Adaptation of the human aryl hydrocarbon receptor to sense microbiota-derived indoles

To investigate the impact of indole upon AHR-mediated gene expression, human HepG2 (40/6) cells stably harboring an AHR responsive luciferase reporter construct were incubated with vehicle, 10 nM TCDD or increasing concentrations (1–100 μM) of re-crystallized indole as indicated (). Exposure to indole resulted in a dose-dependent increase in reporter expression with an EC~ 3 μM. A significant 2-fold induction over vehicle treated was observed at 1 μM and maximal 7-fold expression evident at 100 μM, the highest concentration examined and equivalent to the induction obtained with a saturating dose of the prototypical AHR agonist, TCDD. Such data indicates that indole stimulates canonical dioxin response element (DRE)-dependent AHR-mediated gene expression in the context of human AHR. Fig. 1A 50

The AHR is known to exhibit species-dependent sensitivity with regard to its activation potential. We therefore examined the capacity of indole to influence AHR-mediated gene expression in mouse Hepa1.1 cells stably harboring an AHR responsive luciferase reporter construct. Hepa1.1 cells were incubated with vehicle, 10 nM TCDD or increasing concentrations (1–100 μM) of indole as indicated (). In contrast to the human HepG2 (40/6) cell line, exposure to indole resulted in a modest but significant increase in reporter activity at 100 μM. This represented only 7% of the activity exhibited with 10 nM TCDD, thus suggesting that indole is a weak partial agonist for the mouse AHR. Further examination of the sensitivity of AHR-mediated gene expression by indole was performed using the rat H4IIE1.1 luciferase reporter cell line and revealed that rat AHR-mediated gene expression, similar to mouse is weakly responsive to induction by indole at the doses examined (). Contrary to these data, previous studies have reported that indole is antagonistic with regard to AHR-dependent gene expression. In order to validate the observed inductive capacity of indole and eliminate potential AHR agonist contamination of our indole source, we re-examined the sensitivity of human HepG2 (40/6) using re-crystallized, HPLC-purified and commercial gradeH-NMR validated indole. AHR-mediated reporter expression in HepG2 (40/6) cells exposed to 10 μM recrystallized or HPLC-purified indole yielded essentially identical activity. In contrast, commercial grade indole failed to exhibit significant luciferase reporter activity (). Perhaps explaining the differing results obtained by others. These data indicate that human AHR, in contrast to rodent, is sensitive to indole-mediated activation at low μM concentrations. Fig. 1B Fig. S2 ¹⁷ Fig. S3 ¹⁷ 1

To examine indole-mediated human AHR activation within the context of endogenous gene expression rather than a heterologous reporter system, quantitative AHR target gene expression was assessed in colonic epithelial Caco2 cells exposed to vehicle, 10 nM TCDD, 20 and 100 μM indole, as indicated (). Induction of the AHR target genesandby TCDD and indole was consistent with the previous reporter cell line data with 20 μM indole displaying inductive capacity equivalent to 10 nM TCDD by eliciting significant 400 and 40-fold increases inand, respectively (, B). Increased expression ofby indole was shown to be dependent on AHR activation through competitive antagonism with AHR antagonist GNF351 (). Further analysis ofmRNA expression in response to indole exposure revealed that the observed induction ofandcould not be attributable to enhanced AHR expression (). Similar analysis ofandexpression performed in the mouse Hepa1 cell line exposed to indole failed to recapitulate the induction ofobserved with human Caco2 cells, further demonstrating species-specific AHR activation by indole (). Examination of human Caco2 CYP1A1 enzymatic activity using an EROD-based methodology demonstrated that indole-mediatedinduction is not restricted to mRNA but is reflected at the level of CYP1A1 protein/activity. In response to 12 h exposure to 100 μM indole, CYP1A1 enzymatic activity is significantly enhanced ~4-fold over vehicle treated controls (). This indole-mediated induction proved to be lower than that obtained with 10 nM TCDD, which may be a function of the half-life of indole when compared to the poorly metabolized TCDD. Fig. 2 Fig. 2A Fig. S4 ²⁴ Fig. 2C Fig. S5 Fig. 2D CYP1A1 CYP1B1 CYP1A1 CYP1B1 CYP1A1 AHR CYP1A1 CYP1B1 Cyp1a1 Cyp1b1 CYP1A1/B1 CYP1A1

