A20's linear ubiquitin-binding part limits harmful activation of inflammation-driving Th17 cells and gut inflammation caused by IL-22

To understand the biochemical functions of A20 that regulate intestinal homeostasis, we analyzed intestines from a series of A20 knock-in mice that abrogate A20's DUB (), E3 Ub ligase (), or M1-Ub binding () activities (,). Macroscopically, small intestines from 12-week-oldmice, but not other genotypes, demonstrated visible thickening of the intestinal wall. Histology suggested that the proximal small intestinal mucosa ofmice harbored increased numbers of immune cells in comparison with congenic,, or wild-type (WT) mice (). The expansion of the small intestinal lamina propria (SILP) by lymphocytes is reminiscent of chronic enteritis; however, other features of chronicity (e.g., epithelial metaplasia, villus blunting) were absent. This phenotype was evident in both male and femalemice and was 100% penetrant by 12 weeks of age. This phenotype was not evident in more distal portions of the small intestine and large intestines from 12-week-oldmice. In particular, ileal and colonic tissues frommice expressed normal histology without molecular markers of inflammation (, A and B; supplemental material available online with this article;). The selective enhanced inflammation of proximal small intestines ofmice differs from other spontaneous models of intestinal inflammation such as,, andmice. Hence, A20's M1-Ub binding activity via its zinc finger 7 (ZF7) domain preserves small intestinal immune homeostasis independently of A20's DUB and E3 Ub ligase activities. A20 OTU A20 ZF4 A20 ZF7 A20 ZF7/ZF7 A20 ZF7/ZF7 A20 ZF4/ZF4 A20 OTU/OTU A20 ZF7/ZF7 A20 ZF7/ZF7 A20 ZF7/ZF7 A20 ZF7/ZF7 Il2 –/– Il10 –/– Tnf ΔARE ^{20 21} Figure 1, A and B Supplemental Figure 1 https://doi.org/10.1172/JCI187499DS1↗

A20 is expressed and induced in many cell types, including both hematopoietic and non-hematopoietic cells (). To determine the degree to which hematopoietic cells ofmice (hereafter designatedmice) are sufficient to drive this pathology, we generated radiation chimeras using eitheror WT bone marrow cells to reconstitute irradiated WT mice Chimeras bearinghematopoietic cells spontaneously developed enteritis by 12 weeks after reconstitution, a phenotype that was not seen in chimeras containing WT hematopoietic cells (). Furthermore, thehematopoietic cells drove exaggerated expression of proinflammatory myeloid cytokines such asand chemokines such as(). To assess the relative contributions of T and B cells to enteritis inmice, we interbred these mice withandmice. Twelve-week-oldmice exhibited negligible intestinal inflammation and expressed levels ofandsimilar to those inmice (), suggesting that adaptive lymphocytes are required for enteritis inmice. By contrast, intestines frommice accumulated similar numbers of lamina propria immune cells () and expressed similarly elevated levels of inflammatory markers in comparison withmice (). Hence,-dependent T lymphocytes but not B lymphocytes are required for small intestinal inflammation inmice. ²⁶ Figure 1, C and D Figure 1E Figure 1, F, H, and I Figure 1, G and H Figure 1I A20 ZF7/ZF7 A20 ZF7 A20 ZF7 A20 ZF7 A20 ZF7 Il1b Ccl20 A20 ZF7 Rag1 –/– Ighm –/– A20 ZF7 Rag1 –/– Il1b Ccl20 Rag1 –/– A20 ZF7 A20 ZF7 Ighm –/– A20 ZF7 Rag1 A20 ZF7

To better understand why small intestinal inflammation develops inmice, we analyzed the transcriptomes of intact small intestines from 12-week-old WT andmice by bulk RNA sequencing (RNA-Seq). These studies revealed thatmice upregulated genes involved in NF-κB and inflammasome signaling pathways (). In addition,intestines expressed elevated levels of genes associated with CD4T cell activation, T cell proliferation, and IL-1β production (). IL-17 response genes were among the most enriched groups of genes, which — together with STAT and IL-6 regulation — suggests a prominent type 3 cytokine tone inintestines. Indeed, quantitative mRNA analyses of intact intestines confirmed the exaggerated expression of type 3 cytokines,and(). Meanwhile,levels were decreased inintestines, an expected downregulation in response to elevated IL-17 (). Surprisingly, expression of proinflammatoryandwas diminished andwas not significantly elevated inintestines (). IL-22–dependent transcripts, such as,,, and, were significantly upregulated inintestines (and Figure 5E), while expression ofwas not significantly altered (). As these data suggest that Th17 cells and/or group 3 innate lymphoid cells (ILC3s) might be expanded or hyperfunctional inintestines, we profiled cellular infiltrates from small intestinal tissues. Immunohistochemistry of intact small intestines indicated that CD4T cells were expanded in the lamina propria ofmice (), and flow cytometry of dissociated SILP cells confirmed a relative expansion of CD4T cells (). The increased number of CD4T cells inSILP largely comprised an expansion of Th17 cells in comparison with WT littermates (). In addition, among expandedSILP CD4T cells, IL-17A– and IL-17F–expressing cells were disproportionally increased while IFN-γ–expressing cells were present in similar proportions to those of WT mice. A20 ZF7 A20 ZF7 A20 ZF7 A20 ZF7 A20 ZF7 Il17a Il22 Il17ra A20 ZF7 Il23a Il18 Il23r A20 ZF7 Reg3b Reg3g Saa1 Saa3 A20 ZF7 Il22ra2 A20 ZF7 A20 ZF7 A20 ZF7 A20 ZF7 Figure 2A Figure 2A Figure 2B ²⁷ Figure 2B Figure 2C Figure 2B Figure 2D Figure 2E Figure 2E + + + + +

