Transcription Factor NFAT5 Promotes Glioblastoma Cell-driven Angiogenesis via SBF2-AS1/miR-338-3p-Mediated EGFL7 Expression Change

Glioblastoma (GBM) is the most aggressive primary intracranial tumor of adults and confers a poor prognosis due to high vascularization. Hence anti-angiogenic therapy has become a promising strategy for GBM treatment. In this study, the transcription factor nuclear factor of activated T-cells 5 (NFAT5) was significantly elevated in glioma samples and GBM cell lines, and positively correlated with glioma WHO grades. Knockdown of NFAT5 inhibited GBM cell-driven angiogenesis. Furthermore, long non-coding RNA SBF2 antisense RNA 1 (SBF2-AS1) was upregulated in glioma samples and knockdown of SBF2-AS1 impaired GBM-induced angiogenesis. Downregulation of NFAT5 decreased SBF2-AS1 expression at transcriptional level. In addition, knockdown of SBF2-AS1 repressed GBM cell-driven angiogenesis via enhancing the inhibitory effect of miR-338-3p on EGF like domain multiple 7 (EGFL7). In vivo study demonstrated that the combination of NFAT5 knockdown and SBF2-AS1 knockdown produced the smallest xenograft volume and the lowest microvessel density. NFAT5/SBF2-AS1/miR-338-3p/EGFL7 pathway may provide novel targets for glioma anti-angiogenic treatment.


INTRODUCTION
Glioblastoma (GBM) harbors the highest malignancy of glioma. Despite improvements in therapeutic management, the overall prognosis has remained dismal, with a median survival ranging from 12 to 15 months (Wen and Kesari, 2008). GBM is characterized by its profound vascularization. Deregulated pro-angiogenic factors were produced by GBM cells and secreted into the tumor microenvironment, which stimulates tumor endothelial cell-dependent angiogenesis. Neovascularity brings essential oxygen and nutrition as well as promotes GBM malignancy (Jain et al., 2007). Previous study manifested that microvessel density predicts poor prognosis of glioma patients (Leon et al., 1996) which draws growing attention to anti-angiogenic therapy as an effective therapeutical target of glioma.
Transcription factor nuclear factor of activated T-cells 5 (NFAT5) was originally identified for its involvement in hypertonic kidney inner medulla adaptation (Woo et al., 2002;Lopez-Rodriguez et al., 2004). Accumulated findings have indicated that the expression levels of NFAT5 were aberrant in tumors (Küper et al., 2014;Qin et al., 2017). Moreover, growing publications demonstrated its potential roles in tumor angiogenesis and pro-angiogenic factors regulation Amara et al., 2016). But the expression levels of NFAT5 in gliomas and its potential role in GBM angiogenesis still remain unclear.
In this study, the expression levels of NFAT5 and SBF2-AS1 were investigated in glioma samples and GBM cell lines. In addition, the roles of NFAT5 and SBF2-AS1 in GBM cell-driven angiogenesis were further explored. Furthermore, NFAT5/SBF2-AS1/miR-338-3p/EGFL7 crosstalk in GBM angiogenesis was revealed. Findings in this study may serve as a potential target for glioma treatment.

Clinical Sample
A total of 47 cases paraffin-embedded glioma and five cases normal brain tissues (NBTs) were used for the NFAT5 immunohistochemistry staining. A total of 19 liquid nitrogenstored glioma samples and 5 NBTs were used for NFAT5 Western blot analysis and SBF2-AS1 quantitative real-time PCR analysis. All specimens were obtained from the Department of Neurosurgery, Shengjing Hospital of China Medical University. NBTs were the rejected material from surgeries of brain trauma and epilepsy. Glioma specimens had confirmed pathological diagnosis and were classified according to the World Health Organization (WHO) criteria by two experienced clinical pathologists in a blinded manner. For the use of the above clinical materials for research purposes, approval from the Hospital Ethical Committee was obtained.

