Epithelial TNF controls cell differentiation and CFTR activity to maintain intestinal mucin homeostasis

The gastrointestinal tract relies on the production, maturation, and transit of mucin to protect against pathogens and to lubricate the epithelial lining. Although the molecular and cellular mechanisms that regulate mucin production and movement are beginning to be understood, the upstream epithelial signals that contribute to mucin regulation remain unclear. Here, we report that the inflammatory cytokine tumor necrosis factor (TNF), generated by the epithelium, contributes to mucin homeostasis by regulating both cell differentiation and cystic fibrosis transmembrane conductance regulator (CFTR) activity. We used genetic mouse models and noninflamed samples from patients with inflammatory bowel disease (IBD) undergoing anti-TNF therapy to assess the effect of in vivo perturbation of TNF. We found that inhibition of epithelial TNF promotes the differentiation of secretory progenitor cells into mucus-producing goblet cells. Furthermore, TNF treatment and CFTR inhibition in intestinal organoids demonstrated that TNF promotes ion transport and luminal flow via CFTR. The absence of TNF led to slower gut transit times, which we propose results from increased mucus accumulation coupled with decreased luminal fluid pumping. These findings point to a TNF/CFTR signaling axis in the adult intestine and identify epithelial cell–derived TNF as an upstream regulator of mucin homeostasis.


Introduction
The intestine is exposed to the external environment through ingestion, and its epithelial barrier is required to prevent the entry of toxins and pathogens and restrict resident commensal microbes to the intestinal lumen (1,2). The small intestine is folded into invaginations that house multipotent stem cells, known as crypts, and long finger-like protrusions that house differentiated cells that absorb nutrients (among other functions), known as villi. The epithelium comprises a variety of specialized cell types that derive from Leucine-Rich Repeat Containing G Protein-Coupled Receptor 5 (Lgr5) + stem cells (3). Stem cells give rise to highly proliferative absorptive and secretory progenitor cells, which in turn differentiate into cells that perform specialized functions. Absorptive progenitors produce enterocytes that absorb nutrients (1), whereas secretory progenitors give rise to Paneth, goblet, tuft, and enteroendocrine cells that release protective and regulatory factors (1). Paneth cells at the crypt bottom produce antimicrobial peptides (AMPs) that protect the epithelium from bacteria; enteroendocrine cells in the villus produce hormones that regulate digestion and absorption; tuft cells coordinate host immune responses; and goblet cells in the crypt and villus produce the majority of mucin in the intestine (4).
Goblet cell-derived mucin creates a physical protective barrier that traps pathogens and contaminants (5). In addition, the mucus layer serves as a lubricant for the passage of digested food and waste through the gut (6). Defects in mucin homeostasis, seen in diseases such as cystic fibrosis, result in accumulation of mucin, increased bacterial load, and a significant decrease in the rate of food and waste transit (7,8). Once mucin is produced, the activity of the chloride anion channel cystic fibrosis transmembrane conductance regulator (CFTR) (9) and the protease meprin-β (10) promote mucin unfolding and shedding, respectively. While we are beginning to understand how the production and mobilization of mucin is regulated at the level of individual cells, whether and how upstream epithelial signals contribute to these processes remains unclear.
Various studies point to the cytokine tumor necrosis factor (TNF) as a potential regulator of mucin homeostasis. However, much remains to be learned, as TNF has been predominantly studied in the context of cell death and inflammation in the intestine (11,12), and its function in non-disease states and homeostasis is less clear (13), with 4 conflicting reports about the relationship between TNF and mucin. Contradicting studies report that TNF promotes transcription of Muc2 (14), the major component of mucin, while others assert that it decreases the production and thickness of the mucus layer (15,16).
Developmental studies have also shown divergent results (17,18), and interpretation of data from in vitro studies has been limited by the use of colon cancer cell lines, which lack intestinal cell type diversity and crypt-villus architecture (19). Thus, the role of TNF in adult intestinal homeostasis and mucin production, and the cell types that produce and respond to TNF signals, are unknown.
In this study, we aimed to elucidate the cellular source and functional role of TNF during mucin homeostasis by examining mouse and human adult intestinal tissues and using intestinal organoids that more closely recapitulate in vivo physiology than do 2D cell line models (20). RNA in situ hybridization and immunohistochemical analysis revealed that crypt cells are the major epithelial producers of TNF. Genetic studies demonstrated that loss of TNF led to mucin accumulation and slower gut transit time. By combining genetic ablation of Tnf with lineage tracing of progenitor cells, we found that TNF is required for maintaining the proportion of secretory and absorptive progenitors and for suppressing a goblet cell differentiation bias. Using live imaging, we determined that TNF regulates luminal fluid pumping in organoids by modulation of CFTR activity, establishing a TNF-CFTR signaling axis for mucin flux in the intestine. Finally, to establish the clinical relevance of our mouse and organoid studies to human physiology, we obtained ileal samples from healthy human donors, inflammatory bowel disease (IBD) patients, and IBD patients treated with anti-TNF therapy and performed immunohistochemical analysis for goblet cells. Patients treated with anti-TNF therapy had a higher proportion of goblet cells when compared to non-inflamed IBD samples or even healthy donor samples. This closely matched the increase in goblet cell number we observed when epithelial TNF was deleted in mouse intestine and suggests that anti-TNF therapy induced goblet cell hyperplasia may similarly lead to increased mucin accumulation, bacterial load, and small intestinal transit. We therefore conclude that TNF contributes to mucin homeostasis by regulating both secretory cell differentiation into goblet cells and mucin maturation through the activity of CFTR. Our data identify a previously underappreciated TNF-CFTR signaling axis in intestinal homeostasis and disease.