To better define the SILP lymphocytes inmice, we enriched SILP T cells from 12-week-old WT andmice and profiled these cells by single-cell RNA-Seq (scRNA-Seq). To avoid clustering cells based on cell cycle phases, genes related to cell cycling were removed before clustering analyses. Uniform manifold approximation and projection (UMAP) analyses of these cell cycle–regressed cells identified CD8, Th17, and regulatory T cells, as well as subsets of innate lymphoid cells (ILCs) (and). Genotype-specific UMAP analyses showed a relative expansion of CD4and, to a lesser extent, CD8T cells inmice (). A20 ZF7 A20 ZF7 A20 ZF7 + + + Figure 3A Supplemental Figure 2A Figure 3, B and C

Expanded clusters of CD4T cells inintestines contained many cells expressingand, delineating these cells as Th17 cells (and). Another expanded cluster included regulatory T cells (). By contrast, ILC3s were relatively reduced inmice (). In addition to being more abundant inmice,Th17 cells expressed higher levels ofandthan WT Th17 cells (). The marked expansion of Th17 cells, the increased expression of type 3 cytokines, and the absence of ILC3 expansion inintestines relative to WT intestines suggest that Th17 cells account for the great majority of the increased type 3 cytokine production inintestines. + A20 ZF7 Il17a Il22 A20 ZF7 A20 ZF7 A20 ZF7 Il17a Il22 A20 ZF7 A20 ZF7 Figure 3D Supplemental Figure 2A Figure 3C Figure 3C Figure 3E

The Th17 cells segregated into 2 discrete clusters in UMAP space, and both clusters were heavily represented inintestines (, and). After the contributions of cell cycle genes had been removed as an effect on cell clustering, these two clusters continued to exhibit differential cell cycle states: cells with high G/Gscores (indicating a predominantly non-proliferative state) and cells enriched for high G/M or S scores (suggesting T cell proliferation) (). In addition, this "Th17 proliferative" cluster expressed high levels of, the gene that encodes the proliferative marker Ki67 (). The accumulation of proliferative Th17 cells inSILP aligns with our bulk RNA-Seq analysis, which highlighted T cell proliferation as an enriched category inintestines (). Although the proliferative cluster was transcriptionally distinct from non-proliferating T cells in eitheror WT mice, it consisted of cells that expressed detectable levels ofand/or, confirming they were Th17 cells (and). Hence, increased numbers of proliferating Th17 cells help explain why Th17 cells are more numerous inintestines. A20 ZF7 Mki67 A20 ZF7 A20 ZF7 A20 ZF7 Il17a Il22 A20 ZF7 Figure 3, A and B Supplemental Figure 2A Supplemental Figure 2B Supplemental Figure 2A Figure 2A Figure 3D Supplemental Figure 2A 1 2

Asmouse small intestines contain dramatic expansions of SILP Th17 cells and increased tissue-wide expression of IL-17A and IL-22, we next sought to functionally define the potential roles of these cytokines in regulating intestinal disease inmice. We interbredmice withmice andmice. Small intestines frommice exhibited more (rather than less) severe enteritis than those from IL-17A–competentmice (). Hence, IL-17A plays a protective rather than proinflammatory role in enteritis inmice.intestines expressed morethanintestines but notintestines (). In contrast tomice,mice exhibited less severe enteritis thanmice (). Therefore, IL-22 promotes small intestinal inflammation inmice. To better define the role of IL-22 inintestines, we profiled the genome-wide transcriptomes of small intestines fromand control mice by bulk RNA-Seq. Principal component analysis of bulk RNA-Seq data revealed broad normalization toward wild-type transcriptomic states inmice when compared withmice (). Thus, IL-22 drives transcriptome-wide changes in the proximal small intestine that promote intestinal inflammation inmice. A20 ZF7 A20 ZF7 A20 ZF7 Il17a –/– Il22 –/– A20 ZF7 Il17a –/– A20 ZF7 A20 ZF7 A20 ZF7 Il17a –/– Il22 Il17a –/– A20 ZF7 A20 ZF7 Il17a –/– A20 ZF7 Il22 –/– A20 ZF7 A20 ZF7 A20 ZF7 A20 ZF7 Il22 –/– A20 ZF7 Il22 –/– A20 ZF7 A20 ZF7 Figure 4, A and B Figure 4C Figure 4, A and B Figure 4D

