Characterization of jejunal enteroids in human obesity; a model for studying GLP-1 cells

Abstract

Objectives

Obesity and type 2 diabetes (T2D) are associated with altered secretion of enteroendocrine hormones, including the glucagon-like peptide-1 (GLP-1). The mechanisms underlying this dysregulation remain poorly understood, partly due to the rarity of enteroendocrine cells (EECs) and technical difficulties in humans. Our aim was to generate an in vitro human jejunal enteroids (HJEs) derived from individuals with obesity and T2D as a simplified intestinal epithelium in a controlled environment, in order to explore GLP-1 secretion in metabolic diseases.

Subjects and methods

HJEs were obtained from jejunum fragments sampled during gastric bypass surgery in individuals with severe obesity and normoglycemia (Ob, n = 12), prediabetes (ObPreD, n = 12), or T2D (ObD, n = 10). HJEs were characterized through gene and protein expression analyses and immunofluorescence. To promote the EEC lineage, HJEs were treated with the Notch pathway inhibitor DAPT. Active GLP-1 secretion was assessed by ELISA.

Results

HJEs were successfully generated whatever the metabolic group Ob, ObPreD, ObD, exhibiting epithelial cell lineages and expressing critical genes involved in GLP-1 cell lineage, biosynthesis, and secretion. HJEs secreted active GLP-1 in response to both low and high glucose concentrations, regardless of subject metabolic status. However, HJEs from individuals with severe obesity and T2D exhibited a reduced capacity to release GLP-1 in response to high glucose concentrations compared to those from individuals with severe obesity or with obesity and prediabetes.

Conclusions

HJEs represent a robust human-derived model to investigate EEC function and GLP-1 secretion in metabolic diseases. While HJEs retain functional GLP-1-producing cells, their secretory capacity is impaired in the presence of T2D, confirming functional endocrine alterations. These findings support the use of HJEs in preclinical studies targeting enteroendocrine dysfunction.

Introduction

Obesity and type 2 diabetes (T2D) are major global public health challenges. Obesity is associated with a high prevalence of prediabetes and is a well-established risk factor for T2D [1–3]. Although a modest weight loss (5% to 10% of initial body weight) can reduce the risk of developing T2D [4, 5]. Roux-en-Y gastric bypass is a relevant treatment option for the management of severe obesity when such weight loss is not achieved. Notably, individuals having undergone this type of bariatric surgery present diabetes remission in 30 to 63% of cases within 1–5 years [6, 7]. This surgery improves glycemic control, in part by increasing the secretion of glucagon-like peptide-1 (GLP-1) [8–11], a key hormone involved in glucose metabolism regulation. GLP-1 is an incretin enterohormone that stimulates insulin secretion in response to glucose and reduces appetite [12]. GLP-1 analogs are now widely used in the management of obesity and T2D, showing great efficacy [13]. An increasing gradient of GLP-1 cell density exists along the intestine, with higher densities in distal regions [14, 15], but GLP-1 cells are also present in the jejunum, the major site of nutrient absorption. We previously showed that, at the jejunal level, the lineage of GLP-1 cells can be altered by high-fat consumption, as well as by obesity and T2D status in humans [16, 17].

The cellular mechanisms underlying glucose-stimulation of GLP-1 secretion involve two major pathways: the glucose transport pathway mediated by the sodium-glucose cotransporter 1 (SGLT1) and the glucose sensing pathway mediated by the sweet taste receptor T1R2/T1R3 (encoded by TAS1R2 and TAS1R3). This second pathway, less studied in the gut, involves the G protein α gustducin (encoded by GNAT3). Whatever the signaling pathway, the result is an increase intracellular Ca⁺⁺ causing the release of GLP-1-containing vesicles [18–24].

We previously showed that the glucose-sensing pathway in EEC, was altered in humans with metabolic diseases and in mouse models of obesity and T2D [25]. Similar to insulin exocytosis in pancreatic β cells, exocytosis of GLP-1-containing vesicles involves proteins at the vesicle membrane (v-SNARE), composed of VAMPs (vesicle-associated membrane proteins) and proteins at the plasma membrane t-SNARE (target membrane SNARE), including SNAPs (synaptosomal-associated proteins) and syntaxins [26, 27]. To better understand the mechanisms involved in the GLP-1 secretion in obesity and T2D, EECs need to be studied in a controlled environment.

EECs are rare, constituting approximately 1% of total intestinal epithelial cells and scattered along the intestinal epithelium making them difficult to isolate and study. Enteroids, a 3D culture model derived from intestinal stem cells, closely mimic the structure and function of the intestinal epithelium. First described by Sato et al. in 2009 [28], enteroids retain the expression of the enterohormone profiles of their intestinal segment origin [29–32]. Enteroids are suitable tools to: (1) study the lineage and function of EECs because their transcriptomic profile is similar to that of EECs in vivo [33, 34], (2) analyze genes involved in nutrient sensing and exocytosis [35], and (3) promote the lineage of mucosecretory cells and EECs, using a Notch inhibitor [31, 36, 37], thus facilitating the study of GLP-1 secretion. The term human jejunal enteroids (HJEs) was chosen instead of intestinal organoid, as HJEs are obtained from adult intestinal stem cells whereas intestinal organoids are derived from induced pluripotent stem cells (iPSCs) [38].

