Identification and validation of SOCS1/2/3/4 as potential prognostic biomarkers and correlate with immune infiltration in glioblastoma

Abstract Suppressor of cytokine signalling (SOCS) 1/2/3/4 are involved in the occurrence and progression of multiple malignancies; however, their prognostic and developmental value in patients with glioblastoma (GBM) remains unclear. The present study used TCGA, ONCOMINE, SangerBox3.0, UALCAN, TIMER2.0, GENEMANIA, TISDB, The Human Protein Atlas (HPA) and other databases to analyse the expression profile, clinical value and prognosis of SOCS1/2/3/4 in GBM, and to explore the potential development mechanism of action of SOCS1/2/3/4 in GBM. The majority of analyses showed that SOCS1/2/3/4 transcription and translation levels in GBM tissues were significantly higher than those in normal tissues. qRT‐PCR, western blotting (WB) and immunohistochemical staining were used to verify that SOCS3 was expressed at higher mRNA and protein levels in GBM than in normal tissues or cells. High SOCS1/2/3/4 mRNA expression was associated with poor prognosis in patients with GBM, especially SOCS3. SOCS1/2/3/4 were highly contraindicated, which had few mutations, and were not associated with clinical prognosis. Furthermore, SOCS1/2/3/4 were associated with the infiltration of specific immune cell types. In addition, SOCS3 may affect the prognosis of patients with GBM through JAK/STAT signalling pathway. Analysis of the GBM‐specific protein interaction (PPI) network showed that SOCS1/2/3/4 were involved in multiple potential carcinogenic mechanisms of GBM. In addition, colony formation, Transwell, wound healing and western blotting assays revealed that inhibition of SOCS3 decreased the proliferation, migration and invasion of GBM cells. In conclusion, the present study elucidated the expression profile and prognostic value of SOCS1/2/3/4 in GBM, which may provide potential prognostic biomarkers and therapeutic targets for GBM, especially SOCS3.


| INTRODUC TI ON
Glioblastoma (GBM) is the most common, aggressive primary malignant brain cancer type in humans worldwide. 1 GBM, whose cause is unknown, is common among Caucasians and Asians, with the worst survival and highest morbidity among Caucasians. 2 GBM can occur at any age, and the incidence increases steadily with age. 3 Currently, the overall survival (OS) time of patients with GBM can be extended via radiotherapy, chemotherapy, surgical treatment and immunotherapy, but it is not possible to reduce its high recurrence rate after treatment. Previous clinical trials have shown that the 5year survival rate of GBM is 4%-5%, and the 2-year survival rate is 26%-33%. 4 Hence, it is necessary to further explore the carcinogenic mechanism of GBM and provide better prognostic assessment strategies.
Tumour RNA sequencing (RNA-seq) data are increasingly being used to identify gene signatures and clinical prognostic factors, including gender, age, GBM tumour grade and Karnofsky score. 5 Although previous studies have explored the prognostic factors of glioma, the prognosis of patients with glioma remains poor. Current highthroughput sequencing technology provides rapid analysis of biomarkers and in-depth exploration of potential mechanisms of disease. 6 It is important to find biomarkers that affect the survival and prognosis of glioma patients as soon as possible. The present study explored the influence of four genes in the suppressor of cytokine signalling (SOCS) family on GBM and experimentally evaluated on the expression and functional influence of SOCS3 in the SOCS family in GBM.
The SOCS family consists of eight members, whose dominant role is to inhibit cytokine signal transduction. 7 SOCS1 and SOCS3 can inhibit the signal transduction of various cytokines such as interleukin-6 (IL6), leukaemia inhibitor factor (LIF), OSM, INF gamma and growth hormone (GH), and play regulatory roles in the activation of various immune responses and the pathogenesis of tumours in vivo. 8 SOCS2 is induced by a variety of cytokines that activate STAT5, including GH, IL-6 and LIF, and it is involved in the ubiquitination of target proteins, including GHR and a variety of signalling proteins. 9 SOCS4 is an important regulator of antiviral immunity. 10 As one of the negative feedback loops of the JAK/STAT signalling pathway, SOCS4 can decrease the STAT3 signalling of EGFR by increasing receptor degradation. 11 SOCS5, SOCS6, SOCS7 and CIS play protumour or antitumour roles in a variety of cancers, and may affect the progression of tumours via different mechanisms. For example, SOCS5 can inhibit the migration and invasion of hepatocellular carcinoma cells in vitro by activating PI3K/Akt/mTOR-mediated autophagy. 12 It is worth mentioning that the expression alterations of SOCS1/2/3/4 negatively regulate the JAK/STAT signalling pathway in many tumour cells, so SOCS1/2/3/4 is the main research content of this article. In the current study, SOCS1/2/3/4 were regarded as the key genes affecting the survival and prognosis of patients with GBM, while SOCS3, as a negative regulator, could regulate the JAK/STAT3 signalling pathway and inhibit tumour cell proliferation and tumour development. 13 Therefore, SOCS3 was included in the current study for further experimental verification.

