Genetic and pharmacological inhibition of the nuclear receptor RORα regulates TH17 driven inflammatory disorders

Full development of IL-17 producing CD4+ T helper cells (TH17 cells) requires the transcriptional activity of both orphan nuclear receptors RORα and RORγt. However, RORα is considered functionally redundant to RORγt; therefore, the function and therapeutic value of RORα in TH17 cells is unclear. Here, using mouse models of autoimmune and chronic inflammation, we show that expression of RORα is required for TH17 cell pathogenicity. T-cell-specific deletion of RORα reduces the development of experimental autoimmune encephalomyelitis (EAE) and colitis. Reduced inflammation is associated with decreased TH17 cell development, lower expression of tissue-homing chemokine receptors and integrins, and increased frequencies of Foxp3+ T regulatory cells. Importantly, inhibition of RORα with a selective small molecule antagonist mostly phenocopies our genetic data, showing potent suppression of the in vivo development of both chronic/progressive and relapsing/remitting EAE, but with no effect on overall thymic cellularity. Furthermore, use of the RORα antagonist effectively inhibits human TH17 cell differentiation and memory cytokine secretion. Together, these data suggest that RORα functions independent of RORγt in programming TH17 pathogenicity and identifies RORα as a safer and more selective therapeutic target for the treatment of TH17-mediated autoimmunity. Unlike RORα, which has been thought to be somewhat redundant, RORγt has been well characterized in its function and contribution to the development of Th17 cells. Here the authors show that RORα is important in Th17 differentiation and that RORα deletion or a small molecule inhibitor of RORα can reduce disease in EAE and colitis mouse models.

T helper 17 (T H 17) cells are an important lineage of CD4 + T effector (Teff) cells that survey mucosal barriers, secrete IL-17-family cytokines, and guard against extracellular bacteria and fungi at mucosal surfaces. Patients with defects in the T H 17 pathway, including autosomal dominant Hyper-IgE syndrome (AD-HIES), suffer from recurrent infections to mucocutaneous fungal and bacterial infections, such as Candida albicans and Staphylococcus aureus [1][2][3] . In contrast, over-exuberant T H 17 responses have been implicated in many human autoimmune diseases, including multiple sclerosis, rheumatoid arthritis, and inflammatory bowel disease (IBD) 4 . Genome-wide association studies indicate SNPs in multiple genes associated with the T H 17 axis as susceptibility factors in autoimmunity 5,6 . As a consequence, a large effort has been put forth to target this pathway for the treatment of T H 17-mediated diseases, and in particular to identify pharmacologically accessible mechanisms that block pathogenic, but not protective, T H 17 cell functions.
The nuclear receptor (NR) retinoic acid receptor-related orphan receptor gamma t (RORγt; NR1F3) is considered the lineage defining transcription factor of T H 17 cells 7 . Since its discovery over a decade ago, a significant amount of work has been invested into understanding RORγt's function in T H 17 cells, and to translate this understanding into new treatments for autoimmune and inflammatory syndromes. While RORγt is critical for T H 17 cell development, IL-17A is not completely abolished in RORγt-deficient mice; its close family member, RORα (NR1F1), is also required for full T H 17 cell development 8 . Thus, deletion of both receptors completely abolishes T H 17 cell development 8 . Despite this, RORα is considered redundant to RORγt and as such the functions of RORα in protective and pathogenic T H 17 cell responses have not been systematically explored.
Here we show that RORα is a critical factor required for the development of autoimmunity and chronic inflammatory responses using mouse models of multiple sclerosis and colitis. Importantly, as a member of the nuclear receptor superfamily, RORα is a ligand-regulated transcription factor and targeting of this receptor in vivo not only inhibits the development of autoimmunity, it effectively suppresses active inflammation. These effects extend to human T H 17 cells, both during differentiation and in the regulation of pro-inflammatory cytokine expression in lineage-committed T H 17 cell memory subsets. Unlike pharmacologic inhibition of RORγt, we show that RORα inhibition suppresses pro-inflammatory T H 17-responses while inducing immunoregulatory function, while leaving thymic T cell development intact. Collectively, these data suggest that RORα is a non-redundant factor that is required for pathogenic T H 17 cell function and represents an attractive alternative for treating T H 17-associated inflammatory diseases in humans.

Results
RORα expression is sufficient to drive T H 17 signature genes. While RORα has been reported to be expressed in T H 17 cells, we wanted to establish its kinetic expression pattern during T H 17 cell development 8 . Naïve CD4 + T cells were differentiated under T H 17 polarizing conditions (TGFβ + IL-6) to assess the mRNA and protein expression of RORα versus RORγt. Quantitative realtime polymerase chain reaction (qRT-PCR) analysis demonstrated that RORα (Rora) was upregulated at similar time points as RORγt (Rorc) driving Il17a expression 7 (Fig. 1a). This data is consistent with microarray analysis demonstrating that Rora is upregulated in both pathogenic and non-pathogenic T H 17 cells to levels equal to or greater than Rorc 9 . Western blot analysis confirmed that RORα was expressed at the protein level (Fig. 1b).
To determine whether RORα alone was sufficient to drive IL-17A and T H 17-specific gene expression, we retrovirally overexpressed empty vector (MIGR1), RORα, or RORγt under T H neutral conditions (αIL-4 + αIFNγ, no cytokines) in stimulated naïve CD4 + T cells to assess IL-17A/F expression by flow cytometry. Neutral conditions were used to not confound results since TGFβ and IL-6 upregulate both RORs 8,10,11 . RORα was able to induce IL-17A expression, as was RORγt (Fig. 1c). GFP + cells were sorted to perform qRT-PCR to compare RORα vs. RORγt's ability to drive gene expression. RORα induced the expression of Il17a, Il17f, Il23r, Ccr6, and Csf2, all genes that are considered "core" RORγt target genes 12 (Fig. 1d). These data demonstrate that RORα is capable of driving the T H 17 cell developmental program.
Previous studies assessing RORα's function in the immune system utilized the Staggerer (Rora sg/sg ) mouse, a naturally occurring mutant mouse strain that renders RORα inactive 13 . To more specifically assess the role of RORα in mature CD4 + T cell responses, we developed RORα "floxed" mice and bred them with Cd4 Cre mice to delete RORα from T cells ( Supplementary  Fig. 1A). Differentiation of naive CD4 + T cells under nonpathogenic and pathogenic T H 17 inducing conditions [TGFβ + IL-6 (top panels); TGFβ + IL-6 + IL-23 (middle panels); IL-6 + IL-1β + IL-23 (bottom panels)] from Rora fl/fl and Rora fl/fl x CD4 Cre mice indicated that despite equal amount of RORγt expression between WT and KO conditions, cells lacking RORα were unable to express IL-17A at similar levels as RORα sufficient cells (Fig. 1e). Consistent with previous reports, RORγt-deficient cells had a decreased ability to produce IL-17A 14 indicating that combined expression of both RORs is required for full IL-17A expression ( Supplementary Fig. 1B). We observed no difference in the development of naïve CD4 + T cells into T H 0, T H 1, T H 2, inducible T regulatory (iTregs) or regulatory T (Tr1) cells from Rora fl/fl and Rora fl/fl x CD4 Cre mice ( Supplementary Fig. 1C). qRT-PCR analysis of T H 17 cells cultured under non-pathogenic and pathogenic T H 17 inducing conditions (see Fig. 1e for conditions) demonstrated that RORα deficiency reduced T H 17-specific gene expression, including Il17a, Il17f, Il23r, and Ccr6 whereas no difference was observed in the expression of Abca1 ( Fig. 1f and Supplementary Fig. 2A, B).
