Evaluation of Important Molecular Pathways and Candidate Diagnostic Biomarkers of Noninvasive to Invasive Stages in Gastric Cancer by In Silico Analysis

Gastric cancer affects millions of people each year; it is the fifth deadliest cancer globally. Due to failure to perform routine tests such as endoscopy, it is usually diagnosed in the invasive stages. Therefore, finding diagnostic biomarkers in blood can help to speed up the initial diagnosis of cancer. This study aimed to find appropriate diagnostic biomarkers in the extracellular matrix of noninvasive to invasive stages of gastric cancer patients, using bioinformatics analysis. First, we selected the appropriate datasets from the GEO database. We evaluated the genes' signaling pathways, biological processes, and molecular functions. More accurately, we assessed the genes, in which their protein products are released into the extracellular matrix; we evaluated their protein network. Then, we validated the candidate proteins in the GEPIA and TCGA databases. The extracellular matrix, tyrosine kinase receptors, and immune response pathways are effective factors, which are related to the highly expressed genes and metabolism; cell cycle pathways are also impressive on low-expression genes. 69 highly expressed proteins are released into the extracellular matrix. After drawing the protein network, 5 proteins were selected as more suitable candidates for further studies. These proteins' expression significantly increases in the human samples, and the survival chart showed up to about 80% mortality in the individuals over time. With integrated bioinformatics analysis, BGN, LOX, MMP-9, SERPINE1, and TGFB1 proteins have been selected as suitable diagnostic biomarkers for noninvasive to invasive stages of gastric cancer. Further studies are needed to evaluate more precise mechanisms between these proteins.


Introduction
Gastric cancer is one of the most important and lifethreatening cancers in the world. It is the fifth common cancer, which affects about 1.5 million people in the world annually. On the contrary, it has several risk factors; one of the most important is Helicobacter pylori infection, which is a common disease among people worldwide [1,2]. e patients' age is a treatment challenge. e average age of gastric cancer patients is over 40; even in many countries, it starts from the age of 70 [3,4]. Also, developing countries do not make timely diagnosis for gastric cancer due to the lack of regular use of diagnostic tests to examine the gastrointestinal tract, such as endoscopy and photofluorography; so the cancer becomes apparent in the invasive stages [5]. erefore, the initial diagnosis can be checked using more straightforward screening methods, by finding biomarkers in patients' blood.
en, diagnostic tests can be performed. In this regard, bioinformatics techniques are an effective way to study and select genes and their protein products in various diseases [6,7], such as cancer; they have made acceptable progress.
is study analyzed the gene expression profile in noninvasive to invasive stages of gastric cancer using appropriate datasets. Finally, we further investigated the genes and protein products, which play a significant role in the extracellular matrix.

Selecting the Appropriate Dataset.
Using the GEO database, we selected the appropriate dataset. GSE84437 microarray dataset contains 433 samples, which have been classified according to the pathological staging, from the primary to the invasive stage of the disease. e platform of this dataset is GPL6947 Illumina HumanHT-12 V3.0 expression BeadChip. is section examined the patients with gastric cancer in the two separate groups of noninvasive and invasive stage. In Figure 1, one can see more information about this database.

Preparation of the Data for Bioinformatics Analysis.
Using the GEO database and the GEO2R tool, we isolated the gene expression profiles between the noninvasive and invasive stage of gastric cancer. en, we uploaded the genes into an Excel file and categorized the high-and lowexpressed genes. In this part, P < 0.05 was considered as the statistically significant level; no limit was set for LogFC. No expression differentiation ranges were applied to the genes clustering in the study.

Finding Signaling Pathways and Gene
Ontologies. For this step, we uploaded the high-and low-expressed genes separately into the Enrichr database: KEGG and Reactome libraries were used to isolate the signaling pathways from the gene clusters. To examine the gene ontology, we used the GO library in the Enrichr and PANTHER databases. en, we used the ShinyGO database to draw the interaction network of the gene ontology.