Our aim is to better understand GLP-1 secretion in the context of severe obesity with or without T2D, using the HJEs model. Few models of HJEs derived from individuals with severe obesity have been characterized to date. Hasan et al. described differences in glucose uptake and gluconeogenesis depending on the individual status, severe obesity or lean [39]. In the present study, we generated and characterized HJEs from individuals with three distinct metabolic status: (1) severe obesity with normoglycemia, (2) severe obesity with prediabetes, and (3) severe obesity with T2D. These HJEs retained jejunum characteristics and displayed functional GLP-1 secreting EECs. We showed that HJEs from individuals with obesity and T2D exhibit a reduced capacity to secrete active GLP-1 in response to glucose.

Subjects and methods

Human individuals and jejunum sampling

The study was conducted in accordance with Helsinki Declaration, received approval from the local ethics committee (CPP Ile de France I) and was registered on the Clinical Trials.gov website NCT01454232 and NCT02292121. Informed written consent was obtained from all individuals prior to study inclusion.

Jejunum samples were collected from individuals with severe obesity at time of Roux-en-Y gastric bypass. Individuals were enrolled in the bariatric surgery program at the Nutrition and Visceral Surgery Department of the Pitié-Salpétrière Hospital (Paris, France) as described in [17, 40, 41]. Medical history, clinical and biological variables were recorded for all individuals before the gastric bypass surgery as part of routine management [41]. Importantly, individuals were managed without any specific diet (ketogenic diet or other) or any change in the antidiabetic treatment prior to surgical intervention.

Our study included 28 females and 6 males, aged 20–62 years, BMI from 36.20–58.00 kg/m² (Table 1). According to American Diabetes Association [42] and their metabolic status, subjects were stratified into three groups: (1) severe obesity with fasting normoglycemia (Ob, n = 12, BMI = 44.52 kg/m² ± 1.61, glycemia = 4.92 mmol/L ± 0.15 and Hb1Ac = 5.29% ± 0.06), (2) severe obesity with prediabetes (ObPreD, n = 12, BMI = 46.27 kg/m² ± 1.64, glycemia = 5.43 mmol/L ± 0.17 and Hb1Ac = 5.77% ± 0.18), (3) severe obesity with T2D (ObD, n = 10, BMI = 42.07 kg/m² ± 1.52, glycemia = 6.72 mmol/L ± 0.39 and Hb1Ac = 6.38% ± 0.36) (Table 1).

Table 1.

Clinical and biological characteristics of cohort individuals.

	Individuals	Ob	ObPreD	ObD
Demographic data	n	12	12	10
	Age (years)	33.92 ± 2.00	45.67 ± 2.56**	51.20 ± 2.46§§§
	Gender (F/M)	10/2	10/2	8/2
Corpulence	BMI (kg/m²)	44.52 ± 1.61	46.27 ± 1.64	42.07 ± 1.52
Glucose metabolism (fasting)	Glycemia (mmol/L)	4.92 ± 0.15	5.43 ± 0.17	6.72 ± 0.39§,££
	Insulin (µU/mL)	18.29 ± 2.03	25.20 ± 3.16	27.57 ± 7.46§§
	HbA1c (%)	5.29 ± 0.06	5.77 ± 0.18	6.38 ± 0.36
	HOMA-IR	4.07 ± 0.48	6.20 ± 0.86	7.92 ± 2.27§§
Lipid metabolism (fasting)	Triglycerides (mmol/L)	1.49 ± 0.21	1.27 ± 0.18	1.68 ± 0.27
	Cholesterol (mmol/L)	4.55 ± 0.26	5.05 ± 0.95	4.40 ± 0.35
	HDL (mmol/L)	1.045 ± 0.04	1.21 ± 0.04	1.067 ± 0.08
	LDL (mmol/L)	2.91 ± 0.23	3.28 ± 0.25	2.59 ± 033
Comorbidities	T2D diabetes (n)			10
	T2D treatment (%)			80
	T2D treatment (n)			0–4
	Without treatment (n)			2
	Mono treatment (n)			2
	Combination of treatments (n)			6
	T2D duration (year)			9.71 ± 4.25
	Dyslipidemia (%)	63.64	75	90
	Dyslipidemia treatment (%)	0	0	30
	Hypertension (%)	27.27	50	60
	Hypertension treatment (%)	18.18	16.67	60

Individuals with ObD (fasting blood glucose >7 mmol/L and/or 11.1 mmol/L, 2 h after a 75 g glucose load) were treated with oral antidiabetic drugs (metformin or other antidiabetic drugs) and/or insulin and/or GLP-1 agonists. Statistical tests were adjusted for age and sex (sex only when comparing individual age values).

HbA1c glycated hemoglobin A1, HOMA-IR Homeostatic Model Assessment for Insulin Resistance.

Ob vs ObPreD: **p < 0.01; Ob vs ObD: ^§p < 0.05, ^§§p < 0.01, ^§§§p < 0.001; ObPreD vs ObD: ^££p < 0.01.

