PTTG1 drives intestinal-type gastric cancer progression via G2/M checkpoint-mediated cell-cycle reprogramming Authors: Caihong Hu, Lijun Bian, Lin Luo, Zhonghua Zheng, Yuheng Chen, Xinxiang Gao, Fengyuan Li, Dinghua Yang, Yao Liu, Gang Li, Hao Xu, Caiwang Yan, Guangfu Jin Published: 2026 ABSTRACT SummaryIntestinal-type gastric cancer (IGC) progresses through the Correa cascade, yet the molecular determinants of this transition remain insufficiently defined. Using single-cell RNA sequencing (scRNA-seq) on matched normal, intestinal metaplasia, and tumor tissues from five patients, we identified eight epithelial subtypes, including a transitional enterocyte population that expanded along the disease trajectory. Pseudotime and network analyses revealed a progressive upregulation of PTTG1, which was further supported by public cohorts, bulk transcriptomics, and tissue microarrays. Functional assays demonstrated that PTTG1 knockdown impaired G2/M checkpoint regulation and reduced gastric cancer cell proliferation, migration, and invasion, whereas its overexpression enhanced organoid growth and altered mucin-related features. In xenograft models, PTTG1 depletion suppressed tumor growth and was accompanied by secretory differentiation-associated histological changes. Collectively, these findings suggest that PTTG1 contributes to intestinal-type gastric cancer progression by regulating G2/M checkpoint-associated programs and modulating mucin-associated phenotypes, supporting its potential as a risk biomarker and candidate therapeutic target for IGC prevention.Keywords: Correa cascade, intestinal metaplasia, intestinal-type gastric cancer, PTTG1, G2/M checkpoint Introduction Gastric cancer (GC) remains a major public health burden worldwide, ranking as the fifth most common malignancy and the fifth leading cause of cancer-related mortality.1 Gastric adenocarcinoma (GA) accounts for the majority of GC cases and exhibits substantial heterogeneity at both morphological and molecular levels.2,3 According to Lauren’s classification, GA is categorized into intestinal-type, diffuse-type, and mixed-type,4 which differ in etiological factors, clinical behavior, and prognosis.5,6 Intestinal-type gastric cancer (IGC) is widely considered to develop through a multistep process of gastric mucosal alterations, commonly referred to as the Correa cascade, encompassing non-atrophic gastritis (NAG), chronic atrophic gastritis (CAG), intestinal metaplasia (IM), dysplasia, and adenocarcinoma.7,8,9 Despite strong clinicopathological evidence supporting this model, the molecular mechanisms driving the transition from premalignant lesions to carcinoma remain elusive. Recent advances in single-cell RNA sequencing (scRNA-seq) have enabled detailed interrogation of cellular heterogeneity and cell-state dynamics during tumorigenesis.10,11,12,13,14 In GC, scRNA-seq studies have revealed distinct epithelial subtypes and highlighted dynamic remodeling of the tumor microenvironment across disease stages.15,16,17 However, the mechanisms underlying stepwise malignant transformation of epithelial cells along the Correa model remain incompletely understood. Pituitary tumor-transforming gene 1 (PTTG1), which encodes the securin protein,18 plays a critical role in regulating chromosome segregation, cell-cycle progression, and mitotic fidelity.18,19,20 Dysregulated PTTG1 expression has been observed in multiple malignancies, including hepatocellular, breast, lung, colon, testicular, and ovarian cancers.21,22,23,24,25,26,27,28 Functionally, PTTG1 has been linked to multiple oncogenic behaviors, such as proliferation and metastatic potential, with elevated expression frequently associated with aggressive phenotypes.29,30,31,32 Although increased PTTG1 expression has also been reported in GC tissues compared with adjacent normal mucosa,33 its specific role in gastric carcinogenesis, particularly in epithelial transformation, remains unclear. In this study, we performed scRNA-seq on matched samples from five IGC patients, including five tumor tissues, five IM tissues, and three NAG tissues. Pseudotime trajectory analysis suggested a gradual upregulation of PTTG1 from NAG to IM and IGC, implicating its involvement in the stepwise progression of IGC. To further define the functional role of PTTG1, we conducted loss- and gain-of-function experiments in GC cell lines and gastric organoids, complemented by in vivo validation using xenograft mouse models. Furthermore, transcriptomic and molecular analyses were performed to investigate the mechanisms by which PTTG1 contributes to malignant phenotypes, with a particular focus on its regulation of the G2/M checkpoint. Results Single-cell transcriptomic atlas reveals cellular heterogeneity in multistage IGC To characterize cellular heterogeneity across various lesion stages of IGC at single-cell resolution, we generated scRNA-seq profiles of all viable cells isolated from tumor samples of five IGC patients who underwent gastrectomy without prior treatment, together with matched intestinal metaplasia (IM) tissues and non-atrophic gastritis (NAG) tissues (Figure 1A). All tissue sections were independently evaluated by two pathologists, and the histopathological features of each lesion stage were confirmed by H&E staining (Figure S1). The characteristics of five patients are summarized in Table S1. scRNA-seq overview and identification of major cell types in multistage IGC tissues (A) Schematic diagram of the scRNA-seq workflow. Created with BioRender.com. (B) Uniform manifold approximation and projection (UMAP) of 55,785 single cells from 13 human gastric tissue specimens, colored by major cell type. (C) Dot plot showing the average expression levels and proportions of cells expressing canonical marker genes used for major cell-type identification. (D) UMAP of single cells from five patients, colored by sample origin. (E) UMAP plots of cells from 13 tissue specimens, grouped by histopathological diagnosis and colored by cell type. (F) Proportional distribution of major cell types across different lesion stages. After quality control and filtering the scRNA-seq data to exclude low-quality cells and putative cell doublets, the transcriptomes of a total of 55,785 cells from 13 samples were grouped into nine major cell types, visualized in a uniform manifold approximation and projection (UMAP) plot based on the following markers: B cells (n = 808; expressing CD79A, MS4A1, and CD19), endothelial cells (n = 2,541; expressing VWF, ENG, and PECAM1), epithelial cells (n = 25,753; expressing MUC5AC, TFF1, and TFF2), fibroblast cells (n = 1,276; expressing COL1A2, COL3A1, and DCN); malignant cells (n = 8,418; expressing KRT7, CLDN4, and MET); mast cells (n = 339; expressing TPSAB1, CPA3, and TPSB2); myeloid cells (n = 3,107; expressing CD14, CD163, and CSF1R); plasma cells (n = 11,671; expressing MZB1 and TNFRSF17); and T cells (n = 1,872, expressing CD3D, CD3G, and CD2) (Figures 1B and 1C). To further characterize the cellular composition, we analyzed cell distributions according to sample origin and histopathological diagnosis, as well as cell-type proportions (Figures 1D–1F). Across lesion stages, the proportions of myeloid cells, endothelial cells, and fibroblasts gradually increased, while the proportion of malignant epithelial cells was elevated in IGC tissues (Figure 1F). Epithelial lineage mapping identifies transitional subtypes associated with malignant progression Epithelial cells were extracted and re-clustered, yielding eight subtypes defined by canonical markers, including chief cells, endocrine cells, enterocytes, goblet cells, neck cells, parietal cells, pit cells, and tumor cells (Figures 2A and 2B; Figure S2A). The proportions of chief cells, neck cells, and parietal cells decreased across