Bile acid metabolism and signaling: emerging pharmacological targets of dietary polyphenols

Abstract

Beyond their role as emulsifiers of lipophilic compounds, bile acids (BAs) are signaling endocrine molecules that show differential affinity and specificity for a variety of canonical and non-canonical BA receptors. Primary BAs (PBAs) are synthesized in the liver while secondary BAs (SBAs) are gut microbial metabolites of PBA species. PBAs and SBAs signal to BA receptors that regulate downstream pathways of inflammation and energy metabolism. Dysregulation of BA metabolism or signaling has emerged as a feature of chronic disease. Dietary polyphenols are non-nutritive plant-derived compounds associated with decreased risk of metabolic syndrome, type-2 diabetes, hepatobiliary and cardiovascular disease. Evidence suggests that the health promoting effects of dietary polyphenols are linked to their ability to alter the gut microbial community, the BA pool, and BA signaling. In this review we provide an overview of BA metabolism and summarize studies that link the cardiometabolic improvements of dietary polyphenols to their modulation of BA metabolism and signaling pathways, and the gut microbiota. Finally, we discuss approaches and challenges in deciphering cause-effect relationships between dietary polyphenols, BAs, and gut microbes.

1. Introduction

1.1. Primary BA synthesis and hepatic modifications

Bile acids (BAs) are a heterogeneous group of amphipathic steroidal acids that facilitate emulsification and absorption of dietary fats and other lipophilic compounds (1). BAs signal to nuclear and membrane BA receptors, which regulate immune responses, intestinal permeability, glucose and energy metabolism, and liver BA synthesis (2–8). Greater understanding of BA signaling pathways may provide insights into molecular targets relevant to development of chronic metabolic diseases, including metabolic syndrome (MetS), Type-2 diabetes (T2D), and cardiovascular diseases (CVD).

Serum BA concentrations increase after a meal fluctuating between 2 – 8 μM (9) in healthy human subjects; however, serum levels of BAs and BA precursors may fluctuate differentially in healthy humans compared to patients with obesity, diabetes, or liver dysfunction (10–12). Fasting serum levels of the BA precursor 7α-hydroxy-4-cholesten-3-one were reported to be higher in patients with T2D and MetS than in healthy controls, but median total BA concentrations were not significantly different (10). In T2D patients, postprandial total BA levels were higher than healthy controls after consumption of a high-fat meal and this increase was due mainly to glycine and taurine conjugated BAs (11). A systematic review of 28 clinical studies including subjects with liver diseases indicated that there are no specific BAs or BA ratios that can serve as a reliable biomarker for liver dysfunction, except in patients with intrahepatic cholestasis of pregnancy (12). Moreover, evidence indicates that biological sex and age may also have a profound impact circulating/postprandial BA in healthy and obese individuals (13, 14).

Hepatic cytochrome P450 (CYP) enzymes play an integral role in xenobiotic metabolism of drugs, toxins, and carcinogens, as well as in BA synthesis and steroid hormone metabolism (15). In the liver, CYP family enzymes synthesize primary BAs (PBAs) from cholesterol via the classical or alternative pathways (16, 17) in liver of mice and humans (Figure 1). The classical or neutral BA pathway is responsible for 90% of PBA production, while 10% is synthesized via the alternative or acidic pathway, so called due to acidic intermediate products (18). PBA synthesis requires 17 enzymes across intracellular compartments including the cytosol, endoplasmic reticulum, microsome, and peroxisome (17). The classical pathway is initiated by hydroxylation of cholesterol via cholesterol-7α-hydroxylase (CYP7A1), which is the rate-limiting enzyme for all PBA production, after which sterol 12α-hydroxylase (CYP8B1) produces 7-alpha,12-alpha-dihydroxy-4-cholesten-3-one from 7 alpha-hydroxy-4-cholesten-3-one, an intermediate product that determines the levels of cholic acid (CA) and chenodeoxycholic acid (CDCA) produced (19, 20). Sterol 27-hydroxylase (CYP27A1) is the first enzyme of the alternative pathway, but also plays a role in the classical pathway. In the classical pathway, CYP27A1 hydroxylates sterol intermediates produced by CYP8B1 (21). In the alternative pathway, CYP27A1 initiates BA synthesis via oxidation of cholesterol to 27-hydroxycholesterol (15). 25- and 27-hydroxycholesterol is hydroxylated by oxysterol 7-alpha-hydroxylase (CYP7B1), which leads to the formation of CDCA in humans (21). The alternative pathway can account for more than 10% of the BA pool under conditions where CYP7A1 activity is deficient (22). Cholesterol conversion and BA synthesis may be initiated in other organs as well. Sterol 24-hydroxylase (CYP46A1), which is mainly expressed in the brain, converts cholesterol to 24-S-hydroxycholesterol (23, 24), which is then subsequently hydroxylated by hepatic sterol 7α-hydroxylase (CYP39A1) to form the PBA, CDCA (23–25). In humans, 0.2 – 0.6 mg of BA are synthesized daily to replace similar levels of BAs lost through fecal and urine excretion (26).

Figure 1.

Primary bile acid (PBA) synthesis of unconjugated PBA via classical and alternative pathways in mice versus humans. Bile acid precursor (24-(S)-hydroxycholesterol) derived from modification of cholesterol via CYP46A1 in brain and hepatic CYP39A1 to produce CDCA. Unconjugated PBA are conjugated with glycine (G) or taurine (T) via reactions catalyzed by bile acid-CoA synthetase (BACS) and bile acid-CoA:amino acid N -acyltransferase (BAAT). Conjugated PBA mainly detected in mice (taurine; T) and human (glycine; G) are bolded. Image created with Biorender.com.

BA glucuronidation occurs by UDP-glucuronosyltransferase (UGT) (27) and sulfonation via sulfo-aminotransferases (SULTs) (20, 28). About 70% of BA in urine are sulfonated (28). BA-CoA:amino acid N-acyltransferase (BAAT) can conjugate with both glycine and taurine (29). PBA conjugation with taurine or glycine occurs via a two-step process; first BA–Coenzyme A synthase (BACS) generates BA-CoA then BAAT conjugates BA-CoAs with taurine (predominant in rodents) or glycine (predominant in humans) (29–31). BA conjugation with taurine, glycine, or sulfate groups regulates their toxicity, decreases intestinal re-absorption, enhances solubility, promotes excretion, and alters affinity to BA receptors (32–34).

Differences in BA synthesis between humans and murine species

BA composition differs between humans and murine species as summarized in Figure 1 (35). CA is the major PBA in rodents while CDCA is the predominant PBA in humans (19, 20). Hyocholic acid (HCA) also known as gamma-muricholic acid (γMCA), is considered a minor PBA in humans, synthesized from CDCA by enzyme CYP3A4 (36), making up ~3% of the BA pool (37). CDCA is also reported to be converted to HCA in mice (38); however, is more commonly considered a bacterially-derived BA in rodents (39).

Other PBAs of physiological importance in mice include α-muricholic acid (αMCA) and βMCA, which are derived from oxidation of CDCA and ursodeoxycholic acid (UDCA) by Cyp2c70 in mice (40, 41) and by Cyp2c22 in rats (40, 41). UDCA is another BA which is considered primary in mice, but not humans (42). Interestingly, deletion of hepatic Cyp2c70 in mice promoted a hepatic BA profile similar to humans (43). In mice, CDCA is converted to αMCA and βMCA, with high efficiency (43, 44), however, humans do not express Cyp2c70 and so these MCAs are are not considered to be synthesized in humans (45), though have been detected in human urine in some studies (46, 47). No unconjugated α- or β-MCA were detected in healthy human serum or liver and only minute levels of TαMCA (2.9 ± 2.7 nM) were detected in serum (47). Overall, α- or β-MCA made up < 0.2% of the total BA pool of NASH and non-NASH patients, with significantly less detected in serum NASH patients (48), but more in urine of cholestatic adults (7.2% of total BAs) (49). Greater concentrations of conjugated and unconjugated α- or β-MCA are detected in neonates (~1 μM) (50), suggesting an age-dependent effect on this BA species. Despite their detection in circulation and urine, origins of α- or β-MCA are obscure in humans, and it is not known whether they are primary or bacterially derived in nature.

1.2. Secondary BA synthesis and bidirectional BA-gut microbial dynamics

Once conjugated, PBA enter the gastrointestinal tract (GIT) where they may be metabolized to secondary BAs (SBAs) by bacteria encoding deconjugating, dehydroxylating, epimerizing, and reducing enzymes (42) (Figure 2). Prerequisite to SBA formation is deconjugation of PBA taurine or glycine moieties (e.g., TCA →CA) by microbial bile salt hydrolases (BSH) (51). Gut microbial 7α-dehydroxylases (7αDHs), 7βDHs, reductases, hydroxysteroid dehydrogenases (HSDH), and epimerases (52) subsequently modify unconjugated PBAs to produce diverse SBA species.

Figure 2.

Host bile acid conjugation of primary bile acid (1°) or reabsorbed secondary bile acids (2°) with taurine (T; mainly in rodents) and glycine (G; mainly in humans) in the liver, and metabolism (I.e., deconjugation, dehydroxylation, desulfation, and epimerization) by microbial enzymes, bile salt hydrolase, 7α-, 7β-hydrolase, reductase, hydroxysteroid dehydrogenase, and epimerases (in red), to form 2° bile acids within the intestinal tract. Tilde (~) are placed by bile acids for which synthesis is not clearly delineated. Created with Biorender.com.

Next-generation sequencing analysis revealed that strains from Clostridium, Lactobacillus, Bacteroides, Bifidobacterium, and Enterococcus genera possess multiple bsh genes within and across taxa, encoding BSH enzymes with different specific activities (52–55). Post-deconjugation, other bacterial modifications of unconjugated PBA include dehydroxylation, epimerization and reduction reactions catalyzed by select gut bacteria, including Bacteroides, Peptostreptococcus, Ruminococcus, Eggerthella, Clostridium, Escherichia, and Eubacterium genera (52, 56).

The activity of SBA-producing enzymes may vary between gut microbes; BSH from Lactobacillus strains exhibited superior deconjugation activity compared to BSH found in Bacteroides strains, which had variability in deconjugation activity (55). Clostridium XIVa, Clostridium XI, and Eubacterium taxa produce SBA, such as deoxycholic acid (DCA) and lithocholic acid (LCA), using 7α/βDHs encoded by BA inducible (bai) operon genes (57–59) (Figure 2). Some gut bacteria, such as Clostridium hiranonis TO-931, encode BSH and 7α/βDH and therefore perform both reactions needed to modify conjugated PBA into DCA and LCA (56). BSH genes were also identified in archaeal species, such as Methanobrevibacter smithii ATCC 35061 and Methanosphaera stadtmanae DSM 3091 (60, 61). Human intestinal archaea may also possess 3α/12α-HSHD activity in addition to BSH, which suggests archaea may influence SBA pools and BA signaling (62).

In the intestinal lumen of humans and rodents, the most concentrated SBAs include ωMCA, hyodeoxycholic acid (HDCA), murideoxycholic acid (MDCA), LCA, DCA, HCA, as well as their taurine or glycine conjugates (63, 64). While rodent gut bacteria can convert βMCA to ωMCA, the origin of ωMCA in humans is unclear (65, 66). UDCA is a PBA in rodents and bears, but a SBA in humans (42). Structures of unconjugated and conjugated PBA and SBA species common to humans and mice and are shown on Figure 3. Iso-, allo-, 3-oxo-, 7-oxo-, and isoallo- forms of CA, CDCA, LCA, and DCA, are rare SBA species synthesized by bacteria from Parabacteroides, Bacteroides, Alistipes, Clostridium, Odoribacter, Hungatella, and Lachnospiraceae taxa (61), and may be associated with longevity (67). Due to microbial metabolism of PBAs, human and murine fecal samples contain high concentrations of SBA (68) (Figure 2). Close to 100% of BAs in colon content are derived from microbial modulation (52, 69).

Figure 3.

Molecular structures of primary (1°) and secondary BAs (2°) common to humans and mice. Asterisks (*) are placed by BA for which classification differs between humans and rodents. *Ursodeoxycholic acid is a primary BA (PBA) in mice and a secondary BA (SBA) in humans. * Hyocholic acid may be a PBA in humans and a SBA in mice. CDCA is converted to α/βMCA in mice. PBA and SBA that enter the liver via portal vein are conjugated with taurine (mice) or glycine (humans) in the liver and released into the intestine. In the intestine, PBAs are converted to SBAs via microbial deconjugation, epimerization, and dehydroxylation reactions which lead to a diversity of unique BA species that can enter into enterohepatic circulation in the small intestine (high levels of BAs; ~90–95%) and large intestine (low levels of BAs; <0–2%).

Microbes metabolize BAs and BAs in turn shape the gut microbial community. Due to their amphipathic nature, BAs act as detergents with strong antimicrobial properties (70) that influence composition and diversity of the gut microbiota (54, 71). Unconjugated BAs exhibit more potent antibacterial activity than conjugated BAs and gram-positive gut bacteria are more sensitive to detergent properties of BAs than gram-negative bacteria (71). It was recently shown that pathogenic bacterial species, Klebsiella pneumoniae and Enterococcus faecalis are resilient to high BA levels in vitro (72), suggesting elevated BA levels may not protect against all pathogens. Interestingly, BAs rapidly alter host gut microbial metabolism including amino acid, nucleotide, and carbohydrate metabolic pathways (71). Concomitant alterations in gut microbial communities and BA profiles are associated with metabolic and intestinal diseases (42). Intestinal BAs may benefit metabolic health by inhibiting pro-inflammatory gut bacteria/pathogens and subsequent intestinal inflammation.

