Dopamine receptor D2 confers colonization resistance via microbial metabolites

Abstract

The gut microbiome has major roles in modulating host physiology. One such function is colonization resistance, or the ability of the microbial collective to protect the host against enteric pathogens^1–3, including enterohaemorrhagic Escherichia coli (EHEC) serotype O157:H7, an attaching and effacing (AE) food-borne pathogen that causes severe gastroenteritis, enterocolitis, bloody diarrhea and acute renal failure^4,5 (haemolytic uremic syndrome). Although gut microorganisms can provide colonization resistance by outcompeting some pathogens or modulating host defence provided by the gut barrier and intestinal immune cells^6,7, this phenomenon remains poorly understood. Here, we show that activation of the neurotransmitter receptor dopamine receptor D2 (DRD2) in the intestinal epithelium by gut microbial metabolites produced upon dietary supplementation with the essential amino acid l-tryptophan protects the host against Citrobacter rodentium, a mouse AE pathogen that is widely used as a model for EHEC infection^8,9. We further find that DRD2 activation by these tryptophan-derived metabolites decreases expression of a host actin regulatory protein involved in C. rodentium and EHEC attachment to the gut epithelium via formation of actin pedestals. Our results reveal a noncanonical colonization resistance pathway against AE pathogens that features an unconventional role for DRD2 outside the nervous system in controlling actin cytoskeletal organization in the gut epithelium. Our findings may inspire prophylactic and therapeutic approaches targeting DRD2 with dietary or pharmacological interventions to improve gut health and treat gastrointestinal infections, which afflict millions globally.

EHEC infects hosts by forming attaching and effacing (AE) lesions known as actin pedestals on the gut epithelium, from which it secretes virulence factors, including Shiga toxin, into the host^10–12. The mouse AE pathogen C. rodentium is a widely used model for EHEC infection owing to its highly similar infection strategy, encoded by the locus of enterocyte effacement pathogenicity island shared with EHEC and enteropathogenic E. coli^8,9,13. The gut microbiota provide colonization resistance to C. rodentium by enhancing host defence pathways⁷ and by competing with the pathogen for nutrients within the gut⁶. Additionally, specific small molecules produced by gut microbial metabolism of dietary nutrients can mediate diverse effects, including regulation of immunity, cancer immunotherapy and certain colonization-resistance pathways^14–23. We were thus motivated to examine the roles of bacterial small-molecule metabolites derived from dietary supplementation with the essential amino acid tryptophan (Trp) in protecting against infection with AE pathogens because of their functions in modulating immunity^24–27 and improving gut barrier function during inflammation²⁸.

We pre-treated wild-type C57Bl/6 mice with or without a broad-spectrum antibiotic cocktail to deplete the gut microbiota, followed by a Trp-supplemented diet for seven days, which was well-tolerated (Supplementary Fig. 1a), and infection with C. rodentium (Fig. 1a). Compared with mice fed conventional diet, Trp-fed mice exhibited lower colonic C. rodentium burden (Fig. 1b,c) and decreased colonic inflammation (Fig. 1d–f). These effects were abrogated by antibiotic treatment, suggesting that the microbiota provide protection against severe disease during dietary Trp supplementation.

Fig. 1. Dietary Trp protects against infection with C. rodentium strain DBS100 in a mouse model of EHEC infection.

a, C57Bl/6 mice were pre-treated with antibiotics (ABX; 9 mg kg⁻¹ each metronidazole, ampicillin and neomycin, 4.5 mg kg⁻¹ vancomycin) for 7 days, followed by conventional (2 g kg⁻¹ Trp diet, ad libitum) or Trp (42 g kg⁻¹ Trp diet, ad libitum) diet for 7 days. The mice were then administered C. rodentium (Cr; oral gavage, 10⁸ colony-forming units (CFU)) with continued ABX (except neomycin) and Trp feeding. Faeces were collected before infection and 9 days post-infection to determine Trp metabolite levels (Supplementary Fig. 1). b,c, Bacterial load in faeces (b; measured at 1–2 day intervals for 24 days post-infection) and colon tissue (c; measured at the peak of infection, 10 days post-infection). d–f, Colon sections were stained with haematoxylin and eosin (H&E). d, Sections were scored blindly for submucosal oedema (0–3), goblet cell depletion (0–3), epithelial hyperplasia (0–3), epithelial integrity (0–4) and neutrophil and mononuclear cell infiltration (0–3). Data are expressed as the sum of these individual scores (0–16). See Methods for full description of scoring rubric. e, Crypt heights. f, Representative images. Scale bar, 50 μm. Data are representative of at least 3 independent experiments, n = 10 mice per group. Data are mean ± s.d. Two-tailed Student’s t-test (b) or one-way analysis of variance (ANOVA), followed by post hoc Tukey multiple comparison test (c–e). *P < 0.05, **P < 0.01, ***p < 0.001.

To identify specific Trp metabolites that mediate the effects of the Trp diet, we performed targeted mass spectrometry-based metabolomics before and after C. rodentium infection during Trp feeding and found that the most highly abundant metabolites were indole, indole-3-ethanol (IEt), indole-3-pyruvate (IPyA) and indole-3-aldehyde (I3A) (Supplementary Fig. 1b–l). Although indole can decrease EHEC and C. rodentium virulence, the effects and mechanisms of action of IEt, IPyA and I3A on these pathogens remain mostly unknown^29–31.

To determine the roles of these Trp metabolites on C. rodentium infection, we pre-treated C57Bl/6 mice with or without antibiotics to prevent microbial metabolism of the individual Trp metabolites, followed by administration of IEt, IPyA or I3A for 2 days prior to and during infection (Fig. 2a). Treatment with the individual metabolites lowered colonic C. rodentium burden (Fig. 2b,c) and decreased colonic inflammation (Fig. 2d,e and Extended Data Fig. 1). We verified that levels of IEt, IPyA and I3A in the colon were increased following metabolite administration (Supplementary Figs. 2–4). We also determined that Trp diet, IEt, IPyA and I3A did not significantly alter microbial composition using 16S rRNA gene analysis (Supplementary Fig. 5). Thus, these microbial Trp metabolites mediate colonization resistance against C. rodentium.

Fig. 2. The Trp metabolites I3A, IPyA and IEt protect against C. rodentium infection in mice.

a, C57Bl/6 mice were pre-treated with antibiotics (ABX) for 7 days, followed by the Trp metabolites I3A (1,000 mg kg⁻¹), IPyA (2,900 mg kg⁻¹) or IEt (600 mg kg⁻¹) by oral gavage daily for 2 days. The mice were then administered C. rodentium (oral gavage, 10⁸ CFU) with continued ABX (except neomycin) and metabolite treatment. Faeces were collected before infection and 9 days post-infection to determine Trp metabolite levels (see Supplementary Figs. 2–4). b,c, Bacterial load in faeces (b; measured at 1–2 day intervals for 24 days post-infection) and colon tissue (c; measured at the peak of infection, 10 days post-infection). d,e, Colon sections were stained with H&E. d, Combined scores as described in Fig. 1d. e, Crypt heights. Data are representative of at least 3 independent experiments, n = 10 mice per group. Data are mean ± s.d. Two-tailed Student’s t-test (b) or one-way ANOVA (c–e), followed by post hoc Tukey multiple comparison test.

To determine whether IEt, IPyA and I3A affect the pathogen directly, we treated C. rodentium with each metabolite and found that they did not affect bacterial growth and virulence factor expression (Extended Data Fig. 2a–e). We also treated EHEC with IEt, IPyA and I3A and found that these metabolites also did not affect this pathogen (Extended Data Fig. 2f–k). Because the metabolites did not directly affect bacterial growth or virulence in vitro, we reasoned that IEt, IPyA and I3A may mediate colonization resistance primarily via host pathways.

To identify potential host receptors for these metabolites, we utilized similarity ensemble approach³² (SEA), a computational tool for predicting protein targets of small molecules, with IEt, IPyA and I3A as ligands. All three metabolites were predicted to bind the dopamine receptors DRD2, DRD3 and DRD4, G-protein-coupled receptors with canonical roles as neurotransmitter receptors in the central and peripheral nervous systems^33,34. We found that inhibition of DRD2, DRD3 and DRD4 using the pan-inhibitor haloperidol³⁵ reduced the effects of the metabolites in decreasing bacterial attachment via actin pedestals during EHEC infection of the human intestinal epithelial cell (IEC) line Caco-2 (Extended Data Fig. 3a,b). Further, using CRISPR–Cas9 knockout of DRD2, DRD3, and DRD4 in Caco-2 cells during EHEC infection, we found that this effect depended exclusively on DRD2 (Extended Data Fig. 3c–f). Critically, we found that IEt, IPyA and I3A are bona fide DRD2 ligands, with potencies similar to the endogenous ligand dopamine, using both the GloSensor and Tango G-protein-coupled receptor assays, which quantify cyclic AMP and arrestin recruitment, respectively^36,37 (Extended Data Fig. 4). We also pre-treated C57Bl/6 mice with or without the antibiotic cocktail, followed by dopamine and C. rodentium infection. We found that dopamine also decreased morbidity during infection, albeit with lower potency, probably owing to decreased bioavailability³⁸ (Supplementary Fig. 6a–f). Further, treatment of mice with the pharmacological DRD2 agonist sumanirole had similar effects to the Trp metabolites (Supplementary Fig. 6j–n), whereas the DRD2 antagonist L-741,626 blocked the effects of the Trp diet (Supplementary Fig. 6r–v). Collectively, these results suggest that DRD2 may be a host receptor in the gut that recognizes the Trp metabolites.

