The current study was carried out under relevant approvals by the Central Norway Regional Committee for Medical and Health Research Ethics (reference numbers 5.2007.910 and 2013/212/REKmidt). All patients included in the study provided informed written consent.

Patients were admitted to the Department of Gastroenterology and Hepatology at St Olav’s University Hospital, Trondheim, Norway, for colonoscopy and clinical assessment, and were diagnosed with UC or categorized as non-IBD controls. Pinch biopsies from patients were collected in 10% buffered formalin and histopathology was evaluated as described before (44). Colonoids for functional assays were established using pinch biopsies collected from uninflamed mucosa at the hepatic flexure (28, 43). Patient characteristics are listed in Table 1 .

Colonoid cultures were established and grown in 20% O₂ and 5% CO₂ at 37°C as described in Skovdahl et al. (43), based on protocols of Mahe et al. (40), and Jung et al. (45). Briefly, epithelial crypts isolated from biopsies were resuspended in ice-cold basement membrane matrix Matrigel GFR (Corning^®, #734–1101, Corning, NY, USA) and 50 μL per well was added onto pre-warmed 24-well plates. Complete growth media (CGM) for the colonoids was composed of 50% Wnt-3A conditioned medium (Wnt 3A-CM, #CRL-2647, ATCC, Manassas, VA, USA), 30% Advanced DMEM/F12 (#12634028, Thermo Fischer Scientific, Bremen, Germany), and 20% R-spondin conditioned medium (HA-R-Spondin1-Fc 293T Cells, #AMS.RSPO1-CELLS, AMS Biotechnology, Abington, UK) with 10% bovine serum albumin (BSA), supplemented with N-Acetyl-L-cystein (163.2 µg/mL, #A9165-25G, Sigma-Aldrich, MO, USA), Nicotinamide (1221.2 µg/mL, #N3376-100G, MerckMillipore, Burlington, MA, USA), Recombinant Human Noggin (0.1 µg/mL, #120-10c, PeproTech, Rocky Hill, NJ, USA), TGFβ type 1 activating receptor-like kinase inhibitor (ALK5) A-83-01 (0.211 µg/mL, #SML0788, Sigma-Aldrich), MAPK inhibitor SB202190 (3.31 µg/mL, #S7067, Sigma-Aldrich), Human EGF (0.05 µg/mL, #AF-100-15, PeproTech) and Gastrin (0.02µg/mL, G9145-1MG, Sigma-Aldrich). For colonoid establishment, CGM was supplemented with Glycogen synthase kinase inhibitor CHIRR99021 (1.16 ug/mL, #72052, STEMCELL Technologies, Vancouver, Canada) and Rho kinase (ROCK) inhibitor Thiazovivin (0.78 μg/mL, #72252, STEMCELL Technologies). Colonoids were passaged every 7-8 days at least five times after establishment before carrying out the assays. General experimental outline is shown in Figure 1A . For all experiments, colonoids were passaged into single cells and 10,000 cells per 50 μL Matrigel was used. Colonoids were grown in CGM for ten days with selective ROCK-inhibitor Y-27632 (3.203 μg/mL, #1254, Bio-Techne, Minneapolis, MN, USA) supplemented on day one and day three. Differentiation of colonoids was initiated on day ten by lowering Wnt 3A-CM concentration to 5%, withdrawing Nicotinamide and SB202190 factor from the CGM and adding the pan-Notch inhibitor DAPT (4.324 μg/mL, #2634, Bio-Techne) with media change on alternative days. On day 14, A-83-01 was removed from the differentiation media. Tofacitinib Citrate (CP-690550) (25.2 μg/mL, #S5001, Selleckchem.com) or Budesonide (4.3 μg/mL, # S1286, Selleckchem.com) were added to colonoids in differentiation media on days 11, 13, and 14 ( Figure 1A ). Tofacitinib and Budesonide were dissolved in DMSO and 0.033% DMSO was used as vehicle control. TNF (100 ng/mL, #300-01A, PeproTech) and Poly(I:C) (20 μg/mL, #tlrl-pic, In vivoGen) or IFN-γ (10 ng/mL, # 300-02, PeproTech) were added to the differentiation media and 24-hours later the conditioned media and colonoids were collected for downstream assays. Matrigel was dissolved using cell recovery solution (#734-0107, Corning, NY, United States) and colonoids were washed thrice in ice-cold Phosphate buffered saline (PBS) for immunoblotting or PBS with 0.1% BSA for RNA isolation and stored at -80°C. Alternatively, colonoids were prepared for immunostaining as described below.

