Indole-3-Pyruvic Acid, an Aryl Hydrocarbon Receptor Activator, Suppresses Experimental Colitis in Mice

This article is featured in In This Issue, p.3477

Inflammatory bowel disease (IBD) describes a group of intestinal disorders, including Crohn disease and ulcerative colitis. IBD is a complex multifactorial disorder that depends on genetic susceptibility of the host, the intestinal microbiota, other environmental factors, and the host immune system (1). Recent accumulating evidence suggests that IBD results from an exaggerated immune response against components of the intestinal flora (1–3). Most therapeutic strategies for IBD are aimed at suppressing these abnormal immune responses and reestablishing immune tolerance. Treatment mainly consists of immunosuppressive agents, such as 5-aminosalicylates, azathioprine, 6-mercaptopurine, and steroids (4–7). More recently, biologic agents, such as anti–TNF-α Abs, have been developed and provide great clinical benefit in patients with IBD (8), but their immunogenicity may limit their use (9). There is therefore a need for therapies that maintain immune tolerance persistently without toxicity.

Aryl hydrocarbon receptor (AHR) is a ligand-activated transcription factor that was initially identified as an intracellular 2,3,7,8-tetrachlorodibenzo-p-dioxin (TCDD)–binding protein (10). It has been subsequently shown that many other molecules activate AHR, including dietary and microbial aromatic compounds (11). AHR plays an important role in maintaining homeostasis in the intestinal mucosa (12), and several AHR agonists suppressed intestinal inflammation in mouse models of IBD (13–16). Administration of TCDD was reported to decrease inflammation in a mouse model of 2,4,6-trinitrobenzene sulfonic acid–induced colitis, in part by increasing Foxp3⁺ regulatory T cells (Tregs) (13). Administration of 2-(1′H-indole-3′-carbonyl)-thiazole-4-carboxylic acid methyl ester (ITE), a nontoxic AHR agonist, ameliorated colitis in humanized mice by increasing CD39, granzyme B, and IL-10 expression by Tregs (14). Furthermore, the tryptophan-derived AHR agonists indole-3-aldehyde and 6-formylindolo[3,2-b]carbazole (FICZ) provided mucosal protection in mice via induction of IL-22 (15, 16). Thus, different mechanisms for the protective effect of AHR agonists in the intestinal tract have been postulated.

Indole-3-pyruvic acid (IPA) is an aromatic pyruvic acid and a keto analogue of tryptophan. IPA is a precursor of microbiota-derived AHR agonists, such as indole-3-acetaldehyde, indole-3-aldehyde, and indole-3-acetic acid (17), and is also itself a proagonist of AHR that reacts in aqueous solution to form a variety of AHR agonists (18). IPA is synthesized by the action of aromatic amino acid transaminases. In addition, some microorganisms produce IPA by the deamination of tryptophan (19, 20). IPA reduced damage in the striatum after transient forebrain ischemia in the rat (21). IPA also has the potential to improve sleep disorders (22) and to suppress anxiety in humans (23). We previously demonstrated that IPA prevented UVB-induced damage to HaCaT keratinocytes and to the skin of hairless mice (24). Thus, although IPA is known to have various biological functions, the effect of IPA on colon inflammation has not been addressed.

In this study, we examined the ability of IPA to activate AHR in vitro and in vivo and investigated the anti-inflammatory effect of IPA on chronic colitis induced by adoptive transfer of naive CD4⁺ T cells to SCID mice. To compare IPA with other aromatic pyruvic acids, the effects of phenylpyruvic acid (PPA), a phenylalanine keto analogue that is not an AHR proagonist, and 4-hydroxyphenylpyruvic acid (HPPA), a tyrosine keto analogue with AHR proagonist activity, were also investigated. Furthermore, to clarify the effect of IPA on intestinal immune homeostasis, T cell and dendritic cell (DC) responses in the intestinal immune system were analyzed. This study also examined the inhibitory effect of an AHR antagonist on the observed effects of IPA to confirm whether they were mediated by AHR.

IPA and HPPA were purchased from Sigma-Aldrich (St. Louis, MO). PPA was purchased from Wako Pure Chemical (Osaka, Japan). GNF-351 (an AHR antagonist) was purchased from MilliporeSigma (Billerica, MA), dissolved to a concentration of 20 mg/ml in DMSO, and stored at −20°C until use.

