Protein O-GlcNAc Modification Links Dietary and Gut Microbial Cues to the Differentiation of Enteroendocrine L Cells.

Intestinal L cells regulate a wide range of metabolic processes, and L-cell dysfunction has been implicated in the pathogenesis of obesity and diabetes. However, it is incompletely understood how luminal signals are integrated to control the development of L cells. Here we show that food availability and gut microbiota-produced short-chain fatty acids control the posttranslational modification on intracellular proteins by O-linked β-N-acetylglucosamine (O-GlcNAc) in intestinal epithelial cells. Via FOXO1 O-GlcNAcylation, O-GlcNAc transferase (OGT) suppresses expression of the lineage-specifying transcription factor Neurogenin 3 and, thus, L cell differentiation from enteroendocrine progenitors. Intestinal epithelial ablation of OGT in mice not only causes L cell hyperplasia and increased secretion of glucagon-like peptide 1 (GLP-1) but also disrupts gut microbial compositions, which notably contributes to decreased weight gain and improved glycemic control. Our results identify intestinal epithelial O-GlcNAc signaling as a brake on L cell development and function in response to nutritional and microbial cues.

INTRODUCTION

Enteroendocrine cells (EECs) are specialized epithelial cells of the gastrointestinal tract that form the largest endocrine system in the body. By sensing luminal signals, EECs produce and secret neurotransmitters and hormones to modulate a variety of metabolic functions locally and systemically, such as gut motility, nutrient intake and absorption, glucose metabolism, lipid homeostasis, and energy balance (Mellitzer et al., 2010; Moran-Ramos et al., 2012; Holst, 2013; Gribble and Reimann, 2016). Studies suggest that defects in EEC function and gut hormone secretion are associated with human obesity and diabetes (Toft-Nielsen et al., 2001; Vilsbøll et al., 2001; Faerch et al., 2015). Therefore, gut hormones are excellent therapeutic candidates for treatment of metabolic diseases. Agonists of the glucagonlike peptide 1 (GLP-1) receptor and inhibitors of dipeptidyl peptidase-4 (DPP4), which inactivates GLP-1, are widely used for effective glycemic control in diabetic patients, with demonstrated benefits of weight loss and cardioprotection as well (Cantini et al., 2016; Drucker, 2016). The gut hormone responses to bariatric procedures have been proposed to be a mechanism for weight loss and improvement in glucose metabolism after surgery (Hutch and Sandoval, 2017). Moreover, dual and triple gut hormone receptor agonists are considered next-generation therapies for metabolic diseases (Troke et al., 2014; Brandt et al., 2018). Despite extensive application of enteroendocrine hormones in the clinic, the pathophysiological regulation of EEC development and function is incompletely understood, and direct means to modulate EEC differentiation in vivo are still largely lacking.

Traditionally, EECs are classified on the basis of the hormones they produce: L cells (GLP-1 and peptide YY [PYY]), K cells (gastric inhibitory polypeptide [GIP]), I cells (cholecystokinin [CCK]), D cells (somatostatin [SST]), S cells (secretin [SCT]), and enterochromaffin (EC) cells (serotonin [5-HT]) (Latorre et al., 2016). However, substantial data from recent studies using transcriptional profiling, transgenic reporter lines, immunohistochemistry, and particularly single-cell RNA sequencing have demonstrated that there is significant coexpression of hormones and crossover between EEC subtypes (Egerod et al., 2012; Habib et al., 2012; Gribble and Reimann, 2016; Haber et al., 2017; Gehart et al., 2019). EECs differentiate from the common LGR5+ pluripotent stem cells in the base of the crypt compartment (Cheng and Leblond, 1974; Barker et al., 2007). Three basic helix-loop-helix transcription factors, including ATOH1, Neurogenin 3 (encoded by the Ngn3 gene), and NeuroD1, are expressed sequentially and function in cascades for EEC differentiation (Schonhoff et al., 2004a). ATOH1 is critical for development of all three secretory cell types (Yang et al., 2001; Shroyer et al., 2007), whereas Neurogenin 3 is specific for determination and differentiation into EECs (Jenny et al., 2002; Lee et al.,2002; Mellitzer et al., 2010). NeuroD1, a downstream target of Neurogenin 3, is important for S and I cell differentiation (Mutoh et al., 1997; Naya et al., 1997). In addition, PAX4 and PAX6 have been shown to control L, K, D, and EC cell differentiation (Larsson et al., 1998). Recently, novel general and lineage-specific regulators of EEC differentiation have also been identified by real-time single-cell transcriptional profiling (Gehart et al., 2019). Nonetheless, the upstream molecular mechanisms fine-tuning the transcriptional machinery for EEC development have not been fully elucidated.

Within the last three decades, it was discovered that thousands of cytoplasmic and nuclear proteins are modified by a single O-linked β-N-acetylglucosamine (O-GlcNAc) moiety at serine or threonine residues, termed O-GlcNAcylation (Torres and Hart, 1984; Hart et al., 2007, 2011). O-GlcNAcylation is radically different from other types of glycosylation and, analogous to phosphorylation, plays a central role in signaling pathways relevant to normal cell physiology and chronic human diseases, including cardiovascular disease, diabetes, neurodegeneration, and cancer (Bond and Hanover, 2013; Ruan et al., 2013b; Yang and Qian, 2017). The enzymes O-GlcNAc transferase (OGT) and O-GlcNAcase (OGA) mediate addition and removal of O-GlcNAc, respectively (Vocadlo, 2012; Joiner et al., 2019). O-GlcNAc signaling acts as a hormone and nutrient sensor to control many biological processes, such as gene transcription, translation, protein stability and folding, and cell signaling (Hanover et al., 2012; Ruan et al., 2012, 2013a, 2014, 2017; Liu et al., 2019). Recently, we found that, in the intestinal epithelium, levels of protein O-GlcNAcylation are reduced in patients with inflammatory bowel disease and showed that OGT is required for survival of intestinal epithelial cells and their interplay with gut microbiota (Zhao et al., 2018). Here we further determined that intestinal epithelial O-GlcNAcylation is sensitive to microbiota-derived short-chain fatty acids, controls EEC lineage differentiation and function, and regulates systemic energy and glucose metabolism.

