G), hepatic BA overload, parenchymal neutrophil infiltration, hepatic necrosis, and elevated tumor necrosis factor alpha (TNF-α) mRNA, whereas WT mice did not (Fig. 3D,E). In agreement with the inhibitory effect of TGR5 on cytokine gene induction and production in macrophages and KCs,[10, 12, 14] we found that plasma rise in interleukin (IL)−6, TNF-α and IL-1β concentrations 4 hours after PH was significantly higher in TGR5 KO than in WT mice (Fig. 4A). Hepatic infiltration by neutrophils was also strikingly more intense in TGR5 KO than WT mice after PH or BDL (Fig. 4B). Interestingly, KC depletion with clodronate liposomes, significantly reducing the post-PH induction of cytokine gene, also reduced the occurrence of hepatic necrosis 72 hours after PH (Fig. 4C). Altogether, these data suggest that TGR5 is mandatory for liver protection against BA toxicity and excessive cytokine production, as revealed in different experimental settings of BA overload (PH, BDL, and 1% CA-enriched diet) and BA sequestration (CT diet).
Lack of TGR5 Results in Altered Regulation of Bile Composition and Flow After PH and BDL
Based on these data, we investigated whether TGR5 may contribute to adapt bile flow in circumstances of BA overload after PH or BDL. Interestingly, biliary BA composition was more hydrophobic in TGR5 KO than in WT mice, as calculated on the basis of mass spectrometry analysis, both before and after PH (Fig. 5A). This more hydrophobic BA composition in TGR5 KO mice was also found in plasma and liver before and after PH (Fig. 5B,C) as well as in feces (Supporting Fig. 5A). Namely, muricholic acid (MCA), a hydrophilic primary BA in mice, whose relative concentration is maintained or increased during cholestasis, as well as the MCA/CA ratio taken as a marker of hydrophobicity, were strongly reduced in TGR5 KO mice before and after PH, as compared to WT mice (Fig. 5B,C and Supporting Fig. 5A). Of note, cytochrome P450 (7a1, 8b1, 27a1, 3A11, and 2b10) and sulfotransferase 2a mRNA expressions were similar in WT and TGR5 KO mice (Supporting Fig. 5B). Basal bile flow was slightly smaller in TGR5 KO than in WT mice, as previously reported; however, 48 hours after PH, bile flow increased significantly in WT, as previously described, but not in TGR5 KO mice (Fig. 6A). Similarly, biliary concentrations and outputs of Na+, HCO3−, and Cl− significantly rose 48 hours after PH in WT, but not in TGR5 KO, mice (Fig. 6B and Supporting Fig. 6A), strengthening the idea that TGR5 may control ionic composition of bile after PH. In agreement with these data, biliary pH, although similar in WT and TGR5 KO mice before PH, was maintained or slightly increased in WT, but significantly fell in TGR5 KO mice after PH (Fig. 6C). Of note, the defect in ion secretion in bile observed in TGR5 KO mice after PH was not secondary to cholestasis, because BDL (at 48 hours) did not inhibit, but even increased slightly, bile flow and ionic output in WT, but not in TGR5 KO, mice (Supporting Fig. 6B). Biliary BA concentrations and outputs were not significantly different in WT and TGR5 KO mice, both before and after PH, suggesting that bile flow deregulation in TGR5 KO mice did not result from a reduced BA flow rate (Supporting Fig. 6C).
Together, these data strongly suggest that TGR5-mediated signals may control adaptive ion transport in bile under circumstances in which BA overload occurs, such as after PH and BDL. Although we did not find any significant difference in the basal and post-PH mRNA expression of cystic fibrosis transmembrane conductance regulator (CFTR) and anion exchange isoform 2 (AE2) in livers and gallbladders from WT and TGR5 KO mice (Supporting Fig. 7A and data not shown), CFTR mRNA was significantly less up-regulated, both in liver and gallbladders from TGR5 KO, as compared to WT, mice after BDL (Supporting Fig. 7B,C). However, TGR5 may regulate ion exchange in bile at post-translational steps, through cAMP-mediated mechanisms, as proposed earlier.[15, 23]
Lack of TGR5 Results in Deficient BA Efflux in Urine After PH
To further understand the mechanisms involved in excessive BA accumulation in TGR5 KO mice after PH, BDL, or upon CA feeding, we explored BA efflux at the kidney level, because TGR5 is significantly expressed in this organ (as reported previously[7, 8] and Supporting Fig. 7F). Because a significant increase in BA efflux in urine was observed in WT, but not in TGR5 KO, mice after PH, BDL, or a 1% CA-enriched diet (Fig. 7A,B), we studied the expression of renal BA transporter genes in those experimental settings. Although multidrug resistance-related protein MRP2, MRP3, MRP4, and OST-β mRNAs were significantly up-regulated after PH in WT, MRP2 and MRP4 genes were not, or significantly less, induced in TGR5 KO kidneys (Fig. 7C,D, and Supporting Fig. 7D-E). These data are in line with the observed weaker BA efflux in urine from TGR5 KO mice, because MRP2 and MRP4 have been reported to transport BA into urine in kidney proximal tubule epithelial cells. In line with these data, western blotting analysis of renal MRP2 revealed a weaker expression in TGR5 KO than in WT mice 48 hours after PH (Fig. 7E). Finally, treatment with the natural TGR5 ligand, oleanolic acid (OA), elicited significantly stronger BA elimination in urines in WT than in TGR5 KO mice upon a 1% CA-enriched diet (Fig. 7B). Together, these data suggest that the lack of TGR5 may dampen BA efflux into urines upon BA overload, at least through transcriptional control of the apical BA transporters, MRP2 and MRP4, in the kidney.
