G protein inhibitory α subunit 2 is a molecular oncotarget of human glioma

Identification of novel therapeutic oncotargets for human glioma is extremely important. Here we tested expression, potential functions and underlying mechanisms of G protein inhibitory α subunit 2 (Gαi2) in glioma. Bioinformatics analyses revealed that Gαi2 expression is significantly elevated in human glioma, correlating with poor patients' survival, higher tumor grade and wild-type IDH status. Moreover, increased Gαi2 expression was also in local glioma tissues and different glioma cells. In primary and immortalized (A172) glioma cells, Gαi2 shRNA or knockout (KO, by Cas9-sgRNA) potently suppressed viability, proliferation, and mobility, and induced apoptosis. Ectopic Gαi2 overexpression, using a lentiviral construct, further augmented malignant behaviors in glioma cells. p65 phosphorylation, NFκB activity and expression of NFκB pathway genes were decreased in Gαi2-depleted primary glioma cells, but increased following Gαi2 overexpression. There was an increased binding between Gαi2 promoter and Sp1 (specificity protein 1) transcription factor in glioma tissues and different glioma cells. In primary glioma cells Gαi2 expression was significantly reduced following Sp1 silencing, KO or inhibition. In vivo studies revealed that Gαi2 shRNA-expressing AAV intratumoral injection hindered growth of subcutaneous glioma xenografts in nude mice. Moreover, Gαi2 KO inhibited intracranial glioma xenograft in nude mice. Gαi2 depletion, NFκB inhibition and apoptosis induction were observed in subcutaneous and intracranial glioma xenografts with Gαi2 depletion. Together, overexpressed Gαi2 is important for glioma cell growth possibly by promoting NFκB cascade activation.

Intriguingly, depletion of Gαi2 was unable to prevent downstream signaling activation by RTKs [15][16][17][18][19]. Yet a potential function of Gαi2 in carcinogenesis and tumor progression has been reported. Gαi2 is elevated in colitis-associated cancer (CAC), correlating with decreased relapse-free survival [22]. Conversely, conditional knockdown of Gαi2 in CD11c + cells reduced CAC carcinogenesis [22]. Yin et al., reported that Gαi2 is important for epithelial ovarian cancer cell growth [23]. Conversely, microRNA-222-3p silenced Gαi2 to arrest epithelial ovarian cancer cell growth [23]. Zhang et al., proposed a pivotal role of Gαi2 in the development of nonalcoholic steatohepatitis [24]. Gαi2 expression was upregulated in liver tissues of NASH patients [24]. Importantly, hepatocytes specific Gαi2-deficient mice were resistant to the development of steatohepatitis [24]. Here we will show that overexpressed Gαi2 is important for glioma cell growth possibly by promoting activation of NFκB (nuclear factor kappa B) cascade.

Bioinformatics studies
The RNA-seq data, including 166 GBM (glioblastoma multiforme), 523 LGG (low grade glioma) tissues and 1157 normal tissues, along with the clinical data, were provided from UCSC XENA (https://xenabrowser.net/datapages/). Normalized gene expression was measured as transcripts per million reads plus the log2-based transformation. The overall survival of GBM and LGG patients was assessed through Kaplan-Meier analysis using the "Survival" along with "SurvMiner" R packages. The accuracy evaluation of the prognostic of Gαi2 was carried out by ROC curves using the R packages "Survival ROC" and "time ROC". TCGA LGGGBM cohorts were thereafter analyzed and Gαi2-associated differentially expressed gene (DEGs) were retrieved. KEGG analyses were carried out to explore the enrichment pathways. Chinese glioma functional genomic data were retrieved from the Chinese Glioma Genome Atlas (CGGA) [26]. The RNA sequencing of Diffuse Gliomas was through the Illumina Hiseq 2000. Clinical data were also retrieved from the CGGA data portal. RNA sequencing data and Clinical data were analyzed using R. software. The "Survival" package, "SurvMiner" package and "ggpubr" package were used.

