Construction and verification of matrix stiffness-related prognostic model of gastric cancer based on single-cell analysis and in vitro   experiments     

Construction and verification of matrix stiffness-related prognostic model of gastric cancer based on single-cell analysis and in vitro experiments | Research Square window.SnipcartSettings = { analytics: { enabled: false } }; (function() { var accessVector = localStorage.getItem('access_vector') || ''; window.dataLayer = window.dataLayer || []; if (accessVector) { window.dataLayer.push({ user: { profile: { profileInfo: { snid: accessVector } } } }); } })(); (function(w,d,s,l,i){w[l]=w[l]||[];w[l].push({'gtm.start':new Date().getTime(),event:'gtm.js'});var f=d.getElementsByTagName(s)[0],j=d.createElement(s),dl=l!='dataLayer'?'&l='+l:'';j.async=true;j.src='https://www.googletagmanager.com/gtm.js?id='+i+dl;f.parentNode.insertBefore(j,f);})(window,document,'script','dataLayer','GTM-K279D39R'); Browse Preprints In Review Journals COVID-19 Preprints AJE Video Bytes Research Tools Research Promotion AJE Professional Editing AJE Rubriq About Preprint Platform In Review Editorial Policies Our Team Advisory Board Help Center Sign In Submit a Preprint Cite Share Download PDF Research Article Construction and verification of matrix stiffness-related prognostic model of gastric cancer based on single-cell analysis and in vitro experiments Wei Zhang, Wenzheng Li, Zhengxuan Zhang, Peichun Sun, Gang Wu This is a preprint; it has not been peer reviewed by a journal. https://doi.org/ 10.21203/rs.3.rs-8616596/v1 This work is licensed under a CC BY 4.0 License Status: Posted Version 1 posted You are reading this latest preprint version Abstract Objective Although there is a significant correlation between tumour matrix sclerosis and the progression of gastric cancer, there is still a lack of a prognostic model based on Stiffness-related genes. For this reason, we have developed the matrix Stiffness-related risk score (MSRS) for prognostic prediction. In addition, this study integrates single-cell RNA sequencing technology to identify the cell source of characteristic genes and analyse the heterogeneity of cancer-related fibroblasts (CAF) lines. Methods Candidate stiffness-related genes were isolated by intersecting TCGA-STAD differentially expressed genes with four curated gene sets (e.g., Integrin, YAP). We constructed the MSRS via LASSO Cox regression in the GSE26253 training cohort (n = 432; endpoint: RFS) and validated it in the TCGA-STAD cohort (n = 386; endpoint: OS). Prognostic efficacy was gauged using Kaplan-Meier curves, ROC analyses, and nomograms, while ssGSEA assessed immune infiltration disparities. Furthermore, scRNA-seq data (GSE183904) was leveraged to dissect CAF heterogeneity, and core protein levels were ultimately corroborated by Western blotting. Results A 25-gene MSRS prognostic model was established. In the training cohort, high-risk patients exhibited significantly shorter recurrence-free survival (HR = 4.00, P < 0.0001), with 3-, 5-, and 7-year AUCs of 0.770, 0.791, and 0.770, respectively. The validation cohort confirmed significantly reduced overall survival in the high-risk group ( P = 0.0016; AUCs: 0.648, 0.772, 0.708). Multivariate analysis confirmed the risk score as an independent prognostic indicator (HR = 3.50, P = 1.71×10⁻²¹). Three independent markers were identified: the oncogene MATN3 (HR = 1.559) and tumor suppressors CTSG (HR = 0.693)and MADCAM1 (HR = 0.813). Western blotting confirmed MATN3 upregulation and CTSG/MADCAM1 downregulation in tumor cells. High-risk patients showed increased monocyte infiltration but decreased activated B and NK cells. scRNA-seq revealed that stromal cells exhibited the highest stiffness scores. Specifically, inflammatory CAFs (iCAFs) displayed the highest score ( P = 2.6×10⁻¹⁵) and were enriched in deep tumor regions. gastric cancer matrix stiffness prognostic model scRNA sequencing analysis cancer-related fibroblasts immune infiltration Figures Figure 1 Figure 2 Figure 3 Figure 4 Figure 5 Introduction Gastric cancer (GC) is still the fifth most common malignant tumour and the fourth leading cause of cancer-related death in the world, with more than 1 million new cases every year, resulting in about 650,000 deaths [ 1 ] . Due to the hidden onset and the lack of specific early symptoms, most patients are diagnosed in the late stage, resulting in an extremely unoptimistic 5-year survival rate [ 2 ] . Therefore, clarifying the molecular mechanism of gastric cancer and determining reliable prognostic biomarkers is crucial to improving the clinical treatment effect. Tumour microenvironment (TME) plays a key role in the process of tumour development. The latest research shows that in addition to traditional biochemical signal conduction, the physical properties of TME - especially matrix Stiffness - have a crucial regulatory effect on the behaviour of tumour cells [ 3 ] . Tissue Stiffness is mainly determined by the composition and crosslinking density of the extracellular matrix (ECM). In solid tumours, abnormal ECM deposition and collagen fibre cross-linking lead to a significant increase in Stiffness [ 4 ] . This mechanical sclerosis drives tumour progression through a variety of mechanisms: the cell's perception of the Stiffness of the matrix activates the integrin receptor, thus triggering the downstream sticky spot kinase (FAK) cascascading reaction [ 5 ] . At the same time, the Yes-related protein (YAP), which is a key mechanical conduction factor, will be translocated to the nucleus in a hard environment and regulate the transcription of proliferation and migration-related genes [ 6 ] . In addition, matrix Stiffness assists tumour immune escape by hindering T cell infiltration and promoting the formation of immunosuppressive microenvironment [ 7 ] . In gastric cancer, extracellular matrix remodelling is essentially related to the progression of the disease [ 8 ][ 9 ] . Excessive collagen deposition is significantly related to the depth of tumour infiltration, lymph node metastasis and poor prognosis [ 10 ] . Cancer-related fibroblasts (CAFs) are recognised as the main effect cells for ECM remodelling and hardening [ 11 ][ 12 ] . Although single-cell sequencing technology has revealed the heterogeneity of CAF - subgroups such as myofibroblast type (myCAF), inflammatory type (iCAF) and antigen presentation type (apCAF) [ 13 ][ 14 ] , the specific association mechanism between these subtypes and matrix sclerosis has yet to be clarified. Existing research mostly focusses on isolated signalling pathways and lacks systematic integration of Stiffness-related gene sets. In order to fill this gap, this study integrates four key gene sets: integrin pathways, YAP characteristics, ECM tissue and collagen formation. The matrix Stiffness-related risk score (MSRS) was constructed by LASSO Cox regression. It is worth noting that we use non-recurrent survival (RFS) as the training endpoint and total survival (OS) for verification to strictly evaluate the universality of the model in different clinical outcomes. Independent prognostic genes are identified by multivariable Cox regression, and multi-dimensional methods such as bioinformatics, protein imprinting and single-cell RNA sequencing are used for verification. This study aims to provide new insights into the mechanical biology of gastric cancer and establish potential targets for precise treatment. Materials and Methods Design and Overall Strategy To systematically evaluate the clinical significance and cellular origins of matrix stiffness-associated genes in gastric cancer, we implemented a comprehensive multi-stage analytical framework. The study proceeded through seven key phases: (1) Candidate Identification: Curating potential stiffness-related targets by intersecting multiple established gene sets; (2) Transcriptomic Profiling: Conducting differential expression and functional enrichment analyses within the TCGA cohort; (3) Model Construction: Developing a prognostic risk signature utilizing LASSO regression algorithms; (4) Independence Testing: Confirming the signature’s independent predictive value via multivariate Cox regression; (5) Dual Validation: Verifying model robustness in an external cohort and corroborating core gene expression through in vitro experiments; (6) Clinical Correlation: Assessing associations between risk scores, the immune microenvironment, and chemotherapeutic sensitivity; and (7) Cellular Mapping: Tracing the specific cellular provenance of the signature genes using scRNA-seq data. Matrix Stiffness-Related Gene Sets To comprehensively capture the genetic landscape of matrix stiffness, we curated four distinct gene sets from the Molecular Signatures Database (MSigDB; https://www.gsea-msigdb.org/ ) and published literature. These included the Integrin Pathway, the YAP Signature, ECM Organization, and Collagen Formation gene sets. Following the merging of these datasets and the removal of redundant entries, a total of 398 unique matrix stiffness-associated genes were identified for downstream analysis. TCGA-STAD Dataset Transcriptomic profiles (HTSeq-FPKM format) and associated clinical parameters—including age, gender, tumor stage, and survival outcomes—were retrieved from The Cancer Genome Atlas (TCGA) database for the Stomach Adenocarcinoma (STAD) cohort [ 15 ] . Following rigorous data curation and quality control procedures, a final cohort of 386 patients with complete survival records was retained to facilitate differential expression analysis and model validation. GEO Dataset The training cohort, identified as GSE26253, was retrieved from the NCBI Gene Expression Omnibus (GEO) repository. This dataset contains transcriptomic data derived from the Affymetrix Human Genome U133 Plus 2.0 array, alongside clinical annotations for 432 gastric cancer patients, specifically including recurrence-free survival (RFS) outcomes. [ 16 ] Single-Cell RNA Sequencing Data To determine the cellular origins of the signature genes and characterize CAF heterogeneity, we acquired the single-cell transcriptome dataset GSE167297. This cohort encompasses paired samples of normal tissue and tumor specimens (stratified into shallow and deep regions), providing a high-resolution reference for validating the model’s cellular sources. [ 17 ] TIMER3.0 Database Assessment of tumor immune infiltration was conducted via the TIMER3.0 web server( https://compbio.cn/timer3/ ). Specifically, the “Gene-DE” module was applied to examine the differential expression patterns of MATN3, CTSG, and MADCAM1 by comparing gastric cancer tissues against normal controls. Differential Expression Analysis Use the limma package to analyse the gene expression differences of the TCGA-STAD cohort in the R studio (v4.4.1). In order to achieve the comparability between samples, the original FPKM data is first converted into TPM (Transcripts Per Million) format, and log2 (TPM + 1) standardised processing is carried out. Subsequently, the empirical Bayes statistical framework was used to evaluate the expression level of gene differences [ 18 ] . The conditions for screening significantly differential expression genes (DEGs) are: |log2(FC)| > 1, and the P value (P.adj) after multiple tests and correction by the Benjaminini–Hochberg method is < 0.05. Functional Enrichment Analysis To decipher the underlying biological roles and signaling networks of the identified genes, we leveraged the ‘clusterProfiler’ package in R. This analysis primarily focused on Gene Ontology (GO) annotation, dissecting terms across three dimensions: biological processes (BP), cellular components (CC), and molecular functions (MF). Additionally, pathway enrichment was explored referencing the Kyoto Encyclopedia of Genes and Genomes (KEGG) database. For both analyses, a nominal P-value of less than 0.05 was established as the criterion for statistical significance. LASSO Regression and Prognostic Model Construction Use glmnet packages to implement LASSO regression in the R studio to mine key genes significantly related to prognosis. This method realises variable screening by introducing L1 regularisation terms, which helps to reduce the risk of overfitting in high-dimensional data. Take the GSE26253 cohort as the training data set, and set RFS as the survival outcome indicator. Use ten-fold cross-verification to determine the best penalty coefficient λ (lambda.min), select genes with non-zero regression coefficients according to this parameter, and construct an MSRS prognostic risk model. The MSRS risk score was calculated as: $$\:\varvec{R}\varvec{i}\varvec{s}\varvec{k}\varvec{S}\varvec{c}\varvec{o}\varvec{r}\varvec{e}=\sum\:_{\varvec{i}=1}^{\varvec{n}}(\varvec{C}\varvec{o}\varvec{e}{\varvec{f}}_{\varvec{i}}\times\:\varvec{E}\varvec{x}{\varvec{p}}_{\varvec{i}})$$ Among them, \(\:Coe{f}_{i}\:\) represents the LASSO regression coefficient of the ith gene, \(\:Ex{p}_{i}\:\) is the expression level of the ith gene, and n is the number of core genes. Cox Regression Analysis Use the R language survival package for Cox proportional risk regression analysis. First, single-factor Cox regression was carried out on 25 candidate core genes, and prognostic-related variables were screened ( P < 0.05). Genetics of statistical significance are incorporated into the multi-factor Cox model, and independent prognostic factors are determined by stepwise regression. The results are expressed in terms of risk ratio (HR) and 95% confidence interval (CI), and the difference of P < 0.05 is statistically significant. Survival Analysis and Model Performance Evaluation The Kaplan-Meier method (implemented via ‘survival’ and ‘survminer’ packages) was utilized for survival analysis, with patients stratified into high- and low-risk groups according to the median MSRS score. The statistical significance of survival discrepancies between groups was evaluated by the log-rank test. The prognostic performance of the model was assessed through time-dependent Receiver Operating Characteristic (ROC) analysis (‘timeROC’ package) at 3-, 5-, and 7-year follow-up intervals; an Area Under the Curve (AUC) value closer to 1 indicates superior predictive capability. Additionally, the ‘rms’ package was employed to