L-Cell Differentiation Is Induced by Bile Acids Through GPBAR1 and Paracrine GLP-1 and Serotonin Signaling

Synthetic glucagon-like peptide 1 (GLP-1) analogs and dipeptidyl peptidase 4 (DPP-4) inhibitors today constitute effective components of type 2 diabetes treatment. Both strategies potentiate the action of the GLP-1 system, leading to beneficial effects on insulin and glucagon secretion, food intake, and nutrient metabolism. Consequently, increasing the number of intestinal endocrine L cells, the main source of GLP-1, represents an attractive alternative strategy for augmenting GLP-1 bioavailability (1–3).

Differentiation of L cells is a continuous process in the intestine. Constantly renewing intestinal epithelium has the capacity to expand the population of L cells if the appropriate signals are present (4). In the crypts of the intestine, stem cells expressing Lgr5 give rise to proliferating transit-amplifying cells. Each one of these cells can become a secretory cell progenitor expressing a specific lineage marker Math1 or an absorptive cell progenitor labeled by Hes1 expression—a process controlled by Notch signaling. Next, some of these secretory cell progenitors start expressing Ngn3, which defines endocrine progenitors (5). At the next stage, new transcription factors direct the maturation toward a certain hormonal cell type (6), where NeuroD1, Arx, Foxa 1, and Foxa 2 are associated with differentiation into L cells (7–9). In addition to GLP-1, L cells secrete other hormones important for metabolic regulation, such as cholecystokinin (CCK), neurotensin, and peptide YY (PYY). Somatostatin (SST)-secreting cells and serotonin (5-hydroxytryptamine [HT])-secreting cells, which constitute most of the enteroendocrine cell population, develop as mostly separate branches.

We previously showed that differentiation of L cells can be pharmacologically increased in vitro and in vivo by a Notch inhibitor (1) and Ras homolog family member A-associated coiled-coil–containing protein kinases (ROCK) inhibitor (3) through a general increase in the proportion of secretory cells. The subsequent increase in L-cell number resulted in amplified GLP-1 secretion, which augmented insulin responses and restored glucose tolerance in a mouse model of type 2 diabetes (1,3). However, the therapeutic potential of Notch and ROCK inhibitors used to manipulate L-cell differentiation is limited due to adverse effects during long-term application of these drugs because their effects are not selective to the L-cell lineage. On the other hand, short-chain fatty acids (SCFAs), endogenous products of intestinal microbiota metabolism, selectively enhance differentiation and consequently the number of L cells (9,10). SCFAs are also effective GLP-1 secretagogues (11), and thus, we hypothesized that locally amplified GLP-1 secretion drives L-cell differentiation. Among other natural GLP-1 secretagogues, secondary bile acids, synthesized by intestinal microflora from primary bile acids, are powerful stimulants of GLP-1 release through activation of the G–protein–coupled bile acid receptor 1 (GPBAR1) (12–14). Lithocholic acid (LCA) has the highest affinity to GPBAR1 among secondary bile acids (13). In humans, plasma concentrations of bile acids are at the low micromolar range (15). Intracellularly, GPBAR1 activates the Gs signaling pathway resulting in increased cAMP, which is particularly effective in stimulating release of GLP-1 (16) and other coexpressed peptide hormones from the L cell, including PYY and neurotensin (17). Therefore, synthetic GPBAR1 agonists are promising drug candidates acting locally in the intestine to enhance GLP-1 secretion (18). In this study, the effect of GPBAR1 activation on L-cell number, GLP-1 secretion, and gene expression of L-cell lineage markers was tested in the small intestine organoid platform. The observed positive effect of GPBAR1 agonists on L-cell differentiation was dependent on GLP-1 signaling and, surprisingly, mimicked by downstream serotonin signaling conceivably through paracrine stimulation from neighboring enterochromaffin cells (19).

