GNG5 is a novel oncogene associated with poor prognosis in glioma patients

Background: Although many biomarkers have been reported for detecting glioma, the prognosis for the disease remains poor, and therefore, new biomarkers need to be identied. GNG5, which is part of the G-protein family, has been associated with different malignant tumors, though the role of GNG5 in glioma has not been studied. Therefore, we aimed to identify the relationship between GNG5 expression and glioma prognosis and to identify a new biomarker for the diagnosis and treatment of gliomas. Methods: We used datasets from databases including TCGA and GEO, and results from GEPIA, RT-qPCR, and HPA to determine the expression of GNG5 in glioma. Based on datasets obtained from the CGGA database, we identied the correlation between GNG5 expression and multiple molecular and clinical features as well as clinical prognosis using a variety of analytical methods. Co-expression analysis and GSEA were performed to detect GNG5-related genes in gliomas and possible signaling pathways involved. ESTIMATE, ssGSEA, and TIMER were used to detect the relationship between GNG5 and the immune microenvironment. Results: A total of 1826 glioma related datasets were used in our study, including sequencing data, microarray data, and RT-qPCR data. We found that GNG5 is highly expressed in gliomas, and its expression level is positively correlated with pathological grade, histological type, age, and tumor recurrence and negatively correlated with isocitrate dehydrogenase mutation, 1p/19 co-deletion, and chemotherapy. Moreover, GNG5 as an independent risk factor was negatively correlated with the overall survival time. GSEA analysis revealed the potential signaling pathways involved in GNG5 function in gliomas, such as ECM-receptor interaction and the toll-like receptor signaling pathway. The ssGSEA, ESTIMATE, and TIMER based analysis indicated a correlation between GNG5 expression and various immune cells in glioma, such as B cell, macrophage, and dendritic cells. Conclusions: Based on the large data platform and the use of different databases to corroborate results obtained using various datasets, our study reveals for the rst time that GNG5, as an oncogene, is overexpressed in gliomas and can lead to poor prognosis of patients. Thus, GNG5 is a potential novel biomarker for the clinical diagnosis and treatment of gliomas.

GNG5 is a novel oncogene associated with poor prognosis in glioma patients Wang Zhang First malignant tumors are highly heterogeneous, the use of these widely used biological markers is not su cient for the precise diagnosis and treatment of glioma. Therefore, there is an urgent need for novel molecular markers with high speci city and sensitivity to improve the molecular diagnosis and treatment of glioma.
G-proteins are a family of proteins that participate in multiple cellular functions such as cell division, differentiation, and metastasis during embryonic development through direct interaction with G proteincoupled receptors [3,4]. The formation of gliomas is related to embryonic development; for example, the epithelial-mesenchymal transition (EMT) is a process vital for embryonic development, and it has been shown to regulate the progression and invasion of glioma [5,6]. Multiple studies have con rmed the signi cance of G-protein family members in the pathological progress of cancer. For example, GNG7 is an epigenetic silencing gene involved in the malignant progression of renal clear cell carcinoma and esophageal cancer [7,8]; GNG4 function shows a high degree of correlation with liver cancer and colorectal cancer and is used in their diagnosis and prognosis [9,10]; and GNG11 promotes the adhesion, migration, and invasion of gastric cancer cells [11]. Thus, the G-protein family is associated with tumor progression and may be used as potential biomarkers for the diagnosis and treatment of tumors.
We focused on the GNG5 protein of the G-protein family, and few reports describe the role of GNG5 in tumors. GNG5, as a gamma subunit of G-protein, localized to the human chromosome 1p22, and is composed of 3 exons and 2 introns, and regulates the occurrence and development of many diseases [12]. Additionally, it acts as a regulator of islet function regulating the secretion of insulin, and is also used as a marker of melanosis coli to regulate cellular apoptosis [13,14]. GNG5 has been shown to play an important role in various cancers such as endometrial carcinoma [15] where it is highly expressed, and in invasive ductal carcinoma of the breast [16] where it regulates the secretion of E-cadherin through the Wnt signaling pathway. However, the role of GNG5 in contributing to the pathological mechanism in glioma has not been reported.