Il22ra1, the IL-22–specific receptor chain, is selectively expressed on intestinal epithelial cells (IECs) but not immune cells (). Hence, pathophysiological effects of IL-22 inmice could be mediated by perturbation of IEC functions. By histology, epithelial crypts were disproportionally expanded in small intestines ofmice when compared with WT mice (and). Bulk RNA-Seq highlighted epithelial cell proliferation as an enriched gene set inintestines (), and immunohistochemistry for Ki67 confirmed an expansion of proliferating crypt IECs inintestines (). Interestingly, chimeras bearinghematopoietic cells also expressed elevated levels of IEC-derived defensinsand(). This result supports that radiation-sensitiveTh17 cells express exaggerated amounts of IL-22 that perturb WT IECs. Notably, the IEC proliferation and crypt elongation seen inmice were significantly lessened inmice (). To understand the IL-22–dependent perturbations of IECs inmice, we isolated small intestinal IECs fromand control mice. Transcriptomic analyses of these cells revealed thatepithelia expressed more C-type lectins (,), alarmins (), and chemokines () than WT IECs (). These defects were markedly reduced inIECs, indicating that IL-22 drives exaggerated expression of these proinflammatory mediators inIECs. ²⁸ Figure 4A Figure 5A Figure 2A Figure 5, A and B Figure 5C Figure 5, A and B Figure 5, D and E A20 ZF7 A20 ZF7 A20 ZF7 A20 ZF7 A20 ZF7 Reg3b Reg3g A20 ZF7 A20 ZF7 A20 ZF7 Il22 –/– A20 ZF7 A20 ZF7 Il22 –/– A20 ZF7 Reg3b Reg3g Saa1 Cxcl1 A20 ZF7 Il22 –/– A20 ZF7

As proinflammatory cytokines can perturb epithelial barrier functions (–), we tested intestinal epithelial permeability inmice and control mice by gavaging these mice with FITC-dextran. Increased absorption of FITC-dextran in sera ofmice supports aberrant barrier integrity in these mice. This compromised barrier integrity was similarly exacerbated inmice but normalized inmice (). To understand whyIECs fail to maintain barrier integrity, we assayed expression of occludin () and claudin-4 (), 2 tight junction proteins that support epithelial barrier integrity (,).andexpression were depressed incompared with WT IECs, andwas normalized inIECs (). AlthoughinIECs did not reach levels similar to those inIECs, these were normalized to WT levels. Diminished barrier integrity may allow translocation of microbes and pathogenic products, inciting an inflammatory response (–). To determine whether intestinal microbiota are required for enteritis inmice, these mice were derived into germ-free environments. Germ-freemice exhibited neither intestinal inflammation () nor elevated expression of proinflammatory cytokines (i.e.,,,,) relative to controls (), a stark contrast to themice raised in specific pathogen–free (SPF) environments. Despite the mutation ofin the small intestines of germ-free mice, the absence of commensal microbiota did not lead to alarmin (,) production (), suggesting that microbiota are necessary for IL-22–dependent programs within the small intestine. Segmented filamentous bacteria (SFB) preferentially colonize and induce Th17 responses in the terminal ileum but not duodenum (–). Specific PCR detection of SFB 16S rRNA in our mice infrequently showed SFB in ilea of WT andintestines () and never detected SFB in duodena from either genotype. Therefore, the duodenitis ofmice is microbe dependent but unlikely to be driven by SFB. ^{29 31} Figure 5F ^{32 33} Figure 5G ^{34 36} Figure 5H Figure 5I Figure 5J ^{37 40} Supplemental Figure 3A A20 ZF7 Il22 –/– A20 ZF7 A20 ZF7 Il17a –/– A20 ZF7 Il22 –/– A20 ZF7 Ocln Cldn4 Ocln Cldn4 A20 ZF7 Cldn4 A20 ZF7 Il22 –/– Ocln A20 ZF7 Il22 –/– Il22 –/– A20 ZF7 A20 ZF7 Il1b Ccl20 Tnf Ifng A20 ZF7 A20 ZF7 Saa1 Saa3 A20 ZF7 A20 ZF7

As our SPFand WT mice were cohoused, thereby sharing luminal microbes via coprophagic behavior,intestines may have responded aberrantly to the same commensal microbes that were present in WT cagemates. To further test this notion, we harvested luminal commensal microbes from an unrelated colony of WT C57BL/6 mice, pooled these organisms, and introduced these microbes ("WT" microbes) into cohoused 6-week-oldand WT germ-free mice. These neocolonized, or "conventionalized," mice were maintained as cagemates for 6 weeks. Analyses of intestines from these mice revealed that duodenal tissues from conventionalized WT mice minimally expressedand IL-22–dependent genes compared with those from germ-free WT mice (). By contrast, conventionalizedmice expressed markedly elevated levels of these genes compared with germ-freemice or conventionalized WT mice (). Correspondingly, conventionalizedmice exhibited marked duodenal, but not ileal or colonic, inflammation when compared with conventionalized WT mice (and). Moreover, germ-free and conventionalized mice harbored no detectable SFB (), further supporting that the duodenal inflammation inmice is microbe dependent but, also, independent of SFB. Taken together, these data indicate thatintestines respond aberrantly to "WT" microbiota by expressing markedly elevated expression of IL-22 that, in turn, perturbs IEC functions, disrupts intestinal epithelial barrier integrity, and causes duodenitis. A20 ZF7 A20 ZF7 A20 ZF7 Il22 A20 ZF7 A20 ZF7 A20 ZF7 A20 ZF7 A20 ZF7 Figure 5K Figure 5K Figure 5H Supplemental Figure 1D Supplemental Figure 3B