Nonfasted male GLU-Venus mice with transgenic expression of yellow fluorescent protein (YFP) in L cells (20), GPBAR1 knockout (KO) mice (12), their wild-type littermates, and GLP 1 receptor (GLP-1R) KO mice (21) bred on C57BL/6J background were used for organoids studies at the age of 6–8 months, and GLU-Venus female mice were used for in vivo testing at the age of 3–4 months. Mice of different age were used to reduce costs for mouse colony maintenance. We did not expect a difference in the mechanisms of intestinal cell differentiation in 3- to 8-month-old mice or significant differences between males and females. In our experience, organoids from 3- or 8-month-old males and females show similar characteristics. In all experiments, littermates from the same cage were used as controls. C57BL/6 mice were purchased from Charles River. All animal experiments were approved by the Danish Animal Inspectorate (2018-15-0201-01424).

Human jejunum fragments were obtained from Roux-en-Y gastric bypass operations from four patients with no anticancer treatment (Sydvestjysk Hospital, Esbjerg, Denmark). All patients participated with informed consent, with the research protocol approved by the National Research Ethics Committee (H-18015120).

Mice were sacrificed by cervical dislocation. Small-intestine crypts were isolated from the ileum by EDTA incubation (22) and seeded in Matrigel in advanced DMEM/F12 containing 100 units/mL penicillin/streptomycin, 10 mmol/L HEPES, 2 mmol/L GlutaMAX, supplements N2 and B27, 50 ng/mL murine epidermal growth factor (all from Life Technologies), 1 mmol/L N-acetylcysteine (Sigma-Aldrich), 100 ng/mL murine Noggin, and 500 ng/mL murine R-spondin-1 (PeproTech). For maintenance, organoids were split every 4–6 days and cultured in 48-well plates. Test compounds were added into the culture medium, and the organoids were analyzed after 48 h. By the time of the readout, most of the organoids had three to five crypts. Duodenal organoids were generated and analyzed the same way. Human organoid lines were generated and cultured as previously described (23). The experiments were performed after the first passage.

LCA (Sigma-Aldrich), L3740 (Merck-TGR5-A) (17), tropisetron, RS-39604, BIMU 8 (all from Tocris), GPR40 agonist AM-1638 (24), and GPR119 agonist AR53 (AR231453) (16,24) were first dissolved in DMSO as 1 mmol/L stock, and diluted with the culture medium to 10, 20, or 50 μmol/L for LCA and 1 μmol/L for other compounds. DMSO (0.01%) was added to control wells. After 48 h, the organoids were collected for quantitative (q)PCR analysis or fixed in 4% paraformaldehyde for GLP-1 immunostaining (for human, GPBAR1 KO, wild-type, and GLP-1R KO mouse organoids) (10) and microscopy analysis. The experiment was performed three to six times in duplicates using organoids from six mice. The L-cell percentage in organoids was assessed in whole mount preparation from a three-dimensional stack of 10-µm-thick digital sections, each containing three to eight organoids, acquired on a Zeiss confocal microscope (LSM 780) at ×20 magnification using Zen Software. Label-positive cells from each image were manually counted and expressed as the percentage of cells calculated from nuclear labeling with DAPI, analyzed with ImageJ software. The number of measurements is presented in Figure legends.

Organoids cultured in 96-well plates for 48 h were washed three times in Advanced DMEM/F12 with HEPES and l-glutamine. Medium with stimulants (L3740, LCA, or BIMU 8) and control medium (without stimuli) was added to the wells with five replicates for each condition (unless otherwise indicated) and slowly shaken on an orbital shaker after addition of the medium and before the collection. After 2 h at 37°C, the medium was collected and assayed for GLP-1 content using the GLP-1 V-Plex Meso Scale Discovery kit (Meso Scale Diagnostics). For glucose stimulation, the organoids were washed and preincubated with DMEM supplemented with 2% FBS containing no glucose and glutamine for 1 h. Then DMEM containing 18 mmol/L glucose and 2 mmol/L l-glutamine or DMEM with no stimulants (control) was added. The experiment was performed three to four times. The values were normalized to the DNA content in the wells measured with a DNA quantification kit (Sigma-Aldrich).

