Laquinimod Arrests Experimental Autoimmune Encephalomyelitis By

Laquinimod arrests experimental autoimmune PNAS PLUS encephalomyelitis by activating the aryl hydrocarbon receptor

Joel Kayea,1, Victor Piryatinskya, Tal Birnbergb, Tal Hingalya, Emanuel Raymonda, Rina Kashia, Einat Amit-Romacha, Ignacio S. Caballeroc, Fadi Towficc, Mark A. Atora, Efrat Rubinsteina, Daphna Laifenfeldb, Aric Orbacha, Doron Shinara, Yael Marantza, Iris Grossmanb, Volker Knappertzd, Michael R. Haydene, and Ralph Laufera

aDiscovery & Product Development, Global Research & Development, Teva Pharmaceutical Industries Ltd., Netanya 42504, Israel; bPersonalized & Predictive Medicine, Analytics and Big Data, Global Research & Development, Teva Pharmaceutical Industries Ltd., Netanya 42504, Israel; cImmuneering Corporation, Cambridge, MA 02412; dGlobal Clinical Development, Global Research & Development, Teva Pharmaceutical Industries Ltd., Netanya 42504, Israel; and eGlobal Research & Development, Teva Pharmaceutical Industries Ltd., Netanya 42504, Israel

Edited by Lawrence Steinman, Stanford University School of Medicine, Stanford, CA, and approved August 5, 2016 (received for review May 16, 2016) Laquinimod is an oral drug currently being evaluated for the treat- macrophages, and dendritic cells (DCs). It was reported that treatment of relapsing, remitting, and primary progressive multiple ment of mice with laquinimod is associated with alterations in the sclerosis and Huntington’s disease. Laquinimod exerts beneficial ac- frequency of myeloid subpopulations that included a reduction in + tivities on both the peripheral immune system and the CNS with CD4 DCs. Laquinimod treatment also promoted the development of distinctive changes in CNS resident cell populations, especially as- anti-inflammatory type II monocytes and DCs (6, 7, 9), which are trocytes and microglia. Analysis of genome-wide expression data likely associated with its immunomodulatory activities. These activities revealed activation of the aryl hydrocarbon receptor (AhR) pathway include reduced production of proinflammatory cytokines such as in laquinimod-treated mice. The AhR pathway modulates the differ- IL-17, reduced migration of lymphocytes (4, 7), augmentation of entiation and function of several cell populations, many of which regulatory T-cell numbers (5, 7), and production of brain-derived play an important role in neuroinflammation. We therefore tested neurotrophic factor (5, 8). Although no molecular target has been the consequences of AhR activation in myelin oligodendrocyte glyco- INFLAMMATION

identified for laquinimod, it has been shown to modulate the T-cell IMMUNOLOGY AND protein (MOG)-induced experimental autoimmune encephalomyelitis response probably as a result of its effects on STAT1, MAPK, and (EAE) using AhR knockout mice. We demonstrate that the pronounced NF-κB signaling in APCs (reviewed in ref. 17). To further elucidate effect of laquinimod on clinical score, CNS inflammation, and demye- laquinimod’s immunomodulatory mechanisms of action, in this paper lination in EAE was abolished in AhR−/− mice. Furthermore, using bone we analyzed gene expression levels modulated by laquinimod versus marrow chimeras we show that deletion of AhR in the immune system vehicle-treated mice. We show that laquinimod induces genes known fully abrogates, whereas deletion within the CNS partially abrogates to be associated with the aryl hydrocarbon receptor (AhR). In the the effect of laquinimod in EAE. These data strongly support the idea present study, we investigate whether laquinimod suppresses EAE via that AhR is necessary for the efficacy of laquinimod in EAE and that the AhR pathway by testing its efficacy in myelin oligodendrocyte laquinimod may represent a first-in-class drug targeting AhR for the glycoprotein (MOG)-induced EAE using AhR knockout mice. treatment of multiple sclerosis and other neurodegenerative diseases. Results aryl hydrocarbon receptor | EAE | laquinimod Transcriptome Analysis Reveals That Laquinimod Treatment Induces Genes Associated with Activation of the AhR. To gain insight into ’ aquinimod is an oral drug that is currently in late-stage clinical laquinimod s immunomodulatory mechanisms of action, spleen Ldevelopment for the treatment of relapsing remitting multiple sclerosis (RRMS), primary progressive MS, and Huntington’sdis- Significance ease. Current knowledge indicates that laquinimod exerts activities both on the peripheral immune system and within the CNS. Laquinimod is an oral drug currently being evaluated for the Laquinimod, at the 0.6-mg/d dose, has demonstrated efficacy in treatment of relapsing, remitting, and primary progressive mul- phase II and III MS clinical trials, in which it reduced relapse rate, tiple sclerosis as well as Huntington’s disease. It is thought that disability progression, development of new and active MRI lesions, laquinimod has a primary effect on the peripheral innate immune and brain atrophy (1–3). The clinical efficacy profile of laquinimod is system and also acts directly on resident cells within the CNS. characterized by a dissociation of the moderate magnitude of the However, the exact mechanism of action of laquinimod has not effect on relapse reduction and its associated inflammatory MRI been fully elucidated. We investigated gene expression in laqui- findings and the disproportionally large effect on disability progres- nimod-treated mice and show induction of genes downstream to sion. Such an efficacy profile in patients with RRMS may relate to a activation of the aryl hydrocarbon receptor (AhR). In this paper, distinctive intracerebral activity potentially mediated via changes in we examine the role of the AhR in laquinimod treatment of ex- CNS resident cell populations, potentially astrocytes and microglia. perimental autoimmune encephalomyelitis and demonstrate that The influence of laquinimod on the immune system was studied in AhR is the molecular target of laquinimod in this model. experimental autoimmune encephalomyelitis (EAE) (4–12), an autoimmune disease mediated by proinflammatory myelin-reactive lym- Author contributions: J.K., V.P., T.B., E.A.-R., M.A.A., D.L., A.O., D.S., Y.M., I.G., V.K., and phocytes that cause CNS inflammation leading to demyelination and R.L. designed research; T.H., E. Raymond, and R.K. performed research; I.S.C. and F.T. contributed new reagents/analytic tools; J.K., V.P., T.B., T.H., E.A.-R., I.S.C., F.T., M.A.A., E. Rubinstein, axonal loss. Laquinimod has also been effective in the treatment of D.L., A.O., D.S., M.R.H., and R.L. analyzed data; and J.K., V.K., and R.L. wrote the paper. other models of autoimmune diseases, specifically experimental au- Conflict of interest statement: The authors are employees of Teva Pharmaceutical Indus- toimmune neuritis (13, 14), lupus nephritis (15), and colitis (16). A tries Ltd. or Immuneering Corporation. common characteristic of autoimmune diseases is that autoantigen- This article is a PNAS Direct Submission. reactive T cells must undergo several discrete steps to cause disease. Data deposition: The raw data are available at the National Center for Biotechnology Initial signals that direct T-cell activation and differentiation are Information (accession no. PRJNA319255). provided by antigen-presenting cells (APC), including monocytes, 1To whom correspondence should be addressed. Email: [email protected].