Female BALB/c mice were purchased from Charles River Laboratories Japan (Kanagawa, Japan). Female CB17-SCID mice were purchased from CLEA Japan (Tokyo, Japan). All animal care and use was conducted in accordance with institutional guidelines for the care and use of experimental animals (Institute of Livestock and Grassland Science, National Agriculture and Food Research Organization of Japan). The study protocol was approved by the Animal Care Committee of the Institute of Livestock and Grassland Science (permit no.: 14113018-2, 15113063, 1611D65-2, 1711D018, and 1811D039).

HepG2 cells (American Type Culture Collection, Rockville, MD) were seeded in 96-well plates at 1 × 10⁴ cells per well in 100 μl of Eagle’s MEM (EMEM) supplemented with 10% FCS, 100 U/ml of penicillin, 100 μg/ml of streptomycin, nonessential amino acids, and 2 mM l-glutamine (10% FCS EMEM) and incubated for 24 h. The cells were transfected with 90 ng pGL4.43[luc2P/XRE/Hygro] vector (Promega, Madison, WI) and 9 ng pGL4.74[hRluc/TK] vector (Promega) using Lipofectamine 3000, according to the manufacturer’s instructions (Thermo Fisher Scientific, Waltham, MA). After 18 h, the culture medium was removed, and the cells were treated with PPA, HPPA, or IPA in EMEM containing 0.5% FCS and 0.025% DMSO or with stool supernatants diluted 1/20 in EMEM containing 0.5% FCS. After 24 h, firefly and Renilla luciferase activities were assayed with the Dual-Glo Luciferase Assay System (Promega) and measured on a GloMax Discover (Promega), according to the manufacturer’s instructions.

BALB/c mice (6-wk-old) were given normal MF chow (Oriental Yeast, Tokyo, Japan) or MF chow containing 0.1% PPA, HPPA, or IPA. After 5 d, fresh stool samples were collected from the mice, weighed, and suspended in PBS (100 mg/ml). The supernatants were harvested by centrifugation (5000 × g, 15 min, 4°C), filtered through 0.2-μm filters, and subjected to AHR reporter gene assays as described above. The mice were sacrificed by cervical dislocation, and the colons were subjected to quantitative gene expression analysis.

Naive splenic CD4⁺ T cells from BALB/c mice were purified using the CD4⁺ CD62L⁺ T cell Isolation Kit II (Miltenyi Biotec, Bergisch-Gladbach, Germany). The cells were i.p. transferred to 6-wk-old SCID mice (1 × 10⁶ cells per mouse). Control SCID mice received PBS instead of naive CD4⁺ T cells. For the DC transfer, FACS-sorted CD103⁺ CD11b⁻ CD11c^high cells (5 × 10⁴ cells per mouse) were administered to the mice i.p. along with the naive CD4⁺ T cells. During the experiment, the mice were given normal MF chow or MF chow containing 0.1% PPA, HPPA, or IPA. In the AHR antagonist administration experiments, mice were i.p. injected with GNF-351 (10 μg/mouse, twice daily) or vehicle control (0.5% DMSO). Body weight was monitored every week and the severity of diarrhea was scored at 5 wk after cell transfer (0, normal stools; 1, loose stools; 2, diarrhea; 3, watery stools) in an observer-blinded fashion. The mice were sacrificed by cervical dislocation, and the colons were subjected to histological analysis and quantitative gene expression analysis. Intracellular cytokine analysis was also performed on colonic lamina propria lymphocytes (LPL).

Colons were fixed in 10% formalin PBS for at least 24 h. The colons were embedded in paraffin, sectioned, and stained with H&E. Colon sections were imaged on a Zeiss Axioplan2 microscope (Carl Zeiss, Oberkochen, Germany) and scored histologically in an observer-blinded fashion (0, no obvious inflammation; 1, low inflammation; 2, moderate inflammation; 3, severe inflammation accompanied by intestinal wall thickness; and 4, severe inflammation accompanied by intestinal wall thickness, goblet cell loss, and transmucosal infiltration).

Total RNA was extracted from colon samples preserved in RNAlater (Ambion, Austin, TX) using the RNeasy Mini Kit (QIAGEN, Hilden, Germany). Total RNA was reverse-transcribed using the ReverTra Ace qPCR RT Kit (Toyobo, Osaka, Japan). Quantitative real-time PCR was carried out with THUNDERBIRD SYBR qPCR Mix (Toyobo) on a C1000 Thermal Cycler (Bio-Rad Laboratories, Hercules, CA). All primer sequences employed in this study are shown in Table I. Amplification conditions were as follows: initial denaturation at 95°C for 60 s and then 40 cycles of 95°C for 15 s and 60°C for 30 s. All values were normalized to the expression of Gapdh.