Total RNA was extracted from organoids using the mini RNA extraction kit (Qiagen). cDNA was synthesized using the SuperScript III kit (Invitrogen, Thermo Fisher Scientific, Waltham, MA). Real-time qPCR was performed on a real-time PCR System (Bio-Rad) using SYBR green assays. We used beta-2-microglobulin (B2M) as the endogenous control gene (10). Mouse and human primer sequences were described previously (3,10) and for Gpbar1, 5Htr3a, 5Htr3b, and 5Htr4 are presented in Supplementary Table 1. Each gene expression was tested in two to four organoid lines from different mice generating two to four templates from each line from different passages (biological replicates, n) in independent experiments. The individual measurements were averaged from duplicated qPCR wells (technical replicates).

Two groups of Glu-Venus mice were fed regular feed ground into powder for 4 days before the experiment. The body weight was monitored for potential weight losses due to the feed change. Mice with stable body weight were randomized so that one group received L3740-containing feed for 3 days, and the other continued with the same powder feed (control). Assuming that mice consumed at least 2.8 g of feed during 24 h (25), the mice received ∼60 mg/kg body wt of the compound per day (0.64 mg L3740 per 1 g mouse feed). The mice were fasted for 5 h before an oral glucose tolerance test. Glucose (2 g/kg body wt) was given as an oral gavage, and blood samples were taken before and 15, 30, 60, and 120 min after the bolus. For plasma GLP-1 measurements, blood (∼50 µL) was collected from the retro-orbital sinus from a separate cohort of L3740-treated and control mice before (0 min) and 10 min after the glucose bolus. The total GLP-1 Meso Scale Discovery assay (cat. no. K150JVC-4) was used for determination of plasma GLP-1 concentrations. Mice were euthanized, and fragments of ileum 2 cm from cecum were collected and fixed in 4% paraformaldehyde. The percentage of YFP-positive cells was calculated in six to eight transverse cryosections per mouse as described above. Blood glucose was measured with a Contour-Next Glucometer (Bayer). Area under the curve with baseline subtraction was calculated for glucose dynamics using GraphPad Prism 8 software.

Adult C57BL/6 male mice were euthanized by cervical dislocation, and fragments of intestine were excised, rinsed, and fixated by flushing with Bouin fixative (50% ethanol and 5% acetic acid in deuterium H₂O). The intestines were cut open and rolled onto a toothpick to create so-called Swiss rolls. “Rolls” were fixated 24 h in 4% paraformaldehyde and subsequently embedded in paraffin. In situ hybridization was performed on 5-µm sections using RNAscope Multiplex Fluorescent Assay (ACDBio) with the probes for Htr4 (Mm-Htr4, #408241), Htr3a (Mm-Htr3a, #411141), Lgr5 (Mm-LGR5-C3, #312171-C3), Math1 (Mm-Atoh1-C3, #408791-C3), CD133 (Mm-Prom1-C3, #412221-C3), and Ngn3 (Mm-Neurog3-C2, #422401-C2) following manufacturer’s instruction. Htr4, CD133, and Ngn3 were applied as a triple labeling. Images of staining were acquired on LSM 780 using Zen Software.

Quantitative results are presented as mean values with SEM. Comparison of two groups was done using the nonpaired two-tailed Student t test, assuming normality of data. qPCR data were analyzed using the Wilcoxon matched-pairs signed rank test for data from different organoid lines. P values of <0.05 were considered significant. Data involving more than two groups were assessed by ANOVA with Bonferroni post hoc test using GraphPad Prism 8 software.

The data sets generated during the current study are available from the corresponding author on reasonable request.

