Knockdown of USF1 Inhibits the Vasculogenic Mimicry of Glioma Cells via Stimulating SNHG16/miR-212-3p and linc00667/miR-429 Axis.

The anti-angiogenic treatment of malignant glioma cells is an effective method to treat high-grade gliomas. However, due to the presence of vasculogenic mimicry (VM), the anti-angiogenic treatment of gliomas is not significantly effective in improving overall patient median survival. Therefore, this study investigated the mechanism of mimic formation of angiogenesis in gliomas. The results of this experiment indicate that the expression of upstream transcription factor 1 (USF1) is upregulated in glioma tissues and cells. USF1 knockdown inhibits the proliferation, migration, invasion, VM, and expression of VM-associated proteins in glioma cells by stimulating SNHG16 and linc00667. These two long non-coding RNAs (lncRNAs) regulate ALHD1A1 through the competing endogenous RNA (ceRNA) mechanism influencing the VM of glioma. This study is the first to demonstrate that the USF1/SNHG16/miR-212-3p/ALDH1A1 (aldehyde dehydrogenase-1) and USF1/linc00667/miR-429/ALDH1A1 axis regulates the VM of glioma cells, and these findings might provide a novel strategy for glioma treatment.

Introduction

Malignant gliomas account for approximately 60% of human malignant primary brain tumors. Despite the continuous improvement in different treatment modalities such as surgery, radiotherapy, and chemotherapy in recent years, the prognoses of glioma patients remain poor, with a median survival of approximately 15 months. Anti-angiogenic therapy is currently an effective glioma treatment that has received increasing attention, with it being widely applied and especially effective in the treatment of high-grade gliomas. However, there is yet to be any significant increase in survival rates. This may be because of the presence of vasculogenic mimicry (VM) in gliomas. VM is a tubular structure formed by malignant brain tumor cells that are pluripotent, plastic, and resemble a stem cell phenotype. With this basic structure, VM forms a special microcirculation system that provides the blood supply to solid tumors and participates in the invasion and metastasis of tumor cells. Studies have shown that VM is present in many types of malignant tumors, including human malignant uveal melanoma, hepatocellular carcinoma, osteosarcoma, lung cancer,7, 8 and glioma. In gliomas, VM is closely related to the pathological grade and the malignant biological behavior of the tumor, such as growth, invasion, and metastasis.

The human upstream transcription factor 1 (USF1) gene is located in the q22.3 region of chromosome 1. USF1 contains an HLH-LZ structure (helix-loop-helix leucine zipper) that can bind to the E-box of various gene promoter regions and is involved in regulating genes related to lipid and carbohydrate homeostasis in the human body. During DNA damage, USF1 can stabilize the p53 protein and briefly halt the cell cycle. Recent studies have found that USF1 is involved in regulating the development and progression of various types of tumors. In oral mucosal cancer, USF1 is involved in regulating the expression of the human telomerase gene, which leads to the activation of telomerase and the development of the disease. USF1 can induce colorectal cancer by regulating the human polymeric immunoglobulin (Ig) receptor gene. In gliomas, USF1 can increase the permeability of the blood-brain barrier and downregulate the expression of the tight junction proteins in the glioma microvascular endothelial cells. USF1 regulates the expression of interleukin-10 (IL-10) in glioma cells, and the inhibition of USF1 expression leads to an increase in IL-10 expression. However, the effects of USF1 on the biological behavior of glioma cells, such as VM capacity, remain unclear.

Long non-coding RNAs (lncRNAs) are a class of non-coding RNAs that regulate gene expression and functions at multiple levels. The most representative short non-coding RNAs are microRNAs (miRNAs), which play an important regulatory role in multiple cellular processes, including cell development, proliferation, apoptosis, and differentiation. Perhaps the most well-established mechanism of lncRNA action involves their interaction with miRNA by acting as competing endogenous RNA (ceRNA), and thereby regulating the expression of downstream coding genes. SNHG16 is an lncRNA that is aberrantly expressed in various types of tumors. For example, SNHG16 is highly expressed in breast cancer cells and colon cancer cells.18, 19 In addition, these results suggest that SNHG16 plays the role of a carcinogen in breast and colon cancer. linc00667 is located in the human chr18 (5238100–5246508). It is 8,409 bp in length and contains multiple miRNA response element (MRE) regions. To date, the expression and function of SNGH16 and linc00667 in gliomas have not been reported.

miRNAs are involved in the regulation of the biological behavior of tumor cells, including proliferation, apoptosis, migration, invasion, and angiogenesis. miR-212-3p is aberrantly expressed in various types of tumor cells. For example, miR-212-3p inhibits the proliferation and invasion of intrahepatic cholangiocarcinoma by targeting FOXA1. In gliomas, miR-212-3p is downregulated and can inhibit the proliferation and invasion of glioma cells through the targeted regulation of SGK3. miR-429 can suppress the tumorigenicity of cervical cancer cells by regulating the activity of nuclear factor κB. miR-429 is expressed at low levels in gliomas and can inhibit the migration and invasion of glioma cells by inhibiting the expression of BMK1.

Aldehyde dehydrogenase-1 (ALDH1A1) can oxidize various lipids and aromatic aldehydes. ALDH1A1 is considered a cancer stem cell marker in different types of tumors, such as breast, colon, and lung cancers. The expression of ALDH1A1 in glioma tissues is positively correlated with the pathological grading of the tumor and patient survival.24, 25, 26

This study first determined the expression levels of USF1, SNHG16, linc00667, and ALDH1A1 in glioma tissues and cells, and further investigated the regulatory relationships among these factors and their effects on VM in glioma cells. It aims to provide new evidence on glioma development and progression, and identify new targets for the treatment of gliomas.