lesion stages, while tumor cells were only present in IGC samples (Figure 2C; Table S2; Figures S2B–S2D). Single-cell transcriptomic characterization of epithelial cells (A) UMAP plot showing re-clustered epithelial subtypes (n = 34,171 cells) from 13 human gastric tissue specimens. (B) Dot plot showing the average expression levels and proportions of cells expressing canonical marker genes across epithelial subpopulations. (C) Proportional distribution of epithelial subpopulations across different stages. (D) Trajectory plots showing epithelial subpopulations (left) and pseudotime (right). Cells are colored according to computed pseudotime, with darker colors denoting earlier states and lighter colors denoting later states along the trajectory. Each dot represents a single cell. (E) Boxplot showing incomplete intestinal metaplasia signature scores across epithelial subpopulations, with the highest scores observed in Enterocytes-S3 (Wilcoxon rank-sum test, ∗∗∗p < 0.001). (F) Clustering of co-expressed gene modules. The color bar beneath the dendrogram represents the modules assignment of each gene. (G) Volcano plot showing differential changes across Enterocytes-S3 modules. (H) GO enrichment analysis of genes in Module 4 of Enterocytes-S3. To investigate epithelial cell-state alterations during IGC progression, we performed pseudotime analysis using Monocle to reconstruct potential differentiation trajectories of epithelial subpopulations. The trajectory originated from neck cells and toward enterocytes. Based on gene expression dynamics (Table S3), epithelial cells were categorized into three transcriptional states (states 1–3) using Monocle (Figure S2E). Tumor cells were highly enriched in state 3. This state contained a population of intestinal metaplasia-associated enterocytes, which we designated as Enterocytes-S3. Notably, Enterocytes-S3 was positioned between earlier epithelial populations and IGC cells along the inferred trajectory, suggesting that it may represent an intermediate epithelial state associated with the stepwise transition from non-malignant to malignant epithelial states (Figure S2F; Figure 2D). To further validate the transitional identity of Enterocytes-S3 beyond pseudotime inference, we calculated an incomplete intestinal metaplasia (IIM) signature score based on previously reported IIM-associated genes.34 As shown in Figure 2E, the IIM signature score gradually increased from neck cells to Enterocytes-S2 and Enterocytes-S3 and subsequently decreased in IGC cells. This dynamic pattern was consistent with the pseudotime trajectory and suggested that Enterocytes-S3 represents an IM-like epithelial state along the epithelial transformation continuum, further supporting its association with the stepwise transition from non-malignant to malignant epithelial states. To explore transcriptional programs associated with Enterocytes-S3 cells, we conducted Weighted Gene Co-expression Network Analysis (WGCNA). The identified co-expression network was composed of 25 modules (Figure 2F). Module 4 was significantly upregulated in Enterocytes-S3 subtypes (Figure 2G) and enriched for pathways associated with mitotic nuclear division and mitotic cell-cycle phase transition (Figure 2H). Moreover, Enterocytes-S3 displayed enhanced ligand-receptor interactions with multiple cell types compared with Enterocytes-S2 (Figure S2G). Additionally, Enterocytes-S3 exhibited enhanced proliferative activity compared with earlier epithelial populations, as indicated by increased signature scores for the Hallmark_G2M_Checkpoint and Hallmark_E2F_Targets pathways (Figures S2H and S2I). Together, these findings suggest that Enterocytes-S3 represents a transitional epithelial state with IM-like and cell-cycle-associated features, which is associated with the progression from intestinal metaplasia toward malignant epithelial states. PTTG1 is associated with the Enterocytes-S3 state and malignant progression in IGC To define the molecular profile of the Enterocytes-S3 subpopulation, we determined its set of highly expressed genes (Table S4). Applying the Enterocytes-S3 signature to score the TCGA-STAD transcriptome data, we identified PTTG1 as the most significantly upregulated gene in samples with high Enterocytes-S3 scores (Figure 3A; Table S5). Pseudotime trajectory analysis further showed progressive induction of PTTG1 from enterocytes toward IGC, with the highest expression observed in IGC-S3 cluster (Figures 3B and 3C). Across the TCGA-STAD, ACRG, and HRA008551 datasets, both the Enterocytes-S3 signature score and PTTG1 expression were significantly elevated in GC compared with normal tissues (Figures S3A–S3F). Paired analyses confirmed increased PTTG1 expression in IGC relative to matched normal mucosa (Figures 3D–3F). Identification of PTTG1 as a potential regulatory factor associated with IGC progression (A) Volcano plot showing genes associated with the Enterocytes-S3 signature score in the TCGA-STAD cohort. (B) Pseudotime expression trends of PTTG1. (C) Violin plot of PTTG1 expression in epithelial subpopulations. (D and E) Boxplots showing PTTG1 expression in paired normal and IGC tissues from the TCGA-STAD (D) (n = 8 patients; Wilcoxon signed-rank test; ∗p < 0.05) and ACRG (E) (n = 38 patients; Wilcoxon signed-rank test; ∗∗∗p < 0.001) cohorts. (F) Boxplots showing PTTG1 expression in matched normal, IM, and IGC tissues from the HRA008551 cohort (n = 19 patients; Wilcoxon signed-rank test; ∗p < 0.05, ∗∗∗p < 0.001). (G) Representative IHC images and quantification of PTTG1 in paired normal and IGC tissues from the TMA (n = 64 patients; Wilcoxon signed-rank test; ∗∗∗p < 0.001). Scale bars, 200 μm or 50 μm, as indicated. (H) Representative IHC images and quantification of PTTG1 in IGC (n = 64 patients) and DGC (n = 24 patients) tissues (Wilcoxon rank-sum test; ∗p < 0.05). Scale bars, 200 μm or 50 μm, as indicated. Furthermore, we performed immunohistochemistry (IHC) on a clinical TMA comprising paired normal and tumor samples, including 64 cases of IGC and 24 cases of DGC. IHC demonstrated strong nuclear staining of PTTG1 in IGC samples, in contrast to weak or absent staining in adjacent normal tissues (Figure 3G, left). Quantitative H-Score analysis showed significantly elevated PTTG1 expression in IGC tumors compared with matched normal tissues (p < 0.001, Figure 3G). Moreover, the levels of PTTG1 expression were significantly associated with histological subtype (Table S6), with higher expression in IGC than in DGC (p = 0.034, Figure 3H). Consistently, both the Enterocytes-S3 signature score and PTTG1 expression were elevated in IGC compared with DGC across the TCGA and ACRG cohorts (Figures S3G–S3J). These results identified PTTG1 as a prominent gene associated with the Enterocytes-S3 state and supported its potential involvement in intestinal-type gastric cancer progression. PTTG1 promotes malignant phenotypes of GC cells by regulating the G2/M checkpoint To investigate the functional role of PTTG1 in gastric cancer, we silenced its expression in MKN1 and MKN7 cells using small interfering RNA (siRNA) targeting PTTG1. The knockdown efficiency was confirmed at both the mRNA and protein levels using RT-qPCR (Figure S4A) and western blotting (Figures S4B and S5), respectively. Functional assays, including CCK-8 and EdU assays, showed that PTTG1 silencing significantly inhibited cell proliferation (all p < 0.01, Figures S4C and S4D). Additionally, Transwell assays showed that PTTG1 knockdown reduced the migratory and invasive abilities of GC cells (all p < 0.01, Figures S4E and S4F). To elucidate the underlying mechanism, we performed RNA-seq analysis in GC cells after PTTG1 knockdown to identify molecular programs associated with malignant phenotypes. GO enrichment analysis of differentially