Sult2b1-P2A-EGFP knock-in mice were developed by using the CRISPR/Cas9 genome editing system. A targeting site within the exon 7 of mouse Sult2b1 was selected using the CRISPRdirect web server (http://crispr.dbcls.jp/) (27). Single guide RNA (sgRNA) was transcribed in vitro, and a donor vector was constructed. Cas9, sgRNA, and the donor vector were microinjected into fertilized eggs of C57BL/6J mice. Fertilized eggs were implanted to obtain F0-positive mice and successful implantation was confirmed using PCR and sequencing. F0-positive mice were mated with C57BL/6J mice to generate stable F1-generation mice. Sult2b1^−/− mice were obtained from the Jackson Laboratory (stock no. 018773; Bar Harbor, ME, USA). Dock2^−/− mice have been previously described (11). Sult2b1^−/− and Dock2^−/− mice had been backcrossed with C57BL/6J mice more than 10 generations before use. C57BL/6J mice were purchased from CLEA Japan (Tokyo, Japan). The age-matched male mice (Sult2b1^+/+ , Sult2b1^+/− , and Sult2b1^−/− littermates) were used at 9–12 weeks of age. All mice were maintained under specific-pathogen-free conditions at the animal facility of Kyushu University. All animal experiments were conducted according to the relevant national and international guidelines described in the Act on Welfare and Management of Animals (Ministry of Environment of Japan) and Regulation of Laboratory Animals (Kyushu University) guidelines. The Ethics Committee on Animal Experiments at Kyushu University approved all the animal experiments performed in this study.

Mice were treated with 1.5−2.5% DSS (molecular weight, 36−50 kDa, MP Biomedicals, Solon, OH, USA) dissolved in drinking water for 5 or 6 days, respectively, followed by treatment with normal drinking water until the end of the experiment. The animals were monitored for weight loss (0, none; 1, 1−5%; 2, 5−10%; 3, 10−20%; 4, > 20%), stool consistency (0, normal stool; 2, loose stool; 4, diarrhea), and hemoccult (0, normal; 2, hemoccult positive; 4, gross blood) during the experiments. The Luminol Reaction Experiment Kit (Wako, Osaka, Japan) was used to detect fecal occult blood, as described previously (28).

Intestinal bacteria in the mice were depleted by administering a combination of antibiotics [500 μL per mouse; ampicillin (6.7 mg/mL; Sigma-Aldrich, St Louis, MO, USA), neomycin (6.7 mg/mL; Sigma-Aldrich), vancomycin (3.35 mg/mL; Wako), and metronidazole (6.7 mg/mL; Sigma-Aldrich)] by oral gavage three times a week. The combined antibiotics or sterile distilled water (vehicle control) were administered to the mice for 2 weeks prior to DSS or IND treatment. In the DSS-induced colitis model, antibiotics or water were orally administered until the mice were euthanized.

Mice were decapitated following isoflurane anesthesia; their intestinal tissues were freshly isolated, quick-frozen with liquid nitrogen, and stored at −80°C until analyses. The frozen samples were mixed with an internal standard (IS; deuterium-labeled CS; d7-CS) and homogenized in ice-cold methanol (500 μL) using a homogenizer (Finger Masher AM79330; Sarstedt, Nümbrecht, Germany). The supernatant was filtered using ultrafiltration devices (UltrafreeMCPLHCC; Human Metabolome Technologies, Yamagata, Japan), and the filtrate was directly analyzed using liquid chromatography-tandem MS (LC-MS/MS) for CS content. The triple-quadrupole MS equipped with an electrospray ionization (ESI) ion source (LCMS-8040; Shimadzu Corporation, Kyoto, Japan) was used in the negative-ESI and multiple reaction monitoring modes. The samples were resolved on the Mastro-C18 column (2.1 mm × 100 mm, 3-μm, Shimadzu GLC, Tokyo, Japan) by isocratic flow of mobile phase A (200 mM ammonium acetate) and mobile phase B (methanol) at a ratio of 1:9, flow rate of 0.4 mL/min, and column temperature of 40°C. CS and IS (d7-CS) signals were monitored by ion transitions at m/z 465.3 > 97 and 472.3 > 97, respectively. The absolute content of CS was calculated using peak area ratios of CS against IS.

