PBX3 as a biomarker for the early diagnosis and prediction of prognosis of glioma

Background Increasing evidence have elucidated that PBX3 played a crucial role in cancer initiation and progression. PBX3 was differentially expressed in many cancer types. However, PBX3 potential involvement in gliomas remains to be explored. Methods The expression level of PBX3 in glioma tissues and glioma cells, and its correlation with clinical features were analyzed by data from TCGA, GEPIA, CGGA and CCLE. Univariable survival and Multivariate Cox analysis was used to compare several clinical characteristics with survival. We also analyzed the correlation between PBX3 expression level and survival outcome and survival time of LGG and GBM patients by using linear regression equation. GSEA was used to generate an ordered list of all genes related to PBX3 expression and screening of genes co-expressed with PBX3 mRNA by "limma" package. Results The results showed that PBX3 was highly expressed in gliomas and its expression increased with the increase of malignancy. Survival analysis found that PBX3 is more valuable in predicting the OS and PFI of LGG patients than that of GBM. For further study, TCGA and CGGA data were downloaded for univariate Cox analysis and multivariate Cox analysis which showed that the expression of PBX3 was independent influencing factors for poor prognosis of LGG patients. Meanwhile, Receiver operating characteristic (ROC) curve showed that PBX3 was a predictor of overall survival rate and progression-free survival rate of LGG. Linear regression model analysis indicated that the higher expression of PBX3 the higher the risk of death of LGG patients, and the higher expression of PBX3 the higher the risk of disease progression of LGG patients. Next, TCGA data were downloaded for GSEA and Co-expression analyses, which was performed to study the function of PBX3. Conclusion PBX3 may be involved in the occurrence and development of glioma, and has potential reference value for the early diagnosis and prediction of prognosis of glioma.


Introduction
Glioma which originates from glial cells is the most common malignant tumor in the human central nervous system (CNS), accounting for about 29% of all brain tumors and 80% of intracranial malignant tumors [1].The median survival time of glioblastoma is less than 1.5 years, and the 5-year survival rate is only 0.05%-4.7%[2,3].At present, the diagnosis of gliomas is mainly based on clinical manifestations, imaging examination and pathological analysis.The treatment of glioma mainly includes surgical resection combined with postoperative radiotherapy and chemotherapy.The glioma of G1 and G2 grows slowly, have a good prognosis, and are generally considered to be benign.They can be completely respected and even completely recovered from surgery.The glioma of G3 and G4 which have the highest incidence rate grow rapidly and spread easily to other brain tissues.As a result, it is easy to relapse and has the poor prognosis.Thus, it is quite urgent to investigate the mechanisms underlying the development and progression of glioma in order to identify sensitive and specific early biomarkers for diagnosis and prognosis.
Pre-B cell leukemia homeobox 3 (PBX3) is a member of the PBX family, which can be bind to (HOX) proteins, enhance the DNA binding / transcriptional activity of HOX proteins, and regulate the transcription and expression of target genes.It is reported that PBX3 is highly expressed in a variety of tumors, such as prostate cancer, colorectal cancer, gastric cancer, hepatocellular carcinoma and multiple myeloma.Moreover, the expression level of PBX3 was related to tumor grade and pathological stage [4][5][6].Functional studies have shown that the expression of PBX3 is related to the malignant biological phenotype of tumors.For example, high expression of PBX3 can promote tumor cell proliferation [7][8][9], invasion and migration [4,9], stem cell characteristics [10].Xu et al. reported that PBX3 is highly expressed in GBM and promotes GBM proliferation, invasion, migration and apoptosis [6,11].In our previous studies, we found that PBX3 is highly expressed in gliomas, and promote glioma cell proliferation and inhibits apoptosis.However, the expression level of PBX3 correlation with clinical features, and the correlation between PBX3 and the prognosis of gliomas has not been studied.
Herein, the relationship between the expression level of PBX3 and glioma was investigated from the perspective of bioinformatics.First, we analyzed the expression of PBX3 in glioma tissue, the relationship between PBX3 and glioma patient's clinical characteristics, and the prognostics value of PBX3 by TCGA, CGGA and GEPIA database.Then, TCGA and GTEx data were used to comprehensively analyze the clinical correlations.TCGA data were downloaded for differently expressed genes (DEGs) identification, GSEA and Co-expression analyses was performed to study the function of genes positively related to expression of PBX3.