| Analysis of UALCAN
TCGA gene expression data were analysed by the University of ALabama at Birmingham CANcer (UALCAN) (http://ualcan.path.uab. edu/) 17 database to observe SOCS1/2/3/4 and their hub genes' expression in GBM. In addition, the association between SOCS1/2/3/4 and GBM clinical and pathological features (such as race, IDH status or OS, etc.) and their potential prognostic significance were observed.

| Analysis of biological function
Gene expression data of GBM in HTSeq-FPKM were acquired from TCGA data set to explore 543 patients with GBM. Pearson's correlation coefficient (|r| > 0.4 and p < 0.001) to screen the coexpressed genes of SOCS1/2/3/4. Moreover, Gene Ontology (GO) and Kyoto Encyclopaedia of Genes and Genome (KEGG) analysis 19 of coexpressed genes were evaluated by employing the R 'cluster profiler' software package to search the possible biological functions of SOCS1/2/3/4. Coexpressed genes of the SOCS1/2/3/4 proteinprotein interaction (PPI) network was built by employing the GeneMANIA (http://genem ania.org/) database. 20 The PPI network was put into Cytoscape 3.6.1, 21 and the cytoHubba plug-in was used as the screening criteria for hub genes. In total, 10 genes with the highest correlation were considered to be hub genes, and relevant analysis was subsequently conducted.

| Linked Omics database analysis
Linked Omics Database (http://www.linke domics.org/login.php) 23 contains multiomics data and clinical data for 32 cancer types, as well as data for a total of 11,158 patients from the TCGA project. It is also a multiomics database that integrates proteomics data from mass spectrometry (MS) for selected TCGA tumour sample. The differentially expressed genes (DEGs) related to SOCS3 were screened from TCGA-GBM cohort by the LinkFinder module, and the Pearson correlation coefficient was employed to obtain the results. The results were shown as volcano plots and heat maps.

| TISDB analysis
The association between the immune system and GBM was explored via the TISIDB database (http://cis.hku.hk/TISID B/), 24 and the correlation between SOCS1/2/3/4 expression in GBM and immune infiltration was evaluated by Spearman's correlation analysis.

| Clinical tissue specimens
The tissues of 23 patients with low-grade glioma (LGG) and GBM (including 12 patients with LGG and 11 patients with GBM) used for reverse transcription-quantitative PCR (RT-qPCR) and immunohistochemical staining were obtained from the Fifth

| RT-qPCR assay
Total RNA was extracted from patients' tissue employing TRIzol®

| Western blotting
Cells were washed in PBS, detached with a cell scraper and centrifuged for 10 min at 12,000 × g at 4°C. Cell lysates were boiled for 15 min at 100°C. Total protein (15-20 μg)

| Colony formation assay
Transfected cells were digested with trypsin and counted with cell technology plates. Next, 2 mL medium containing 10% FBS was placed into a 6-well plate, and 500 cells were added to the 6-well plate. The cells were collected after 10 days of culture, fixed in paraformaldehyde for 30 min, stained with crystal violet for 30 min, washed with PBS, observed and counted.

