Enterohepatic cycle of bile acids and their importance in intestinal postprandial lipid absorption

Once synthesised, bile acids are secreted into the canalicular space between hepatocytes and reach gallbladder (Fig. 1). Then bile acids are mixed with potassium and sodium ions thereby forming bile salts. The arrival of a meal in the proximal duodenum induces the secretion of cholecystokinin by enteroendocrine I-cells, which subsequently binds to cholecystokinin A receptors on cholangiocytes, induces gallbladder contraction and bile release into the duodenal lumen. There, bile acids facilitate dietary lipid and fat-soluble vitamin absorption and transport by forming, together with phospholipids, TAG, lipid-soluble vitamins and cholesterol, the postprandial mixed micelles. Indeed, it has been shown using the human differentiated Caco-2 cell line, a model frequently used to study apical-to-basolateral lipid transport, that there is no transport of postprandial micelles without taurocholate (TCA)⁽ Reference Luchoomun and Hussain ^{4 )}. Bile acids are not absorbed and continue until the distal intestine (i.e. ileum) where they are re-absorbed involving the apical sodium-dependent bile salt transporter, and the basolateral heterodimer organic solute transporter α/β (OSTα/OSTβ) and reach portal blood (Fig. 1). From the portal circulation, bile acids are taken up by the liver via different transporters belonging to the organic anion transporting polypeptide family⁽ Reference Hagenbuch and Meier ^{5 )} or via the sodium-TCA cotransporting polypeptide (NTCP/SLC10A1) thus closing their enterohepatic cycle. In human subjects, and depending on when the measure was made but also on the regimen, this cycle occurs six to twelve times daily thus limiting bile acid faecal loss⁽ Reference Mazuy, Helleboid and Staels ^{6 ,} Reference Prawitt, Caron and Staels ^{7 )}. It is reported that 95 % of bile acids are re-absorbed in the distal ileum and 5 % of bile acids arrive in the colon⁽ Reference de Aguiar Vallim, Tarling and Edwards ^{8 )}. In the ileum and mainly in the colon, primary bile acids are deconjugated and de-hydroxylated by bacteria belonging to the gut microbiota thus generating secondary bile acids. Indeed, some bacteria mainly belonging to the Clostridiales and Bacteroidales orders display bile salt hydrolase activity transforming (tauro- and/or glyco-) CA and (tauro- and/or glyco-) CDCA acids into deoxycholic acid (DCA) and lithocholic acids (LCA), respectively⁽ Reference Dey, Wagner and Blanton ^{9 )}. These biochemical reactions increase colonic bile acid re-absorption.

Fig. 1. (Colour online) Enterohepatic cycle of bile acids. Bile acids are produced from cholesterol in the liver. In the fasted state, bile acids are stored in the gallbladder. After meal ingestion, bile acids are expulsed in the intestinal lumen where they emulsify dietary fat. In the ileum, 95 % of bile acids are reabsorbed by the apical sodium-dependent bile salt transporter (ASBT) and basolateral heterodimer organic solute transporter α/β (OSTα/OSTβ). Through the portal circulation, bile acids return to the liver where, by their binding to FXR, they decrease gene expression of the rate-limiting enzymes in bile acid synthesis, i.e. Cyp7a1 and Cyp8b1. In enterocytes, the activated farnesoid X receptor (FXR) increases Fgf15/19 and Shp gene expression thus participating to bile acid metabolism regulation. In L-cells, bile acids bind and activate G-protein-coupled bile acid receptor 1 (TGR5) leading to the secretion of the incretin glucagon-like peptide 1 (GLP-1). By contrast, the activated FXR in L-cells decreases glucose-induced GLP-1 secretion.

Regulation of bile acid metabolism: the role of bile acid receptors

At high concentrations bile acids are cytotoxic molecules. The organism has developed numerous regulatory mechanisms to avoid their overproduction and their accumulation in organs⁽ Reference Rolo, Palmeira and Holy ^{10 –} Reference Perez and Briz ^{12 )}. These mechanisms, driven by a negative feedback by bile acids themselves, involve both transmembrane and nuclear receptors (NR). Among them, the membrane G-protein-coupled bile acid receptor 1 (TGR5) and the  NR pregnane X receptor and farnesoid X receptor (FXR) participate to alleviate hepatic and intestinal bile acid overload (Fig. 1). The most studied receptors in term of bile acid and glucose metabolism are TGR5 and FXR, we will only focus this review on these two receptors.

