Microbially conjugated bile salts found in human bile activate the bile salt receptors TGR5 and FXR

Abstract

Background:

Bile salts of hepatic and microbial origin mediate interorgan cross talk in the gut-liver axis. Here, we assessed whether the newly discovered class of microbial bile salt conjugates (MBSCs) activate the main host bile salt receptors (Takeda G protein-coupled receptor 5 [TGR5] and farnesoid X receptor [FXR]) and enter the human systemic and enterohepatic circulation.

Methods:

N-amidates of (chenodeoxy) cholic acid and leucine, tyrosine, and phenylalanine were synthesized. Receptor activation was studied in cell-free and cell-based assays. MBSCs were quantified in mesenteric and portal blood and bile of patients undergoing pancreatic surgery.

Results:

MBSCs were activating ligands of TGR5 as evidenced by recruitment of G_sα protein, activation of a cAMP-driven reporter, and diminution of lipopolysaccharide-induced cytokine release from macrophages. Intestine-enriched and liver-enriched FXR isoforms were both activated by MBSCs, provided that a bile salt importer was present. The affinity of MBSCs for TGR5 and FXR was not superior to host-derived bile salt conjugates. Individual MBSCs were generally not detected (ie, < 2.5 nmol/L) in human mesenteric or portal blood, but Leu-variant and Phe-variant were readily measurable in bile, where MBSCs comprised up to 213 ppm of biliary bile salts.

Conclusions:

MBSCs activate the cell surface receptor TGR5 and the transcription factor FXR and are substrates for intestinal (apical sodium-dependent bile acid transporter) and hepatic (Na⁺ taurocholate co-transporting protein) transporters. Their entry into the human circulation is, however, nonsubstantial. Given low systemic levels and a surplus of other equipotent bile salt species, the studied MBSCs are unlikely to have an impact on enterohepatic TGR5/FXR signaling in humans. The origin and function of biliary MBSCs remain to be determined.

INTRODUCTION

The liver plays a central role in the handling of bile salts, a class of cholesterol-derived, amphipathic molecules with digestive and signaling actions.¹ Bile salt homeostasis is deranged in numerous liver diseases, and direct pathogenic contributions of bile salts are evident in cholestatic liver disorders.^2–5 Bile salts are molecular mediators of interorgan cross talk in the gut-liver axis, whose proper function is key to maintaining intestinal and liver health.^6,7

The microbial community in the gut has the unique ability to transform liver-derived (ie, primary) bile salts during their enterohepatic cycle. Microbial conversion of bile salts appears confined to bacteria, with a range of gram-negative and gram-positive species harboring one or multiple enzymes that act on the side chain conjugate group or at specific positions within the steroid nucleus.⁸ Some of these secondary bile salts, notably deoxycholic acid (DCA) and conjugates thereof, can become a substantial part of the recirculating pool, impacting host physiology by affecting the digestive, bactericidal, and signaling properties of bile salts.⁸ Less prominent secondary species may act more locally, for example, close to the site of formation, as appears the case for newly recognized bile salt metabolites that modulate T cell differentiation in the intestines.^9–11

A new class of bacterial bile salt metabolites was first reported in 2020,¹² and comprises C₂₄ bile salts that are N-amidated with amino acids other than those used by the host conjugation system (ie, glycine and taurine).¹³ These microbial bile salt conjugates (MBSCs) were present in human and mouse feces, could not be detected in germ-free mice, and were formed in vitro by Clostridium bolteae after presenting with cholic acid (CA) and amino acids. The initial mass spectrometry–based approach led to the identification of three MBSCs, viz. leucocholic acid (LeuCA), phenylalanocholic acid (PheCA), and tyrosocholic acid (TyrCA), but this number rapidly expanded to > 100 by targeted efforts that revealed broad diversity at the level of both bile acid and amino acid precursor.^14–16 Moreover, numerous additional bacterial producers were identified^14,17 and led to the notion that MBSC formation is common among human gut bacteria. Amine N-acyl transferase activity of bile salt hydrolase (bsh), a bacterial enzyme known to catalyze bile salt deconjugation, has been implicated in the generation of MBSCs. Additional biosynthetic routes likely exist as bacteria devoid of bsh can form MBSCs.¹⁴

The biological and pathological functions of MBSCs remain enigmatic.^18,19 MBSCs likely act as ligands for host bile salt receptors, including farnesoid X receptor (FXR) and pregnane X receptor, and accordingly have the potential to modulate processes controlled by these nuclear receptors.^18,20 A recent study demonstrated that several MBSCs inhibit the germination of Clostridium difficile spores in vitro, pointing toward the a role of MBSCs in bacterial communication.¹⁴ Analysis of public mass spectrometry datasets revealed an association between fecal MBSC abundance (viz. LeuCA, PheCA, TyrCA) and inflammatory bowel disease, notably Crohn’s disease with dysbiosis.¹² Mice given a high-fat diet had a higher abundance of the aforementioned MBSCs in their feces, hinting at a relation with obesity-related disorders.¹²

Quantitative data on levels in biospecimens other than intestinal content is lacking, and it is unresolved if MBSCs enter the host circulation. Of note, repeated oral gavage of mice with a considerable dose of LeuCA or TyrCA, did not lead to detection of these MBSCs in portal blood or bile.¹² We previously reported that MBSCs comprise a minor fraction (up to 40 ppm) of total bile salts in chyme of patients with intestinal failure.¹⁸ Clearly, insight into affinities for bile salt receptors and quantitative data on levels in the circulation are pivotal to further our understanding of the role of MBSCs in health and disease. In this study, we aimed to determine whether MBSCs are substrates for bile salt uptake transporters, study their interaction with the main bile salt receptors Takeda G protein–coupled receptor 5 (TGR5) and FXR, and assess their entry into the human circulation.

