Expression and significance of SOX B1 genes in glioblastoma multiforme patients

Abstract The overall survival of glioblastoma multiforme (GBM) patients remains poor. To improve patient outcomes, effective diagnostic and prognostic biomarkers for GBM are needed. In this study, we first applied bioinformatic analyses to identify biomarkers for GBM, focusing on SOX (sex‐determining region on the Y chromosome (SRY)‐related high mobility group (HMG) box) B1 family members. The ONCOMINE, GEPIA, LinkedOmics and CCLE databases were used to assess mRNA expression levels of the SOX B1 family members in different cancers and normal tissue. Further bioinformatic analysis was performed using the ONCOMINE database in combination with the LinkedOmics data set to identify the prognostic value of SOX B1 family members for GBM. We found mRNA expression levels of all tested SOX B1 genes were significantly increased in GBM. In the LinkedOmics database, increased expression of SOX3 indicated a better overall survival. In GEPIA databases, increased expression of all SOX B1 family members suggested an improved overall survival, but none of them were statistically different. Then, Transwell assays and wound healing were employed to evaluate the motility and invasive captivity of U251 cells when silencing SOX2 and SOX3. We found exogenous inhibition of SOX2 appeared to reduce the migration and invasion of U251 cells in vitro. Collectively, our research suggested that SOX2 might serve as a cancer‐promoting gene to identify high‐risk GBM patients, and SOX3 had the potential to be a prognostic biomarker for GBM patients.

the use of surgery, radiotherapy, chemotherapy and immunotherapy. 5 To date, no specific biomarkers that offer improvements to GBM patient survival have been found. Therefore, it is necessary to identify strategies and targets for early diagnosis and prognosis of GBM.
SOX (sex-determining region on the Y chromosome-related high mobility group box) B1 genes consist of SOX1, SOX2 and SOX3, sharing a high degree of sequence similarity, both within and outside the high mobility group box. 6 Previous studies have discovered that SOX B1 members are widely expressed in neural tissue, embryonic stem cells and testes. 7 They are involved in various physiological processes, such as maintaining embryonic stem cell function, the occurrence of neural tissues, controlling male sex determination and maintaining neural stem cells. 8,9 In addition, dysregulated expression of SOX B1 family members affect the occurrence and prognosis of various cancers. Accumulating evidence suggests that SOX B1 genes have both antiproliferative and prosurvival effects, depending on cancer. On the one hand, SOX B1 members have been shown to be tumour suppressors in nasopharyngeal carcinoma, 10 ovarian cancer 11 and metrocarcinoma. 12,13 On the other hand, SOX B1 genes have been shown to play an oncogenic role in breast cancer, 14,15 squamous cell carcinoma, 16 hepatocellular carcinoma, 17 osteosarcoma 18,19 and brain cancer. 20,21 Consequently, a greater understanding of their role in various cancers is needed.
In recent years, proteomics, transcriptomics and high-throughput sequencing technologies have developed rapidly, generating a large amount of genomics data in the field of cancer research. Although the relationship between the SOX B1 genes and many cancers has been partly reported, no study has fully summarized their role via bioinformatics. Based on multiple published databases, we analysed the expression of SOX B1 family members to determine their expression levels in various cancers, with focus on their diagnostic and prognostic value in GBM.

| ONCOMINE database
ONCOMINE is the largest and most comprehensive gene chip database and data extraction platform (https://www.oncom ine.org/ resou rce/login.html), containing 715 databases and 86,733 samples that can be used to compare differences between cancer and normal tissue. 22 It was used to compare the transcription levels of SOX B1 members in different cancers. The mRNA expression levels of the SOX B1 genes in clinical cancer specimens were compared with levels in normal specimens (controls), using a Student's t-test to generate a p-value. The p-value cut-off was defined as 0.05.

| GEPIA database
GEPIA database (Gene Expression Profiling Interactive Analysis) (http://gepia.cance r-pku.cn/) is an online database developed by Peking University that performs a dynamic analysis of gene expression spectrum data. The expression of genes in different tumours was analysed in combination with TCGA (The Cancer Genome Atlas) and GTEx databases (Genotype-Tissue Expression). SOX B1 gene survival analysis was carried out by the method of total survival rate and disease-free survival rate.

