Bile acid metabolism and signaling in health and disease: molecular mechanisms and therapeutic targets

Abstract

Bile acids, once considered mere dietary surfactants, now emerge as critical modulators of macronutrient (lipid, carbohydrate, protein) metabolism and the systemic pro-inflammatory/anti-inflammatory balance. Bile acid metabolism and signaling pathways play a crucial role in protecting against, or if aberrant, inducing cardiometabolic, inflammatory, and neoplastic conditions, strongly influencing health and disease. No curative treatment exists for any bile acid influenced disease, while the most promising and well-developed bile acid therapeutic was recently rejected by the FDA. Here, we provide a bottom-up approach on bile acids, mechanistically explaining their biochemistry, physiology, and pharmacology at canonical and non-canonical receptors. Using this mechanistic model of bile acids, we explain how abnormal bile acid physiology drives disease pathogenesis, emphasizing how ceramide synthesis may serve as a unifying pathogenic feature for cardiometabolic diseases. We provide an in-depth summary on pre-existing bile acid receptor modulators, explain their shortcomings, and propose solutions for how they may be remedied. Lastly, we rationalize novel targets for further translational drug discovery and provide future perspectives. Rather than dismissing bile acid therapeutics due to recent setbacks, we believe that there is immense clinical potential and a high likelihood for the future success of bile acid therapeutics.

Introduction

Globally, cardiometabolic diseases present a substantial health risk, positively correlating with most, if not all of the top CDC-listed causes of American death.¹ The prevalence of these conditions has increased rapidly from an American morbidity of 37.5% in 2011 to 41.8% in 2018, primarily comprised of type 2 diabetes mellitus (T2DM), obesity, non-alcoholic fatty liver disease (NAFLD), non-alcoholic steatohepatitis (NASH), and atherosclerotic cardiovascular disease (ASCVD).^2,3 Alongside cardiometabolic diseases, non-metabolic conditions such as inflammatory bowel disease (IBD) and cancer have also been realized as intricate challenges to human health. Despite the apparent heterogeneity, these diseases share the commonality of complex biochemistry, all lack fully deterministic models for disease pathogenesis, and have no well-defined curative treatments. This shared complexity imposes a significant burden on humanity, diminishing quality of life and leading to mortality. Novel treatments are imperative to both better control and potentially cure these diseases.

Bile acids (BAs) are hepatically synthesized cholesterol derivatives that function as amphipathic surfactants and systemic endocrine hormones. Alongside regulating their own synthesis and enterohepatic circulation, BAs are potent modulators of macronutrient (lipid, carbohydrate, protein) metabolism and the systemic pro-inflammatory/anti-inflammatory balance. BAs provide complex physiological modulation by binding to Farnesoid X Receptor (FXR) and Takeda G Protein-Coupled Receptor 5 (TGR5), the canonical BA receptors, while exerting effects at other more recently characterized non-canonical BA receptors.^4,5 Alterations in BA physiology are directly correlated to the pathogenesis of cardiometabolic, inflammatory, and neoplastic diseases.⁶ Hence, the complex role of BAs on physiology provides ample targets for drug discovery.

The most developed BA therapeutic is the FXR agonist candidate Obeticholic Acid (OCA) by Intercept Pharmaceuticals. Although already approved for primary biliary cholangitis (PBC) at lower doses, it was rejected for NASH indication approval by the FDA due to significant dose-dependent toxicities.⁷ Rather than dismissing BA therapeutics due to recent setbacks, we believe that there is immense clinical potential and a high likelihood for the future success of BA therapeutics. In this review, we provide a bottom-up approach on BAs, mechanistically explaining their biochemistry, physiology, and pharmacology at canonical and non-canonical receptors. Using this mechanistic model of BAs, we explain how abnormal BA physiology drives disease pathogenesis, emphasizing how ceramide synthesis may serve as a unifying pathogenic feature for cardiometabolic diseases. We provide an in-depth summary on pre-existing BA receptor modulators, explaining their shortcomings and theorizing how they be remedied. Lastly, we rationalize novel targets for further translational drug discovery and provide future perspectives.

BA biochemistry

Primary BA synthesis

Pathway overviews

Hepatic BA synthesis tends to follow a common four-stage pattern: 7α-hydroxylation initiation, sterol ring modification, side chain truncation, and phase II conjugation (Fig. 1).⁸ Each of the previously mentioned metabolic steps takes place in the: endoplasmic reticulum (ER), cytosol, mitochondria, and peroxisome, respectively.⁹ There are two parallel metabolic pathways, the classical pathway and the alternative pathway, which both perform the first three steps of BA synthesis but use the same enzymes for phase II conjugation. ~90% of human BAs and ~75% of mice BAs are products of the classical pathway. The remaining ~10% in humans and ~25% in mice are synthesized via the alternative pathway, otherwise known as the acidic pathway.¹⁰ Metabolic flux through the alternative pathway has been correlated with the upregulation of Cytochrome P450 Family 7 Subfamily B Member 1 (CYP7B1) expression in response to adaptive physiological responses, such as those in response to liver disease, a high fat (HF)/high cholesterol (HC) diet, or cold exposure.^11,12,13

Fig. 1

Biochemical synthesis and maturation of BAs. Schematic diagram depicting the biochemical synthesis of BAs. A prefix of G represents a glycine conjugate, while a prefix of T represents a taurine conjugate. Far left and far right labels describe the enzymatic reactions shown parallel within the tree. Top numbering keeps track of which conjugated primary BAs become which secondary BAs. This figure was created with BioRender.com

The classical pathway of BA synthesis

The classical pathway for BA synthesis begins with 7α-hydroxylation initiation of cholesterol through the pathway’s rate-limiting enzyme Cytochrome P450 Family 7 Subfamily A Member 1 (CYP7A1).¹⁴ For the first branch of the classical pathway, following initiation, C₃, C₄, and C₅ are oxidized to a 4α,5β-enone by 3-Beta-Hydroxysteroid Dehydrogenase Type 7 (HSD3β7) and are 12α-hydroxylated by Cytochrome P450 Family 8 Subfamily B Member 1 (CYP8B1). Subsequently, the A ring is reduced in the 5β and 3β positions by Aldo-keto Reductase Family 1 Member D1 (AKR1D1) and Aldo-keto Reductase Family 1 Member C4 (AKR1C4), respectively. Lastly, C₂₇ is oxidized three times into an alcohol, aldehyde, and eventually a carboxylic acid by Cytochrome P450 Family 27 Subfamily A Member 1 (CYP27A1). This first metabolic branch produces Trihydroxycholestanoic Acid (THCA), the precursor for the BA Cholic Acid (CA). In comparison, after CYP7A1 metabolism, BAs may partake in a second metabolic branch that skips 12α-hydroxylation by CYP8B1. Instead, 7α-hydroxycholesterol is metabolized by HSD3β7, AKR1D1, AKR1C4, and CYP27A1 to produce Dihydroxycholestanoic Acid (DHCA), the precursor for the BA Chenodeoxycholic Acid (CDCA).¹⁵

The alternative pathway of BA synthesis

The alternative pathway for BA synthesis begins with C₂₇ oxidation of cholesterol to a spectrum of alcohol, aldehyde, and carboxylic acid metabolites by CYP27A1. Subsequently, the alternative pathway 7α-hydroxylates with CYP7B1, the alternative pathway analogous enzyme to CYP7A1.¹⁴ The remainder of the alternative pathway is identical only to the DHCA-synthetic branch of the classical pathway: HSD3β7, AKR1D1, AKR1C4, and CYP27A1.^14,15 However, the final CYP27A1 mediated step for the alternative pathway is different compared to the classical pathway, in which only one or two oxidations are performed to complete the oxidation of C₂₇ into a carboxylic acid.¹⁴

Convergent synthesis—BA conjugation

Convergently, both the classical and alternative pathways meet at the common conjugation phase of synthesis. The carboxylates of THCA and DHCA are activated to form thioesters with Coenzyme-A (CoA) by BA-CoA Synthase (BACS) and are C₂₅ epimerized from R to S stereochemistry by α-Methylacyl-CoA Racemase (AMACR). The thioester is α,β desaturated at C₂₄ and C₂₅ by Acyl-CoA Oxidase 2 (ACOX2), is oxidized to a β-keto thioester by D-Bifunctional Protein (HSD17β4), and is thiolysed by Peroxisomal Thiolase 2 (SCP2) to release propionyl-CoA.¹⁵ These last three reactions heavily mimic the β-oxidation of odd-numbered fatty acids: desaturation, β-keto thioester formation, and thiolysis. The resulting CA-CoA and CDCA-CoA are then C₂₄ phase II conjugated by BA-CoA:amino acid N-acetyltransferase (BAT) to produce C₂₄ glycine or taurine conjugated BAs in humans or only taurine conjugates in rodents.¹⁶ In small quantities sulfation at C₃ or C₆ by Sulfotransferase Family 2 A Member 1 (SULT2A1) or Family 2B Member 8 (SULT2B8) may occur. In addition, small quantities of glucuronidated metabolites at C₃, C₆, or C₂₄ by UDP-Glucuronosyltransferase Family 1 Member A3 (UGT1A3), Family 2 Member B4 (UGT2B4), or Family 2 Member B7 (UGT2B7) may occur.^17,18 Products synthesized after 7α-hydroxylation initiation, sterol ring modification, side chain truncation, and phase II conjugation are denoted as primary conjugated BAs: glycine/taurine conjugates of CA and CDCA in the classical pathway and only CDCA in the alternative pathway.¹⁹ CDCA may be further metabolized by Cytochrome P450 Family 3 Subfamily A Member 4 (CYP3A4) to produce Hyocholic Acid (HCA), alternatively known as γ-Muricholic Acid (γMCA), in trace quantities in humans and substantial quantities in rodents. Only rodents metabolize CDCA with Cytochrome P450 Family 2 Subfamily C Member 70 (CYP2C70) to produce α-Muricholic Acid (αMCA), an intermediate whose 7α-alcohol is further epimerized by CYP2C70 to 7β to produce β-Muricholic Acid (βMCA).^20,21,22,23

Microbial maturation of primary BAs to secondary BAs

A brief overview of BA maturation

Enterohepatic circulation and gut exposure provide ample opportunities for BA modification and diversification (Fig. 2). The gut microbiome can modify primary BAs by deconjugation with Bile Salt Hydrolase (BSH), 7α/β-dehydroxylation with proteins of the bai operon, and epimerization by the family of Hydroxysteroid Dehydrogenases (HSDHs) to produce what are known as secondary BAs.²⁴ Once systemically reabsorbed, a topic developed in further sections, secondary and primary unconjugated BAs may be reconjugated. Absent in humans, mice express Cytochrome P450 Family 2 Subfamily A Member 12 (CYP2A12) and are able to 7α-rehydroxylate 7α/β-dehydroxylated secondary BAs.²⁵ In humans, the main secondary BAs are Deoxycholic Acid (DCA), Lithocholic Acid (LCA), and Ursodeoxycholic Acid (UDCA), while Hyodeoxycholic Acid (HDCA) is found in trace quantities.¹⁹ Similar to humans, rodents synthesize DCA, LCA, but generate larger quantities of UDCA. Rodents additionally metabolize αMCA and βMCA to create substantial quantities of HDCA and ω-Muricholic Acid (ωMCA).²⁶

Fig. 2

Microbial maturation of primary BAs to secondary BAs. Schematic diagram depicting the gut maturation of primary conjugated BAs. 1* and 2* are used as shorthand notation to represent primary and secondary, respectively. NAD+ and NADH are the biochemical cofactors Nicotinamide Adenine Dinucleotide (NAD+) and Nicotinamide Adenine Dinucleotide Hydride (NADH), respectively. Red circled regions and atoms highlight the motifs changed by the prior performed metabolic reaction. This figure was created with BioRender.com