construct a prognostic nomogram that integrates MSRS scores with clinical stages. The calibration degree of the model was evaluated via calibration diagrams, where optimal prediction accuracy is indicated when the predicted probability closely aligns with observed outcomes along the 45-degree reference line. Immune Cell Infiltration Analysis To estimate the infiltration density of various immune cell populations within the tumor microenvironment, we utilized the ‘GSVA’ R package to execute the single-sample gene set enrichment analysis (ssGSEA) algorithm. The specific marker signatures for these immune subsets were curated from established literature. Comparison of infiltration profiles across the high- and low-risk cohorts was achieved via the Wilcoxon rank-sum test. Furthermore, the potential linkage between the MSRS and immune cell abundance was explored using Spearman’s rank correlation. Statistical significance was defined by a P -value 0.1 deemed meaningful; positive ( r > 0) and negative ( r < 0) values indicated proportional and inverse relationships, respectively. Drug Sensitivity Analysis Target genes of commonly used targeted therapies for gastric cancer were compiled from published literature, including key targets such as Mesenchymal-Epithelial Transition factor (MET), Thymidylate Synthase (TYMS), Topoisomerase IIα (TOP2A), and erb-b2 Receptor Tyrosine Kinase 2 (ERBB2). Differential expression of these target genes between high- and low-risk groups was analyzed to infer potential drug sensitivity profiles in each risk group. Single-Cell RNA Sequencing Data Analysis The Seurat package was utilized to facilitate single-cell data processing. Following rigorous quality control filtration, a final dataset comprising 23,069 high-quality cells was selected for subsequent analysis. The data underwent normalization and unsupervised clustering to define distinct cellular subpopulations, which were projected visually using t-distributed Stochastic Neighbor Embedding (t-SNE). Cluster annotation was performed by referencing established marker genes, resulting in the characterization of seven primary cell lineages: T cells (CD3D, CD3E), B cells (CD79A, MS4A1), myeloid cells (CD68, CD14), epithelial cells (EPCAM, KRT18), endothelial cells (PECAM1, VWF), stromal cells (DCN, COL1A1), and mast cells (TPSAB1, CPA3).Compositional changes in cell types across different tissue regions (normal, superficial tumor, and deep tumor) were examined. The AddModuleScore function was used to calculate a stromal stiffness score for each cell, with the 25 core MSRS genes serving as the scoring gene set. Differences in stiffness scores among cell types and between normal and tumor tissues were compared using the Kruskal-Wallis test. Subclustering of stromal cells was performed to classify cancer-associated fibroblasts (CAFs) into three subtypes based on published marker genes: myCAF (ACTA2, TAGLN, MYL9), iCAF (IL6, CXCL12, PDGFRA), and apCAF (CD74, HLA-DRA, HLA-DRB1). Stiffness scores were compared across CAF subtypes, and proportional changes in CAF subtypes across tissue regions were analyzed. Cell Culture We obtained the immortalized human gastric epithelial cell line (GES-1) alongside three gastric adenocarcinoma lines (AGS, HGC-27, and MKN-45) from Wuhan Saiweier Biotechnology Co., Ltd. Cell propagation was conducted using Gibco’s RPMI-1640 basal medium, which was enriched with 10% fetal bovine serum (FBS) and a 1% penicillin-streptomycin mixture. Incubation took place in a humidified chamber regulated at 37°C with a 5% CO2 atmosphere. Routine subculturing was performed at intervals of 2–3 days. For all downstream assays, cells were harvested exclusively during their exponential growth phase. Western Blot Analysis Cells in the logarithmic growth were lysed on the ice with RIPA lysate (Beyotime) containing protease inhibitors and phosphatase inhibitors for 30 minutes. Centrifuge at 12,000 rpm for 25 minutes, take the supernatant, and use the BCA protein quantitative kit (Thermo Fisher) to determine the protein concentration. According to the quantitative results, take an equal sample for SDS-PAGE separation, and transfer the protein electricity to the PVDF membrane. After turning the membrane, use the Sanying Rapid WB antibody dilution solution (Proteintech PR20039) for dilution operation. After 5 minutes of rinsing the membrane with TBST, place it in the diluent at room temperature or 37°C and gently shake it for about 5 minutes. Then dilute anti-MATN3 1:1000, anti-CTSG 1:1000, anti-MADCAM1 1:1000, anti-GAPDH 1:5000) with this dilution and incubate at room temperature for 25 minutes. After incubation, wash with TBST 5 times for 1 minute each time. After that, use the same dilution solution to dilute HRP-labelled dianti, incubate at room temperature for 15 minutes, and wash again 3 times for 1 minute each time. After washing, add ECL development reagent for chemiluminescence detection, and use the gel imaging system to collect the results. GAPDH is used as an internal reference control for protein quantitative standardisation. Statistical Analysis All statistical computations and data management were executed within the R statistical framework (version 4.4.1). To differentiate continuous variables between two distinct groups, the Wilcoxon rank-sum test was employed; conversely, the Kruskal-Wallis test was selected for comparisons spanning three or more cohorts. The correlation intensity between variables was assessed using Spearman’s coefficient. We adopted a two-sided testing approach for all analyses, establishing statistical significance at P < 0.05. Visual representations denote significance levels as follows: ‘ns’ for P ≥ 0.05, with increasing significance marked by * ( P < 0.05), ** ( P < 0.01), *** ( P < 0.001), and **** ( P < 0.0001). Results Identification of Matrix Stiffness-Related Genes and Construction of the MSRS Model In order to identify matrix Matrix Stiffness-related genes in gastric cancer, this study integrated four signalling pathway gene sets of YAP, integrin, extracellular matrix tissue and collagen formation, and obtained a total of 398 matrix Matrix Stiffness-related genes (Fig. 1 A). The differential expression analysis of the TCGA-STAD data set identified 8,248 differential expression genes (DEGs), including 4,147 up-regulation genes and 4,101 down-regulation genes (Fig. 1 B). 237 candidate genes were obtained by intersecting DEGs with matrix Matrix Stiffness-related genes (Fig. 1 C). Figure 1 D shows the top 40 candidate genes sorted by difference multiples. Functional enrichment analysis shows that 237 candidate genes are significantly enriched in ECM-receptor interaction, adhesion spots and PI3K-Akt signalling pathways (KEGG analysis, Fig. 1 G). GO analysis shows that these genes are mainly involved in biological processes such as extracellular matrix tissue, collagen fibre assembly and cell-matric adhesion (Fig. 1 H). GSEA analysis further confirms the significant enrichment of adhesive spot pathways (NES = 1.93, P 0) and 7 were protective genes (LASSO coefficient < 0). Prognostic Significance and Experimental Validation of MSRS Core Genes To isolate genes with independent prognostic power, we subjected the 25 candidate genes to multivariate Cox regression analysis (Fig. 2 A). This screening identified three pivotal markers: MATN3 was characterized as a risk factor (HR = 1.56, 95% CI: 1.14–2.13, P = 0.005), whereas CTSG (HR = 0.69, 95% CI: 0.57–0.84, P < 0.001) and MADCAM1 (HR = 0.81, 95% CI: 0.68–0.98, P = 0.028) functioned as protective elements. Subsequent expression profiling via the TIMER3.0 database revealed that MATN3 levels were significantly upregulated in gastric cancer tissues compared to normal controls ( P < 0.001, Fig. 2 B). Conversely, observing the tumor samples showed a marked downregulation of CTSG ( P < 0.001, Fig. 2 C) and MADCAM1 ( P < 0.01, Fig. 2 D). To corroborate these in silico findings, Western blotting was performed to assess protein abundance in vitro. As anticipated, MATN3 protein expression was elevated in gastric cancer cell lines (AGS, HGC-27, and MKN-45) relative to the normal gastric epithelial line GES-1 (Fig. 2 E). In contrast, protein levels of both CTSG (Fig. 2 F) and MADCAM1 (Fig. 2 G) were substantially diminished in the cancer cells, aligning perfectly with the bioinformatic predictions. Validation of MSRS Prognostic Efficacy and Clinical Utility The GSE26253 cohort, comprising 432 patients, functioned as the training set to evaluate the predictive power of the MSRS. Subjects were dichotomized into high- or low-risk tiers utilizing the median score as the specific threshold. Kaplan-Meier plotting demonstrated that individuals in the high-risk bracket faced a notably inferior recurrence-free survival (RFS) relative to their low-risk counterparts ( P < 0.0001, Fig. 3 A). This prognostic trend was subsequently validated within the independent TCGA-STAD cohort (n = 386), where elevated risk scores were significantly associated with compromised overall survival (OS) ( P = 0.0016, Fig. 3 B).To gauge the model’s temporal accuracy, time-dependent ROC analysis was employed. Within the training dataset, the signature yielded Area Under the Curve (AUC) statistics of 0.770, 0.791, and 0.770 for the 3-, 5-, and 7-year intervals, respectively (Fig. 3 C). Parallel analysis in the validation group produced corresponding AUCs of 0.648, 0.772, and 0.708 (Fig. 3 D). Enhancing clinical utility, a composite nomogram was generated by merging the MSRS with clinical staging data to forecast individual outcomes (Fig. 3 E). Subsequent calibration assays revealed a high degree of concordance between the predicted probabilities and the actual survival events (Fig. 3 F). Interplay of MSRS with Immune Infiltration and Therapeutic Susceptibility ·We subsequently explored how MSRS stratification impacts the immunological landscape and chemotherapeutic potential. Infiltration analysis unveiled distinct cellular compositions: the high-risk cohort was characterized by a marked enrichment of monocytes. In contrast, the abundance of activated B cells, natural killer (NK) cells, and myeloid-derived suppressor cells (MDSCs) was substantially depleted in these patients (Fig. 4 A). Spearman correlation assays validated these observations, establishing a positive linear relationship between MSRS scores and both immature dendritic cells ( r = 0.153, P = 0.001) and monocytes ( r = 0.128, P = 0.008). Conversely, an inverse association was identified for MDSCs ( r = -0.155, P = 0.001), NK cells ( r = -0.147, P = 0.002), and activated B cells ( r = -0.136, P = 0.005) (Figs. 4 B–C). Regarding therapeutic implications, we analyzed the expression of critical drug targets. The results demonstrated that MET, TYMS, TOP2A (all P < 0.01), and ERBB2 ( P < 0.05) were significantly upregulated in the low-risk group relative to the high-risk counterparts (Fig. 4 D). These findings imply that patients falling into the low-risk category might be more favorable candidates for targeted treatments against these markers. Validation of MSRS Core Genes at Single-Cell Resolution To elucidate the cellular origins of MSRS core genes, we analyzed the GSE167297 single-cell RNA sequencing dataset. Following quality control, 23,069 cells were retained for downstream analysis. Unsupervised clustering identified 21 distinct cell clusters (Cluster 0–20) (Fig. 5 A), which were annotated into seven major cell types based on canonical marker genes: T cells, B cells, myeloid cells, epithelial cells, endothelial cells, stromal cells, and mast cells (Fig. 5 B). Cellular composition analysis revealed that stromal cells were most abundant in deep tumor regions, followed by normal tissues, and least prevalent in superficial regions (Fig. 5 C). Stromal stiffness scoring demonstrated that stromal cells exhibited the highest stiffness scores among all cell types, significantly surpassing other populations (Fig. 5 D). t-SNE visualization further confirmed that high-stiffness cells (red) predominantly localized within stromal cell clusters (Fig. 5 E). Notably, tumor tissues displayed significantly higher overall stiffness scores than normal tissues (Fig. 5 F). Moreover, stiffness scores progressively increased with tumor invasion depth (Normal < Superficial < Deep) (Fig. 5 G). Expression profiling of the 25 core genes revealed that the majority (including COL10A1, SPARC, ELN, COL11A1, COMP, COL1A1, and COL5A2) were predominantly enriched in stromal cells (Fig. 5 H). To further characterize the stromal compartment, we performed subclustering analysis on cancer-associated fibroblasts (CAFs). Based on established markers, CAFs can be divided into three subtypes: myofibroblast CAFs (myCAFs), inflammatory CAFs (iCAFs) and antigen-presenting CAFs (apCAFs) (Fig. 5 I). Stiffness scoring demonstrated that iCAFs exhibited the highest scores, followed by apCAFs and myCAFs (Kruskal-Wallis test, P = 2.6e-15) (Fig. 5 J). Subtype-specific marker expression patterns were as follows: myCAFs highly expressed ACTA2, TAGLN, and MYL9; iCAFs upregulated IL6, CXCL12, and PDGFRA; and apCAFs were enriched for CD74, HLA-DRA, and HLA-DRB1 (Fig. 5 K). Dynamic shifts in CAF subtype composition were observed across tissue regions (Fig. 5 L). From normal to superficial tumor regions, the proportions of myCAFs and apCAFs increased significantly, while iCAFs decreased. Conversely, from superficial to deep tumor regions, iCAF prevalence gradually rebounded, whereas myCAF and apCAF proportions declined. Discussion This study systematically analysed the prognostic value and cell source of these genes in gastric cancer by integrating multiple matrix Matrix Stiffness-related gene sets. Based on 25 matrix Matrix Stiffness-related genes, the MSRS prognosis model was successfully constructed, and three independent prognostic markers, MATN3, CTSG and MADCAM1, were identified. Single-cell analysis shows that these genes are mainly expressed in matrix cells, among which the iCAF subtype has the highest matrix Matrix Stiffness score. These findings provide a new perspective for understanding the role of matrix Matrix Stiffness in the development of gastric cancer, and provide potential targets for prognostic evaluation and targeted treatment.The