Neuronal-driven glioma growth requires Gαi1 and Gαi3

Neuroligin-3 (NLGN3) is necessary and sufficient to promote glioma cell growth. The recruitment of Gαi1/3 to the ligand-activated receptor tyrosine kinases (RTKs) is essential for mediating oncogenic signaling. Methods: Various genetic strategies were utilized to examine the requirement of Gαi1/3 in NLGN3-driven glioma cell growth. Results: NLGN3-induced Akt-mTORC1 and Erk activation was inhibited by decreasing Gαi1/3 expression. In contrast ectopic Gαi1/3 overexpression enhanced NLGN3-induced signaling. In glioma cells, NLGN3-induced cell growth, proliferation and migration were attenuated by Gαi1/3 depletion with shRNA, but facilitated with Gαi1/3 overexpression. Significantly, Gαi1/3 silencing inhibited orthotopic growth of patient-derived glioma xenografts in mouse brain, whereas forced Gαi1/3-overexpression in primary glioma xenografts significantly enhanced growth. The growth of brain-metastatic human lung cancer cells in mouse brain was largely inhibited with Gαi1/3 silencing. It was however expedited with ectopic Gαi1/3 overexpression. In human glioma Gαi3 upregulation was detected, correlating with poor prognosis. Conclusion: Gαi1/3 mediation of NLGN3-induced signaling is essential for neuronal-driven glioma growth.

Recent studies have revealed that neuronal activity and neuron-glioma circuits are vital for the malignancy of human glioma, and that the neuronal activity could be a key determinant for glioma tumorigenesis and progression [14][15][16]. Studies highlighted a crucial role of neuroligin-3 (NLGN3) in the mechanism of glioma progression [17][18][19][20]. NLGN3, a synaptic protein cleaved and secreted in an activity-dependent manner, was identified as the primary factor responsible for glioma progression [17][18][19][20]. Xenografts derived from high-grade primary human glioma cells failed to grow in a NLGN3-deficient brain [18], demonstrating that NLGN3 is required for glioma cell growth. Furthermore, inhibition of ADAM10 (A Disintegrin And Metalloproteinase 10) blocked activitydependent NLGN3 secretion to inhibit orthotopic growth of primary glioma xenografts [19].
NLGN3 secreted from neurons was found to induce phosphorylation and activation of several key receptor tyrosine kinases (RTKs) on glioma cells to promote glioma progression [17][18][19]. Additionally, NLGN3-activated glioma cells produce feedforward expression of NLGN3 to further increase glioma cell growth and proliferation [17][18][19]. Recent studies found that NLGN3 induces the expression of synaptic genes required for neuron-glioma synapse formation, as well as genes essential for glioma proliferation and invasiveness [14,15]. These studies identify NLGN3 as a key neuronal-derived factor for promoting activity-dependent glioma growth [17][18][19].
Gαi proteins, the inhibitory α subunits of G proteins (heterotrimeric guanine nucleotide-binding proteins), have three primary family subunits, Gαi1, Gαi2 and Gαi3 [21]. Through binding to GPCRs (G protein-coupled receptors) Gαi proteins and the βγ complexes can suppress adenylate cyclase (AC) to deplete cyclic AMP (cAMP) levels [21]. Gαi proteins are inhibited by pertussis toxin. Glider et al., showed that pertussis toxin potently inhibited migration and invasion of high grade glioma cells [22]. In addition, pertussis toxin and temozolomide co-treatment resulted in better anti-glioma efficiency in a rat intracerebral glioma model [23]. Moreover, a D2 dopamine receptor agonist stimulated mitogenesis of C6 glioma cells through activation of pertussis toxin-sensitive Gαi proteins [24].
As neuron-secreted NLGN3 activates RTKs to promote glioma progression, we explored the potential function of Gαi1 and Gαi3 in NLGN3-induced signaling and glioma cell progression.

Methods
Ethics. Protocols of this study were approved by the Ethics Committee of Soochow University.
Cell lines. U251MG glioma cell line was obtained from the Cell Bank of Shanghai Biological Institution (Shanghai, China), cultivated as described.
Human glioma tissues. According to principles of the Declaration of Helsinki, this study was approved by the Ethics Committee of Soochow University. Human tissues, including the glioma tissues and surrounding normal brain tissues, were described previously [26]. Tissues were obtained from The First, Second, Third Affiliated and Children Hospitals of Soochow University. All participates provided written-informed consent.