We focused on exploring the expression pattern of GNG5 in glioma from gene level to protein level, examining the underlying relationship between GNG5 and the clinical and molecular characteristics of glioma in patients, and revealing the potential biological function of GNG5 in the pathological progression of glioma. The results obtained could enhance our understanding of the complex molecular mechanisms of glioma and provide a promising novel target for combating the disease.

Data collection
A comparison of the transcript levels of GNG5 in different tumors was performed using Gene Expression Pro ling Interactive Analysis (GEPIA, http://gepia.cancer-pku.cn). To identify differences in expression for GNG5 between gliomas and the control group, RNAseq data from 698 gliomas and 5 normal brain tissues were downloaded from The Cancer Genome Atlas (TCGA, https://portal.gdc.cancer.gov/), and microarray data of the GSE131273 dataset from the Gene Expression Omnibus (GEO, https://www.ncbi.nlm.nih.gov/geo/) database. A Mann-Whitney test was performed to analyze the difference in expression for GNG5 between glioma and normal brain tissues. RNA sequencing data from 1018 gliomas and their corresponding clinical information was downloaded from the Chinese Glioma Genome Atlas (CGGA, http://www.cgga.org.cn/help.jsp) database. The relationship between the expression level of GNG5 and clinicopathological features was further analyzed using the data obtained from CGGA. The patients were divided into older and younger groups according to the median age of the patients, and high and low expression groups according to the average expression level of GNG5 in the samples. Additionally, we obtained the overall survival (OS) information for glioma patients from the GSE53733 dataset, including 23 patients who showed long-term survival (> 36 months), 16 patients that showed short-term survival (< 12 months), and 31 patients who displayed intermediate survival. We analyzed the difference in expression of GNG5 among the three groups of patients with different survival phenotypes. The expression of GNG5 at the protein level was analyzed using data from the Human Protein Atlas (HPA, https://www.proteinatlas.org/) database.

Patients, tissue preparation and RT-qPCR
A total of 40 primary glioma samples and 5 non-glioma samples were selected from the Department of Neurosurgery of the First A liated Hospital of Harbin Medical University. The tissues were cut into bean grains and put into liquid nitrogen after removal, as soon as possible. Total RNA was collected using Trizol Reagent ( Thermo Fisher Scienti c, Waltham, MA, USA) and RNA was reverse transcribed using the transcriptor rst-strand cDNA synthesis kit (Hoffmann-La Roche, Basel, Switzerland), according to the manufacturer's instructions as described previously [17]. RNA-speci c primers (Ruibo Beijing) were used for the GNG5 reverse transcription (GNG5-forward: CGGACTCAACCGCGTAAA, GNG5-reverse: GGGTCTGAAGGGATTTGTACTT). RT-PCR was conducted with SYBR Green on a 7500HT Fast Realtime System (Applied Biosystems, Foster City, CA, USA). The glyceraldehyde-3-phosphate dehydrogenase (GAPDH, forward: TCCAAAAATCAAGTGGGGCGA, reverse: TGATGACCCTTTTGGCTCCC) was used for the normalization of GNG5. The relative fold-change in the expression of GNG5 was determined by using the -ΔCT method.
Gene set enrichment analysis (GSEA) and s immune correlation analysis GSEA is a tool developed by research teams at MIT and Harvard University's Broad Institute to analyze genome-wide expression pro ling from microarray data [18], and we used it for analyzing GNG5 related pathways and molecular mechanisms in glioma. We used 1000 gene set permutations for each analysis using the expression level of GNG5 as a reference phenotype tag. The resulting enriched pathways were analyzed based on nominal (NOM) P-values and normalized enrichment scores (NES). We examined the relationship between the expression of GNG5 in glioma tissues and the immune microenvironment using the single-sample gene set enrichment analysis (ssGSEA) to calculate the enrichment of 29 immune cell geneset signatures in each glioma sample based on data downloaded from the CGGA database [19][20][21]. Further, a cluster analysis of glioma samples was performed based on the quantitative results obtained from the ssGSEA analysis. The ESTIMATE method (estimate R package) was used to evaluate the immune scores, tumor purity, and the stromal score for each glioma sample [22]. Tumor IMmune Estimation Resource (TIMER, https://cistrome.shinyapps.io/timer/) is an ideal public database for studying the in ltration abundance of tumor-in ltrating immune cells. The database pre-evaluated the immune in ltration levels of six immune in ltrating cells (including B cells, CD4 + T cells, CD8 + T cells, neutrophils, macrophages and dendritic cells) in 10897 cancer samples by complex statistical methods.