study found that MATN3 was highly expressed in gastric cancer tissue and significantly associated with poor prognosis (HR = 1.56). As a member of the Matrilin protein family, MATN3 is an important ECM protein, which is mainly involved in the development and homeostasis maintenance of cartilage and bone tissue [ 19 ] . Under normal circumstances, MATN3 maintains the integrity and mechanical properties of the tissue structure by interacting with other ECM components (such as collagen and protean) [ 20 ] . Single cell analysis confirms that MATN3 is mainly highly expressed in tromal cells, which is consistent with the functional positioning of its ECM components. MATN3 may promote the progression of gastric cancer through two mechanisms: first, excessive deposition increases the density and Matrix Stiffness of ECM, providing a favourable mechanical microenvironment for tumour cells [ 20 ] ; second, activate the cancer-promoting signalling pathway by interacting with integtin receptors [ 21 ] . In contrast, CTSG and MADCAM1 as protective genes are lowly expressed in most cancers [ 22 ][ 23 ] . CTSG, biologically classified as a serine protease, is derived primarily from neutrophils and mast cells, playing a pivotal role in modulating inflammatory cascades and immune responses [ 24 ] . Prior investigations have highlighted its tumor-suppressive potential across various malignancies. For instance, in non-small cell lung cancer (NSCLC), downregulation of CTSG is correlated with lymphatic metastasis, advanced pathological staging, and a scarcity of infiltrating immune cells, patterns that are indicative of compromised immune surveillance [ 25 ] . Similarly, studies on head and neck squamous cell carcinoma (HNSCC) have established CTSG as a standalone prognostic indicator; its depletion reportedly disrupts immuno-inflammatory axes, such as the IL-17 pathway, thereby impairing the host’s ability to monitor tumor growth [ 26 ][ 27 ] . Furthermore, evidence from colorectal cancer (CRC) suggests that CTSG overexpression can impede tumorigenesis by blocking the Akt/mTOR signaling axis while simultaneously promoting the expression of apoptotic regulators [ 28 ] . Synthesizing these observations, we postulate that the diminished expression of CTSG may facilitate gastric cancer progression through multifaceted mechanisms. Mucosal vascular addressin cell adhesion molecule 1 (MAdCAM-1) is predominantly expressed on mucosal vascular endothelial cells, where it functions as a critical mediator of lymphocyte homing to the intestinal mucosa via interactions with the α4β7 integrin [ 29 ] . This interaction is fundamental to the maintenance of mucosal immunity. Consistent with this, our single-cell analysis corroborated that MADCAM1 is primarily enriched in endothelial populations. Clinical evidence suggests that the MAdCAM-1-α4β7 axis may serve as a vital checkpoint for cancer immunosurveillance; specifically, diminished levels of serum soluble MAdCAM-1 have been correlated with adverse prognoses in malignancies such as lung, kidney, and bladder cancer [ 23 ] . Furthermore, mutations in MADCAM1 are significantly associated with reduced metastasis-free survival. These mutations not only facilitate direct cancer cell migration but also foster an immunosuppressive microenvironment, potentially through mechanisms involving PD-L1-mediated immune escape and reprogramming [ 30 ] . Consequently, therapeutic strategies aimed at restoring or potentiating MAdCAM-1 expression hold promise for cancer immunotherapy. It is worth noting that this study found that the iCAF subtype has the highest matrix Matrix Stiffness score, significantly higher than myCAF and apCAF. This result is different from the traditional view. Previous studies believe that myCAF is the main effect cell of ECM remodelling and matrix sclerosis. Because of its high expression of α-SMA and a variety of collagen, it has a strong ability to synthesise and contract ECM [ 31 ] . However, this study suggests that the role of iCAF in matrix Matrix Stiffness regulation may be more important than previous understanding. The matrix Matrix Stiffness score used in this study is based on the calculation of 25 MSRS core genes, which not only include ECM structural components, but also include integrated element signalling pathways, YAP signalling pathways and other mechanical force perception and transduction-related genes, so it can more comprehensively reflect the response ability of cells to matrix Matrix Stiffness. Although iCAF does not directly produce a large number of ECM components, its highly expressed inflammatory factors (such as IL-6, CXCL12, etc.) can promote the ECM synthesis and crosslinking of neighbouring cells (including myCAF and tumour cells) through parasecretion. Studies show that IL-6 can upregulate the expression of aminoyl oxidase (LOX) by activating the STAT3 signalling pathway to promote collagen fibre cross-linking and matrix sclerosis [ 32 ] . In addition, the inflammatory factors secreted by iCAF can also recruit immune cells and reshape the physicochemical characteristics of the tumour microenvironment. Another intriguing finding was the distinct spatial distribution pattern of stromal cells, characterized by the highest density in deep tumor regions, intermediate levels in normal tissues, and the lowest density in superficial tumor areas. This pattern deviates from the simplistic "tumor > normal" paradigm, suggesting complex spatial heterogeneity. We propose the following explanations: (1) superficial tumor regions may represent the invasive front, where tumor cells actively degrade and remodel the ECM, resulting in limited stromal cell recruitment; (2) deep tumor regions may constitute the established tumor core, where stromal cells are stably integrated into a supportive matrix network; (3) normal tissues maintain physiological stromal cell levels, falling between these two extremes. This observation provides valuable insights into the dynamic evolution of the tumor microenvironment during cancer progression. Conclusions This study integrates multiple matrix Matrix Stiffness-related gene sets and constructs an MSRS prognostic model based on 25 core genes. The model showed good predictive performance in both the training set (with no recurrence survival as the end point) and the verification set (with total survival as the end point), and the consistency verification between different endpoints confirmed the generalisation ability of the model. The study found that the three molecular markers MATN3, CTSG and MADCAM1 have independent prognostic value. Single-cell analysis shows that the core genes are mainly expressed in matrix cells, among which the iCAF subtype has the highest matrix Matrix Stiffness score. This study provides new insights into understanding the role of matrix Matrix Stiffness in the progression of gastric cancer. MSRS can be used as a prognostic assessment tool for gastric cancer, and the core gene may become a potential therapeutic target. Declarations Funding ： The authors declare that no funds, grants, or other support were received during the preparation of this manuscript. Competing interests: The authors declare no competing interests. Data Availability ： Publicly available datasets: The gastric cancer microarray data (Training set) are available in the Gene Expression Omnibus (GEO) under accession number GSE26253, accessible at: https://identifiers.org/geo:GSE26253. The single-cell RNA sequencing data are available in the Gene Expression Omnibus (GEO) under accession number GSE183904, accessible at: https://identifiers.org/geo:GSE183904. The RNA-sequencing data (Validation set) are available in The Cancer Genome Atlas (TCGA-STAD), accessible via the GDC Data Portal at: https://portal.gdc.cancer.gov/projects/TCGA-STAD. Data generated in this study: The original contributions presented in the study (including Western Blot images and raw data) are included in the article/Supplementary Material. Further inquiries can be directed to the corresponding author. Author Contribution All authors conceived and designed the research; Wenzheng Li collected data; Gang Wu conducted research and analyzed and interpreted data; Wei Zhang and Wenzheng Li wrote the initial draft; Wei Zhang and Gang Wu revised the manuscript; Wei Zhang and Wenzheng Li had primary responsibility for final content. All authors read and approved the final version of the manuscript. References Patel AK, Sethi NS, Park H (2026) Gastric cancer: a review. JAMA Published online January 7 Wang J, Du L, Chen X (2022) Adenosine signaling: optimal target for gastric cancer immunotherapy. Front Immunol 13:1027838 Peerani E, Candido JB, Maniati E et al (2025) Matrix stiffness shapes transcriptional profiles and drug responses of pancreatic cancer cells. Acta Biomater Published online December 10 Huang J, Zhang L, Wan D et al (2021) Extracellular matrix and its therapeutic potential for cancer treatment. Signal Transduct Target Ther 6(1):153 Zheng Y, Zhou R, Cai J et al (2023) Matrix stiffness triggers lipid metabolic cross-talk between tumor and stromal cells to mediate bevacizumab resistance in colorectal cancer liver metastases. Cancer Res 83(21):3577–3592 Jiang C, Centonze A, Song Y et al (2024) Collagen signaling and matrix stiffness regulate multipotency in glandular epithelial stem cells in mice. Nat Commun 15(1):10482 Chen E, Zeng Z, Zhou W (2024) The key role of matrix stiffness in colorectal cancer immunotherapy: mechanisms and therapeutic strategies. Biochim Biophys Acta Rev Cancer 1879(6):189198 Zhang T, Li X, He Y et al (2022) Cancer-associated fibroblasts-derived HAPLN1 promotes tumour invasion through extracellular matrix remodeling in gastric cancer. Gastric Cancer 25(2):346–359 Lee D, Ham IH, Oh HJ et al (2024) Tubulointerstitial nephritis antigen-like 1 from cancer-associated fibroblasts contribute to the progression of diffuse-type gastric cancers through the interaction with integrin β1. J Transl Med 22(1):154 Yang Y, Sun H, Yu H et al (2025) Tumor-associated-fibrosis and active collagen-CD44 axis characterize a poor-prognosis subtype of gastric cancer and contribute to tumor immunosuppression. J Transl Med 23(1):123 Emon B, Song YJ, Joy MSH, Kovour MV, Prasanth KV, Saif MTA (2023) Mechanosensitive changes in the expression of genes in colorectal cancer-associated fibroblasts. Sci Data 10(1):350 Zheng S, Wang J, Ding N et al (2021) Prodrug polymeric micelles integrating cancer-associated fibroblasts deactivation and synergistic chemotherapy for gastric cancer. J Nanobiotechnol 19(1):381 Xu Q, Yang C, Ning J et al (2025) SMOC2high myofibroblastic cancer-associated fibroblast drives primary cilia-associated tumor microenvironment remodeling and poor prognosis in gastric cancer. Chin J Cancer Res 37(4):603–623 Lin Y, Sun X, Li C et al Three-year follow-up of neoadjuvant tislelizumab with chemotherapy in locally advanced gastric and gastroesophageal junction adenocarcinoma: revealing cancer-associated fibroblast heterogeneity corresponding to PD-1 blockade efficacy. Adv Sci (Weinh). Published online December 1, 2025. Blum A, Wang P, Zenklusen JC, Snapshot (2018) TCGA-analyzed tumors. Cell 173(2):530 Oh SC, Sohn BH, Cheong JH et al (2018) Clinical and genomic landscape of gastric cancer with a mesenchymal phenotype. Nat Commun 9(1):1777 Jeong HY, Ham IH, Lee SH et al (2021) Spatially distinct reprogramming of the tumor microenvironment based on tumor invasion in diffuse-type gastric cancers. Clin Cancer Res 27(23):6529–6542 Hossain MA, Asa TA, Islam MS, Rahman MZ, Moni MA (2025) Identification of key candidate genes for ovarian cancer using integrated statistical and machine learning approaches. Brief Bioinform 26(6):bbaf602 Pretemer Y, Kawai S, Nagata S et al (2021) Differentiation of hypertrophic chondrocytes from human iPSCs for the in vitro modeling of chondrodysplasias. Stem Cell Rep 16(3):610–625 Guo Z, Su W, Zhou R et al (2021) Exosomal MATN3 of urine-derived stem cells ameliorates intervertebral disc degeneration by antisenescence effects and promotes NPC proliferation and ECM synthesis by activating TGF-β. Oxid Med Cell Longev 2021:5542241 Long L, Zou G, Cheng Y, Li F, Wu H, Shen Y (2023) MATN3 delivered by exosome from synovial mesenchymal stem cells relieves knee osteoarthritis: evidence from in vitro and in vivo studies. J Orthop Translat 41:20–32 Chan S, Wang X, Wang Z et al (2023) CTSG suppresses colorectal cancer progression through negative regulation of Akt/mTOR/Bcl2 signaling pathway. Int J Biol Sci 19(7):2220–2233 Fidelle M, Rauber C, Alves Costa Silva C et al (2023) A microbiota-modulated checkpoint directs immunosuppressive intestinal T cells into cancers. Science 380(6649):eabo2296 Li Y, Wu S, Zhao Y et al (2024) Neutrophil extracellular traps induced by chemotherapy inhibit tumor growth in murine models of colorectal cancer. J Clin Invest 134(5):e175031 Dai Q, Yao X, Zhang Y et al (2024) CTSG is a prognostic marker involved in immune infiltration and inhibits tumor progression through the MAPK signaling pathway in non-small cell lung cancer. J Cancer Res Clin Oncol 151(1):21 Shen Y, Chen H, Gong X, Wang Z, Chen M, Chen D (2023) Identification of lysosome-related genes in connection with prognosis and immune cell infiltration for drug candidates in head and neck cancer. Open Life Sci 18(1):20220660 Huang GZ, Wu QQ, Zheng ZN et al (2021) Bioinformatics analyses indicate that cathepsin G (CTSG) is a potential immune-related biomarker in oral squamous cell carcinoma (OSCC). Onco Targets Ther 14:1275–1289 Chan S, Wang X, Wang Z et al (2023) CTSG suppresses colorectal cancer progression through negative regulation of Akt/mTOR/Bcl2 signaling pathway. Int J Biol Sci 19(7):2220–2233 Ozawa N, Yokobori T, Osone K et al (2024) MAdCAM-1 targeting strategy can prevent colitic cancer carcinogenesis and progression via suppression of immune cell infiltration and inflammatory signals. Int J Cancer 154(2):359–371 Zhang J, Liu F, Yang Y et al (2022) Integrated DNA and RNA sequencing reveals early drivers involved in metastasis of gastric cancer. Cell Death Dis 13(4):392 Croizer H, Mhaidly R, Kieffer Y et al (2024) Deciphering the spatial landscape and plasticity of immunosuppressive fibroblasts in breast cancer. Nat