Primary culture of human cancer cells. The source and culturing of primary human glioma cells were previously described [26]. From a lung cancer patient with brain metastases (female, 50 years old, brain metastatic lung papillary adenocarcinoma, CK7 + , CK20 -, Villin -, TTF-1 + , CDX2 -, PAX8 -, CD10 -, CAIX -, PR -, EMA + , GFAP -), surgical removed brain metastatic cancer tissues were washed, minced into small pieces, and digested with Collagenase I and DNase (Sigma-Aldrich). Single-cell suspensions were then pelleted and washed. Fibroblasts, blood vessel cells, immune cells and other non-cancerous cells were abandoned. Purified brain-metastatic human lung cancer cells (bmLCs) were ex vivo cultured in the described medium [31]. The study protocols were reviewed and approved by the Ethics Committee of Soochow University, and conformed to the guidelines of Helsinki declaration. The informed consent was obtained from all subjects before their participation.
Quantitative real-time reverse transcriptase polymerase chain reaction (qPCR). As described, qPCR was carried out by an ABI 7600 Prism equipment through utilizing a SYBR Green PCR kit. Gαi3 and NLGN3 mRNA levels were quantified by a ΔΔ Ct protocol [32], using GAPDH as the internal control [26]. Primers are listed in Table 1. Table 1. Primers and sgRNA sequences utilized in the study.

Genes qPCR primers
Western blotting and co-immunoprecipitation. Protocols for Western blotting and co-immunoprecipitation (co-IP), as well as data quantification, have been extensively described previously [25,28]. For all Western blotting assays, each lane was loaded with exact same amount of quantified protein lysates (30-40 μg in each treatment). Same set of lysate samples were run in parallel ("sister") gels to test different proteins when necessary. Expression of indicated proteins was quantified via ImageJ software, with results normalized to the equal loadings.
Endosome fractions. Cells with the applied treatments were harvested and re-suspended in the hypotonic swelling buffer [33], and lysed with 30 strokes in a Dounce homogenizer using a tight pestle, and swelling was stopped by the addition of two fold homogenization buffer [33]. Lysates were centrifuged to obtain the post-nuclear supernatants, and centrifuged [33]. The resulting supernatants were centrifuged, and the pellet solubilized in the homogenization buffer [33]. Insoluble particles were removed by short centrifugation and the supernatant loaded onto a 5-20% continuous OptiprepTM (Sigma-Aldrich), poured using homogenization buffer. The gradient was further centrifuged at 60,000 g for 24h, with total 10 endosomal fractions collected, and proteins precipitated with 12% TCA for 1h. Fractions were centrifuged at 12,000 g for 1h. The protein pellets, combining all ten endosomal fractions, were dissolved in SDS-sample buffer for analysis by Western blotting.
CRISPR/Cas9 knockout of Gαi1 and Gαi3. The lentiviral CRISPR/Cas-9 Gαi1 KO construct and the lentiviral CRISPR/Cas-9 Gαi3 KO construct were provided by Shanghai Genechem (Shanghai, China), transfected into MEFs, and selected with puromycin. Gαi1/3 knockout was confirmed by Western blotting. Control cells were treated with the empty vector with nonsense sgRNA (Santa Cruz Biotech). The sgRNA sequences used for Gαi1 and Gαi3 KO are listed in Table 1.
The orthotopic primary-derived xenograft assay. For intracranial tumor implantation, primary human glioma cells or bmLCs (5 × 10 5 cells of each mouse in 200 µL of Matrigel gel/10% FBS medium, with different genetic treatments) were implanted using the previously described coordinates [26,38]. Magnetic resonance imaging (MRI) was carried out to visualize tumor xenografts. On the day when the first mouse in any group exhibited symptoms (severe fever, vomiting, or greater than 15% body weight loss), all groups were sacrificed and tumors isolated through surgery. Tumor volumes were measured by the formula: π/6 × larger diameter × (smaller diameter) 2 . Immunohistochemistry (IHC) was performed using the previously described procedures [26]. All animal procedures were approved by Soochow University Ethics Review Board.
Statistical analysis. The in vitro experiments were replicated three times or more, and data expressed as means ± standard deviation (SD). To examine statistical differences among different groups one-way ANOVA was carried out with multiple comparisons performed by post hoc Bonferroni test (SPSS 18.0). P values < 0.05 were considered statistically significant. A two-tailed unpaired t test (Excel 2007) was applied to examine significance between two treatment groups.