TGR5 is a membrane receptor which is activated by oleanolic acid, a triterpenoid molecule, and bile acids (affinity for TGR5: tauro-LCA = LCA > DCA > CDCA = CA)⁽ Reference Prawitt and Staels ^{13 )}. First described as a regulator of cytokine production in a human monocyte cell line⁽ Reference Kawamata, Fujii and Hosoya ^{14 )}, TGR5 is expressed in Küpffer cells, cholangiocytes, adipocytes, myocytes and enteroendocrine cells⁽ Reference Maruyama, Tanaka and Suzuki ^{15 –} Reference Briere, Ruan and Cheng ^{17 )}. It has been reported that whole body TGR5 KO mice have a decreased bile acid pool size⁽ Reference Maruyama, Tanaka and Suzuki ^{15 ,} Reference Harach, Pols and Nomura ^{18 ,} Reference Li, Holmstrom and Kir ^{19 )}. These animals also display more TCA and less tauro-beta MCA and have decreased expression of Cyp7b1 and Cyp27a1 gene expression than wild-type (WT) animals⁽ Reference Donepudi, Boehme and Li ^{20 )}. Incubation of murine hepatocytes with culture media from TGR5-activated macrophages decreases Cyp7a1 gene expression⁽ Reference Lou, Ma and Fu ^{21 )}, thus highlighting a possible paracrine function of TGR5 in Küpffer cells in hepatic bile acid synthesis regulation. Moreover, the gallbladder is the organ with the highest TGR5 expression levels⁽ Reference Briere, Ruan and Cheng ^{17 )}. TGR5 agonists stimulate gallbladder filling by a mechanism involving cAMP and muscle relaxation⁽ Reference Briere, Ruan and Cheng ^{17 ,} Reference Li, Holmstrom and Kir ^{19 )}. Finally, TGR5 is also expressed in the colon in enterochromaffin cells and myenteric neurons where its activation decreases colonic contractility and motility through a 5-hydroxytryptamin/calcitonin-gene-related peptide pathway, a well-known regulatory mechanism of intestinal peristaltism⁽ Reference Alemi, Poole and Chiu ^{22 –} Reference Grider, Piland and Gulick ^{25 )}. By increasing the delay before defecation, this mechanism can participate to better intestinal bile acid reabsorption but further studies are needed to fully decipher the role of TGR5 in bile acid metabolism.

Bile acids also regulate their own synthesis through binding and activation of the nuclear bile acid receptor FXR, firstly identified in 1995 in rodents⁽ Reference Seol, Choi and Moore ^{26 ,} Reference Forman, Goode and Chen ^{27 )} as a receptor for farnesol⁽ Reference Wang, Chen and Hollister ^{28 )}. In eukaryotes, the NR superfamily is the largest transcription factor family. Forty-nine NR have been identified so far⁽ Reference Mangelsdorf, Thummel and Beato ^{29 )}. FXR is encoded by the NR1H4 gene. Almost all NR share a common structure with five functional domains. As the other NR, the N-terminal domain of FXR is constituted by a ligand-independent activation site called activated function-1 and a DNA-binding domain. These regions are separated from the ligand-binding domain and the ligand-dependent activation site (activated function-2) by a hinge region. Due to alternative splicing and the utilisation of different promoters, four FXR isoforms (FXRα1–4) have been reported (for review⁽ Reference Prawitt, Caron and Staels ^{7 )}). The most transcriptionally active FXR isoforms are FXRα2 and FXRα4 that differ from FXRα1 and FXRα3, respectively, by an introduction of four extra amino acids in the hinge region⁽ Reference Zhang, Kast-Woelbern and Edwards ^{30 )}. FXR is highly expressed in the intestine, liver and kidney and at lower levels in adipose tissue and pancreas. Primary bile acids are the most potent activators of FXRα (called thereafter FXR; affinity for FXR: CDCA > TCA > DCA = tauro-LCA; for review⁽ Reference Prawitt and Staels ^{13 )}). Moreover,  tauro-alpha MCA, tauro-beta MCA and ursodeoxycholic acid have been identified recently as FXR antagonists⁽ Reference Hu, Bonde and Eggertsen ^{31 –} Reference Campana, Pasini and Roda ^{33 )}. The development early in the 21st century of whole-body FXR KO mice highlighted the crucial role of FXR in energy homeostasis and more specifically in bile acid metabolism⁽ Reference Sinal, Tohkin and Miyata ^{34 )}. Indeed, FXR KO mice display an increase in plasma bile acid, as well as TAG and cholesterol levels compared with WT littermates. Once activated by postprandial bile acids, hepatic FXR increases Nr0b2 gene expression (small heterodimer partner), an orphan NR with co-repressor activities, which decreases Cyp7a1 and Cyp8b1 gene expression by direct interaction with liver receptor homologue-1) and the recruitment of corepressors thus decreasing bile acid synthesis⁽ Reference Lu, Makishima and Repa ^{35 ,} Reference Sanyal, Båvner and Haroniti ^{36 )} (Fig. 1). Hepatic activated FXR also decreases Cyp7b1 and Ntcp and increases Bsep, Ostα, Ostβ and Mdr3 (multi drug resistance 3) gene expression thus enhancing hepatic bile acid drain (for review⁽ Reference Lefebvre, Cariou and Lien ^{37 )}). In the intestine, FXR activation up-regulates Ibabp and Ostα/Ostβ and decreases apical sodium-dependent bile salt transporter gene expression thus enhancing the enteroportal circulation of bile acids⁽ Reference Chen, Ma and Dawson ^{38 –} Reference Lee, Zhang and Lee ^{41 )}. In enterocytes, FXR upregulates the expression and the secretion in the portal blood of fibroblast growth factor (FGF; 15 in mice, 19 in human subjects)⁽ Reference Markan and Potthoff ^{42 )}. Through a pathway not yet fully identified involving β-Klotho, FGF15/19 activates hepatic FGFR4 and decreases Cyp7a1 gene expression in the liver⁽ Reference Owen, Mangelsdorf and Kliewer ^{43 )}. Thus, FXR in both hepatocytes and enterocytes controls bile acid metabolism and decreases their cellular toxicity (Fig. 1).

