Gut bacterial indole improves blood sugar control in diabetic mice by boosting hormone release and gut cell growth

All mice used were 8–12 weeks old and free of specific pathogens. They weighed 20–32 g, apart from TALLYHO mice, which weighed 30–40 g. Male C57Bl/6J mice (Charles River) were used unless otherwise stated. Mice on a 129svJ background were bred inhouse. GluCreERT2 mice were generated by F. Gribble and F. Reimann, and crossed with Ai95(RCL-GCaMP6f)-D mice (The Jackson Laboratory, USA; RRID:IMSR_JAX:028865; https://www.jax.org/strain/028865↗). GLP-1 receptor (GLP-1R fl/fl) mice were generated by R. Seeley at the University of Michigan and crossed with Pdx1-Cre/Esr1 mice (The Jackson Laboratory, USA; RRID:IMSR_JAX:024968; https://www.jax.org/strain/024968↗) to generate Pdx1-CreERT; Glp1r fl/fl mice. Tamoxifen (100 mg/kg) was i.p. injected for 4 days to induce beta cell knockout of the GLP-1R. Both male and female GluCreERT2;Ai95-D and Pdx1-Cre;GLP-1R fl/fl mice were used. Only male TALLYHO/JnG mice (The Jackson Laboratory, USA; RRID:IMSR_JAX:005314; https://www.jax.org/strain/005314↗), purchased from the Jackson Laboratory, were used because only the male mice develop characteristics of diabetes, including glucose intolerance and insulin resistance. A total of 388 mice were used. Each experimental group consisted of a minimum of six mice. The numbers of animals/experiments/cells for each study are given in the results and figure legends. Group sizes were determined based on prior studies and on power calculations (where applicable) to ensure adequate statistical power.

The mice were housed in groups in individually ventilated cages containing standard bedding with a 12 h light/dark cycle (lights on at 07:00 hours, lights off at 19:00 hours) with ad libitum access to standard laboratory RM1 chow and water, and cardboard tubes for environmental enrichment. Procedures were carried out during the light cycle. All procedures were carried out according to the UK Animals (Scientific Procedure) Act 1986, and approved by the Imperial College London Animal Welfare and Ethical Review Body. Our study followed ARRIVE reporting guidelines. The manipulations used mice from different groups (randomised by body weight) in a mixed sequence to avoid confounders, and investigators were blinded regarding mouse genotype and treatment.

Glucose tolerance tests were performed on mice after a 5 h fast with ad libitum access to water. After baseline blood glucose measurement, mice were i.p. injected with 20% glucose (2 g/kg). Changes in blood glucose were measured over 120 min using tail-tip venesection and an Accu-Chek glucometer (Roche). Glucose was administered intraperitoneally rather than orally to bypass the gastrointestinal tract, allowing us to study glucose clearance and insulin secretion independently of glucose-stimulated GLP-1 release and any potentially confounding effects on gastric emptying.

To investigate the effect of indole on acute glucose tolerance, indole (20 mg/kg; I3408; Sigma-Aldrich) or vehicle (water) was orally gavaged 15 min before i.p. injection of glucose. This dose of indole was selected based on a previous study showing an effect of indole on reducing mucosal inflammation [10]. Saline (154 mmol/l NaCl) was used as a vehicle control for all i.p. treatments except FICZ treatment (a synthetic aryl hydrocarbon receptor [AhR] agonist), for which 1% DMSO in saline was used.

Due to the rapid but variable progression of diabetes in TALLYHO mice, there was considerable variability in fasting blood glucose levels across individual mice. The raw IPGTT data were therefore normalised to each mouse’s baseline value (t₀) for meaningful comparisons of glucose excursion.

Tests were performed in the early light phase on individually caged mice after a 16 h fast. Mice received an oral gavage of indole (20 mg/kg) or vehicle (water), following which pre-weighed amounts of RM1 chow were returned to the cages. Changes in food intake were measured over 2 h.

Insulin tolerance tests were performed on mice after a 5 h fast with ad libitum access to water. After baseline blood glucose measurement, mice were injected with insulin at 1 U/kg (Actrapid, Novo Nordisk). Changes in blood glucose were measured over 120 min using tail-tip venesection and an Accu-Chek glucometer.

