Gut-to-bile transfer of microbially amidated minor bile acids in patients with hepatopancreatobiliary disorders

Abstract

Background and Aims:

During bile acid (BA) intestinal transit, microbially amidated BAs (MABAs) are produced. This study investigated their cholephilic behavior and their presence in the bile of patients with hepatopancreatobiliary diseases.

Approach and Results:

Bile samples were collected during surgical or endoscopic procedures and analyzed using high-performance liquid chromatography–tandem mass spectrometry (HPLC–MS/MS), with cholic acid (CA) and chenodeoxycholic acid (CDCA) chemically amidated with leucine (Leu), phenylalanine (Phe), or tyrosine (Tyr) as standards. Gut-to-bile transfer was investigated in cellular and animal models.

MABAs (Leu>Phe>Tyr) were detected (<1 µM) in the bile of ≈50% of patients with hepatopancreatobiliary disorders. Their levels were positively correlated with total BA concentrations and inversely correlated with the proportion of major conjugated BAs, but not with age, fat-soluble vitamin levels, or disease outcomes. Oral gavage of D- and L-enantiomers of Tyr-CA in mice resulted in intestinal hydrolysis and limited access of L-Tyr-CA to the enterohepatic circulation. In rats, the intravenous injection of glycocholic acid (GCA) and MABAs resulted in similarly rapid biliary outputs. The time course of biliary secretion after infusing MABAs and GCA into the microbiota-free rat ileum in situ was also similar. Docking studies predicted the interaction of BA transporters and MABAs with binding energies comparable to those of taurocholic acid (TCA) and GCA. In cells expressing BA transporters, MABA uptake was efficient (NTCP>ASBT>OATP1B3) and inhibitable by TCA.

Conclusions:

Like major conjugated BAs, MABAs are transferred from the gut, where they are produced, to the bile of patients with hepatopancreatobiliary diseases, suggesting gut dysbiosis that favors species generating these compounds.

INTRODUCTION

Cholesterol is biotransformed by hepatocytes into primary bile acids (BAs), such as cholic acid (CA) and chenodeoxycholic acid (CDCA), through double-bond reduction, epimerization of the C3 hydroxyl group, additional hydroxylation(s) of the steroid nucleus, and shortening of the side chain, ending in a carboxylic group that can be subsequently N-amidated with glycine or taurine.¹ BAs are mostly secreted across the canalicular membrane of hepatocytes into nascent bile through the bile salt export pump (BSEP),^2,3 although other ATP-binding cassette (ABC) proteins, such as MRP2^4,5 and BCRP⁶ can also be involved. BAs are cholephilic compounds that are efficiently maintained in the enterohepatic circulation⁷ and play a crucial role in various biological functions. In bile, their ability to form mixed micelles facilitates the transport of hydrophobic substances such as cholesterol, biliary pigments, and protoporphyrins. In the intestine, these polymolecular structures are essential for emulsifying and solubilizing digested fats and fat-soluble vitamins (A, D, E, and K), thereby promoting their absorption.^3,8 Most BA molecules reaching the terminal ileum are efficiently reabsorbed mainly through active transport mediated by the apical sodium-dependent BA transporter (ASBT).⁹ The subsequent translocation across the basolateral membrane into the portal circulation occurs via facilitated passive transport, which occurs through the heteromeric organic solute transporter (OSTα/β).¹⁰ Once transferred to the portal circulation, partly bound to serum proteins such as albumin and lipoproteins,¹¹ BAs reach the liver parenchyma sinusoidal blood from where they are efficiently taken up by hepatocytes through the Na⁺-taurocholate cotransporting polypeptide (NTCP) and, to a lesser extent, by sodium-independent organic anion transporting polypeptides (OATPs), mainly OATP1B1 and OATP1B3.¹² However, ~15% of the BAs entering the small intestine is not recovered by the ileum and continues its journey through the large intestine to reach the colon, where less efficient reabsorption of more hydrophobic species generated by microbial metabolism occurs. Eventually, ~5% of the total BA pool is eliminated daily via feces.⁷ During intestinal transit, BAs undergo microbial conversion, resulting in “damaged” molecular species.³ The most common biotransformations include deconjugation, 7α-dehydroxylation, and oxidation/epimerization of the hydroxyl groups at the C3, C7, and C12 positions.¹³ Certain anaerobic bacteria expressing 7α-dehydroxylases produce the secondary BAs, deoxycholic acid (DCA), and lithocholic acid (LCA), which are the most prevalent species found in human feces.^14,15 Epimerization by the successive action of 7α and 7β hydroxysteroid dehydrogenases (HSDs) transforms CDCA into ursodeoxycholic acid (UDCA). Moreover, oxidation by HSDHs can create more hydrophilic keto groups at the C3, C7, and C12 positions.¹³ The other primary intestinal biotransformation of conjugated BAs is hydrolysis of the C24 N-acyl amide bond, resulting in unconjugated species. This reaction is catalyzed by bile salt hydrolases (BSHs) present in bacteria commonly residing in the small intestine and colon.^16,17 Upon reabsorption in the intestine, unconjugated BA molecules are “repaired” in the liver through reconjugation with glycine or taurine. The additional “repair” through re-hydroxylation of C7 is species-specific but does not occur in humans.