Our data above show that T cell–autonomousfunctions and IL-22 are integral to enteritis inmice. Accordingly, we investigated howregulates T cell production of IL-22. We enriched naive CD4T cells from 8-week-oldand WT mice and differentiated these cells into Th17 cells with recombinant IL-6 and TGF-β. The efficiency of in vitro Th17 differentiation (as defined by the expression of RORγt, the key transcription factor that coordinates Th17 differentiation) () was consistently greater than 99% (), regardless of A20 mutation status. Notably,Th17 cells exhibited greater amounts of RORγt localized to the nucleus than WT Th17 cells, suggesting thatrestrains RORγt expression and/or nuclear localization in a cell-autonomous fashion ().Th17 cells also expressed more phosphorylated Stat3, reflecting substantial activation of this pathway (). In the absence of PMA/ionomycin stimulation,Th17 cells also expressed moremRNA than WT cells (). Sinceexpression is dependent on the aryl hydrocarbon receptor (Ahr) (), these cells were treated with the Ahr agonist FICZ (,). FICZ further exaggerated enhancedexpression inTh17 cells relative to WT cells (). Similarly, ELISA of supernatants from these cells revealed thatT cells secreted more IL-22 protein than WT cells (). Notably, these results were obtained from cells that were not stimulated with PMA or ionomycin, avoiding potential caveats associated with possible A20-dependent regulation of these stimuli. Aligned with these findings, flow cytometry studies of PMA/ionomycin-stimulated cells showed increased numbers of IL-22T cells incultures when compared with WT cultures (). Hence, mutation ofwithin T cells leads to greater RORγt expression and increased expression of IL-22. A20 ZF7 A20 ZF7 A20 ZF7 A20 ZF7 A20 ZF7 A20 ZF7 A20 ZF7 A20 ZF7 Il22 Il22 Il22 A20 ZF7 A20 ZF7 A20 ZF7 A20 ZF7 + + ⁴¹ Figure 6A Figure 6A Figure 6B Figure 6C ^{42 42 43} Figure 6C Figure 6D Supplemental Figure 4B

AsTh17 cells express moremRNA and protein than WT cells, we hypothesized thatrestrains epigenetic regulation oftranscription. To identify potential genomic loci that may regulatein, we surveyed chromatin accessibility via assay for transposase-accessible chromatin sequencing (ATAC-seq) followed by massively-parallel sequencing of Th17 cells differentiated from naive WT andCD4T cells. ATAC-seq revealed increased accessibility across thegene inTh17 cells (, region g), supporting open chromatin at's promoter and enhancedtranscription. In addition, multiple other sites upstream ofdemonstrated DNA accessibility, predicting potential enhancer regions (, regions a–f). Notably, ATAC-seq highlighted a conserved region 32 kb upstream of's transcription start site (, region b), a site recently characterized as anenhancer (). To define this conserved region as a bona fide functional enhancer and to quantify the activation state of this enhancer, naive WT andCD4T cells were differentiated into Th17 cells, and chromatin immunoprecipitation (ChIP) of acetylated lysine 27 of histone H3 (H3K27ac) was performed at these genomic loci. H3K27ac was increased inTh17 cells at the enhancer for(, region b), supporting increased enhancer activation and enhancedtranscription inTh17 cells. By contrast, sequences located on the immediate shoulders of the enhancer (, regions a and c) as well as other DNA accessible sites (, regions d–f) were less decorated with H3K27ac and similarly marked in WT andTh17 cells. The enhancement of H3K27ac marks at the conserved enhancer with FICZ treatment suggests that Ahr activation facilitates the acetylation of this locus. These results provide a molecular underpinning for Ahr in promotingexpression (,). As our flow studies of isolated nuclei showed thatTh17 cells harbor increased accumulation of nuclear RORγt compared with WT Th17 cells (), we next tested whether this increased RORγt mediates increased transcriptional activity at theenhancer. We differentiated naive CD4WT andcells into Th17 cells in vitro and acutely treated cells with GSK805 (a RORγt-specific inhibitor) (,) overnight beginning on day 6 of differentiation. Subsequent chromatin analyses at the enhancer showed that RORγt inhibition inTh17 cells normalized nucleosome occupancy/density () and H3K27ac () to WT levels. These findings support RORγt's role in reducing nucleosomal occupancy, facilitating chromatin and DNA accessibility, and enhancing acetylation of H3K27 at theenhancer inTh17 cells. Therefore, elevated RORγt in differentiatedTh17 cells actively remodels chromatin at theenhancer to drive increasedtranscription. Taken together, these studies suggest thatrestrains RORγt activity to limit chromatin accessibility and hyperacetylation of a specificenhancer, thus restrictingtranscription in CD4Th17 cells. A20 ZF7 Il22 A20 ZF7 Il22 Il22 cis A20 ZF7 Il22 A20 ZF7 Il22 Il22 Il22 Il22 Il22 A20 ZF7 A20 ZF7 Il22 Il22 A20 ZF7 A20 ZF7 Il22 A20 ZF7 Il22 A20 ZF7 A20 ZF7 Il22 A20 ZF7 A20 ZF7 Il22 Il22 A20 ZF7 Il22 Il22 + + + + Figure 6E Figure 6E Figure 6E ⁴⁴ Figure 6F Figure 6F Supplemental Figure 4C ^{42 43} Figure 6A ^{45 46} Figure 6G Figure 6H