In addition to stimulating direct DRE-mediated transcription, activated AHR has recently been demonstrated to act in a combinatorial fashion with inflammatory cytokine signaling to facilitate synergistic induction of interleukin-6 (). We therefore examined the capacity of indole to activate AHR-mediated gene expression within the more complex context ofsynergy. Caco2 cells were exposed to vehicle, 10 ng/ml IL1B, 20 μM indole, or a combination of IL1B together with indole, as indicated, and followed by quantitative PCR analysis ofmRNA (). Exposure to indole failed to elicit a significant induction of; however combinatorial treatment with indole and IL1B prompted a robust and significant 3-fold synergistic induction ofexpression when compared to IL1B treatment alone. In order to demonstrate AHR-dependency with regard to synergistic IL6 expression by indole and IL1B, we utilized the AHR competitive antagonist GNF351. Caco2 cells were pre-treated with 200 nM GNF351 for 1 h prior to 20 μM indole and 10 ng/ml IL1B exposure for an additional 4 h (). Treatment with GNF351 significantly suppressed the indole/IL1B-mediated synergistic induction ofby 50% without influencing the stimulatory action of IL1B in isolation. This observation indicates that the indole component ofsynergy is dependent upon human AHR. IL6 IL6 IL6 IL6 IL6 IL6 IL6 ²⁵ Fig. 2E ²⁴ Fig. 2E

To establish whether indole/IL1B-mediatedinduction by Caco2 cells is reflected at the protein level, an IL6 ELISA assay was performed. Caco2 cells were treated with vehicle, 10 nM TCDD or 100 μM indole, in isolation or in combination with 10 ng/ml IL1B, as indicated (). Following 24 h treatment, conditioned media was collected and assayed for secreted IL6 protein. Data obtained were consistent with AHR agonist-mediated synergistic induction ofmRNA. In the context of IL1B exposure, both TCDD and indole stimulated IL6 protein secretion ~3-fold over that observed with IL1B alone, indicating that IL6 protein synthesis is influenced by indole exposure. IL6 IL6 Fig. 2F

To examine human AHR activation by indole in non-transformed cell lines, we utilized primary peritoneal macrophages (Mϕ) derived from commercially available ‘humanized’ AHR mice compared with wild-type C57BL6 mice expressing the mouse AHR. Expression of human and mouse AHR by ‘humanized’ AHR and control mice, respectively, was established through AHR-specific immunoblot analysis (). Peritoneal Mϕ derived from ‘humanized’ and wild-type controls were exposed (4 h) to vehicle, 10 μM indole or 500 pM indirubin andmRNA expression quantified though PCR (). Exposure to indirubin elicited a significant 20-fold increase inexpression by ‘humanized’ AHR Mϕ but limited ability to induce in wild-type Mϕ, consistent with previous reports demonstrating human AHR-selective activation by indirubin, thus demonstrating ‘humanized’ AHR Mϕ are fully permissive for AHR-mediated transcription by human-selective activators. Exposure to indole stimulatedexpression in ‘humanized’ AHR Mϕ by 5-fold but exhibited an attenuated ability to induce expression in identically treated wild-type Mϕ. These data further demonstrate that indole exhibits a capacity to activate AHR-mediated transcription in a species-selective manner and support that the hypothesis that indole would modulate immune cell activity through AHR activation in humans and to a lesser extent in rodents. Fig. S6 Fig. 2G ²⁶ Cyp1a1 Cyp1a1 Cyp1a1

Species-specific activation of human AHR-mediated transcription indicated that indole is a putative AHR ligand. To investigate this notion further, competitive ligand binding assays were performed utilizing hepatic cytosol derived from mice expressing the humantransgene under the control of the hepatocyte-specificpromoter (). Human AHR liver cytosol, incubated under saturating conditions with the AHR photoaffinity ligand (PAL) and increasing concentrations of indole, revealed a dose-dependent decrease in PAL binding consistent with indole being a competitor and direct ligand for the human AHR. A comparison of the binding affinities of indole with the known AHR ligand beta-napthoflavone (βNF) suggests that the relative affinity of human AHR for indole is orders of magnitude lower than for βNF. Competitive ligand binding assays performed using hepatic cytosol derived from wild-type C57BL6 mice demonstrated the expected competition with βNF, but showed a lack of indole-mediated competition at the doses examined. Higher concentrations of indole were not examined due to non-specific effects upon PAL binding. The absence of indole-mediated competition with wild-type murine AHR is consistent with previous reporter gene expression and quantitative PCR analysis, which demonstrate that indole is not an effective activator of the mouse AHR (). AHR albumin Fig. 3 Fig. 3