Immunohistochemistry
All paraffin-embedded specimens were sliced into serial 4 µm sections and sections were labeled with primary antibodies against human NFAT5 (1:100; ab3446, Abcam, Cambridge, UK), followed by incubation with biotinylated secondary antibody included in an immunohistochemical labeling kit MaxVision,Fu Zhou,China). The NFAT5 expression was scored according to the proportion of positive cells and the staining intensity by two independent investigators who were blinded to tumor grade. The proportion of positively stained tumor cells was graded for 0 (<10% positive tumor cells), 1(10-50% positive tumor cells), 2 (50-90% positive tumor cells), and 3 (>90% positive tumor cells). The intensity of staining were scored 0 for no staining, 1 for weak staining, 2 for moderate staining, and 3 for strong staining. A combined staining index was calculated by multiplying the proportion of positive staining and the intensity of staining. The stained sections were defined as high expression (staining index>4) or low expression (staining index≤4).

Cell Culture and Preparation for Glioblastoma (GBM) Cell-Conditioned Medium (GCM)
Human GBM cell lines U87, U118, and human embryonic kidney 293T (HEK293T) cells were purchased from the Shanghai Institutes for Biological Sciences Cell Resource Center (Shanghai, China). Normal human astrocytes (NHA) were obtained from Sciencell Research Laboratories (Carlsbad, CA, USA) and cultured in astrocyte medium (Carlsbad, CA, USA). The human brain microvessel endothelial cell (ECs) line was gifted Dr Couraud (Institute Cochin, Paris, France). U87, U118, and HEK293T cells were cultured in Dulbecco's modified Eagle medium of high glucose supplemented with 10% fetal bovine serum. ECs were cultured as described previously (Guo et al., 2014). All cells were maintained in a humidified incubator at 37 • C with 5% CO 2 . For the exposure of cells to hypoxia, U87 and U118 cells were incubated in a hypoxic chamber containing 0.3% O 2 , 5% CO 2 , and 94.7% N 2 . GBM cell-conditioned medium was prepared as described previously (Cai et al., 2017).

RNA Extraction and Quantitative Reverse Transcription-PCR (qRT-PCR)
Total RNA was isolated with Trizol (Life Technologies Corporation). TaqMan PCR (TaqMan MicroRNA Reverse Transcription kit and Taqman Universal Master Mix II, Applied Biosystems) and SYBR Green quantitative PCR (One-Step SYBR PrimeScript RT-PCR Kit, Takara, Dalian, China) were carried out in at least triplicate for the target genes. Expression levels of target genes were determined using the 2 − Ct method and normalized to GAPDH. Expression levels of miR-338-3p were normalized to U6. Probes for TaqMan PCR assays and primer sets for SYBR Green assays are listed in Table S1.

Cell Viability Analysis
ECs cell viability was determined by the Cell Counting kit-8 (CCK-8) assay (Beyotime Institute of Biotechnology, Jiangsu, China). Cells (2 × 10 3 ) were seeded in 96-well plates in triplicate and incubated in GBM cell-conditioned medium for 24 h. Each well was incubated with 10 µl CCK-8 for 2 h, and the absorbance was measured at 450 nm using a spectrophotometer. All experiments were performed at least three times.

Cell Migration Assay
ECs migration ability was assessed using 6.5-mm transwell chambers with a pore size of 8 µm (#3422 Costar, Corning, NY, USA). Cells (2 × 10 5 ) were suspended in serum-free medium and were seeded into the upper chamber. The lower chamber was filled with 600 µl GBM cell-conditioned medium. Migrated cells were fixed, stained and counted under a microscope. Five randomly chosen fields were counted for each well, and the average cell number was determined. The experiments were performed at least in triplicate.

Tube Formation Assay
Prechilled 96-well plates were coated with 100 µl Matrigel (BD Biosciences, Bedford, MA, USA). The plates were then incubated at 37 • C for 30 min to allow the matrix to solidify. ECs were suspended in GBM cell-conditioned medium and 4 × 10 4 cells were added to the plates, followed by incubation at 37 • C for 6 h. Tube counts were quantified by Chemi Imager 5500 V2.03 software (Alpha Innotech, San Leondro, CA, USA).

Enzyme-Linked Immunosorbent Assay (ELISA)
EGFL7 levels in glioma-conditioned medium were measured with ELISA. Briefly, 96-well microplates were coated with 100 µl capture antibody (2 µg/ml) against EGFL7 (Santa Cruz Biotechnology), incubated overnight at 4 • C. The wells were then blocked with 5% BSA in PBS for 1 h at room temperature. Then samples were applied to wells and incubated for 1 h at room temperature. After washed with PBS with 0.1% Tween-20, horseradish peroxidase conjugated goat anti-mouse immunoglobulin G was added to each well and incubated for 1 h at room temperature. Subsequently, tetramethylbenzidine (TMB) solution was added to the wells, then reaction was stopped by applying sulfuric acid. The absorbance was then measured at 450 nm on a spectrophotometer.