Deep-dive, BSH - BA deconjugation & microbially conjugated BAs

BSH is a N-terminal nucleophilic hydrolase that uses a N-terminal cysteine residue to deconjugate primary BAs, commonly thought to serve as gatekeeper reaction for further BA maturation.^27,28 Control over BA deconjugation via BSH prevalence is important since the conversion of CA/CDCA conjugates into their free form greatly enhances their antimicrobial nature, restricting Clostridium difficile growth.^29,30 BSH is found across most major phyla such as gram-positive Bifidobacterium, Lactobacillus, Clostridium, Enterococcus, Listeria, and gram-negative Stenotrophomonas, Brucella, and Bacteroides, which is suggested to play a substantial role in total deconjugation.^{31,32,33,34,35,36,37,38,39} This widespread prevalence of BSH suggests that BSH may be horizontally transferable.^28,40,41

Contrary to original belief, BA conjugation is not a process that is exclusive to the liver. In addition to deconjugating glycine/taurine conjugates, microbial BSH may also conjugate microbially synthesized (CA/CDCA/DCA/UDCA)-CoA to other amino acids in a microbiota-dependent manner. In addition, BSH may catalyze trans-amidation reactions, where glycine/taurine C₂₄ motifs are exchanged for that of other amino acids.⁴² All proteinaceous amino acid conjugates have been found except those containing aspartate or proline. This process has been identified in Enterocloster bolteae, Clostridium perfringens, and the resulting molecules are referred to as microbially conjugated BAs (MCBAs).^28,43,44

In comparison to the role of glycine/taurine conjugates, MCBAs contain varied lipophilicities and carry modified steric bulk, modulating BA physiochemical and pharmacological properties.⁴⁵ The more hydrophobic the MCBA, the more potent it acts as an antimicrobial agent, perhaps acting as a regulatory feedback loop against MCBA synthesis.⁴³ MCBAs have already been characterized to both increase and decrease the pharmacological properties of canonical BAs at nuclear transcription factor receptors.⁴⁴ The differential pharmacological activity of MCBAs compared to classical BAs is strongly species dependent. Comparing humans and mice, some MCBAs provide similar pharmacological effects relative to canonical BAs, while others provide entirely different or neutral effects.⁴⁴ Microbial conjugates of CA have been found to retain FXR agonism and are enriched in IBD and cystic fibrosis patients, although it is unknown if it is causative or correlative. More research is needed to see how this new form of conjugation modifies the properties of other BAs and impacts disease pathogenesis.

Deep-dive, 7α/β-dehydroxylation and the bai operon

7α/β-dehydroxylation of BAs has been strongly associated with the Clostridium family of bacteria, including C. scindens, C. hylemonae, C. hiranonis, and C. sordellii, encoded by the 9 gene bai operon.^{46,47,48,49,50} Through the process of 7α-dehydroxylation, CA is converted into DCA and CDCA/UDCA are converted into LCA. C₇ hydroxylated BAs such as CA and CDCA/UDCA are transported into bacteria by baiG, and from such are C₂₄ conjugated into a thioester with CoA by baiB.⁵¹ BaiA2 then oxidizes the C₃ alcohol to form a ketone. BaiCD oxidizes C₄/C₅ of 7α-hydroxy BAs, while baiH performs the same for 7β-hydroxy BAs, both of which creating 4α,5β-enones, similar in outcome to the reaction performed by HSD3B7 in primary BA synthesis.^50,52,53 The enzyme baiF then performs a trans-thioesterification, in which the CoA of the oxidized 7-hydroxy metabolite is transferred to a new substrate BA.⁴⁶

The rate limiting step of this pathway is performed by the 7-dehydratase baiE, which although not specific for CoA conjugation, is enzymatically more efficient with CoA conjugates.⁵³ Catalytically a D35/H83 catalytic dyad abstracts the 6α-H, allowing for electron delocalization within the adjacent conjugated enone π-system. Through a suspected E1cb mechanism, the 7-hydroxyl (OH) is eliminated as a leaving group which retrieves the proton originally abstracted by H83 and exits the enzyme as water, resulting in the formation of a 4α,5β,6γ,7δ-enone.⁵⁴ Two steps of reduction are performed by baiN to convert the substrate from a 4α,5β,6γ,7δ-enone to a 4α,5β-enone and to a 3-oxo-BA, while baiA2/baiO performs the final step in reducing C₃ back to an alcohol.^55,56

Bacteria conserve nine genes to perform BA 7α/β-dehydroxylation

For what reason do bacteria conserve the entire bai operon if it is just baiE/baiN that perform the net reaction? In other words, what value do the other genes add to the operon that cannot be found purely with baiE/baiN? BaiB likely serves as a commitment enzyme analogous to that of thiokinase in the process of β-oxidation. In this case, baiB likely provides the bacteria with a way to regulate pathway activity besides that intrinsic to operon transcriptional regulation. BaiF likely serves multiple valuable functions. In situations of high bacterial BA uptake, it may enhance the steady-state rate at which BAs are committed to the pathway. If baiB does exhibit product inhibition or post-translational regulation, baiF may allow the pathway to continue at a basal rate regardless of baiB activity. Lastly, baiF by performing a trans-thioesterification is energetically thrifty, preventing the bacterial expenditure of more ATP for CoA ligation.

In metabolism it is uncommon if not unseen to see direct deoxygenation of a monofunctional alcohol to an alkane. The rest of the bai operon proves its value by breaking this single complex problem down to multiple standard biochemical reactions. BaiA2 oxidizes the C₃ alcohol to a ketone to enhance its electron-withdrawing power, using the standard reversible biochemistry of alcohol to ketone oxidations (Ex: Lactate Dehydrogenase). This new ketone increases the ease at which baiCD/baiH is able to further oxidize, resembling the reverse reaction of enzymes commonly associated with steroid metabolism (Ex: 5α-Reductase). BaiE dehydrates and baiN reduces in close proximity to the baiA2-baiCD/baiH installed electron-withdrawing enone, very similar to fatty acid synthesis (Ex: Fatty Acid Synthase). BaiO/baiA2 finishes the pathway by reversing the initial alcohol to ketone oxidation. At first glance, this pathway seems redundant, but upon further observation it is incredible to see how metabolism repurposes various standard archetypal reactions to obtain new outcomes.

Deep-dive, HSDH epimerization

BAs may be oxidized and epimerized by a two-step reaction, in which the existing alcohol is oxidized to a ketone by the removal of a hydride ion, followed by reduction of the ketone on the opposite face. These two reactions are enzymatically performed by different position-specific HSDHs, likely not originating from the same species of bacteria.⁵⁷ BAs may be oxidized at the 3, 7, or 12 positions to generate oxo-intermediates and are then reduced to form 3, 7, or 12 epimers. 3-dehydrogenation is associated with Blautia producta, Eggerthella genus, Enterorhabdus mucosicola, and Acinetobacter lwoffii.^{56,58,59,60,61} Eggerthella lenta 3α-HSDH is a rare exception that is able to metabolize BAs in their conjugate form, challenging the general idea that BSH metabolism is necessary for further maturation into a secondary BA.⁶² 7α-dehydrogenation is confirmed in Clostridium baratii, while 7β-dehydrogenation is found in Ruminococcus gnavus, Clostridium absonum, Stenotrophomonas maltophilia, and Collinsella aerofaciens.^{63,64,65,66,67} 12α-dehydrogenation has been found in Eggerthella lenta, Enterorhabdus mucosicola, Clostridium scindens, Peptacetopacter hiranonis, Clostridium hylemonae, and Bacteroides.^{41,56,60,61,68} 12β-dehydrogenation has been found in Clostridium paraputrificium, Clostridium tertium, and Clostridium difficile.⁶⁹ Through the process of epimerization, CA may be C₇ epimerized to Ursocholic acid (UCA), C₃ epimerized to isoCA, or C₁₂ epimerized to Epicholic acid (ECA). CDCA may be C₇ epimerized to form UDCA or C₃ epimerized to form isoCDCA, while LCA is C₃ epimerized to isoLCA.

BA pool composition: structure/functionality

Structural diversity provides varying degrees of hydrophobicity

The tetracyclic fused steroidal backbone of BAs has both a concave hydrophilic α-face and a hydrophobic convex β-face, in which hydrophilicity is correlated to both the number and positioning of OH groups on C₆, C₇, and C₁₂ (Fig. 3). Experimentation with hydrophilic-hydrophobic indices has revealed that the addition of a single β-OH group to an already existing poly-α-hydroxylated scaffold can greatly increase hydrophilicity. The addition of polar surface area to the hydrophobic β-face disrupts pure hydrophobic interactions, resulting in a BA that is less amphipathic and more hydrophilic. The relative physiochemical hydrophilicities of unconjugated free BAs are as follows: ωMCA > βMCA > HCA > αMCA > UDCA > HDCA > CA > CDCA > DCA > LCA. Relative rankings remain the same regardless of the type of conjugation. As a whole, the hydrophilicities of BA conjugation types are as follows: taurine conjugates > glycine conjugates > free BAs.⁷⁰

Fig. 3

Structural, physiochemical, and pharmacological properties of BAs: Representation of the structural scaffold, conjugates, physiochemical properties, and ranking of endogenous agonist pharmacology at the canonical BA receptors. Sites R₁ through R₆ represent locations for enzymatic hydroxylation. α-orientation groups are located beneath the steroidal plane, while β-orientation hydroxyl groups are located above the steroidal plane. Subscripted values next to BA names represent the species of origin: (H = Human), (R = Rodent), (Ht = Human Traces), (Rt = Rodent Traces). This figure was created with BioRender.com

Individual BAs work together to produce the BA pool

The BA pool is defined as all the BAs within enterohepatic circulation, disregarding those in systemic circulation. Breaking this down by tissue, ~85–90% of BAs are in the gut, 10–15% in the gallbladder, and 1% in the liver.⁷¹ As a guiding principle, BAs found in the liver are strongly conjugated. Those within the biliary tree, small intestine, and systemic circulation are moderately conjugated, while large fractions of free BAs are found in the large intestine and feces.

BA pool composition is defined as the following four metrics: percent occurrence per BA, percent 12α-hydroxylated, percent glycine/taurine conjugated, and percent primary BAs vs. secondary BAs. The BA pool composition of different biological tissues/fluids is frequently measured by liquid chromatography-mass spectrometry analysis (Table 1). The human hepatic BA pool is mostly comprised of CA and CDCA. Once exposed to the gut and excreted as feces, the human BA pool composition shifts strongly more towards the hydrophobic BAs DCA and LCA. The human serum BA composition, derived from hepatic excretion and gut BA reabsorption, is mostly comprised of CA, CDCA, DCA, and UDCA.