Gαi1 and Gαi3 are required for NLGN3-induced Akt, Erk and mTORC1 activation in MEFs
To further confirm that Gαi1/3 are required for NLGN3-induced signaling the shRNA method was utilized. As described previously [29], Gαi1 shRNA lentiviral particles and/or Gαi3 shRNA lentiviral particles were added to WT MEFs to silence Gαi1 and/or Gαi3. As shown, shRNA-mediated single knockdown of Gαi1 or Gαi3 resulted in partial inhibition of Akt, GSK3α/β and Erk1/2 phosphorylation in response to NLGN3 (P < 0.05 vs. MEFs with control shRNA, Figure 2A). Significantly, Gαi1 and Gαi3 double knockdown by targeted shRNAs ("Gαi1/3 DshRNA") led to further inhibition of NLGN3-induced signaling in MEFs (Figure 2A).

Gαi1 and Gαi3 are required for orthotopic growth of primary glioma xenografts in mouse brain
We performed orthotopic primary human glioma xenograft experiments using a previouslydescribed protocol [26]. P1 glioma cells, derived from a GBM patient (see MRI diagnosis in Figure 6A), were intracranially (using the described parameters [38]) injected into brains of nude mice. At experimental day-26, when the first mouse in the control group exhibited obvious symptoms, all the mice were sacrificed and tumors isolated [26]. MRI imaging results confirmed that the growth of orthotopic primary glioma xenografts ( Figure 6B, yellow boxes) was significantly inhibited when expressing Gαi1 shRNA plus Gαi3 shRNA ("Gαi1/3 DshRNA"). Gαi1 The P1 primary glioma cells or U251MG cells, with Gαi1 shRNA plus Gαi3 shRNA ("Gαi1/3 DshRNA") or the scramble control shRNA ("scr-shRNA"), were treated with or without NLGN3 (50 ng/mL) for applied time periods, cell proliferation (A and E), growth (B), and migration (C and E) were examined by the assays mentioned in the text, with Gαi1 and Gαi3 expression tested as well (D, for P1 glioma cells). P1 glioma cells, with the adenovirus Gαi1 construct plus the adenovirus Gαi3 construct ("OE-Gαi1/3", two lines: "sL1/sL2") (F-I), were treated with or without NLGN3 (50 ng/mL) for applied time period, cell proliferation (F), growth (G), migration (H) and Gαi1/3 expression (I) were tested, with results quantified. For the EdU staining assay, ten random views were included to calculate EdU/DAPI ratios. For the "Transwell" assays, ten random views of each condition were included to calculate the average number of migrated cells. For all functional assays exact same number of viable cells of different genetic treatments were initially seeded into each well/dish (at 0h). "C" stands for medium control. Blotting data was repeated five times (D and I). *P < 0.05 (A-E). *P < 0.05 vs. "Ad-Vec" cells (F-H). "n.s." stands for non-statistical difference (A-C). Scale bar=100 μm (A and C). and Gαi3 double silencing resulted in an 80% reduction of orthotopic glioma xenograft growth (Figure 6C), and was significantly more effective than Gαi1 single silencing (50-60% inhibition, see our previous study [26]). Mouse body weights were equivalent between the Gαi1/3 DshRNA and control shRNA groups (Figure 6D). Gαi1/3 protein expression levels were significantly decreased in orthotopic primary xenografts with Gαi1/3 DshRNA, and levels of p-Akt, p-S6K and p-Erk1/2 inhibited (Figure 6E). Immunohistochemistry (IHC) results, shown in Figure 6F, further confirmed reductions in p-Akt and p-S6 in Gαi1/3-silenced orthotopic tumors (vs. control tumors). Therefore Gαi1/3 silencing significantly inhibited orthotopic growth of primary glioma xenografts.
In contrast, ectopic Gαi1/3 overexpression promoted orthotopic growth of primary glioma xenografts. Primary glioma cells transduced with Ad-Gαi1 plus Ad-Gαi3 ("OE-Gαi1/3") or the vector control ("Ad-Vec") were injected into the brains of the nude mice. At experimental day-20, when the first mouse in the OE-Gαi1/3 group exhibited obvious symptoms, all mice were sacrificed and the tumors isolated. MRI images showed that Gαi1/3overexpressing primary glioma xenografts grew faster than the control tumors (with Ad-Vec) ( Figure  6G), presenting with higher tumor volumes ( Figure  6H). Mouse body weights were unchanged ( Figure  6I). Western blotting of tumor lysates confirmed Gαi1/3 overexpression in OE-Gαi1/3-tumor tissues, where levels of p-Akt, p-S6K and p-Erk1/2 increased (Figure 6J). Taken together, Gαi1 and Gαi3 are required for orthotopic primary glioma xenograft growth in mouse brain.