NFκB activity
Briefly, the nuclear proteins were extracted through high-speed centrifugation. The TransAM™ ELISA kit (Active Motif) was utilized to examine the NFκB (p65) DNA-binding activity. In brief, 0.25 μg of nuclear extracts were tested for the binding of p65 to the specific DNA sequence. Following colorimetric reaction, the optical density (OD) value was tested through ELISA at 450 nm.

Chromatin immunoprecipitation (ChIP)
The detailed protocols of ChIP assay were reported in our previous study [25,30]. Briefly, the cell and human tissue lysates were homogenized [36] and fragmented DNA was achieved. Lysates were immunoprecipitated (IP) with the anti-Sp1 antibody. Sp1-bound DNA with the Gαi2 promoter site [37] was tested by the quantitative PCR method.

Xenograft studies
The nude mice were reported previously [17,21]. P1 cells or A172 cells were subcutaneously injected into the right flanks of nude mice and glioma xenografts were formed. The mice were intratumorally injected with the Gαi2 shRNAcontaining AAV or control shRNA AAV (1×10 9 PFU). Alternatively, P1 glioma cells were intracranially injected to the brains of the nude mice as described [38] and intracranial P1 glioma xenografts were formed. The protocols of this study were approved by Soochow University's IACUC and Ethics Committee.

Statistical analyses
Statistical methods were reported in our previous studies [20,25]. The data in this study were normally distributed and were shown as mean ± standard deviation (SD). All in vitro experiments were repeated five times, and each time similar results obtained.

Gαi2 overexpression in human glioma
Gαi2 expression data were retrieved from TCGA and Genotype-Tissue Expression (GTEx) project through UCSC XENA. A total of 689 glioma tissues ("Tumor"), including 166 glioblastoma (GBM) tissues and 523 low grade glioma (LGG) tissues, as well as 1157 normal brain tissues ("Normal") were retrieved. Gαi2 transcript number in 689 glioma tissues (GBMLGG, "Tumor") was higher than it in the normal brain tissues (P < 0.001, Figure 1A). Further analyses revealed that Gαi2 transcripts were significantly elevated in both GBM tissues ( Figure 1B) and LGG tissues ( Figure 1C).
The Kaplan-Meier survival and univariate Cox analysis from TCGA revealed that high Gαi2 expression in glioma tissues (GBMLGG) was correlated with poor patients' overall survival (P < 0.001, Figure 1D). Compared to Gαi2-low GBM patients, Gαi2-high GBM patients tend to have poor overall survival (Figure 1E), although no significant different was detected (P = 0. 218, Figure 1E). Importantly, Gαi2-high LGG patients' overall survival is significantly lower ( Figure 1F). Moreover, Gαi2 overexpression in GBMLGG tissues is correlated with patients' age and higher tumor grade ( Figure 1G), but was not correlated with patients' gender ( Figure 1G).
Alignment Diagram (Nomogram) prediction map based on Gαi2 expression showed that Gαi2 overexpression had a significant value in predicting poor survival probability of patients with GBMLGG ( Figure 1H), with area under the survival curve (AUC) at: 0.971 ( Figure 1H). Moreover, Gαi2 overexpression could predict poor 1/3/5-year survival probability of patients with GBMLGG, GBM or LGG ( Figure 1I). Bioinformatics analyses show that Gαi2 is significantly elevated in human glioma, correlating with patients' poor survival and higher grade of the tumors.
In local human glioma tissues Gαi2 expression was tested. Sixteen (n = 16) HGG tissues (labeled as "T") and matched adjacent normal brain tissues (labeled as "N") [17,20,21] were tested. Figure 2D revealed that Gαi2 mRNA levels in glioma tissues were increased significantly. In five representative HGG patients (Patient-1#-5#), Gαi2 protein expression was elevated in the glioma tissues ( Figure  2E and F). Combining all 16 pairs of Gαi2 expression data showed that Gαi2 protein was significantly upregulated in glioma tissues ( Figure 2G).
We next studied whether Gαi2 was upregulated in different human glioma cells, including primary human glioma cells ("P1-P5", derived from five patients [21]) and A172 cells. Gαi2 mRNA levels in the tested glioma cells were significantly higher than those in the primary human astrocytes ("Astro-cytes1/2") ( Figure 2H). Gαi2 protein upregulation was also in primary and immortalized glioma cells ( Figure  2I). Gαi2 mRNA ( Figure 2J) and protein ( Figure 2K) expression was also significantly elevated in other primary glioma cells-derived from other patients ("P4"/"P5") and in immortalized cell lines, including U87MG, U251MG and SHG-44. These results clearly show that Gαi2 is overexpressed in local glioma tissues/cells.
were added to P1 glioma cells with dCas9, and stable cells formed and named "koGαi2" cells. The control P1 glioma cells were with the scramble control shRNA (non-sense) plus the CRISPR/dCas9 empty construct ("shC+Cas9C"). As shown, Gαi2 mRNA and protein ( Figure 3A and B) levels were significantly decreased in shGαi2 and koGαi2 P1 glioma cells, and Gαi1/3 expression unchanged (Figure 3A and B). CCK-8 viability was decreased significantly in shGαi2 and koGαi2 P1 glioma cells (Figure 3C). Gαi2 shRNA or KO largely inhibited P1 glioma cell proliferation and significantly decreased the EdU-stained nuclei ( Figure   3D). In addition, genetic depletion of Gαi2 prevented P1 glioma cell colony formation (Figure 3E), further supporting the anti-proliferative activity.
Other primary human glioma cells, "P2" and "P3" (see our previous studies [17,20,21,25]), as well as the immortalized A172 glioma cells were again infected Gαi2 shRNA-expressing lentiviral particles, and stable cells (labeled as "shGαi2") were established after selection, showing depleted Gαi2 mRNA ( Figure  3H). Gαi2 shRNA significantly inhibited cell proliferation and decreased EdU incorporation in these primary and immortalized glioma cells ( Figure  3I). Moreover, in vitro cell migration was significantly slowed following Gαi2 silencing (Figure 3J). These results clearly supported that Gαi2 depletion led to robust anti-glioma cell activity.