TNF, TNFR1, and TNFR2 are expressed in defined spatial domains along the cryptvillus axis
To identify possible sender-receiver signaling relationships between cells in the small intestine, we assessed the mRNA and protein expression of the ligand TNF and its receptors TNFR1 and TNFR2. We utilized probes for single-molecule RNA in situ hybridization (RNAscope) targeting Tnf (encodes TNF), Tnfrsf1a (encodes TNFR1), and  Figure 1C). By FACS sorting epithelial crypt and villus cells from the mouse ileum and performing qPCR, we quantitatively compared the expression of Tnf, Tnfrsf1a, and Tnfrsf1b in both intestinal compartments ( Figure 1D).

Compared to villus cells, crypt cells expressed about 6-fold more Tnf, comparable
Tnfrsf1a, and 15-fold more Tnfrsf1b transcripts, closely matching the RNAscope data.
We next examined protein abundance using immunohistochemistry and found similar protein localization along the crypt-villus axis (Figure 1, E-G). TNF was broadly present in intestinal epithelium and enriched in epithelial crypts. ( Figure 1E). TNFR1 expression was highest at the villus tip and gradually decreased towards the crypt ( Figure 1F) and TNFR2 was strongly enriched in crypts ( Figure 1G). When we deleted Tnf (Supplemental Figure 1A), Tnfrsf1a (Supplemental Figure 1B), or Tnfrsf1b (Supplemental Figure 1C), expression of the proteins encoded by these genes was no longer detected, validating the in situ and immunofluorescence results and demonstrating that the genetic tools are robust.
These findings demonstrate a defined expression pattern of TNF ligand and its receptors, which may permit different cellular responses to the same TNF signal across the crypt-villus axis.
The absence of TNF causes increased luminal mucin, goblet cell number, gut transit time, and bacterial load 6 Elevated levels of TNF in Crohn's disease tissue samples or treatment of colon cell lines with exogenous TNF are associated with decreased abundance and thickness of the mucus barrier (16,21,22), suggestive of a role for TNF in maintaining intestinal mucus. However, whether and how TNF might regulate mucin production under homeostatic conditions is unclear. Therefore, we evaluated the role of TNF in mucin homeostasis by analyzing constitutive Tnf −/− mice (23), which lack TNF in all tissues, including the intestinal epithelium. In comparison with age-matched control mice, luminal mucin was elevated in Tnf −/− mice ( Figure 2, A and B). However, intracellular epithelial mucin measured by mean mucin fluorescence ( Figure 2C) and granule size per goblet cell ( Figure 2D) were unaffected. These findings suggested that increased luminal mucin levels in mutant mice could be explained, at least in part, by an increase in mucus-producing goblet cells.
We then quantified goblet cell number in control and mutant villi and crypts and found Prior work has correlated increased luminal mucin with increased gut transit time (8).
Therefore, we tested if gut transit time was delayed in Tnf −/− mice. Using a 70 kD FITC-Dextran dye, we tracked anteroposterior dye displacement along the gut length over time ( Figure 2H). After 1 hour, FITC-Dextran traveled 80 percent of the total gut length of control mice and 70 percent of the total gut of mutant mice ( Figure 2I), while total gut length remained constant between both groups ( Figure 2J). These findings indicate that, in the absence of TNF, the expulsion of digested food and waste is impaired.
Together with slower gut transit, previous work has shown that increased luminal mucin is associated with higher levels of bacterial load (24). We assessed the bacterial load of control and mutant mice by isolating genomic DNA from feces and performing qPCR for the universal bacterial 16s rRNA gene, which showed that bacterial load in mutant mice was elevated ~ 20-fold compared to control ( Figure 2K). Thus, during homeostasis, TNF contributes to the regulation of intestinal mucin levels, goblet cell number, gut transit, and bacterial load.