Thegene is well conserved between murine and human genomes, and A20 polymorphisms and mutations are associated with human IBD and celiac disease (,,). To determine whether A20's ZF7 domain regulates Th17 cell functions and IL-22 expression in human T cells, we generatedmutant human CD4T cells using CRISPR/Cas9. Since A20's ZF7 domain comprises the most C-terminal residues of the A20 protein, targetingis unlikely to affect the stability or structure of the A20 protein. Our recent studies of murinemutant proteins suggest that these proteins are indeed expressed at supranormal levels in TNF-stimulated fibroblasts (). We thus designed CRISPR guide RNAs (gRNAs) to delete the N-terminal half ofdomain. We isolated naive CD4T cells from healthy donor peripheral blood, activated the cells via TCR stimulation, and electroporated CRISPR/Cas9 along with either these ZF7-targeting or control gRNAs to generatemutant and isogenic control human T cells. Sanger sequencing demonstrated that more than 85% of alleles were targeted by-specific gRNAs (, A and B). These cells were then differentiated toward Th17 cells for 9 days, at which time more than 99% expressed RORγt in both-mutated and control cells (), confirming efficient Th17 differentiation. Expression ofmRNA is markedly induced by NF-κB activity, and A20 mediates negative feedback on NF-κB signaling (). In keeping with the compromised ability ofcells to restrain NF-κB signaling, expression ofmRNA was increased inhuman T cells when compared with paired isogenic control T cells (). The NF-κB family members c-Rel and Rela/p65 bind RORγt promoters and drive RORγt transcription (). Consistent with increased NF-κB signaling and with our findings in murine Th17 cells, ablation ofin human Th17 cells led to higher expression of RORγt (). Finally,-deleted human Th17 cells led to elevated expression ofandtranscripts relative to paired, isogenic, A20-competent control cells (). Hence, A20's ZF7 motif restrains RORγt expression andexpression in human Th17 cells. TNFAIP3 A20 ZF7 A20 ZF7 A20 ZF7 A20's ZF7 A20 ZF7 A20 ZF7 A20 ZF7 Tnfaip3 A20 ZF7 TNFAIP3 A20 ZF7 A20 ZF7 A20 ZF7 IL17A IL22 IL22 ^{7 10 47 21} Supplemental Figure 5 Figure 6I ²⁶ Figure 6J ⁴⁸ Figure 6I Figure 6J + +

Both male and female mice were used in this study, and similar results were obtained with both. Sex was not considered a biological variable in this study.

Animal studies were conducted under an approved Institutional Animal Care and Use Committee protocol at UCSF. Mice were bred and housed at 22°C under a 12-hour light/12-hour dark cycle with ad libitum access to food and water.(OTU),(ZF4), and(ZF7) knockin mice were generated in our laboratory and previously described (,). B6.129S2 Ighm/J (μMT; 002288), B6.129S7 RAG-1/J (Rag1; 002216), and C57BL/6 Il22/J (IL22; 027524) mice were purchased from The Jackson Laboratory.mice were provided by Y. Iwakura (Tokyo University of Science, Tokyo, Japan) via S. Gaffen, University of Pittsburgh, Pittsburgh, Pennsylvania, USA). A20 C103/C103 A20 ZF4/ZF4 A20 ZF7/ZF7 Il17a –/– ^{20 21} tm1Cgn tm1Mom tm1.1(icre)Stck

Mice at the indicated ages were euthanized, and the proximal small intestine was collected in 4% paraformaldehyde. Samples were subsequently processed and stained by HistoWiz to produce H&E-, CD4-, or Ki67-stained slides. Pathology scores of intestinal inflammation were generated by assessment of total tissue inflammation and epithelial changes in the colon. The total inflammation score (scale 0–3) was based on the overall severity or extent of inflammation in the intestine. An epithelial change score (scale 0–4) was assigned for each of the following categories: goblet cell loss, intraepithelial neutrophils and/or cryptitis, abscesses, and crypt loss. The scores were summed to generate a pathology score for each mouse.

At 6 weeks of age, WT recipient mice were irradiated with 12 Gy total-body radiation. The same day, recipients were injected intravenously with 5 × 10bone marrow cells isolated from femora and tibiae of corresponding donors (WT ormice). Animals were kept on antibiotics in the drinking water for 1 week after irradiation. Bone marrow recipients were sacrificed 12 weeks after bone marrow transfer. 6 A20 ZF7

Mice (11 weeks old) were fasted for 5 hours before oral gavage of FITC-dextran (average MW 4,000; Sigma-Aldrich 46944) at 500 mg/kg body weight. Serum was collected 4 hours after gavage, and the fluorescence intensity was measured on a Molecular Devices fluorescence microplate reader.

Single lamina propria cells were isolated from small intestinal lamina propria as previously described (). Briefly, the proximal 10 cm of small intestines from euthanized mice were flushed with cold PBS to clear feces and mucus. Excess mesenteric fat and Peyer patches were removed. Intestines were opened longitudinally and incubated in pre-digestion buffer (calcium- and magnesium-free HBSS, 3% FBS, 10 mM HEPES [pH 7.4], 1 mM dithiothreitol, 5 mM EDTA) twice for 15 minutes each, rinse buffer (calcium- and magnesium-free HBSS, 3% FBS, 10 mM HEPES [pH 7.4]) once for 5 minutes, and digestion buffer (RPMI with 3% FBS, 10 mM HEPES [pH 7.4], 0.03 mg/mL DNase I, 0.1 mg/mL Liberase TM (Sigma 5401119001) for 10 minutes, all at 37°C with agitation at 220 rpm. Enzyme-digested tissue was further dissociated in a gentleMACS C-tube using the gentleMACS (Miltenyi Biotec) Dissociator m_intestine program, added to RPMI with 10 mM HEPES (pH 7.4) and 3% FBS, and filtered through 70- or 100-μm filters. Leukocytes were enriched from the interface of a 40%/80% Percoll (GE Healthcare 17-081-01) gradient after centrifugation. ⁵⁹