Interestingly, it has been recently found that several intestinal bacteria (eg, Enterocloster bolteae) can conjugate BAs with atypical amino acids such as leucine (Leu), phenylalanine (Phe), and tyrosine (Tyr) in mice.¹⁸ Further investigations in vitro have shown that 27 of the 70 bacterial species (both Gram-negative and Gram-positive) commonly found in the human intestine can perform conjugation reactions between CA and these amino acids to generate Leu-CA, Phe-CA, and Tyr-CA.¹⁹ Furthermore, these microbially amidated BAs (MABAs) have been found in fecal samples from patients with inflammatory bowel disease (IBD) and cystic fibrosis,¹⁸ suggesting that under certain pathological conditions, accompanied by changes in the size or composition of the intestinal microbiota, the synthesis of these BA derivatives may be promoted. Moreover, their degradation could be reduced owing to faster intestinal transit time in IBD, shortening exposure to deconjugating enzymes like BSHs, and to pancreatic insufficiency, resulting in fewer pancreatic carboxypeptidases in the intestinal lumen.

Following the initial description, other MABAs differing in their BA scaffold and linked amino acids have been reported.²⁰ To explore the potential effects of these compounds on host physiology, the interaction of prototypic MABAs with the main BA receptors TGR5 and FXR was assessed in vitro. The tested MABAs did not outperform the much more abundant host-derived BAs.²¹ However, many important questions regarding these fascinating molecules require further investigation. We found it particularly intriguing to clarify the relationship between the presence of MABAs in human bile and various hepatopancreatobiliary disorders affecting BA enterohepatic circulation. Moreover, this study assessed the cholephilic properties of MABAs, particularly whether these molecules are absorbed by the intestine and secreted into bile with effectiveness comparable to that of major conjugated BAs.

METHODS

Patients and sample collection

Bile samples from a cohort of 155 patients prescribed to undergo endoscopic retrograde cholangiopancreatography (ERCP) with a diagnosis of several hepatopancreatobiliary disorders were prospectively collected from January 2017 to December 2019 at Navarra University Hospital. A second group of 57 bile samples was collected during surgical procedures at Salamanca University Hospital from February 2020 to November 2024. These cohorts included patients with benign strictures, choledocholithiasis, pancreatic ductal adenocarcinoma (PDAC), extrahepatic cholangiocarcinoma (eCCA), intrahepatic cholangiocarcinoma (iCCA), gallbladder cancer (GBC), ampullary cancer (AC), primary sclerosing cholangitis (PSC), and HCC. All patients were older than 18 years and provided written informed consent for the examination of their samples and the use of their clinical data. All research was conducted in accordance with both the Declarations of Helsinki and Istanbul, and written consent was given by all subjects. The study protocol was approved by the ethics committees of both institutions (protocols #2016/91 and #PI2020-02-422).

Chemicals, cell cultures, and in vitro uptake assays

All information about chemicals used in this study is available in the Supplemental information, http://links.lww.com/HEP/J862. Cell sources and culture protocols, as well as the methods to carry out uptake studies following previously established protocols,^22–24 are available in “Detailed Methods” of the Supplemental information, http://links.lww.com/HEP/J862.

In vivo studies

MABA transfer from systemic blood to bile or from in situ isolated ileum to bile was investigated in anesthetized rats. The entry of D or L enantiomers into the enterohepatic circulation was investigated in mice after oral gavage. In vivo experiments were performed following previously established protocols,²⁵ which are fully described in “Detailed Methods” of the Supplemental information, http://links.lww.com/HEP/J862. Animal procedures were approved by the Ethical Committees for Laboratory Animals of the involved institutions, following European guidelines and in accordance with the NIH Guide for the Care and Use of Laboratory Animals.

HPLC–MS/MS analyses

Bile and cell lysates were diluted with methanol and centrifuged before their chromatographic analysis. The 16 most abundant species in human BA pool (tauro-conjugated, glyco-conjugated, and non-amidated primary and secondary BAs) and MABAs (Phe-CA, Leu-CA, and Tyr-CA) were analyzed by high-performance liquid chromatography–tandem mass spectrometry (HPLC–MS/MS) as previously reported²⁶ and have been described in “Detailed Methods” of the Supplemental information, http://links.lww.com/HEP/J862.

Molecular simulations

The amino acid sequences of human ASBT, NTCP, OATP1B1, and OATP1B3 were retrieved from the UniProtKB database. The SWISS-MODEL Workspace/GMQE tool identified structural templates for homology modeling. Specifically, the active-state crystallographic structure of ASBT was selected as a template for constructing agonistic conformations of ASBT and NTCP. For OATP1B1 and OATP1B3, the structure of the multidrug transporter MdfA was used. All procedures and software used for molecular simulation are described in “Detailed Methods” of the Supplemental information, http://links.lww.com/HEP/J862.