TNF does not affect secretory cell turnover, but controls goblet cell number by regulating secretory progenitor cell differentiation
Goblet cell number depends on the rate of loss of mature goblet cells through cell death and/or extrusion from the villi and the rate of differentiation from secretory progenitors.
We investigated how goblet cell numbers increase in the crypts of Tnf −/− mice using a lineage tracing approach, focusing on the fate of secretory progenitors. We bred Atoh1 CreERT2 ; Rosa26 tdTomato mice (25)(26)(27), which label secretory progenitors upon tamoxifen induction, with Tnf −/− mice, and then lineage traced secretory progenitors in the absence of TNF.
Crypt cells give rise to differentiated cells that transit along the crypt-villus axis to replace more mature cells that ultimately die and are shed at villus tips (27). This continuous epithelial turnover takes approximately 3-5 days in the small intestine (1,28). We predicted that lower rates of differentiated cell turnover would lead to an increase in goblet cell number due to the persistence of differentiated goblet cells in the epithelium. To explore this possibility, we measured the movement of labeled proliferating cells out of the crypt in Atoh1 CreERT2 ; Rosa26 tdTomato ; Tnf −/− mice as a proxy for secretory cell turnover (27,29). We treated mutant mice with a single tamoxifen dose followed by a single injection of the thymidine analog EdU 24 hours later ( Figure 3A). Over the course of 72 hours, proliferative secretory progenitors were uniquely labeled with both tdTomato and EdU, enabling us to map the appearance of nascent mature secretory cells arising from secretory progenitors and to measure the movement of double-positive tdTomato + /EdU + cells out of the crypt and up the villus. We found that after 48 hours of EdU labeling, control and mutant secretory progenitor cells alike were displaced ~ 50 μm from the hinge region that separates the crypt and villus compartments ( Figure 3, B-C), suggesting that secretory cell turnover was unaltered.
Next, we assessed the number of secretory progenitors in crypts, which would affect the overall rate of differentiation into goblet cells. In our previous experiments, both control and mutant secretory progenitors were similarly displaced ~ 50 μm from the hinge region over the course of 48 hours. Therefore, we chose to chase for 36 hours to preferentially label secretory progenitors while minimizing lineage tracing of differentiated secretory cells in the villus. We induced Atoh1 CreERT2 ; Rosa26 tdTomato ; Tnf −/− mice with a single tamoxifen pulse followed by a short chase of 36 hours ( Figure 3A). We found that labeled Atoh1 CreERT2 ; Rosa26 tdTomato ; Tnf −/− mice had increased numbers of tdTomato + cells within crypts. The increase in the number of tdTomato + cells indicates an increase in ATOH1 + secretory progenitors, supporting a role for a secretory differentiation bias in the absence of TNF ( Figure 3D). Since the crypt houses secretory (ATOH1 + ) and absorptive (NICD + ) progenitors (26), we next assessed if the increase in ATOH1 + /tdTomato + secretory progenitors was specific to this lineage and at the expense of the number of absorptive progenitors by staining absorptive progenitors for NICD ( Figure 3E). We found that Tnf −/− mice have an increase in tdTomato + secretory progenitors with a corresponding decrease in absorptive progenitors (Figure 3, F and G). To determine if there was a specific increase in secretory progenitors and not a general increase in all cells, we used crypt and villus length to estimate any changes in total cell number. We found that the total number of progenitors per crypt increased ( Figure 3H