| LinkedOmics data set
LinkedOmics data set (http://linke domics.org/login.php) is an online tool for analysing TCGA databases. This data set was used to examine the relationship between mRNA levels of SOX B1 genes and the overall survival of GBM patients.

| Cancer Cell Line Encyclopaedia database
The Cancer Cell Line Encyclopaedia (CCLE) database (www.broad insti tute.org/ccle/home) is a project to develop the integrated computational analysis of human cancer models and molecular spectrum of nearly 1,000 human cancer cell lines used in global drug research and development. 23 The SOX B1 members' expression is affirmed via the CCLE database.  The old culture medium was sucked and washed with PBS twice.

| Lipofection transfection
Finally, the mixture was placed into the 6-well plates, gently blended and then placed into a 37 °C 5% CO 2 incubator for further culture.
After 4 h, the mixture was removed and replaced with medium containing 10% serum but no antibiotics.

| Wound-healing migration assay
Before cell seeding, a vertical line and 5 horizontal lines were drawn

| Transwell invasion assays
Twenty-four hours after transfection, 1 × 10 5 cells in serum-free DMEM were plated in the upper chamber (the total volume was After washing with PBS twice, cells were mounted under microscope observation.

| Quantitative real-time polymerase chain reaction
For RNA extraction, the Trizol reagent (Life Technologies) was used in accordance with the manufacturer's instructions. Real-time quantitative PCR (RT-qPCR) was performed with SYBR Green Real-time PCR Master Mixes (Thermo Fisher Scientific, USA), and β-actin served as an internal control. Our results were analysed using 2 −ΔΔC T method. All primer sequences are listed in detail in Table S1.

| Statistical analysis
Statistical analyses were carried out using Graph Pad Prism 8. The data are presented as the mean ±SD. Statistical testing was performed using an unpaired t-test for two group comparisons, One-way ANOVA with a post hoc test was applied for multigroup comparisons. p < 0.05 was considered to indicate statistical significance.

| Assessing expression levels of SOX B1 genes in different cancers using the ONCOMINE database
We compared the expression levels (mRNA) of SOX B1 family members (SOX1, SOX2 and SOX3) in various cancers and normal tissue using the ONCOMINE database. SOX B1 genes were found to be differentially expressed in different cancers ( Figure 1). Collectively, SOX B1 genes have been most studied in the brain and central nervous system. In the ONCOMINE database, there are 12 databases that record the mRNA levels of the SOX B1 family in the brain and central nervous system. Six databases indicated that SOX2 was overexpressed in the brain and the central nervous system, making it the most studied SOX B1 family gene here.

| Assessing expression levels of the SOX B1 genes in different cancers using the GEPIA database
We compared the mRNA levels of SOX B1 family members in pancancer using the GEPIA database. The results suggested that SOX B1 gene expression was higher in GBM and low-grade glioma (LGG) patients than in normal tissue samples. Because LGG patients have always been characterized by a relatively good prognosis, 24,25 we focused on the expression levels of SOX B1 genes in GBM. The results showed that GBM patients had higher SOX2 expression levels than healthy patients (p < 0.01) (Figure 2A-2E).

| Assessing expression levels of the SOX B1 genes in different cancers using the CCLE database
We compared the expression levels of SOX B1 genes using the CCLE database. Consistent with the previous findings (above), the mRNA levels of SOX2 and SOX3 were higher in glioma cell lines compared to other cancer types ( Figure 3A-3C).

| Assessing changes in SOX B1 family member expression in GBM patients
We compared the mRNA expression levels of SOX B1 genes between GBM and normal tissue using the ONCOMINE database. In all statistically significant data sets, all SOX B1 genes were upregulated, to varying degrees, compared to normal tissue (

| Prognostic analysis of SOX B1 subgroups
We conducted a prognostic analysis for SOX B1 genes by using LinkedOmics and GEPIA databases. The data indicated that decreased SOX1 expression levels predicted better overall survival in GBM, while decreased SOX2 indicated poor overall survival, but neither result was statistically significant. To our surprise, increased SOX3 showed better overall survival (Logrank p = 0.0432) ( Figure 4A). The survival rate of high SOX3 patients is much higher than low SOX3 patients (HR = 0.825). In the GEPIA database, increased SOX B1 expression indicated a better overall survival, but the results were not statistically significant ( Figure 4B).
We analysed the correlation among SOX1, SOX2, and SOX3 via the LinkedOmics database. We found there was no significant correction between SOX1 and SOX2 or also SOX1 and SOX3. However, SOX2 and SOX3 expression was positively correlated. (R = 0.3001, p < 0.05) ( Figure 5D).