Next, we assessed whether DRD2 expression in the gut epithelium–the initial target tissue of AE pathogens–mediates the effects of the Trp metabolites in promoting colonization resistance to C. rodentium infection. We generated Drd2^fl/fl × Vil1-cre mice, which lacked Drd2 expression in IECs (Fig. 3a,b), and treated these mice with Trp diet, IEt, IPyA or I3A followed by infection with C. rodentium (Fig. 3c and Extended Data Figs. 4–6). We found that Drd2^fl/fl control littermates that were treated with Trp diet or individual metabolites exhibited decreased colonic C. rodentium burden (Fig. 3d,e and Extended Data Fig. 6a,b) and colonic inflammation as measured by histology (Fig. 3f–h and Extended Data Fig. 6c–e). However, these effects were abrogated in the Drd2^fl/fl × Vil1-cre mice, indicating that DRD2 expression in IECs mediates the effects of the Trp metabolites.

Fig. 3. Effects of Trp diet and metabolites in protecting against C. rodentium infection depend on DRD2 in IECs.

a, IECs were isolated from Drd2^fl/fl × Vil1-cre or Drd2^fl/fl mice, and cell lysates were analysed by western blot with the indicated antibodies. Source data are provided in Supplementary Fig. 9. b, Intestinal cryosections from Drrd2^fl/fl × Vil1-cre or Drd2^fl/fl mice were stained with DAPI and DRD2 antibody, followed by an anti-mouse Alexa Fluor 594. Representative z-slices are shown. Scale bar, 20 μm. c–h, Drd2^fl/fl × Vil1-cre or Drd2^fl/fl mice were fed a conventional (2 g kg⁻¹ Trp diet, ad libitum) or Trp (42 g kg⁻¹ Trp diet, ad libitum) diet for 7 days or IEt (600 mg kg⁻¹) by oral gavage daily for 2 days, and then infected with C. rodentium (oral gavage, 10⁸ CFU) with continued Trp feeding or metabolite treatment. Results for I3A and IPyA are shown in Extended Data Fig. 6. c, Experimental design. d,e, Bacterial load in faeces (d; measured every 1–2 days for 24 days post-infection) and colon tissue (e; measured at the peak of infection, 10 days post-infection). f–h, Colon sections were stained with H&E. f, Combined scores as described in Fig. 1d. g, Crypt heights. h, Representative images. Scale bar, 50 μm. Data are representative of at least 3 independent experiments, n = 10 mice per group. Data are mean ± s.d. Two-tailed Student’s test (d) or one-way ANOVA (e–g), followed by post hoc Tukey multiple comparison test.

Administration of the DRD2 agonist sumanirole had similar effects to the Trp metabolites (Extended Data Fig. 7a–f), whereas the DRD2 antagonist L-741,626 blocked the effects of the Trp diet (Extended Data Fig. 7j–o). However, the effects of sumanirole were eliminated in the Drd2^fl/fl × Vil1-cre mice, whereas L-741,626 did not have an effect in these mice, indicating that DRD2 activation in IECs is responsible for the effects of the Trp diet and metabolites. Because serotonin can affect neuronal dopaminergic signalling^39,40 and gut luminal serotonin decreases C. rodentium virulence⁴¹, we treated mice with inhibitors of the serotonin selective reuptake transporter or tryptophan hydroxylase to increase or decrease serotonin levels, respectively, and found that the Trp diet enabled colonization resistance to C. rodentium irrespective of serotonin levels (Supplementary Fig. 7).

We further examined the effects of the Trp diet, IEt, IPyA and I3A in C. rodentium-infected mice lacking Drd2 expression in immune cells that, in wild-type mice, express DRD2 and are involved in host immunity against C. rodentium, including macrophages, dendritic cells and CD4⁺ T cells, using Drd2^fl/fl × LysM-cre (LysM is also known as Lyz2), Drd2^fl/fl × Cd11c-cre (Cd11c is also known as Itgax) and Drd2^fl/fl × Cd4-cre mice, respectively. The effects of the Trp diet were not eliminated in the mice lacking Drd2 in macrophages, dendritic cells and CD4⁺ T cells, indicating that the metabolites act via DRD2 expressed in IECs (Supplementary Fig. 8).

We then sought to determine mechanisms by which DRD2 activation mediates colonization resistance in IECs. The actin regulatory protein N-WASP is an initial host protein that is hijacked by AE pathogens during intestinal colonization and is required for pedestal formation and mucosal colonization of C. rodentium^11,12. We therefore examined the effects of the Trp diet and metabolites on the formation of actin pedestals by this protein during infection. We found that the Trp diet, IEt, IPyA and I3A all reduced pedestal formation during C. rodentium infection in wild-type or Drd2^fl/fl mice, and that this decrease was eliminated in Trp diet-fed, antibiotic-treated wild-type (Fig. 4a and Extended Data Fig. 6f) or Drd2^fl/fl × Vil1-cre mice (Fig. 4b and Extended Data Fig. 6g). We further found that the Trp diet and metabolites decreased N-WASP protein levels in the gut epithelium before and during C. rodentium infection in Drd2^fl//fl mice, and these effects were also abrogated in the Drd2^fl/fl × Vil1-cre mice (Extended Data Fig. 6h,i). Dopamine and the DRD2 agonist sumanirole also decreased actin pedestal formation and N-WASP levels in C. rodentium infection (Extended Data Fig. 7g–i and Supplementary Fig. 6g–i,o–q), whereas the DRD2 antagonist L-741,626 inhibited the effects of the Trp diet (Extended Data Fig. 7p–r and Supplementary Fig. 6w–y), and these effects were eliminated in the Drd2^fl/fl × Vil1-cre mice (Extended Data Fig. 7g–i). Further, we determined that IEt, IPyA and I3A act via DRD2 to protect against EHEC infection of Caco-2 monolayers through a similar mechanism (Extended Data Fig. 3f,k,l), suggesting that our findings may be generalizable to related AE human pathogens.

Fig. 4. Trp metabolites decrease actin pedestal formation in IECs during C. rodentium and EHEC O157:H7 infection via DRD2.

a, C57Bl/6 mice were pre-treated with antibiotics (ABX; 9 mg kg⁻¹ each metronidazole, ampicillin and neomycin, 4.5 mg kg⁻¹ vancomycin) for 7 days. The mice were then fed a conventional (2 g kg⁻¹ Trp diet, ad libitum) or Trp (42 g kg⁻¹ Trp diet, ad libitum) diet for 7 days or the Trp metabolites I3A (1,000 mg kg⁻¹), IPyA (2,900 mg kg⁻¹) or IEt (600 mg kg⁻¹) by oral gavage daily for 2 days. b, Drd2^fl/fl × Vil1-cre or Drd2^fl/fl mice were fed a conventional (2 g kg⁻¹ Trp diet, ad libitum) or Trp (42 g kg⁻¹ Trp diet, ad libitum) diet for 7 days or IEt (600 mg kg⁻¹) by oral gavage daily for 2 days. a,b, The mice were then administered C. rodentium (oral gavage, 10⁸ CFU) with continued antibiotics (except neomycin), Trp feeding or metabolite treatment (a) or Trp feeding or IEt treatment (b). Samples were stained with DAPI and Alexa Fluor 647-phalloidin. Pedestal formation was measured as the number of pedestals per host cell (ABX + Trp diet, n = 1,039; ABX + I3A, n = 1,044; ABX + IPyA, n = 934; ABX + IEt, n = 974; Vil1-cre: Trp diet, n = 1,028; IEt, n = 869 cells examined over 3 independent experiments). In box plots, box edges delineate interquartile range, the centre line is the median value, whiskers show lowest and highest values within 1.5 times the interquartile range from the first and third quartiles, and outliers beyond the whiskers are shown as dots. c–i, Caco-2 cells (DRD2 wild type (WT) versus knockout (KO)) were pre-treated with metabolites (100 μM) for 24 h and then infected with EHEC O157:H7 (multiplicity of infection (MOI) of 50) for 12 h. Cell lysates were analysed by western blot with the indicated antibodies. Source data are provided in Supplementary Fig. 9. d,f,h,i, Densitometry was performed using FIJI. n = 3 biological replicates. Data are representative of at least 3 independent experiments, n = 10 mice per group. Data are mean ± s.d. One-way ANOVA, followed by post hoc Tukey multiple comparison test. j, Model of activation of DRD2 in IECs by gut microbial Trp metabolites and the pathway to reduced pathogen burden via inhibition of actin pedestal formation. Dark blue, host actin regulatory proteins; grey, bacterial proteins.