To investigate whether GPBAR1 activation could increase L-cell number, we applied LCA to organoids generated from GLU-Venus mouse ileum crypts, where L cells are labeled by transgenic YFP expression (20). The addition of 10 μmol/L LCA to the culture medium for 48 h increased the L-cell number twofold (Fig. 1A and B). Increasing the LCA concentration did not increase L-cell number further (Supplementary Fig. 1A and B); however, 20 and 50 μmol/L LCA impaired crypt formation in organoids, indicating toxicity, which was not observed with application of 10 μmol/L LCA (Supplementary Fig. 1A). To avoid potential nonspecific effects, we used L3740, a potent, synthetic, nondetergent GPBAR1 agonist (17) (Supplementary Fig. 1C). Incubation with 1 μmol/L L3740 increased the number of L cells by 2.5-fold within 48 h (Fig. 1C and D), while maintaining normal morphology of the organoids (data not shown). Immunostaining for GLP-1 confirmed that YFP-positive cells produced GLP-1 (Supplementary Fig. 1D). L cells were mostly localized to the crypt domains in the control and L3740-treated groups. We also tested the effect of L3740 on L-cell numbers in duodenal organoids and found a 1.7-fold increase in the L-cell percentage in the treated group compared with the control group (Supplementary Fig. 1E).

Next, we tested GLP-1 secretion from GPBAR1-treated organoids in response to a combination of natural stimulants of GLP-1 release—glucose and l-glutamine—and observed a 1.8-fold increase in the secretory capacity of the cultures compared with the control (Fig. 1E), which is in agreement with the observed increase in L-cell number.

Real-time qPCR analysis of specific cell markers was used as an indirect measure of representation of different cell types. We found increased expression of Gcg and Cck transcripts in L3740-treated organoids but not of other L-cell hormones such as Pyy and Nts (Fig. 1F). We tested the expression of Tph1, a marker of serotonin-secreting enterochromaffin cells, and found no difference between the control and treated group. Because the Tph1 gene encodes a key enzyme for serotonin synthesis and not the final product, we also determined the enterochromaffin cell representation and found no changes after the treatment with L3740 (Supplementary Fig. 1F). Levels of mRNA transcripts encoding the transcription factors NeuroD1 and Ngn3 associated with L-cell endocrine specification was also elevated, whereas expression of Arx, Foxa1, and Foxa2 remained unchanged (Fig. 1G). Expression of secretory cell progenitor markers Math1 and Dll1 did not change, indicating that the increase in L-cell number was modulated after the Notch-mediated secretory cell specification (26). Accordingly, the expression of Hes-1, a marker of progenitor of nonsecretory absorptive cells, enterocytes, was the same in control and L3740 groups. No changes in the expression levels of I-fabp and Itf (enterocyte and goblet cell markers, respectively) were observed, and the expression of Lyz1, a Paneth cell marker, was slightly reduced (Fig. 1G). The expression of Lgr5, a stem cell marker, was unchanged (Fig. 1G).

To directly test whether GPBAR1 is required for bile acid-induced L-cell differentiation, we generated organoids from GPBAR1 KO mice and their wild-type littermates. Exposure to L3740 did not affect the number of L cells (Fig. 1H) or the expression of Gcg, Nd1, and Ngn3 (Fig. 1I) in the absence of functional GPBAR1 signaling. Consistent with these findings, L3740 and LCA did not induce GLP-1 secretion from organoids lacking GPBAR1 (Fig. 1J). Thus, LCA and L3740 both selectively induced L-cell differentiation through mechanisms requiring functional GPBAR1 signaling.

To investigate whether GPBAR1 stimulation can amplify L-cell numbers in vivo, L3740 was administered o GLU-Venus mice for 3 days in their feed. L3740-treated mice showed a modest increase in glucose tolerance after an oral glucose challenge (Fig. 2A and B) and elevated basal GLP-1 plasma levels (0 min) and 10 min after the gavage (Fig. 2C). Body weight in the control and treated group before and after the treatment was the same (Supplementary Fig. 1G). There was no difference in feed intake between the groups (2.6 ± 0.1 vs. 2.7 ± 0.1 g/mouse/day in control and treated groups, respectively; NS). L-cell number in the distal ileum was increased (Fig. 2D and E) and accompanied by elevated expression of Gcg and Ngn3 compared with controls (Fig. 2F). This indicates that the GPBAR1 agonist L3740 stimulates formation of new endocrine-committed intestinal cells in vivo in mice.