Chromatin Immunoprecipitation (ChIP) Assay
U118 chromatin were incubated with 3 µg anti-NFAT5 (ab3446, Abcam). Purified immunoprecipitated DNA was prepared for the PCR with the specific primers. The primers used for PCR are presented in Table S3.

Nude Mouse Xenograft Model
All animal experiments were complied with the guidelines of the Animal Welfare Act and were reviewed and approved by the Ethics Committee of Shengjing Hospital. Four weeks old female BALB/C nude mice were purchased from Cancer Institute of the Chinese Academy of Medical Science. Mice were randomly divided into each double-blind group by two performers, n = 7 per group. A suspension of 5 × 10 5 cells in a 100 µl volume was injected subcutaneously into the right flank of the mice. Tumor nodules were estimated with caliper at a 5-day interval. Mice were scarified on the 45th day after injection, and the tumors were excised and collected for the CD31 staining. Tumor volumes = (length × width 2 )/2.

Microvessel Density (MVD) Assay
MVD of xenograft gliomas were determined by the immunohistochemical staining of endothelial marker CD31 and evaluated according to the previous description (Weidner et al., 1991;Cai et al., 2017). Areas of highest vascularization were selected by scanning the tumor sections at low magnification. Stained microvessels were counted in a single 200× field within the selected filed by three observers without previous knowledge of tumor groups. The following cellular structures were considered as countable microvessel: (a) stained lumen, (b) stained endothelial cell, (c) stained endothelial cell cluster (a and b are clearly separated from adjacent strained lumens, tumor cells and other connective tissue elements). The MVD value was calculated as the average vessel counts in three selected areas within a microscopic field.

Statistical Analysis
All statistical analyses were performed with GraphPad Prism 5 (GraphPad Software, La Jolla, CA, USA). All values are presented as the mean± standard deviation (SD) from at least three independent experiments. Student's t-test was used for comparisons between two groups. One-way ANOVA was used for multi-group comparisons followed by Bonferroni post-hoc test. A chi-square test was applied to determine the association of NFAT5 levels with WHO grades. The difference was considered statistically significant when P < 0.05.

NFAT5 Expression Positively Correlated with Glioma Grade
Immunohistochemistry results showed that NBTs presented weak NFAT5 staining. However, the proportion of NFAT5positive cells and the staining intensity were higher in glioma sections ( Figure 1A) and presented a tumor pathological grade-dependent pattern ( Figure 1B). Consistent with the immunohistochemistry results, Western blot assay showed that NFAT5 levels were elevated in glioma tissues and were correlated with the pathological grade ( Figure 1C). In addition, compared with NHA, significant upregulation of NFAT5 was observed in U87 and U118 glioblastoma (GBM) cell lines ( Figure 1D). It is FIGURE 1 | NFAT5 was upregulated in glioma tissues and GBM cell lines. (A) Representative patterns of NFAT5 expression levels that were determined by immunohistochemistry staining in glioma tissues and normal brain tissues. A1-A2 normal brain tissues (n = 5), B1-B2 grade I gliomas (n = 8), C1-C2 grade II gliomas (n = 10), D1-D2 grade III gliomas (n = 14), E1-E2 grade IV gliomas (n = 15). Original magnification, × 100 in A1-E1, × 400 in A2-E2. (B) Association of NFAT5 expression levels with WHO grade. P value was estimated by chi-square test. (C) Representative patterns of NFAT5 expression levels that were detected by Western blot in glioma tissues (Grades I-II, 7 cases; Grades III-IV, 12 cases) and NBTs (5 cases). Data represent mean ± s.d. (each case was determined by Western blot for three times), # P and *P < 0.05, ## P < 0.01. (D) Expression of NFAT5 was detected by Western blot in normal human astrocytes, U87 and U118 cell lines. Data represent mean ± s.d. (n = 3, each). *P < 0.05.
Frontiers in Molecular Neuroscience | www.frontiersin.org well-established that hypoxia plays an important role for GBM angiogenesis. Therefore, we tested the NFAT5 levels under the hypoxic conditions. As shown in Figure S1 NFAT5 levels were upregulated in U87 and U118 cells under hypoxic conditions compared with under normoxic conditions.