1.3. BA transport and circulation

BA transport is a highly regulated process crucial to maintenance of BA homeostasis (Figure 4). BAs are secreted into the bile duct canaliculus via ATP-dependent bile salt export pump (BSEP) and multi-drug resistance protein 2 (MRP2), and stored in the gallbladder prior to being secreted post-prandially into the duodenum via mechanisms driven by cholecystokinin (20, 73). Substrates for BSEP includes monovalent BAs like taurine- and glycine-conjugated BAs while less abundant divalent BAs like sulfate- or glucuronide-conjugated BA are transferred by MRP2 (74). Sulfated BAs inhibits BSEP transport via allosteric binding (74).

Figure 4.

Hepatic and enterocyte BA transporters. Bile acid (BA), Ileal bile-acid-binding protein (IBABP), apical Na+-dependent bile salt transporter (ASBT), organic solute transporter-alpha (OSTα) and -beta (OSTβ), multi-drug resistance protein (MRP)-2(MRP2), −3(MRP3), and −4 (MRP4), organic anion transporting polypeptide 1(OATP1) and 2 (OATP2), bile salt export pump (BSEP), NA-taurocholate co-transporting polypeptide (NTCP), fibroblast growth factor receptor 4 (FGFR4), fibroblast growth factor-15/19 (Fgf15/19), small heterodimer partner (SHP). Image created with Biorender.com.

The majority of BAs (~95%) secreted into the intestine are reabsorbed into enterohepatic circulation via BA-specific transporters, namely apical Na+-dependent bile salt transporter (ASBT) on the apical side of enterocytes (75). Most BA recirculation occurs in the ileum, where ASBT is highly expressed (4, 76). Ileal ASBT facilitates uptake of predominantly conjugated BA while deconjugated BA are reabsorbed via passive diffusion throughout the intestinal tract (75). Located in cytoplasm of enterocytes, ileal bile-acid-binding protein (IBABP), also known as fatty acid binding protein-6 (FABP6), promotes BA flux to protect enterocytes from BA accumulation (77).

Within enterocytes, BAs can interact with intracellular nuclear or membrane-bound G-protein-coupled BA receptors prior to entering circulation or being excreted (76). Nuclear transcription factor farnesoid X-receptor (FXR) is a BA receptor classically known for its role in regulating hepatic BA synthesis and efflux (78–80). Intestinal FXR activation by BAs initiates transcription of fibroblast growth factor-15 (Fgf15) in mice or FGF19 in humans, which enters circulation and binds to the fibroblast growth factor receptor 4 (FGFR4)/βKlotho receptor in the liver to trigger a cascade of reactions to inhibit BA synthesis (5, 51).

On the basolateral side of enterocytes, enterohepatic BA transport is mainly mediated by MRP3 and MRP4 (81), as well as organic anion transporting polypeptide 2 (OATP2), and heteromeric organic solute transporter-alpha (OSTα) and -beta (OSTβ) (82, 83). Conjugated and unconjugated BAs are absorbed during BA uptake in the terminal ileum; however, conjugated PBAs (e.g., TCA) are preferentially re-absorbed (84). BA can also be re-excreted into the intestine via MRP2, which is expressed on the apical side of enterocytes (85). Unabsorbed BAs enter the colon and some unconjugated BA (e.g., CA, CDCA) passively diffuse across colonocytes (4). Lower concentrations of unconjugated BA are reabsorbed into the portal vein compared to conjugated BA species, most unconjugated BA are excreted in feces while conjugated BA recirculate back to the liver (4, 20, 76).

BAs from enterohepatic circulation can enter the liver via Na+-taurocholate co-transporting polypeptide (NTCP) or OATP1 and activate hepatic FXR signaling (86). Hepatic FXR activation initiates transcription of small heterodimer partner (SHP), which inhibits expression of CYP7A1 in the liver (87), although intestinal FXR signaling is the predominant pathway for suppressing BA synthesis. In the liver, BAs can enter systemic circulation via hepatic MRP3, MRP4, and by bi-directional transporter OSTα-OSTβ heterodimer (88). In normal physiological conditions, less than 1% of the total BA pool reaches systemic circulation (89). Liver injury causes increased efflux of BA into circulation, a feature associated with T2D, cardiomyopathies, cholestasis, and liver diseases (11, 90–93). Preclinical and clinical data highlight the relevance of BA signaling and BA receptor modulation in mediating metabolic health (94–96).

1.4. Canonical and non-canonical BA receptors

Uncovering the signaling properties of BAs has revolutionized understanding of BA biology (32). FXR and Takeda G-protein coupled receptor 5 (TGR5) (97, 98) are the best studied BA receptors. Constitutive androstane receptor (CAR), pregnane X-receptor (PXR), retinoid X-receptor (RXR), liver X-receptor-α and -β (LXRα/β) and vitamin D receptor (VDR) are non-canonical nuclear BA receptors (33), while sphingosine-1-phosphate receptor-2 (S1PR2) (99), cholinergic receptor muscarinic 2 (M2R) and 3 (M3R) (100, 101) are non-canonical G protein-coupled BA receptors found on cell membranes (Table 1 and Figure 5).

Table 1.

BA receptors, ligands, and functions

Bile acid receptors (BAR)	Ligands (BA or synthetic)	Function	Refs.
Nuclear BAR FXR	CDCA, DCA, LCA, CA, UDCA 5β-cholanoic acid, 5β-norcholanoic acid, 5α-cholanoic acid, obeticholic acid, GW-4064	Bile acid homeostasis, basolateral BA uptake in hepatocytes, intracellular BA transport in intestine, Fgf15-Shp signaling axes. Expression of ceramide biosynthesis enzymes and other genes related to energy metabolism. Anti-inflammation.	(Sepe et al. 2016; Makishima et al. 1999; Parks et al. 1999; Pellicciari et al. 2002; Jiang et al. 2015)
CAR	No clear BA ligand Phenobarbital, TCPOBOP, Androstanol	Phase I/II bile acid and drug detoxification pathways, and expression of BA transporters. Cell growth and apoptosis, glucose metabolism.	(Tzameli et al. 2000; Li and Chiang 2013; Guo et al. 2003)
PXR	LCA PCN (mouse PXR), Rifampicin (human PXR)	Involved in phase I/II bile acid and drug detoxification pathways. Lipid metabolism and inflammation signaling.	(Staudinger et al. 2001; Xie et al. 2001; Hofmann 2002)
LXR	No clear BA ligand, indirectly activated by FXR. Oxysterols T0901317, GW3965, Dexamethasone	Promotes conversion of cholesterol to BA. Regulation of Lipid metabolism. Expression of SREBP, CYP7A1, and genes in lipid metabolism.	(Landrier et al. 2003; Lehmann et al. 1997; Ulven et al. 2005; Houck et al. 2004)
RXR	Indirectly activated by BA through FXR CDCA, DCA, 3,4-diketocholanic acid	Forms a heterodimer with other nuclear receptors to regulate hepatic gene expression. Regulation if apical BA transporters and lipogenesis through other receptors (PPARs).	(Wang et al. 1999; Perez et al. 2012; Li, Cai, and Boyer 2021)
VDR	LCA, 3-keto-LCA, Calcitriol (1α-25-OH vitamin D3) Alfacalcidol	Calcium homeostasis. Cell proliferation and differentiation. Phase I drug and BA metabolism. Anti-inflammation.	(Makishima et al. 2002; Neimark et al. 2004; Haussler et al. 2011; Wu and Sun 2011)
Membrane BAR TGR5	LCA, DCA, CDCA, CA, UDCA, TLCA, INT-777	Energy metabolism, Anti-inflammation. Intestinal motility and insulin signaling. Cell proliferation and liver regeneration.	(Maruyama et al. 2002; Kawamata et al. 2003; Baghdasaryan et al. 2011)
S1PR2	TCA, TDCA, TUDCA, GCA, GDCA JTE-013	Lipid and glucose metabolism. Inhibit liver regeneration. Regulate allergy and immune cell signaling and differentiation. Myelination.	(Studer et al. 2012; Zhang et al. 2013; Seyedsadr et al. 2019)
M2R, M3R	Acetylcholine. GDCA, TDCA, TCA, TCLA	Cardiac contractions, smooth muscle activity, heart rate.	(Sheikh Abdul Kadir et al. 2010; Raufman, Cheng, and Zimniak 2003; Raufman, Chen, Zimniak, et al. 2002; Raufman, Chen, Cheng, et al. 2002)

1,4-Bis[2-(3,5-dichloropyridyloxy)] benzene (TCPOBOP), Pregnenolone 16α-carbonitrile (PCN), N-(2,2,2-trifluoroethyl)-N-[4-[2,2,2-trifluoro-1-hydroxy-1-(trifluoromethyl)ethyl]phenyl]-benzenesulfonamide (T0901317), Alfacalcidol (1alpha-hydroxyvitamin D3), (3α,5β,6α,7α,12α,23S)-6-ethyl-3,7,12-trihydroxycholane-23-carboxylic acid (INT-777), N-(2,6-dichloro-4-pyridinyl)-2-[1,3-dimethyl-4-(1-methylethyl)-1H-pyrazolo[3,4-b]pyridin-6-yl]-hydrazinecarboxamide (JTE-013). Synthetic BA receptor ligands are in blue text.

Figure 5. Nuclear and membrane-bound BA receptors.

Graphic of nuclear and membrane bound BA receptors. G protein coupled BA receptors including Takeda G-protein coupled receptor 5 (TGR5), Sphingosine 1-phosphate receptor-2 (S1PR2), and cholinergic muscarinic receptors 2/3 (M2R/M3R). Nuclear BA receptors shown below include Farnesoid X receptor (FXR), Constitutive androstane receptor (CAR), Pregnane X Receptor (PXR), Liver X receptor (LXR), and Vitamin D receptor (VDR) which heterodimerize with retinoid X receptor (RXR) to regulate transcription of several genes. Genes regulated by nuclear/membrane BA receptors are shown below images along with a short list of known pathways altered by these receptors.

Canonical BA receptors

Discovered as a BA receptor 20 years ago (78–80), FXR is expressed in many tissue and cell types and is involved in multiple metabolic processes (5). Interestingly, both suppression and activation of FXR have been reported to improve metabolic health in models of metabolic diseases. Suppression of intestinal FXR was shown to improve metabolic phenotypes via an intestinal FXR-ceramide signaling axis which lowered circulating ceramides to promote insulin sensitivity and/or glucose metabolism in murine models of genetic and diet-induced obesity (102–104). FXR activation may indirectly activate TGR5, which is expressed in metabolic tissues including heart, kidney, small intestine, brown and white adipose tissue (BAT, WAT). Tgr5 activation has been shown to increased energy expenditure and mitochondrial markers of non-shivering thermogenesis (98, 105, 106). Unlike the other GPCRs, TGR5 was reported to be activated by unconjugated BA (e.g., LCA, DCA, CDCA, CA, UDCA) (97). TGR5 may aid in processes related to innate immune regulation (107), regulation of glucose metabolism via GLP-1 secretion (108), and enhanced energy expenditure (109). Due to the metabolic potential of TGR5, it is frequently analyzed using models of metabolic illnesses. TGR5 also promotes GLP-1 secretion, which stimulates insulin production (108).

Activation of intestinal FXR inhibited LPS-induced intestinal inflammation in rodent models of colitis as well as human and murine macrophages (110). Hepatic FXR activation reduced lipogenesis and gluconeogenesis in db/db mice leading to less hyperlipidemia and hyperglycemia (111, 112). Activation of FXR and SHP may protect against deregulated inflammatory pathways characteristic of metabolic diseases (113). FXR transactivates SHP (114) which inhibits assembly of NOD-like receptor pyrin domain containing 3 (NLRP3) (115). NLRP3 is predominately expressed in macrophages and promotes transcription of inflammatory cytokines (116). Blocking NLRP3 inflammasome activation was shown to protect against HFD-induced obesity in rodents (117, 118); polymorphisms of NLRP3 was associated with T2D in Chinese Han populations (119). Alterations in hepatic FXR and SHP signaling were shown to regulate expression of lipogenic genes, such as Srebf1, to reduce HFD-induced hypertriglyceridemia (120). The activation or suppression of FXR in metabolic tissues was shown to promote metabolic resilience via different mechanisms; however, these studies did not investigate how other BA receptors are affected.

Non-canonical BA receptors

Nuclear BA receptor RXR heterodimerizes with FXR or LXR to alter expression of PBA biosynthesis and cholesterol transport genes (121). RXR may also heterodimerize with PXR, VDR, peroxisome proliferator-activated receptor-gamma (PPARγ), or CAR to regulate lipid and glucose metabolism, BA synthesis, and protect against toxic BA species (e.g., LCA) in the intestine and liver (122–125). Activation of PXR by LCA was shown to induce expression of CYP3A and promote detoxification pathways (126). PXR may also regulate expression of hepatic Oatp2 and Cyp7a1, which suggests PXR may influence hepatic BA transport and synthesis mechanisms as well (122). LXR is a transcription factor involved in regulating metabolism of cholesterol to oxysterols in the liver (127). FXR may also mitigate inflammatory responses and lipogenesis via PPARγ signaling (128, 129).

M2R and M3R were reported to interact with conjugated PBAs and SBAs including TCA, TLCA, TDCA, and GDCA (130, 131). M2R and M3R predominate in epithelial and smooth muscle tissues (132). A study using neonatal rat cardiomyocytes showed BA-induced arrhythmia was driven via M2R, which may help regulate cardiac contractions (100). M2R and M3R may also play a role in regulating blood glucose and insulin levels (133, 134). Known BA ligands for S1PR2 includes TCA TDCA, TUDCA, GCA, and GDCA (99). Hepatic activation of S1PR2 by conjugated BAs promoted extracellular regulated kinase (ERK)1/2 and protein kinase B (PKB, also called AKT) signaling pathways, which may increase susceptibility to hepatocyte damage or exacerbate cholangiocarcinoma (99, 135). TCA-induced activation of S1PR2 upregulated expression of sphingosine kinase 2 (Sphk2) and genes involved in hepatic nutrient metabolism within primary mouse hepatocytes in a S1PR2- and Sphk2-dependent manner (136). Collectively, these findings highlight the potential of S1PR2 to interfere in hepatic metabolism and cellular processes.