Table 1 Interspecies variation in BA pool composition: human vs. mice

It is incredibly important to note that tissue-specific BA pool composition is highly species-dependent. To contrast the BA pool of humans with rodents such as mice, an overwhelming proportion of the mice BA pool in all tissues is composed of the highly polar MCA family. This is imperative to note since as we shall discuss the MCA family of BAs are potent FXR antagonists. Conjugation of human BAs mostly favors glycine conjugation, while in rodents is exclusive to taurine conjugation. Both in MCA prevalence and conjugation preference, interspecies differences between human and rodent BA pools make it incredibly challenging to translate research from rodent models to human applications. From such it is evident that wild-type mice models do not perfectly replicate the properties of the human BA pool.⁷²

A balanced BA pool is essential for gut lipid absorption

The specific physiochemical properties and amphipathic nature of BAs allows for them to act as digestive surfactants, assisting in the solubilization and absorption of lipid nutrients and lipid-soluble vitamins (A, D, E, K). Deviations from a normal BA pool composition by either impaired synthetic or impaired enterohepatic circulation mechanisms will result in an impaired ability to absorb adequate nutrition and have normal growth patterns.^14,73,74 Phase II conjugation with glycine or taurine greatly increases the acidity of BA carboxylate groups, lowering their unconjugated pKa from ~5 to ~4 in the case of glycine conjugates and ~1 in the case of taurine conjugates.⁷⁵ The lower the average pKa of the BA pool, the higher the overall ratio of salt to protonated acid. This effectively allows for the biochemical optimization of BA pool composition to obtain physiological pH solubilization of fat-soluble nutrients by assisting in micelle formation.⁷⁶ BA pool compositions that are too hydrophobic are hepatotoxic, causative for cholestasis, and are poor surfactants. In comparison, an overly hydrophilic composition cannot stabilize BA-nutrient micelles.⁷⁷ Therefore, the maintenance of proper species-specific BA pool composition is essential to ensure the optimal surfactant-mediated absorption of lipid-nutrients.

BA pool composition drives balanced pharmacological profiles

Alongside surfactant properties, different BAs have varying pharmacological profiles at the two canonical BA receptors FXR and TGR5. The efficacy and potency rankings for common human BAs as FXR agonists are: CDCA > DCA > LCA > CA >> UDCA/MCAs, and CDCA > LCA > DCA > CA, respectively.^78,79,80,81 The efficacy of UDCA and all MCAs are so low that functionally in the presence of other more potent BAs they act as FXR partial-agonists or antagonists.^82,83 The analogous rankings for efficacy and potency are identical for TGR5 and is: LCA > DCA > CDCA > CA.^84,85 Disregarding trace quantities, within the human BA pool CDCA represents the most potent/efficacious FXR agonist, while DCA and LCA are the most potent/efficacious TGR5 agonists. Both most potent/efficacious TGR5 agonists are secondary BAs, highlighting the importance of BA maturation in modulating TGR5 agonism. In healthy human physiology the BA pool consists of more CDCA than DCA/LCA. Accordingly, any factor that manipulates the relative ratio between classical and alternative pathway metabolic flux, or gut conversion of CA to DCA or CDCA to LCA is key in influencing BA pool composition and hence the BA pool agonist profile. Increasing metabolic flux through the alternative pathway by inhibiting or transcriptionally repressing the classical pathway gatekeeper enzymes CYP7A1 or CYP8B1 will produce a BA pool that has higher FXR agonism and decreased TGR5 agonism.^10,86

BA physiology

Mechanisms of enterohepatic circulation & maturation of BAs

The steady-state BA pool composition is strongly influenced by the tightly regulated physiological processes of enterohepatic circulation, maturation, and fecal excretion of BAs (Fig. 4). Conjugated BAs are secreted from the basolateral surface of hepatocytes into the bile canaliculi for storage in the gallbladder.⁸⁷ Glycine and taurine conjugated BAs are secreted by Bile Salt Export Pump (BSEP), while sulfate and glucuronate conjugated BAs are secreted by Multidrug Resistance-Associated Protein 2 (MRP2).⁸⁸ The post-prandial release of Cholecystokinin (CCK) contracts the gallbladder, leading to the release of bile into the duodenum.⁸⁹ Within the duodenum, BAs assist in the previously mentioned solubilization of fat-soluble nutrients and mature into secondary BAs.^88,90 Approximately 95% of intestinal BAs are reabsorbed into the hepatic portal vein due to terminal ileal reuptake by Apical Sodium-Dependent BA Transporter (ASBT) and passive diffusion.^91,92 Residual BAs that escape ileal reuptake can be excreted in feces, establishing the only known route for cholesterol excretion.⁹³

Fig. 4

Physiological mechanisms behind BA enterohepatic circulation, maturation, and fecal excretion: Schematic diagram representation of the physiology behind BA enterohepatic circulation, maturation, and fecal excretion. BAs are synthesized in the liver and are excreted into the duodenum for maturation. Matured BAs may either be systemically reabsorbed or excreted in feces. This figure was created with BioRender.com

Once within terminal ileal enterocytes, the cytosolic Ileal BA Binding Protein (I-BABP) chaperones BAs from the apical to basolateral membrane, where Organic Solute Transporter α/β (OST (α/β)) and Multidrug Resistance-Associated Protein 3 (MRP3) allow for BA entry into hepatic portal circulation.^94,95 In contrast, enteric apical MRP2 effluxes BAs back into the ileal lumen. From the hepatic portal vein, conjugated BAs enter sinusoidal hepatocytes by the use of apical Sodium-Taurocholate co-Co-transporting Polypeptide (NTCP), while free BAs enter by use of apical Organic Anion-Transporting Polypeptide 1 (OATP1), both of which complete the cycle of enterohepatic circulation.^75,90 In situations of hepatic BA overload, the liver can efflux BAs into the systemic circulation via basolateral MRP3, Multidrug Resistance-Associated Protein 4 (MRP4), or OST (α/β).¹⁰ The hepatic suppression of BSEP/MRP2 expression will limit the quantity of primary BAs that reach the gut for maturation. The ileal suppression of ASBT, MRP3, OST (α/β) expression or induction of MRP2 will enhance fecal loss of BAs and prevent the uptake of gut-modified secondary BAs. Lastly, modifications to the microbial composition of the gut may alter which biosynthetic pathways are utilized for BA maturation.

BA signaling cascades

Canonical BA receptor—FXR, a peripheral signal for nutrient density

Molecular biology of FXR

FXR is a nuclear transcription factor that is encoded by the gene FXRα/NR1H4 and is expressed as four separate isoforms (FXRα1-α4). FXR isoforms dimerize with Retinoid X Receptor Alpha (RXRα) and bind to two distinct FXR Response Elements (FXREs), the Inverted Hexamer Spaced by 1 Nucleotide (IR-1) motif (5’-AGGTCA-X-TGACCT-3’) and the Everted Hexamer Repeat Spaced by 2 Nucleotides (ER-2) motif (5’-TGACCT-XX-GGGTCA-3’) to regulate the transcriptional activity of target genes (Fig. 5a).^80,96,97,98 The ENSEMBL database identifies 388 sequence orthologues for the FXR receptor, where the sequence homologies of Mus musculus and Rattus norvegicus compared to the human sequence are 87.3% and 81.64%, respectively.⁹⁹ Due to sequence divergence between species, results from animal models may not directly correlate to the human FXR receptor. Cumulative reports from the Protein Atlas database identify the main subcellular location of FXR as within the nucleoplasm. mRNA-seq expression reveals that FXR is expressed minorly in most organs. The following list ranks the top 8 FXR-expressing organs in terms of nTPM (normalized mRNA transcripts per million): liver, small intestine, kidney, adrenal glands, ovary, gallbladder, colon, bladder, pancreas.¹⁰⁰

Fig. 5

Transcriptional and epigenetic roles of FXR and the FXR-SHP axis. a The FXR/RXRα heterodimer binds to either an IR-1 or ER-2 containing FXRE to transcriptionally regulate target gene activity. The FXR/RXRα crystal structure shown is PDB 5Z12.⁷²⁸ b The FXR-SHP axis inhibits LRH-1/HNF4α-driven gene expression. SHP directly dimerizes with LRH-1/HNF4α to prevent their transcriptional activity. In addition, SHP recruits Swi/Snf complex-mediated epigenetic repression. LRH-1 inhibition represses SHP transcription, forming a negative feedback loop. c Epigenetic regulation of FXR target genes by the ASCOM complex. BA agonized FXR/RXRα recruits the ASCOM complex to enhance the expression of FXRE-containing genes. Without FXR agonism, p53 and p53BP1 provide basal SHP expression. This figure was created with BioRender.com

Epigenetic regulation of FXR transcription

Epigenetic methylation has been found to impact FXR activity, in which CpG methylation found in both mouse and human colon cancers directly reduces transcription of FXR.^101,102,103 13 sites for CpG methylation have been found in adenomatous polyposis coli deficient mice models.¹⁰³ A clear relationship has been found between FXR CpG methylation status and the BA pool comparing healthy pregnant women to those with intrahepatic cholestasis, in which patients with reduced methylation and higher FXR transcription were associated with cholestasis, likely due to the repression of trans-biliary bile flux.¹⁰⁴

Post-translational and post-transcriptional regulation of FXR activity

The transcriptional activity of FXR at FXREs is directly reduced by acetylation at K217, in which p300 acetylates and Sirtuin 1 (SIRT1) performs NAD+-dependent deacetylation.^105,106 SIRT1 further modulates the transcription of FXR by activating Hepatocyte Nuclear Factor 1 Alpha (HNF1α).¹⁰⁷ This regulation is further complicated by the expression of miRNA. miR-34a and miR-22 are both known to silence the expression of SIRT1, hence reducing FXR expression. In addition, FXR exhibits negative autoregulation by inducing the transcription of miR-22.^108,109,110 Lastly, FXR has been found to be SUMOylated by Small Ubiquitin-Related Modifier 1 (SUMO1) and 2 (SUMO2), inhibiting its ability to trans-activate FXRE-containing genes.¹¹¹

Molecular biology of the FXR-SHP axis

The nuclear FXR receptor has four main functions: to signal for incoming nutrient density (lipid, carbohydrate, amino acid), to enforce negative autoregulation of BA synthesis to prevent BA overload, to reduce the extent of enteric cholesterol absorption, and to inhibit inflammatory responses. Direct FXR agonism induces the expression of Small Heterodimer Partner (SHP), a corepressor which without a DNA binding domain dimerizes with transcription factors to directly inhibit them and/or epigenetically repress their target genes.¹¹² This is known as the FXR-SHP axis. All FXR isoforms induce SHP expression by binding to a IR-1 motif, while FXRα2 and FXRα4 may additionally bind to a lone ER-2 motif to induce stronger SHP expression.⁹⁸

The FXR-SHP axis transcriptionally and epigenetically regulates target genes

There are two major epigenetic transcriptional complexes associated with FXR-SHP axis activity. The first of which are the ATP-dependent Swi/Snf chromatin remodeling complexes (Fig. 5b). Swi/Snf chromatin remodeling complexes require an ATPase such as Probable Global Transcriptional Activator SNF2L2 (BRM) or Transcriptional Activator BRG1 (BRG1), other proteins known as BRM or BRG1 associated factors (BAFs), the Paired Amphipathic Helix Protein Sin3a (Sin3a)/Histone Deacetylase 1/2 (HDAC-1/2) corepressor complex, and the histone methyltransferase G9a to modulate chromatin compaction.^113,114,115 In such, BA-activated FXR/RXRα dimers recruit the BRG1 Swi/Snf complex, providing chromatin relaxation and allowing for FXR-stimulated transcription, such as that which occurs with SHP. SHP may recruit the BRM Swi/Snf complex, which allows for H3K9/K14 deacetylation, H3K9 methylation, direct inhibition of Liver Receptor Homolog 1 (LRH-1)/Hepatocyte Nuclear Factor 4 Alpha (HNF4α) transcriptional activation, and epigenetic compaction of LRH-1/HNF4α target genes.^113,116,117