Knockdown of NFAT5 Suppressed GBM Cell-Driven Angiogenesis in Vitro
Since GBM cells have a strong capacity for stimulating angiogenesis, we used GBM cell-conditioned medium as a means to induce angiogenesis in human brain microvessel endothelial cells to determine in vitro angiogenesis ability as previous description (Wurdinger et al., 2008). In order to characterize the influence of NFAT5 on GBM cell-driven angiogenesis, stable NFAT5 knockdown U87 and U118 cells were conducted. Then, effects of GBM cell-conditioned medium on endothelial cells (ECs) were tested. As shown in Figure 2A, there was no significant difference in ECs cell viability between the control group and the NFAT5 knockdown negative control (NFAT5(-)NC) group. However, cell viability was significantly decreased in the NFAT5 knockdown (NFAT5(-)) group compared with that in the NFAT5(-)NC group. Similarly, knockdown of NFAT5 inhibited the migration ( Figure 2B) and tube formation (Figure 2C) of ECs. The above data indicated that downregulation of NFAT5 suppressed GBM cell-driven angiogenesis in vitro.

SBF2-AS1 Was Upregulated in Gliomas and Knockdown of SBF2-AS1 Inhibited GBM Cell-Driven Angiogenesis in Vitro
Long non-coding RNAs (lncRNAs) are important mediators for tumor angiogenesis. SBF2-AS1 is an lncRNA that was proved to be upregulated in non-small cell lung cancer (Lv et al., 2016). Expression levels of SBF2-AS1 in gliomas and its potential roles in GBM cell-driven angiogenesis were unclear. As shown in Figure 3A, SBF2-AS1 was upregulated in glioma tissues and was positively correlated with pathological grade. In addition, compared with in NHA, SBF2-AS1 levels were increased in U87 and U118 cells ( Figure 3B). To clarify the potential role of SBF2-AS1 in GBM cell-driven angiogenesis, we next examined the effects of SBF2-AS1 silencing on GBM cell-driven angiogenesis. As shown in Figures 3C-E, ECs were  Expression levels of SBF2-AS1 were detected by qRT-PCR in glioma tissues (Grades I-II, seven cases; Grade III, six cases; Grade IV, six cases) and NBTs (five cases). Data represent mean ± s.d. (n = 3, each), # P and *P < 0.05,**P, && P, and ## P < 0.01. (B) Expression levels of SBF2-AS1 were detected by qRT-PCR in normal human astrocytes, U87 and U118 cell lines. Data represent mean ± s.d. (n = 4), ** P < 0.01. SBF2-AS1 was knockdown in U87 and U118 cells. Effects of GBM cell-conditioned medium on ECs were detected. (C) ECs cell viability was measured by CCK-8 assay. (D) ECs migration was measured by Transwell assay. (E) ECs tube formation was measured by Matrigel tube formation assay. Data represent mean ± s.d. (n = 4, each). *P < 0.05, **P < 0.01. Scale bar represents 30 µm.
suppressed with supernatant from SBF2-AS1 knockdown GBM cells leading to a significant reduction of cell viability, migration, and tube formation. These results revealed that SBF2-AS1 inhibition impaired GBM cell-driven angiogenesis.
NFAT5 Regulated SBF2-AS1 at Transcriptional Level, and SBF2-AS1 Was Involved in NFAT5-Mediated GBM Cell-Driven Angiogenesis The correlation analysis of NFAT5 and SBF2-AS1 showed that SBF2-AS1 level was positively correlated with NFAT5 level (Figure 4A). Thus, we hypothesized that SBF2-AS1 might be involved in NFAT5-mediated GBM cell-driven angiogenesis. To confirm the hypothesis, we detected SBF2-AS1 levels in NFAT5 knockdown U87 and U118 cells by qRT-PCR. As shown in Figure 4B, knockdown of NFAT5 significantly decreased SBF2-AS1 levels