TNF expressed by epithelial cells regulates mucin homeostasis and goblet cell differentiation
Prior work (31,32) and our own immunohistochemical analysis of TNF suggests that the epithelium is a significant source of TNF that drives the observed mucin phenotypes.
However, it has been shown that mesenchymal cells are also an intestinal source of TNF (33). To determine if epithelial-derived TNF is required for proper mucin homeostasis, we generated mice harboring both the Vil CreERT2 and Tnf flox/flox alleles (34,35), which allowed for spatio-temporal deletion of TNF as opposed to constitutive deletion in Tnf −/− mice. In Vil CreERT2 ; Tnf flox/flox mice, we deleted TNF in intestinal epithelial cells by administering daily doses of tamoxifen over 6 days, which is ~ 2 epithelial turnover cycles ( Figure 4A). and bacterial load (7,8,24). CFTR function is essential for proper mucin flux and maturation in the intestine (9,10). We therefore asked if epithelial TNF acted as an upstream regulator of CFTR, in addition to regulating epithelial cell differentiation. We first characterized the spatial expression of CFTR within the intestine using RNAscope and found that Cftr transcripts were spatially restricted to epithelial crypts in the ileum and a subpopulation of villus cells. Among cells in the crypt, LGR5 + stem cells highly expressed Cftr (Supplemental Figure 2A). These data are consistent with previous reports of CFTR expression (36)(37)(38) and suggest that the crypt base is a major site of fluid and ion pumping downstream of CFTR activity. Furthermore, the zonated expression of CFTR, TNF, and TNF receptors (Figure 1) in epithelial crypts indicate a potential interaction between CFTR and TNF in crypts.
Intestinal organoids provide an ideal system to dissect molecular mechanisms intrinsic to the epithelium by enabling the analysis of genetic and pharmacological perturbations coupled with live imaging. During normal organoid growth, fluid flows into organoid lumens and causes them to periodically inflate and collapse in a CFTR-dependent manner (39). Indeed, inflation of organoids has been used to predict patient response to drugs in cystic fibrosis, which is caused by loss-of-function mutations in CFTR (40,41). We assessed the impact of TNF on CFTR activity by first measuring the rates of inflation of control and Tnf −/− organoids cultured in normal growth conditions. The rate of inflation of Tnf −/− organoids was significantly reduced in normal growth media, ~ 2.5 times slower than controls ( Figure 5, A and B). Consistent with a role for TNF upstream of CFTR, recombinant TNF rescued the lumen size of Tnf −/− organoids at 24 h of growth, and combined treatment with rTNF and the CFTR inhibitor (CFTRinh-172) blocked the previously observed rescue ( Figure 5C). Using a Villin CreERT2 ; Cftr flox/flox conditional knockout (8), we were able to further validate that TNF acts upstream of CFTR to induce luminal flow as Cftr loss-of-function reversed the organoid inflation induced by rTNF treatment ( Figure   5D).
We next aimed to determine the mechanism by which TNF modulates CFTR activity.
We hypothesized that TNF could increase the total amount of CFTR and thus increase CFTR activity. To test this, we performed qPCR on control or Tnf −/− organoids grown for 24 h in culture. We found that there was no change in Cftr transcripts in Tnf −/− organoids (Supplemental Figure 2C), suggesting that TNF acts on CFTR post-transcriptionally.
Therefore, we hypothesized that rTNF was increasing the quantity of active CFTR in organoids. To get a more accurate read out of maximum CFTR function we used forskolin to elevate cyclic AMP levels and stimulate CFTR activity and swelling in intestinal organoids (40). Previous work showed that rTNF stimulates CFTR activity through a PKC-dependent mechanism in human bronchial epithelial cells (42). Therefore, we co-stimulated control organoids with forskolin and combinations of rTNF, CFTR-inh172, and the PKC inhibitor Bisindolylmaleimide I (GF109203X). We found that TNF had an additive effect with forskolin in inducing organoid swelling and caused an overall 2-fold increase in lumen swelling. This additive effect was abrogated with CFTR inhibition, confirming that rTNFinduced swelling acts through CFTR. Additionally, rTNF co-treatment with the PKC inhibitor GF109203X abrogated the additive effect of rTNF, pointing to a PKC-dependent mechanism ( Figure 5, E and F). Thus, we conclude that rTNF modulates CFTR-induced fluid pumping through PKC.
To identify whether TNFR1 or TNFR2 was required for TNF modulation of CFTR-in- Ions released by CFTR promote mucin unfolding, exposing sites where the protease meprin-β can then cleave allowing mucin to be processed and released (5). Failure to shed meprin-β from epithelial cell membranes leads to unprocessed MUC2 and more dense mucus (10). Therefore, we asked whether epithelial TNF also acts upstream of meprin-β and assessed the activity of meprin-β in Vil CreERT2 ; Tnf flox/flox mice by using protease localization as a readout. Vil CreERT2 ;Tnf flox/flox mice showed more intense and dense meprin-β staining compared to controls (Supplemental Figure 2) indicative of less shed meprin-β and consequently a decreased ability to cleave mucin. Combined, these results expand the role of epithelial TNF in mucin homeostasis beyond control of goblet cell number to include modulation of two key processes in mucin flux: regulation of CFTR activity and meprin-β shedding.

