Background
Gastric cancer (GC) is a malignant disease posing a serious threat to human health. Over one million new gastric cancer cases were reported in 2020, resulting in an estimated 769,000 deaths worldwide [
1]. Currently, it has the fifth highest incidence and accounts for the fourth highest mortality among all cancers globally. Recurrence of peritoneal metastasis is one of the leading causes of death in patients with advanced gastric cancer. Peritoneal metastasis refers to a form of cancer metastasis caused by the direct growth of cancer cells in the primary tumor of gastric cancer through blood, lymph or peritoneum [
2]. Nearly 20% of gastric cancer patients were diagnosed with peritoneal metastasis before or during surgery, and more than 50% of patients with T3 and T4 stages developed peritoneal metastasis after radical resection. The higher the degree of peritoneal metastasis, the shorter the survival time [
3]. Consequently, searching for novel treatments for gastric cancer peritoneal metastasis is of utmost importance.
Long non-coding RNAs (lncRNAs) are a class of RNAs over 200 nucleotides in length and are generally considered to have no coding capacity [
4,
5]. In recent years, lncRNAs have been found to play multiple roles in the initiation, development and metastasis of cancers [
6]. Specifically, Lnc030 has been reported to be involved in maintaining cancer cell stemness and promoting cancer initiation and progression in breast cancer [
7]. LncRNA H19 has also been found to play a crucial role in bone metastasis of hepatocellular carcinoma [
8]. In gastric cancer, lncRNA GCMA was found to promote growth, invasion, and metastasis through a competing-endogenous-RNAs pattern [
9].
Weighted Gene Co-expression Network Analysis (WGCNA) is a more systematic and comprehensive R package for high-dimensional data processing than conventional differentially-expressed-gene analysis. Using WGCNA, it is possible to identify and screen key biomarkers and co-expressed gene modules [
10]. Zhang et al. identified methionine sulfoxide reductase B3 (MSRB3) as a key gene up-regulated in peritoneal metastasis of gastric cancer using WGCNA [
11]. Zhang et al. identified estrogen receptor 1 (ESR1), histone deacetylase 1 (HDAC1) and clathrin heavy chain (CLTC) associated with the occurrence of stomach adenocarcinoma using integrated bioinformatics analysis [
12]. However, to our knowledge, there is no report identifying key lncRNAs for gastric cancer peritoneal metastasis using WGCNA.
In this study, we performed whole-transcriptome sequencing on tissue samples of gastric cancer peritoneal metastasis, and identified GCPM-related lncRNAs (lnc-RFNG-1 and lnc-TRIM28-14) and mRNAs (CD93, COL3A1, and COL4A1). Intriguingly, lnc-TRIM28-14 was positively correlated with the expression levels of CD93 and COL4A1 in gastric cancer peritoneal metastasis, suggesting a regulatory relationship between them. Consequently, we hypothesized that lnc-TRIM28-14, CD93 and COL4A1 could regulate GCPM and could be novel tumor markers and potential therapeutic targets for GCPM.
Discussion
Peritoneal metastasis is a common metastasis mode and an end state of gastric cancer usually associated with poor prognosis. Based on the seed-soil theory, gastric cancer peritoneal metastasis is a polymorphic and multi-molecular regulated process, including changes in seeds and soil and the regulation of multiple RNAs [
15,
16]. Many scholars have made efforts to explore the molecular mechanism underlying GCPM and find new methods for the diagnosis and treatment of GCPM. LIMK1 was reported to mediate GCPM [
17], whereas miR-30a-5p, -659-3p, and -3917 were reported to be potential biomarkers of GCPM [
18]. Recent studies have found several lncRNAs with important roles in GCPM [
19‐
21].
However, no research has been conducted on GCPM-related lncRNA based on WGCNA, which is a more systematic biological algorithm. We performed RNA sequencing, constructed a co-expression network with as many lncRNAs as possible using WGCNA and found GCPM-related blue module and identified six candidate hub lncRNAs (DNM3OS, lnc-MFAP2-53, lnc-PPIAL4C-4, lnc-RFNG-1, lnc-TRIM28-14, lnc-YARS2-4) associated with GCPM. Subsequently, we performed tissue qRT-PCR, RT-PCR validation and ROC curve, which revealed that the expression of lnc-TRIM28-14 and lnc-RFNG-1 was associated with GCPM. Moreover, high expression of lnc-TRIM28-14 indicated a poor prognosis. However, we did not demonstrate that expression of lnc-RFNG-1 was associated with prognosis, probably due to the low number of cases used, individual differences and the proportion of patients with and without peritoneal metastasis in our cases.