Orthotopic growth of brain-metastatic human lung cancer cells requires Gαi1 and Gαi3
Next we tested whether Gαi1 and Gαi3 were also required for the growth of metastatic brain tumor. Brain-metastatic lung cancer cells, bmLCs, from a lung cancer patient with brain metastasis (see the MRI image of brain metastatic tumor, Figure 7A) were obtained and primary-cultured. The bmLCs were Figure 6. Gαi1 and Gαi3 are required for orthotopic growth of primary glioma xenografts in mouse brain. Brain MRI images of a primary glioma patient providing primary human glioma cells ("P1" cells, A). The exact same amount of glioma cells (5 × 10 5 cells of each mouse), expressing Gαi1 shRNA plus Gαi3 shRNA ("Gαi1/3 DshRNA"), scramble non-sense shRNA ("scr-shRNA"), the adenovirus Gαi1 plus Gαi3 constructs ("OE-Gαi1/3"), or the empty vector ("Ad-Vec"), were intracranially injected to brains of nude mice (5-6 week old), after 26/20 days, representative MRI images of orthotopic glioma xenografts were presented (B and G); Animals were decapitated and tumors were isolated by surgery, tumor volumes (C and H) and mice body weights (D and I) were recorded. Tumor tissue lysates were tested by Western blot assay of listed proteins (E and J). Immunohistochemistry (IHC) images of p-Akt and p-S6K were shown (F). Western blotting quantifications were from five replicate blot data. IHC experiments were repeated in three pairs of tissues. # P < 0.05. "n.s." stands for non-statistical difference (D and I). Scale bar=100 μm (F).
Thereafter, cells were intracranially injected into brains of nude mice. At experimental day-25, when the first mouse in the control group exhibited obvious symptoms, all the mice were sacrificed. MRI image results confirmed that orthotopic bmLCs xenografts bearing Gαi1/3 DshRNA grew significantly slower than control bmLCs xenografts with scr-shRNA ( Figure 7B). Gαi1 and Gαi3 double shRNA led to over 80% reduction of orthotopic bmLCs xenograft growth ( Figure 7C). The mice body weights were indifferent between the two groups ( Figure 7D). Next, signaling proteins in orthotopic bmLCs xenografts were examined. Western blotting assay results, Figure 7E, confirmed Gαi1 and Gαi3 silencing in bmLCs xenografts with Gαi1/3 DshRNA, while Gαi2 expression was unchanged ( Figure 7E). Importantly, phosphorylations of Akt, S6K and Erk1/2 were largely decreased in Gαi1/3 DshRNA-expressing orthotopic bmLCs xenografts (Figure 7E). IHC images further confirmed decreased Akt-S6K phosphorylations in orthotopic bmLCs xenografts with Gαi1/3 DshRNA (Figure 7F). These results showed that silencing of Gαi1 and Gαi3 inhibited orthotopic growth of bmLCs in mouse brain. Figure 7. Orthotopic growth of brain-metastatic human lung cancer cells requires Gαi1 and Gαi3. Brain MRI images of a primary lung cancer patient with brain metastasis (A) providing the brain-metastatic human lung cancer cells ("bmLCs"). bmLCs (5 × 10 5 cells of each mouse), expressing Gαi1 shRNA plus Gαi3 shRNA ("Gαi1/3 DshRNA"), scramble non-sense shRNA ("scr-shRNA"), the adenovirus Gαi1 plus Gαi3 constructs ("OE-Gαi1/3"), or the empty vector ("Ad-Vec"), were intracranially injected to brains of nude mice (5-6 week old); After 25/17 days, representative MRI images of orthotopic bmLCs xenografts were presented (B and G); Animals were decapitated and tumors were isolated by surgery, tumor volumes (C and H) and mice body weights (D and I) were recorded. The orthotopic bmLCs xenograft tissue lysates were tested by Western blot assay of listed proteins (E and J). Immunohistochemistry (IHC) images of p-Akt and p-S6K were shown (F). Western blotting quantifications were from five replicate blot data. IHC experiments were repeated in three pairs of tissues. # P < 0.05. "n.s." stands for non-statistical difference (D and I). Scale bar=100 μm (F).
