Figures and data
![](https://prod--epp.elifesciences.org/iiif/2/99337%2Fv1%2Fcontent%2F596846v1_fig1.tif/full/max/0/default.jpg)
Global analysis of the tumor microenvironment and malignant features of H. pylori infection associated GC.
(A) A general workflow of GC sample preparation and processing of single-cell suspensions for scRNA-seq analysis. In total, 27 gastric samples, including gastric tissues of healthy control (HC, n=6), gastric cancer without H. pylori infection (non-HpGC, n=6), gastric cancer with previous H. pylori infection (ex-HpGC, n=6) and gastric cancer with current H. pylori infection (HpGC, n=9), were collected to perform scRNA-seq. (B) Uniform Manifold Approximation and Projection (UMAP) plot for unbiased clustering and cell type annotation of 86,637 high-quality cells. EPI_M: malignant epithelium; EPI: non-malignant epithelium; Endo: endothelium; Fib: fibroblast; Mono/Mac: monocyte/macrophage; Plasma: plasma cells; Mast: mast cells; B: B cells; T: T cells. (C) the absolute quantities of nine different cell types. (D) Heatmap showing the top eight differentially expressed genes (DEGs) of nine main cell types. (E) Violin plots showing the expression of signature genes of nine cell types. (F) Pie plot revealing the proportions of nine cell types in HC, non-HpGC, ex-HpGC and HpGC. (G) Box plot showing statistical analysis of proportion of nine cell types in HC, non-HpGC, ex-HpGC and HpGC. The P value of Student’s t test is shown. (H) Volcano plot displaying the differential upregulated genes in EPI and EPI_M. (I) GSEA showing the pathway activity (scored per cell by Gene Set Enrichment Analysis) in malignant epithelium (EPI_M) and non-malignant epithelium (EPI).
![](https://prod--epp.elifesciences.org/iiif/2/99337%2Fv1%2Fcontent%2F596846v1_tbl1.tif/full/max/0/default.jpg)
![](https://prod--epp.elifesciences.org/iiif/2/99337%2Fv1%2Fcontent%2F596846v1_tbl1a.tif/full/max/0/default.jpg)
Patient characteristics of each sample in GC scRNA-seq
![](https://prod--epp.elifesciences.org/iiif/2/99337%2Fv1%2Fcontent%2F596846v1_fig2.tif/full/max/0/default.jpg)
Characterization of the malignant epithelium within GC with different H. pylori infection status.
(A) UMAP plot showing six subpopulations of malignant epithelium, colored by different cell types (upper) and different H. pylori infection status (lower). (B) Pie plot showing the proportion of six subset of malignant epithelium in non-HpGC, ex-HpGC, and HpGC. (C) Heatmap displaying the differentially expressed genes (DEGs) among the six subsets of malignant epithelium. (D) Bubble plot showing the difference of representative molecular among the non-HpGC, ex-HpGC, and HpGC. (E) GSEA showing the pathway activity (scored per cell by Gene Set Enrichment Analysis) in malignant epithelium of HpGC compared to that of nonHpGC. NES, normalized enrichment score. (F) The ridge plot showing the differentiation score (Diff Score) of malignant epithelium within each sample. (G) Heatmap showing the Diff Score of six subpopulations of malignant epithelium. (H) Box plot showing differentiation score among non-HpGC, ex-HpGC, and HpGC, P values were assessed by Wilcoxon test, Two-way ANOVA test is used for comparison of multiple groups. (I) Kaplan-Meier survival analysis of TCGA STAD patients stratified by tumor sample differentiation scores, which was used to group samples into high and low groups based on 33th and 67th percentile. The P value of two-sided log-rank test is shown.
![](https://prod--epp.elifesciences.org/iiif/2/99337%2Fv1%2Fcontent%2F596846v1_fig3.tif/full/max/0/default.jpg)
Characterization of the non-malignant epithelium within GC with different H. pylori infection status.
(A) Unbiased clustering of non-malignant epithelium generated nine subtypes. (B) Heatmap showing the molecular feature of non-malignant epithelium according to the top five DEGs. (C) Box plot showing the dynamic proportion of different cell types in non-malignant epithelium with different H. pylori infection status. (D) Heatmap showing the relative abundance (estimated by GSVA) of non-malignant epithelium subtypes in normal control (NC), chronic gastritis (CG), intestinal metaplasia (IM), GC samples (GSE2669). (E) Boxplot showing the relative abundance (estimated by GSVA) of non-malignant epithelium subtypes SPEM, IM, Enterocytes, Chief cells, Neck cells and Parietal cells within NC, CG, IM, and GC samples (GSE2669). The P value of Student’s t test is shown. (F) The trajectory analysis showing potential differentiation and transition trajectories in non-malignant epithelium clusters. (G) The ridge plot showing the pseudotime of non-malignant epithelium revealed the gastric pre-lesion process. (H−I) Heatmap showing scaled expression of dynamic genes (G) and TFs (H) along cell pseudotime.
![](https://prod--epp.elifesciences.org/iiif/2/99337%2Fv1%2Fcontent%2F596846v1_fig4.tif/full/max/0/default.jpg)
Characterization of tumor-infiltrating T and natural killer (NK) cells in H. pylori infection associated GC.