After PH, the remnant liver adapts to an immediate BA overload by regulating BA synthesis and transport to protect liver cells from BA toxicity.[5, 6] Moreover BA, increasingly viewed as signaling molecules, affect both this adaptive process and liver regeneration itself, mainly through binding to FXR.[4, 6] We investigated the previously unexplored impact of the membrane-bound BA receptor, TGR5, during liver regeneration after PH in mice. Although PH induces a transient BA hepatic overload in WT mice, massive hepatocyte necrosis and cholestasis were observed with delayed liver regeneration only in TGR5 KO mice after PH, suggesting that the ability to challenge BA overload before significant cell damage occurs was, in some way, exceeded in TGR5 KO mice. The lack of TGR5 resulted in more hydrophobic bile and in excessive hepatic inflammation after PH, associated with deficient adaptation of bile composition and flow, as well as insufficient BA efflux in urine, all these factors contributing to excessive BA overload.
Cytokine production and release are finely tuned after PH, in a balanced way, to both protect liver cells and promote them for growth-factor–dependent progression into the cell cycle. The exacerbated post-PH induction of cytokines observed in TGR5 KO mice may thus have contributed to delay regeneration, but also to enhance cholestasis,[6, 26] and to favor hepatocyte necrosis, as suggested by KC depletion experiments. However, because early post-PH liver injury was not affected by KC depletion (Fig. 4), inflammation appears more as a worsening, rather than as a triggering, factor in the PH-induced TGR5 KO phenotype.
We observed that plasma, liver, bile, and feces from TGR5 KO mice exhibited a more hydrophobic BA composition, as suggested previously. Interestingly, small heterodimer partner null mice are more susceptible to BDL-induced liver damage than WT mice, because they have a more hydrophobic BA pool. In the same line, mice fed with a lithocholic acid–enriched diet exhibit a hydrophobic bile composition and bile duct obstruction leading to destructive cholangitis with bile infarcts. More recently, FXR-dependent production of fibroblast growth factor 15 has been proposed to protect liver from BA overload by switching BA composition toward a more hydrophilic profile. Thus, too much hydrophobic BA accumulating in the TGR5 KO liver immediately after PH may have led, by itself, to liver injury. This hypothesis is supported by the rescued post-PH phenotype in experiments with BA resin (CT) on the one hand and by the severe phenotypes observed in TGR5 KO mice after BDL or CA-enriched feeding on the other hand. In line with our data, bile from TGR5-overexpressing mice was less hydrophobic than WT bile, strengthening the idea that TGR5 controls bile hydrophobicity (Supporting Fig. 5C).
Although, after PH, hepatocytes adapted BA synthesis and transport in both genotypes, we found that bile composition in ions did not significantly adapt in TGR5 KO mice. Because TGR5 is not significantly detected in hepatocytes, TGR5-dependent biliary adaptation after PH most likely reflects processes occurring in cholangiocytes. Our data keep in line with the proposition that TGR5 would control CFTR-dependent Cl− secretion in cholangiocytes, because TGR5 KO exhibited less Cl− secretion in bile than WT mice after PH or BDL. The underlying mechanisms may involve cAMP-dependent membrane targeting of apical sodium-dependent bile salt transporter and CFTR, as previously proposed, although TGR5-dependent transcriptional control of CFTR mRNA remains possible (Supporting Fig. 7B,C). The post-PH increase in HCO3− biliary output, together with biliary pH regulation, may be part of a TGR5-dependent adaptive mechanism enhancing bile secretion and protecting the overloaded remnant liver from BA toxicity.[30, 31] In line with this idea, we observed a post-PH rise in bile viscosity in TGR5 KO mice that may be related to this deficient adaptive response impairing bile flow (Fig. 6D). In addition to the striking phenotype observed in TGR5 KO mice upon BA overload, further work will be needed to understand how the lack of TGR5 affects basal liver homeostasis (Supporting Fig. 1).
We finally found that TGR5 may contribute to BA elimination in urine, at least through the control of MRP2 and MRP4 gene expression in conditions of BA overload. Although nothing has been reported on yet about the role of TGR5 in the kidney,[7, 18] deficient urinary BA elimination worsens liver injury after BDL.[24, 32] In our study, because hepatic necrosis occurs very early on after PH, the default in urinary BA elimination, significantly observed in the days after PH, may more likely result in a worsening of BA overload, rather than in the initiation of liver injury. Interestingly, cAMP is reported as a crucial regulator for MRP2 targeting at the bile canaliculus, raising the possibility that TGR5-mediated (and cAMP-mediated) post-translational regulation of MRP2 may occur also in kidney epithelial cells. Further studies are needed to identify mechanisms involved in TGR5-mediated regulation of BA efflux in urine.
In conclusion, we found that TGR5 protects the liver against BA overload after PH, thereby preserving its regeneration capacity. After PH, BDL, or upon CA-enriched feeding, intrahepatic stasis of abnormally hydrophobic bile may be one of the primary factors involved in liver injury observed in TGR5 KO mice. Moreover, in the setting of BA overload, excessive inflammation as well as impaired urinary BA efflux observed in the absence of TGR5 may worsen liver injury.
The authors thank Patrick Pham, Nathalie Samson, Pascale Leblanc-Veyrac, and Noémie Dherbe for their technical help. The authors also thank Mélanie Durth and Christophe Clanet (LADHYX–Hydrodynamique, Ecole Polytechnique, Palaiseau, France). The authors thank Raoul Poupon for his critical reading and fruitful discussion of the manuscript.