To determine the molecular mechanism underlying how DRD2 activation in IECs leads to decreased pedestal formation and AE pathogen attachment, we assessed the dependence of the metabolite-induced effects on signalling pathways downstream of DRD2 using inhibitors of Gα_i, Gβγ and β-arrestin signalling. We found that inhibition of Gβγ, but not Gα_i or β-arrestin signalling (Extended Data Fig. 8), prevented the metabolite-induced decrease in N-WASP levels and increase in N-WASP phosphorylation, the latter of which leads to its subsequent proteasomal degradation⁴² (Extended Data Fig. 9a,f–h). Blockade of phospholipase C (PLC) and protein kinase C-θ (PKCθ), which are downstream of Gβγ³⁴, also eliminated these effects of the Trp metabolites on N-WASP, including prevention of the phosphorylation of Wiskot–Aldrich syndrome protein (WASP)-interacting protein (WIP), a PKCθ substrate⁴³ whose phosphorylation causes N-WASP⁴⁴ activation, leading to subsequent proteasomal degradation of N-WASP⁴² (Extended Data Figs. 9–12). Critically, this pathway was dependent on DRD2, indicating that the activation of DRD2 in IECs by Trp metabolites leads to decreased N-WASP levels, and thus decreased AE pathogen attachment and colonization via sequential activation of Gβγ, PLC and PKCθ (Fig. 4c–i and Extended Data Fig. 3g–m).

Here we have demonstrated a role in the gut for the neurotransmitter receptor DRD2 in promoting colonization resistance against enteric AE pathogens, including EHEC and C. rodentium, upon activation either by indole metabolites produced by the gut microbiota during dietary Trp supplementation or by pharmacological agonism (Fig. 4j). Although relatively high levels of Trp supplementation were necessary to reveal these effects, we found the Trp metabolites to be selective for DRD2 activation in IECs in this model, leading to decreased expression of a key host actin regulatory protein that is hijacked by these pathogens during initial colonization of the gut. These findings represent a unique pathway of colonization resistance that differs from previously described immune-mediated pathways because it depends on a neurotransmitter receptor that controls the actin cytoskeleton in the intestinal epithelium^45,46. Of note, recent high-throughput screens have shown that DRD2 can be activated by additional microbial metabolites^47–49, suggesting that this receptor may function more generally as a sensor of aromatic amino acid metabolites and that other metabolites may also activate this pathway. Thus, our studies reveal that DRD2 is an important mediator of host–microorganism interactions in the gut and can have an unconventional role in conferring host defence in the intestines beyond its canonical roles in the nervous system. DRD2 and its downstream pathways could serve as targets for prophylactics or therapeutic agents for improving gut health and treating gastrointestinal infection with certain AE pathogens.

Methods

Materials

Bacterial strains and culture.

EHEC EDL-933 and TUV93-0 were gifts from T. Doerr and J. Leong, respectively. C. rodentium DBS100 was obtained from G. Sonnenberg. Difco MacConkey agar was purchased from BD Biosciences. Bacteriological agar was purchased from VWR, yeast extract and tryptone were purchased from IBI Scientific, and sodium chloride was purchased from Fisher Scientific.

Tissue culture.

Caco-2 and HEK293T cells were obtained from and authenticated by the American Tissue Culture Company and cultured mycoplasma-free, according to their guidelines. HEK293T cells stably expressing a GloSensor cAMP reporter were obtained from T. Gardella. DMEM, penicillin/streptomycin, 0.05% trypsin, and DPBS were obtained from Corning, and Seradigm Premium Grade Fetal Bovine Serum (FBS) and Transfectagro were purchased from VWR. Cell culture Transwell inserts (transparent PET membrane, 12-well, 0.4-μm pore size) and Falcon 12-well companion plates were obtained from BD Falcon. Lipofectamine 2000, Lipofectamine RNAiMAX, and DMEM (low glucose) were purchased from Thermo Fisher Scientific. Polybrene and puromycin were purchased from EMD Millipore Sigma.

Metabolites and compounds.

Trp was purchased from Chem-Impex International. I3A and IPyA were obtained from Biosynth. IEt, indole-3-propionic acid and dopamine hydrochloride were obtained from Alfa Aesar. Kynurenine and sotrastaurin were purchased from Cayman Chemical Company. Indole, indole-3-acetamide, dl-indole-3-lactate, 5-hydroxytryptamine, indole-3-acetic acid and gallein were purchased from Sigma Aldrich. Tryptamine hydrochloride was obtained from TCI America, and indole-3-acrylate was purchased from Santa Cruz Biotechnology. Sumanirole maleate, L-741,626, fluoxetine hydrochloride (Prozac) and 4-chloro-dl-phenylalanine methyl ester hydrochloride were purchased from Neta Scientific. Pertussis toxin was purchased from Thermo Fisher Scientific. Barbadin was obtained from MedChem Express. PKCθ inhibitor and MG-132 were purchased from Selleck Chemicals.

Histology.

Haematoxylin was purchased from VWR, and eosin Y was obtained from Acros Organics. Canada Balsam was obtained from Ward’s Science, and xylenes was obtained from Macron Fine Chemicals.

Antibiotics.

Ampicillin and vancomycin hydrochloride were purchased from Sigma Aldrich. Neomycin sulfate hydrate and metronidazole were obtained from Alfa Aesar.

Western blotting.

DC Protein Assay kit was purchased from Bio-Rad. SuperSignal West Pico Chemiluminescent Substrate was purchased from Thermo Fisher Scientific. Bovine serum albumin (BSA) was purchased from VWR, and non-fat dry milk was purchased from Laboratory Product Sales. Protease inhibitor (complete) tablets were obtained from Roche. Sodium β-glycerophosphate was obtained from Alfa Aesar, and sodium orthovanadate was purchased from MP Biomedicals. Sodium fluoride was purchased from Chem-Impex International, and sodium pyrophosphate decahydrate was purchased from Fisher Scientific. Anti-DRD2 (clone B-10, sc-5303), anti-DRD3 (clone 9F4, sc-136170), anti-DRD4 (clone 2B9, sc-136169), anti-PKCθ (clone E-7, sc-1680), anti-WIP (clone A-7, sc-271113) and anti-N-WASP (clone C-1, sc-271484) were purchased from Santa Cruz Biotechnology. Alexa Fluor 647 mouse anti-WIP (pSer488, clone K32-824, 558674) was purchased from BD Biosciences. Anti-N-WASP (pTyr256, WP2601) was purchased from ECM Biosciences. Anti-mouse horseradish peroxidase conjugate (170-5947) and anti-GAPDH (clone 6C5, MCA4739) were purchased from Bio-Rad. Anti-β-actin (clone C-4, MA5-11869) was purchased from Thermo Fisher Scientific.

Immunofluorescence.

Paraformaldehyde (PFA) 32% solution, electron microscopy grade, was purchased from Electron Microscopy Sciences, and DAPI Prolong Diamond was purchased from Thermo Fisher Scientific. Fisher HealthCare Tissue Plus OCT Compound was obtained from Fisher Scientific. Donkey anti-mouse Alexa Fluor 594 (A21293) and Alexa Fluor 647-phalloidin (A22287) were purchased from Thermo Fisher Scientific.

LC–MS.

Liquid chromatography–mass spectrometry (LC–MS) grade methanol, water, acetonitrile and formic acid were obtained from Fisher Scientific.

Quantitative PCR.

RNABee was obtained from Tel-Test. Diethyl-pyrocarbonate and chloroform were purchased from Sigma Aldrich. Isopropanol and ethanol were purchased from VWR. Glycogen (Roche) was purchased from Krackeler Scientific. Random hexamers were purchased from Thermo Fisher Scientific. dNTP was purchased from BioBasic. MMLV reverse transcriptase was purchased from Clontech, and PerfeCta SYBR Green SuperMix, Low ROX, were obtained from Quanta Biosciences.

Primers for quantitative PCR.

EHEC tir forward (GAAGTCGGCACCTGCGAATCA) and reverse (GCATAGGGACCGTGCAGAATC), eae (intimin) forward (TGTCGCACTAACAGTCGCTT) and reverse (GCAACCACGGGAAATGATGG), espF forward (TTCACCGGAGTAAGACGCAC) and reverse (CTGCTTCTACACTAGGGCGG), tccp forward (TAGCTCCATCAGCGCAACAA) and reverse (GCGCTGCCTCACATTAGGA) and rpoA forward (GCGCTCATCTTCTTCCGAAT) and reverse (CGCGGTCGTGGTTATGTG) primers were purchased from IDT. C. rodentium tir forward (CAGGCTAAACGTCAGCAGGA) and reverse (TCGGCGGATTTCGTCTATGG), eae (intimin) forward (TCAGCATAGCGGAAGCCAAA) and reverse (TGCTACCGCCTTGCACATAA), espF forward (AATGGAATTGGTCAGGCCGT) and reverse (ACTGAAAAGCTCGCACCTCC) and rpoA forward (GCCCTGTTGACGATCTGGAA) and reverse (GCTCAACCTCAGTACGCTGT) primers were purchased from IDT.