RESULTS

Deficiency in Intestinal Epithelial OGT Promotes L Cell Differentiation and Metabolic Health in Mice

To determine whether protein O-GlcNAcylation negatively controls EEC development, we genetically deleted OGT specifically in intestinal epithelial cells (Vil-Ogt knockout [KO]) by crossing the Villin-Cre and Ogt-floxed mouse lines (Ruan et al., 2014; Zhao et al., 2018). The Ogt gene is located on the X chromosome; thus, male mice are hemizygous. Levels of Ngn3 and various gut hormone genes, including Gcg (L cells), Gip (K cells), and Cck (I cells), were elevated in the ileum and colon of Vil-Ogt KO males compared with littermate controls (Figures 1A and 1B). Segment-specific changes were also observed. For example, Pax6 and the EC cell marker Chga were only increased in the Vil-Ogt KO ileum (Figure 1A). Expression of the L cell-specific regulator Hoxb9 and the hormone Pyy was upregulated in the colon of Vil-Ogt KO mice (Figure 1B). We then specifically examined L cells and found that GLP-1+ and PYY+ cell numbers in the intestine (Figures 1C–1E) and GLP-1 levels in the serum (Figure 1F) were increased in Vil-Ogt KO males.

Figure 1.

Loss of OGT in the Intestinal Epithelium Promotes EEC Development

(A and B) Expression of EEC marker genes in the ileum (A) and colon (B) of 2-month-old WT and Vil-Ogt KO male mice (n = 4–10).

(C–E) Representative images (left) and quantified numbers (right) of GLP-1+ cells in the ileum (C, 133 and 92 crypt/villus units counted from 5 WT and 4 KO) and colon (D, 176 and 138 crypts counted from 5 WT and 5 KO) and PYY+ cells in the colon (E, 1,005 and 867 crypts counted from 5 WT and 4 KO) of 10-week-old WT and Vil-Ogt KO male mice.

(F) Serum GLP-1 levels of WT (n = 9) and Vil-Ogt KO (n = 10) male mice at the age of 8–10 weeks.

Data are represented as mean ± SEM. *p < 0.05, **p < 0.01, ***p < 0.001 by two-tailed unpaired Student’s t test.

Concomitantly, hemizygous Vil-Ogt KO male mice showed lower body weight (Figure 2A), less fat and lean mass (Figure 2B), decreased random blood glucose (Figure 2C), and enhanced glucose clearance rate during an oral glucose tolerance test (Figures 2D and 2E) compared with littermate control males. Similarly, homozygous Vil-Ogt KO females also had decreased body weight (Figure 2F) and improved glucose tolerance (Figures 2G and 2H). Collectively, these data demonstrate that OGT deficiency in the intestinal epithelium promotes enteroendocrine L cell development and improves body weight and glucose metabolism.

Figure 2.

Improved Metabolism in Vil-Ogt KO Mice

(A and B) Body weight (A) and body composition (B) of 7-week-old control (n = 6) and Vil-Ogt KO (n = 5) male mice.

(C) Random blood glucose levels of control and Vil-Ogt KO male mice at the age of 4 (n = 2-5), 8 (n = 6), and 10 (n = 6) weeks.

(D and E) Oral glucose tolerance test of 7-week-old control (n = 6) and Vil-Ogt KO (n = 4) male mice (D). Area under the curve of (D) is shown in (E).

(F) Body weight of 9-week-old control (n = 11), heterozygous (n = 4), and homozygous (n = 5) Vil-Ogt KO female mice.

(G and H) Oral glucose tolerance test of 10- to 14-week-old control (n = 11), heterozygous (n = 3), and homozygous (n = 5) Vil-Ogt KO female mice (G). Area under the curve of (G) is shown in (H).

Data are represented as mean ± SEM. *p < 0.05, **p < 0.01, ***p < 0.001 by two-tailed unpaired Student’s t test (A, B, and E), one-way ANOVA with post hoc Tukey’s test (F and H), or two-way ANOVA with post hoc Sidak’s test (C, D and G).

A Cell-Autonomous Action of O-GlcNAcylation on L Cell Development

Vil-Ogt KO mice develop intestinal damage and inflammation (Zhao et al., 2018), which may confound our findings regarding L cell development. We then performed ex vivo intestinal organoid culture and found that OGT inhibition by OSMI-1 decreased global protein O-GlcNAcylation (Ortiz-Meoz et al., 2015; Figure S1A); increased the expression of Ngn3, Hoxb9, and the L cell hormone genes Gcg and Pyy (Figure 3A); and notably enhanced GLP-1 secretion from organoids (Figure 3B).

Figure 3.

Protein O-GlcNAcylation Inhibits L Cell Development

(A and B) Ileal organoids were treated with vehicle or OSMI-1 for 2 days, and then gene expression (A) and GLP-1 secretion to the medium (B) were determined (n = 3).

(C) Schematic view of the Rosa26-rOGT locus and mating strategy to generate Vil-Cre+;R26-rOGT mice.

(D) RNA levels of rat Ogt gene expression in the ileum and colon of 4-month-old control (n = 5) and Vil-Cre+;R26-rOGT (n = 4) mice.

(E) Immunoblotting showing total protein O-GlcNAcylation, hemagglutinin (HA) tag, and OGT in the colon of control and Vil-Cre+;R26-rOGT mice.

(F and G) Representative images (F) and quantified numbers of GLP-1+ L cells (G) in the colon of 4-month-old control (756 crypts of 5 mice) and Vil-Cre+;R26-rOGT (611 crypts of 4 animals) mice.

(H) Serum GLP-1 levels of control and Vil-Cre+;R26-rOGT male mice (n = 3).

(I) Growth curve of control (n = 8) and Vil-Cre+;R26-rOGT (n = 4) male mice fed normal chow.

(J–L) Levels of blood glucose (J), blood insulin (K), and calculated HOMA-IR (L) of overnight-fasted control (n = 8) and Vil-Cre+;R26-rOGT (n = 4) male mice at the age of 4 months.

Data are represented as mean ± SEM. *p < 0.05, **p < 0.01, ***p < 0.001 by two-tailed unpaired Student’s t test.

To extend these findings in vivo, we then specifically overexpressed OGT in intestinal epithelial cells by crossing Villin-Cre with a mouse line harboring an inducible rat OGT transgene in the Rosa26 locus (Yang et al., 2020; Figure 3C). qRT-PCR showed that expression of the rat Ogt gene was significantly up-regulated in intestinal tissues (Figure 3D). Immunoblotting further confirmed OGT overexpression and elevation of global protein O-GlcNAcylation (Figure 3E). Inflammatory gene expression and intestinal epithelial permeability were comparable in control and Vil-Cre+;R26-rOGT mice (Figures S1B–S1D), suggesting that OGT overexpression did not overtly cause epithelial damage or inflammation. As expected, Vil-Cre+;R26-rOGT mice displayed reduced L cell numbers in the colon (Figures 3F and 3G) and GLP-1 levels in the serum (Figure 3H) compared with controls. Although similar body weight (Figure 3I) and fasting glucose levels (Figure 3J) were observed in controls and Vil-Cre+;R26-rOGT mice, Vil-Cre+;R26-rOGT mice had higher fasting insulin levels in the blood (Figure 3K) and showed a trending increase in homeostatic model assessment of insulin resistance (HOMA-IR, Figure 3L). These results demonstrate that OGT and protein O-GlcNAcylation negatively regulate L cell development and glucose metabolism in a cell-autonomous manner.