We examined the relationship between the expression of GNG5 and the in ltration abundance of the above six immune cells in glioma using TIMER [23].

Statistical analysis
We used R (v3.5.1) for statistical analysis. The Wilcoxon rank-sum test was used to analyze the correlation between the expression level of GNG5 and clinical data of glioma patients. A Cox's regression model and the Kaplan-Meier method was used to analyze the relationship between the expression level of GNG5 and the OS of the patients. COX test was used for univariate and multivariate analysis to reveal the risk factors affecting the prognosis of glioma. The Mann-Whitney test was used to analyze the differences in the quantitative reverse transcription PCR (RT-qPCR) results using the Graphpad Prism 8.0 software (GraphPad Software Inc., San Diego, U.S.A.).

Clinical characteristics of the patients involved in the study
We used data from the CGGA database that contains a large number of glioma gene expression pro les and a vast amount of clinical information data to further analyze the relationship between GNG5 expression and the clinical features of glioma. RNA sequencing data of 749 patients, including 442 males and 307 females, with complete clinical information were screened from 1018 samples based on the CGGA database. The mean age and survival range were 43.26 years and 8-79 years for males, and 3.22 years and 0-11.99 years for females, respectively. In total, we analyzed 502, 222, and 25 cases of primary, recurrent, and secondary gliomas, respectively. An analysis of the pathological results showed that there were 218, 240, and 291 cases with WHO grade II, grade III, and grade IV gliomas, respectively. In addition, 40.28% of the patients (n = 410) showed mutations in IDH, and 33.30% patients (n = 339) were wildtype for IDH. Only 15.23% of patients (n = 155) showed a 1p19q co-deletion, whereas 58.35% (n = 594) did not have this genotype. These results are summarized in Table S1.

GNG5 is highly expressed in gliomas
We analyzed GNG5 expression in different tumors and control groups using GEPIA. A total of 15,203 patient samples were included, including 9,663 samples from 33 tumors and 5,540 corresponding control samples. GNG5 expression was elevated in tumor tissues including glioblastoma (GBM), lower-grade glioma (LGG), ovarian serous cystadenocarcinoma (OV), pancreatic adenocarcinoma (PAAD), skin cutaneous melanoma (SKCM), and uterine carcinosarcoma (UCS), and showed signi cantly lower expression in acute myeloid leukemia (LAML) relative to the control group (Fig. 1A). These ndings suggest that GNG5 is highly expressed in a variety of tumors. To further analyze the expression of GNG5 in gliomas, we analyzed the expression level of GNG5 in glioma and normal brain tissues based on transcriptome data in TCGA database and chip data in GSE131273 dataset, respectively. And the results showed that GNG5 expression was signi cantly higher in gliomas relative to normal brain tissue (Fig. 1B, C, P < 0.001), which was consistent with the results of RT-qPCR ( Fig. 1D, P < 0.05). Furthermore, immunohistochemical data from HPA revealed that GNG5 expression in glioma was higher than that in non-glioma tissues ( Fig. 1E-F).
Relationship between GNG5 and the underlying molecular and clinical characteristics in glioma patients We analyzed the relationship between GNG5 expression level in the 749 samples from the CGGA database and the various tumor subtypes, their pathological classi cation, molecular classi cation, tumor treatment, and the age of the patients. The results indicated that the GNG5 expression level in gliomas was signi cantly correlated with the pathological grade and the age of the patients ( Fig. 2A, C, P < 0.001). The average level of GNG5 expression in tissue from patients with recurrent gliomas was higher than in patients with primary and secondary tumors ( Fig. 2B, P < 0.001). Similar results were found based on a correlation analysis between GNG5 expression level in gliomas and different pathological subtypes, such as the expression levels of GNG5 in recurrent astrocytoma, recurrent anaplastic astrocytoma and recurrent anaplastic oligodendroglioma were higher than that in corresponding primary pathological subtypes, as shown in Fig. 2G.