Results

USF1 Positively Correlated with Glioma Levels and Promotes the Occurrence of VM of Glioma

First, CD34-periodic acid-Schiff (PAS) immunohistochemistry was used to prove that the number of VMs increased with the increase of the case levels of glioma (Figure S1A). The Cancer Genome Atlas (TCGA) database was used to find the high expression of USF1 in 665 cases of glioma (Figure S1B). Western blot found that, compared with normal brain tissues (NBTs), the expression of USF1 in glioma tissues was significantly increased, and the expression levels were positively correlated with histopathological grading. Compared with human astrocyte (HA) cells, the expression of USF1 in U87 and U251 cells was significantly upregulated (Figures 1A and 1B).

Figure 1

Knockdown of USF1 Inhibited VM Formation, and USF1 Targeted and Positively Regulated SNHG16 (or linc00667)

(A) Relative expression levels of USF1 protein in NBTs, LGGTs (WHO I–II), and HGGTs (WHO III–IV) (data are presented as the mean ± SD; NBTs group, n = 3; LGGs group, n = 3; HGGs group, n = 3). **p < 0.01 versus NBTs group; ##p < 0.01 versus LGGs group. (B) Relative expression levels of USF1 protein in HA, U87, and U251 cells (data are presented as the mean ± SD; n = 3, each group). **p < 0.01 versus HA group. (C) Cell Counting Kit-8 (CCK-8) assay was used to measure the effect of USF1 on the proliferation of U87 and U251 cells (data are presented as the mean ± SD; n = 4, each group). **p < 0.01 versus USF1(−)NC group. (D) Three-dimensional culture of U87 and U251 cells after USF1(−) was calculated (original magnification, ×200; scale bar, 100 μm; data are presented as the mean ± SD; n = 3, each group). **p < 0.01 versus USF1(−)NC group. (E) Transwell assays were used to measure the effect of USF1 on cell migration and invasion of U87 and U251 cells (original magnification, ×200; scale bar, 100 μm; data are presented as the mean ± SD; n = 3, each group). **p < 0.01 versus USF1(−)NC group. (F) The relative expression level of VM protein in U87 and U251 cells after USF1(−) (data are presented as the mean ± SD; n = 3, each group). **p < 0.01 versus USF1(−)NC group. (G) The relative expression levels of SNHG16 and linc00667 after USF1(−) (data were presented as the mean ± SD; n = 5, each group). **p < 0.01 versus USF1(−)NC. (H) Schematic representation of USF1 promoter region in 3,000 bp upstream of the transcription start site (TSS; designated as +1). Chromatin immunoprecipitation (ChIP) PCR products for putative SNHG16 and linc00667 binding sites and an upstream region not expected to associate with SNHG16 and linc00667 were depicted with bold lines. Dashed lines represent the primers used for each PCR. The image was representative of independent ChIP experiments.

To further elucidate the potential mechanisms in regulating VM, we then assessed the effects of USF1(−) on U87 and U251 cell proliferation, VM, migration, and invasion. The results showed that the cell proliferation, VM, migration, and invasion of the USF1(−) group significantly decreased compared with the USF1(−)NC (negative control) group. VM-associated proteins, vascular endothelial cadherin (VE-cadherin), EPH receptor A2 (EphA2), matrix metallopeptidase 2 (MMP-2), and matrix metallopeptidase 14 (MMP-14) expression decreased significantly (Figures 1C–1F).

USF1 was inhibited in glioma cells. qRT-PCR showed that SNHG16 and linc00667 expression were decreased (Figure 1G). JASPAR CORE, a bioinformatics software, found the potential binding sites of USF1 in the upstream promoter region of the transcription starting point of SNHG16 and linc00667 (1,000 bp). The results of the chromatin immunoprecipitation (ChIP) showed that USF1 has binding sites in the promoter region of SNHG16 (5′-AGCTCGTGACA-3′) and linc00667 (5′-CGCAAATGACA-3′) (Figure 1H).

SNHG16 and linc00667 Were Positively Correlated with Glioma Levels and Promote the Occurrence of VM of Glioma

The results of qRT-PCR showed that the expression levels of SNHG16 in brain glioma tissues were significantly increased compared with NBTs, and the expression levels were positively correlated with histopathological grading. Compared with HA cells, the expression levels of SNHG16 in U87 and U251 cells were significantly increased (Figures 2A and 2B). In the U87 and U251 cells of inhibit SNHG16, compared with the SNHG16(−)NC group, SNHG16 expression of the SNHG16(−) group was obviously decreased (Figure S1E). The cell proliferation, VM, migration, invasion, and expression of VM-associated proteins of the SNHG16(−) group significantly decreased compared with the SNHG16(−)NC group (Figures 2C–2F).