OGT Suppresses Ngn3 Gene Transcription via FOXO1 O-GlcNAcylation

NGN3 plays a pivotal role in controlling the commitment of pluripotent progenitor cells toward the EEC lineage (Jenny et al., 2002; Mellitzer et al., 2010). We could not detect O-GlcNAcylation of overexpressed NGN3 (Figure S2A); thus we sought to test whether OGT controls Ngn3 gene transcription. Consistent with the observation that the expression of Ngn3 and its downstream genes was upregulated in Vil-Ogt KO intestine (Figures 1A and 1B), OGT was able to suppress Ngn3 promoter activity in cultured cells (Figure 4A). Previous reports have shown that FOXO1, an O-GlcNAc-modified protein, negatively regulates Ngn3 transcription (Al-Masri et al., 2010; Bouchi et al., 2014). We sought to test whether OGT represses Ngn3 transcription via FOXO1 O-GlcNAcylation. Consistent with previous reports (Housley et al., 2008; Fardini et al., 2014), we found that OGT overexpression (Figure 4B) and specific inhibition of OGA by 6-Ac-CAS (Figure 4C) substantially increased FOXO1’s transactivation activity on insulin-responsive elements. We then mutated four known O-GlcNAc sites on FOXO1 (T314, S547, T645, and S651) to alanine (FOXO1-4A) (Housley et al., 2008). FOXO1-4A showed impairment in O-GlcNAcylation compared with wild-type FOXO1, particularly when OGA was inhibited by PUGNAc (Figure 4D). As a result, the 4A mutation diminished the transcriptional activator function of FOXO1 when co-expressed with OGT (Figure 4E).

Figure 4.

O-GlcNAcylation Modulates the Transcriptional Activity of FOXO1

(A) Transcriptional repressor of Ngn3-luciferase activity by OGT in Caco-2 cells (n = 6). RLU, relative luminescence unit.

(B and C) Transcriptional activation of FOXO1 on the insulin-responsive element (IRE) with OGT coexpression (B) or OGA inhibition (C). Luciferase activity was normalized to co-transfected β-galactosidase (β-gal) activity in HepG2 cells (n = 4).

(D) O-GlcNAcylation levels of WT- and 4A-FOXO1 expressed in 293 cells in the absence or presence of PUGNAc, an OGA inhibitor.

(E) Changes in the transcriptional activation activity of FOXO1 on the IRE-luciferase (luc) reporter after co-expression with OGT in HepG2 cells (n = 4).

(F) Transcriptional repression of Ngn3-luciferase activity by WT and 4A-FOXO1 in Caco-2 cells (n = 4–5).

(G–J) Ileal organoids were treated with or without OSMI-1 in the presence or absence of AS1842856 for 2 days (n = 4). Expression of the Nng3 (G), Gcg (H), and Pyy (I) genes and GLP-1 secretion into the medium (J) were determined. Log2 values of fold change (FC) after OSMI-1 treatment were calculated and plotted.

Data are represented as mean ± SEM. *p < 0.05, **p < 0.01, ***p < 0.001 by two-tailed unpaired Student’s t test (A and G–J), one-way ANOVA with post hoc Turkey’s (B and E) or Dunnett’s (F) test, or two-way ANOVA with post hoc Tukey’s test (C).

On the other hand, FOXO1 acted as a transcriptional repressor on the Ngn3 promoter (Figure 4F), and such inhibitory activity was absent when FOXO1 O-GlcNAcylation was ablated (Figure 4F). To test whether the effect of OGT on L cell development is dependent on FOXO1, we treated ileal organoids with the OGT inhibitor OSMI-1 in the presence or absence of a FOXO1 inhibitor, AS1842856. Regarding FOXO1’s positive target genes, such as Gopc, suppression oi transcription by OSMI-1 was abolished when AS1842856 was present (Figure S2B). Notably, the ability of OGT inhibition by OSMI-1 to promote expression of Ngn3, Gcg, and Pyy genes (Figure 4G–I) and secretion of GLP-1 (Figure 4J) was diminished when the transcriptional activity of FOXO1 was blocked by AS1842856. Together, these results suggest that FOXO1 O-GlcNAcylation mediates the transcriptional suppression of the Ngn3 gene by OGT.

OGT in the Ngn3 Lineage Controls Differentiation of L Cells

Next we sought to determine whether O-GlcNAc signaling in Ngn3+ progenitors controls L cell differentiation. To do so, we generated EEC-specific OGT KO mice using the Ngn3-Cre line (Schonhoff et al., 2004b). No changes in body weight were observed between wild-type and Ngn3-Ogt KO mice (Figure 5A). Pancreatic endocrine progenitors also express Ngn3, and deletion of OGT in islet progenitors (E.U.A., unpublished data) or insulin+ cells (Alejandro et al., 2015) induces diabetes as a result of β-cell dysfunction as early as 4 weeks of age (Figures 5B and S3A). Therefore, we analyzed Ngn3-Ogt KO mice before weaning, when they had similar body weight, slightly lower blood glucose (Figure 5A, B), and normal β-cell mass in Ngn3-Ogt KO compared with littermate controls (E.U.A., unpublished data). Immunostaining showed that GLP-1+ and PYY+ L cells (Figures 5C and 5D) and ChgA+ EC cells (Figures S3B and S3C) modestly increased their numbers in the Ngn3-Ogt KO intestine. At this age, when assessed for oral glucose tolerance, Ngn3-Ogt KO mice had higher initial blood glucose levels, probably because of defects in β-cells, but faster glucose clearance, which could be attributed to GLP-1-mediated insulin hypersensitivity (Figure 5E).

Figure 5.

Loss of OGT in the Ngn3 Lineage Promotes L cell Development

(A and B) Body weight (A) and random blood glucose (B) of WT and Ngn3-Ogt KO male mice at the indicated ages.

(C) Representative images (left) and quantified numbers (right) of GLP-1+ L cells in the colon (187 and 235 crypts counted from 4 WT and 5 KO) of 3-week-old WT and Ngn3-OGT KO male mice.

(D) Representative images (left) and quantified numbers (right) of PYY+ L cells in the colon (1,430 and 1,141 crypts counted from 4 WT and 5 KO) of 3-week-old WT and Ngn3-Ogt KO male mice.

(E) Oral glucose tolerance test of 3-week-old (n = 4–5) Ngn3-Ogt KO male mice and their littermate controls after 4 h of fasting.

(F) Ileal organoids from WT and Ngn3-Ogt KO male mice were cultured and collected to determine the expression of L cell marker genes by qRT-PCR (n = 4).