www.pnas.org/cgi/doi/10.1073/pnas.1607843113 PNAS Early Edition | 1of8 Downloaded by guest on October 1, 2021 gene expression profiles were compared between laquinimod- AB and vehicle-treated EAE mice, 6 d postdisease induction. Our EAE Veh EAE Laq Naive Veh Naive Laq analysis showed that 610 genes were differentially modulated Wnt4 2 Cyp1a1 2 Cyp1a1 Ahrr Ahrr Gprc5d between laquinimod- and vehicle-treated EAE mice. Of the 610 Amer3 Adamts15 Gpnmb 1 Ak4 1 genes, 227 genes were down-regulated by laquinimod treatment, Dnah2 Ido2 Scin Cx3cl1 including key Th17-related cytokine genes, specifically IL-17a, IL- 1810041L15Rik Hic1 Adamts15 Tph1 Ripk4 0 Mpzl2 Hic1 0 17re, and IL-22Ra1 (Fig. 1), which were significantly reduced in Nfasc Nfasc Mmp27 laquinimod-treated mice [log fold-changes of −2.91, −1.64, and Mcpt1 2 Spint1 Noxa1 Apol10b − Eva1a −1 −1 1.81, respectively; false-discovery rate (FDR) values of 4.4e-4, Ntf3 Pdzph1 Col13a1 Ido1 7e-3, and 2.7e-3]. These findings are in line with published data Mpzl2 Dnah2 Adamts9 Spint1 Plet1 −2 Apol7c −2 showing the beneficial effect of laquinimod on Th17 (4, 7). A total Ppp1r14d RP23−329A21.3 Mmp13 Scin of 383 genes were up-regulated by laquinimod in this analysis, and Olfml2a Gpnmb Palmd Mmp13 cytochrome P450 family member A1 (Cyp1a1) and Ahrr, proto- Pdzph1 Tiparp Pgf Cables1 Ppp1r3c Il1r2 typical genes associated with the AhR pathway (18, 19), were Cyp1b1 Fndc5 1700057G04Rik Dusp14 among the highest fold-change genes induced by laquinimod Ido2 Dpep2 Dpep3 Gpr15 A Muc6 F10 treatment in EAE mice (Fig. 2 ). Analysis of laquinimod-induced Serpinb10 Mmp8 Ifitm7 Ston1 genes in naive mice revealed a similar pattern (Fig. 2B), indicating Olfr334−ps1 Ctla2b Gm8221 Ttc9 Slfn4 that activation of AhR was inherent to drug and independent of Phactr1 Hdc Gzmc Ccl22 disease state. However, many other AhR-associated genes, in- Wnt2 Fkbp5 Gm15759 Ctla2b cluding Cyp1b1 (20), Tiparp (19), Ido1 and Ido2 (21, 22), Spint1, Mb21d2 Msr1 Rims4 Slc7a11 Gm26578 Lad1 and Serpins (23) were induced by laquinimod in both naive and Ppp1r3g Pik3r6 Cnga3 Smim3 EAE mice. Laquinimod-induced activation of the AhR pathway Slc9a3 Ccr1 RP23−326P16.3 Serpinb2 was confirmed by assessing gene expression levels of the AhR bio- Lpar1 C Thrsp marker, Cyp1a1 in mouse liver. As shown in Fig. 2 , the endoge- Inhbb C Fgf3 nous AhR ligand 2-(1H-Indol-3-ylcarbonyl)-4-thiazolecarboxylic Crhr1 Aox3 Kcnj16 acid methyl ester (ITE) (24) caused increases of up to 74- and 4.2- Syt15 Ido1 fold in hepatic Cyp1a1 and Cyp1a2 mRNA levels, respectively. Gm28548 Cxcl3 RP23−329A21.3 Treatment of mice with 25 mg/kg laquinimod caused 539- and Tiparp Gsg1 21-fold increases in hepatic Cyp1a1 and Cyp1a2 mRNA expres- Acpp Fndc5 sion levels, respectively, compared with vehicle-treated mice. Ltc4s Pmp22 Slc27a2 These data verified the initial genomic findings and demon- Ms4a7 Mmp27 strated that laquinimod consistently induces the expression of Trnp1 Mgarp Msr1 genes downstream to the activation of AhR. Dab2 Gpr82 Additional experiments were performed to determine the Gpt2 Dpep2 concentration dependence of the laquinimod effect by quanti- Ugt1a6a Cd93 Rusc2 fying the level of mRNA changes in primary hepatocytes, the Tsku Adamdec1 most AhR-responsive tissue. Treatment of human hepatocytes Serpinb8 Amz1 ∼ ∼ 2210408F21Rik with laquinimod for 24 h resulted in 180- and 80-fold in- Stbd1 ± Cdh1 duction of CYP1A1 and CYP1A2 mRNA, with an EC50 of 0.2 Fbxo27 0.04 μM and 0.3 ± 0.03 μM, respectively (Fig. 3A). Treatment with 100 μM omeprazole for 24 h was used as a positive control Fig. 2. Transcriptome analysis reveals that laquinimod induces genes in the ± ± AhR pathway. (A) This heatmap was generated using the patterns of ex- and resulted in 1,487 34 and 188 11 fold-induction of pression of splenocyte samples from EAE mice at day 6 postinduction. It CYP1A1 and CYP1A2 mRNA. Treatment of primary mouse shows the patterns of expression of 88 genes with higher levels of expression hepatocytes with laquinimod resulted in induction of Cyp1a1 in EAE mice treated with laquinimod (green, n = 6) than in EAE mice treated = mRNA with an EC50 of 0.06 ± 0.02 μM and 5.70 ± 0.20 μMat4 with vehicle (orange, n 6). (B) This heatmap was generated using the and 24 h, respectively (Fig. 3B). Both the potency and magnitude patterns of expression of splenocyte samples from naive mice at 6 d post- of induction by laquinimod in mouse hepatocytes were larger at induction. It shows the patterns of expression of 45 genes with higher levels of expression in naive mice treated with laquinimod (red, n = 6) than in naive mice treated with vehicle (blue, n = 5). For both A and B, the filters used to define statistical significance were a fold-change > 2 and an FDR < 0.01. Rows

Il17a Il17re Il22ra1 were ordered from highest to lowest average fold-change. (C)Invivoin- 2 * FDR=4.4e-4 2 * FDR=7e-3 2 * FDR=2.7e-3 duction of hepatic Cyp1a mRNA following 5-d treatment with 25 mg/kg laquinimod in C57BL/6 mice. This graph is taken from a representative ex- 0 0 0 periment, which has been repeated at least four times.