Figure 2

Knockdown of SNHG16 or linc00667 Inhibited VM Formation

(A) Relative expression levels of SNHG16 in NBTs, LGGTs (WHO I–II), and HGGTs (WHO III–IV) (data are presented as the mean ± SD; NBTs group, n = 15; LGGs group, n = 16; HGGs group, n = 10). **p < 0.01 versus NBTs group; #p < 0.05 versus LGGs group. (B) Relative expression levels of SNHG16 in HA, U87, and U251 cells (data are presented as the mean ± SD; n = 3, each group). **p < 0.01 versus HA group). (C) CCK-8 assay was used to measure the effect of SNHG16 on the proliferation of U87 and U251 cells (data are presented as the mean ± SD; n = 4, each group). **p < 0.01 versus SNHG16(−)NC group. (D) Three-dimensional culture of U87 and U251 cells after SNHG16(−) was calculated (original magnification, ×200; scale bar, 100 μm; data are presented as the mean ± SD; n = 3, each group). **p < 0.01 versus SNHG16(−)NC group. (E) Transwell assays were used to measure the effect of SNHG16 on cell migration and invasion of U87 and U251 cells (original magnification, ×200; scale bar, 100 μm; data are presented as the mean ± SD; n = 3, each group). **p < 0.01 versus SNHG16(−)NC group. (F) The relative expression level of VM protein in U87 and U251 cells after SNHG16(−) (data are presented as the mean ± SD; n = 4, each group). **p < 0.01 versus SNHG16(−)NC group. (G) Relative expression levels of linc00667 in NBTs, LGGTs (WHO I–II), and HGGTs (WHO III–IV) (data are presented as the mean ± SD; NBTs group, n = 15; LGGs group, n = 16; HGGs group, n = 10). **p < 0.01 versus NBTs group; #p < 0.05 versus LGGs group. (H) Relative expression levels of linc00667 in HA, U87, and U251 cells (data are presented as the mean ± SD (n = 3, each group). **p < 0.01 versus HA group. (I) CCK-8 assay was used to measure the effect of linc00667 on the proliferation of U87 and U251 cells (data are presented as the mean ± SD; n = 4, each group). **p < 0.01 versus linc00667(−)NC group. (J) Three-dimensional culture of U87 and U251 cells after linc00667(−) was calculated (original magnification, ×200; scale bar, 100 μm; data are presented as the mean ± SD; n = 3, each group). **p < 0.01 versus linc00667(−)NC group. (K) Transwell assays were used to measure the effect of linc00667 on cell migration and invasion of U87 and U251 cells (original magnification, ×200; scale bar, 100 μm; data are presented as the mean ± SD; n = 3, each group). **p < 0.01 versus linc00667(−)NC group). (L) The relative expression level of VM protein in U87 and U251 cells after linc00667(−) (data are presented as the mean ± SD; n = 4, each group). **p < 0.01 versus linc00667(−)NC group.

Meanwhile, the expression levels of linc00667 in brain glioma tissues were significantly increased compared with NBTs, and the expression levels were positively correlated with histopathological grading. Compared with HA cells, the expression levels of linc00667 in U87 and U251 cells were significantly increased (Figures 2G and 2H). In the U87 and U251 cells of inhibit linc00667, compared with the linc00667(−)NC group, linc00667 expression of the linc00667(−) group was obviously decreased (Figure S1G). The cell proliferation, VM, migration, invasion, and expression of VM-associated proteins of the linc00667(−) group significantly decreased compared with the linc00667(−)NC group (Figures 2I–2L).

miR-212-3p and miR-429 Show Low Expression in Glioma Tissues and Cells, and Overexpression of miR-212-3p and miR-429 Inhibited VM of Glioma Cells

The results of qRT-PCR showed that the expression levels of miR-212-3p in brain glioma tissues were significantly decreased compared with NBTs, and the expression levels were negatively correlated with histopathological grading. Compared with HA cells, the expression levels of miR-212-3p in U87 and U251 cells were significantly decreased (Figure 3A). qRT-PCR assay results showed that the expression levels of miR-212-3p in the SNHG16(−)group were significantly increased compared with the SNHG16(−)NC group (Figure 3B). The binding sites of SNHG16 and miR-212-3p were predicted by bioinformatics software (starBase2.0). The results of relative luciferase activity showed that the SNHG16-mutation (Mut)+pre-miR-212-3p-NC group and the SNHG16-Mut+pre-miR-212-3p group were not statistically significant compared with the SNHG16-WT (wild-type)+pre-miR-212-3p-NC group, but the relative luciferase activity of the SNHG16-WT+pre-miR-212-3p group was significantly decreased compared with the SNHG16-WT+pre-miR-212-3p-NC group (Figure 3C). To confirm whether SNHG16 and miR-212-3p were present in the RNA-induced silencing complex (RISC), we carried out an RNA-binding protein immunoprecipitation (RIP) assay. qRT-PCR was performed to measure RNA levels in the immunoprecipitates. The expression of SNHG16 and miR-212-3p were both increased in the anti-Ago2 group compared with the anti-IgG group (Figures S1I and S1J). The cell proliferation, VM, migration, invasion, and expression of VM-associated proteins of the pre-miR-212-3p group significantly decreased compared with the pre-miR-212-3p-NC group. The cell proliferation, VM, migration, invasion, and expression of VM-associated proteins of the anti-miR-212-3p group significantly decreased compared with the anti-miR-212-3p-NC group (Figures 3D–3G).