The colon was excised and cut open along its longitudinal axis. After washing with PBS, the colon was cut into 3–4-cm segments and placed into a tube filled with 30 ml of Ca²⁺/Mg²⁺-free HBSS supplemented with 5 mM EDTA, 0.04% NaHCO₃, and 5% FCS. The segments were shaken in a water bath at 150 rpm for 20 min. The tube was shaken vigorously, and the segments were filtered through gauze. This process was repeated two additional times to remove the epithelial cells from the colon segments. The segments were finely fragmented into ∼2-mm segments and placed into 10 ml of RPMI 1640 medium supplemented with 10% FCS, 100 μg/ml DNase I (Roche Diagnostics, Mannheim, Germany), and 100 U/ml collagenase type I (Invitrogen, Carlsbad, CA). Colonic fragments were shaken at 37°C for 60 min to separate cells from the tissue. After filtration through gauze, the cells were harvested by centrifugation at 300 × g for 10 min. The cells were suspended in 40% Percoll and layered onto 2 ml of 63% Percoll. After centrifugation at 580 × g for 20 min, colonic LPLs were collected from the interface and subjected to intracellular cytokine analysis.

Naive splenic CD4⁺ T cells were purified from BALB/c mice using the CD4⁺ CD62L⁺ T cell Isolation Kit II (Miltenyi Biotec). The cells (1 × 10⁶ cells) were activated with 5 μg/ml of plate-bound anti-CD3 and 2 μg/ml anti-CD28 with 12.5 ng/ml murine rIL-12 (R&D Systems, Minneapolis, MN) for Th1 differentiation or 50 ng/ml murine rIL-27 (R&D Systems) and 2 ng/ml human rTGF-β (R&D Systems) for type 1 Treg (Tr1) differentiation, in the presence or absence of 50 μM IPA. All cells were plated in 24-well plates in 500 μl of RPMI 1640 supplemented with 10% heat-inactivated FCS, 100 U/ml of penicillin, 100 μg/ml of streptomycin, 10 mM HEPES, and 50 μM 2-ME (complete RPMI 1640). In the AHR antagonist addition experiments, GNF-351 (1 μM final concentration) was added to the culture. After 4 d, the cells were harvested and subjected to intracellular cytokine analysis. IFN-γ and IL-10 concentrations in the culture supernatants were also determined by ELISA (eBioscience/Thermo Fisher Scientific).

To examine the ability of DCs in mesenteric lymph nodes (MLN) to induce T cell differentiation, DCs were purified from MLN using CD11c microbeads (Miltenyi Biotec). Naive CD4⁺ T cells (1 × 10⁵ cells) from BALB/c mice were cocultured with purified MLN CD11c⁺ cells (1 × 10⁵ cells) and were activated with anti-CD3 for 4 d. IL-10 and IFN-γ levels in the culture supernatants were determined by ELISA.

LPLs or CD4⁺ T cells differentiated in vitro were incubated with cell stimulation mixture (eBioscience) in complete RPMI 1640 at 37°C for 4–12 h. The cells were harvested, and dead cells were stained with Fixable Viability Dye (eBioscience) for 15 min on ice. After washing, the cell surface was stained with anti-mouse CD4 allophycocyanin-eFluor 780 (eBioscience) in the presence of anti-CD16/32 (BD Pharmingen) for 20 min at 4°C. The cells were washed, fixed with IC Fixation Buffer (eBioscience), and permeabilized with permeabilization buffer (eBioscience). Intracellular cytokine staining was performed with Alexa Fluor 488–conjugated anti–IFN-γ (eBioscience) and PE-conjugated anti–IL-10 (eBioscience) Abs for 20 min at room temperature. Data were acquired on a Gallios Flow Cytometer (Beckman Coulter, Miami, FL) and analyzed using FlowJo software.

BALB/c mice were fed MF chow or MF chow containing 0.1% IPA. In the AHR antagonist administration experiments, mice were i.p. injected with GNF-351 (10 μg/mouse, twice daily) or vehicle control (0.5% DMSO). After 2 wk, the mice were sacrificed, and MLN were excised. Cells were isolated from MLN and stained with PE-conjugated CD11b, allophycocyanin-conjugated CD103, and PE-Cy7–conjugated CD11c Abs in the presence of anti-CD16/32 for 20 min at 4°C. Data were acquired on a Gallios Flow Cytometer and analyzed using FlowJo software.