We next investigated the effect of GPBAR1 activation in human intestinal epithelium. Human ileal organoids were treated with 1 μmol/L L3740 for 48 h. Human organoid culture requires the addition of Wnt3A and mitogen-activated protein kinase inhibitors, which interfere with normal cell differentiation, and the percentage of mature cells in human organoids consequently is much lower than in an intact intestine or in mouse organoids (23). In mouse organoids, L cells were usually found in crypt domains, whereas in human organoids, we did not observe a clear pattern for L-cell location (Fig. 3B). GPBAR1 stimulation induced a 1.6-fold increase in the L-cell number compared with control cells (i.e., similar to what was observed in the murine organoids) (Fig. 3A). Only 48% of vehicle-stimulated organoids contained L cells, and most of those had only one or two L cells regardless of the organoid size (Fig. 3B). After L3740 treatment, L cells were observed in 71% of organoids (Fig. 3C). Considering the irregular scattering of L cells, we compared the frequency distribution of L cells in the organoid population. A shift toward more cells per organoid was detected in the L3740-treated group (Fig. 3C). The expression of GCG was increased in the L3740-treated group as well as expression of NEUROD1, but we did not detect changes in NGN3 expression, probably because the overall expression of NGN3 was surprisingly low in the human organoids (Fig. 3D). These data indicate that L3740 stimulates L-cell development also in human intestinal epithelium.

We hypothesized that activation of GPBAR1 promotes L-cell differentiation by chronically increasing secretion of GLP-1. To test this, the effect of L3740 was probed in intestinal organoids from GLP-1R KO mice (21). The increase in L-cell number determined by immunostaining was completely eliminated in the absence of functional GLP-1R signaling (Fig. 4A). This indicates that the effect of GPBAR1 activation requires the endogenous GLP-1R and that L-cell number is regulated by a positive feed forward mechanism.

In an attempt to maximize the proposed effect of GLP-1R activation, we used 100 μmol/L liraglutide, a synthetic GLP-1 analog, in the organoid culture medium and assessed the L-cell number. Liraglutide increased the number of L cells (Supplementary Fig. 2A), but interpreting this effect is difficult because the peptide also had a pronounced stimulatory effect on organoid growth (Supplementary Fig. 2B).

In the intestinal epithelium, the GLP-1R is almost exclusively and highly expressed by the large population of enterochromaffin cells producing serotonin, or 5-hydroxytryptamine (5-HT) (19). Thus, we hypothesized that differentiation of L cells could be mediated through serotonin release. We stimulated L-cell differentiation in mouse organoids with L3740 in the presence of two different serotonin receptor inhibitors—tropisetron, a prototype rather broad-spectrum serotonin antagonist blocking 5-HT receptor 3 (5-HTR3) and 5-HT receptor 4 (5-HTR4) (26), and 5-HTR4 blocker RS-39604. Both drugs abolished L3740-induced L-cell differentiation (Fig. 4B). In situ hybridization demonstrated that the Htr4 is expressed in most epithelial cells of the small intestinal crypts (Fig. 4C), and Htr3, as expected (27,28), is expressed mainly in submucosal ganglia (Fig. 4C). In view of these facts and because 5-HTR4 inhibition was sufficient to inhibit the effect of L3740, we focused on the 5-HT4 receptor. Correspondingly, we found that L3740 treatment increased the expression of Ht4r in the organoids (Fig. 4D). To determine which differentiating cells in the crypt express Htr4, we performed in situ hybridization costaining for Htr4 and stem cell marker Lgr5, transit-amplifying cell marker CD133, secretory progenitor marker Math1, and endocrine progenitor marker Ngn3. We found that Htr4 was present in all these cell types (Fig. 4E). This is in accordance with the capacity of these cells to differentiate into L cells, which supports our discovery of a role for serotonin in regulating the number of L cells.