To gain a more global understanding of how RORα affects T H 17 cell development, we performed RNA-sequencing on T H 17 cells from Rora fl/fl and Rora fl/fl × CD4 Cre mice (Fig. 1g). KEGG pathway analysis indicated RORα was involved in regulation of cytokine-cytokine receptor interactions, inflammatory bowel disease, TNF signaling pathways, and Jak-STAT signaling pathways. We compared our data to RORγ-deficient T H 17 cells and found that RORα had a smaller gene signature than RORγt 12 , with only 79 genes in common between the two receptors in T H 17 cells, including "core" RORγt target genes 12 (Il17a, Ccr6, Ccl20, etc.) suggesting RORα and RORγt shared regulation of these genes, which is consistent with the little that is known about RORα's role in T H 17 cells 8 (Fig. 1h). To begin to address RORαand RORγt-direct vs. indirect target genes, we compared gene expression in RORα-and RORγt-deficient T H 17 cells relative to RORα vs. RORγt cistromes from liver and T H 17 cells, respectively 12,15 . The RORα liver cistrome will still give some indication of RORα direct target genes in tissues. Using this approach, we identified a small number of shared RORα/γt-direct target genes, including Furin, Il17ra, and Il1r1. While our analysis indicated a large number of RORα indirect target genes, including Il17a, Ccl20, Socs3, and Ccr6, this may be a consequence of probing the RORα cistrome in a tissue other than T cells. However, a number of interesting direct RORα (only) target genes were uncovered, including those located in the mitochondrial membrane, Slc11a2, Idh3a, Gpd2, and Ndufb2, suggesting RORα may play a role in the regulation of mitochondrial metabolism. T H 17 pathogenicity has been shown to be regulated by high rates of cellular metabolism and changes in glycolysis, OXPHOS, fatty acid or glutamine metabolism can prevent T H 17-mediated autoimmune disease [16][17][18][19][20][21][22] . Given most of these genes were downregulated, decreased OXPHOS could be a contributing factor to the decreased T H 17 responses in vitro. Interestingly, Fosl2, previously identified as a key T H 17 regulator 12 , also appears to be directly regulated by RORα. Given it acts as a repressor, regulation of Fosl2 expression could also account for the decreased T H 17 responses observed in RORα-deficient T H 17 cells.
These data indicate that RORα may regulate genes in diverse categories (i.e. ion transport, NADH metabolic processes, cell migration, etc.) that drive its function regulating cellular T H 17 responses.
RORα is required for the development of chronic inflammatory diseases. Recently, polymorphisms in RORA have been identified and associated with an increased risk for multiple sclerosis (MS) 23 . While found in non-coding regions, these SNPs were predicted to affect the affinity of transcription factor binding in regulatory regions of the RORA gene 23 . To establish how T-cell specific loss of RORα affected autoimmune disease development, we immunized Rora fl/fl and Rora fl/fl × CD4 Cre mice with MOG  peptide/complete freund's adjuvant (CFA) plus pertussis toxin to induce experimental autoimmune encephalomyelitis (EAE), a commonly used T H 17-dependent mouse model of chronicprogressive MS 24,25 . While the incidence of disease between the two groups was not significantly different, Rora fl/fl × CD4 Cre mice displayed attenuated disease severity relative to Rora fl/fl mice (Fig. 2a). Analysis of CD4 + T cells from mice killed at the peak of disease (~Day 15) demonstrated a decreased percentage and number of CCR6 + cells in the draining lymph nodes (LNs) and central nervous system (CNS) in the Rora fl/fl × CD4 Cre mice ( Fig. 2b and Supplementary Fig. 3). CCR6 is a chemokine receptor that permits T H 17 cell migration into inflamed peripheral tissues, including the CNS and gut 26 and is thought to be a RORγt target gene 12 . Genetic deletion or inhibition of CCR6 function has been shown to significantly reduce the severity of EAE 27 . Interestingly, there appeared to be an increase in the percent of CD25 + Foxp3 + cells in the LNs and CNS of Rora fl/fl × CD4 Cre mice compared to Rora fl/fl control mice (Fig. 2c). This correlated with the decreased percent and number of RORγt + cells in LNs and CNS (Fig. 2d). In line with this, there were significantly less IL-17A + cells, and importantly, significantly less IL-17A + IFNγ + cells in Rora fl/fl × CD4 Cre mice compared to Rora fl/fl control mice (Fig. 2e). IL-17A + IFNγ + cells have been linked to pathogenicity in several models of chronic disease, including EAE/MS [28][29][30][31] . These data suggest that RORα plays an important, non-redundant role to RORγt in the development of EAE.
We next assessed whether loss of RORα affects the development of colitis, another chronic inflammatory disorder associated with increased T H 17 differentiation and function 29,32 . We sorted naïve CD4 + T cells from Rora fl/fl and Rora fl/fl x CD4 Cre mice and transferred them (i.p.) into Rag1 −/− recipients and monitored their weight over time. PBS/sham injected mice were used as a control. Mice receiving Rora fl/fl × CD4 Cre T cells developed minimal signs of disease (weight loss, diarrhea, changes in colon length), unlike mice receiving Rora fl/fl naïve CD4 + T cells (Fig. 3a). Pro-inflammatory cytokine gene expression from proximal colons, including Il17a, Il17f, Il1b, Il6, Tnf, Ifng, Il22, and Il23p19, was also significantly decreased in mice receiving Rora fl/fl × CD4 Cre T cells compared to Rora fl/fl recipients (Fig. 3b). We also observed decreased expression of Cxcl10, a gene that is activated downstream of IL-17 and IL-22 signaling, which is consistent with the overall decreased T H 17 proinflammatory profile observed in vivo. As expected, histological analysis revealed that mice receiving Rora fl/fl T cells developed more severe colitis, evidenced by the increased cellular infiltration, broadening of crypts, and epithelial hyperplasia (Fig. 3c). In contrast, mice receiving Rora fl/fl × CD4 Cre T cells showed mild signs of inflammation in the proximal and distal colons. Analysis of mice at 6 weeks post-transfer indicated there were less Rora fl/fl × CD4 Cre cells infiltrating the colon, likely due to Rora fl/fl × CD4 Cre cells expressing significantly less α4β7, an integrin that mediates lymphocyte extravasion to the intestinal lamina propria 33 (Fig. 3d, e). Similar to what was observed in the EAE model, Rag1 −/− recipients receiving Rora fl/fl × CD4 Cre cells showed an increase in the percent of CD25 + Foxp3 + cells in the mesenteric LNs (mLNs) and colons compared to mice receiving Rora fl/fl cells (Fig. 3f). Mice receiving Rora fl/fl × CD4 Cre T cells also had significantly less RORγt, IL-17A + , and IL-17A + IFNγ + cells (Fig. 3g). Populations of T cells simultaneously expressing both cytokines have been described in intestinal biopsies of IBD patients 34,35 . Collectively, our genetic data suggests that RORα plays a significant role in the development of T H 17 cells, pathogenicity, and T H 17-mediated pro-inflammatory diseases.