METHODS

Materials

Chenodeoxycholic acid (CDCA), cholic acid (CA), obeticholic acid (OCA), taurolithocholic acid (TLCA), glycochenodeoxycholic acid (GCDCA), and Forskolin A were purchased from Sigma. Stocks were prepared in molecular biology grade DMSO (Sigma, D8418). MBSCs employed in this study were those initially discovered by Quinn et al,¹² viz. LeuCA, PheCA, and TyrCA, as well as the respective N-amidates with CDCA (leucochenodeoxycholic acid [LeuCDCA], phenylalanochenodeoxycholic acid [PheCDCA], tyrosochenodeoxycholic acid [TyrCDCA]). Both D-stereoisomers and L-stereoisomers were used for N-amidation with tyrosine. Chemical synthesis and purification of the above MBSCs was essentially according to Quinn et al,¹² and all compounds were verified by ¹H and ¹³C nuclear magnetic resonance in deuterated DMSO.

Patient samples

Blood (superior mesenteric vein, inferior mesenteric vein, and portal vein), gallbladder bile, and urine samples were collected intraoperatively from patients undergoing pancreaticoduodenectomy for treatment of head of pancreas tumors at the University Hospital Aachen. These samples were collected in the framework of the FOCUS study (Ethical Commission University Hospital Aachen approval: EK 172/17), following written informed consent of the patient. Furthermore, plasma samples (systemic blood) were analyzed from a previously reported cohort of patients with acute intestinal failure undergoing chyme reinfusion.²¹ All research was conducted in accordance with both the Declarations of Helsinki and Istanbul.

Cell culture

Human embryonic kidney cells (293T, ACC 635) were directly purchased from the German Collection of Microorganisms and Cell Cultures (DSMZ, Braunschweig, Germany). For the nano-luciferase bioluminescence resonance energy transfer assay, 293 cells were directly acquired from the American Type Culture Collection (CRL-1573). Murine macrophages (RAW264.7) were straight from the American Type Culture Collection (TIB-71). All cells were cultured in a humidified atmosphere containing 5% CO₂ at 37°C in high-glucose DMEM supplemented with 10% fetal bovine serum, 1.0 mM pyruvate, 100 U/mL penicillin G, and 100 μg/mL streptomycin. For serum-free conditions, the serum component was substituted for 0.2% bovine albumin serum. All cell culture reagents were purchased from ThermoFisher Scientific.

In initial tests, cell viability was evaluated at the end of incubations by a dehydrogenase activity assay (Cell Counting Kit 8, Abcam #ab228554) and expressed relative to the solvent control.

Plasmids

Expression vectors for human TGR5 (pCMV3-GPBAR1 C-GFPSpark, hereafter pCMV3-TGR5) and its control (pCMV3-C-GFPSpark, hereafter pCMV3-GFP), human apical sodium-dependent bile acid transporter (ASBT) (pCMV3-SLC10A2), and human Na⁺ taurocholate co-transporting protein (NTCP) (pCMV3-SLC10A1) were purchased from SinoBiological. Reporter plasmids were obtained from Promega and encompassed Firefly luciferase under the control of a cAMP-responsive element (pGL4.29[luc2P/CRE/Hygro], hereafter pGL4-CRE-Fluc) and Renilla luciferase (pRL-TK). An FXR reporter (pGL3-Shp_e-FLuc) and pcDNA3.1-based plasmids for expression of human FXRα2, FXRα4, and RXRα were kind gifts of Prof. Saskia van Mil (University Medical Center Utrecht, The Netherlands) and have been described elsewhere.²² Plasmids were propagated in TOP10 E. coli (Thermo Fisher Scientific) and endotoxin-free plasmid DNA was isolated using anion exchange columns (NucleoBond, Macherey-Nagel).

Nano bioluminescence resonance energy transfer assay

Ligand-induced recruitment of a mini-G_s protein to TGR5 was assessed by a nano-luciferase bioluminescence resonance energy transfer assay as detailed elsewhere.^23,24 In brief, 293 cells were transiently transfected (polyethylenimine, Sigma-Aldrich) with a mini-G_s protein and human TGR5 C-terminally fused with nano-luciferase and seeded into white opaque 96-well plates the next day. Test compounds, including solvent (0.5% DMSO) and positive controls (100 µM TLCA), were added 24 hours later, and donor and acceptor emissions were measured directly. Bioluminescence resonance energy transfer ratios (acceptor:donor emission) are expressed relative to the ratio of the solvent control (ie, 0.5% DMSO = 1.0). Dose-response curves were fitted by nonlinear regression [agonist] versus response, variable slope using GraphPad Prism 9. Each condition was analyzed in triplicate, with four independent replications.

Coactivator recruitment assay

Ligand-induced recruitment of steroid receptor coactivator 1 coactivator peptide to the ligand-binding domain of FXR was assayed by time-resolved fluorescence resonance energy transfer (FRET) using LanthaScreen technology according to the manufacturer’s instructions (Thermo Fisher Scientific, cat# A15140). The concentration of test compounds ranged from 10 nM to 100 µM in 1.0% DMSO. time-resolved FRET signals were assessed after 5.5 hours of incubation at room temperature, at wavelengths of 490 nm (donor) and 520 nm (acceptor) using a Tecan Spark microplate reader. FRET ratios (acceptor:donor emission) are expressed relative to the ratio of the solvent control (ie, 1.0% DMSO = 1.0). Dose-response curves were fitted by nonlinear regression [agonist] versus response, variable slope using GraphPad Prism 9. Each condition was analyzed in quadruplicate, with a single replication.