For studies in TALLYHO mice, raw blood glucose data were normalised to t₀ to account for variability in diabetes progression.

Pancreatic islets were isolated from adult C57Bl6/J mice by collagenase digestion (Universal Biologicals, Cambridge, UK) as previously described [11], and maintained at 37°C, 5% CO₂ in RPMI medium (Sigma-Aldrich) with 10% FBS (Gibco/Thermo Fisher) and 1% penicillin/streptomycin (Gibco/Thermo Fisher) overnight.

Islets were incubated at 37°C for 1 h in HEPES Krebs Ringer bicarbonate buffer (HKRB) (pH 7.4, bubbled with CO₂) with 1% BSA and 3 mmol/l glucose (i.e. low glucose). Next, the islets were stimulated with HKRB with 11 mmol/l glucose (i.e. high glucose) in the presence or absence of 1 mmol/l indole. Supernatant samples (which contain secreted insulin) and lysed islet samples were separately collected and stored at −20°C until further use. Insulin levels were assayed using a mouse insulin ELISA kit (Cisbio Bioassays, Codolet, France) according to the manufacturer’s instructions.

All cell and organoid lines were routinely tested for mycoplasma contamination. STC-1 cells were maintained in DMEM (Gibco/Thermo Fisher) with 10% FBS, 1% penicillin/streptomycin at 37°C, 5% CO₂. Colonic crypts were isolated from 6–8-week-old C57Bl6/J mice as previously described [12, 13] and plated onto 24-well, 2% Matrigel-coated plates. The crypts were cultured in DMEM (with 10% FBS, 1% penicillin/streptomycin and 10 μmol/l Y-27632 [Bio-Techne, Abingdon, UK]) and incubated overnight at 37°C, 5% CO₂.

Hormone secretion assays were performed using STC-1 cells and primary murine colonic crypt cultures. The cells were treated using test reagents (made up in DMEM) for 2 h at 37°C, 5% CO₂. Supernatant and lysed cell samples were collected and stored at −20°C. The levels of GLP-1 and peptide YY (PYY) in cell supernatants and lysates and in plasma were measured using previously described sensitive and specific in-house radioimmunoassays [14, 15].

Ileal organoids were derived from the last 6 cm of the ileum of 6–12-week-old C57Bl6/J and PPG-Cre;GCaMP6 mice, as previously described [16]. Dissociated crypts were resuspended in Cultrex Pathclear reduced growth factor basement membrane extract (Bio-Techne, Abingdon, UK) and cultured in Intesticult medium (STEMCELL Technologies, Cambridge, UK) at 37°C with 5% CO₂, with medium changes every 2–3 days. Organoids were passaged every 5–7 days at ratios of 1:2–1:3 using Gentle Cell Dissociation Reagent (STEMCELL Technologies, Cambridge, UK) and mechanical disruption.

Organoids from PPG-Cre;GCaMP6 mice were seeded onto 96-well black-walled plates. After 3 days, they were treated with 1 μmol/l 4-hydroxytamoxifen for 24 h to induce GCaMP6f expression in L cells. Timelapse recordings were used to assess calcium response following addition of the treatment. Forskolin (10 μmol/l) and ATP (100 μmol/l) were added after 10 min to confirm cell viability. Imaging was performed using a Nikon ECLIPSE Ti2-E widefield microscope, capturing one frame per second.

Images were analysed using ImageJ with a custom macro package (Intensity_2) from Imperial FILM facility. Regions of interest (ROIs) were drawn around cell bodies to measure fluorescence intensity (f). Baseline fluorescence was calculated from 10–30 s prior to treatment addition. Fluorescence change was quantified as ΔF/F_baseline, with the maximum value calculated over 10 min post-treatment. Cells were then stratified as responders if the maximum ΔF/F_baseline for the treatment was greater than 20% of F_max − F_min, and as non-responders if not [17].