Figure 3

The Expression and Effect of miR-212-3p and miR-429 in Glioma

SNHG16 and linc00667 endogenous expression, effect on proliferation, VM formation, migration, invasion, and proteins of VM in glioma. (A) miR-212-3p expression levels in NBTs, LGGs (WHO I–II), and HGGs (WHO III–IV) were shown (data are presented as the mean ± SD; NBTs group, n = 5; LGGs group, n = 6; HGGs group, n = 5). **p < 0.01 versus NBTs group; #p < 0.05 versus LGGs group. Relative expression levels of miR-212-3p in HA, U87, and U251 cells are also shown (data are presented as the mean ± SD; n = 3, each group). **p < 0.01 versus HA group. (B) The expression of miR-212-3p after cell transfection with SNHG16(−) plasmid (data are presented as the mean ± SD; n = 3, each group). **p < 0.01 versus SNHG16(−)NC group. (C) The predicted miR-212-3p binding sites in SNHG16-3′ UTR-WT and the designed mutant sequence SNHG16-3′ UTR-Mut are indicated (data are presented as the mean ± SD; n = 4, each group). **p < 0.01 versus SNHG16-3′ UTR-Mut+pre-miR 212-3p-NC group. (D) CCK-8 assay was used to measure the effect of miR-212-3p on the proliferation of U87 and U251 (data are presented as the mean ± SD; n = 4, each group). **p < 0.01 versus pre-miR-212-3p-NC group; ##p < 0.01 versus anti-miR-212-3p-NC group. (E) Three-dimensional culture of U87 and U251 cells after pre-miR-212-3p and anti-miR-212-3p was calculated (original magnification, ×200; scale bar, 100 μm; data are presented as the mean ± SD; n = 3, each group). **p < 0.01 versus pre-miR-212-3p-NC group; ##p < 0.01 versus anti-miR-212-3p-NC group. (F) Transwell assays were used to measure the effect of miR-212-3p on cell migration and invasion of U87 and U251 (original magnification, ×200; scale bar, 100 μm; data are presented as the mean ± SD; n = 3, each group). **p < 0.01 versus pre-miR-212-3p-NC group; ##p < 0.01 versus anti-miR-212-3p-NC group. (G) The relative expression level of VM protein in U87 and U251 cells after pre-miR-212-3p and anti-miR-212-3p (data are presented as the mean ± SD; n = 4, each group). **p < 0.01 versus pre-miR-212-3p-NC group; ##p < 0.01 versus anti-miR-212-3p-NC group. (H) miR-429 expression levels in NBTs, LGGs (WHO I–II), and HGGs (WHO III–IV) were shown (data are presented as the mean ± SD; NBTs group, n = 15; LGGs group, n = 16; HGGs group, n = 10). **p < 0.01 versus NBTs group; #p < 0.05 versus LGGs group. Also shown were relative expression levels of miR-429 in HA, U87, and U251 cells (data are presented as the mean ± SD; n = 3, each group). **p < 0.01 versus HA group. (I) The expression of miR-429 after cell transfection with linc00667(−) plasmid (data are presented as the mean ± SD; n = 4, each group). **p < 0.01 versus linc00667(−)NC group. (J) The predicted miR-429 binding sites in linc00667-3′ UTR-WT and the designed mutant sequence linc00667 3′ UTR-Mut are indicated (data are presented as the mean ± SD; n = 4, each group). **p < 0.01 versus linc00667-3′ UTR-Mut+pre-miR-429-NC group. (K) CCK-8 assay was used to measure the effect of miR-429 on the proliferation of U87 and U251 (data are presented as the mean ± SD; n = 4, each group). **p < 0.01 versus pre-miR-429-NC group; ##p < 0.01 versus anti-miR-429-NC group. (L) Three-dimensional culture of U87 and U251 cells after pre-miR-429 and anti-miR-429 was calculated (original magnification, ×200; scale bar, 100 μm; data are presented as the mean ± SD; n = 4, each group). **p < 0.01 versus pre-miR-429-NC group; ##p < 0.01 versus anti-miR-429-NC group. (M) Transwell assays were used to measure the effect of miR-429 on cell migration and invasion of U87 and U251 (original magnification, ×200; scale bar, 100 μm; data are presented as the mean ± SD; n = 4, each group). **p < 0.01 versus pre-miR-429-NC group; ##p < 0.01 versus anti-miR-429-NC group. (N) The relative expression level of VM protein in U87 and U251 cells after pre-miR-429 and anti-miR-429 (data are presented as the mean ± SD; n = 4, each group). **p < 0.01 versus pre-miR-429-NC group; ##p < 0.01 versus anti-miR-429-NC group.

In addition, the results of qRT-PCR showed that the expression levels of miR-429 in brain glioma tissues were significantly decreased compared with NBTs, and the expression levels were negatively correlated with histopathological grading. Compared with HA cells, the expression levels of miR-429 in U87 and U251 cells were significantly decreased (Figure 3H). qRT-PCR assay results showed that the expression levels of miR-429 in the linc00667(−) group were significantly increased compared with the linc00667(−)NC group (Figure 3I). The binding sites of linc00667 and miR-429 were predicted by bioinformatics software (starBase2.0). The results of relative luciferase activity showed that the linc00667-Mut+pre-miR-429-NC group and the linc00667-Mut+pre-miR-429 group were not statistically significant compared with the linc00667-WT+pre-miR-429-NC group, but the relative luciferase activity of the linc00667-WT+pre-miR-miR-429 group was significantly reduced compared with the linc00667-WT+pre-miR-429-NC group (Figure 3J). In the RIP assay, the expression of linc00667 and miR-429 were both increased in the anti-Ago2 group compared with the anti-IgG group (Figures S1L and S1M). Similar results were also observed when we detected the effect of miR-429 on the cell proliferation, VM, migration, invasion, and expression of VM-associated proteins of U87 and U251 cells. These results demonstrated that miR-212-3p and miR-429 exerted the inhibition of the VM role in the glioma (Figures 3K–3N).

Inhibiting SNHG16 and Overexpressing mir-212-3p, as well as Inhibiting linc00667 and Overexpressing miR-429, Inhibited VM of Glioma Cells

miR-212-3p overexpression and inhibition were implemented in U87 and U251 cells that stably inhibited SNHG16, and cell viability was determined 48 h later. Experimental results showed that, compared with the SNHG16(−)NC+pre-miR-212-3p-NC group, cell proliferation, VM, migration, and invasion of the SNHG16(−)+pre-miR-212-3p group significantly decreased. The ALHD1A1 protein expression and the expression of VM-associated proteins were significantly reduced (Figures 3A–3D).

In addition, miR-429 overexpression and inhibition were implemented in U87 and U251 cells that stably inhibited linc00667, and cell viability was determined 48 h later. Experimental results showed that, compared with the linc00667(−)NC+pre-miR-429-NC group, cell proliferation, VM, migration, and invasion of the linc00667(−)+pre-miR-429 group significantly decreased. The ALHD1A1 protein expression and the expression of VM-associated proteins were significantly reduced (Figures 4E–4H).