Reporter assays

cAMP

293T cells, grown in 24-well plates until 70%–80% confluency, were transfected (Lipofectamine LTX, Thermo Fisher Scientific) with a mixture of pCMV3-TGR5 (0.35 ng), pGL4-CRE-Fluc (15 ng), pRL-TK (1.5 ng), and pCMV3-GFP (83 ng). After 24 hours, cells were exposed for 5 hours to solvent (0.1% DMSO), positive control (10 µM Forskolin A), or test compounds in serum-free medium, and subsequently lysed by repeated freeze-thawing.

Farnesoid X receptor

293T cells, grown in 24-well plates until 70–80% confluency, were transfected (Lipofectamine LTX, Thermo Fisher Scientific) with a mixture of pCMV3-SLC10A1/pCMV3-SLC10A2 (0 or 1 ng), pcDNA3.1-FXRα2 or pcDNA3.1-FXRα4 (10 ng), pcDNA3.1-RXRα (2 ng), pGL3-Shp_e-FLuc (5 ng), pRL-TK (2 ng), and 81 ng or 80 ng pCMV3-GFP (added to reach 100 ng plasmid per well). Twenty-four hours after transfection, cells were exposed to solvent (0.1% DMSO), positive control (10 µM OCA), or test compounds in serum-free medium for a further 24 hours and lysed.

Firefly and Renilla luciferase activity was measured in cell lysates using the Dual-Luciferase Reporter assay system (Promega) in a GloMax navigator luminometer (Promega). Data are expressed as normalized ratios (firefly:Renilla signal), relative to the solvent control (ie, DMSO = 1.0). Conditions were tested in quadruplicate, with 3 independent replications.

Cytokine release

RAW264.7 cells at 70%–80% confluency were pretreated for 1 hour with solvent (0.1% DMSO), positive control (50 µM TLCA), or the indicated test compounds in serum-free medium. Next, an optimized, nontoxic dose of lipopolysaccharide (LPS) (3 ng/mL; from E. coli O128:B12, Sigma-Aldrich) was added and incubations continued for 23 hours. Conditioned media were harvested, and supernatants (10 min, 300 g) were assayed for IL-6 and TNFα content by ELISA (RnD Systems, DY406 & DY410). Cytokine release was normalized to cellular protein mass. Conditions were tested in quadruplicate, and experiments were repeated thrice.

Quantification of MBSCs

MBSCs were incorporated into an existing liquid chromatography–mass spectrometry assay for quantifying bile salts, essentially according to García-Cañaveras et al.²⁵ In brief, plasma and bile (prediluted with water) samples were deproteinized with 2 volumes of methanol and 1 volume of a mix of deuterated internal standards in isopropanol. Following vigorous mixing and centrifugation (15 min., 50,000 g, 4°C), the supernatant was transferred to sealable microinsert glass vials. 5.0 µL sample was injected onto a reverse phase column (Acquity UPLC BEH Shield RP18 column, 2.1 × 100 mm, 1.7 µm particle size, Waters) equilibrated with 85%A (95% H₂O, 5% acetonitrile containing 10 mmol/L ammonium acetate) and operated at 50°C. Gradient elution to 100%B (acetonitrile) was performed using an UltimateTM 3000 quaternary UPLC pump (Thermo Scientific) at a flow rate of 0.7 mL/min at 50°C. For detection, a Xevo TQ-S mass spectrometer (Waters) with an electrospray ionization source was employed in negative ion mode. MBSCs were monitored as tyrosine/phenylalanine/leucine daughter ions using transitions to 180, 164, and 130, respectively. GCDCA-D4 was used as an internal standard for MBSCs. Unweighted linear regression was used to calculate sample concentrations from a 12-point standard curve spiked in plasma with a low endogenous bile salt content.

Statistical analysis

A linear mixed model was used to analyze data from replicated experiments, with replications modeled as a correlated random effect. For FXR reporter assays, OCA was included as a positive control to verify the proper functioning of the cellular assay. OCA had a notably larger effect than the other compounds, thus impacting on variance structure in the statistical approach. Grubb’s test identified OCA as a significant outlier (p < 0.01) in each of the conditions (+/− transporter). Data points from OCA treatment were therefore not included in the linear mixed model analysis, but are still presented graphically. Treatment effects were compared versus the control situation, applying Bonferroni correction for multiple testing. Data are presented as the average (± SD) of the means of independent replications. A p-value ≤ 0.05 was considered statistically significant. Statistical analyses were conducted in SPSS version 29.0 (IBM) and GraphPad Prism 9.0 (GraphPad Software, La Jolla, CA, USA).

RESULTS

MBSCs are activating ligands for TGR5

TGR5 is a cell surface G protein–coupled receptor for both unconjugated and conjugated bile salts, displaying the highest affinity for secondary species.²⁶ To test whether MBSCs act as ligands of TGR5, we first assessed receptor activation via the recruitment of a mini-G_s protein, using a proximity-based bioluminescence resonance energy transfer assay.^23,24 TLCA, the most potent endogenous TGR5 ligand known thus far, elicits concentration-dependent recruitment of a mini-G_s protein, with an EC₅₀ value (1.7 µM) similar to reported values.²⁷ (Figure 1A). Each of the CDCA-based MBSCs activated TGR5 with similar efficacy as TLCA and host-derived glycine/taurine conjugates of CDCA, albeit with lower potency (in order of decreasing affinity: TLCA>TCDCA=GCDCA=CDCA>LeuCDCA=PheCDCA>D-TyrCDCA=L-TyrCDCA, Figure 1A, Table 1). CA-based MBSCs were weaker TGR5 agonists compared to their CDCA-based counterparts, with EC₅₀ values ~7-fold higher (potency order: TLCA>taurocholic acid=PheCA=glycocholic acid>CA=D-TyrCA> LeuCA, where glycocholic acid denotes glycocholic acid and taurocholic acid denotes taurocholic acid, Figure 1A, Table 1).