A secondary epigenetic mechanism is associated with FXR activity, known as the Activating Signal Cointegrator-2 (ASC-2) – Containing Coactivator Complex (ASCOM) (Fig. 5c).¹¹⁸ ASCOM is additionally composed of either Histone Methyltransferase Mixed-Lineage Leukemia Proteins 3 (MLL3) or 4 (MLL4) histone methyltransferases and the histone demethylase Ubiquitously Transcribed Tetratricopeptide Repeat, X Chromosome (UTX). Respectively, these methylate H3K4 and remove methylation at H3K27, cumulatively increasing transcriptional activity.^119,120 BA-activated FXR/RXRα dimers recruit the ASCOM complex to enhance the transcription of SHP, BSEP, MRP2, and NTCP.^120,121 The ASCOM complex has an additional role with respect to the Tumor Suppressor Transcription Factor p53 (p53). Under normal or stressed conditions the SHP promoter may be bound by p53, which by the scaffolding protein p53 Binding Protein 1 (p53BP1) recruits the ASCOM complex and transcribes SHP.¹²²

Molecular physiology of the FGF15/19-[FGFR4/βK]-FXR axis

Alongside direct BA agonism of FXR, the hepatic FXR signaling axis is also modulated by downstream signaling from a complex of the integral membrane protein β-Klotho (βK) and the receptor tyrosine kinase Fibroblast Growth Factor Receptor 4 (FGFR4) (Fig. 6).¹²³ Gut FXR agonism stimulates the systemic secretion of the FGFR4/βK complex agonists Fibroblast Growth Factor 15 (FGF15 – mice orthologue) and Fibroblast Growth Factor 19 (FGF19 – human orthologue), serving as long-distance signals for high enteric levels of FXR agonists.^124,125 Specific to the hunger center of the hypothalamus, a secondary role of FGF15/19 is to suppress appetite by inhibiting Agouti-Related Peptide (AgRP) and Neuropeptide Y (NPY) neurons.^126,127

Fig. 6

Gut FXR signaling. Gut FXR signaling within enterocytes induces BA/cholesterol efflux, ceramide synthesis, and FGF15/19 secretion. Proteins with a yellow outline represent kinases and phosphatases, while yellow arrows represent phosphorylation/dephosphorylation. Yellow P’s represent phosphate groups. Italicized text refers to genes while normal text refers to proteins. Question marks represent unknown mechanisms. This figure was created with BioRender.com

In the liver, FGFR4/βK complex agonism phosphorylates the adaptor protein Fibroblast Growth Factor Receptor Substrate 2 (FRS2) and recruits Growth Factor Receptor Bound Protein 2 (Grb2), Grb2 Associated Binding Protein 1 (Gab1), and Src Homology Region 2 Domain-Containing Phosphatase-2 (SHP2).¹²⁸ SHP2 binding to GAB1 promotes the dephosphorylation of Phosphoprotein Associated with Glycosphingolipid-Enriched Microdomains (PAG), preventing the membrane anchoring of the Proto-Oncogene Tyrosine-Protein Kinase Src (Src) inhibitory kinase C-Terminal Src Kinase (Csk).¹²⁹ The release of Csk-mediated inhibition allows for Src to phosphorylate Y67 of FXR and promote nuclear translocation and transcriptional regulation.^128,130 This is denoted as the [FGFR4/βK]-FXR axis. Grb2 additionally recruits the Ras guanine nucleotide exchange factor Son of Sevenless (Sos), which in conjunction with Src activity activates the Ras-Raf-MEK-ERK signaling cascade, resulting in the phosphorylation and activation of Extracellular Signal-Related Kinase (ERK).^123,131 This is known as the [FGFR4/βK]-ERK axis.

Alongside FGF15/19 secretion, gut FXR agonism induces the systemic secretion of ceramides. Gut FXR transcriptionally enhances the expression of the following ceramide synthetic proteins: Serine Palmitoyltransferase Long Chain Base Subunit 3,4 (SPTLC3, SPTLC4), Sphingomyelin Phosphodiesterase 3,4 (SMPD3, SMPD4), Ceramide Synthase 2,4 (CERS2, CERS4), and Delta 4-Desaturase Sphingolipid 1,2 (DEGS1, DEGS2).¹³²

Master regulators of BA homeostasis—FXR and the FGF15/19-[FGFR4/βK]-FXR axis

Feedback inhibition for BA synthesis is mediated by two mechanisms, one directing from FXR itself, and another originating from agonized FGFR4/βK complex (Fig. 7). BA agonized FXR induces the expression of SHP, while the downstream actions of agonized [FGFR4/βK]-ERK axis mediated phosphorylation of SHP prevents its proteasomal degradation, both leading to increased SHP levels.^133,134 SHP dimerizes and directly inhibits the activity of the transcription factors LRH-1 and HNF4α, in addition to inducing the epigenetic repression of their target genes, preventing the transcription of CYP7A1 and CYP8B1. SHP-mediated inhibition and epigenetic repression of LRH-1 target genes represses its own transcription, forming a negative feedback loop.^{135,136,137,138} HNF4α-induced CYP7A1 and CYP8B1 expression is dependent on the recruitment of COUP Transcription Factor 2 (COUP-TFII), the histone acetyl transferase CREB Binding Protein (CBP), Transcription Factors IID (TFIID), and IIB (TFIIB).^139,140,141 In addition, Sos activates Ras, which agonizes Ras-like protein (Ral) and subsequently Mammalian Target of Rapamycin Complex 1 (mTORC1). mTORC1 and ERK phosphorylate and prevent the nuclear translocation of Transcription Factor EB (TFEB), preventing TFEB-induced CYP7A1 expression.^142,143

Fig. 7

Hepatic FXR signaling—BA autoregulation. Hepatic FXR signaling pertaining to BA autoregulation. Hepatic FXR agonism increases BA/cholesterol efflux, primary BA conjugation, and represses BA synthesis. Dashed mechanism arrows represent the Ras-Raf-MEK signaling cascade, while dashed arrows through a protein represent the direction of substrate flux. Proteins with a yellow outline represent kinases and phosphatases, while yellow arrows represent phosphorylation/dephosphorylation. Yellow P’s represent phosphate groups. Italicized text refers to genes while normal text refers to proteins. Question marks represent unknown mechanisms. This figure was created with BioRender.com

Repression of CYP7A1 and CYP8B1 directly lowers metabolic flux through the classic pathway of BA synthesis.^{137,144,145,146} This two-sided negative-feedback loop lowers the overall BA pool size via CYP7A1 repression and composition via CYP8B1 repression, shunting BA synthesis into the alternative pathway and producing the strong FXR agonist CDCA. This results in a smaller BA pool that is more biased in composition towards FXR agonism. FXR is therefore referred to as the master regulator of BA synthesis.^135,147 In addition, FXR agonism can help prevent systemic BA overload-induced toxicity. High concentrations of BAs and FXR-SHP axis agonism represses hepatic NTCP expression and gut ASBT expression to prevent further BA entry, while FXR induces BSEP, MRP2, I-BABP, OSTα (liver-specific), and OSTβ to increase BA fecal excretion.^92,148 BACS, BAT, UGT2B4, and SULT2A1 are also upregulated by FXR to increase the rate at which hepatic BAs efflux into the biliary tree.¹⁴⁹ Sulfation reduces gut BA reabsorption, assisting in the elimination of excess BAs.¹⁷

The FGF15/19-[FGFR4/βK]-FXR axis modulates triglyceride metabolism

Low FXR agonism enhances the rate of cholesterol excretion while high FXR agonism represses the synthesis of triglycerides (Fig. 8). FXR-induced SHP can dimerize with the transcription factor liver X receptor (LXR) and inhibits its induction of sterol regulatory element-binding protein 1C (SREBP-1C), the master transcriptional regulator of triglyceride synthesis.^150,151 In addition, [FGFR4/βK]-ERK axis activation represses SREBP-1C activity by inducing ERK phosphorylation of SHP, which recruits DNA Methyltransferase-3A (DNMT3A) to epigenetically repress SREBP-1C target genes.¹⁵² Repression of SREBP-1C activity directly reduces the transcription of Fatty Acid Synthase (FAS), Acetyl-CoA Carboxylase (ACC), Stearoyl-CoA Desaturase enzyme 1 (SCD1), and ATP Citrate Lyase (ACLY) genes, key players in triglyceride synthesis. In addition, FXR-SHP axis activation inhibits carbohydrate response element binding protein (ChREBP), another transcription factor which when activated by glucose metabolites induces lipogenesis by upregulating FAS, SCD1, ACC, and ACLY.¹⁵³ FXR has been shown to induce the expression of Apolipoprotein CII (ApoC-II) and competitively inhibit the HNF4α-mediated transcription of Apolipoprotein CIII (ApoC-III). ApoC-II expression and ApoC-III repression increase the activity of lipoprotein lipase (LPL), allowing for a more efficient removal of triglycerides out of the systemic circulation into peripheral tissue.^154,155,156 Furthermore, FXR has been experimentally shown to induce peroxisomal proliferator-activated receptor alpha (PPARα), a key transcription factor responsible for the catabolism of fatty acids.^157,158

Fig. 8

Hepatic FXR signaling—lipid metabolism. Hepatic FXR signaling pertaining to lipid metabolism represses triglyceride synthesis, VLDL packaging, increases lipid utilization, chylomicron uptake, and promotes a pro-atherogenic serum lipid profile. Dashed mechanism arrows represent the Ras-Raf-MEK signaling cascade, while dashed arrows through a protein represent the direction of substrate flux. Proteins with a yellow outline represent kinases and phosphatases, while yellow arrows represent phosphorylation/dephosphorylation. Yellow P’s represent phosphate groups. Italicized text refers to genes while normal text refers to proteins. Question marks represent unknown mechanisms. This figure was created with BioRender.com

FXR modulation of cholesterol metabolism

FXR-SHP axis activation additionally represses very-low density lipoprotein (VLDL) secretion by SHP dimerization with HNF4α, directly inhibiting it and leading to the epigenetic repression of its target genes. This decreases the rate at which the liver distributes cholesterol and triglycerides. When inhibited, HNF4α fails to induce the expression of downstream Apolipoprotein B100 (ApoB100) and Microsomal Triglyceride Transfer Protein (MTTP) genes, both of which are essential to VLDL packaging and synthesis.¹⁵⁹ Although VLDL excretion is reduced, FXR-mediated repression of BA synthesis inhibits cholesterol excretion, allowing for it to accumulate hepatically. Resulting from such, the reduced quantity of VLDL produced will now be cholesterol rich, eventually producing cholesterol-rich low-density lipoproteins (LDL). Long-term FXR agonism contributes towards a potentially pro-atherogenic state due to an RXRα-independent negative regulatory FXRE within the Apolipoprotein A1 (Apo-A1) gene. Apo-A1 is the core structural lipoprotein required for high-density lipoprotein (HDL)-mediated scavenging of systemic cholesterol, necessarily fighting against atherosclerotic plaque development.¹⁶⁰ FXR activity additionally stimulates the expression of Apolipoprotein E (ApoE) and phospholipid transfer protein (PLTP). Functionally this results in enhanced chylomicron/LDL uptake into the liver, in addition to increasing the rate at which HDL particles uptake phospholipids from VLDL, converting them into LDL.¹⁶¹

FXR-mediated changes to serum cholesterol are further complicated by the FXR-induced expression of HDL Scavenger Receptor Class B Type 1 (SR-B1), which increases the hepatic uptake of HDL cholesterol for biliary excretion, further reducing serum HDL cholesterol.^162,163 FXR signaling attempts to ameliorate this systemic cholesterol excess by increasing the expression of hepatic and gut ATP Binding Cassette Subfamily G Members 5 and 8 (ABCG5) (ABCG8), transporters which serve to efflux hepatic cholesterol into the bile canaliculi and to return cholesterol absorbed by enterocytes back into the gut lumen.^{163,164,165,166} Although this may offset some further cholesterol accumulation, increased cholesterol content within the biliary tree without BAs may increase the risk for gallstones.¹⁶⁷ Another compensatory activity of highly agonized FXR-SHP-axis activity is to reduce the expression of Proprotein Convertase Subtilisin/Kexin Type 9 (PCSK9), increasing the activity of the LDL Receptor (LDLr).^168,169 However, it needs to be emphasized that even if more LDL cholesterol is absorbed into the liver, the same problem remains, the liver has no active metabolic pathway for large quantity cholesterol excretion.