The lncRNA lnc-TRIM28-14 is an intergenic sequence transcribed from chromosome 19. It occurs adjacent to tripartite motif-containing protein 28 (TRIM28) gene. The TRIM28 gene belonged to Transcriptional Intermediary Factor 1 family and worked as one of key regulators of development and differentiation [
22]. In recent years, studies have found that TRIM28 could regulate tumor progression and promote tumor growth by enhancing TRIM24 stability [
23]. Another study found that TRIM28 inhibited immunostimulatory cytokine expression [
24]. A recent study found that lncRNA could bind to TRIM28 to inhibit its function [
25]. Interestingly, TRIM28 was reported to be associated with GCPM [
26]. Therefore, it was worth considering whether there was a potential association between lnc-TRIM28-14 and TRIM28 in expression or function. Lnc-RFNG-1 is an intronic lncRNA derived from Ofucosylpeptide 3-beta-N-acetylglucosaminyltransferase (RFNG) gene. RFNG, a member of Fringe proteins, has been reported to regulate cell growth and differentiation in a NOTCH-dependent manner [
27‐
29]. To date, little research has been done on RFNG, and even less is known about its role in tumors. Only one study reported that RFNG might be involved in the progression of hepatocellular carcinoma [
30]. Therefore, Lnc-RFNG-1 opens up new insights into how genes work independently of proteins.
lncRNAs are often considered to play a role in regulating gene expression. Therefore, we also performed WGCNA on mRNA sequencing data. We found GCPM-related magenta modules and screened six candidate hub genes (CD93, COL1A2, COL3A1, COL4A1, COL4A2, and COL6A1). GCPM-related CD93, COL3A1 and COL4A1 were identified using tissue qRT-PCR and ROC curve analysis. Kaplan–Meier survival analysis suggested that high expression of CD93, COL3A1 and COL4A1 indicated a poor prognosis. Unfortunately, we did not verify that the expression of COL1A2, COL4A2, and COL6A1 was associated with GCPM and prognosis, most likely due to our limited sample size. CD93 is a transmembrane protein, which has received increasing research attention in recent years. Studies found that CD93 promoted endothelial cells β1 integrin activation during tumor angiogenesis [
31] and blockade of CD93 can promote tumor vascular maturation, improving drug delivery and immune microenvironment [
32]. COL3A1 and COL4A1 are important components of the extracellular matrix. COL3A1 was reported to promote renal cell carcinoma growth and metastasis [
33] whereas COL4A1 facilitated hepatocellular carcinoma cells proliferation, migration and invasion by activating FAK-Src signaling [
34]. Bioinformatics analysis suggested that H19 promoted gastric carcinogenesis by promoting COL3A1 and COL4A1 expression [
35]. However, the association of these molecules with GCPM has not been reported.
We considered how ln-TRIM28-14 and lnc-RFNG-1 mediated GCPM. Subsequent GO and KEGG analyses suggested that hub lncRNAs and genes were enriched in similar pathways. Using GSEA, we identified ECM-receptor interaction and focal adhesion as the potential key pathways of GCPM. Therefore, we speculated that the “seed” gastric cancer cells enhanced invasiveness by interacting with the matrix, thus mediating peritoneal metastasis. We considered whether there was an expression regulation relationship between hub lncRNAs and genes. Consequently, we analyzed the qRT-PCR results. Finally, we verified that expression of lnc-TRIM28-14 was positively correlated with that of CD93 and COL4A1. In recent years, studies found that lncRNAs can regulate gene expression and present various patterns [
36], such as competing-endogenous-RNAs pattern [
37], transcription regulation [
38] and alternative splicing [
39].
Unfortunately, due to time constraints, we did not conduct more experiments to verify the regulatory relationship of lnc-TRIM28-14 with CD93 and COL4A1. We considered to verify the specific roles and molecular mechanisms of these lncRNAs and genes mediating GCPM in our follow-up study. In addition, we found no association between the expression of ln-RFNG-1 and that of hub genes. We speculated that ln-RFNG-1 might regulate the biological behavior of tumor cells in an unconventional pattern, promoting seed invasion and colonization into the abdominal cavity.
Publisher’s Note
Springer Nature remains neutral with regard to jurisdictional claims in published maps and institutional affiliations.