in U87 and U118 cells. Transcription factors may bind to the specific region of promoters to regulated gene expression. TGGAAA is the consensus binding sequence of NFAT5 (Kultz and Csonka, 1999). Combined with scanning the SBF2-AS1 promoter region by using the JASPAR database (http:// jaspar.binf.ku.dk/), we found three putative NFAT5 binding sites ( Figure 4C). Then chromatin immunoprecipitation assay was performed to clarify the association of NFAT5 and SBF2-S1 promoter region. As shown in Figure 4C primers were designed to amplify the 2,000 bp upstream region of the putative NFAT5 binding site that was not expected to associate with NFAT5. PCR products were observed in the NFAT5 immunoprecipitation in the PCR3 group, which indicated that NFAT5 bound to SBF2-AS1 promoter at -1,453 to -1,448 bp region. The above data indicated that NFAT5 regulated SBF2-AS1 at transcriptional level. Moreover, the combination of NFAT5 and SBF2-AS1 knockdown enhanced the inhibitory effects on ECs cell viability ( Figure 5A), migration (Figure 5B), and tube formation ( Figure 5C), which were induced by NFAT5 knockdown alone.
The above findings indicated that SBF2-AS1 might be involved in NFAT5-mediated GBM cell-driven angiogenesis.
Having found that miR-338-3p was regulated by SBF2-AS1, we speculated that it might participate in SBF2-AS1-mediated GBM cell-driven angiogenesis. To test this, miR-338-3p was overexpressed or knocked down in GBM cells with SBF2-AS1 silencing. Then effects of the GBM cell supernatant on ECs were determined. As shown in Figure 6E, overexpression of miR-338-3p promoted the suppressive effect on EC cell viability which was induced by SBF2-AS1 knockdown. While downregulation of miR-338-3p reversed the effect induced by SBF2-AS1 inhibition. Similar effects were observed on ECs migration ( Figure 6F) and tube formation (Figure 6G). Collectively, these results supported an involvement of miR-338-3p in SBF2-AS1-mediated GBM cell-driven angiogenesis.
Knockdown of NFAT5 and SBF2-AS1 Inhibited EGFL7 Expression and Secretion without Affecting EGFL7 mRNA After confirming that EGFL7 was involved in miR-338-3pregulated GBM cell-driven angiogenesis, we hypothesized that EGFL7 was a mediator of NFAT5/SBF2-AS1-regualted GBM cell-driven angiogenesis. As shown in Figure 9A, EGFL7 expression levels were lower in the NFAT5(-) group than that in the NFAT5(-)NC group. In addition EGFL7 levels were decreased in the NFAT5(-) GBM cell-conditioned medium ( Figure 9B). However, there was no significant difference in the EGFL7 mRNA levels between the NFAT5(-)NC group and NFAT5(-) group ( Figure 9C). ChIP assay showed that NFAT5 did not associate with the putative binding site in EGFL7 promoter region ( Figure S2), which indicated that NFAT5 might regulate EGFL7 in an indirect way. Similarly, SBF2-AS1 inhibition decreased EGFL7 expression ( Figure 9D) and secretion ( Figure 9E) without affecting EGFL7 mRNA level ( Figure 9F). Moreover, the combination of NFAT5 and SBF2-AS1 knockdown enhanced the inhibitory effects induced by individual knockdown (Figures 9G,H). Also, EGFL7 mRNA levels were not changed ( Figure 9I).
CD31 immunohistochemistry staining was performed to determined the microvessel density of the xenograft gliomas. As shown in Figure 10C, downregulation of NFAT5 or SBF2-AS1 decreased the microvessel density of xenograft gliomas and NFAT5 knockdown combined with SBF2-AS1 knockdown presented the lowest microvessel density.

