Bile salt hydrolase catalyses formation of amine-conjugated bile acids – Nature

Correlation of BSH and ability to produce BBAAs

To check for the association of BSH and BBAA production, we leveraged liquid chromatography with tandem mass spectrometry (LC–MS/MS) data from two published datasets involving 202 Human Microbiome Project (HMP) bacterial isolates in ref. ³ and 69 representative gut bacteria members in ref. ². We combined the two datasets. For comparison between bacteria from the studies, we normalized the dataset on the basis of the proportion of BBAAs detected from the total number of BBAAs tested in each study. Next, we downloaded the assembled genomes of the bacteria from the National Center for Biotechnology Information (NCBI) genome database. A phylogenetic tree was obtained from the NCBI taxonomy database using taxids in the study and visualized using the ggtree²⁸ package in R. We annotated the genomes using Prokka²⁹ and checked for the presence or absence of the bsh gene. Briefly, Prokka uses several databases (Uniprot, RefSeq and Pfam) in a hierarchical manner to annotate coding sequences in the genome. Genomes with coding sequences annotated as choloylglycine hydrolase (cbh) or having its domain were deemed as having the bsh gene or its homologue, whereas genomes lacking these annotations were deemed lacking the gene. To test for correlation, we used the phylolm³⁰ package in R to fit a following phylogenetic linear regression model:

where BBAAs is the vector of proportions of the BBAAs detected for bacteria in each study and bsh is the vector of presence or absence of the bsh gene.

RNA sequencing sample preparation and analysis

B. longum subsp. longum NCTC 11818 was acquired from ATCC, and 30 µl of an overnight culture of B. longum was inoculated into 3 ml of brain–heart infusion (BHI) broth (BD Difco) supplemented with 0.05% w/v cysteine, 5 µg ml⁻¹ hemin and 1 µg ml⁻¹ vitamin K (Sigma). Equivalent volumes of 1 M NaOH (control) or CA and DCA in 1 M NaOH at a final concentration of 200 µM were added. Cultures were grown to mid-exponential phase (approximately 6 h) at 37 °C in an anaerobic chamber filled with 20% CO₂ and 5% H₂ in N₂ gas. Then 4 ml of RNAprotect (Qiagen) was added to 2 ml culture and incubated for 5 min at room temperature. Samples were centrifuged at 5,000g for 5 min at 4 °C, and the bacterial pellets were flash-frozen in liquid nitrogen and stored at −80 °C. Pellets were resuspended in 1 ml TRIzol Reagent and transferred to Lysing Matrix E tubes for homogenization at 2 × 20 s cycles at 6,500g. Next, 200 µl chloroform was added, vortexed for 15 s and incubated at room temperature for 10 min. Samples were centrifuged for 15 min at 16,000g, 4 °C, and 500 µl upper aqueous phase was transferred to a fresh tube containing 500 µl of 100% ethanol. Samples were then transferred to a Purelink spin column, centrifuging for 15 s at 12,000g to elute flowthrough. Columns were washed with 350 µl Wash Buffer I and removed by centrifugation for 15 s at 12,000g. Contaminating DNA was degraded by adding 80 µl of deoxyribonuclease (DNase) solution (10 µl DNase in 1× reaction buffer) and incubating at room temperature for 15 min. A second wash with Wash Buffer I was then performed. Columns were washed twice with 500 µl Wash Buffer II, eluting each time by centrifugation for 15 s at 12,000g. Purified RNA was eluted with 30 µl RNase-free water by centrifugation for 1 min at 12,000g into a fresh tube.

RNA samples were submitted to the Microbial Genome Sequencing Center for sequencing. Samples were DNase treated with Invitrogen DNase (RNAse-free). We prepared the library using Illumina’s Stranded Total RNA Prep Ligation with Ribo-Zero Plus Kit and 10 bp IDT for Illumina indices. Sequencing was done on a NextSeq2000 to yield 2 × 50 bp reads. Demultiplexing, quality control and adaptor trimming were performed with bcl-convert (v.3.9.3).