−2 −2 −2

−4 −4 −4 4 h than at 24 h, presumably due to metabolism following longer −6 −6 −6 incubation. Omeprazole does not activate mouse AhR, so EAE EAE Naive EAE EAE Naive EAE EAE Naive

Normalized expression (log2 CPM) Normalized expression Vehicle Laquinimod Vehicle Vehicle Laquinimod Vehicle Vehicle Laquinimod Vehicle 3-methylcholanthrene (3-MC) treatment (2 μM) for 24 h was * Significant (FDR < 0.05) used as a positive control and resulted in a 116- ± 1.5-fold in- Fig. 1. Laquinimod treatment down-regulates genes associated with the duction of Cyp1a1 mRNA. In mouse hepatocytes, basal levels of Th17 pathway. These boxplots show the patterns of the expression of indi- Cyp1a2 mRNA were high and were not significantly augmented vidual genes from splenocyte samples taken from naive and EAE mice at day by treatment with either laquinimod or 3-MC. 6 postinduction (n = 6 per treatment). It shows the patterns of expression across three conditions: EAE mice treated with vehicle, EAE mice treated The Efficacy of Laquinimod in EAE Is Dependent on the AhR. Given with laquinimod, and naive mice treated with vehicle (n = 5). Laquinimod that laquinimod activates AhR, we wanted to determine the role treatment significantly reduced the expression of IL-17a [log2 fold change (FC): −2.91; FDR: 4.4e-4], Il17re (log2 FC: −1.64; FDR 7e-3), and Il22ra1 (log2 of AhR in the efficacy of laquinimod in MOG-induced EAE. We −/− FC: −1.81; FDR: 2.7e-3). found that AhR mice were susceptible to EAE, albeit with

2of8 | www.pnas.org/cgi/doi/10.1073/pnas.1607843113 Kaye et al. Downloaded by guest on October 1, 2021 PNAS PLUS

Fig. 3. In vitro verification of induction of CYP1A mRNA by laquinimod. (A) CYP1A1 (Left) and CYP1A2 (Right) fold-induction in human hepatocytes treated with laquinimod for 24 h compared with vehicle controls. Experiments were repeated at least four times with cells from different donors. (B) Cyp1a1 fold- induction in mouse hepatocytes treated with laquinimod for 4 (Left)or24h(Right). Experiments were repeated at least four times with different donors. INFLAMMATION IMMUNOLOGY AND slightly less severe disease. Whereas WT mice developed EAE with 0.6 (93% inhibition, P < 0.0001). In contrast, laquinimod com- − − − − a mean day of onset of 12.8 ± 1.3, AhR / mice developed EAE pletely lost its efficacy in AhR / mice (Fig. 4) that had comparable with delayed onset (15.7 ± 4.5 d) and less severe disease (mean disease severity (3.3 ± 1.7) and a profile undistinguishable from − − − − maximal score of 3.5 ± 1.1, compared with 4.2 ± 0.7 in AhR / and vehicle-treated AhR / mice (3.5 ± 1.1). WT, respectively). As previously reported, prophylactic daily oral As previously shown (4), prophylactic treatment with laquini- treatment with 25 mg/kg laquinimod inhibited MOG-induced EAE mod abrogated the extent of inflammation, demyelination, and in WT C57BL/6 mice (Fig. 4). Laquinimod treatment reduced acute axonal injury within the spinal cords from WT mice with disease incidence from 100 to 20% and decreased the severity of MOG-induced EAE. The extent of damage in spinal cords from − − disease in mice that show clinical severity from 4.2 ± 0.7 to 0.3 ± WT and AhR / mice was assessed by histological analysis.

− − Fig. 4. Prophylactic treatment with laquinimod has no effect on clinical score in AhR / mice. The graph shows mean disease scores with SE from vehicle- − − treated (closed symbols) and laquinimod-treated (open symbols) WT or AhR / mice (n = 15 per group). The embedded table shows disease incidence, mean day of onset, and mean maximal score of disease. The data are representative of four independent experiments.

Kaye et al. PNAS Early Edition | 3of8 Downloaded by guest on October 1, 2021 Quantification of inflammatory demyelination as determined tokines and the frequency of regulatory T cells in spleens taken − − in luxol-fast-blue (LFB)/periodic acid-Schiff (PAS) with H&E from WT and AhR / mice. Treatment with laquinimod in WT + + counterstaining revealed a significant protective effect of laqui- mice significantly increased the percentage of CD4 CD25 − − + nimod in WT animals but not in AhR / animals (Fig. 5). Al- FoxP3 regulatory T cells (Fig. 6A) compared with vehicle-treated −/− though widespread inflammatory demyelination was evident mice (0.80 ± 0.06 vs. 0.37 ± 0.04; P < 0.05), whereas in AhR in vehicle-treated WT mice, these histopathological alterations mice no difference was observed (0.33 ± 0.04 vs. 0.38 ± 0.06). As were virtually absent in laquinimod-treated WT animals (P = expected, laquinimod treatment resulted in decreased production γ 0.0003). Unpaired t test with Welch’s correction showed no of IL-17, GM-CSF, and IFN in MOG-reactivated splenocytes P < significant difference between vehicle-treated and laquinimod- from WT mice, albeit only significantly for GM-CSF ( 0.04). In − − treated AhR / animals (P = 0.31). Quantification of acute ax- contrast, laquinimod treatment did not reduce MOG-specific IL- 17, GM-CSF, and IFNγ production in splenocytes taken from onal damage in amyloid precursor protein (APP)-stained slides −/− revealed extensive axonal damage in vehicle-treated mice. As AhR mice. In all mice, the recall response to purified protein expected, the number of APP spheroids was most intense in derivative (PPD) was not affected by laquinimod, and the level of cytokine release was not statistically different between WT and regions with white matter inflammation. Laquinimod was pro- − − AhR / mice (Fig. 6B). These data demonstrate that laquinimod tective in WT mice (P < 0.0001), but had no effect on acute − − expands regulatory T cells and limits T-effector cells in an AhR- axonal damage in AhR / mice (Fig. 5). Semiquantification of dependent manner in EAE. the extent of microglial and astrocytic activation was performed in Iba1- and GFAP-stained sections. Extensive microglial acti- AhR Deletion in the Immune System Is Necessary and Sufficient to vation and astrogliosis were present in vehicle-treated mice, and Negate the Effect of Laquinimod in EAE. Current knowledge indi- both were significantly reduced by laquinimod in WT mice. In cates that laquinimod exerts activities both on the peripheral −/− contrast, laquinimod had no effect on either parameter in AhR immune system and within the CNS. The data above clearly show mice (Fig. 5). that AhR is required for the efficacy of laquinimod in EAE; Treatment with laquinimod has been shown to decrease Th1 + however, it is not clear whether AhR is required in the periph- and Th17 responses with a corresponding increase in CD4 + + eral immune system, in the CNS, or in both. To address this, we CD25 FoxP3 regulatory T cells both in the periphery and made chimeras by transplanting bone marrow (BM) cells into − − within the CNS (4, 5, 7, 11). To investigate whether these im- busulfan-conditioned recipients: In one set of animals, AhR / munomodulatory effects of laquinimod are also mediated by the (CD45.2) BM cells were transplanted into congenic B6 CD45.1 − − AhR pathway, we analyzed the production of Th1 and Th17 cy- recipients to create mice with an AhR / peripheral immune