Reports from research conducted previously shows that patients with mutations in the IDH gene or a 1p/19q co-deletion had longer survival and better prognosis [24]. Interestingly, we found that the expression level of GNG5 in patients with mutations in the IDH gene was lower than in patients with wildtype IDH (Fig. 2D). Similarly, patients with a 1p/19q co-deletion had signi cantly lower expression of GNG5 than in patients without the co-deletion (Fig. 2E). Together, these results suggest that GNG5 expression may correlate with the prognosis of glioma patients. When a grouping of patients based on whether they received chemotherapy or not was performed, we found that tissue from patients who received chemotherapy showed an upregulation of GNG5 expression relative to those who did not receive chemotherapy (Fig. 2F). However, radiotherapy did not affect GNG5 expression (data was not shown).
One explanation for this observation may be that the surgical resection of the tumor and subsequent sequencing data analysis are from samples collected from patients before they underwent chemotherapy. Taken together, these results indicate that a correlation exists between GNG5 expression and the molecular and clinical characteristics of glioma.
High expression of GNG5 is a predictor of poor prognosis The average follow-up time for the 749 patients whose data for survival analysis is included in this study was 3.22 years. The results from the survival analysis showed that high expression of GNG5 is signi cantly correlated with reduced survival of glioma patients (Fig. 3A, P < 0.001); these results were veri ed by Kaplan-Meier method besed on TCGA database and the GNG5 expression analysis based on GSE53733 dataset ( Figure S1A, B). Moreover, a time-dependent receiver operated characteristic (ROC) analysis showed that the values for area under the ROC curve (AUC) were 0.714, 0.792, and 0.821 for one, three, and ve-year OS respectively (Fig. 3B). Therefore, our results indicate that GNG5 may serve as a biomarker for glioma patients, especially for the ve-year OS group.
The results of a univariate analysis suggest that the high expression level of GNG5 in glioma patients is associated with reduced OS, primary recurrence or secondary (PRS) type, histological type, pathological grade, age, chemotherapy status, presence of mutations in the IDH gene, and a 1p/19q co-deletion (Fig. 4A). Additionally, results from the multivariate analysis indicate that the expression of GNG5, PRS type, pathological grade, age, chemotherapy status, presence of mutations in the IDH gene, and 1p/9q codeletion were independently correlated with OS (Fig. 4B). The above data indicate GNG5 may serve as a prognostic factor and increased GNG5 expression is associated with poor OS.

Co-expression analysis of GNG5
To further explore the function of GNG5, the limma package (Bioconductor) was used in the R statistical software to analyze the coexpression genes with GNG5. We found 4517 genes associated with GNG5 GNG5 related signaling pathways based on GSEA GSEA was used to identify GNG5 related signaling pathways involved in gliomas. A P < 0.05 and a false discovery rate (FDR) < 0.25 represented a signi cant enrichment in the results (in the enrichment of MSigDB Collection). Six pathways, including those involving the ECM-receptor interaction, focal adhesion, cell adhesion molecules, toll-like receptor signaling pathway, nod-like receptor signaling pathway, and the rig-like receptor signaling pathway showed signi cantly differential enrichment in samples from patients showing the GNG5 high expression phenotype based on NES, NOM P-values, and FDR values ( Fig. 6; Table 1), indicating a potential role for GNG5 in the development of glioma. GNG5 is related to the immune microenvironment Next, we explored whether there is a correlation between GNG5 expression and the tumor immune microenvironment. We analyzed 29 immune-system related gene sets characterizing different types and functions of immune cells (Table S2). Based on data from the CGGA database, ssGSEA was used to quantify and hierarchically cluster immune cells in tumor samples into three groups. According to the clustering heat map of immune cell gene sets in the three groups, a high immunity group (high-immune), medium immunity group (mid-immune), and low immune activity group (low-immune) were de ned ( Fig. 7A). Moreover, the high-immune group had a signi cantly higher immune score than that of the lowimmune group, though the tumor purity showed the opposite characteristics (Fig. 7B, C). Interestingly, we found that the expression of GNG5 was signi cantly increased in the high-immune group, and decreased in the low-immune group (P < 0.001, Fig. 7D). Further, we used TIMER to analyze the relationship between GNG5 expression and in ltration abundance of six immune cells (B cells, CD4 + T cells, CD8 + T cells, neutrophils, macrophages, and dendritic cells) after purity correction in glioma [23,25]. We found that GNG5 expression was positively correlated with the in ltrating abundance of CD8 + T cell (Cor = were corrected with GNG5 expression in GBM (Fig. 7E).