For the analysis, the complete reference genome and annotation of B. longum NCTC 11818 were obtained from the NCBI Assembly database. Obtained demultiplexed reads were checked for quality and filtered using fastp (v.0.12.4) (ref. ³¹). Filtered reads were aligned with the reference genome using bowtie2 (v.2.2.5) (ref. ³²). Obtained alignments were sorted using samtools (v.1.10) (ref. ³³). Qualimap (v.2.2.2) (ref. ³⁴) was used to check for the quality of alignments. featureCounts (v.2.0.1) (ref. ³⁵) was used to generate counts of reads uniquely mapped to annotated genes. These raw counts were then normalized and analysed for differential gene expression using the DESeq2 package (v.1.35.0) (ref. ³⁶) in RStudio.

qScript cDNA SuperMix (Quantabio) was used for the reverse transcription of 250 ng RNA template in a 20 µl reaction for RT-qPCR. Primer sequences were designed using PrimerBlast for B. longum bsh and lactate dehydrogenase (ldhl) (Supplementary Table 2). We performed the RT-qPCR with 1 µl complementary DNA (cDNA) template and 0.4 µM of each primer on a QuantStudio 3 using fast cycling parameters with PowerUp SYBR detection reagent.

B.
longum BSH inhibition

B. longum subsp. longum NCTC 11818 was acquired from ATCC, and 30 µl of an overnight culture of B. longum was inoculated into 3 ml of BHI broth (BD Difco) supplemented with 0.05% w/v cysteine, 5 µg ml⁻¹ hemin and 1 µg ml⁻¹ vitamin K (Sigma). Next, 2 µl of the overnight culture was added to 196 µl fresh supplemented-BHI media, and 1 µl of 1 M NaOH (vehicle) or 20 mM CA and DCA in 1 M NaOH at a final concentration of 100 µM were added to the plate. To inhibit BSH, 1 µl of 20 mM GR-7 in dimethyl sulfoxide (DMSO) was added to the treatment wells to a final concentration of 100 µM. Equivalent volume of DMSO was added to the control wells. To monitor the production of BBAAs, samples were taken out at 4, 6, 8, 10, 14, 16 and 24 h after incubation and flash-frozen in liquid nitrogen until extraction for LC–MS/MS quantification of bile acids.

Heterologous expression and purification of B.
longum BSH in E. coli

The gene for B. longum BSH (NC_015067.1, codon-optimized) cloned into pET-28b(+) using NdeI and XhoI restriction sites was synthesized by GenScript. The resulting plasmid was transformed into E. coli Tuner (DE3) cells (Novagen) by means of electroporation at 2,500 V, and positive transformants were selected on Luria–Bertani media containing 50 µg ml⁻¹ kanamycin. Expression strains were grown in yeast extract media (45 g yeast extract (Research Products Int.), 1.6 g KH₂PO₄ (Research Products Int.), 13.0 g K₂HPO₄ (Research Products Int.) and 1% glycerol (v/v) (Sigma-Aldrich) per 1 l) at 37 °C to late exponential phase (optical density at 600 nm (OD₆₀₀) = 0.8). The cells were cooled to 18 °C before induction with 0.1 mM isopropyl β-d-1-thiogalactopyranoside (Research Products Int.) and allowed to express BSH for 18–20 h before being collected by centrifugation (4,000g at 4 °C, 20 min). The resulting cell pellets were frozen at −80 °C until use.

The cells were resuspended in buffer A (50 mM Tris (Research Products Int.), 300 mM NaCl (Research Products Int.), 20 mM imidazole (Sigma-Aldrich), pH 7.4) with protease inhibitors benzamidine HCl (Research Products Int.) and Pefabloc SC (Sigma-Aldrich) and disrupted by means of sonication (QSonica Sonicator Q500) for 6 min on ice (30 s pulse, 30 s pause, at 60% amplitude). The resulting lysate was centrifuged at 130,000g in a Beckman Optima L-90X ultracentrifuge at 4 °C for 1 h.