Commun 15(1):2806 Xu S, Deng KQ, Lu C et al (2024) Interleukin-6 classic and trans-signaling utilize glucose metabolism reprogramming to achieve anti- or pro-inflammatory effects. Metabolism 155:155832 Additional Declarations No competing interests reported. Cite Share Download PDF Status: Posted Version 1 posted You are reading this latest preprint version Research Square lets you share your work early, gain feedback from the community, and start making changes to your manuscript prior to peer review in a journal. As a division of Research Square Company, we’re committed to making research communication faster, fairer, and more useful. We do this by developing innovative software and high quality services for the global research community. Our growing team is made up of researchers and industry professionals working together to solve the most critical problems facing scientific publishing. Also discoverable on Platform About Our Team In Review Editorial Policies Advisory Board Help Center Resources Author Services Accessibility API Access RSS feed Manage Cookie Preferences © Research Square 2026 | ISSN 2693-5015 (online) Privacy Policy Terms of Service Do Not Sell My Personal Information {"props":{"pageProps":{"initialData":{"identity":"rs-8616596","acceptedTermsAndConditions":true,"allowDirectSubmit":true,"archivedVersions":[],"articleType":"Research Article","associatedPublications":[],"authors":[{"id":588588956,"identity":"ce992a34-0f96-4c6d-b882-cd26bcf7ea38","order_by":0,"name":"Wei Zhang","email":"","orcid":"","institution":"Henan Provincial People's Hospital","correspondingAuthor":false,"prefix":"","firstName":"Wei","middleName":"","lastName":"Zhang","suffix":""},{"id":588588958,"identity":"1c8be367-88a7-4ec7-8566-5c53c0c10865","order_by":1,"name":"Wenzheng Li","email":"","orcid":"","institution":"People’s Hospital of Henan University, Henan University","correspondingAuthor":false,"prefix":"","firstName":"Wenzheng","middleName":"","lastName":"Li","suffix":""},{"id":588588962,"identity":"2960d01c-91cb-4b00-bf93-4c2a5dbeef42","order_by":2,"name":"Zhengxuan Zhang","email":"","orcid":"","institution":"Henan Provincial People's Hospital","correspondingAuthor":false,"prefix":"","firstName":"Zhengxuan","middleName":"","lastName":"Zhang","suffix":""},{"id":588588968,"identity":"01680d6d-7875-4d47-ae49-2b1af61bdf81","order_by":3,"name":"Peichun Sun","email":"","orcid":"","institution":"Henan Provincial People's Hospital","correspondingAuthor":false,"prefix":"","firstName":"Peichun","middleName":"","lastName":"Sun","suffix":""},{"id":588588969,"identity":"4b3f8394-018c-4a89-a3c4-303ba44b3817","order_by":4,"name":"Gang Wu","email":"data:image/png;base64,iVBORw0KGgoAAAANSUhEUgAAAZAAAAAyAQMAAABI0h/eAAAABlBMVEX///8AAABVwtN+AAAACXBIWXMAAA7EAAAOxAGVKw4bAAAA3UlEQVRIiWNgGAWjYBACNvbG5gcfKtjk5OUfHzjw4QcRWvh4Dh8znHGGz9iwIS3x4MweIrTISaQlSHO2ySUyHMgxPszBRozDGHIMjBnbzBIYG858OMzAwyDPL3aAkJYzBo8LzqXlsTP2bjhcYMFgOHN2AgEtjD0GxjPKjhUzNvNuODyDhyHB4DYhLcw8BtI8bP8TG47xPDjMw0aMFja2BGmeNrbEhjM8DERq4WEGBTKbseEMNgNgIEsQ9ov8/IfQqJRgfvzhww8beX5pAlrQgQRpykfBKBgFo2AUYAcA8DdGXqL0a5QAAAAASUVORK5CYII=","orcid":"","institution":"Henan Provincial People's Hospital","correspondingAuthor":true,"prefix":"","firstName":"Gang","middleName":"","lastName":"Wu","suffix":""}],"badges":[],"createdAt":"2026-01-16 08:39:07","currentVersionCode":1,"declarations":"","doi":"10.21203/rs.3.rs-8616596/v1","doiUrl":"https://doi.org/10.21203/rs.3.rs-8616596/v1","draftVersion":[],"editorialEvents":[],"editorialNote":"","failedWorkflow":false,"files":[{"id":102598017,"identity":"06a6df82-c8b6-4567-abab-4bda44a6fa34","added_by":"auto","created_at":"2026-02-13 12:27:01","extension":"png","order_by":1,"title":"Figure 1","display":"","copyAsset":false,"role":"figure","size":3386684,"visible":true,"origin":"","legend":"\u003cp\u003e\u003cstrong\u003eIdentification of matrix stiffness-associated genes and establishment of the MSRS prognostic model.\u003c/strong\u003e\u003c/p\u003e\n\u003cp\u003e(A) Venn diagram illustrating the integration of four distinct matrix stiffness-related gene sets. (B) Volcano plot depicting differentially expressed genes (DEGs) between gastric cancer and adjacent normal tissues in the TCGA-STAD cohort.（C）Venn diagram demonstrating the intersection of matrix stiffness-associated genes with DEGs, resulting in 237 candidate genes. (D) Bar plot displaying the top 40 differentially expressed matrix stiffness-related genes. (E) LASSO regression analysis with 10-fold cross-validation to determine optimal parameter selection. (F) LASSO coefficient plot of the 25 selected MSRS genes. (G) KEGG pathway enrichment analysis of the 237 candidate genes. (H) Gene Ontology (GO) enrichment analysis of the candidate genes. (I) Gene Set Enrichment Analysis (GSEA) revealing significant enrichment of the focal adhesion pathway.\u003c/p\u003e","description":"","filename":"Figure1.png","url":"https://assets-eu.researchsquare.com/files/rs-8616596/v1/1de28ba1b28ce93d898a8997.png"},{"id":102597863,"identity":"5df41bd4-fda0-4e43-bda8-4cee90ca4b76","added_by":"auto","created_at":"2026-02-13 12:26:47","extension":"png","order_by":2,"title":"Figure 2","display":"","copyAsset":false,"role":"figure","size":2769641,"visible":true,"origin":"","legend":"\u003cp\u003e\u003cstrong\u003eIdentification and experimental validation of independent prognostic genes.\u003c/strong\u003e\u003c/p\u003e\n\u003cp\u003e(A) Multivariate Cox regression analysis of the 25 MSRS genes revealed three independent prognostic factors: MATN3, CTSG, and MADCAM1 (Forest plot). (B–D) Violin plots depicting differential expression of MATN3, CTSG, and MADCAM1 between tumor and adjacent normal tissues in the TCGA-STAD cohort (***\u003cem\u003eP \u003c/em\u003e\u0026lt; 0.001, ** \u003cem\u003eP \u003c/em\u003e\u0026lt; 0.01). (E–G) Western blot analysis confirmed protein expression levels of MATN3, CTSG, and MADCAM1 in normal gastric epithelial cells (GES-1) and gastric cancer cell lines (AGS, HGC-27, and MKN-45), with GAPDH serving as the loading control.\u003c/p\u003e","description":"","filename":"Figure2.png","url":"https://assets-eu.researchsquare.com/files/rs-8616596/v1/9453c963ea93588688043bce.png"},{"id":102597958,"identity":"0d648784-5934-4fe5-afa3-9cfbcb19bf88","added_by":"auto","created_at":"2026-02-13 12:26:57","extension":"png","order_by":3,"title":"Figure 3","display":"","copyAsset":false,"role":"figure","size":2785239,"visible":true,"origin":"","legend":"\u003cp\u003e\u003cstrong\u003ePrognostic evaluation of the MSRS model in training and validation cohorts.\u003c/strong\u003e\u003c/p\u003e\n\u003cp\u003e(A) Kaplan-Meier analysis of recurrence-free survival in the training cohort (GSE26253, n=432). Statistical significance was determined by log-rank test (\u003cem\u003eP\u003c/em\u003e \u0026lt; 0.0001). (B) Kaplan-Meier analysis of overall survival in the validation cohort (TCGA-STAD, n=386), with log-rank test demonstrating significance (\u003cem\u003eP\u003c/em\u003e = 0.0016). (C) Time-dependent receiver operating characteristic (ROC) curves evaluating 3-, 5-, and 7-year predictive performance in the training cohort. (D) Time-dependent ROC curves assessing 3-, 5-, and 7-year predictive accuracy in the validation cohort. (E) Nomogram incorporating MSRS risk score and clinical stage for survival probability prediction. (F) Calibration analysis evaluating the concordance between predicted and observed survival outcomes.\u003c/p\u003e","description":"","filename":"Figure3.png","url":"https://assets-eu.researchsquare.com/files/rs-8616596/v1/a965b78afc7d04ac423e9c60.png"},{"id":102598024,"identity":"a46e623f-d855-4e7c-8315-d3c15545c3d4","added_by":"auto","created_at":"2026-02-13 12:27:06","extension":"png","order_by":4,"title":"Figure 4","display":"","copyAsset":false,"role":"figure","size":213657,"visible":true,"origin":"","legend":"\u003cp\u003e\u003cstrong\u003eAssociation of MSRS with Immune Cell Infiltration and Drug Target Gene Expression.\u003c/strong\u003e\u003c/p\u003e\n\u003cp\u003e(A) Box plots showing differences in immune cell infiltration between high‑ and low‑risk groups. (B) Bar charts summarizing Spearman correlation coefficients between MSRS risk scores and immune cell infiltration levels. (C) Scatter plots illustrating the relationships between MSRS risk scores and specific immune cell subsets. (D) Box plots displaying the differential expression of drug‑target genes in high‑ and low‑risk groups.Significance levels: * \u003cem\u003eP\u003c/em\u003e \u0026lt; 0.05; **\u003cem\u003e P\u003c/em\u003e \u0026lt; 0.01; *** \u003cem\u003e P\u003c/em\u003e \u0026lt; 0.001; **** \u003cem\u003eP\u003c/em\u003e \u0026lt; 0.0001.\u003c/p\u003e","description":"","filename":"Figure4.png","url":"https://assets-eu.researchsquare.com/files/rs-8616596/v1/edf1fbd04c2ba538ac42f452.png"},{"id":102597926,"identity":"4280175a-3691-409c-ae3c-e2ba62f6a5f1","added_by":"auto","created_at":"2026-02-13 12:26:55","extension":"png","order_by":5,"title":"Figure 5","display":"","copyAsset":false,"role":"figure","size":4883145,"visible":true,"origin":"","legend":"\u003cp\u003e\u003cstrong\u003eSingle-cell RNA sequencing analysis of MSRS gene cellular origins and cancer-associated fibroblast (CAF) subtype heterogeneity.\u003c/strong\u003e\u003c/p\u003e\n\u003cp\u003e(A) t-distributed stochastic neighbor embedding (t-SNE) visualization of cellular clusters derived from gastric cancer single-cell RNA sequencing data. (B) t-SNE projection with cells color-coded according to annotated cell types. (C) Proportional representation of distinct cell populations across normal, superficial tumor, and deep tumor tissue regions. (D) Comparative assessment of matrix stiffness scores among different cell types, with stromal cells demonstrating the highest values (Kruskal-Wallis test, \u003cem\u003eP\u003c/em\u003e \u0026lt; 2.2 × 10⁻¹⁶). (E) t-SNE plot illustrating spatial distribution of stiffness scores. (F) Differential stiffness scores between normal and tumor tissues. (G) Variation in stiffness scores across tissue depths (Kruskal-Wallis test, \u003cem\u003eP\u003c/em\u003e \u0026lt; 2.2 × 10⁻¹⁶). (H) Dot plot representing the expression patterns of MSRS genes across diverse cell populations. (I) t-SNE visualization delineating CAF subtypes: antigen-presenting CAFs (apCAFs), inflammatory CAFs (iCAFs), and myofibroblastic CAFs (myCAFs). (J) Comparative stiffness scores among CAF subtypes, revealing significantly elevated scores in iCAFs (Kruskal-Wallis test, \u003cem\u003eP\u003c/em\u003e = 2.6 × 10⁻¹⁵). (K) Dot plot depicting marker gene expression profiles of CAF subtypes. (L) Relative abundance of CAF subtypes across different tissue regions.\u003c/p\u003e","description":"","filename":"Figure5.png","url":"https://assets-eu.researchsquare.com/files/rs-8616596/v1/78e10d8bd7b8ffce9feea5c1.png"},{"id":102651092,"identity":"cf044f79-66a3-4b65-a080-ff2526053073","added_by":"auto","created_at":"2026-02-14 06:40:48","extension":"pdf","order_by":0,"title":"","display":"","copyAsset":false,"role":"manuscript-pdf","size":14403361,"visible":true,"origin":"","legend":"","description":"","filename":"manuscript.pdf","url":"https://assets-eu.researchsquare.com/files/rs-8616596/v1/689b508d-9001-4236-b28b-7db3a42590f7.pdf"}],"financialInterests":"No competing interests reported.","formattedTitle":"Construction and verification of matrix stiffness-related prognostic model of gastric cancer based on single-cell analysis and in vitro experiments ","fulltext":[{"header":"Introduction","content":"\u003cp\u003eGastric cancer (GC) is still the fifth most common malignant tumour and the fourth leading cause of cancer-related death in the world, with more than 1\u0026nbsp;million new cases every year, resulting in about 650,000 deaths\u003csup\u003e[\u003cspan citationid=\"CR1\" class=\"CitationRef\"\u003e1\u003c/span\u003e]\u003c/sup\u003e. Due to the hidden onset and the lack of specific early symptoms, most patients are diagnosed in the late stage, resulting in an extremely unoptimistic 5-year survival rate\u003csup\u003e[\u003cspan citationid=\"CR2\" class=\"CitationRef\"\u003e2\u003c/span\u003e]\u003c/sup\u003e. Therefore, clarifying the molecular mechanism of gastric cancer and determining reliable prognostic biomarkers is crucial to improving the clinical treatment effect.\u003c/p\u003e \u003cp\u003eTumour microenvironment (TME) plays a key role in the process of tumour development. The latest research shows that in addition to traditional biochemical signal conduction, the physical properties of TME - especially matrix Stiffness - have a crucial regulatory effect on the behaviour of tumour cells\u003csup\u003e[\u003cspan citationid=\"CR3\" class=\"CitationRef\"\u003e3\u003c/span\u003e]\u003c/sup\u003e. Tissue Stiffness is mainly determined by the composition and crosslinking density of the extracellular matrix (ECM). In solid tumours, abnormal ECM deposition and collagen fibre cross-linking lead to a significant increase in Stiffness \u003csup\u003e[\u003cspan citationid=\"CR4\" class=\"CitationRef\"\u003e4\u003c/span\u003e]\u003c/sup\u003e. This mechanical sclerosis drives tumour progression through a variety of mechanisms: the cell's perception of the Stiffness of the matrix activates the integrin receptor, thus triggering the downstream sticky spot kinase (FAK) cascascading reaction\u003csup\u003e[\u003cspan citationid=\"CR5\" class=\"CitationRef\"\u003e5\u003c/span\u003e]\u003c/sup\u003e. At the same time, the Yes-related protein (YAP), which is a key mechanical conduction factor, will be translocated to the nucleus in a hard environment and regulate the transcription of proliferation and migration-related genes \u003csup\u003e[\u003cspan citationid=\"CR6\" class=\"CitationRef\"\u003e6\u003c/span\u003e]\u003c/sup\u003e. In addition, matrix Stiffness assists tumour immune escape by hindering T cell infiltration and promoting the formation of immunosuppressive microenvironment \u003csup\u003e[\u003cspan citationid=\"CR7\" class=\"CitationRef\"\u003e7\u003c/span\u003e]\u003c/sup\u003e.