RORα controls T H 17 cell responses through a cell intrinsic manner in vivo. To address whether RORα affects T H 17 cell function through cell-intrinsic or cell-extrinsic mechanisms in vivo, we took advantage of a co-transfer model. We transferred equal numbers of congenic CD45.1 + (WT) or CD45.2 + RORα KO (KO) naïve CD4 + T cells (i.p.) into Rag1 −/− recipients (Fig. 4a). After 5 weeks of expansion, there was no observable difference in the ratio between WT and KO T cells in the spleen. However, there was a significant difference in the ratio of WT to KO T cells in the mLNs and colon lamina propria (LP) with WT markedly outcompeting KO T cells (Fig. 4a). Despite this phenotype observed in the spleen, we observed a decreased percentage of CCR6 + cells in all KO T cells relative to WT in all tissues (spleen, mLNs, and colon LP; Fig. 4b). RORα KO T cells also expressed significantly decreased amounts of RORγt + cells as well as IL-17A + and IL-17A + IFNγ + cells in mLN and colon LP relative to WT T cells (Fig. 4c, d). These results demonstrate that RORα is required for T cell accumulation in the intestine and the downstream T H 17 cell pro-inflammatory response. Furthermore, it cannot be rescued by the presence of RORα-sufficient T cells. Therefore, RORα activity in T H 17 cells is required for their full effector function in vivo through a cell-intrinsic manner.  A RORα-selective small molecule regulates T H 17 cell development. We previously identified and characterized SR3335, a proof-of-concept RORα-selective inverse agonist 36 . To date, SR3335 is the only RORα inverse agonist in the primary literature that has been demonstrated to bind RORα and modulate its activity 36 . When tested against all 48 human nuclear receptors, SR3335 exhibited modest activity at LXRβ and PXR, but this activity was negligible compared to full LXR and PXR agonists T0901317 37 and Rifampicin 38 , respectively (Supplementary Fig. 4A-C). To determine whether pharmacological modulation of RORα's activity can affect IL-17A expression, naïve CD4 + T cells were differentiated under several T H 17 conditions (non-pathogenic and pathogenic) in the presence of SR3335 or vehicle control. SR3335 dose dependently suppressed IL-17A expression without affecting cell viability (Fig. 5a). This effect was specific to RORα and T H 17 cells as the effect was lost when tested in Rora fl/fl × CD4 Cre T H 17 cells nor did SR3335 affect the development of T H 1 or iTregs cells ( Supplementary Fig. 4D, E). Gene expression analysis indicated that SR3335 inhibited the expression of Il17a, Il17f, Il21, Il23r, and Ccr6 without affecting the expression of the LXR target gene Abca1 (Fig. 5b). Since loss of RORγ and many recently developed RORγ modulators induce thymic apoptosis, we wanted to determine if genetic deletion and/ or targeting RORα induced similar effects. Unlike RORγ 39 , T-cellspecific deletion of RORα had no effect on T cell development in the thymus (Fig. 5c, upper panels). To determine if SR3335 had any effects on thymic apoptosis, we treated C57BL/6 mice for 3 days with SR3335 (30 mg/kg, b.i.d, i.p.), the RORγ-selective inverse agonist SR2211 40 (20 mg/kg, b.i.d., i.p.), or vehicle controls. After 3 days, mice were killed to collect thymi to assess T cell development and total thymic cell numbers. Unlike mice treated with SR2211, which demonstrated decreased overall thymic cell number, percent of cells in the double positive stage, and total number of cells in the double positive stage, mice treated with SR3335 resembled vehicle control treated mice, indicating SR3335 had little effect on the thymus (Fig. 3c, lower panels). These data indicate that RORα-selective inverse agonists can effectively target the T H 17 pathway in vitro but do not appear to carry the associated risks of thymic apoptosis that RORγ inverse agonists do.
SR3335 inhibits the development of autoimmunity. To establish whether SR3335 was effective at inhibiting the development of autoimmunity in vivo, we immunized C57BL/6 mice to induce EAE and treated them with SR3335 (i.p. 30 mg/kg, B.I.D.) or vehicle control following immunization for the duration of the experiment. Dosing was based on drug metabolism and pharmacokinetic experiments performed in-house ( Supplementary  Fig. 5A). Disease development and incidence was delayed and significantly reduced in SR3335-treated animals relative to vehicle controls (Fig. 6a). Weights of the animals indicated no overt toxicity of the drug. Analysis of CD4 + T cells at the peak of disease indicated a significantly decreased frequency of CD3 + CD4 + T cells in the CNS whereas the numbers of cells were decreased in the LNs and the CNS (Fig. 6b). The decreased frequency and number of cells in the CNS was likely due to the decreased percent of CCR6 + cells in the CNS (Fig. 6c). SR3335 treatment led to a significant decrease in the frequency and number of RORγt + cells in the LNs and CNS (Fig. 6d). This correlated with the decreased frequency and number of GM-CSF + and IL-17A + cells in the LNs and CNS (Fig. 6e, f). GM-CSF-expressing T H 17 cells have been shown to be highly encephalitogenic 41,42 (Fig. 6e). While there was only a very small difference in the frequency of IL-17A + IFNγ + cells in the SR3335-treated mice, the absolute numbers of these cells were significantly lower in the CNS of the drug-treated mice compared to vehicle controls (Fig. 6f). These data suggest that treatment with a RORα-selective ligand largely phenocopy's the genetic data and is effective at inhibiting the development of autoimmunity.
RORα-selective ligands are inhibitory during active inflammation. To evaluate whether SR3335 was effective during active inflammation we tested SR3335 in the PLP 139-151 -induced relapsing-remitting EAE (R-EAE) model. SR3335 (30 mg/kg, B.I.D.) or vehicle was administered once mice had recovered from the first wave of disease (day 18 post-immunization). SR3335 significantly reduced the relapse rate and severity of relapse compared to animals receiving vehicle (Fig. 7a). Similar to the chronic EAE model, SR3335 appeared to inhibit cellular infiltration into the CNS, indicated by the decreased number of cells in the CNS relative to the draining lymph nodes (Fig. 7b). This was likely due to the decreased expression of CCR6 on CD4 + T cells (Fig. 7c). Intracellular FACS analysis demonstrated a decreased frequency and number of RORγt + cells in the LNs and CNS of the SR3335-treated mice relative to vehicle controls ( Fig. 7d). While we did observe a small decrease in IL-17A expression in the LNs of these mice, we did not observe a difference in the frequency of RORγt + IL-17A expression in the CNS of vehicle versus SR3335-treated mice. However, there was a significant reduction of the number of IL-17A + and pathogenic IL-17A + IFNγ + cells in CNS compared to mice receiving vehicle control (Fig. 7e). These results offer proof-of-concept that pharmacological modulation of RORα activity post-disease onset (therapeutic mode) can suppress the progression of T H 17-driven autoimmunity and chronic inflammatory disorders.