Single-cell suspensions were twice rinsed with PBS, stained with an amine-reactive viability dye in PBS for 15 minutes, quenched, and washed with FACS wash buffer (FWB: HBSS or PBS with 0.5% BSA). Cells were then blocked with 2 μg FcBlock (per 1 million cells) for 15 minutes and stained with antibodies against cell surface antigens for 30 minutes. For nuclear assays, nuclei were isolated as previously described () with slight modifications: cells were incubated in ice-cold isolation buffer (375 mM sucrose, 10 mM HEPES [pH 7.9], 10 mM potassium chloride, 5 mM magnesium chloride, 0.1% vol/vol Triton X-100, protease and phosphatase inhibitors) on ice for 15 minutes, and nuclei were collected by centrifugation at 1,300for 5 minutes at 4°C. For intracellular antigens, cells/nuclei were subsequently fixed in 5% neutral-buffered formalin for 30 minutes at room temperature, permeabilized with eBioscience Perm Buffer (Thermo Fisher Scientific 00-8333-56), blocked with normal rat serum (STEMCELL Technologies) for 15 minutes, and rocked with primary antibodies overnight at 4°C. Single cells/nuclei were washed twice with FWB before analysis on a flow cytometer. ⁶⁰ g

The following antibodies were used for flow cytometry: anti-CD45 (30-F11, BD Biosciences), anti-CD90.2 (30-H12, BD Biosciences), anti-TCRβ (H57-597, BioLegend), and anti-CD4 (GK1.5, BioLegend) for staining of mouse cell surface proteins; and anti–IL-17A (TC11-8H4, BioLegend), anti–IL-22 (poly1564, BioLegend), anti-RORγt (B2D, Thermo Scientific; or REA278, Miltenyi Biotec), and anti–phosphorylated Stat3 (4/P-STAT3, BD Biosciences) for staining of human or mouse intracellular proteins. For intracellular cytokine staining, cells were first incubated for 4 hours in medium with 100 ng/mL phorbol 12-myristate 13-acetate (PMA), 1 μg/mL ionomycin, and 5 μg/mL brefeldin A.

Naive CD4T cells were isolated from murine splenocytes using a Mouse CD4 naive T cell negative selection kit (STEMCELL Technologies 19765) per the manufacturer's instructions. Isolated cells were subsequently differentiated in plates precoated with 2 μg/mL anti-CD3 (overnight at 4°C, or 2 hours at 37°C) in Th17 medium (IMDM with 10% FBS, 50 μM β-mercaptoethanol, 2 μg/mL anti-CD28 [Bio X Cell BE0015, clone 37.51], 20 ng/mL recombinant murine IL-6 [R&D Systems 406-ML], 5 ng/mL recombinant human TGF-β1 [PeproTech 100-21], 10 μg/mL anti–IL-4 [Bio X Cell BE0045, clone 11B11], and 10 μg/mL anti–IFN-γ [Bio X Cell BE0055, clone XMG1.2]) at a density of 1 million cells/mL. Medium was replenished on day 2, and cells were split into fresh medium on day 3 of differentiation to maintain a cell density of 1–2 million cells/mL. For RORγt inhibition experiments, naive CD4T cells were differentiated for 6 days before incubation with 1 μM GSK805 (MedChemExpress HY-12776) in fresh Th17 medium for 16 hours prior to harvesting. + +

Total RNA was isolated from whole intestinal tissue or single-cell suspensions (epithelial fractions or lamina propria) using QIAGEN RNeasy extraction kits or TRIzol (Invitrogen) per the respective manufacturer's instructions. Whole tissues were lysed in either ice-cold TRIzol or QIAGEN RLT Buffer using Lysing Matrix D (MP Biomedical 116913050-CF) with MP Biomedical FastPrep-24 Homogenizer (1 cycle of 20 seconds at 4 m/s). Single-cell suspensions were lysed directly in TRIzol. RNA was reverse-transcribed using High Capacity cDNA Reverse Transcription kit (Thermo Fisher Scientific 4368814) per the manufacturer's instructions and assayed by quantitative real-time PCR using TaqMan probes (Thermo Scientific) or the SYBR Green (Thermo Scientific) method. Relative mRNA expression was calculated using the Livak (ΔΔCt) method, withas a loading control, unless otherwise noted. In experiments using human cells, the housekeeping genes,,, andwere collectively used, and the geometric mean was used in calculating relative mRNA expression. To detect segmented filamentous bacteria (SFB), 16S ribosomal RNA (rRNA) was amplified using primer sequences specific to SFB 16S rRNA: 5′-GATCCTGGCTCAGGACGAAC-3′ and 5′-TTCATCGGGCTATCCCCCA-3′. PCR products from ileal samples with detectable levels of SFB 16S rRNA were subjected to agarose electrophoresis, and the 146-bp products were excised, purified using QIAGEN Gel Extraction kit, and Sanger-sequenced. NCBI BLAST of Sanger sequences confirmed the PCR to specifically detect SFB 16S rRNA. 16S rRNA copies were calculated based on a standard curve generated using a double-stranded DNA gBlock template (Integrated DNA Technologies) encoding a portion of the SFB 16S rRNA sequence. The limit of detection of this assay is routinely less than 1 copy of SFB 16S rRNA. Actb ACTB GAPDH EIF4G2 RPLP0