Ectopic Gαi2 overexpression promotes glioma cell growth
Ectopic Gαi2 overexpression could possibly further enhance glioma cell progression. Therefore, the lentiviral particles packaging the Gαi2-expressing construct were transfected to P1 cells. Stable cells were again formed after selection: "oeGαi2". Gαi2 mRNA level was significantly augmented in oeGαi2 cells and was over 13-fold higher than that of vector control P1 glioma cells ("Vec") ( Figure 5A). Figure 5B confirmed Gαi2 protein upregulation in oeGαi2 P1 glioma cells. Gαi1/3 expression was not changed (Figure 5A-B). oeGαi2 promoted P1 glioma cell proliferation and increased nuclear EdU incorporation ( Figure 5C). Cell in vitro migration and invasion (Figure 5D-E) were accelerated following ectopic Gαi2 overexpression. The Gαi2-expressing construct lentiviral particles were also added to P2/P3 primary cells and A172 cells, resulting in robust Gαi2 mRNA elevation ("oeGαi2") ( Figure 5F). In the tested primary and immortalized glioma cells, ectopic Gαi2 overexpression augmented cell proliferation (EdU incorporation, Figure 5G) and accelerated in vitro cell migration ( Figure 5H). Thus, Gαi2 overexpression resulted in pro-glioma cell activity. Whereas in the "Astryocyte1" and "Astryocyte2", ectopic Gαi2 overexpression, by the same lentiviral Gαi2expressing construct, led to Gαi2 mRNA upregulation ("oeGαi2", Figure 5I). It however failed to increase proliferation (EdU incorporation, Figure 5J) in the astrocytes.

