For the development of the intestinal organoid cultures, small intestinal tissues from at least 3 C57BL6/J mice with the age of 5–9 days from both sexes were pooled for stem cell isolation. Overall, 3 independent organoid cultures from 11 animals were used for the experiments described in the present study. The animals were bred and housed at the Institute for Veterinary Physiology and Biochemistry of the Justus Liebig University Giessen at an ambient temperature of 23°C–25°C and air humidity of 50%–55% on a 12:12 h light-dark cycle with free access to water and food. Animals were killed by decapitation. Experiments were approved by the named animal welfare officers of the Justus Liebig University (administrative number: 679_M) and performed according to the German and European animal welfare law.

For preparation of intestinal stem cells, 100 mM phosphate buffered saline (PBS) was used; for stem cell isolation 2 mM EDTA (ethylenediaminetetraacetic acid) was dissolved in PBS (EDTA/PBS). Organoid growth medium (IntestiCult™ Organoid Growth Medium Mouse) was obtained from StemCell Technologies (Vancouver, Canada).

Tyrode solution, which was used for Ca²⁺ imaging experiments, consisted of (in mM): 140 NaCl, 5.4 KCl, 10 HEPES [N-(2-hydroxyethyl)piperazine-N′-2-ethanesulfonic acid], 12.2 glucose, 1.25 CaCl₂, and 1 MgCl₂. For Ca²⁺ imaging experiments with Ca²⁺-free Tyrode solution, CaCl₂ was omitted. The immunofluorescence experiments were carried out in phosphate buffer (PB) containing 80 mM Na₂HPO₄ and 20 mM NaH₂PO₄. For Tyrode solution and PB the pH was set to 7.4 with 1 N HCl/NaOH.

For preparation of intestinal organoids, a modified protocol was used based on the protocols of Sato et al. (2009) and StemCell Technologies (StemCell Technologies, 2016). For one organoid culture, the small intestines from 3–5 neonatal mice were dissected, opened longitudinally and cut into 2 × 2 mm squares followed by several washing steps with ice-cold PBS (30 s sinking time of the fragments, discarding the supernatant and refilling with 10 mL ice-cold PBS) until the supernatant became clear. Ice-cold EDTA/PBS solution was used to isolate crypts (shaking on a hinged plate at 40 rpm for 20–30 min on ice until a sufficient number of single cells and crypt fragments, including stem cells, was observed).

Under sterile conditions, the EDTA/PBS was exchanged with 10 mL ice-cold PBS containing 0.1% (w/v) bovine serum albumin (BSA) (BSA/PBS) and pipetted up and down three times. After 30 s, the cell-containing supernatant was transferred into a 50 mL tube through a 70 μm cell strainer. This step was repeated with the origin pellet up to four times until a purified cell suspension was obtained. The fraction with the highest number of cells/crypt fragments (and a low contamination by cell debris) was selected, centrifuged (5 min at 280 g), mixed with 10 mL ice-cold BSA/PBS, and transferred into a fresh 15 mL tube before it was centrifuged again (5 min at 280 g). The supernatant was removed and the pellet was resuspended with 150 µL IntestiCult™ medium (tempered at 15°C–20°C, StemCell Technologies). The same volume of ice-cold Matrigel^® (#356231, growth-factor reduced, Corning, New York, United States) was added to the cell suspension and mixed carefully. 50 μL of the cell/Matrigel^® suspension were transferred into the center of a preheated 24-well plate and put into an incubator (37°C, 5% (v/v) CO₂) for approximately 30 min before each well was covered with 500 µL IntestiCult™ medium (37°C, StemCell Technologies). Medium was exchanged every 2–3 days.

For passaging of intestinal organoids, the medium was removed, the Matrigel^® drop was liquefied with ice-cold BSA/PBS and the organoids were disrupted mechanically by up and down pipetting. The cell suspension was transferred into a 15 mL tube and centrifuged at 200 g for 5 min. The pellet was resuspended with fresh IntestiCult™ medium (StemCell Technologies), the same volume of Matrigel^® was added and cells were seeded on plates as described above.