(G) Secretion of GLP-1 from WT and Ngn3-Ogt KO organoids into the medium, determined by ELISA (n = 6).

(H and I) Representative images (H) and quantified numbers (I) of GLP-1+ L cells in WT and Ngn3-Ogt KO organoids treated with vehicle or OSMI-1 (n = 3–4). See also Table S1.

Data are represented as mean ± SEM. *p < 0.05, **p < 0.01, ***p < 0.001 by two-tailed unpaired Student’s t test (C, D, F, and G) or two-way ANOVA with post hoc Sidak’s test (B and E) or Tukey’s test (I).

To eliminate potential confounding effects from Ngn3-expressing pancreas and neurons, we isolated and cultured ileal organoids from control and Ngn3-Ogt KO mice to assess L cell development. qPCR revealed no changes in the expression of markers for enterocytes (Slc5a1), goblet cells (Muc2), Paneth cells (Defa6), or tuft cells (Dclk1) (Figure S3D). Among EEC marker genes, L cell-specific Hoxb9, Gcg, and Pyy were significantly upregulated in Ngn3-Ogt KO organoids, whereas Ngn3 and genes encoding other EEC hormones were unaffected (Figures 5F and S3E). As a result, more GLP-1 peptides could be detected in the medium of Ngn3-Ogt KO organoids than in a wild-type culture (Figure 5G). OGT inhibition by OSMI-1 and Ngn3-Ogt KO significantly increased the number of GLP-1+ cells (Figures 5H and 4I; Table S1). However, OSMI-1 could not further promote L cell hyperplasia in Ngn3-Ogt KO organoids (Figures 5H and 5I), suggesting that OSMI-1 acted on Ngn3+ EEC progenitors to promote L cell development.

Protein O-GlcNAcylation and L Cell Function Are Inversely Regulated

Nutrients like glucose and fatty acids drive the hexosamine synthetic pathway and protein O-GlcNAcylation. We then sought to test whether levels of intestinal epithelial O-GlcNAcylation are regulated by food availability. Overnight fasting in mice significantly reduced the level of global protein O-GlcNAcylation in colonic epithelial cells, which was replenished by 4 h of refeeding (Figure 6A). In rodents, high-fat diet (HFD)-induced obesity impairs L cell function (Gil-Lozano et al., 2016; Richards et al., 2016; Zheng et al., 2017). However, we found that intestinal protein O-GlcNAcylation was substantially increased by HFD in mice (Figure 6B).

Figure 6.

Regulation of L Cells and Protein O-GlcNAcylation by Gut Microbiota and Diet

(A) Representative images of O-GlcNAc immunohistochemistry of the colon from ad-libitum-fed, overnight-fasted, and 4-h-refed mice (n = 3).

(B) Immunoblotting of global protein O-GlcNAcylation in the colon of mice fed normal chow (NC; n = 4) or a high-fat diet (HFD; n = 5) for 2 months.

(C) Expression of the Gcg and Pyy genes in isolated colonic epithelial cells from 2-month-old SPF and GF mice (n = 5).

(D) GLP-1 levels in the serum of 2-month-old SPF and GF mice (n = 6), determined by ELISA.

(E and F) Representative images (E) and quantification (F) of O-GlcNAc immunostaining of the colon from 2-month-old SPF and GF mice (n = 5).

(G and H) Representative images (G) and quantification (H) of O-GlcNAc immunostaining of the colon from 2-month-old mice gavaged with H2O or antibiotics (ABX) for 2 weeks (n = 6).

(I) Four-month-old mice were treated with saline or ABX for 1 week, followed by 2 doses (24 h and 4 h before sacrifice) of sodium butyrate (BUT) or SCFAs (3 g/kg body weight). Colon protein was extracted for immunoblotting with an anti-O-GlcNAc antibody.

(J) Caco-2 cells were serum-starved overnight, treated with SCFAs (8 μM each) for 24 h, and subjected to O-GlcNAc immunoblotting.

(K and L) Ileal organoids were treated with SCFAs and OSMI-1, as indicated, for 2 days. Gcg gene levels (K, n = 5–7) and GLP-1 secretion (L, n = 4–5) were determined.

Data are represented as mean ± SEM. *p < 0.05, **p < 0.01, ***p < 0.001 by two-tailed unpaired Student’s t test (C, D, F, and H) or one-way ANOVA with post hoc Dunnett’s (K and L) test.

The gut microbiota has a significant effect on host metabolism, partially through synthesis of short-chain fatty acids (SCFAs) (Tremaroli and Bäckhed, 2012; Rosenbaum et al., 2015; Knip and Siljander, 2016). L cell hyperplasia was observed in germ-free (GF) and antibiotic-treated mice and could be suppressed by microbiota colonization and SCFAs (Wichmann et al., 2013; Arora et al., 2018; Zarrinpar et al., 2018). Similarly, we observed increased levels of Gcg and Pyy RNA (Figure 6C) and serum GLP-1 in GF mice (Figure 6D) compared with age-matched, conventionally housed specific pathogen-free (SPF) mice purchased from the same vendor. Immunostaining showed that global protein O-GlcNAcylation in the cytoplasm of colonocytes was dramatically inhibited in GF mice (Figures 6E and 6F). We also treated C57BL/6 mice with multiple broad-spectrum antibiotics (ampicillin, gentamicin, metronidazole, neomycin, and vancomycin) to deplete the gut microbiota by oral gavage for 2 weeks (Hill et al., 2010; Reikvam et al., 2011). 16 s rRNA sequencing of fecal DNA confirmed that antibiotic treatment significantly reduced gut microbial diversity (Figures S4A and S4B). A substantial reduction in cytoplasmic levels of colonic O-GlcNAcylation was observed by immunostaining (Figures 6G and 6H). Moreover, oral administration of a mixture of SCFAs (acetate, propionate, and butyrate) to mice largely restored the levels of O-GlcNAcylation in colonic epithelial cells (Figures 6I and S4C). In addition, in cultured human intestinal Caco-2 cells, SCFA treatment was sufficient to raise protein O-GlcNAcylation levels (Figure 6J). These data demonstrate that dietary nutrients and gut microbiota-derived SCFAs promote epithelial O-GlcNAcylation.

To test whether SCFAs suppress L cell function via O-GlcNAcylation, we treated ileal organoids with SCFAs and were able to observe repression of Gcg gene expression (Figure 6K) and GLP-1 secretion (Figure 6L) by SCFAs. Notably, treating cells simultaneously with the OGT inhibitor OSMI-1 could modestly reverse the repressive effect of SCFAs (Figures 6K and 6L). It is worth noting that SCFAs inhibited GLP-1 secretion in wild-type and Ngn3-Ogt KO organoids (Figure S4D), indicating that OGT-independent pathways may also exist. Collectively, these results established O-GlcNAcylation as a molecular link between SCFAs and L cell function.