The following purification steps were performed at 4 °C. Supernatant was applied to a HisPur Ni-column (Fisher Scientific) pre-equilibrated with buffer A. The column was washed with 20 column volumes of buffer, and then the BSH protein was eluted with 5 column volumes of buffer B (50 mM Tris, 300 mM NaCl, 250 mM imidazole, pH 7.4). The eluent was dialysed into the final storage buffer (50 mM Tris, 50 mM NaCl, 5% glycerol (v/v), pH 7.0) overnight at 4 °C. The resulting protein solutions were concentrated using Amicon Ultra 15 ml centrifugal filters (Millipore) with a molecular weight cutoff of 10 kDa, flash-frozen in liquid nitrogen and stored at −80 °C.

In vitro purified enzyme assay

Purified BlBSH enzyme was diluted to 10 mg ml⁻¹ in phosphate buffer (pH 7.0). An equimolar mixture of 21 amino acids was prepared in H₂O (5 mM each of l-alanine, l-arginine, l-asparagine, l-aspartic acid, l-cysteine, l-glutamine, l-glutamic acid, glycine, l-histidine, l-isoleucine, l-leucine, l-lysine, l-methionine, l-phenylalanine, l-proline, l-serine, l-threonine, l-tryptophan, l-tyrosine, l-taurine and l-valine). Each reaction mixture contained 178 µl phosphate buffer (pH 6.0), 10 µl of 5 mM amino acids mixture, 2 µl of 100 mM bile acid (CA or TCA in DMSO) or vehicle (DMSO) and 10 µl of the 10 mg ml⁻¹ enzyme. The reaction mixture was incubated for 90 min at 37 °C. The reaction was stopped by adding 800 µl ice-cold 100% methanol containing 0.5 µM deuterated bile acid internal standards.

Construction of B. fragilis NCTC 9343 Δbsh mutant

A 1,797 bp DNA fragment containing the N-terminal region of the BSH gene was PCR amplified from the B. fragilis NCTC 9343 using primers BSH-NT-FOR and BSH-NT-REV. The N-terminal fragment was cloned into the SphI/BamHI sites of the pFD516 vector. Then a 1,734 bp DNA fragment containing the C-terminal region was PCR amplified with primers BSH-CT-FOR and BSH-CT-REV2 and cloned into the SacI site of pFD516. The new construct, pER363, contains a deletion of 765 bp DNA fragment from the bsh gene. A 3,533 bp BglII/SalI DNA fragment from pER363 containing the Δbsh construct was cloned into the BamHI/SalI sites of the suicide vector for allelic replacement in Bacteroides, pLGB13 to construct pER-366. pER-366 was mobilized from E. coli S17-1 λpir into B. fragilis NCTC 9343 by biparental mating as described previously. Transconjugants were selected on brain–heart infusion supplemented (BHIS) plus 0.5% yeast extract plates containing 200 μg ml⁻¹ gentamycin and 10 μg ml⁻¹ erythromycin. Four transconjugant colonies were grown overnight on BHIS broth with 200 μg ml⁻¹ gentamycin (erythromycin was omitted). Then 10 μl of an overnight culture was spread on BHIS with 0.5% yeast extract plates containing 200 μg ml⁻¹ gentamycin and 100 ng ml⁻¹ anhydrotetracycline. Transconjugant colonies were streaked on fresh BHIS with 0.5% yeast extract plates containing 200 μg ml⁻¹ gentamycin and 100 ng ml⁻¹ anhydrotetracycline. PCR amplification using BSH-mutcheck-FOR and BSH-mutcheck-REV primers was used to identify chromosomal double-crossed over genetic recombination. The B. fragilis NCTC 9343 Δbsh mutant isolates (BER-279) were also tested for loss of erythromycin resistance and respiratory test.

Genetic complementation of Δbsh mutant

pER-300 containing bsh operon in the pNBU2-bla-ermGb vector was mobilized into BER-279 by biparental mating. Transconjugants were selected on BHIS plus 0.5% yeast extract plates containing 200 μg ml⁻¹ gentamycin and 10 μg ml⁻¹ erythromycin (BER-280).