FIGURE 1.

TGR5-activating potential of microbial bile salt conjugates. (A) Recruitment of a mini-G_s protein to ligand-activated TGR5 was studied in 293 cells expressing tagged protein versions, allowing their interaction to be monitored by bioluminescence resonance energy transfer. TGR5 activation was evaluated for both CDCA-based (left panel) and CA-based MBSCs (right panel). Experiments were replicated four times, with three technical replicates per condition in each experiment. (B) Signaling downstream of TGR5 was assessed in 293T cells carrying a cAMP-driven luciferase reporter in the absence (−TGR5) or presence of TGR5 overexpression (+TGR5). TGR5 activation was evaluated for both CDCA-based (left panel) and CA-based MBSCs (right panel) at 100 µM. Experiments were repeated 3 times, with four technical replicates per condition in each replication. (C) Inhibition of NF-κB signaling was tested in RAW264.7 cells treated with the indicated test compounds for 1 hour prior to exposure to solvent (−LPS) or 3 ng/mL LPS (+LPS). Cytokine levels in the medium were normalized to cellular protein content. Experiments were replicated twice, with 4 technical replicates per condition in each test. Data are presented as the average±SD of independent replications and were statistically evaluated using a linear mixed model. Treatment effects were compared versus the control situation, applying Bonferroni correction for multiple testing. Significance is depicted as *p < 0.05, **p < 0.01 and ***p < 0.001. Abbreviations: CA, cholic acid; CDCA, chenodeoxycholic acid; G(CD)CA, glyco(chenodeoxy)cholic acid; LPS, lipopolysaccharide; MBSCs, microbial bile salt conjugates; ns, not significant; T(CD)CA, tauro(chenodeoxy)cholic acid; TLCA, taurolitocholic acid.

TABLE 1.

Overview of affinity constants of microbial bile salt conjugates for the main host bile salt receptors

Bile salt species	EC50 for TGR5 (µM)	EC50 for FXR (µM)
TLCA	1.7 ± 0.1	N.D.
CDCA	32.7 ± 3.2	2.6
GCDCA	29.4 ± 0.9	8.6
TCDCA	26.1 ± 1.3	12.3
LeuCDCA	50.1 ± 7.4	N.D.
PheCDCA	59.2 ± 2.5	N.D.
L-TyrCDCA	88.9 ± 5.3	2.8/7.8 ± 2.9
D-TyrCDCA	93.7 ± 14.6	N.D./6.8
CA	423.0 ± 39.6	N.D.
GCA	342.2 ± 26.7	N.D.
TCA	276.6 ± 16.6	N.D.
LeuCA	1149.8 ± 60.8	N.D.
PheCA	265.4 ± 67.5	N.D.
L-TyrCA	Not calculable	N.D./46.3
D-TyrCA	528.5 ± 61.0	N.D./9.9

Notes: EC50 values for TGR5 were derived from 4 independent cell-based nanoBRET assays. Those for FXR were obtained from a cell-free LanthaScreen coactivator assay, or in the case of a second column entry, a cell-based reporter assay. This reporter assay was repeated thrice for L-TyrCDCA. Data are presented as the average (± SD) of the means of independent replications. Abbreviations: CA, cholic acid; CDCA, chenodeoxycholic acid; D-TyrCA, D-Tyrosocholic acid; D-TyrCDCA, D-Tyrosochenodeoxycholic acid; FXR, farnesoid X receptor; GCA, glycocholic acid; GCDCA, glycochenodeoxycholic acid; LeuCA, leucocholic acid; LeuCDCA, leucochenodeoxycholic acid; L-TyrCA, L-Tyrosocholic acid; L-TyrCDCA, L-Tyrosochenodeoxycholic acid; nanoBRET, nano-luciferase bioluminescence resonance energy transfer; N.D., not determined; PheCA, phenylalanocholic acid; PheCDCA, Phenylalanochenodeoxycholic acid; TCA, Taurocholic acid; TCDCA, taurochenodeoxycholic acid; TGR5, Takeda G protein-coupled receptor 5; TLCA, taurolithocholic acid; TyrCA, tyrosocholic acid.

Downstream consequences of TGR5 activation were next evaluated using a reporter assay for cAMP. The effectiveness of the cellular assay was confirmed by incubation with the adenylate cyclase stimulator forskolin, which activated the reporter independent of TGR5 overexpression (Figure 1B). MBSCs elicited strong activation of the reporter in a strict TGR5-dependent manner. Differences in effectiveness among CDCA-based and CA-based variants were not apparent at the test concentration (100 µM).

In macrophages, TGR5 activation inhibits nuclear translocation of NF-κB and, consequentially, has anti-inflammatory effects.²⁸ NF-κB pathway antagonism by MBSCs was assessed in RAW264.7 macrophages, which endogenously express TGR5.²⁹ Activation of NF-κB signaling through the LPS-toll-lie receptor 4 route resulted in strongly enhanced medium levels of the pro-inflammatory mediators TNFα and IL-6 (Figure 1C). LPS-induced cytokine level was diminished (TNFα: -3.6-fold, IL-6: -4.6-fold) in cells pretreated with TLCA, in line with the reported anti-inflammatory action of TGR5.^28,30 Likewise, prior exposure to PheCDCA and PheCA (the most hydrophobic CDCA-based and CA-based MBSCs in this study) counteracted LPS-enhanced medium levels of TNFα and IL-6 (Figure 1C). Note that the tested MBSCs somewhat reduced medium TNFα levels in the absence of LPS, which may relate to slight effects on cellular viability (Supplemental Figure S1, http://links.lww.com/HC9/A814).