Microbial Disruption in Vil-Ogt KO Mice Contributes to Metabolic Improvement

Intestinal epithelial O-GlcNAcylation not only responds to microbial cues but also modulates the gut microbiota; we recently demonstrated that OGT deficiency in intestinal epithelial cells is associated with microbial dysbiosis in mice (Zhao et al., 2018). We then sought to determine whether the microbial changes in Vil-Ogt KO mice contribute to the systemic improvement in metabolism through fecal microbiota transplantation (FMT; Figure S5A). Microbiota-depleted or GF mice receiving FMT from Vil-Ogt KO mice gained significantly less body weight compared with their counterparts receiving FMT from wild-type (WT) mice (Figures 7A and 7B). GF mice colonized with Vil-Ogt KO microbiota had much less fat composition (Figure 7C) and reduced white adipose tissue (WAT) weight (Figure 7D). Food intake (Figure 7E) and locomotor activity (Figure 7F) were similar in mice receiving FMT from WT and Vil-Ogt KO mice. Mice receiving Vil-Ogt KO FMT had a trending increase in respiratory exchange rate (RER; Figures 7G and 7H), suggesting that glucose was used preferentially in these mice. Moreover, total energy expenditure, determined by heat production normalized to body weight (Figure 7I) or adjusted to body weight using analysis of covariance (ANCOVA) (Figure 7J), were increased remarkably in GF mice colonized with Vil-Ogt KO microbiota (Tschöp et al., 2011). Intriguingly, Vil-Ogt KO microbiota was able to marginally increase L cell numbers in recipient mice (Figure S5B). However, no substantial differences in random blood glucose (Figure S5C), oral glucose tolerance (Figure S5D), or insulin tolerance (Figure S5E) were observed in mice receiving FMT from WT or Vil-Ogt KO mice. These data indicate that changes in gut microbiota associated with Vil-Ogt KO are sufficient to drive L cell hyperplasia and reduced weight gain but not improvement in glucose metabolism.

Figure 7.

Gut Microbiota in Vil-Ogt KO Mice Promotes Metabolic Health

(A) 10-week-old C57BL/6 male mice were subjected to ABX gavage for 2 weeks to deplete gut microbiota and then received FMT (on days 0 and 8) from wild-type (WT) or Vil-Ogt KO mice. Body weight gain was monitored for 5 weeks (n = 10).

(B) Changes in body weight of 10-week-old GF male mice receiving FMT (on days 0 and 14) from WT or Vil-Ogt KO mice (n = 5).

(C) Body composition analysis showing percentages of fat and lean mass in mice 2 months after FMT (n = 5).

(D) Weight of intrascapular brown adipose tissue (BAT), inguinal WAT (iWAT), gonadal WAT (gWAT), liver, and quadriceps muscle in mice was determined 15 weeks after FMT (n = 5).

(E) Average daily food intake of FMT mice over 15 days (10–11 weeks after FMT, n = 5).

(F and J) Metabolic cage study of FMT mice (2 months after FMT, n = 5) showing daily locomotor activity (F), circadian RER (G), average RER during day and night (H), heat production normalized to body weight (I), and body weight-adjusted heat production (J) using ANCOVA.

Data are represented as mean –SEM. *p < 0.05, **p < 0.01, ***p < 0.001 by two-tailed unpaired Student’s t test (C, D) or two-way ANOVA with post hoc Sidak’s test (A, B, H, and J).

DISCUSSION

Enteroendocrine L cells evolve multiple sensory mechanisms to detect different classes of luminal nutrients, including carbohydrates, lipids, and proteins (Gribble and Reimann, 2016). Recently, growing evidence has demonstrated that the gut microbiota influences intestinal homeostasis, glucose metabolism, and energy balance. Here we show that microbiota-derived SCFAs drive the levels of protein O-GlcNAcylation in intestinal epithelial cells, which serve as a negative regulator of L cell development and function. Loss of intestinal epithelial OGT and protein O-GlcNAcylation in mice results in L cell hyperplasia, improved metabolism, and microbial changes that contribute to metabolic health. Mechanistically, we determined that OGT-mediated FOXO1 O-GlcNAcylation suppresses Ngn3 gene expression and inhibits differentiation of Ngn3+ EEC progenitors toward L cells in vivo.

In addition to aiding digestion and producing vitamins, the microbiota benefits the host through synthesis of SCFAs, including acetate, propionate, and butyrate (Tremaroli and Bäckhed, 2012). SCFAs bind to G-protein-coupled receptor 43 (GPR43, also known as free fatty acid receptor 2 [FFAR2]) and GPR41 (FFAR3) to regulate lipid and glucose metabolism (den Besten et al., 2013). Butyrate and propionate are also potent inhibitors of histone deacetylases (HDACs) (Kasubuchi et al., 2015). More importantly, SCFAs are preferred metabolic substrates for intestinal epithelial cells (den Besten et al., 2013); however, regulation of L cell development and function by SCFAs remains controversial. SCFAs have been shown to increase the number of L cells and induce PYY and GLP-1 secretion via GPR43 and GPR41 (Tolhurst et al., 2012; Petersen et al., 2014; Psichas et al., 2015). However, intestinal KO of GPR43 and GPR41 does not alter glucose tolerance in diabetic animals (Tang et al., 2015), indicating that other receptors or sensing mechanisms for SCFAs might be involved. On the other hand, we and others have shown that GLP-1 transcription and secretion are more than 2-fold higher in GF mice than in conventionally raised mice. Importantly, intestinal colonization of the gut microbiota increases SCFAs and suppresses L cell function (Wichmann et al., 2013; Arora et al., 2018). More recently, Larraufie et al. (2018) demonstrated that SCFAs, particularly propionate and butyrate, strongly increase PYY gene expression but slightly suppress GCG expression in human L cells. The reasons for these discrepancies among the reported results are still unclear but could be manifold: different assay systems used (cell lines versus organoids versus whole organisms, rodents versus humans), non-linear but biphasic responses to SCFA concentrations, shifting the differentiation of progenitors toward other EEC lineages (K, I, S, N, EC cells, etc.), divergent effects on L cells versus other mucosal cell types (Nøhr et al., 2013, 2015), and complex interplay of SCFAs with other microbiota-derived metabolites and signals. Further studies are required to fully elucidate the microbial control of L cell development and function.