To further support out hypothesis, primary bmLCs, stably transduced with Ad-Gαi1 plus Ad-Gαi3 ("OE-Gαi1/3"), as well as the control cells with the vector control ("Ad-Vec"), were injected into the brains of the nude mice. At day-17, the first mice in the OE-Gαi1/3 group exhibited obvious neurological symptoms. MRI images showed that the orthotopic bmLCs xenografts with OE-Gαi1/3 were larger than the control xenografts ( Figure 7G). Tumor volumes of OE-Gαi1/3 bmLCs xenografts were significantly higher than those of Ad-Vec xenografts ( Figure 7H). The mice body weights were indifferent ( Figure 7I). Western blotting assaying of orthotopic bmLCs xenograft tissues confirmed Gαi1/3 overexpression as well as increased phosphorylations of Akt-S6K1 and Erk1/2 in OE-Gαi1/3 xenograft tissues ( Figure 7J). Together, these results show the orthotopic growth of bmLCs requires Gαi1 and Gαi3.

Gαi3 upregulation in human glioma tissues
We have previously shown that Gαi1 mRNA and protein levels are upregulated in human glioma, correlating with high tumor grade [26]. To examine Gαi3 expression in human glioma, we first consulted the TCGA database to examine RNA-Seq data. As shown, in human glioma tissues Gαi3 mRNA expression is significantly upregulated [P <0.05 vs. normal tissues] (Figure 8A). The overall survival of the low grade glioma (LGG) patients with Gαi3-high gliomas is significantly lower than those with Gαi3-high gliomas (P < 0.001, Figure 8B). There is however no significant difference in the overall survival of Gαi3-low GBM patients and Gαi3-high GBM patients (Figure 8B). This could be due to the extremely low overall survival of GBM patients ( Figure 8B). When combining all glioma data, we showed that the average survival of Gαi3-high glioma patients is significantly lower than those with Gαi3-low gliomas (P < 0.001, Figure 8B). These results indicate that Gαi3 upregulation in human glioma is correlated with a poor overall survival.
To confirm the significance of the bioinformatics observations, we examined Gαi3 expression in human glioma specimens ("T") and surrounding normal brain ("N") tissues, from sixteen (16) low-grade (Grade I-II) tumors and another sixteen high-grade (Grade III-IV) tumors [26]. qPCR results, Figure 8C, demonstrated that Gαi3 mRNA levels are significantly upregulated in human glioma tissue (vs. normal brain tissue, Figure 8C). In late-stage gliomas (Grade III-IV, n=16) Gαi3 mRNA is more highly upregulated than in early-stage gliomas (Grade I-II, n=16) ( Figure 8C). In the late-stage gliomas (Grade III-IV, n=16) Gαi3 mRNA is highly upregulated than that in early-stage gliomas (Grade I-II, n=16) ( Figure 8C). Western blotting quantification results confirmed Gαi3 protein upregulation in human glioma tissues (Figure 8D), being more significant in the high-grade tumors ( Figure 8D). Human glioma tissues ("T", 16 high-grade and 16 low-grade) and the paired normal brain tissues ("N") were homogenized and dissolved in the tissue lysis buffer, Gαi3 mRNA and protein levels were tested by qPCR (C) and Western blotting (D), respectively, with results quantified. NLGN3 mRNA was tested by qPCR (E). * P < 0.05 vs."N" group. # P < 0.05.
In line with previous studies [18,19,40], NLGN3 mRNA levels were increased in human glioma tissues (Figure 8E), and levels were higher in the late-stage gliomas ( Figure 8E). TCGA cohorts showed that NLGN3 transcripts are higher in glioma tissues when compared to normal brain tissues (Figure 8F), being significant in LGG tissues ( Figure 8F). Therefore, Gαi3, similar to Gαi1 [26], is upregulated in human glioma tissues, correlating with poor survival, high tumor grade and NLGN3 upregulation.
Sustained and constitutive activation of multiple key RTKs is a primary driving force for glioma cell progression [53,54], causing hyperactivation of downstream mitogenic pathways, including PI3K-Akt-mTOR and Erk-MAPK [54,55]. There are reports of GPCR-RTKs crosstalk in glioma cells. For example, Huang et al., found that activation of the formylpeptide receptor induced trans-activation of EGFR via Gαi proteins, exacerbating malignant behaviors in glioblastoma cells [56]. Ghosh et al., found that Gαi-GIV (a novel non-receptor GEF for Gαi [57]) complex directly associated with and activated EGFR to promote cancer cell migration and proliferation [57]. Studies have shown that NLGN3 induces phosphorylations of multiple RTKs as well as downstream PI3K-Akt-mTOR and Erk-MAPK cascades, responsible for glioma cell growth and proliferation [19]. We found that Gαi1/3 proteins associate with activated RTKs (VEGFR and TrkB ( [28,29]) and are required for their endocytosis and downstream signaling. Here in glioma cells, Gαi1/3 silencing, by targeted shRNAs, blocked NLGN3induced RTKs endosomal translocation and subsequent Gab1 activation, without affecting RTK phosphorylation and expression. Importantly, Gab1 silencing inhibited NLGN3-induced downstream signaling activation in glioma cells. These results suggest that Gαi1 and Gαi3 mediate NLGN3-induced signal transduction by promoting RTK endocytosis and adaptor protein Gab1 activation.