The matrix-assisted laser desorption/ionization (MALDI)-linear ion trap MS (MALDI LTQ XL; Thermo Fisher Scientific, Waltham, MA, USA) and Ultraflextreme MALDI-TOF/TOF (Bruker Daltonics, Bremen, Germany) were used for MALDI imaging analysis of CS and d7-CS, as described previously (29). d7-CS was purchased from Sigma-Aldrich (#903752). Data were acquired on the TOF/TOF and LTQ instruments in negative reflectron mode or negative selected ion monitoring mode, respectively, with raster scans at a pitch distance of 30 μm. Image reconstructions of data obtained with the TOF/TOF instrument were performed using the FlexImaging 4.1 software (Bruker Daltonics), and data obtained with the LTQ instrument were performed using ImageQuest v.1.0.1 software (Thermo Fisher Scientific).

Total cell lysates were prepared and separated using sodium dodecyl-sulfate polyacrylamide gel electrophoresis (SDS-PAGE), as previously described (30). Briefly, intestinal tissues were homogenized in a 1.5-mL tube containing 250 μL of 1× cell lysis buffer (#9803; CST, Danvers, MS, USA) supplemented with a cocktail of complete protease inhibitors (Roche, Basel, Switzerland) using an electric homogenizer for 1 min on ice. After centrifugation, supernatants were mixed with an equal volume of 2× sample buffer [125 mM Tris-HCl, 0.01% bromophenol blue, 4% SDS, 20% glycerol, and 200 mM dithiothreitol (DTT)] and boiled for 10 min. Total protein concentration was measured using the DC™ Protein Assay Reagent (Bio-Rad, Hercules, CA, USA). Tissue extracts were separated using SDS-PAGE and immunoblotted with the following antibodies: rabbit anti-SULT2B1b [custom-made (26); 1:1,000], and goat anti-β-actin (#sc-1616; 1:2,000, Santa Cruz Biotechnology, TX, USA). The following horseradish peroxidase-conjugated secondary antibodies were used: mouse anti-rabbit IgG (#sc-2357; 1:2,000, Santa Cruz Biotechnology), and mouse anti-goat IgG (#sc-2354; 1:2,000, Santa Cruz Biotechnology).

Colon tissues were fixed in 4% (w/v) paraformaldehyde (Wako) for 18 h at 4°C and embedded in paraffin blocks. Sections were stained using hematoxylin and eosin (H&E) and histological scores were assigned by a trained and blinded pathologist, as previously described (31). Eight pathological changes, including the extent of inflammatory cell infiltration, goblet cell reduction, decreased crypt density, crypt hyperplasia, thickening of the muscle layer, extent of submucosal tissue inflammation, crypt abscess, and ulceration, were rated on a 0–3 scale from normal to severe. The sum of each score (maximum 24) was used as the histological score. For immunofluorescence analyses, fixed tissues were incubated with 30% sucrose (Wako) in phosphate-buffered saline (PBS) for 18 h at 4°C and embedded in O.C.T. compound (Sakura Finetek, Osaka, Japan). After freezing at −80 °C, cryostat sections were blocked with G-Block (GenoStaff, Tokyo, Japan) for 15 min at room temperature. Samples were stained using Alexa Fluor™ 488-conjugated WGA overnight at 4°C. Nucleus staining was performed using 4′,6-diamidino-2-phenylindole (DAPI; Dojindo, Kumamoto, Japan). All images were obtained using a laser scanning confocal microscope (FV3000; Olympus).

Colon tissues were homogenized in TRIzol™ reagent (Thermo Fisher Scientific) on ice for 1 min using an electric homogenizer. Next, total RNA was extracted using the TRIzol™ Plus RNA Purification Kit (Thermo Fisher Scientific). The purity and concentration of RNA were assessed using the NanoDrop™ device (ND-1000; Thermo Fisher Scientific). RNA samples were reverse-transcribed using PCR, as previously described (30). Bacterial DNA was isolated from fecal samples using the QIAamp Fast DNA Stool Mini Kit (QIAGEN, Valencia, CA, USA). Real-time PCR was performed on a CFX Connect™ Real-Time PCR Detection System (Bio-Rad) using SYBR Green PCR Master Mix (Thermo Fisher Scientific). Target gene expression was normalized to that of Hprt. Primer sequences are listed in Supplementary Table 1 . Bacteroides spp. and total bacteria in fecal samples were detected using the following primer pairs for 16S rRNA (32): Bacteroides (5′-GAGAGGAAGGTCCCCCAC-3′ and 5′-CGCTACTTGGCTGGTTCAG-3′) and total bacteria (Eubacteria: 5′-CGGTGAATACGTTCCCGG-3′ and 5′-TACGGCTACCTTGTTACGACTT-3′). Melt curve analysis was performed to ensure specificity of the amplification products.