Mice.

C57Bl/6, Drd2^fl/fl (020631, Drd2loxP), Vil1-cre (021504), Cd11c-cre (008068), LysM-cre (004781) and Cd4-cre (022071) mice were acquired from Jackson Laboratories. All mice were subsequently bred and maintained at the animal facility of Cornell University and used at 8–12 weeks of age. Age- and sex-matched male or female mice (5–10 per group) were co-housed for 7 days prior to use and fed Envigo Teklad global irradiated 18% protein rodent diet meal 2918 as the conventional diet. Mice were blindly assigned to groups in a random fashion.

Ethics statement.

Animal studies were approved by and performed in accordance with the guidelines of the Institutional Animal Care and Use Committee (protocol number 2015-0069).

Plasmids.

DRD2-Tango (plasmid #66269), pCDNA3.1(+)-CMV-bArres-tin2-TEV (plasmid #107245) and lentiCRISPRv2 puro (plasmid #98290) were obtained from Addgene. pCDNA3.1-DRD2 was generated by sub-cloning DRD2 into the pCDNA3.1 backbone using EcoRI and XhoI. VSVg and PAX2 packaging plasmids were obtained from J. Baskin.

In vivo experiments

Antibiotic treatment.

Mice were administered an antibiotic cocktail of ampicillin (9 mg kg⁻¹), metronidazole (9 mg kg⁻¹), neomycin (9 mg kg⁻¹) and vancomycin (4.5 mg kg⁻¹) via oral gavage every 12 h. Mice were pre-treated with antibiotics for 7 days prior to receiving tryptophan diet, metabolites, or dopamine hydrochloride. Mice were continued to be administered the antibiotic cocktail while receiving tryptophan diet, metabolites, or dopamine hydrochloride. Neomycin treatment was discontinued 24 h prior to C. rodentium infection, and ampicillin, metronidazole and vancomycin were continued to be administered for the remainder of the experiment.

Tryptophan diet.

Conventional diet (Envigo Teklad global irradiated 18% protein rodent diet meal 2918) or tryptophan (Trp) diet was provided to mice ad libitum in feeding jars. The tryptophan diet was prepared by supplementing conventional diet (2 g kg⁻¹ Trp diet) with an additional 40 g kg⁻¹ Trp diet. Mice received tryptophan diet for 7 days prior to C. rodentium infection and then for the remainder of the experiment.

Metabolite treatment.

Mice were administered I3A (1,000 mg kg⁻¹), IPyA (2,900 mg kg⁻¹) or IEt (600 mg kg⁻¹). I3A, IPyA and IEt were dissolved in vehicle (dimethylsulfoxide (DMSO)) and administered via oral gavage every 12 h. Mice were pre-treated with metabolite or vehicle for two days prior to C. rodentium infection and then for the remainder of the experiment by oral gavage.

Dopamine treatment.

Mice were pre-treated with dopamine hydrochloride (100 mg kg⁻¹) or vehicle (PBS) via intraperitoneal injection every 24 h for 2 days prior to C. rodentium infection and for the remainder of the experiment.

DRD2 agonist treatment.

Mice were pre-treated with sumanirole (4 mg kg⁻¹) or vehicle (PBS) via intraperitoneal injection every 24 h for 2 days prior to C. rodentium infection and for the remainder of the experiment.

DRD2 antagonist treatment.

Mice were pre-treated with L-741,626 (1 mg kg⁻¹) or vehicle (PBS) via intraperitoneal injection every 24 h for 2 days prior to Trp diet and for the remainder of the experiment.

SERT inhibitor treatment.

Mice were pre-treated with Prozac (20 mg kg⁻¹) or vehicle (water) via oral gavage every 24 h for 2 days prior to Trp diet and for the remainder of the experiment.

TpH1 inhibitor treatment.

Mice were pre-treated with 4-chloro-dl-phenylalanine methyl ester hydrochloride (150 mg kg⁻¹) or vehicle (PBS) via intraperitoneal injection every 24 h for 2 days prior to Trp diet and for the remainder of the experiment.

C. rodentium infection.

C. rodentium was grown in LB broth overnight with shaking at 37 °C. The next day, bacteria were subcultured until mid-log (OD₆₀₀ = 0.4–0.6) phase, and mice were infected with 10⁸ CFU by oral gavage. Mice were euthanized 10 days or 24 days post-infection.

Weight loss.

Mice were weighed daily at the same time each day. Each mouse weight was normalized to itself and control mice on day 0.

CFU quantification.

Faeces and colon tissue were weighed and homogenized using a Corning LSE vortex mixer, and a Brinkmann KINEMATICA Ch-6010 KRIENS-LU benchtop homogenizer, respectively. Homogenates were plated onto MacConkey agar and incubated for 24 h at 37 °C. C. rodentium colonies were counted to determine CFU per gram of faeces and colon tissue.

Histopathology and crypt height measurement.

The distal colon (2-cm segment) was flushed with PBS, excised and fixed in 10% neutral buffered formalin, paraffin-embedded, sectioned (5 μm) and stained with Harris haematoxylin and eosin Y. Samples were blinded, imaged using an Olympus CX41RF microscope equipped with Olympus cellSens Entry version 4.2. Tissue sections were assessed for submucosal oedema (scoring: 0, no change; 1, mild; 2, moderate; 3, profound), goblet cell depletion (scored based on numbers of goblet cells per high-power field: 0, ≥ 50; 1, 25–50; 2, 10–25; 3, ≤10), epithelial hyperplasia (scored based on percentage above the height of the control: 0, no change; 1, 1–50%; 2, 51–100%; 3, ≥100%), epithelial integrity (scoring: 0, no change; 1, ≤10 epithelial cells shedding per lesion; 2, 11–20 epithelial cells shedding per lesion; 3, epithelial ulceration; 4, epithelial ulceration with severe crypt destruction) and neutrophil and mononuclear cell infiltration (scoring: 0, none; 1, mild; 2, moderate; 3, severe). Data are expressed as the sum of these individual scores (range: 0–16). Crypt height was measured using Olympus cellSens Entry software for ten well-oriented crypts per mouse.

Cryosection immunofluorescence preparation.

A portion of the distal colon was frozen in OCT, and blocks were sectioned (5 μm) on a Thermo Scientific Microm HM 525 cryostat, adhered to a glass slide, washed with PBS, and fixed with 4% PFA prior to staining. Images are single z-slices of portions of the intestinal epithelial monolayer generated by cryosectioning of the epithelium proximal the apical surface.

Pedestal formation.

Using distal colon cryosections stained with DAPI Prolong Diamond and Alexa Fluor 647-phalloidin, bacteria-associated actin pedestals were visualized. The number of pedestals per host cell were quantified for each of the three replicate images per mouse. For box plots, interquartile ranges (boxes), median values (line within box), whiskers (lowest and highest values within 1.5 times the interquartile range from the first and third quartiles), and outliers beyond whiskers (dots), are shown.

Lysates for western blots.

Epithelial cells from the distal colon were isolated by mechanical disruption and lysed with 1× RIPA lysis buffer (150 mM sodium chloride, 1.0% Triton X-100, 0.5% sodium deoxycholate, 0.1% SDS, 25 mM Tris, 1 mM EDTA) with 1× protease inhibitor (cOmplete tablets) and sodium β-glycerophosphate (17.5 mM), sodium orthovanadate (1 mM), sodium fluoride (20 mM), and sodium pyrophosphate decahydrate (5 mM) added immediately prior to use. Samples were quantified using the DC Protein Assay kit and BioTek PowerWave XS2 plate reader with Molecular Devices SoftMax Pro software v 5.2 rev C.

LC–MS.

Faecal contents were collected fresh from mice and immediately flash frozen in liquid nitrogen. Frozen samples were dried on a VirTis Benchtop K Series Manifold Freeze Dryer. Dried samples were crushed and resolubilized in methanol (10 times the volume of the dry weight of the samples) and rocked at room temperature for 1 h before collecting the supernatant, which was then dried down. Immediately prior to LC–MS analysis, the samples were resuspended in methanol (10 times the volume of the dry weight of the samples) and filtered. LC–MS analysis was performed on an Agilent 6230 electrospray ionization–time-of-flight (ESI–TOF) mass spectrometer coupled to an Agilent 1260 HPLC equipped with an Agilent Poroshell 120 ECC18 reverse phase column (3 × 50 mm, 2.7 μm) using a flow rate of 0.5 ml min⁻¹. The gradient was ramped from 90% water and 0.1% formic acid (solvent A) and 10% acetonitrile and 0.1% formic acid (solvent B) to 50% A and 50% B for 0.5 min. The gradient was then ramped to 35% A and 65% B for an additional 0.5 min, then to 15% A and 85% B for 4.5 min, followed by 0% A and 100% B for 0.75 min. The gradient was then held constant at 0% A and 100% B for an additional minute. For detection, the mass spectrometer was equipped with a dual ESI source operating in positive or negative mode, acquiring in extended dynamic range from m/z 100–3,200 at 1 spectrum per s; gas temperature: 325 °C; drying gas 10 l min⁻¹; nebulizer: 20 psi; fragmentor: 80 V. Data were acquired using Agilent MassHunter Workstation Software LC/MS Data Acquisition for 6200 series TOF/6500 series Q-TOF version B.08.00 build 8.00.8058.3 SP1, and data analysis was carried out using Agilent MassHunter Workstation Software Qualitative Analysis version B.07.00 service pack 2 build 7.0.7024.29. Quantification of metabolites was determined by integrating the extracted ion count of the exact masses of the metabolites, which were determined using commercial standards. Standard curves in which known amounts of metabolite were utilized to determine the amount of each metabolite in each sample, which was normalized to the dry weight of the faecal samples.