Additional evidence for the human AHR agonist potential of indole was obtained by performing nuclear translocation and DNA binding assays. Sub-cellular localization of AHR in human-derived HepG2 cells incubated (1 h) with vehicle or 100 μM indole was assessed by immunoblotting and demonstrated a redistribution of AHR from the cytoplasm into the nucleus upon treatment with indole, consistent with the action of an AHR agonist (). Similar localization studies performed using mouse-derived Hepa1 cells failed to exhibit significant nuclear enrichment of AHR following incubation with indole (). Consistent with the observed human-specific nuclear redistribution of AHR following exposure to indole, DNA retardation assays exhibited the capacity of indole to facilitate binding oftranslated human AHR/ARNT but not mouse AHR/ARNT to its cognate DNA response element (). Fig. 4A Fig. 4A Fig. 4B in vitro

The establishment of indole as an agonist for the human AHR raised the question: are there other biologically relevant indole derivatives that are ligands for the human AHR? To address this question, HepG2 (40/6) cells were incubated with vehicle, 10 nM TCDD, 10 μM indole, 1–10 μM 3-methyl indole (skatole), 2-oxindole or 3-indole propionic acid, as indicated and luciferase reporter activity determined. The data obtained demonstrates that 3-methyl indole and 2-oxindole can stimulate the human AHR, with both indole derivatives exhibiting a dose-dependent increase in luciferase reporter activity and an efficacy equivalent to indole (). In contrast, 3-indole propionic acid stimulated reporter activity but failed to exhibit dose-dependency. Complementary studies performed using Hepa 1.1 cells largely failed to identify a stimulatory effect associated with the indole derivatives (). However, 3-methyl indole does exhibit modest dose-dependent activation of the mouse AHR. Structure-activity relationships of human AHR for indole derivatives were investigated further by exposure of human HepG2 (40/6) luciferase reporter cells to isomers of methyl indole (1-methyl indole, 2-methyl indole and 3-methyl indole). Cells were treated with vehicle, 10 nM TCDD, 10 μM indole or 1–10 μM methyl indole isomers, as indicated (). The data reveal a dose-dependent increase in AHR activity associated with exposure to 3-methyl indole that is equivalent to that observed with indole. However, no significant AHR activity was evident with either 1-methyl or 2-methyl indole. Identical structure-activity associations were observed by analyzing/mRNA expression together with CYP1A1 enzyme activity () and synergisticmRNA/protein expression in Caco2 cells (). Electrophoretic mobility shift assays (EMSA) demonstrated that the elevated AHR activity observed with 3-methyl indole, like indole was associated with enhanced binding of AHR/ARNT to its cognate response element (). Complementary studies performed using Hepa1 cells revealed no significant AHR activity following exposure to indole or any of the isomers of methyl indole tested (). Fig. 5 Fig. S7 Fig. 6A Fig. 6B,C Fig. 6D Fig. 6E Fig. S7 CYP1A1 B1 IL6