Fig. 5. Prophylactic treatment with laquinimod has no effect on inflammatory demyelination, acute axonal damage, or microglial and astroglial activation in − − AhR / mice. All stains were performed using n = 15 per treatment group. The inflammatory index was quantified using LFB/PAS and H&E staining of spinal cords taken from vehicle- or laquinimod-treated WT or AhR−/− mice. (Upper panels) A cross-section of the whole spinal cord using 4× objective (Nikon Eclipse E200). (Lower panels) A higher magnification (40×) taken from the area marked by a rectangle. Quantification of acute axonal damage using APP staining + − − showed a significant reduction (*P < 0.0001) in the number of APP spheroids in laquinimod-treated WT mice, but not in AhR / mice. Quantification of microglial and astrocyte activation using Iba-1 or GFAP staining showed a significant reduction (*P < 0.0001) in the number of Iba-1 (black bars) and GFAP − − (striped bars) cells in laquinimod-treated WT mice, but not in AhR / mice.

4of8 | www.pnas.org/cgi/doi/10.1073/pnas.1607843113 Kaye et al. Downloaded by guest on October 1, 2021 PNAS PLUS

Fig. 6. Laquinimod immunomodulation in EAE is AhR-dependent. (A) MOG EAE mice were treated daily with laquinimod (25 mg/kg, n = 6 per group) or + vehicle (water, n = 6 per group), and 15 d after immunization, spleen cells were removed and evaluated by FACS for expression of CD25 and Foxp3 by CD4 + + + cells. (Left) A representative dot plot from a vehicle-treated mouse, where CD4 CD25 cells were gated, and the percentage of FoxP3 cells was calculated + + from within that gate (Middle). The individual data from each animal demonstrates that laquinimod significantly increases the percentage of CD4 CD25 + − − FoxP3 regulatory T cells in WT and not AHR / mice. Data are representative of two independent experiments. (B) The remaining spleen cells were restimulated with PPD or MOG in vitro, and after 48 h the culture supernatants were analyzed for IL-17, GM-CSF, and IFNγ release. Under some conditions, the INFLAMMATION IMMUNOLOGY AND cytokine levels were below the level of detection and are marked as “BQL.” Laquinimod treatment reduced MOG-specific GM-CSF (P < 0.04), IL-17, and IFNγ. Data are representative of two independent experiments.

+ + system and AhR / CNS; and in another set of animals B6 depicted by clinical score (Fig. 4) and histopathological findings − − CD45.1 BM cells were transplanted into AhR / recipients to (Fig. 5). These data indicate that deletion of AhR is necessary and + + create mice with an AhR / peripheral immune system and an sufficient to abrogate the effect of laquinimod in EAE. − − AhR / CNS. Chimerism was tracked using FACS analysis of Treatment with laquinimod has been shown to shift the bal- peripheral blood, and at 13 wk posttransplant recipient mice had ance from pathogenic Th17 cells toward an increase in the + + + >80% donor-derived blood cells. At that point, chimeric mice number of CD4 CD25 FoxP3 regulatory T cells both in the were induced with MOG EAE and treated with vehicle or periphery and in situ within the CNS (4, 5, 7). Our RNA-Seq data + + laquinimod. In chimeras with an AhR / immune system, showed 227 genes had decreased expression levels in laquinimod- laquinimod treatment reduced disease severity from a mean treated mice, including key Th17-related cytokine genes (Fig. 1). clinical score of 2.5 ± 0.8–0.7 ± 0.8 (72% inhibition, Fig. 7), The mechanism by which AhR signaling modulates T-regulatory although the effect of laquinimod was not as efficacious as seen (T-reg) biology includes shaping T-reg differentiation by dictating in WT C57BL/6 mice where it inhibited disease by 93% (Fig. 4). the state of FoxP3 promoter methylation and enhancing FoxP3 In contrast, laquinimod completely lost its efficacy in chimeric expression; mediating methylation of the il17 promoter and de- − − mice with an AhR / immune system (Fig. 7) where disease creasing the expression of IL-17; and inducing tolerogenic DCs − − severity (2.9 ± 1.1) was comparable to vehicle-treated AhR / that promote the generation of T regs (25–30). Although we do mice (3.5 ± 0.5). These data indicate that AhR is fully required not have experimental evidence regarding the effect of laquinimod in the peripheral immune system for the effect of laquinimod on FoxP3 promoter methylation, laquinimod has been shown to and may be partially dependent on the expression of AhR for the reduce IL-17 production (5, 7) and induce tolerogenic DCs (7, 9). effect of laquinimod within the CNS. Furthermore, we show that both the reduction of proinflamma- + + + tory cytokines and the increase in CD4 CD25 FoxP3 regulatory Discussion T-cell numbers by laquinimod is AhR-mediated (Fig. 6). Consid- In this paper we used transcriptome analysis to further elucidate ering the link of AhR with other regulatory T-cell populations, the mechanism of action (MoA) of laquinimod, and we dem- specifically Tr1 cells (31–34), it will be interesting to investigate onstrate that laquinimod induces genes associated with the AhR whether laquinimod also modulates Tr1 cells. Various compounds pathway. The prototypical AhR genes Cyp1a1 and Ahrr were acting on the AhR pathway regulate adaptive immune responses among those with the highest average fold-change in both naive through effects on both APCs and T cells (29). In contrast, no and EAE mice treated with laquinimod (Fig. 2 A and B). We direct effects of laquinimod on T cells have been detected (7). verified the in vivo induction of Cyp1a1 using qPCR of mRNA Laquinimod, like other AhR activators, can induce the differen- levels from livers of treated mice (Fig. 2C) and demonstrated tiation of APCs with a tolerogenic phenotype. The mechanisms that laquinimod is a potent inducer of Cyp1a in vitro (Fig. 3). responsible for this tolerogenic phenotype may result from the up- Together, these findings demonstrate that laquinimod is an acti- regulation of indoleamine 2,3-dioxygenase expression, resulting in vator of AhR, and the exact molecular events that result in acti- increased synthesis of immunosuppressive kynurenines (21, 22). vation are currently being investigated. This raised the possibility Indeed, our transcriptome data showed up-regulation of both Ido1 that the therapeutic effect of laquinimod is dependent on AhR and Ido2 (Fig. 2). Taken together, these data suggest that the ef- activation. Indeed, we report here that the effect of laquinimod fects of laquinimod on the encephalitogenic T-cell response in EAE − − was completely lost in MOG-induced EAE in AhR / mice, as involve different cellular and potentially molecular mechanisms