A large body of literature has shown that inflammatory bowel disease (IBD), including ulcerative colitis (UC) and Crohn’s disease (CD), is generally associated with increased numbers of intestinal EECs and blood levels of gut hormones in humans (Worthington et al., 2018). These enteroendocrine responses are thought to be involved in tissue repair and responsible for the reduced appetite and altered gut mobility associated with IBD. However, current IBD animal models that are induced by chemicals, including 2,4,6-trinitrobenzenesulfonic acid (TNBS) and dextran sulfate sodium (DSS), and genetic alterations (such as Tcra−/− and II2−/−) cannot fully recapitulate the EEC alterations found in IBD patients (Worthington et al., 2018). Our previous study has shown that levels of OGT and protein O-GlcNAcylation are reduced in inflamed intestinal epithelium in humans and that loss of intestinal epithelial OGT leads to microbial dysbiosis and intestinal inflammation in mice (Zhao et al., 2018). In this study, we further show that OGT deficiency and associated microbial changes promote L cell differentiation and metabolic improvements. Proinflammatory cytokines, such as interferon γ (IFNγ) and tumor necrosis factor alpha (TNF-α) can act on intestinal stem/progenitors to drive alterations in EEC development. We then took advantage of organoid cultures and Vil-Cre+;R26-rOGT mice to modulate OGT levels without affecting the inflammatory gene program and were able to demonstrate that OGT directly inhibits L cell development. Taken together, these data indicate that defective protein O-GlcNAcylation can be a unifying pathogenic mechanism for microbial dysbiosis, intestinal inflammation, and EEC hyperplasia in IBD.

The Ngn3+ lineage in the intestine is essential for life because loss of EECs in mice impairs lipid absorption, alters glucose metabolism, and leads to frequent perinatal death (Mellitzer et al., 2010). On the other hand, L cell hyperplasia, hormone hyper-secretion, and receptor agonism reduce food intake and ameliorate diabetes. Consistently, improved glucose metabolism and decreased body weight were observed in Vil-Ogt KO mice, in which EEC development was boosted. However, we only found very mild metabolic phenotypes in Vil-Cre+;R26-rOGT mice, in which L cell numbers were reduced. No significant changes in body weight, glucose tolerance, insulin tolerance, or gastrointestinal motility were observed in mice fed with normal chow or an HFD (Figure 5; data not shown). This disparity could be due to the fact that L cells and GLP-1 levels were the primary targets of OGT overexpression in Vil-Cre+;R26-rOGT mice, whereas most EEC types were affected in Vil-Ogt KO mice. In fact, when gut-derived GLP-1 was genetically deleted in mice, relatively mild impairments of glucose tolerance and gastric emptying were noticed (Chambers et al., 2017; Song et al., 2019). Therefore, we do not think that the L cell hyperplasia observed in Vil-Ogt KO mice is enough to drive the substantial improvement in body weight and glucose metabolism. Microbial dysbiosis caused by Vil-Ogt KO could reduce the gain of body weight and fat when transplanted into GF mice. Thus, we speculate that GLP-1 hypersecretion and perturbed gut microbiota together result in the metabolic changes observed in Vil-Ogt KO mice. It is also possible that intestinal epithelial damage and inflammation caused by OGT deletion contribute to the evident weight loss in Vil-Ogt KO mice.

In summary, we show that O-GlcNAc signaling in intestinal epithelial cells, controlled by the gut microbiota and nutrient availability, suppresses Ngn3 transcription via FOXO1 O-GlcNAcylation, restrains EEC development, and modulates whole-body energy and glucose metabolism. Therefore, harnessing the hexosamine synthetic pathway and protein O-GlcNAcylation can be a future strategy for treatment of obesity and diabetes.

STAR★METHODS

RESOURCE AVAILABILITY

Lead Contact

Requests for further information and resources can be directed to the Lead Contact, Hai-Bin Ruan (hruan@umn.edu).

Materials Availability

Unique reagents generated in this study are available with a completed Material Transfer Agreement.

Data and Code Availability

The accession number for the 16S-seq data reported in this paper is Sequence Read Archive (SRA) repository: PRJNA644017.

EXPERIMENTAL MODEL AND SUBJECT DETAILS

Animals

Ogt-floxed mice (Shafi et al., 2000) and Rosa26-rOGT-floxed on the C57BL/6 background were kindly provided by Dr. Xiaoyong Yang (Yang et al., 2020). Villin-Cre mice (stock no. 004586) and Neurog3-Cre mice (stock no. 006333) were purchased from the Jackson Laboratory. Germ free mice on the C57BL/6 background were purchased from Taconic Biosciences together with age and sex matched C57BL/6 SPF mice as control. 5 pairs of them were euthanized immediately upon arrival for tissue collection. 10 extra germ free mice were used in fecal microbiota colonization experiments. Homozygous floxed Ogt mice were bred to Villin-Cre and Neurog3-Cre mice to generate Vil-Ogt KO and Ngn3-Ogt KO mice, respectively. Homozygous floxed Rosa26-rOGT mice were bred to Villin-Cre mice to generate Vil-rOGT Tg mice. 3-week to 4-month old mice were used for experiments. Age of mice in each experiment was specified in figure legends. All animals were kept on a 14 h: 10 h light: dark cycle in the animal facility at the University of Minnesota. Mice were group-housed unless otherwise mentioned, with free to access water and standard chow diet. All procedures involving animals were conducted within IACUC guidelines under approved protocols.

Ileal organoid and Caco-2 cell culture

Isolated ileal crypts were counted and cultured in Matrigel following a published protocol (Mahe et al., 2013). Growth medium consists of Advanced DMEM/F12 with Glutamax, HEPES, Pen/Strep, 1x N2 supplement, 1x B27 supplement, EGF (50 ng/ml), Noggin (100 ng/ml), and R-spondin conditioned media. Organoids were maintained at 37°C and medium was changed every 3 days. Cultures were split weekly by mechanical disruption of organoids. Passage 3-5 were used for experiments. Organoids were treated with 50 μM OSMI-1, SCFAs (5 mM acetate, 1 mM butyrate and propionate) (Pearce et al., 2020) and/or AS1842856 (Calbiochem, 1 μM) (Galley et al., 2019) for 2 days. Then medium was collected for GLP-1 measurement and organoids were subjected to RNA extraction or GLP-1 staining. Caco-2 cells were cultured as described previously (Natoli et al., 2012). Differentiated Caco-2 cells were serum starved overnight and treated with SCFAs (8 μM each) for 24 h before being subjected to protein extraction.

METHOD DETAILS

Metabolic assays

Body weights were monitored every week. Body composition was assessed using an EchoMRI system (Wang et al., 2018). For food intake measurement, mice were individually housed and food consumption was weighed every morning for 7 consecutive days. For the metabolic cage study, mice were first acclimated in metabolic chambers (Columbus Instruments), and then physical activity and energy expenditure were measured continuously for at least 2 days (Huang et al., 2014). For oral glucose-, and insulin-tolerance tests. 16 h fasted mice were given glucose (1.5 g/kg body weight) by gavage; 6 h fasted mice were injected with insulin (1 U/kg body weight) intraperitoneally. Tail-vein blood collected at the designated times was used to measure blood glucose level using a Contour Glucometer (Bayer). HOMA-IR was calculated as fasting insulin level x fasting glucose level. Serum insulin (EZRMI-13K) and active GLP-1 (EGLP-35K) were determined using ELISA kit (Millipore) following manufacture’s instruction. To measure GI transit time, mice were fasted overnight and given a semiliquid methyl cellulose solution containing 0.1% Trypan blue by gavage. The time for expulsion of the fist blue feces pellet was determined as gastrointestinal transit time.