expressed genes (adjusted p < 0.05, |log2FC| > 1.5) revealed significant enrichment of cell-cycle-related processes, including mitotic nuclear division and sister chromatid segregation (Figures S6A and S6B). Gene set enrichment analysis (GSEA) further indicated alteration of the G2/M checkpoint pathway (both p < 0.001, Figure S6C; Figure 4A). Consistently, flow cytometry demonstrated a reduction of G2-phase cells (both p < 0.01) and accumulation in G1-phase (both p < 0.05) upon PTTG1 silencing in MKN1 and MKN7 (Figure 4B). Western blotting confirmed downregulation of Cyclin B1 and p-CDK1, along with increased γH2AX levels (Figure 4C; Figure S7). Immunofluorescence (IF) further revealed increased γH2AX foci, consistent with enhanced DNA-damage-associated signaling (Figure 4D). Collectively, these findings indicated that PTTG1 supports the proliferative and invasive phenotypes of GC cells, at least in part, through G2/M checkpoint-associated regulation and DNA-damage-related responses. PTTG1 knockdown impaired G2/M progression in GC cells and increased DNA damage (A) Representative GSEA plots showing enrichment of G2/M checkpoint pathway. (B) Flow cytometry analysis of cell-cycle distribution in GC cells following PTTG1 knockdown (n = 3 biological replicates; Student’s t test; ∗p < 0.05, ∗∗p < 0.01 and ∗∗∗p < 0.001; ns, not significant). (C) Western blot analysis of cell-cycle-related and DNA-damage-associated proteins in GC cells with or without PTTG1 knockdown. Blots are representative of three independent experiments. (D) IF analysis showing increased γH2AX foci following PTTG1 knockdown. Images are representative of three independent experiments. Scale bars, 50 μm. (E) RT-qPCR analysis of PTTG1, E2F1, and CCNB1 expression (n = 3 biological replicates; Student’s t test; ∗∗∗p < 0.001). (F) Western blot analysis of PTTG1 and E2F1 protein levels. Blots are representative of three independent experiments. (G) Motif prediction and CUT&RUN-qPCR analysis of E2F1 binding to the CCNB1 promoter (n = 3 biological replicates; Student’s t test; ∗p < 0.05 and ∗∗p < 0.01; ns, not significant). Quantitative data are presented as mean ± SEM. To further investigate the molecular mechanism underlying PTTG1-mediated regulation of the G2/M checkpoint, we examined whether PTTG1 modulates Cyclin B1 expression through intermediate transcriptional regulators. Given that E2F1 has been reported to directly bind the CCNB1 promoter,35,36 we next assessed the involvement of E2F1 in this regulatory axis. PTTG1 knockdown significantly reduced E2F1 expression at both the mRNA and protein levels in MKN1 and MKN7 cells (Figures 4E and 4F; Figure S8), accompanied by a concomitant decrease in CCNB1 mRNA expression (Figure 4E). To further determine whether E2F1 directly regulates CCNB1 transcription in this context, we performed CUT&RUN-qPCR analysis. Motif prediction identified a putative E2F1-binding site within the CCNB1 promoter region. Notably, CUT&RUN-qPCR revealed significant enrichment of E2F1 at the CCNB1 promoter in control cells, which was markedly reduced upon PTTG1 knockdown (Figure 4G). These findings support a model in which PTTG1 regulates Cyclin B1 expression, at least in part, through modulation of E2F1-dependent transcriptional activity. PTTG1 overexpression enhances proliferation and mucin-related intestinal metaplasia features in gastric organoids To further evaluate the functional role of PTTG1 in gastric epithelial cells, we overexpressed PTTG1 in human paracancerous gastric organoids derived from IGC patients. Overexpression efficiency was confirmed by RT-qPCR, western blotting, and IF staining (Figures 5A–5C; Figure S9). Organoids with PTTG1 overexpression displayed significantly larger diameters on days 6, 10, and 12 compared with controls (all p < 0.01, Figure 5D). IF staining analysis revealed increased Ki67 expression (p = 0.043, Figures 5E and 5F), consistent with enhanced proliferative activity. Cyclin B1 expression was also elevated in the overexpression group (p < 0.001, Figures 5G and 5H), and nuclear localization of Cyclin B1 was more frequently observed in a subset of epithelial cells. Histological assessment demonstrated stronger Alcian blue staining in PTTG1-overexpressing organoids, suggesting increased acidic mucin-related features (Figure 5I). Given that mucin-related alterations may accompany intestinal metaplasia-like differentiation, we further examined MUC2 expression in gastric organoids. MUC2 IF staining showed stronger cell-associated MUC2 staining in PTTG1-overexpressing organoids (Figure 5J). Similarly, western blotting showed that the MUC2-immunoreactive band was increased in PTTG1-overexpressing organoids compared with control organoids (Figures 5K and S10). Together, these results suggest that PTTG1 overexpression promotes organoid growth and is associated with increased MUC2 expression and mucin-related intestinal metaplasia-like features in gastric organoids. PTTG1 overexpression promotes proliferation and intestinal metaplasia-like features in organoids (A–C) Paracancerous gastric organoids were transduced with a PTTG1-overexpressing lentiviral vector, and PTTG1 expression was assessed by RT-qPCR (A) (n = 3 biological replicates; Student’s t test; ∗∗p < 0.01), western blotting (B), and IF (C). Scale bars, 100 μm or 50 μm, as indicated. (D) Representative images of control and PTTG1-overexpressing organoids and quantification of organoid diameters on days 1, 6, 10, and 12 (n = 3 biological replicates; Student’s t test; ∗∗∗p < 0.001; ns, not significant). (E–H) Representative IF images of Ki67 (E) and Cyclin B1 (G) and quantification of the percentages of Ki67-positive (F) and Cyclin-B1-positive cells (H) in control and PTTG1-overexpressing organoids (n = 5; Student’s t test; ∗p < 0.05 and ∗∗∗p < 0.001). Scale bars, 100 μm or 50 μm, as indicated. (I) Representative images of Alcian blue staining in control and PTTG1-overexpressing organoids. Scale bars, 100 μm or 50 μm, as indicated. (J) Representative IF staining of MUC2 in control and PTTG1-overexpressing organoids. Scale bars, 50 μm. (K) Western blot analysis of PTTG1 and MUC2 protein levels in control and PTTG1-overexpressing organoids. Blots and images are representative of three independent experiments. Quantitative data are presented as mean ± SEM. PTTG1 silencing suppresses tumor growth and is associated with differentiation-related mucin changes in xenograft models Stable PTTG1-knockdown GC cell lines were generated by lentiviral transduction with two independent short hairpin RNAs (shRNAs) (shPTTG1#1 and shPTTG1#2) or a negative control (shNC). Knockdown efficiency was confirmed by RT-qPCR (Figure S11A) and western blotting (Figure 6A; Figure S12). Functionally, PTTG1 knockdown suppressed cell proliferation, migration, and invasion abilities in vitro (Figures S11B–S11D). Knockdown of PTTG1 suppresses tumor growth in vivo (A) MKN1 cells were transduced with shPTTG1 lentivirus to establish stable PTTG1-knockdown cells, and PTTG1 protein levels were determined by western blotting. Blots are representative of three independent experiments. (B) Tumor growth curves based on tumor volumes measured twice weekly (n = 10 tumors from 5 mice per group; two tumors per mouse; Student’s t test at the experimental endpoint; ∗∗p < 0.01 and ∗∗∗p < 0.001). (C) Representative gross images of the excised subcutaneous tumors at the experimental endpoint (n = 10 tumors per group). (D) Tumor weights at the experimental endpoint (n = 10 tumors per group; Student’s t test; ∗p < 0.05 and ∗∗p < 0.01). (E) Representative IHC images and quantification of the percentages of Ki67-positive cells in xenograft tumors (n = 10 tumors per group; Student’s t test; ∗p < 0.05). Scale bars, 100 μm or 50 μm, as indicated. (F and G) Representative