Sp1 and Gαi2 promoter binding increases in glioma tissues and cells
Since Gαi2 mRNA/protein levels were both elevated in glioma, it could be due to the transcriptional mechanism. Recent studies have implied that Sp1 (specificity protein 1) could be an important transcription factor of Gαi2 [37,50]. We first tested whether Sp1 was important for Gαi2 expression in glioma cells. To this purpose, lentiviral particles with Sp1 shRNA were added to P1 cells, and stable cells ("shSp1") formed. Alternatively, the lentiviral particles with the CRISPR/dCas9-Sp1-KO construct was added to the dCas9-expressing P1 cells, and stable Sp1 KO cells ("koSp1") formed after selection. Sp1 mRNA ( Figure 7A) and protein ( Figure  7B) expression was robustly decreased in shSp1 and koSp1 P1 glioma cells. Importantly, Gαi2 mRNA/ protein (Figure 7B and C) levels were reduced in Sp1-depleted P1 glioma cells. Mithramycin A, a compound that can prevent Sp1 binding to GC boxes in DNA [50], also decreased Gαi2 mRNA/protein levels in P1 cells (Figure 7D and E).
Next, the lentiviral particles packaging Sp1-overexpressing construct were added to P1 glioma cells, and stable cells established ("oeSp1"). Sp1 protein levels were remarkably upregulated in oeSp1 P1 glioma cells (Figure 7F). Following Sp1 overexpression, Gαi2 protein expression was increased as well ( Figure 7F). Remarkably, Sp1 ChIP results revealed that Sp1-Gαi2 promoter binding [51] in various glioma cells ("P1-P3" primary cells and A172 cells) was substantially higher than it in Astrocytes1/2 ( Figure 7G). Moreover, in human glioma tissues of five representative GBM patients, Sp1 binding to the Gαi2 promoter was robustly increased (Figure 7H). Therefore, Sp1 and Gαi2 promoter binding increasing could be an important mechanism of Gαi2 upregulation in human glioma tissues and cells.

P1 glioma cells (five million cells per mouse)
were subcutaneously (s.c.) injected to nude mice. Twenty days after cell injection, the subcutaneous P1 glioma xenografts were formed and each xenograft was close to 100 mm 3 ("Day-0"). AAV with Gαi2 shRNA ("AAV-sh-Gαi2") were intratumorally injected to P1 glioma xenografts daily (for ten days), and control mice intratumorally injected with AAV-shC. Every six days tumor volumes were recorded. As shown, AAV-sh-Gαi2 injection remarkably hindered subcutaneous P1 glioma xenograft growth ( Figure  8A) and reduced the estimated daily tumor growth [33,52]. Intratumoral AAV-sh-Gαi2 injection slowed P1 glioma xenograft growth ( Figure 8B). All P1 xenografts were carefully isolated at Day-42 and were tested. AAV-sh-Gαi2 xenografts were much smaller and lighter than AAV-shC xenografts (Figure 8C). No significant difference was observed in the mice body weights (Figure 8D). Thus, Gαi2 silencing inhibited subcutaneous P1 glioma xenograft growth in nude mice. Figure 8. Gαi2 silencing inhibits subcutaneous glioma xenograft growth in nude mice. The subcutaneous P1 glioma xenograft-bearing nude mice were daily intratumorally injected with Gαi2 shRNA-expressing AAV ("AAV-sh-Gαi2") or AAV-shC. The volumes of the xenografts (A) and animal body weights (D) were recorded. The estimated daily growth was calculated and was expressed at mm 3 per day (B). At Day-42, all P1 glioma xenografts were isolated and measured (C). Listed mRNAs and proteins in the described P1 glioma xenograft tissues were tested (E, F, H and I). The representative IHC images of Gαi2 in the described P1 glioma xenograft slides were presented (G). Nuclear TUNEL fluorescence staining in the described P1 xenograft slides were presented (J). The subcutaneous A172 glioma xenograft-bearing nude mice were subject to daily intratumoral injection of AAV-sh-Gαi2 or AAV-shC. At Day-35, all A172 glioma xenografts were isolated, and were measured (K and L). In the xenograft tissues listed mRNAs and proteins were examined, and results quantified (M and N). *P < 0.05 versus "AAV-shC" group. "N.S." means P > 0.05. Scale bar = 100 μm.