Mice were euthanized by decapitation and blood samples were obtained after a cardiac puncture to measure the plasma IL-6 concentration using the Mouse IL-6 ELISA kit (Thermo Fisher Scientific). To quantify S100A8 in feces, fecal samples (2–3 feces per mouse) were resuspended in cold fecal protein extraction buffer (50 mM Tris, pH 7.5, 150 mM NaCl). The samples were homogenized and incubated on ice for 30 min, briefly vortexed every 5 min, and then centrifuged at 2000 × g at 4 °C for 10 min. The supernatants were diluted to 1:10 using PBS. The fecal concentration of S100A8 was measured using the Mouse S100A8 DuoSet ELISA kit (R&D Systems, Minneapolis, MN, USA).

Intraepithelial and lamina propria cells were isolated using the Lamina Propria Dissociation Kit (Myltenyi Biotec, Bergishe Gladbach, Germany), according to the manufacturer’s instructions. Briefly, the colon or SI was washed with PBS and diced. Tissues were incubated with 1× Hanks’ Balanced Salt Solution (HBSS) containing 5 mM ethylenediaminetetraacetic acid (Sigma-Aldrich), 5% fetal calf serum (Thermo Fisher Scientific), and 1 mM DTT (Sigma-Aldrich) for 20 min at 37 °C with continuous rotation using the MACSmix Tube Rotator (Myltenyi Biotec). After vortexing for 10 s, tissues were passed through a 100-μm MACS SmartStrainer (Myltenyi Biotec). The flow-through containing intraepithelial cells was collected and viable cells were recovered using density gradient centrifugation with the Lympholyte-M Cell Separation Media (Cedarlane, Hornby, Ontario, Canada). The lamina propria tissue samples were transferred into a fresh tube and incubated for 30 min at 37 °C with the Lamina Propria Dissociation Kit and gentleMACS™ Dissociators (Myltenyi Biotec). Cell suspensions were stained for viability using Cell-ID Intercalator-103Rh (1:500, Standard BioTools, South San Francisco, CA, USA) and blocked with anti-mouse CD16/32 antibody (1:1000, 2.4G2, TONBO Biosciences, San Diego, CA, USA) for 10 min prior to staining with metal-conjugated antibodies ( Supplementary Table 2 ); cells were then prepared, as previously described (33), and acquired on a Helios CyTOF Mass Cytometer (Standard BioTools). CD45⁺ live singlets were subjected to viSNE analysis using Cytobank Premium (Cytobank Inc., Santa Clara, CA, USA).

For flow cytometry, intraepithelial and lamina propria cells were prepared as described above. Cells were incubated for 10 min at room temperature with Fixable Viability Stain 510 (BD Biosciences, San Jose, CA, USA), washed, and re-incubated for 10 min on ice with anti-mouse CD16/32 (TONBO Biosciences) to block the Fc receptors. The cells were subsequently stained with the following antibodies: anti-mouse CD45 (1:100, 30-F11, BioLegend, San Diego, CA, USA), anti-mouse CD11b (1:100, M1/70, BD Biosciences), and anti-mouse Gr-1 (1:100, RB6-8C5, TONBO Biosciences). Flow cytometric analysis was performed with the BD FACSVerse™ equipped with BD FACSuite™ software (BD Biosciences).