MBSCs are substrates for ASBT and NTCP and activate FXR

To assess if MBSCs are FXR agonists, we first determined the effect of L-TyrCDCA in a cell-free system to bypass the potential requirement for an uptake transporter. Using a time-resolved FRET assay, we observed concentration-dependent recruitment of steroid receptor coactivator 1 coactivator peptide to the ligand-binding domain of FXR upon incubation with L-TyrCDCA (Figure 2A). The affinity for FXR was similar for L-TyrCDCA (EC₅₀ = 2.8 µM) and its unconjugated parent bile salt CDCA (EC₅₀ = 2.6 µM) and somewhat higher relative to host-conjugated CDCA variants (GCDCA: EC₅₀ = 8.6 µM, taurochenodeoxycholic acid: 12.3 µM) (Figure 2A, Table 1). The affinity of Tyr-MBSCs for FXRα2 was also studied in reporter cells expressing an uptake transporter, revealing composite EC₅₀ values of 6.8 µM and 7.8 µM for D-TyrCDCA and L-TyrCDCA, respectively (Figure 2B). A 6-fold lower affinity was noted for L-TyrCA (composite EC₅₀ = 46.3 µM), while the corresponding D stereoisomer had an affinity (composite EC₅₀ = 9.9 µM) for FXR α2 akin to D/L-TyrCDCA (Supplemental Figure S2, http://links.lww.com/HC9/A814).

FIGURE 2.

FXR-activating potential of microbial bile salt conjugates. (A) Ligand-induced recruitment of SRC1 coactivator peptide to the ligand-binding domain of FXR was studied using a TR-FRET–based, cell-free assay. Test compounds were evaluated at concentrations ranging from 10 nM to 100 µM, with four technical replicates per condition. (B) The affinity of D-TyrCDCA and L-TyrCDCA to activate FXRα2 was studied in 293T cells expressing ASBT and an FXR-responsive element-driven luciferase reporter. Transiently transfected cells were exposed to various doses of L-TyrCDCA for 16–18 hours, with four technical replicates per condition. (C-F) Activation of FXRα2 and FXRα4 isoforms by MBSCs was investigated using a cell-based reporter assay. Hereto, 293T cells were transiently transfected with the indicated FXRα isoform, in the absence or presence of a bile salt importer (ASBT, NTCP), and incubated overnight with 50 µM CDCA-based MBSCs (C, E) or their CA-based equivalents (D, F). All conditions were tested in quadruplicate, with 3 independent replications. A linear mixed model was used to analyze data from replicated experiments. Treatment effects were compared versus the control situation, applying Bonferroni correction for multiple testing. Significance is depicted as *p < 0.05, **p < 0.01 and ***p < 0.001. Abbreviations: ASBT, apical sodium-dependent bile acid transporter; GCDCA, glycochenodeoxycholic acid; MBSC, microbial bile salt conjugate; NTCP, Na⁺ taurocholate co-transporting polypeptide; ns, not significant; OCA, obeticholic acid; SRC1, steroid receptor coactivator 1; TR-FRET, time-resolved fluorescence resonance energy transfer.

FXR activation by each of the 8 MBSC variants available for this study, was further evaluated using a cell-based reporter approach. To this end, 293T cells, which have negligible endogenous ASBT and NTCP mRNA expression, were transiently transfected with the liver-enriched (FXRα2) or intestinal-enriched (FXRα4) isoform of FXR in the absence or presence of a sodium-dependent bile salt uptake transporter (Figure 2C-F).^22,31 The membrane-permeable FXR agonist OCA strongly activated the FXR reporter in cells overexpressing FXRα2 or FXRα4, independent of the presence of a bile salt importer. In contrast, notable reporter activation by MBSCs required exogenous expression of either the intestinal (ASBT) or hepatic (NTCP) bile salt uptake transporter. All CDCA-based and CA-based MBSCs were able to activate both FXRα2 and FXRα4 isoforms in the presence of ASBT or NTCP (Figure 2CE). The magnitudes of the effect of individual MBSCs on FXRα2 or FXRα4 were similar at the tested concentration (Figure 2C-F). In contrast with the findings of Quinn et al,¹² FXR reporter activity in 293T cells required exogenous FXR expression (Supplemental Figure S3, http://links.lww.com/HC9/A814).

No substantial entry of MBSCs in the human circulation

To assess if MBSCs enter the host circulation, mesenteric venous and portal blood obtained during abdominal pancreatic surgery was assayed. Demographics and perioperative serum biochemistry of studied patients are presented in Supplemental Table S1, http://links.lww.com/HC9/A814. Despite the sensitivity of the liquid chromatography–mass spectrometry assay in the low nanomolar range (limit of detection: 0.31–0.63 nmol/L), one or more individual MBSCs could be detected in only 6 out of 30 blood samples (Table 2). Levels were above the limit of quantification in a single (cholestatic) patient only, revealing low levels of LeuCA (2.5 nmol/L) and LeuCDCA (1.7 nmol/L) in the venous output from the colon. Note that unconjugated and host-conjugated bile salt species could be quantified in all samples and totaled to a median of 15–36 µmol/L, depending on the originating blood vessel.

TABLE 2.