IHC images (F) and quantification (G) of PTTG1 and Cyclin B1 expression in xenograft tumors (n = 10 tumors per group; Student’s t test; ∗p < 0.05 and ∗∗∗p < 0.001). Scale bars, 100 μm or 50 μm, as indicated. (H) Representative AB-PAS staining of xenograft tumors (10 tumors from 5 mice per group). Scale bars, 100 μm or 50 μm, as indicated. Quantitative data are presented as mean ± SEM. Equal numbers of shPTTG1 or shNC cells were subcutaneously injected into 5-week-old nude mice. Tumor growth was monitored every 3–4 days (Figure 6B), and PTTG1 knockdown significantly suppressed tumor progression (Figure 6C), as evidenced by reduced tumor volume (both p < 0.01, Figure 6B) and weight (both p < 0.05, Figure 6D). Immunohistochemistry further showed attenuated proliferation in shPTTG1 tumors, with reduced Ki67 and Cyclin B1 expression and lower H-Score (all p < 0.05, Figures 6E–6G). Alcian blue-periodic acid-Schiff (AB-PAS) staining revealed increased mucin-related staining in shPTTG1 tumors, suggesting differentiation-associated mucin changes after PTTG1 depletion (Figure 6H). However, MUC2 immunohistochemistry showed no significant difference between shPTTG1 and control tumors (Figure S11E), indicating that the mucin-related histological changes observed by AB-PAS were not accompanied by a significant change in MUC2 expression. To further characterize the underlying mechanisms, we assessed apoptosis in the xenograft tumors by cleaved caspase-3 staining, which showed no significant change following PTTG1 knockdown (Figure S11F). Collectively, these results demonstrated that PTTG1 promotes gastric cancer progression in vivo by regulating G2/M checkpoint-related proteins expression, while its depletion impairs tumor growth and induces differentiation-associated mucin changes. Consistent with its role in tumor-associated biological processes, survival analyses in independent cohorts showed no significant association between PTTG1 expression and overall survival (p = 0.76 for TCGA-STAD and p = 0.23 for ACRG; Figure S13), suggesting that PTTG1 may primarily reflect tumor-intrinsic regulatory programs rather than serve as an independent prognostic factor. Discussion scRNA-seq has emerged as a transformative technology to resolve tumor heterogeneity and infer cell-state dynamics in gastric cancer.37,38 In intestinal-type gastric cancer, it has proven particularly valuable for delineating transcriptomic alterations along the Correa cascade, spanning premalignant to malignant lesions.39,40 For instance, Zhang et al. profiled mucous-secreting cell and enteroendocrine cell populations, and identified two distinct subsets of goblet cells.41 Similarly, Su et al. investigated dynamic changes in the immune microenvironment and elucidated mechanisms of cell-cell interactions across four disease stages of IGC progression.42 In our study, we established a single-cell transcriptomic atlas of NAG, IM, and IGC tissues, revealing expansion of myeloid, endothelial, and fibroblast populations, together with an increased fraction of malignant epithelial cells. Within the epithelial compartment, pseudotime analysis highlighted a subset of enterocytes, designated Enterocytes-S3, which exhibited transcriptional features intermediate between non-malignant and malignant epithelial states. Notably, the incomplete intestinal metaplasia signature score increased from neck cells to Enterocytes-S3 and declined in IGC cells, providing support for its transcriptionally intermediate state. These findings suggest that Enterocytes-S3 may represent a putative transitional epithelial state along the gastric carcinogenesis continuum, although its direct tumor-initiating potential remains to be determined. WGCNA revealed a cell-cycle-related gene module that was highly activated in Enterocytes-S3. Our study identified PTTG1 as a key hub candidate, with steadily increasing expression along the pseudotime trajectory toward malignancy. Originally identified in rat pituitary tumor cells,43 PTTG1 is minimally expressed in most normal tissue except testis44 but is markedly overexpressed across multiple cancer types.45,46 Consistent with these observations, analyses of public datasets (TCGA-STAD and ACRG), together with our in-house transcriptomic data, demonstrated that PTTG1 expression was significantly elevated in IGC compared with normal or metaplastic tissues and was higher in IGC than in DGC. These findings were further validated by IHC on tissue microarrays, indicating that PTTG1 represents a molecular feature associated with IGC. Functional assays showed that PTTG1 is involved in promoting proliferation and invasion of gastric cancer cells, including MKN1 and MKN7 cell lines derived from intestinal-type gastric adenocarcinoma, whereas its depletion suppressed tumor growth both in vitro and in vivo. Previous studies have shown that PTTG1 participates in multiple signaling pathways, including PI3K/AKT signaling,47,48,49 epithelial-mesenchymal transition (EMT),50 Wnt/β-catenin signaling,51 cGMP-PKG signaling52 and mitogen-activated protein kinase (MAPK) signaling31 pathways. Transcriptomic profiling revealed that PTTG1 silencing led to downregulation of cell-cycle-related pathways, particularly the G2/M checkpoint. Importantly, PTTG1 depletion resulted in a reduced proportion of G2-phase cells, consistent with impaired G2/M progression rather than classical G2/M arrest. We further explored a potential regulatory axis linking PTTG1 to Cyclin B1 via E2F1, a known transcriptional regulator of CCNB1.35,36 PTTG1 knockdown reduced E2F1 expression, and CUT&RUN-qPCR demonstrated decreased E2F1 occupancy at the CCNB1 promoter. These findings suggest that PTTG1 may regulate Cyclin B1 expression, at least in part, through modulation of E2F1-dependent transcription, although the precise upstream mechanism by which PTTG1 regulates E2F1 remains unclear. In addition, PTTG1 has been reported to participate in DNA damage response and repair.53 Consistently, increased γH2AX levels indicated accumulation of DNA-damage-associated signals, suggesting a potential role of PTTG1 in coordinating cell-cycle progression with genomic stability. Notably, PTTG1 expression was not significantly associated with patient survival or clinicopathological parameters in public datasets, suggesting that it may primarily be involved in tumor-intrinsic biological processes rather than serve as an independent prognostic factor. Moreover, our organoid and xenograft models suggest that PTTG1 may be associated with mucin-related epithelial features in a context-dependent manner. In gastric organoids, PTTG1 overexpression was associated with intestinal metaplasia (IM)-like features, as indicated by increased Alcian blue staining and elevated MUC2 expression.54,55 In contrast, PTTG1 knockdown in xenograft tumors enhanced AB-PAS staining without a corresponding change in MUC2 expression. As AB-PAS detects both acidic and neutral mucins and does not distinguish mucin subtypes, this increase may reflect broader mucin-related histological changes rather than MUC2-specific intestinal metaplasia-like differentiation. These findings suggest that PTTG1 may influence mucin-associated epithelial phenotypes in a model-dependent manner. However, the precise relationship between PTTG1, mucin composition, and epithelial differentiation requires further investigation using more physiologically relevant systems. In conclusion, our study provides a single-cell-resolved view of epithelial state transitions during gastric carcinogenesis and identifies PTTG1 as a regulator associated with epithelial state transitions and IGC progression. These findings offer insight into the cellular dynamics of the Correa cascade and highlight PTTG1 as a candidate molecular feature warranting further investigation. Limitations