Physiological implications—double-edged FXR lipid phenotypes

Cumulatively, it is then very helpful to characterize the impact of FXR agonism on lipid metabolism into two different states of FXR activity. Within a state of low total FXR activity SREBP-1C activity is high, HDL synthesis is not repressed, and cholesterol is being metabolized into BAs and excreted, producing a state of high triglycerides, VLDL cholesterol, HDL cholesterol, and low LDL cholesterol. In a state of high total FXR activity, SREBP-1C activity is low, HDL synthesis is repressed, and cholesterol is accumulating, producing a state of low triglycerides, VLDL cholesterol, HDL cholesterol, and high LDL cholesterol.

The FGF15/19-[FGFR4/βK]-FXR axis modulates carbohydrate metabolism

FXR agonism activates glycogenesis and inhibits gluconeogenesis (Fig. 9). FXR agonism through unknown mechanisms has been found to mimic insulin and increase insulin sensitivity by inducing the downstream phosphorylation of S473 on Protein Kinase B (Akt). In addition, the dephosphorylation of S312 and the phosphorylation of Tyr residues on Insulin Receptor Substrates 1/2 (IRS1/2) have been observed.^{170,171,172,173} FXR activated Akt in addition to [FGFR4/βK]-ERK axis activated Ribosomal-S6-kinase (RSK) phosphorylates and inhibits Glycogen Synthase Kinase 3 Beta (GSK3β).^174,175 GSK3β is a physiological inhibitor of glycogenesis, whose inhibition leads to positive glycogenic signaling.

Fig. 9

Hepatic/gut FXR signaling—carbohydrate metabolism. Hepatic and gut enteroendocrine L-cell FXR signaling pertaining to carbohydrate metabolism. FXR agonism increases glycogenesis, insulin sensitivity, represses gluconeogenesis, and blunts glycolysis. Gut FXR agonism represses GLP-1 synthesis and secretion. Dashed mechanism arrows represent the Ras-Raf-MEK signaling cascade, while dashed arrows through a protein represent the direction of substrate flux. Proteins with a yellow outline represent kinases and phosphatases, while yellow arrows represent phosphorylation/dephosphorylation. Yellow P’s represent phosphate groups. Italicized text refers to genes while normal text refers to proteins. Question marks represent unknown mechanisms. This figure was created with BioRender.com

FXR-SHP axis activation inhibits HNF4α−induced transcription of Phosphoenolpyruvate Carboxykinase (PEPCK), Fructose-1,6-Bisphosphatase (FBPase), and Glucose-6-Phosphatase (G6Pase) genes. FXR-SHP axis activation also inhibits the transcription factor Forkhead Box Protein O1 (FOXO1) and its induced expression of PEPCK and G6Pase.^176,177,178 FXR-SHP axis activation therefore has an inhibitory role on the expression of three of the four rate-limiting proteins within hepatic gluconeogenesis (G6Pase, FBPase, PEPCK). In contrast to the general trend of enhancing glucose tolerance, FXR agonism blunts both hepatic and enteroendocrine intestinal L-cell glycolytic flux. This is mechanistically mediated by FXR-SHP axis inhibition of ChREBP, repressing the transcription of Pyruvate Kinase (PK).¹⁵³ Within enteroendocrine intestinal L-cells this decreases the synthesis and secretion of Glucagon-Like Peptide-1 (GLP-1). Decreased GLP-1 synthesis is explained by reduced ChREBP-induced transcription of GLP-1’s precursor Proglucagon (GCG), while hindered GLP-1 secretion is caused by reduced glycolytic flux and hence reduced ATP-driven exocytosis.¹⁷⁹

FGF15/19-[FGFR4/βK] signaling represses hepatic gluconeogenesis

An unknown protein downstream of [FGFR4/βK] complex agonism dephosphorylates the transcription factor cyclic adenosine monophosphate (cAMP) response element binding protein (CREB). The dephosphorylation of pCREB to produce CREB reduces glucagon or adrenergic-induced gluconeogenesis in the short-term by not trans-activating cAMP response element (CRE) mediated expression of G6Pase, PEPCK, and FOXO1. The histone acetyl transferase CBP, Histone Acetyltransferase p300 (p300), and the coactivator CREB regulated transcription coactivator 2 (CRTC2) must be bound by pCREB to epigenetically activate gene expresson.^178,180 FGF15/19 serves as a long-distance signal for nutrition by further inhibiting starvation-induced gluconeogenesis, in which inactivated CREB cannot induce the expression of the FOXO1 coactivator peroxisome proliferator-activated receptor-gamma coactivator 1 alpha (PGC-1α), drastically suppressing the expression of G6Pase and PEPCK.^181,182,183

Viewing carbohydrates, the hepatic FGF15/19-[FGFR4/βK]-FXR axis centrally functions to potentiate insulin-induced glucose uptake, signals for the short-term storage of dietary carbohydrates as glycogen, serves to prevent the unneeded synthesis of carbohydrates in the presence of recently consumed nutrients, hinders glycolytic flux, and intestinally blunts incretin-like activity.

FXR modulation of amino acid metabolism

FXR agonism increases the flux of excess amino acids through the urea cycle and signals for the utilization of recently ingested amino acids within protein synthesis (Fig. 10). In mice, hepatic FXR agonism plays a wide role in sustaining the proper metabolism of amino acids. This occurs by improving glutamate shunting of ammonia via Glutamine Synthetase (GLUL) induction and flux through the urea cycle by inducing Carbamoyl Phosphate Synthetase-1 (CPS1), Arginosuccinate Synthase 1 (ASS1), and Arginosuccinate Lyase (ASL).^184,185 [FGFR4/βK]-ERK axis activation phosphorylates and activates RSK and MAP Kinases-Interacting Serine/Threonine-Protein Kinase 1 (MKNK1), resulting in the phosphorylation of Eukaryotic Translation Initiation Factors 4B (EIF4B) and 4E (EIF4E), respectively.¹⁷⁵ The phosphorylation of EIF4B and EIF4E are crucial for the initiation of eukaryotic cap-dependent translation.¹⁸⁶ RSK also phosphorylates Ribosomal Protein S6 (rpS6), a modulator of ribosomal translation which additionally increases the efficacy of cap-dependent translation.¹⁷⁵

Fig. 10

Hepatic FXR signaling—amino acid metabolism. Hepatic FXR signaling pertaining to nitrogen and protein metabolism. Hepatic FXR agonism increases urea cycle flux, glutamate shunting, and increases cap-dependent translation. Dashed mechanism arrows represent the Ras-Raf-MEK signaling cascade, while dashed arrows through a protein represent the direction of substrate flux. Proteins with a yellow outline represent kinases and phosphatases, while yellow arrows represent phosphorylation/dephosphorylation. Yellow P’s represent phosphate groups. Italicized text refers to genes while normal text refers to proteins. Question marks represent unknown mechanisms. This figure was created with BioRender.com

General mechanisms—FXR dampens hepatic and systemic inflammation

Alongside high expression within the gut and liver, FXR is also expressed within immune cells.¹⁸⁷ FXR exhibits potent anti-inflammatory activity in peripheral monocytes, in human and mouse myeloid cells, dendritic cells, and in hepatic natural killer T cells.^{188,189,190,191} FXR agonism exerts anti-inflammatory properties by physically inhibiting the formation of the NLR Family Pyrin Domain Containing 3 (NLRP3) inflammasome (Fig. 11a). This is accomplished by physically preventing the association of NLRP3 and pro-caspase-1, preventing the maturation of the pro-inflammatory cytokines Interleukin 1 Beta (IL-1β) and Interleukin 18 (IL-18).¹⁹² In addition, FXR agonism has been found to inhibit the binding of the transcription factor Nuclear Factor Kappa-Light-Chain-Enhancer of Activated B Cell (NF-κB) to cis-regulatory κB sites of pro-inflammatory genes, such as IL-1β, Interleukin 6 (IL-6), and Tumor Necrosis Factor Alpha (TNFα).^146,193 Although FXR mediates hepatic anti-inflammatory properties by the two mechanisms mentioned, it is still unknown if the total anti-inflammatory role of FXR is entirely receptor-mediated or a downstream effect of reducing hepatic lipid content.

Fig. 11

FXR signaling is anti-inflammatory, anti-apoptotic, and anti-autophagic. a FXR activity inhibits NLRP3/Caspase-1 and NF-κB induced inflammation. b Non-ligand-bound FXR is anti-apoptotic by preventing extrinsic apoptosis, while ligand-bound FXR prevents BA-overload and deoxysphingolipid induced apoptosis. c FXR activity is anti-autophagic by preventing pCREB/PPARα-mediated autophagic gene expression. d Starvation-induced AMPK enhances BA synthesis and BA mobilization to ensure adequate nutrient absorption. Proteins with a golden outline indicate kinases, while golden arrows represent phosphorylation. This figure was created with BioRender.com

FXR is anti-apoptotic and disrupts autophagy

Hepatocytes under significant cellular stress may be vulnerable to apoptosis, especially under hepatic conditions such as NAFLD or NASH.¹⁹⁴ FXR plays a significant hepatoprotective role against apoptosis in which non-ligand-bound FXR binds to the death effector domain of caspase-8, inhibiting extrinsic apoptosis (Fig. 11b).¹⁹⁵ FXR additionally prevents apoptosis by inducing Cytochrome P450 Family 4 Subfamily F (CYP4F)-mediated metabolism of pro-apoptotic deoxysphingolipids and by preventing BA-induced apoptosis by repressing CYP7A1/CYP8B1 expression.^196,197,198

Under extensive whole body starvation, autophagy recruits intracellular macromolecules to the lysosome to be broken down for energy to ensure physiological survival.^199,200 Within the liver pCREB and CRTC2 greatly contribute to the starvation-induced expression of autophagic genes, likely induced by glucagon or adrenergic signaling. Agonized FXR functionally disrupts the pCREB/CRTC2 interaction preventing autophagy, even during fasting (Fig. 11c).²⁰¹ FXR may also directly bind to autophagic gene promoters and suppress their transcriptional activity. FXR’s suppression over autophagic gene expression is nuanced since the same binding site is regulated by pro-autophagic PPARα, allowing for nutrient-sensitive control over autophagy in addition to that provided by Adenosine Monophosphate (AMP)-Activated Protein Kinase (AMPK) sensing the ATP/AMP balance and mTORC1 sensing free branched chain amino acids.²⁰² Further interplay between FXR and the starvation response has been discovered, in which activated AMPK transcriptionally represses FXR and stimulates Steroid Receptor Coactivator-2 (SRC-2) to coactivate FXR-induced expression of BSEP (Fig. 11d).^203,204 In anticipation of upcoming dietary nutrition, this allows for a transient release from BA synthesis autoregulation and increases BA mobilization into the biliary tree, cumulatively increasing the efficacy in which BA-mediated absorption replenishes current depleted nutrient stores.