Statistical analyses

Post hoc analyses, such as paired or unpaired Student t tests, were applied appropriately to calculate the statistical significance of differences among groups. Differences were considered statistically significant at p<0.05. Microsoft Excel (version 15.32) and GraphPad (Prism5) were used for analyses. The Correl analysis was used to calculate the Pearson correlation coefficients.

RESULTS

MABA determination in human bile

Analysis of bile samples revealed the presence of MABAs formed by the conjugation of primary BAs, CA, and CDCA, with Leu, Phe, or Tyr in patients with hepatopancreatobiliary disorders (Figure 1). In nearly all samples measured at the University of Salamanca, where MABAs could be determined, the most abundant conjugates were those with Leu. Tyr-conjugated MABAs were the least prevalent (Figure 1). An independent analysis performed at the University of Maastricht on a subset of samples, which included additional bile samples from patients with PSC, confirmed these findings. Moreover, as this second laboratory incorporated the measurement of CDCA-based MABAs, it demonstrated a comparable abundance of MABAs derived from CA and CDCA.

FIGURE 1.

The circles depict the number and percentage of patients with (yellow sectors) and without (green sectors) measurable levels of microbially amidated bile acids (MABAs) determined in human bile by 2 different laboratories (Salamanca and Maastricht) (A). Molecular species of MABAs as determined by HPLC–MS/MS (B). Bars depict the proportion of molecular species of MABAs in bile samples where at least one of them was measurable (C). Abbreviations: AC, ampullary cancer; CA, cholic acid; CDCA, chenodeoxycholic acid; eCCA, extrahepatic cholangiocarcinoma; GBC, gallbladder cancer; HPLC–MS/MS, high-performance liquid chromatography–tandem mass spectrometry; iCCA, intrahepatic cholangiocarcinoma; Leu, leucine; PDAC, pancreatic ductal adenocarcinoma; Phe, phenylalanine; PSC, primary sclerosing cholangitis; Tyr, tyrosine.

The range of MABA concentrations in the positive samples (Figure 2) was similar in both laboratories and comparable to previously reported levels.²¹ Overall, the MABAs were measurable in approximately half of the patients, with only modest differences between the studied patient groups. Absolute levels of biliary MABAs showed considerable variation within patient groups, without significant differences among them (Supplemental Figure S1, http://links.lww.com/HEP/J862).

FIGURE 2.

Relationship in human bile between the concentration (A, C, E) and relative abundance (B, D, F) of microbially amidated bile acids (MABAs) and total bile acids (BAs) concentration (A, B), the conjugated/unconjugated ratio of major BAs (C, D), and primary/secondary ratio of major BAs (E, F). Values for MABAs were calculated by adding Tyr-CA plus Phe-CA and Leu-CA. Samples in which MABAs could not be accurately quantified were excluded from this analysis. Bile samples were from 73 patients with benign strictures (n=4), choledocholithiasis (n=8), pancreatic ductal adenocarcinoma (n=29), extrahepatic cholangiocarcinoma (n=9), intrahepatic cholangiocarcinoma (n=8), HCC (n=8), ampullary cancer (n=4), and gallbladder cancer (n=3). The best significant Pearson correlations were exponential: for A (y=60.678x ^0.350), B (y=60.669x ^−0.650), and D (y=167.050x ^−0.489). Abbreviations: CA, cholic acid; Leu, leucine; Phe, phenylalanine; Tyr, tyrosine.

Due to the lack of commercial compounds, only the predominant MABAs initially reported¹⁸ were synthesized and used as standards in HPLC–MS/MS determinations. Therefore, the assumption that the measured forms are representative of all MABAs may be incorrect for those patients considered to have a negative bile sample.

Regarding BAs produced by the liver, a significant positive correlation was observed between the total BA concentration and the ratio of conjugated to unconjugated major BA species (Supplemental Figure S2, http://links.lww.com/HEP/J862). The absolute abundance of MABAs was proportional to that of total BAs (Figure 2A). However, the relative proportion of MABAs showed an inverse relationship, being higher in samples with lower total BA concentrations (Figure 2B). Considering absolute amounts, a trend to an inverse relationship between the proportion of conjugated major BAs and the abundance of MABAs was observed (Figure 2C). This correlation was statistically significant when the relative proportion of MABAs was plotted versus the levels of conjugated major BAs (Figure 2D). In contrast, no significant relationship was detected between the abundance (Figure 2E) or the proportion (Figure 2F) of MABAs and the ratio of primary to secondary major BAs.

No significant correlations were found between MABA abundance and age at diagnosis; survival time in cancer patients; serum levels of fat-soluble vitamins A, D, and E; or the indirect assessment of vitamin K homeostasis, as determined by the coagulation-related parameter international normalized ratio (INR) (Figure 3).

FIGURE 3.