16S rRNA gene sequencing and analysis.

Following 7 days of Trp diet or 2 days of Trp metabolite treatment, genomic DNA was extracted from mouse faeces using a QIAGEN DNeasy 96 PowerSoil Pro Kit. Amplicon libraries were created by PCR using universal bacterial primers targeting the V4 region of the 16S rRNA gene (515F and 806R)⁵⁰ and Classic++ Hot Start Taq DNA Polymerase Master Mix (TONBO Biosciences). Samples were amplified in duplicate using an Applied Biosystems SimpliAmp Thermal Cycler and an Eppendorf Mastercycler Gradient Thermal Cycler. PCR amplification conditions were as follows: 94 °C (3 min), 25 cycles of 94 °C (45 s), 50 °C (1 min), 72 °C (1.5 min), and 72 °C (10 min). The duplicate amplified products were pooled, and 16S rRNA gene amplicons were purified with Mag-Bind TotalPure NGS (Omega Bio-tek). Paired-end sequencing (2 × 250 bp) was performed on the Illumina MiSeq platform at the Cornell Institute of Biotechnology.

The 16S rRNA gene sequences were imported into Quantitative Insights into Microbial Ecology (QIIME2 version 2021.4) for analysis⁵¹. Sequences were demultiplexed, trimmed to 250 bp, and denoised using DADA2⁵². Sequences were clustered into de novo operational taxonomic units (OTUs) with 97% similarity using the Greengenes database (version 13_8, https://ftp.microbio.me/greengenes_release/gg_13_8_otus/)⁵³. Alpha and beta diversity analyses were computed using the core-metrics-phylogenetic function and Shannon diversity plots and weighted UniFrac distance matrices were generated using Vega Editor (QIIME2) to generate principal coordinate analysis plots. Permutational multivariate analysis of variance (PERMANOVA) was used to assess differences in beta diversity. Taxonomic classification using a pre-trained naive Bayes machine-learning classifier was performed and visualized using taxonomic bar graphs⁵³.

In vitro experiments

Caco-2 monolayers.

Caco-2 cells were seeded at 15,000 cells per transwell insert in 12-well companion plates. Monolayers were grown in DMEM supplemented with 10% FBS and penicillin/streptomycin at 37 °C. Medium was replenished every 2–3 days. On day 18, medium was replenished, and I3A, IPyA, IEt (100 μM) or vehicle (DMSO) was added. On day 20, medium and metabolites were replenished, and cells were infected with EHEC (MOI 50). After 16 h, cells were collected for western blot or immunofluorescence.

Haloperidol treatment.

Caco-2 monolayers were grown as above. Haloperidol (10 μM) or vehicle (DMSO) was added initially to monolayers on day 17 and replenished on day 18 and 20.

Inhibitor treatments.

Caco-2 cells were seeded at 10⁴ cells per well in 12-well plates or on glass coverslips for immunofluorescence. Cells were grown in DMEM supplemented with 10% FBS and penicillin/streptomycin at 37 °C. The Gα₁ inhibitor pertussis toxin (100 ng ml⁻¹) was added to cells 18 h prior to metabolite treatment. The β-arrestin inhibitor barbadin (100 μM), Gβγ inhibitor gallein (10 μM), PLC inhibitor U-73122 (10 μM) and PKC inhibitor sotrastaurin (5 μM) were added to cells 30 min prior to metabolite treatment. PKCθ inhibitor (5 μM) was added to cells 24 h prior to metabolite treatment. The proteasome inhibitor MG-132 (10 μM) was added to cells 1 h prior to metabolite treatment. Following pre-treatment with inhibitor, I3A, IPyA or IEt (100 μM) was added. After 24 h, cells were infected with EHEC (MOI 50), and 12 h post-infection, cells were collected for western blot or immunofluorescence.

PKCθ siRNA.

Control siRNA (sense): 5′-rCrGrUrUrArArUrCrGrCrGrUrArUrArArUrArCrGrCrGrUAT-3′, (antisense): 5′-rArUrArCrGrCrGrUrArUrUrArUrArCrGrCrGrArUrUrArArCrGrArC-3′, PKCθ (encoded by Prkcq) siRNA 1 (sense): 5′-rCrArArArGrArArUrUrCrUrUrArArArCrGArGrArArGrCCC-3′, (antisense): 5′-rGrGrGrCrUrUrCrUrCrGrUrUrUrArArGrArArUrUrCrUrUrUrGrUrC-3′, and PKCθ siRNA 2 (sense): 5′-rArGrGrUrUrUrCrArArGrArCrUrUrGrArUrArCrUrGrCrAAT-3′, (antisense): 5′-rArUrUrGrCrArGrUrArUrCrArArGrUrCrUrUrGrArArArCrCrUrUrU-3 were each incubated with Transfectagro and Lipofectamine RNAiMAX for 10 min prior to addition to Caco-2 cells. After 48 h, medium was replenished, and I3A, IPyA or IEt (100 μM) was applied to cells for 24 h, followed by infection with EHEC (MOI 50). After 12 h, cells were collected for western blot or immunofluorescence.

CRISPR–Cas9 knockout of Drd2, Drd3 and Drd4.

DRD2 forward (5′-ATGGGAGTTTCCCAGTGAAC-3′) and reverse (5′-GTTCACTGGGAAACTCCCAT-3′), DRD3 forward (5′-CAGGCCATTGCCGAAGACGA-3′) and reverse (5′-TCGTCTTCGGCAATGGCCTG-3′), and DRD4 forward (5′-CAACCTGTGCGCCATCAGCG-3′) and reverse (5′-CGCTGATGGCGCACAGGTTG-3′) guide RNAs were cloned into the lentiviral CRISPR plasmid lentiCRISPRv2 puro, following digestion with BsmBI. HEK293T cells were seeded at 10⁵ cells per well in 6-well plates. At 90–95% confluency, the cells were co-transfected with the lentiCRISPR plasmid and VSVg and PAX2 packaging plasmids using Lipofectamine 2000. The cells were incubated at 37 °C for 2 days. Afterwards, the HEK293T cell supernatant was collected and syringe-filtered using a 0.45-μm filter. For lentiviral transduction, filtered supernatant was added dropwise to Caco-2 cells seeded in a 10-cm dish, and the cells were incubated for 2 days at 37 °C. Polybrene (4 μg ml⁻¹) was added, and transduced cells were split 1:1 into medium containing puromycin (2 μg ml⁻¹). After selection, successful transduction was confirmed by western blotting.

Caco-2 immunofluorescence.

Caco-2 monolayers or cells grown on glass coverslips were cultured as above, washed with PBS, and fixed with 4% PFA prior to immunofluorescence.

Pedestal formation.

Using Caco-2 monolayers or cells grown on glass coverslips stained with DAPI Prolong Diamond and Alexa Fluor 647-phalloidin, bacteria-associated actin pedestals were visualized. The number of pedestals per Caco-2 cell were quantified for each of the three replicate samples per treatment. For box plots, interquartile ranges (IQRs, boxes), median values (line within box), whiskers (lowest and highest values within 1.5 times interquartile range from the first and third quartiles), and outliers beyond whiskers (dots), are shown.

Western blot lysate preparation.

Caco-2 monolayers or cells were grown as above, washed with PBS, and lysed with 1× RIPA lysis buffer (150 mM sodium chloride, 1.0% Triton X-100, 0.5% sodium deoxycholate, 0.1% SDS, 25 mM Tris, 1 mM EDTA) with 1× protease inhibitor (cOmplete tablets) and sodium β-glycerophosphate (17.5 mM), sodium orthovanadate (1 mM), sodium fluoride (20 mM), and sodium pyrophosphate decahydrate (5 mM) added immediately prior to use. Samples from biological replicates were pooled prior to quantification using the DC Protein Assay kit and BioTek PowerWave XS2 plate reader.

Growth curves and viable CFU counts.