Canonical BA receptor—TGR5, a metabolic and immunological modulator

Molecular biology of TGR5

TGR5 is a G-protein coupled receptor encoded by the GPBAR1 gene. The G_α subunit for TGR5 is coupled to Adenylate Cyclase (AC), designating TGR5 as a G_αs G-protein.⁸⁴ The ENSEMBL database identifies 75 sequence orthologues for TGR5, where the sequence homology of Mus musculus and Rattus norvegicus compared to the human sequence are 83.59% and 82.37%, respectively.⁹⁹ As previously stated for FXR, the sequence divergence between species may not allow for a direct correlation between animal models and results seen in human physiology. The following list ranks the top 8 baseline TGR5-expressing organs in terms of nTPM: adipose tissue, breast, gallbladder, smooth muscle, spleen, stomach, kidney, and heart. Although low in documented nTPM, it is also of use to note that TGR5 is expressed within the small intestine, colon, was recently discovered in mice hepatocytes, and in mice is exercise-inducible via the transcription factor Activating Transcription Factor 6 (ATF6) within skeletal muscle. The membrane-bound TGR5 receptor has many metabolic functions, such as increasing basal metabolic rate, whole-body nutrient tolerance, post-prandial satiety, and via Adiponectin (ADIPOQ) signaling induces insulin-sensitivity, reduces gluconeogenesis, reduces lipogenesis, and increases ceramide metabolism. Non-metabolic functions of TGR5 are to protect against hepatic pathology, reduce hepatic and adipose exposure to inflammatory signaling, and to maintain proper biliary function.²⁰⁵

Epigenetic regulation of TGR5 transcription

CpG methylation of the TGR5 promoter has been found to reduce transcription and is associated with both liver-failure and chronic hepatitis B.²⁰⁶ Hyper-methylation of the TGR5 promoter has also been found associated with hepatocellular carcinoma. Although TGR5 promoter methylation is age-related and higher in patients older than 60, methylation may serve as a potential diagnostic for both acute-on-chronic hepatitis B liver failure and hepatocellular carcinoma.²⁰⁷ Besides direct phenotypic association between hepatic pathology and CpG methylation of the TGR5 promoter, not much research has been conducted directly on the factors driving TGR5 epigenetic regulation. It is currently theorized that a significant portion of TGR5 transcriptional regulation is secondary to FXRE activity within its promoter, being directly driven by the epigenetics, post-transcriptional, and post-translational modifications modulating FXR activity.²⁰⁸ TGR5 agonists have been found to induce SIRT1 and Sirtuin 3 (SIRT3) transcription via activation of PGC-1α.²⁰⁹ TGR5-induced SIRT1 expression may further modulate FXR activity, establishing a mutual regulatory relationship.

TGR5 agonism drives energy utilization and mitochondrial biogenesis

Metabolically relevant TGR5 agonism stimulates AC to produce cAMP and activates Protein Kinase A (PKA) (Fig. 12).^{85,210,211,212} CREB is phosphorylated by PKA to produce pCREB. pCREB binds to the CRE and induces the adipocyte/skeletal muscle expression of Type 2 Iodothyronine Deiodinase 2 (DIO2) and PGC-1α.^213,214 Increased expression of DIO2 directly increases the rate of conversion of Tetraiodothyronine (T4) into Triiodothyronine (T3). When the more active form of the hormone T3 binds to the Thyroid Hormone Receptor (TR), TR acts as a transcription factor for genes with Thyroid Response Elements (TREs). In mice, TGR5-induced TR-agonism represses CYP8B1 expression without the presence of a TRE, while in humans TR-agonism induces the expression of CYP7A1.^{215,216,217,218,219} In brown adipose tissue (BADT) TR agonism induces Uncoupling Protein 1 (UCP1) which by establishing a futile cycle of mitochondrial proton shuttling increases energy usage via thermogenesis.^220,221 In white adipose tissue (WADT) TR agonism induces the expression of β1 and β2 Adrenergic Receptors (β1-AR) (β2-AR) along with UCP1. Spare adrenergic receptors increase WADT sensitivity to adrenergic-induced Hormone-sensitive Lipase (HSL) activity.^222,223,224 In skeletal muscle (SKM) Uncoupling Protein 3 (UCP3) is induced by TR agonism along with Sarcoplasmic/ER Calcium-Dependent ATPase (SERCA), resulting in energetically futile cycles of protons at the mitochondria and calcium at the sarcoplasmic reticulum.^225,226,227 In all three target tissues, TGR5 agonism increases energy consumption, upregulates mitochondrial oxidative phosphorylation, and hence increases basal metabolic rate.^213,228

Fig. 12

TGR5 signaling—metabolic effects. Metabolic effects of TGR5 signaling, crystal structure shown is PDB 7CFM.⁶³¹ TGR5 agonism induces incretin release, upregulates mitochondrial biogenesis and basal metabolic rate, and leads to ADIPOQ synthesis, increasing carbohydrate/lipid tolerance and reducing ceramide concentrations. Proteins with a yellow outline represent kinases and phosphatases, while yellow arrows represent phosphorylation/dephosphorylation. Yellow P’s represent phosphate groups. Question marks represent unknown mechanisms. Dashed mechanism arrows represent vesicle exocytosis, while dashed arrows through a protein describe the direction of ion flow. Italicized text refers to genes while normal text refers to proteins. This figure was created with BioRender.com

Following the same pattern of energy usage, PGC-1α functions as the master regulator of mitochondrial biogenesis, leads to the adipose exocytosis of ADIPOQ, and enhances adipose UCP1 expression.^229,230,231 The coactivator function of PGC-1α in adipose tissue increases the transcriptional activity of Nuclear Respiratory Factor 2 (NRF2), which induces the expression of Nuclear Respiratory Factor 1 (NRF1) by binding to the antioxidant response element (ARE).²³² NRF1 and NRF2 induce the nuclear expression of mitochondrial respiratory chain proteins.²³³ NRF1 induces the mitochondrial transcription of Transcription Factors A (TFAM), B1 (TFB1M), and B2 (TFB2M), resulting in preparation for mitochondrial biogenesis.^234,235,236 PGC-1α is additionally a coactivator for Peroxisomal Proliferator-Activated Receptor Gamma (PPARγ)/RXRα dimers. When the ternary complex binds to Peroxisomal Proliferator-Activated Receptor Response Elements (PPREs), adipocytes synthesize and secrete the adipokine ADIPOQ.^230,231,237 Systemically, Adiponectin Receptor 1 (AdipoR1) agonism increases whole-body insulin sensitivity by activating Adaptor Protein, Phosphotyrosine Interacting with PH Domain and Leucine Zipper 1 (APPL1), promoting the binding of IRS1/2 to the insulin receptor. In addition, AdipoR1 agonism activates Liver Kinase B1 (LKB1) leading to the phosphorylation and activation of AMPK, which represses hepatic gluconeogenesis, lipogenesis, and cholesterogenesis.^{238,239,240,241} Adiponectin Receptor 2 (AdipoR2) agonism increases β-oxidation of fatty acids within the liver and adipose tissue by inducing PPARα.^242,243 Agonism of both AdipoR1/R2 activates receptor intrinsic ceramidase activity, protecting against the negative metabolic side effects of ceramides.^{244,245,246,247,248}

TGR5 agonism drives incretin release

Within pancreatic β-islets, α-islets, and enteroendocrine L-cells, TGR5 agonism leads to calcium-induced exocytosis by two mechanisms. The first of which involves cAMP activation of Exchange Protein Activated by cAMP (EPAC), the recruitment and activation of Receptor-Associated Protein (RAP), Phospholipase C (PLC) mediated hydrolysis of Phosphoinositide (PIP₂) to Inositol Triphosphate (IP₃) / Diacylglycerol (DAG), and ER calcium release.^249,250 Specific to the pancreas, EPAC additionally stimulates Rab-interacting Molecule 2 (RIM2) and Piccolo to increase calcium-induced exocytotic granule trafficking and fusion.^251,252,253 The second mechanism involves PKA-mediated phosphorylation of Sulfonylurea Receptor 1 (SUR1), inhibition of ATP-Sensitive Potassium Channels (K_ATP), membrane depolarization, Voltage-gated Calcium Channel (VGCC) activation, and calcium-induced calcium release from the ER.^{254,255,256,257} In β-islets, this leads to the exocytosis of insulin.^249,254,258 In enteroendocrine intestinal L-cells this leads to the exocytosis of GLP-1 and Peptide YY (PYY), resulting in enhanced insulin secretion, reduced rates of gastric emptying, and appetite suppression.^{259,260,261,262,263} By the same two mechanisms pancreatic α-islets secrete GLP-1 instead of glucagon. This transition is explained by α-islet-specific EPAC activity, which induces the expression of proprotein convertase 1 (PC1), biasing alternative splicing of proglucagon mRNA to favor the GLP-1 transcript instead of the glucagon transcript.²⁴⁹ Cumulatively under a metabolic lens, TGR5 agonism enhances basal metabolic rate by increasing thermogenic futile cycling, lipolysis, and mitochondrial biogenesis, while leading to increased nutrient tolerance, insulin sensitivity, satiety, and ceramide metabolism.

TGR5 promotes hepatic mitochondrial function, vasodilation, and biliary flow

TGR5-induced hepatic endothelial cAMP synthesis results in EPAC/RAP-mediated stimulation of Phosphoinositide 3-Kinase (PI3K) and phosphorylation of PIP₂ into phosphatidylinositol trisphosphate (PIP₃) (Fig. 13).²⁶⁴ PIP₃ synthesis activates Pyruvate Dehydrogenase Kinase 1 (PDK1) leading to the phosphorylation of Akt at T308. Two models currently exist, in which either PIP₃ or pT308-Akt activate Mammalian Target of Rapamycin Complex 2 (mTORC2), resulting in the phosphorylation of Akt at S473.²⁶⁵ pT308/pS473-Akt along with PKA directly phosphorylate and activate Endothelial Nitric Oxide Synthase (eNOS) at S1117, maintaining hepatic endothelial mitochondrial function and vasodilation.^266,267,268 TGR5-induced cAMP-based agonism provides additional benefits by activating Cystic Fibrosis Transmembrane Conductance Regulator (CFTR). CFTR agonism leads to chloride secretion and soon after bicarbonate exchange by Anion Exchange Protein 2 (AE2) into the bile duct, maintaining bile flow and protecting against cholangitis/cirrhosis.^{269,270,271,272,273}

Fig. 13

TGR5 signaling—non-metabolic and immunologic effects. Non-metabolic and immunologic effects of TGR5 signaling, crystal structure shown is PDB 7CFM. TGR5 agonism represses macrophage/adipocyte-specific inflammation, enhances hepatic vasodilation, biliary flow, represses pro-inflammatory NF-κB activity, and induces IL-10/TGFβ expression. Proteins with a yellow outline represent kinases and phosphatases, while yellow arrows represent phosphorylation/dephosphorylation. Yellow P’s represent phosphate groups. Question marks represent unknown mechanisms. Dashed mechanism arrows represent vesicle exocytosis, while dashed arrows through a protein describe the direction of ion flow. Italicized text refers to genes while normal text refers to proteins. This figure was created with BioRender.com