Enteroendocrine cells

Even if enteroendocrine cells represent only 1 % of the total intestinal epithelial cells, the length of the intestine makes it the largest endocrine organ. Based on the peptide they secrete and on their expression profile all along the intestine, at least thirteen different enteroendocrine cell types have been identified. Among them, enteroendocrine K- and L-cells secrete the incretins glucose-insulinotropic polypeptide (IP) and glucagon-like peptide (GLP)-1. The incretin effect is based on the observation that oral glucose administration induces a more pronounced insulin secretion than an isoglycaemic intraveinous injection. The enteroendocrine L-cells are present all along the upper and the lower intestine following a cephalocaudal gradient with a maximum abundance in the colon. The proglucagon gene, the same gene that produces pancreatic glucagon, encodes GLP-1. After transcription and translation into proglucagon, the action of prohormone convertase 1/3 in L-cells leads to GLP-1, GLP-2, oxyntomodulin and IP2, whereas the action of prohormone convertase 2 in pancreatic α-cells leads to glucagon, glicentin-related polypeptide, IP1 and major proglucagon fragment (for review⁽ Reference Baggio and Drucker ^{44 )}). In L-cells, the main bioactivity on glucose metabolism is linked to GLP-1. Indeed, in the pancreas, GLP-1 potentiates glucose-induced insulin secretion thus increasing insulin sensitivity of key metabolic organs such as skeletal muscle, adipose tissue and the liver. GLP-1 also inhibits gastric emptying, increases satiety and cardiac function. In blood, GLP-1 half-life is about 1·5–5 min due to a rapid degradation by dipeptidyl peptidase 4. Therapeutic strategies leading to more stable GLP-1 or to a lower GLP-1 degradation have been developed (for review⁽ Reference Baggio and Drucker ^{44 )}). These drugs are the non-hydrolysable GLP-1 mimetics and  dipeptidyl peptidase 4 inhibitors that are successfully used to treat type 2 diabetic patients. Another strategy to increase GLP-1 activity would be to increase its endogenous production and secretion by L-cells.

Glucose is a regulator of glucagon-like peptide 1 production and secretion

Many diet-derived metabolites such as oleoylethanolamine, n-3 PUFA, the SCFA butyrate and propionate, glutamine and L-ornithine drive GLP-1 secretion mainly through binding and activation of diverse G-protein-coupled receptors (for review⁽ Reference Ward ^{45 )}). Glucose also enhances GLP-1 biosynthesis and secretion. Two distinct mechanisms both leading to an increase of intracellular calcium concentrations and membrane fusion of GLP-1 containing vesicles are involved in glucose-induced GLP-1 secretion. The first mechanism involves the sodium-glucose cotransporter 1 where the entry of two sodium ions, concomitantly with one molecule of glucose, induces a difference of potential leading to the opening of a voltage-dependent calcium channel⁽ Reference Parker, Adriaenssens and Rogers ^{46 ,} Reference Reimann, Habib and Tolhurst ^{47 )}. Although this mechanism seems to be the driving force for glucose-induced GLP-1 secretion⁽ Reference Gorboulev, Schürmann and Vallon ^{48 )}, a second mechanism identified only recently and involving GLUT2-mediated intracellular glucose catabolism through glycolysis pathway has been described⁽ Reference Kuhre, Frost and Svendsen ^{49 )}. At high extracellular glucose concentrations, the increase in intracellular glucose levels to millimolar range induces glucose catabolism into pyruvate through the glycolysis pathway. Then pyruvate is decarboxylated and conjugated to CoA to form acyl-CoA. By entering in the mitochondrial citrate cycle, acyl-CoA increases the ATP:ADP ratio, which leads to the closure of potassium ATP-dependent channels. The subsequent accumulation of potassium in the intracellular space leads to membrane depolarisation thus opening voltage-dependent calcium channels and the intracellular accumulation of calcium leads to GLP-1 vesicle release⁽ Reference Ward ^{45 ,} Reference Kuhre, Frost and Svendsen ^{49 )}. A few years ago glucose was identified as a proglucagon gene expression enhancer⁽ Reference Daoudi, Hennuyer and Borland ^{50 )}. Recently, a role for the carbohydrate responsive element-binding protein (ChREBP) in the glucose-mediated proglucagon gene increase has been proposed⁽ Reference Trabelsi, Daoudi and Prawitt ^{51 )}. ChREBP is a transcription factor activated by glucose metabolites and is highly expressed in enteroendocrine L-cells⁽ Reference Habib, Richards and Cairns ^{52 ,} Reference Filhoulaud, Guilmeau and Dentin ^{53 )}. We have shown that glucose increases proglucagon gene expression in small interference (si)Ctrl, but not in siChREBP enteroendocrine murine L-cells⁽ Reference Trabelsi, Daoudi and Prawitt ^{51 )}. Moreover, incubation with lactate or 2-deoxyglucose, a non-metabolisable glucose analogue, does not increase proglucagon gene expression. Thus, both ChREBP and glucose catabolism are mandatory for the observed glucose-mediated proglucagon gene expression⁽ Reference Trabelsi, Daoudi and Prawitt ^{51 )}. However, further studies are needed to fully address the mechanisms behind this ChREBP-dependent glucose-mediated proglucagon gene increase (Fig. 2).