Human jejunal enteroid culture

HJEs were generated from surgical specimens of adult jejunum. After removing the muscularis propria from jejunal sample, intestinal crypts were obtained by dissociating the mucosa in a chelating buffer according to the protocol previously published by Mahé et al. [43]. A lysis of red blood cells was added to this protocol, as the tissues of ObD individuals are highly vascularized. A total of 300 crypts were seeded in Matrigel (Corning) into 24-well tissue culture plates. During the first week, the HJEs (passage 0) are incubated in the Intesticult Organoid Growth Medium (Stem Cell) with a ROCK pathway inhibitor (Y-27632, Biotechne, 10 μM) in a 37 °C, 5% CO₂ incubator. The culture medium was changed every 2–3 days. HJEs were passaged in a 1:2 split after 1 week.

HJEs were amplified with passages every 7 to 10 days after growth. For passage, HJEs were removed from the Matrigel with Cell Recovery Solution (Corning), mechanically dissociated and replated in Matrigel in 24-well plates. HJEs (passage 1) were cultivated in a growth culture medium containing 50% primary culture medium (Advanced DMEM/F-12 (Life Technologies), 20% fetal bovine serum (Eurobio), Glutamax 1X (Life Technologies), 1% penicillin/streptomycin (Life Technologies) and 50% conditioned medium from L-WRN cell culture, secreting proliferation factors, Wnt3a, R-Spondin 3 and Noggin (CRL-3276™, ATCC®) supplemented with a ROCK pathway inhibitor (Y-27632, Biotechne, 10 μM) and a TGFBR1 inhibitor (SB431542, Biotechne, 10 μM) [36]. The plate was incubated in a 37 °C, 5% CO₂ incubator. The culture medium was changed every 2–3 days. The conditioned medium was prepared according to the protocol of Miyoshi et al. [44].

Characteristics of the individual from whom the HJE were derived and their experimental applications were described in Supplementary Tables 1 and 2. All analyses were performed in passage 1 except for active GLP-1 secretion experiments, which were performed in passage 2.

Quantigene and RT-qPCR for gene expression analyses

Gene expression analysis is performed using the Quantigene Plex Assay Kit (Thermo Fisher Scientific). HJEs were extracted with buffer and gene expression was analyzed according to the manufacturer’s instructions. Relative quantification was determined by median fluorescence intensity. For DAPT kinetic, total RNA was extracted from HJEs using TRIzol Reagent (Life Technologies) according to the manufacturer’s instructions, followed by the RNeasy Mini Kit (Qiagen). Reverse transcription (RT) was performed with 500 ng RNA using a High-Capacity cDNA Reverse Transcriptase Kit (Applied Biosystems, Thermo Fisher Scientific). Q-PCR was performed with Step One (Thermo Fisher Scientific) using SYBR Green (Eurogentec). RPLP0 was used as the reference gene. Sequences of the oligonucleotide primers used are reported in Supplementary Table 3.

Simple Western assay for protein expression

HJEs were scraped into lysis buffer (Tris HCl 20 mM pH 7.4, NaCl 150 mM, EDTA 5 mM, Triton X-100 1% v/v, sodium deoxycholate 0.5% w/v) supplemented with Protease Inhibitor Cocktail (Sigma). Protein concentrations were determined using the Protein Quantitation Kit (Interchim). Protein homogenates were analyzed by Simple Western assay (Protein Simple). The primary antibodies used were: anti-TFF3 (#Ab4407, Abcam), anti-FABP1 (#Ab7366, Abcam), anti-Chromogranin A (#Ab15160, Abcam), anti-PC1/3 (#Ab220363, Abcam) and anti-Lysozyme (#Ab108508, Abcam). Data were analyzed using Compass for SW2.5 software (ProteinSimple). Protein levels were determined by chemiluminescence signal (AUC) and normalized to total proteins (Protein Simple).

Human enteroids transparisation for immunofluorescence analyses

For whole-mount staining, HJEs were collected in 1.5 mL tubes and fixed for 30 min in 4% paraformaldehyde. Then, they were permeabilized in PBS with 0.5% Triton X-100 and 2% donkey serum for 30 min. For immunofluorescence labeling, the antibodies used were: anti-Chromogranin A (#Ab15160, Abcam), anti-Mucin 2 (#BSB6159, Diagomics), anti-Lysozyme (#Ab108508, Abcam), anti-E-cadherin (#M108, Takara), anti-Ki67 (#M7240, Dako), anti-GLP-1 (#T4057, Peninsula). Alexa 488, 546 or 647-conjugated anti-immunoglobulin G were used as secondary antibody (Thermofisher Scientific). Nuclei were stained by 4′-6-diamidino-2-phenylindole (DAPI). HJEs were embedded in a clearing solution (RapiClear 1.49, Nikon) to facilitate observation. All images were collected with a 3i upright spinning disk microscope equipped with a Yokogawa CSU-W1 spinning disk head and a multi-immersion 25X/0.8 objective. Images were acquired with a Hamamatsu Orca Flash 4.0 sCMOS camera controlled with 3i Slidebook software. Z-series optical sections were collected with a step-size of 1.1 micrometer.