Site-directed mutagenesis of BlBSH active site amino acid residues

The pET-28b(+)-BlBSH vector constructed above was mutated using the Q5 Site-Directed Mutagenesis Kit according to the manufacturer’s instructions (New England Biolabs). The NEBaseChanger tool was used to design primers (Supplementary Table 2) and calculate the appropriate annealing temperatures for PCR. Following PCR mutagenesis and treatment with kinase, ligase and DpnI, the products were transformed into NEB 5-alpha-competent E. coli by heat shock at 42 °C, and positive transformants were selected on Luria–Bertani media containing 50 µg ml⁻¹ kanamycin. Individual colonies were selected, and successful mutations were confirmed by QIAprep Spin Miniprep kit (Qiagen) preparation of plasmid DNA followed by Sanger sequencing from the T7 promoter region. Positively identified plasmid DNA was stored at −20 °C. To express the proteins, mutant pET-28B(+)-BlBSH plasmids were transformed into E. coli BL21(DE3) cells by electroporation at 2,500 V, and positive transformants were selected on Luria–Bertani media containing 50 µg ml⁻¹ kanamycin. To measure TCA deconjugation, transformed E. coli were cultured in M9 media supplemented with 1 mM TCA. To measure CA conjugation, transformants were cultured in Luria–Bertani media supplemented with 1 mM CA. All transformants were grown at 37 °C and 250 rpm until OD₆₀₀ ≈ 0.6 was reached; then protein was induced by the addition of 0.1 M isopropyl β-d-1-thiogalactopyranoside and grown for a further 4 h. E. coli were extracted for LC–MS/MS according to the protocol below.

Extraction and LC–MS/MS quantification of bile acids from bacterial cultures

For bacterial cultures, 200 µl of the culture following growth was lysed by three freeze–thaw cycles in liquid nitrogen. Next, 150 µl of each lysate was transferred to a 1 ml capacity 96-well plate, and 600 µl of ice-cold 100% high-performance liquid chromatography (HPLC)-grade methanol containing 0.5 µM deuterated bile acid internal standards was added and mixed thoroughly by pipetting. The plate was then incubated on ice for 20 min. Samples were sonicated for 10 min in a sonicating water bath and then centrifuged for 15 min at 2,000 rpm at 4 °C. Then 200 µl of each supernatant was transferred to a fresh 1.5 ml microcentrifuge tube and dried using a speedvac. Dried extracts were resuspended in 200 µl HPLC-grade methanol and sonicated for 10 min before transfer to autosampler vials for LC–MS/MS.

Quantitative LC–MS/MS analyses were performed using a Waters ACQUITY Ultra-performance liquid chromatography (UPLC) system coupled with a Waters Xevo TQ-S Triple Quadrupole mass spectrometer (Waters). Chromatographic separation was achieved using an ACQUITY BEH C8 (2.1 × 100 mm, 1.7 µm) UPLC column (Waters) heated to 60 °C. Samples (1 µl) were injected into the column and eluted with 90% mobile phase A (1 mM ammonium acetate in 9% acetonitrile, pH 4.15) and 10% mobile phase B (1:1 acetonitrile:isopropanol) for 0.1 min. Mobile phase B was increased in consecutive linear gradients from 10% to 35% over 9.15 min, 35% to 85% B over 2.25 min and 85% to 100% B over 0.3 min. Then 100% B was held for 0.6 min followed by a linear gradient to 10% B over 0.1 min and held for 2.5 min, for a total of 15 min. The flow rate was modified throughout the run starting with 300 µl per min for 9.25 min and then increasing to 325 µl per min over 2.25 min and 500 µl per min over 0.6 min. The flow rate was held at 500 µl per min for 0.3 min and then reduced to 300 µl per min over 0.4 min and held for 2.2 min. MS analyses were carried out using electrospray ionization in positive ion mode using the following settings: capillary voltage 3.5 kV, cone voltage 20 V, source temperature 150 °C, desolvation temperature 300 °C, desolvation gas flow 540 l h⁻¹ and cone gas flow 150 l h⁻¹. BBAAs were identified by multiple reaction monitoring transitions from parent [M + H]⁺ ions to daughter [M + H]⁺ ions corresponding to the amino acid fragment (Supplementary Table 3). Peak identification and processing was done using MassLynx software (Waters). Peak areas for each conjugated bile acid were normalized to peak areas from deuterated internal standards of GCA-d4 or GDCA-d4, matching the bile acid. Quantification was performed against a standard curve of authentic standards at eight concentrations ranging from 7.81 nM to 1 µM.