General mechanisms—TGR5 exerts potent anti-inflammatory activity

TGR5 is extensively expressed within cells derived from the myeloid lineage and is anti-inflammatory.²⁷⁴ Immunologically, TGR5-induced pT308/pS473-Akt activates mTORC1, resulting in the differential translation of CCAAT/enhancer-binding protein β (CEBP/β) mRNA to produce Liver-Enriched Inhibitory Protein (LIP).^275,276,277 LIP production reduces the expression of pro-inflammatory TNFα and IL-6, reducing both macrophage migration and insulin resistance-associated inflammation within adipose tissue.²⁷⁸ TGR5 agonism inhibits NF-κB pathway activation within macrophages, Kupffer cells, and hepatocytes, reducing the expression of the following cytokines: Interferon Gama (IFNγ), Interleukin-1β (IL-1β), IL-2, IL-6, and TNFα.^279,280,281 Mechanistically this is accomplished by inhibiting the phosphorylation of NF-κB Inhibitor Alpha (IκBα) and inhibiting the translocation of the NF-κB transcription complex p50 and p65 (p50/p65). TGR5 activation recruits β-Arrestin 2 and inhibits the phosphorylation and proteasomal degradation of IκBα.²⁸² TGR5 provoked pCREB activation induces the expression of the transcription factor c-Fos (c-Fos), which when phosphorylated by PKA prevents p50/p65 from binding to κB sites.²⁸³ In addition, pCREB activation induces the expression of Interleukin 10 (IL-10), and Transforming Growth Factor Beta (TGFβ).^284,285 Further reductions in pro-inflammatory cytokines are seen by TGR5-AC-PKA axis activity, leading to the ubiquitination and destruction of the NLRP3 inflammasome.²⁸⁶ NLRP3 cessation directly leads to decreased concentrations of macrophage-relevant pro-inflammatory cytokines IL-1β and IL-18.^287,288 Phenotypically, TGR5 agonism has been found to switch macrophages away from the pro-inflammatory M1 phenotype towards the anti-inflammatory M2 phenotype.^{278,284,289,290,291}

Non-canonical BA receptors

Sphingosine-1-phosphate receptor 2

Sphingosine-1-phosphate Receptor 2 (S1PR2) is a G_i coupled G-protein coupled receptor commonly expressed within hepatocytes, cholangiocytes, stellate cells, intestinal epithelial cells, macrophages, and is directly agonized by conjugated BAs.^{292,293,294,295} Downstream of such S1PR2 induces the expression of Sphingosine Kinase 2 (SphK2). Sphk2 converts sphingosine to Sphingosine-1-phosphate (S1P), inhibiting HDAC-1/2, and enhancing the expression of genes associated with hepatic glucose and lipid metabolism such as: SREBP-1C, FAS, Carnitine Palmitoyl Transferase 1A (CPT-1A), PPARγ, PGC-1α, ApoB100, and FXR. Mice S1PR2^−/− and SphK2^−/− models found that the lack of S1PR2-mediated induction of ApoB100 and CPT-1A directly enhances the extent of hepatic steatosis, indicating that S1PR2 plays an important modulatory role in NAFLD pathogenesis.²⁹⁶ In addition, S1PR2 may activate ERK, Akt, and c-Jun N-terminal Kinase (JNK) signaling cascades, if bound to glycine conjugated CDCA may induce apoptosis, and stimulates the pro-inflammatory M1 polarization of macrophages via the Gi-PI3K-JNK axis.^{293,294,297,298,299} Inhibition of S1PR2 activity was found to decrease total serum BAs, reduce cholestatic injury in mice bile duct ligation mice models, restore intestinal mucosal barrier function, and reverse pro-inflammatory M1 macrophage polarization in mice Dextran Sulfate Sodium (DSS)-colitis models.^300,301

Muscarinic receptors 2 & 3

Muscarinic receptors 2 (CHRM2) and 3 (CHRM3) are G_i and G_q coupled G-protein coupled receptors, respectively, and both may be agonized by taurine conjugated BAs.³⁰² CHRM2 and CHRM3 overexpression is associated with the initiation and progression of colon, gastric, and pancreatic cancers.^{303,304,305,306} In addition, taurine conjugated LCA has been found to induce cholangiocarcinoma as a downstream result of muscarinic agonism.³⁰⁷

Vitamin D receptor

The vitamin D receptor (VDR) is a nuclear transcription factor that dimerizes with RXRα, is canonically agonized by vitamin D3, and is expressed extensively within the bones, intestine, kidney, and liver.^308,309,310 As time progressed it was discovered that VDR may be agonized by LCA, serving as an intestinal BA receptor.³¹¹ LCA agonized VDR suppresses CYP7A1 expression by binding to HNF4α and preventing its association with CYP7A1 or by inhibiting its interaction with COUP-TFII, CBP, TFIID, and TFIIB.^{136,139,140,141,312} LCA agonized VDR additionally induces the expression of detoxifying metabolic enzymes such as CYP3A4, SULT2A1, MRP3, and ASBT, which cumulatively enhance the metabolism and excretion of BA acids.^{313,314,315,316} VDR agonism suppresses FXR-SHP-axis activity independent of its DNA binding domain, preventing FXR from transactivating FXRE-containing promoters.³¹⁷ In addition, VDR agonism transcriptionally represses SHP.^318,319 Lastly, VDR expression and polymorphism within VDR has been found to significantly affect microbial beta diversity, directly modulating the efficacy of secondary BA synthesis.^320,321 A lack of VDR agonism has been associated with decreased Lactobacillus presence, and enhanced Clostridium and Bacteroidetes presence.

Pregnane X receptor

Pregnane X receptor (PXR) is a RXRα heterodimeric transcription factor that is largely expressed within the liver and gut.³²² PXR is commonly referred to as the xenobiotic sensor since it regulates the expression of many classes of enzymes associated with xenobiotic metabolism.³²³ Upon binding to a xenobiotic agonist, LCA, or 3-oxo-LCA, PXR induces the expression of Cytochrome P450 3 A, 2B, and 2 C subfamilies of phase I metabolic enzymes, phase II enzymes such as glutathione S-transferases, UGTs, and SULTs, and phase III enzymes such as Multidrug Resistance 1 (MDR1), MRP2, and Organic Anion Transporting Polypeptide 2 (OATP2). PXR therefore has a direct role in preventing lipophilic BA-mediated toxicity, such as that which occurs with high LCA concentrations.^324,325,326 Activated PXR sequesters HNF4α, preventing it from interacting with PGC-1α, transcriptionally repressing CYP7A1.^327,328 PXR agonism enhances de novo lipogenesis by enhancing expression of SCD1 and various long fatty acid elongases, in addition to sequestering the transcription factor Forkhead Box Protein A2 (FOXA2) and preventing the expression of CPT-1A.^329,330 Lastly, agonized PXR may directly dimerize with pCREB, reducing its ability to induce gluconeogenesis.³³¹

Constitutive androstane receptor

The hepatically expressed RXRα heterodimeric transcription factor constitutive androstane receptor (CAR) has high homology to that of PXR. Different from that of PXR, CAR is constitutively active, where androstane metabolites were found to be inverse agonists.^332,333 Analogous to PXR, CAR induces the expression of enzymes responsible for metabolizing BAs, but it is unsure which BAs modulate CAR transcriptional regulation and to what extent.^323,334,335

Liver X receptors

The LXRs are heterodimer transcription factors with RXRα that are endogenously agonized by oxysterols.³³⁶ Two genes encode the highly homologous LXR family, NR1H3 for LXRα and NR1H2 for LXRβ.³³⁷ LXRα is found extensively within the liver, adipose tissue, the gut, and macrophages, while LXRβ is expressed ubiquitously.³³⁸ In rats, LXR activation serves as a sensor for cholesterol over-abundance, enhancing the transcription of CYP7A1, ABCG5, and ABCG8 to increase cholesterol excretion and decrease its absorption.^339,340 In humans, LXR activation has similar effects on ABCG5 and ABCG5 but reduces BA synthesis by inducing SHP transcription.^341,342,343 HDCA has been found to be a weak LXRα agonist.³⁴⁴

Peroxisomal proliferator-activated receptor alpha

PPARα is a transcription factor that under starved conditions and adipose tissue lipolysis induces fatty acid uptake and catabolism.^345,346 HDCA has been found to directly displace Chromosomal Region Maintenance 1 (CRM1) from the Ran GTPase (Ran)/CRM1/PPARα cytosolic shuttling heterotrimer, allowing for PPARα to locate to the nucleus and modulate the transcription of lipid-catabolic genes.³⁴⁷ PPARα agonism has been found to suppress expression of NTCP, OATP1, and upregulate BSEP/MRP3/MRP4 expression, cumulatively enhancing BA efflux from the liver.^348,349 In addition PPARα reduces NLRP3 inflammasome activation, caspase-1 cleavage, and downstream maturation of IL-1β by inducing the transcription of lncRNA Gm15441.³⁵⁰

Ceramides, a unifying hypothesis against BA-induced cardiometabolic disease

Abnormal biochemical or physiological changes to the BA pool composition directly alter the ratio between ceramide synthetic gut FXR agonism and ceramide catabolic systemic TGR5 agonism. Resulting increases in ceramide synthesis are directly deterministic for the pathogenesis of cardiometabolic diseases. Ceramides negatively impact the liver by inducing lipogenesis and gluconeogenesis (Fig. 14). Hepatic uptake of gut FXR-induced ceramides has been found to lead to hepatocyte ER stress, the activation of JNK, the phosphorylation of the transcription factor c-Jun, and the inhibition of HNF1α, a key transcription factor responsible for the transcription of the FXR gene. This chain of events prevents proper FXR-mediated SREBP-1C repression, leading to increases in lipogenesis and overall promotion of the steatotic phenotype of NAFLD/NASH.^124,691,692 Similar to other cases of chemotherapeutic resistance, mutations have been found in FGFR4 such as FGFR4-V550L or FGFR4-V550M which limit the design and applicability of current FGFR4 antagonists. In line with parallel research on FGFR2, PROTAC-mediated FGFR4 degradation may allow for one to target regions of the receptor not intrinsically associated with inhibition, providing more opportunities against mutation.^693,694,695

BA sequestrants increase TGR5 agonism and GLP-1 release

Besides centrally acting targets, effort has been put into finding gut-restricted non-BA-receptor targets. The first of which is a repurposing of the already FDA-approved BA sequestrant anionic exchange resins: colesevelam, cholestyramine, and colestipol. BA sequestrants prevent the enteric reabsorption of BAs, increasing the luminal concentration of BAs present in the distal gastrointestinal tract. BA sequestrants increase the fecal excretion of BAs, lowering systemic LDL cholesterol by shunting it into BA synthesis. This leads to increased BA exposure and agonism of enteroendocrine L-cell TGR5, enhancing systemic GLP-1 secretion in mice.⁶⁹⁶ The enhanced GLP-1 secretion was correlated with improved insulin sensitivity and suppressed hepatic glycogenolysis in a DIO rat model.^697,698 Alongside DIO rat models, this GLP-1 secretory outcome was found and validated in a systemic review of phase II human trials, serving as evidence for the potential usage of BA sequestrants not just for hypercholesterolemia but as an insulin-sensitizing and anorexic therapeutic.^699,700,701

ASBT inhibitors increase BA synthesis and reduce cholestatic risk

Analogous to the role of BA sequestrants, ASBT inhibitors prevent the ileal reuptake of BAs. Decreased ileal reuptake of BAs reduces hepatic FXR agonism and enhances hepatic BA synthesis and flux through the biliary tree, lowering serum cholesterol and cholestatic risk.⁷⁰² Preliminary results in MRP2^−/− mice models have indicated that ASBT inhibitors have a functional role in the treatment of PSC, inhibiting inflammatory/profibrotic gene expression.⁷⁰³ The ASBT inhibitor A4250 also known as Odevixibat (Alibero Pharma) was already FDA-approved as a first-in-class drug used to treat familial intrahepatic cholestasis, while the second compound SC-435 (University of Connecticut) has been used purely as a test compound.^704,705,706 Alongside these current compounds, two new ASBT inhibitors are in human clinical trials, volixibat (Mirium Pharmaceuticals) and maralixibat (Mirium Pharmaceuticals). Maralixibat is currently approved for Alagille syndrome and recently completed the phase 2b/3 MARCH trial for progressive familial intrahepatic cholestasis, showing positive safety and efficacy data.⁷⁰⁷ Volixibat is currently in in two phase 2b/3b trials, VISTAS for PSC, and VANTAGE for PBC.^708,709 Both of which are likely to be approved by 2027-2028.