SR3335 regulates memory T H 17 and T H 17.1 cells.
It has been difficult to differentiate naïve human CD4 + T cells into T H 17 cells that secrete a substantial amount of IL-17A. However, recent work has demonstrated that anti-CD3 stimulation in the presence of IL-23 and IL-1β, without anti-CD28, is sufficient to drive an abundant IL-17 response in human T cells 43 . To determine if SR3335 was effective at inhibiting human T H 17 cell differentiation, we isolated human naïve CD4 + T cells from peripheral blood mononuclear cells (PBMCs) and differentiated them under conditions described by Revu et al. 43 . Similar to mouse T H 17 cells, SR3335 dose dependently inhibited the development of human T H 17 cells without affecting cell viability (Fig. 8a). To establish whether inhibition of RORα activity would affect T H 17 cells derived from patients with IBD, specifically Ulcerative Colitis (UC), we differentiated naïve CD4 + T cells from PBMCs from UC patients vs. healthy, age-matched controls (HC) under conditions described in Fig. 8a. While PBMCs from UC patients consistently yielded greater amounts of IL-17A and IFNγ, SR3335 was still able to inhibit IL-17A in a dose dependent manner ( Supplementary Fig. 5b).
IL-17A expressing memory cells (CD45RO + ) are enriched within PBMCs and can be identified by their co-expression of CCR6 and CCR4 44 . However, CCR6 + human memory cells can be further subdivided based on expression of CXCR3, which describes a subset of cells that expresses both IL-17A and IFNγ upon ex-vivo stimulation 45 . These two populations of cells have been demonstrated to be stable and are referred to as T H 17 (CCR6 + CCR4 hi CXCR3 lo ) and T H 17.1 (CCR6 + CCR4 lo CXCR3 hi ) 46 (Fig. 8d, e). This SR3335-mediated increase in IL-22 extends to mouse memory cells as culture of sorted mouse CCR6 + cells treated with SR3335 induced a similar response ( Supplementary Fig. 5C). Gene expression analysis from these samples revealed that SR3335 inhibited expression of many "core" T H 17 genes, including IL17A, IL17F, CCR6, and IL12RB1, which was consistent with much of the mouse data (Fig. 8f). Interestingly, SR3335 treatment also led to an increase in the transcription factor MAF in T H 17 and T H 17.1 populations. Collectively, these results indicate that RORα-selective inverse agonists can inhibit T H 17 cell functions in lineage-committed human memory T cells as well as from patients with T H 17 inflammatory diseases, like UC.

Discussion
Here we used a combination of genetic, molecular biology, and pharmacological approaches to demonstrate that RORα is a    Despite its clear effects and synergism with RORγt on T H 17 cell development 8 , the functional role of RORα was never expanded upon. Importantly, the T-cell specific role of RORα or its effects in other models of chronic inflammation were never explored because RORα was considered "redundant" to RORγt. In nuclear receptor research, there are many receptor sub-families where one member is considered "dominant" and the other "redundant." However, recent studies have started to tease apart the distinct signaling pathways for receptors in these sub-families demonstrating that they each possess a unique, requisite function in each cell type. For example, RXRα (NR2B1) and RXRβ (NR2B2) were recently shown to have distinct functions in retinoic acid signaling and neuronal differentiation 48 . Both LXRα (NR1H2) and LXRβ (NR1H3) were each found selectively associated with sequence motifs of specific transcription factors demonstrating LXRα-or LXRβ-selective gene regulation in macrophages 49 . Since redundancy suggests dispensable or unnecessary, our data indicates that RORα is a requisite factor for T H 17 cell differentiation and the development of autoimmunity. Indeed, a recent publication has hinted at RORα's function in T H 17 cell pathogenicity in a Rip2-dependent fashion. In this report, Rip2 negatively regulated the expression of RORα as Rip2 −/− CD4 + T cells expressed more RORα than WT CD4 + T cells 50 . The authors reported that RORα drove the expression of Il1r1, which they determined was RORα-dependent and was the basis for enhanced T H 17 cell pathogenicity. While the increase in IL-1β signaling is consistent with increased pathogenicity observed in T H 17 cells 51,52 , our data analysis has determined that Il1r1 is regulated by both RORα and RORγt. Nevertheless, it still needs to be determined how and why Rip2 is regulating RORα expression in T H 17 cells and what the biological significance of this is. Our genomics analysis did, however, identify some unique direct RORα target genes, including those that are important for mitochondrial metabolism (Idh3a, Gpd2, and Ndufb2), ion transport (Slc11a2), and other T H 17 cell transcriptional regulators (Fosl2), underscoring the need to further probe the function of this transcription factor. What is interesting to note is that numerous genes, like Ccr6, Il17a, Il23r, did not appear to be direct RORα target genes. This is likely a function of the cell type used for ChIP-seq analysis and further analysis assessing RORα's cistrome in T H 17 cells is warranted to understand its full regulatory functions.
Interestingly, deletion of RORα increased Treg frequencies in vivo. However, treatment with SR3335 trended towards an increase, but did not significantly phenocopy the knock out mice. This may be due, in part, to the pharmacokinetic properties of SR3335 in vivo. The K i of SR3335 was reported as 220 nM with an IC 50 of 480 nM in a Gal4-RORα ligand binding domain cotransfection assay 36 . With the plasma exposure at 230 nM at 6 h and falling quickly thereafter, the activity on target is likely not 100% throughout a 24-h period, even with B.I.D. dosing. Perhaps use of a more potent inverse agonist, with better in vivo exposure, would generate a more significant response and phenocopy the genetic models. Alternatively, a complete deletion of RORα may be necessary to observe the phenotype as changes in chromatin structure are likely to happen in the absence of a gene more so than through pharmacological modification. Regardless, RORα has been demonstrated to be expressed in skin-resident Tregs and its expression is required to restrain allergic skin inflammation 53 . Acquisition of RORα expression appears to occur once the cells are in the skin as RORα expression is low in skin draining lymph nodes 53 . This is intriguing given a previous report described an antagonistic role for Foxp3 in RORα-mediated gene transcription. In this scenario, Foxp3, via its LXXLL domain, directly interacted with RORα to inhibit Il17a transcription 10 . While some of the genes differentially expressed in skin-resident RORα-expressing Tregs are similar to those in RORα-deficient T H 17 cells (Ccr6, Cd44, Nr1d1, etc.), perhaps this molecular antagonism between RORα and Foxp3 may not occur in skin-resident Tregs, or only at select target genes. Based on the increased frequencies of Tregs in RORα-deficient mice with EAE and colitis, this molecular antagonism might occur at sites with different inflammatory stimuli than in the skin. A similar cooperativity has been described for RORγt and STAT3 in the differentiation of T H 17 vs. Tregs 11 . In the absence of pro-inflammatory cytokines, TGFβmediated Foxp3 induction drives Tregs by inhibiting RORγt. However, in the presence of pro-inflammatory cytokines, including IL-6, RORγt is stabilized, promoting T H 17 differentiation 11 . RORγt-deficient mice have increased Foxp3 + Tregs 7 and are protected from EAE, similar to what we observe in RORαdeficient T cells. It is possible that a similar mechanism occurs between RORα and Foxp3 in the regulation of T H 17 vs. Treg development, or cooperation occurs between all three transcription factors. However, further in-depth studies are required to understand how RORα may be regulating the T H 17/Treg balance.