Functional studies demonstrate that Gαi1/3 are essential in mediating NLGN3-induced glioma cell progression in vitro and orthotopic primary glioma xenograft growth in vivo.NLGN3-stimulated glioma cell growth, proliferation and migration were largely inhibited by Gαi1/3 silencing, but enhanced by ectopic Gαi1/3 overexpression. Glioma xenografts are unable to grow in NLGN3 KO brain suggesting an essential dependency on NLGN3 for orthotopic glioma growth [28,29]. Significantly, our results show that the growth of orthotopic primary glioma xenografts was largely inhibited with Gαi1/3 silencing. Conversely the Gαi1/3-overexpressed primary glioma xenografts grew significantly faster than the control tumors. We concluded that possibly by mediating NLGN3-induced signaling transduction, Gαi1/3 are essential for glioma cell progression in vitro and in vivo.
One important finding of this study is that Gαi1 and Gαi3 are required for the orthotopic growth of bmLCs in mouse brain. One possibility is that neuron-secreted NLGN3 is also important for orthotopic growth of bmLCs. Indeed, ex vivo cultured bmLCs were responsible to NLGN3. NLGN3 activated Akt-mTOR and Erk1/2 signalings, and promoted bmLCs proliferation and migration in vitro. Such actions were almost abolished with Gαi1/3 DshRNA. Importantly, orthotopic growth of bmLCs in mouse brain was largely inhibited with Gαi1/3 DshRNA, but augmented with ectopic Gαi1/3 overexpression. In addition, Akt-mTOR and Erk1/2 activation in orthotopic bmLCs xenograft tissues was inhibited with Gαi1/3 DshRNA, but enhanced with Gαi1/3 overexpression. Other mitogens, including the BDNF and VEGF, could also be important contributors for the growth of brain-metastatic cancer cells [18]. Interestingly, we have previously shown that BDNF-and VEGF-induced mitogenic signalings rely on Gαi1/3 [28,29]. This could also demonstrate the essential role of Gαi1/3 in mediating orthotopic growth of bmLCs in mouse brain.
NLGN3 is a crucial factor for neuron-glioma synapse formation and neuronal activity-dependent glioma progression [14,15,20]. Interestingly, Hanahan and his colleagues have shown that brain metastatic breast-cancer cells can form a specialized type of pseudo-tripartite synapse with glutamatergic neurons [58]. This will allow cancer cells to take up glutamate and activate GluN2B-mediated NMDAR (N-methyl-D-aspartate receptor) signaling, required for metastatic cancer cell growth [58]. Thus, another possibility is that bmLCs might form synapses with neurons to activate glutamate signaling, and neuron-secreted NLGN3 could be essential for the process. The underlying mechanisms warrant further characterizations.
NLGN3 expression levels negatively correlate with overall survival of glioblastoma patients [18,19]. Furthermore high NLGN3 expression is associated with glioma recurrence [40]. Our current and previous results [26] have demonstrated that Gαi1 and Gαi3 levels are significantly upregulated in human glioma tissues, and their upregulation is correlated with poor patients survival, high tumor grade and NLGN3 upregulation. Gliomas, especially the high grade GBM, are composed of a pathologically heterogeneous mixture of cells exhibiting different cellular and nuclear polymorphism [59][60][61]. Therefore, Gαi1/3 and NLGN3 upregulation in glioma tissues might be derived from both glioma cells and possible other cells in the tumor bulk (immune cells, endothelial cells and cancer stem cells, etc). Although the underlying mechanisms of Gαi1/3-mediated NLGN3 signaling and glioma cell progression need further characterization, the results of the present and previous studies [26] strongly indicate that Gαi1 and Gαi3 could be valuable therapeutic targets for treating human glioma.

Conclusions
Gαi1/3 mediation of NLGN3-induced signaling is essential for glioma growth in vitro and in vivo.