Identification and Validation of Prognostic Biomarker Signatures Associated with Overall Survival in Colorectal Cancer: Evidence from Bioinformatics Analysis and an in vivo Study

Abstract

Background:

Colorectal cancer (CRC) is one of the most common gastrointestinal cancers worldwide. Thus, genes targeting is useful for the prognosis and therapy of CRC. This study aimed to identify a promising and valuable signature model of prognostic biomarkers associated with overall survival (OS) in CRC.

Methods:

Two mRNA microarray datasets (GSE18105 and GSE113513) from the Gene Expression Omnibus (GEO) database were screened to extract key genes from differentially expressed genes (DEGs) to establish a multiscale embedded gene co-expression network, protein-protein interaction network, and survival analysis. Univariate Cox analysis was conducted to construct a prognostic signature for OS using Kaplan–Meier analysis. Then, we constructed and analyzed the protein-protein interaction network using STRING and Cytoscape, respectively to establish the key genes. Finally, the selected potential prognostic genes were validated in tissue samples of CRC by quantitative real-time PCR (qRT-PCR).

Results:

In the present study, among 340 identified DEGs, four key genes (SPP1, CHEK1, KIF18A, and MAD2L1) were detected. The prognostic gene signature model demonstrated strong performance in the prognosis of CRC (AUC > 0.9). Moreover, the four key genes were also used to construct a risk-score prognostic model for OS and the findings showed that the prognostic gene signature model was highly effective in predicting the OS in CRC patients. The Gene Ontology (GO) enrichment analysis indicated the key genes were significantly associated with several CRC-related signaling pathways such as calcium-independent cell-cell adhesion. Finally, the results of qRT-PCR showed that the upregulation of SPP1, CHEK1, KIF18A, and MAD2L1 was associated with poor prognosis and served as risk factors for CRC patients compared to controls.

Conclusion:

The findings of the present study provided a set of four key genes with valid clinical utility that can serve as an alternative tool for prognosis and identification of new targets in CRC treatment.

Introduction

Colorectal cancer (CRC) is considered the third most common cancer worldwide, and it is predicted to have more than 1.1 million deaths by 2030 [1]. On the other hand, CRC is one of the most prevalent gastrointestinal malignancies [2]. Although significant improvements have been made in the diagnosis and treatment of CRC, global findings indicate that its high mortality remains unsatisfactory for patients with CRC due to cancer recurrence, metastasis, and resistance to radiotherapy and chemotherapy [1]. Therefore, further research is needed to decipher the molecular mechanisms and novel biomarkers associated with the development and progression of CRC. Therefore, this research can provide new perspectives and insights to identify new diagnostic and therapeutic targets and monitor disease progression.

CRC is defined as a heterogeneous disease, which is mostly caused by genetic alterations and interactions of environmental factors [3]. Based on literatures, several genes and molecular pathways such as RACK1 and the lncRNA BCAR4 play an important role in the development of CRC [4, 5]. For instance, it has been reported that the expression of RACK1 is significantly up-regulated in cancerous tissues compared to adjacent normal tissues in patients with CRC. The increased expression of RACK1 mainly leads to increased cell proliferation, migration, and cell invasion [4]. Over the recent years, bioinformatics analysis has served as a valuable tool for several life science applications that could be applied for early diagnosis, prognosis, and treatment of CRC. In addition, advances in high throughput technologies have led to the generation of an unprecedented volume of biological data at different levels of genomics, transcriptomics and proteomics. Gene microarray profiling, a high-throughput method to detect mRNA expression in tissues, has dramatically become a promising tool in medical oncology [6]. By analyzing the differential gene expression between tumoral and control tissues, a better understanding about the molecular pathogenesis of various cancers including CRC can be achieved, which facilitates the identification of potential target genes and signaling pathways for precise therapy. Despite detailed investigations to identify novel targets for CRC management, there is limited comprehensive analysis of gene expression profiles [7, 8]. This analysis can lead to introduce key genes and signaling pathways involved in the CRC progression. Therefore, here, we analyzed GEO datasets to identify potential biomarkers and signaling pathways involved in the progression of CRC and the identified key genes via bioinformatics analysis were then validated by Real Time- PCR.

