Single-Cell RNA Sequencing Unifies Developmental Programs of Esophageal and Gastric Intestinal Metaplasia

Abstract

Intestinal metaplasia in the esophagus (Barrett's esophagus IM, or BE-IM) and stomach (GIM) are considered precursors for esophageal and gastric adenocarcinoma, respectively. We hypothesize that BE-IM and GIM follow parallel developmental trajectories in response to differing inflammatory insults. Here, we construct a single-cell RNA-sequencing atlas, supported by protein expression studies, of the entire gastrointestinal tract spanning physiologically normal and pathologic states including gastric metaplasia in the esophagus (E-GM), BE-IM, atrophic gastritis, and GIM. We demonstrate that BE-IM and GIM share molecular features, and individual cells simultaneously possess transcriptional properties of gastric and intestinal epithelia, suggesting phenotypic mosaicism. Transcriptionally E-GM resembles atrophic gastritis; genetically, it is clonal and has a lower mutational burden than BE-IM. Finally, we show that GIM and BE-IM acquire a protumorigenic, activated fibroblast microenvironment. These findings suggest that BE-IM and GIM can be considered molecularly similar entities in adjacent organs, opening the path for shared detection and treatment strategies.

Significance: Our data capture the gradual molecular and phenotypic transition from a gastric to intestinal phenotype (IM) in the esophagus and stomach. Because BE-IM and GIM can predispose to cancer, this new understanding of a common developmental trajectory could pave the way for a more unified approach to detection and treatment. See related commentary by Stachler, p. 1291. This article is highlighted in the In This Issue feature, p. 1275.

Figure 1.

Overview of the scRNA-seq atlas of the GI tract. A, Overview of the samples analyzed in the study. For each sample, the approximate location of tissue is indicated. Where indicated, text in brackets denotes the study from which samples originate [Wang et al. (37), Zhang et al. (61), and Sathe et al. (62)]. The remaining samples were collected in the current study or originate from Nowicki-Osuch and Zhuang et al. (4). B, UMAP of all high-quality cells used in the study. The main plot shows all tissue types overlay (point order is randomized). The insets show selected tissue types grouped according to their anatomic location and disease state. BSCJ, squamocolumnar junction between NE and BE-IM.

Figure 2.

Cell-type distribution across the tissues of the GI tract. A, UMAP of all high-quality cells (point order is randomized) used in the study, with the cell types assigned to the epithelial compartment in color. The line indicates a TSCAN-derived trajectory between the clusters of epithelial cells. Nonepithelial cells are in gray. B, Stacked bar chart of the proportion of cell types assigned to individual tissue types. The color scale is the same as in A. BSCJ, squamocolumnar junction between NE and BE-IM.

Figure 3.

Columnar cells of the stomach and esophageal IM have mosaic gastric and intestinal phenotypes. A, UMAP of columnar cells isolated from samples of NSCJ, NGC, NGB, ND, ileum, colon, rectum, BE, gastric metaplasia of esophagus, squamocolumnar junction between NE and BE, CIM, GIM, nonchronic atrophic gastritis, and chronic atrophic gastritis with color denoting cell types. The cell types were assigned using cells from normal tissue types. B, Scatter plot of log-normalized expression of intestinal (GPA33) and gastric (MUC5AC) markers in the selected tissue types. C, Coimmunofluorescent staining of the esophagus with BE with BE-IM and GIM shows coexpression of intestinal and gastric markers in both types of IM using lineage markers MUC5AC (gastric) and GPA33 (intestinal), and progenitor markers MUC6 (gastric) and OLFM4 (intestinal). White arrowheads denote selected goblet cells; dashed white lines indicate selected BE-IM and GIM crypts, for GIM samples: 1, a crypt with mixed gastric and intestinal phenotype; 2, a crypt with cells showing mosaic phenotype; 3, a crypt with features of complete IM; scale bar, 100 μm. See supplementary Figs. S8 and S9 for ND and NGC. Images are representative of 12 patients (NGC = 2, ND = 2, BE-IM = 4, and GIM = 4). D, UMAP with MuSiC-derived contribution of gastric and intestinal phenotypes to individual cells of the esophageal IM (top) and stomach IM (bottom). E, Violin plots of squamous, SMG, gastric, or intestinal phenotype contribution to cells from NGC, E-GM, BE-IM, BSCJ, NAG, CAG, GIM, CIM, and ND samples.

Figure 4.

E-GM and atrophic gastritis share early features of developing IM. A, UMAP of columnar cells of gastric cell types (NGC and NGB, top left), E-GM (top right), and atrophic gastritis (NAG and CAG, bottom left) with cell-type annotation highlighted. Bottom right, UMAP with clusters 6 and 16 (identified during reclustering of columnar cells and overlapping gastric neck cells) highlighted. EE, enteroendocrine. B, Stacked bar chart of tissue contribution to the cells assigned to clusters 6 and 16. C, Gene set enrichment analysis (GSEA) using the C3 gene set database of differentially expressed genes between neck-like cells from E-GM and NGC samples. HNF4A- and MYC-related pathways are highlighted. The differential analysis was done between E-GM cells and NGC + NSCJ neck-like cells with each patient treated as an individual replicate. NES, normalized enrichment score. D, GSEA using the C3 gene set database of differentially expressed genes between NAG/CAG and NGC/NGB/NSCJ sample neck-like cells. AP-1–related pathways are highlighted. The differential analysis was done between cluster 16 cells (including NAG, CAG, and E-GM cells) and cluster 6 (including NGC + NGB + NSCJ neck-like cells) with each patient treated as an individual replicate. E, Venn diagram demonstrating the overlap between genes enriched in the comparison of atrophic gastritis (NAG and CAG) and normal gastric samples or E-GM and normal gastric samples. Bottom left: bubble plot of top 25 expressed genes shared between the comparison. Bottom right, violin plots of selected genes. F, WGS-based analysis of E-GM and BE-IM. Top, mutational burden [single-nucleotide variants/megabase (SNV/Mb)] of NGC, E-GM, and BE-IM samples. Bottom, distribution of COSMIC SNV signatures in the E-GM and BE-IM samples.

Figure 5.

Gastric and esophageal IM are enriched for extracellular matrix–depositing subtypes of fibroblasts. A, UMAP of fibroblast-like cells (cluster 13 in Supplementary Fig. S1) with fibroblast-specific clusters highlighted. B, UMAP of fibroblast-like cells (cluster 13 in Supplementary Fig. S1) with manually annotated cell types derived from Davidson et al. (38). C, UMAP of fibroblast-like cells (cluster 13 in Supplementary Fig. S1) with the contribution of individual tissue type highlighted. D, Bubble plot of top five genes identified in the differential analysis between the fibroblast-specific cell clusters. E, Bubble plot of genes identified as markers of early fibroblast development by Davidson et al. (38). F, Contribution of cell types derived from Davidson et al. (38) to cell counts from individual tissue types. The tissues originate from three studies: (i) Nowicki-Osuch and Zhuang et al. (4) and this study, (ii) Sathe et al. (62), and (iii) Zhang et al. (61). For tissue for which patient status could be identified, the tissue was split into healthy and disease states. NGB samples are adjacent to gastric cancer samples. G, Proportion of each subtype in total S1/S2/S3 fibroblasts derived from IF staining of E-GM, BE-IM, and GIM. H, Representative IF staining of E-GM, BE-IM, and GIM of S1 (CD34⁺CD31⁻), S2 (POSTN⁺aSMA⁻), and S3 (aSMA⁺) fibroblasts. I, Schematic summary of unified developmental trajectories between BE-IM and GIM.