| Transwell assay
The Transwell chamber and the Matrigel were purchased from Corning Inc and Biozellen, respectively. For migration experiments, the upper layer of each chamber was inoculated with 30,000 cells, and 100 μL serum-free medium and 800 μL 10% FBS medium were added to the lower layer of the chamber. For the invasion assay, Matrigel was added to the upper layer of the chamber, and the subsequent steps were performed as aforementioned.

| Immunohistochemical staining analysis
Tissue samples of 12 patients with LGG and 11 patients with GBM were fixed with formaldehyde, paraffin embedded, sectioned and incubated with anti-SOCS3 antibodies overnight at 4°C, followed by incubation with a sheep antirabbit secondary antibody for 20 min at room temperature. Sample was then washed with PBS and incubated with a streptomycin working solution at room temperature for 15 min. Next, DAB chromogenic solution was used, and areas with a strong immune response were selected and observed at ×400 magnification.

| Statistical analysis
GraphPad Prism 9 (GraphPad Software, Inc.) was employed for creating graphs, while SPSS 26.0 (IBM Corp) was employed for statistical analysis. Normally distributed data were expressed as the mean ± standard deviation. Student's t-test was employed to compare SOCS1/2/3/4 expression in cancer and normal tissues, as well as the proliferation, invasion and migration of GBM cells before and after knocking down SOCS3. Kaplan-Meier curves were employed to analyse patients' survival according to their SOCS1/2/3/4 expression levels. Univariate Cox regression analysis was employed to calculate hazard ratio (HR) and 95% confidence interval (CI) in survival analysis. All R packages used were R software version V4.0.3.

| SOCS1/2/3/4 expression is increased in GBM
The Oncomine database was employed to compare SOCS1/2/3/4 mRNA expression between brain and CNS cancer and normal tissues. The results revealed that SOCS1, SOCS2, SOCS3 and SOCS4 mRNA levels were upregulated in patients with brain and CNS cancer ( Figure 1A). Indepth analysis revealed that SOCS1 was overexpressed in patients with GBM, with a fold-change of 2.133 and a p = 5.37 × 10 −9 . In the Sun database, the SOCS1 expression level was high in GBM with a fold-change  (Table 1). In addition, TCGA-GTEx database and GSE16011 data set in GEO database 27 were used to observe SOCS1/2/3/4 expression differences between GBM and normal tissues. All the results showed that mRNA levels of SOCS1/2/3/4 were significantly higher in GBM than in normal tissues ( Figure 1B survival rate is about 32% and the patient's 5-year survival rate is 0. The results also revealed that IDH status and SOCS3 contributed the most to the total points and survival probability relative to the other factors in the multivariate regression model ( Figure 2E). codeletion, primary therapeutic effects and age and negatively correlated with OS. High SOCS4 expression was associated with WHO status, 1p/19q co-deletion and primary therapeutic effects (p < 0.05; Table 2) ( Figure 3A,B). Furthermore, we analysed the prognostic F I G U R E 1 (A). SOCS1/2/3/4 mRNA expression levels in the Oncomine database in tumours. Red represents overexpressed genes, while blue represents underexpressed genes. The threshold parameters were as follows: p = 1 × 10 −4 and fold-change =1.5. (B). TCGA-GTEx databases showed that SOCS1/2/3/4 expression in normal and GBM tissues. (C-F). GSE16011 data set was used to analyse SOCS1/2/3/4 expression in normal and tumour tissues. The horizontal line in the middle of each box plot represents the median. When the median in the red box plot is higher than the median in the blue box plot, it means that the gene expression level in the cancer tissue is higher than that in the normal tissue, otherwise it is the opposite. (*p < 0.05; **p < 0.01; ***p < 0.001). SOCS, suppressor of cytokine signalling.  which also prepares us for further experimental research on SOCS3. ( Figure 5A-D). Furthermore, a Venn diagram showed seven intersections of SOCS1/3/4 coexpressed genes ( Figure 5E). The PPI network, which was generated by the GeneMANIA website, revealed that 20 potential target genes interacted with SOCS1/2/3/4 ( Figure 5F). Next, correlation analysis of SOCS1/2/3/4 was performed using the cBioPortal database, 28 and the results revealed that SOCS1 and SOCS3 were negatively correlated with SOCS4, while other genes were positively correlated ( Figure 5G). As  (Table S1). GEPIA 29 analysis of these 10 genes showed increased expression of IL6ST, IFNGR1, IFNGR2, IFNAR1, EGFR and CISH in GBM (p < 0.05) ( Figure S1). In addition , CUL5, IL6R, IFNGR1,   IFNGR2, IFNAR1,