To extract RNA from colonoids, the RNeasy mini kit (#74106, Qiagen, Hilden, Germany) was used as per the manufacturer’s instructions. RNA sequencing (RNA-seq) of colonoids was performed as described before (43). RNA-seq libraries were generated using SENSE total RNASeq library prep kit with RiboCop rRNA depletion (Lexogen GmbH, Vienna, Austria). Sequencing consisting of 75 single-end reads was performed using Illumina HiSeq4000 instrument (Illumina, Inc., San Diego, CA, USA), and FASTQ files were generated using bcl2fastq 2.18 (Illumina) as per manufacturer’s protocols.

All RNA-seq data analysis was performed in R software. LIMMA linear models with least squares regression and empirical Bayes moderated t statistics were used to identify differential gene expression between experimental groups, and Benjamini–Hochberg false discovery rate (FDR) correction-adjusted P values ≤ 0.05 were considered statistically significant. Enrichment analyses were generated with MetaCore+MetaDrug™ version 21.3 build 7060. Normalized reads and log2 values for all experiments are mentioned in Supplementary File SF2 . The RNA-seq dataset for TNF and unstimulated conditions is available under the GEO accession number GSE172404.

Patient colonic biopsies were fixed in 10% buffered formalin for 3-6 days as described previously (44). Colonoids collected in 50 μL Richard-Allan Scientific™ HistoGel™ Specimen Processing Gel (#HG-4000-012, Thermo Fisher Scientific, Massachusetts, USA) were fixed for 24-48 hours in 10% buffered formalin as described previously (28, 43). Formalin-fixed paraffin-embedded sections from biopsies or colonoids were treated with Neo-Clear (#1.09843.5000, MerckMillipore) and a series of ethanol for de-paraffinization. For immunohistochemistry (IHC), endogenous peroxidase was quenched with hydrogen peroxide and washed once with deionized water. Antigen retrieval was performed by boiling with citrate buffer (pH 6.0) for 15 minutes in a microwave oven. All samples were blocked with 3% BSA in 0.25% Triton-X in PBS (PBST) for 30 minutes at room temperature, washed twice in PBST, incubated overnight at 4°C with pan MHC-II-Mouse monoclonal Anti-HLA DR + DP + DQ antibody [CR3/43] (1:400 dilution in 1% BSA in PBST, #ab7856, Abcam, MA, USA) in a hydrated chamber. The following day, for secondary staining in immunofluorescence, sections were processed with MaxFluor™ 488 Immunofluorescence Mouse Detection Kit (#MF31-M, Dianova, Hamburg, Germany) as per manufacturer’s protocol and incubated 15 minutes with DAPI (1:1000 dilution in PBS, 1 mg/ml solution, #62248, Thermo Fisher Scientific, Massachusetts, USA). The rabbit/mouse EnVision-HRP/DAB+ kit (#K5007, Dako, CA, USA) was used for IHC, and sections were counterstained with haematoxylin.