Measuring the Relationship between the Protein Products.
We used the STRING database to study the protein products of the genes. en, we examined the proteins, which were most related to each other, at the center of the protein network.

Confirmation of Candidate Genes in the Gastric Cancer
Samples. To more accurately confirm the selected genes from the previous steps and especially the relationship between the protein network, the final genes were evaluated in the GEPIA database; it uses the TCGA and GTEx databases to examine the gene expression in patients' samples. Survival and gene expression diagrams were drawn as box plot; they were illustrated by the GEPIA database.

Examination of the Expression of Differential Genes in the Period between the Noninvasive and Invasive Stage of Gastric
Cancer.
e categorized genes' expression profile showed 1347 high-expressed and 3281 low-expressed genes. For this evaluation, 10 patients in the noninvasive and invasive stage were selected based on the pathological stage; their expression profiles were evaluated. e data quality and classification performed in the PCA diagram are shown in Signaling pathways of the extracellular matrix, tyrosine kinase receptors, immune response, metabolism, and cell cycle were significantly observed in the invasive and noninvasive stage of gastric cancer.
After examining the high-and low-expressed genes in the Enrichr database, the following factors were found to be more prominent: the extracellular matrix, assembly of collagen fibrils and other multimeric structures, receptor tyrosine kinases, clathrin-mediated endocytosis, signaling by NOTCH3, interleukins and immune system, membrane trafficking, and semaphorin interaction signaling pathways in the high-expressed genes and the cell cycle, metabolism of proteins, translation, processing of capped intron-containing pre-mRNA, cellular responses to external stimuli, mitochondrial translation, and separation of sister chromatids' signaling pathways in the low-expressed genes. Figure 2 shows the number of genes involved in each of the signaling pathways separately.

Gene Ontology in Noninvasive and Invasive Gastric
Cancer. Different approaches have examined the genes' nature. In this section, genes were evaluated by three approaches including biological processes, molecular functions, and cellular components. As one can see in Table 1 and Figure 3, biological processes and molecular functions were studied with more focus. Anatomical structure morphogenesis, regulation of the cellular component organization, regulation of the developmental process, actin filamentbased process, multicellular organismal process, and vasculature development pathways in biological processes and cytoskeletal protein binding, actin binding, enzyme binding, extracellular matrix structural constituent, cell adhesion molecule binding, and kinase binding in molecular functions with high expression were observed among them.
RNA processing, organonitrogen compound biosynthetic process, mitotic cell cycle, cellular response to stress, and peptide metabolic process in biological processes and nucleic acid binding, structural constituent of the ribosome, transferase activity, mRNA binding, and small-molecule binding in molecular functions were observed in the bioinformatics data of low-expressed genes. In the cellular components, the focus was on the 69 involved genes in the extracellular matrix. ey have been shown in Figure 4(a).

e Correlation of the Extracellular Matrix Protein Network in the Noninvasive to Invasive Gastric Cancer.
In this part of the study, a significant protein network was obtained between the genes, in which their protein products were released into the extracellular matrix. is protein network had 69 nodes, 380 edges, and protein-protein interaction enrichment (P value: <1.0E − 16). Based on the analyses performed by the STRING database, it was found that these genes play an effective role in the ECM-receptor interaction, focal adhesion, TGF-beta, and the PI3K-Akt signaling pathway. In Figure 4(b), one can see more detailed information on how proteins are related and their associated pathways. Accordingly, in this section, the proteins with the most connection with other proteins in the network were selected; the selected proteins are as follows: BGN, LOX, MMP-9, SERPINE1, and TGFB1.

Candidate Genes Significantly Showed Higher Expression in the Patients with Gastric Cancer.
We evaluated the identification of hub genes and protein products, which play a significant role in gastric cancer's invasive state using the GEPIA database. e expression of BGN, LOX, MMP-9, SERPINE1, and TGFB1 proteins was significantly higher in the samples of gastric cancer patients compared to the healthy people. e survival chart for cited proteins showed 60-80% mortality increment in patients with gastric malignancy ( Figure 5); it indicated the importance of these genes' role.