Relationship between the abundance of microbially amidated bile acids (MABAs) in bile samples from patients with various hepatopancreatobiliary disorders (benign strictures, choledocholithiasis, pancreatic ductal adenocarcinoma, extrahepatic cholangiocarcinoma, intrahepatic cholangiocarcinoma, HCC, ampullary cancer, and gallbladder cancer) and the age at diagnosis (A), survival time in cancer patients (B), and serum levels of liposoluble vitamins A, D, and E (C–E), and the international normalized ratio (INR), a coagulation index commonly used as an indirect indication of vitamin K homeostasis (F). Values are expressed as molecules of MABAs per million molecules of total bile acids (mpm). Note that n values vary because the clinical data were not available for all patients. The contingency table shows the number of patients treated with antibiotics before bile collection and whether MABAs were detected in their bile (G). Statistical analysis was performed using both the Chi-square test of independence and the Fisher exact test.

Due to the diversity of diseases included in this study, a non-uniform pharmacological treatment can be expected. Among the only 6 patients who received UDCA, MABAs were detected in the bile of 3 patients, but were absent in the other 3. Some patients (26%) had received prophylactic treatment with antibiotics (amoxicillin, amoxicillin-clavulanic acid, cefditoren, cefotaxime, ceftriaxone, cefuroxime, ciprofloxacin, metronidazole, norfloxacin, or piperacillin-tazobactam) before bile sampling. The percentage of samples with detectable MABAs was lower in patients receiving antibiotics (22% vs. 45%; p=0.007) (Figure 3G).

To obtain an approximate value of the proportion of MABAs versus total BAs in both bile and the intestine, we measured 6 paired samples of bile and feces from patients with PDAC. MABAs were detected in 3 fecal samples (average value =383 pmol/g wet weight). The mean amount of total BAs in these samples was 2.3 µmol/g wet weight. The proportion ranged from 70 to 3000 molecules of MABAs per million molecules of total BAs. The analysis of the 6 paired bile samples revealed the presence of MABAs in 3 samples. Interestingly, there was no strict correspondence between MABA status in paired fecal and bile samples; that is, MABA-positive bile samples were not precisely the ones paired with MABA-positive fecal samples. The average abundance of MABAs in bile was 4 molecules of MABAs per million molecules of total BAs, which suggests that the relative abundance of MABAs versus total BAs is markedly higher in feces than in bile.

In vivo studies

MABAs were not detected in systemic circulation at baseline in either mice or rats (data not shown). To evaluate the entry of D and L enantiomers into the enterohepatic circuit, L-Tyr-CA or D-Tyr-CA was administered via oral gavage to the mice. Generally, administration of L-Tyr-CA did not yield quantifiable levels (ie, <4.8 nmol/L) in the serum (Figure 4A). In contrast, D-Tyr-CA was consistently detected in systemic serum from the earliest time point after gavage, with average levels reaching values on the order of 100 nM (Figure 4A). Consistently, a higher concentration (in the mM range, representing 0.7% of total biliary BAs) of Tyr-CA in bile was observed after D-Tyr-CA gavage (Figure 4B). Other MABAs of endogenous origin, such as Leu-CA (Figure 4C) and Phe-CA (Figure 4D), constituted minor fractions of the biliary BA pool, and their abundance was unaffected by D-Tyr-CA or L-Tyr-CA administration. In contrast, a transient increase in plasma CA concentration was observed after gavage of L-Tyr-CA but not D-Tyr-CA (Figure 4E), suggesting cleavage of the L enantiomer but not the D enantiomer in the mouse intestine.

FIGURE 4.

Differential fate of D and L enantiomers of Tyr-CA in the enterohepatic circulation of mice (n=3 per group) receiving D-Tyr-CA (orange symbols) or L-Tyr-CA (blue symbols) via oral gavage. Systemic plasma concentrations of Tyr-CA (A) and cholic acid (CA) (E) were assessed by LC–MS in sequential blood samples. Tyr-CA (B), Leu-CA (C), and Phe-CA (D) concentrations were determined in gallbladder bile collected at the end of the experimental time (240 min) and are expressed as molecules of microbially amidated bile acids (MABAs) per million molecules (mpm) of total bile acids. In 2 animals per group, blood samples were collected from the portal and hepatic veins at the end of the experiments (4 h after gavage). The differential concentrations of Tyr-CA and total BAs between portal and hepatic venous plasma were determined 240 minutes after oral administration of D-Tyr-CA (F, G) or L-Tyr-CA (H, I); *p<0.05. Abbreviations: LC–MS, liquid chromatography–mass spectrometry; Leu, leucine; Phe, phenylalanine; Tyr, tyrosine.