Materials and Methods

System biology approaches and bioinformatics analysis

Microarray data analysis

After searching NCBI/GEO series (http://www.ncbi.nlm.nih.gov/geo/) using keywords [“Colorectal Cancer” AND “Microarray” AND “Gene OR Transcript OR mRNA”], 1725 results were obtained. By restricting the results to “Homo sapiens” and “Tissue”, 327 data series remained. From the 327 series, 6 were selected and analyzed with GEO2R platform and finally 2 datasets with the highest number of meaningful results were picked up. These included 2 gene transcript datasets, GSE18105 and GSE113513. The expression datasets were all normalized with GEO2R and the statistically differentiating genes (DEGs) were obtained by the cutoff criteria of p-value<0.05 and │Fold-change│>1.5. The details of each dataset including the number of samples and microarray platform type are summarized in Table 1. The volcano plots showing the significant genes are presented in Figure 1.

Table 1.

Details of the Selected Microarray Platforms

GEO series	Platform	#Tumor samples	#Control samples	Ref.
GSE18105	GPL570 [HG-U133_Plus_2] Affymetrix Human Genome U133 Plus 2.0 Array	94	17	Matsuyama T. et al. [9]
GSE113513	GPL15207 [PrimeView] Affymetrix Human Gene Expression Array	14	14	Shen A. et al. [10]

Figure 1.

Normalized GEO Datasets and the Volcano Plots Showing the Significant up/down Regulated Genes in Each Dataset

Integration of datasets results

The differentially expressed genes were analyzed using venny online platform [11] to find the intersection between the datasets. Accordingly, 340 common genes between the gene microarrays were obtained.

Construction of protein-protein interaction network

Common genes were searched in STRING database to find all the possible interactions of the 340 DEGs and construct the protein-protein interaction network. The resulting network was further analyzed in Cytoscape software.

Gene ontology and Pathway enrichment analysis

ClueGO plug-in in Cytoscape was used to find the significant enriched biological processes (BP), molecular functions (MF), cellular components (CC), and KEGG pathways in the DEG network. The Bonferroni-corrected p-value<0.05 was set as statistically significant.

Hub or key genes selection

The constructed network was analyzed in Cytoscape and the nodes with the highest numbers of interactions were selected as the hub genes. Hub nodes usually play key roles in the regulation of the network-related pathways. These nodes might also serve as potential biomarkers of the diseases.

Survival analysis

Survival analysis was performed for the hub genes in the network. To check the possibility of the DEGs correlation with overall survival in colorectal cancer patients, the Human Protein Atlas (HPA) and the UALCAN databases were used. The HPA is an open-source online platform for mapping of human proteins in tissues, cells, and organs by using the data derived from omics studies and antibody-based imaging. In the “pathology” section of HPA, we sought the impact of the DEG expression levels on survival of colorectal cancer patients. The proteins with meaningful correlation with overall survival were selected according to the Kaplan-Meier survival curves. The cutoff p-value was set 0.05.

ROC analysis

After selection of the hub genes which were involved in survival of colorectal cancer patients, the ROC curves were plotted in GraphPad prism 8.0, according to the nodes normalized expression data. The ROC curves are shown in Figure 2. The p-value <0.05 and the area under the curve (AUC) > 0.90 was set as the cutoff for selection of potential biomarkers.

Figure 2.

ROC Curve Analysis Results for Genes with Area under the Curve (AUC) ≥ 0.9 and p-vale<0.05 (including SPP1, CHEK1, KIF18A, and MAD2L1)

Experimental validation

Demographic characteristics of patients

Twelve CRC patients referred to Ayatollah Mousavi Hospital of Zanjan University of Medical Sciences between November 2022 and September 2023 provided the cancerous colon and rectal tissues and adjacent normal tissues. Patients with diabetes, autoimmune disease, cardiovascular disease, and chemotherapy were excluded from the study. This study was approved by the Ethic Committee of Zanjan University of Medical Sciences. Written informed consent was obtained from each patient prior to precipitation. (Ethical Code: IR.ZUMS.REC.1401.089). Demographic information of included patients was presented in Table 2.