\u003c/p\u003e \u003cp\u003eIn gastric cancer, extracellular matrix remodelling is essentially related to the progression of the disease \u003csup\u003e[\u003cspan citationid=\"CR8\" class=\"CitationRef\"\u003e8\u003c/span\u003e][\u003cspan citationid=\"CR9\" class=\"CitationRef\"\u003e9\u003c/span\u003e]\u003c/sup\u003e. Excessive collagen deposition is significantly related to the depth of tumour infiltration, lymph node metastasis and poor prognosis\u003csup\u003e[\u003cspan citationid=\"CR10\" class=\"CitationRef\"\u003e10\u003c/span\u003e]\u003c/sup\u003e. Cancer-related fibroblasts (CAFs) are recognised as the main effect cells for ECM remodelling and hardening \u003csup\u003e[\u003cspan citationid=\"CR11\" class=\"CitationRef\"\u003e11\u003c/span\u003e][\u003cspan citationid=\"CR12\" class=\"CitationRef\"\u003e12\u003c/span\u003e]\u003c/sup\u003e. Although single-cell sequencing technology has revealed the heterogeneity of CAF - subgroups such as myofibroblast type (myCAF), inflammatory type (iCAF) and antigen presentation type (apCAF) \u003csup\u003e[\u003cspan citationid=\"CR13\" class=\"CitationRef\"\u003e13\u003c/span\u003e][\u003cspan citationid=\"CR14\" class=\"CitationRef\"\u003e14\u003c/span\u003e]\u003c/sup\u003e, the specific association mechanism between these subtypes and matrix sclerosis has yet to be clarified. Existing research mostly focusses on isolated signalling pathways and lacks systematic integration of Stiffness-related gene sets.\u003c/p\u003e \u003cp\u003eIn order to fill this gap, this study integrates four key gene sets: integrin pathways, YAP characteristics, ECM tissue and collagen formation. The matrix Stiffness-related risk score (MSRS) was constructed by LASSO Cox regression. It is worth noting that we use non-recurrent survival (RFS) as the training endpoint and total survival (OS) for verification to strictly evaluate the universality of the model in different clinical outcomes. Independent prognostic genes are identified by multivariable Cox regression, and multi-dimensional methods such as bioinformatics, protein imprinting and single-cell RNA sequencing are used for verification. This study aims to provide new insights into the mechanical biology of gastric cancer and establish potential targets for precise treatment.\u003c/p\u003e"},{"header":"Materials and Methods","content":"\u003cdiv id=\"Sec3\" class=\"Section2\"\u003e \u003ch2\u003eDesign and Overall Strategy\u003c/h2\u003e \u003cp\u003eTo systematically evaluate the clinical significance and cellular origins of matrix stiffness-associated genes in gastric cancer, we implemented a comprehensive multi-stage analytical framework. The study proceeded through seven key phases: (1) Candidate Identification: Curating potential stiffness-related targets by intersecting multiple established gene sets; (2) Transcriptomic Profiling: Conducting differential expression and functional enrichment analyses within the TCGA cohort; (3) Model Construction: Developing a prognostic risk signature utilizing LASSO regression algorithms; (4) Independence Testing: Confirming the signature\u0026rsquo;s independent predictive value via multivariate Cox regression; (5) Dual Validation: Verifying model robustness in an external cohort and corroborating core gene expression through in vitro experiments; (6) Clinical Correlation: Assessing associations between risk scores, the immune microenvironment, and chemotherapeutic sensitivity; and (7) Cellular Mapping: Tracing the specific cellular provenance of the signature genes using scRNA-seq data.\u003c/p\u003e \u003c/div\u003e\n\u003ch3\u003eMatrix Stiffness-Related Gene Sets\u003c/h3\u003e\n\u003cp\u003eTo comprehensively capture the genetic landscape of matrix stiffness, we curated four distinct gene sets from the Molecular Signatures Database (MSigDB; \u003cspan class=\"ExternalRef\"\u003e\u003cspan class=\"RefSource\"\u003ehttps://www.gsea-msigdb.org/\u003c/span\u003e\u003cspan address=\"https://www.gsea-msigdb.org/\" targettype=\"URL\" class=\"RefTarget\"\u003e\u003c/span\u003e\u003c/span\u003e) and published literature. These included the Integrin Pathway, the YAP Signature, ECM Organization, and Collagen Formation gene sets. Following the merging of these datasets and the removal of redundant entries, a total of 398 unique matrix stiffness-associated genes were identified for downstream analysis.\u003c/p\u003e\n\u003ch3\u003eTCGA-STAD Dataset\u003c/h3\u003e\n\u003cp\u003eTranscriptomic profiles (HTSeq-FPKM format) and associated clinical parameters\u0026mdash;including age, gender, tumor stage, and survival outcomes\u0026mdash;were retrieved from The Cancer Genome Atlas (TCGA) database for the Stomach Adenocarcinoma (STAD) cohort\u003csup\u003e[\u003cspan citationid=\"CR15\" class=\"CitationRef\"\u003e15\u003c/span\u003e]\u003c/sup\u003e. Following rigorous data curation and quality control procedures, a final cohort of 386 patients with complete survival records was retained to facilitate differential expression analysis and model validation.\u003c/p\u003e\n\u003ch3\u003eGEO Dataset\u003c/h3\u003e\n\u003cp\u003eThe training cohort, identified as GSE26253, was retrieved from the NCBI Gene Expression Omnibus (GEO) repository. This dataset contains transcriptomic data derived from the Affymetrix Human Genome U133 Plus 2.0 array, alongside clinical annotations for 432 gastric cancer patients, specifically including recurrence-free survival (RFS) outcomes.\u003csup\u003e[\u003cspan citationid=\"CR16\" class=\"CitationRef\"\u003e16\u003c/span\u003e]\u003c/sup\u003e\u003c/p\u003e\n\u003ch3\u003eSingle-Cell RNA Sequencing Data\u003c/h3\u003e\n\u003cp\u003eTo determine the cellular origins of the signature genes and characterize CAF heterogeneity, we acquired the single-cell transcriptome dataset GSE167297. This cohort encompasses paired samples of normal tissue and tumor specimens (stratified into shallow and deep regions), providing a high-resolution reference for validating the model\u0026rsquo;s cellular sources.\u003csup\u003e[\u003cspan citationid=\"CR17\" class=\"CitationRef\"\u003e17\u003c/span\u003e]\u003c/sup\u003e\u003c/p\u003e \u003cdiv id=\"Sec8\" class=\"Section2\"\u003e \u003ch2\u003eTIMER3.0 Database\u003c/h2\u003e \u003cp\u003eAssessment of tumor immune infiltration was conducted via the TIMER3.0 web server(\u003cspan class=\"ExternalRef\"\u003e\u003cspan class=\"RefSource\"\u003ehttps://compbio.cn/timer3/\u003c/span\u003e\u003cspan address=\"https://compbio.cn/timer3/\" targettype=\"URL\" class=\"RefTarget\"\u003e\u003c/span\u003e\u003c/span\u003e). Specifically, the \u0026ldquo;Gene-DE\u0026rdquo; module was applied to examine the differential expression patterns of MATN3, CTSG, and MADCAM1 by comparing gastric cancer tissues against normal controls.\u003c/p\u003e \u003c/div\u003e\n\u003ch3\u003eDifferential Expression Analysis\u003c/h3\u003e\n\u003cp\u003eUse the limma package to analyse the gene expression differences of the TCGA-STAD cohort in the R studio (v4.4.1). In order to achieve the comparability between samples, the original FPKM data is first converted into TPM (Transcripts Per Million) format, and log2 (TPM\u0026thinsp;+\u0026thinsp;1) standardised processing is carried out. Subsequently, the empirical Bayes statistical framework was used to evaluate the expression level of gene differences\u003csup\u003e[\u003cspan citationid=\"CR18\" class=\"CitationRef\"\u003e18\u003c/span\u003e]\u003c/sup\u003e. The conditions for screening significantly differential expression genes (DEGs) are: |log2(FC)| \u0026gt; 1, and the P value (P.adj) after multiple tests and correction by the Benjaminini\u0026ndash;Hochberg method is \u0026lt;\u0026thinsp;0.05.\u003c/p\u003e\n\u003ch3\u003eFunctional Enrichment Analysis\u003c/h3\u003e\n\u003cp\u003eTo decipher the underlying biological roles and signaling networks of the identified genes, we leveraged the \u0026lsquo;clusterProfiler\u0026rsquo; package in R. This analysis primarily focused on Gene Ontology (GO) annotation, dissecting terms across three dimensions: biological processes (BP), cellular components (CC), and molecular functions (MF). Additionally, pathway enrichment was explored referencing the Kyoto Encyclopedia of Genes and Genomes (KEGG) database. For both analyses, a nominal P-value of less than 0.05 was established as the criterion for statistical significance.\u003c/p\u003e \u003cdiv id=\"Sec11\" class=\"Section2\"\u003e \u003ch2\u003eLASSO Regression and Prognostic Model Construction\u003c/h2\u003e \u003cp\u003eUse glmnet packages to implement LASSO regression in the R studio to mine key genes significantly related to prognosis. This method realises variable screening by introducing L1 regularisation terms, which helps to reduce the risk of overfitting in high-dimensional data. Take the GSE26253 cohort as the training data set, and set RFS as the survival outcome indicator. Use ten-fold cross-verification to determine the best penalty coefficient λ (lambda.min), select genes with non-zero regression coefficients according to this parameter, and construct an MSRS prognostic risk model.\u003c/p\u003e \u003cp\u003eThe MSRS risk score was calculated as:\u003cdiv id=\"Equa\" class=\"Equation\"\u003e\u003cdiv format=\"TEX\" class=\"mathdisplay\" id=\"FileID_Equa\" name=\"EquationSource\"\u003e\n$$\\:\\varvec{R}\\varvec{i}\\varvec{s}\\varvec{k}\\varvec{S}\\varvec{c}\\varvec{o}\\varvec{r}\\varvec{e}=\\sum\\:_{\\varvec{i}=1}^{\\varvec{n}}(\\varvec{C}\\varvec{o}\\varvec{e}{\\varvec{f}}_{\\varvec{i}}\\times\\:\\varvec{E}\\varvec{x}{\\varvec{p}}_{\\varvec{i}})$$\u003c/div\u003e\u003c/div\u003e\u003c/p\u003e \u003cp\u003eAmong them, \u003cspan class=\"InlineEquation\"\u003e\u003cspan class=\"mathinline\"\u003e\\(\\:Coe{f}_{i}\\:\\)\u003c/span\u003e\u003c/span\u003erepresents the LASSO regression coefficient of the ith gene, \u003cspan class=\"InlineEquation\"\u003e\u003cspan class=\"mathinline\"\u003e\\(\\:Ex{p}_{i}\\:\\)\u003c/span\u003e\u003c/span\u003e is the expression level of the ith gene, and n is the number of core genes.\u003c/p\u003e \u003c/div\u003e \u003cdiv id=\"Sec12\" class=\"Section2\"\u003e \u003ch2\u003eCox Regression Analysis\u003c/h2\u003e \u003cp\u003eUse the R language survival package for Cox proportional risk regression analysis. First, single-factor Cox regression was carried out on 25 candidate core genes, and prognostic-related variables were screened (\u003cem\u003eP\u003c/em\u003e\u0026thinsp;\u0026lt;\u0026thinsp;0.05). Genetics of statistical significance are incorporated into the multi-factor Cox model, and independent prognostic factors are determined by stepwise regression. The results are expressed in terms of risk ratio (HR) and 95% confidence interval (CI), and the difference of \u003cem\u003eP\u003c/em\u003e\u0026thinsp;\u0026lt;\u0026thinsp;0.05 is statistically significant.\u003c/p\u003e \u003c/div\u003e \u003cdiv id=\"Sec13\" class=\"Section2\"\u003e \u003ch2\u003eSurvival Analysis and Model Performance Evaluation\u003c/h2\u003e \u003cp\u003eThe Kaplan-Meier method (implemented via \u0026lsquo;survival\u0026rsquo; and \u0026lsquo;survminer\u0026rsquo; packages) was utilized for survival analysis, with patients stratified into high- and low-risk groups according to the median MSRS score. The statistical significance of survival discrepancies between groups was evaluated by the log-rank test. The prognostic performance of the model was assessed through time-dependent Receiver Operating Characteristic (ROC) analysis (\u0026lsquo;timeROC\u0026rsquo; package) at 3-, 5-, and 7-year follow-up intervals; an Area Under the Curve (AUC) value closer to 1 indicates superior predictive capability. Additionally, the \u0026lsquo;rms\u0026rsquo; package was employed to construct a prognostic nomogram that integrates MSRS scores with clinical stages. The calibration degree of the model was evaluated via calibration diagrams, where optimal prediction accuracy is indicated when the predicted probability closely aligns with observed outcomes along the 45-degree reference line.\u003c/p\u003e \u003c/div\u003e \u003cdiv id=\"Sec14\" class=\"Section2\"\u003e \u003ch2\u003eImmune Cell Infiltration Analysis\u003c/h2\u003e \u003cp\u003eTo estimate the infiltration density of various immune cell populations within the tumor microenvironment, we utilized the \u0026lsquo;GSVA\u0026rsquo; R package to execute the single-sample gene set enrichment analysis (ssGSEA) algorithm. The specific marker signatures for these immune subsets were curated from established literature. Comparison of infiltration profiles across the high- and low-risk cohorts was achieved via the Wilcoxon rank-sum test. Furthermore, the potential linkage between the MSRS and immune cell abundance was explored using Spearman\u0026rsquo;s rank correlation. Statistical significance was defined by a \u003cem\u003eP\u003c/em\u003e -value\u0026thinsp;\u0026lt;\u0026thinsp;0.05, with a coefficient threshold of |\u003cem\u003er\u003c/em\u003e| \u0026gt; 0.1 deemed meaningful; positive (\u003cem\u003er\u003c/em\u003e\u0026thinsp;\u0026gt;\u0026thinsp;0) and negative (\u003cem\u003er\u003c/em\u003e\u0026thinsp;\u0026lt;\u0026thinsp;0) values indicated proportional and inverse relationships, respectively.\u003c/p\u003e \u003c/div\u003e \u003cdiv id=\"Sec15\" class=\"Section2\"\u003e \u003ch2\u003eDrug Sensitivity Analysis\u003c/h2\u003e \u003cp\u003eTarget genes of commonly used targeted therapies for gastric cancer were compiled from published literature, including key targets such as Mesenchymal-Epithelial Transition factor (MET), Thymidylate Synthase (TYMS), Topoisomerase IIα (TOP2A), and erb-b2 Receptor Tyrosine Kinase 2 (ERBB2). Differential expression of these target genes between high- and low-risk groups was analyzed to infer potential drug sensitivity profiles in each risk group.