To purify CD103⁺ CD11b⁻ DCs, MLN cells from mice administered 0.1% IPA MF for 2 wk were depleted of T cells by negative isolation using CD90.2 microbeads (Miltenyi Biotec). The CD90.2⁻ cells were stained with PE-conjugated CD11b, allophycocyanin-conjugated CD103, and PE-Cy7–conjugated CD11c Abs in the presence of anti-CD16/32 for 20 min at 4°C and subsequently sorted using a MoFlo (Beckman Coulter). The sorted CD103⁺ CD11b⁻ CD11c^high cells were used immediately for adoptive transfer.

The Mann–Whitney U test and the Steel–Dwass test were used to analyze the diarrhea score and histological score, respectively. Student t tests and Tukey test were used to analyze body weight gain and to compare data from the AHR reporter gene assays, gene expression analyses, and flow cytometric analyses. The p values < 0.05 were considered significant.

First, the ability of aromatic pyruvic acids (PPA, HPPA, and IPA) to activate AHR was examined using an AHR reporter gene system. Consistent with previous reports (18), IPA (both 50 and 250 μM) and HPPA (250 μM) activated AHR, and IPA had a stronger effect than HPPA (Fig. 1A). Next, we investigated whether oral administration of aromatic pyruvic acid can activate AHR in the colon. BALB/c mice were fed with aromatic pyruvic acids in MF chow (0.1%) for 5 d, and the ability of their stool samples to activate AHR was investigated. Stool supernatants from the mice administered IPA significantly activated AHR, but samples from HPPA- or PPA-treated mice did not (Fig. 1B). In agreement with this result, only administration of IPA significantly upregulated the expression of Cyp1a1, a biomarker for AHR activation, in the colon (Fig. 1C). These results indicate that orally administered IPA acts as an AHR activator in the intestinal tract.

We next examined the effects of aromatic pyruvic acids on chronic inflammation caused by transfer of naive CD4⁺ T cells into SCID mice. SCID mice were peritoneally injected with naive CD4⁺ T cells to induce colon inflammation, and then aromatic pyruvic acids were orally administered to the mice. Adoptive transfer of naive CD4⁺ T cells into SCID mice suppressed body weight gain (Fig. 2A) and induced diarrhea (Fig. 2B) and severe colon inflammation accompanied by intestinal wall thickness, goblet cell loss, and transmucosal infiltration (Fig. 2C, 2D) at 5 wk after cell transfer. Although administration of IPA did not significantly improve body weight gain (Fig. 2A), it suppressed diarrhea (Fig. 2B) and improved colon inflammation (Fig. 2C, 2D), similar to other AHR agonists that suppress intestinal inflammation in IBD mouse models (13–16). Administration of PPA or HPPA did not show these effects.

Th1 and proinflammatory cytokines play a crucial role in chronic inflammation caused by the transfer of naive CD4⁺ T cells into SCID mice (25–28). Therefore, we evaluated whether IPA administration downregulates the expression of Th1 cytokines (Il-12a, Il-12b, Ifng, Tnfa) and proinflammatory cytokines (Il-1b, Il-6) in the T cell–mediated colitis model. The expression of Il-12b, Ifng, Tnfa, and Il-1b was significantly decreased in the colons of SCID mice fed IPA (Fig. 3). To investigate the effect of IPA on anti-inflammatory factors, the expression of Il-22, Foxp3, and Il-10 was analyzed. Administration of IPA did not upregulate the expression of Il-22 or Foxp3 but did significantly upregulate the expression of Il-10 (Fig. 3). These results indicate that IPA administration decreases Th1 cytokines and increases IL-10 in the colons of SCID mice following adoptive transfer of naive CD4⁺ T cells. To corroborate these findings, intracellular cytokine analysis was performed on colonic LPL from SCID mice that received naive CD4⁺ T cells. Five weeks after naive T cell transfer, the colons were removed, and colonic LPL were prepared. Administration of IPA reduced the total numbers of colonic LPL (Fig. 4A). In agreement with the results of the gene expression analysis (Fig. 3), administration of IPA decreased the frequency of IFN-γ⁺ IL-10⁻ CD4⁺ T cells and increased the frequency of IFN-γ⁻ IL-10⁺ CD4⁺ T cells (Fig. 4B, 4C). In contrast, administration of IPA did not affect the frequency of IFN-γ⁺ IL-10⁺ CD4⁺ T cells (Fig. 4B, 4C). These results indicate that oral administration of IPA decreases Th1 cells and increases IFN-γ–nonproducing Tr1 in the colon. We also investigated the effect of IPA on IL-22 and Foxp3 expression in CD4⁺ T cells isolated from LPLs. Consistent with the results of gene expression analysis, administration of IPA did not affect the frequency of IL-22⁺ CD4⁺ T cells or Foxp3⁺ CD4⁺ T cells (Supplemental Fig. 1A–D). In addition, administration of IPA did not affect the frequency of IL-17⁺ CD4⁺ T cells (Supplemental Fig. 1E, 1F), which are termed Th17 cells and were reported to play a key role in the intestinal immune system (29).