Lamina propria cells were isolated from small intestines of 2 mice of each genotype as described above. Isolated cells were negatively selected for T cells and ILCs (STEMCELL Technologies 19851) and stained with TotalSeq-C anti-mouse hashtagged antibodies (BioLegend). Cells from each animal were separately hashtagged. For each genotype, a total of 60,000 cells were collected for processing using the 10x Genomics Chromium Next GEM Single Cell 5′ Kit v1.1. Sequences were aligned and processed with Cell Ranger v7.1 using the mm10 reference genome and default parameters. Cell Ranger output was further processed with R version 4.3.3 and Seurat version 4.3 (). Seurat objects were created using only genes that appeared in at least 3 cells. Cells were further filtered to exclude multiplets (defined as having 2 or more different hashtags) and low-quality/multiplet cells (cells with fewer than 200 detected genes, more than 2,500 detected genes, or more than 15% mitochondrial reads). Read counts were then normalized using NormalizeData. After viewing of UMAP clusters, T and innate lymphocyte clusters were selected based on the presence of T/innate lymphocyte genes (,,,,,,,) as well as the absence of non-T/non-innate lymphocyte genes (,,,,,,,,). ⁴⁹ Ptprc Trac Trdc Cd3e Cd4 Cd8a Ncr1 Klrb1b Des Acta2 Col1a2 Pecam1 Cdh5 Epcam Cd79a Mcpt1 Apoe

Cell clusters were generated using the Louvain algorithm implemented by the FindClusters function. Marker genes for each cluster were determined using Wilcoxon's test on the raw counts, implemented by the function FindAllMarkers, and clusters of cell types were additionally determined by manual inspection of the lists of cluster marker genes. Dimensionality reduction by UMAP was performed using the RunUMAP function with the 30 largest principal components. Visualization of all scRNA-Seq data was generated using the Seurat package and/or ggplot2. To remove cell cycle phases as a source of clustering heterogeneity, Seurat's CellCycleScoring function was used to score each cell and regress out the effects of cell cycle genes/phases.

Total RNA was isolated from intact proximal small intestine as described above. One microgram of total RNA was depleted of rRNA (NEBNext rRNA Depletion Kit v2, New England Biolabs E7405) and subsequently fragmented, reverse-transcribed into cDNA, and amplified into barcoded libraries with NEBNext RNA Library Prep (New England Biolabs E7765) using custom barcoded Illumina-compatible primers. Libraries were pooled and sequenced on an Illumina NovaSeq X Plus instrument as 150-bp paired-end reads.

Sequenced reads (40 million to 70 million paired reads per sample) were processed with fastp () for quality control and adapter sequence trimming. Trimmed reads were aligned to themm10 genome using HISAT2 (), and transcriptomes for each sample were assembled with StringTie2 () and UCSC's mm10 genes.gtf annotation. After construction of non-redundant transcriptomes across all samples (stringtie --merge), expression counts were tabulated (stringtie -e) and used as input for downstream gene expression analyses. DESeq2 () was used for differential gene expression analyses. ^{61 62 63 64} Mus musculus

In vitro–differentiated Th17 cells were processed for chromatin immunoprecipitation (ChIP) as previously described (). Briefly, 10 million Th17 cells were cross-linked at room temperature in 1% formaldehyde for 8 minutes, quenched with 125 mM glycine for 5 minutes, and resuspended in ChIP lysis buffer (10 mM Tris [pH 8], 1 mM EDTA, 0.5 mM EGTA, 0.5% wt/vol-lauroyl sarcosine, and protease and phosphatase inhibitors). Each 250- to 300-μL aliquot was individually sonicated in a Bioruptor Pico to generate an average DNA fragment length of 250 bp as determined by agarose electrophoresis. Debris was spun down at 20,000for 15 minutes, and the cleared chromatin supernatants were quantified and used for subsequent ChIP experiments. Sonicated chromatin was diluted to 500 μL in ChIP lysis buffer with 1% Triton X-100, 0.1% sodium deoxycholate, 1 mM EDTA, and protease and phosphatase inhibitors. Chromatin was precleared with 10 μL protein G Dynabeads (Thermo Fisher Scientific 10003D) for 4 hours at 4°C before incubation with 1 μg of primary antibodies overnight at 4°C. For anti–histone H3 (Abcam ab1791) or anti–acetyl histone H3 lysine 27 (anti-H3K27ac; Active Motif 39133) ChIP, 3 or 10 μg of total chromatin, respectively, was used. Protein G Dynabeads were incubated with immune complexes for 2 hours at 4°C before magnetic capture. Immunoprecipitates were washed 5 times with 1 mL RIPA wash buffer (50 mM HEPES [pH 7.6], 10 mM EDTA, 0.7% wt/vol sodium deoxycholate, 1% vol/vol IGEPAL CA-630 (Sigma-Aldrich), 0.5 M lithium chloride, and protease inhibitors) and once with TE (10 mM Tris with 1 mM EDTA [pH 8]) with 50 mM sodium chloride and eluted in elution buffer (10 mM Tris, 1 mM EDTA, 1% wt/vol sodium dodecyl sulfate, pH 8) at 56°C for 15 minutes. To reverse cross-links, eluates were incubated overnight at 65°C. Samples were diluted 2-fold in TE and sequentially digested with 80 μg of DNase-free RNase A at 37°C for 1 hour, and then 80 μg proteinase K at 56°C for 1 hour. ChIP DNA was purified by phenol/chloroform/isoamyl alcohol (25:24:1) and ethanol precipitation. DNA pellets were resuspended in 10 mM Tris (pH 8) and assayed by quantitative real-time PCR using the SYBR Green method. Antibodies were titrated to ensure they were not limiting. ⁶⁵ N g