BA signaling plays a pivotal role in mediating energy homeostasis (125); therefore, BA-receptor pathways gained attention as promising targets for therapeutics (3, 32). Consumption of food stimulates BA secretion and influences gut microbial composition and in-turn BA profiles; therefore, the impact of dietary compounds on BA receptor activity may occur via modulation of gut microbiome (137). Dietary compounds targeting BA receptors or downstream signaling pathways may lead to development of novel therapeutics for metabolic, liver, and cardiovascular diseases (4, 138, 139).

1.5. Circadian Regulation of BA

To ensure cholesterol homeostasis, cholesterol conversion to BAs is tightly coordinated during the day/night to align with feeding-fasting times, and light-dark cycles. Cyp7a1 expression oscillates during the day in rodents and humans and increases in response to feeding, supporting a food- and circadian-dependent influence on BA synthesis (140, 141). Circadian clockwork is coordinated by the suprachiasmatic nuclei (SCN) of the hypothalamus and peripheral organs via a feedback loop between a complex of heterodimerized transcriptional activators, i.e., circadian locomotor output cycles Kaput (CLOCK) complexed to brain and muscle-ARNT-Like 1 (BMAL1), and period proteins complexed to cryptochromes (142). Increased levels of CLOCK:BMAL1 during waking hours activate transcription of orphan nuclear receptor, reverse erythroblastosis virus-α/β (REV-ERBα/β), a regulator of lipid metabolizing genes (143). REV-ERB genes recruit histone deacetylase 3 to different regulatory genes and change their expression by deacetylation (144). Expression of BA regulating genes were shown to correlate with increased REV-ERBα/β during waking hours (145, 146), suggesting REV-ERBα/β may regulate BA homeostasis genes. REV-ERBα deficient mice had reduced hepatic Cyp7a1 and increased Shp mRNA levels (145, 146). Knock-down studies have postulated that REV-ERBα accumulation may modulate BA synthesis by inhibiting insulin-induced gene 2 (Insig-2) gene expression and preventing its inhibition of SREBP (146). SREBP promotes synthesis of cholesterol precursors, like oxysterols, which can function as ligands to LXR and promote Cyp7a1 transcription (146). Relative mRNA levels of Fxr, BA synthesis regulatory genes (Fgf15 and Shp), and BA transporters (Asbt, Ostα, Mrp2) in livers and ileal tissue were differentially altered by circadian rhythms (147). The circadian serum and liver BAs profiles in C57BL/6 mice indicated enhanced dehydroxylation during fasted state, and increased host re-conjugation of hepatic BAs post-feeding (147), which suggested that gut bacterial BA metabolism also responded to circadian rhythm (147).

In humans, shift work-induced disruption of circadian rhythms has been linked to MetS and CVD (13). HFD daytime restricted feeding regimen in mice was employed to mimic shiftwork eating and was shown to abolish the circadian fluctuation of hepatic CYP7A1 and CYP8B1 genes, as well as period proteins and cryptochrome genes in the SCN, and of CLOCK and CRY genes in the ileum (138). In a model of metabolic disease, HFD-feeding was shown to disrupt the circadian regulation of the gut microbiome (11, 12), which could further influence circadian rhythms of BA profile (147). Supplementing mice with high-fat milk for four weeks in the evenings promoted fatty liver and hypercholesteremia in association with reduced hepatic Clock and Bmal1, and increased Cry2 mRNA levels(148), implicating that genes involved in regulating circadian rhythm may influence mechanisms underlying hepatobiliary diseases. These alterations occurred with reductions to hepatic Cyp7a1, which was hypothesized to contribute to increased liver cholesterol (148). Thus, both obesogenic diets and arhythmic eating may impair BA homeostasis by disrupting circadian rhythm signaling. The impact of diet and other drug therapies on circadian rhythm signaling should be considered in future preclinical studies and translational research.

1.6. Polyphenols and cardiometabolic health

Global prevalence of obesity, MetS, T2D, non-alcoholic fatty liver disease (NAFLD), and CVD are all increasing (149–152). CVD is the most common non-communicable disease globally; in 2019, the WHO reported CVD was associated with 18.6 million deaths globally (249) with 80% of deaths occurring in low to middle income countries (250). The increased incidence of cardiometabolic disease during this time is largely due to increased consumption of highly processed foods that are rich in sugars and seed oils, but low in fiber and phytonutrients, combined with more sedentary lifestyles (153). Evidence suggests that whole and minimally processed foods rich in dietary polyphenols and fiber is protective against metabolic disease in a manner involving the gut bacteria (154).

Polyphenols are secondary metabolites produced by plants to protect against oxidative and environmental stressors; however, humans benefit from consumption of polyphenols as well (155–157). Polyphenols are categorized into several families and comprised of a diversity of monomeric and polymeric subunits containing one or more phenolic rings in their chemical structure (158). Plant foods rich in polyphenols (e.g., fruits, berries, vegetables, tea, coffee, spices, nuts, legumes, and whole grains) have been reported to promote metabolic resilience and may protect against developing cardiometabolic diseases, such as T2D, hepatobiliary diseases, and CVD (159–161). Diet is the main determinant of gut microbial composition (162). Polyphenols generally have poor bioavailability, and their metabolic benefits appear to be mediated by their modulation of the gut microbiota and polyphenol-derived microbial metabolites (163–165). In addition, many studies suggest polyphenols alter a gut microbiome-BA signaling axis to promote metabolic health (Table 2).

Table 2.