We next investigated whether IPA can directly affect the differentiation of Th1 and Tr1. Naive CD4⁺ T cells from BALB/c mice were activated with plate-bound anti-CD3 and anti-CD28 in media containing murine rIL-12 (Th1 differentiation conditions) or murine rIL27 and human rTGF-β (Tr1 differentiation conditions) in the presence or absence of IPA. Under Th1 differentiation conditions, IPA treatment slightly increased the frequency of IFN-γ⁺ cells (Fig. 5A, 5B) but did not affect the number of viable cells (Fig. 5C) or the amount of IFN-γ in the culture supernatant (Fig. 5D). In contrast, under Trl differentiation conditions, IPA treatment increased the frequency of IL-10⁺ cells (Fig. 5E, 5F), the number of viable cells (Fig. 5G), and the amount of IL-10 in the culture supernatant (Fig. 5H). IPA treatment also enhanced the suppressive activity of Tr1 (Supplemental Fig. 2). To determine whether IPA treatment induces IL-10 production under Th0 conditions, the effect of IPA in the absence of TGF-β and IL-27 was evaluated. However, IL-10 production was not enhanced by IPA treatment under these conditions (data not shown). These results demonstrate that IPA does not inhibit Th1 cell differentiation but promotes Tr1 differentiation directly.

Intestinal DCs continuously migrate from the intestine to the MLN (30–32) and play a central role in inducing T cell differentiation and maintaining immune homeostasis (33–35). Therefore, we investigated the effect of IPA administration on the ability of MLN DCs to promote T cell differentiation. MLN CD11c⁺ cells were isolated from BALB/c mice fed MF chow containing 0.1% IPA for 2 wk and then cocultured with naive splenic CD4⁺ T cells in the presence of an anti-CD3 Ab. CD4⁺ T cells cocultured with MLN DCs from IPA-fed mice produced less IFN-γ (Fig. 6A) but similar amounts of IL-10 (Fig. 6B) as DCs from control mice, suggesting that IPA attenuated the ability of MLN DCs to induce Th1 cell differentiation. Consistent with this finding, the frequency of Th1 cells in the colon LPL (Fig. 6C, 6D) and in the MLN (Fig. 6E, 6F) were decreased in IPA-fed BALB/c mice.

MLN DCs can be divided into three subsets based on their expression of CD103 and CD11b: CD103⁺ CD11b⁻ DCs, CD103⁺ CD11b⁺ DCs, and CD103⁻ CD11b⁺ DCs. To investigate the effect of IPA administration on DC subsets in the MLN, MLN cells prepared from BALB/c mice fed IPA were analyzed for the expression of CD11c, CD103, and CD11b by flow cytometry. IPA administration did not affect the total number of DCs in the MLN (Fig. 7A, 7B) but significantly increased the frequency of CD103⁺ CD11b⁻ DCs and decreased the frequency of CD103⁻ CD11b⁺ DCs (Fig. 7A, 7C). To determine whether this alteration of the composition of DC subsets in the MLN was involved in the decline in the Th1-inducing capacity of MLN DCs, we assessed the ability of these DC subsets to induce Th1 cell differentiation. CD103⁺ CD11b⁻ CD11c^high cells, CD103⁺ CD11b⁺ CD11c^high cells, and CD103⁻ CD11b⁺ CD11c^high cells were FACS-sorted from BALB/c mice fed IPA and then separately cocultured with naive splenic CD4⁺ T cells in the presence of an anti-CD3 Ab. Production of IFN-γ was highest by CD4⁺ T cells cocultured with CD103⁻ CD11b⁺ CD11c^high cells (Supplemental Fig. 3). In the absence of overt stimulation such as TLR stimulation, only CD103⁻ CD11b⁺ DCs can induce IFN-γ production (36). These findings suggest that the decrease in CD103⁻ CD11b⁺ DCs is one mechanism underlying the decline in the Th1-inducing capacity of MLN DCs.