For all subjects, all GLP-1 cells from 5 HJEs were quantified by 2 or 3 experimenters using Fiji software (2.14.0/1.54j). The volume of all nuclei was measured using Imaris (10.2) software. The number of cells was determined from the average volume of a cell nucleus.

Glucose-stimulated active GLP-1 secretion in human enteroids

For the active GLP-1 secretion experiments, HJEs at passage 2 were plated in 12-well plates with two domes of HJEs per well. The secretory cell linage was promoted with the growth culture medium as described above in which Notch inhibitor (N-[N-(3,5-difluorophenacetyl)-l-alanyl]-s-phenylglycine butyl ester (DAPT), Biotechne, 10 μM) [36] was added for 96 h. The growth culture medium supplemented with DAPT, was changed every 2 days.

For the secretion experiment, HJEs were starved during 1 h at 37 °C in Hanks Buffered Salt Solution (HBSS) (Life Technologies) supplemented with 10 mM HEPES, 0.1% fatty acid-free bovine serum albumin, pH 7.4. Then, HJEs were incubated in HBSS low glucose (5 mM) or high glucose (50 mM) with DPP-4 inhibitor (Millipore) for 2 h in 37 °C, 5% CO₂ incubator. A mixture of 10 µM forskolin (F) (Sigma) and 10 µM 3-Isobutyl-1-methylxanthine or IBMX (I) (Sigma) was used to amplify cAMP intracellular concentration. Supernatants were collected for cytotoxicity assay (LDH Cytotoxicity Detection Kit, Sigma) and active GLP-1 was determined by High Sensitivity GLP-1 active Chemiluminescent ELISA Kit (Millipore). DNA was extracted from HJEs with Nucleospin Tissue Kit (Macherey Nagel) and quantified using the Nanodrop Spectrophotometer (Thermofisher) to normalize GLP-1 release.

Statistical analyses

Data are expressed as mean ± standard error of the mean (SEM) unless otherwise indicated. We used R software for all statistical analysis. All tests were performed using linear model fitting by adjusting for age and sex. When using binary features, generalized models (GLM) were used with binomial functions. The relevant p values were obtained using ANOVA and post hoc Tukey’s multiple comparisons tests on the models. GLM were computed using glmer and ANOVA functions in lme4 and car R package. p < 0.05 was considered statistically significant.

Results

Successful generation of enteroids from jejunum of individuals with severe obesity

We investigated whether we could develop HJEs from individuals with graded severity metabolic pathology, severe obesity without diabetes (Ob), severe obesity with prediabetes (ObPreD) and severe obesity with T2D (ObD) (Table 1).

The intestinal crypts were prepared from samples taken at the time of RYGBP in individuals with severe obesity. From these jejunal crypts, we generated HJEs after 10 days in Matrigel culture. The resulting structures exhibited proliferative cells and joined epithelial cells, as demonstrated by Ki67 staining and E-Cadherin labeling, a marker of adherens cell-cell junction (Fig. 1A). HJEs also displayed differentiated cells, Paneth cells (Lysozyme labeling), mucosecreting cells (Mucin 2 labeling) and EECs (Chromogranin A labeling) (Fig. 1A). We measured protein expression of different epithelial cell type markers in HJEs from Ob (n = 11), ObPreD (n = 10) and ObD individuals (n = 10). Figure 1B showed a representative reconstructed image of a western in capillary. We quantified various cell markers and showed that severe obesity did not prevent their expression in HJEs. However, the expression levels of FABP1 (liver fatty acid-binding protein), Lysozyme, TFF3 (Trefoil Factor 3), and CGA (Chromogranin A) were not significantly altered by increasing severity of metabolic pathology (Fig. 1C). Furthermore, neither severe obesity nor associated metabolic complications such as prediabetes or T2D impaired the capacity of enteroids to produce diverse epithelial cell types, including enterocytes, mucus-secreting cells, Paneth cells, and EECs.

Fig. 1. Cell composition of jejunal enteroids from individuals with severe obesity.

A Representative immunofluorescence images of cellular composition of HJEs (Ob, n = 2) cultured for 10 days. Chromogranin A (CGA), mucin 2 (MUC2), lysozyme (LYZ), E-cadherin (ECAD) labeling. Ki67 labeling indicates cell proliferation. Nuclei are visualized using DAPI. Scale bar: 100 μm. B Total protein (left), Fatty acid binding protein 1 (FABP1), LYZ, Trefoil factor 3 (TFF3) and CGA (right) levels were analyzed by Simple Western assay. Reconstructed images are displayed, based on the area under the curve from chemiluminescence signal obtained for 2 individuals in each group. C Quantification of FABP1, LYZ, TFF3 and CGA protein levels. Areas under the curve from chemiluminescence results are normalized to respective total protein levels, and expressed as ratio values normalized to the mean of Ob. Ob (n = 11, blue circle), ObPreD (n = 10, orange circle) and ObD (n = 10, green circle) individuals.