\u003c/p\u003e \u003c/div\u003e \u003cdiv id=\"Sec16\" class=\"Section2\"\u003e \u003ch2\u003eSingle-Cell RNA Sequencing Data Analysis\u003c/h2\u003e \u003cp\u003eThe Seurat package was utilized to facilitate single-cell data processing. Following rigorous quality control filtration, a final dataset comprising 23,069 high-quality cells was selected for subsequent analysis. The data underwent normalization and unsupervised clustering to define distinct cellular subpopulations, which were projected visually using t-distributed Stochastic Neighbor Embedding (t-SNE). Cluster annotation was performed by referencing established marker genes, resulting in the characterization of seven primary cell lineages: T cells (CD3D, CD3E), B cells (CD79A, MS4A1), myeloid cells (CD68, CD14), epithelial cells (EPCAM, KRT18), endothelial cells (PECAM1, VWF), stromal cells (DCN, COL1A1), and mast cells (TPSAB1, CPA3).Compositional changes in cell types across different tissue regions (normal, superficial tumor, and deep tumor) were examined. The AddModuleScore function was used to calculate a stromal stiffness score for each cell, with the 25 core MSRS genes serving as the scoring gene set. Differences in stiffness scores among cell types and between normal and tumor tissues were compared using the Kruskal-Wallis test. Subclustering of stromal cells was performed to classify cancer-associated fibroblasts (CAFs) into three subtypes based on published marker genes: myCAF (ACTA2, TAGLN, MYL9), iCAF (IL6, CXCL12, PDGFRA), and apCAF (CD74, HLA-DRA, HLA-DRB1). Stiffness scores were compared across CAF subtypes, and proportional changes in CAF subtypes across tissue regions were analyzed.\u003c/p\u003e \u003c/div\u003e \u003cdiv id=\"Sec17\" class=\"Section2\"\u003e \u003ch2\u003eCell Culture\u003c/h2\u003e \u003cp\u003eWe obtained the immortalized human gastric epithelial cell line (GES-1) alongside three gastric adenocarcinoma lines (AGS, HGC-27, and MKN-45) from Wuhan Saiweier Biotechnology Co., Ltd. Cell propagation was conducted using Gibco\u0026rsquo;s RPMI-1640 basal medium, which was enriched with 10% fetal bovine serum (FBS) and a 1% penicillin-streptomycin mixture. Incubation took place in a humidified chamber regulated at 37\u0026deg;C with a 5% CO2 atmosphere. Routine subculturing was performed at intervals of 2\u0026ndash;3 days. For all downstream assays, cells were harvested exclusively during their exponential growth phase.\u003c/p\u003e \u003c/div\u003e \u003cdiv id=\"Sec18\" class=\"Section2\"\u003e \u003ch2\u003eWestern Blot Analysis\u003c/h2\u003e \u003cp\u003eCells in the logarithmic growth were lysed on the ice with RIPA lysate (Beyotime) containing protease inhibitors and phosphatase inhibitors for 30 minutes. Centrifuge at 12,000 rpm for 25 minutes, take the supernatant, and use the BCA protein quantitative kit (Thermo Fisher) to determine the protein concentration. According to the quantitative results, take an equal sample for SDS-PAGE separation, and transfer the protein electricity to the PVDF membrane. After turning the membrane, use the Sanying Rapid WB antibody dilution solution (Proteintech PR20039) for dilution operation. After 5 minutes of rinsing the membrane with TBST, place it in the diluent at room temperature or 37\u0026deg;C and gently shake it for about 5 minutes. Then dilute anti-MATN3 1:1000, anti-CTSG 1:1000, anti-MADCAM1 1:1000, anti-GAPDH 1:5000) with this dilution and incubate at room temperature for 25 minutes. After incubation, wash with TBST 5 times for 1 minute each time. After that, use the same dilution solution to dilute HRP-labelled dianti, incubate at room temperature for 15 minutes, and wash again 3 times for 1 minute each time. After washing, add ECL development reagent for chemiluminescence detection, and use the gel imaging system to collect the results. GAPDH is used as an internal reference control for protein quantitative standardisation.\u003c/p\u003e \u003c/div\u003e \u003cdiv id=\"Sec19\" class=\"Section2\"\u003e \u003ch2\u003eStatistical Analysis\u003c/h2\u003e \u003cp\u003eAll statistical computations and data management were executed within the R statistical framework (version 4.4.1). To differentiate continuous variables between two distinct groups, the Wilcoxon rank-sum test was employed; conversely, the Kruskal-Wallis test was selected for comparisons spanning three or more cohorts. The correlation intensity between variables was assessed using Spearman\u0026rsquo;s coefficient. We adopted a two-sided testing approach for all analyses, establishing statistical significance at \u003cem\u003eP\u003c/em\u003e\u0026thinsp;\u0026lt;\u0026thinsp;0.05. Visual representations denote significance levels as follows: \u0026lsquo;ns\u0026rsquo; for \u003cem\u003eP\u003c/em\u003e\u0026thinsp;\u0026ge;\u0026thinsp;0.05, with increasing significance marked by * (\u003cem\u003eP\u003c/em\u003e\u0026thinsp;\u0026lt;\u0026thinsp;0.05), ** (\u003cem\u003eP\u003c/em\u003e\u0026thinsp;\u0026lt;\u0026thinsp;0.01), *** (\u003cem\u003eP\u003c/em\u003e\u0026thinsp;\u0026lt;\u0026thinsp;0.001), and **** (\u003cem\u003eP\u003c/em\u003e\u0026thinsp;\u0026lt;\u0026thinsp;0.0001).\u003c/p\u003e \u003c/div\u003e"},{"header":"Results","content":"\u003cdiv id=\"Sec21\" class=\"Section2\"\u003e \u003ch2\u003eIdentification of Matrix Stiffness-Related Genes and Construction of the MSRS Model\u003c/h2\u003e \u003cp\u003eIn order to identify matrix Matrix Stiffness-related genes in gastric cancer, this study integrated four signalling pathway gene sets of YAP, integrin, extracellular matrix tissue and collagen formation, and obtained a total of 398 matrix Matrix Stiffness-related genes (Fig.\u0026nbsp;\u003cspan refid=\"Fig1\" class=\"InternalRef\"\u003e1\u003c/span\u003eA). The differential expression analysis of the TCGA-STAD data set identified 8,248 differential expression genes (DEGs), including 4,147 up-regulation genes and 4,101 down-regulation genes (Fig.\u0026nbsp;\u003cspan refid=\"Fig1\" class=\"InternalRef\"\u003e1\u003c/span\u003eB). 237 candidate genes were obtained by intersecting DEGs with matrix Matrix Stiffness-related genes (Fig.\u0026nbsp;\u003cspan refid=\"Fig1\" class=\"InternalRef\"\u003e1\u003c/span\u003eC). Figure\u0026nbsp;\u003cspan refid=\"Fig1\" class=\"InternalRef\"\u003e1\u003c/span\u003eD shows the top 40 candidate genes sorted by difference multiples. Functional enrichment analysis shows that 237 candidate genes are significantly enriched in ECM-receptor interaction, adhesion spots and PI3K-Akt signalling pathways (KEGG analysis, Fig.\u0026nbsp;\u003cspan refid=\"Fig1\" class=\"InternalRef\"\u003e1\u003c/span\u003eG). GO analysis shows that these genes are mainly involved in biological processes such as extracellular matrix tissue, collagen fibre assembly and cell-matric adhesion (Fig.\u0026nbsp;\u003cspan refid=\"Fig1\" class=\"InternalRef\"\u003e1\u003c/span\u003eH). GSEA analysis further confirms the significant enrichment of adhesive spot pathways (NES\u0026thinsp;=\u0026thinsp;1.93, \u003cem\u003eP\u003c/em\u003e\u0026thinsp;\u0026lt;\u0026thinsp;0.05).Through LASSO regression analysis, 15 key genes were screened from 237 genes, of which 8 were risk genes (LASSO coefficient\u0026thinsp;\u0026gt;\u0026thinsp;0) and 7 were protective genes (LASSO coefficient\u0026thinsp;\u0026lt;\u0026thinsp;0).\u003c/p\u003e \u003cp\u003e \u003c/p\u003e \u003c/div\u003e \u003cdiv id=\"Sec22\" class=\"Section2\"\u003e \u003ch2\u003ePrognostic Significance and Experimental Validation of MSRS Core Genes\u003c/h2\u003e \u003cp\u003eTo isolate genes with independent prognostic power, we subjected the 25 candidate genes to multivariate Cox regression analysis (Fig.\u0026nbsp;\u003cspan refid=\"Fig2\" class=\"InternalRef\"\u003e2\u003c/span\u003eA). This screening identified three pivotal markers: MATN3 was characterized as a risk factor (HR\u0026thinsp;=\u0026thinsp;1.56, 95% CI: 1.14\u0026ndash;2.13, \u003cem\u003eP\u003c/em\u003e\u0026thinsp;=\u0026thinsp;0.005), whereas CTSG (HR\u0026thinsp;=\u0026thinsp;0.69, 95% CI: 0.57\u0026ndash;0.84, \u003cem\u003eP\u003c/em\u003e\u0026thinsp;\u0026lt;\u0026thinsp;0.001) and MADCAM1 (HR\u0026thinsp;=\u0026thinsp;0.81, 95% CI: 0.68\u0026ndash;0.98, \u003cem\u003eP\u003c/em\u003e\u0026thinsp;=\u0026thinsp;0.028) functioned as protective elements. Subsequent expression profiling via the TIMER3.0 database revealed that MATN3 levels were significantly upregulated in gastric cancer tissues compared to normal controls (\u003cem\u003eP\u003c/em\u003e\u0026thinsp;\u0026lt;\u0026thinsp;0.001, Fig.\u0026nbsp;\u003cspan refid=\"Fig2\" class=\"InternalRef\"\u003e2\u003c/span\u003eB). Conversely, observing the tumor samples showed a marked downregulation of CTSG (\u003cem\u003eP\u003c/em\u003e\u0026thinsp;\u0026lt;\u0026thinsp;0.001, Fig.\u0026nbsp;\u003cspan refid=\"Fig2\" class=\"InternalRef\"\u003e2\u003c/span\u003eC) and MADCAM1 (\u003cem\u003eP\u003c/em\u003e\u0026thinsp;\u0026lt;\u0026thinsp;0.01, Fig.\u0026nbsp;\u003cspan refid=\"Fig2\" class=\"InternalRef\"\u003e2\u003c/span\u003eD). To corroborate these in silico findings, Western blotting was performed to assess protein abundance in vitro. As anticipated, MATN3 protein expression was elevated in gastric cancer cell lines (AGS, HGC-27, and MKN-45) relative to the normal gastric epithelial line GES-1 (Fig.\u0026nbsp;\u003cspan refid=\"Fig2\" class=\"InternalRef\"\u003e2\u003c/span\u003eE). In contrast, protein levels of both CTSG (Fig.\u0026nbsp;\u003cspan refid=\"Fig2\" class=\"InternalRef\"\u003e2\u003c/span\u003eF) and MADCAM1 (Fig.\u0026nbsp;\u003cspan refid=\"Fig2\" class=\"InternalRef\"\u003e2\u003c/span\u003eG) were substantially diminished in the cancer cells, aligning perfectly with the bioinformatic predictions.\u003c/p\u003e \u003cp\u003e \u003c/p\u003e \u003cp\u003e \u003c/p\u003e \u003cdiv id=\"Sec23\" class=\"Section3\"\u003e \u003ch2\u003eValidation of MSRS Prognostic Efficacy and Clinical Utility\u003c/h2\u003e \u003cp\u003eThe GSE26253 cohort, comprising 432 patients, functioned as the training set to evaluate the predictive power of the MSRS. Subjects were dichotomized into high- or low-risk tiers utilizing the median score as the specific threshold. Kaplan-Meier plotting demonstrated that individuals in the high-risk bracket faced a notably inferior recurrence-free survival (RFS) relative to their low-risk counterparts (\u003cem\u003eP\u003c/em\u003e\u0026thinsp;\u0026lt;\u0026thinsp;0.0001, Fig.\u0026nbsp;\u003cspan refid=\"Fig3\" class=\"InternalRef\"\u003e3\u003c/span\u003eA). This prognostic trend was subsequently validated within the independent TCGA-STAD cohort (n\u0026thinsp;=\u0026thinsp;386), where elevated risk scores were significantly associated with compromised overall survival (OS) (\u003cem\u003eP\u003c/em\u003e\u0026thinsp;=\u0026thinsp;0.0016, Fig.\u0026nbsp;\u003cspan refid=\"Fig3\" class=\"InternalRef\"\u003e3\u003c/span\u003eB).To gauge the model\u0026rsquo;s temporal accuracy, time-dependent ROC analysis was employed. Within the training dataset, the signature yielded Area Under the Curve (AUC) statistics of 0.770, 0.791, and 0.770 for the 3-, 5-, and 7-year intervals, respectively (Fig.\u0026nbsp;\u003cspan refid=\"Fig3\" class=\"InternalRef\"\u003e3\u003c/span\u003eC). Parallel analysis in the validation group produced corresponding AUCs of 0.648, 0.772, and 0.708 (Fig.\u0026nbsp;\u003cspan refid=\"Fig3\" class=\"InternalRef\"\u003e3\u003c/span\u003eD). Enhancing clinical utility, a composite nomogram was generated by merging the MSRS with clinical staging data to forecast individual outcomes (Fig.\u0026nbsp;\u003cspan refid=\"Fig3\" class=\"InternalRef\"\u003e3\u003c/span\u003eE). Subsequent calibration assays revealed a high degree of concordance between the predicted probabilities and the actual survival events (Fig.\u0026nbsp;\u003cspan refid=\"Fig3\" class=\"InternalRef\"\u003e3\u003c/span\u003eF).\u003c/p\u003e \u003cp\u003e \u003c/p\u003e \u003cp\u003e \u003c/p\u003e \u003c/div\u003e \u003c/div\u003e \u003cdiv id=\"Sec24\" class=\"Section2\"\u003e \u003ch2\u003eInterplay of MSRS with Immune Infiltration and Therapeutic Susceptibility\u003c/h2\u003e \u003cp\u003e\u0026middot;We subsequently explored how MSRS stratification impacts the immunological landscape and chemotherapeutic potential. Infiltration analysis unveiled distinct cellular compositions: the high-risk cohort was characterized by a marked enrichment of monocytes. In contrast, the abundance of activated B cells, natural killer (NK) cells, and myeloid-derived suppressor cells (MDSCs) was substantially depleted in these patients (Fig.\u0026nbsp;\u003cspan refid=\"Fig4\" class=\"InternalRef\"\u003e4\u003c/span\u003eA). Spearman correlation assays validated these observations, establishing a positive linear relationship between MSRS scores and both immature dendritic cells (\u003cem\u003er\u003c/em\u003e\u0026thinsp;=\u0026thinsp;0.153, \u003cem\u003eP\u003c/em\u003e\u0026thinsp;=\u0026thinsp;0.001) and monocytes (\u003cem\u003er\u003c/em\u003e\u0026thinsp;=\u0026thinsp;0.128, \u003cem\u003eP\u003c/em\u003e\u0026thinsp;=\u0026thinsp;0.008). Conversely, an inverse association was identified for MDSCs (\u003cem\u003er\u003c/em\u003e = -0.155, \u003cem\u003eP\u003c/em\u003e\u0026thinsp;=\u0026thinsp;0.001), NK cells (\u003cem\u003er\u003c/em\u003e = -0.147, \u003cem\u003eP\u003c/em\u003e\u0026thinsp;=\u0026thinsp;0.002), and activated B cells (\u003cem\u003er\u003c/em\u003e = -0.136, \u003cem\u003eP\u003c/em\u003e\u0026thinsp;=\u0026thinsp;0.005) (Figs.\u0026nbsp;\u003cspan refid=\"Fig4\" class=\"InternalRef\"\u003e4\u003c/span\u003eB\u0026ndash;C). Regarding therapeutic implications, we analyzed the expression of critical drug targets. The results demonstrated that MET, TYMS, TOP2A (all \u003cem\u003eP\u003c/em\u003e\u0026thinsp;\u0026lt;\u0026thinsp;0.01), and ERBB2 (\u003cem\u003eP\u003c/em\u003e\u0026thinsp;\u0026lt;\u0026thinsp;0.05) were significantly upregulated in the low-risk group relative to the high-risk counterparts (Fig.\u0026nbsp;\u003cspan refid=\"Fig4\" class=\"InternalRef\"\u003e4\u003c/span\u003eD). These findings imply that patients falling into the low-risk category might be more favorable candidates for targeted treatments against these markers.\u003c/p\u003e \u003cp\u003e \u003c/p\u003e \u003cdiv id=\"Sec25\" class=\"Section3\"\u003e \u003ch2\u003eValidation of MSRS Core Genes at Single-Cell Resolution\u003c/h2\u003e \u003cp\u003eTo elucidate the cellular origins of MSRS core genes, we analyzed the GSE167297 single-cell RNA sequencing dataset. Following quality control, 23,069 cells were retained for downstream analysis. Unsupervised clustering identified 21 distinct cell clusters (Cluster 0\u0026ndash;20) (Fig.\u0026nbsp;\u003cspan refid=\"Fig5\" class=\"InternalRef\"\u003e5\u003c/span\u003eA), which were annotated into seven major cell types based on canonical marker genes: T cells, B cells, myeloid cells, epithelial cells, endothelial cells, stromal cells, and mast cells (Fig.