Murine ileal organoids for imaging were plated onto 8-well chamber slides (Thermo Fisher). After treatment, organoids were fixed using 4% paraformaldehyde, permeabilised using 0.1% Triton-X PBS, and blocked using 5% BSA for 1 h at room temperature. Primary antibody incubation was performed overnight at 4°C using mouse anti-GLP-1 antibody (ab23468, 1:500; Abcam; verified by relative expression, reduced mouse ileal GLP-1 detected when GLP-1 mRNA and circulating protein also reduced). Subsequently, the organoids were washed and incubated with the secondary antibody (anti-mouse Alexa Fluor 647-conjugated IgG, 1:1000; Thermo Fisher) for 1 h at room temperature. DAPI was used as a nuclear stain. Images were acquired using a Leica SP5 confocal laser scanning microscope.

Immunohistochemistry was performed according to standard protocols. Tissues were fixed overnight in 4% paraformaldehyde at 4°C, and processed into paraffin blocks. After sectioning using a Leitz 1512 microtome, 5 μm sections were deparaffinised, rehydrated and subjected to antigen retrieval by heating to 90°C in 1 mmol/l EDTA pH 8 solution for 30 min. Next, sections were blocked using 3% BSA in PBS-T (Sigma-Aldrich) for 1 h at room temperature. Primary antibody incubation was performed overnight at 4°C using mouse anti-GLP-1 antibody (ab23468, 1:100; Abcam) and anti-Villin 1 (MA5-16408, 1:100; Thermo Fisher; verified by relative expression, high in Caco-2, COLO 205 and HT-29 cells compared with low or negative in Jurkat and PANC-1 cells and high in mouse jejunum compared with low or negative in mouse colon, kidney and liver). Sections were washed with PBS, and incubated with the secondary antibodies (goat anti-mouse Alexa Fluor 555-conjugated and goat anti-rabbit Alexa Fluor 647-conjugated IgG, 1:300; Thermo Fisher) for 3 h at room temperature. DAPI was used to counterstain the nuclei. Ileum sections were imaged using a Zeiss LSM-780 inverted confocal microscope. The number of GLP-1-immunoreactive cells in stained ileum samples was quantified manually using the ‘Cell Counter’ plug-in in ImageJ, and presented as a proportion (expressed in arbitrary units, AU) of the mean number of GLP-1-positive cells to the mean number of villi counted per field of view. The person counting was blinded to treatment.

Organoids were cultured and treated as described in the ‘Ileal organoid cultures’ section above, and snap frozen at −80°C. Total RNA was isolated using TRIsure (Bioline/Meridian Bioscience, Cincinnati, OH, USA) –chloroform extraction after homogenisation with TissueLyser II. After cDNA synthesis using a high-capacity cDNA reverse transcription kit (Applied Biosystems) according to the manufacturer’s instructions, qPCR was performed using Maxima Probe Master Mix (Life Technologies) and Taqman PCR core reagents (Applied Biosystems) in a C1000 CFX384 real-time system thermocycler. Data were analysed using themethod. 2 Δ Δ C t

For bulk RNA sequencing, the right liver lobe was dissected from mice and snap frozen at −80°C. Total RNA was purified using the RNeasy Plus Mini kit (Qiagen) according to the manufacturer’s instructions.

Purified total RNA samples were analysed using Nanodrop (A₂₆₀/A₂₈₀ >1.8, suggesting good RNA purity), and RNA integrity was checked using an Agilent Bioanalyzer RNA 6000 Nano assay (RNA integrity number [RIN] >7). Strand-specific mRNA (polyA) libraries were constructed at Novogene UK, and sequenced using an Illumina NovaSeq 6000 sequencing system with 5.8–7.9 billion bases per sample. Adapters were trimmed using Novogene’s bioinformatics pipeline, and alignments were performed using HISAT2 [18] and mouse genome GRCm38 Ensembl version 102. Uniquely mapped reads were carried forward for gene-level counts and differential gene expression analysis using DESeq2 [19].

The dataset was submitted for Ingenuity Pathway Analysis (http://www.ingenuity.com/↗). A core analysis was conducted using a cut-off p value of <0.05, which reduced the analysis-ready dataset to 716 genes. Stringent filters (Benjamini–Hochberg p<0.05, z score >2) were applied to detect enriched canonical pathways. For upstream regulator analysis, a z score cut-off of >2 and a p value for overlap of <0.05 was used.