Levels of individual microbial bile salt conjugates in human biosamples

Study	Sample type	LeuCA (nmol/L)	LeuCDCA (nmol/L)	PheCA (nmol/L)	PheCDCA (nmol/L)	TyrCA (nmol/L)	TyrCDCA (nmol/L)	Total bile salts
FOCUS	SMV (n=10)	1/1 [3.4]	0/0	0/0	0/0	1/0	0/0	35.9 µmol/L [5.8–266]
	IMV (n=10)	1/1 [2.5]	1/1 [1.7]	0/0	1/0	0/0	0/0	15.3 µmol/L [3.1–220]
	PV (n=10)	0/0	0/0	0/0	1/0	0/0	0/0	28.6 µmol/L [5.6–216]
	bile (n=14)	7/7 498 [71–2163]	10/9 411 [65–5111]	4/3 1061 [473–2628]	8/5 464 [88–1327]	1/0	0/0	33.8 mmol/L [2.3–370]
	urine (n=14)	0/0	0/0	0/0	0/0	0/0	0/0	9.5 µmol/L [2.0–158]
RESCUE	CV (n=8)	0/0	0/0	0/0	0/0	0/0	0/0	1.5 µmol/L [0.4–3.4]
	Jejunal chyme (n=8)	8/8 10.2 [1.1–56.4]	8/7 6.5 [0.8–31.0]	8/8 39.8 [3.9–101.3]	8/7 13.1 [1.6–56.4]	5/8 4.4 [2.6–7.0]	6/8 6.0 [2.9–11.9]	4.4 mmol/L [0.8–7.3]

Notes: The first data entry in columns of individual MBSCs states the number of samples in which the particular MBSC could be detected and quantified, respectively. The limit of detection was 0.31 nmol/L for Leu-based MBSCs and 0.63 nmol/L for the other MBSCs. The limit of quantification was 0.63 nmol/L for Leu-based MBSCs and 1.25 nmol/L for the other MBSCs.

If quantification was feasible, the second data entry provides absolute values, presented as median and—if applicable—[range] (third entry). Note that total bile salt levels in plasma and bile/chyme are in micromolar and millimolar units, respectively, and are depicted as median [range]. Abbreviations: CV, cubital vein; IMV, inferior mesenteric vein; LeuCA, leucocholic acid; LeuCDCA, leucochenodeoxycholic acid; MBSC, microbial bile salt conjugate; PheCA, phenylalanocholic acid; PheCDCA, Phenylalanochenodeoxycholic acid; PV, portal vein; SMV, superior mesenteric vein; TyrCA, tyrosocholic acid; TyrCDCA, tyrosochenodeoxycholic acid.

Unexpectedly, we noted that MBSCs were present in the bile of these patients. Leu-based MBSCs could be quantified in over half of the bile samples, with—in that case—median levels of 498 nmol/L (LeuCA) and 411 nmol/L (LeuCDCA) (Table 2). Phe-based MBSCs were detected somewhat less frequently and were quantifiable in 21% (PheCA, median 1061 nmol/mL) and 36% (PheCDCA, median 464 nmol/mL) of the instances. Generally, MBSC variants of Tyr were not present in bile. There was considerable interindividual variation in biliary MBSC levels that remained when expressed relative to the total amount of bile salts. In samples with quantifiable levels, total MBSCs comprised a minor fraction (14.9 [1.5–213] ppm) of total bile salts in bile. MBSCs were not detected in the urine samples of these patients (Table 2).

MBSCs were assayed in a second study, involving patients with acute intestinal failure and verified occurrence of MBSCs in their small intestinal lumen.^18,21 Despite examining those patients with the highest MBSC levels in their chyme, none of the individual MBSCs could be detected in corresponding systemic venous blood (Table 2).

DISCUSSION

In this study, we examined whether microbially conjugated bile salts enter the human circulation and can activate the main host bile salt receptors at the cell surface (TGR5) and in the nucleus (FXR). The key findings are that the tested MBSCs are substrates for the bile salt uptake transporters in the intestine (ASBT) and the liver (NTCP) and are activating ligands of TGR5 and FXR in vitro. Despite the potential for intestinal absorption, the studied MBSCs were generally not detected in the venous output from the small and large intestines in the studied patients. Surprisingly, specific MBSCs were readily detected in human bile, where they comprised a minor fraction of total bile salts. Given their low systemic levels and surplus of other bile salt species with similar affinities for these receptors, the studied MBSCs are unlikely to have an impact on enterohepatic TGR5/FXR signaling in the host.

N-amidates of cholic acid with noncanonical amino acids (viz. Leu, Phe, and Tyr) were first reported 3 years ago and shown to be of gut microbial origin, with 2 Enterocloster boltae strains capable of producing them in vitro.¹² The number of MBSCs (also referred to as microbially conjugated bile acids¹⁴ and bacterial bile acid amidates³²) has vastly expanded since, with variation in parent bile salt and linked amino acid (both proteinaceous and nonproteinaceous, as well as dipeptides) giving rise to recognition of over hundred species nowadays.^15,16 This was paralleled by the identification of additional bacterial producers, with MBSC formation appearing as a common feature of gut bacteria. Previously unrecognized amine N-acyl transferase activity of bsh, a ubiquitous bacterial enzyme that catalyzes deconjugation of host bile salts, appears to be involved in the formation of MBSCs.³³ Additional biosynthetic pathways likely exist, as suggested by MBSC production by bacteria seemingly devoid of bsh.³³ Notwithstanding the advances in basic aspects of MBSCs, insight into their biological function(s) is limited. To appreciate a role in host bile salt signaling, we studied the interaction of MBSCs with the key bile salt receptors TGR5 and FXR. MBSCs are not commercially available and were prepared by chemical synthesis. Studied MBSCs encompassed the three initially discovered MBSCs, their CDCA-based equivalents, and the respective conjugates with the D enantiomer of tyrosine. Previously, we showed that the latter are resistant to cleavage by bacterial bsh and pancreatic carboxypeptidases.¹⁸