Polyphenol-induced changes in BA signaling, species, and receptors

Polyphenol Treatment (type/administration)	Models	△ BA species	△ BA genes	△ Microbiome	Endpoints	Refs.
Metabolic/T2D
Grape polyphenols (GP) 1. 1% GP (w/w) in LFD for 4 weeks 2. 1% GP in HFD for 10 weeks 3. Serum BA at endpoint.	1. db/db mice (LFD) 2. C576BL/J mice (HFD) 3. Ileal organoids from C576BL/J mice (BA treatments)	↑ Serum TCA, DCA ↓ Serum THDC A, ωMCA and TωMC A	↓ Intestinal Tgr5 ↓ Intestinal Fg15 ↓ Intestinal and hepatic Shp ↑ Hepatic Cyp7a1	↑ A. muciniphila and Blautia ↓ SBA-producing gut bacteria: Clostridium, Butyricicoccus, Ruminococcus, Lactococcus, Oscillospira	Improved glucose metabolism	(Tveter et al. 2020)
Blueberry extract (BE) 1. 0.5% BE in drinking water with chow or HFD for 15 weeks 2. 0.5% BE in drinking water with chow for 10 weeks 3. 0.5% BE in drinking water with chow or HFD ± antibiotics in 10 weeks 4. Serum BA concentrations	1. C576BL/J mice (Chow or HFD) 2. db/db mice (Chow) 3. Antibiotic treated mice (Chow or HFD) 4. Primary mouse hepatocytes (Serum BA treatments)	↑ Serum CDCA, LCA ↓ Serum TβMC A, TCA and TαMC A	↑ BAT Tgr5 ↑ Hepatic Fxr ↑ Hepatic Shp	↑ A. muciniphila, Bifidobacterium, and Lactobacillus ↓ Ruminoclostridium and Desulfovibrio	Improved glucose metabolism Reduced inflammation in intestine, circulation, and liver Reduced bodyweight and adiposity Enhanced energy expenditure	(Guo et al. 2019)
Camu camu (CC) crude extract 1. Daily gavage of 200 mg CC/kg bodyweight in HFHS for 8 weeks 2. Microbiota transplant of CC or HFHS over 14 days	1. C576BL/J mice (Chow or HFHS) 2. Fecal microbiota transplant (CC or HFHS microbiota)	Fecal and Serum βMCA, ωMCA HDCA, UDCA, and DCA ↑ Fecal and Serum TβMC A and TαMC A ↓	BAT Tgr5 ↑ Hepatic Fxr ↓ Intestinal Shp ↓	↑ A. muciniphila ↓ Lactobacillus	Enhanced glucose tolerance and energy expenditure Lowered bodyweight gain and tissue adiposity, blunted inflammation, and alleviated LPS-induced endotoxemia	(Anhê et al. 2019)
Grape extract (GE) 1. 1% (w/w) GE in drinking water with LFD or HFFD for 12 weeks 2.Individual and mixtures of GE metabolites for 6 days	1. Male C57BL/6Cnc mice (LFD or HFFD) 2. C3H10T1/2 (Mouse embryonic fibroblasts) differentiated into brown adipocytes	Plasma HDCA, DCA, αMCA ↑ Plasma TCA and TαMC A ↓ Greater plasma SBA/P BA ratio	BAT Tgr5 ↑	↑ Bifidobacteria, Akkermansia, and Clostridia ↓ Lachnoclostridium, Desulfovibrio, Proteobacteria, and Bacteroidetes	Reduced body weight, adiposity, tissue inflammation and improved glucose metabolism	(Han et al. 2020)
Resveratrol (RSV) 1. At 0.8, 4, 20, 100 or 500 mg/kg bodyweight in NCD or HFD for 5 days 2. At 100 and 500 mg/kg body weight (0.1% and 0.5% w/w, respectively) in NCD, HFD, or HCD for 12 weeks 3. RSV doses (0.08, 0.4, 2, 10, 50μm)	1. Male C57BL/6J mice (NCD or HFD) 2. Male C57BL/6J mice (NCD, HFD, or HCD) 3. Caco-2 cells (RSV)	Plasma BA pool ↑	Hepatic Cyp7a1 ↑ Intestinal Lxrα ↑	N.A.	Reduced circulating cholesterol levels Lower hepatic lipid accumulation Higher intestinal excretion of cholesterol	(Pang et al. 2021)
RSV 1. 200 mg/kg bodyweight in HFD for 8 weeks 2. RSV treatment (20μm 50 and 100 μM) with and without CDCA (100μm)	1. C57BL/6J mice (HFD) 2. Hep2G cells (RSV)	Plasma BA pool ↑	Hepatic Cyp7a1 ↑ Hepatic Lxrα ↑	N.A.	Reduce d body weight Total plasma cholesterol, and LDL cholesterol	(Chen et al. 2012)
RSV 1. 0.4% RSV in normal chow diet for 10 weeks 2. Abx FMT from chow and RSV fed	1. db/db mice (Chow) 2. db/db mice (antibiotic cocktail, chow, RSV)		BAT Tgr5 ↑ WAT Tgr5 ↑		Improved glucose clearance Reduced epididymal white adipose tissue	(Hui et al. 2020)
Raspberry Pomace 1. At 7% w/w with or without seeds prepared using standard or fine granulation supplemented into HFD or AIN-93 for 8 weeks	1. Male Wistar rats (AIN-93 or HFD)	Cecal CA and βMCA ↓ Cecal DCA in raspber ry pomace with seeds ↓ Cecal ωMCA, LCA, TDCA, and GCA ↓ in raspber ry pomace, fine ground with seeds Liver BA pool ↓	Hepatic Fgfr4 ↓ in rats suppleme nted with seeds in pomace Hepatic Shp ↑ in rats suppleme nted with seedless pomace	N.A.	Lowered hepatic cholesterol levels	(Fotschki et al. 2017)
Chlorogenic acid 1. Oral gavage at 150mg/kg bodyweight with normal chow or HFD feeding for 16 weeks	1. C57BL/6 mice (HFD)	Serum BA ↓ in normal chow Serum BA ↑ in HFD Serum LCA, TCDC A, THDC A, TUDC A, and TDCA ↑	Intestinal Fxr and Fgf15↓ Hepatic Cyp7a1 ↑	N.A.	Reduced body weight, total plasma cholesterol, and LDL cholesterol Increased energy expenditure	(Ye et al. 2022)
Apple, grapes, or red beet pomace extracts 1. Supplemented in drinking water to rats fed a semisynthetic control diet (SCD)	1. Male Wistar rats	Cecal TCA, DCA, and HDCA ↓ Cecal LCA, αMCA, βMCA, and CA ↑ Fecal DCA and LCA ↓ Fecal αMCA, βMCA, and CA ↑	Not measured	↑ Lactobacillus and Bifidobacterium in apple extract.	Higher concentrations of intestinal cholesterol, and cholesterol metabolites	(Sembries et al. 2006)
Hepatobiliary
Meriva® tablet 250 mg/day delivering 50 mg/day curcumin by mouth 1x/day for 8 weeks	1. NAFLD-patients	↓ Serum CDCA, TCA, LCA	Not measured	N.A.	Reduced BA-load in NAFLD-patients	(Chashmniam et al. 2019)
Curcumin 1. 120mg cucumin /kg by gavage for 5 days 2. 15 tablets of Danning tablet (dnt)/day for 12 weeks 3. 5–20uM curcumin or 5–20 mg/mL DNT	1. alphanaphthylisothiocyana te (ANIT)-induced cholestasis model in C57BL/6 and whole-body FXR−/− mice 2. Patients given DNT after cholecystectomy surgery due to cholecystitis compared to only antibiotic treatment after surgery 3. Mouse primary hepatocytes (HEK293T cells)	↓ Total bile acids in serum, i.e., TCA, THDC A, TCDC A, TUDC A in wildtype mice serum ↓ Hepatic CA, TCA, BMCA, ωMCA, TαMC A, TβMC A, TωMC A in wildtype mice ↓ Serum GCDC A, GCA, and CA in patients	↑ Hepatic Bsep, Cyp7a1, Cyp8b1, Oatp1a1 in wildtype mice ↓ Hepatic Mrp4, Ostβ in wild-type mice ↑ Ileal Ostα in wild-type mice ↓ Ileal Ibabp in wild-type mice	N.A.	Attenuated liver inflammation and necrosis in mice, and symptoms of cholestasis in humans. Dose-dependent activation of FXR by DNT and curcumin	(Yang et al. 2016)
Xanthohumol 60 mg/kg body weight via diet for 12 weeks	1. Male and female, FXRliver−/− mice or wild-type (WT) mice (HFD)	↓ Serum total bile acids, i.e., CA, DCA and ωMCA in FXRliver −/− ↓ Hepatic CA, DCA, TCA and TDCA in FXRliver −/− ↑ Serum and hepatic DCA, TCA, βMCA, TαMC A and TβMC A in wildtype mice	↑ Hepatic Cyp7a1, Mrp3, Mrp4, Bsep, Car, Pxr, Gr	N.A.	Reduced hepatic lipids, and ceramides in serum of both WT and Liver FXR −/− Reduced serum AST, serum cholesterol in Liver FXR/−/− only Increased xanthohumol metabolite s detected in females	(Paraiso et al. 2021)
Resveratrol 10 mg kg/body weight by gavage 1x/day for 28 days	1. Male, Wistar Rats, sham operated, or bile duct obstructed (BDO) groups	↑ bile flow in healthy and BDO mice ↓ Serum total bile acids in BDO-mice	↑ Hepatic Ntcp, Oatp1a1 in BDO mice ↓ Hepatic Mrp3, Cyp7a1 in BDO mice ↑ Hepatic BSEP, MRP2, MRP4, CYP7A1, p-AMPKα, SIRT1 in healthy mice Agonism of PXR detected via TRFRET PXR Competiti ve Binding Assay kit	N.A.	Increased bile flow in healthy mice and biliary excretion of bile acids and glutathione Reduced bile flow, bile duct proliferation, hepatic fibrosis, AST and ALT in BDO-mice	(Eva Dolezelova 2017)
Ellagic acid or trans-resveratrol ~50 mg/kg body weight via drinking water for 4 weeks	1.Female, C57BL/6 (WT) or whole-body CAR−/− mice, given ethanol	N.A.	↑ Hepatic Cyp7a1, Cyp7b1, Hsd3b5 in wildtype mice	N.A.	Reduced liver lipid in WT ethanol fed mice, but not CAR−/− mice Increased HDL in WT ethanol fed mice	(Yao et al. 2014)
Freeze dried grape extract 300 ug/mL in 10% DMSO	1. Primary mouse hepatocytes damaged by TDCA (50–800uM)	N.A.	Upregulat ed Akt/NFkB pathway	N.A.	Reduced cellular apoptosis Protected from oxidative stress via increased glutathione content	(Xu et al. 2009)
Pomegranate peel polyphenols or pomegranate ellagic acids or punicalagin 10, 20, 40 ug/mL	1. Steatosis model of human L-02 hepatocytes induce by fetal bovine serum (50%)	↑ Total bile acid levels	↑ Hepatic PPARγ, CYP7A1, ABCA1 and Pparγ, Cyp7a1, Abca1	N.A.	Reduced cholesterol content in cells Increased bile acid levels in cells	(Lv et al. 2016)
Whole grape powder 46 g orally 1x/day for 4 weeks	1. Healthy adults on a low-polyphenol diet	↓ Serum total bile acids, i.e., species GCDC A, TCDC A, GDCA, and TDCA	Not measured	↑ α-diversity of the gut microbiome ↑ Verrucomicrobia (Akkermansia), Lachnospiraceae_UCG-010 ↓ Bifidobacterium, Dialister	Reduced serum cholesterol, HDL, LDL	(Yang et al. 2021)
Apple polyphenol extract 100 or 500mg/kg body weight by gavage for 12 weeks	1. Male, C57BL/6 mice (HFD)	↓ Fecal total bile acids, i.e., ωMCA, UDCA, TCA, GCDC A, GUDC A, THCA	↓ Hepatic Cyp7b1, Cyp8b1, Fxr ↑ Hepatic FXR ↓ Hepatic CYP7B1, CYP27A1 ↓ Ileal Npc1l1, Abst ↑ Ileal Lxr, Abca1, Abcg1, Abcg5, Abcg8	↑ Akkermansia ↓ Lactobacillus	Reduced mucosal injury Reduced body weight, hepatic triglyceride and cholesterol Ameliorated hepatic steatosis	(Li et al. 2021) (Li et al. 2021)
(-)-epigallocatechin-3-gallate 0.32% of diet weight via diet for 8 weeks	1. Male, C57BL/6 mice (HFD)t	↑ Serum total bile acids, i.e., CA ↓ Serum TCA, TβMC A, TDCA	N.A.	↑ Aldercreutzia, Akkermansia, Allobaculum, Parabacteroides, ↓ Desulfovibrionaceae, Mucispirrillum, Anaerotruncus ↓Shannon index	Reduced body and liver weight Reduced hepatic triglycerides, lipid accumulation and area of fatty lesions	(Chihiro Ushiroda 2019)
C. aurantium L. 8.7 g by gavage for 4 weeks	1. Male C57BL/6 mice, antibiotics treated compared antibiotic + probiotic (containing Lactobacillus casei)	↑ Cecal DCA ↓ Cecal TDCA ↓ Hepatic TDCA, TCA, CA, CDCA, GCDC A, GCA ↑ Hepatic TUDC A, βMCA, GUDC A, and GHDC A ↓ Fecal TCDC A, TCA, CA, CDCA ↑ Fecal TUDC A, UDCA, HDCA, DCA, βMCA, LCA, GLCA	↑ Ileal FXR and FGF15 ↑ Hepatic BSEP ↓ Hepatic CYP7a1, NTCP and OATP	↑ α-diversity of the gut microbiome ↑Lactobacillus, Roseburia, Romboutsia, Ruminiclostridium_9, Lachnospiraceae_NK4A136_ group and Ruminococcaceae_NK4A214_group	Increased body weight Reduced serum endotoxin Increased genetic markers of gut barrier function	(Liu et al. 2020)
Tea polyphenols or Resveratrol 1. 200 mg tea polyphenols/kg or 100mg resveratrol/kg body weight by gavage 1x/day for 8 weeks 2. 20 μmol tea polyphenols/L	1. Male, C57BL/6 mice, high fat diet with or without probiotic (Lactobacillus casei YRL577 and L. paracasei X11) 2. HepG2 cells	N.A.	↑ intestinal Fxr, Fgf15 ↓ intestinal Abst	N.A.	Reduced liver cholesterol and triglyceride levels, and fat vacuole density Slowed body weight gain Reduced serum triglycerides, total cholesterol, LDL, AST, and ALT Increased serum HDL Reduced fecal propionate, and increased butyric acid and valeric acid Reduced lipid accumulation in HepG2 cells	(Zhang et al. 2020)
(-)-epigallocatechin-3-gallate 3.2 g/kg diet via diet for 17 weeks	1. Male, C57BL/6 mice, High-fat western (HFW) style diet	↓ Small intestine total bile acids ↑ Fecal total bile acids in feces	↑ Hepatic Cyp7a1, Cyp27a1, Fxr, PPARα, LDLR, SR-B1	N.A.	Reduced liver weight and mesenteric adipose tissue at week 17 but not 33 Reduced microvesicular steatosis, macrovesicular steatosis, and hepatocellular hypertrophy at week 17 Reduced serum total lipids and cholesterol Reduced ALT at week 16 Reduced body weight gain at weeks 3–16 but not later Decreased serum fasting blood glucose and insulin	(Huang et al. 2018)
Cardiovascu lar
Quercetin or Leucodelphinidin 1. Both at 100mg/kg/day in 2% cholesterol diet (HCD) for 90 days	1. Male albino Sprague Dawley rats (HCD)	↑ Hepatic BA and fecal BA pools	N.A.	N.A.	Decreased atherosclerotic index Serum total- and LDL-cholesterol levels	(Mathew et al. 2012)
Quercetin 1. 0.4%, 4g/kg bodyweight in control diet AIN-93 for 5 weeks 2. Doses of quercetin (0.4% w/w) for 20 min	1. Male Wistar rats (AIN-93) 2.Rat liver microsomes (quercetin)	↑ Serum and fecal BA pools	↑ Hepatic Cyp7a1 ↑ Hepatic Lxrα	N.A	Enhanced cholesterol conversion to BA Increased hepatic cholesterol efflux	(Zhang et al. 2016)
Apple polyphenols 1. Lyophilized; 20% (w/w) in control diet with 0.25% cholesterol diet (HCD) for 4 weeks	1. Male lean and obese (fa/fa) Zucker rats (HCD)	↑ Small intestine and fecal BA pools	N.A.	N.A.	Lowered plasma triglycerides, total- and LDL-cholesterol Heart concentrations of malondial dehyde were reduced	(Aprikian et al. 2002)
Apple pomace with and without seeds 1. 2.1% (w/w) and 6.5% (w/w) apple pomace with and without seeds in control diet for 4 weeks	1. Male F344 rats (Control diet)	↑ Cecal Total and PBA pools	Hepatic Cyp7a1 was unchanged	N.A.	Decreased total-, LDL-, and IDL-cholesterol concentrations	(Ravn-Haren et al. 2018)
RSV 1. Chow, chow with RSV (0.4%; w/w), 1% choline with or without Abs, or RSV and 1% choline with or without Abs for 30 days 2. Chow, Z-Gug (100 mg/kg body weight) with or without RSV, and GW4064 (75 mg/kg body weight) with or without RSV. GW4064 and ZGug given 7 days prior to 30-day endpoint 3. Chow, chow with RSV (0.4%; w/w), 1% choline with or without Abs, or RSV and 1% choline with or without Abs for 4 months	1. Female C576BL/J mice (Chow, RSV, Choline ±Abs, RSV and Choline ± Abs) 2. Female C576BL/J mice (Chow, Z-Gug ± RSV, GW4064± RSV) 3. Female ApoE−/− mice (Chow, RSV, Choline ±Abs, RSV and Choline ± Abs)	TCA, CDCA, and TβMC A↓ CA, βMCA, DCA, and LCA ↑ Ileal and fecal BA pools ↑	Hepatic Cyp7a1↑ Intestinal Fxr ↓ Intestinal Fgf15 ↓	↑ Bifidobacterium, Akkermansia, Bacteroides, and Lactobacillus ↓ Bilophila and Prevotella	Reduced atherosclerotic index, aortic plaque, and circulating total cholesterol after 4 months	(Chen et al. 2016)
Grape seed procyanidin extract (GSPE) 1. 14 hours after a one-time oral gavage of 250mg/kg bodyweight 2. 0, 20, 50, 100μM GSPE with or without CDCA (100μM)	1. Male C576BL/J (WT) and Fxr−/− mice 2. Caco-2 cells (GSPE, GPSE+CDCA)	↓ Serum BA ↑ Fecal BA	↑ Hepatic Cyp7a1 ↓ Intestinal Fxr ↓ Intestinal Fgf15 ↓ Hepatic Srebp1	N.A.	Decreased serum triglyceride and cholesterol levels	(Heidker, Caiozzi, and Ricketts 2016)
GSPE 1. 5h after a onetime oral gavage of 250mg/kg bodyweight with Chow	1. Male Wistar rats (Chow)	N.A.	↑ Hepatic Cyp7a1 and Shp	N.A.	Lower plasma triglycerides, fatty acids, lipoproteins, and LDL-cholesterol	(Del Bas et al. 2005)
GSPE 1. After oral gavage of 250mg/kg bodyweight 2 times: baseline over 14 hours period 2. 0, 20, 50, 100μM GSPE with or without CDCA (100μM) and GPSE with or without GW4064 (1μM)	1. Male C576BL/J and Fxr−/− mice 2. Hela and CV-1 cells (GSPE and CDCA)	N.A.	↓ Hepatic Srebp1 No change in hepatic Shp and Cyp7a1	N.A.	Decreased serum triglycerides and BA	(Del Bas et al. 2009)
GSPE 1. Oral gavage of 250mg/kg bodyweight once 2.	1. Male C576BL/J and Shp−/− 2. HepG2 cells (GSPE treatment)	N.A.	↑ Hepatic Shp ↓ Hepatic Srebp1c	N.A.	Decreased ApoB and triglyceride synthesis Lower serum triglycerides	(Del Bas et al. 2008)
GSPE 1.Oral gavage of 25 mg/kg bodyweight or low-fat condensed milk on pregnant rats once daily, dams were fed with standard diet of 2.99 kcal/g	1. Female virgin rats (Low-fat condensed milk and GPSE)	↓ Fecal total BA, PBA, and SBA pools ↓ DCA and LCA	↓ Hepatic Cyp7a1 ↑ Hepatic Lxrα, Srebp1, Srebp2	N.A.	Decreased HDL-C levels, increased total cholesterol-to-HDL-C ratios and exacerbated fasting triglyceride-to-HDL-C ratios (an atherogenic index marker)	(Del Bas et al. 2015)
Apples 1. Fed 2 whole apples/day for 8 weeks	1. Healthy adults (23 women/17 men; age 51±11 years) with mild hypercholesterolemia	LCA and GUDC A were correlat ed with total cholesterol levels in women No differen ces in other BA species	N.A.	N.A	Decreased serum total and LDL-cholesterol, triacylglycerol, and intercellular cell adhesion molecule-1	(Koutsos et al. 2020)
Litchi pericarp procyanidins (LPPC) (Litchi chinensis) 1. Oral gavage of 100mg LPPC /kg body mass daily in ApoE KO group only for 24 weeks	1. Male ApoE KO and C57BL/6J mice	N.A.	↑ Hepatic Fxr, Lxrα, and Shp	N.A.	Lower plasma total and LDL-cholesterol Glucose levels Reduced body and liver weights Decreased aortic lesions	(Rong et al., 2018)

Polyphenol intervention studies reporting alterations in BA, species, or receptors in the context of metabolic diseases including diabetes, obesity, hepatic disorders, and cardiovascular diseases. High-fat diet (HFD), Grape polyphenols (GP), Blueberry extract (BE), Camu camu extract (CC), High-fat high-sugar (HFHS), high-fat high-fructose diet (HFFD), Not analyzed (N.A), Semisynthetic control diet (SCD), Resveratrol (RSV), normal cholesterol diet (NCD), High-density lipoproteincholesterol (HDL-C), Antibiotic cocktail (Abs, Abx)

Polyphenol-induced alterations of BA pathways represent novel mechanisms of action with respect to resilience against metabolic disease (104, 166–168). SHP and FXR are orphan receptors that may respond to ligands other than BAs (80, 169). Research suggests polyphenols may act as ligands for BA nuclear receptors (170), while other studies indicate polyphenol metabolites do not bind BA receptors (167, 171). Other studies report polyphenols modulate cholesterol metabolism by interacting directly with BA receptors or through indirect promotion of BA synthesis (171–173). Moreover, there is evidence that polyphenols may act as co-agonists of BA receptor activity in the presence of known BA agonists, such as CDCA (171, 174). Based on preclinical and clinical studies, the metabolic benefits of polyphenols may be mediated through modulation of gut microbial communities (165, 175–177) and BA signaling (104, 166–168, 178–180). The following sections review the literature covering the relationships between polyphenol-induced changes in gut microbial communities, BA profiles, and signaling to BA receptors in the context of MetS, T2D, hepatobiliary diseases, and CVD.