Microbial modification changes BA pool composition

The last gut-restricted non-BA-receptor target is that of the gut microbiome. Modification of the gut microbiome directly impacts both the primary to secondary and conjugated to unconjugated ratios within the BA pool.^710,711 Current efforts are to either pharmacologically reduce BSH-containing bacterial or administer probiotic/prebiotic mixes to enrich for BSH-containing bacteria.⁷¹² An example of the first technique is exemplified with metformin, which by unknown mechanisms decreases Bacteroides fragilis, which in mice increased gut UDCA levels and provided the positive benefit of reducing ceramide synthesis.⁷¹³ The alternative approach is highlighted by the probiotic supplement VSL#3 (VSL3 Pharma), which has been shown to induce BA deconjugation and maturation into secondary BAs, induce BA synthesis, enhance insulin sensitivity by driving BA pool composition to favor TGR5 agonism, and protect mice from NASH.^714,715 Alongside BA pool composition changes it is theorized that a higher prevalence of BSH-containing bacteria reduces the surfactant efficacy of gut BAs, reducing systemic LDL cholesterol by inhibiting cholesterol absorption and increasing cholesterol excretion as BAs.⁵⁴

Conclusions and future perspectives

Continuing with BAs—recommendations for future success

It is essential to acknowledge the pharmacological complexity of BA receptor modulators, exemplified by OCA’s capacity to impact numerous physiological circuits and induce severe side effects. Messy pharmacology, low bioavailability, dose-dependent cytotoxicity, and non-linear pharmacokinetics have plagued BA-centric drug discovery. However, we believe that dismissing BA research prematurely is unwarranted, since significant progress has been made addressing these concerns and unlocking new potential for BA receptor modulators.

Messy receptor pharmacology has been substantially improved by develo** tissue-specific drugs. Many of the original side effects of OCA have been directly remedied, or at least mechanistically understood to pave the way for future innovation. For instance, hepatic FXR-mediated IL-31 pruritis has been strongly mitigated by the usage of gut-restricted FXR agonists. Although challenges such as a pro-atherogenic serum lipid profile persist, emerging strategies such as SFXRMs, the discovery and usage of new non-steroidal scaffolds, and scaffold-specific structure activity relationships have been found to meaningfully reduce adverse effect incidence and severity. Similarly, positive developments have been seen in the case of TGR5 agonists, in which the initial alarming concern about gallbladder non-contractility has been solved by the implementation of gut-restricted TGR5 agonists. Despite the intrinsic complications of modulating FXR and TGR5, tremendous preclinical experimentation and clinical validation has given us the tools to remove less favorable outcomes, maximizing the chance of therapeutic success (Table 3).

Table 3 Approved or candidate BA-centric drugs and their clinical trials

Improving low oral bioavailability, dose-dependent cytotoxicity, and non-linear pharmacokinetics have been the strongest forces behind non-steroidal scaffold research. Besides the negative outcomes of WAY-362450 for FXR and SB-756050 for TGR5, most if not all non-steroidal scaffolds have shown dramatically improved pharmacokinetic parameters. Diverging from endogenous BAs, non-steroidal or heavily modified steroidal scaffolds provide a refined approach for clinical development, removing most of the amphipathic surfactant-like effects that trigger dose-dependent cytotoxicity.

Recent advancements in both tissue-targeted and structurally diverse approaches have provided ample opportunities to wrangle the originally untamable beast of BA physiology. Combining this with the incredibly important role of BA physiology in the pathogenesis of cardiometabolic, inflammatory, and neoplastic diseases, there is immense clinical potential and a high likelihood for the future success of BA-centric therapeutics.

How may BA therapeutics be better or worse than the existing therapeutic paradigm?

The current landscape of T2DM treatment predominantly revolves around generating insulin analogs or potentiating endogenous insulin signaling. Besides recent developments in GLP-1 receptor agonists and Sodium-Glucose Cotransporter-2 (SGLT2) inhibitors, existing therapies either require injections or have negative side effects such as weight gain. Progress has been made against obesity, but approved therapeutics only focus on hunger repression or inhibiting the dietary uptake of lipids. While the potential approval of a Thyroid Receptor Beta (THβ) agonist such as resmetirom (Madrigal Pharmaceuticals) may offer promise as a first NAFLD/NASH drug, there are currently no FDA-approved drugs for these conditions.⁷¹⁶ ASCVD lacks a cure, relying on slowing the rate of the inevitable by pharmacologically lowering LDL cholesterol. In the realm of IBD, there is no curative treatment, only the usage of aminosalicylate, corticosteroid, or biologic therapy to reduce intestinal inflammation. Hepatocellular carcinoma chemotherapy resistance employs small molecule antagonists against FGFR4, but most of these approaches are strongly impacted by patient-specific mutations.

Contrasting with the status-quo, BA-centric approaches offer some distinct advantages over the traditional approaches. BA therapeutics offer significant benefits for T2DM, obesity, and NAFLD/NASH treatment relative to existing therapeutics, inhibiting ceramide synthesis, strongly modulating hepatic metabolism, and mirroring the existing therapeutic viability of GLP-1 receptor pharmacology. While BA therapeutics may not surpass the immense benefit of statin therapy, their ability to prevent ceramide synthesis is likely to strongly impact the inflammatory aspects of ASCVD disease progression. IBD management with BA therapeutics has a high benefit compared to the traditional approach directly targeting the Th17 and Treg centric pathway that drives disease pathogenesis. Lastly, a more advanced BA therapeutic approach such as PROTAC development may provide more profound FGFR4 suppression, better outcomes, and a lower likelihood for patient-specific resistance.

Navigating preclinical to clinical translational challenges

BA research encounters many challenges that prevent the direct translation of preclinical findings into clinical applications. There are four major translational challenges seen between preclinical research and clinical outcomes that must be minimized to ensure translational success. The first challenge arises from the substantial differences in lipid metabolism between rodents and humans, potentially biasing both cholesterol availability for BA synthesis and interpretability of lipid-centric outcomes. Rodents, unlike humans, predominantly store cholesterol within HDL, intrinsically having a lower risk for ASCVD before humanization techniques are applied.⁷¹⁷ Addressing this, known solutions involve the development of rodent genetic variants or human hepatocyte xenotransplantation models that express lipid profiles very similar to that of humans.

Similarly, the second translational challenge stems from species-specific differences between the BA pools of rodents and humans. For example, rodents produce large quantities of FXR-antagonistic MCAs while humans do not. Although currently unresolved, genetic ablation of CYP2C70 with or without CYP2A12 has been found to produce a BA pool more akin to that of humans.^{22,25,718,719,720} Further genetic and biochemical research is crucial for refining the BA pools of rodents to more similarly match that of humans.

The third translational challenge revolves around species-specific differences in the coding sequence of BA receptors. Beyond the impacts of varying the composition of a single BA within the BA pool, small differences in the coding sequence of BA receptors strongly modulate the potency and efficacy of all known receptor agonists/antagonists. A prime example of such are MCBAs, whose modulatory activity on canonical BA scaffold receptor pharmacology in some cases is completely different species-to-species. To further highlight the issue, many reports have shown that species-specific differences to the BA receptor coding sequence can cause considerable shifts up to a half-order in magnitude in assay EC₅₀/IC₅₀ results when comparing human and rodent models.^654,721 To address this challenge, the most straightforward solution may be to humanize rodent BA receptor coding sequences to match those of humans.

The final translational challenge in BA research relates to intrinsic species-specific differences in signaling cascades, in which post-translational protein modifications, transcription factor cis-elements, and their biological outcomes may differ from species to species.⁷²² While altering the substrate selectivity and enzymatic activity of an entire signaling cascade proves challenging due to the multitude of targets, adding, removing, or changing the identity of cis-element sequences emerges as a viable option.

Future directions for BA-centric research

BA-centric research has developed tremendously since its inception, but as more questions are answered many new questions are asked. From the perspective of biochemistry many questions need to be answered pertaining to BA synthesis and maturation. It is crucial to reveal what factors drive BA synthesis between the classical and alternative pathway of synthesis, highlighted in importance by the abnormal 12α-OH BA levels seen in T2DM, obese, and NAFLD/NASH patients. From such, more research needs to be done on the mechanisms connecting TGR5 and CYP8B1, especially how this may be modulated during insulin resistance.

From the perspective of physiology, perhaps the most imperative question to further investigate is the effect of gut FXR-induced ceramides on disease pathogenesis. Alongside such, more work is needed to pinpoint human organ-specific BA pool compositions, which in tandem with a better understanding of the gut/liver FGF15/19-[FGFR4/βK]-FXR cascades would prove incredibly important. The impacts of FXR and TGR5 within mucosal immunity also warrant further investigation, in which there needs to be a better consensus on how these two signaling pathways regulate Treg and Th17 differentiation. The discovery of non-canonical BA receptors opens many questions, importantly opening the door for new modulatory points within the FXR/TGR5 signaling cascades and BA physiology as a whole. Lastly, more research is needed to investigate the physiological purpose and functions of MCBAs.

BA pharmacology is evolving rapidly. Gut restricted receptor modulators allow for potent pharmacological effects without systemic side effects that almost discarded the therapeutic field as a whole. Further research is needed to develop additional ways to circumvent the traditional pitfalls of BA receptor modulators. SFXRMs should be thoroughly investigated, alongside new design strategies for gut-restriction, non-steroidal scaffolds, tissue-specific formulations, and strongly said it is imperative to see if the potent preclinical effects of gut-restricted FXR agonists/antagonists and TGR5 agonists translate to humans.

Don’t give up on BAs

If developed correctly, systemic and gut-restricted FXR/TGR5 modulators, upstream regulators of BA signaling cascades, proteins that control BA pool size/composition, and microbiome-targeted therapies provide very attractive pharmacological targets for drug discovery against cardiometabolic, inflammatory, and neoplastic diseases. BAs are widespread physiological regulators, modulating carbohydrate, lipid, protein, inflammatory, and basal metabolic pathways. Abnormal BA pool size and/or composition can physiologically predispose for cardiometabolic disease, notably by upregulating the synthesis of ceramides. Abnormal BA pool size or composition additionally influences the ensemble biochemical properties of the BA pool, such as surfactant efficacy and relative agonism between FXR/TGR5.

There is a dire need for new innovations in drug discovery to more effectively treat cardiometabolic, inflammatory, and neoplastic diseases. In this review, we provided a bottom-up approach on BAs, explaining their biochemistry, physiology, and pharmacology at canonical and non-canonical receptors (Fig. 18). Using such, we thoroughly dissected how abnormal BA physiology directly induces disease pathogenesis and in the case of cardiometabolic disease is highly driven by gut-synthesized ceramides. Lastly, we rationalized novel targets for further translational drug discovery and provided future perspectives. Cardiometabolic, inflammatory, and neoplastic diseases are hard to drug due to complex biochemistry, the lack of deterministic models for disease pathogenesis, and have no well-defined curative treatments, leaving a large gap in the market for new therapies to ease the burden and treat unmet needs. BAs and their associated signaling cascades are crucial for proper health and homeostasis. Although BA therapeutics had a tumultuous start, they have now been refined to reveal a high therapeutic potential, simply said, don’t give up on BAs.

Fig. 18

Summary—BA signaling in health and disease. BAs and their associated signaling pathways are central regulators of health. Their biochemistry, physiology, and pharmacology can protect against or if aberrant induce the pathogenesis of disease. BA therapeutics offer immense clinical potential, which if further developed may provide novel treatments for multiple high morbidity/mortality conditions. This figure was created with BioRender.com