Fig. 2. (Colour online) Activation of bile acid receptors in L-cells modulates glucagon-like peptide-1 (GLP-1) production and secretion. Activation of L-cell G-protein-coupled bile acid receptor 1 (TGR5) increases intracellular cAMP levels, thus leading to an increase in both GLP-1 production and secretion through the protein kinase (PK) A/cAMP responsive elements binding protein pathway and exchange protein directly activated by cAMP (EPAC2)/diacylglycerol (DAG)/PKCζ and EPAC2/phospholipase C (PLCε)/insulinotropic polypeptide (IP)3 pathways, respectively. Activation of L-cell farnesoid X receptor (FXR) in the presence of glucose decreases the glycolysis pathway, thus leading to lower intracellular ATP levels. This decrease is associated with lower levels of GLP-1. Moreover, FXR is in the same complex as carbohydrate responsive element-binding protein (ChREBP) and decreases glucose-induced proglucagon gene expression. The bile acid sequestrant (BAS) colesevelam, by inhibiting FXR activation, prevents these decreases. Furthermore, bile acid in complexes with BAS are still able to activate TGR5, thus further increasing L-cell GLP-1 secretion.

Bile acids are regulators of glucagon-like peptide 1 production and secretion via the bile acid receptors G-protein-coupled bile acid receptor 1 and farnesoid X receptor

As shown for glucose, bile acids also modulate both GLP-1 secretion and proglucagon gene expression. TGR5 is expressed all along the intestine with the maximum levels in the colon, where the enteroendocrine L-cells are predominant⁽ Reference Harach, Pols and Nomura ^{18 )}. In experiments on L-cells isolated using the transgenic GLU-Venus mouse model, it has been shown that TGR5 expression is mainly restricted to enteroendocrine L-cells⁽ Reference Reimann, Habib and Tolhurst ^{47 )}. Binding of bile acids to TGR5 in L-cells dissociates the Gαs subunit of the heterotrimeric protein from the Gβ/γ subunits. Thus, activated Gαs activates adenylate cyclase that converts ATP into cAMP. After binding of cAMP to the two regulatory subunits of protein kinase A, the catalytic subunits of protein kinase A are dissociated and shuttle to the nucleus. There, protein kinase A phosphorylates and activates cAMP responsive elements binding protein, which in turn binds to cAMP responsive elements in the promoter of target genes, including proglucagon, thus regulating their expression⁽ Reference Harach, Pols and Nomura ^{18 ,} Reference Vallim and Edwards ^{54 ,} Reference Thomas, Gioiello and Noriega ^{55 )} (Fig. 2). cAMP produced upon TGR5 activation also triggers GLP-1 secretion through the EPAC2/phospholipase Cε/IP3 and EPAC2/diacylglycerol/protein kinase Cζ pathways⁽ Reference Harach, Pols and Nomura ^{18 ,} Reference Thomas, Gioiello and Noriega ^{55 –} Reference Bala, Rajagopal and Kumar ^{57 )} (EPAC, exchange protein directly activated by cAMP) (Fig. 2). Very recently, it has been shown that bile acid-induced GLP-1 secretion is mediated mostly through TGR5 located at the basolateral side of L-cells⁽ Reference Brighton, Rievaj and Kuhre ^{58 )}.