Next, we incubated organoids with the 5-HTR4 agonist BIMU 8 alone and in combination with L3740. BIMU 8 produced a similar increase in the number of L cells as L3740, and combination of the two did not result in an additive or synergistic effect (Fig. 4F). Importantly, BIMU 8 had no acute effect on GLP-1 release (Supplementary Fig. 2C). When tested in GLP-1R KO mouse organoids and GPBAR1 KO mouse organoids, BIMU 8 still increased the number of L cells (Fig. 4G and H), further supporting that the increase in L-cell numbers by L3740 was an indirect effect mediated by activation of serotonin receptors.

Based on the key role of GLP-1 signaling in the effect of GPBAR1 activation on L-cell differentiation, we tested other efficient L-cell secretagogues: a synthetic agonist of free fatty receptor 1 (FFAR1 or GPR40), AM-1638, and a lipid amide receptor GPR119 agonist, AR53. These two secretagogues were tested alone or in combination with the 5-HTR4 antagonist. Both agents alone robustly increased the number of L cells in organoids to a similar degree as L3740 (Fig. 4I), while inhibiting 5-HTR4 abolished the effect of both the FFAR1 agonist and the GPR119 agonist (Fig. 4I). Thus, these findings indicate that serotonin signaling acts downstream of GLP-1R signaling, enabling regulation of intestinal cell differentiation into L cells.

Based on studies on secondary bile acids produced by intestinal microbiota, we here propose a general mechanism through which GLP-1 secretagogues stimulate L-cell differentiation and thereby increase L-cell number in the intestinal epithelium. This mechanism involves paracrine GLP-1 stimulation of serotonin secretion and downstream activation of serotonin receptors conceivably on progenitor cells in the intestinal crypts (Supplementary Fig. 3). Activation of GPBAR1 increased L-cell differentiation in human and mouse intestinal organoids, supported by increased expression of transcription factors guiding L-cell differentiation. Interestingly, the GPBAR1 agonist L3740 only increased the expression of Gcg and Cck (which is known to be produced by L-cells to some degree), but not PYY and Nts, which are expressed in certain populations of L cells, and these results may reflect the prolonged maturation time for these populations (17). As a large fraction of enteroendocrine cell subsets express Gcg (29), GPBAR1 activation can further stimulate Gcg expression and prohormone convertase 1/3 (30), thereby converting more cells into GLP-1–producing cells. Thus, GBPAR1 modulates L-cell differentiation selectively in contrast to earlier methods to increase the number of L cells (1–3). Using GPBAR1-deficient mouse organoids, we demonstrate that the effect indeed was specific for GPBAR1 when L3740 was applied. However, other efficient GLP-1 secretagogues acting on FFAR1 (GPR40) and GPR119 mimicked the effect of L3740 on the L-cell number. This is likely due to high enrichment of GPBAR1, FFAR1, and GPR119 (20) in L cells and the ability of L cells to secrete large amounts of GLP-1 upon its Gs activation. Previously, similar effects on L-cell abundance have been shown for FFAR2 and FFAR3, receptors for other microbial metabolites, SCFAs (10).

GLP-1R has been implicated in cell proliferation and differentiation (31). As data from GLP1R-deficient mice indicate, GLP-1R signaling specifically regulates L-cell differentiation. GLP-1R is mainly expressed within intestinal intraepithelial lymphocytes, enteric neurons, and enterochromaffin cells in the intestinal epithelium, and GLP-1 stimulates release of serotonin (19). The role of serotonin in growth and maturation of intestine has previously been described, although these effects were attributed to the interaction of enterochromaffin cells with enteric neurons (32,33). Here we show that 5-HTR4 is predominantly expressed in the crypt in stem cells, supporting earlier data (34), but also in secretory and endocrine progenitor cells. Importantly, the serotonin receptor antagonist abolished the increase in L-cell number induced by GLP-1 secretagogues. In agreement, activation of serotonin signaling increased L-cell number regardless of the presence of GPBAR1 and GLP-1R, indicating that serotonin acts downstream of GLP-1.