Figure 4

SNHG16 Regulated Tumor-Induced VM via Binding to miR-212-3p, and linc00667 Regulated Tumor-Induced VM via Binding to miR-429

(A) CCK-8 assay was used to evaluate the effect of miR-212-3p and SNHG16 on cell proliferation (data are presented as the mean ± SD; n = 4, each group). **p < 0.01 versus SNHG16(−)+pre-miR-212-3p-NC group. (B) Three-dimensional culture of U87 and U251 cells after cell transfection with SNHG16 and miR 212-3p plasmid was calculated (original magnification, ×200; scale bar, 100 μm; data are presented as the mean ± SD; n = 3, each group). **p < 0.01 versus SNHG16(−)+pre-miR-212-3p-NC group. (C) Transwell assay was used to evaluate the effect of miR-212-3p and SNHG16 on cell migration and invasion (data are presented as the mean ± SD; n = 4, each group). **p < 0.01 versus SNHG16(−)+pre-miR-212-3p-NC group. (D) The expression of VM-associated proteins of U87 and U251 cells after cell transfection with SNHG16 and miR-212-3p plasmid were shown (data are presented as the mean ± SD; n = 3, each group). **p < 0.01 versus SNHG16(−)+pre-miR-212-3p-NC group. (E) CCK-8 assay was used to evaluate the effect of miR-429 and linc00667 on cell proliferation (data are presented as the mean ± SD; n = 4, each group). **p < 0.01 versus linc00667(−)+pre-miR-429-NC group. (F) Three-dimensional culture of U87 and U251 cells after cell transfection with linc00667 and miR-429 plasmid was calculated (original magnification, ×200; scale bar, 100 μm; data are presented as the mean ± SD; n = 3, each group). **p < 0.01 versus linc00667(−)+pre-miR-429-NC group. (G) Transwell assay was used to evaluate the effect of miR-429 and linc00667 on cell migration and invasion (data are presented as the mean ± SD; n = 4, each group). **p < 0.01 versus linc00667(−)+pre-miR-429-NC group. (H) The expression of VM-associated proteins of U87 and U251 cells after cell transfection with linc00667 and miR-429 plasmid was shown (data are presented as the mean ± SD; n = 4, each group). **p < 0.01 versus linc00667(−)+pre-miR-429-NC group.

ALDH1A1 Positively Correlated with Glioma Levels and Promotes the Occurrence of VM of Glioma

Western blot found that, compared with NBTs, the expression of ALDH1A1 in glioma tissues was significantly increased, and the expression levels were positively correlated with histopathological grading. Compared with HA cells, the expression of ALDH1A1 in U87 and U251 cells was significantly upregulated (Figure 5A). Expression changes of ALDH1A1 in U87 and U251 cell lines overexpressed and inhibited miR-212-3p and miR-429, respectively. Western blot results showed that the pre-miR-212-3p group ALDH1A1 expression levels were significantly reduced compared with the pre-miR-212-3p-NC group. On the contrary, compared with the anti-miR-212-3p-NC group, the expression levels of ALDH1A1 in the anti-mir-212-3p group increased significantly. Similar results were also observed when detecting the effect of miR-429 (Figure 5B).

Figure 5

Knockdown of ALDH1A1 Inhibited VM Formation and miR-212-3p (or miR-429) Targeted ALDH1A1

(A) ALDH1A1 protein expression levels in NBTs, LGGs (WHO I–II), and HGGs (WHO III–IV) (data are presented as the mean ± SD; NBTs group, n = 3; LGGs group, n = 3; HGGs group, n = 3). **p < 0.01 versus NBTs group; ##p < 0.01 versus LGGs group. Also shown is the expression of ALDH1A1 in HA, U87, and U251 cells (data are presented as the mean ± SD; n = 3, each group). **p < 0.01 versus HA group. (B) The expression of ALDH1A1 after cell transfection with pre-miR-212-3p and anti-miR-212-3p plasmid. The expression of ALDH1A1 after cell transfection with pre-miR-429 and anti-miR-429 plasmid (data are presented as the mean ± SD; n = 3, each group). **p < 0.01 versus pre-miR-429-NC and pre-miR-212-3p-NC group; ##p < 0.01 versus anti-miR-429-NC group and pre-miR-212-3p-NC group. (C) The predicted miR-212-3p binding sites in ALDH1A1-3′ UTR-WT and the designed mutant sequence ALDH1A1-3′ UTR-Mut were indicated. The predicted miR-429 binding sites in ALDH1A1-3′ UTR-WT and the designed mutant sequence ALDH1A1-3′ UTR-Mut were indicated (data are presented as the mean ± SD; n = 3, each group). **p < 0.01 versus ALDH1A1-3′ UTR-Mut+pre-miR-212-3p-NC and ALDH1A1-3′ UTR-Mut+pre-miR-429-NC group. (D) CCK-8 assay was used to measure the effect of ALDH1A1 on the proliferation of U87 and U251 cells (data are presented as the mean ± SD; n = 3, each group). **p < 0.01 versus ALDH1A1(+)NC group; ##p < 0.01 versus ALDH1A1(−)NC group. (E) Three-dimensional culture of U87 and U251 cells after ALDH1A1(+) and ALDH1A1(−) was calculated (original magnification, ×200; scale bar, 100 μm; data are presented as the mean ± SD; n = 3, each group). **p < 0.01 versus ALDH1A1(+)NC group; ##p < 0.01 versus ALDH1A1(−)NC group. (F) Transwell assay was used to measure the effect of ALDH1A1 on cell migration and invasion of U87 and U251 cells (original magnification, ×200; scale bar, 100 μm; data are presented as the mean ± SD; n = 3, each group). **p < 0.01 versus ALDH1A1(+)NC group; ##p < 0.01 versus ALDH1A1(−)NC group. (G) The expression of VM-associated proteins of U87 and U251 cells after ALDH1A1(+) and ALDH1A1(−) were shown (data are presented as the mean ± SD; n = 3, each group). **p < 0.01 versus ALDH1A1(+)NC group; ##p < 0.01 versus ALDH1A1(−)NC group. (H) The p-ERK/ERK expression was regulated by ALDH1A1 (data are presented as the mean ± SD; n = 4, each group). **p < 0.01 versus ALDH1A1(+)NC group; ##p < 0.01 versus ALDH1A1(−)NC group.