All brightfield images were captured with Nikon E400 microscope, DS-Fil U2 camera, and NIS-Elements BR imaging software (Nikon Corporation, Tokyo, Japan) with 20x objective and processed with Fiji (46). Confocal fluorescence microscopy images were captured with 20x or 63x objective using a Zeiss LSM 880 Fast Airyscan Confocal microscope (Zeiss, Oberkochen, Germany). To minimize photobleaching, laser power typically was 20% under maximum, and the pinhole was set to 0.8–1.5. Images were processed with Zen Blue 2.6 software (Zeiss). Quantitative confocal image analyses of MHC-II staining in colonoids were performed using Fiji (46). The fluorescence intensity of MHC-II staining was determined by measuring the integrated density. Otsu’s thresholding algorithm was applied to convert images with DAPI staining to binary images, and the nuclei (cell counts) were quantified using the Analyze Measure plugin in ImageJ. The corrected total cell fluorescence (CTCF) was calculated: CTCF = Integrated density-(Area of selection X Mean fluorescence of background readings). The average of CTCF/cell counts for five power fields per condition for each experiment is represented in the graphs. A total of 2000-6000 cells in the colonoids were counted per treatment condition. Quantification results are presented in Supplementary File SF1 , sheet no 3.

Colonoids were taken from -80 °C, washed twice in ice-cold PBS before they were lysed for 2 hours on ice in lysis buffer containing 50 mM Tris-HCl pH 7.5, 150 mM NaCl, 5 mM EDTA, 1% NP-4#0, 1 mM DTT, 1x Complete^® EDTA-free protease inhibitor (#11873580001, Sigma-Aldrich), and 1x phosphatase inhibitor cocktail I (#P2850, Sigma-Aldrich) and III (#P0044, Sigma-Aldrich), respectively. Protein lysates were clarified by centrifugation at 14,000 × g for 20 minutes at 4°C, and protein concentration was measured using the Bradford protein assay (Bio-Rad, California, USA). Protein lysates were denatured in 1 × NuPage lithium dodecyl sulfate (LDS) sample buffer supplemented with 40 mM DTT for 10 minutes at 70°C before they were separated on 4–12% NuPage Bis-Tris gels (# NP0321BOX, Invitrogen, MA, USA) and electroblotted onto nitrocellulose membranes (Bio-Rad). The membranes were blocked in Rockland Blocking Buffer for fluorescent Western Blotting (#MB-070, Rockland, PA, USA) for 1 hour at room temperature before incubation with the indicated antibodies overnight at 4°C. The blots were incubated with Dylight secondary antibodies (Invitrogen) for visualization. Images were obtained with LI‐COR Odyssey and analyzed using Image Studio Software (LI‐COR Biosciences, NE, USA). Total protein levels were normalized to GAPDH and expressed as fold change to untreated DMSO control. The following antibodies were used: HLA DR + DP + DQ mouse monoclonal antibody [CR3/43] (1:400 dilution, #ab7856, Abcam), HLA Class 1 ABC mouse monoclonal antibody [EMR8-5] (1:200 dilution, #ab70328 Abcam), STAT11 (9H2) mouse monoclonal antibody (1:1000 dilution, #9176, Cell Signaling Technology, Danvers MA, USA), STAT1 Y701:Phospho-Stat1 (Tyr701) (D4A7) rabbit monoclonal antibody (1:1000 dilution, #7649 Cell Signaling Technology), and GAPDH (D16H11) XP rabbit monoclonal antibody (1:5000 dilution, #5174, Cell Signaling Technology). Quantification results are presented in Supplementary File SF1 , sheet number 4.

IFNγ measurement by ELISA was carried out using IFN gamma Human Uncoated ELISA Kit (#88-7316-86, Thermo Fischer Scientific) as per the manufacturer’s instructions.

Data from RNA sequencing was analyzed as described above. For all other datasets, statistical analyses were performed in GraphPad Prism 9.0. For normally distributed datasets containing two groups, paired T-test was used, and for normally distributed datasets with more than two groups, one-way ANOVA with Geisser-Greenhouse correction followed by Šidák multiple comparison testing was used. For data not following normal distribution, nonparametric Friedman’s test followed by Dunn´s multiple comparison tests were done. Comparisons between UC and non-IBD controls are made with Two-way ANOVA followed by Tukey’s multiple comparison tests. All datasets with P < 0.05 was considered statistically significant and is indicated by *.