of the study Our study also has several limitations. First, the spatial origin of individual cells along the gastric gland axis is lost during tissue dissociation for scRNA-seq. Second, although we sampled NAG, IM and IGC tissues from the same patients to approximate lesion progression, the cross-sectional approach cannot fully capture temporal dynamics; longitudinal sampling would more accurately reflect the stepwise transformation from preneoplastic to malignant states. Third, the scRNA-seq cohort included two female and three male participants, and the limited sample size precluded a robust assessment of sex-associated differences. Sex-stratified analyses were also not systematically performed in the TCGA, ACRG, or in-house bulk transcriptomic cohorts. Therefore, the potential influence of sex on the observed cellular and molecular alterations warrants further evaluation in larger, adequately powered cohorts. Fourth, the Enterocytes-S3 population was defined primarily based on transcriptomic features, and its direct functional role in tumor initiation or progression remains to be established. Fifth, although organoid and xenograft models suggested mucin-related epithelial changes associated with PTTG1 perturbation, more physiologically relevant culture systems, such as air-liquid interface organoid models, will be needed to further evaluate mature mucin production, localization, and epithelial differentiation. Finally, although our data suggest an indirect regulatory axis linking PTTG1 to Cyclin B1 via E2F1, the upstream mechanism by which PTTG1 regulates E2F1 requires further investigation. Resource availability Lead contact Further information and requests for resources and reagents should be directed to and will be fulfilled by the lead contact, Guangfu Jin (guangfujin@njmu.edu.cn). Materials availability This study did not generate new unique reagents. Data and code availability •Single-cell RNA-seq data have been deposited in the Genome Sequence Archive (GSA-Human) at National Genomics Data Center, China National Center for Bioinformation/Beijing Institute of Genomics, Chinese Academy of Sciences (accession no. HRA015134, https://ngdc.cncb.ac.cn/gsa-human/browse/HRA015134). All data reported in this study are available from the lead contact upon request. Single-cell RNA-seq data have been deposited in the Genome Sequence Archive (GSA-Human) at National Genomics Data Center, China National Center for Bioinformation/Beijing Institute of Genomics, Chinese Academy of Sciences (accession no. HRA015134, https://ngdc.cncb.ac.cn/gsa-human/browse/HRA015134). All data reported in this study are available from the lead contact upon request. •Bulk RNA-seq data derived from gastric cell lines have also been submitted to GSA-Human accession number HRA015866 (https://ngdc.cncb.ac.cn/gsa-human/browse/HRA015866). All data reported in this study are available from the lead contact upon request. Bulk RNA-seq data derived from gastric cell lines have also been submitted to GSA-Human accession number HRA015866 (https://ngdc.cncb.ac.cn/gsa-human/browse/HRA015866). All data reported in this study are available from the lead contact upon request. •No novel computational methods were developed in this study. All analyses were performed using publicly available tools, and the analysis scripts are available at https://github.com/NMUGCEPI/IGC_code. No novel computational methods were developed in this study. All analyses were performed using publicly available tools, and the analysis scripts are available at https://github.com/NMUGCEPI/IGC_code. •Additional information required to reanalyze the data reported in this paper is available from the lead contact upon reasonable request. Additional information required to reanalyze the data reported in this paper is available from the lead contact upon reasonable request. Acknowledgments The authors gratefully acknowledge the TCGA and ACRG for providing data access, as well as the developers of open-source tools that facilitated this research. This work was supported by grants from Noncommunicable Chronic Diseases-National Science and Technology Major Project (2025ZD0545200 and 2023ZD0501500), The National Natural Science Foundation of China (82125033, 82230110, 82388102, and 82574185), and Major Program of Wuxi Medical Center, Nanjing Medical University. We are grateful to all study participants and research staff for their essential contributions and commitment. Author contributions Writing – original draft, C.H.; writing – review & editing, G.J., C.H., and L.B.; investigation, C.H. and L.B.; visualization, C.H., L.B., L.L., and Z.Z.; validation, C.H., L.L., Z.Z., and X.G.; methodology, C.H., Z.Z., and Y.C.; data curation, C.H. and C.Y.; formal analysis, C.H., L.B., L.L., and Z.Z.; software, L.B., L.L., and Z.Z.; resources, G.J., F.L., D.Y., Y.L., G.L., and H.X.; conceptualization, C.Y. and G.J.; supervision, C.Y. and G.J.; project administration, C.Y. and G.J.; funding acquisition, G.J. All authors read and approved the final manuscript. Declaration of interests The authors declare no competing interests. STAR★Methods Key resources table REAGENT or RESOURCESOURCEIDENTIFIERAntibodiesanti-PTTG1 antibodyCSTCat# 13445Tanti-Ki67 antibodyCSTCat# 9027Santi-Cyclin B1 antibodyCSTCat# 12231Santi-p-CDK1 antibodyCSTCat# 4539Tanti-MUC2 antibodyAbcamCat# ab272692; RRID: AB_2888616anti-γH2AX antibodyCSTCat# 9718Santi-E2F1 antibodyCSTCat# 3742S; RRID: AB_2096936anti-β-ACTIN antibodySigma-AldrichCat# A5441; RRID: AB_476744Alexa 594 donkey anti-rabbit secondary antibodiesThermo Fisher ScientificCat# A11012; RRID: AB_141359Alexa 488 donkey anti-rabbit secondary antibodiesThermo Fisher ScientificCat# A32731; RRID: AB_2633280Alexa 647 donkey anti-rabbit secondary antibodiesThermo Fisher ScientificCat# A32733; RRID: AB_2633282DAPIThermo Fisher ScientificCat# 62248Goat-anti-mouse IgGJackson ImmunoResearchCat# 115-035-166; RRID: AB_2338511Goat-anti-rabbit IgGJackson ImmunoResearchCat# 111-035-144; RRID: AB_2307391HRP-conjugated anti-rabbit secondary antibodiesZSGB-BioCat# PV-6001; RRID: AB_2864333Bacterial and virus strainssiNCRiboBioN/AsiPTTG1#1 (5′-3′): GAAGACUGUUAAAGCAAAARiboBioN/AsiPTTG1#2 (5′-3′): GAUGAUGCCUAUCCAGAAARiboBioN/AshNCShanghai GenechemN/AshPTTG1#1 (5′-3′): GAAGACUGUUAAAGCAAAAShanghai GenechemN/AshPTTG1#2 (5′-3′): GAUGAUGCCUAUCCAGAAAShanghai GenechemN/APTTG1 overexpressionShanghai GenechemN/ABiological samplesHuman gastric tissue samplesThe Affiliated Cancer Hospital of Nanjing Medical University, The First Affiliated Hospital of Nanjing Medical UniversityThis studyHuman gastric organoidThe First Affiliated Hospital of Nanjing Medical UniversityThis studyNude mice tissue samplesthe Experimental Animal Center of Nanjing Medical UniversityThis studyChemicals, peptides, and recombinant proteinssCelLiveTM Tissue Preservation SolutionSingleronCat# 1010012sCelLiveTM Tissue Dissociation SolutionSingleronCat# 1020012GEXSCOPE® red blood cell lysis bufferSingleronCat# 1200050002TrypLE™ Express EnzymeInvitrogenCat# 12605010TRIzolTM ReagentInvitrogenCat# 15596018PrimeScriptTM RT Master MixTaKaRaCat# RR036ATB Green Premix Ex TaqTaKaRaCat# RR420ARIPA lysis bufferBeyotimeCat# P0013Jphosphatase inhibitor cocktailBimakeCat# B15002protease inhibitor cocktailMedChemExpressCat# HY-K0010-1MLChemiluminescent HRP SubstrateMerck MilliporeCat# WBKLS0500Cultrex Organoid Harvesting SolutionTrevigenCat# 3700-100-01Matrigel Growth Factor Reduced (GFR) Basement Membrane MatrixCorningCat# 356231Lipofectamine™ RNAiMAX Transfection ReagentInvitrogenCat# 13778150RPMI1640BICat# 01-100-1ACSHuman Gastric Epithelial Organoid KitbioGenusCat# K2004-HGCritical commercial assaysCell Counting Kit-8DojindoCat# CK04Cell LightTM EdU DNA Cell Proliferation KitRiboBioCat# C10310-1(PI)/RNase staining