| GSEA analysis
Since SOCS3 has the greatest prognostic value in GBM, we also further performed enrichment analysis for SOCS3, GSEA (gene   Figure 6A). Notably, we found that SOCS3 were associated with IL-6/JAK/STAT3 signalling pathway in GBM. Our further enrichment analysis of SOCS3 found that high-risk group was significantly associated with IL-6/JAK/STAT3 ( Figure 6B). In order to confirm this idea, we used siRNA and plasmid to knock down and overexpress

| Correlation of SOCS1/2/3/4 expression with immune characteristics
Previous studies have shown that tumour infiltrating lymphocytes (TILs) are independent predictors in tumours. 30 The present study an-

| Analysis of SOCS3 and immunoregulationassociated genes and immune cell infiltration
The above analysis revealed that SOCS1/2/3/4 expression was significantly correlated with multiple TILs. In order to further analyse the relationship between SOCS3 and immune regulatory genes, immune checkpoint genes and immune cell infiltration, we searched Sangerbox3.0 website based on TCGA and GTEx databases. Since immune checkpoint genes provide important targets for tumour immunotherapy, we specifically analysed the correlation between immunomodulatory genes and immune checkpoint genes and glioma.
We found that SOCS3 was positively correlated with most of the related genes in glioma, including immune activation genes, immune checkpoint genes, chemokine genes and chemokine receptor genes ( Figure 9A,B). We downloaded the agreed standardized pan-cancer data set from the UCSC database (http://xenab rowser.

| SOCS3 expression and epigenetic modification and mutation landscape
To explore the role of SOCS3 in epigenetic modification, SOCS3 was analysed at the epigenetic genome level. We used the cBio-Portal database to explore the categories and frequencies of SOCS3 genetic alteration in glioma ( Figure S7A). The main types of SOCS3 mutations are missense and truncation mutation ( Figure S7B).
Copy number values were significantly associated with diploid, gain and shallow deletion alterations of SOCS3 ( Figure S7C). RNA methylation is a common epigenetic modification and includes N1methyladenosine (m1A), cytosine-5-methylation (m5C) and N6methyladenosine (m6A). We analysed their correlation with SOCS3 expression levels. As shown in Figure 8A, there was a significant positive correlation between SOCS3 and 44 RNA modifications increased, suggesting that SOCS3 may promote the occurrence and development of GBM.

| Verification of SOCS1/2/3/4 mRNA expression in glioma tissue samples
The mRNA expression of SOCS1/2/3/4 in 174 cases of GBM and noncancer tissues was analysed by TCGA database, and the association between SOCS1/2/3/4 mRNA expression and clinicopathological features is shown in Table 3. It was found that age and IDH status were the common influencing factors of SOCS1/2/3/4 mRNA levels.
In addition, the SOCS1 mRNA levels were associated with OS and DFS, while the SOCS2 mRNA levels were associated with gender, race, Karnofsky performance score, OS and DFS. The mRNA level of SOCS3 was associated with Karnofsky performance score, OS and DFS, while the mRNA level of SOCS4 was markedly associated with Karnofsky performance score.

| Analysis of SOCS1/2/3/4 protein expression level by IHC staining
The HPA database for immunohistochemical staining was employed to explore the protein levels of SOCS1/2/3/4 in glioma samples (low grade and high grade) and normal brain tissue. The results revealed that SOCS1/2/3/4 protein expression in high-grade glioma tissues were higher than those in LGG and normal tissues. The expression level of SOCS1/2/3/4 in normal tissues was the lowest (Figure 10).
These findings suggested that upregulation of SOCS1/2/3/4 may predict advanced malignancies such as GBM.