MHC-II expression by IECs is particularly described in the small intestine, where it is expressed in differentiated cells at constitutive conditions (4) and in a subset of small intestinal undifferentiated epithelial stem cells (12). In colonic epithelium, it is considered to be absent in constitutive conditions and upregulated during inflammation (4, 16, 18). It is unclear if colonoids as a model system express MHC-II genes at constitutive conditions. Therefore, we characterized the expression of MHC-II related genes in differentiated and undifferentiated colonoids using an in-house RNA sequencing (RNA-seq) dataset from 3 non-IBD controls (D1-D3, Table 1 ). Although we noted large interindividual differences in normalized mRNA reads, we found that differentiated human colonoids expressed significantly higher levels of MHC-II related genes HLA-DMA, HLA-DMB, HLA-DRA, HLA-DRB1, and HLA-DRB6 when compared to undifferentiated colonoids. Further, HLA-DRB5, HLA-DQB1, and the MHC-II transactivator CIITA and MHC-II invariant chain CD74 had a tendency towards increased expression in differentiated epithelium compared to undifferentiated epithelium ( Figure 1B ). Since MHC-I gene expression is not well characterized in colonoids, we also explored its expression in differentiated vs. undifferentiated conditions. MHC-I gene subtype HLA-C was significantly upregulated in the differentiated epithelium, and other MHC-I related genes HLA-A, HLA-B, non-classical HLA-E and HLA-F, class I component Beta-2-Microglobulin (B2M) showed a similar tendency of increased expression in the differentiated epithelium compared to undifferentiated epithelium ( Figure 1C ). Taken together, differentiated colonoids express higher levels of MHC-II and MHC-I genes than undifferentiated colonoids.

Next, we looked at another in-house RNA-seq data set from colonoids derived from 3 non-IBD and 3 UC donors (D1-D3 and D7-D9, Table 1 ), stimulated with either TNF alone or in combination with Poly (I:C) (TNF+Poly(I:C)) for 24 hours. Pathway network analysis of the gene expression data revealed that IFN-dependent type I signaling pathway and IFNγ-dependent antigen presentation via the JAK/STAT pathways were the most upregulated pathways by the combined stimulation with TNF+Poly(I:C) ( Figure 2A ). Upon further examination of the IFNγ-dependent antigen presentation pathway ( Supplementary Figure 1A ), we observed no detectable levels of IFNγ gene expression. IFNγ receptors IFNGR1 and IFNGR2 were upregulated as well as Janus kinase JAK1, JAK2, and transcription factor STAT1 were significantly upregulated ( Figure 2B ). Notably, HLA-DQB1, HLA-DRA, HLA-DRB1, HLA-DR5, and CD74, and MHC-I related genes HLA-A, HLA-B, HLA-C, B2M, non-classical genes HLA-E and HLA-F, were upregulated upon stimulation with the combination of TNF+Poly(I:C). Similar tendency for a subset of these genes was observed with TNF-stimulation alone when compared to unstimulated conditions ( Figure 2B ). Costimulatory CD40 gene expression was also enhanced ( Figure 2B ), and a similar pattern of increased expression was observed for CIITA (P=0.089) in colonoids stimulated with TNF+Poly(I:C) ( Supplementary Figure 1B ). Since mucosa of UC patients has been shown to express costimulatory CD80 and CD86 that are important for T-cell activation, we looked at their expression in colonoids and found no expression in our RNA-seq data (data not shown). Several other genes relevant to antigen presentation pathway such as CTSS, CD74, IRF1, transcription factor USF1, transactivator of MHC-I NLRC5, and several proteosome subunits and transporter proteins, were upregulated in the colonoids stimulated with TNF+Poly(I:C) when compared to unstimulated controls ( Supplementary Figure 1C ). Together, TNF+Poly(I:C) stimulation of colonoids leads to enhanced expression of genes relevant to antigen presentation, including MHC-II expression. Interestingly, these effects were in line with upregulation of antigen presentation machinery observed in microdissected epithelium from IBD patients with active inflammation ( Supplementary Figure 1D ) (15).

