Comprehensive transcriptomic profiling and mutational landscape of primary gastric linitis plastica

Abstract

Background: Primary gastric linitis plastica (GLP) is a distinct phenotype of gastric cancer with poor survival. Comprehensive molecular profiles and putative therapeutic targets of GLP remain undetermined.

Methods: We subjected 10 tumor-normal tissue pairs to whole exome sequencing (WES) and whole transcriptome sequencing (WTS). 10 tumor samples were all GLP which involves 100% of the gastric wall macroscopically. TCGA data were compared to generate the top mutated genes and the overexpressed genes in GLP.

Results: Our results reveal that GLP has distinctive genomic and transcriptomic features, dysfunction in the Hippo pathway is likely to be a key step during GLP development. 6 genes were identified as significantly highly mutated genes in GLP, including AOX1, ANKRD36C, CPXM1, PTPN14, RPAP1, and DCDC1). MUC6, as a previously identified gastric cancer driver gene, has a high mutation rate (20%) in GLP. 20% of patients in our GLP cohort had CDH1 mutations, while none had RHOA mutations. GLP exhibits high immunodeficiency and low AMPK pathway activity. Our WTS results showed that 3 PI3K-AKT pathway-related genes (PIK3R2, AKT3, and IGF1) were significantly up-regulated in GLP. Two genes were identified using immunohistochemistry (IHC), IGF2BP3 and MUC16, which specifically expressed in diffuse-type-related gastric cancer cell lines, and its knockdown inhibits PI3K-AKT pathway activity.

Conclusions: We provide the first integrative genomic and transcriptomic profiles of GLP, which may facilitate its diagnosis, prognosis, and treatment.

Keywords: Gastric cancer (GC); Gastric linitis plastica (GLP); Scirrhous gastric cancer (SGC); Whole exome sequencing (WES); Whole transcriptome sequencing (WTS).

Fig. 1

Research schematic A Methodology workflow. B Hematoxylin and eosin (H&E) staining of a typical gastric linitis plastica (GLP) sample. Scale bar, 400 μm, Left: × 10; Right × 400

Fig. 2

Somatic mutations analysis A Genomic landscape of GLP. The top bar plot shows the mutation burden for each sample. The body of the graph displays details about each gene, including different mutation types in different samples and the total mutation frequency across all patients with available WES samples. B Genomic landscape of GC without diffuse-type gastric cancer. The bottom of the graph shows the Pathological type of each sample. C Genomic landscape of DGC. D Comparison result of previously reported seven driver genes of gastric cancer between GLP and non-diffuse GC. Different colors indicated different mutation types. E Comparison result of previously reported seven driver genes of gastric cancer between GLP and DGC. F Comparison result of frequently mutated genes of GLP between GLP and GC. G Comparison result of frequently mutated genes of GLP between GLP and non-diffuse DGC. H Significantly differential mutated genes between GLP and GC, *P ≤ 0.05. Number of patients in each group, Odds ratio (OR), and p value are shown in the plot. I In each bar plot, the x-axis represented each one of the 96 types of the 3-bp sequence context and the y-axis indicated the frequency of the 96 substitution patterns. Mutational signatures were deciphered and cosine similarity was used as a metric to compare the similarity of deciphered signatures to known COSMIC SBS signatures. G TCGA and ACRG molecular subtypes of GLP patients, The top part of the graph shows the mutation status of TP53, CDH1, and RHOA, while the bottom part shows the information used for the molecular subtypes decision tree

Fig. 3

Transcriptomic analysis A Heat map of the hierarchical clustering of the top 2000 highly variable genes (HVGs) across GLP and GC datasets. The Spearman correlation coefficient was used as the distance metric for clustering. B Bar plot of Kyoto Encyclopedia of Genes and Genomes (KEGG) enrichment analysis showing the KEGG terms of genes up-regulated in GLP as compared to GC. C Two functional modules found from KEGG enrichment analysis. D Up-regulated genes in GLP-related to module B. E, F Genes that are differentially expressed between GLP and GC, p-adjust values (BH method) are shown in each result

Fig. 4

Potential targeting-genes in GLP. A Venn diagram for genes which up-regulated in GLP compared to both GC samples and normal samples. B IGF2BP3 and MUC16 expression between GLP and GC samples. C Gene Set Enrichment Analysis (GSEA) enrichment analysis between GLP and GC group. Normalized enrichment score (NES) and p values are shown in each result

Fig. 5

MUC16 and IGF2BP3 affect PI3K-AKT pathway activity. A Western blot result of targeted genes expression in GES1, AGS, NUGC4, and KATO III cell lines. B qRT-PCR results of IGF2BP3 and MUC16 in GES1, AGS, NUGC4, and KATO III cell lines. C Western blot result of targeted genes expression in NUGC4 and KATO III cell lines after MUC16 or IGF2BP3 knockdown using siRNAs

Fig. 6

IGF2BP3 and MUC16 expression detected by IHC and WB analysis A–C Typical immunohistochemistry images and corresponding IHC scores of GLP and GC sample. A IGF2BP3 staining. B MUC16 staining. C Corresponding IHC scores. D Western blot result of IGF2BP3 and MUC16 expression levels in five GLP and five GC tissue samples, odd numbered for GLP samples and even numbered for GC samples