We next investigated whether the CD103⁺ CD11b⁻ DCs induced by the administration of IPA could suppress colonic inflammation. CD103⁺ CD11b⁻ DCs were isolated from the MLN of BALB/c mice fed MF chow containing 0.1% IPA, and SCID mice were injected i.p. with naive CD4⁺ T cells alone or in combination with these sorted CD103⁺ CD11b⁻ DCs. Although transfer of CD103⁺ CD11b⁻ DCs did not significantly improve body weight gain (Fig. 7D), it significantly reduced the severity of colon inflammation (Fig. 7E–G). These results suggest that the MLN CD103⁺ CD11b⁻ DCs induced by IPA administration are regulatory cells that maintain intestinal immune homeostasis.

It was unknown whether the observed effects of IPA were mediated by AHR. Therefore, we first investigated whether an AHR antagonist inhibited the ability of IPA to induce Tr1 differentiation. We used GNF-351, which is an AHR antagonist that inhibits both dioxin response element-dependent and -independent activity and lacks partial agonist potential (37). Addition of GNF-351 inhibited the IPA-induced increases in the frequency of IL-10⁺ cells (Fig. 8A, 8B), the number of viable cells (Fig. 8C), and the concentration of IL-10 in the culture supernatant (Fig. 8D). Addition of GNF-351 also inhibited the IPA-induced enhancement of the suppressive activity of Tr1 (Supplemental Fig. 2).

We next investigated whether GNF-351 inhibited the ability of IPA to alter the composition of MLN DC subsets. GNF-351 acts as an AHR antagonist for up to 12 h (37); therefore, BALB/c mice were i.p. injected with GNF-351 twice daily and orally administered IPA for 2 wk. Thereafter, expression of CD11c, CD103, and CD11b in MLN cells was analyzed by flow cytometry. GNF-351 treatment significantly inhibited the IPA-induced increase in the frequency of CD103⁺ CD11b⁻ DCs (Fig. 9A, 9B). In contrast, GNF-351 treatment did not significantly inhibit the IPA-induced decrease in the frequency of CD103⁻ CD11b⁺ DCs (Fig. 9A, 9B).

Finally, we investigated whether GNF-351 inhibited the anti-inflammatory activity of IPA. SCID mice were peritoneally injected with naive CD4⁺ T cells to induce colon inflammation. Thereafter, the mice were orally administered IPA and i.p. injected with GNF-351 twice daily for 5 wk. Although GNF-351 treatment did not significantly affect body weight gain (Fig. 9C), it significantly attenuated the anti-inflammatory effect of IPA (Fig. 9D–F).

Altogether, these results indicate that AHR underlies the abilities of IPA to induce Tr1 differentiation, to alter the composition of MLN DC subsets, and to suppress colon inflammation.

In the current study, we show that among aromatic pyruvic acids, only IPA has a significant anti-inflammatory effect on the colon in a mouse model of IBD (Fig. 2). HPPA, which is also an AHR proagonist, had no effect on colon inflammation (Fig. 2). This may be due to loss of the ability of HPPA to activate AHR in the colon. In fact, administration of HPPA did not upregulate Cyp1a1 expression in the colon (Fig. 1C), and the ability of stool supernatant to activate AHR was elevated by administration of IPA but not by administration of HPPA (Fig. 1B). IPA activated AHR more strongly than HPPA (Fig. 1A), and, unlike HPPA, IPA may be converted to AHR agonists by microbiota in the intestinal tract. Therefore, IPA is a promising agent to effectively deliver AHR agonists to the intestinal tract. It was reported that fecal samples from patients with IBD induced less activation of AHR than those from healthy subjects, which was attributed to decreased levels of indole-3-acetic acid, a microbial AHR agonist, in fecal samples from IBD patients (17). Furthermore, analysis of mice lacking Card9, a susceptibility gene for IBD, revealed that altered microbiota that lost the ability to produce AHR agonists were associated with increased susceptibility to colitis (17). These data indicate that delivering appropriate AHR agonists to the intestinal tract is an attractive strategy for the treatment of IBD. Thus, IPA, an effective AHR activator, may be a potent immunomodulating therapy for IBD patients.