Survival and expression analysis by GEPIA
Gene Expression Profiling Interactive Analysis (GEPIA) (http://gepia.cancer-pku.cn/index.html), an online database was used to confirm the correlation between PBX3 expression and clinicopathologic information in glioma.GEPIA uses a standard processing pipeline that analyzes 9,736 tumors from the RNA sequencing expression data and 8,587 normal samples from projects known as the Genotype-Tissue Expression project (GTEx) and The Cancer Genome Atlas (TCGA).Box Plots using disease state (Tumor or Normal) as variable was graphed to calculate differential expression of PBX3.Meanwhile, Survival curve of differential PBX3 expression were analyzed to find the correlation of the gene expression with glioma patients' prognosis.

Expression level of PBX3 in normal tissues by GTEx data
GTEx were downloaded from University California santa cruz (UCSC) Xena project (http:// xena.ucsc.edu/).The data include TOIL RNA-Seq by Expectation Maximization (RSEM) TPM (n = 7,862) and GTEx phenotype (n = 9,783).TCGA data were downloaded from UCSC Xena project.The data include TOIL RSEM TPM (n = 10,535) and Curated clinical data (n = 12,591).The Ensembl number of the gene in RNAseq data was transformed into the Official Symbol of the gene by R 4.2.0.In addition, the GTEx and TCGA data was used to analyze the differential expression of PBX3 in normal brain tissue and glioma tissue through ggpubr package.

Clinical correlation analyses between PBX3 expression and glioma patients from TCGA database
The relevant clinical information of 1111 glioma patients (LGG+GBM) (Low-Grade Glioma and Glioblastoma), was download from TCGA database by RTCGA.clinical package.Meanwhile, we obtained 1131 clinical information of glioma patients from Jianfang's study [12].Then, after combining the two sets of data, a total of 1131 patients with gliomas were included in this study.The expression value of PBX3 in glioma patients with clinical information were integrated to obtain 693 patients via perl script (Table 1), and the expression difference of PBX3 in discrete variables of clinical information was analyzed by using R 4.2.0.Finally, among these 693 patients, we eliminated some patients with incomplete clinical information and obtained 452 patients for COX analysis.

Clinical correlation analyses between PBX3 expression and glioma patients from CGGA database
The relevant clinical information of 1019 glioma patients (LGG+GBM) were download from Chinese Glioma Genome Atlas (CGGA) database.The expression value of PBX3 in glioma patients with clinical information were integrated to obtain 929 patients via perl script (Table 2), and the expression difference of PBX3 in discrete variables of clinical information was analyzed by using R 4.2.0.Finally, among these 929 patients, we eliminated some patients with incomplete clinical information and obtained 499 patients for COX analysis.

Expression of PBX3 in glioma cell lines
The gene expression data of 57 glioma cell lines were downloaded from Cancer Cell Line Encyclopedia (CCLE)databases(https://sites.broadinstitute.org/ccle).The expression difference of PBX3 in glioma cells was analyzed by using R 4.2.0 ggplot2 packages.

Linear regression analysis
Based on the data from TCGA and CGGA databases, we analyzed the correlation between PBX3 expression level and survival outcome and survival time of LGG by using linear regression equation.The analysis was performed with the statistical software packages R and EmpowerStats (http://www.empowerstats.com,X&Y Solutions, Inc., Boston, MA).

GSEA enrichment analysis
Gene Set Enrichment Analysis (GSEA) is a computational method to determine whether a group of genes are differentially expressed in two biological states.In this study, GSEA was used to generate an ordered list of all genes related to PBX3 expression, and then to identify the differential expression of signal pathways between gliomas with high PBX3 expression and low PBX3 expression in TCGA database.Gene set permutations were performed 1000 times for each analysis.The expression level of PBX3 was used as a phenotype label.The nominal P value and normalized enrichment score (NES) were used to sort the pathways enriched in each phenotype.

Analysis of gene co-expression of PBX3 mRNA
Screening of genes co-expressed with PBX3 mRNA by "limma" package in R4.2.0 software.The threshold of co-expression was correlation coefficient > 0.5 and P < 0.001.Meanwhile, draw a single gene correlation map.The first five genes with positive and negative correlation with PBX3mRNA expression were screened, and the correlation circle diagram was drawn.