2. Polyphenols and metabolic resilience

2.1. Metabolic syndrome and type-2 diabetes

MetS is characterized by coexistence of at least three of five risk factors (i.e., obesity, hypertension, dyslipidemia, insulin resistance, and hyperglycemia), which increases the risk of developing T2D (181). If uncorrected MetS, may progress to T2D, which is characterized by loss of insulin sensitivity, poor glycemic control, and increasing beta cell dysfunction which promotes loss of insulin production (182). Reviewed below is literature suggesting that polyphenol-induced improvements of MetS and T2D symptoms are linked to modulating the gut microbiota, BA profiles, and BA receptor signaling. The metabolic benefits associated with polyphenols discussed herein, such as decreased serum triglycerides, cholesterol levels, and improved glycemia and other phenotypic data, are often reported. However, there is controversy regarding mechanisms by which purified extracts of polyphenols as well as purified polyphenolic compounds improve metabolic phenotypes. Specifically, polyphenol-induced changes in BA receptor activity and markers of BA signaling pathways and BA profiles have limited reproducibility across studies. In addition, data regarding the impact of dietary polyphenols on 1) gut microbiota, 2) BA profiles, and 3) BA signaling is limited as many studies are lacking one or more of these variables needed to understand the mechanisms related to BA signaling.

2.2. Metabolic benefits of polyphenols associated with altered gut microbiota and BA signaling

Wild-type (WT) mice fed a high-fat high-fructose diet (HFFD) and supplemented with 1% GPs in drinking water for 13 weeks showed reductions in body weight gain, adiposity, insulin insensitivity, glucose intolerance, and tissue and serum inflammation (167). These metabolic improvements were associated with a serum BA profile that promoted Tgr5 activation in WAT and BAT along with increased markers of thermogenesis (167). GPs induced a bloom in cecal levels of Bidifdobacteria, Akkermansia, and Clostridia, which were negatively correlated with PBA (TαMCA, TβMCA, TCA) levels, but positively correlated with the SBA, DCA (167). As GPs are not well absorbed, bacterial-derived polyphenol metabolites quantified in serum by LC-MS were hypothesized to be the bioactive compounds responsible for the observed metabolic changes. Treating brown adipocytes differentiated from C3H10T1/2 cells with polyphenol metabolites (individually or mixture) showed that they could increase mRNA levels of (uncoupling protein 1) Ucp1 but not Tgr5, suggesting effects on Tgr5 were indirect (167).

Compared to ingredient-matched control diet, db/db mice fed low-fat diet (LFD) supplemented with GPs rich in proanthocyanidins (PACs) for 4 weeks had reduced hyperglycemia, a bloom in fecal and cecal Akkermansia muciniphila, reduced relative abundance of SBA-producing genera (Clostridium, Lactococcus, Blautia, Butyricicoccus, Ruminococcus, Streptococcus, Dorea), and decreased SBAs (TωMCA, THDCA, and ωMCA) in serum (104). Moreover, GP supplementation of db/db mice reduced intestinal expression of FXR-responsive genes (Tgr5, Shp, Fgf15) including ceramide biosynthesis genes (Cers4, Smpd3, and Sptlc2), suggesting GP-induced intestinal FXR suppression as a mechanism behind the reduced hyperglycemia in db/db mice (104).

Another study reported that mice fed a high-fat high-sucrose (HFHS) diet supplemented with camu camu (CC) fruit extract for 8 weeks had improved glucose metabolism and insulin sensitivity, and increased energy expenditure, which may be explained by higher BAT mRNA levels of Ucp1 (166). In addition, CC supplementation had increased relative abundance of fecal A. muciniphila, reduced Lactobacillus sp., decreased fecal and serum TαMCA and TβMCA concentrations, and higher levels of ωMCA, HDCA, DCA, UDCA, and βMCA (166). Compared to unsupplemented HFHS-fed mice, Tgr5 gene expression was not changed by CC supplementation; however, the authors suggested CC-induced changes in BA profile and energy expenditure may be due to increased Tgr5 activity (166). In addition to high levels of PACs (1855 mg/100g dry weight), the CC extract contained small phenolic compounds (100mg/100g dry weight) and a high level of fiber (34.29 g/100g dry weight) (166). A. muciniphila, a mucin-metabolizing species of phylum Verrucomicrobiota (183), has been associated with improved cardiometabolic outcomes and gut barrier integrity (184–195) with some exceptions (196, 197). Despite numerous differences in polyphenol profile, murine diet base, and duration of diet intervention, both GP and CC extracts promoted a bloom in A. muciniphila and improved glucose metabolism, likely due to PAC compounds (198).

Compared to unsupplemented controls, db/db mice fed a chow diet supplemented with resveratrol, a stilbene found in grape skins, for 10 weeks had improved glucose tolerance and elevated expression of thermogenic markers (Ucp1, Cidea, Prmd16, Ppargc1a, and Dio2) in BAT and WAT browning (199). Resveratrol supplementation reduced fecal relative abundance of Firmicutes, Lactobacillus, Candidatus Saccharimonas and Ruminococcus but increased Bacteroides, Parasutterella, and Mucispirillum (199). Serum LCA, a TGR5 agonist (200), was elevated in resveratrol-supplemented db/db mice, which may explain the observed increased Ucp1 expression in WAT and BAT (199). The gut microbiota may be required for the glucoregulatory effects of resveratrol, as antibiotics partially abolished its anti-hyperglycemic effects (199).

In db/db or HFD-fed male mice provided with blueberry extract in drinking water for 15 weeks, metabolic phenotypes were improved in association with alterations in gut microbial communities, BA profiles, as well as FXR and Tgr5 signaling (201). Compared to HFD-fed mice, blueberry extract supplementation improved energy expenditure while reducing bodyweight gain, plasma and hepatic triglycerides, inguinal and epididymal WAT weight, liver weight, and markers of inflammation (serum LPS, TNFα, and IL-6 levels; Tlr4, Tnfα, and Il6 gene expression) in intestine, liver, and adipose tissue (201). Blueberry extract increased thermogenesis in BAT and promoted browning of WAT in a gut microbiota-dependent manner as these metabolic benefits were abolished by antibiotic treatment (201). Blueberry extract supplementation increased fecal relative abundance of A. muciniphila, Lactobacillus, and Romboutsia, while Desulfovibrio and Ruminoclostridium were decreased (201). In db/db mice, blueberry extract treatment decreased body weight gain and improved glucose tolerance and insulin sensitivity (201). In db/db and HFD-fed mice supplemented with blueberry extract, serum levels of TCA, TβMCA, and TαMCA were reduced while levels of TGR5 agonist LCA and FXR agonist CDCA were elevated in serum, and this BA profile was shown to activate hepatic FXR signaling (201).

Collectively the above studies suggest that dietary polyphenols alter a gut microbiota-BA signaling axis resulting in improved metabolic outcomes. Studies reporting the relationships between polyphenols, BA profile, and BA receptor activity are limited and require further investigation. Elucidating cause-effect relationships between polyphenols and BA receptors in tissue specific knock-out models of BA receptors (e.g., TGR5, FXR, CAR, PXR, S1PR2) are merited.

2.3. Metabolic benefits of polyphenols linked to modified BA signaling and/or BA profile

Studies have identified relationships underlying the metabolic benefits of polyphenols linked to BA signaling and profile independent of analyzing polyphenol-induced changes in gut microbiota.

Polyphenols that modify BA signaling

Unlike FXR, Lxrα/β transactivates Cyp7a1 to induce BA synthesis, Lxrβ, but not Lxrα, deficient mice had impaired glucose-stimulated insulin secretion and increased lipids in pancreatic β-cells (202). Lxrα/β double knockout mice fed a high cholesterol diet were resistant to body weight gain, had reduced plasma triglyceride levels, and improved glucose tolerance (203). Compared to HFD-fed controls, C57BL/6J mice fed HFD supplementation with resveratrol (200 mg/kg per day for 8 weeks) had reduced total cholesterol, increased high-density lipoprotein (HDL) cholesterol, and increased hepatic gene and protein expression of Lxrα and Cyp7a1 (204). Resveratrol treatment of HepG2 cells also elevated Lxrα and Cyp7a1 expression in a dose-dependent manner (204). These data suggest the cholesterol-lowering properties of resveratrol may occur via hepatic BA production through cholesterol turnover (204).

A 12-week diet intervention study showed that mice fed HFD and high-cholesterol diet (HCD) supplemented with resveratrol at two doses (100 or 500 mg resveratrol/kg bodyweight) lowered plasma total cholesterol levels in a dose-dependent manner (205). Resveratrol supplementation reduced serum cholesterol in HFD-fed and HCD-fed mice by increasing intestinal cholesterol efflux. Independent of Lxr-mediated hepatic cholesterol metabolism, cholesterol levels can be regulated via intestinal cholesterol efflux, a process referred to as trans-intestinal cholesterol excretion (206). Ablation of LXRα in Caco-2 cells blocked resveratrol-induced cholesterol excretion (205). These data suggest that the anti-hypercholesteremic benefit of resveratrol requires Lxrα signaling; however, additional studies using tissue-specific LXR knockout mice would be useful to delineate mechanisms.

Oral administration of PAC-rich GPs in rats promoted a healthier lipoprotein profile, increased gene expression of Cyp7a1 and, counterintuitively, its inhibitor Shp (171). GP treatment of HeLa epithelial and CV-1 fibroblast cell lines transfected with FXR promoter-driven luciferase reporter plasmid showed GP increased FXR transcriptional activity in a dose-dependent manner when in the presence of natural FXR agonist CDCA, but not synthetic FXR agonist GW4064 (171). Decreased serum triglycerides and cholesterol and was observed in WT mice fed standard chow diet 14 h after gavage with a single dose of grape seed procyanidin extract (GSPE, 250 mg/kg bodyweight); however, this was not observed in FXR full-body KO mice, indicating that GSPE-induced improvements were FXR-dependent (207). In WT mice, but not Fxr whole-body KO mice, GSPE decreased intestinal BA absorption, intestinal expression of BA transport genes (Ibabp, Asbt) and Fgf15, markers of hepatic hypertriglyceridemia (HMG-CoA reductase and synthase), and Cyp7a1 expression, further suggesting GSPE alters cholesterol and BA homeostasis in an Fxr-dependent manner (207). Consistent with these in vivo data and supporting the idea that GPs inhibit FXR activity, GSPE-treated Caco-2 cells had lower gene expression of Fgf19 as well as apical (Asbt) and basolateral (Ostα/β) BA-transporters (207).

Presence of seeds and food processing method can influence metabolic and BA measures. Male Wistar rats fed HFD supplemented with raspberry pomace with or without seeds and prepared with standard or fine granulation had different effects on cecal BA profiles and hepatic markers of BA signaling (208). Compared to rats fed HFD, HFD supplemented with raspberry polyphenols did not impact cecal BA load or hepatic FXR levels; however, cecal CA and βMCA were reduced (208). Rats fed seed-free raspberry pomace diet had elevated hepatic protein levels of Shp, but comparable hepatic protein amounts of Fgf19 and Fgfr4 as the rats fed AIN93 control diet; while rats fed raspberry pomace containing seeds had lower hepatic Fgfr4 protein levels, total hepatic BAs, and cholesterol levels compared to HFD-fed rats (208). Rats fed HFD with raspberry pomace containing seeds had lower cecal levels of DCA compared to rats fed HFD or HFD supplemented with seed-free raspberry pomace. HFD supplemented with fine-ground raspberry pomace with seeds had lower cecal levels of ωMCA, LCA, TDCA, and GCA compared to all other groups (208). Collectively, these findings suggests that seed polyphenols and pomace particle size extracts are variables that can differentially influence BA signaling pathways, and BA concentrations in intestine and liver (208). Due to the presence/absence of seeds, different raspberry pomace preparations had different levels of fiber and polyphenols incorporated into the final diets limiting the interpretation of the findings (208, 209).