EHEC or C. rodentium were grown in LB broth overnight with shaking at 37 °C. The next day, bacteria were cultured in fresh medium with I3A, IPyA or IEt (100 μM). Cultures were incubated with shaking at 37 °C. OD₆₀₀ absorbance readings were measured after 2, 4, 6, 8, 12 and 24 h using a BioTek PowerWave XS2 plate reader. At 24 h, cultures were plated on LB agar and incubated overnight at 37 °C. Colonies were counted to determine viable CFU.

RNA isolation and quantitative PCR.

EHEC or C. rodentium were grown in low glucose DMEM overnight with shaking at 37 °C with I3A, IPyA, or IEt (100 μM). The next day, bacteria were subcultured until late-log phase (OD₆₀₀ = 0.6–0.8) in low glucose DMEM with I3A, IPyA, or IEt (100 μM). Cultures were grown shaking at 250 rpm at 37 °C for aerobic conditions, standing at 37 °C for microaerophilic conditions, or standing at 37 °C under anaerobic conditions using an anaerobic chamber (Coy Lab AC16-113, maintained with the gas composition 3% hydrogen, 20% carbon dioxide, and 77% nitrogen, and equipped with an incubator (Coy Lab 2000)). Cells were lysed with RNABee prior to RNA purification according to the manufacturer’s instructions. RNA was quantified using a GE Nanovue. Using a Bio-Rad C1000 Touch Thermal Cycler, RNA was reverse transcribed using random hexamers and MMLV reverse transcriptase. cDNA samples were analysed using PerfeCta SYBR Green SuperMix, Low ROX, and a Bio-Rad CFX96 Real-Time PCR Detection System equipped with Bio-Rad CFX Maestro 1.1 version 4.1.2433.1219. PCR amplification conditions were as follows: 95 °C (3 min) and 40 cycles of 95 °C (15 s) and 60 °C (45 s). Relative expression of mRNA transcripts was normalized to the RNA polymerase subunit alpha rpoA. Data are represented as the fold induction over control samples.

Luciferase activity assays

Passive lysis buffer (5× PLB) was prepared with 125 mM Tris, pH 7.8, 10 mM 1,2-diaminocyclohexane tetraacetic acid (CDTA), 10 mM DTT, 5 mg ml⁻¹ BSA, 5% (vol/vol) Triton X-100, and 50% (vol/vol) glycerol in ddH₂O. An aqueous solution of 1× firefly luciferase substrate was prepared containing 75 mM HEPES, pH 8.0, 4 mM MgSO₄, 20 mM DTT, 0.1 mM EDTA, 0.53 mM ATP, 0.27 mM coenzyme A, and 0.47 mM D-luciferin (firefly) in ddH₂O. An aqueous solution of 1× Renilla luciferase buffer was prepared containing 7.5 mM sodium acetate, pH 5.0, 400 mM sodium sulfate, 10 mM CDTA, 15 mM sodium pyrophosphate, and 0.025 mM 2-(4-aminophenyl)-6-methylbenzothiazole. A 100× Renilla luciferase substrate was prepared by diluting coelenterazine to 0.55 mM in anhydrous methanol and added to 1× Renilla luciferase buffer immediately prior to the assay. Luminescence was collected on a Veritas microplate luminometer (Turner Biosystems) using software version 1.4.0.

DRD2-Tango assay.

HEK293T cells were transfected overnight with DRD2-Tango and pCDNA3.1(+)-CMV-bArrestin2-TEV. The next day, medium was replenished, and cells were treated for 24 h with I3A, IPyA, IET or dopamine hydrochloride (100 μM) or vehicle (DMSO). After 24 h, cells were washed with PBS, lysed with 1× PLB, and 20 μl of lysate was added to a 96-well white flat-bottom plate. Afterwards, 50 μl of 1× firefly luciferase substrate was added to each well, and luminescence was measured for 10 min using a Turner BioSystems Veritas Microplate Luminometer. Immediately after, 50 μl of the 1× Renilla substrate was added to each well, and luminescence was measured for 10 min. Luciferase activity was determined by calculating the ratio of the firefly luciferase signal to the Renilla luciferase signal.

GloSensor assay.

HEK293T cells stably expressing a GloSensor cAMP reporter were transfected overnight with pCDNA3.1-DRD2. The next day, medium was replenished, and the cells were incubated with d-luciferin (0.47 mM) for 30 min at room temperature. Cells were then treated for 15 min with I3A, IPyA, IEt or dopamine hydrochloride (100 μM, unless otherwise indicated for half-maximal effective concentration (EC₅₀) and dissociation constant (K_d) determination). After 15 min, cells were washed with PBS, lysed with 1× PLB, and 20 μl of lysate was added to a 96-well white flat-bottom plate. Afterwards, luminescence was measured as above using a Turner BioSystems Veritas Microplate Luminometer. EC₅₀ and K_d values were calculated using GraphPad Prism version 9.5.1 (733).

Western blot

Lysates were sonicated using a Heat Systems Ultrasonic Processor XL sonicator. Protein concentrations were determined using the DC Protein Assay kit and BioTek PowerWave XS2 plate reader. Lysates were resolved on SDS–polyacrylamide gels and transferred to nitrocellulose. Membranes were blocked with 5% BSA in 25 mM Tris, 150 mM sodium chloride, and 0.1% Tween-20 solution (TBS-T) rocking at room temperature for 1 h, then probed with the appropriate antibodies (DRD2, DRD3, DRD4, PKCθ, pWIP, WIP, pN-WASP and N-WASP primary antibodies, 1:1:000; anti-β-actin (1:20,000); and anti-GAPDH (1:2,500)) in 5% milk or BSA in TBS-T with 0.05% sodium azide with rocking at 4 °C overnight. Overnight incubation in primary antibody was followed by washing (3× TBS-T) and incubation with the appropriate species-specific HRP antibody (1:10,000) in 5% milk in TBS-T with rocking at room temperature for 1 h. Western blots were developed using SuperSignal West Pico Chemiluminescent Substrate on a Bio-Rad ChemiDoc MP using Image Lab Touch software version 2.4.0.03. Densitometry was performed using FIJI Image J 1.53c, Java 1.8.0_172 (64-bit) and normalized to the housekeeping protein and control lysate proteins. Source data are provided in the Supplementary Information (Supplementary Fig. 9).

Immunofluorescence

Fixed samples were permeabilized with 0.5% Triton X-100 in PBS for 15 min, blocked with 5% BSA in PBS for 1 h, and then incubated with the anti-DRD2 (1:200 dilution) in 5% BSA in PBS at room temperature for 2 h. Samples were incubated with donkey anti-mouse Alexa Fluor 594 antibody (1:500) in 5% BSA in PBS in the dark at room temperature for 1 h, then mounted with DAPI Prolong Diamond overnight. Alternatively, the samples were incubated with Alexa Fluor 647-phalloidin (1:500) in 5% BSA in PBS in the dark at room temperature for 1 h, then mounted with DAPI Prolong Diamond overnight. Samples were imaged with a Zeiss LSM 800 confocal laser scanning microscope equipped with 20× 0.8 NA and 40× 1.4 NA Plan Apochromat objectives, 405, 488, 561 and 640 nm solid-state lasers, and two GaAsP PMT detectors or a Zeiss LSM 880 confocal laser scanning microscope equipped with a 40× 1.4 NA Plan Apochromat objective, 405, 458, 488, 514, 561 and 633 nm solid-state lasers, two PMT channels, and a 32 channel GaAsP detector array. Fluorescence images were acquired using Zeiss ZEN 2.3 version 2.3.69.1018. Relative brightness of stained cells was quantified using FIJI and normalized to control cells.

Similarity ensemble approach

SEA³² was used to predict protein targets for I3A, IPyA and IEt. Nonhuman protein targets were excluded.

Statistical analysis

Experiments were completed at least three independent times. Error bars signify standard deviation from the mean. Unless otherwise indicated, statistical significance between two groups was evaluated with a two-tailed Student’s t-test using Microsoft Excel version 2108 (build 14326.20238) or determined using one-way ANOVA followed by post hoc Tukey test for multiple comparison analyses using R 4.2.2.

Extended Data

Extended Data Fig. 1. The Trp metabolites I3A, IPyA, and IEt protect against Citrobacter rodentium infection in mice.

C57Bl/6 mice were pre-treated with antibiotics (ABX) for 7 d, followed by Trp metabolites, I3A (1000 mg/kg), IPyA (2900 mg/kg), or IEt (600 mg/kg) by oral gavage daily for 2 d. The mice were then administered C. rodentium (CR, oral gavage, 10⁸ CFU) with continued ABX (except neomycin) and metabolite treatment for 10 d. Colon sections were stained with H&E. Shown are representative images. Scale bar: 50 μm. Data are representative of at least 3 independent experiments, n = 10 mice per group.

Extended Data Fig. 2. Tryptophan metabolites do not affect C. rodentium and EHEC growth and virulence in vitro.