Data were analysed using GraphPad Prism version 9.0, and are expressed as means ± SD. An unpaired, two tailed Student’s t test was used for two-group comparisons. One-way ANOVA with Tukey’s post hoc test was used for single independent variable experiments with more than two groups. Two-way ANOVA with Tukey’s post hoc test was used for analyses with two independent variables. Significance was set at p<0.05.

An oral indole pre-load significantly improved acute intraperitoneal glucose tolerance (vehicle: 20.4±1.8 mmol/l; indole: 17.1±2.1 mmol/l at t₁₅; n=8–10, p<0.001) (Fig. 1a), but an i.p. indole pre-load did not (Fig. 1b), suggesting that the effects of indole are mediated via the gut, potentially by modulating gut hormones. Furthermore, other tryptophan metabolites did not have similar effects on glucose tolerance. Administration of the major tryptophan metabolite kynurenic acid via oral gavage or i.p. injection, at doses that have previously been shown to affect inflammasome activation in colitis [20], did not affect glucose tolerance (Fig. 1c, d). Although previous studies have reported that chronic indole suppresses food intake [21, 22], we did not identify any acute effects of indole on food intake or insulin sensitivity in lean mice (Fig. 1e, f).

Given previous reports that indole modulates gut hormone secretion in vitro [8, 9], we hypothesised that the acute glucoregulatory effects of indole were mediated via enhanced GLP-1 secretion. We found that increasing concentrations of indole stimulated GLP-1 release from the enteroendocrine STC-1 cell line (10 mmol/l: 2.93±0.43-fold change vs vehicle; n=3–8, p<0.001) (Fig. 2a) and murine colonic L cell cultures (3 mmol/l: 5.76±1.70; 10 mmol/l: 7.07±1.73-fold change vs vehicle; n=3–7, p<0.001) (Fig. 2b). Live cell imaging on transgenic organoids generated from PPG-Cre;GCaMP6 murine ileal crypts, in which L cells express the fluorescent Ca²⁺ indicator GCaMP6f (Fig. 2c), showed that indole induced a rapid and reversible dose-dependent elevation of intracellular Ca²⁺, with 10 mmol/l indole increasing the maximum fluorescence intensity by 1.86±1.14-fold (n=15–90, p<0.001) and also increasing the percentage of responsive L cells (Fig. 2d–f). Correspondingly, oral administration of indole in mice significantly increased plasma total GLP-1 levels after 15 min (vehicle: 23.30±8.10 pmol/l; indole: 36.45±5.73 pmol/l; n=5–6, p<0.05) (Fig. 2g). Together, these results extend previous in vitro findings showing that indole stimulates GLP-1 release [8, 9]. Furthermore, following oral indole administration in mice, insulin secretion (vehicle: 217.9±64.5 pmol/l; indole: 312.0±67.7 pmol/l at t₁₅; n=7–9, p<0.05) increased in parallel with the improvement in glucose tolerance (vehicle: 21.8±2.4 mmol/l; indole: 18.0±2.2 mmol/l at t₁₅; n=7–9, p<0.001) (Fig. 2h). However, indole did not potentiate insulin secretion from murine islets in vitro (Fig. 2i), suggesting that it does not directly modulate pancreatic hormone secretion. The acute glucose-lowering effects of oral indole administration were dependent on GLP-1 signalling, as these effects were significantly attenuated by pre-treatment with the GLP-1 receptor antagonist exendin (9–39) (glucose iAUC: vehicle + indole: 299.3±73.82 mmol/l × 120 min; exendin (9–39) + indole: 517.5±170.5 mmol/l × 120 min; n=12–14, p<0.001) (Fig. 2j). Additionally, there was no longer a beneficial effect of indole on glucose tolerance in Pdx1-CreERT;Glp1r fl/fl transgenic mice, in which GLP-1 receptor expression is inducibly knocked down in Pdx1-expressing pancreatic beta cells (Fig. 2k, l). Together, these data suggest that the acute glucoregulatory effects of indole are mediated by increased GLP-1 release driving insulin secretion.