Compared to normal-fat diet fed mice, HFD-fed mice gavaged daily with chlorogenic acid (150 mg/kg bodyweight) for 16 weeks had lower body weight gain, which was associated with inhibition of intestinal FXR-FGF15 and increased Cyp7a1, as determined by assessment of mRNA and protein expression (210). Chlorogenic acid supplementation increased energy expenditures in HFD-fed and normal-fat diet fed mice, which was associated with increased BAT markers for mitochondrial biogenesis and thermogenesis (Ucp1, PPARα, and Pgc1α) (210). chlorogenic acid corrected HFD-induced dyslipidemia by decreasing serum cholesterol and triglycerides and increasing serum levels of LCA, TCDCA, THDCA, TUDCA, and TDCA (210). Oral gavage of synthetic Fxr agonist (GW4064) for one week prior to the 16 week endpoint reversed the protective effects of chlorogenic acid in HFD-fed mice (210). Overall, several studies suggest that dietary polyphenols can improve metabolic phenotypes by altering BA profiles and FXR signaling.

Polyphenol extracts that modify BA profiles

The gut microbiota of rats fed polyphenol-rich extracts of white grapes, apple, or red beets led to changes in gut bacteria which correlated with increased levels of PBAs (mostly CA) and decreased SBAs in fecal and cecal content (211). Wistar rats supplemented with drinking water containing extracts of white grape pomace, apple, or red beet for 4 weeks showed different fecal BA profiles although common changes were observed in cecal content, specifically, decreased levels of TCA, DCA, and HDCA and increased LCA, αMCA, βMCA, and CA (211). Polyphenol supplementation resulted in decreased fecal DCA and LCA while fecal excretion of CA, αMCA, and βMCA were increased (211). Apple, grape and red beet pomace extracts may increase intestinal PBAs while decreasing SBAs, cholesterol, and cholesterol metabolites (211). Male Sprague Dawley rats fed HFD supplemented with black tea (BT) polyphenols (400 mg/kg bodyweight) for 45 days had reduced body weight, visceral adiposity, and fecal fatty acid content (212). Moreover, BT polyphenols downregulated genes involved in bile and pancreatic secretion (213). In summary, the above reports all point to dietary polyphenol-induced changes to BA signaling as an important mechanism for promoting metabolic resilience in rodent models of obesity and T2D.

3. Hepatoprotective effects

3.1. Liver disease and BA homeostasis

Non-alcoholic fatty liver disease (NAFLD) is characterized by fat accumulation in the liver, and is often due to poor diet and sedentary lifestyle factors (214). Almost half of NAFLD-patients present with intrahepatic cholestasis, a condition where bile builds up in the liver(215). The liver directs BA synthesis, conversions, conjugation, and excretion; therefore, disruption of BA homeostasis is associated with disturbances in hepatic carbohydrate and lipid metabolism and liver disease progression (215–217). Hydrophobic BAs become cytotoxic to hepatocytes at high concentrations (218). Both intrinsic toxicity of certain BA as well as increases to total BA levels due to bile duct impairments, drugs, diet, or dysbiosis induces production of inflammatory cytokines in hepatocytes (219–222).

Increased total serum BAs have been detected in patients with NAFLD compared to healthy subjects (216, 217, 223–225). NAFLD-patient and HFD-fed rodent livers had increased gene expression of hepatic BA uptake proteins (NTCP, OATP1B1, OATP1B3) and BA synthesis proteins (CYP7A1, CYP8B1, CYP27A1), consistent with increased serum BA (225). However, evidence also indicates inhibition of intestinal FXR may be beneficial in preventing NAFLD (103, 226, 227). It was proposed that FXR-inhibition may be hepatoprotective by reducing cholesterol accumulation via increased BA synthesis and excretion (227), or by reducing FXR-mediated ceramide synthesis in the intestine (103). Additionally, interventions that reduce BA reabsorption by inhibiting the ABST (228), or sequestering BAs (229), have been proposed as possible NAFLD and cholestasis therapies. As discussed previously, there is no clear signature BA profile for NAFLD (12), despite several studies reporting BA profiles in NAFLD patients and rodent models, with support for DCA, CDCA, CA, UDCA, TUDCA as biomarkers for NAFLD (219, 220, 225, 230–232).An extensive review of toxic effects of different BA species identified challenges in ascertaining the physiological relevance of BA profile and concentrations used in in vitro studies verses in vivo studies and the difficulties in elucidating the effects of single BA species verses the total BA pool (233).

More recently, attention has turned to the gut microbiota for its impact on BA profile and signaling. NAFLD incidence was shown to be correlated with increased abundance of taurine-metabolizing bacteria in the gut, suggesting the role of gut microbiota in BA-related NAFLD progression (225). Levels of unconjugated PBAs in feces of NAFLD-patients correlated with NAFLD score and serum concentrations of alanine transferase (ALT) and triglycerides (219), which may be due to increased bacterial conjugation. While exact mechanisms are yet to be delineated, targeting gut microbiota, or genes regulating BA synthesis and transport may reduce the BA-associated toxicity observed in NAFLD.

3.2. Polyphenols are hepatoprotective in association with BA signaling

Clinical studies on NAFLD- or cholestasis-patients and animal models suggest that hepatoprotective effects may be due to alterations to BA profile and expression of BA regulating genes. Several, but not all (234), polyphenol-intervention studies reported reductions to serum BA in mice or humans (230, 235–238). Some studies hypothesized an FXR-mediated suppression of BA synthesis was integral to hepatoprotective effects (230, 239–241) and others counterintuitively reported increased gene markers of BA synthesis (230, 237, 242–244), which are markers of FXR inhibition. The lack of predictive insight between BA synthetic genes and BA load revealed a necessity for studies which investigate the relation between expression of BA genes and BA profile and the necessity for caution in using gene expression as support for true functions.. This is emphasized by studies reporting increases to BA synthesis proteins but not mRNA from polyphenol-supplementation (237). Increases to gene or protein markers of BA synthesis (CYP7A1, CYP8B1), excretion (BSEP, MRP2), or secretion into circulation (MRP3–4) observed across multiple studies (230, 236, 237, 242–244) together suggests a mechanism for increased elimination of cholesterol form the liver, and BA excretion; however, mechanisms for induction of these genes is not clear, and several receptors are proposed, i.e. FXR, CAR and PXR, PPARs. While the role of BA receptors behind hepatoprotective effects is not yet delineated, there is significant evidence for hepatoprotective effects of polyphenols involving changes to BA signaling. Here we discuss details from polyphenol intervention studies which measured markers of liver health in association with BA homeostasis. See Table 2 for a list of BA species and genes altered by supplementation.

Curcumin containing Danning tablets (DNT), were an effective post-cholecystectomy therapy in humans to prevent infection and inflammation compared to standard antibiotics (230). DNT-supplementation lowered inflammatory cytokines in serum more than antibiotics in association with reduced serum BAs (230). A curcumin-rich supplement, called Meriva^® (250 mg tablets delivering 50 mg curcumin/day) also reduced certain BAs in serum of men and women with NAFLD after 8 weeks (235). In a mouse model of cholestasis, induced by alpha-naphthylisothiocyanate (ANIT), curcumin alleviated symptoms in WT mice in association with reduced BA, though driven by reductions to different BA species than observed in humans (230). Curcumin also reverted gene changes induced by ANIT by increasing mRNA levels of Cyp7a1, Cyp8b1, Oatp1a1, Bsep, and Mrp4 and reducing inflammatory cytokines, and ileal Ibat and Ostβ (230). Interestingly, curcumin effects on BA profile and regulatory genes in ANIT-exposed mice were observed in WT mice, but not whole-body FXR-knockouts, leading authors to hypothesize that curcumin normalized the BA pool via FXR regulation (230). While gene markers in vivo did not definitively show FXR activation or inhibition, both curcumin and a polyphenol-rich date palm fruit extract were shown to activate FXR via in vitro studies in HEK293T or Caco-2 cells (230, 245). Reduced fatty liver HFD-fed mice supplemented with (−)-epigallocatechin-3-gallate (EGCG; 3.2 g/kg diet for 17 weeks) also occurred with increased Cyp7a1, Cyp27a1, Fxr as well as Pparα (242). Reduced BAs in intestinal content and increased BA levels in fecal content also suggested that EGCGs could attenuate NAFLD by reducing BA reabsorption within the gut (242). Interestingly, EGCG-supplemented mice did not exhibit reduced fatty liver after 33 weeks of the diet intervention, indicating that EGCG could reduce diet-induced fatty liver in early stage but not late stage steatosis (242). Increases to Fxr, and Pparα with increased Cyp7a1, Cyp27a1 obscure understanding of the effects of these receptors known to inhibit BA synthesis. Unlike other studies, supplementation with apple polyphenols (500 mg/kg body weight/day) exerted gene and protein markers in accordance with FXR activation, i.e., reduced hepatic Cyp7b1, and Cyp8b1, along with reduced fecal BA after 12 weeks (239). A dose-dependent positive effect of apple polyphenols on FXR activity was detected which suggests an FXR-mediated inhibition of BA synthesis (239). Also, in accordance with increased ileal FXR and FGF15 levels observed in mice treated with mice given bitter orange or probiotics, hepatic CYP7a1, NTCP, and OATPs were decreased and BSEP was increased (240). While it is not clear whether polyphenol-induced reported reductions to BA load in the discussed polyphenol-intervention studies were due to FXR activation or inactivation, these studies suggest polyphenol-mediated hepatoprotection may be due to reduced BA load in a potentially FXR-dependent manner.

Protection from BA toxicity may also be due to improved efficiency of phase II metabolism and phase III hepatic clearance of BAs by PXR and CAR activity. CAR and PXR activation by polyphenols can protect from inflammation by promoting synthesis of glutathione-S-transferases, which catalyze conjugation of glutathione to toxic compounds to facilitate their elimination (246). Several polyphenol-intervention studies in animals and cell models reported increases in glutathione levels (237, 241, 247) suggesting polyphenols can promote elimination of noxious stimuli via increased conjugation to glutathione by CAR and PXR activation. Hops-derived polyphenol xanthohumol (60 mg/kg body weight) reduced fatty liver, inflammation, and AST levels induced by HFD in FXR-liver knockout mice, presenting an FXR-independent mechanism for hepatoprotective effects (236). Authors hypothesized xanthohumol promoted liver health via CAR and PXR, due to observed increases in their mRNA levels in FXR liver knockouts (236). As with curcumin studies, metabolic improvements from xanthohumol in FXR liver knockouts were associated with increased Cyp7a1, Bsep, Mrp3, and Mrp4 mRNA levels (236). Increased expression of Bsep by xanthohumol suggested a mechanism for observed reductions to serum BAs (236). A resveratrol intervention study also supported PXR activation, exhibiting increased mRNA levels of the PXR-target, Mdr1a in healthy mice gavaged with resveratrol (10 mg/kg body weight) for 28 days, and weak agonism of PXR by resveratrol via competitive binding assays (237). Resveratrol increased protein, but not mRNA, levels of BSEP, MRP2, MRP4, and CYP7A1 in healthy mice (237). Again, despite increased expression of BA synthesis proteins, serum BA levels were reduced in resveratrol-treated mice, possibly due to increased bile excretion via BSEP (237). In support of increased BA synthesis, in vitro studies on steatotic human hepatocytes showed resveratrol directly induced BA synthesis (237, 243). Pomegranate polyphenols increased total intracellular BA levels and reduced lipid and cholesterol accumulation in hepatocytes, suggesting that supplementation reduces cholesterol load by increasing BA synthesis (243). Pomegranate polyphenols activated the PPARy-ABCA1/CYP7A1 pathway in a dose-dependent manner (10, 20, and 40 μg/mL), providing a mechanism by which pomegranate polyphenols may increase cholesterol metabolism and BA synthesis (243). A CAR-dependent mechanism was presented in a study of alcohol-induced liver disease, where 4 weeks of ellagic acid or trans-resveratrol water (50 mg/kg body weight/day) reduced fatty liver in association with Cyp7a1, Cyp7b1, and Hsd3b5 in alcohol-fed WT female mice but not whole-body CAR knockouts (244).

A recent polyphenol intervention study has presented that hepatoprotective effects are not solely due to increased or decreased transcription of BA genes or activation/deactivation of BA receptors, but also due to their preserved circadian expression. To study the protective effects of polyphenols on circadian rhythm, the circadian clock was disrupted by limiting feeding to daytime hours in mice which are naturally nocturnal (248). Five weeks of apple polyphenol-supplementation counteracted daytime feeding mediated hepatic lipid accumulation and reduced steatosis and ballooning scores in association with preserved circadian fluctuations of triglycerides and levels of hepatic Cyp7a1 and Cyp8b1 (248). This was the first demonstration of that polyphenol-mediated improvement in liver health was associated with improved circadian regulation of BA synthesis genes. Furthermore, the circadian nature of BA genes and pool may explain why not all studies report markers for increased BA synthesis or suppression, as there may be exaggerated peaks and troughs in expression depending on time of day.

3.3. Polyphenols are hepatoprotective in association with BA signaling and gut microbiota

Several polyphenol intervention studies report changes to the gut microbiota in association with changes to BA profile, with a recurring bloom of Akkermansia muciniphila, a species often associated with metabolic health. Treatment with A. muciniphila improved metabolic health in overweight and obese subjects, therefore dietary strategies to promote its abundance are of interest (249). Polyphenol rich-grape powder was shown to significantly modulate the intestinal microbiota (increased Akkermansia, Lachnospiraceae UCF-010, and Flavonifractor) and reduce total BA levels in serum, particularly conjugated BA (238). Separate polyphenol-intervention studies investigating apple polyphenols or (−)-epigallocatechin-3-gallate (EGCG) also reported increases to Akkermansia, in association with attenuation of HFD-induced NAFLD (234, 239). Unlike most other polyphenol supplementation studies, male mice fed HFD containing 0.32% EGCG for 8 weeks increased total BAs in serum but decreased taurine-conjugated species (234). The ratios of deconjugated to conjugated BA suggested reduced deconjugation of CA and DCA and increased βMCA deconjugation in HFD-fed mice supplemented with EGCG (234), revealing an interesting selectivity of polyphenol-treated bacteria on BA metabolism.