Kaye et al. PNAS Early Edition | 5of8 Downloaded by guest on October 1, 2021 − − Fig. 7. Lack of AhR in the peripheral immune system is sufficient to negate the effect of laquinimod. Bone marrow chimeras were made from WT or AhR / − − mice into preconditioned AhR / or WT recipients. MOG EAE was induced in the chimeric mice 13 wk posttransplant, and the graph shows mean disease scores with SE from vehicle-treated (closed symbols) and laquinimod-treated (open symbols) mice (n = 9 or 10 per group). The embedded table shows disease incidence, mean day of onset, and group mean score of disease. The data are representative of a single experiment.

+ + from other AhR ligands. Similar to other ligand-activated efficacy of laquinimod in chimeras with an AhR / immune sys- transcription factors, AhR-mediated biological responses have tem. Indeed, our data would seem to corroborate a very recent been shown to be compound, cell type, and species-dependent publication that reports that expression of AhR within astrocytes (35). Studies to delineate in which target cells and by what mech- limits CNS inflammation (39). Although our data strongly suggest anism laquinimod activates AhR are currently ongoing in our that the anti-inflammatory effects of laquinimod are mediated by laboratory. AhR, the precise mechanism of its neuroprotective activities re- Laquinimod inhibits astrogliosis in EAE (Fig. 5), and astrocytes mains to be further investigated. have been proposed to play an important role in the protective In conclusion, we have demonstrated that laquinimod acti- effect of laquinimod in cuprizone-induced CNS demyelination vates AhR, which is necessary for its therapeutic efficacy in the (36). Laquinimod treatment reverses cuprizone-induced astro- MOG-induced EAE model of MS. AhR has been known for its gliosis and leads to decreased production of proinflammatory ability to mediate the biochemical, metabolic, and toxic effects factors and reduced NF-κB activation in cultured murine and of environmental chemicals. Recently, a paradigm shift has oc- human astrocytes (36). Interestingly, cross-talk between AhR and curred with the understanding that AhR has endogenous roles κ NF- B has been described in other cell types (37, 38), suggesting and is an important regulator of cell development, differentia- that laquinimod-mediated AhR activation may lead to down- tion, and function (40). AhR is an important regulator of the κ regulation of NF- B in astrocytes and possibly other immune- development and function of both innate and adaptive immune related cells. In an attempt to elucidate the importance of cells, mediated by the ability of AhR to respond to endogenous AhR in the peripheral immune system versus in resident CNS ligands generated from the host cell, diet, and microbiota (41– cells like astrocytes, we performed a cross-over bone marrow −/− 44). This recent paradigm shift has opened new avenues of re- transplant experiment. In chimeras with an AhR immune search in the possibility of targeting AhR to treat inflammatory system, laquinimod completely lost efficacy (Fig. 7), indicating diseases in which a role of AhR has been found, including lupus that the AhR in the immune system is sufficient and necessary (45) and colitis (46). Laquinimod may represent a first-in-class for the effect of laquinimod. The exact population of immune drug targeting AhR for MS and other diseases with inflammatory cells that requires AhR is currently being investigated. In chi- + + or neuroinflammatory components. meras with an AhR / immune system, laquinimod treatment inhibited disease by 72% (Fig. 7), although the effect of Materials and Methods laquinimod was not as strong as the almost complete inhibition Test Compounds and Formulation. Laquinimod was synthesized at Teva Phar- seen in WT C57BL/6 mice (Fig. 4). This finding supports the maceutical Industries, Ltd. The compound was dissolved at 2.5 mg/mL in pu- idea that laquinimod may be partially dependent on the ex- rified water and administered orally by gavage in a volume of 0.2 mL. ITE was pression of AhR within the CNS, where lack of expression of purchased from Tocris Biosciences and was freshly prepared daily in corn oil AhR in resident astrocytes may explain the less-than-expected and then administered intraperitoneally at a volume of 0.2 mL. Omeprazole