MBSCs were found to activate the G protein–coupled receptor TGR5, as evidenced by the recruitment of stimulatory G_s subunit surrogate (mini-G_s protein) and enhanced expression of a cAMP-driven reporter (Figure 1AB). CDCA-based MBSCs were generally more potent TGR5 agonists compared to their CA-based equivalents (Table 1), in line with the reported higher affinity of this receptor for hydrophobic and secondary bile salts.²⁸ This was also reflected in the potency of individual CDCA-based MBSCs (in order of decreasing affinity: LeuCDCA=PheCDCA>D-TyrCDCA=L-TyrCDCA) and their elution order in reverse phase chromatography (TyrCDCA: R_t 6.89 min, LeuCDCA: R_t 8.20 min, PheCDCA: R_t 8.70 min). Phe-based MBSCs were further evaluated for their anti-inflammatory potential and found to diminish the LPS-induced release of pro-inflammatory cytokines TNFα and IL-6 (Figure 1C). This is congruent with an NF-ĸB inhibitory effect of TGR5 activation by these MBSCs.³⁰

The initially discovered MBSCs were reported to activate hepatic FXR in mice.¹² Despite repeated oral gavage, MBSCs could not be detected in the portal or systemic blood in this study, and it could not be ruled out that the observed repression of FXR target genes was due to cleavage of MBSCs and effects of the cholic acid moiety itself.¹² Here, we observed that L-TyrCDCA activated FXR, as supported by the recruitment of a coactivator peptide to its ligand-binding domain in a cell-free system and activation of a cellular FXR reporter (Figure 2AB). The EC₅₀ values (2.8 and 7.5 µM, respectively) were comparable to values for unconjugated and host-conjugated CDCA (Table 1)^34,35 Moreover, each of the tested MBSCs, either with CA or CDCA backbone, profoundly activated the cellular FXR reporter, provided that ASBT or NTCP was present (Figure 2CD). In a pilot experiment, MBSCs were fully recovered in intact form following O/N exposure to 293T cells. Minute amounts (0.0%–0.5% of total) of unconjugated (CD) CA were detected in conditioned media, suggesting that minor degradation may occur after cell contact (Supplemental Table S2, http://links.lww.com/HC9/A814).

It is plausible that MBSCs must be absorbed to initiate signaling through TGR5 or FXR, although there is debate on whether TGR5 in enteroendocrine cells is expressed at the apical membrane and can, thus, be directly activated by luminal bile salts. Our findings support that the tested MBSCs can enter the portal circulation through ASBT, which in the intestines is primarily expressed in the terminal ileum and may then undergo hepatic extraction by NTCP. Quantitative information on MBSC levels is largely lacking, with limited—but valuable—insights from analysis of fecal matter,^36,37 human jejunal chyme,¹⁸ and luminal content sampled at various sites along the intestines in mice^12,14 and humans.^36,38 Apart from revealing associations with inflammatory bowel disease and cystic fibrosis, the available data indicate that MBSCs are also present in the small intestinal lumen in at least healthy subjects and patients with acute intestinal failure. Hence, it is likely that MBSCs are available as substrates for uptake by ASBT. To the best of our knowledge, we present the first data on the occurrence of MBSCs in the human circulation. Using a liquid chromatography–mass spectrometry assay with sensitivity in the low nanomolar range, we first examined MBSCs in blood not yet subjected to first-pass hepatic clearance, thus, offering the best prospects for the detection of MBSCs. Despite notable total bile salt levels in these vessels, individual MBSCs were generally not detected, let alone at levels that allowed quantification (Table 2). Patients without cholestasis were included in this initial analysis (Supplemental Table S1, http://links.lww.com/HC9/A814), making it unlikely that abrogated intestinal bile inflow common to this patient category, accounted for the negligible MBSC levels in the (portal) circulation. We previously reported that MBSC levels in jejunal chyme of patients with acute intestinal failure add up to 39 ppm of total bile salts. In matched plasma of “chyme MBSC-positive” patients, individual MBSCs were again not detectable (Table 2). Considering an obligate 4-fold dilution of samples as part of the analytical workup and a limit of detection of 0.31–0.63 nmol/L, individual MBSCs at concentrations of 2.5 nmol/L or higher can be detected with our assay. Hence, the entry of MBSCs into the host circulation appears nonsubstantial. Cell surface or intracellular degradation of MBSCs by enterocytic carboxypeptidases could be at play. Given the substrate promiscuity of OSTαβ,³⁹ it is unlikely that MBSCs are not recognized by this transporter and are retained in the enterocyte rather than being released into the mesenteric venules. In preliminary tests, we observed that MBSCs activated the cellular FXR reporter if OSTαβ was overexpressed, indicating that this bidirectional transporter imported MBSCs from the medium (Supplemental Figure S4, http://links.lww.com/HC9/A814). In mock-transfected cells, only membrane-permeant OCA was able to activate FXR. The apparent lack of MBSCs in the circulation of human subjects is reminiscent of the failure to detect LeuCA and TyrCA in the portal blood of mice receiving these MBSCs by gavage.¹²