Some AHR agonists enhance mucosal barrier function through the induction of IL-22 (15, 16). However, in our study, oral administration of IPA did not upregulate the gene expression of IL-22 (Fig. 3) and did not increase IL-22⁺ CD4⁺ T cells (Supplemental Fig. 1A, 1B) in the colons of SCID mice receiving naive CD4⁺ T cells, suggesting that IPA abrogates colon inflammation through IL-22–independent mechanisms. In this study, administration of IPA altered the composition of T cell subsets in the colon LPL and the composition of DC subsets in the MLN.

Recent studies show that AHR agonists control the differentiation and activation of specific T cell subsets (38–41). TCDD induced Foxp3⁺ Treg differentiation, whereas FICZ induced Th17 cell differentiation (38). Furthermore, AHR acted in synergy with c-Maf to promote the generation of Tr1 (41). Thus, AHR affects T cell responses differently depending on the experimental conditions and the type of agonist used. In the current study, IPA increased IL-10–producing T cells in the colon LPL in a T cell–mediated colitis model (Fig. 4B, 4C). Unlike ITE, which increases not only IL-10 but also Foxp3 expression in LPL T cells (14), IPA did not upregulate Foxp3 gene expression (Fig. 3) and did not increase Foxp3⁺ CD4⁺ T cells in the colon (Supplemental Fig. 1C, 1D). In addition, IPA promoted Tr1 differentiation in vitro (Fig. 5E–H, Supplemental Fig. 2). These findings suggest that IPA promotes Tr1 differentiation but not Foxp3⁺ Treg differentiation in the colon LPL in a T cell–mediated colitis model. Tr1 are characterized by high levels of IL-10 production, which plays a critical role in maintaining intestinal immune homeostasis (42). Transfer of Tr1 suppressed T cell–mediated colitis (43), and Bifidobacterium breve prevented intestinal inflammation by inducing Tr1 (44). In addition, a phase 1/2a clinical study showed that administration of Ag-specific Tr1 to patients with refractory Crohn disease was well tolerated (45). Therefore, increasing Tr1 might be one of the mechanisms through which IPA suppresses colon inflammation.

IPA decreased the frequency of Th1 cells in the colon LPL in mice with T cell–mediated colitis (Fig. 4B). Because Th1 cells are involved in the pathogenesis of colitis in this model (26), inhibition of Th1 cell development appears to be an important mechanism for suppression of colon inflammation by IPA. Th1 cells in the LPL and MLN were also decreased in normal BALB/c mice administered IPA (Fig. 6C–F). IPA did not inhibit the differentiation of Th1 cells directly (Fig. 5A–D) but attenuated the ability of MLN DC to induce Th1 cell differentiation (Fig. 6A). Therefore, the decrease in Th1 cells in the colon LPL may be due in part to a decline in the Th1-inducing ability of MLN DCs.

The present study showed that IPA administration decreased the frequency of CD103⁻ CD11b⁺ DCs and increased the frequency of CD103⁺ CD11b⁻ DCs in the MLN (Fig. 7A, 7C). Additionally, the current study suggested that this alteration in the composition of MLN DC subsets is one mechanism underlying the decrease in the Th1-inducing ability of MLN DCs. Different subsets of DCs induce different types of adaptive immune responses. CD103⁺ CD11b⁺ DCs are involved in the generation of Th17 cells and Foxp3⁺ Tregs (46–49). CD103⁺ CD11b⁻ DCs are unique cells that can cross-present Ag to CD8⁺ T cells (50). CD103⁻ CD11b⁺ DCs induce the differentiation of both IFN-γ and IL-17–producing effector T cells (36, 51, 52). CD103⁻ DCs, mainly composed of CD103⁻ CD11b⁺ DCs, are highly inflammatory during chronic colitis (53, 54). In addition, Kourepini et al. (55) revealed that acute colitis caused an increase in MLN CD103⁻ DCs and a decrease in MLN CD103⁺ DCs. They further revealed that adoptive transfer of MLN CD103⁻ DCs exacerbated colitis, whereas adoptive transfer of MLN CD103⁺ DCs did not (55). These findings suggest that the composition of the MLN DC subset affects the progression of intestinal inflammation.