Characterization of enteroendocrine and GLP-1 cells in jejunal enteroids from individuals with severe obesity

We next focused our analysis on EECs in HJEs (Fig. 2). The expression of two housekeeping genes, HPTR1 and RPLP0, was evaluated in relation to the phenotypic characteristics of patients. Statistical analysis revealed no significant differences in RPLP0 expression with respect to metabolic status, age, or sex. In contrast, HPTR1 expression varied significantly with patient age (p = 0.020) and showed a trend with metabolic status (p = 0.066) and sex (p = 0.056). These results suggest that HPTR1 expression levels may be influenced by patient characteristics. Therefore, RPLP0 was selected as the normalization reference gene. The gene expression analysis of transcription factors involved in EEC lineage showed that early transcription factors (NEUROG3 and PAX4) and late transcription factors (PAX6, FOXA1 and FOXA2) were significantly expressed indicating that EEC lineage is maintained in these HJEs (Fig. 2A). However, the presence of aggravated metabolic condition (e.g., prediabetes or T2D), did not alter the expression of this transcription factor, suggesting that the EEC lineage remains preserved in HJEs from individuals with severe metabolic deterioration (Fig. 2A). Gene expression of Chromogranin A and enterohormones was analyzed as marker of differentiated EECs (Fig. 2B, C). GIP, GCG (gene encoding GLP-1 in intestine) and PYY are expressed in these HJEs and it is noteworthy that the in vivo expression gradient of these enterohormones is also preserved, i.e., with a decreasing expression pattern of GIP, GCG and PYY in jejunum (Fig. 2C, upper left panel). Again, gene expression of these various EEC markers is not altered by the aggravated metabolic conditions.

Fig. 2. Enteroendocrine cells in jejunal enteroids from individual with severe obesity.

A Expression of genes encoding transcription factors involved in EEC differentiation. B CHGA gene expression. Results are given as fold expression normalized to Ob individuals. RNA levels are normalized to RPLP0. C Gene expression analysis of enterohormones. The gradient of enterohormone expression is maintained (top left). Gene expression of GIP (top right), GCG (bottom left) and PYY (bottom right) are normalized to subjects with obesity and without diabetes (Ob). Ob (n = 6, blue circle), ObPreD (n = 5, orange circle) and ObD (n = 7, green circle) individuals. ***p < 0.001 and ****p < 0.0001.

We then focused on GLP-1 cells (Fig. 3) and analyzed the presence of the prohormone convertase 1/3 (PC1/3). PC1/3 encoded by PCSK1 gene, is involved in the post-translational process essential for the cleavage of proglucagon into GLP-1 in the intestine. We showed that PC1/3 is expressed in HJEs from individuals with severe obesity both at mRNA (Fig. 3A) and protein level (Fig. 3B). We next investigated the GLP-1 cell density in HJEs. We showed that HJEs displayed GLP-1 cells (Fig. 3C upper panels) and we determined their abundance (Fig. 3C, lower left panel). Due to the heterogeneity in HJE size, the total number of cells per HJE was estimated by calculating the average nuclear volume within each HJE (Fig. 3C, lower middle panel). The number of GLP-1–positive cells was then normalized per 1000 epithelial cells. On average, we counted approximately 2 GLP-1–positive cells per 1000 epithelial cells, corresponding to about 0.2% (Fig. 3C, lower right panel).

Fig. 3. GLP-1 cells in jejunal enteroids from individuals with severe obesity.

A PCSK1 gene expression, a GLP-1 cell marker. RNA level is normalized to RPLP0. Result is given as fold expression normalized to individuals with obesity without diabetes (Ob). Ob (n = 6, blue circle), ObPreD (n = 5, orange circle) and ObD (n = 7, green circle) individuals. B Total protein and PC1/3 levels were quantified by Simple Western assay. Reconstructed images are displayed for two individuals in each group. Results are normalized to respective total protein levels, and expressed as ratio values normalized to the mean of Ob individuals (down right). Ob (n = 11, blue circle), ObPreD (n = 10, orange circle) and ObD (n = 10, green circle) individuals. Molecular weight markers in kDa. C Representative Z-stack projections of 36 optical slices covering the entire HJE. The representative image shows the distribution of GLP-1 cells (green) in HJE, with nuclei labeled with DAPI (blue) (top). The number of GLP-1 cells per enteroid (down left), the nuclei volume per enteroid (down middle) and the number of GLP-1 cells per 1000 cells in enteroids (down right). For each experiment, 5 HJEs were analyzed. Quantifications were performed by 3 different experimenters. Ob (n = 6, blue circle), ObPreD (n = 9, orange circle) and ObD (n = 6, green circle) individuals. Scale bar: 50 µm.

We found that GLP-1 cell differentiation (GCG and PCSK1 gene expression), GLP-1 biosynthesis (PC1/3 protein expression and GLP-1 labeled cells) markers are all expressed in HJEs from individuals with severe obesity and prediabetes and T2D did not alter GLP-1 cell markers and GLP-1 cell density.