Table 2.

Demographic Characteristics of Patients Used for RT-qPCR Validation.

Variable	Number (n= 12)
Sex (M, F)	(6, 6)
Age	61.6 ± 16.9
Tumor size (cm)
< 7	6
> 7	6
TNM Stage	N (%)
I	4 (33.3)
II	3 (25)
III	4 (33.3)
IV	1 (8.3)

M, male; F, female, Variable of age reported as mean ± SD

Validation of the selected genes by qRT-PCR

To validate the bioinformatics analysis results, we employed reverse-transcription quantitative-real-time PCR (qRT-PCR) analyses to assess expression level of SPP1, CHEK1, KIF18A, and MAD2L1 on CRC tissues and adjacent normal. Total RNA was extracted using RiboEx reagent (GeneAll Biotech, Korea) as described by the manufacturer. The quality (based on the appearance of the spectra) and quantity of RNA were assessed using Nano Drop (ND-1000, Thermo Scientific Fisher, US). Three independent RNA samples were used for each real-time PCR experiment. Complementary DNA (cDNA) was synthesized from 3 µg of total RNA using RevertAid™ First Strand cDNA Synthesis Kit (Thermo Fisher Scientific Inc.) following the manufacturer’s instructions. The sequence of the primers is following: I) SPP1: 5ʹ- CGAGGTGATAGTGTGGTTTATGG-3ʹ (F); 5ʹ-GCACCATTCAACTCCTCGCTTTC-3ʹ (R), II) CHEK1: 5ʹ-GTGTCAGAGTCTCCCAGTGGAT-3ʹ (F); 5ʹ-GTTCTGGCTGAGAACTGGAG TAC-3ʹ (R), III) KIF18A: 5ʹ-CAGTTCAGCCTATTCCTT-3′ (F), 5′ TATCACTGTTTATGTTT GAGC-3′ (R), IV) MAD2L1: 5ʹ-TTGAGTGTGACAAGACTGCAAAAG-3′ (F); 5ʹ-CAGTGGCAGAAATGTCACCGTAG-3′ (R), and Glyceraldehyde 3-phosphate dehydrogenase (GAPDH) gene (Biosystems, Life Technologies, USA) was used as internal control. The relative expression value of each gene was determined based on the threshold cycle (Ct) value of the target genes, normalized to that of reference genes (GAPDH) using the 2^-∆∆Ct method. RT-qPCR results were analyzed using SPSS, and plots were designed with GraphPad Prism Software. Data are reported as the mean ± SD and comparison between groups were analyzed by student’s t-test. A p-value <0.05 was considered statistically significant.

Results

Differentially expressed genes (DEGs)

To find the differentially expressed genes, the intersection between datasets was extracted from the Venn diagram. The gene expression microarray GSE18105 resulted in 569, where GSE113513 resulted in 1946 significantly altered genes respectively according to the cutoff criteria. The intersection between these 2 datasets yielded in 340 common genes (Figure 3). The common DEGs were used for further analysis. Supplementary Table 1 provides the detailed names of the common DEGs.

Figure 3.

Venn Diagram Showing Common Genes between the Selected GEO Datasets.

Network analysis and hub genes

The protein-protein interaction (PPI) network consisted of 340 nodes and 783 edges (Figure 4). The network was analyzed in Cytoscape software and the hub nodes with the highest connectivity degrees were selected as potential biomarkers for colorectal cancer. The top 20 hub DEGs included CD44, CDK1, ASPM, BMP2, CHEK1, KIF23, BUB1, MAD2L1, SPP1, WNT5A, CXCL12, TNFSF11, NUF2, SLC26A3, WNT2, CLCA1, KIF18A, LEF1, MMP3, and RFC3 (Table 3). Among these hub nodes, CD44 and SPP1 also served as bottlenecks in the network with the highest betweenness centrality.

Figure 4.