RORα appears to regulate the expression of several genes important for T H 17 cell migration and function, specifically CCR6 26 . Decreased expression of CCR6 may underlie the decreased number of cells observed in inflamed sites, including the colon and CNS. While CCR6 is classically considered a RORγt target gene 12 , its expression was significantly decreased in vitro and in vivo in RORα-deficient or SR3335-treated T cells. While this may be a consequence of decreased RORγt expression, we do not believe this to be the case because in vitro, protein expression of RORγt present between wild-type and RORαdeficient T H 17 cells was similar (Fig. 1e). These data would suggest that CCR6 is either a shared target gene with RORγt or, relies on RORα for expression. While Ccr6 did not appear to be a direct RORα target gene in our analysis, again this could be due to the tissue type analyzed. Interestingly, expression of Ccr6 is altered in RORα −/− skin Tregs 53 , so we believe the latter is the case. The situation with IL-17A is a little more complicated. It too is decreased in vivo and one could argue that its decreased expression is a function of decreased RORγt expression. However, Fig. 6 SR3335 suppresses the development and severity of chronic EAE. a Clinical EAE scores (left) from mice subjected to MOG-induced EAE and treated with Vehicle (10/10/80 formulation of DMSO/Tween80/H 2 O) or SR3335 (i.p., 30 mg/kg, b.i.d.) for the duration of the experiment. Middle graph demonstrates the percent incidence of disease. Right graph demonstrates the percent change in body weight over time between the two groups. (Vehicle, n = 10; SR3335, n = 10) Data representative of three separate, independent experiments with similar results. b Frequencies and cell number of CD3 + CD4 + cells in the draining lymph nodes (LN) and CNS of mice at peak of disease. c Representative FACS plots and graphs (right) summarizing the frequency and cell number of CCR6 + and T cells in the LNs and CNS of vehicle vs. drug treated mice at the peak of disease. d FACS analysis, frequencies, and cell counts of RORγt + cells in LNs and CNS of mice at peak of disease. e Graphs depicting the frequencies and cell counts of CD4 + GM-CSF + cells in LN and CNS of mice at peak of disease. f FACS analysis, frequencies, and cell counts of IL-17A + IFNγ − , IL-17A + IFNγ + , and IL-17A − IFNγ + cells in the LN and CNS of mice at the peak of disease. Cells were gated on live, CD45 + CD3 + CD4 + CD44 + cells. [V = Vehicle (n = 12), SR = SR3335-treated mice (n = 11) for panels b-f]. Data are presented as mean values ± s.e.m. Data are representative of two combined independent experiments with similar results. Mann-Whitney U test for a, Student's two-tailed t tests were performed for statistical analysis, b-f. Source data are provided as a Source Data file.  IL-17A is also decreased in vitro despite equal amounts of RORγt. The answer could lie with the fact that RORγt is a member of the nuclear receptor superfamily and thus a ligand-regulated transcription factor. There have been several reports describing natural, endogenous ligands for RORγt expressed in T H 17 cells [54][55][56] . Therefore, expression of a receptor does not always correlate with activity. It is possible that RORα-deficiency leads to dysregulation of pathways that invariably affect the production of endogenous RORγ-ligands. This would manifest as downstream changes in gene expression rather than expression of the receptor itself. Consequently, RORα could affect T H 17 cell development through mechanisms regulating cellular metabolic processes.
While initially characterized in a diet induced obese (DIO) mouse model of type 2 diabetes 36 , our efforts to examine SR3335's effects in mouse models of autoimmunity and chronic inflammation have demonstrated that it effectively and largely phenocopies much of the genetic effects observed in the Rora fl/fl × CD4 Cre mice. We also found it interesting that SR3335 increased expression of IL-22 in both mouse and human memory T H 17 cells whereas Nanostring analysis indicated that SR3335 increased the mRNA expression of both MAF and IL22 in human memory T H 17 and T H 17.1 cells. This is confounding given that c-Maf has been demonstrated to inhibit IL-22 expression 57 . However, the change in gene expression is minor relative to protein expression via FACS. Given c-Maf's role in T cell effector function and cytokine expression has been well documented [58][59][60] , this needs to be interrogated further. This will be particularly important since production of IL-22 by T H 17 cells has been shown to be upregulated during tissue injury to promote repair 61 . Thus, the observed increase in IL-22 production coupled with decreased IL-17A and TNFα with SR3335 highlight the benefits of targeting RORα in chronic inflammatory diseases like IBD where such actions would prove beneficial to patients. Importantly, the effects on IL-22 appears to be specific to the RORα ligand as other, previously published RORγ inverse agonists did not have similar effects 14 . In contrast to beneficial effects in IBD, SR3335 inhibited TNFα in T H 17 and T H 17.1 cells; inhibition of TNF signaling, via use of non-selective TNF inhibitors, exacerbated MS 62,63 and induced new cases of demyelinating disease and neuropathies in patients treated for other inflammatory diseases 64,65 . These effects were likely due to full abrogation of TNF signaling whereas SR3335 selectively inhibited TNFα in T H 17 and T H 17.1 cells, leaving T H 1 and T H 2 cells intact. While the effects on TNFα were modest, as were those on IL-22, they need to be further explored to determine if the effects are a cause for concern with potential MS therapies, or the mechanism can be further exploited with more potent RORα inverse agonists for IBD.
While our studies have focused on the positive benefits of inhibiting RORα in T H 17 cells under chronic inflammatory conditions, tissue-resident "homeostatic" T H 17 cells play key, non-pathogenic roles in protection of the mucosal barrier 4 . Furthermore, evidence suggests these homeostatic T H 17 cells may contribute to the development of chronic inflammatory diseases 4 . Therefore, it is possible that T-cell-specific deletion of RORα, and/or inhibition of RORα activity may render the host more susceptible to dissemination of enteric bacteria or infection by pathogenic bacteria, including Citrobacter rodentium. However, studies demonstrating transient inhibition or deletion of RORγ activity suggest the host would only be mildly susceptible to C. rodentium infection as ILC3s remained intact in mouse hosts 66 . Conversely, a recent report suggests that RORα is required for the maintenance of ILC2s and ILC3s 67 . Given that conditional T-cell deletion of RORα would render ILCs intact and use of a RORα small molecule inhibitor would not completely ablate RORα activity in vivo, it is possible that RORα function would still be sufficient in ILC3s to control C. rodentium infection. However, this would have to be tested. Thus, we believe that in the absence of, or inhibition of RORα in T H 17 cells, the host would be mildly susceptible to bacterial infection at mucosal barriers. Furthermore, a recent report demonstrated that the induction of inflammatory T H 17 cell responses in the periphery are not influenced by "homeostatic" Th17 cells 68 . This data suggests that any RORα-mediated effects on homeostatic T H 17 cells would likely have no effect on susceptibility to chronic inflammatory conditions, including EAE. One caveat to these experiments, however, were that they were focused on inflammatory cells generated by C. rodentium, not EAE. Therefore, EAE experiments will need to be performed to definitively prove this point.
