PART1 and hsa-miR-429-Mediated SHCBP1 Expression Is an Independent Predictor of Poor Prognosis in Glioma Patients.

Gliomas are the most common primary brain tumors. Because of their high degree of malignancy, patient survival rates are unsatisfactory. Therefore, exploring glioma biomarkers will play a key role in early diagnosis, guiding treatment, and monitoring the prognosis of gliomas. We found two lncRNAs, six miRNAs, and nine mRNAs that were differentially expressed by analyzing genomic data of glioma patients. The diagnostic value of mRNA expression levels in gliomas was determined by receiver operating characteristic (ROC) curve analysis. Among the nine mRNAs, the area under the ROC curve values of only CEP55 and SHCBP1 were >0.7, specifically 0.834 and 0.816, respectively. Additionally, CEP55 and SHCBP1 were highly expressed in glioma specimens and showed increased expression according to the glioma grade, and outcomes of high expression patients were poor. CEP55 was enriched in the cell cycle, DNA replication, mismatch repair, and P53 signaling pathway. SHCBP1 was enriched in the cell cycle, DNA replication, ECM receptor interaction, and P53 signaling pathway. Age, grade, IDH status, chromosome 19/20 cogain, and SHCBP1 were independent factors for prognosis. Our findings suggest the PART1-hsa-miR-429-SHCBP1 regulatory network plays an important role in gliomas.


Introduction
Recently, gliomas have received more attention due to the unsatisfactory 5-year survival rates and median survival times of glioma patients [1][2][3][4][5]. However, current comprehensive treatments, comprising surgery, radiotherapy, and chemotherapy, have not achieved the desired therapeutic effects [2,3]. With the continuous development of genomics, researchers are paying closer attention to the molecular expression changes of glioma patients, such as IDH status, MGMT promoter methylation, and 1p/19q codeletion [6,7]. Therefore, we explored biomarkers that could guide the early diagnosis, treatment, and prognosis of gliomas through genomic data.
Increasing evidence has indicated that altered interactions between lncRNAs, miRNAs, and downstream target genes are closely related to tumor development [11][12][13][14]. The lncRNAs that regulate miRNA activity by means of "sponge" adsorption are also known as competing endogenous RNAs (ceRNAs) [15]. These lncRNAs act as ceRNAs by competitively binding to miRNAs, thereby regulating protein levels and subsequently cellular behaviors [16,17]. However, the lncRNAs that act as ceRNAs in tumors are poorly understood.
SHC binding and spindle associated 1 (SHCBP1) is located on 16q11.2 and encodes a protein in the Src homolog and collagen homolog (SHC) family that is a target of the SHC adaptor protein [18]. Studies have shown that SHCBP1 plays a role in tumor development that may be related to activation of the FGF [18], NF-κB [19], and/or TGF-β1/Smad signaling pathways [20]. Studies have also shown that SHCBP1 is involved in the development of various tumors such as hepatocellular carcinoma [21], glioma [19], breast cancer [22], and gastric cancer [23].
A centrosomal protein of 55 kDa (CEP55) is located at 10q23 and is a member of the centrosome-associated protein family [24]. CEP55 is involved in cytokinesis [25]; during late-stage cytokinesis, CEP55 recruits proteins that are directly related to cell membrane separation, such as tumorsusceptibility gene 101 (TSG101) and ALG2-interacting protein X (ALIX) [26,27]. Regarding tumor development, CEP55 is involved in the development of glioma [28], hepatocellular carcinoma [29], breast cancer [30], lung cancer [31], and ovarian cancer [32]. Although there are reports of a relationship between SHCBP1 and CEP55 in glioma, these have only focused on their roles in the migration and invasion of glioma cells. We used glioma lncRNA and mRNA databases to analyze the role of SHCBP1 and CEP55 in glioma and their associations with prognosis.
We started with the lncRNA database to obtain differentially expressed lncRNAs, miRNAs, and mRNAs and to construct lncRNA-miRNA-mRNA regulatory networks. The receiver operating characteristic (ROC) curve showed that CEP55 and SHCBP1 had higher predictive values for patient prognosis. Furthermore, CEP55 and SHCBP1 were highly expressed in gliomas, showing higher expression in higher grade cases, and high expression patients had poorer prognoses. KEGG enrichment analysis revealed that CEP55 and SHCBP1 are enriched in multiple signaling pathways involved in tumorigenesis and development. Cox analysis found that SHCBP1 along with age, grade, IDH status, and chromosome 19/20 cogain were independent factors for determining patient prognosis.
2.2. Identifying Differentially Expressed lncRNAs, miRNAs, mRNAs, and Overlapping Genes. To obtain differentially expressed lncRNAs, miRNAs, and mRNAs, we selected the lncRNA and mRNA databases of GSE103227 and GSE103229 and obtained differentially expressed lncRNAs and mRNAs. GSE103227 was set to logFC ≥ 3, and p values < 0.01 were considered statistically significant. GSE103229 was set to logFC ≥ 2, and p values < 0.01 were considered statistically significant. Next, we obtained miRNAs that interacted with the differentially expressed lncRNAs using miRcode software (http://www.mircode.org/). We predicted the target genes of these miRNAs using miRDB (http:// mirdb.org/), miRTarBase (https://bio.tools/mirtarbase), and TargetScan (http://www.targetscan.org/vert_72/). We only accepted target genes if they were predicted by all three programs. We retained the overlapping mRNAs from these miRNA target genes and the differentially expressed mRNAs from the databases. Then, based on these overlapping mRNAs, we retained lncRNAs and miRNAs to map the ceRNA network using Cytoscape (https://cytoscape.org/). To increase the credibility of these data, we obtained a corresponding lncRNA-miRNA-mRNA regulatory network through the intersections of the respective lncRNAs, miRNAs, and mRNAs of GSE103227 and GSE103229 in R software (https://www.r-project.org/).