To measure the effects of the proinflammatory cytokine TNFα on intestinal organoids, the organoids were incubated with TNFα (100 ng/mL) for 1, 3 or 5 days. Untreated controls were handled in parallel for all experimental series.

Due to the findings that intestinal organoids start changing their morphology after treatment with TNFα (see Results), a long-term record over a period of 24 h was performed. The morphological changes were recorded with an Olympus IX81 microscope equipped with a MT illumination system, an Olympus FireWire camera (V2.1.30), and a climatic chamber (Evotec AUC04/B). Live imaging experiments with organoids covered with IntestiCult™ medium in a 24-well plate were run under the following conditions: 37°C, 5% (v/v) CO₂ and 50% air humidity. 100 ng/mL TNFα was added at the beginning of each experiment. The cAMP-dependent secretagogue forskolin (9 µM) was administered as viability control at the end of each experiment. Every 5 min a bright-field microscopical image was taken and transferred into a video file with 12 frames per second (Olympus xcellence rt software, version 2.0.1). In order to quantify the time course of TNFα-induced changes in morphology, relative lumen size, expressed as luminal area in % of the total organoid’s area, was analyzed every hour using ImageJ (1.53e, W. Rasband and contributors, National Institutes of Health, United States).

To elucidate the underlying mechanisms of TNFα-induced organoid swelling, changes in the relative lumen size of organoids were determined after pretreatment with TNFα (100 ng/mL) + bumetanide (100 μM, a blocker of NKCC), TNFα + indometacin [100 μM, a cyclooxygenase (COX) inhibitor], or TNFα + ethanol [0.5% (v/v), used as solvent of both drugs]. Drugs were administered to the medium and organoids were kept at 37°C and 5% (v/v) CO₂ for 24 h. Pictures were taken with a light microscope (Nikon ECLIPSE Ts2R, HAMAMATSU ORCA-spark) before and 24 h after the respective treatment.

For Ca²⁺ imaging experiments (carried out in the same setup used for the live imaging experiments), Matrigel^® was removed by gently pipetting with ice-cold BSA/PBS. Organoids were collected and centrifuged for 1 min at 1,000 rpm before 30 µL of the cell suspension was transferred to a glass bottom microwell dish (MatTek Corporation, United States) covered with poly-L-lysin (0.1 mg/L, 70,000–150,000 kDa, Cell Systems, Troisdorf, Germany). Untreated controls or TNFα-treated organoids (100 ng/mL; for 1, 3 or 5 days, respectively) were loaded at room temperature with the membrane permeable form of fura-2, fura-2 acetoxymethylester (fura-2/AM, 6 µM), combined with an equal volume of the nonionic detergent pluronic acid [20% (w/v) stock solution in DMSO] for 60 min. After washing with 500 µL Tyrode solution, the organoids were covered with 4 mL Tyrode solution (either with or without CaCl₂) and transferred to the imaging setup. The cells were excited alternately with 340 nm or 380 nm and the emission ratio was calculated (340 nm/380 nm) in different regions of interest (ROI), each with the size of one individual cell. An increase in the fura-2 ratio represents an increase in the cytosolic Ca²⁺ concentration.

After a stabilization phase of 3 min to measure the basal fura-2 ratio, the cells were stimulated with the cholinergic agonist carbachol (50 µM). In some experiments La³⁺ (1 mM), a blocker of non-selective cation channels, was administered previous to carbachol. Viability control was set at the end of each experiment by administration of cyclopiazonic acid (CPA, 5 µM), a blocker of the sarcoplasmic-endoplasmic reticulum Ca²⁺ ATPase (SERCA). A response to the respective drug was accepted when two conditions were fulfilled simultaneously: 1.) the amplitude of the change exceeded the 4-fold standard deviation of the scattering in the fura-2 ratio during the control period just prior to addition of the drug. 2.) The amplitude of the change in the fura-2 ratio exceeded an absolute value of 0.1. Cells which did not respond to carbachol or CPA were excluded from further analysis.