Using L-cells sorted by fluorescence-activated cell sorting from GLU-Venus mice, it has been shown that FXR is also expressed in L-cells⁽ Reference Trabelsi, Daoudi and Prawitt ^{51 )}. Moreover, FXR mRNA levels are higher in L-cells than in non-L-cells. In fresh human jejunal biopsies, GLP-1 immunoreactive cells are also immunoreactive for FXR showing that FXR is expressed in human L-cells. In both human subjects and mice, FXR activation by either bile acids or the specific FXR agonist GW4064 decreases GLP-1 production. More precisely, L-cell-activated FXR is in the same protein complexes containing ChREBP and inhibits glucose-induced ChREBP-mediated proglucagon gene expression as shown by using siChREBP transfected cells and a non-metabolisable glucose analogue. Moreover, FXR activation also decreases glucose-induced GLP-1 secretion. Specifically, in glucose-containing media, FXR activation overall decreases the glycolysis pathway at the gene expression level, lowers intracellular ATP levels and finally lowers glucose-induced GLP-1 release. KCl-induced GLP-1 secretion is not altered by FXR activation, showing the glucose dependency of FXR action on GLP-1 secretion. This glucose dependency is more specifically due to the glycolysis pathway since the glucose-induced GLP-1 secretion is not decreased after GW4064 treatment of murine biopsies challenged with a GLUT-2 inhibitor. These results show that FXR activation decreases both proglucagon gene expression and GLP-1 secretion by interfering with pathways activated by glucose⁽ Reference Trabelsi, Daoudi and Prawitt ^{51 )} (Fig. 2). The importance of this FXR–GLP-1 pathway as a potential therapeutic target will be discussed later.

Obese and type 2 diabetic patients have altered bile acid metabolism

Obesity and type 2 diabetes (T2D) have been shown to have reached pandemic levels in industrial countries. The WHO estimated 600 million obese and 422 million diabetic individuals in 2014^{( 59 , 60 )}. T2D is characterised by fasting hyperglycaemia and insulin resistance which participate together with hypertriglyceridaemia, hypercholesterolaemia, abdominal obesity and hypertension, in the so-called metabolic syndrome⁽ Reference Huang ^{61 )}. Early and recent studies highlight a disrupted bile acid pool size and/or composition in T2D⁽ Reference Bennion and Grundy ^{62 –} Reference Sonne, van Nierop and Kulik ^{66 )}. Patients with uncontrolled T2D have an increase in bile acid pool size, which disappears after insulin treatment⁽ Reference Bennion and Grundy ^{62 )}. Other studies show no differences in bile acid pool size between uncontrolled and insulin treated T2D patients but a change in bile acid pool composition with an increase in the proportion of the secondary bile acid DCA⁽ Reference Abrams, Ginsberg and Grundy ^{63 ,} Reference Brufau, Stellaard and Prado ^{64 )}. Another study shows that an increase of plasma DCA occurs together with a decrease in plasma CA⁽ Reference Suhre, Meisinger and Döring ^{65 )}. Very recently, it has been reported in human subjects that the amplitude in plasmatic postprandial bile acid levels is positively correlated with meal fat content. Moreover, and when compared to age-, sex- and BMI-matched normoglycemic subjects, T2D patients display an increase in total plasma bile acids, mainly due to increases in glycine conjugated bile acids, in DCA and in ursodeoxycholic acid during both oral glucose- and meal tolerance tests⁽ Reference Sonne, van Nierop and Kulik ^{66 )}. Finally, 12α-hydroxylated bile acids are negatively associated with insulin sensitivity⁽ Reference Haeusler, Astiarraga and Camastra ^{67 )}. Even though some discrepancies exist, these studies clearly support the notion of change in bile acid metabolism in T2D. Moreover, in rats fed a high-fat diet, bile diversion to distal intestine improved glucose tolerance⁽ Reference Goncalves, Barataud and De Vadder ^{68 )}. Such an improvement is also observed in mice after bile diversion⁽ Reference Flynn, Albaugh and Cai ^{69 ,} Reference Pierre, Martinez and Ye ^{70 )} or in mice after ileal interposition, which bypasses bile acid cycling⁽ Reference Kohli, Kirby and Setchell ^{71 )}. An increase in GLP-1 secretion is one of the mechanisms evocated to explain these beneficial effects. Furthermore, in healthy subjects, TCA administration stimulates the secretion of GLP-1 by enteroendocrine L-cells and increases fullness sensation⁽ Reference Wu, Bound and Standfield ^{72 )}. The same year, another team has shown that intrarectal TCA administration increases GLP-1 secretion and decreases blood glucose without hypoglycaemia in obese T2D patients⁽ Reference Adrian, Gariballa and Parekh ^{73 )}. A better understanding of how bile acids act as signalling molecules through TGR5 and FXR can thus be of interest to develop specific molecules to treat T2D.

It has been difficult to appreciate the exact contribution of each bile acid receptor to energy homeostasis since bile acids are ligands for both TGR5 and FXR. The development of specific synthetic (such as GW4064) or semi-synthetic (such as obeticholic acid (INT-747) and fexaramine) FXR agonists, as well as specific TGR5 agonists (such as INT-777), allow the study of the impact of each bile acid receptor activation to energy homeostasis. Moreover, the development of whole body as well as organ-specific FXR and TGR5 KO animals allows further discrimination of the relative contribution of each bile acid receptor in a specific tissue on glucose metabolism in the pathophysiological context of obesity and T2D (for review⁽ Reference Luchoomun and Hussain ^{4 ,} Reference Hagenbuch and Meier ^{5 ,} Reference Lefebvre, Cariou and Lien ^{37 )}).