6of8 | www.pnas.org/cgi/doi/10.1073/pnas.1607843113 Kaye et al. Downloaded by guest on October 1, 2021 and 3-MC were purchased from Sigma-Aldrich. Busuflex (Busulfan) (6 mg/mL) The thresholds used to define statistical significance were a Benjamini- PNAS PLUS was purchased from Otsuka America Pharmaceutical Inc. Hochberg–corrected (FDR) P < 0.05 and an absolute fold-change > 2.0. The raw sequence data can be accessed with the accession no. PRJNA319255 at Mice. Healthy C57BL/6 mice at 6–8 wk of age were obtained from the Harlan www.ncbi.nlm.nih.gov/bioproject/319255. Animal Breeding Center, Rehovot, Israel. AhR knockout (AhR−/−) mice on a C57BL/6 background (C57BL/6-Ahrtm1.2Arte) were obtained from Taconic. Histopathology of Spinal Cords from EAE. At the end of the EAE study, animals Congenic C57BL/6 B6.SJL-Ptprca Pepcb/BoyJ (B6 CD45.1) mice were obtained were perfused with PBS solution containing 4% (vol/vol) paraformaldehyde. from Jackson Laboratories. The mice were housed at 22–24 °C, and food and Thereafter, the entire vertebral column was carefully dissected, and isolated water were available ad libitum. All experimental procedures conformed to vertebral columns were incubated overnight at 4 °C for the purpose of tissue accepted ethical standards for use of animals in research and were in ac- post fixation. For histological and immunohistochemical studies, spinal cords cordance with Committee for the Care and Use of Experimental Animal were decalcified for 48 h (37 °C), and then samples were washed for 12 h guidelines and approved by the Teva Institutional Animal Care and Use under running water to remove the decalcification solution. Spinal cords Committee. were embedded in paraffin, and 5-μm-thick transverse sections were prepared and stained for myelination/inflammatory index, microglial and Bone Marrow Transplantation. Ten congenic B6 CD45.1 or AhR−/− (CD45.2) astrocytic activation, and acute axonal damage. Intact and damaged myelin recipient mice were preconditioned with three intraperitoneal injections of plus inflammatory infiltrates were visualized using LFB/PAS stains and 10 mg/kg busulfan on days −5, −3, and −1 before transplantation. Donor counterstained with H&E. The extent of inflammatory demyelination was mice were killed, and bone marrow cells were isolated from the femurs quantified by assessing the inflammatory-demyelination index, which is and tibias of all four limbs. Following lysis of RBCs, 9–10 × 106 BM cells defined as the area covered by inflammatory demyelination in relation to were injected i.v. into the recipient mice. Peripheral blood was removed, the entire white matter area of each slide. For the visualization and quan- lysed, and resuspended in FACS buffer and then stained using CD45.1-PE tification of microglia/monocyte and astrocyte activation, paraffin-embed- or CD45.2-APC (Miltenyi Biotec). After 13 wk, mice were induced with ded sections were dewaxed, washed in PBS, and incubated overnight with MOG EAE. the respective primary antibody diluted in blocking solution. For the visualization of epitope-primary antibody complexes, HRP-coupled polymer sec- Induction of EAE and Clinical Evaluation. Mice were immunized in the flanks ondary antibodies were used (EnVision, Dako), and 3,3′-diaminobenzidine with 300 μg/mouse pMOG35–55 peptide (Novetide) in normal saline emul- (DAB) was used as a chromogenic substrate. Quantification of the extent sified in an equal volume of Complete Freund’s Adjuvant containing 500 μg of microgliosis/monocytosis (anti-Iba1) and astrogliosis (anti-GFAP) was per mouse Mycobacterium tuberculosis. Two injections of pertussis toxin performed using a blinded staging approach. The following scoring system (175 ng/0.2 mL per mouse, intraperitoneally) were given at the time of im- was used: 0 = normal cellular density; 1 = moderate increase in cellular munization and 48 h later. Animals were scored for clinical signs of disease density; 2 = intermediate increase in cellular density; 3 = high cellular density;

on a daily basis using the following scores: 0 = normal behavior, 1 = distal and 4 = maximum cellular density. Anti-APP stains were performed to visu- INFLAMMATION IMMUNOLOGY AND limp tail, 1.5 = complete limp tail, 2 = disturbed righting reflex, 3 = ataxia, alize acute axonal damage. To quantify the extent of acute axonal damage, 4 = early paralysis (hind legs), 5 = full paralysis, and 6 = moribund or death. the number of APP+ spheroids was counted in four randomly chosen fields of To calculate mean disease onset, animals that did not develop disease were the white matter part of the corpus callosum irrespective of the presence or considered to have onset on day 31. Laquinimod was dosed orally at absence of lesions. The number of positive spheroids was counted under high- 25 mg·kg·d beginning on the day of immunization until the end of the ex- power magnification and quantified as the number of spheroids per square periment (days 0–30). Placebo-treated mice were similarly administered a millimeter. volume of 0.2 mL water orally 6 d a week. Each group contained between 12 and 15 mice. CYP1A Induction in Primary Human and Murine Hepatocytes. Cryopreserved murine (Bioreclamation In Vitro Technologies) and human (Celsis In Vitro RNA Sequencing of Splenocytes from Naive and EAE Mice. Naive female C57BL/6 Technologies) hepatocytes were thawed and plated on 24 multiwell plates mice or mice with MOG EAE were treated with vehicle or 25 mg/kg coated with collagen type I substratum in William’s E medium (GIBCO-32551) laquinimod for 25 d. After 6 d, spleens from six mice per treatment arm supplemented with 5 μg/mL insulin, 0.1 μM dexamethasone, and 10% (vol/vol) were removed, and RNA was isolated using the Qiagen miRNeasy Mini Kit. FBS. Cells were for 3–4 h, and the medium was changed to Hepatocyte Basal Globin mRNA-depleted RNA samples were converted into cDNA libraries Medium (HBM-Lonza CC-3199) supplemented with CC-4182 (Lonza). Plates using the TruSeq Stranded mRNA Sample Prep Kit (Illumina, #RS-122-2103). were maintained at 37 °C for 24 h before treatment with compounds. Pro- Final cDNA libraries were analyzed for size distribution and, using an totypical CYP1A inducers 3-MC and omeprazole were dissolved in DMSO and Agilent 2200 TapeStation (D1000 Screentape, Agilent # 5067–5582), added to the culture medium at a final solvent concentration of 0.1% (vol/vol). quantitated by qPCR (KAPA Library Quant Kit, KAPA Biosystems # KK4824) After 4 or 24 h of treatment, total RNA was extracted from duplicate samples and then normalized to 2 nM in preparation for sequencing. Sequencing using an RNA extraction kit (RNeasy 96 kit, 74181, Qiagen). Quantitative was performed using an Illumina TruSeq Paired-End Cluster Kit V4 (Illu- mRNA analysis was performed by real-time qRT-PCR in the ABI Prism 7900 HT mina # PE-401-4001), and a clustered flowcell was generated using the Sequence Detection System (TaqMan, Perkin-Elmer-Applied Biosystem). EC50 normalized cDNA libraries as templates. The cDNA templates were dena- values were calculated by nonlinear regression using XLfit 4.2 (IDBS). turedusingfresh0.1NNaOH,dilutedtoafinalloadingconcentrationof 13 pM, and placed on an Illumina cBot (v1.5.12.0) for cluster generation. CYP1A mRNA Induction in Livers of WT Mice. Female C57BL/6 mice were Templates were attached to the flowcell via a dense lawn of oligonucle- treated for 5 d with vehicle, laquinimod (25 mg/kg, orally), or ITE (10 mg/kg, otides that bind to the sequencing adapters added during sample prepa- i.p.). Mouse liver samples were taken 4 or 24 h after the final dose of ration, which are extended and then denatured. The flowcell was then compound, and tissue samples were lysed and total RNA purified. Single- sequenced through 51 bases, paired end, with an 8-base index cycle on an stranded cDNA was prepared from RNA with the RT Master Mix using the AB Illumina HiSEq. 2000 (HiSeq Control Software v1.5.15.1). During sequenc- 7900HT Fast Real Time PCR System thermocycling program (Applied Biosys- ing cycles, fluorescent reversible terminator dNTPs were added to the tems). Probes for murine Beta-Actin (Mm01205647_g1), Cyp1a1 (Mm00487218_m1), clusters with only a single base per target being incorporated. Following and Cyp1a2 (Mm00487224_m1) were purchased from Thermo Fischer Sci- imaging of the clusters, the terminator and fluorescent tags were cleaved entific. The relative quantity of the target cDNA compared with that of the