| Experimental verification of the fact that SOCS3 is highly expressed in GBM cell lines and patients' tissues
To further verify the aforementioned bioinformatics analysis results, a series of experiments were carried out. RNA was extracted from 14 tissue specimens for the PT-qPCR analysis, and it was found that patients with GBM had higher SOCS3 mRNA expression than those found in normal tissues ( Figure 11A). Western blotting was performed on seven human GBM cell lines, and it was found that A172, U87, U118, LN229 and U251 cells had higher SOCS3 expression levels ( Figure 11B). Therefore, A172 and LN229 cells were transfected with siRNAs and the knockdown efficiency was statistically significant, particularly for LN229 ( Figure 11C,D). IHC staining revealed that SOCS3 was highly expressed in nine of 11 tissues derived from patients with GBM, while SOCS3 expression was low in all 12 patients with LGG ( Figure 11E,F). In conclusion, the present study

| Knockdown of SOCS3 attenuates GBM cells proliferation, migration and invasiveness
To identify the functional role of SOCS3 in GBM, LN229 cells with higher transfection efficiency were selected for further experiments. As aforementioned above, siRNA was used for transfection and Western blotting was performed to verify the knockdown efficiency. Numerous studies have revealed that SOCS3 expression is closely associated with the proliferation of tumour cells. 33,34 The current study used SOCS3 as a marker of cell proliferation for knockdown experiments, and detected PCNA expression by Western blotting, which revealed that the expression level of PCNA was reduced ( Figure 12A). In addition, transfected cells were plated into 6-well plates for colony formation assay, and the cell proliferation of groups subjected to different treatments was observed 10 days later. It was found that the cell proliferation of SOCS3 knockdown group was significantly reduced ( Figure 12B). Wound healing and Transwell assays revealed that the migration and invasion abilities of LN229 cells with SOCS3 knockdown were markedly reduced ( Figure 12C-E). In conclusion, inhibition of SOCS3 expression could reduce the proliferation, migration and invasion of GBM cell lines.

| DISCUSS ION
GBM is a malignant brain tumour with a high mortality rate and a  Genetic and epigenetic alterations are known to affect gene expression and may also be associated with adverse clinical outcomes. 22 We mainly analysed the types and frequencies of SOCS3  43 It has also been found that EYA2 is downregulated in hepatocellular carcinoma (HCC), and EYA2 can regulate SOCS3 expression in combination with DACH1 transcription, thereby inhibiting HCC progression via SOCS3-mediated blockade of the JAK/STAT signalling pathway. 44 Current studies have found that the high expression of SOCS3 can promote the proliferation of GBM cells. 34 Therefore, we want to further study the other functions of SOCS3, including migration and invasion, and explore the potential mechanisms of SOCS3's possible regulatory role by bioinformatics analysis, so as to prepare for further exploration of SOCS3 mechanisms in GBM in the future.
The current study found that GBM with high SOCS1/2/3/4 expression had a poor prognosis. Our group previously found a potential correlation between SOCS3 expression and GBM. 45 The current study verified the presence of high SOCS3 expression in a variety of cell lines through cell experiments, and the RT-qPCR and IHC analyses of human tissue specimens demonstrated the high expression of SOCS3 in GBM. To further explore the potential role of SOCS3 as a marker for poor prognosis of GBM, SOCS3 was knocked down by siRNA, and it was found that cell lines with reduced SOCS3 expression had lower proliferation levels, and reduced migration and invasion abilities. In conclusion, SOCS3 may be a potential prognostic marker for patients with GBM. In

FU N D I N G I N FO R M ATI O N
The present study was supported by grants from the National Natural Science Foundation of China (grant nos. 81972361 and 81874068) and Medical Science and Technology Project of Henan Province (grant nos. LHGJ20210487 and 222102310039).

CO N FLI C T O F I NTE R E S T S TATE M E NT
The authors declare no potential competing interests.

DATA AVA I L A B I L I T Y S TAT E M E N T
All data included in this study are available by contacting the corresponding authors.

CO N S E NT
The patients/participants provided their written informed consent to participate in this study.