Monocolonization of GF mice

Fifteen male 8- to 10-week-old C57BL/6 J mice were maintained under germ-free conditions in positive-pressure isolators (Class Biologically Clean). Gnotobiotic experiments were conducted under the Pennsylvania State University Institutional Animal Care and Use Committee approved protocol 202101826. Mice were singly housed with ad libitum access to water and food. Sterility of mice was ensured by qPCR and aerobic/anaerobic culture with rich media. Mice were administered 10⁸ CFUs of either B. fragilis NCTC 9343 or the ∆bsh mutant, with mice monitored daily. No adverse effects on mouse weight or behaviour were observed. After 7 days of colonization, mice were euthanized and ileal and caecal tissue and contents collected. To quantify B. fragilis colonization, faecal samples were resuspended in phosphate-buffered saline (PBS) and serially diluted before quantification on BHI plates using the drop-plate method.

Extraction and untargeted LC–MS/MS profiling of monocolonized mice

Bile acids were extracted from mouse ileal contents according to ref. ³⁷. Briefly, 25 mg of ileal contents were weighed in 1.5 ml screw-cap homogenization tubes, and approximately 50 µl of 0.1 mm zirconia beads were added. Next, 1 ml of ice-cold methanol containing 0.5 µM deuterated internal bile acid standards was added. Samples were then thoroughly homogenized in Precellys 24 tissue homogenizer at 20 s × 2 at 6,500 rpm. Homogenized samples were then freeze–thawed three times in liquid nitrogen and centrifuged at 15,000 rpm at 4 °C. Then 200 µl of the supernatant was transferred to autosampler vials for LC–MS/MS analysis.

We performed LC–MS/MS analyses using a Thermo Fisher Scientific Orbitrap Exploris 120 Mass Spectrometer system coupled with a Vanquish ultra-high-performance liquid chromatography system (Thermo Fisher Scientific). Chromatographic separation was carried out using an ACQUITY BEH C18 UPLC column (2.1 × 100 mm, 1.7 µm; Waters). The column was maintained at a temperature of 60 °C. Mobile phase solvent A was a mixture of 9% acetonitrile and 1 mM ammonium acetate with a pH of 4.15, and solvent B was a 1:1 ratio of acetonitrile and isopropanol. Samples (5 µl) were injected into the column, and the initial mobile phase was 10% B, kept for 0.1 min. Mobile phase B increased in consecutive linear gradients to 35% at 9.25 min and to 85% B at 11.5 min and was held at 85% for 3.4 min. It then decreased to 10% B at 15.1 min and was held at 10% B for 2.9 min. The flow rate of the mobile phase was 0.3 ml min⁻¹. Heated electrospray ionization was used with the following settings: spray voltages of 3.5 kV for positive charge and 2.5 kV for negative charge, sheath gas at 50 Arb, auxiliary gas at 10 Arb, sweep gas at 1 Arb, ion transfer tube temperature 325 °C and vaporize temperature 350 °C. MS settings included an Orbitrap resolution of 120,000, a scan range (m/z) from 150 to 1,200 and an RF lens of 60%. MS2 spectra were data-dependently acquired using higher-energy collisional dissociation with normalized collision energies of 20%, 35% and 50% and a resolution of 60,000. Data were analysed using MS-DIAL (v.4.9.221218).

Amino acid quantification of caecal content by ¹H NMR

Roughly 50 mg of caecal content was weighed in a 2.0 ml screw-cap microcentrifuge tube and mixed with 1.2 ml of 0.1 M phosphate buffer (K₂HPO₄:NaH₂PO₄ = 4:1, pH = 7.4) containing 50% D₂O and 0.005% (w/v) of sodium 3-(trimethylsilyl) propionate-2,2,3,3-d4 as internal standard. Samples were homogenized and then subjected to two freeze–thaw cycles in liquid nitrogen. After centrifuging at 17,000g for 10 min at 4 °C, 550 µl of clear supernatant was transferred to 5 mm NMR tubes for NMR analysis.