G-protein-coupled bile acid receptor 1

TGR5 activation is important in energy metabolism regulation via its capacity to increase energy expenditure and to promote GLP-1 production. Indeed, TGR5 is expressed in brown adipose tissue and skeletal muscle where, through the cAMP/deiodinase 2 pathway, it catalyses the conversion of inactive prohormone thyroxine to active 3,5,3′-tri-iodothyronine⁽ Reference Watanabe, Houten and Mataki ^{16 ,} Reference Bianco, Maia and da Silva ^{74 )}. This hormone thus enhances brown adipose tissue lipolysis and increases thermogenesis⁽ Reference Watanabe, Houten and Mataki ^{16 )}. TGR5 is also involved in lipid metabolism regulation. Whereas both male and female TGR5 KO mice display similar body weight compared to their WT littermates when fed a high-fat diet⁽ Reference Maruyama, Tanaka and Suzuki ^{15 ,} Reference Vassileva, Hu and Hoos ^{75 )}, TGR5 KO female animals have less cholesterol in VLDL, LDL and HDL lipoprotein fractions⁽ Reference Vassileva, Hu and Hoos ^{75 )}. These mice also display less TAG in VLDL fraction than TGR5 WT mice⁽ Reference Vassileva, Hu and Hoos ^{75 )}. Very recently, it has been shown by Donepudi et al.⁽ Reference Donepudi, Boehme and Li ^{20 )} that TGR5 KO mice are protected against fasting-induced steatosis. Thomas et al.⁽ Reference Thomas, Gioiello and Noriega ^{55 )} have shown that high-fat-fed mice containing a constitutively active form of TGR5 (TGR5-Tg) display a better glucose tolerance, whereas high-fat-fed TGR5 KO mice have a worsened glycaemic profile. These improvements in glucose metabolism are due to the GLP-1-mediated incretin effect. Indeed, high-fat diet fed TGR5-Tg mice display more GLP-1 and insulin after an oral glucose tolerance test than WT mice thus highlighting the importance of the TGR5/GLP-1 pathway in the improvement of glycaemia by bile acids.

Farnesoid X receptor. Different studies have revealed that, depending on the organ, FXR activation can be beneficial or deleterious for glucose control in obesity. In 2006, Zhang et al.⁽ Reference Zhang, Lee and Barrera ^{76 )} demonstrated that overexpressing a constitutively active form of FXR in the liver of db/db mice improves glucose tolerance. Moreover, ob/ob mice also display an improved glucose tolerance after intra-peritoneal injection of GW4064 (IP, 30 mk/kg mouse, once daily for 10 d) and have a decrease in insulin secretion⁽ Reference Cariou, van Harmelen and Duran-Sandoval ^{77 )}. A recent study shows that GW4064 treated mice (50 mk/kg mouse, by IP, twice weekly for 6 weeks) gain less body weight when fed a high-fat diet than vehicle-treated mice. GW4064 treated mice also display a better glucose tolerance than vehicle-treated mice⁽ Reference Ma, Huang and Yan ^{78 )}. These studies demonstrate that hepatic FXR activation decreases gluconeogenic genes expression. Moreover FXR activation also leads to the induction by glucose of glycolysis gene expression by interfering with the ChREBP pathway⁽ Reference Caron, Huaman Samanez and Dehondt ^{79 )}. Altogether, these results show that activating hepatic FXR seems beneficial for glucose control. Conversely, mice fed with a high-fat diet mixed with GW4064 display a lower bile acid pool size with a decrease in TCA proportion. They are also more obese and hyperglycaemic than mice fed with the control diet or fed with a high-fat diet enriched with CA 0·1 % showing a deleterious impact of oral GW4064 administration⁽ Reference Watanabe, Horai and Houten ^{80 )}. The involvement of the GLP-1 pathway in this phenotype is unclear since no GLP-1 measurements were performed. However, T2D obese patients treated with TCA have increased GLP-1 and lower blood glucose⁽ Reference Adrian, Gariballa and Parekh ^{73 )}. Moreover, we have shown in mice that oral administration of GW4064 (by gavage, 30 mk/kg mouse, once daily for 5 d) decreases proglucagon gene expression and intestinal biopsies from mice treated following this protocol failed to secrete GLP-1 in response to glucose⁽ Reference Trabelsi, Daoudi and Prawitt ^{51 )}. It should be noted that GW4064, as well as fexaramine⁽ Reference Fang, Suh and Reilly ^{81 )}, are not well absorbed by the intestine. Thus, the decreased GLP-1 pathway after intestinal FXR activation can be involved in the deleterious effect of oral GW4064 on glucose metabolism. As indicated earlier, four different isoforms of FXR (FXRα1–4) have been identified (for review⁽ Reference Schonewille, de Boer and Groen ^{82 )}). Whereas the expression levels of FXRα1–2 and FXRα3–4 are similar in the liver, the intestine expresses higher levels of FXRα3–4 than FXRα1–2⁽ Reference Zhang, Kast-Woelbern and Edwards ^{30 )}. Differences in FXR isoform expression profiles may be involved in the differences between the beneficial effect of FXR activation in the liver and its harmful action in the intestine. Further studies are needed to fully address the importance of each FXR isoform, and especially in the intestine, in energy homeostasis.