The expression of PBX3 in glioma tissues, glioma cells and normal tissues
In order to determine the expression level of PBX3 in normal human brain tissues, GTEx data analysis was conducted.As shown in Fig 1A -1C, the expression of PBX3 was observably upregulated in full-grade gliomas, LGG and GBM than that in normal tissues.According to GEPIA database, we found that the expression of PBX3 was observably upregulated in LGG and GBM than that in normal control (Fig 1D).The expression of PBX3 was observably upregulated in GBM than that in LGG according to TCGA and CGGA databases (Fig 1E and 1F).For further study, we found that the expression of PBX3 was related to histological types.There were GBM, astrocytoma, oligoastrocytoma and oligodendroglioma histological types from TCGA databases, the expression of PBX3 was highest in GBM, followed by astrocytoma, and the expression of PBX3 in oligoastrocytoma and oligodendroglioma was significantly lower than that of GBM and astrocytoma (Fig 1G).In addition, there were GBM(glioblastoma), A (astrocytoma),O(oligodendroglioma),OA(oligo-astrocytoma), AOA(anaplastic oligo-astrocytoma),AA(anaplastic astrocytoma) histological types from CGGA databases.The expression of PBX3 was highest in GBM, followed by AA, and the expression of PBX3 in O and OA was significantly lower than that other five histological types, which was consistent with TCGA databases (Fig 1H ).By analyzing the expression level of PBX3 in 57 glioma cell lines from CCLE database, it was found that the expression of PBX3 was increased in glioma cell lines (Fig 1I).PBX3 might be an important molecular mechanism for the occurrence of gliomas.

Relationship of PBX3 expression with clinical parameters of patients with glioma
Based on TCGA and CGGA databases, the expression of PBX3 increased with the increase of glioma histologic-grade (Fig 2A and 2B).The expression of PBX3 in recurrent glioma patients was significantly higher than that in primary glioma patients, and the expression of PBX3 was increased with the increase of recurrence times (Fig 2C and 2D).The expression of PBX3 was higher in IDH wildtype glioma patients than Isocitrate Dehydrogenase (IDH) mutant (Fig 2E and 2F), and the prognosis of IDH wildtype is worse than IDH mutant type.The expression level of PBX3 in glioma patients age�55 years was significantly higher than that in patients age <55 years (Fig 2G and 2H).In addition, the expression level of PBX3 in glioma patients with tumor was higher than that in patients without tumor and the expression level of PBX3 in glioma patients ((Fig 2I) and karnofsky performance score ≦60 was significantly higher than that in patients karnofsky performance score >60 according to TCGA(Fig 2J).Studies have confirmed that glioma patients with 1p19q-codel have a better prognosis than that non-codeled.At the same time, it is found that the expression level of PBX3 in glioma patients with 1p19qcodel deletion is significantly lower than that non-codel according to CGGA database (Fig 2K).This shows that PBX3 is associated with tumor progression and has a certain prognostic value in patients with glioma.

Relationship between PBX3 expression and overall survival time of patients with glioma
In TCGA and CGGA database, the OS (Overall survival time) of PBX3 high expression group was shorter than that of low expression group in full-grade gliomas (Fig 3A and 3B).And, in TCGA, CGGA and GEPIA database, the OS of PBX3 high expression group was shorter than that of low expression group in LGG patients (Fig 3C -3E).In addition, the OS of GBM in low expression of PBX3 was better than that in high expression of PBX3 according to CGGA, but there was no statistically significant from GEPIA and TGGA analysis (Fig 3F -3H).This shows that PBX3 is more valuable in predicting the OS of LGG patients than that of GBM.

Relationship between PBX3 expression and PFI (Progression-free interval) of patients with glioma
In TCGA database, the PFI (Progression-free interval) of PBX3 high expression group was shorter than that of low expression group in full-grade gliomas (Fig 4A).And, in TCGA and GEPIA database, the PFI of PBX3 high expression group was shorter than that of low expression group in LGG patients (Fig 4B and 4C).In addition, the PFI of GBM in low expression of PBX3 was better than that in high expression of PBX3 according to GEPIA, but there was no statistically significant from TGGA analysis (Fig 4D and 4E).This shows that PBX3 is more valuable in predicting the PFI of LGG patients than that of GBM.

ROC curve survival analysis of PBX3
According to TCGA database, ROC curve analysis showed that the AUC of PBX3 mRNA predicting 1-year, 3-year and 5-year overall survival rate of LGG was 0.811, 0.791 and 0.698(Fig 5A).And, in CGGA database, ROC curve analysis showed that the AUC of PBX3 mRNA predicting 1-year, 3-year and 5-year overall survival rate of LGG was 0.695,0.715and 0.726(Fig 5B).In addition, ROC curve analysis also showed that the AUC of PBX3 mRNA predicting 1-year, 3-year and 5-year progression-free survival rate of LGG was 0.732,0.693and 0.658(Fig 5C).This result from TCGA and CGGA database confirmed that the expression of PBX3 was a predictor of overall survival rate and progression-free survival rate of LGG.