For immunofluorescence experiments, intestinal organoids embedded in Matrigel^® were fixed with preheated 4% (w/v) paraformaldehyde (PFA, 37°C) in PBS at room temperature for 30 min. After removing of PFA and rinsing with PBS, the organoids were suspended in BSA/PBS and collected in a 15 mL tube. The supernatant was removed after 15 min and a remaining amount of approximately 1 mL suspension including the organoids was transferred into a 1.5 mL tube, which was shortly spun down for 1 s with a microlitre centrifuge (Roth, Karlsruhe, Germany). The supernatant was removed, the pellet was resuspended with 300 µL liquid gelatin (100 g/L) and transferred into a plastic form for 30 min at 4°C to solidify before it was embedded with tissue freezing medium (Leica Biosystems, Wetzlar, Germany) and cryofixed in N₂-cooled isopentane. 4 μm thick sections were mounted on microscope slides (Superfrost^® Plus, Thermo Fisher Scientific). After rehydration in PB, the sections were incubated for 2 h in a blocking solution containing PB with 0.2% (v/v) triton-X-100, 3% (w/v) BSA, and 10% (v/v) donkey serum. Each primary antibody was dissolved in PB containing 0.1% (v/v) Triton-X-100, 1% (w/v) BSA, 0.5% (w/v) milk powder, and 1% (v/v) donkey serum. After incubation of the sections overnight at 4°C (for sources and final dilutions, see Table 1), the slices were rinsed in PB three times for 5 min each and loaded with the respective secondary antibody for 1 h at room temperature (for combinations see the legend of the respective figure). For nucleus staining and embedding, RotiFluo with DAPI (4,6-diamidio-2-phenylindoldilactate; Roth) was used. Negative staining was performed in parallel without the respective primary antibody. Pictures were taken with the fluorescence microscope Nikon 80i (Nikon, Düsseldorf, Germany). For quantitative analysis, cell counting (e.g., cells with immunoreactivity for cleaved caspase 3) was carried out in a blinded fashion.

For histological staining, intestinal organoids were fixed with the same protocol as used for immunohistochemistry, sliced in 4 µm thick sections and rehydrated in PB three times for 5 min each. For hematoxylin-eosin (HE) staining the slides were stained in Mayer’s hemalum for 7 min, followed by rinsing in running tap water for further 7 min and staining with 0.1% (v/v) eosin [addition of 0.01% (v/v) glacial acetic acid] for 5 min. For Periodic Acid Schiff (PAS) staining, the rehydrated slides were transferred into 1% (w/v) periodic acid solution for 10 min, followed by rinsing in tap water for 10 min and in distilled water for 2 min twice. Further staining with Schiff’s reagent for 15 min was followed by washing in 35°C warm tap water before the slides were stained with Mayer’s hemalum for 5 min and rinsed in running tap water for 10 min. After HE and PAS staining, the slides were transferred into staining dishes with increasing ethanol concentrations [70% < 80% < 96%, (v/v)] for a few seconds each. Finally, the slides were treated with Roti-Histol for 3 min twice and mounted with Roti-Histokit. All staining reagents were obtained from Carl Roth GmbH, Karlsruhe, Germany. Pictures were taken with the Nikon 80i light microscope.