IBD patients treated with anti-TNF have an increased number of goblet cells in crypts
Given the importance of TNF levels in IBD as well as many other diseases (43,44), we next aimed to determine the clinical relevance of our findings. In IBD, goblet cell numbers have been reported to decrease (23,45), while anti-TNF treatment has been shown to lead to an increase in goblet cell numbers in mouse colitis (46). Here, we showed that loss of TNF causes an increase in goblet cell number under homeostatic conditions. Thus, we set out to determine the effect of TNF under conditions that mimicked homeostasis and disease in patients. We obtained non-inflamed ileal tissue samples from IBD patients, IBD patients undergoing anti-TNF therapy, and healthy donors, and quantified goblet cell proportion in crypt epithelium ( Figure 6). We found that non-inflamed tissue from IBD patients and healthy donors had a similar proportion of goblet cells, which supports the notion of non-inflamed IBD tissue as a good model of normal goblet cell homeostasis in patients. Non-inflamed IBD tissue thus allowed for the analysis of the effect of anti-TNF therapy in a condition resembling steady state in humans.
Strikingly, IBD patients undergoing anti-TNF therapy had a higher proportion of crypt goblet cells than IBD patients and healthy donors ( Figure 6, B and C). This reproduces the goblet cell hyperplasia phenotypes in TNF loss-of-function mice ( Figures 2G and 4F), suggesting a similar mechanism of TNF control of goblet cell differentiation in human physiology as in mouse.