Univariate and multivariate cox regression analysis of OS prognostic factors in LGG
To further confirm our conjecture, univariate cox analysis was employed in TCGA database.The analysis found that PBX3 is a highly risk factor (HR = 1.15) for predicting the OS of LGG patients.Besides, age, histological type, caner status, histologic grade were all high-risk factors (Table 3).Then multivariate analysis was performed, and it was unearthed that among these factors, PBX3(HR = 1.14) remained independently related to OS, suggesting that PBX3 might be an independent OS prognostic factor for LGG patients (Table 3).And, we further confirmed above result by univariate and multivariate analysis according to CGGA.Univariate analysis found that PBX3 is a highly risk factor (HR = 2.43) for predicting the OS of LGG patients.Besides, age, histological type, caner status, histologic grade were all high-risk factors.Besides, PRS-type (primary, recurrent), IDH-mutation (mutant, wildtype), histological type, WHO grade, 1p19q-codeletion status and age (Table 4).And subsequent multivariate analysis shows that PBX3 (HR = 1.95) remained independently related to OS, which also suggested that PBX3 might be an independent OS prognostic factor for LGG patients (Table 4).

Univariate and multivariate cox regression analysis of PFI prognostic factors in LGG
According to TCGA database, univariate analysis found that PBX3 is a highly risk factor (HR = 1.15) for predicting the PFI of LGG patients.Besides, age, histological type, caner status, histologic grade were all high-risk factors (Table 5).And subsequent multivariate analysis shows that PBX3 (HR = 1.12) remained independently related to PFI, which suggested that PBX3 might be an independent PFI prognostic factor for LGG patients (Table 5).

The correlation between PBX3 expression and the OS of LGG by linear regression model
According to TCGA database, the expression of PBX3 was positively associated with incident death events in LGG patients.The OR for the death of LGG patients was increased 3.1 times in the high PBX3 expression group compared with those in the low group (Table 6).Meanwhile, the expression of PBX3 was negatively correlated with the overall survival time of LGG patients (Table 6).The overall survival time in high group of PBX3 expression was decreased by 1.1years compared with those in the low group (Table 6).In addition, we further analyzed the correlation between PBX3 expression and the OS from CGGA database.the expression of PBX3 was positively associated with incident death events in LGG patients.The OR for the death events of LGG patients was increased 0.8 and 3.6 times in the middle and high PBX3 expression group compared with those in the low group (Table 5).Meanwhile, the expression of PBX3 was negatively correlated with the OS of LGG patients (Table 7).The overall survival time of LGG patients in middle and high group of PBX3 expression was decreased by 2.0 years compared with those in the low group (Table 7).

The correlation between PBX3 expression and the PFI of LGG by linear regression model
According to TCGA database, the expression of PBX3 was positively associated with disease progression events in LGG patients.The OR for disease progression events 2.6 times in the high PBX3 expression group compared with those in the low group (Table 8).Meanwhile, the expression of PBX3 was negatively correlated with the progression free interval time of LGG patients (Table 6).The progression free interval time in high group of PBX3 expression was decreased by 1 years compared with those in the low group (Table 8).

Analysis of gene Co-expression of PBX3
Gene correlation analysis by limma package showed that the first five genes with positive correlation with PBX3 mRNA expression include GNG12, WEE1, PDPN, FNDC3B and TGIF1,