The one-dimensional ¹H spectra of caecal content extracts were recorded at 298 K by using a Bruker Avance NEO 600 MHz NMR spectrometer (Bruker Biospin) equipped with a 5 mm TCI cryoprobe with enhanced sensitivity for ¹H. The noesygppr1d pulse sequence was used, and spectral acquisition parameters were as follows: 64 scans were collected into 64,000 data points using 90° pulses, four dummy scans, 20 ppm spectral width. All spectra were analysed with Chenomx NMR Suite (v.9.02). After automatic processing, the phase, baseline and internal standard were checked and modified manually for each spectrum for quality assurance. The amino acids were identified and quantified using a built-in metabolite library.

Cell-based reporter assays

Reporter assay kits for human farnesoid X receptor (hFXR, cat. no. IB00601); vitamin D receptor (hVDR, IB00701); constitutive androstane receptor variant 3 (hCAR3, IB00901); pregnane X receptor (hPXR, IB07001); aryl hydrocarbon receptor (hAhR, IB06001); peroxisome proliferator-activated receptors alpha (hPPARα, IB0011), beta/delta (hPPARβ/δ, IB0012) and gamma (hPPARγ, IB0010); and mouse FXR (mFXR, M0060) and PPAR alpha (mPPARα, M0011), beta/delta (mPPARβ/δ, M0012) and gamma (mPPARγ, MR0010) were purchased from INDIGO Biosciences, Inc. Agonism assays were performed with test compounds at a single concentration (50 µM) with a concurrent dose-response of reference ligands following the manufacturer’s instructions. Briefly, each reporter assay cell suspension was thawed by the addition of cell recovery medium (CRM), and 100 µl of the cell suspension were used per well. Then 100 µl of compound screening medium supplemented with test compounds were added and incubated at 37 °C, 5% CO₂ for 24 h. The next day, the medium was removed, detection reagents were added and luminescence was measured (Synergy H4 Hybrid Multi-Mode Microplate Reader). Data were expressed relative to vehicle-treated cells and log₁₀-transformed for visualization. Differences between treatments were determined using analysis of variance (ANOVA) followed by Dunnett’s post hoc test (JMP Pro 15, SAS Institute). Significant differences were determined when P < 0.05. Non-linear regression and EC50 calculations were performed with Prism v.6.0 (GraphPad Software, Inc.).

Enteroid model

Normal (de-identified) human small intestinal samples were established and cultured in high-Wnt-containing medium (human colonoid medium) as previously described³⁸. Enteroids were treated with 50 µM and 100 µM Glu-CA or Glu-CDCA for 18 h, the medium was removed, and enteroids were washed twice and collected in PBS. Matrigel and PBS were carefully removed, and enteroids were lysed and extracted using a PureLink RNA mini kit according to the manufacturer’s instructions. qScript cDNA SuperMix (Quantabio) was used for the reverse transcription of 500 ng RNA template in a 40 µl reaction for RT-qPCR. We performed the RT-qPCR with 1 µl cDNA template and 0.4 µM of each primer (Supplementary Table 2) on a QuantStudio 3 using fast cycling parameters with PowerUp SYBR detection reagent.

IGRAM cohort

African American women in their third trimester of pregnancy were enroled as part of a prospective, longitudinal, observational maternal–infant cohort study (Infant Growth and Microbiome (IGRAM))²⁵. The study was approved by the Institutional Review Board at Children’s Hospital of Philadelphia. Stool samples from infants (56 male and 52 females) were collected at birth, 1 month and 12 months of life.