We then explored the potential of these HJEs to secrete GLP-1 in response to glucose and evaluated the expression of genes involved in glucose transport (Fig. 4A), in glucose sensing (Fig. 4B) and in exocytosis (Fig. 4C). We found that genes necessary to GLP-1 secretion via glucose transport pathway or via glucose sensing pathway and genes necessary to GLP-1 granule exocytosis process are expressed in these HJEs. The presence of severe obesity did not prevent the expression of genes involved in the different steps of GLP-1 secretion in response to glucose in HJEs. We observed that the only gene with a decreased expression in HJEs from individuals with aggravated pathology prediabetes, was GNAT3, encoding the small α subunit of G protein gustducin, a key protein involved in the sweet taste detection in salivary papilla and in EECs.

Fig. 4. Expression of genes involved in glucose sensing and GLP-1 exocytosis in jejunal enteroids from individuals with severe obesity.

Gene expression encoding A glucose transporters and potassium channels, B the sweet taste transduction pathway, C GLP-1 exocytosis. RNA level is normalized to RPLP0. Results are given as fold expression normalized to Ob individuals. Ob (n = 6, blue circle), ObPreD (n = 5, orange circle) and ObD (n = 7, green circle) individuals. *p < 0.05. Cotransporter sodium/glucose SGLT1 (SLC5A1), Glucose transporter GLUT2 (SLC2A2), Fructose transporter GLUT5 (SLC2A5), Voltage dependent K⁺ channel α subunit (KCNH2), ATP sensitive K⁺ channel (KCNJ11). G protein α Gustducin (GNAT3), IP3 Receptor (ITPR3), Ca⁺⁺-activated channel (TRPM5), Voltage-dependent Ca⁺⁺ channel subunit α1A1 (CACNA1A). t-SNARE plasma membrane proteins: Syntaxin 1A (STX1A) and Synaptosome associated protein 25 (SNAP25). v-SNARE vesicle membrane proteins: Synatptotagmin-7 (SYT7), Vesicule associated membrane protein 1 (VAMP1), Unc-13 Homolog D (UNC13D), Syntaxin binding protein 1 (STXBP1). Vesicular fusion protein: N-Ethylmaleimide Sensitive Factor, Vesicle Fusing ATPase (NSF).

Functionality of GLP-1 cells in jejunal enteroids from individuals with severe obesity

Before assessing the functionality of GLP-1 secreting cells, we promoted secretory cell lineage in HJEs using a Notch inhibitor (DAPT) as previously reported [31, 36, 37]. After 96 h of treatment with 10 µM DAPT, we observed a protein expression increase of EEC marker CGA and GLP-1 cell marker PC1/3 (Supplementary Fig. 1A, B). Gene expression of GCG, CHGA and PCSK1was also increased by the addition of DAPT during kinetics (Supplementary Fig. 1C). Thus, the DAPT was added to the growth culture medium for the GLP-1 secretion assays only. In HJEs derived from individuals, regardless of their metabolic status, increasing the glucose concentration from basal (5 mM) to high (50 mM) significantly stimulated active GLP-1 secretion, both in absolute and normalized values (Fig. 5A, B). However, the fold change of glucose-stimulated active GLP-1 secretion was lower in HJEs from ObD compared to the Ob (Fig. 5C).

Fig. 5. GLP-1 secretion in jejunal enteroids from individuals with severe obesity.

Enteroids were incubated for 2 h with low (5 mM) or high (50 mM) glucose concentration in absence or presence of the mix 10 µM Forskolin/10 µM IBMX (F/I) as positive control. A Active GLP-1 secretion in response to glucose. B Active GLP-1 secretion was normalized to 5 mM glucose for each individual. C Active GLP-1 secretion is calculated as fold change between 50 mM and 5 mM glucose. Ob (n = 7, blue circle), ObPreD (n = 7, orange circle) and ObD (n = 8, green circle) individuals. *p < 0.05, **p < 0.01, ***p < 0.001.

The elevation of intracellular cAMP by the F/I mix amplified active GLP-1 secretion in response to low or high glucose in HJEs derived from Ob individuals, irrespective of metabolic status (Fig. 5A, B).

We noticed that treatments with basal or high glucose concentration, with or without the F/I mix did not increase the cell mortality measured by LDH activity (Supplementary Fig. 2).

In summary, HJEs secrete active GLP-1 at both low and high glucose concentrations, regardless of the metabolic status of individuals with obesity. However, HJEs from ObD exhibited a reduced capacity to secrete active GLP-1 at high glucose concentration (Fig. 5A–C).

Discussion

We successfully generated and characterized here HJEs from individuals with severe obesity with or without prediabetes or T2D in order to study GLP-1 cells. HJEs were derived from jejunal crypts in contrast to intestinal organoids derived from induced pluripotent stem cells (iPSCs), that display a transcriptomic profile similar to that of a human fetal intestine [45]. The originality of our model lies in modeling the jejunal epithelium, a key site of nutrient absorption, using human jejunum samples obtained during bariatric surgery. This HJE model enables the investigation of the endocrine cell characteristics and GLP-1 secretion with deterioration of metabolic conditions.