The Protein-Protein Interaction Network, Constructed in Cytoscape based on All the Possible Interactions of DEGs. Larger circles denote higher degree hub genes.

Table 3.

Top 20 Genes with the Highest Degrees in the Protein-Protein Interaction Network. (↑: increased in tumor group, ↓: decreased in tumor group).

No.	ID_REF	P-value	FC	Direction of gene expression (Tumor/control)	Hub Degree
1	CD44	6.91E-09	2.6	↑	30
2	CDK1	0.000000115	2.7	↑	25
3	ASPM	0.0000115	2.3	↑	22
4	BMP2	0.00000246	2.8	↓	22
5	CHEK1	0.00000391	2.0	↑	22
6	KIF23	0.00000576	2.4	↑	22
7	BUB1	0.0000164	2.0	↑	21
8	MAD2L1	0.00000129	2.4	↑	21
9	SPP1	0.017	2.6	↑	20
10	WNT5A	0.0000381	2.5	↑	19
11	CXCL12	1.2E-10	19	↓	18
12	TNFSF11	0.00246	2.1	↑	18
13	NUF2	0.000000445	2.4	↑	17
14	SLC26A3	2.97E-08	9.4	↓	17
15	WNT2	0.00000011	2.0	↑	17
16	CLCA1	0.000188	13.5	↓	16
17	KIF18A	0.000000722	2.1	↑	16
18	LEF1	0.000012	3.5	↑	16
19	MMP3	0.0000343	7.4	↑	16
20	RFC3	8.5E-09	3.0	↑	16

FC. Fold Change

Gene ontology

Based on ClueGO Cytoscape plugin, the meaningful enriched gene ontology (GO) terms and KEGG pathways were determined for the DEGs (Table 4). The most significant biological processes included calcium-independent cell-cell adhesion via plasma membrane cell-adhesion molecules, response to prostaglandin, collagen catabolic process, epithelial to mesenchymal transition, interleukin-1 beta secretion, atrial cardiac muscle tissue morphogenesis, and mesenchymal cell development, negative regulation of cell proliferation involved in contact inhibition, creatine metabolic process, and excretion. The top enriched molecular functions included beta-amyloid binding, glycolipid binding, and alcohol dehydrogenase (NAD) activity. Golgi lumen was the most significant cellular component. KEGG pathways enrichment results also showed that Nitrogen metabolism, pancreatic secretion, and bile secretion were the most meaningful pathways related to colorectal cancer.

Table 4.

Gene Ontology and KEGG Pathway Enrichment Results

GO ID	GO Term	Bonferroni-corrected P-value	Associated Genes (%)	Associated Genes Found
Biological Process
GO:0016338	Calcium-independent cell-cell adhesion via plasma membrane cell-adhesion molecules	260.0E-6	25.00	BMP2, CLDN1, CLDN11, CLDN23, CLDN5, CLDN8
GO:0034694	Response to prostaglandin	47.0E-3	8.33	CCL19, PTGDR, SFRP1
GO:0030574	collagen catabolic process	6.8E-3	9.88	CEL, COL10A1, COL11A1, COL12A1, FAP, MMP1, MMP3, MMP7
GO:0001837	epithelial to mesenchymal transition	2.4E-3	8.85	BMP2, GCNT2, LEF1, MSX2, OLFM1, SFRP1, SFRP2, TMEM100, WNT2, WNT5A
GO:0050702	interleukin-1 beta secretion	25.0E-3	8.11	CCL19, NLRP2, WNT5A
GO:0055009	atrial cardiac muscle tissue morphogenesis	8.4E-3	50.00	PITX2, PROX1, WNT2
GO:0014031	mesenchymal cell development	85.0E-6	8.06	BMP2, CITED2, EDNRA, GCNT2, LEF1, MSX2, OLFM1, PITX2, SEMA6A, SEMA6D, SFRP1, SFRP2, TMEM100, WNT2, WNT5A
GO:0060244	negative regulation of cell proliferation involved in contact inhibition	290.0E-6	57.14	CEL, DACH1, FAP, SRPX
GO:0006600	creatine metabolic process	45.0E-3	27.27	CKB, CKMT2, GHR
GO:0007588	excretion	30.0E-3	8.86	ANPEP, GUCA2B, KCNMA1, NEDD4L, SCNN1B, SCNN1G, SLC26A3
Molecular Function
GO:0001540	beta-amyloid binding	47.0E-3	8.33	BCHE, LDLRAD3, OLFM1
GO:0051861	glycolipid binding	49.0E-3	16.00	CEL, DPEP1, GLTP, LAMA1
GO:0004022	alcohol dehydrogenase (NAD) activity	20.0E-3	37.50	ADH1B, ADH1C, DHRS9
Cellular Component
GO:0005796	Golgi lumen	3.5E-3	9.47	CHGA, MUC12, MUC2, MUC4, MUC5B, OGN, PRELP, WNT5A, ZG16
KEGG Pathway
GO:0000910	Nitrogen metabolism	740.0E-6	29.41	CA1, CA12, CA2, CA4, CA7
GO:0004972	Pancreatic secretion	590.0E-6	10.42	ATP2A3, CA2, CEL, CLCA1, CLCA4, KCNMA1, PLA2G2A, PRKCB, SLC26A3, SLC4A4
GO:0004976	Bile secretion	17.0E-3	9.86	ABCG2, AQP8, CA2, NR1H4, SLC4A4, SLC51B, SLCO1B3