ROC Curve Analysis of mRNA.
After obtaining differentially expressed mRNAs, we performed ROC curve analysis using R software to obtain a molecular marker that could be used to predict patient prognosis.

Expression Level and Survival
Analysis of CEP55 and SHCBP1. To investigate CEP55 and SHCBP1 expression levels in nontumor brain tissue and glioma, we used the GSE4290 and CGGA databases to analyze their expression in nontumor brain tissue and different grades of gliomas. Relationships between these variables and patient prognosis were analyzed using GraphPad Prism software (GraphPad Software, Inc., San Diego, CA, USA).
2.5. Gene Set Enrichment Analysis (GSEA) and Cox Analysis of CEP55 and SHCBP1. To analyze the signaling pathways regulated by CEP55 and SHCBP1 in glioma, we performed KEGG enrichment analysis by GSEA (http://software .broadinstitute.org/gsea/index.jsp). To further determine whether CEP55 and SHCBP1 can be used as molecular markers to predict patient outcomes, we performed singlefactor and multivariate Cox analyses in TCGA glioma database using R software. Multivariate Cox analysis results were presented as forest plots using R software.
2.6. Statistical Analysis. GraphPad Prism 6 was used for all statistical analyses. Statistical significance was defined as a two-tailed p value < 0.05. Univariate Cox analysis and multivariate Cox analysis were performed using R software, with p values < 0.05 considered statistically significant. GSEA was performed using GSEA software, where |normalized enrichment score ðNESÞ| > 1, NOM p value < 0.05, and FDR q value < 0.25 were considered to indicate statistical significance.

Predicting CEP55 and SHCBP1 Expression, Patient Prognosis, and Participation in KEGG Signaling Pathways.
To investigate the accuracy of differential gene prediction for patient prognosis, we plotted the ROC curves. Among    Figures 3(a)-3(i)). By ROC curve analysis, we obtained two lncRNA-miRNA-mRNA regulatory networks: PART1-hsa-miR-429-SHCBP1 and PART1-hsa-miR-301b-3p-CEP55. By analyzing the expression of CEP55 and SHCBP1 in glioma patients and its correlations to patient prognosis, we found higher expression of CEP55 and SHCBP1 in glioma tissues than in nonmalignant tissue (Figures 4(a) and 4(b)) and increased expression with increased glioma grade (Figures 4(c) and 4(d)), and survival analysis found that patients with low CEP55 and SHCBP1 expression had better prognoses (Figures 4(e) and 4(f)). To determine the role of CEP55 and SHCBP1 in glioma, we performed KEGG analyses.
These revealed that CEP55 is enriched in the cell cycle, DNA replication, mismatch repair, and P53 signaling pathway ( Figure 5(a)), while SHCBP1 is enriched in the cell cycle, DNA replication, ECM receptor interaction, and P53 signaling pathway ( Figure 5(b)).