The data suggest that the AHR activities associated with both indole and 3-methyl indole are selective for the human AHR. Furthermore, direct ligand binding and subsequent nuclear translocation and DNA binding facilitate the enhanced transcriptional activity of human AHR elicited by indole. In an effort to understand the molecular basis for such ligand binding and selectivity,modeling of human and mouse AHR ligand binding domains (LBD) (amino acid residues 247–290 or 241–284, human or mouse respectively) was conducted in the context of indole and 3-methyl indole (). Homology modeling of human and mouse AHR LBD docked with indirubin provided an optimized model exhibiting the most energetically favorable LBD conformation for ligand binding (). Subsequent docking simulations, using these optimized LBD conformations in the context of indole, 3-methyl, or 2-methyl indole binding, revealed no significant difference in LBD conformation or free energy calculations that can account for the observedexperimental evidence indicating human AHR selectivity (). The previously established specificity of the human AHR for indirubin, which closely resembles two covalently linked indole moieties, suggested the novel concept that stoichiometry of human AHR/indole binding may be 2:1 rather than 1:1. This two indole binding hypothesis was examined using theLBD models of human or mouse AHR and identified a favorable conformation associated with two molecules within the human but not the mouse AHR LBD (;). Similar modeling predictions were performed using 3-methyl indole, which exhibits human AHR-selective activity(). Data obtained substantiate the two indole binding hypothesis, a favorable conformation was observed for human AHR and 3-methyl indole with a simulated stoichiometry of 2:1. However, this stoichiometry was not permissive when modeled with the mouse AHR LBD (). Binding of 2-methyl indole moieties at this ratio proved to be energetically unfavorable in the context of human AHR. in silico in vitro in silico in vitro Fig. 7 Fig. 7A,B Fig. S8 Fig. 7C,D Table S1 Fig. 7E Fig. 7F

Activation of human AHR by indole but not mouse AHR represents a gain of function for the human AHR and indicates an evolutionary divergence within thelocus. In order to examine whether this divergence is a specialization restricted to humans and therefore a recent adaptation, or a characteristic of thefamily of primates in general and therefore a more distant mammalian divergence, we investigated the capacity of indole to activate(Chimpanzee) AHR. Treatment oftranslated mouse or chimpanzee AHR with vehicle, 10 nM TCDD or 20 μM indole, followed by EMSA, revealed chimpanzee AHR to be sensitive to both TCDD and indole-mediated activation, as evidenced by a robust level of AHR/ARNT/DRE complex formation ().modeling of the chimpanzee AHR ligand binding domain in the context of indole identified the favorable bimolecular binding conformation previously observed with human AHR (). A comparison between the amino acid sequences that comprise the ligand binding domain of chimpanzee, human and mouse AHR reveals near 100% homology between chimpanzee and human, with only a single residue substitution at position 381 (valine and alanine, human, chimpanzee respectively) (). However, the homologous residue in mouse AHR is also an alanine residue and therefore is unlikely to be a determinant of indole binding. Ahr Hominidae Pan troglodytes in vitro In silico Fig. S9 Fig. S9 Fig. S9

C57BL/6J,, and TaconicC57BL/6-mice were housed on corncob bedding in a temperature- and light-controlled facility and given access to food and water. Mice were maintained in a pathogen-free facility and treated humanely with approval from the Animal Care and Use Committee of the Pennsylvania State University and methods were carried out in accordance with approved guidelines. Adult (10–12 weeks) mice were used for macrophage isolation experiments. AHR Ahr Cre Ahrtm1.1(AHR)Arte ad libitum Ttr fx/fx Alb ©

Hepa1, HepG2 and their respective AHR-reporter derivatives harboring the stably integrated pGudluc 1.1 or 6.1 constructs were maintained in α-modified essential media (Sigma-Aldrich, St. Louis, MO) supplemented with 8% fetal bovine serum (Hyclone Laboratories, Logan,). The Caco-2 human colon carcinoma cell line was maintained in α-MEM with 20% FBS. Primary peritoneal Mϕ cells were maintained in DMEM (Gibco, Carlsbad, CA) supplemented with 10% fetal bovine serum (FBS) (Hyclone Labs, Logan, UT), 2 mM L-Glutamine, and 1 mM sodium pyruvate (Sigma, St. Louis, MO). Cells were cultured at 37 °C in a humidified atmosphere composed of 95% air and 5% COin the presence of 100 IU/ml penicillin/100 μg/ml streptomycin (Sigma-Aldrich). 2

Mice (mand h) were injected with 3 ml of 3% thioglycolate media intraperitoneally on day one. Approximately 72 h post-thioglycolate injection, mice were euthanized. Primary Mϕ were isolated by peritoneal lavage in ice-cold phosphate buffered saline (PBS). Cells were centrifuged and re-suspended in Mϕ culture media for 4 h. After 4 h cells were washed with PBS and incubated overnight in Mϕ media. Cells were treated the following day for 4 h unless otherwise described in figure legends. Ahr Ahr b ⁴¹