LC–MS/MS quantification of bile acids in IGRAM stool samples

A separate method was used for the quantification of bile acids in the IGRAM cohort. Bile acids were extracted from infant stool according to ref. ³⁷. Briefly, 25 mg of stool was weighed in 1.5 ml screw-cap homogenization tubes, and approximately 50 µl of 0.1 mm zirconia beads were added. Next, 1 ml of ice-cold methanol containing 0.5 µM deuterated internal bile acid standards was added. Samples were then thoroughly homogenized in Precellys 24 tissue homogenizer at 20 s × 2 at 6,500 rpm. Homogenized samples were then freeze–thawed three times in liquid nitrogen and centrifuged at 15,000 rpm at 4 °C. Then 200 µl of the supernatant was transferred to autosampler vials for LC–MS/MS analysis in a Thermo Fisher Scientific TSQ Quantis Plus mass spectrometer.

UPLC parameters were modified from the method of ref. ³⁹. LC–MS/MS analyses were performed using a Vanquish ultra-high-performance liquid chromatography system coupled with a TSQ Quantis Plus mass spectrometer (Thermo Fisher Scientific). Chromatographic separation was achieved using a BEH C18 (2.1 × 100 mm, 1.7 µm) UPLC column (Waters) heated to 60 °C. Then 2 µl of samples was injected into the column with a flow rate of 0.4 ml min⁻¹. Mobile phase A was 1 mM ammonium acetate and 0.1% formic acid in water, and mobile phase B was 1 mM ammonium acetate and 0.1% formic acid in 225:25:3 methanol:acetonitrile:water. The chromatographic gradient for a total 22 min method was 65% A at 0 min, 65% A at 1 min, 40% A at 2 min, 35% A at 9 min, 0% A at 15 min, 0% A at 20 min and 65% A at 20.5 min until 22 min. MS analyses were carried out using a triple quadrupole mass spectrometer with electrospray ionization in negative ion mode using the following settings: spray voltage 3.5 kV, sheath gas-flow rate 45 arb, aux gas-flow rate 10 arb, sweep gas-flow rate 1 arb, ion transfer tube temperature 325 °C and vaporizer temperature 350 °C. BBAAs were identified by selected reaction monitoring (SRM) mode with a chromatographic peak width of 12 s, Q1 and Q3 resolution of 1.2 full-width at half-maximum and collision gas pressure of 1.5 mTorr. SRM transitions from parent [M-H]⁻ ions to daughter [M-H]⁻ ions corresponded to the loss of bile acids (Supplementary Table 4). Peak areas for each conjugated bile acid were normalized to peak areas from deuterated internal standards of GCA-d4, GUDCA-d4 or GDCA-d4, matching the bile acid. Quantification was performed against a standard curve of authentic standards at eight concentrations ranging from 7.81 nM to 1 µM.

Shotgun metagenomics of IGRAM stool samples

DNA was extracted from faecal samples and from negative controls using the PowerSoil-htp kit (MO BIO Laboratories) following the manufacturer’s instructions, with the optional heating step included (note that MO BIO has since been purchased by QIAGEN; the extraction kit is now sold as the DNeasy PowerSoil HTP 96 Kit). Shotgun libraries were generated from 1 ng of DNA using the NexteraXT kit (Illumina). Libraries were sequenced on the Illumina HiSeq using 2 × 125 bp chemistry in High Output mode. In each plate, two positive controls containing a mix of Vibrio campbellii and PhiX lambda phage were added.

Shotgun metagenomic data were analysed using Sunbeam v.2.0.1 (ref. ⁴⁰). Low-quality sequences were trimmed using Trimmomatic with default parameters⁴¹. Sequences with low complexity and sequences that align to the human and phiX genomes were removed from further analysis. The abundance of bacteria was estimated using Kraken⁴². Sample similarity was assessed by Bray–Curtis dissimilarity and visualized on a principal coordinate plot. Difference in community composition was assessed using the PERMANOVA test. The Shannon diversity metric was used to calculate the alpha diversity. Differences in bile acid levels across the time points were calculated using linear mixed effects models. Log-transformed bile acid levels were used as the outcome, the time points were modelled as the predictor, and the subject information was added as the random effect.

Reporting summary

Further information on research design is available in the Nature Portfolio Reporting Summary linked to this article.