Analysis of blood samples collected from the portal and hepatic vein at the end of the experiments (4 h after gavage) revealed a marked increase in portal Tyr-CA concentration after D-Tyr-CA gavage. The levels of D-Tyr-CA in the portal plasma averaged ~1.5 µM, constituting 6% of the total portal BAs. (Figure 4F). A portohepatic venous difference of more than 90% was apparent for D-Tyr-CA, indicating substantial hepatic extraction of the D enantiomer (Figure 4F). This proportion was similar to that determined for the endogenous total BAs (Figure 4G). In contrast, negligible levels of Tyr-CA in portal plasma were observed after L-Tyr-CA gavage (Figure 4H), despite normal concentrations in portal plasma and efficient hepatic extraction of total BAs (Figure 4I). Based on these results, and because the expected predominant enantiomers generated in the intestine are those formed by N-amidation with L-amino acids, only L-MABAs were used in subsequent studies.

When GCA was intravenously injected alongside MABAs (Tyr-CA, Phe-CA, and Leu-CA) at 0.5 µmol each into anesthetized rats, the liver promptly took up all 4 BAs and secreted them into bile with a very similar time course (Figure 5A). The magnitude of the induced biliary output during the first 10 minutes was similar for all MABAs and GCA (Figure 5B). The total bile recovery rate of administered MABAs was ~80% for each (Figure 5C). Continuous endogenous production of GCA by the rat liver throughout the experimental period hindered the accurate calculation of GCA recovery.

FIGURE 5.

Time course of the biliary output of GCA and the microbially amidated bile acids (MABAs), Tyr-CA, Phe-CA, and Leu-CA in rats, after i.v. administration of a mix containing 0.5 µmol of each one (A). Induced output for the first 10 minutes after bile acid (BA) administration (B). Total biliary recovery of exogenously administered MABAs for 50 minutes after injection (C). Values are mean±SEM of 5 separate experiments. *p<0.05 as compared to biliary output at minute 0 versus minute 10 for all BAs. N.S., no significant difference among groups. Abbreviations: CA, cholic acid; GCA, glycocholic acid; Leu, leucine; Phe, phenylalanine; Tyr, tyrosine.

When GCA, along with MABAs (Tyr-CA, Phe-CA, and Leu-CA), was administered (1 µmol each) to the in situ isolated ileum of anesthetized rats, the tested BAs were absorbed from the intraluminal fluid where they were added and subsequently recovered in the bile. As expected, the time course of biliary secretion was delayed compared to when the intestinal step was bypassed due to direct administration into the blood (Figure 5). Nevertheless, the pattern remained similar for all MABAs and GCA (Figure 6A). No significant difference was noted among the tested BAs regarding induced output during the first 60 minutes following their administration (Figure 6B). The total recovery over 90 minutes was similar for all MABAs, ~50%–60% of the administered amount (Figure 6C).

FIGURE 6.

Time course of the biliary output of GCA and the microbially amidated bile acids (MABAs), Tyr-CA, Phe-CA, and Leu-CA, after the intraileal administration of a mix containing 1 µmol of each one into the isolated ileum of anesthetized rats (A). Induced output for the first 60 minutes after bile acid (BA) administration (B). Total biliary recovery of exogenously administered MABAs for 90 minutes after injection (C). Values are mean±SEM of 5 separate experiments. *p<0.05 as compared to biliary output at minute 0 versus minute 20 for all BAs. N.S., no significant difference among groups. Abbreviations: CA, cholic acid; GCA, glycocholic acid; Leu, leucine; Phe, phenylalanine; Tyr, tyrosine.

In silico studies

To predict whether MABAs interact with the main BA uptake carriers, the structures of NTCP, ASBT, OATP1B1, and OATP1B3 were modeled and used to analyze by molecular docking their interactions with potential substrates in the pocket domain (Figure 7). The interactions of the amino acids that comprise this transporter region were similar for both major BAs and MABAs (Figure 7).

FIGURE 7.

Modeling of the predicted structure (green ribbons) of the main transporters involved in bile acid uptake, namely NTCP (A), ASBT (B), OATP1B1 (C), and OATP1B3 (D), as well as the interaction of their pocket region (gray ribbons) with TCA and the 3 most abundant CA-based microbially amidated bile acids (MABAs) found in human bile, specifically Leu-CA, Phe-CA, and Tyr-CA. The bar graphs depict the binding energy between the substrate and the transporter. Bile acids are ordered from top to bottom by increasing binding energy for each transporter, with colored bars representing major bile acids produced by the host. Abbreviations: ASBT, apical sodium-dependent bile acid transporter; CA, cholic acid; CDCA, chenodeoxycholic acid; GCA, glycocholic acid; Leu, leucine; NTCP, Na⁺-taurocholate cotransporting polypeptide; OATP, organic anion transporting polypeptide; Phe, phenylalanine; Tyr, tyrosine; TCA, taurocholic acid.