Bioinformatics software (miRanda) predicted that ALDH1A1 and miR-212-3p (or ALDH1A1 and miR-429) existed as binding sites. The results of relative luciferase activity showed that luciferase activity of the ALDH1A1-Mut+pre-miR-212-3p-NC (ALDH1A1-Mut+pre-miR-429-NC) group and ALDH1A1-Mut+pre-miR-212-3p (ALDH1A1-Mut+pre-miR-429) group was not statistically significantly different compared with the ALDH1A1-WT+pre-miR-212-3p-NC (ALDH1A1-WT+pre-miR-429-NC) group. Compared with the ALDH1A1-WT+pre-miR-212-3p-NC (ALDH1A1-WT+pre-miR-429-NC) group, the luciferase activity of the ALDH1A1-WT+pre-miR-212-3p (ALDH1A1-WT+pre-miR-429) group was significantly reduced (Figure 5C). We successfully transfected U87 and U251 cells with ALHD1A1 inhibition and overexpression, and tested changes in the biological behavior of cells. Compared with the ALDH1A1(+)NC group, the cell proliferation, VM, migration, invasion, and expression of VM-associated proteins of the ALDH1A1(+) group significantly increased. On the contrary, compared with the ALDH1A1(−)NC group, the cell proliferation, VM, migration, invasion, and expression of VM-associated proteins of the ALDH1A1(−) group significantly decreased. In addition, the changes of phosphorylation and total ERK were detected, and the phosphorylated ERK (p-ERK) was significantly increased in the ALDH1A1(+) group compared with the ALDH1A1(+)NC group. Compared with the ALDH1A1(−)NC group, the p-ERK was significantly decreased in the ALDH1A1(−) group (Figures 5D–5H).

The Effect of Co-transfection with mir-212-3p and ALDH1A1 (or Co-transfection of miR-429 and ALDH1A1) on the VM in Glioma Cells

The changes of cell biological behavior were detected by Cell Counting Kit-8 (CCK-8), Matrigel experiment, and Transwell assay, when the overexpression of miR-212-3p or miR-429 was performed in the U87 and U251 cells of the stable ALDH1A1(+). The results showed that the proliferation, VM, migration, invasion, and expression of VM-associated proteins of the ALDH1A1(+)+pre-miR-212-3p-NC group significantly increased (Figures 6A–6D), compared with the ALDH1A1(+)NC+pre-miR-212-3p-NC group. Similar results were also observed when detecting the effect of ALDH1A1 and miR-429 (Figures 6E–6H).

Figure 6

miR-212-3p (or miR-429) Regulated Tumor-Induced VM via Binding to ALDH1A1

(A) CCK-8 assay to evaluate the effect of miR-212-3p and ALHD1A1 on cell proliferation (data are presented as the mean ± SD; n = 3, each group). **p < 0.01 versus ALDH1A1(+)NC+pre-miR-212-3p-NC group; ##p < 0.01 versus ALDH1A1(+)+pre-miR-212-3p-NC group. (B) Three-dimensional culture of U87 and U251 cells after cell transfection with ALHD1A1 and miR-212-3p plasmid was calculated (magnification, ×200; scale bar, 100 μm; data are presented as the mean ± SD (n = 3, each group). **p < 0.01 versus ALDH1A1(+)NC+pre-miR-212-3p-NC group; ##p < 0.01 versus ALDH1A1(+)+pre-miR-212-3p-NC group. (C) Transwell assay to evaluate the effect of miR-212-3p and ALHD1A1 on cell migration and invasion (data are presented as the mean ± SD; n = 4, each group). **p < 0.01 versus ALDH1A1(+)NC+pre-miR-212-3p-NC group; ##p < 0.01 versus ALDH1A1(+)+pre-miR-212-3p-NC group). (D) The expression of VM-associated proteins of U87 and U251 cells after cell transfection with ALHD1A1 and miR-212-3p plasmid were shown (data are presented as the mean ± SD; n = 3, each group). **p < 0.01 versus ALDH1A1(+)NC+pre-miR-212-3p-NC group; ##p < 0.01 versus ALDH1A1(+)+pre-miR-212-3p-NC group. (E) CCK-8 assay to evaluate the effect of miR-429 and ALHD1A1 on cell proliferation (data are presented as the mean ± SD; n = 3, each group). **p < 0.01 versus ALDH1A1(+)NC+pre-miR-429-NC group; ##p < 0.01 versus ALDH1A1(+)+pre-miR-429-NC group. (F) Three-dimensional culture of U87 and U251 cells after cell transfection with ALHD1A1 and miR-429 plasmid was calculated (original magnification, ×200; scale bar, 100 μm; data are presented as the mean ± SD; n = 3, each group). **p < 0.01 versus ALDH1A1(+)NC+pre-miR-429-NC group; ##p < 0.01 versus ALDH1A1(+)+pre-miR-429-NC group. (G) Transwell assay to evaluate the effect of miR-429 and ALHD1A1 on cell migration and invasion (data are presented as the mean ± SD; n = 4, each group). **p < 0.01 versus ALDH1A1(+)NC+pre-miR-429-NC group; ##p < 0.01 versus ALDH1A1(+)+pre-miR-429-NC group. (H) The expression of VM-associated proteins of U87 and U251 cells after cell transfection with ALHD1A1 and miR-429p plasmid were shown (data are presented as the mean ± SD; n = 3, each group). **p < 0.01 versus ALDH1A1(+)NC+pre-miR-429-NC group; ##p < 0.01 versus ALDH1A1(+)+pre-miR-429-NC group.

