Sheng-Hua Zhuo, Liang-Wang Yang, Shen-Bo Chen, Jin-Ben Zhang, Zhao-Teng Zhang,Zheng-Zheng Li, Kun Yang

Department of Neurosurgery,First Affiliated Hospital of Hainan Medical College,Haikou 570102,China

Keywords:Glioma miR-29b-3p Target gene Prognosis Bioinformatics

ABSTRACT Objective: To investigate the core target genes of miR-29b-3p, and analyze the clinical significance of the core target genes in glioma. Methods: Bioinformatics analysis was used to predict and screen the target genes of miR-29b-3p. STRING and Cytoscape software were used to analyze the protein-protein interaction (PPI) of target genes. the differences expression and survival prognosis in glioma were analyzed by GEPIA and CGGA. Independent prognostic factors analyzed by univariate and multivariate Cox proportional hazards regression model. Results: 22 target genes of miR-29b-3p were predicted using LinkedOmics, miRDB,miRTarBase, TargetScan, and starbase databases. Through the construction of the PPI network,genes out of the network were removed, and a total of 16 genes were screened for further study of their clinical significance. Based on analysis of GEPIA and CGGA databases, COL2A1,DNMT3A, and DNMT3B were excluded. Through further analysis of the univariate and multivariate Cox proportional hazard regression model, finally identified three core target genes: SERPINH1, LOXL2, CDK6. Conclusion: Bioinformatics analysis showed that miR-29b-3p targeted three core genes such as SERPINH1, LOXL2, and CDK6 in glioma. The expression of these genes was different between brain normal tissues and gliomas, between different grades of tumor, IDH mutation status and 1p/19q codeletion status. Its high expression had adverse effects on overall survival and recurrence-free survival. These core target genes can be used as an independent prognostic factor.

1. Introduction

Glioma as the most common primary central nervous system(CNS) tumor, especially glioblastoma (GBM), its therapeutic effect is not satisfactory, the prognosis is extremely poor, and some lowgrade gliomas (LGG) is still prone to relapse despite treatment.Therefore, it has important clinical significance to explore more valuable therapeutic targets for glioma. MicroRNAs (miRNAs,miR) are endogenous small single-stranded RNAs, with a length of about 22 nucleotides, which acts on the target messenger RNA(mRNA) 3'-untranslated region (3'UTR) and negatively regulates

its expression at the post-transcriptional level. miRNAs have been shown to play a vital role in all stages of cancer from origin to metastasis[1]. As a tumor suppressor, miR-29b-3p is abnormally expressed in a variety of tumors, which is closely related to the occurrence and development of tumors[2]. It can target multiple genes, and the target genes are not the same in different tumors.The expression of miR-29b-3p is low, and the restoration of its expression has an anticancer effect on gliomas[3,4].

However, whether the target genes of miR-29b-3p are highly expressed in gliomas, can affect the prognosis of patients and are independent prognostic factors have not been fully elucidated.Herein, we combined the clinical database and miRNA target gene prediction database to construct the protein-protein interaction (PPI)network of the target genes, and further comprehensively analyzed the difference of target gene expression in LGG and GBM under different conditions and its effect on survival prognosis. Through univariate and multivariate Cox proportional hazards regression model analysis, independent prognostic factors were determined, and the core target genes of miR-29b-3p in glioma were screened out.

2. Materials and methods

2.1 Screening of genes that are negatively related to miR-29b-3p expression and survival prognosis in gliomas

The LinkedOmics database (http://www.linkedomics.org/) contains multi-omics data and clinical data on 32 cancers and cancers across tumor types from the Cancer Genome Atlas (TCGA) project. It is also the first multi-omics database that integrates mass spectrometry(MS)-based global proteomics data generated by the Clinical Protein Tumor Analysis Alliance (CPTAC) on selected TCGA tumor samples[5]. We use LinkedOmics database and LGGGBM data set to analyze the expression of genes negatively related to miR-29b-3p and clinical prognosis (high expression, poor prognosis), download the data for follow-up analysis.

2.2 Intersection of multiple databases of miR-29b-3p target genes and construction of PPI network

Through four miRNA target gene prediction databases miRDB(http://mirdb.org/index.html), miRTarBase (http://mirtarbase.mbc.nctu.edu.tw/php/index.php), TargetScan (http: //www.targetscan.org/vert_72/), starbase (starbase.sysu.edu.cn/), download and organize miR-29b-3p target gene data; The Venn diagram was made on the online Venn diagram production website (http://bioinformatics.psb.ugent.be/webtools/Venn/) to obtain the intersection genes, and the overlapping genes continued to be intersected with the negative correlation genes of miR-29b-3p and the positive correlation genes of clinical prognosis. The target gene protein interaction was analyzed by STRING (https://string-db.org/) and the relevant data were downloaded. The PPI map was constructed by Cytoscape software (https://www.cytoscape.org). Genes that did not participate in the network interaction relationship were removed, and the remaining genes were screened out for subsequent analysis.