The transmembrane domains of ASBT and NTCP contain hydrophobic grooves within their binding pockets^27–29, which may facilitate binding of MABAs. This structural feature likely contributes to the high affinity of MABAs for these transporters, as indicated by their favorable binding energies. MABA-binding interactions closely resemble those of the major BAs, suggesting an effective interaction with sodium-dependent transporters. All assayed MABAs formed hydrogen bonds with NTCP residue N103, and residue N262 was near the hydroxyl group at position 3. Interestingly, H4 was located near the side chain region of the BAs, suggesting potential π-stacking interactions with the steroid ring of Phe-CA and Tyr-CA. The binding interactions with ASBT were similarly characterized by a hydrogen bond involving the hydroxyl group at position 3 and residue T110, with N266 in close proximity. Additionally, the hydroxyl group at position 7 forms a hydrogen bond with A111. In ASBT, the carbonyl group of the amide bond forms a hydrogen bond with W118, reinforcing the possibility of π-stacking interactions between Phe-CA, Tyr-CA, and the protein, analogous to the NTCP interactions.

The interactions of MABAs and the major BAs with OATP1B1 and OATP1B3 were comparable. The hydroxyl group at position 3 consistently formed a hydrogen bond with D236 in both transporters, whereas the anionic carboxylic acid group interacted with multiple hydrogen bond donors in the highly charged region. Additionally, in OATP1B1, the hydroxyl group at position 7 established an additional interaction with residue S231.

Furthermore, the calculated binding energy values were comparable for both major BAs and MABAs across all the assayed transporters. Nevertheless, the transporter exhibiting the highest binding energy values, reflecting a higher binding affinity, was NTCP (Figure 7A), followed by ASBT (Figure 7B), with both displaying values exceeding 10 kcal/mol for both major BAs and MABAs. In contrast, the binding energies of all the compounds to OATP1B1 (Figure 7C) and OATP1B3 (Figure 7D) were below this threshold.

Additionally, while our study primarily focuses on evaluating the interactions of L-series MABA derivatives, certain gut bacteria are known to produce D enantiomeric amino acids.³⁰ To determine whether this stereochemistry might influence transporter interaction, we conducted docking studies using the corresponding D-series MABA compounds (Supplemental Table S1, http://links.lww.com/HEP/J862). These in silico analyses revealed no marked differences in binding energies compared to their L-series counterparts, suggesting that the interaction and likely transport of D-MABAs would be essentially equivalent. Based on these results, we selected 2 major sodium-dependent BA transporters (NTCP and ASBT), along with 1 quantitatively less relevant sodium-independent BA transporter (OATP1B3), for subsequent in vitro evaluation of their ability to mediate the uptake of L-MABAs.

In vitro studies

To further compare the behavior of MABAs and major BAs in relation to their handling by hepatic and intestinal BA uptake transporters, in vitro experiments were conducted using cells that stably expressed the most relevant BA importers. The 2 most abundant CA-based species in human bile, Phe-CA and Leu-CA, were also tested. For comparison, GCA was used. Regarding sodium-dependent transporters, NTCP expression induced a marked and similar uptake of GCA (Figure 8A), Leu-CA (Figure 8B), and Phe-CA (Figure 8C). In all cases, coincubation with TCA at a 5-fold higher concentration resulted in a significant reduction in their uptake. However, inhibitory effects varied across species. Specifically, TCA-induced BA transport inhibition followed the order GCA>Leu-CA>Phe-CA. As the method used to quantify BAs in cell lysates allowed for the determination of TCA, this BA was also measured in cells incubated with TCA, along with GCA or MABAs. As expected, when NTCP-expressing cells were coincubated with GCA and TCA at a 1:5 ratio, their uptake was approximately proportional to this ratio. Interestingly, TCA exhibited a lower inhibitory effect on Leu-CA uptake, and TCA uptake was also reduced (Figure 8D). This effect was even more pronounced in the case of Phe-CA, which was less sensitive to TCA-induced inhibition. Moreover, the TCA uptake was significantly lower in the presence of Phe-CA (Figure 8D). These findings were consistent with the order of the binding energies of NTCP with these BAs (TCA<GCA<Leu-CA<Phe-CA) (Figure 7A).

FIGURE 8.

Uptake of GCA (A, E, I) and the 2 most abundant microbially amidated bile acids (MABAs) detected in human bile, that is, Leu-CA (B, F, J) and Phe-CA (C, G, K), by CHO cells transfected with empty vectors (Mock) or lentiviral vectors bearing the coding sequence of the 2 major sodium-dependent bile acid (BA) transporters NTCP (A–D) and ASBT (E–H), and the minor sodium-independent bile acid transporter OATP1B3 (I–L). The cells were incubated with 50 µM of the indicated BA without or with 250 µM taurocholate (TCA). The uptake after 60 minutes incubation was measured by HPLC–MS/MS after 60 minutes incubation. In cells coincubated with TCA, the uptake of this BA was also measured (D, H, L). Values are mean±SEM of 6 incubations carried out in 3 separate cell culture experiments. *p<0.05, by paired t test. Abbreviations: ASBT, apical sodium-dependent BA transporter; CA, cholic acid; CHO, Chinese hamster ovary; GCA, glycocholic acid; HPLC–MS/MS, high-performance liquid chromatography–tandem mass spectrometry; Leu, leucine; NTCP, Na⁺-taurocholate cotransporting polypeptide; OATP, organic anion transporting polypeptide; Phe, phenylalanine; Tyr, tyrosine.