Mouse BM-derived neutrophils were isolated from femurs and tibias of mice using HBSS (Thermo Fisher Scientific) containing 0.5% bovine serum albumin (BSA; Sigma-Aldrich). Cells were resuspended with 0.5 mL HBSS containing 0.5% BSA and treated under hypotonic conditions for 10 s by adding 3 mL of sterile distilled water, and then supplemented with 0.3 mL of 10% NaCl to restore the osmolarity. Cells were layered on a discontinuous Percoll (Sigma-Aldrich) gradient. After centrifugation, cells at the 62/81% interface were recovered and washed with HBSS and then with RPMI (Wako). Cells were resuspended in RPMI containing 2 μL of luminol (50 mg/mL; Wako) with 12.5 μM CS (#C9523; Sigma-Aldrich), 1.54 μM SOD (Sigma-Aldrich), or vehicle (0.2% dimethyl sulfoxide; Wako) at 37°C for 60 min. Cells were stimulated with 100 nM PMA (Sigma-Aldrich) and luminol luminescence due to nicotinamide adenine dinucleotide phosphate activity was measured using an IVIS Imaging System (IVIS Spectrum; PerkinElmer, Waltham, MA, USA) at each time point.

Neutrophil depletion in blood was performed, as described previously (34). Anti-Ly6G (1A8, #BP0075-1, BioXCell, West Lebanon, NH, USA), anti-rat Kappa immunoglobulin (MAR 18.5, #BE0122, BioXCell), and the corresponding isotype control (2A3, #BP0089, BioXCell) were injected intraperitoneally into the mice.

To induce SI injury, IND (Wako) dissolved in 5% NaHCO₃ was subcutaneously administered in the back neck of the mice (5 mg/kg body weight), as described previously (35). CS (200 mg/kg body weight; Sigma-Aldrich) or vehicle alone was orally administered to mice three times at 4 h intervals. d7-CS (330 μg per mouse; Sigma-Aldrich) was orally administered to Sult2b1^+/+ mice and assessed 4 h later. CS and d7-CS were dissolved in 40 mM 2-hydroxypropyl-beta-cyclodextrin (HPβCD; Sigma-Aldrich) before use. Twenty-four hours after IND administration, mice were intravenously administered 150 μL of 0.4% Evans blue dye (Wako) dissolved in PBS, and the mice were euthanized by decapitation 30 min later. The number of ulcer lesions in the SI stained with blue dye were counted under the stereomicroscope (SZ-PT; Olympus) and these areas were quantified using Image J (National Institutes of Health, Bethesda, MD, USA). Ulcer lesions in the SI were also sampled for flow cytometry analysis to calculate the absolute number of neutrophils per wet weight.

Graphing and statistical analyses were performed using Prism 8 (GraphPad Software, La Jolla, CA). The data were initially tested using a Kolmogorov–Smirnov test for normal distribution. For comparison between the two groups, parametric and nonparametric data were analyzed using a two-tailed unpaired Student’s t-test and a two-tailed Mann–Whitney test, respectively. Statistical differences between more than two experimental groups were evaluated using analysis of variance (ANOVA) with Dunnett’s multiple comparison test. Survival analysis was performed using Kaplan–Meier curves, with comparisons between groups made using the log-rank test. Data are expressed as mean ± standard deviation (SD), and P-values of less than 0.05 were considered significant. All sample sizes and statistical tests employed are documented in figure legends.

We have previously found that the SI expresses SULT2B1 and produces CS in mice (26). To quantify CS produced in the intestinal tract, we measured CS levels in whole tissues of the SI and colon among Sult2b1^+/+ , Sult2b1^+/− , and Sult2b1^−/− mice using MS analysis. Although CS was absent throughout the intestinal tract of Sult2b1^−/− mice, it was produced in the duodenum, jejunum, ileum, and the whole colon of Sult2b1^+/+ mice ( Figure 1A ). Moreover, MS analysis revealed markedly decreased CS levels, even in Sult2b1^+/− mice ( Figure 1A ). Recent microbiological studies have reported that Bacteroides spp. produce CS in humans and mice via a specific enzyme with sulfotransferase-like activity (36, 37). Consistently, CS and Bacteroides spp. were detected in the feces of Sult2b1^−/− mice ( Supplementary Figures 1A, B ), indicating that intestinal CS was produced by both intestinal tissues and commensal bacteria. However, CS concentration in feces collected from Sult2b1^−/− mice was markedly reduced to 21% of that from Sult2b1^+/+ mice, even though Bacteroides spp. increased in the gut of Sult2b1^−/− mice ( Supplementary Figures 1A, B ). These findings suggest that IECs are the primary source of intestinal CS.