Characterization of competitive binding within the AHR ligand binding pocket between the AHR photoaffinity ligand, 2-azido-3-[I]iodo-7,8-dibromodibenzo--dioxin and indole was performed essentially as described previously. 125 p ²⁶

The reporter cells (Hepa 1.1/Hep G2 40/6) were seeded in twelve-well plates and cultured to 90% confluence. Cells were treated as indicated for 4 h then lysed in 400 μl of lysis buffer [25 mM Tris-phosphate, pH 7.8, 2 mM dithiothreitol, 2 mM 1,2-diaminocyclohexane-,,′,′-tetraacetic acid, 10% (v/v) glycerol, and 1% (v/v) Triton X-100]. Lysate (20 μl) was combined with 80 μl of Luciferase Reporter Substrate (Promega, Madison, WI), and luciferase activity was measured with a TD-20e luminometer (Turner Designs, Sunnyvale, CA). N N N N

Total RNA was isolated from cells using TRI Reagent (Sigma-Aldrich), followed by reverse transcription using the High Capacity cDNA Archive kit (Applied Biosystems, Foster City, CA) according to the manufacturer’s protocols. PerfeCTa SYBR Green SuperMix for iQ (Quanta Biosciences, Gaithersburg, MD) was used to determine mRNA levels, and analysis was conducted using MyIQ software, in conjunction with a MyIQ-single-color PCR detection system (Bio-Rad Laboratories, Hercules, CA).

To quantify protein expression, media was collected from Caco-2 cells at 12 and 24 h post treatment and stored at −80 °C. IL6 content of media samples was determined via LEGEND MAX™ Human IL6 ELISA kit with pre-coated plates (Biolegend, San Diego, Ca.) according to manufacturer’s protocol.

Activity of CYP1A1 was determined using P450-Glo™ CYP1A1 Assay following the manufacturer’s intstructions (Promega, Madison, WI). Caco-2 cells were cultured to 90% confluence, then treated with ligand as indicated for 12 h, followed by the addition of 5 μL (1:100 v/v) of Luc-CEE for an additional 3 h. This was followed by the addition of 150 μL of lysis buffer, chemiluminescence was determined by mixing 50 μL of lysate plus 50 μL of luciferase reporter substrate and measured on a TD-20e luminometer. The luciferase activity was normalized to protein content of the lysate as determined by BCA protein assay (Thermo Fisher Scientific, Waltham, MA).

HepG2 and Hepa1 cells were treated with vehicle or indole (100 μM) followed by nuclear/cytosol protein isolation and protein blot analysis, as described previously. ⁴²

Gel retardation assays were performed usingtranslated human AHR, mouse AHR and ARNT, as described previously. in vitro ⁴³

The homology model of mouse and human AhR-PASB-LBD based on the NMRof the HIF-2α-PASB (PDB 1P97) was prepared and optimized as recently described. Indirubin was docked into the mouse and human optimized model. Then, the complexes were submitted to 1.5 × 10steps MC ligand-protein side chain optimization to reach the most energetically favorable conformations. Molecular docking was run as previously reported. In the ICM-VLS (Molsoft ICM) screening procedure, the ligand scoring is optimized to obtain maximal separation between the binders and non-binders. Each compound is assigned a score according to its fit within the receptor; this ICM score accounts for continuum and discreet electrostatics, hydrophobicity and entropy parameter. Surface energy of binding is based on atomic solvent-accessible surface using special water molecule probe radii designed for calculations of the solvation energy. This term based on “” is a product of atomic accessibilities by the atomic energy density parameters similar to those proposed in literature. The non-hydrogen atomic accessible surfaces are calculated using a faster modification of the Shrake & Rupley-algorithm. The energy value calculated is the difference between the two molecules and the sum of the one molecule-free state (ΔH, kcal/mol). apo atomic solvation ^{44 45 46 47 48 49 50} 4

Data analysis was conducted using Prism 4 software, GraphPad Software Inc. (San Diego, CA). One Way ANOVA analysis was completed using Bonferroni post test. T-test parameters were unpaired and two-tailed analysis. Data represent mean ± S.E.M. and are representative of three independent experiments-value ≤0.05 (*),-value ≤0.01 (**),-value ≤0.001 (***). p p p