The expression of the major transporter responsible for intestinal BA uptake, ASBT, enhanced the uptake of GCA (Figure 8E), Leu-CA (Figure 8F), and Phe-CA (Figure 8G). This effect was less pronounced than that of NTCP and was partially prevented by coincubation with TCA, which was itself taken up by ASBT (Figure 8H). Cross-competition with TCA resulted in varying degrees of TCA-induced inhibition, which was similar for GCA and Leu-CA; however, the least pronounced effect was observed for Phe-CA. Conversely, the highest TCA uptake occurred when coincubated with GCA, whereas the lowest uptake was observed with Phe-CA (Figure 8H). These results were also consistent with the highest binding energy value found for the interaction of Phe-CA with ASBT (Figure 7B).

For comparison, a quantitatively less relevant carrier, the sodium-independent BA transporter OATP1B3, was also included in this study. The uptake of GCA was enhanced in cells expressing OATP1B3 (Figure 8I); however, this effect was less than 8% of that observed in cells expressing NTCP (Figure 8A). The magnitude of OATP1B3-induced uptake of Phe-CA (Figure 8J) and Leu-CA (Figure 8K) was much smaller than that mediated by NTCP and ASBT. However, uptake was inhibited by TCA. Furthermore, both Phe-CA and Leu-CA inhibited TCA uptake (Figure 8L). In the case of OATP1B3, the highest binding energy was that of Phe-CA (Figure 7D), which was the least sensitive to TCA-induced uptake inhibition and induced the strongest inhibition of TCA uptake (Figure 8L).

DISCUSSION

Since their discovery in human and mouse bile just a few years ago, the list of microbially conjugated BAs has grown to include over 100 distinct species.^19,31 Among N-amidated BAs, conjugated with amino acids other than taurine and glycine, such as the amino acids Glu, Ile, Leu, Phe, Thr, Trp, and Tyr have been identified.^20,32–34 This study confirmed that the most abundant MABAs in human bile, as previously described in samples from the gastrointestinal tract^18,35 and bile,²¹ were BAs amidated with Leu and Phe. We used the combined levels of Leu-CA and Phe-CA, and the less abundant Tyr-CA, as a representative value of the total MABA levels. However, we cannot rule out the presence of other MABAs in samples classified here as negative.

Our findings showed that biliary MABA levels were extremely low compared to liver-derived conjugated BAs. Moreover, MABA detection rates declined in patients who had received antibiotic treatment. Conversely, their relative abundance in bile increased as total BA concentrations decreased, suggesting a weakened capacity of the bile environment to suppress gut microbial growth. This is consistent with the fact that hepatopancreatobiliary disorders are commonly accompanied by impaired bile flow into the intestine, which implies a reduction in the well-documented bactericidal effect of BAs.^36,37 Unconjugated BAs are more effective antimicrobials than their conjugated counterparts, primarily due to their ability to disrupt bacterial membranes, leading to the leakage of cellular contents.³⁸ In contrast, conjugated primary BAs exhibit a lower antimicrobial effect.^38,39 Accordingly, microbial BA conjugation has been suggested as a mechanism for protection from BA toxicity, and the bactericidal activity can vary based on amino acid conjugation. In this sense, distinct effects have been observed between hydrophobic conjugates (Phe-CA and Leu-CA) and major BA conjugates produced in the liver, such as GCA or TCA.¹⁸

Under impaired biliary function conditions, such as those commonly present in patients with hepatopancreatobiliary disorders, significant changes in the gut microbiota, comprising trillions of microorganisms, have been reported.⁴⁰ Cholestasis has been associated with reduced microbial diversity, an increase in potential pathogens, and a decrease in beneficial species.⁴¹ Moreover, the presence of biliary MABAs in patients with pancreatic cancer was not surprising because, in addition to the biliary stricture that can be caused by the tumor, an impaired exocrine pancreatic secretion can alter intestinal microbiota. This dysbiosis resembles that seen in patients with cystic fibrosis, with insufficient pancreatic lipase production resulting in the accumulation of non-absorbed fat in the gut.⁴² Additionally, many patients with hepatopancreatobiliary disorders commonly are on complex drug regimens, further destabilizing gut microbiota and altering BA metabolism, pool size, and composition.⁴³ Supporting the potential relationship between changes in intestinal microbiota and the biliary appearance of MABAs is that fecal levels of these metabolites are elevated in the dysbiotic state of Crohn disease but not in ulcerative colitis.¹⁸

Interestingly, the appearance of MABAs is not a universal feature across all patients with hepatopancreatobiliary disorders, suggesting that additional factors influence their generation beyond microbiota alterations.