For qPCR experiments, Matrigel^® was liquified with ice-cold BSA/PBS. The organoid suspension was centrifuged for 5 min at 280 g and supernatant was discarded. The pellet was resuspended with lysis buffer (Macherey-Nagel, Düren, Germany), homogenized with a mixer mill (for 2 min at 30 Hz), and the total RNA was extracted by using the RNA Plus kit including columns to remove genomic DNA (Macherey-Nagel). The concentration of RNA (ng/µL) in each sample and their purity (OD260/280) was determined with the Nanodrop One^© (Thermo Fisher Scientific). An equal amount of RNA was transcribed to cDNA with the High-Capacity RNA-to-cDNA Kit (Thermo Fisher Scientific). The qPCR experiments were performed with TaqMan^® Gene Expression Master Mix (Thermo Fisher Scientific). All qPCR steps were performed in accordance to manufactures’ instructions.

Overall, seven reference genes were selected either from Thermo Fisher Scientific [glyceraldehyde-3-phosphate dehydrogenase (Gapdh, Mm99999915_g1); ribosomal protein lateral stalk subunit P0 (Rplp0), Mm00725448_s1; eukaryotic translation elongation factor 2 (Eef2), Mm05700170_g1; β-glucuronidase (Gusb), Mm01197698_m1] or from PrimerDesign geNorm 12 gene kit mouse (mitochondrial ATP synthase (ATP5B) and 18s; Z-ge-SY-12-month). Gapdh and Gusb were chosen as the most stable reference genes expressed in the control and TNFα-treated organoids after analysis of inter- and intragroup variances with NormFinder (MOMA, Denmark; Andersen et al., 2004).

Final primers used for qPCR were obtained from Thermo Fisher Scientific (for detailed information see Table 2). For each primer, 3 technical replicates and at least 3 biological replicates were used (3 independent organoid passages), and each qPCR protocol was run twice. The individual efficiency for each amplicon was calculated with LinRegPCR (http://www.hartfaalcentrum.nl). A mean efficiency per amplicon group was determined and outliers (more than 5% of the mean) were excluded. The baseline corrected C_t values were used for comparison of expression levels with the efficiency-corrected ΔC_t method.

Carbachol and LaCl₃ were dissolved in aqueous stock solutions. Forskolin, bumetanide, and indometacin were dissolved in ethanol [final maximal ethanol concentration 0.5% (v/v)]. Dimethyl sulfoxide (DMSO) was used as solvent for cyclopiazonic acid [CPA; maximal DMSO concentration 0.1% (v/v)]. Recombinant TNFα from mouse (activity 20·10⁶ units/mg) was dissolved in sterile BSA/PBS. All drugs were obtained from Sigma, Taufkirchen, Germany.

In general, results are given as mean ± standard error of the mean (SEM) with number (n) of investigated organoids or cells. Relative expression ratios (Figure 7) are given as mean and 95% confidence interval. Experiments were repeated at least 3 times with 3 independent organoid passages. For comparison of two groups Student’s t-test or Mann-Whitney U-test was performed. In order to find out which test method had to be used an F-test was applied. For comparison of more than two groups, analysis of variance (ANOVA) followed by Tukey’s post hoc test was used. Linear regression was calculated with GraphPad Prism 5.01 (GraphPad Software, La Jolla, CA, United States). Statistical analysis of C_t values was performed with REST^© (REST 2009) for group-wise comparison of qPCR data (Pfaffl et al., 2002). p < 0.05 was considered to be statistically significant.

Conventional and immunofluorescent stainings were performed to validate the morphology and epithelial barrier function of the intestinal organoid culture. HE-stained organoids revealed the typical morphology of the intestinal epithelium, i.e., columnar enterocytes arranged around a central lumen, whereas PAS staining confirmed the presence of mucus producing goblet cells (Figure 1). The polarity of the epithelial cells in the organoids was confirmed by the distribution of villin, an actin-binding protein in the intestinal brush border, which was found on the apical pole of the cells (Figure 2). The establishment of epithelial cell contacts was confirmed by the expression of zonula occludens 1 (ZO-1), a tight junction-associated scaffolding protein, and claudin-3, a barrier-forming tight junction protein, which were located on the subapical or basolateral side, respectively (Figure 2). Negative staining without the respective primary antibody did not show any signal (Figure 2).