DISCUSSION
In this study, NFAT5 and SBF2-AS1 was highly expressed in glioma samples and positively associated with glioma WHO grades. Knockdown of NFAT5 in GBM cell lines inhibited tumor-driven angiogenesis. Similarly, downregulation of SBF2-AS1 impaired GBM cell-driven angiogenesis. NFAT5 positively regulated SBF2-AS1 via binding to its promoter region in a sequence-dependent manner. In addition, SBF2-AS1 sponged miR-338-3p to release the post-translational inhibition of EGFL7, which stimulating the cell viability, migration and tube formation of endothelial cells (Figure 11). Furthermore, knockdown of both NFAT5 and SBF2-AS1 inhibited the xenograft glioma growth and reduced the microvessel density.
NFAT5 is a member of the nuclear factor of activated T cells (NFAT) transcription factors. Recent studies have pointed to an important role for NFAT5 in modulating tumor biology. Küper et al. (2014) found that knockdown of NFAT5 is accompanied by a significant decrease in the proliferation and migration of renal carcinoma cells. Guo et al. (Guo and Jin, 2015) reported that NFAT5 inhibition impairs the proliferation and migration of lung adenocarcinoma cells by downregulating aquaporin-5. In breast cancer, NFAT5 was demonstrated to be a putative biomarker of inflammatory breast cancer phenotype (Remo et al., 2015). Moreover, the expression of NFAT5 promotes breast cancer cells migration (Jauliac et al., 2002) and tumor-driven angiogenesis . On the contrary, NFAT5 acts as a tumor suppressor in hepatocellular carcinoma (Qin et al., 2017). In this study, we found that NFAT5 was deregulated in GBM and knockdown of NFAT5 inhibited GBM cell-driven angiogenesis in in vitro and in vivo studies.
NFAT5 was shown to promote pro-angiogenic factor expression. Amara et al. (2016) clarified that NFAT5 interacts with STAT3, which stimulates VEGF-A expression via binding to its promoter region in breast cancer cells. Moreover, Veltmann et al. (2016) found that knockdown of NFAT5 reduces bFGF transcripts in retinal pigment epithelium cells, which indicates that bFGF gene were transcriptionally activated by NFAT5 under hyperosmotic conditions. In addition, Madonna et al. (2016) demonstrated that high glucose induces human aortic endothelial cells angiogenesis and COX-2 expression possibly through the activation of NFAT5. In this study, downregulation of NFAT5 in GBM cell lines blocked the expression and secretion of EGFL7 via the SBF2-AS1/miR-338-3p axis.
An increasing number of studies have showed that aberrant expressed lncRNAs are vital regulatory factors in GBM. In particular, lncRNA H19 , HULC (Zhu et al., 2016), and TUG1 (Cai et al., 2017) were demonstrated to regulate GBM angiogenesis. SBF2-AS1 is primarily associated with the poor prognosis in patients with non-small cell lung cancer (Zhao et al., 2016). In this study, we found that SBF2-AS1 was highly expressed in GBM and associated with GBM cell-driven angiogenesis. Mechanisms underlying the deregulated lncRNAs in tumors are under investigation. Zhang et al. (2017) uncovered that lncRNA Orilnc1 is regulated by RAS/RAF/MEK/ERK signaling via the transcriptional activation of AP1. Similar, Teng et al. (2016) reported that an increase in lncRNA HCP5 expression can be observed by the introduction of transcription factor RUNX1. In line with the above findings, our results clarified that NFAT5 blockade reduced the expression of SBF2-AS1, which possibly attributed to the absent of NFAT5 in SBF2-AS1 promoter region. Recent publications indicated that RNA binding proteins regulate lncRNAs expression via stabilizing them (Chai et al., 2016;Wang L. et al., 2017). Whether RNA binding proteins are involved in NFAT5-regulated SBF2-AS1 expression change deserves further study.
The mechanisms of lncRNAs in tumor biology have not been elucidated very clearly. The well-researched function pattern for lncRNAs is acting as miRNA sponges to release the posttranslational inhibition of target mRNAs. For instance, TUG1 maintains stemness features of glioma stem cells via sponging miR-145 to promote SOX2 and MYC expression (Katsushima et al., 2016). ZFAS1 was reported to promote hepatocellular carcinoma metastasis via sponging miR-150 to activate ZEB1, MMP14, and MMP16 . Consistent with the above findings, we demonstrated that SBF2-AS1 regulates EGFL7 may via sponging miR-338-3p. FIGURE 9 | Knockdown of NFAT5 and SBF2-AS1 inhibited EGFL7 protein level and secretion without affecting EGFL7 mRNA level. Effects of NFAT5 knockdown on EGFL7 protein level (A), secretion (B), and mRNA