Effects of Single or Combined Use of Inhibitors of Multiple Targets of USF1, SNHG16, and LINC00667 on the Growth of Transplanted Tumors in Nude Mice

The nude mouse transplanted tumor model further detected the effects of USF1, SNHG16, and linc00667 for glioma, divided into the control group, the USF1(−) group, the SNHG16 (−) group, the linc00667(−) group, and the USF1(−)+SNHG16(−)+linc00667(−) group. For subcutaneous implantation, the cells were injected subcutaneously into the flank area of each animal. For survival analysis in orthotopic inoculations, cells were stereotactically implanted into the right striatum of the mice. The results showed that, compared with the control group, the transplanted tumor volumes of the USF1(−) group, SNHG16(−) group, linc00667(−) group, and USF1(−)+SNHG16(−)+linc00667(−) group were significantly reduced. Further, compared with the USF1(−) group, SNHG16(−) group, and linc00667(−) group, the tumor volume of the USF1(−)+SNHG16(−)+linc00667(−) group was smaller (Figure 7A). The results of survival analysis were consistent with the results of subcutaneous transplantation. Compared with the control group, the USF1(−), SNHG16(−), linc00667(−), and USF1(−)+SNHG16(−)+linc00667(−) groups had a longer survival period, whereas the survival period of the USF1(−)+SNHG16(−)+linc00667(−) group was the longest (Figure 7B). As shown in Figure 7C, there was no significant difference in the number of VMs between control and USF1(−)NC+SNHG16(−)NC+linc00667(−)NC groups. However, the number of VMs in the other four groups was significantly reduced, and the number in the USF1(−)+SNHG16(−)+linc00667(−) group was the least (Figure 7C).

Figure 7

In Vivo Tumor Xenograft Study

(A) For subcutaneous xenograft study, cells were injected subcutaneously into nude mice. A sample tumor from the representative group was shown. Tumor volume was measured and calculated. (B) Survival curves from representing nude mice injected into the right striatum were shown (data are presented as the mean ± SD; n = 6, each group). **p < 0.01 versus USF1(−)NC+SNHG16(−)NC+linc00667(−)NC group; #p < 0.05 versus USF1(−) group; Δp < 0.05 versus SNHG16(−) group; &p < 0.05 versus linc00667(−) group. (C) CD34-PAS was applied to detect the VM in xenografted tumor, and results were expressed as the number of VMs per field (original magnification, ×400; scale bar, 50 μm).

Discussion

This study first demonstrated that as the grade of malignant gliomas increases, the number of VMs also increases. This finding was consistent with previous studies. Studies have reported that the expression levels of many proteins are positively correlated with the VM of the tumor, including matrix metalloproteinases (MMPs), vascular endothelial cadherin (VE-cadherin), and EphA2. Therefore, these proteins are considered VM-associated proteins.28, 29 The present study used the expression of these proteins as a positive indicator for VM.

USF1 is located on chromosome 1 q22.3 and encodes a multifunctional transcription factor that is widely expressed in eukaryotes. It plays an important role in the regulation of the cell cycle, glucose, lipid metabolism, cell proliferation, and the development and progression of multiple types of tumors.30, 31, 32, 33 We first searched TCGA database and found that USF1 was highly expressed in 665 glioma cases. Our results concurred with this search, because USF1 was found to be highly expressed in the glioma tissues, as well as cells, and its expression levels increased with the pathological grade of the tumor. The inhibition of USF1 expression inhibited the proliferation, migration, invasion, and VM capacity of glioma cells. The expression of VM-associated proteins VE-cadherin, EphA2, MMP-2, and MMP-14 also decreased. These results suggest that USF1 plays a role in promoting VM in glioma tissues and cells. Similarly, USF1 has been shown to be highly expressed in colorectal cancer tissues and cells, and the inhibition of USF1 expression hinders the proliferation, migration, and invasion of colorectal cancer cells. USF1 has also been shown to be highly expressed in melanoma cells, and the inhibition of USF1 expression decreases the capacity of melanoma cells for epithelial-mesenchymal transition (EMT) and migration.

This study demonstrated that SNHG16 and linc00667 were highly expressed in glioma tissues and cells, and their expression increased in tumors with higher pathological grades. The respective inhibition of SNHG16 and linc00667 inhibited the proliferation, migration, invasion, and VM capacity of U87 and U251 cells. Furthermore, the expression of VM-associated proteins such as VE-cadherin, EphA2, MMP-2, and MMP-14 also decreased, which suggested their role as tumor-promoting factors in glioma tissues and cells. Recently, the role of SNHG16 in tumors has attracted the attention of researchers. Similar to the results of this study, in which the high expression of SNHG16 in glioma tissues and cells promoted VM and tumor development, SNHG16 has been shown to be highly expressed in breast and intestinal cancer cells, and inhibiting its expression can inhibit the proliferation and migration of breast cancer cells.18, 19 Moreover, SNHG16 is highly expressed in gastric cancer tissues, and its expression levels are related to the degree of gastric cancer invasion. Thus, SNHG16 can be used as a target for monitoring the prognosis and treatment of gastric cancer.

linc00667 contains multiple MRE regions. Through analysis of the database, these regions were found to contain potential binding sites for multiple miRNAs. There is currently no report of linc00667 expression in tumors. This study showed that linc00667 was expressed in small amounts in normal human brain tissues and highly expressed in glioma cells. The inhibition of linc00667 expression inhibited the proliferation, migration, invasion, and VM capacity of glioma cells, and reduced the expression of VM-associated proteins VE-cadherin, EphA2, MMP-2, and MMP-14, which suggests that linc00667 had a tumor-promoting effect in gliomas and that inhibition linc00667 expression could inhibit the malignant behavior of gliomas.