In vitro–differentiated Th17 cells were used to generate Tn5-tagmented DNA fragments as previously described (). Briefly, to isolate nuclei, cells were lysed in ATAC lysis buffer (10 mM Tris-HCl [pH 7.5], 10 mM sodium chloride, 3 mM magnesium chloride, 0.1% vol/vol IGEPAL CA-630, 0.1% vol/vol Tween-20, 0.01% wt/vol digitonin) for 3 minutes on ice. The nuclei were briefly rinsed in wash buffer (10 mM Tris-HCl [pH 7.5], 10 mM sodium chloride, 3 mM magnesium chloride, 0.1% vol/vol Tween-20) before incubation in transposition buffer (10 mM Tris-HCl [pH 7.6], 5 mM magnesium chloride, 10% vol/vol dimethyl formamide, 0.1% vol/vol Tween-20, 0.01% wt/vol digitonin) with 100 nM loaded hyperactive Tn5 transposase (Diagenode C01070012) for 30 minutes at 37°C at 1,000 rpm. Transposed DNA was isolated from QIAGEN MinElute columns. Libraries were generated using published Illumina-compatible indexed oligonucleotides and 8 cycles of PCR amplification using NEBNext High-Fidelity 2X PCR Master Mix (New England Biolabs M0541). Libraries were purified from SPRIselect beads (Beckman Coulter B23317), and 51-bp paired-end sequences were sequenced on a NovaSeq X instrument. Sequences were aligned to the GRCm38/mm10 mouse reference genome using Bowtie 2 v2.4.5 (), and PCR duplicates were removed by SAMtools v1.15 () followed by Picard v2.27.1 MarkDuplicates (). BigWig files were generated using UCSC's bedGraphToBigWig script. DNA accessible regions and sites of differential accessibility were determined by MACS2 v2.2.7.1 () (). ^{66 67 68 69} https://broadinstitute.github.io/picard/↗ https://pypi.org/project/MACS2/↗

Naive CD4T cells were isolated from healthy donor PBMCs using a human CD4 Naive T Cell negative selection kit (BioLegend 480041) per the manufacturer's instructions. Isolated cells were stimulated in stimulation medium (Immunocult XF serum-free medium plus 50 μM β-mercaptoethanol, Immunocult anti-CD3/-CD28 cocktail [STEMCELL Technologies], and 50 IU/mL human recombinant IL-2 [NIH Biological Resources Branch. Preclinical Biologics Repository]) for 2–3 days. Stimulated cells were subsequently electroporated in a total of 20 μL P3 solution (Lonza) with Cas9 ribonucleoprotein complexes (3.1 μM TruCut Cas9 v2 [Thermo Fisher Scientific] with 9 μM CRISPR guides [Integrated DNA Technologies]) using the Lonza Amaxa 4D-Nucleofector and program EH-115. Target sequences recognized by CRISPR guides were GATACGTCGGTACCGGACCG for the control/non-targeting guide and TTTGGCAATGCCAAGTGCAA and ACCCCCCCAAGCAGCGTTGC forablation. Electroporated cells were immediately placed into pre-warmed stimulation medium and incubated for 3 days. Genomic editing was confirmed to be at least 80% efficient by Sanger sequencing. Cells were subsequently differentiated into Th17 cells in plates precoated with 5 μg/mL anti-CD3 (Bio X Cell BE0001-2, clone OKT3) in Th17 medium (Immunocult XF serum-free medium with 50 μM β-mercaptoethanol, 30 ng/mL recombinant human IL-6 [Proteintech HZ-1019], 2.5 ng/mL recombinant human TGF-β1 [PeproTech 100-21], 10 ng/mL recombinant human IL-1β [Proteintech HZ-1164], 10 ng/mL recombinant human IL-23 [Proteintech HZ-1254], 10 μg/mL anti–IL-4 [Bio X Cell BE0240, clone MP4-25D2], and 10 μg/mL anti–IFN-γ [Bio X Cell BE0235, clone B133.5]) at a density of 1 million cells/mL. Cells were split and medium replenished on days 3, 6, and 8 of Th17 differentiation to maintain a cell density of 1–2 million cells/mL. Cells were harvested on day 9 of differentiation for analysis. + A20 ZF7

Statistical analyses were performed using Prism 10 (GraphPad Software). Pathology scores were assessed by an unpaired Mann-Whitneytest. Quantitative real-time PCRs were assessed by a 2-tailed unpaired Student'stest with Welch's correction, unless otherwise noted. A significance level of 0.05 was used as the threshold for statistical significance. All data shown are representative of at least 2 independent experiments. U t

All animal studies were conducted under an approved Institutional Animal Care and Use Committee protocol at the University of California, San Francisco (UCSF).

Raw bulk RNA-Seq, scRNA-Seq, and ATAC-seq datasets generated for this study were deposited under Gene Expression Omnibus primary accession numbers GSE296200, GSE296201, and GSE296203. The values displayed as individual data points within the quantitative graphs are available in thefile. Supporting Data Values