These results suggest that GNG5 may be a potential factor in uencing the immune microenvironment in glioma.
Discussion G-proteins consisting of α, β and γ subunits are also known as GTP binding proteins, which mediate the transduction of a variety of hormones and neurotransmitters across the membrane and subsequently trigger a series of physiological and biochemical reactions in cells [26,27]. G-proteins regulate basic life processes such as cell metabolism, secretion, growth, proliferation, differentiation, distortion, pathological changes, and cell death [4]. GNG5, a subunit of G-protein, has been reported to promote the proliferation and migration of tumor cells [28]. However, to date, few studies have reported the role and mechanism of action for GNG5 in tumors.
To explore the expression patterns of GNG5 in various tumors, we searched for GNG5 in the GEPIA database. We found that GNG5 was signi cantly highly expressed in a variety of cancers, including gliomas. (Fig. 1A). To further understand the role of GNG5 in tumors, we focused on its role in glioma. We obtained GNG5 expression data from TCGA and GEO databases and clinical samples. Our results showed that GNG5 expression was increased in glioma based on data from sequencing, microarrays, and PCR data (Fig. 1B-D). Previous studies have suggested that GNG5 shows elevated expression in endometrial cancer and in in ltrating ductal carcinoma of the breast [15,16]. Furthermore, POP4, a GNG5 coexpressing promoter gene, is highly expressed in prostate cancer cells, while RIMS1, a GNG5 coexpressing suppressor gene, is lowly expressed in craniopharyngioma [29,30]. Further, we searched the HPA database for GNG5 protein expression, and the results indicate that a high level of GNG5 is observed in glioma (Fig. 1E-F). Interestingly, we found that GNG5 expression levels are closely related to the clinical features associated with glioma prognosis, especially with the pathological grade and primary recurrence status of glioma (Fig. 2). Some studies have reported that glioma patients with a mutation in the IDH gene and a 1p/19q co-deletion have a better prognosis [31,32]. However, we found that the average expression level of GNG5 was lower in samples from patients with mutations in the IDH gene and those with a 1p/19q co-deletion. Therefore, we hypothesized that GNG5 may be an oncogene in glioma.
To clarify whether GNG5 is an oncogene and may lead to a poor prognosis for glioma, we obtained clinical data from thousands of glioma samples from TCGA, CGGA, and GEO databases. The results from our analysis show that increased expression of GNG5 could indeed cause poor prognosis of glioma patients (as shown in Fig. 3A, S1). Moreover, as shown in Fig. 3B, GNG5 has a high diagnostic value and can be used as a biomarker for predicting the prognosis of glioma. Additionally, genes coexpressed with GNG5 such as AK2 are highly expressed in lung cancer and can lead to poor prognosis in patients [33]; besides, RER1 as a coexpressed gene with GNG5, could promote the progression of pancreatic cancer and reduce the survival rate of patients [34]. Although, the high expression of GNG5 in glioma could reduce the OS rate of patients, whether the expression of GNG5 is necessarily correlated with such a prognosis for glioma patients was further examined. Results from univariate and multivariate analyses showed that increased expression of GNG5 is an independent risk factor for glioma patients. However, the mechanism explaining how GNG5 leads to a poor prognosis for glioma patients is yet to be elucidated.
We next undertook a GSEA to understand the functioning of GNG5 in glioma. We found that high expression of GNG5 promotes the activation of a series of cancer-related signaling pathways (Fig. 6). Similar interactions have been described previously; for example, the ECM-receptor interaction is involved in the proliferation and invasion of cancer cells [35,36], and focal adhesion is involved in the metastasis and invasion of cancer cells [37]. Additionally, the toll-like receptor promotes the immune evasion of glioma by down-regulating the MHC II molecules in microglia [38], and cell adhesion molecules and the nod-like receptor play a signi cant role in the proliferation, invasion, and metastasis of glioma cells [39][40][41][42]. A particular gene may play a role in a variety of signaling pathways in the pathological process of diseases. Used as a tool to reveal the molecular mechanisms underlying diseases, GSEA is widely used and has high reliability. Compared with traditional methods of analysis, GSEA has the advantage of using large sample sizes and can avoid the biases inherent in experimental results caused by arti cial threshold setting [18]. Previous studies have suggested the reliability of GSEA as a predictive tool. For example, Liu Z et al. reported that overexpression of HOXA2 can activate the JAK-STAT and focal adhesion signaling pathways in glioma [43], and Wang et al. elucidated the autophagy-related signaling pathway in glioblastoma through the GSEA method [44]. Therefore, we believe that the signaling pathways enriched with GNG5 in glioma as revealed in our study could provide the basis for further investigations into the speci c mechanisms underlying GNG5 functioning in glioma.