C. rodentium (CR) (a–e) and EHEC O157:H7 (f–k) were cultured in the presence of I3A, IPyA, or IEt (100 μM). a, f, Growth was monitored by measuring OD600 absorbance readings over 24 h. b, g, Cultures were plated after 24 h, and CFUs were counted. (c–e, h–k) Bacteria were cultured in low glucose DMEM under anaerobic (ana), microaerophilic (micro), and aerobic (aero) conditions to activate locus of enterocyte effacement-pathogenicity island expression with I3A, IPyA, or IEt (100 μM). RNA was isolated after cultures reached late-log phase (OD600 = 0.6–0.8), and cDNA was synthesized and analyzed by qPCR for the indicated genes. Relative expression of mRNA transcripts was normalized to the RNA polymerase subunit alpha rpoA. Data are represented as the fold induction over control samples, bars = mean, error bars = standard deviation. Statistical analysis was performed using a two-tailed Student’s t-test (a–b, f–g), or one-way ANOVA, followed by post-hoc Tukey multiple comparison test (c–e, h–k), n = 3 biological replicates examined over 3 independent experiments.

Extended Data Fig. 3. Effects of I3A, IPyA, and IEt depend on dopamine receptor D2 (DRD2).

a–b, Polarized Caco-2 monolayers were pre-treated with haloperidol (HAL, 10 μM) for 24 h, followed by metabolites (I3A, IPyA, or IEt, 100 μM) for 2 d, and then infection with EHEC O157:H7 for 16 h. b, Representative images of pedestals (denoted by arrows) from Caco-2 cells stained with DAPI and Alexa Fluor 647-phalloidin and imaged by confocal microscopy. Shown are maximum intensity z-projections. Scale bar: 5 μm. c, Western blot analysis of Caco-2 monolayers to verify CRISPR/Cas9-mediated knockout (KO) of Drd2, Drd3, and Drd4. GAPDH is shown as a loading control. d–f, Caco-2 monolayers (WT vs. KO) were pre-treated with metabolites (I3A, IPyA, or IEt, 100 μM) for 2 d and then infected with EHEC O157:H7 for 16 h. (a, d–f) Pedestal formation = # of pedestals per Caco-2 cell (HAL: I3A, n = 955; IPyA, n = 970; IEt, n = 1024; Drd3 KO: I3A, n = 1089; IPyA, n = 1094; IEt, n = 1012; Drd4 KO: I3A, n = 1032; IPyA, n = 887; IEt, n = 1026; Drd2 KO: I3A, n = 1032; IPyA, n = 1026; IEt, n = 1073 cells examined over 3 independent experiments). For box plots, interquartile ranges (IQRs, boxes), median values (line within box), whiskers (lowest and highest values within 1.5 times IQR from the first and third quartiles), and outliers beyond whiskers (dots), are shown. g-m, Caco-2 cells (WT vs KO) were pre-treated with metabolites (I3A, IPyA, or IEt, 100 μM) for 24 h and then infected with EHEC O157:H7 for 12 h. Cell lysates were analyzed by Western blotting with the indicated antibodies. Source data are provided in Supplementary Fig. 9. (h, j, l, m) Densitometry was performed using FIJI, bars = mean, error bars = standard deviation. Data are representative of at least 3 independent experiments, n = 3 biological replicates. Statistical analysis was performed using one-way ANOVA, followed by post-hoc Tukey multiple comparison test.

Extended Data Fig. 4. I3A, IPyA, and IEt are ligands of dopamine receptor D2 (DRD2), which signals via Gαi, and Drd2 is knocked out in intestinal epithelial cells in Drd2^fl/fl × Villin-Cre mice fed Trp diet during C. rodentium infection.

a, c–f, HEK 293 T cells overexpressing either DRD2 and a split luciferase-based cAMP sensor (GloSensor) or b, DRD2-Tango and a β-arrestin-TEV fusion were incubated with dopamine (DA), I3A, IPyA, or IEt (1 mM each in a–b; concentrations indicated in c–f) for 15 min (a, c–f) or 24 h (b), after which luminescence was measured to quantify ligand-induced (a) decrease in cAMP or (b) increase in β-arrestin recruitment. RLU = relative luminescence units. EC50 and Kd values were calculated using GraphPad Prism. (g–h) Drd2^fl/fl × Villin (Vil)-Cre or Drd2^fl/fl mice were fed a conventional (2 g Trp/kg diet, ad libitum) or Trp (42 g Trp/kg diet, ad libitum) diet for 7 d and then infected with C. rodentium (CR, oral gavage, 10⁸ CFU) with continued Trp feeding. Ten days post-infection, intestinal cryosections were stained with DAPI and an anti-DRD2 antibody, followed by an anti-mouse Alexa Fluor 594 antibody. (g) Shown are representative z-slices. Scale bar: 20 μm. (h) Image brightness was quantified using FIJI. Statistical analysis was performed using one-way ANOVA, followed by post-hoc Tukey multiple comparison test; bars = mean, error bars = standard deviation, (a–f) n = 3 biological replicates and (g–h) n = 10 mice per group examined over 3 independent experiments.

Extended Data Fig. 5. Drd2 is knocked out in intestinal epithelial cells in Drd2^fl/fl x Villin-Cre mice administered Trp metabolites during C. rodentium infection.

Drd2^fl/fl x Villin (Vil)-Cre or Drd2^fl/fl mice were treated with Trp metabolites, I3A (1000 mg/kg), IPyA (2900 mg/kg), or IEt (600 mg/kg), by oral gavage daily for 2 d, and then infected with C. rodentium (CR, oral gavage, 10⁸ CFU) with continued metabolite treatment. Ten days post-infection, intestinal cryosections were stained with DAPI and an anti-DRD2 antibody, followed by an anti-mouse Alexa Fluor 594 antibody. (a) Shown are representative z-slices. Scale bar: 20 μm. (b–d) Image brightness was quantified using FIJI. Data are representative of at least 3 independent experiments, n = 10 mice per group, bars = mean, error bars = standard deviation. Statistical analysis was performed using one-way ANOVA, followed by post-hoc Tukey multiple comparison test.

Extended Data Fig. 6. Effects of the tryptophan (Trp) diet and metabolites I3A, IPyA, and IEt in protecting against C. rodentium infection depend on dopamine receptor D2 (DRD2) in intestinal epithelial cells (IECs).

Drd2^fl/fl x Villin (Vil)-Cre or Drd2^fl/fl mice were fed a conventional (2 g Trp/kg diet, ad libitum) or Trp (42 g Trp/kg diet, ad libitum) diet for 7 d or Trp metabolites, I3A (1000 mg/kg), IPyA (2900 mg/kg), or IEt (600 mg/kg), by oral gavage daily for 2 d, and then infected with C. rodentium (CR, oral gavage, 10⁸ CFU) with continued Trp feeding or metabolite treatment. a–b, Bacterial load in (a) feces and (b) colon tissue was measured (a) every 1–2 d for 24 d post-infection and (b) at the peak of infection, 10 d post-infection. c–e, Colon sections were stained with H&E and (c) blindly scored for submucosal edema (0-3), goblet cell depletion (0-3), epithelial hyperplasia (0-3), epithelial integrity (0-4), and neutrophil and mononuclear cell infiltration (0-3). Data are expressed as the sum of these individual scores (0-16). See Methods for full description of scoring rubric. (d) Crypt heights were measured. (e) Representative images. Scale bar: 50 μm. (f) Representative images of pedestals from Fig. 4a (denoted by arrows) stained with DAPI and Alexa Fluor 647-phalloidin and imaged by confocal microscopy. Shown are maximum intensity z-projections. Scale bar: 5 μm. (g) Pedestal formation = # of pedestals per host cell (I3A, n = 1079; IPyA, n = 1027 cells examined over 3 independent experiments). For box plots, interquartile ranges (IQRs, boxes), median values (line within box), whiskers (lowest and highest values within 1.5 times IQR from the first and third quartiles), and outliers beyond whiskers (dots), are shown. h, i, Intestinal epithelial cells were isolated, and cell lysates were analyzed by Western blotting with the indicated antibodies. Source data are provided in Supplementary Fig. 9. (i) Densitometry was performed using FIJI, n = 3 biological replicates. Data are representative of at least 3 independent experiments, n = 10 mice per group, bars = mean, error bars = standard deviation. Statistical analysis was performed using the two-tailed Student’s t-test (a) or one-way ANOVA, followed by post-hoc Tukey multiple comparison test: *p < 0.05, **p < 0.01, ***p < 0.001.