Discussion
GBM and other HGG are most aggressive and lethal malignant tumors that originate in the brain [53,54]. Currently, there is a lack of effective treatments [5,6,55,56]. Compared with traditional treatment methods, molecular targeted therapies [57,58] could have better selectivity and specificity against glioma [5,6,55,56]. We showed that Gαi2 could be an important therapeutic oncotarget of glioma. Bioinformatics analyses revealed that Gαi2 transcripts are significantly elevated in human glioma, and its overexpression correlates with poor patients' survival, higher tumor grade and WT-IDH status. Moreover, Gαi2 upregulation is also detected in local glioma tissues and various human glioma cells.
In primary and immortalized (A172) glioma cells, Gαi2 shRNA or KO substantially suppressed viability, cell proliferation and mobility. Silence of Gαi2 by targeted shRNA however failed to inhibit viability and proliferation in non-cancerous human astrocytes. In addition, Gαi2 shRNA or KO provoked caspase activation, mitochondrial depolarization and apoptosis in the primary and A172 glioma cells. Whereas Gαi2 silencing failed to provoke caspase-apoptosis activation in human astrocytes. Contrarily, ectopic Gαi2 overexpression, using the lentiviral construct, further promoted malignant behaviors of primary and immortalized glioma cells, enhancing cell proliferation, migration and invasion. Gαi2 overexpression was however not effective in human astrocytes. Importantly, daily intratumoral Gαi2 shRNA AAV injection largely hindered subcutaneous P1 xenograft growth in nude mice. Moreover, the growth of intracranial P1 xenografts was largely inhibited after Gαi2 KO. Therefore, overexpressed Gαi2 is important for glioma cell growth.
Early studies have implied that Gαi2 could be important for NF-κB cascade activation. Conditional disruption of Gαi2 in CD11c + DCs and MDSCs prevented NF-κB and STAT3 activation [22]. Gαi2-depletion-induced NF-κB inactivation was possibly due to blocking IL-6 signaling [22]. Under hepatic ischemia-reperfusion injury, increased Gαi2 expression promoted NF-κB pathway activation through phosphorylating mixed-lineage protein kinase 3 (MLK3) [60]. In the present study, we found that Gαi2 was important for NFκB activation in glioma cells. Indeed, p65 phosphorylation, NFκB (p65) DNA-binding activity and expression of NFκB-dependent genes (cIAP2 and survivin) were significantly decreased in Gαi2-depleted primary glioma cells, but were increased following Gαi2 overexpression. BAY-11-7082, the NFκB inhibitor, largely suppressed proliferation and migration of Gαi2-overexpressed P1 glioma cells. Importantly, decreased p65 phosphorylation was observed in subcutaneous and intracranial glioma xenografts with Gαi2 depletion. Therefore, promoting NFκB cascade activation should be one important mechanism of Gαi2-driven glioma cell growth.
Studies have proposed that Sp1 is an important transcription factor for the malignant progression of glioma. Yu et al., found that Sp1 enhanced NLR family pyrin domain containing 6 (NLRP6) transcription to promote immune evasion, malignant behaviors and radio-resistance in glioma cells. Contrarily, Sp1 silencing suppressed in vitro glioma cell growth and tumorigenesis in vivo [61]. Li et al., reported that Sp1 upregulated the LncRNA LBX2-AS1 to promote proliferation and EMT in glioma cells [62]. Tan et al., discovered that miR-150-3p silenced Sp1 to hinder glioma cell growth [63]. Our results supported that Sp1-dependent Gαi2 transcription was increased in glioma tissues and cells, which might be one primary mechanism of Gαi2 upregulation in glioma. In glioma cells Gαi2 expression was downregulated after Sp1 silencing, KO or inhibition. It was however increased following Sp1 overexpression. Therefore, the increase of Sp1-dependent transcription should be one key mechanism of Gαi2 overexpression in human glioma.
Here, TCGA LGGGBM cohorts were analyzed and Gαi2-asscoateid DEGs in glioma tissues were retrieved, including a significant number of genes with unknown functions in human glioma. Moreover, KEGG analyses showed that Gαi2-asscoateid DEGs were enriched in NFκB and other signaling cascades. Further studies will be needed to explore expression and potential functions of these Gαi2-asscoateid DEGs in glioma, and to test these enriched pathways in the progression of glioma. Their connection with Gαi2 should also be analyzed.

Conclusion
Together, overexpressed Gαi2 is important for glioma cell growth possibly by promoting NFκB cascade activation. Gαi2 is possibly a novel and promising therapeutic oncotarget of glioma.