After incubation with TNFα (100 ng/mL, 24 h), the shape of the organoids shifted from the typical branched to a more rounded form with an enlarged lumen filled with fluid, cells or cell fragments (Figure 3). In addition, the epithelium flattened continuously during incubation time (see also time-lapse Supplementary Video S1).

The observed TNFα-induced “organoid swelling” (Figure 3; Supplementary Video S1) could have the following reasons:

1. Passive extension by enhanced luminal cell content due to increased number of apoptotic cells

In order to differentiate between these possibilities, immunofluorescent staining, Ca²⁺ imaging, morphometric analyses, and qPCR for expression of tight junctions and Cl⁻ transporters were performed.

Tumor necrosis factor α, administered with the same protocol as for the swelling experiments depicted in Figure 3 (100 ng/mL, exposure time 1 day), increased the amount of desquamated material within the lumen of intestinal organoids (Figure 3). To investigate if this is caused by a higher number of apoptotic cells in the lumen, sections from fixed organoids were stained with an antibody against cleaved caspase-3, a key enzyme in the apoptosis cascade. The number of caspase-3 positive cells per image, which also showed a positive nucleus staining with DAPI, was counted (Figure 4). It reached 5.21 ± 1.76 cells per image in the untreated control group (n = 14 analyzed images containing a total of 22 organoids) and was more than doubled after 1 day of TNFα incubation (11.81 ± 2.41 cells per image, n = 16 analyzed images containing a total of 25 organoids, p < 0.05, Figure 4B).

Not only the number of apoptotic cells in the lumen was increased after TNFα treatment, but also the relative lumen size of the intestinal organoids. To differentiate if this observation is a direct effect of TNFα, time-lapse video recordings from the intestinal organoids were taken over a period of 24 h under control conditions or in the presence of 100 ng/mL TNFα (Supplementary Video S1). Linear regression analysis exhibited a constant and significant increase in the relative lumen size of TNFα-treated organoids compared to untreated controls (Figure 5). The cAMP-dependent secretagogue forskolin (9 µM) was used as viability control at the end of each experiment (Supplementary Video S1). Forskolin led to an increase of the relative lumen size from 20% to 54% in the control group (n = 5 organoids), which did not differ from the increase after 24 h incubation with 100 ng/mL TNFα (from 34% to 63%, n = 7 organoids, Figure 5).

Inhibitor experiments were carried out to get insights into the mechanisms involved in the TNFα-induced organoid swelling. In the absence of any inhibitors, 100 ng/mL TNFα caused an increase in relative lumen size by 19.4 ± 2.1 percentage points after 24 h incubation (Figure 6). In order to find out whether this swelling was due to Cl⁻ secretion, the organoids were pretreated with bumetanide (100 µM), which blocks the Na⁺-K⁺-2Cl⁻ cotransporter (NKCC1) responsible for basolateral uptake of Cl⁻ to be secreted across the apical membrane. Indeed, TNFα-induced swelling was significantly diminished by about half in the presence of bumetanide (p < 0.05). To see if prostaglandins, which are known to mediate TNFα-induced secretion in human colon (Schmitz et al., 1996; Schmitz et al., 1999), might be involved in the TNFα-induced swelling, the COX inhibitor indometacin (100 µM) was administered in combination with TNFα for 24 h. Indometacin as well as ethanol, which was used as solvent control, did not affect the TNFα-induced swelling (Figure 6). Hence, the bumetanide-sensitive NKCC1 seems to be, at least in part, a target of the fluid secretion enhanced by TNFα.

qPCR was performed to analyze long-term effects of TNFα on the expression of different epithelial barrier proteins (ZO-1, claudin-2 and -3) and transporters which are involved in chloride secretion (CFTR, NKCC1). The relative expression in TNFα-treated organoids (100 ng/mL, 5 days) was compared to the expression in untreated controls (n = 3 technical and biological replicates, respectively). NKCC1 was significantly higher expressed in TNFα-treated organoids compared to the respective controls, whereas the goblet cell marker mucin 2 (Muc2) was reduced; all other genes showed neither an up- nor a downregulation (Figure 7). The Nfκb inhibitor alpha, Iκbα (Nfκbia), which is increased, e.g., after 48 h TNFα incubation in chromaffin cells (Ait-Ali et al., 2008), served as a positive control for TNFα treatment and tended to be upregulated without reaching statistical significance due to the high variance (Figure 7).