Survival analysis

To assess the relationship between the proposed biomarkers and overall survival of colorectal cancer patients, survival data were extracted from the databases Human Protein Atlas and UALCAN (Figure 5). Among the putative gene/protein markers, CLCA1, WNT5A, and CHEK1 showed correlation with overall survival of colorectal cancer patients according to Human Protein Atlas. SPP1, LEF1, KIF18A, and MAD2L1 also showed correlation with overall survival of patients with colon and/or rectal adenocarcinomas according to UALCAN database. High expression of CLCA1, WNT5A, and CHEK1 are favorable to survival in colorectal cancer. This means the overall survival of patients with CRC decreases as the expression of these 3 genes goes up. Higher expression of MAD2L1 and KIF18A showed favorable and LEF1 showed unfavorable correlation with rectal adenocarcinoma according to UALCAN database. SPP1 expression also had favorable correlation with both colon and rectum adenocarcinomas according to UALCAN database. SPP1 and CHEK1 can serve as prognostic markers for colorectal cancer where all CHEK1, SPP1, MAD2L1, and KIF18A can serve as prognostic markers for rectum adenocarcinoma.

Figure 6.

Relative Gene Expression pattern of A) SPP1, B) CHEK1, C) Kinesin 18A, and D) MAD2L1 between CRC tissues and normal tissues by real-time PCR. The expression of GAPDH was used as internal control gene.

ROC curves

ROC curves were plotted using GraphPad prism software. The results are shown in Figure 2. ROC (Receiver-Operating Characteristic) analysis is a valuable tool for evaluating the performance of a diagnostic test or the accuracy of a statistical model. It is a plot of Sensitivity (the true positive rate) in function of 100-Specificity (the false positive rate) for different cut-off points of a variable. The Area Under the curve (AUC) is a measure of how well a variable can distinguish between diagnostic groups (here, between gastric cancer and control samples). A curve closer to the upper left corner, shows higher overall accuracy of the test. We performed ROC curve analysis for top hub genes which had correlation with the overall survival of colorectal cancer patients according to human protein atlas and UALCAN databases. AUC cutoff for selection of putative biomarkers was 0.90. According to the results, SPP1, CHEK1, KIF18A, and MAD2L1 showed the highest accuracy and had AUC values more than 90 percent. These nodes are proposed as prognostic markers for colorectal cancer (Table 5). They might also have important biological roles in colorectal cancer pathogenesis. The ROC curve analysis results are shown in Table 5.

Table 5.