To identify CS-producing regions in the intestinal tract, we further investigated CS localization using MS imaging. CS was specifically concentrated at the top of the villi close to the intestinal lumen in the SI and colon of Sult2b1^+/+ mice ( Figures 1B, C ). In contrast, CS production was not detected in Sult2b1^−/− mice ( Supplementary Figure 1C ). Cholesterol sulfation is mediated by SULT2B1b produced from the gene Sult2b1 in mice and SULT2B1 in humans (22, 23, 26). Western blot analyses showed the presence of SULT2B1b in the SI and colon of Sult2b1^+/+ mice ( Figure 1D ). On the other hand, SULT2B1b expression levels in Sult2b1^+/− mice were reduced to 51 and 41% of that in Sult2b1^+/+ mice in the SI and colon, respectively ( Figures 1D, E ), demonstrating that gene dosage affects SULT2B1b expression. The publicly available single-cell RNA-seq data [Tabula Muris (38)] revealed upregulation of the Sult2b1 gene in the enterocytes of the colon tissue, which are nutrient-absorbing intestinal epithelial cells (IECs; Figure 1F and Supplementary Figure 1D ). Additionally, we generated Sult2b1-P2A-EGFP knock-in mice ( Supplementary Figure 2 ) and confirmed the presence of Sult2b1-expressing enhanced green fluorescent protein (EGFP)⁺ cells in the upper villi of the SI and colon ( Figure 1G ). Therefore, IEC-derived CS may have physiological roles at the interface between the intestinal lumen and mucosa.

To examine whether CS affects gut inflammation, we compared the severity of DSS-induced colitis among Sult2b1^+/+ , Sult2b1^+/− , and Sult2b1^−/− mice. DSS-induced colitis is widely practiced, owing to several similarities with human UC (39). Macroscopic analyses of the colon, SI, and spleen revealed no significant difference between Sult2b1^+/+ and Sult2b1^−/− mice at steady state ( Supplementary Figure 3 ). However, on treating these mice with 2.5% DSS for 6 days, the survival rate on day 18 was markedly reduced in the absence of Sult2b1 ( Figure 2A : Sult2b1^+/+ mice, 100%; and Sult2b1^−/− mice, 0%). Consistent with the CS production levels ( Figure 1A ), the survival rate of Sult2b1^+/− mice was also reduced to 32% ( Figure 2A ).

Based on these findings, we speculated that even short exposures to and/or low concentrations of DSS would induce severe colitis in Sult2b1^+/− and Sult2b1^−/− mice. To test this hypothesis, we treated Sult2b1^+/+ , Sult2b1^+/− , and Sult2b1^−/− mice with 1.5% DSS for 5 days. Sult2b1^+/− and Sult2b1^−/− mice exhibited exacerbated weight loss and increased disease activity index (DAI) scores, compared to that in Sult2b1^+/+ mice ( Figures 2B, C ). Consistently, on day 9, the colon length in Sult2b1^+/− and Sult2b1^−/− mice was shorter than that in Sult2b1^+/+ mice ( Figure 2D ). After the DSS challenge, CS diffused throughout the crypts from the luminal side toward the muscularis mucosae ( Figure 2E ). Although the gene expression levels of inflammatory cytokines including interleukin (Il)6, Il17a, Il1b, and tumor necrosis factor-α (Tnfa) gradually increased with colitis progression in Sult2b1^+/+ mice, these levels were considerably higher in Sult2b1^+/− and Sult2b1^−/− mice than those in Sult2b1^+/+ mice on day 7 ( Figure 2F ). Similarly, plasma IL-6 levels on day 7 increased in Sult2b1^+/− and Sult2b1^−/− mice, compared to those in Sult2b1^+/+ mice ( Figure 2G ). Moreover, histological analysis revealed severe colitis in Sult2b1^−/− mice, characterized by immune cell infiltration, goblet cell loss, and ulceration ( Figures 2H, I ). Thus, the lack of CS exacerbated DSS-induced colitis.