Since MHC-II expression in intestinal epithelium is known to be induced by the IFNγ from the immune cells (19) and there are no immune cells present in our model, we next asked whether IFNγ could be secreted from the colonoids. We measured IFNγ in conditioned medium from unstimulated and TNF+poly(I:C) stimulated colonoids from n = 5 donors (D3, D5, D7, D9 and D10, Table 1 ) by ELISA but could not detect any IFNγ protein (data not shown). This is in line with our findings that IFNγ mRNA was not detected in the colonoids.

MHC-II protein expression and localization in IECs is altered during IBD, with enhanced expression observed at basolateral sides of the epithelium that may be functionally relevant for IECs to present antigens to immune cells in lamina propria (47). Therefore, we wanted to evaluate the protein expression and localization of MHC-II expression upon TNF+Poly(I:C) stimulation. Colonoids stimulated with TNF+Poly(I:C) or IFNγ for 24 hours were stained for MHC-II by immunofluorescence ( Figure 3A ) and immunohistochemistry ( Figure 3B ). In accordance with the gene expression data, we found that TNF+Poly(I:C) stimulation had a tendency towards upregulation of MHC-II protein levels when compared to unstimulated colonoids ( Figure 3A ). Immunohistochemistry showed staining of differentiated colonoids from donor D2, confiring that this donor expressed MHC-II protein at untreated conditions ( Figure 3B , left panels). The extent of MHC-II protein expression was enhanced upon stimulation with TNF+Poly(I:C) and IFNγ ( Figure 3B , middle and right panels). In line with observation by others (19) the colonoids showed mainly enhanced apical expression of MHC-II after 24 hours of IFNγ stimulation. MHC-II protein appeared to be expressed in both basolateral and apical expression of MHC-II after 24 hours TNF+Poly(I:C)-stimulation ( Figure 3B , middle panels). Basolateral and apical MHC-II expression was also observed in donors D2, D3, D7, and D8 by high resolution confocal imaging ( Supplementary Figure 2 ). Thus, MHC-II protein expression can be induced directly by TNF+Poly(I:C) stimulation at basolateral sides as reported for epithelial MHC-II expression in active IBD (47).