Although 10 mmol/l indole increased the release of PYY, which is co-expressed with GLP-1 in L cells [23], from murine colonic crypt cultures by 5.26±3.94 fold (n=7, p<0.01) (Fig. 3a), plasma PYY levels were unchanged 15 min after oral indole administration in mice (Fig. 3b). This is in accordance with the lack of effect of indole on food intake in vivo (Fig. 1e), and suggests that any effect on PYY release in vivo was probably short-lived. It may also reflect the relatively low density of PYY-expressing cells in the proximal small intestine.

Indole and its derivatives can modulate intestinal homeostasis and immune functions via the AhR [5], while impaired AhR ligand production has been associated with an increased risk of developing metabolic syndrome [24]. Given that many AhR antagonists have partial agonist actions, and exhibit different efficacies depending on the specific ligand being blocked and the tissue that the receptor is expressed in, we investigated whether AhR agonism plays a role in glucoregulation. However, at concentrations and doses previously shown to stimulate hormone secretion and reduce obesity-induced hyperglycaemia [24], the synthetic AhR agonist FICZ did not stimulate GLP-1 secretion from STC-1 cells (Fig. 3c), nor did it modify glucose tolerance in C57Bl6/J mice (Fig. 3d). This suggests that the AhR is not involved in the acute effects of indole on glucoregulation and GLP-1 secretion, perhaps unsurprisingly given that indole is a relatively weak agonist of the mouse AhR [25].

The excitatory calcium-permeable cation channel transient receptor potential channel subtype A1 (TRPA1) is widely expressed in enteroendocrine cells (EECs), and has been implicated in gut microbial EEC activation in zebrafish, as well as in indole-mediated serotonin secretion from human and murine primary intestinal cultures [9]. TRPA1 activation has also previously been linked to GLP-1 secretion from murine L cells in vitro [26]. We found that indole-induced GLP-1 release from STC-1 cells (Fig. 3e) and murine primary colonic crypt cultures (Fig. 3f) was attenuated in the presence of the TRPA1 inhibitor HC030031↗, while pre-treatment of PPG-Cre;GCaMP6 organoids with HC030031↗ blocked indole-stimulated L cell activation (Fig. 3g, h). Additionally, TRPA1 inhibition in mice using the TRPA1 inhibitor A-967079 also blocked indole-mediated improvements in glucose tolerance (glucose iAUC: vehicle + indole: 462.0±108.2 mmol/l × 120 min; A-967079 + indole: 586.9±126.3 mmol/l × 120 min; n=12, p<0.05) (Fig. 3i). When PPG-Cre;GCaMP6 organoids were imaged in Ca²⁺-free buffer, indole-stimulated Ca²⁺ mobilisation was blocked, reflecting the need for extracellular Ca²⁺ influx for indole to activate L cells (Fig. 3j, k), in accordance with the action of TRPA1.

The indole precursor molecule L-tryptophan can also stimulate GLP-1 release, an effect that is thought to be mediated in part via the Ca²⁺-sensing receptor [27]. While L-tryptophan-induced Ca²⁺ mobilisation in L cells was expectedly attenuated by the Ca²⁺-sensing receptor antagonist NPS-2143, it was unmodified by the TRPA1 inhibitor HC030031↗ (Fig. 3l, m). Together, these data suggest that the TRPA1 ion channel mediates indole-stimulated GLP-1 secretion and glucose management via a process distinct from that which mediates tryptophan-induced GLP-1 secretion.