so that the next base could be incorporated. control cDNA (β-actin) was determined by the ΔΔCt method. The results of this method are expressed as fold-change with respect to the target tran- RNA Sequencing Analysis. Gene expression values were obtained by aligning script expression in the untreated control. the sequencing reads to the mouse genome (GRCm38) using STAR (47) and − − using featureCounts (48) to quantify the number of reads that aligned T-Cell Phenotyping in WT and AhR / Mice. Following EAE induction, WT and − − uniquely to genes specified in the GENCODE mouse transcriptome M7 (49). AhR / mice (n = 6 per group) were treated with vehicle or laquinimod for The resulting count matrix was filtered by removing genes with less than 20 15 d. Spleens were removed under sterile conditions, crushed in PBS to cumulative counts across splenocyte samples. The filtered matrix was nor- create a suspension, and lysed with ammonium-chloride-potassium (ACK) malized with Limma Voom (50) using the mouse type (EAE or naive) and lysis buffer, and the cell pellet was resuspended in FACS buffer or cell + + + treatment (laquinimod or vehicle) variables to estimate the model coeffi- culture medium. Cells (1 × 106) were stained for CD4 CD25 Foxp3 using cients. Differential expression between EAE treated with laquinimod and the eBioscience mouse regulatory T-cell staining kit according to the EAE treated with vehicle was performed using Limma-moderated t tests (51). manufacturer’s instructions. The percentage of FoxP3+ cells was calculated

Kaye et al. PNAS Early Edition | 7of8 Downloaded by guest on October 1, 2021 + from within the CD4 CD25+ lymphocyte gate. The remaining cells were using the R&D mouse magnetic luminex kit LXSAMSM-14 according to the resuspended in Stimulation Medium (RPMI; 2.5% FCS, Antibiotics, manufacturer’s instructions. 6 L-glutamine and β−mercaptoethanol) and seeded at 0.5 × 10 cells per well in a 96-well, flat-bottom plate to a final volume of 0.25 mL/well. Cells ACKNOWLEDGMENTS. We thank Prof. Markus Kipp of ProMyelo GmbH, who were exposed to medium, PPD (2 μg per well), or MOG35–55 (10 μg per well) performed all of the histology analysis, and Dr. Annalise Di Marco from IRBM for 48 h. Cell culture supernatants were then analyzed for cytokine levels Science Park, who performed the AhR induction experiments in hepatocytes.