To generate HJEs, we took advantage of the bariatric surgery model with a cohort of 34 clinical characterized individuals divided into three groups according to the presence or absence of prediabetes or T2D. First, severe obesity with prediabetes or T2D do not prevent the generation of HJEs and we demonstrated that these HJEs feature the cell types of an in vivo intestinal epithelium, with enterocytes, mucosecretory cells, Paneth cells and EECs, with GLP-1 cells in a controlled environment. Second, despite the fact that obesity with prediabetes or T2D has little effect on the general characteristics of these HJEs, we also demonstrated an impairment in GLP-1 secretion in HJEs obtained from jejunum of subjects with obesity and T2D, confirming functional endocrine alteration.

In order to minimize the risk of phenotypic drift in HJEs derived from patient with obesity and/or T2D, we performed the experiments in HJEs at passage 1 and at passage 2 only for secretion experiments. Although one study described phenotypic preservation of HJEs after numerous passages [39], other reports have described important alterations with extended culture. For example, decreased expression of SGLT1, SLC2A2, CHGA, and GCG has been described in human duodenal organoids and mouse enteroids [46]. In addition, culture conditions accounted for up to 60% of the variance between transcriptomes of crypts and 2-week organoids [47], and an epigenetic drift was observed between mouse enteroids cultured for 3 vs. 20 weeks [48]. Numerous studies show that enteroids retain the characteristics of the tissues from which they are derived. We showed that HJEs from individuals with severe obesity displays in particular EECs, L-cells expressing GCG and PYY and K-cells expressing GIP, with a gradient of enterohormone expression as seen [31, 32, 49]. However, we did not analyze a co-localization of GLP-1 and PYY in EEC as it has been suggested in some in vivo studies [50, 51].

In our earlier studies, we reported reduced GLP-1 cell density, GLP-1-cell lineage markers (NEUROG3, PAX4, PAX6, NKX2.2, NEUROD1, ISL1, FOXA1, FOXA2) and proglucagon processing (PCSK1, GCG) in EECs from T2D patients [17]. However, in this study, we found no difference in GLP-1 cell density nor in NEUROG3, PAX4, PAX6, ISL1, FOXA1, FOXA2), and proglucagon processing (PCSK1) or hormone (GCG, GIP, PYY) or CEE marker (CHGA). Zietek et al. showed that gene expression of several genes could be lost between HJEs passage such as SGLT1, CHGA, GLUT2, GCG [46]. The authors suggest that enteroid culture conditions impact the gene expression after the first passage [46]. In our model, proper gene expression could be measured but with no impact of T2D. We can hypothesize that the loss of T2D effect on gene expression could be due the culture conditions. Indeed, enteroids were grown in the same medium whatever the patient’s metabolic defect, whereas in vivo, the epithelial cell environment is different and probably specific to the metabolic defect. Post-translational regulation cannot be excluded either. Furthermore, we here could not compare our observation with control individuals with a normal BMI range as in our previous study [17].

We have previously shown that T2D causes a decrease in GLP-1 cell density in individuals with obesity [17]. This GLP-1 cell density decrease was not retrieved in HJEs from individuals with metabolic diseases. Filippello et al., in murine enteroids cultivated in presence to high glucose concentration, demonstrated a decrease in the number of GLP-1 cells per enteroid [52]. In our model, the lack of impact of prediabetes or T2D on GLP-1 cell density could be explained by culturing HJEs in an equal glucose concentration for each metabolic phenotype.

Nevertheless, we observed that GNAT3 gene expression was decreased in HJEs derived from subjects with ObPreD compared to Ob HJEs. GNAT3 encodes the G protein α gustducin subunit associated to sweet taste receptor. The intestinal sweet taste detection pathway in EECs, similar to that of taste buds, is involved in GLP-1 secretion [25, 53]. We previously reported a down-regulation of GNAT3 in jejunal enteroendocrine cell sorted in both humans and mice by T2D [25]. In mice the intestinal Gnat3 expression is partially restored after entero-gastric anastomosis, that mimic bypass gastric surgery in humans [25]. Moreover, a meta-analysis in an African population demonstrated a link between GNAT3 expression and waist-to-hip ratio [54]. Further studies are required to determine the involvement of the sweet taste transduction pathway in the alteration of GLP-1 secretion in individuals with metabolic diseases.

GLP-1 secretion was studied in response to a high concentration of glucose (50 mM), which approximates postprandial glucose concentration in the human small intestine in vivo [55, 56]. Zietek et al. showed that high glucose concentration, which is similar to post-prandial glucose levels in the gut in vivo, enhanced the secretion of active GLP-1 in murine enteroids [49]. In our model, EECs in HJEs derived from individuals with severe obesity, regardless of their metabolic status, remain functional, as they secrete active GLP-1 in response to high glucose concentration (50 mM) and to elevated intracellular cAMP level (F/I mix). The response to 50 mM glucose was heterogeneous across patients, regardless of metabolic status (Ob, ObPreD, ObD). This variability is related to the inter-individual variability in patient phenotype and may reflect the fact that HJEs were derived from individuals who had been severely obese for many years, contributing to impaired enteroendocrine cell function. In addition, we demonstrated that HJEs from ObD individuals exhibit a deterioration in the secretion of active GLP-1 in response to glucose compared to Ob HJEs, suggesting a functional alteration of GLP-1 secretion independently of GLP-1 biosynthesis and cell density. It has been reported that SNARE proteins involved in secretory-granule exocytosis may contribute to the GLP-1 secretion defect. Indeed, a defect in GLP-1 release has been demonstrated when genes required for the distal steps of exocytosis, such as syntaxins, are absent [26, 27]. Although we did not observe a reduction in the expression of these genes in ObD HJEs, we cannot rule out the possibility that the number of secretory granules containing GLP-1 is decreased, thereby contributing to the secretion defect.