Recently, immunotherapy has been shown to be a new treatment strategy for glioma, and genes related to glioma immunity have been reported frequently [45,46]. However, the association of GNG5 with glioma immunity has not been reported. Our results suggest that GNG5 may participate in the malignant progression of glioma through ECM-receptor interaction, toll-like receptor pathway, and the nod-like receptor signaling pathways, which are reported to be related to the immune response process [47,48]. Therefore, we hypothesized that GNG5 may be involved in the immune response in glioma. To verify this, we analyzed 29 immune-associated gene sets in glioma samples using ssGSEA and ESTIMATE. The results showed that the expression of GNG5 in glioma was correlated with immune activity (Fig. 7D).
Previous studies have reported a correlation between programmed death-1/ligand-1 (PD-1/PD-L1) and immune activity in tumors in triple-negative breast cancer [20]; Moreover, a large number of studies have shown that ssGSEA and ESTIMATE are highly reliable in the evaluation of tumor immune strati cation and immune invasion analysis [22,[49][50][51][52]. Furthermore, we analyzed the correlation between the expression of GNG5 and in ltration level of tumor-in ltrating immune cells using TIMER. The results suggest that GNG5 is correlated with the in ltration abundace of various immune cells in glioma, especially macrophage, dendritic cells and neutrophils in low-grade glioma (Fig. 7E). Though there are currently no reports describing the relationship between the G-protein family and cancer immunity, correlations between single genes and cancer immunity have been observed. For example, CD70, which is highly expressed in glioma, can promote macrophage in ltration into a tumor [45], and JAK1 was shown to be positively correlated with the in ltration of various immune cells in breast cancer, such as CD8 + T cells and dendritic cells, etc. [53]. Additionally, the expression of LAYN was found to be positively correlated with the in ltration of CD4 + T and CD8 + T cells, macrophages, neutrophils, and dendritic cells in colon and gastric adenocarcinomas [54]. These results indicate that GNG5 is a highly expressed gene in glioma, is associated with immune activity, and future research could help establish the immune response related to GNG5 function in glioma.
Although we used bioinformatics analysis and large datasets to uncover the role of GNG5 in glioma, our study has a few limitations. First, as extensive clinical sample information in the public databases are not always available, it is di cult to comprehensively evaluate the correlation between GNG5 and the clinical features of glioma patients. However, it is di cult to avoid the problem of the lack of clinical sample information in public databases, as this data is collated from different research centers. Second, although multicenter research in public databases can make up for the de ciencies inherent in singlecenter research, there are still some shortcomings, especially the inconsistency of intervention measures and the lack of clinical information. Thus, prospective studies implemented in the future will need to minimize bias in experimental results due to retrospective analysis.

Conclusion
Our study shows that GNG5 is highly expressed in gliomas, and this expression is correlated with various molecular and clinical features of glioma patients. High expression levels of GNG5 predict a poor prognosis in glioma patients. Additionally, GNG5 may participate in the pathological progress of glioma through the signaling pathways related to cancer, such as ECM-receptor interaction, the toll-like receptor signaling pathway, and the nod-like receptor signaling pathway. These results indicate that GNG5 is a novel oncogene in glioma and could provide a potential biomarker for the diagnosis and treatment of gliomas. The public data involved in this study were all downloaded from public databases, therefore there was no new informed consent required. Immunohistochemical results of GNG5 in glioma and normal tissues based on human protein atlas. ***P < 0.01, *P < 0.05.

Figure 2
Expression characteristics of GNG5 in gliomas based on CGGA.