Extended Data Fig. 7. Synthetic DRD2 agonist protects against a mouse model of EHEC infection using C. rodentium, strain DBS100, whereas a DRD2 antagonist blocks the effects.

a–i, Drd2^fl/fl x Villin (Vil)-Cre or Drd2^fl/fl mice were pre-treated with DRD2 agonist sumanirole (SUM, 4 mg/kg, IP injection) daily for 2 d. j–r, Drd2^fl/fl x Vil-Cre or Drd2^fl/fl mice were pre-treated with DRD2 antagonist L-741,626 (1 mg/kg, IP injection) daily for 2 d, followed by conventional (2 g Trp/kg diet, ad libitum) or Trp (42 g Trp/kg diet, ad libitum) diet for 7 d. a–r, The mice were then administered C. rodentium (CR, oral gavage, 10⁸ colony-forming units, CFU) with continued (a–i) SUM treatment or (j–r) L-741,626 and Trp feeding. a, j Timeline for (a) SUM and (j) L-741,626 study. b–c, k–l, Bacterial load in (b, k) feces and (c, l) colon tissue was measured (b, k) every 1–2 d for 10 d post-infection and (c, l) at the peak of infection, 10 d post-infection. (d–f, m–o) Colon sections were stained with H&E and (d, m) blindly scored for submucosal edema (0-3), goblet cell depletion (0-3), epithelial hyperplasia (0-3), epithelial integrity (0-4), and neutrophil and mononuclear cell infiltration (0-3). Data are expressed as the sum of these individual scores (0-16). See Methods for full description of scoring rubric. (e, n) Crypt heights were measured. (f, o) Representative images. Scale bar: 50 μm. (g, p) Intestinal cryosections were stained with DAPI and Alexa Fluor 647-phalloidin. Pedestal formation = # of pedestals per host cell (SUM, n = 883; L-741,626, n = 1775 cells examined over 3 independent experiments). For box plots, interquartile ranges (IQRs, boxes), median values (line within box), whiskers (lowest and highest values within 1.5 times IQR from the first and third quartiles), and outliers beyond whiskers (dots), are shown. h–i, q–r, Intestinal epithelial cells were isolated, and cell lysates were analyzed by Western blotting with the indicated antibodies. Source data are provided in Supplementary Fig. 9. (q) Samples derive from the same experiment, and Western blots were processed in parallel. (i, r) Densitometry was performed using FIJI, n = 3 biological replicates. Data are representative of at least 3 independent experiments, n = 10 mice per group, bars = mean, error bars = standard deviation. Statistical analysis was performed using the two-tailed Student’s t-test (b, k) or one-way ANOVA, followed by post-hoc Tukey multiple comparison test: ***p < 0.001.

Extended Data Fig. 8. Effects of Trp metabolites do not depend on Gαi and β-arrestin signaling.

Caco-2 cells were pre-treated with (a–d) Gαi inhibitor pertussis toxin (PTx, 100 ng/mL) for 18 h or (e–h) β-arrestin inhibitor barbadin (100 μM) for 30 min, followed by Trp metabolite (I3A, IPyA, or IEt, 100 μM) for 24 h. The cells were then infected with EHEC (MOI 50) for 12 h, (a, e) fixed, and stained with DAPI and Alexa Fluor 647-phalloidin. Pedestal formation = # of pedestals per Caco-2 cell (PTx: I3A, n = 994; IPyA, n = 916; IEt, n = 905; barbadin: I3A, n = 934; IPyA, n = 993; IEt, n = 1087 cells examined over 3 independent experiments). For box plots, interquartile ranges (IQRs, boxes), median values (line within box), whiskers (lowest and highest values within 1.5 times IQR from the first and third quartiles), and outliers beyond whiskers (dots), are shown. b–d, f–h, Alternatively, cells were lysed and analyzed by Western blotting with the indicated antibodies. Source data are provided in Supplementary Fig. 9. (c–d, g–h) Densitometry was performed using FIJI. Data are representative of at least 3 independent experiments, n = 3 biological replicates, bars = mean, error bars = standard deviation. Statistical analysis was performed using one-way ANOVA, followed by post-hoc Tukey multiple comparison test.

Extended Data Fig. 9. Effects of Trp metabolites depend on Gβγ and phospholipase C.

Caco-2 cells were pre-treated with (a–h) Gβγ inhibitor gallein (10 μΜ) for 30 min or (i–p) PLC inhibitor U-73122 (10 μM) for 30 min, followed by Trp metabolite (I3A, IPyA, or IEt, 100 μM) for 24 h. The cells were then infected with EHEC (MOI 50) for 12 h, (a, i) fixed, and stained with DAPI and Alexa Fluor 647-phalloidin. Pedestal formation = # of pedestals per Caco-2 cell (gallein: I3A, n = 944; IPyA, n = 1073; IEt, n = 905; U-73122: I3A, n = 977; IPyA, n = 1068; IEt, n = 1075 cells examined over 3 independent experiments). For box plots, interquartile ranges (IQRs, boxes), median values (line within box), whiskers (lowest and highest values within 1.5 times IQR from the first and third quartiles), and outliers beyond whiskers (dots), are shown. b–h, j–p, Alternatively, cells were lysed and analyzed by Western blotting with the indicated antibodies. Source data are provided in Supplementary Fig. 9. (c, e, g, h, k, m, o, p) Densitometry was performed using FIJI. Data are representative of at least 3 independent experiments, n = 3 biological replicates, bars = mean, error bars = standard deviation. Statistical analysis was performed using one-way ANOVA, followed by post-hoc Tukey multiple comparison test.

Extended Data Fig. 10. Effects of Trp metabolites depend on protein kinase C (PKC).

Caco-2 cells were pre-treated with pan-PKC inhibitor sotrastaurin (Sotra, 5 μM) for 30 min, followed by Trp metabolite (I3A, IPyA, or IEt, 100 μM) for 24 h. The cells were then infected with EHEC (MOI 50) for 12 h, (a) fixed, and stained with DAPI and Alexa Fluor 647-phalloidin. Pedestal formation = # of pedestals per Caco-2 cell (I3A, n = 912; IPyA, n = 1138; IEt, n = 1086 cells examined over 3 independent experiments). For box plots, interquartile ranges (IQRs, boxes), median values (line within box), whiskers (lowest and highest values within 1.5 times IQR from the first and third quartiles), and outliers beyond whiskers (dots), are shown. b–f, Alternatively, cells were lysed and analyzed by Western blotting with the indicated antibodies. Source data are provided in Supplementary Fig. 9. (c, e, f) Densitometry was performed using FIJI. Data are representative of at least 3 independent experiments, n = 3 biological replicates, bars = mean, error bars = standard deviation. Statistical analysis was performed using one-way ANOVA, followed by post-hoc Tukey multiple comparison test.

Extended Data Fig. 11. Effects of Trp metabolites depend on protein kinase C (PKC)-θ.

Caco-2 cells were pre-treated with (a–f) isoform-selective PKC-θ inhibitor PKC-θi, 5 μM) for 24 h, or (g–l) PKC-θ was knocked down using two different siRNA duplexes (1 and 2) or a negative control siRNA duplex (C), followed by Trp metabolite (I3A, IPyA, or IEt, 100 μM) for 24 h. The cells were then infected with EHEC (MOI 50) for 12 h, (a, g) fixed, and stained with DAPI and Alexa Fluor 647-phalloidin. Pedestal formation = # of pedestals per Caco-2 cell (PKCθi: I3A, n = 1063; IPyA, n = 1005; IEt, n = 1033; siRNA: I3A, n = 1680; IPyA, n = 1462; IEt, n = 1502 cells examined over 3 independent experiments). For box plots, interquartile ranges (IQRs, boxes), median values (line within box), whiskers (lowest and highest values within 1.5 times IQR from the first and third quartiles), and outliers beyond whiskers (dots), are shown. b–f, h–l, Alternatively, cells were lysed and analyzed by Western blotting with the indicated antibodies. Source data are provided in Supplementary Fig. 9. (c, e, f, i, k, l) Densitometry was performed using FIJI. Data are representative of at least 3 independent experiments, n = 3 biological replicates, bars = mean, error bars = standard deviation. Statistical analysis was performed using one-way ANOVA, followed by post-hoc Tukey multiple comparison test.

Extended Data Fig. 12. Effects of Trp metabolites depend on proteasomal degradation.

Caco-2 cells were pre-treated with proteasomal inhibitor MG-132 (10 μM) for 1 h, followed by Trp metabolite (I3A, IPyA, or IEt, 100 μM) for 24 h. The cells were then infected with EHEC (MOI 50) for 12 h, (a) fixed, and stained with DAPI and Alexa Fluor 647-phalloidin. Pedestal formation = # of pedestals per Caco-2 cell (I3A, n = 1079; IPyA, n = 999; IEt, n = 911 cells examined over 3 independent experiments). For box plots, interquartile ranges (IQRs, boxes), median values (line within box), whiskers (lowest and highest values within 1.5 times IQR from the first and third quartiles), and outliers beyond whiskers (dots), are shown. b–h, Alternatively, cells were lysed and analyzed by Western blotting with the indicated antibodies. Source data are provided in Supplementary Fig. 9. (c, e, g, h) Densitometry was performed using FIJI. Data are representative of at least 3 independent experiments, n = 3 biological replicates, bars = mean, error bars = standard deviation. Statistical analysis was performed using one-way ANOVA, followed by post-hoc Tukey multiple comparison test.

Data availability

Next-generation sequencing reads have been deposited at NCBI BioProject under accession number PRJNA1049399. Source data are provided with this paper.