solutionDojindoCat# C543Alcian Blue-Periodic acid-Schiff Stain KitSolarbioCat# G1285-6∗50mLHyperactive pG-MNase CUT&RUN Assay Kit for PCR/qPCRVazymeCat# HD101-01Deposited dataCell lines RNA-seq analysisThis paperHRA015866https://ngdc.cncb.ac.cn/gsa-human/browse/HRA015866Gastric tissue samples scRNA-seq analysisThis paperHRA015134https://ngdc.cncb.ac.cn/gsa-human/browse/HRA015134The Cancer Genome Archive (TCGA) RNA-seq dataUCSC XenaTCGA-STADAsian Cancer Research Group (ACRG) RNA-seq dataGEOGSE66229NJGC RNA-seq dataGSA-HumanHRA008551Experimental models: Cell linesMKN1 (RRID: CVCL_1415)the Institute of Biochemistry and Cell Biology of the Chinese Academy of SciencesN/AMKN7 (RRID: CVCL_1417)the Institute of Biochemistry and Cell Biology of the Chinese Academy of SciencesN/AExperimental models: Organisms/strainsBALB/c nude miceGemPharmatech Co., Ltd (Nanjing, China)N/AOligonucleotidesPTTG1Forward5′-TGAATGCGGCTGTTAAGACCT-3′Reverse5′-TTTGATTGAAGGTCCAGACCCC-3′Sangon, Shanghai, ChinaN/AE2F1Forward5′-ACGTGACGTGTCAGGACCT-3′Reverse5′-GATCGGGCCTTGTTTGCTCTT-3′Sangon, Shanghai, ChinaN/ACCNB1Forward5′-AACTTTCGCCTGAGCCTATTTT-3′Reverse5′-TTGGTCTGACTGCTTGCTCTT-3′Sangon, Shanghai, ChinaN/AACTBForward5′-TAAGGAGAAGCTGTGCTACG-3′Reverse5′-GACTCGTCATACTCCTGCTT-3′Sangon, Shanghai, ChinaN/ACCNB1 (CUT&RUN-qPCR)Forward5′-GCAGAGGAGAGCAAGGGTAAG-3′Reverse5′-GGCTGAAGGACAAAAACATTGCT-3′Sangon, Shanghai, ChinaN/ASoftware and algorithmsImageJ softwareImageJhttps://imagej.net/ij/GraphPad PrismGraphpadhttps://www.graphpad.comNovaSeq 6000 platformIlluminaN/AQuantStudio 7 Flex Real-time qPCR SystemApplied BiosystemsN/ABD FACSCanto II flow cytometerBD BiosciencesN/ALSM700 confocal microscopeZEISSN/AR software (v4.1.1)The R Foundationhttps://www.r-project.org/CeleScope (v1.1.8)Singleronhttps://github.com/singleron-RD/CeleScopeCutadapt (v1.17)Martin et al.56https://github.com/marcelm/cutadaptSTAR v2.5.3aDobin et al.57https://github.com/alexdobin/STARfeatureCounts (v2.0.1)Liao et al.58https://subread.sourceforge.net/Seurat (v4.4.0)Hao et al.59https://satijalab.org/seurat/inferCNV (v1.11.1)Tickle et al.60https://github.com/broadinstitute/inferCNVMonocle (v2.30.1)Trapnell et al.61https://cole-trapnell-lab.github.io/monocle-release/hdWGCNA (v0.2.18)Morabito et al.62https://smorabit.github.io/hdWGCNA/clusterProfiler (v4.10.0)Wu et al.63https://github.com/YuLab-SMU/clusterProfilerCellPhoneDB (v3.0.0)Garcia-Alonso et al.64https://github.com/ventolab/CellphoneDBGSVA (v1.50.0)Hanzelmann et al.65https://github.com/rcastelo/GSVACodes and scriptsThis paperhttps://github.com/NMUGCEPI/IGC_code Experimental model and study participant details Sample collection Gastric tissue specimens were obtained from patients diagnosed with IGC who underwent surgical resection, including IGC tissues, premalignant lesions (≤ 2 cm to tumor margin), and matched non-atrophic gastritis tissues (≥ 5 cm away from tumor margin). For each sample, one portion was snap-frozen for histopathological evaluation, while the remaining portion was freshly processed for scRNA-seq. All histopathological diagnoses were independently confirmed by two pathologists. A total of 13 samples were collected from five IGC patients (detailed demographic and clinical information, including sex and age, is provided in Table S1), comprising 5 tumor tissues, 5 IM tissues and 3 NAG tissues. Written informed consent was obtained from all individuals prior to sample collection. The study was approved by the Ethics Committee of the First Affiliated Hospital of Nanjing Medical University (approval ID: 2023-SR-389) and the Affiliated Cancer Hospital of Nanjing Medical University (approval ID: 2020K-050). Cell culture Human gastric cancer cell lines MKN1 (RRID: CVCL_1415) and MKN7 (RRID: CVCL_1417) were obtained from the Institute of Biochemistry and Cell Biology of the Chinese Academy of Sciences (Shanghai, China) and authenticated by Short Tandem Repeat (STR) profiling analysis at Genetic Testing Biotechnology Corporation (Suzhou, China) within the past three years. All cells were maintained in RPMI 1640 medium supplemented with 10% FBS and cultured at 37°C in a humidified incubator containing 5% CO2. Routine mycoplasma testing confirmed that all cell cultures used in this study were free of contamination. Organoid isolation and culture Paracancerous tissue samples were obtained from patients who underwent surgical resection for intestinal-type gastric cancer at the First Affiliated Hospital of Nanjing Medical University, with pathological confirmation. All patients provided written informed consent, and the study was approved by the Ethics Committee of the First Affiliated Hospital of Nanjing Medical University (approval ID: 2023-SR-389). Gastric organoids derived from paracancerous tissues were established and maintained in a standard organoid culture system.66 For tissue preparation, gastric samples were rinsed and dissected into fragments of approximately 5 mm2. Tissue fragments were incubated in a chelating buffer consisting of PBS supplemented with 10 mM EDTA, 0.5 mM DL-dithiothreitol, antibiotics and Primocin at 4°C for 45 minutes. Following mechanical dissociation, released gastric glands were collected, washed with PBS, and resuspended in Matrigel. A drop of 50 μL Matrigel-cell suspension was dispensed into a 24-well plate and overlaid with 500 μL of gastric organoid medium. Organoids were cultured at 37°C in a humidified incubator containing 5% CO2 and the culture medium was replaced every 2-3 days. Mouse xenograft experiment Male BALB/c nude mice (4-6 weeks old) were randomly divided into three groups (n = 5 per group, 10 tumors per group). Mice were housed in a SPF facility under a 12-hour light/dark cycle with free access to autoclaved food and water. MKN1 cells stably expressing shNC, shPTTG1#1, or shPTTG1#2 (2 × 106 cells per group) were subcutaneously injected into the bilateral axillary regions of each mouse. Tumor size was measured twice weekly, and tumor volume was calculated as (width2×length)/2. Mice were sacrificed 5 weeks after injection, and all tumors were excised and weighed. All animal experiments complied with the National Institutes of Health’s Guide for the Care and Use of Laboratory Animals, and were approved by the Committee on the Ethics of Animal Experiments of Nanjing Medical University (Approval number: 1809026-2). Method details Tissue dissociation and preparation Freshly resected specimens were transferred to sCelLiveTM Tissue Preservation Solution (Singleron, 1010012) on ice within 30 minutes after resection and rinsed with phosphate-buffered saline (PBS). The specimens were washed three times with Hanks Balanced Salt Solution (HBSS) and dissociated in sCelLiveTM Tissue Dissociation Solution (Singleron, 1020012) using Singleron PythoN™ Tissue Dissociation System at 37 °C for 15 minutes. The cell suspension was filtered through a 40 μm sterile strainer, followed by red blood cell removal using GEXSCOPE® red blood cell lysis buffer (RCLB, Singleron, 1200050002) at room temperature for 5-8 minutes. Cells were pelleted by centrifugation at 300×g for 5 minutes at 4 °C, and gently resuspended in PBS. Single-cell RNA sequencing Single-cell suspensions were loaded onto microwell chips using the Singleron Matrix Single Cell Processing System. Barcoding beads were collected, and captured mRNA was reverse transcribed into cDNA, followed by PCR amplification. The amplified cDNA was fragmented and ligated to sequencing adapters. Libraries were constructed using the GEXSCOPE Single Cell RNA Library Kits according to the manufacturer’s protocol.67 Individual libraries were diluted to 4 nM, pooled, and sequenced on an Illumina NovaSeq 6000 platform with 150 bp paired-end reads. Raw reads from scRNA-seq were processed to generate gene expression matrices using CeleScope (https://github.com/singleron-RD/CeleScope) (v1.1.8) pipeline. Low-quality reads and adapter sequences were removed with Cutadapt (v1.17),56 followed by extraction of cell barcodes and UMIs. After that, we used STAR v2.5.3a to map reads to the reference genome GRCh37 (Ensembl version 75 annotation).57 UMI counts and gene counts of each cell were acquired with featureCounts (v2.0.1) software, and used to generate expression matrix files for subsequent analyses.58 Quality control, cell clustering and annotation Cells with UMI counts < 1,000 or > 20,000, gene counts < 200 or > 4,000, and mitochondrial content > 50% were excluded.37 After filtering, a total of 55,785 cells were retained for subsequent analyses. Data preprocessing and unsupervised clustering were conducted using Seurat (v4.4.0).59 In the first round of clustering, the top 2,000 variable genes were identified and used for principal component analysis (PCA). Clustering was performed based on the top 30 principal components, with batch effects corrected using canonical correlation analysis (CCA) by treating individual patients as the source of batch effects. As all cells from different patients mixed well, we did not observe obvious batch effect. Then we used a resolution of 0.3 to identify clusters and annotated them as major cell types. Cell clusters were visualized in two-dimensional space using UMAP. The cell type identity of each cluster was assigned based on the expression of the known marker genes reported previously.68 Cells expressing marker genes associated with more than one major lineage were classified as undefined cells and excluded from further analyses. Major cell types were re-clustered and annotated based on the unsupervised clustering and marker gene expression profiles. Next, we performed a second round of clustering to further characterize subpopulations of epithelial cells. Among these epithelial cells, malignant cells were further distinguished from non-malignant cells by inferring large-scale copy-number variations (CNVs) of each cell using inferCNV (v1.11.1) R package as described.60 For the clustering of epithelial cells, top 30 principal components were selected on the basis of 2,000 HVGs (resolution = 0.3). Trajectory analysis Cell differentiation trajectory was inferred using Monocle (v2.30.1).61 New CellDataSet object from cluster-annotated Seurat object was used by the newCellDataSet function. Cells were ordered along the pseudotime trajectory based on the top 800 most significantly differentially expressed genes (q value) identified from comparisons across distinct cell types. Dimensionality reduction was performed using DDRTree and trajectories were visualized with “plot_cell_trajectory” function. Weighted gene co-expression network analysis (WGCNA) and gene ontology (GO) enrichment analysis Co-expression network analysis was constructed using the hdWGCNA R package (v0.2.18) to identify gene regulatory modules associated with cellular differentiation.62 A soft-thresholding power was selected using scale-free topology criteria. Co-expression modules were defined by hierarchical clustering coupled with the dynamic tree cut method. Module eigengenes (MEs) were calculated as the first principal component of each module, and hub genes were identified according to high intramodular connectivity (kME) and correlation with the module’s eigengene. Functional enrichment of hub genes was performed using the “clusterProfiler” R package (v4.10.0).63 Cell-cell interaction analysis Ligand-receptor interactions among different cell populations were analyzed using CellPhoneDB v3.0.0.64 Gene expression thresholds for ligands and receptors were determined based on the distribution of average log-transformed expression values across all cell types. The analysis was performed using the following command with default parameters for the counts data. The significant cell-cell interactions were defined as p < 0.05 and average log expression > 0.1. Furthermore, the specific numbers of significant ligand-receptor pairs between Enterocytes-S2/Enterocytes-S3 and other major cell types were calculated. For each directed interaction pair, the total count of significant crosstalk was aggregated. These interaction frequencies were subsequently visualized to characterize the communication intensity between enterocytes subtypes and the surrounding microenvironment. Differential expression analysis of PTTG1 Public gastric cancer transcriptomic datasets, including The Cancer Genome Atlas (TCGA) cohort (32 normal, 375 GC samples) and Asian Cancer Research Group (ACRG) dataset (GSE66229 contains 98 normal, 146 IGC and 135 DGC samples),69 were analyzed for PTTG1 expression. In addition, we used our own transcriptome data comprising tissues from 48 gastric cancer patients (36 NAG, 48 IM, 48 GC samples).70 Total RNA was isolated using the RNA Nano 6000 Assay Kit and sequenced on a NovaSeq 6000 platform (Illumina). Raw reads were aligned to the GENCODE v19 reference genome using STAR v2.5.3a, and gene expression levels were normalized using the trimmed mean of M-values (TMM). Construction and scoring of the Enterocytes-S3 gene signature To construct the Enterocytes-S3 gene signature, differential expression analysis was first performed to identify genes that were specifically highly expressed in Enterocytes-S3 cells compared with other cell types (logFC > 1). Subsequently, the ssGSEA method implemented in GSVA (v1.50.0) was used to calculate the Enterocytes-S3 signature score for each sample in bulk transcriptomic data based on this gene set.65 Tissue microarray (TMA) analysis TMA was performed as described previously.71 Tissue microarrays were generated using 99 matched pairs of gastric adenocarcinoma samples and adjacent paracancerous tissues obtained from the Second Affiliated Hospital of Soochow University, assembled from 1.6 mm diameter tissue cores on a single slide. Immunohistochemistry was conducted following established protocols. Sections were incubated with a rabbit anti-PTTG1 antibody, scanned with an OLYMPUS VS200 microscope, and analyzed with a Servicebio image analysis system. Protein expression levels were quantified using histochemical score (H-score).72 Owing to tissue loss or detachment during experiment processing, 88 paired samples with evaluable staining were included in the final analysis, comprising 64 IGC and 24 DGC. Infections and transfections The siRNA sequences targeting PTTG1 were designed and synthesized by RiboBio (Guangzhou, China). Transient transfection was performed using Lipofectamine™ RNAiMAX according to the manufacturer’s instructions. Lentiviral shRNAs and overexpression constructs were obtained from GeneChen (Shanghai, China) and used to generate stable cell lines, with empty vectors serving as controls. For lentiviral infection, cells were incubated with concentrated lentiviral particles in the presence of 8 μg/mL polybrene. After 6 hours of incubation, the medium was replaced with fresh complete medium. For organoid transduction, organoids were dissociated into small clusters using TrypLE, incubated with lentivirus-containing medium supplemented with polybrene for 4 hours, and replated in Matrigel. Real-time quantitative PCR and western blotting Total RNA was isolated from cell lysates using TRIzolTM Reagent (Invitrogen) according to the manufacturer’s instructions, followed by reverse transcription using the PrimeScriptTM RT Master Mix (TaKaRa, Tokyo, Japan). Real-time qPCR was performed in duplicate using TB Green Premix Ex Taq (TaKaRa) on the QuantStudio 7 Flex Real-time qPCR System (Applied Biosystems). Relative gene expression levels were normalized using the 2-ΔΔCt method, with ACTB serving as internal control. Cell lysates were prepared using RIPA lysis buffer (Beyotime) supplemented with