Fecal microbiota colonization

Donor mice were euthanized and ceca were aseptically removed immediately. The content was diluted 1:10 in a 50% glycerol/PBS solution and frozen at −80°C. On the day of inoculation, the cecal content was further diluted 1:5 in autoclaved PBS prior to oral gavage of 0.15 mL per mouse. Prior to inoculation, recipient animals were treated with a cocktail of broad-spectrum antibiotics to deplete gut microbiota (Hill et al., 2010; Reikvam et al., 2011). Specifically, animals having free access to autoclaved food and water were subjected to oral gavage daily for 14 days with 100 ul of autoclaved water supplemented with ampicillin (2 mg/ml), gentamicin (2 mg/ml), metronidazole (2 mg/ml), neomycin (2 mg/ml), and vancomycin (1 mg/ml). The germ-free recipient mice were inoculated immediately upon arrival at the University of Minnesota animal facility, and housed in SPF environment thereafter. A second inoculation was given 2 weeks later. Metabolic assays were carried out at indicated time points after transplantation.

Short chain fatty acids (SCFAs) treatment

Mice were treated with a cocktail of broad-spectrum antibiotics to deplete gut microbiota as mentioned above. Then sodium butyrate alone (3g/kg BW) or a mixture of sodium acetate, sodium propionate and sodium butyrate (1g of each SCFA/kg BW) were dissolved into saline and given by gavage 24h and 4 h before sacrifice.

In vivo intestinal barrier function assays

Mice were fasted for 2-h and orally gavaged with fluorescein isothiocyanate (FITC)-dextran (average molecular weight: 3,000–5,000, 0.6 mg/g; Sigma) diluted in PBS. Fluorescence intensity of plasma samples was measured (excitation 492 nm/emission 520 nm) 4 h after the gavage.

16S rRNA gene sequencing

Total DNA in stool and cecal contents was extracted using the PowerFecal DNA Isolation Kit (Mo Bio). The V4 region of the bacterial 16S rRNA gene was amplified by triplicate PCR (F515/R806) using barcoded fusion primers. Samples were pooled in sets with a maximum of 96 samples in equal quantities. Paired-end sequencing of the amplicon library was performed on the Illumina MiSeq 300-bp paired-end platform at the University of Minnesota Genomics Center (Gohl et al., 2016). A multistep bioinformatics analysis was performed using the QIIME 1.9.1 software, including filtering raw fastq files for primer and adaptor dimer sequences, removing contaminating host sequences and chimeric sequences, clustering sequences into operational taxonomic units (OTUs) using the open reference OTU calling method with the green genes 16S reference, and calculating alpha and beta diversity metrics. Linear discriminant analysis (LDA) effect size (LEfSe) method was used for microbial biomarker discovery (Segata et al., 2011).

Chemicals, plasmids, transfection, and luciferase assay

OSMI-1 (Sigma, 50 μM), 6-Ac-CAS (GlycoSyn, 10 μM), PUGNAc (Toronto Research Chemicals, 10 μM) were used when indicated. The Myc-hOGT plasmid was provided by Dr. Xiaochun Yu at the University of Michigan (Chen et al., 2013). Ngn3-luc was generated by inserting a ~4.1 kb promoter of the mouse Ngn3 gene into the pGL3-basic vector. IRE-luc was generated by inserting the −1985 to −280 bp upstream of the mouse Irs2 gene that contains two insulin-response elements (IREs) into the pGL3-basic vector. pCMV6-mNgn3-Myc-DDK (#MR212139) was purchased from Origene. pCMV-FOXO1 (#12148) was purchased from Addgene. pCMV-FOXO-4A was generated with the QuikChange XL II Site-Directed Mutagenesis Kit (Agilent). HepG2 and Caco-2 cells were transfected with expression plasmids, luciferase reporters, and β-galactosidase or Renilla-luciferase using Lipofectamine 3000 (Invitrogen) or FuGENE HD (Promega). Cells were lysed and luciferase and β-galactosidase enzyme activities were measured using kits from Promega. Relative luciferase activity was determined by normalizing to β-galactosidase or Renilla-luc activity (Ruan et al., 2012; Wang et al., 2018).

Histology, immunohistochemistry, and immunofluorescence

Tissues and organoids were fixed in 10% neutral buffered formalin. Organoids were then embedded in 5% gelatin before paraffin embedding. Immunohistochemistry was carried out using Histostain-Plus 3rd Gen IHC Detection Kit (Life technologies) following manufacturer’s instruction. Antigen retrieval was performed in Citric buffer using a 2100 Retriever (Aptum Biologics). For immunofluorescence, tissue slides were blocked with 3% BSA, 0.2% TWEEN 20 in PBS, incubated with primary antibodies (1:100 to 1:200 dilution) overnight, and secondary antibodies (Alexa Fluor 488 anti-Rabbit IgG and anti-Mouse IgG, 1:400) for 1h. A Nikon system was used for fluorescence detection. For cell number quantification, 3-5 random fields per sample were captured. Stained cell specifically on the epithelium were counted. In the intestine, cell number was normalized to the number of crypt/villus units (ileum) or crypts (colon). In the organoid, cell number was normalized to nuclei number as indicated by DAPI staining. For the quantification of O-GlcNAcylation intensity in intestinal epithelium, the nucleus and cytoplasm region in the epithelium were selected and quantification was performed by FIJI.

Immunoprecipitation and Western Blot

Tissues were lysed in RIPA buffer containing proteinase inhibitors, protein phosphatase inhibitors and an OGA inhibitor. For immunoprecipitation, whole-cell lysates were incubated with and precipitated by anti-Flag beads (Millipore Sigma). Equal amounts of whole lysates or immunoprecipitation samples were electrophoresed on TGX precast gels (Bio-Rad) and transferred to nitrocellulose membrane. Membranes were incubated with primary antibodies at 4°C for overnight. Western blotting was visualized by using IRDye secondary antibodies and the Odyssey imaging system (LI-COR Biosciences).

RNA and real time PCR

Total RNA was extracted from mouse tissues and organoids using TRIzol reagent (Invitrogen). cDNA was reverse transcribed (BioRad) and amplified with SYBR Green Supermix (Bio-Rad) using a C1000 Thermal Cycler (Bio-Rad). All data were normalized to the expression of the Rplp0 gene. Primer sequences are listed in Table S2.