\u0026nbsp;\u003cspan refid=\"Fig5\" class=\"InternalRef\"\u003e5\u003c/span\u003eB). Cellular composition analysis revealed that stromal cells were most abundant in deep tumor regions, followed by normal tissues, and least prevalent in superficial regions (Fig.\u0026nbsp;\u003cspan refid=\"Fig5\" class=\"InternalRef\"\u003e5\u003c/span\u003eC). Stromal stiffness scoring demonstrated that stromal cells exhibited the highest stiffness scores among all cell types, significantly surpassing other populations (Fig.\u0026nbsp;\u003cspan refid=\"Fig5\" class=\"InternalRef\"\u003e5\u003c/span\u003eD). t-SNE visualization further confirmed that high-stiffness cells (red) predominantly localized within stromal cell clusters (Fig.\u0026nbsp;\u003cspan refid=\"Fig5\" class=\"InternalRef\"\u003e5\u003c/span\u003eE). Notably, tumor tissues displayed significantly higher overall stiffness scores than normal tissues (Fig.\u0026nbsp;\u003cspan refid=\"Fig5\" class=\"InternalRef\"\u003e5\u003c/span\u003eF). Moreover, stiffness scores progressively increased with tumor invasion depth (Normal\u0026thinsp;\u0026lt;\u0026thinsp;Superficial\u0026thinsp;\u0026lt;\u0026thinsp;Deep) (Fig.\u0026nbsp;\u003cspan refid=\"Fig5\" class=\"InternalRef\"\u003e5\u003c/span\u003eG). Expression profiling of the 25 core genes revealed that the majority (including COL10A1, SPARC, ELN, COL11A1, COMP, COL1A1, and COL5A2) were predominantly enriched in stromal cells (Fig.\u0026nbsp;\u003cspan refid=\"Fig5\" class=\"InternalRef\"\u003e5\u003c/span\u003eH). To further characterize the stromal compartment, we performed subclustering analysis on cancer-associated fibroblasts (CAFs). Based on established markers, CAFs can be divided into three subtypes: myofibroblast CAFs (myCAFs), inflammatory CAFs (iCAFs) and antigen-presenting CAFs (apCAFs) (Fig.\u0026nbsp;\u003cspan refid=\"Fig5\" class=\"InternalRef\"\u003e5\u003c/span\u003eI). Stiffness scoring demonstrated that iCAFs exhibited the highest scores, followed by apCAFs and myCAFs (Kruskal-Wallis test, \u003cem\u003eP\u003c/em\u003e\u0026thinsp;=\u0026thinsp;2.6e-15) (Fig.\u0026nbsp;\u003cspan refid=\"Fig5\" class=\"InternalRef\"\u003e5\u003c/span\u003eJ). Subtype-specific marker expression patterns were as follows: myCAFs highly expressed ACTA2, TAGLN, and MYL9; iCAFs upregulated IL6, CXCL12, and PDGFRA; and apCAFs were enriched for CD74, HLA-DRA, and HLA-DRB1 (Fig.\u0026nbsp;\u003cspan refid=\"Fig5\" class=\"InternalRef\"\u003e5\u003c/span\u003eK). Dynamic shifts in CAF subtype composition were observed across tissue regions (Fig.\u0026nbsp;\u003cspan refid=\"Fig5\" class=\"InternalRef\"\u003e5\u003c/span\u003eL). From normal to superficial tumor regions, the proportions of myCAFs and apCAFs increased significantly, while iCAFs decreased. Conversely, from superficial to deep tumor regions, iCAF prevalence gradually rebounded, whereas myCAF and apCAF proportions declined.\u003c/p\u003e \u003cp\u003e \u003c/p\u003e \u003c/div\u003e \u003c/div\u003e"},{"header":"Discussion","content":"\u003cp\u003eThis study systematically analysed the prognostic value and cell source of these genes in gastric cancer by integrating multiple matrix Matrix Stiffness-related gene sets. Based on 25 matrix Matrix Stiffness-related genes, the MSRS prognosis model was successfully constructed, and three independent prognostic markers, MATN3, CTSG and MADCAM1, were identified. Single-cell analysis shows that these genes are mainly expressed in matrix cells, among which the iCAF subtype has the highest matrix Matrix Stiffness score. These findings provide a new perspective for understanding the role of matrix Matrix Stiffness in the development of gastric cancer, and provide potential targets for prognostic evaluation and targeted treatment.The study found that MATN3 was highly expressed in gastric cancer tissue and significantly associated with poor prognosis (HR\u0026thinsp;=\u0026thinsp;1.56). As a member of the Matrilin protein family, MATN3 is an important ECM protein, which is mainly involved in the development and homeostasis maintenance of cartilage and bone tissue\u003csup\u003e[\u003cspan citationid=\"CR19\" class=\"CitationRef\"\u003e19\u003c/span\u003e]\u003c/sup\u003e. Under normal circumstances, MATN3 maintains the integrity and mechanical properties of the tissue structure by interacting with other ECM components (such as collagen and protean) \u003csup\u003e[\u003cspan citationid=\"CR20\" class=\"CitationRef\"\u003e20\u003c/span\u003e]\u003c/sup\u003e. Single cell analysis confirms that MATN3 is mainly highly expressed in tromal cells, which is consistent with the functional positioning of its ECM components. MATN3 may promote the progression of gastric cancer through two mechanisms: first, excessive deposition increases the density and Matrix Stiffness of ECM, providing a favourable mechanical microenvironment for tumour cells\u003csup\u003e[\u003cspan citationid=\"CR20\" class=\"CitationRef\"\u003e20\u003c/span\u003e]\u003c/sup\u003e ; second, activate the cancer-promoting signalling pathway by interacting with integtin receptors\u003csup\u003e[\u003cspan citationid=\"CR21\" class=\"CitationRef\"\u003e21\u003c/span\u003e]\u003c/sup\u003e. In contrast, CTSG and MADCAM1 as protective genes are lowly expressed in most cancers\u003csup\u003e[\u003cspan citationid=\"CR22\" class=\"CitationRef\"\u003e22\u003c/span\u003e][\u003cspan citationid=\"CR23\" class=\"CitationRef\"\u003e23\u003c/span\u003e]\u003c/sup\u003e.\u003c/p\u003e \u003cp\u003eCTSG, biologically classified as a serine protease, is derived primarily from neutrophils and mast cells, playing a pivotal role in modulating inflammatory cascades and immune responses\u003csup\u003e[\u003cspan citationid=\"CR24\" class=\"CitationRef\"\u003e24\u003c/span\u003e]\u003c/sup\u003e. Prior investigations have highlighted its tumor-suppressive potential across various malignancies. For instance, in non-small cell lung cancer (NSCLC), downregulation of CTSG is correlated with lymphatic metastasis, advanced pathological staging, and a scarcity of infiltrating immune cells, patterns that are indicative of compromised immune surveillance\u003csup\u003e[\u003cspan citationid=\"CR25\" class=\"CitationRef\"\u003e25\u003c/span\u003e]\u003c/sup\u003e. Similarly, studies on head and neck squamous cell carcinoma (HNSCC) have established CTSG as a standalone prognostic indicator; its depletion reportedly disrupts immuno-inflammatory axes, such as the IL-17 pathway, thereby impairing the host\u0026rsquo;s ability to monitor tumor growth\u003csup\u003e[\u003cspan citationid=\"CR26\" class=\"CitationRef\"\u003e26\u003c/span\u003e][\u003cspan citationid=\"CR27\" class=\"CitationRef\"\u003e27\u003c/span\u003e]\u003c/sup\u003e. Furthermore, evidence from colorectal cancer (CRC) suggests that CTSG overexpression can impede tumorigenesis by blocking the Akt/mTOR signaling axis while simultaneously promoting the expression of apoptotic regulators\u003csup\u003e[\u003cspan citationid=\"CR28\" class=\"CitationRef\"\u003e28\u003c/span\u003e]\u003c/sup\u003e. Synthesizing these observations, we postulate that the diminished expression of CTSG may facilitate gastric cancer progression through multifaceted mechanisms.\u003c/p\u003e \u003cp\u003eMucosal vascular addressin cell adhesion molecule 1 (MAdCAM-1) is predominantly expressed on mucosal vascular endothelial cells, where it functions as a critical mediator of lymphocyte homing to the intestinal mucosa via interactions with the α4β7 integrin \u003csup\u003e[\u003cspan citationid=\"CR29\" class=\"CitationRef\"\u003e29\u003c/span\u003e]\u003c/sup\u003e. This interaction is fundamental to the maintenance of mucosal immunity. Consistent with this, our single-cell analysis corroborated that MADCAM1 is primarily enriched in endothelial populations. Clinical evidence suggests that the MAdCAM-1-α4β7 axis may serve as a vital checkpoint for cancer immunosurveillance; specifically, diminished levels of serum soluble MAdCAM-1 have been correlated with adverse prognoses in malignancies such as lung, kidney, and bladder cancer\u003csup\u003e[\u003cspan citationid=\"CR23\" class=\"CitationRef\"\u003e23\u003c/span\u003e]\u003c/sup\u003e. Furthermore, mutations in MADCAM1 are significantly associated with reduced metastasis-free survival. These mutations not only facilitate direct cancer cell migration but also foster an immunosuppressive microenvironment, potentially through mechanisms involving PD-L1-mediated immune escape and reprogramming \u003csup\u003e[\u003cspan citationid=\"CR30\" class=\"CitationRef\"\u003e30\u003c/span\u003e]\u003c/sup\u003e. Consequently, therapeutic strategies aimed at restoring or potentiating MAdCAM-1 expression hold promise for cancer immunotherapy.\u003c/p\u003e \u003cp\u003eIt is worth noting that this study found that the iCAF subtype has the highest matrix Matrix Stiffness score, significantly higher than myCAF and apCAF. This result is different from the traditional view. Previous studies believe that myCAF is the main effect cell of ECM remodelling and matrix sclerosis. Because of its high expression of α-SMA and a variety of collagen, it has a strong ability to synthesise and contract ECM\u003csup\u003e[\u003cspan citationid=\"CR31\" class=\"CitationRef\"\u003e31\u003c/span\u003e]\u003c/sup\u003e. However, this study suggests that the role of iCAF in matrix Matrix Stiffness regulation may be more important than previous understanding. The matrix Matrix Stiffness score used in this study is based on the calculation of 25 MSRS core genes, which not only include ECM structural components, but also include integrated element signalling pathways, YAP signalling pathways and other mechanical force perception and transduction-related genes, so it can more comprehensively reflect the response ability of cells to matrix Matrix Stiffness. Although iCAF does not directly produce a large number of ECM components, its highly expressed inflammatory factors (such as IL-6, CXCL12, etc.) can promote the ECM synthesis and crosslinking of neighbouring cells (including myCAF and tumour cells) through parasecretion. Studies show that IL-6 can upregulate the expression of aminoyl oxidase (LOX) by activating the STAT3 signalling pathway to promote collagen fibre cross-linking and matrix sclerosis\u003csup\u003e[\u003cspan citationid=\"CR32\" class=\"CitationRef\"\u003e32\u003c/span\u003e]\u003c/sup\u003e. In addition, the inflammatory factors secreted by iCAF can also recruit immune cells and reshape the physicochemical characteristics of the tumour microenvironment.\u003c/p\u003e \u003cp\u003eAnother intriguing finding was the distinct spatial distribution pattern of stromal cells, characterized by the highest density in deep tumor regions, intermediate levels in normal tissues, and the lowest density in superficial tumor areas. This pattern deviates from the simplistic \"tumor\u0026thinsp;\u0026gt;\u0026thinsp;normal\" paradigm, suggesting complex spatial heterogeneity. We propose the following explanations: (1) superficial tumor regions may represent the invasive front, where tumor cells actively degrade and remodel the ECM, resulting in limited stromal cell recruitment; (2) deep tumor regions may constitute the established tumor core, where stromal cells are stably integrated into a supportive matrix network; (3) normal tissues maintain physiological stromal cell levels, falling between these two extremes. This observation provides valuable insights into the dynamic evolution of the tumor microenvironment during cancer progression.\u003c/p\u003e"},{"header":"Conclusions","content":"\u003cp\u003eThis study integrates multiple matrix Matrix Stiffness-related gene sets and constructs an MSRS prognostic model based on 25 core genes. The model showed good predictive performance in both the training set (with no recurrence survival as the end point) and the verification set (with total survival as the end point), and the consistency verification between different endpoints confirmed the generalisation ability of the model. The study found that the three molecular markers MATN3, CTSG and MADCAM1 have independent prognostic value. Single-cell analysis shows that the core genes are mainly expressed in matrix cells, among which the iCAF subtype has the highest matrix Matrix Stiffness score. This study provides new insights into understanding the role of matrix Matrix Stiffness in the progression of gastric cancer. MSRS can be used as a prognostic assessment tool for gastric cancer, and the core gene may become a potential therapeutic target.\u003c/p\u003e"},{"header":"Declarations","content":"\u003cp\u003e\u003cstrong\u003eFunding\u003c/strong\u003e\u003cstrong\u003e：\u003c/strong\u003e\u003c/p\u003e\n\u003cp\u003eThe authors declare that no funds, grants, or other support were received during the preparation of this manuscript.\u003c/p\u003e\n\u003cp\u003e\u003cstrong\u003eCompeting interests:\u003c/strong\u003e\u0026nbsp;\u003c/p\u003e\n\u003cp\u003eThe authors declare no competing interests.\u0026nbsp;\u003c/p\u003e\n\u003cp\u003e\u003cstrong\u003eData Availability\u003c/strong\u003e\u003cstrong\u003e：\u003c/strong\u003e\u003c/p\u003e\n\u003cp\u003ePublicly available datasets:\u0026nbsp;The gastric cancer microarray data (Training set) are available in the Gene Expression Omnibus (GEO) under accession number\u0026nbsp;GSE26253, accessible at: https://identifiers.org/geo:GSE26253. The single-cell RNA sequencing data are available in the Gene Expression Omnibus (GEO) under accession number\u0026nbsp;GSE183904, accessible at: https://identifiers.org/geo:GSE183904. The RNA-sequencing data (Validation set) are available in The Cancer Genome Atlas (TCGA-STAD), accessible via the GDC Data Portal at: https://portal.gdc.cancer.gov/projects/TCGA-STAD.\u003c/p\u003e\n\u003cp\u003eData generated in this study: The original contributions presented in the study (including Western Blot images and raw data) are included in the article/Supplementary Material. Further inquiries can be directed to the corresponding author.