phosphatase inhibitor cocktail (Selleck) and protease inhibitor cocktail (MedChemExpress). Equal amounts of protein were separated by SDS-PAGE and transferred onto PVDF membranes. Membranes were blocked with 5% non-fat milk and incubated overnight at 4°C with primary antibodies, followed by incubation with HRP-conjugated anti-mouse (Jackson ImmunoResearch, 115-035-166) and anti-rabbit (Jackson ImmunoResearch, 111-035-144) secondary antibodies. Signals were visualized using Chemiluminescent HRP Substrate (Merck Millipore) and imaged on a ChemiDocTM XRS+ scanner (Bio-Rad). CUT&RUN-qPCR assays CUT&RUN-qPCR was performed using a commercial kit. Briefly, gastric cancer cells were collected and incubated with ConA beads, followed by overnight incubation with specific antibodies. After incubation with pG-MNase, chromatin digestion was initiated by CaCl2 and subsequently stopped with stop buffer. DNA fragments were then collected and analyzed by qPCR. Cell proliferation assays Cell proliferation was assessed using the Cell Counting Kit-8 (CCK-8; Dojindo, Japan) and the Cell LightTM EdU DNA Cell Proliferation Kit (EdU, RiboBio, China). MKN1 and MKN7 cells were plated into 96-well plates at 3,000 and 1,500 cells per well, respectively. For CCK-8 assay, the OD values were recorded at 0, 24, 48, 72 and 96 hours using an Infinite M200pro microplate reader (Tecan, Switzerland). EdU incorporation was quantified as the proportion of EdU-labeled cells using the ArrayScan VTI high-content imaging system (Thermo Fisher Scientific). Transwell migration and invasion assay Cell migratory and invasive abilities were examined using Transwell chambers (8.0 μm pore size; Corning). For invasion assays, the upper chambers were pre-coated with 50 μL Matrigel, while migration assays were performed without coating. Transfected cells were suspended in serum-free medium and seeded into the upper chambers (200 μL per chamber), while complete medium was added to the lower chambers as a chemoattractant. After 48 h of incubation, cells remaining on the upper membrane surface were removed. Migrated or invaded cells on the underside were fixed with 4% paraformaldehyde, stained with 1% crystal violet, and photographed in five randomly selected fields. Cell numbers were quantified to evaluate migratory and invasive capacities. Cell cycle assay Cells were collected, washed with PBS, and fixed in ethanol at −20 °C for 2 hours. Following fixation, cells were incubated with a propidium iodide (PI)/RNase staining solution at room temperature for 15 minutes in the dark. DNA content was subsequently analyzed using BD FACSCanto II flow cytometer (BD Biosciences). Immunohistochemistry, immunofluorescence and immunocytochemistry staining Tumor tissues were fixed in 4% paraformaldehyde, embedded in paraffin, and sectioned at a thickness of 5 μm. For immunostaining, the following primary antibodies were used: rabbit anti-PTTG1 antibody (Cell Signaling Technology, 13445T, 1:200), rabbit anti-Ki67 antibody (Cell Signaling Technology, 9027S, 1:200), rabbit anti-Cyclin B1 antibody (Cell Signaling Technology, 12231S, 1:400), rabbit anti-MUC2 antibody (Abcam, ab272692, 1:500), rabbit anti-γH2AX antibody (Cell Signaling Technology, 9718S, 1:500). Paraffin sections were stained with HRP-conjugated anti-rabbit secondary antibodies (ZSGB-Bio), and all slides were scanned using OLYMPUS VS200 microscope and analyzed by a Servicebio image analysis system. Fluorescent staining was done using Alexa 594 donkey anti-rabbit (A11012), Alexa 488 donkey anti-rabbit (A32731) secondary antibodies and Alexa 647 donkey anti-rabbit (A32733) secondary antibodies (Thermo Fisher Scientific, 1:200). Nuclei were counterstained with DAPI (Thermo Fisher Scientific, 62248, 1:1000). Images were captured using a confocal microscope LSM700 (ZEISS). Alcian blue-periodic acid-Schiff (AB-PAS) staining Organoids were released from Matrigel using CultrexTM Organoid Harvesting Solution (Trevigen) and washed with PBS. Organoids were fixed with 4% paraformaldehyde at 4°C overnight, embedded in HistoGelTM (Thermo Fisher Scientific),54 and transferred to 70% ethanol for paraffin embedding. Sections were deparaffinized, rehydrated, and stained with Alcian Blue-Periodic acid-Schiff Stain Kit (Solarbio) following the manufacturer’s instructions. Quantification and statistical analysis Differences between independent groups were assessed using the Wilcoxon rank-sum test, whereas paired comparisons were performed using the Wilcoxon signed-rank test. Experimental data are presented as the mean ± standard error of the mean (SEM), and differences between two experimental groups were assessed using unpaired Student’s t test. The exact values of n, definitions of n, and numbers of independent experimental replicates are provided in the corresponding figure legends. All statistical tests were two-sided, and p < 0.05 was considered statistically significant. Statistical significance was denoted as follows: ∗p < 0.05, ∗∗p < 0.01 and ∗∗∗p < 0.001; ns, not significant (p ≥ 0.05). All statistical analyses were performed with R software (v4.1.1). Footnotes Supplemental information can be found online at https://doi.org/10.1016/j.isci.2026.117742. Supplemental information Document S1. Figures S1–S13 and Tables S1, S2, S4, and S6 Table S3. Differentially expressed genes among epithelial cell states (states 1–3) identified by pseudotime analysis Table S5. Differentially expressed genes in TCGA-STAD samples with high Enterocytes-S3 signature scores References Associated Data Supplementary Materials Document S1. Figures S1–S13 and Tables S1, S2, S4, and S6 Table S3. Differentially expressed genes among epithelial cell states (states 1–3) identified by pseudotime analysis Table S5. Differentially expressed genes in TCGA-STAD samples with high Enterocytes-S3 signature scores Data Availability Statement •Single-cell RNA-seq data have been deposited in the Genome Sequence Archive (GSA-Human) at National Genomics Data Center, China National Center for Bioinformation/Beijing Institute of Genomics, Chinese Academy of Sciences (accession no. HRA015134, https://ngdc.cncb.ac.cn/gsa-human/browse/HRA015134). All data reported in this study are available from the lead contact upon request. Single-cell RNA-seq data have been deposited in the Genome Sequence Archive (GSA-Human) at National Genomics Data Center, China National Center for Bioinformation/Beijing Institute of Genomics, Chinese Academy of Sciences (accession no. HRA015134, https://ngdc.cncb.ac.cn/gsa-human/browse/HRA015134). All data reported in this study are available from the lead contact upon request. •Bulk RNA-seq data derived from gastric cell lines have also been submitted to GSA-Human accession number HRA015866 (https://ngdc.cncb.ac.cn/gsa-human/browse/HRA015866). All data reported in this study are available from the lead contact upon request. Bulk RNA-seq data derived from gastric cell lines have also been submitted to GSA-Human accession number HRA015866 (https://ngdc.cncb.ac.cn/gsa-human/browse/HRA015866). All data reported in this study are available from the lead contact upon request. •No novel computational methods were developed in this study. All analyses were performed using publicly available tools, and the analysis scripts are available at https://github.com/NMUGCEPI/IGC_code. No novel computational methods were developed in this study. All analyses were performed using publicly available tools, and the analysis scripts are available at https://github.com/NMUGCEPI/IGC_code. •Additional information required to reanalyze the data reported in this paper is available from the lead contact upon reasonable request. Additional information required to reanalyze the data reported in this paper is available from the lead contact upon reasonable request.