Molecular subtype of gastric cancer based on apoptosis-related genes reveals differential immune microenvironment and intratumoral microorganisms distribution

Abstract

Background: Gastric cancer (GC) is known for its high heterogeneity, presenting challenges in current clinical treatment strategies. Accurate subtyping and in-depth analysis of the molecular heterogeneity of GC at the molecular level are still not fully understood.

Methods: This study categorized GC into two subtypes based on apoptosis-related genes (ARGs) and investigated differences in tumor immune microenvironment, intratumoral microorganisms distribution, gene expression, and signaling pathways. Key prognostic genes related to apoptosis in GC were identified through random survival forest analysis, and their specific signaling mechanisms were explored. Expression levels of key genes were validated through PCR in paired GC tissues and cancer cell lines. Moreover, biological functions of these key genes were verified in vitro experiments.

Results: A consistent clustering of GC was conducted using 161 apoptosis-related genes (ARGs), resulting in the identification of two subtypes, C1 and C2. Subsequently, significant differences were found in the tumor immune microenvironment, intratumoral microorganisms, gene expression, signaling pathways, and protein interaction networks between the two subtypes. GPX3, PLAT, and CAV1 were identified as key prognostic genes related to apoptosis in GC, with a focus on their impact on disease progression-related pathways. Furthermore, PCR assays validated that these three key genes exhibited significantly low expression levels in both GC cell lines and tissues. Finally, knocking down key genes expression significantly promoted cell proliferation, colony formation and invasion of GC.

Conclusions: Our study conducted a comprehensive analysis of the molecular characteristics of ARGs in GC, revealed their association with the tumor immune microenvironment and intratumoral microorganisms. These findings provide new ideas for the molecular classification of GC.

Keywords: Apoptosis; Gastric cancer; Immune microenvironment; Intratumoral microorganisms; Molecular subtype.

Fig. 1

Identification of subtypes of GC based on Consensus clustering. A Heatmap of the consensus matrix when k = 2. B Cumulative distribution function (CDF) curves from k = 2 to 7. C CDF delta area curves from k = 2 to 7. D Heatmap visualizing the different expression pattern of the 161 apoptosis genes in the two clusters. E Survival curve of the patients in the two clusters. F Apoptotic scores for two subtypes

Fig. 2

Analysis of immune characteristics of GC subtypes. A Quantification of immune processes for metacluster in The Cancer Genome Atlas (TCGA). B-E Immune feature analysis. Boxplots for immune pathways (B), differential immune cell distribution (C), immune score (D) and stromal score (E) in metacluster of TCGA, respectively

Fig. 3

Assessment of the immune levels and molecular characteristics of GC subtypes. A Comprehensive heatmap for microenvironment cell population (MCP) counter and single sample gene set enrichment analysis (ssGSEA) immune score. B Boxplots for immune signatures in metacluster of TCGA. C Differential expression of transcripts per million (TPM) of different immune checkpoints in metacluster of TCGA. D, E: GO (D) and KEGG (E) enrichment analysis of subtype molecular features

Fig. 4

Performance validation of GC molecular subtypes. A Biomarker heatmap for metacluster in TCGA using upregulated genes. B Consistency heatmap for TCGA metacluster and nearest template prediction (NTP) predicted subtype. C Sensitivity analysis of two GC molecular subtypes to common chemotherapy drugs. D Heat map of mutation spectra between two GC subtypes. Percentage is the mutation frequency of the total sample. E-G Microsatellite Instability (MSI) (E), Neoantigen (F) and Tumor Mutational Burden (TMB) (G) between expression groups of metacluster. H Submap heatmap of predicted response to immunotherapy

Fig. 5

Tumor immune dysfunction and exclusion (TIDE) analysis. A Differences in immune dysfunction between subtypes. B Differences in immune exclusion between subtypes. C Differences in immune no-benefit between subtypes. D Differences in immune responder between subtypes

Fig. 6

Differences in the distribution of clinical indicators in C1 and C2 subtypes. Differences in the distribution of age (A), gender (B), grade (C), metastasis (D), node (E), stage (F), tumor (G) between C1 and C2 subtypes

Fig. 7

Microbial distribution between GC subtypes. A Distribution analysis of two GC subtypes and different microbial abundances. B Heatmap of correlation between microbiome and immune infiltration. C, D Association between immunomodulator genes and microbes in C1 (C), C2 (D) subtypes. E, F Association between CTL evasion genes and microbes in C1 (E), C2 (F) subtypes. G Correlation between intratumor microbes and TIDE response. H Infiltration level of M2 macrophages between the response group and non-response group

Fig. 8

Identification of differentially expressed genes (DEGs) between subtypes. A Volcano plot of DEGs. Purple and pink indicate the down-regulation and upregulation of differential expression, respectively (screening conditions: P < 0.05 and |LogFC|>0.585). B Venn diagram of DEGs and apoptosis-related genes (ARGs). CluGen refers to ARGs that used for GC subtype analysis. C Random Survival Forest analysis. D-F Kaplan-Meier analysis of CAV1 (D), GPX3 (E) and PLAT (F). Hexp: group with higher than median value of gene expression. Lexp: group with lower than median value of gene expression

Fig. 9

Functional enrichment of differential genes between subtypes and proteomics analysis. A GO-KEGG enrichment analysis of differential genes from the Metascape database. A cluster network of enriched pathways, in which nodes that share the same cluster are often located close to each other. B Protein interaction network analysis

Fig. 10

GSVA analysis of three key genes. CAV1 (A), GPX3 (B), and PLAT (C)

Fig. 11

GSEA analysis of three key genes. CAV1 (A, B), GPX3 (C, D), and PLAT (E, F)

Fig. 12

Validation of differential expression and clinical significance of key genes in GC. A-C CAV1 (A), GPX3 (B) and PLAT (C) expression in GC cell lines. D-F Expression levels of CAV1 (D), GPX3 (E) and PLAT (F) in cancer tissues and adjacent normal tissues (n = 37). A lower ΔCt value indicates a higher gene expression level. G-I Overall survival (OS) analysis. (Student’s t-test, paired t-test and Kaplan-Meier analysis, *P < 0.05, **P < 0.01, and ***P < 0.001)

Fig. 13

Validation of biological function of key genes in gastric carcinogenesis. A knockdown efficiency of CAV1, GPX3 and PLAT in AGS and HGC-27 cells were verified by PCR analysis. Data are shown as the means ± SDs of three independent experiments. B CCK-8 assay was performed to detect cell proliferation after cell transfection. Each sample had six replicates. C Plate colony formation assay. D Transwell invasion assay. (Student’s t-test, *P < 0.05, **P < 0.01, and ***P < 0.001)