The importance of FXR on glucose homeostasis reached a milestone thanks to the development of both whole-body and tissue-specific FXR KO animals. Indeed, many studies demonstrate that whole-body FXR KO mice are protected against diet-induced or genetically induced obesity. These mice also have an improved glucose tolerance⁽ Reference van Dijk, Grefhorst and Oosterveer ^{83 –} Reference Parséus, Sommer and Sommer ^{87 )}. Van Dijk et al.⁽ Reference van Dijk, Grefhorst and Oosterveer ^{83 )} have demonstrated that whole-body FXR KO mice present a delay in intestinal glucose absorption due to enterocytic accumulation of glucose-6-phosphate. In the hyperphagic ob/ob mice FXR gene expression deficiency improves all metabolic parameters and in particular glucose metabolism through an enhancement of peripheral glucose disposal and an increased adipose tissue insulin sensitivity⁽ Reference Prawitt, Abdelkarim and Stroeve ^{84 )}. These improvements are not observed in ob/ob mice where FXR gene is specifically invalidated in the hepatocyte thus demonstrating that FXR in extra-hepatic tissues drives the beneficial effects of FXR gene expression deficiency on glucose homeostasis.

A recent study shows that mRNA levels of FXR and its target genes small heterodimer partner and FGF19, are increased in the ileum of obese v. lean human subjects⁽ Reference Jiang, Xie and Lv ^{88 )}. Moreover, the expression of these genes positively correlates with BMI. Furthermore, bile diversion to the ileum in obese mice inhibits the expression of these genes and improves glucose metabolism⁽ Reference Flynn, Albaugh and Cai ^{69 )}. Jiang et al.⁽ Reference Jiang, Xie and Lv ^{88 )} further demonstrate that administration of a bile acid with specific intestinal FXR antagonism improves metabolic parameters, such as triglyceridaemia and glucose clearance, in obese mice. These improvements are due to a decrease in intestinal ceramide production. Moreover, and according to the phenotype observed in whole-body FXR KO mice, intestinal specific FXR KO mice are protected against diet-induced obesity also through a reduction of intestinal ceramide production⁽ Reference Jiang, Xie and Lv ^{88 ,} Reference Li, Jiang and Krausz ^{89 )}. Finally, whereas FXR KO mice fed a high-fat diet have a reduced glycaemia after an oral glucose tolerance test. This improvement is lost after GLP-1R antagonism showing that the beneficial effect of FXR gene invalidation on glycaemia is in part mediated through a GLP-1/GLP-1R pathway⁽ Reference Trabelsi, Daoudi and Prawitt ^{51 )}. Altogether, these results highlight a role of intestinal FXR in glucose metabolism in a pathophysiological context of obesity and T2D.

Inactivation of intestinal farnesoid X receptor transcriptional activity using bile acid sequestrants

Bile acid sequestrants (BAS) are anionic exchange resins first used to decrease hypercholesterolaemia. Indeed, BAS increase HDL-cholesterol levels by 3–5 % and decrease by 15–30 % LDL-cholesterol without changing or slightly increasing TAG levels⁽ Reference Brufau, Stellaard and Prado ^{64 ,} Reference Herrema, Meissner and van Dijk ^{90 )}. By trapping bile acids in the intestinal lumen, these molecules increase their faecal output. Thus, bile acids cannot activate intestinal FXR leading to the inhibition of the negative feedback loop driven by FGF15/19. Hepatocytes continue to convert cholesterol into bile acids thus decreasing plasma cholesterol⁽ Reference Brufau, Stellaard and Prado ^{64 ,} Reference Herrema, Meissner and van Dijk ^{90 )}. In the USA, BAS are also used as antidiabetic drugs. Indeed, cholestyramine administration for 5 d to diabetic patients decreases glycaemia by 20 mg/dl and glucosuria by 40 g compared with placebo treated patients⁽ Reference Garg and Grundy ^{91 ,} Reference Staels and Kuipers ^{92 )}. The mechanisms behind such improvements are multiple and not fully identified. Among them, an increase in splanchnic glucose utilisation and an increase in GLP-1 secretion can participate in the improvements (for review⁽ Reference Prawitt, Caron and Staels ^{93 )}). Splanchnic glucose utilisation is defined by the hepatic absorption of portal glucose, coming from the intestine, and its metabolisation through glycogenesis or glycolysis. As mentioned earlier, FXR KO mice have a delay in intestinal glucose absorption⁽ Reference van Dijk, Grefhorst and Oosterveer ^{83 )}. Moreover, ob/ob FXR KO mice have an improved glucose homeostasis due to an improved glucose clearance and increased insulin sensitivity in adipose tissue. Colesevelam improves glucose homeostasis only in ob/ob FXR WT but not in FXR KO ob/ob mice showing that the beneficial effect of colesevelam is dependent on FXR⁽ Reference Prawitt, Abdelkarim and Stroeve ^{84 )}. Finally, hepatic FXR activation inhibits glucose-induced glycolytic gene expression⁽ Reference Caron, Huaman Samanez and Dehondt ^{79 )}. BAS, through lifting these repressions, increases hepatic glucose utilisation. These results are in accordance with the fact that BAS administered to T2D patients increase glucose clearance and insulin sensitivity. This study also demonstrated an increase in incretin secretion after colesevelam⁽ Reference Beysen, Murphy and Deines ^{94 )}.