QUANTIFICATION AND STATISTICAL ANALYSIS

N values ≥ 3 and ≥ 4 were used for in vitro and in vivo experiments, respectively. The statistical comparisons were carried out using two-tailed unpaired Student’s t test and one-way or two-way ANOVA with indicated post hoc tests with Prism 7 (Graphpad). Differences were considered significant when p < 0.05. Results are shown as mean ± SEM. *, p < 0.05; **, p < 0.01; ***, p < 0.001.

KEY RESOURCES TABLE

REAGENT or RESOURCE	SOURCE	IDENTIFIER
Antibodies
Anti-O-GlcNAc antibody	Abcam	Cat# ab2739; RRID: AB_303264
Anti-GLP-1 antibody	Abcam	Cat# ab22625; RRID: AB_447206
Anti-ChgA antibody	Protein tech	Cat# 10529-1-AP; RRID: AB_2081122
Anti-PYY antibody	Cell Signaling Technology	Cat# 24895; RRID: AB_2798887
Anti-HA tag antibody	Cell Signaling Technology	Cat# 3724S; RRID: AB_1549585
Anti-OGT antibody	Cell Signaling Technology	Cat# 24083; RRID: AB_2716710
Anti-β-Actin antibody	Millipore Sigma	Cat# A5441; RRID: AB_476744
Anti-Flag M2 antibody	Millipore Sigma	Cat# F3165; RRID: AB_259529
Anti-Tubulin antibody	Santa Cruz	Cat# SC-8035; RRID: AB_628408
IRDye® 800CW Goat anti-Rat IgG Secondary Antibody	Li-Cor	Cat# 925-32219; RRID: AB_2721932
IRDye® 680RD Goat anti-Mouse IgG Secondary Antibody	LI-Cor	Cat# 926-68070; RRID: AB_10956588
Alexa Fluor 488 Donkey anti-Mouse Secondary Antibody	Invitrogen	Cat# R37114; RRID: AB_2556542
Alexa Fluor 488 Donkey anti-Rabbit Secondary Antibody	Invitrogen	Cat# R37118; RRID: AB_2556546
Chemicals, Peptides, and Recombinant Proteins
Anti-FLAG® M2 Affinity Gel	Millipore Sigma	A2220
TRIzol reagent	Invitrogen	Cat# 15596018
iTaq Universal SYBR Green Supermix	Bio-Rad	Cat# 172-5124
RIPA buffer	Millipore Sigma	Cat# 20-188
Trypan blue solution	Millipore Sigma	Cat# 501613349
Matrigel	Corning	354230
Advanced DMEM/F12	Thermo Scientific	Cat# 12634-010
EGF	Peprotech	CaN/At # AF-100-15
Noggin	Peprotech	Cat# 250-38-100ug
R-spondin 1 conditioned medium	In-house production	N/A
Y-27632	Millipore Sigma	Cat# Y0503
HEPES	Thermo Scientific	Cat # 15630-56
Glutamax	Thermo Scientific	Cat # 35050-038
Penicillin/Streptomycin	Thermo Scientific	Cat # 15140-122
B-27 supplement (50X)	GIBCO	Cat # 17504-044
N2 supplement (100 X)	GIBCO	Cat # 17502-048
PBS	Corning	Cat # 21040CV
Vectashield Mounting Medium with DAPI	Vector Laboratories	Cat # H-1500
Insulin	Millipore Sigma	Cat# 91077C
Ampicillin Sodium Salt	Teknova	Cat# A9510
Gentamicin Sulfate	Teknova	Cat# G3610
Metronidazole	Spectrum	Cat# M1511
Neomycin Sulfate	Amresco	Cat# 0558
Vancomycin	VWR	Cat# 0990-5G
Sodium butyrate	Millipore Sigma	Cat# ARK2161
Sodium acetate	Millipore Sigma	Cat# S1429
Sodium propionate	Millipore Sigma	Cat# P5436
AS1842856	Calbiochem	Cat# 344355
6-Ac-CAS	GlycoSyn	Cat# FC-002
PUGNAc	Toronto Research Chemicals	Cat# A157250
OSMI-1	Millipore Sigma	Cat# SML1621
FITC-Dextran	Millipore Sigma	Cat# FD4
Lipofectamine 2000 Transfection Reagent	Invitrogen	Cat# 11668019
FuGENE® HD Transfection Reagent	Promega	Cat# E2311
Critical Commercial Assays
Histostain-Plus 3rd Gen IHC Detection Kit	Life technologies	Cat# 859073
Rat/Mouse Insulin ELISA kit	Millipore Sigma	Cat# EZRMI-13K
Glucagon Like Peptide-1 (Active) ELISA kit	Millipore Sigma	Cat# EGLP-35K
PowerFecal DNA Isolation Kit	Mo Bio	Cat# 12830-50
Deposited Data
16S sequencing	This paper	BioProject ID: PRJNA644017
Experimental Models: Cell Lines
Human: HepG2	ATCC	HB-8065
Human: Caco-2	ATCC; share by Dr. Daniel A. Vallera	HTB-37
R-spondin stable cell line	from Dr. Noah Shroyer	N/A
Experimental Models: Organisms/Strains
Mouse: Ogt-floxed mice	Jackson Laboratory	004860
Mouse: Rosa26-rOGT-floxed mice	Yang et al., 2020	N/A
Mouse: Villin-Cre	Jackson Laboratory	# 004586
Mouse: Neurog3-Cre mice	Jackson Laboratory	# 006333
Mouse: Vil-Ogt KO mice	This paper	N/A
Mouse: Ngn3-Ogt KO mice	This paper	N/A
Mouse: Vil-rOGTTg mice	This paper	N/A
Mouse: SPF C57BL/6 mice	Taconic Biosciences	B6-M MPF
Mouse: Germ-free C57BL/6 mice	Taconic Biosciences	B6-M GF
Oligonucleotides
Table S2	This paper	N/A
Recombinant DNA
Myc-hOGT plasmid	Chen et al., 2013	N/A
Ngn3-luc plasmid	This paper	N/A
IRE-luc plasmid	This paper	N/A
pCMV6-mNgn3-Myc-DDK	Origene	Cat# MR212139
pCMV-FOXO1	Addgene	Cat# 12148
pCMV-FOXO-4A	This paper	N/A
β-galactosidase	Ruan et al., 2012	N/A
Renilla-luciferase	Wang et al., 2018	N/A
Software and Algorithms
Graphpad Prism 7	https://www.graphpad.com	N/A
FIJI	https://imagej.net/Fiji	N/A
Other
Contour glucometer	Bayer	Cat# 9545C
Blood glucose test strips	Bayer	Cat# 7097C