Discussion
Identifying the gastric cancer biomarkers can improve the process of diagnosis and treatment. Usually, a high percentage of people in a community do not undergo endoscopic and photofluorography tests, unless they have a specific digestive problem [8,9]. Also, because the average age of patients with gastric cancer is 40, many people do not desire to perform these tests routinely [10]. For this reason, careful examination and biomarkers' selection in the body secretions such as blood, saliva, or urine can be helpful in the first step of analysis in the gastric cancer candidates.
In this study, in the first step, bioinformatics analysis could help to predict the candidacy of effective diagnostic biomarkers in gastric cancer. In addition, the gene expression profile of individuals was evaluated in the noninvasive to invasive stage of gastric cancer. In the first step, the associated signaling pathways of gastric cancer invasion were obtained.
One of the important signaling pathways of cancer cell metastasis is the NOTCH3 pathway, which is one of the main angiogenesis actors, but this gene works in many other ways and can facilitate the invasion of cancer cells [11]. e study by Kang et al. showed a direct link between NOTCH3 and Jagged2; NOTCH3 is effective in the development and recurrence of gastric cancer [12]. A study by Wu et al., which assessed patients with gastric cancer, showed that 4 NOTCH receptors play an effective role in stimulating gastric cancer tumor cells; they can be selected as a prognostic biomarker in gastric cancer [13]. Another study by Du et al., a metaanalysis of 1547 gastric cancer patients and 450 control samples, showed higher expression of NOTCH1, NOTCH3, and Jagged1 in the cancer samples than healthy individuals Journal of Oncology 5 [14]. In general, more studies are needed to find NOTCH3related pathways in increasing gastric cancer invasion. One of the most attractive signaling pathways observed in this study was the semaphorin pathway. e semaphorin protein family plays a major role in axonal guidance in neurons. Still, several studies have shown that these neurotransmitters can also play a role in the growth, division, and migration of cancer cells. In a survey by Miyato et al., it was shown that increased SEMA3C expression is directly related to the increased invasion of gastric cancer cells [15].   Journal of Oncology Maejima et al.'s study also indicated that increased SEMA3E expression could play a key role in the development and metastasis of the gastric cancer cells [16]. e study by Pan et al., who worked on the gastric cancer cell line SGC7901, showed that SEMA5A was significantly overexpressed in these cells and effectively invades them [17]. en, siRNAs decrease SEMA5A expression, which significantly reduces angiogenesis and metastasis in these cells; on the contrary, it increases the apoptosis induction. Another intriguing study again by Pan et al. showed that SEMA5A could affect MMP-9 by acting on the MEK/ERK pathway and could play a key role in gastric cell invasion [18]. Another study revealed the SEMA5A important role in metastasis and invasion of gastric cancer cells by uPA regulatory activity and the PI3K/Akt pathway [19]. ese studies showed that the semaphorin family of proteins play a key role in the nervous system and can also be effective in the cancer development and progression.
To select a suitable biomarker with a bioinformatics approach, it is better to evaluate the high-expressed genes more carefully and observe which protein products of these genes release into the extracellular matrix [20]. en, in the next steps, the path and the mechanism of these genes or their protein products can be confirmed, and the appropriate strategies can be achieved for timely detection. is study selected the BGN, LOX, MMP-9, SERPINE1, and TGFB1 proteins with more and more accurate relationships in different signaling pathways based on this method. e drawn protein network shows the critical connection between these proteins and other proteins.
Lysyl oxidase (LOX) was one of the selected proteins in this study. It plays a vital role in organizing the ECM by regulating posttranslational processes in fibrous proteins, such as elastin and collagen [21]. Several studies have examined LOX in gastric cancer. Zhao et al. identified a significant association between LOX and MMP-2 and MMP-9 in the gastric cancer cell line and tumor tissue. Decreased expression of MMP-2 and MMP-9 was also observed in the presence of inhibited LOX [22]. e study by Han et al. also showed that increased LOX expression was directly related to HIF-1; increased expression of this gene indicated the hypoxia role in increasing the activity of tumor cells in gastric cancer [23].