1. Comi G, et al.; ALLEGRO Study Group (2012) Placebo-controlled trial of oral laquini- 25. Quintana FJ, et al. (2008) Control of T(reg) and T(H)17 cell differentiation by the aryl mod for multiple sclerosis. N Engl J Med 366(11):1000–1009. hydrocarbon receptor. Nature 453(7191):65–71. 2. Filippi M, et al.; ALLEGRO Study Group (2014) Placebo-controlled trial of oral laqui- 26. Kimura A, Naka T, Nohara K, Fujii-Kuriyama Y, Kishimoto T (2008) Aryl hydrocarbon nimod in multiple sclerosis: MRI evidence of an effect on brain tissue damage. receptor regulates Stat1 activation and participates in the development of Th17 cells. – J Neurol Neurosurg Psychiatry 85(8):851 858. Proc Natl Acad Sci USA 105(28):9721–9726. 3. Vollmer TL, et al.; BRAVO Study Group (2014) A randomized placebo-controlled phase 27. Duarte JH, Di Meglio P, Hirota K, Ahlfors H, Stockinger B (2013) Differential influences – III trial of oral laquinimod for multiple sclerosis. J Neurol 261(4):773 783. of the aryl hydrocarbon receptor on Th17 mediated responses in vitro and in vivo. 4. Wegner C, et al. (2010) Laquinimod interferes with migratory capacity of T cells and PLoS One 8(11):e79819. reduces IL-17 levels, inflammatory demyelination and acute axonal damage in 28. Hayes MD, Ovcinnikovs V, Smith AG, Kimber I, Dearman RJ (2014) The aryl hydro- mice with experimental autoimmune encephalomyelitis. J Neuroimmunol 227(1-2): carbon receptor: Differential contribution to T helper 17 and T cytotoxic 17 cell de- 133–143. velopment. PLoS One 9(9):e106955. 5. Aharoni R, et al. (2012) Oral treatment with laquinimod augments regulatory T-cells 29. Quintana FJ (2013) Regulation of central nervous system autoimmunity by the aryl and brain-derived neurotrophic factor expression and reduces injury in the CNS of – mice with experimental autoimmune encephalomyelitis. J Neuroimmunol 251(1-2): hydrocarbon receptor. Semin Immunopathol 35(6):627 635. + 14–24. 30. Pot C (2012) Aryl hydrocarbon receptor controls regulatory CD4 T cell function. Swiss 6. Mishra MK, Wang J, Silva C, Mack M, Yong VW (2012) Kinetics of proinflammatory Med Wkly 142:w13592. monocytes in a model of multiple sclerosis and its perturbation by laquinimod. Am J 31. Apetoh L, et al. (2010) The aryl hydrocarbon receptor interacts with c-Maf to promote Pathol 181(2):642–651. the differentiation of type 1 regulatory T cells induced by IL-27. Nat Immunol 11(9): 7. Schulze-Topphoff U, et al. (2012) Laquinimod, a quinoline-3-carboxamide, induces 854–861. type II myeloid cells that modulate central nervous system autoimmunity. PLoS One 32. Gandhi R, et al. (2010) Activation of the aryl hydrocarbon receptor induces human 7(3):e33797. type 1 regulatory T cell-like and Foxp3(+) regulatory T cells. Nat Immunol 11(9): 8. Thöne J, et al. (2012) Modulation of autoimmune demyelination by laquinimod via 846–853. induction of brain-derived neurotrophic factor. Am J Pathol 180(1):267–274. 33. Wu HY, et al. (2011) In vivo induction of Tr1 cells via mucosal dendritic cells and AHR 9. Jolivel V, et al. (2013) Modulation of dendritic cell properties by laquinimod as a signaling. PLoS One 6(8):e23618. mechanism for modulating multiple sclerosis. Brain 136(Pt 4):1048–1066. 34. Mascanfroni ID, et al. (2015) Metabolic control of type 1 regulatory T cell differen- 10. Ruffini F, et al. (2013) Laquinimod prevents inflammation-induced synaptic alter- tiation by AHR and HIF1-α. Nat Med 21(6):638–646. ations occurring in experimental autoimmune encephalomyelitis. Mult Scler 19(8): 35. Denison MS, Soshilov AA, He G, DeGroot DE, Zhao B (2011) Exactly the same but – 1084 1094. different: Promiscuity and diversity in the molecular mechanisms of action of the aryl 11. Moore S, et al. (2013) Therapeutic laquinimod treatment decreases inflammation, hydrocarbon (dioxin) receptor. Toxicol Sci 124(1):1–22. initiates axon remyelination, and improves motor deficit in a mouse model of mul- 36. Brück W, et al. (2012) Reduced astrocytic NF-κB activation by laquinimod protects tiple sclerosis. Brain Behav 3(6):664–682. from cuprizone-induced demyelination. Acta Neuropathol 124(3):411–424. 12. Mishra MK, et al. (2014) Laquinimod reduces neuroaxonal injury through inhibiting 37. Ruby CE, Leid M, Kerkvliet NI (2002) 2,3,7,8-Tetrachlorodibenzo-p-dioxin suppresses microglial activation. Ann Clin Transl Neurol 1(6):409–422. tumor necrosis factor-alpha and anti-CD40-induced activation of NF-kappaB/Rel in 13. Zou LP, et al. (2002) Suppression of experimental autoimmune neuritis by ABR-215062 dendritic cells: p50 homodimer activation is not affected. Mol Pharmacol 62(3): is associated with altered Th1/Th2 balance and inhibited migration of inflammatory – cells into the peripheral nerve tissue. Neuropharmacology 42(5):731–739. 722 728. 14. Pitarokoili K, et al. (2014) Laquinimod exerts strong clinical and immunomodulatory 38. Vogel CF, et al. (2014) Cross-talk between aryl hydrocarbon receptor and the in- κ – effects in Lewis rat experimental autoimmune neuritis. J Neuroimmunol 274(1-2): flammatory response: A role for nuclear factor- B. J Biol Chem 289(3):1866 1875. 38–45. 39. Rothhammer V, et al. (2016) Type I interferons and microbial metabolites of trypto- 15. Lourenço EV, Wong M, Hahn BH, Palma-Diaz MF, Skaggs BJ (2014) Laquinimod delays phan modulate astrocyte activity and central nervous system inflammation via the and suppresses nephritis in lupus-prone mice and affects both myeloid and lymphoid aryl hydrocarbon receptor. Nat Med 22(6):586–597. immune cells. Arthritis Rheumatol 66(3):674–685. 40. Esser C, Rannug A (2015) The aryl hydrocarbon receptor in barrier organ physiology, 16. Sun J, et al. (2015) Laquinimod ameliorates spontaneous colitis in interleukin-10- immunology, and toxicology. Pharmacol Rev 67(2):259–279. gene-deficient mice with improved barrier function. Int Immunopharmacol 29(2): 41. Nguyen NT, Hanieh H, Nakahama T, Kishimoto T (2013) The roles of aryl hydrocarbon 423–432. receptor in immune responses. Int Immunol 25(6):335–343. 17. Varrin-Doyer M, Zamvil SS, Schulze-Topphoff U (2014) Laquinimod, an up-and-coming 42. Quintana FJ, Sherr DH (2013) Aryl hydrocarbon receptor control of adaptive immu- immunomodulatory agent for treatment of multiple sclerosis. Exp Neurol 262(Pt A): nity. Pharmacol Rev 65(4):1148–1161. 66–71. 43. Cella M, Colonna M (2015) Aryl hydrocarbon receptor: Linking environment to im- 18. Hu W, Sorrentino C, Denison MS, Kolaja K, Fielden MR (2007) Induction of cyp1a1 is a munity. Semin Immunol 27(5):310–314. nonspecific biomarker of aryl hydrocarbon receptor activation: Results of large scale 44. Zhou L (2016) AHR function in lymphocytes: Emerging concepts. Trends Immunol screening of pharmaceuticals and toxicants in vivo and in vitro. Mol Pharmacol 71(6): 37(1):17–31. – 1475 1486. 45. Dorgham K, et al. (2015) Ultraviolet light converts propranolol, a nonselective 19. MacPherson L, et al. (2014) Aryl hydrocarbon receptor repressor and TiPARP (ARTD14) β-blocker and potential lupus-inducing drug, into a proinflammatory AhR ligand. Eur use similar, but also distinct mechanisms to repress aryl hydrocarbon receptor sig- J Immunol 45(11):3174–3187. naling. Int J Mol Sci 15(5):7939–7957. 46. Lamas B, et al. (2016) CARD9 impacts colitis by altering gut microbiota metabolism of 20. Boutros PC, et al. (2011) Hepatic transcriptomic responses to TCDD in dioxin-sensitive tryptophan into aryl hydrocarbon receptor ligands. Nat Med 22(6):598–605. and dioxin-resistant rats during the onset of toxicity. Toxicol Appl Pharmacol 251(2): 47. Dobin A, et al. (2013) STAR: Ultrafast universal RNA-seq aligner. Bioinformatics 29(1): 119–129. – 21. Jaronen M, Quintana FJ (2014) Immunological relevance of the coevolution of IDO1 15 21. and AHR. Front Immunol 5:521. 48. Liao Y, Smyth GK, Shi W (2014) featureCounts: An efficient general purpose program – 22. Nguyen NT, et al. (2014) Aryl hydrocarbon receptor and kynurenine: Recent advances for assigning sequence reads to genomic features. Bioinformatics 30(7):923 930. in autoimmune disease research. Front Immunol 5:551. 49. Harrow J, et al. (2012) GENCODE: The reference human genome annotation for The 23. Hao N, et al. (2012) Xenobiotics and loss of cell adhesion drive distinct transcriptional ENCODE Project. Genome Res 22(9):1760–1774. outcomes by aryl hydrocarbon receptor signaling. Mol Pharmacol 82(6):1082–1093. 50. Law CW, Chen Y, Shi W, Smyth GK (2014) voom: Precision weights unlock linear 24. Quintana FJ, et al. (2010) An endogenous aryl hydrocarbon receptor ligand acts on model analysis tools for RNA-seq read counts. Genome Biol 15(2):R29. dendritic cells and T cells to suppress experimental autoimmune encephalomyelitis. 51. Ritchie ME, et al. (2015) limma powers differential expression analyses for RNA-sequencing Proc Natl Acad Sci USA 107(48):20768–20773. and microarray studies. Nucleic Acids Res 43(7):e47.

8of8 | www.pnas.org/cgi/doi/10.1073/pnas.1607843113 Kaye et al. Downloaded by guest on October 1, 2021