A Panel of Potential Prognostic Biomarkers Related to Overall Survival of Colorectal Cancer, based on Kaplan-Meier Survival Curves in HPA and UALCAN databases. ROC curve analysis results are also shown. (*Markers with AUC ≥ 0.90 were selected as a prognostic panel.) SPP1 and CHEK1 can serve as prognostic markers for colorectal cancer where all CHEK1, SPP1, MAD2L1, and KIF18A can serve as prognostic markers for rectum adenocarcinoma.

Gene Biomarkers
Gene symbol	KM-Plot p-value	AUC	ROC curve p-value	Survival data-related tissue
SPP1 *	0.019	0.9898 *	<0.0001	Colon, Rectum
CLCA1	0.0003	0.8622	0.0011	Colon, Rectum
WNT5A	0.00003	0.8214	0.0038	Colon, Rectum
CHEK1 *	0.00036	0.9643 *	<0.0001	Colon, Rectum
LEF1	0.024	0.8929	0.0004	Rectum
KIF18A *	0.033	0.9592 *	<0.0001	Rectum
MAD2L1 *	0.0035	0.949 *	<0.0001	Rectum

Real-time qPCR Validation Results

The expression of CHEK1, SPP1, MAD2L1, and KIF18A were evaluated in CRC tissues compare to adjacent normal tissues. The results indicated that the expression levels of all genes were up-regulated significantly in tumor tissues compared to normal tissues (Figure 6).

Discussion

Colorectal cancer is associated with broadly high mortality and morbidity. In addition, the failure to have early screening and diagnosis in patients with CRC leads to poor overall survival rates and prognosis. Therefore, there is an urgent requirement to identify sensitive and specific biomarkers for the management of CRC. Bioinformatics analyses enable to detect the gene expression alterations during CRC and could be an effective approach to identify novel biomarkers. An initial aim of this project was to identify a panel of prognostic biomarkers associated with overall survival in patients with CRC by PPI network analysis. We identified 340 common DEGs from GSE18105 and GSE113513 datasets in CRC tissues compared to adjacent normal tissues. The GO enrichment analysis suggested shared DEGs were significantly enriched in calcium-independent cell-cell adhesion via plasma membrane cell-adhesion molecules, response to prostaglandin, collagen catabolic process, epithelial to mesenchymal transition, interleukin-1 beta secretion, atrial cardiac muscle tissue morphogenesis, mesenchymal cell development, and negative regulation of cell proliferation involved in contact inhibition. In accordance with the present results, previous studies have reported that the cell adhesion is an important component of malignant transformation, cancer progression, and the development of chemo-resistance [12]. Also, according to other studies, prostaglandins exert a profound influence over the adhesive, migratory, and invasive behavior of cells during the development and progression of cancer [13]. Additionally, it is demonstrated that microsomal prostaglandin E2 synthase-1 (mPGES-1) becomes upregulated in cancer [14].

Our results showed that SPP1, CHEK1, CLCA1, WNT5A, LEF1, KIF18A, and MAD2L1 can serve as prognostic markers for CRC, and rectum adenocarcinoma based on HPA and UALCAN analysis. Interestingly, some nodes including SPP1, CHEK1, KIF18A, and MAD2L1 were also detected as prognostic markers for CRC based on AUC values more than 90 percent in ROC curve analysis. In addition, all of these genes were presented as a hub with high degrees in the PPI network. Therefore, this genes panel can play an important role in diagnosis, and prognosis of CRC. Therefore, due to passing through several screening analyses including PPI network, survival and ROC analyses, these 4 genes have a higher value in subsequent evaluations.

