Fig 2: Stromal cell remodeling in GC progression. (A) UMAP showing the CAF subcluster. (B) Representative marker genes of the CAF subcluster. (C) Proportions of each subset of CAFs. (D) Proportions of four representative CAF subpopulations (CAFmat‐POSTN, CAFinfla‐PRSS35, CAFinfla‐MFAP5, and CAFEMT‐KRT19) across tissue groups. (E) Expression specificity of PRSS35 and MFAP5 in CAFs. (F) Survival analysis of PRSS35 and MFAP5 in GC. (G) UMAP view of 7 endothelial cell clusters. (H) Proportions of each subset of endothelial cell clusters. (I) Proportions of representative IL‐33 + Venous‐1 subpopulations across tissue groups. (J) Heatmap of marker gene expression in endothelial cell subpopulations. Red signifies increased expression. (K) Survival analysis of IL‐33 in GC. (L) Expression specificity of IL‐33 in endothelial cells. (M) KEGG bubble map of DEGs between IL‐33 + Venous‐1 and other endothelial cell subpopulations. p values were calculated via one‐way Kruskal‒Wallis rank‒sum tests. *p < 0.05; **p < 0.01; ***p < 0.001; ****p < 0.0001. CAFs, cancer‐associated fibroblasts; DEGs, differentially expressed genes; KEGG, Kyoto Encyclopedia of Genes and Genomes; UMAP, uniform manifold approximation and projection.

Fig 3: IL‐33+ ECs promote EGC and AGC angiogenesis and growth in vivo. (A) Diagram of a mouse subcutaneous transplanted tumor (EC cells and organoids). (B, C). Macroscopic images of the excised subcutaneous tumor mass upon sacrifice. No tumors were observed at the injection sites of the only EGC and AGC organoids (H&E‐confirmed inflammatory tissue). EC overexpression of IL‐33 promoted the growth of organoids (both EGC and AGC) subcutaneously (B). EC knockdown of IL‐33 inhibited the growth of subcutaneous organoids (both EGC and AGC) (both EGC and AGC) (C). (D, E) Tumor volume was monitored every four days, and tumor growth curves were drawn. EC‐OE‐NC had no significant effect on the growth of tumors compared with that of the only EGC organoid group (D). EC‐OE‐IL‐33 and EC‐OE‐NC had no significant effect on the growth of AGC organoids (E). (F, G) Tumor weights of the extracted subcutaneous tumors at the endpoint. ECs‐OE‐NC had no significant effect on the weight of tumors compared with the weight of tumors in the group with only EGC organoids. EC‐OE‐IL‐33 and EC‐OE‐NC had no significant effect on the weight of AGC organoids (F). The rate of subcutaneous tumor formation in the EC‐OE‐IL‐33 group of EGCs was greater than that in the EC‐OE‐NC group, and there was no difference in the other groups (G). (H) Ultrasound image of a mouse subcutaneous tumor. The EC‐sh‐IL‐33 group had more dark areas of fluid in the center of the subcutaneous tumors and a lower central blood supply. (I) Ultrasonic PSV detection of subcutaneous tumors in vivo at the endpoint. (J) Ultrasonic RI detection of subcutaneous tumors in vivo at the endpoint. The data represent the means ± SDs from six mice per group. ns, not significant; *p < 0.05; **p < 0.01; ***p < 0.001; ****p < 0.0001 versus empty vector control (two‐sided Wilcoxon rank‐sum test). AGC, advanced gastric cancer; B mode, brightness mode; CD mode, color Doppler mode; EGC, early gastric cancer; ECs, endothelial cells; H&E, hematoxylin‐eosin staining; NC, negative control; OE‐IL‐33, overexpression of IL‐33; PW mode, pulse‒wave Doppler mode; PSV, peak systolic velocity; RI, resistance index; sh‐IL‐33, knockdown of IL‐33.