In summary, we have demonstrated a pathogenic role for RORα in T H 17 cell development and chronic inflammation. Our work indicates that targeting this receptor for the treatment of T H 17-mediated autoimmunity may be an alternative strategy that warrants further investigation.

Methods
Mice. C57BL/6, SJL/J, B6.129S7-Rag1 tm1Mom /J, stock # 003145 (Rag1 −/− ), B6(Cg)-Rorc tm3Litt /J, stock # 008771 (Rorc fl/fl ), and Tg(Cd4-Cre)1Cwi/BfluJ, stock# 017336 (CD4-Cre) mice, and B6.SJL-Ptprc a Pepc b /BoyJ, stock # 002014 (CD45.1) were purchased from the Jackson Laboratory. The Rora fl/fl mice were generated on a C57BL/6 background using a previously published targeting strategy: loxp sites were inserted flanking exon 3 of the Rora gene, which causes a premature stop codon early in exon 4 69,70 . Rora fl/fl × CD4-Cre and Rorc fl/fl × CD4-Cre mice were generated by cross breeding Rora fl/fl or Rorc fl/fl with CD4-Cre mice to produce littermates homozygous for loxp allele and hemizygous for Cre. All groups of experimental mice were matched for age and sex and randomly allocated to treatments within cage groups to eliminate litter effects. All experiments were conducted at controlled temperature (22-23°C), humidity~60%, and 12 h:12 h light:dark cycles. Mice had access to regular chow (Harlan 2920X) and water, ad libitum. For all in vitro experiments, mice were killed between 8 and 10 a.m. For all in vivo experiments, mice were killed between 7 and 11 a.m. For EAE experiments, mice were immunized between 11 a.m. and 1 p.m. All mice were maintained under specific pathogen free conditions and the Scripps Florida Institutional Animal Care and Use Committee (IACUC) approved all experiments.
Human samples. Human blood samples were conducted in accordance with IRB protocols approved by the Institutional Review Board at the Scripps Research Institute (TSRI), OneBlood (Orlando, Florida), and the University of Miami. Blood was obtained following informed written consent from healthy adults or patients with ulcerative colitis. Consenting patients provided clinical history and demographic data at the time of phlebotomy. Institutional review boards at OneBlood and the University of Miami approved all procedures. All forms and documentation that were used obtaining informed consent is stored at OneBlood or the University of Miami. PMBCs were isolated following ficoll density centrifugation and frozen in liquid nitrogen until further analysis. Cryopreserved PBMCs were stored in de-identified and barcoded vials.
Isolation of human PBMCs and T cells. Human blood samples were collected from OneBlood (Orlando, FL) and processed in accordance with protocols approved by Institutional Review Boards at Scripps Florida. Peripheral blood mononuclear cells (PBMCs) were isolated from healthy donor buffy coats by Ficoll-Paque (Sigma) density gradient centrifugation. Naïve CD4 + T cells or effector memory CD4 + T cells were separated using the EasySep Human Naïve CD4 + T cell isolation kit II or EasySep Human Memory CD4 + T cell Enrichment Kit, respectively (STEMCELL Technologies, Canada) according to the manufacturer's instructions. Cell purity of >95% was obtained in the majority of donors for the experiments. For UC/healthy donor samples, frozen PBMCs were thawed and incubated overnight in fresh media. Cell viability was assessed the following day following by isolation of naïve CD4 + T cells following methods outlined above.
Human in vitro CD4 + T cell differentiation. Human T H 17 cell differentiations were performed as follows: naïve CD4 + T cells were isolated from buffy coats and cultured in 96-well plates at a density of 2 × 10 5 per well. Cells were activated with 5 μg/mL plate-bound human anti-CD3 (clone OKT3; Bio X Cell) in complete RPMI 43 . For T H 17 differentiation, recombinant human IL-23 (50 ng/mL) and IL-1β (50 ng/mL) were added at the start of culture. Sorted effector memory T cell populations were cultured as follows: cells were cultured in DMEM supplemented with 10% FBS, 100 IU/mL penicillin, 100 μg/ml streptomycin, 50 μM β-mercaptoethanol, and 2 mM L-glutamine 10 mM HEPES, 1% non-essential amino acids, and 1%Na pyruvate. Purified T cells were activated with anti-CD3 and anti-CD28coated beads (Dynabeads; Invitrogen) for 3 days. After 48 h, IL-2 (30 U/mL) was added to the culture. On day 3, anti-CD3/anti-CD28 coated beads were removed using Dynabeads magnets (Invitrogen), and cells were maintained with IL-2 for an additional 3 days 47 . DMSO or SR3335 was added at the start of the cultures.
Flow cytometry. Surface staining: single cell suspensions prepared from spleen, lymph nodes, CNS, etc. were washed and stained with fluorescence-conjugated antibodies for 20 minutes, washed, then resuspended in FACS buffer (0.5% BSA, 2 mM EDTA in PBS). Intracellular cytokine staining: cells were restimulated with 50 ng/mL PMA and 1 μg/ml ionomycin for 2 h with the addition of GolgiStop (BD Bioscience) for an additional 2 h. Cells were then surface stained using procedures outlined above, fixed and permeabilized using the Foxp3 staining kit (eBioscience). Flow cytometric analysis was performed on a BD LSRII (BD Biosciences) instrument and analyzed using FlowJo software (TreeStar). All antibodies used in experiments are described in Supplementary Table 1. FACS sorting of human and mouse effector memory T cells. All FACS sorting was performed using a FACS Fusion (BD). For sorting of human memory T cell subsets from healthy adult PBMCs, enriched memory CD4 + T cells (STEMCELL Tech) were stained using the following antibodies: anti-CD4 (BV605); anti-CD25 (PE-CF594); anti-CD45RO (Alexa700); anti-CCR6 (APC); anti-CCR4 (PE-Cy7); anti-CXCR3 (PerCP-Cy5.5); anti-CRTH2, anti-CCR10 (PE; dump); viability (BV510). Stained cells were resuspended in FACS buffer, filtered through a 40-μm filter, and sorted. Sorting of mouse CCR6 + cells was performed as follows: T cells were isolated from the spleens and lymph nodes and enriched using CD4 + negative isolation kits from STEMCELL Technologies according to the manufacturer's instructions. Memory (CD25 − CD62L lo CD44 hi ) were further sorted for CCR6 expression by FACS (FACSAria II) 71 . For more information on antibodies, please see Supplementary Table 1.