Secreted phosphoprotein 1 (SPP1) is involved in immune regulation, cell survival, and tumor progression. Studies have demonstrated that SPP1 plays an important role in certain individual tumors. On the other hand, the results of numerous studies indicated that SSP1 is a prognostic biomarker and also plays an oncogenic role in various types of human cancers [15-21]. In a comprehensive study, Yuan et al., discovered that SPP1 expression was higher in the majority of the human cancers and also the highest expression of SSP1 was remarkably associated with poor survival in various cancers based on Kaplan-Meier plotter and the PrognoScan database [16]. These data suggested that correlated genes with SPP1 were mainly involved in positive regulation of immune cell activation and infiltration. Also, SPP1 has shown a significant positive relationship with the immunocyte and immune marker sets infiltrating degrees. All of these findings provide strong evidence that SPP1 may promote cancer progression via linking with carcinogenic genes and facilitating immune cells’ infiltration in colon adenocarcinoma and other types of cancers [16]. However, the expression profile and oncogenic features of SPP1 in multiple human cancers are remaining unknown. In this regard, Zeng et al., (2022) analyzed the expression of SPP1 and its correlation with tumor prognosis, immune invasion, tumor microenvironment, and immunotherapy in 33 cancer types [17]. In conclusion, they reported that SPP1 contributed to tumorigenesis, tumor progression, and regulated tumor immune microenvironment, revealing SPP1 may be a new and potential target for assessment of prognosis and immunotherapy in diverse cancers [17]. Recently, another research group has found that the SPP1 expression level was positively associated with the infiltration level of dendritic cells, neutrophils, and macrophages in multiple cancer types. It was also significantly positively correlated with hepatitis A virus cellular receptor 2 (HAVCR2), which was observed in most tumor types.

Checkpoint kinase 1 (CHEK1) is an important serine/threonine kinase that blocks damaged DNA from being copied and passed on to offspring, therefore it is an area of increasing interest in cancer development and treatment [22, 23]. It has been reported that there is a correlation between CHEK1 expression and tumor grade and also its recurrence [24, 25]. There is growing evidence that CHEK1 is involved in tumorigenesis and plays an important role in surviving cancer cells after chemotherapy [26]. Further, studies’ results suggest that CHEK1 does not appear to be a tumor suppressor; but it assists tumor growth and might enhance resistance to anticancer therapy. According to Al-Kaabi et al., the up-regulation of CHEK1 may be related to poor prognosis in breast cancer [27]. Furthermore, the up-regulation of CHEK1 was observed in various human cancers such as hepatocellular carcinoma, breast and colon cancer [28-30]. In regard of CRC, Fang et al., have reported that the repression of CHEK1 caused a significant reduction in cell proliferation and CCNB1, an important member of the cyclin family, expression in CRC cells [31]. In another study, Stawinska et al. evaluated alterations in CHEK1 and CHEK2 expression levels in colon cancer and reported that decreased expression of CHEK2 may be a key mechanism involved in the development of colorectal neoplasm [32]. On the other hand, there are limited therapeutic strategy for advanced CRC. A study has shown that CHEK1 inhibitor sensitizes resistant CRC stem cells to nortopsentin [33]. Therefore, these findings could provide a basis to develop an efficient option for CRC treatment. Recently, Tozaki et al. have reported that the combined inhibition of ataxia telangiectasia-mutated serine/threonine kinase (ATM) and CHEK1 appeared synergistic antitumor effects and induced synergistic lethality in CRC cells at a low dose, which can be the basis for creating new treatments for colorectal cancer [34]. In addition, by integrated bioinformatics analysis, Yu et al., have presented key genes and signaling pathways in CRC that CHEK1 were found to be promising prognostic biomarkers among CRC patients [35]. Recent treatment methods have gone towards the identification of proteins involved in mitotic regulation [36]. Drugs with mitotic inhibitor characteristics (such as taxanes and vinca alkaloids) target microtubules, and some positive results have been obtained for the treatment of multiple human cancers. Recently, kinesin motor proteins have been found to be important proteins regulating mitotic processes and also potential targets of carcinoma treatment [37, 38]. Kinesin family member 18A (KIF18A) is a microtubule-associated motor that contributed to cell division [39]. Some investigations have shown that RNA interference suppressed KIF18A expression in cells, and cells were stopped at G2/M. this means that KIF18A regulates the cell cycle [40]. Additionally, according to the previous studies, the low expression of KIF18A has been observed in normal tissues, while it has increased expression in solid tumors, including breast cancer [41], hepatocellular carcinoma [42], colorectal cancer [43], and other types of cancer. For example, Nagahara et al. evaluated the role of KIF18A in the CRC progression. For this purpose, they were assessed the mRNA level of KIF18A by qRT-PCR in patients with CRC [43]. The authors found that the KIF18A was upregulated significantly in CRC compared to the normal colon tissue. In addition, overexpression of KIF18A in CRC is significantly associated with clinicopathologic factors including tumor stage and metastasis, which indicate that KIF18A has a major role in CRC progression. Further in vitro and translational studies by this group demonstrated that KIF18A expression was correlated with metastasis and was presented as a key factor for CRC progression [43]. Additionally, Zhu et al. have conducted a bioinformatics study in order to introduce biomarkers in colon cancer [44]. The results of their analysis indicated that KIF18A may serve as a biomarker for the early diagnosis and progression of colon cancer [44]. Chromosomal instability (CIN) is a hallmark of cancer, and targeting of CIN-associated vulnerabilities is a newly therapeutic option in drug discovery. Recently, Tamayo et al. discovered some analogs that could be used as chemical probes to interrogate the role of KIF18A inhibition [45]. This was the first disclosure of KIF18A inhibitors with in vivo activity that could be hopeful in cancer treatment such as CRC.