BAS-stimulated GLP-1 secretion has been proposed to occur by inhibiting bile acid ileal reabsorption. Thus, BAS drive bile acids to the colon where L-cell density is the highest. More precisely, the bile acid in complexes with BAS are still able to bind and to activate L-cell-TGR5, thus increasing GLP-1 secretion⁽ Reference Harach, Pols and Nomura ^{18 )}. Potthoff et al.⁽ Reference Potthoff, Potts and He ^{56 )} further demonstrated that the improvement of glycaemia after BAS is driven through a TGR5/GLP-1-induced reduction in hepatic glycogenolysis. Using colesevelam, Shang et al.⁽ Reference Shang, Saumoy and Holst ^{95 )} have shown in Zucker diabetic fatty rats a decrease in glucose clearance together with an increase in GLP-1 secretion. As indicated earlier, BAS de-activate intestinal FXR activity by inhibiting bile acid flux through enterocytes. Together with an enhanced glucose clearance and a decrease in intestinal small heterodimer partner gene expression, Zucker diabetic fatty rats treated with BAS have an increased glucose-induced GLP-1 secretion⁽ Reference Chen, McNulty and Anderson ^{96 )}. We have shown that colesevelam treatment of ob/ob mice improves glucose clearance after an oral glucose tolerance test and increases GLP-1 production through FXR, because such improvements are not observed in ob/ob FXR KO mice⁽ Reference Trabelsi, Daoudi and Prawitt ^{51 )} (Fig. 2).

Inactivation of intestinal farnesoid X receptor via microbiota manipulation

In recent years, a clear role of the gut microbiota in energy homeostasis regulation has emerged. Mice without intestinal microbiota (germ-free (GF) mice) are protected against diet-induced obesity and have an improved glycaemia compared with conventionally raised mice fed the same diet⁽ Reference Bäckhed, Manchester and Semenkovich ^{97 )}. Moreover, these GF mice have profound changes in bile acid metabolism. Indeed, GF mice have higher levels of  tauro-alpha MCA and tauro-beta MCA, two bile acids with FXR antagonist properties⁽ Reference Sayin, Wahlström and Felin ^{32 )}. Proglucagon mRNA levels and GLP-1 positive cells are also increased in GF mice compared with conventionally-raised mice⁽ Reference Wichmann, Allahyar and Greiner ^{98 )}. To link these two observations, we measured ileal proglucagon mRNA in both GF and conventionally raised FXR WT and FXR KO mice. FXR KO mice have increased proglucagon mRNA levels only in conventionally raised mice showing that the impact of FXR gene deficiency on proglucagon gene expression needs gut microbiota⁽ Reference Trabelsi, Daoudi and Prawitt ^{51 )}. Recent observations show that FXR WT and FXR KO mice have different gut microbiota⁽ Reference Parséus, Sommer and Sommer ^{87 )}. Moreover, high-fat fed GF mice transplanted with intestinal microbiota from FXR KO obese mice gain less body weight and display a better glucose tolerance than high-fat diet fed GF mice colonised with intestinal microbiota from FXR WT obese mice⁽ Reference Parséus, Sommer and Sommer ^{87 )}. Therefore, the beneficial effect of FXR gene deficiency on glucose tolerance involves the GLP-1 pathway and gut microbiota. Finally, treatment of obese mice with the prebiotic tempol enhances glucose metabolism through increased levels in bile acids with FXR antagonist properties. These improvements due to tempol treatment are not observed in intestinal FXR KO mice⁽ Reference Li, Jiang and Krausz ^{89 )} and are in line with a study showing that feeding mice with the FXR antagonist GβMCA improves glucose tolerance through intestinal FXR⁽ Reference Jiang, Xie and Lv ^{88 )}. Thus, prebiotics which increase the levels of bile acids with FXR antagonist properties improve glucose metabolism.