Retroviral transduction.
To generate the RORα retroviral vector, RORα was inserted into a modified MIGR1 vector (Addgene) with additional multiple cloning sites inserted, using the BamHI site and EcoRI site. MIGR1 empty vector was used as a control. MIGR1 RORγt retroviral construct was a gift from Dan Littman (Addgene Plasmid #24069). Virus production: Plat-E cells (Cell Biolabs, Inc.) were cultured in DMEM containing 10% fetal bovine serum, 2mM L-glutamine, and 1% penicillin/streptomycin at 37°C under standard culture conditions. Plat-E cells were seeded at 350,000 cells/ml in a 6-well plate the day before transfection. In all, 3 μg total retroviral plasmid DNA (1.5 μg MIGR1 plus 1.5 μg pCL-Eco) was transfected using Fugene6 reagent (Promega) according to manufacturer's protocol. Viral supernatant was harvested 48 h post-transfection and used immediately for transduction. For retroviral transduction, naïve CD4 + T cells were stimulated as indicated with anti-CD3 and anti-CD28 and cultured under T H 17 conditions. At 24 h post TCR priming, the culture medium was replaced with virus supplemented with 8 μg/ml polybrene. Plates were centrifuged at 1800 rpm for 75 min at 37°C and then incubated at 37°C for 3-4 h. After this time, the medium was replaced with the original media removed before addition of virus.
Induction and clinical evaluation of EAE. Chronic (MOG) EAE was induced in 10-week-old, female Rora fl/fl or Rora fl/fl x Cd4-Cre mice, or WT mice (drug treatments) using EAE induction kits (EK-2110; Hooke Laboratories, Lawrence, MA, USA) according to manufacturer's instructions. Briefly, mice were injected subcutaneously over two sites in the flank with 200 μg/mouse of mouse MOG  peptide in an emulsion of Complete Freund's Adjuvant (CFA), containing killed M. tuberculosis, strain H37Ra. Concentration of killed M.tuberculosis was adjusted by lot depending on EAE induction and ranged from~2-5 mg/mL emulsion. Pertussis toxin was dissolved in PBS and injected i.p. at a range of 200-400 ng/mouse 4 h post-immunization (day 0) and 24 h later. The concentration of pertussis toxin used varied by lot and was determined by dose response studies. For relapsingremitting EAE (R-EAE), 10-week-old, female SJL/J mice were immunized using EAE induction kits (EK-2120; Hooke Laboratories, Lawrence, MA, USA) according to manufacturer's instructions. Similar induction protocols were used for R-EAE with the exception that 100 μg/mouse of mouse PLP 139-151 peptide in an emulsion of Complete Freund's Adjuvant (CFA), containing 1 mg/mL of killed M.tuberculosis, strain H37Ra, was used. Furthermore, 1 injection of 400 ng/mouse of pertussis toxin following immunization (day 0) occurred. As described above, all concentrations were adjusted by lot for consistent EAE induction. Clinical scoring started 7 days post-immunization after which mice were scored daily according to the following criteria: 0, no clinical disease; 1, limp/flaccid tail; 2, limp tail and hind leg weakness; 3, limp tail and complete paralysis of hind limbs; 4, limp tail, complete hind limb and partial front limb paralysis; and 5, quadriplegia or premoribund state. Gradations of 0.5 were used when mice exhibited signs that fell between two scores. All scoring was blinded to genotype or treatment and previous scores for each mouse. For drug treatment experiments, SR3335 was dissolved in a 10% DMSO, 10% Tween 80, and 80% H 2 O solution equaling 3 mg/mL and administered intraperitoneally (i.p.) at 30 mg/kg as was vehicle control (10% DMSO, 10% Tween 80, and 80% H2O) twice per day (b.i.d). For chronic MOGinduced EAE, the treatment was started the evening of the immunization and continued for the duration of the experiment. Disease was monitored for 28 days or as indicated in the figure legend. For R-EAE, the treatment was started the after the animals recovered from the first wave of disease and continued for the duration of the experiment. For all b.i.d. dosing, animals were dosed at 7 a.m. and 7 p.m. (lights on/lights off). At the end of the experiment, mice were anaesthetized and transcardially perfused with PBS, and spleen, peripheral lymph nodes together with spinal cords were removed for single cell isolation for FACS analysis.
T-cell transfer model of colitis. Spleens were collected from 8-10-week-old, female Rora fl/fl or Rora fl/fl × Cd4-Cre mice to sort naïve CD4 + T cells. In all, 5 × 10 5 CD4 + CD25 − CD62L hi CD44 lo cells suspended in PBS were adoptively transferred i.p. (100 μl/mouse) into 8-week-old, female Rag1 −/− recipient mice 72 . For congenic transfers, similar protocols were used with the exception that naïve female CD45.1 CD4 + T cells were sorted in lieu of Rora fl/fl T cells. Post sort, cells were counted, adjusted for cell number, mixed at a 50/50 ratio, and analyzed by flow cytometry to determine equal frequency of the two experimental groups. Once verified, the mixture was injected into female Rag1 −/− recipients. Mice were monitored biweekly for body weight change. Beddings were transferred between cages to minimize the impact of microflora on disease development. Mice were killed to assess histological inflammation due to ethical requirements if they reached 80% or less of their original body weight. For the therapeutic evaluation of SR3335 in the T-cell transfer model of colitis, spleens were collected from 8-10-week-old female C57BL/6 mice to sort naïve CD4 + T cells as described above. In total, 5 × 10 5 sorted naïve CD4 + T cells were adoptively transferred i.p. (100 μl/mouse) into female Rag1 −/− recipient mice as described above. Three weeks post-transfer, SR3335 was dissolved in a 10% DMSO, 10% Tween 80, and 80% H 2 O solution at 3 mg/mL and administered i.p. at 30 mg/kg as was vehicle control (10% DMSO, 10% Tween 80, and 80% H2O) twice per day (b.i.d). Mice were treated for 2 weeks. After the two-week period, mice were killed to collect spleens, mesenteric lymph nodes, and intestines for FACS analysis.
Mouse intestine tissue end-point collection. At the termination of the 12-week experiments, the whole colon and ileum (distal 1/3 part of the small intestine) were removed from the mouse. Each segment of the intestine was opened longitudinally, and the fecal contents were gently removed. Colon length and weight were measured and then the colon was dissected in equal halves, designated as 'proximal' and 'distal' colon and snap frozen in dry ice for protein and RNA analysis.
Histological assessment of intestinal inflammation. Following overnight fixation in 10% buffered formalin, proximal and distal colon samples were paraffinembedded, cut in 5 mm sections, and stained with hematoxylin and eosin (H&E). Inflammation was scored in a blinded fashion using a modified scoring system described in Wang et al. 72 . Briefly, each sample was graded on a scale of 0-3 to assess the following four criteria: assessments of markers of severe inflammation (crypt abscesses, ulcers), tissue damage, degree of epithelial hyperplasia and goblet NATURE COMMUNICATIONS | https://doi.org/10.1038/s41467-020-20385-9 ARTICLE cell depletion, and inflammatory cell infiltration. Scores for each criteria were added to give an overall inflammation score of 0-12.