2.3 Differential expression of target gene sets in different states of gliomas and analysis of survival prognosis

Gene Expression Profiling Interactive Analysis (GEPIA) web server is a database for gene expression analysis based on tumor and normal samples in the TCGA and the GTEx projects. GEPIA2 is an updated and enhanced version that expands gene expression quantification from the gene level to the transcript level, and supports the analysis of specific cancer subtypes and the comparison between subtypes[6].In the GEPIA2 database, we set the [log2FC (fold-change)] ＞ 1, and the P value ＜ 0.01 to analyze the differences expression of target gene sets in LGG and GBM compared with normal tissues. At the same time, we select LGG and GBM to group by the median, and use the overall survival (OS) and recurrence-free survival (RFS)were analyzed for survival. Further select mRNAseq_693 data in the Chinese Glioma Genome Atlas Project (CGGA: http://cgga.org.cn/) to compare the expressions between II, III and IV grades, IDH mutant and IDH wildtype, and 1p/19q codeletion and 1p/19q noncodeletion. Analyze the OS of primary and recurrent gliomas, sorted out and plotted, P value showed: ＜0.05: *, ＜0.01: **, ＜0.001: ***,＜0.0001: ****.

2.4 Univariate and multivariate Cox regression analyses of target genes

We use the mRNAseq_693 data set in CGGA to eliminate cases without survival time and status, and finally enroll 655 patients, and include WHO grade, gender, age, radiotherapy status, chemotherapy status, IDH mutation status, 1p/19q codeletion status, and MGMT promoter Methylation status was included in the analysis, and the expression of each gene was divided into high and low expression groups according to the median. SPSS25.0 was used to perform univariate Cox regression analysis on all variables; variables with P value ＜0.01 were selected for further multivariate Cox regression analysis. The Cox model implemented the forward stepwise regression method ground on partial maximum likelihood estimation.

3. Results

3.1 Genes negatively related to miR-29b-3p expression and survival prognosis in gliomas

A total of 6404 genes negatively related to miR-29b-3p were screened from the LGGGBM data set of LinkedOmics database. In addition, the expression of 8507 genes has been shown to negatively correlate with the survival prognosis of glioma patients.

3.2 Prediction and primary screening of miR-29b-3p target genes and protein-protein interactions

71 target genes of miR-29b-3p were predicted and intersected from miRDB, miRTarBase, TargetScan and starbase (Fig. 1A), following,and then 22 genes were obtained after intersection of genes that were negatively related to miR-29b-3p expression and survival prognosis,namely BMF, CD276, ELN, COL3A1, NASP, DNMT3A, COL4A2,C1QTNF6, NKIRAS2, CDK6, SERPINH1, LOXL2, COL2A1,VEGFA, ABCE1, MCL1, TDG, DNMT3B, COL4A1, DYNLT1,ISG20L2, LAMC1 (Fig. 1B). The PPI of target genes were analyzed by STRING, and the genes that did not participate in the network relationship were deleted such as NASP, C1QTNF6, NKIRAS2,ABCE1, DYNLT1 and ISG20L2. The PPI network was visualized by Cytoscape software (Fig. 1C).

Figure 1 Multi-database intersection of miR-29b-3p target genes and protein-protein interaction network of target genes. (A) miRDB, miRTarBase,TargetScan, starbase predicted miR-29b-3p target gene intersection;(B) predicted intersection genes with miR-29b-3p negative correlation genes and clinical prognosis negative correlation genes. (C) PPI network diagram.1:Negatively correlated significant genes of miR-29b. 2:Negatively correlated significant genes of overall survival.

3.3 Differential expression and survival analysis of target genes in different states of gliomas

We analyzed the differential expression of LGG, GBM and normal tissues of each gene in GEPIA, and analyzed the influence of each gene on OS and RFS in gliomas (Table 1). The expression differences of genes in CGGA at different grades, IDH mutation status and 1p/19q codeletion status and their influence on the OS of primary and recurrent gliomas (Table 2). Based on the analysis of Table 1 and Table 2, GEPIA shows that the expression levels of COL2A1, DNMT3A and DNMT3B in normal brain tissue are not different from those in LGG and GBM. COL2A1 does not affect the prognosis of disease-free survival. CGGA shows that there is no difference in the expression of COL2A1 between grade III and IV,different IDH mutation states, and different 1p/19q codeletion states.It has no effect on the prognosis in primary and recurrent gliomas.The expression of DNMT3A and DNMT3B are no difference in grade II and III. The expression of DNMT3B has no difference in IDH mutation states, so it is not included in the next analysis. The genes namely SERPINH1, VEGFA, MCL1, COL4A2, CD276,COL4A1, ELN, COL3A1, LAMC1, LOXL2, TDG, CDK6 and BMF included in the next analysis.