(A) Unbiased clustering of T and NK cells generated 11 clusters. (B−C) Molecular features annotations according to the top five DEGs (B), and representative genes (C). (D) Pie plot showing the T/NK cell subtype abundance distribution in HC, non-HpGC, ex-HpGC, and HpGC. (E) The percentage contribution of T/NK cell subtype in HC, non-HpGC, ex-HpGC, and HpGC samples. P values were assessed by Student t test. (F) The deconvolution analysis showing the relative abundance of CD8_CXCL13 and Tregs in GC with different H. pylori infection status with TCGA STAD dataset, P values were assessed by Wilcoxon test. (G) The cytotoxic and inhibitory expression scores in T and NK clusters. (H) Kaplan-Meier survival analysis of TCGA STAD patients stratified by CD8_CXCL13 and Tregs relative abundance, which was used to group samples into high and low groups based on 33th and 67th percentile. The P value of two-sided log-rank test is shown. (I) Bubble plot showing intercellular interactions between suppressive T cells and malignant cells. (J) Dotplot showing the expression of NECTIN2 and PVR in malignant non-HpGC, ex-HpGC, and HpGC cells. (K) Immunostaining showing the ligand TIGIT expressed in suppressive T cells and the receptor NECTIN2 expressed on the malignant epithelium, respectively, in one HpGC sample.
![](https://prod--epp.elifesciences.org/iiif/2/99337%2Fv1%2Fcontent%2F596846v1_fig5.tif/full/max/0/default.jpg)
Characterization of tumor-infiltrating myeloid cells by scRNA-seq in H. pylori infection associated GC.
(A–C) Unbiased clustering of myeloid cells generated nine clusters (A), and molecular features were annotated according to the top seven DEGs (B) and representative genes (C). (D) The percentage contribution of each myeloid cells cluster in HC, non-HpGC, ex-HpGC, and HpGC. P values were assessed by Student t test. (E) Volcano plot showing the DEGs between Angio-TAM and TREM2+ TAM. (F) Violin plot showing the expression of functional gene sets in myeloid clusters. (G) Immunostaining showing the distribution of Angio-TAM and TREM2+ TAM in one HpGC sample. (H) Barplot showing the enriched signaling pathway between Angio-TAM and TREM2+ TAM. (I) The correlation of cell type (percentage) between Tregs and Angio-TAM and TREM2+ TAM. (J) Dotplot showing intercellular interactions among suppressive T cell and Angio-TAM and TREM2+ TAM. (K) The relative abundance of Angio-TAM in HpGC and non-HpGC in TCGA STAD dataset, P values were assessed by Wilcoxon test. (L) Kaplan-Meier survival analysis of TCGA STAD patients stratified by Angio-TAM relative abundance, which was used to group samples into high and low groups based on 33th and 67th percentile. The P value of two-sided log-rank test is shown.
![](https://prod--epp.elifesciences.org/iiif/2/99337%2Fv1%2Fcontent%2F596846v1_fig6.tif/full/max/0/default.jpg)
Characterization of cancer-associated fibroblasts (CAFs) by scRNA-seq in H. pylor infection associated GC.
(A) Unbiased clustering of CAF generated 6 clusters. (B–C) Molecular features annotations according to the top ten DEGs (B) and representative genes (C). (D) The pie plot showing the abundance distribution of six CAF subset in HC, non-HpGC, ex-HpGC, and HpGC. (E) The percentage contribution of each CAF cluster in HC, non-HpGC, ex-HpGC, and HpGC. P values were assessed by Student t test. (F) Heatmap showing the enriched signaling pathway among six CAF clusters. (G) The cell type percentage correlation between iCAF and Treg, Angio-TAM, and TREM2+ TAM. (H) Heatmap showing intercellular interaction strength among immune cells and different subsets of CAF. (I) Enriched signaling pathway among suppressive T cell, Angio-TAM, and TREM2+ TAM with iCAF. (J) The relative abundance of iCAF in HpGC and non-HpGC in TCGA STAD dataset. P values were assessed by Wilcoxon test. (K) Kaplan-Meier plot shows that the abundance of iCAF predicts poor prognosis of GC using two public bulk RNA sequencing dataset (left, TCGA; right, ACRG). The abundance of iCAF was used to group samples into high and low groups based on 33th and 67th percentile. The P value of two-sided log-rank test is shown. ACRG: Asian Cancer Research Group.
![](https://prod--epp.elifesciences.org/iiif/2/99337%2Fv1%2Fcontent%2F596846v1_fig7.tif/full/max/0/default.jpg)
Single cell TME composition associated with GC immunotherapy outcome.
(A) GSEA plot showing the enrichment of iCAF, CD8_MKI67, Angio-TAM, and CD8_CXCL13 in anti-PD1 responsive or non-responsive GC. NES, normalized enrichment score. (B–C) Bar chart showing the cell subtypes relative abundance, the immune signature, angiogenesis signature derived from scRNA-seq predicted GC immunotherapy outcome, progression-free survival (B), and overall survival (C). (D) Heatmap showing the expression of immune and angiogenesis signature derived from scRNA-seq in immunotherapy responsive and non-responsive GC. (E) Violin plot showing the expression of immune signature and angiogenesis signature in immunotherapy responsive and non-responsive GC. The P value of Wilcoxon test is shown. (F) A model for evaluating the GC immunotherapy sensitivity and specificity using the immune signature and angiogenesis signature derived from scRNA-seq. (G) The Kaplan-Meier plot showing the immune signature and angiogenesis signature could efficiently predict prognosis of GC anti-PD-1 therapy. The P value of two-sided log-rank test is shown.