Antibiotic studies have shed insight on the importance of the gut microbiota in BA homeostasis. While diminishing certain taxa is beneficial, depletion of entire niches by antibiotics can induce dysbiosis in association with altered BA profile and signaling (240). Supplementing antibiotic-treated mice with bitter orange (C. aurantium L.) polyphenols (8.7 g) for 4 weeks had similar effects on liver and fecal BA profile and BA-regulating gene expression as treatment with Lactobacillus probiotics (240), suggesting that BA modulation by bitter orange potentially occur via a prebiotic effect. Both bitter orange and probiotics changed hepatic and fecal BA in antibiotic-treated mice to a profile more like control mice which were not treated with antibiotics (240). Bitter orange increased cecal DCA and reduced TDCA in antibiotic-treated mice suggesting increased bile salt hydrolase activity (240). While authors of an EGCG-intervention study predicted reduced bacterial metabolism of BA was beneficial to liver health in context of HFD (234), the antibiotic and bitter orange intervention study suggests some bacterial BA metabolism is important for BA homeostasis.

Polyphenol-induced reduction in Lactobacillus relative abundance was associated with reduced NAFLD (239), while another study showed attenuation of NAFLD when Lactobacillus casei YRL577 supplementation was combined with tea polyphenols (200 mg/kg body weight) or resveratrol (100mg/kg body weight) (241). Such differences indicate that while overall reductions to the Lactobacillus genera may be beneficial, specific probiotic species may benefit BA metabolism. Mice supplemented with Lactobacillus casei YRL577 and tea polyphenols expressed upregulated intestinal Fxr and Fgf15 mRNA and decreased Asbt (241). The effects of combined application of probiotics and polyphenols on expression of intestinal BA pathway genes was proposed to improve markers of NAFLD (241). Further exploration of enhanced benefits of co-administration of probiotic and prebiotic polyphenols vs. independent treatment would provide needed insight into benefits of symbiotics.

In summary, there is compelling evidence for polyphenol’s hepatoprotective effects, though our understanding of both NAFLD etiology and mechanisms of polyphenol benefits are still obscure. It is likely that polyphenols are hepatoprotective and influence BA homeostasis via mechanisms. This is supported by a previous review that noted that the reproducible upregulation of CYP7A1 expression in several polyphenol-intervention studies was attributed to several different mechanisms (250). In the discussed studies, CAR, PXR, and PPARs, and changes to gut bacteria have been implicated behind changes to BA homeostasis. Overall, a combination of increased BA synthesis and excretion induced by polyphenols likely contributes to both reduced cholesterol accumulation BA hepatotoxicity in NAFLD.

4. Cardioprotective effects

CVD covers an array of disorders of the heart or blood vessels including atherosclerosis, coronary heart disease, angina, stroke, heart failure, and many others (251). Metabolic syndrome symptoms such as hypertension, hyperglycemia, dyslipidemia, and high body mass index as well as diabetes, smoking, sedentary lifestyle, and unhealthy diet are all risk factors for CVD (152). Epidemiological evidence suggests that consumption of a variety of whole and minimally processed polyphenols-rich foods can reduce CVD risk (252). Clinical studies have reported that dietary polyphenols reduced blood pressure, blood glucose, total cholesterol, triglycerides, insulin resistance, LDL-cholesterol, lipoprotein oxidation, and inflammatory cytokines while increasing HDL cholesterol, serum antioxidant levels, and intestinal epithelium barrier (252–254). Polyphenols may protect cardiovascular health via mechanisms related to gut microbiota, as the gastrointestinal tract is the site where polyphenols are most concentrated (154, 255). Recent studies reviewed below indicate that cardioprotective effects of polyphenols are due to modulation of a gut microbiota-BA axis.

4.1. Cardioprotective effects of Polyphenols linked to gut microbiota-BA signaling

Recent studies have shown that dietary polyphenols alter a gut microbiome-BA signaling axis in association with altered expression of markers associated with CVD (256). Studies suggest polyphenols modulate a gut microbiota-BA signaling axis which may regulate susceptibility to CVD. Polyphenol consumption is frequently associated with decreased serum triglycerides, cholesterol, LDL, VLDL, and atherosclerotic index, but increased HDL levels. However, there is controversy regarding BA-related mechanisms by which purified extracts of polyphenols as well as purified polyphenolic compounds improve cardiovascular phenotypes. Data suggests that there may be multiple mechanisms by which polyphenol-induced cardioprotective effects may be driven by BA signaling. In addition, studies reporting the impact of dietary polyphenols on 1) the gut microbiota, 2) BA profiles, and 3) signaling to BA receptors is limited as most studies lack data from one or more of these critical variables involved in regulating BA signaling.

Gut bacteria metabolize dietary choline and phosphatidylcholine to trimethylamine (TMA), which is further metabolized by hepatic flavin monooxygenases to trimethylamine-N-oxide (TMAO), a compound that inhibits BA synthesis and induces atherosclerosis (257, 258). Despite its ability to increase expression and activity of FMOs in liver, resveratrol (0.4% in diet) was found to reduce TMAO-induced atherosclerosis in female C576BL/J and ApoE^−/− mice (fed diet with 1% choline) by decreasing gut microbial TMA production (259). Resveratrol-supplemented mice had increased cecal levels of BSH-producing bacteria from Lactobacillus and Bifidobacterium genera and increased fecal BSH enzyme activity (259). BSH activity may have contributed to the observed increase in BA deconjugation and fecal excretion as well as the decreased small intestinal BA content in resveratrol-treated C576BL/J and ApoE^−/− mice (259). Although FXR expression was not altered in resveratrol-treated mice, its activity was suppressed as Fgf15 was decreased and Cyp7a1 expression was increased, which together indicated higher hepatic BA synthesis and turnover (259). Antibiotic treatment was shown to inhibit resveratrol-mediated protection from atherosclerosis in ApoE^−/− indicating that, although TMA-producing bacteria are reduced, other microbial functions may contribute to the anti-atherogenic effects of resveratrol (259). Future studies that investigate how different types of polyphenols may alter specific species/strains of bacteria to promote atherosclerosis protection vs. promotion are needed.

4.2. Cardioprotective effects of polyphenols linked to BA signaling and profile

Male Wistar rats fed AIN-93G supplemented with quercetin (4 g/kg bodyweight) for five weeks had increased fecal and serum BA levels and increased serum LDL-cholesterol (260). Quercetin-supplemented mice had increased gene and protein expression of hepatic LDL cholesterol receptor, a glycoprotein located on cell membranes to scavenge LDL cholesterol from circulation. Liver tissue of quercetin-supplemented rats showed increased mRNA and protein expression of LXRα, but not FXR, as well as increased Cyp7a1 activity, gene and protein expression, indicating increased BA synthesis (260). Contrary to other quercetin-intervention studies (261), no differences in serum triglycerides or cholesterol was observed (260); however, these mice were fed a control diet with normal cholesterol levels which may not represent mechanisms related to the anti-hypercholesteremic properties of quercetin in high-cholesterol diets (260).

Proanthocyanidins are the most abundant polyphenols in red wine which exert anti-atherogenic actions through antioxidation and modulating lipoprotein metabolism (262). An acute study using male Wistar rats gavaged with either a grape seed proanthocyanidin extract (GSPE, 250 mg/kg bodyweight) or water control had lower plasma triglyceride, free fatty acids, apolipoprotein, and LDL-cholesterol levels (262). Hepatic expression of Shp, Cyp7a1, and cholesterol synthesis enzymes were elevated in GSPE-supplemented rats (262). Expression of other markers of lipoprotein metabolism (apoAII, apoCII, and apoCII) were lower in liver of GSPE treated rats, which suggests GSPE may reduce the atherosclerotic risk in association with Shp signaling mechanisms (262).

In another acute study, administration of GSPE (two doses of 250 mg/kg bodyweight within 12 hours) to male C57BL6J WT mice, but not whole-body Fxr^−/− mice, resulted in reduced serum triglycerides and reduced expression of steroid response element binding protein 1 (SREBP1) and several of its target lipogenic genes (171). In FXR-promoter driven luciferase reporter assays, GSPE was shown to increase FXR activity in the presence of CDCA (171).

Acute supplementation of C576BL/J mice with GSPE had decreased intestinal expression of BA transporters (Asbt, Ibabp) and suppressed intestinal FXR signaling, as evidenced by decreased intestinal Fgf15 expression (207). GSPE-treated mice had lower serum and intestinal BA pools, but GSPE treated Fxr−/− mice did not affect serum or fecal BA levels (207). Moreover, gene expression analysis in Caco-2 cells treated with GSPE for 4 – 12 hours suggested GSPE inhibited intestinal expression of ASBT, OSTα/β, FGF19, and IBABP alone or in the presence of 100 μM of CDCA activator. Collectively, these findings suggest that GSPE may rapidly increase BA synthesis to alleviate triglyceridemia in an FXR-dependent manner. Additional studies analyzing how GSPE may alter the gut microbiota may reveal insight into mechanisms by which gut microbial metabolism may explain differences observed between rats/mice, as well as variations between in vitro and in vivo results. Another study using HepG2 cells, male C576BL/J, and whole-body Shp^−/− knockout mice showed the hypotriglyceridemic properties of GSPE may be linked to hepatic Shp signaling to lipogenic markers such as steroid response element binding protein 1c (SREBP1c) and lipoproteins (263). Treatment of HepG2 cells with GSPE (0 – 100 mg/L) inhibited secretion of lipoprotein ApoB and triglyceride synthesis concomitant with elevated mRNA expression and protein levels of SHP (263). Treatment of HepG2 cells with siRNA to silence Shp abrogated the GSPE-induced increase in SHP mRNA levels and decrease in triglyceride concentrations; however, apoB levels were still reduced by GSPE (263). Moreover, 12 hours after oral gavage of GSPE (250 mg/kg bodyweight) GSPE decreased plasma triglycerides in male C576BL/J mice but not in Shp^−/− mice; hepatic expression of Srebp1c was decreased by GSPE treatment in a Shp-dependent manner (263). Litchi pericarp (commonly known as lychee) is a fruit from south China rich in proanthocyanidins and recently reported to have anti-atherosclerotic and hypolipidemic benefits (264). HFD-fed male ApoE KO mice were gavaged daily with lychee procyanidins (100 mg/kg body mass) for 24 weeks and showed lowered plasma total and LDL cholesterol, glucose levels, and reduced body and liver weights compared to unsupplemented mice (264). Lychee procyanidins reduced aortic lesions in ApoE KO mice and elevated hepatic Lxrα, Fxr, and Shp gene expression which may have attenuate lipogenesis of lipoproteins and cholesterol (264).

Additional studies to confirm the conditional anti-CVD properties of polyphenols are also merited. A study using pregnant virgin rats showed that daily oral gavage of GSPE (25 mg/kg bodyweight) resulted in offspring being born with higher atherogenic index markers when compared to control groups (265). Interestingly, GSPE may have impaired cholesterol turnover to BA in adult male offspring evidenced by elevated hepatic Lxrα expression, decreased hepatic Cyp7a1 expression, and lower excretion of fecal BA (265). Moreover, lipogenic genes (Srebf1 and Srebf2) were elevated in liver from offspring of GSPE-gavaged rats, suggesting GSPE during pregnancy may increase CVD-risk of offspring (265).

Supplementing obese male Zucker rats fed control diet (0.25% cholesterol) with lyophilized apple (20% w/w) for 21 days reduced oxidative stress, lowered plasma, liver, and heart triglyceride levels as well as plasma concentrations of total- and LDL- cholesterol (266). Apple feeding of lean and obese Zucker rats had elevated intestinal BA pools and superior excretion of fecal BA which suggested elevated hepatic Cyp7a1 expression; although the authors did not measure Cyp7a1 expression (266). These authors also acknowledge that fiber may have been a confounding variable in this study since they investigated whole apple matrices and not purified apple polyphenols (266). Male F344 rats that were diets supplemented with 2.1% or 6.5% (w/w) apple pomace extract with or without seeds for 4 weeks showed decreased total- and LDL-cholesterol levels in plasma when compared to unsupplemented controls (267). Although apple pomace extract elevated cecal elimination of total and PBAs, fecal BA levels between apple pomace-fed and control-fed rats were not different and changes in cholesterol or hepatic BA synthesis genes (Cyp7a1 and Hmgcr) were not observed (267). Different fiber levels in the two doses of apple pomace may have been a confounding variable (267). A separate study using male albino Sprague-Dawley rats fed a control diet, high-cholesterol (2%), or high-cholesterol diet supplemented with either leucodelphinidin or quercetin (100mg/kg/day) for 90 days showed supplemented rats had a reduced atherogenic index, significantly lower serum total cholesterol and LDL-cholesterol, and increased HDL-cholesterol (261). Although BA receptor signaling was not investigated, leucodelphinidin and quercetin administration significantly increased hepatic BA and fecal BA levels when compared to unsupplemented high-cholesterol diet control (261).

Clinical studies evaluating polyphenols and BA signaling in CVD subjects are limited. A randomized, controlled, crossover intervention study was performed in healthy adults (23 women/17 men; age 51±11 years) with mild hypercholesterolemia where subjects consumed either two apples per day or a sugar- and energy-matched apple control beverage for 8 weeks. Compared to control beverage, whole-apple consumption decreased serum total- and LDL-cholesterol, triacylglycerol, and intercellular cell adhesion molecule-1 levels while endothelial-dependent microvascular vasodilation was increased (268). Whole apples did not significantly alter BA levels but higher levels of total cholesterol in women were positively correlated to LCA and negatively correlated to GUDCA concentrations (268). Collectively, these findings indicate whole apple consumption may reduce atherosclerosis risk in association with improved microvascular function in endothelial cells and by reducing plasma cholesterol levels (268).