\u003c/p\u003e\n\u003cp\u003e\u003cstrong\u003eAuthor Contribution\u003c/strong\u003e\u003c/p\u003e\n\u003cp\u003eAll authors conceived and designed the research; Wenzheng Li collected data; Gang Wu conducted research and analyzed and interpreted data; Wei Zhang and Wenzheng Li wrote the initial draft; Wei Zhang and Gang Wu revised the manuscript; Wei Zhang and Wenzheng Li had primary responsibility for final content. All authors read and approved the final version of the manuscript.\u003c/p\u003e"},{"header":"References","content":"\u003col\u003e\u003cli\u003e\u003cspan\u003ePatel AK, Sethi NS, Park H (2026) Gastric cancer: a review. JAMA Published online January 7\u003c/span\u003e\u003c/li\u003e \u003cli\u003e\u003cspan\u003eWang J, Du L, Chen X (2022) Adenosine signaling: optimal target for gastric cancer immunotherapy. Front Immunol 13:1027838\u003c/span\u003e\u003c/li\u003e \u003cli\u003e\u003cspan\u003ePeerani E, Candido JB, Maniati E et al (2025) Matrix stiffness shapes transcriptional profiles and drug responses of pancreatic cancer cells. Acta Biomater Published online December 10\u003c/span\u003e\u003c/li\u003e \u003cli\u003e\u003cspan\u003eHuang J, Zhang L, Wan D et al (2021) Extracellular matrix and its therapeutic potential for cancer treatment. Signal Transduct Target Ther 6(1):153\u003c/span\u003e\u003c/li\u003e \u003cli\u003e\u003cspan\u003eZheng Y, Zhou R, Cai J et al (2023) Matrix stiffness triggers lipid metabolic cross-talk between tumor and stromal cells to mediate bevacizumab resistance in colorectal cancer liver metastases. Cancer Res 83(21):3577\u0026ndash;3592\u003c/span\u003e\u003c/li\u003e \u003cli\u003e\u003cspan\u003eJiang C, Centonze A, Song Y et al (2024) Collagen signaling and matrix stiffness regulate multipotency in glandular epithelial stem cells in mice. Nat Commun 15(1):10482\u003c/span\u003e\u003c/li\u003e \u003cli\u003e\u003cspan\u003eChen E, Zeng Z, Zhou W (2024) The key role of matrix stiffness in colorectal cancer immunotherapy: mechanisms and therapeutic strategies. Biochim Biophys Acta Rev Cancer 1879(6):189198\u003c/span\u003e\u003c/li\u003e \u003cli\u003e\u003cspan\u003eZhang T, Li X, He Y et al (2022) Cancer-associated fibroblasts-derived HAPLN1 promotes tumour invasion through extracellular matrix remodeling in gastric cancer. Gastric Cancer 25(2):346\u0026ndash;359\u003c/span\u003e\u003c/li\u003e \u003cli\u003e\u003cspan\u003eLee D, Ham IH, Oh HJ et al (2024) Tubulointerstitial nephritis antigen-like 1 from cancer-associated fibroblasts contribute to the progression of diffuse-type gastric cancers through the interaction with integrin β1. J Transl Med 22(1):154\u003c/span\u003e\u003c/li\u003e \u003cli\u003e\u003cspan\u003eYang Y, Sun H, Yu H et al (2025) Tumor-associated-fibrosis and active collagen-CD44 axis characterize a poor-prognosis subtype of gastric cancer and contribute to tumor immunosuppression. J Transl Med 23(1):123\u003c/span\u003e\u003c/li\u003e \u003cli\u003e\u003cspan\u003eEmon B, Song YJ, Joy MSH, Kovour MV, Prasanth KV, Saif MTA (2023) Mechanosensitive changes in the expression of genes in colorectal cancer-associated fibroblasts. Sci Data 10(1):350\u003c/span\u003e\u003c/li\u003e \u003cli\u003e\u003cspan\u003eZheng S, Wang J, Ding N et al (2021) Prodrug polymeric micelles integrating cancer-associated fibroblasts deactivation and synergistic chemotherapy for gastric cancer. J Nanobiotechnol 19(1):381\u003c/span\u003e\u003c/li\u003e \u003cli\u003e\u003cspan\u003eXu Q, Yang C, Ning J et al (2025) SMOC2high myofibroblastic cancer-associated fibroblast drives primary cilia-associated tumor microenvironment remodeling and poor prognosis in gastric cancer. Chin J Cancer Res 37(4):603\u0026ndash;623\u003c/span\u003e\u003c/li\u003e \u003cli\u003e\u003cspan\u003eLin Y, Sun X, Li C et al Three-year follow-up of neoadjuvant tislelizumab with chemotherapy in locally advanced gastric and gastroesophageal junction adenocarcinoma: revealing cancer-associated fibroblast heterogeneity corresponding to PD-1 blockade efficacy. Adv Sci (Weinh). Published online December 1, 2025.\u003c/span\u003e\u003c/li\u003e \u003cli\u003e\u003cspan\u003eBlum A, Wang P, Zenklusen JC, Snapshot (2018) TCGA-analyzed tumors. Cell 173(2):530\u003c/span\u003e\u003c/li\u003e \u003cli\u003e\u003cspan\u003eOh SC, Sohn BH, Cheong JH et al (2018) Clinical and genomic landscape of gastric cancer with a mesenchymal phenotype. Nat Commun 9(1):1777\u003c/span\u003e\u003c/li\u003e \u003cli\u003e\u003cspan\u003eJeong HY, Ham IH, Lee SH et al (2021) Spatially distinct reprogramming of the tumor microenvironment based on tumor invasion in diffuse-type gastric cancers. Clin Cancer Res 27(23):6529\u0026ndash;6542\u003c/span\u003e\u003c/li\u003e \u003cli\u003e\u003cspan\u003eHossain MA, Asa TA, Islam MS, Rahman MZ, Moni MA (2025) Identification of key candidate genes for ovarian cancer using integrated statistical and machine learning approaches. Brief Bioinform 26(6):bbaf602\u003c/span\u003e\u003c/li\u003e \u003cli\u003e\u003cspan\u003ePretemer Y, Kawai S, Nagata S et al (2021) Differentiation of hypertrophic chondrocytes from human iPSCs for the in vitro modeling of chondrodysplasias. Stem Cell Rep 16(3):610\u0026ndash;625\u003c/span\u003e\u003c/li\u003e \u003cli\u003e\u003cspan\u003eGuo Z, Su W, Zhou R et al (2021) Exosomal MATN3 of urine-derived stem cells ameliorates intervertebral disc degeneration by antisenescence effects and promotes NPC proliferation and ECM synthesis by activating TGF-β. Oxid Med Cell Longev 2021:5542241\u003c/span\u003e\u003c/li\u003e \u003cli\u003e\u003cspan\u003eLong L, Zou G, Cheng Y, Li F, Wu H, Shen Y (2023) MATN3 delivered by exosome from synovial mesenchymal stem cells relieves knee osteoarthritis: evidence from in vitro and in vivo studies. J Orthop Translat 41:20\u0026ndash;32\u003c/span\u003e\u003c/li\u003e \u003cli\u003e\u003cspan\u003eChan S, Wang X, Wang Z et al (2023) CTSG suppresses colorectal cancer progression through negative regulation of Akt/mTOR/Bcl2 signaling pathway. Int J Biol Sci 19(7):2220\u0026ndash;2233\u003c/span\u003e\u003c/li\u003e \u003cli\u003e\u003cspan\u003eFidelle M, Rauber C, Alves Costa Silva C et al (2023) A microbiota-modulated checkpoint directs immunosuppressive intestinal T cells into cancers. Science 380(6649):eabo2296\u003c/span\u003e\u003c/li\u003e \u003cli\u003e\u003cspan\u003eLi Y, Wu S, Zhao Y et al (2024) Neutrophil extracellular traps induced by chemotherapy inhibit tumor growth in murine models of colorectal cancer. J Clin Invest 134(5):e175031\u003c/span\u003e\u003c/li\u003e \u003cli\u003e\u003cspan\u003eDai Q, Yao X, Zhang Y et al (2024) CTSG is a prognostic marker involved in immune infiltration and inhibits tumor progression through the MAPK signaling pathway in non-small cell lung cancer. J Cancer Res Clin Oncol 151(1):21\u003c/span\u003e\u003c/li\u003e \u003cli\u003e\u003cspan\u003eShen Y, Chen H, Gong X, Wang Z, Chen M, Chen D (2023) Identification of lysosome-related genes in connection with prognosis and immune cell infiltration for drug candidates in head and neck cancer. Open Life Sci 18(1):20220660\u003c/span\u003e\u003c/li\u003e \u003cli\u003e\u003cspan\u003eHuang GZ, Wu QQ, Zheng ZN et al (2021) Bioinformatics analyses indicate that cathepsin G (CTSG) is a potential immune-related biomarker in oral squamous cell carcinoma (OSCC). Onco Targets Ther 14:1275\u0026ndash;1289\u003c/span\u003e\u003c/li\u003e \u003cli\u003e\u003cspan\u003eChan S, Wang X, Wang Z et al (2023) CTSG suppresses colorectal cancer progression through negative regulation of Akt/mTOR/Bcl2 signaling pathway. Int J Biol Sci 19(7):2220\u0026ndash;2233\u003c/span\u003e\u003c/li\u003e \u003cli\u003e\u003cspan\u003eOzawa N, Yokobori T, Osone K et al (2024) MAdCAM-1 targeting strategy can prevent colitic cancer carcinogenesis and progression via suppression of immune cell infiltration and inflammatory signals. Int J Cancer 154(2):359\u0026ndash;371\u003c/span\u003e\u003c/li\u003e \u003cli\u003e\u003cspan\u003eZhang J, Liu F, Yang Y et al (2022) Integrated DNA and RNA sequencing reveals early drivers involved in metastasis of gastric cancer. Cell Death Dis 13(4):392\u003c/span\u003e\u003c/li\u003e \u003cli\u003e\u003cspan\u003eCroizer H, Mhaidly R, Kieffer Y et al (2024) Deciphering the spatial landscape and plasticity of immunosuppressive fibroblasts in breast cancer. Nat Commun 15(1):2806\u003c/span\u003e\u003c/li\u003e \u003cli\u003e\u003cspan\u003eXu S, Deng KQ, Lu C et al (2024) Interleukin-6 classic and trans-signaling utilize glucose metabolism reprogramming to achieve anti- or pro-inflammatory effects. Metabolism 155:155832\u003c/span\u003e\u003c/li\u003e\u003c/ol\u003e"}],"fulltextSource":"","fullText":"","funders":[],"hasAdminPriorityOnWorkflow":false,"hasManuscriptDocX":true,"hasOptedInToPreprint":true,"hasPassedJournalQc":"","hasAnyPriority":false,"hideJournal":true,"highlight":"","institution":"","isAcceptedByJournal":false,"isAuthorSuppliedPdf":false,"isDeskRejected":"","isHiddenFromSearch":false,"isInQc":false,"isInWorkflow":false,"isPdf":false,"isPdfUpToDate":true,"isWithdrawnOrRetracted":false,"journal":{"display":true,"email":"[email protected]","identity":"researchsquare","isNatureJournal":false,"hasQc":true,"allowDirectSubmit":true,"externalIdentity":"","sideBox":"","snPcode":"","submissionUrl":"/submission","title":"Research Square","twitterHandle":"researchsquare","acdcEnabled":true,"dfaEnabled":false,"editorialSystem":"","reportingPortfolio":"","inReviewEnabled":false,"inReviewRevisionsEnabled":true},"keywords":"gastric cancer, matrix stiffness, prognostic model, scRNA sequencing analysis, cancer-related fibroblasts, immune infiltration","lastPublishedDoi":"10.21203/rs.3.rs-8616596/v1","lastPublishedDoiUrl":"https://doi.org/10.21203/rs.3.rs-8616596/v1","license":{"name":"CC BY 4.0","url":"https://creativecommons.org/licenses/by/4.0/"},"manuscriptAbstract":"\u003ch2\u003eObjective\u003c/h2\u003e \u003cp\u003eAlthough there is a significant correlation between tumour matrix sclerosis and the progression of gastric cancer, there is still a lack of a prognostic model based on Stiffness-related genes. For this reason, we have developed the matrix Stiffness-related risk score (MSRS) for prognostic prediction. In addition, this study integrates single-cell RNA sequencing technology to identify the cell source of characteristic genes and analyse the heterogeneity of cancer-related fibroblasts (CAF) lines.\u003c/p\u003e\u003ch2\u003eMethods\u003c/h2\u003e \u003cp\u003eCandidate stiffness-related genes were isolated by intersecting TCGA-STAD differentially expressed genes with four curated gene sets (e.g., Integrin, YAP). We constructed the MSRS via LASSO Cox regression in the GSE26253 training cohort (n\u0026thinsp;=\u0026thinsp;432; endpoint: RFS) and validated it in the TCGA-STAD cohort (n\u0026thinsp;=\u0026thinsp;386; endpoint: OS). Prognostic efficacy was gauged using Kaplan-Meier curves, ROC analyses, and nomograms, while ssGSEA assessed immune infiltration disparities. Furthermore, scRNA-seq data (GSE183904) was leveraged to dissect CAF heterogeneity, and core protein levels were ultimately corroborated by Western blotting.\u003c/p\u003e\u003ch2\u003eResults\u003c/h2\u003e \u003cp\u003eA 25-gene MSRS prognostic model was established. In the training cohort, high-risk patients exhibited significantly shorter recurrence-free survival (HR\u0026thinsp;=\u0026thinsp;4.00, \u003cem\u003eP\u003c/em\u003e\u0026thinsp;\u0026lt;\u0026thinsp;0.0001), with 3-, 5-, and 7-year AUCs of 0.770, 0.791, and 0.770, respectively. The validation cohort confirmed significantly reduced overall survival in the high-risk group (\u003cem\u003eP\u003c/em\u003e\u0026thinsp;=\u0026thinsp;0.0016; AUCs: 0.648, 0.772, 0.708). Multivariate analysis confirmed the risk score as an independent prognostic indicator (HR\u0026thinsp;=\u0026thinsp;3.50, \u003cem\u003eP\u003c/em\u003e\u0026thinsp;=\u0026thinsp;1.71\u0026times;10⁻\u0026sup2;\u0026sup1;). Three independent markers were identified: the oncogene \u003cem\u003eMATN3\u003c/em\u003e (HR\u0026thinsp;=\u0026thinsp;1.559) and tumor suppressors \u003cem\u003eCTSG\u003c/em\u003e (HR\u0026thinsp;=\u0026thinsp;0.693)and \u003cem\u003eMADCAM1\u003c/em\u003e (HR\u0026thinsp;=\u0026thinsp;0.813). Western blotting confirmed \u003cem\u003eMATN3\u003c/em\u003e upregulation and \u003cem\u003eCTSG/MADCAM1\u003c/em\u003e downregulation in tumor cells. High-risk patients showed increased monocyte infiltration but decreased activated B and NK cells. scRNA-seq revealed that stromal cells exhibited the highest stiffness scores. Specifically, inflammatory CAFs (iCAFs) displayed the highest score (\u003cem\u003eP\u003c/em\u003e\u0026thinsp;=\u0026thinsp;2.6\u0026times;10⁻\u0026sup1;⁵) and were enriched in deep tumor regions.\u003c/p\u003e","manuscriptTitle":"Construction and verification of matrix stiffness-related prognostic model of gastric cancer based on single-cell analysis and in vitro experiments ","msid":"","msnumber":"","nonDraftVersions":[{"code":1,"date":"2026-02-13 12:25:35","doi":"10.21203/rs.3.rs-8616596/v1","editorialEvents":[{"type":"communityComments","content":0}],"status":"published","journal":{"display":true,"email":"[email protected]","identity":"researchsquare","isNatureJournal":false,"hasQc":true,"allowDirectSubmit":true,"externalIdentity":"","sideBox":"","snPcode":"","submissionUrl":"/submission","title":"Research Square","twitterHandle":"researchsquare","acdcEnabled":true,"dfaEnabled":false,"editorialSystem":"","reportingPortfolio":"","inReviewEnabled":false,"inReviewRevisionsEnabled":true}}],"origin":"","ownerIdentity":"ef4130c3-6c42-4360-a0b4-7ffa0b5acdee","owner":[],"postedDate":"February 13th, 2026","published":true,"recentEditorialEvents":[],"rejectedJournal":[],"revision":"","amendment":"","status":"posted","subjectAreas":[],"tags":[],"updatedAt":"2026-03-29T19:38:18+00:00","versionOfRecord":[],"versionCreatedAt":"2026-02-13 12:25:35","video":"","vorDoi":"","vorDoiUrl":"","workflowStages":[]},"version":"v1","identity":"rs-8616596","journalConfig":"researchsquare"},"__N_SSP":true},"page":"/article/[identity]/[[...version]]","query":{"redirect":"/article/rs-8616596","identity":"rs-8616596","version":["v1"]},"buildId":"XKTyCvWXoU3ODBz1xrDgd","isFallback":false,"isExperimentalCompile":false,"dynamicIds":[84888],"gssp":true,"scriptLoader":[]}