To elucidate the functional role of CS, we examined whether it affects the barrier function of IECs in vivo. Staining intestinal sections with wheat germ agglutinin (WGA), a lectin bound to mucins on the luminal surface epithelium and in goblet cells, revealed similar mucin distribution between Sult2b1^+/+ and Sult2b1^−/− mice ( Figure 3A ). In addition, mRNA expression levels of Tjp1, Ocln, Muc2, and Lypd8, which are critical for IEC barrier functions (40–42), were comparable between Sult2b1^+/+ and Sult2b1^−/− mice at steady state and after 1.5% DSS challenge ( Figure 3B ). Therefore, CS function may not be mediated by enhancing intestinal barrier function. We further determined whether CS-deficient mice exhibit severe colitis in the absence of commensal bacteria during DSS-induced mucosal injury. By bacterial depletion using oral antibiotics prior to the 1.5% DSS challenge ( Figure 3C ), the clinical signs of colitis, such as body weight loss, DAI score, and colon length shortening, were recovered in Sult2b1^−/− mice to the levels similar to those in Sult2b1^+/+ mice ( Figures 3D–F ). Thus, these results showed that the lack of CS does not exacerbate colitis in the absence of intestinal bacteria.

Then, we hypothesized that DOCK2 in immune cells is the functional target of CS during colitis because CS binds to the catalytic domain of DOCK2 and inhibits its Rac-GEF activity (26). To address this hypothesis, we crossed Sult2b1^−/− with Dock2^−/− mice to obtain Sult2b1 and Dock2 double knockout mice (Sult2b1^−/−Dock2^−/− ). When mice were treated with 2.0% DSS for 5 days, Sult2b1^−/−Dock2^+/+ mice exhibited exacerbated weight loss compared to Sult2b1^+/+Dock2^+/+ mice ( Figure 3G ). In contrast, there was no difference among Sult2b1^+/+Dock2^+/+ , Sult2b1^+/+Dock2^−/− , and Sult2b1^−/−Dock2^−/− mice in the degree of weight loss ( Figure 3G ), indicating that CS alleviates DSS-induced colitis by inhibiting DOCK2.

To comprehensively analyze the immune cell profiles in the colon from Sult2b1^+/+ and Sult2b1^−/− mice during DSS-induced colitis, we performed high-dimensional phenotyping of intraepithelial and lamina propria immune cells using cytometry by time-of-flight (CyTOF). After gating on singlets and live CD45⁺ cells, colonic immune cells were subdivided into 14 populations ( Figure 4A ), which were identified by the differential expression of individual lineage markers and the visualization of t-distributed stochastic neighbor embedding (viSNE) algorithms ( Figure 4B ). On day 7 after the 1.5% DSS challenge, CyTOF analyses revealed increased proportions of neutrophils, monocytes, and macrophages in the lamina propria and intraepithelial CD45⁺ cells of Sult2b1^−/− mice, compared to those in Sult2b1^+/+ mice ( Figure 4C and Supplementary Figure 4 ). In contrast, the proportions of conventional dendritic cells, eosinophils, natural killer cells, and CD8α⁺ T cells decreased in Sult2b1^−/− mice, with no significant difference in the percentage of mast cells, CD4⁺ T cells, γδ T cells, and B cells ( Supplementary Figure 4 ). Thus, CS deficiency altered the composition of immune cells during DSS-induced mucosal injury.