Specific GLP-1 secretagogues can enhance L cell differentiation [28]. Furthermore, the tryptophan derivative indole-3-acetate has also been shown to modify intestinal lineage differentiation [29]. Having established the acute effects of indole on glucoregulation and GLP-1 secretion, we investigated whether indole mediated longer-term effects on the GLP-1 system. Wild-type (WT) murine ileal organoids were exposed to a 3 h treatment pulse of various concentrations of indole (Fig. 4a) and multiple treatment lengths for 100 µmol/l indole (Fig. 4b), and the effect on expression of Gcg, which encodes GLP-1, was determined 2 days later. We subsequently repeated the 3 h treatment pulse study using only the most effective concentration of indole, 100 µmol/l (Fig. 4c). We found a 1.24±0.36-fold (n=7, p<0.01) upregulation in mRNA expression of Gcg in indole-treated organoids, in addition to elevated transcript levels for Pyy and Gip (Fig. 4d). Indole treatment also increased mRNA expression of the transcription factors NeuroD1 and Neurog3 (also known as Ngn3), which are associated with L cell endocrine specification (Fig. 4e), indicating enhanced development of enteroen-docrine secretory progenitors. Expression of the enteroendocrine differentiation factor genes Pax4 and Pax6 was also increased but not significantly (Fig. 4e). In line with the upregulated Gcg mRNA expression, addition of 100 µmol/l indole to organoid culture medium for 3 h increased L cell number 1.55-fold (vehicle: 1.14±0.53%; indole: 1.78±0.49%; n=27–30, p<0.001) (Fig. 4f–h). Thus, a short indole treatment exposure can augment the L cell population in mouse intestinal organoids.

We next investigated whether the effects of indole on epithelial cell fate translated into metabolic benefits in vivo. Prior to a single oral bolus of indole treatment intervention, both groups of C57Bl6/J mice were similarly glucose-tolerant (Fig. 5a). As expected, mice showed improved glucose tolerance after acute oral indole administration (Fig. 5b). This improved glucose tolerance was sustained, with a reduced peak in blood glucose levels following an i.p. bolus of glucose 3 days after oral indole administration (Fig. 5c), despite the half-life of indole previously being reported to be less than 8 h [21]. This effect of indole on glucose tolerance was lost by 7 days post-administration (Fig. 5d). Next, we observed that, while i.p. indole administration (which does not affect GLP-1 secretion) did not modify acute glucoregulation (Fig. 5e), indole-treated mice displayed improved glucose tolerance 3 days post i.p. administration (Fig. 5f). These data suggest that the effect of indole on long-term glucose tolerance is mediated via a mechanism that is distinct from that mediating the acute stimulation of GLP-1 release. However, in Pdx1-CreERT;Glp1r fl/fl mice, these effects on glucose tolerance 3 days after oral indole administration were lost (Fig. 5g), suggesting that the mechanism for these longer effects on glucose tolerance is also dependent on GLP-1 receptor signalling. These findings suggest that indole improves acute and long-term glucose tolerance via two distinct mechanisms, but that both involve GLP-1 signalling in pancreatic beta cells.

The intestinal epithelium undergoes rapid renewal, regenerating every 3–5 days. Given that the improvement in glucose tolerance observed in indole-treated mice was lost after approximately a week, and in line with the in vitro findings, we hypothesised that a single dose of indole would increase differentiation of GLP-1 secreting L cells to improve glycaemic control. Ileum samples collected 3 days later from C57Bl6/J mice that received a single oral indole dose showed an increase in L cell density (vehicle: 0.53±0.11 AU; indole: 1.05±0.46 AU; n=5, p<0.05) (Fig. 5h–j).

Taken together, these data suggest that indole modulates EEC fate specification and increases L cell abundance, improving glucose homeostasis in mice. These findings position indole as a potential glucose-lowering therapy.