level (C). Effects of SBF2-AS1 knockdown on EGFL7 protein level (D), secretion (E), and mRNA level (F). Effects of combination knockdown of NFAT5 and SBF2-AS1 on EGFL7 protein level (G), secretion (H), and mRNA level (I). Data represent mean ± s.d. (n = 4). *P < 0.05, **P < 0.01. miR-338-3p was verified to be a tumor suppressor in gastric cancer (Guo et al., 2014), hepatocellular carcinoma (Huang et al., 2011), and so forth. Moreover, miR-338-3p exerts anti-angiogenic effect in hepatocellular carcinoma . In GBM, miR-338-3p was shown lowly expressed and restoration of miR-338-3p inhibited GBM cells malignant behaviors (Howe et al., 2017). In this study, our finding manifested that ectopic expression of miR-338-3p in FIGURE 10 | Combination of NFAT5 knockdown and SBF2-AS1 knockdown inhibited xenograft glioma growth and the microvessel density in vivo. The stable expressing U87 and U118 cells were used for the in vivo study. (A) The nude mice carrying tumors from respective groups and a sample tumor from each group were shown. (B) Xenograft glioma growth curves in nude mice. Tumor volume was measured every 5 days after injection. Data represent mean ± s.d. (n = 7, each group). (C) Tumors were harvested on day 45 and paraffin-embedded tumor sections were prepared. CD-31 immunohistochemistry staining was performed to determine the microvessel density (MVD). (magnification, × 400; scale bar = 50 µm). Data represent mean ± s.d. (n = 4, each group). *P, & P, and # P < 0.05, && P and **P < 0.01. GBM cells reduced GBM cell-driven angiogenesis. miRNAs regulate gene expression by degradation of target mRNAs or inhibition of translation (Bartel, 2004). In this study, we found that miR-338-3p reduced EGFL7 protein levels without affecting EGFL7 mRNA level, perhaps through the translation inhibition.
Previous studies have well-established the involvement of EGFL7 in endothelial cell behaviors. Knockdown of EGFL7 in zebrafish embryos specifically blocks vascular tubulogenesis (Parker et al., 2004). EGFL7 also acts as a chemoattractant for embryonic endothelial cells migration (Campagnolo et al., 2005). In addition, EGFL7 protects endothelial cells from hyperoxiainduced cell death (Xu et al., 2008). Recent evidence focuses on the pro-angiogenic role of EGFL7 in cancer. Shen et al. (2016) reported that EGFL7 is high expressed in pancreatic carcinoma and promotes cancer invasion and angiogenesis. Hu et al. (2016) demonstrated that miR-126 inhibited hepatocellular carcinoma angiogenesis through reducing EGFL7. In addition, ERK activation plays an important role in EGFL7-mediated pro-angiogenic effect (Takeuchi et al., 2014;Chim et al., 2015;Gong et al., 2017). Increased EGFL7 levels promote glioma angiogenesis and may predict poor prognosis of GBM patients (Wang F. Y. et al., 2017). Consistent with the above findings, we demonstrated that enhanced EGFL7 levels in GBM cells promoted ECs cell viability, migration, and tube formation via activating ERK.
GBM angiogenesis is a complicated process. Endothelial celldependent manner, which was focused on in this study, plays a predominant role (Zhang et al., 2015;Gilbert, 2016). Moreover, macrophages in the tumor microenvironment are key regulators of GBM angiognensis (Lu-Emerson et al., 2013;Zhu et al., 2017). Recent evidence indicated that vasculogenic mimicry, which is characterized as endothelium-independent tubular structure, may be a novel blood supply manner in malignant glioma (Yue and Chen, 2005;Liu et al., 2011). In addition, glioblastoma stem-like cells could differentiate into endothelial cells, which promotes the angiogenetic malignancy (Chroscinski et al., 2015). An interesting line of investigation would be to evaluate whether NFAT5 is involved in the above cellular process.

AUTHOR CONTRIBUTIONS
YL and YX conceived and designed this study. HY, XL and JZ performed the main experiments. LZ and ZL helped with the Immunohistochemistry assay. SS and LZ helped with the animal experiments. JZ, SS, and ZL collected the clinical data and analyzed the data. HY and XL drafted the manuscript and performed the literature review. YL and YX revised the article critically. All authors had final approval of the submitted versions.

FUNDING
This work is supported by grants from the Natural Science Foundation of China (Nos. 81672511 and 81573010), the Liaoning Science and Technology Plan Project (No. 2015225007), the Shenyang Science and Technology Plan Projects (Nos. F15-199-1-30 and F15-199-1-57), and the outstanding scientific fund of Shengjing hospital (No. 201304).