Lamina propria mononuclear cell Isolation from mouse intestine. At the termination of the 4-6-week experiments, to collect mononuclear cells from the lamina propria of intestine, mouse ileum or colon were dissected, opened longitudinally, rinsed three times in cold PBS, and then DMEM phenol-free medium with 200 U/mL penicillin and 200 mg/mL streptomycin one time to remove the fecal content. Washed intestines were incubated in DMEM with 1 mM DTT at room temperature (RT) on a rocker for 30 min then in DMEM with 0.5 M EDTA for 30 min at RT to remove the mucus layer. After washing twice in DMEM, intestines were then digested in DMEM with Liberase and DNase I at 37°C in a shaking water bath for 30 min. The cells were then dislodged from the lamina propria and submucosa by vigorous shaking for 30 s. The supernatant containing leukocytes was pooled together by centrifugation at 500 g for 5 min. Cells were further cleaned up by 30%/70% Percoll (Sigma) gradient separation before further study.
Analysis of thymocytes. For the measurement of double positive (DP) thymocyte survival kinetics, 7-9-week-old male C57BL/6 mice were treated with compound or vehicle control for 3 days. All compounds were formulated in 15% cremophor at the following concentrations: SR2211 -2 mg/mL; SR3335 -3 mg/mL; and administered i.p., b.i.d. Thymii were collected at the termination of the experiment, minced and passed through 70 μm mesh cell strainers. Cell counts were performed prior to cell staining for flow cytometry analysis.
Quantitative reverse-transcription polymerase chain reaction. Tissue samples (snap frozen and homogenized) were extracted into TRIzol before RNA purification using RNeasy columns (Zymo Research, CA), followed by cDNA synthesis using iScript (BioRad, CA) containing oligo (dT) and random hexamer primers. qPCR SYBR Green (Roche) was used for quantitative polymerase chain reaction using a HT7900 machine (Life Technologies, CA); passive reference dye ROX was used. Primer efficiencies were determined using complementary DNA and primer dilutions for each gene of interest. Primers used to determine gene expressions can be found in Supplementary Table 2. All gene expression data were normalized to the housekeeping gene β-actin or 18 s.
Nanostring. Gene expression was quantified using a custom nCounter probe set containing 55 probe pairs with additional probes for positive and negative controls as well as housekeeping genes. (Nanostring Technologies). Briefly, 30,000 T cells were lysed in 5ul of freshly made buffer RLT (Qiagen) containing β-ME. Biotinconjugated capture probes and fluorescent-barcoded reporter probes were hybridized to cell lysates overnight at 65°C in a thermocycler. The following day, posthybridization processing occurred on all lysates which were then run on an nCounter SPRINT (Nanostring Technologies) according to the manufacturer's instructions. Raw data were normalized and analyzed using nSolver software (Nanostring Technologies).
RNA−sequencing and data analysis. mRNA was extracted from T H 17 cells on Day 2 (WT vs. RORα KO cells) or Day 3 (MIGR1 transduced cells) of in vitro differentiation. Total RNA was extracted using Qiagen RNeasy kits, quantified using the Qubit 2.0 Fluorometer (Invitrogen, Carlsbad, CA), and run on the Agilent 2100 Bioanalyzer (Agilent Technologies, Santa Clara, CA) for quality assessment. DNasetreated total RNA (300 ng) was depleted of ribosomal RNA (rRNA) using appropriate probes provided by Illumina (TruSeq Total RNA-seq kit) and further assessed on the bioanalyzer to confirm 18S and 28S rRNA peaks are depleted. rRNA-depleted RNA is processed using the TruSeq Stranded Total RNA sample prep kit (Illumina, San Diego, CA). Briefly, RNA samples are chemically fragmented in a buffer containing divalent cations and heating at 94°C for 8 min. The fragmented RNA is random hexamer primed and reverse transcribed to generate first strand cDNA. The second strand is synthesized after removing the RNA template and incorporating dUTP in place of dTTP. cDNA is then end repaired and adenylated at their 3′ ends. A corresponding 'T' nucleotide on the adapters is utilized for ligating the adapter sequences to the cDNA. The adapter-ligated DNA is purified using magnetic Ampure XP beads and PCR amplified using 12-13 cycles to generate the final libraries. The final libraries are size selected and purified using 1.0 × Ampure XP beads to remove any primer dimers. The final library size is typically 200-600 bp with insert sizes ranging from 80 to 450 bp. Final libraries are validated using bioanalyzer DNA chips and qPCR quantified using primers recognizing the Illumina adapters. Libraries are pooled at equimolar ratios, quantified using qPCR (quantification of only the adapter-ligated libraries) and loaded onto the NextSeq 500 flow cell at 1.8 pM final concentration for pair end 75 bp reads. In total, 20-25 million mappable reads per sample were collected. Demultiplexed and quality filtered raw reads (fastq) generated from the NextSeq 500 were trimmed (adapter sequences) using Flexbar 2.4 and aligned to the reference genome using TopHat version 2.0.9 73 . HT seq-count version 0.6.1 was used to generate gene counts and differential gene expression analysis was performed using Deseq2 74 . The normalized gene counts were used to plot the heatmaps using the heatmap function in Prism. To determine enriched functional groups in the RNA-seq data, KEGG pathway analysis was performed using DAVID 75,76 .
Luciferase reporter assays. HEK293 cells were plated 24 h prior to transfection in 96-well plates at a density of 15 × 10 3 cells/well. Transfections were performed using Lipofectamine 3000 (Invitrogen) according to manufacturer's protocol. For drug treatments -16 h post-transfection, cells were treated with vehicle or compound. Twenty-four hours post-treatment luciferase activity was measured using BriteLite (PerkinElmer Life and Analytical Sciences) and read using an Envision multilabel plate reader (PerkinElmer Life and Analytical Sciences). All values were normalized to DMSO to produce fold induction values.
Statistical analyses. Samples sizes, statistical analysis, and P values for every experiment are stated in the figures. Raw data from independent experiments can be found in the Source data. Statistical analyses were performed using GraphPad PRISM version 6 and 8 (GraphPad Software, Inc., La Jolla, CA). Student's twotailed t-tests were used for comparison between two groups. To compare differences between groups (i.e. EAE or transfer colitis), unpaired two-tailed nonparametric Mann-Whitney U tests with a post hoc test were performed as indicated in the Figure legends. p < 0.05 was considered significant.
Reporting summary. Further information on research design is available in the Nature Research Reporting Summary linked to this article.

Data availability
Sequence data have been deposited in the Gene Expression Omnibus (GEO) under accession code GSE160327. All other data are in the article and Supplementary Information files or from the corresponding author upon reasonable request. Source data are provided with this paper.