In the current analysis, the mitotic arrest deficient 2-like 1 (MAD2L1) is identified as prognostic biomarker for CRC. MAD2L1 as a component of spindle checkpoint has an important role in mitosis [46]. Dysregulation of MAD2L1 could lead to CIN and aneuploidy, which may facilitate the development of human cancers. Several studies have reported the overexpression of MAD2L1 in various human gastrointestinal cancers such as liver and gastric cancer [47-49]. Specifically, some studies have mentioned this protein as a prognostic signature in colon cancer [50]. For example, Rimkus’ results demonstrate that overexpression of MAD2L2 associated with poor prognosis in the CRC [51]. In another study by Ding et al., it has been revealed that the expression of MAD2L1 in CRC tissues is higher than that in normal tissues. In addition, knockdown of MAD2L1 remarkably inhibited CRC cell growth through impairing cell cycle progression and promoting apoptosis [52]. Therefore, MAD2L1 can serve as a novel oncogenic gene, which has a role in the regulation of cancer cell growth and apoptosis and could be used as an emerging diagnostic and therapeutic biomarker for CRC. Totally, these results show the prognostic significance and expression characteristics of SPP1, CHEK1, KIF18A, and MAD2L1 in the CRC that have required further verification in larger populations with CRC. To validation of our results, we finally performed mRNA level alteration analysis of four genes (including SPP1, CHEK1, KIF18A, and MAD2L1) by qRT-PCR in twelve CRC patients. The results showed the up-regulation of them in cancerous tissues compared to normal tissues. In this regard, the previous studies also reported the high expression of these genes in several types of tumor tissues compared to normal tissues that were consistent with the current study [29, 43, 52, 53]. Several notable strengths of this study were the experimental investigation of identified prognostic genes in CRC patients, all the data analyzed in our study was extracted from GEO databases. Then, further experiments with biological tissue samples were performed to validate our findings.

In conclusion, in the present study, we found four key genes involved in prognosis of CRC. Functional analysis showed those key genes were significantly associated with CRC-related signaling pathways such as calcium-independent cell-cell adhesion. On the other hand, the up-regulation of key genes is associated with the poor prognosis in CRC patients. Taken together, this study provides useful insight on the understanding of carcinogenesis and helps in early detection and prognosis of CRC.

Figure 5.

Kaplan-Meier Survival Curves of CHEK1, SPP1, MAD2L1, and KIF18A from the Human Protein Atlas and UALCAN database. Expression of SPP1 shows unfavorable correlation, where the expression of CHEK1, KIF18A, and MAD2L1 show favorable correlation with the survival. SPP1 and CHEK1 can serve as prognostic markers for colorectal cancer where all CHEK1, SPP1, MAD2L1, and KIF18A can serve as prognostic markers for rectum adenocarcinoma.

Author Contribution Statement

MB. J, M. K and N.AD contributed to the conception, and design and critically reviewed and approved the final manuscript as submitted. N.AD contributed to data collection, statistical analyses, and interpretation and wrote the paper draft. All authors approved the final version for submission.