Table 1 Differential expression analysis and survival analysis of target genes in GEPIA.

3.4 Independent prognostic factors were determined by univariate and multivariate Cox regression analysis for patients with glioma

Through univariate Cox regression analysis, gender, radiotherapy status, chemotherapy status, and MGMT promoter methylation status were eliminated; after WHO grade, age, IDH mutation status, 1p/19q codeletion status and each gene were subjected to multivariate Cox regression analysis. As a result, WHO grade, IDH mutation status,1p/19q codeletion status, SERPINH1, LOXL2, and CDK6 were independent prognostic factors (Table 3).

Table 2 Differential expression analysis and survival analysis of target gene set in CGGA

Table 3 Univariate and multivariate Cox regression analysis of prognosis in 655 patients with CGGA

4. Discussion

In this study, we screened the targeted core gene set of miR-29b-3p in glioma through multi-database for the first time and analyzed the expression differences of these genes between normal and tumor tissues, different grades, IDH mutation status and 1p/19q codeletion status. Its effect on the prognosis of OS and RFS, and can be used as independent prognostic factors. In previous studies, serpin family H member 1(SERPINH1) is a collagen-binding protein and collagenspecific chaperone, which is regulated by miR-29 in the occurrence and development of breast cancer [7]. SERPINH1 is related to endothelial cell migration and matrix remodelling in gliomas and can promote angiogenesis[8]. Moreover, SERPINH1 regulated by miR-29a can promote the growth and invasion of gliomas[9]. The high expression of lysyl oxidase-like 2 (LOXL2) is related to the large size and high grade of gliomas. The overexpression of LOXL2 promotes the proliferation and invasion of glioma cells[10]. In hepatocellular carcinoma, miR-26 and miR-29 target LOXL2 to coinhibit its expression[11]. In mantle cell lymphoma, miR-29 inhibits its expression by directly binding to cyclin-dependent kinase 6(CDK6) 3'UTR[12]. MiR-29b targets CDK6 to inhibit osteosarcoma cell proliferation and migration[13]. In gliomas, neurotensin can regulate CDK6 expression through c-Myc/miR-29b-1, thus promoting tumor occurrence and development[3]. Based on bioinformatics methods, this study screened out the core target genes of tumor suppressor miR-29b-3p in glioma. Although some genes have not been verified to be targeted by miR-29b-3p in gliomas.Previous studies have shown that these genes are closely related to the occurrence and development of glioma[3,8,10], which provides an important basis for miR-29b-3p to become a key therapeutic target.

At present, the main treatment methods for glioma include surgical resection with maximum safety and concurrent radiotherapy and chemotherapy, but the patient's prognosis is still very poor. Due to the existence of blood-brain tumor barrier (BBTB) and tumor heterogeneity, the development and application of anti-tumor drugs have increased the difficulty. Recently, the research on the combination therapy of glioblastoma drugs based on nano-carriers has greatly increased, which is conducive to the development of new combinations of nano-drugs and nano-thermochemotherapy to combat gliomas[14,15]. Nanomedicine provides a new way for site-specific delivery of miRNA or anti-miRNA alone or in combination with other therapeutic drugs in the brain to achieve the best therapeutic response. The nano-delivery system can not only protect miRNA from nuclease degradation but also prolong its halflife in blood, making it easier to play a role through BBTB[16]. This provides a new idea for the treatment of gliomas. A study confirmed the anticancer effect of miR-29b-conjugated nanoparticles in a human GBM tissue section culture system[17]. miR-29b is expected to become a new target for GBM treatment.

In conclusion, this study used multiple databases such as LinkedOmics, miRDB, miRTarBase, TargetScan, starbase, GEPIA,and CGGA to analyze the relationship between the target genes of miR-29b-3p and clinical characteristics and its impact on clinical prognosis. T Through univariate and multivariate Cox proportional hazard regression model analysis to determine independent prognostic factors, we finally identified 3 genes, SERPINH1,LOXL2 and CDK6, as the core target genes of miR-29b-3p. The expression of each gene was different between normal and tumor tissues, different grades, IDH mutation status and 1p/19q codeletion status. Its high expression has an adverse effect on the prognosis of OS and RFS and can be used as an independent prognostic factor for glioma.