Introduction
Bladder cancer (BLCA) is the most lethal malignancy of the urinary tract and the most common nonskin, solid cancer. In 2020, GLOBOCAN estimated 573278 new cases and 212536 deaths, making BLCA the tenth most diagnosed cancer worldwide [
1,
2]. According to reports, with variable risks of recurrence and progression, the mortality and morbidity of BLCA have gradually increased in recent years [
3,
4]. Transurethral resection of bladder tumor (TURBt) was the gold standard for the initial diagnosis and treatment of non-muscle invasive bladder cancer (NMIBC) [
5]. Due to the high recurrence rate of NMIBC, patients need to undergo disproportionately invasive and unpleasant cystoscopy 4 times each year [
6]. Therefore, a simple and reliable biomarker is necessary for accurate diagnosis of BLCA.
As heritable gene expression alterations, one of the most widespread epigenetic alterations is DNA methylation, which can affect the function of tumor suppressor genes and change their expression [
7‐
13]. Because DNA methylation is conventionally regarded as a silencing epigenetic marker, several methylation markers have been reported in the detection of BLCA and prediction of the risk of disease prognosis and progression in recent years [
14‐
16]. Hence, further research on methylated differentially expressed genes (MeDEGs) using high-throughput data has great significance for discovering novel cancer biomarkers. With the development of bioinformatics, many excellent software and online tools have emerged. These bioinformatics tools provided rapid and convenient analysis methods for the large amount of data from diverse gene-sequencing platforms and accurately screened potential novel genes as biomarkers [
17,
18].
The existing literatures on DNA methylation considered imperfect because the analytical and validated methods used in these studies lacked systematicity and integrity. In this study, the potential biomarker which had strength relation with BLCA were screened from different database used a series of advanced bioinformatics tools. In addition, the results were identified by several online platform to ensure the validation. The aim of these research was to identify the hub MeDEGs that were greatly associated with BLCA. We hope that this research will provide valuable biomarker candidate genes for BLCA diagnosis.
Discussion
BLCA is a common malignancy of the urinary tract and a significant cause of cancer morbidity and mortality worldwide. The five-year survival rate is only 5% in patients with distant metastasis [
29].In recent years, some methods for predication postoperative survival and recurrent of BLCA were reported [
30]. Epigenetic mechanisms take part in an important role in the pathogenesis of BLCA. Identifying accurate biomarkers for primary BLCA is a key clinical need for BLCA diagnostics. At meantime, the effective biomarkers are also important for the therapy of BLCA and healthcare [
31,
32]. Many studies have exploited aberrant DNA expression signatures or methylation signatures to predict the characteristics or prognosis and drug resistance of different type cancer, such as BLCA [
33‐
35] and prostate cancer [
36,
37].
In this study, several bioinformatics analysis methods were applied to identify potential key MeDEGs associated with BLCA. Using two DEG profiles of BLCA obtained from the GEO database, 72 upregulated and 138 downregulated DEGs were observed. By comparing the MeDEG profile retrieved from the GEO database with these DEGs, 8 hypomethylated and highly expressed genes and 17 hypermethylated and lowly expressed genes were identified.
GO enrichment analysis showed that hypermethylated and low expression genes were mainly enriched in organ development and morphogenesis-related BP, especially in neural nucleus and gland development. KEGG enrichment analysis indicated that metabolism for CYP450, several amino acids metabolism and signaling pathways were significantly enriched. Interestingly, these signaling pathways and substances which were closely related to cell proliferation and the pharmacodynamics of antitumor drugs. For instance, PI3K-Akt activation was also found in breast cancer [
38], gastric cancer [
39], and thyroid carcinoma [
40]. Activation of Hedgehog (Hh) signal resulted in tumorigenesis, malignancy, such as basal cell carcinoma, pancreatic cancer, prostate cancer [
41‐
43]. Hypermethylated genes were also related to focal adhesion in the research, which potentially promotes tumor cell proliferation and mobility [
44].
The 17 hypermethylated low expression genes, including ISL1, ABCA8, MFAP4, COX7A1, SPARC1, ALDH1A3, ACOX2, HOXA9, PLSCR4, CPXM2, BCL2, MAMD2, CKB, EFEMP1, SNRPN, GSTM5, and LAMA3 were analyzed using Cytoscape software. EFEMP1, SPARCL1, ABCA8, MFAP4, PLSCR4, MAMDC2, COX7A1, CPXM2, ALDH1A3 and LAMA3 were identified as hub genes. Among these genes, ALDH1A3, HOXA9 and ISL1 methylation patterns have been reported to be related to the clinical outcomes of BLCA [
45‐
47]. SPARCL1 was a prognostic biomarker for colorectal cancer because its expression was downregulated through DNA methylation [
48,
49]. Many genes such as ABCA8, MFAP4 and MAMDC2 also been potential diagnostic and prognostic biomarkers in hepatocellular carcinoma, breast cancer and ovarian cancer [
50‐
53]. Because these genes were related to BLCA at the mechanistic level, it was possible to be a potential biomarker for BLCA.
The most of chosen hub genes were correct by four online platform tools validated. Through multiple genes comparison using the GEPIA online platform, the MFAP4, MAMDC2, SPARCL1, ABCA8 and EFEMP1 had highly difference expression level between tumors and normal tissue. Among these five genes, the SPARCL1, EFEMP1 and MFAP4 had significant highly methylation between normal and tumor tissues using GSCA online platform. The co-efficient >-0.5 between the mRNA expression levels and methylation expression were COX7A1, EFEMP1 and MFAP4. Through analysis, the MFAP4, SPARCL1, EFEMP1, COX7A1, ABCA8 and MAMDC2 would be more likely to become potential biomarker.
As was well known, CpGs were hot-shot regions of the genome, one-third of all point mutations causing genetic diseases in human result from mutation at CpG site [
54]. The DNA methylation was changed during the initiation and progression of cancer with hypomethylation of CpG poor intergenic regions and hypermethylation of CpG islands associated with gene silencing and reduced plasticity [
55]. In the genome of normal cells, promoter CpG islands were hypomethylated. However, tumor cell hypermethylation of the CpG island in the tumor suppressor promoter region was associated with malignant formation and progression [
56,
57]. The methylation alternation of hub genes in BLCA and normal tissues were compared using MEXPRESS visualization tool. The results illustrated there were significant negative correlation in expression and methylation around the CpG and promoter region. The hypermethylation around promoter and CpG region of hub genes may led to down-regulate expression. The hub genes were related with PI3K-Akt and Hedgehog signal transduction which were also associated with cancer cell proliferation and survival. Hence hypermethylation would be associated with hub gene repression and initiate BLCA.
Conclusion
In this study, several differentially methylated genes associated with BLCA were identified. The characteristics of the signatures were confirmed by a series of systematic bioinformatics analysis tools. We hoped these genes, especially the MFAP4, SPARCL1, EFEMP1, COX7A1, ABCA8 and MAMDC2, would be an effective biomarker for BLCA diagnostics.
This study was mainly based on bioinformatic analysis of the GEO database. The amount of data and verification of identified genes were insufficient. In addition, some of hypermethylated genes had been observed not only in BLCA but also in many other cancers. Future research will be needed to confirm the performance of these aberrantly methylated genes in clinical practice.
Publisher’s Note
Springer Nature remains neutral with regard to jurisdictional claims in published maps and institutional affiliations.