Kasashima et al. expressed the inhibition of LOX by siRNAs could reduce E-cadherin expression and increase vimentin expression, thereby activating the EMT pathway [24]. e EMT pathway is one of the most key and effective pathways to enhance the growth, proliferation, and invasion of tumor cells [25]. Zhang et al.'s study also showed that even after surgery in patients with gastric cancer and tumor tissue isolation, LOX expression was high, which led to the recurrence of the disease [26]. Based on the bioinformatics analysis in this study, it was also shown that LOX expression was significantly higher in patients with gastric cancer. Also,  Another studied gene was BGN. It is a member of the proteoglycan group and is leucine rich, which plays a role in regulating fibrous proteins such as elastin and collagen fibers [27]. A study by Hu et al. showed that high BGN expression could play a role in metastasis and invasion of gastric cells into lymph nodes in both laboratory and animal phases. To induce gastric cell invasion, BGN also stimulates the FAK/paxillin signaling pathway, which significantly increases metastasis [28]. Other studies have been performed on BGN and gastric cancer, most notably in bioinformatics analysis; they have related gastric cancer patients with healthy individuals. High expression of BGN is found in gastric cancer [29,30]. e present study examined the noninvasive to invasive stage of gastric cancer; BGN was found by examining the ECM proteins. As one can see in Figure 5, the survival graph for BGN also decreases dramatically over time.
SERPINE1 in the extracellular matrix revealed another gene association with the noninvasive to invasive stage of gastric cancer. is gene plays an essential role in the cell adhesion, migration of tumor cells to other organs, and cellular aging [31]. e study by He et al. showed that SERPINE1 inhibition using long noncoding RNAs reduces the invasion of gastric tumor cells significantly [32]. Transcriptome studies have shown that SERPNE1 can play an important role in the EMT pathway [33]. Based on the survival chart obtained from this gene, it can be concluded that mortality is directly related to other candidate genes due to increased expression of SERPNE1 and mortality in gastric cancer patients.
Matrix metalloproteinase plays a significant role in the invasion of cancer cells. To date, numerous studies have proven the effective role of this family to invade cancer cells in many malignancies [34]. However, an interesting examined relationship in this study was the role of semaphorin and LOX with MMP-9; these two different signaling pathways can play an important regulatory role for MMP-9.
e study by Wang et al. also showed that MMP-9 is effectively involved in the EMT pathway in gastric cancer. When melatonin was used, the IL-1β/NF-κB/MMP-2/MMP-9 genes' expression and also invasion of gastric cancer cells reduced significantly [35]. Due to the crucial role of MMP-9 in ECM regeneration, it seems that several factors are associated with this gene. HOXC6, as an important transcription factor, enhances the MMP-9 activity and is involved in increasing the metastasis of gastric cancer [36]. Also, the RUNX3 signaling pathway regulating TIMP3 can be effective in MMP-9 activity in gastric cancer invasion [37].
In association with TGFB1 in gastric cancer, there are limited studies on this gene's biomarker role. Most studies are about the mutations, variants, and polymorphisms of TGFB1 with gastric cancer. For example, the C-509T [38] and 509C/T [39] mutations were significantly associated with gastric cancer invasion. erefore, in this study, by performing bioinformatics analysis, we showed that TGFB1 could also be considered as a diagnostic biomarker for gastric cancer.

Conclusion
Finally, it can be mentioned that, with the help of continuous bioinformatics analyses, BGN, LOX, MMP-9, SERPINE1, and TGFB1 proteins enhance the progression of gastric cancer; they play a significant role in the organization and communication of cells in the extracellular matrix. Also, a closer look at the mechanism of action between semaphorin, LOX, and MMP-9 can reveal new pathways in gastric cancer invasion and its association with the EMT pathway.

Data Availability
We selected the appropriate datasets from the GEO database. We evaluated the genes' signaling pathways, biological processes, and molecular functions more accurately.