5. Challenges and future perspectives

Here, we highlighted research showing the health-promoting effects of dietary supplementation with either purified phenolic compounds or extracts of polyphenols linked to altered gut microbiome, BA profiles, and downstream signaling to BA receptors in preclinical and clinical models. Despite data supporting the efficacy of polyphenols to combat chronic metabolic diseases, limitations in data-processing, lack of studies differentiating effects of extracts vs. purified phenolic compounds, variations in pharmacokinetic properties of polyphenols at different dosages, differences due to biological sex, incomparable effects on BA profile due to measurements form different compartments (serum, liver, or feces, etc.), and variable reproducibilityto changes of gut microbial and BA profile all impede mechanistic insight underlying the reported health benefits that may be linked to BA receptor activity. Variations in pharmacokinetic properties of the different types of polyphenols may explain variability in reported BA profiles and signaling between studies. Efforts to identify which BA species exhibit affinity for nuclear and membrane bound BA receptors have been investigated for decades; however, more thorough investigations are needed to uncover how individual and/or mixtures of BA species may drive BA signaling (32, 200). Mechanistic studies in BA receptor knockout models, especially models with tissue-specific deletions, are needed to delineate how polyphenol-induced BA changes may differentially influence BA receptor signaling.Despite numerous publications on BA profile from liver disease, there is no BA signature for liver injury (269). Due to the diurnal nature of BA synthesis, factors such as time of collection may be a major variable for BA profiles obtained across human and animal studies (270). Studies which take special notice of the time of sample collection are needed to reduce circadian variation as a variable. Additionally, it is not clear whether increased or altered BA profile and signaling are causes or consequences of NAFLD pathogenesis, which complicates interpretation of trends from diet-interventions. Further, identification of cytotoxic BA is difficult due to use of different models and doses across studies. An extensive review of toxic effects of different BA species identified challenges in pinpointing physiological relevance of in vitro studies to in vivo work due to discrepancies between BA concentrations used and effects of single BAs vs. a physiological BA pool (233). Thus, more in vivo work is warranted. Furthermore, inferences of the effects on BA signaling and transfer are difficult to draw from studies which often measure BA in different compartments. Analysis of BA in serum, liver, intestine, feces, etc. is required for a comprehensive overview of how BA synthesis vs. transfer is affected. Therefore, caution must be taken in interpreting the physiological relevance of BA profile across intervention studies.

Sex-related differences in polyphenol absorption and BA profile are also a significant variable. For instance, absorption of xanthohumol was greater in females, and BA were distributed differently in males and females, though differences in metabolic phenotypes attributed to absorption and BA profile were not investigated (271). Such differences could obscure take-aways about effects of polyphenols on BA signaling and species, especially in small clinical studies where biological sex is not always considered for its confounding effects. Thus, more animal studies on females are required to obtain a better fingerprint on effects of polyphenols, since male-based findings are not necessarily applicable.

Comprehensive studies controlling sources of variation for polyphenol-induced alterations in gut microbial communities are imperative to elucidate the role of the gut microbial communities and for reproducibility. Additional experiments studying how gut microbial composition at baseline of the intervention, how strain or model may attribute to differences in gut microbiota, the impact of the environment, stress, and diet on gut microbiota are important considerations for future work. In addition to confounding variables of each model; external factors in analysis of gut microbiota like the sequencing machine used, sample type (e.g., fecal vs. cecal), sequencing depth of each sample, and downstream bioinformatics analysis can vary between studies. Additional work is required to standardize approaches to delineate which changes in gut microbes may conditionally promote superior metabolic outcomes. Identification and classification of bacteria using different databases (e.g., SILVA vs. Greengenes) also attributes to the disparity in information reported between microbiome studies (272). If the identification of rare gut microbes differs between each database, identifying which communities may promote healthier phenotypes would be obstructed. Experiments using standardized approaches for collecting, analyzing, and reporting changes in the gut microbial communities are needed, especially in clinical studies with polyphenol interventions. Furthermore, more studies that leverage multi-omics approaches like whole-genome (shotgun sequencing), and meta-transcriptomics would be influential in providing more accurate resolution and detailed classification of specific microbes (strain-level resolution) and how these microbes may communicate with the host. Currently, most research utilizes 16S rRNA amplicon sequencing to interpret polyphenol-induced changes in gut microbial communities; however, this approach is limited due to low resolution for classification of microbes, and lack of full bacterial genomes to enable better functional annotation of metabolic pathways associated with specific microbes (273).

Since polyphenols exhibit antimicrobial properties (274); higher doses can dramatically alter gut microbial composition (275). In addition to the direct effects of polyphenols at high concentrations, different polyphenolic compounds exhibit differential capacity to alter gut microbial communities and may exert more potent effects on BA synthesis (275–277). The delivery method of polyphenols is a critical component of study design that may explain variations in reported results. Studies delivering extracts or purified compounds apart from a food matrix may yield different results from studies using whole foods or diets fortified with the same composition of polyphenols. In addition to form of delivery, the duration of the polyphenol intervention may influence results. Repeated oral dosing of grape seed extracts improved bioavailability of polyphenols in vivo (278). Variations in these factors can affect downstream impacts of polyphenols on gut microbiome metabolites and should be considered a variable when studying the downstream impact of polyphenols on BA receptor signaling. For instance, if a delivery method reduces bioaccessibility of a polyphenol with reported affinity for a BA receptor (e.g., FXR with ECGC) (170) then this must be considered in the study design and controlled for. There is limited information regarding the affinity of polyphenol metabolites for understudied BA receptors including PXR, VDR, and CAR; however, some evidence suggests polyphenols may promote metabolic resilience through these receptors (279, 280).

Alternative mechanisms reported should be considered as potential sources of indirect interactions between polyphenols and BA receptor signaling. Evidence showing that extracts of polyphenols, and extracts containing fiber can directly bind BAs or interfere in expression of BA transporters was reported (209, 211, 281–283). Select types of polyphenols, such as PACs, and monomers of flavan-3-ols (catechins, epicatechin) sequestered BA (TCA, GDCA, and TDCA) in vitro in a dose-dependent manner (283). Black tea theaflavins (BTT) inhibited cholesterol micelles concentrations in vitro and when fed to rats; however BTT did not inhibit micellular concentrations of bile salts in vitro (284), unlike green tea catechins which were reported to reduce bile salts in cholesterol micelles (285). Interestingly, select theaflavins from black tea, inhibited TCA binding to ASBT in Caco-2 cells via binding of theaflavins to cysteine residues on ASBT (281). Therefore, these data may collectively indicate polyphenols may differentially interfere in BA pathways via their accumulation in lipid micelles, via BA sequestration, and direct binding of polyphenols to BA transporters. The presence of fiber in polyphenol extracts is a confounding variable to elucidate direct effects of polyphenols on altering BA signaling. A recent paper demonstrated the differential BA-binding capacity in vitro of separated extracts of phenolics and fibers in raw kale (209). Moreover, fiber-enriched extracts bound significantly greater percentages of DCA, CDCA, and total BAs in vitro than phenolic-rich extracts of kale (209).

Polyphenols, especially large polymeric compounds such as PACs, are also known to interact with and precipitate proteins (286). The low absorption of parent polyphenols into circulation may therefore be an inherent safety mechanism. Polyphenols and polyphenol-derived microbial metabolites have the potential for off-target or promiscuous binding effects, as with pan-assay interference compounds (PAINS) (287). Web-based applications and databases (e.g., ZINC, ChEMBL) are used to flag compounds for PAIN alerts so they are excluded from bioassays (288, 289). A comprehensive study of publicly available, published data demonstrated that many compounds without PAINS alerts are as promiscuous as those with the alerts, including FDA-approved drugs (290). Whether individual polyphenol compounds will show useful bioactivity or act as an interference compound in a particular assay must be empirically determined. For example, it was recently shown that whether individual polyphenols caused interference or not depended on specific assay conditions, such as solvent composition (291).

We reviewed work demonstrating patterns by which polyphenols induce changes in gut microbiota to promote metabolic improvements in models of T2D, MetS, CVD, and liver diseases, but factors varying across intervention studies include polyphenol dosage, delivery method (oral gavage vs. supplemented in diet), types of polyphenols and their respective bioavailability in vivo, properties of individual polyphenols and their characteristics when mixed in a complex extract are variables to consider for polyphenol intervention studies. Comprehensive experiments that characterize the types/classes of phenolic compounds being fed, the concentrations or individual polyphenols being supplemented, and confirmation of what polyphenols are present in the gut to interact with the gut microbiome, and complementary in vitro studies are needed to investigate how polyphenols can influence the intestine and liver independent of the gut microbiome and gut microbial metabolites. The requirements of the gut microbiota and BA receptors for the observed metabolic benefits of polyphenols is also poorly understood. Experiments incorporating BA receptor KO models and in vitro gut organoid models may reveal insight into mechanisms of action for polyphenols to combat metabolic diseases. Despite these ongoing limitations and challenges, patterns linking polyphenol-induced changes in gut microbiota, BA profiles, and BA receptor signaling have emerged and warrant further investigation.

Abbreviations

7αDHs

7α-dehydroxylases

7βDHs

7β-dehydroxylase

αMCA

α-muricholic acid

βMCA

β-muricholic acid

ωMCA

ω-muricholic acid

ALT

Alanine transferase

ANIT

Alpha-naphthyl-isothiocyanate

ASBT

Apical Na+-dependent bile salt transporter

BAs

Bile acids

BAAT

BA-CoA:amino acid N-acyltransferase

BACS

BA–Coenzyme A synthase

BSEP

Bile salt export pump

BSH

Bile salt hydrolases

BMAL-1

Brain and muscle-ARNT-Like 1

BAT

Brown adipose tissue

CC

Camu camu

CVD

Cardiovascular disease

CDCA

Chenodeoxycholic acid

CYP7A1

Cholesterol-7α-hydroxylase

CA

Cholic acid

M2R

Cholinergic receptor muscarinic 2

M3R

Cholinergic receptor muscarinic 3

CLOCK

Circadian locomotor output cycles Kaput

CAR

Constitutive androstane receptor

CYP

Cytochrome P450

DCA

Deoxycholic acid

EGCG

Epigallocatechin-3-gallate

ERK

Extracellular regulated kinase

FXR

Farnesoid X-receptor

FABP-6

Fatty acid binding protein 6

FGFR4

Fibroblast growth factor receptor 4

FGF15

Fibroblast growth factor-15

FGF19

Fibroblast growth factor-19

γMCA

Gamma-muricholic acid

GCDCA

Glycochenodeoxycholic acid

GCA

Glycocholic acid

GDCA

Glycodeoxycholic acid

HCD

High cholesterol diet

HDL

High-density lipoproteins

HFD

High-fat diet

HFFD

High-fat high-fructose diet

HSDH

Hydroxysteroid dehydrogenases

HCA

Hyocholic acid

HDCA

Hyodeoxycholic acid

IBABP

Ileal bile-acid-binding protein

Insig-2

Insulin-induced gene 2

LCA

Lithocholic acid

LXRα/β

Liver X-receptor-α and -β

LDL

Low-density lipoproteins

LFD

Low-fat diet

MetS

Metabolic syndrome

MRP2

Multi-drug resistance protein 2

MRP3

Multi-drug resistance protein 3

MRP4

Multi-drug resistance protein 4

MDCA

Murideoxycholic acid

BMAL1

Muscle-ARNT-Like 1

NTCP

Na+-taurocholate co-transporting polypeptide

NLRP3

NOD-like receptor pyrin domain containing 3

NAFLD

Non-alcoholic fatty liver disease

NASH

Non-alcoholic steatohepatitis

OATP1

Organic anion transporting polypeptide 1

OATP2

Organic anion transporting polypeptide 2

OSTα

Organic solute transporter-alpha

OSTβ

Organic solute transporter-beta

CYP7B1

Oxysterol 7-alpha-hydroxylase

PAINS

Pan-assay interference compounds

PPARγ

Peroxisome proliferator-activated receptor-gamma

PXR

Pregnane X-receptor

PBAs

Primary bile acids

PACs

Proanthocyanidins

PKB alias AKT

Protein kinase B

RXR

Retinoid X-receptor

REV-ERBα/β

Reverse erythroblastosis virus-α/β

SBAs

Secondary bile acids

SHP

Small heterodimer partner

Sphk2

Sphingosine kinase 2

S1PR2

Sphingosine-1-phosphate receptor-2

SREBP1c

Steroid response element binding protein 1c

CYP8B1

Sterol 12α-hydroxylase

CYP46A1

Sterol 24-hydroxylase

CYP27A1

Sterol 27-hydroxylase

CYP39A1

Sterol 7α-hydroxylase

SREBP

Sterol regulatory element-binding protein

SULTs

Sulfo-aminotransferases

SCN

Suprachiasmatic nuclei

TGR5

Takeda G-protein coupled receptor 5

TCDCA

Taurochenodeoxycholic acid

TCA

Taurocholic acid

TDCA

Taurodeoxycholic acid

THDCA

Taurohyodeoxycholic acid

TαMCA

Tauro-α-muricholic acid

TβMCA

Tauro-β-muricholic acid

TG

Triglycerides

TMA

Trimethylamine

TMAO

Trimethylamine-N-oxide

T2D

Type 2-diabetes

UGT

UDP-glucuronosyltransferase

UDCA

Ursodeoxycholic acid

VLDL

Very-low density lipoproteins

VDR

Vitamin D receptor

WAT

White adipose tissue