We showed that transfer of CD103⁺ CD11b⁻ DCs, the frequency of which was increased in the MLN by IPA administration, significantly reduced the severity of colon inflammation (Fig. 7D–G). This result suggests that the increase in CD103⁺ CD11b⁻ DCs in the MLN is related to the ability of IPA to suppress colonic inflammation. Although the mechanism by which IPA alters the composition of DC subsets in the MLN remains unclear, this is, to our knowledge, the first report demonstrating that an AHR activator can increase regulatory CD103⁺ CD11b⁻ DCs in the MLN. Several previous reports suggest a relationship between CD103 and AHR. Chng et al. (56) showed that CD103⁺ DCs were decreased in the MLN in CD11c⁺ cell-specific AHR-deficient mice. Interestingly, these mice were highly sensitive to dextran sulfate sodium (DSS)–induced colitis, suggesting that AHR expression in DCs plays an important role in homeostasis of mucosal immunity (56). In addition, administration of TCDD was reported to increase the frequency of CD103⁺ DCs in the MLN and colons of mice treated with 2,4,6-trinitrobenzenesulfonic acid (13). Elucidating how IPA regulates the number of CD103⁺ CD11b⁻ DCs in the MLN is an interesting topic for future studies.

Muzaki et al. (57) revealed that CD103⁺ CD11b⁻ DCs restrain colitis via intestinal T cell–derived IFN-γ, which triggers a reversible early anti-inflammatory response in intestinal epithelial cells. This seems to disagree with our finding that CD103⁺ CD11b⁻ DCs had a lower Th1 cell–inducing ability (Supplemental Fig. 3). However, this discrepancy may be explained by the difference in the function of CD103⁺ CD11b⁻ DCs between the steady and inflammatory states. The intestinal T cell–mediated IFN-γ response is induced upon DSS treatment but not at steady-state (57). Whereas a robust Th1 cell population is induced by CD103⁺ CD11b⁻ DCs from DC-specific conditional A20-knockout mice, which develop intestinal inflammation, CD103⁺ CD11b⁻ DCs from wild-type mice have a much lower Th1-inducing ability (52). Muzaki et al. (57) mentioned that CD103⁺ CD11b⁻ DCs contribute to maintenance of homeostasis via the balance of intestinal T cells at steady-state. Indeed, mice lacking DCs that are dependent on the transcription factor IFN regulatory factor 8 (mainly CD103⁺ CD11b⁻ DCs) exhibit a decrease in Th1 cells and a slight decrease in Foxp3⁺ Tregs in the large intestine (58). In addition, CD103⁺ CD11b⁻ DCs highly expressing programmed death-ligand 1 have a high Foxp3⁺ Treg-inducing ability (59). Thus, CD103⁺ CD11b⁻ DCs probably maintain intestinal homeostasis via the balance between effecter Th1 cells and Foxp3⁺ Tregs at steady-state. In contrast with the DSS model, the T cell–mediated colitis model gradually changes from steady-state to the inflammatory state with an increase in pathogenic T cells. In our model, CD103⁺ CD11b⁻ DCs may suppress intestinal inflammation via a homeostasis mechanism that operates at steady-state rather than via an acute IFN-γ response.

It is also possible that IPA directly affected the function of DCs. Many studies have revealed that AHR agonists alter the innate and regulatory immune function of DCs (60–63). Endogenous AHR agonists, such as FICZ and ITE, inhibit the ability of human monocyte-derived DCs to induce Th1 and Th17 cell differentiation in vitro (62). AHR agonists, such as TCDD, indole-3-carbinol, and indirubin-3′-oxime, enhance the ability of bone marrow–derived DCs to induce differentiation of Foxp3⁺ Tregs in vitro (61, 63), whereas TCDD does not affect the T cell–activating capacity of bone marrow–derived DCs in vivo (60, 61). IPA has been reported to produce FICZ (64); therefore, it is possible that IPA directly inhibits the Th1-inducing ability of DCs via FICZ. It must be clarified whether oral administration of IPA affects the function of DCs in the future.

In summary, IPA suppressed colon inflammation by increasing IL-10–producing T cells and decreasing Th1 cells in the colon lamina propria. IPA administration altered composition of MLN DC subsets, which may explain the suppressive effects of IPA. IPA may therefore be an effective therapeutic agent for IBD.

We thank Dr. Tadatoshi Furusawa (Institute of Agrobiological Sciences, National Agriculture and Food Research Organization) for expertise and help with the FACS analysis and Masaru Kobayashi (National Institute of Animal Health, National Agriculture and Food Research Organization) for expertise in the preparation of histological sections. We also thank Yuko Kurosawa and Chie Koitabashi for outstanding technical assistance.

The authors have no financial conflicts of interest.