Discussion
PBX3 plays an important role in embryonic development, organogenesis and steroid synthesis.
There are many similar biological processes in embryonic development and tumorigenesis, such as growth, differentiation and tissues with similar cell groups [13,14].Recent studies have shown that PBX3 is highly expressed in a variety of malignant tumors and is related to tumor cell proliferation, invasion, metastasis and resistance to radiotherapy and chemotherapy [15].However, the effect of PBX3 expression on the prognosis of gliomas and the molecular mechanism of its involvement in the occurrence and development of gliomas need to be further studied.The purpose of this study is to explore the possible prognostic value of PBX3 in gliomas and the underlying role of PBX3 in gliomas.Related studies have reported that PBX3 is highly expressed in glioma tissues [6,11,16].In our study, we found that PBX3 was highly expressed in glioma tissued by analyzing data from TCGA and GTEx databases.This finding is consistent with the GEPIA database analyses.The result reveals that PBX3 might acts as a crucial regulatory role in glioma occurrence and progression.The expression of PBX3 increased with the increase of pathological grade of glioma, this unearthed that PBX3 might sever as a possible molecular marker for predicting the degree of glioma malignancy.The expression of PBX3 in recurrent gliomas is significantly higher than that in primary gliomas, indicating that PBX3 can be used as a tumor marker to predict glioma recurrence.The expression of PBX3 is related to other clinicopathological features such as idh-mutation,1p19q-codeletion status, age and karnofsky performance score.These results suggest that PBX3 is associated with tumor progression and is of great value in the diagnosis and prognosis of gliomas.
In addition, we further found that PBX3 is more valuable in predicting the OS and PFI of LGG patients than that of GBM.Univariate and multivariate Cox regression analysis showed that PBX3 was a high-risk factor and could be used as an independent prognostic index for patients with LGG glioma.ROC curve analysis indicated that PBX3 was a predictor of overall survival rate and progression-free survival rate of LGG. the higher the expression of PBX3, the higher the risk of death and disease progression of LGG patients by linear regression model analysis.These results suggest that the expression of PBX3 can be used as a risk indicator to assess death and disease progression of LGG patients.
To further investigate the functions of PBX3 in glioma, we performed GSEA using TCGA data, GSEA showed that regulation of NOS1 pathway, CK1 pathway, HDAC pathway, Phosphatidylinositol system, Neurotransmitter receptor complex were differentially concentrated in PBX3 high expression phenotype.CK1 has been reported involved in the regulation of Wnt pathways, cell proliferation, apoptosis and autophagy [17], dysregulated expression CK1 isoforms have previously been linked to tumorigenesis [18].Histone deacetylases (HDAC) deacetylate the histone and induce gene repression thereby leading to cancer [19,20].PI3K signaling pathway has been as an important signaling pathway involved in the regulation of various aspects of cancers [21][22][23].A study by Kotake et al. [24] has reported that low levels of NO formed by NOS1, triggers cell proliferation primarily via the soluble guanylate cyclasecyclic guanosine monophosphate (sGC-cGMP) dependent mechanism, and NOS1 was also observed to be aberrantly expressed in glioma [25].Diverse neurotransmitters not only maintain normal brain functions but also influence glioma progression [26].To some extent, these pathways also indicate that PBX3 is closely related to glioma.The analysis of genetic co-expression of PBX3 mRNA showed that PBX3 mRNA was positively correlated with the expression of GNG12, WEE1, PDPN, FNDC3B and TGIF1 genes, and negatively correlated with the expression of ATP6V1G2, AC120036.4,PDE2A, TEF and DNAJC12 genes.It is reported that GNG12 [27], WEE1 [28], TEF [29] and DNAJC12 [30] participate in cell cycle regulation, cell proliferation and cell apoptosis.PDPN [31], FNDC3B [32], TGIF1 [33] and DNAJC12 [30] are involved in cell migration and metastasis.In addition, there are other studies have confirmed that GNG12 participate in inflammation signals, related to tumor immunity [26].PDE2A is involved in glioma stem cells (GSCs) stemness and tumorigenesis [34].Therefore, we speculate that the high expression of PBX3 in glioma cells may be participated in a variety of biological processes, including cell proliferation, cell apoptosis, cell migration and metastasis, tumor immunity and stem cells stemness through the above signal pathways and co-expressed molecules.
Our study demonstrated that the expression level of PBX3 is related to the prognosis of glioma patients, and the high expression of PBX3 is associated with poor prognosis and is a highrisk factor.Compared with GBM patients, PBX3 is more valuable in predicting the survival time of LGG patients.PBX3 can be used as an independent prognostic index for patients with LGG glioma.We further explored the signal pathways that PBX3 may be involved in and the genes co-expressed with PBX3.The study explored the relationship between PBX3 expression and glioma via bioinformatics, and needed further study with function and mechanism of PBX3 by a range of vivo and vitro experiments.

Fig 1 .
Fig 1.Expression level of PBX3 in normal brain, glioma tissues and glioma cells by analyzing data from TCGA, GTEx, GEPIA and CCLE databases.(A-C) Expression level of PBX3 in full-grade gliomas, LGG and GBM in comparison with the normal brain tissues according to TCGA and GTEx.(D) Expression level of PBX3 in LGG and GBM in comparison with the normal brain tissues according to GEPIA.(E) and (F) The expression level of PBX3 between LGG and GBM according to TCGA and CGGA databases.(G) and (H) The expression level of PBX3 in histological types of gliomas according to TCGA and CGGA databases.(I) The expression level of PBX3 in glioma cell lines according to CCLE databases.https://doi.org/10.1371/journal.pone.0293647.g001