Moreover, although neither gene nor protein expression indicated a functional alteration of GLP-1 secretion according to metabolic status, we cannot exclude the hypothesis that a reduced PC1/3 enzymatic activity contributes to the impaired GLP-1 secretion observed in ObD HJEs. Indeed, PCSK1 polymorphisms, associated with PC1/3 deficiency, are linked with an increased risk of obesity [57]. Another possible explanation is related to the observation that GNAT3 expression is decreased in ObPreD vs. Ob (p < 0.05), although this difference did not reach statistical significance in ObD vs Ob, in contrast to our previous work [25]. As GNAT3 is involved in the sweet taste detection pathway, its reduced expression could also contribute to the GLP-1 secretion defect.

Despite the rarity of GLP-1 secreting cells in jejunum, we showed the relevance of HJE model from individuals with metabolic disease as an in vitro model to study GLP-1 cells. However, some limitations of our study should be noted. We had access to rare samples from gastric bypass surgery from individuals with severe obesity established with BMI, associated with co morbidities well defined, prediabetes, T2D, treatments, dyslipidemia. Although the cohort size is limited, we stratified individuals into three groups Ob, ObPreD and ObD, based on their glycemia status, ensuring a low but balanced number of individuals each group. Despite a consistent distribution of individuals by sex across groups (10 females vs. 2 males per group), this imbalance may limit the interpretation of the results. However, statistical analysis was adjusted for both age and sex. Given the small number of patients ObD included in this study, it is impossible to differentiate between the results according to treatment. However, we cannot rule out any effect of these treatments on gene expression or GLP-1 secretion without it being possible to determine their impact. The absence of control lean individuals is a limitation for interpreting the impact of obesity on EECs in an HJE model. In addition, EECs in HJEs are physiologically scarce, which necessitated the use of Notch inhibitor to enrich this cell type for studying GLP-1 secretion in response to glucose. Epithelial cells in primary culture have the advantage of being native without the addition of molecules that modify their proliferative or differentiated phenotype. However, these native cells cannot be amplified or maintained for more than a few hours. Organoids or enteroids, by contrast, can be amplified over several passages. The drawback is that the addition of morphogens, growth factors, or Notch inhibitors interferes with the native differentiation process. In our GLP-1 secretion experiments, the use of a Notch inhibitor represents a limitation, since the system is artificially manipulated to promote the secretory cell lineage. This manipulation may vary between each point and could lead to a drift in the metabolic phenotype of HJEs. Furthermore, while HJEs recapitulate intestinal epithelial cells, they lack the immune cells that play a key role in the low-grade chronic inflammation associated with obesity. In metabolic diseases, the gut microbiota is altered, and these changes can contribute to the low-grade inflammation that characterizes these conditions. On the other hand, it is well established that low-grade inflammation promotes the development of insulin resistance and the progression of T2D [58]. Thus, an altered gut microbiota, immunological changes, and a low-grade inflammation may all contribute to insulin resistance in individuals with obesity and T2D. In this context, the enteroid culture system used in this study does not fully reproduce the pathophysiological environment of the native intestinal tissue, which can be considered a limitation. To minimize this limitation, the low-grade inflammation component could be introduced by co-culturing the organoids with immune cells or by adding cytokines to the culture medium. This field could not be addressed with this model.

In conclusion, in vitro models such as HJEs represent valuable tools to investigate the dysregulation of GLP-1 secretion in obesity and T2D, and to identify pharmacological or natural molecules capable of enhancing enterohormone release.

Author contributions

CO, PS and AR designed experiments, acquired, analyzed data, contributed to the discussion and wrote the manuscript. DM and PMO acquired data. HAS analyzed data and performed statistical analyzes. CP and KC contributed to the cohort recruitment and patient phenotyping and LG (surgeon) participated to jejunal sampling. KC contributed to data interpretation, discussion and manuscript editing.

Funding

This work was supported by INSERM, Sorbonne Université, MSD Avenir, PMO was funded by EFSD fellowships. The clinical study was promoted by the Assistance Publique-Hôpitaux de Paris (APHP) and Direction of Clinical research, which promoted the clinical investigations (Microbaria and LEAKY GUT projects). This work has benefited from funding from by Benjamin Delessert Institute, and PEPR-SAMS JEMINI.

Data availability

Data described in the manuscript will be made available upon request pending application and approval from the corresponding author.

Competing interests

The authors declare no competing interests.

Supplementary information

The online version contains supplementary material available at 10.1038/s41366-026-02024-3.