Discussion
Mucin provides a layer of protection against external pathogens and contaminants (5,6). In the respiratory tract, the epithelium is protected from pathogens in part by active ciliary movement that clears mucin away from the epithelium. In contrast, the digestive tract lacks ciliary movement but still maintains flux of mucin away from the epithelium (5).
In healthy intestines, the mucus layer facilitates waste expulsion, protects the host against pathogens, and houses commensal microbes that aid in nutrient absorption (2,47). However, mucus can accumulate in disease states, leading to increased bacterial load, inflammation, and tissue damage. Loss of the mucosal barrier can also lead to disease by induction of persistent intestinal inflammation, as seen in inflammatory bowel disease (IBD) (48). Thus, a balance between the maintenance and turnover of mucus is essential for gut homeostasis. However, the upstream molecular mechanisms regulating the production and flux of mucin are poorly understood.
While TNF is a cytokine with a well-appreciated role in inflammation, here, we revealed previously underappreciated functions of TNF in epithelial mucin homeostasis. First, TNF regulates mucin production by regulating the differentiation of goblet cells, the mucinproducing cells in the respiratory and digestive tracts. Both TNFR1 and TNFR2 contribute to changes in goblet cell number (Supplemental Figure 2, G and H), and therefore, both may also have roles in regulating differentiation. Second, TNF controls the flux of mucin along the epithelial surface through TNFR1 by PKC-dependent induction of CFTR activity.
These newly identified roles expand our understanding of the functional repertoire of TNF.
Developmental and in vitro studies have shown that TNFR2 signaling promotes transcription of Muc2 (15,19). Our work in adult mice found that TNF does not affect MUC2 protein expression but rather suppresses goblet cell differentiation, the primary cell type tasked with mucin biosynthesis. Therefore, changes in Muc2 expression levels in other systems may reflect changes in the number of Muc2-expressing cells as opposed to changes in transcript levels on a per-cell basis. While our studies show that loss of TNF function expands the number of ATOH1 + cells, another group has shown that TNF can stabilize ATOH1 in the context of colon cancer (49). This may suggest that TNF has different roles in the colon during cancer, when compared to the small intestine during homeostasis. Our expression analysis revealed that both TNF and TNFR2 are enriched in the adult crypt, where stem cells and progenitors are localized. We also observed increased numbers of secretory progenitors and goblet cells when TNF is lost. Taken together, these data indicate that TNF signaling functions as a regulator of goblet cell differentiation in the small intestine.
A number of studies have described TNF as a pro-differentiation factor of immune cells (50), skeletal muscle cells (51), and bone cells (52). Furthermore, NF-κB, which is a downstream effector of TNF signaling, is important for the Paneth vs. goblet cell fate decision.
Genetic ablation of NF-κB signaling in the intestine leads to a significant decrease in Paneth cell numbers, defective Paneth cell maturation, and an increase in goblet cell number (53). We identified TNF as a differentiation factor in the intestine that has no effect on Paneth cell number but does control goblet cell number. This suggests that TNF may be acting independently of NF-κB or on tangential signaling pathways in addition to NF-κB.
Both Wnt and Notch signaling have been shown to be important for cell fate decisions in the intestine (26,54) and TNF has been shown to promote Wnt and Notch signaling in certain contexts (55,56). An intriguing hypothesis is that TNF acts together with Notch to regulate goblet cell differentiation in the intestine.
Our observation that TNF controls goblet cell number is supported by several reports.
Mouse studies have shown that TNF promotes the loss of goblet cells (15,18,57), while clinical studies of IBD patient samples, which typically have elevated TNF levels, also revealed a lower number of goblet cells and partial loss of mucus barriers (22,45). Consistent with these findings, anti-TNF treatment in mouse colitis causes goblet cell hyperplasia (46). Furthermore, in non-inflamed human ileal samples we found that anti-TNF therapy results in an increased proportion of goblet cells. Like our mouse models, goblet cell hyperplasia in patients may promote increased accumulation of mucin, increased bacterial load, and prolonged gut transit. Our data highlight important considerations during anti-TNF therapy in IBD, mainly the collateral damage to surrounding non-inflamed tissue, which may exacerbate inflammation. In organoids, we discovered that the pleiotropic effects of TNF on mucin homeostasis include induction of CFTR channel activity and flow, resulting in inflation of organoid lumens. TNF regulates the CFTR-dependent inflation-collapse dynamics observed in organoids, which is an important contributor to organoid cell patterning and morphogenesis (39,59). A link between TNF and CFTR is also suggested by studies of human bronchial epithelial and HeLa cell monolayer cultures, in which TNF promotes CFTR apical localization and boosts cell chloride currents (42). Our studies are consistent with this body of work, suggesting a general mechanism of TNF modulation of CFTR activity through TNFR1 across various organs, including the intestine. Interestingly, pathologies associated with TNF or CFTR dysfunction exhibit several phenotypic similarities. For example, Cftr knockout mice exhibit increased mucin, slower gut transit, and increased bacterial load (8,24). In addition, patients with cystic fibrosis are 7 times more likely to have IBD, a disease characterized by elevated levels of TNF (60). This association implies that the phenotypes seen during cystic fibrosis may lead to IBD and thus provides clinical support for a model where increased goblet cells and mucin in patients undergoing anti-TNF therapy exacerbates inflammation and IBD.
In conclusion, the data presented here demonstrate that epithelial TNF is an essential regulator of mucin production and flux in the intestine. In addition to the canonical functions of TNF in inflammation and cell death, we reveal new roles for TNF in intestinal cell differentiation and modulation of CFTR activity. Our findings provide new considerations for the homeostatic roles of TNF in healthy tissue in developing future interventions for IBD using anti-TNF therapy. Tissue preparation for mucin staining For mucin staining, the ileum (distal third of small intestine) was isolated. Fragments of 1.5 cm length containing fecal pellets were then isolated and placed directly into methacarn fixative (60% methanol, 30% chloroform, 10% acetic acid). Fragments were fixed for 7 days rocking at room temperature. Following fixation, fragments were washed twice in methanol for 20 minutes each wash, twice in absolute ethanol for 20 minutes each, and twice in xylene for 10 minutes each. The tissue was then embedded in paraffin blocks following standard procedures and cut into 4 μm sections.