In addition to various other microbially performed BA transformations,⁴⁴ the predominant metabolic changes in primary BAs result from the activity of HSDs, which remove the 7α-OH group, leading to secondary BAs, along with deamidation of the side chain, resulting in unconjugated BAs. Our results suggested that the ability of the intestinal microbiota to conjugate BAs with Leu, Phe, or Tyr is not directly linked to the production of secondary BAs. This may reflect differences in the abundance of bacterial species capable of each transformation.³⁵ Conversely, we found a correlation between deamidation capacity and MABA formation. Recent reports attribute MABA synthesis to the amine N-acyltransferase activity of BSH,^35,45 which is ubiquitously expressed in the human gut microbiome.⁴⁶ Beyond its known deconjugating activity,¹⁷ BSH can also conjugate BAs with various amino acids. This may explain the inverse relationship we observed in bile between MABA concentrations and the conjugated/unconjugated BA ratio. The acyltransferase activity of BSH varies among bacterial species and is influenced by environmental pH, affecting amino acid-specific conjugation across different gut segments and microbial communities.³⁵

A crucial question addressed by this study was whether MABAs are generated as an indirect consequence of the underlying hepatopancreatobiliary disorder, such as benign strictures or various hepatopancreatobiliary cancers, since these conditions can affect microbiota composition and metabolism. However, the capacity to generate MABAs was not uniformly present across disease types, suggesting that while these conditions promote dysbiosis, other factors determine the emergence of this conjugating activity.

As for the potential impact of MABAs on host physiology, previous in vitro studies have demonstrated that the interaction of MABAs with BA receptors is equipotent to that of major BAs. This finding, along with the dramatically lower levels of MABAs compared to other molecular species present in the BA pool, suggests that it is unlikely for MABAs to have a relevant role in the enterohepatic TGR5/FXR signaling in humans.²¹ The present study demonstrates that MABAs reach cellular targets in the host and behave as major BAs in entering the enterohepatic circuit. This confirms previous indirect evidence,²¹ suggesting that MABAs interact with BA transporters like BA conjugated with glycine or taurine.

The ileal epithelium takes up BAs, which subsequently reach the liver, where they are efficiently extracted and secreted into bile. Docking studies have indicated that MABAs can interact effectively with crucial BA transporters, exhibiting interaction energies comparable to those of the most abundant primary BAs. This predicted that MABAs are mainly transported by sodium-dependent carriers (NTCP and ASBT), with a minor contribution from organic anion transporting polypeptides (OATP1B1 and OATP1B3). In vitro studies have confirmed these predictions. NTCP and ASBT displayed the highest ability to transport Leu-CA and Phe-CA. However, they differed in their sensitivity to inhibition by TCA (Leu-CA>Phe-CA), and MABAs inhibited NTCP- and ASBT-mediated TCA transport with different efficacies (Phe-CA>Leu-CA).

Despite these transport differences, all MABAs were secreted into bile at similar rates following i.v. administration in rats, comparable to GCA. This suggests that BSEP-mediated canalicular secretion functions efficiently for MABAs as well. No significant differences in biliary output were observed following ileal administration, indicating that variations in ASBT interaction did not impact MABA recirculation, at least at the concentration used in this study.

We previously demonstrated that MABAs can be degraded in vitro by pancreatic carboxypeptidases and BSH from Clostridium perfringens.²⁰ Quantitative information on MABAs in human plasma is scarce, but it suggests sub-nanomolar MABA concentrations in mesenteric, portal, and systemic blood.²¹ Thus, enzymatic degradation in the gut likely limits MABA entry into the enterohepatic circulation. However, the readily detectable presence of MABAs in human bile may reflect the pronounced concentration that BAs achieve after portal extraction and subsequent canalicular secretion. The results obtained in mouse experiments clarified this point. Marked differences were seen depending on whether the animals received oral gavage of L-Tyr-CA or D-Tyr-CA, susceptible or resistant to enzymatic breakdown in vitro. Systemic appearance was only relevant in the case of D-Tyr-CA, with further support for the efficient hepatic extraction and biliary secretion of D-Tyr-CA. Conversely, the systemic entry of L-Tyr-CA was minimal, whereas the transient elevation of systemic CA levels was aligned with the luminal degradation and subsequent absorption of unconjugated CA. Under non-manipulated diets in humans and rodents, amino acids available in the intestinal lumen for bacterial re-amidation of deconjugated BAs are expected to be primarily L enantiomers. Accordingly, the low amount of MABAs found in bile is due to the low amidation activity of the microbiota compared to that of the liver, and the susceptibility of L-MABAs to the deconjugating activities present in the intestinal lumen. Whether D-amino acids also serve as natural substrates for MABA formation remains unclear.

In conclusion, MABAs are transferred in small amounts from the gut to bile via the same mechanisms accounting for the enterohepatic circulation of major conjugated BAs. Their presence in the bile of certain patients with hepatopancreatobiliary dysfunction likely reflects microbiota shifts that favor bacteria capable of producing these rare BA derivatives.

Supplementary Material

hep-83-1158-s001.pdf ^{(490KB, pdf)}