Cox Analysis of CEP55 and SHCBP1 for Predicting
Patient Prognosis. Through ROC curve analysis, we found that CEP55 and SHCBP1 had low false-positive rates as predictors of prognosis. To illustrate the relationship between CEP55, SHCBP1, and patient prognosis, we performed univariate and multivariate analyses. Univariate Cox analysis showed that except for gender and number of genetic mutations, patient prognosis increased the HR of each additional unit of the corresponding indicator, which was statistically   (Figures 6 and 7, Table 1). Based on these results and the developed lncRNA-miRNA-mRNA regulatory network, we found that the PART1-hsa-miRNA-429-SHCBP1 signaling pathway plays an important role in glioma development.

Discussion
Through analyzing the human glioma databases GSE103227 and GSE103229, we obtained differentially expressed lncRNAs and mRNAs. We then found the miRNAs that interacted with these differentially expressed lncRNAs. To increase data, we cross-validated the two database results and obtained two lncRNAs, 11 miRNAs, and 20 mRNAs. After ROC curve analysis, we found that only CEP55 and SHCBP1 were prognostic with high accuracy, and the  BioMed Research International Previous studies have shown that PART1 plays a tumor suppressive function in gliomas [35], but has a procancer effect in bladder cancer [36], non-small-cell lung cancer [37], and colorectal cancer [38]. hsa-miRNA-429 plays an inhibitory role in a number of tumors [39][40][41], including glioma [42][43][44]. SHCBP1 is involved in the development of hepatocellular carcinoma [21], glioma [19], breast cancer [22], and gastric cancer [23]. However, the lncRNAs and miRNAs with which SHCBP1 interacts and whether it could be used as a prognostic molecular biomarker had been unknown. Based on the bioinformatics analysis of a large number of glioma specimens, we found that PART1 and hsa-miRNA-429 could regulate SHCBP1 expression and that SHCBP1 can be used as a molecular biomarker to judge the prognosis of glioma patients. Previous studies have found that SHCBP1 acts as a cancer-promoting gene in a variety of tumors [19][20][21][22][23], which is consistent with our findings. We also discovered multiple tumor-associated signaling pathways, such as cell cycle and DNA replication, that may lead to glioma development. This is consistent with the results of SHCBP1 activity in hepatocellular carcinoma cells and breast cancer cells, where it regulates cell cycle [21,22]; SHCBP1 also regulates cell migration via EMT receptor interactions in synovial sarcoma cells [20]. Additionally, we found that SHCBP1 is low expressed in nontumor brain tissue, but as the level of glioma increases, its expression level is correspondingly increased. Finally, the prognosis of patients with high SHCBP1 expression is poor, which is consistent with the study by Zhou et al. [19]. In addition to the relationship between SHCBP1 and patient prognosis, by Cox analysis, we found that SHCBP1 expression, age, grade, IDH status, and chromosome 19/20 cogain were independent factors associated with patient prognosis. When analyzing IDH (wild) long-term (more than 36 months) and short-term (less than 10 months) survivors of primary glioblastoma, the author found that chromosome 19/20 cogain is a favorable prognostic marker for non-G-CIMP glioma (G-CIMP glioma, CpG island methylator phenotype) patients [45]. Its related mechanism still needs our further research. We obtained results similar to SHCBP1 for CEP55, but it could not be used as an independent factor to judge patient prognosis.
In conclusion, we found that CEP55 and SHCBP1 can be used to predict the prognosis of glioma patients with high accuracy and that SHCBP1 is an independent prognostic factor for glioma patients. SHCBP1 is highly expressed in glioma patients, later-staged gliomas had higher SHCBP1 expression, and SHCBP1 expression levels were negatively correlated with patient survival. SHCBP1 is enriched in multiple signaling pathways such as the cell cycle, DNA replication, ECM receptor interaction, and P53 signaling pathway. The PART1-hsa-miRNA-429-SHCBP1 signaling pathway may play an important role in glioma and provides a new direction for basic research into the underlying biology of gliomas.

Data Availability
The authors confirm that all data underlying the findings are fully available without restriction. All relevant data are within the paper.

Conflicts of Interest
The authors declare that they have no conflicts of interest.

Authors' Contributions
Chengmin Xuan, Mingwei Jin, and Lei Wang contributed equally to this paper. The corresponding author Lei Wang is the affiliated hospital of Xuzhou Medical University, which is the fourth unit. Lei Wang is a member of Xuzhou Children's Hospital.