| Effects of SOX2 and SOX3 downregulation on migration ability and invasion abilities of U251 cells
The silencing effect was confirmed by western blotting ( Figure S1 and Figure 6A). The downregulation of SOX2 and SOX3 to influence tumour cell migration was evaluated using wound-healing migration assay. Figure 6B and C showed representative microscopy images at 0h and 48h, together with the relative quantitative analysis

| DISCUSS ION
The prognosis of gliomas is closely related to its histological type and tumour grade. Therefore, early and precise diagnosis of GBM is a in pan-cancer through ONCOMINE and GEPIA data sets, SOX2 was overexpressed in GBM and LGG and SOX3 had elevated expression in LGG. By using the CCLE data set, we further confirmed that SOX B1 genes were overexpressed in gliomas. Since the prognosis of LGG patients is relatively good, we focused on analysing the relationship between SOX B1 genes and GBM. In 2007, Schmitz M et al. found that SOX2 expression levels (mRNA and protein) were increased in human brain tumour biopsies. 33 SOX2 plays a vital role in the carcinogenesis and maintenance of GBM stem cells. 34,35 Subsequently, an increasing number of studies have shown that overexpression of SOX2 is closely related to tumour invasiveness and poor prognosis. [36][37][38] Silenced SOX2 can inhibit proliferation and induce loss of tumorigenicity in GBM tumour-initiating cells in immunosuppressant mice, 39 and knockdown studies of SOX2 reduced cellular proliferation and colony formation in a GBM cell line. 40,41 Our results showed SOX2 silencing significantly decreased proliferation of GBM cells, so SOX2 overexpression may contribute to tumour progression of GBM.
Furthermore, we analysed the prognostic value of SOX B1 mRNA levels in GBM patients by using LinkedOmics and GEPIA databases.
To our surprise, increased expression of SOX2 had no influence on the prognosis of GBM patients. And, overexpressed SOX3 indicated better overall survival in GBM patients (Logrank p = 0.0432, HR high = 0.825), suggesting that SOX3 is an antioncogene. In contrast, previous studies have shown that overexpression of SOX3 is associated with poor overall survival in gastric cancer, 42 breast cancer 43 and adult de novo acute myeloid leukaemia. 44 Lu S et al. 45 indicated that overexpression of SOX3 predicts a poor outcome in GBM patients; and, Sa JK et al. 46 found SOX3 is associated with tumour invasiveness, malignancy and poor prognosis in GBM patients. Vicentic et al. 47 found SOX3 can accelerate the malignant behaviour of GBM cells. Hence, the definite role of SOX3 in GBM was unclear. Finally, we conducted gene co-expression analysis using the ONCOMINE database. The results indicated that SOX B1 members were closely related to many different genes. However, SOX1, SOX2 and SOX3 did not share any co-expressed genes. By using the LinkedOmics database, we found SOX2 and SOX3 expression was positively related.

| CON CLUS ION
In conclusion, we investigated the expression and prognostic value of SOX B1 genes in GBM. We concluded that the expression of SOX1, SOX2 and SOX3 in GBM might result in tumorigenesis.
Overexpressed SOX2 could serve as a biomarker to identify highrisk GBM patients. Moreover, SOX2 may enhance the migratory and invasive capacity of glioma cells. Furthermore, SOX3 may serve as a prognostic biomarker set for GBM patients. Hence, SOX2 may serve as a potential therapeutic target in GBM patients, and more experiments are needed to clearly identify the specific mechanism of GBM formation pathway involved in SOX2 and SOX3.

ACK N OWLED G EM ENTS
We thank the experts who carefully read our manuscript and the editors who gave our article a chance to be published.

CO N FLI C T O F I NTE R E S T
The authors declare no conflict of interest.

DATA AVA I L A B I L I T Y S TAT E M E N T
All data, models and code generated or used during the study appear in the submitted article.