Gene expression data from colonoids indicated that upregulation of the MHC-I and MHC-II gene expression was associated with increased JAK1/2 and STAT1 expression ( Figure 2B ). Therefore, we investigated whether the pan-JAK inhibitor Tofacitinib could alter MHC-I and MHC-II expression. In comparison, we also investigated the effect of the anti-inflammatory glucocorticoid Budesonide on MHC-I and MHC-II expression. Colonoids from both non-IBD (n = 4) and UC (n = 4) donors (D3-D6 and D7-D10, Table 1 ) were pre-treated with 50 µM of Tofacitinib or 10 µM of Budesonide on day 11 in differentiation media and refreshed with new differentiation media and drugs on day 13 and one-hour before stimulation with TNF+Poly(I:C) on day 14. On day 15, 24-hours post-stimulation ( Figure 1A ), colonoids were collected and assessed for MHC-II and MHC-I expression by immunoblotting ( Table 1 ). Colonoids treated with TNF+Poly(I:C) showed enhanced levels of MHC-II protein ( Figures 4A, B ) in donors with MHC-II expression at constitutive conditions (D3, D7-D10, dashed boxes), verifying gene expression data ( Figure 2B ). In donors lacking MHC-II expression at constitutive conditions (D4-D6), we did not observe enhanced TNF+Poly(I:C)-dependent MHC-II protein expression. There was a significant upregulation of MHC-II protein expression in TNF+Poly(I:C) stimulated colonoids derived from UC donors (red circle) when compared to non-IBD donors (blue circle) ( Figure 4B ). Treatment with TNF+Poly(I:C) in colonoids was also associated with increased activation of STAT1 protein in all donors ( Figure 4C ), in line with the significant upregulation of STAT1 mRNA detected by RNA-seq ( Figure 2B ). Tofacitinib pre-treatment inhibited phosphorylation of STAT1 and reduced the expression of total STAT1 protein levels in colonoids treated with TNF+Poly(I:C) ( Figures 4A, C ). Importantly, Tofacitinib significantly down-regulated expression of TNF+Poly(I:C)-induced MHC-II protein levels in those donors that had MHC-II expression (D3, D7-D10, dashed box) by immunoblotting ( Figures 4A, B ). Budesonide pre-treatment in colonoids neither influenced the TNF+Poly(I:C)-depedent activation of STAT1 protein nor MHC-II protein expression. We also observed TNF+Poly(I:C)-dependent increase in MHC-I protein levels when stained with a pan MHC-I antibody recognizing HLA-A, HLA-B and HLA-C ( Figures 4A , Figure 4D ), verifying gene expression data ( Figure 2B ). Some donors showed a tendency towards downregulation of TNF+Poly(I:C)-depdent MHC-I protein levels due to Tofacitinib pre-teatment, but this was not statistically significant ( Figure 4D ). Further, Budesonide pre-treatment did not have any effect on TNF+Poly(I:C)-depdent MHC-I protein levels. Thus, TNF+Poly(I:C)-dependent increase of MHC-II protein expression in colonoids was downregulated with Tofacitinib pre-treatment possibly by repressing activation of STAT1, while pre-treatment with Budesonide had no effect.

Next, we investigated whether Tofacitinib and Budesonide pre-treatment influence TNF+Poly(I:C)-dependent protein localization of MHC-II. We performed endpoint immunofluorescence staining with the experimental setup described in Figure 1A , using a new passage of colonoids from the same donors used for endpoint immunoblotting analysis ( Figure 4 ). Tofacitinib pre-treatment downregulated and Budesonide pre-treatment did not affect TNF+Poly(I:C)-dependent MHC-II protein expression ( Figure 5A ) in those donors where MHC-II expression was observed (D3, D7, D8, and D10, dashed box), in line with immunoblotting results ( Figures 4A, B ). We observed apical and basolateral expression of TNF+Poly(I:C)-dependent MHC-II protein expression, as shown in Figure 3 , but did not find any alterations in protein localization due to the drug treatments.

In our endpoint immunoblotting ( Figure 4 ) and immunofluorescense ( Figure 5 ) experiments we noticed that MHC-II protein was expressed in colonoids from all UC donors (D7-D10) and one non-IBD donor (D3). Colonoids from the non-IBD donors D4-D6 did not have any MHC-II protein expression ( Figure 4A ). In order to better understand the interindividual differences in MHC-II expression, we examined whether the presence of MHC-II expression in differentiated colonoids mimicked in vivo MHC-II expression in colonic epithelium from the same donors ( Table 1 ). Amongst the 6 donors compared (4 non-IBD and 2 UC), we observed that colonoids from 1 non-IBD donor (D3) and 2 UC donors (D7 and D8) with MHC-II protein expression in colonoids ( Figure 4 ) mimicked MHC-II expression in the colonic epithelium in vivo ( Figure 5B ). Further, donors D4-D6 that did not express MHC-II in colonoids also lacked MHC-II expression in the colonic epithelium in vivo ( Figure 5B ). Thus, the reproducible interindividual differences in MHC-II expression observed in colonoid experiments recapitulate in vivo epithelial MHC-II expression.