To further investigate the therapeutic utility of indole in type 2 diabetes treatment, we next studied the effects of sub-chronic indole administration in WT and a type 2 diabetes mouse model (Fig. 6a). WT C57Bl6/J mice that received sub-chronic indole administration showed reduced fasting blood glucose levels (vehicle: 7.9±1.1 mmol/l; indole: 6.8±0.7 mmol/l; n=5, p<0.05) (Fig. 6b), in addition to improvements in intraperitoneal glucose tolerance (vehicle: 19.2±3.2 mmol/l; indole: 16.4±2.3 mmol/l at t₁₅; n=8–10, p<0.05) and oral glucose tolerance (Fig. 6c, d); a modest increase in insulin sensitivity was also noted (Fig. 6e). Furthermore, these effects were independent of any changes in body weight (Fig. 6f). Mice given sub-chronic indole treatment also demonstrated a trend towards elevated basal circulating GLP-1 levels and higher Gcg expression in the ileum, although these changes did not reach statistical significance (Fig. 6g, h). These data are in line with findings that chronic indole supplementation in mice reduced high-fat diet-associated systemic insulin resistance and glucose intolerance, although this is the first study to report metabolic improvements after a relatively short treatment intervention and in WT mice. Given these metabolic benefits of indole, and the preservation of GLP-1’s insulinotropic effect in type 2 diabetes, the same sub-chronic indole treatment protocol was repeated in TALLYHO/JnG mice, a polygenic type 2 diabetes mouse model that exhibits hyperglycaemia and insulin resistance, reflecting the multifactorial nature of human type 2 diabetes [30]. Indole-treated TALLYHO/JnG mice showed improved glucose tolerance (Fig. 6i), and, interestingly, an improvement in insulin sensitivity (net glucose iAUC: vehicle 464.2±249.2 mmol/l × 120 min; indole 795.6±114.6 mmol/l × 120 min; n=7, p<0.01) (Fig. 6j), demonstrating the potential translatability of indole’s metabolic effects to a disease state. To further investigate this insulin sensitivity phenotype, we profiled whole-liver transcriptomes from indole- and vehicle-treated TALLYHO mice. A total of 719 differentially expressed genes (unadjusted p<0.05) were found (Fig. 6k), including upregulation of Hk3, which encodes the rate-limiting enzyme of many glucose metabolism pathways, representing a possible mechanism by which indole improves hepatic insulin sensitivity. Ingenuity Pathway Analysis further revealed that indole induces mild transcriptional alterations in liver metabolism; specifically, an inhibition of IL-6 cytokine signalling (Fig. 6l–n). The anti-inflammatory effects of indole in the liver could potentially also enhance insulin sensitivity, consistent with work showing that indole and related compounds exert hepatoprotective effects in metabolic dysfunction-associated steatotic liver disease (previously known as non-alcoholic fatty liver disease) by reducing the production of proinflammatory mediators [21, 31, 32]. Furthermore, modulation of cell–cell and tight-junction assembly pathways was observed, in line with previous reports of indole’s role in maintaining epithelial barrier function (Fig. 6n) [6, 7].

The Section of Endocrinology and Investigative Medicine, Hammersmith Hospital, London, is funded by grants from the MRC, BBSRC and NIHR, and is supported by the NIHR Biomedical Research Centre Funding Scheme. The views expressed are those of the authors and not necessarily those of the MRC or BBSRC, the NHS, the NIHR or the Department of Health and Social Care. PP is funded by a National Science Scholarship from A*STAR, Singapore. KGM is supported by Diabetes UK (18/0005886 and 20/0006295), the BBSRC (BB/W001497/1 and BB/X017273/1), the MRC (MR/Y013980/1) and the Wellcome Trust (310835/Z/24/Z). BJ acknowledges funding support from the Medical Research Council (MR/Y00132X/1 and MR/X021467/1), the Wellcome Trust (301619/Z/23/Z), Diabetes UK, the Eli Lilly LRAP programme and Metsera Inc. BYHL and GSHY are supported by BBSRC project grant BB/S017593/1 and the MRC Metabolic Diseases Unit (MC_UU_00014/1). Next-generation sequencing was performed at the IMS Genomics and Bioinformatics Core, supported by the MRC (MC_UU_00014/5 and RGAG/542 MC_UU_00039), the Wellcome Trust (208363/Z/17/Z and RGAG/546 226800/Z/22/Z), and the Cancer Research UK Cambridge Institute Genomics Core. The FR/FMG laboratories were funded by the Wellcome Trust (220271/Z/20/Z) and the MRC (MRC_MC_UU_12012/3). The study’s funders were not involved in the design of the study, the collection, analysis and interpretation of data, or writing the report, and did not impose any restrictions regarding publication of the report.

RNA sequencing data are available from the Gene Expression Omnibus under accession number GSE306720↗.

All data, methods and unique/stable reagents generated in this study will be made available from the corresponding author upon reasonable request, but may require a completed Materials Transfer Agreement. RNA sequencing data are available from the Gene Expression Omnibus under accession number GSE306720↗.