In order to confirm the expression of NKCC1 on the protein level, immunofluorescent staining of untreated and TNFα-treated organoids (100 ng/mL, 1 day and 5 days) was performed. As it is typical for the intestinal epithelium, the Na⁺-K⁺-2Cl^- cotransporter was expressed on the basolateral membrane of organoids (Figure 8). This pattern did not change in an obvious manner after 1 day (Figure 8) nor after 5 days (data not shown as it was identical to 1 day) of TNFα incubation.

The second messenger Ca²⁺ is known to be centrally involved in the induction of Cl⁻ secretion, e.g., after stimulation with acetylcholine. Experiments with fura-2 loaded organoids were carried out in order to find out whether TNFα (100 ng/mL) might interfere with this Ca²⁺ signaling. Hence, untreated or TNFα-treated organoids were stimulated with the cholinergic agonist carbachol (50 µM) and the SERCA blocker cyclopiazonic acid (CPA, 5 µM) as viability control. Basal fura-2 ratio changed in dependence on the incubation time with TNFα. After 1 day cytokine exposure, the fura-2 ratio of the organoids, kept in Ca²⁺-containing buffer, was significantly increased compared to the untreated control group (3.55 ± 0.06 versus 3.09 ± 0.07, p < 0.05, Table 3) indicating a higher basal cytosolic Ca²⁺ concentration. After 3 days incubation, the difference between the TNFα and the control group was lost and after 5 days the TNFα-treated organoids showed lower basal fura-2 ratios than the control organoids (Table 3). The carbachol-induced increase in the cytosolic Ca²⁺ concentration exhibited a different time pattern: it increased significantly from 1.17 ± 0.14 in the control group to 1.65 ± 0.07 (p < 0.05) after 3 days exposure to the cytokine, whereas no significant differences were observed at the other two time points (1 day or 5 days; Table 3). Interestingly, the carbachol-induced rise in fura-2 ratio reached in general higher values with increasing age of organoid culture (Table 3).

In order to find out whether the transient enhancement of the carbachol response by 3 days TNFα incubation might be caused by a stronger Ca²⁺ influx from the extracellular space or by an enhanced release from intracellular stores, experiments with Ca²⁺-free Tyrode solution were performed. Under these conditions, the carbachol-induced Δfura-2 ratio was significantly reduced by more than 40% in comparison to the response in Ca²⁺-containing buffer (Table 3; Figure 9). In the absence of extracellular Ca²⁺, at no time point TNFα was able to enhance the carbachol response significantly above the value of the control group (Table 3).

To investigate if Ca²⁺-permeable cation channels in the cell membrane are responsible for the enhancement of the carbachol-response, experiments with the lanthanide, La³⁺, a broad blocker of non-selective cation channels, were performed in Ca²⁺-containing buffer. Indeed, when La³⁺ (1 mM) was administered to the organoids after 3 days TNFα incubation, the carbachol response was significantly reduced by approximately 20% compared to the untreated controls and only 50% of the cells responded to the stimulation with carbachol (Table 4; Figure 9). Unexpectedly, the lanthanide alone induced a transient increase in the fura-2 signal (see Discussion). In conclusion, TNFα seems to have time-dependent effects on Ca²⁺ signaling in intestinal organoids and enhances carbachol-induced secretion via Ca²⁺-permeable cation channels in the membrane.