It has been shown that neutrophils are commonly involved in several inflammatory diseases through infiltration into the tissue and ROS production (43). Therefore, we focused on neutrophils and compared their prevalence in the colon of Sult2b1^+/+ , Sult2b1^+/− , and Sult2b1^−/− mice by using flow cytometry. We found that, irrespectively of Sult2b1 expression, the percentage of CD45⁺ cells in the total live cells gradually increased in the colon during the progression of colitis; however, the percentages of CD45⁺ cells were unchanged among these mice ( Supplementary Figures 5A, B ). In contrast, the neutrophil prevalence in CD45⁺ cells of the lamina propria was considerably higher in Sult2b1^+/− and Sult2b1^−/− mice than that in Sult2b1^+/+ mice on days 5 and 7 ( Figure 5A and Supplementary Figure 5C ). Similarly, on day 7, the prevalence of intraepithelial neutrophils in CD45⁺ cells increased in Sult2b1^−/− mice compared to that in Sult2b1^+/+ mice ( Figure 5B and Supplementary Figure 5C ). During colitis, neutrophils abundantly express and secrete S100A8 protein, which forms an S100A8/S100A9 heterodimer (also called calprotectin; a non-invasive biomarker in IBD) (44). As expected, Sult2b1^−/− mice exhibited a notably higher fecal level of S100A8 protein than that in Sult2b1^+/+ mice on day 7 after the DSS challenge ( Figure 5C ). Furthermore, we evaluated whether CS could inhibit reactive oxygen species (ROS) production in neutrophils, because ROS production by murine and human neutrophils largely depends on DOCK2-mediated Rac activation (18, 20). When wild-type bone marrow (BM)-derived neutrophils were stimulated with phorbol myristate acetate (PMA), they produced ROS in a superoxide dismutase (SOD)-inhibitable manner, and this ROS production was also inhibited by CS treatment ( Figure 5D ). These results suggest that CS could suppress infiltration of and ROS production by neutrophils in the intestinal tract.

As neutrophil recruitment was enhanced in CS-deficient mice after the DSS challenge, we assessed the contribution of circulating neutrophils to colitis using anti-Ly6G neutralizing antibodies. Based on the results of a previously reported neutrophil depletion protocol with some modifications (34), we confirmed that circulating neutrophils were reduced in the blood ( Figures 5E, F ). Although neutrophils in the lamina propria were not reduced ( Figure 5F ), anti-Ly6G-treated Sult2b1^−/− mice exhibited significant recovery of the clinical signs including body weight loss, increased DAI scores, and colon length shortening, compared to isotype control-treated Sult2b1^−/− mice ( Figures 5G-I ). Thus, these results indicate that reducing the number of circulating neutrophils leads to the alleviation of DSS-induced colitis in Sult2b1^−/− mice.

Although NSAIDs are one of the most commonly used drugs owing to their analgesic properties, their adverse effects, such as gastrointestinal tract ulcers, are concerning (9, 10). NSAID-induced gastric ulcers can be controlled by consuming acid-inhibitory drugs; no medicines can prevent NSAID-induced SI ulcers (9, 10). As IECs in the SI expressed Sult2b1 and produced CS abundantly ( Figure 1 ), CS might also act in the SI to alleviate excessive inflammation.

To determine whether CS affects inflammatory status in the SI, we subcutaneously administered a high dose of IND (5 mg/kg body weight) to Sult2b1^+/+ and Sult2b1^−/− mice with or without the oral treatment of CS ( Figure 6A ). MS imaging revealed the absence of CS in the crypts around ulcerative lesions of Sult2b1^+/+ mice after IND administration ( Figure 6B ). Furthermore, Sult2b1^−/− mice exhibited markedly increased counts and areas of SI ulcers, compared with those in Sult2b1^+/+ mice, and these ulcerative lesions were dramatically ameliorated by CS administration ( Figures 6C, D ). Similarly, compared with Sult2b1^+/+ mice, the absolute number of neutrophils in ulcerative lesions increased in Sult2b1^−/− mice, which was also suppressed by CS administration ( Figure 6E ). To better understand localization of the administered CS, we orally administered deuterium-labeled CS (d7-CS) to Sult2b1^+/+ mice and found that d7-CS was detected in the blood at 1 h after oral administration and localized in the lamina propria and ulcerative lesions lacking endogenous CS 4 h later ( Figure 6F and Supplementary Figure 6A ). In addition, when circulating neutrophils were depleted by intraperitoneal injection of anti-Ly6G antibodies ( Supplementary Figures 6B, C ), Sult2b1^−/− mice showed a significant decrease in the counts and areas of SI ulcers ( Figure 6G and Supplementary Figure 6D ). Taken together, these results suggest that CS contributes to protecting SI tissues from ulcer formation by suppressing the infiltration of inflammatory neutrophils. In addition, NSAID-induced ulcers are known to be exacerbated by the invasion of commensal bacteria (9, 10). Consistent with this, antibiotic treatment markedly suppressed IND-induced ulcer formation even in Sult2b1^−/− mice ( Figure 6H and Supplementary Figure 6E ).