Mice
Tissue preparation for immunofluorescence and immunohistochemistry Animals were anesthetized via intraperitoneal (i.p) injection of 250 mg/kg of body weight avertin (2,2,2-tribromoethanol) and transcardially perfused with 4% paraformaldehyde (PFA) diluted in 1x phosphate buffered saline (PBS). Tissues were post-fixed for 24 h at 4 °C, and processed for paraffin embedding following standard procedures.

Tissue preparation for RNAscope
The ileum was isolated and immersion fixed in 4% PFA for 24 hours at room temperature. The tissue was then processed for paraffin embedding following standard procedures. For swelling experiments, induced organoids were treated with 1 ng/ml rTNF immediately after plating and the relative lumen area was measured at 24 h. For qPCR validation of Cftr knockout, induced organoids were grown in normal media for 4 days after plating and RNA was harvested for qPCR.
Live imaging of organoids Ileal organoids were passaged at day 7 in culture by mechanical dissociation and plated in a 96-well plate format. Organoids were plated at a density of 50 organoids per every 3 μl droplet of Matrigel. After the droplet solidified, each well was overlayed with 200 μl ENR medium supplemented with the appropriate drug treatments. The plate was immediately transferred to a Zeiss Cell Observer spinning disc confocal system with Yokogawa CSU-X1M and incubation chamber. The system was pre-equilibrated for 30 minutes to 37 °C and 5% CO2. Images were acquired every 20 minutes for 48 hours, at 5x magnification using brightfield trans-luminescence. A 90 μm z-stack was taken per region (z-step = 30 μm).

Forskolin-induced swelling assay
Organoids were passaged into a 96-well plate at a density of 50 organoids per well and allowed to grow in normal ENR media for 2 days. At the start of the experiment, old media was replaced with 100 μl of fresh media containing 1 ng/ml rTNF, 20 μM CFTR-inh172, and 5 μM GF109203X. The plate was then placed back in the incubator for 4 h at 37 ℃ 5% CO2. Afterward, 100 μl of media containing additives plus 0.8 μM forskolin was quickly added to wells to give a total well volume of 200 μl and a final concentration of 0.4 20 μM forskolin. The plate was immediately transferred to a Zeiss Cell Observer spinning disc confocal system with Yokogawa CSU-X1M and incubation chamber. The system was pre-equilibrated for 30 minutes to 37 °C and 5% CO2. Tiled images of the center plane of each well were acquired every 10 m for 120 m at 5x magnification.

Imaging of tissue sections
Image acquisition of immunostained intestinal swiss rolls used for quantitative analysis was performed on a Leica DMi8 inverted microscope at 20x magnification. Image acquisition of representative regions of immunostained slides was performed using a Zeiss LSM 900 using a 40x objective. A 4 or 7 μm z-stack was taken per region (z-step= 1 μm).

Human Intestinal Tissue
Healthy human adult intestinal tissue was obtained via an IRB-approved research protocol with the organ procurement organization for Northern California, Donor Network West, in collaboration with UCSF Viable Tissue Acquisition Lab (VITAL) Core. At the time of organ acquisition for clinical transplantation, intestinal tissue was also collected from an organ donor that consented to research. After the clinical procurement process, full length intestinal tissue was stored and transported in University of Wisconsin preservation media on ice. As tissues are from de-identified deceased individuals lacking associated health information, the study is IRB designated as non-human subjects research. Specimens from IBD and non-affected individuals were obtained from standard of care archives, facilitated through Cedars-Sinai Medical Center (CSMC) MIRIAD IRB #3358.

Statistics
Normally distributed data were analyzed using parametric two-tailed Student's t-test with Welch's correction unless otherwise noted. The non-parametric Mann-Whitney Utest was used if the data did not fit a normal distribution. Significance was taken as p < 0.05 with a confidence interval of 95%. Data are presented as mean ± SD for parametric data or as median ± interquartile range for non-parametric data. For organoid rTNF and CFTR-inh172 treatment studies, a one-way ANOVA with Dunnet's multiple comparison's test was used to compare all treatment groups to the control group. Statistical information is otherwise provided in the figure legends.