The GPBAR1 agonist L3740 stimulated formation of new endocrine-committed intestinal cells in vivo in mice. The increased glucose tolerance and GLP-1 levels in L3740-treated mice indicates that the increase in L-cell abundance was physiologically significant. It is possible that positive influence of bile acids on energy expenditure and improved glucose metabolism (35) are partially caused by increased L-cell numbers. Together with our data from human organoids, this suggests that GPBAR1 may be a potential therapeutic target for increased endogenous GLP-1 production in humans with type 2 diabetes. The current limitation of use of GPBAR1 agonists as antidiabetic agents is gall bladder dysfunction (18,36). However, altering the intestinal microbiota in favor of microorganisms producing bile acids (37–39) to activate GPBAR1 locally appears to be an intriguing possibility.

In conclusion, bile acids may act as regulators of L-cell number in the intestinal epithelium. Here we show a specific target, GPBAR1, which increases the differentiation into L cells but does not alter other cell ratios in intestinal epithelium. We propose a model whereby GLP-1 levels contribute to control of L-cell abundance through serotonin signaling, explaining the ability of the body to adjust the L-cell numbers to compensate for changes in nutrient availability. Finally, our study demonstrates that directing the differentiation in the intestinal epithelium into specific cell types is possible and that GPBAR1 (and possibly other G-protein–coupled receptors) could be a target for modulation of L-cell number as a therapeutic strategy against type 2 diabetes.

Acknowledgments. The authors would like to thank Stine Lindberg Vedersø and Barbara Thaysen for technical assistance with the in vivo studies and Anna Sofie Husted for help with graphical design (all from the Novo Nordisk Foundation Center for Basic Metabolic Research). The authors gratefully acknowledge the support of the Core Facility for Integrated Microscopy (University of Copenhagen).

Funding. G.S. is supported by a long-term Federation of European Biochemical Societies fellowship. Work in the Reimann/Gribble laboratory is supported by the Wellcome Trust (106262/Z/14/Z, 106263/Z/14/Z) and the UK Medical Research Council (MRC_MC_UU_12012/3). The Novo Nordisk Foundation Center for Basic Metabolic Research (https://www.metabol.ku.dk) is supported by an unconditional grant from the Novo Nordisk Foundation to the University of Copenhagen (NNF10CC1016515). The study was supported by the Novo Nordisk Foundation Challenge grant NNF14OC0016798 to F.B. and the University of Copenhagen Challenge grant NNF15OC0013655 to T.W.S. This work was supported by the Swiss National Science Foundation (Sinergia CRSII5_180317/1_125487) and the Ecole Polytechnique Fédérale de Lausanne (EPFL).

Duality of Interest. No potential conflicts of interest relevant to this article were reported.

Author Contributions. M.L.L., G.S., K.L.E., C.K., and N.P. performed the research and analyzed and interpreted the experiments. M.L.L., F.B., T.W.S., and N.P. wrote the manuscript with input from all of the other authors. M.L.L., T.W.S., and N.P. conceived the study and designed experiments. B.M., F.K.K., and F.B. provided key samples and expertise. F.R., F.M.G., D.J.D., and K.S. provided essential mouse lines for the study and edited the manuscript. E.J.P.d.K., K.S., and F.B. designed studies. N.P. is the guarantor of this work and, as such, had full access to all the data in the study and takes responsibility for the integrity of the data and the accuracy of the data analysis.

Prior Presentation. Parts of this study were presented in abstract form at the 17th Copenhagen Bioscience Conference – Intestinal Organoids, Hillerød, Denmark, 29 September–2 October 2019.