Sleeve gastrectomy reveals the plasticity of the human gastric epithelium

Abstract

Gastrin is secreted following a rise in gastric pH, leading to gastric acid secretion. Sleeve gastrectomy (SG), a bariatric surgery where 80% of the gastric corpus is excised, presents a challenge for gastric pH homeostasis. Using histology, and single-cell RNA sequencing of the gastric epithelium in 12 women, we observed that SG is associated with an increase in a sub-population of acid-secreting parietal cells that overexpress respiratory enzymes and an increase in histamine-secreting enterochromaffin-like cells (ECLs). ECLs of SG-operated patients overexpressed genes coding for biosynthesis of neuropeptides and serotonin. Mathematical modeling showed that pH homeostasis by gastrin is analogous to non-linear proportional and integral control, that drives adaptation of the epithelium to acid-secretion demand. Quantitative model predictions were validated in patients. The results demonstrate human gastric epithelium remodeling following SG at the molecular and cellular levels, and more generally how trophic hormones enable robust adaptation of tissue function to meet physiological demand.

Fig. 1. The gastric epithelial response to SG.

A The naive stomach (pink) and following sleeve gastrectomy (SG, cyan). Most of the gastric corpus is removed in the gastrectomy, and the stomach assumes the shape of a sleeve. B UMAP plot of 16581 cells of human gastric antrum color-coded according to cell type. C The same UMAP plot as in (B), color-coded in pink for naive patients, and cyan for SG-operated patients. All cell types were detected in SG and naive patients. D Percentage of cell types from the entire epithelial population in the gastric corpus from naive (pink) and SG-operated patients (cyan). n = 6,5 patients. Data are presented as quartiles, mean (box) and minima-maxima (whiskers). E UMAP plot of 37,850 cells of the human gastric corpus color-coded according to cell type. F The same UMAP plot as in (E), color-coded in pink for naive patients, and cyan for SG-operated patients. All cell types were detected in SG and naive patients. G Depth of the gastric corpus mucosa gland base (between the isthmus and bottom of the gland), pit (above the isthmus) and the entire gland in naive and SG-operated patients. n = 11,8 patients. H The number of KI67 positive cells per gastric oxyntic gland in the corpus of naive and SG-operated patients. n = 19,10 patients. KI67 immunostaining in the gastric corpus of naive (I) and SG-operated patients (J). Scale bar = 0.5 mm. Blow-up scale bar = 0.1 mm. Same scale in (I, J). ns non-significant *p < 0.05, **p < 0.01. Data in (G, H) are presented as mean ± standard error of the mean (SEM). 2-sided Mann–Whitney U-test as used in (D, G, H). Source data and p values are provided in the source data file.

Fig. 2. Effects of SG on gene expression in corpus parietal and endocrine cells.

A 3-dimensional UMAP projection of parietal cells of the gastric corpus; parietal cells (purple) and parietal^resp cells (dark purple). B Gene ontology enrichment in parietal^resp cells relative to standard parietal cells. C CA2 expression in parietal and parietal^resp cells in naive (pink) and SG-operated patients (cyan). D Percentage of parietal^resp cells out of all parietal cells in naive (pink) and SG-operated patients (cyan). n = 5,5 patients. E Percentage of ECLs, X-cells, and other endocrine cell types out of the corpus endocrine population in naive (pink) and SG-operated patients (cyan). n = 6,5 patients. F Expression levels of enzymes, CCKBR, and prohormones in ECLs of naive (pink) and SG-operated patients (cyan). G Percentage of 5HT⁺ positive cells out of endocrine CHGA⁺ cells in corpus of naive (pink) and SG-operated patients (cyan). n = 9,8 patients. Data are presented as mean ± SEM. Representative images of 5HT (red) and CHGA (blue) staining in the gastric corpus epithelium of naive (H) and SG-operated patients (I). J GHRL expression in X-cells of the corpus in naive (pink) and SG-operated patients (cyan). K GAST expression in G-cells of the antrum in naive (pink) and SG-operated patients (cyan). L SST expression in D-cells of the antrum in naive (pink) and SG-operated patients (cyan). M CCKBR expression in the endocrine and parietal cells of the gastric corpus in naive (pink) and SG-operated patients (cyan). *^,**^,***p < 0.05, p < 0.01, p < 0.001^{##,& &}. p < 0.01 in Friedman’s Test by cell type and surgery accordingly. Scale bar in H = 20 μm. Data in (D, E) are presented as quartiles, mean (box) and minima-maxima (whiskers). 2-sided Student’s t test was used in (D, G) and 2-way ANOVA with Tukey’s post-hoc analysis in (E). Significance in (C, F, J–M) calculated using Wilcoxson rank sum test. 2-sided Friedman’s test used in 2 C. Source data and p values are provided in the source data file.

Fig. 3. A model for gastrin-mediated gastric pH homeostasis.

A Illustration of pH regulation by gastrin. A low concentration of Hydrogen ions enters the gastric lumen from the esophagus (H_in). A decrease in Hydrogen levels (H) induces the secretion of gastrin (G), which induces histamine secretion from ECLs (E). Both gastrin and histamine stimulate Hydrogen secretion from parietal cells (P) at concentration H₀. Gastrin also promotes the generation of parietal cells and ECLs. Hydrogen ions then leave the stomach through the pylorus. B Steady-state level of H, H_st, is determined by H₁, the threshold for gastrin secretion. H_st levels are robust to sensitivity to gastrin and histamine (s) (C), parietal cell generation rate (β_P) (D), rate of incoming Hydrogen through the esophagus (H_in) (E) and the threshold for ECL and parietal cell generation in response to gastrin (G₁) (F). G A diagram showing gastrin (G) control over acid secretion as a product of two integral feedbacks over the number of parietal cells (P) and ECLs (E) and proportional feedback by gastrin. The sensitivity of H_st to u for a range of values of s and u is high in a system without integral feedback over P and E (H), and very low when the integral feedbacks are incorporated (I).

Fig. 4. Model predictions and validations for the gastric response to SG.

A Simulation of the levels of H, the quasi steady state levels of Hydrogen before SG (pink) and years following SG (cyan). The same parameters were used in the simulation, except for reducing the number of corpus glands by 5-fold. Dashed vertical line marks the time of surgery (t = 0). B As in A, for G, the levels of gastrin. C Fasting gastrin levels before, 1 day and 10 days after surgery in SG patients. n = 27. D Fold change in gastrin levels in the same patients as in (C). E Simulation of P, the number of parietal cells per gland before and years following SG. F The number of parietal cells in naive stomachs and stomachs from patients that underwent SG years earlier. n = 13,5 patients. Representative images of ATP4B staining in biopsies from the gastric corpus of naive and SG-operated patients. Same scale used in (G, H). I Simulation of E, the number of ECLs per gland before and following SG. J The number of ECLs in the naive stomachs and stomachs of patients who underwent SG years earlier. n = 9,6 patients. K Simulation of the acid secretion capacity in the stomach, defined as the product of n,s,G,P and E before surgery and after surgery. *p < 0.05; ***p < 0.001. Data in (C, D, F, J) are presented as mean ± SEM. 2-sided DW with Dunn’s post-hoc used in (C), ANOVA with Tukey’s post-hoc used in (D), 2-sided Student’s t test in (F) and 2-sided MW in (J). Source data and p values are provided in the source data file.