Surprisingly, we detected MBSCs in bile samples from patients undergoing pancreatic surgery (Table 2). Leucine-based MBSCs appeared most abundant and were detectable in more than half of the bile samples. It is conceivable that biliary release of MBSCs is mediated through the canalicular bile salt efflux pump bile salt export pump, which transports canonical bile salt amidates (viz. glycine and taurine conjugates) but, allegedly, also bile salts amidated with the atypical amino acid 2-fluoro-beta-alanine, a catabolite of 5-fluorouracil.⁴⁰ The source of biliary MBSCs is unclear. Ascending bacteria or bile duct–colonizing species, introduced by, for example, endoscopic procedures, may synthesize them from abundant precursor pools in bile. It cannot be excluded, however, that these atypical bile salt conjugates may be produced by the liver. MBSCs comprise up to 213 ppm of total biliary bile salts, and this low abundant fraction may have been overlooked in prior research employing targeted analytics. Anecdotical reports on atypical bile salt conjugates of hepatic origin (eg, ornithinocholic acid) are available in the older literature.⁴¹ Independent of their origin, the presence of MBSCs in bile raises the question of whether these bile salt species recirculate enterohepatically, like host-derived bile salt conjugates. Improved analytical sensitivity will be important to address this point, as bile is presumably a highly concentrated source. In contrast, MBSCs in the bloodstream and/or urine may simply be overlooked by current analytical limitations.

We demonstrated that the affinity of the tested MBSCs for TGR5 and FXR is comparable to, but certainly not higher than, host-derived bile salt conjugates or unconjugated species. Even if the efficacy of MBSCs is somewhat higher, trivial levels in the circulation and surplus of equipotent bile salt species disfavor a major contribution of the tested MBSCs to signaling through TGR5 and FXR. Additional bile salt receptors exist and may be targeted by MBSCs, as already shown for pregnane X receptor.³² Still, above arguments suggest that the primary role of MBSCs may lie outside the direct modulation of host signaling.

MBSCs may contribute to intermicrobial communication. Initial tests indicated that MBSCs (viz. LeuCA, PheCA, and TyrCA), however, did not affect microbial composition in vitro.¹² Microbial bile salt modification is regarded as a protective mechanism that offers adapted bacterial species an advantage.⁸ The hallmark reaction in secondary bile salt formation is 7α-dehydroxylation, with few bacterial species harboring the bai operon that encodes the relevant enzymatic machinery.⁴² One example is Clostridium scindens, which inhibits germination of the pathogen C. difficile by producing secondary bile salts.⁴³ Recently, Foley and colleagues reported that certain MBSCs (viz. TyrCA, PheCA, Pheβ muricholic acid) inhibit spore germination of C. difficile in vivo.¹⁴ Growth of C. difficile was not inhibited in the presence of TyrCA and PheCA but was significantly delayed by the Tyr- and Phe-variant of β-muricholic acid. Bacterial reconjugation may be a broad mechanism to thwart the growth-inhibitory effects of unconjugated bile acids.⁴³

Our study has several strengths and limitations. Analysis of patient materials allowed a first glimpse into systemic MBSC levels, or apparent lack thereof, in a human context. The enigmatic finding of MBSCs in bile raises the provoking question of whether noncanonical bile salt amidates are exclusively produced by microbes. A limitation of our study is that it is unclear if our selection of 6 MBSCs is representative of the vastly expanding repertoire of MBSCs. Without a commercial source, our choice for chemical synthesis was based on the three initially discovered MBSCs. It can be reasoned that these three were discovered in the first place because they represented abundant MBSC species, as indeed reflected in later studies.^14,37,38 Moreover, the studied MBSCs presumably behave as classical, monovalent bile salt anions. Amidation with acidic or basic amino acids, however, will result in divalent or zwitterionic species, with presumably distinct biophysical (eg, membrane permeating) and biological properties. Hence, certain MBSCs may be less susceptible to degradation by host and microbial enzymes and may be absorbed in amounts relevant to host bile salt signaling. Along this line, gavage of mice with D-TyrC(DC)A, which is refractory to in vitro degradation by pancreatic carboxypeptidases and bsh,¹⁸ may shed light on the enzymatic breakdown as the cause of low systemic levels. Note that it is currently unresolved if D-amino acids are actual substrates for bacterial bile salt reconjugation. The above notions offer exciting prospects for experimental and analytical follow-up efforts.

In conclusion, we demonstrated that the tested MBSCs activate the cell surface bile salt receptor TGR5. Moreover, they are activating ligands of the transcription factor FXR, with a cellular bile salt uptake system required for FXR activation. The studied MBSCs are substrates for both ASBT and NTCP, but their entry into the portal circulation is nonsubstantial in humans. MBSCs are readily detected in human bile, where they represent a minor portion of total bile salts. Since the affinity of the evaluated MBSCs for TGR5 and FXR is not superior to host-derived bile salt conjugates, the trivial levels are unlikely to contribute to host bile salt signaling via these receptors. The biological functions of MBSCs remain to be defined and may be restricted to an ecological role in the gut lumen.

Supplementary Material

hc9-8-e0383-s001.docx ^{(423.1KB, docx)}

AUTHOR CONTRIBUTIONS

Steven W.M. Olde Damink and Frank G. Schaap: Conceptualization and Funding acquisition; Ümran Ay, Martin Leníček, Hans van Eijk, and Frank G. Schaap: Methodology; Ümran Ay, Martin Leníček, Raphael S. Haider, Hans van Eijk, and Frank G. Schaap: Formal analysis; Arno Classen, Kiran V.K. Koelfat, Gregory van der Kroft, Ulf. P. Neumann, Carsten Hoffmann, and Carsten Bolm: Resources; Ümran Ay, and Frank G. Schaap: Writing—original draft preparation.

FUNDING INFORMATION

This work was funded by grants from the Netherlands Association for the Study of the Liver (NVH, (Gastrostart #28-2021) and the Deutsche Forschungsgemeinschaft (DFG, German Research Foundation), Project-ID 403224013—SFB 1382.

CONFLICTS OF INTEREST

The authors have no conflicts to report.