Further, the JASPAR CORE database was used to predict the presence of USF1 binding sites in the promoter regions of SNHG16 and linc00667. Next, ChIP experiments were performed, and they confirmed the binding of USF1 to both SNHG16 and linc00667. Furthermore, the inhibition of USF1 expression downregulated the expression of SNHG16 and linc00667. These results suggest that USF1 could regulate the expression of SNHG16 and linc00667 at the transcriptional level. This in turn indicated that the inhibition of USF1 expression could inhibit the proliferation, migration, invasion, and VM capacity of glioma cells and downregulate the expression of VM-associated proteins by inhibiting the transcription of SNHG16 and linc00667.

This study further demonstrated that there was little expression of miR-212-3p and miR-429 in the glioma tissues and cells, and that their expression levels were lower in high-grade glioma tissues (HGGTs). The respective overexpression of miR-212-3p and miR-429 could inhibit the proliferation, migration, invasion, and VM capacity of glioma cells and downregulate the expression of VM-associated proteins. The inhibition of miR-212-3p and miR-429 resulted in the opposite effects. These results suggest that miR-212-3p and miR-429 were tumor-suppressing factors in glioma tissues and cells. Previous studies have shown similar results, with the overexpression of miR-212-3p being shown to inhibit the proliferation and invasion of intrahepatic cholangiocarcinoma cells and glioma cells, which indicates that miR-212-3p has a tumor-suppressing role in intrahepatic cholangiocarcinoma and gliomas.20, 21 Similarly, the overexpression of miR-429 can inhibit the tumorigenicity of cervical cancer cells, which is consistent with the tumor-suppressor effects of this miRNA on gliomas found here.22, 23

In this study, the binding sites of SNHG16 and miR-212-3p and those of linc00667 and miR-429 were predicted using the starBase2.0 database. Based on the prediction, dual-luciferase reporter gene assays were performed to confirm the binding of SNHG16 and miR-212-3p and that of linc00667 and miR-429. RNA immunoprecipitation (IP) experiments confirmed that SNHG16 and miR-212-3p were enriched in Ago2 protein, suggesting SNHG16 decreased miR-212-3p expression in a RISC-dependent manner. linc00667 and miR-429 were the same as SNHG16 and miR-212-3p. Furthermore, the inhibition of SNHG16 and linc00667 expression was found to upregulate the expression of miR-212-3p and miR-429, respectively. This suggests that inhibition of SNHG16 expression could inhibit the proliferation, migration, invasion, and VM capacity of glioma cells and downregulate the expression of VM-associated proteins through upregulating the expression of miR-212-3p. Similarly, inhibition of linc00667 expression could inhibit the proliferation, migration, invasion, and VM capacity of glioma cells and downregulate the expression of VM-associated proteins through upregulating the expression of miR-429. Similar results showing the effects of the binding of lncRNAs and miRNAs on the biological behavior of glioma cells have also been previously reported. For example, it has been shown that the binding of colorectal neoplasia differentially expressed (CRNDE) and miR-384 and the binding of SOX2OT to miR-194-5p and miR-122 modulate the malignant behavior of glioma cells. lncRNA can bind to miRNAs and act as ceRNA, which in turn affects the miRNA-mediated regulation of target genes.38, 39 In glioma cells, HOTAIR has been shown to bind to miR-326 as its ceRNA and upregulate fibroblast growth factor 1 (FGF1) to promote the proliferation, migration, and invasion of glioma cells. In esophageal cancer cells, HOTAIR has been shown to bind to miR-148a as its ceRNA and upregulate the expression of Snail2, thereby promoting the invasion and metastasis of esophageal cancer cells. In pancreatic cancer cells, HOTAIR acts as a ceRNA and binds to miR-613, upregulating the expression of Notch3 and promoting the proliferation, migration, and invasion of pancreatic cancer cells.

ALDH1A1 plays an important role in cell division and drug resistance. ALDH1A1+ tumor cells are more frequently associated with high tumor microvessel density in breast cancer. In ovarian cancer tissues, ALDH1A1 is involved in VM formation. It can promote the proliferation, migration, and invasion of cancer cells and inhibit their apoptosis.46, 47 Temozolomide, as a traditional anti-angiogenic drug, can inhibit angiogenesis in glioma tissue. Experiments have shown that ALDH1A1 can mediate glioma cell resistance to temozolomide, thereby reducing temozolomide inhibition of glioma angiogenesis. This study demonstrated that ALDH1A1 was highly expressed in glioma tissues and cells, and its expression increased with the pathological grade of the glioma. Inhibition of ALDH1A1 expression inhibited the proliferation, migration, invasion, and VM capacity of glioma cells and downregulated the expression of VM-associated proteins, whereas the overexpression of ALDH1A1 exerted the opposite effects. The results suggest that ALDH1A1 played an oncogenic role in glioma tissues and cells. Additionally, we found that the overexpression of ALDH1A1 in gliomas significantly increased the expression of p-ERK, and that inhibition of ALDH1A1 exerted the opposite effects. In gliomas, ERK can facilitate VM formation via promoting MMP2, MMP9, and MT1-MMP. Many studies have indicated that ERK is involved in the regulation of multiple biological processes in human cells, such as cell proliferation, transformation, and differentiation, through the phosphorylation of its target proteins. At the same time, ERK can also have tumor-promoting effects. For instance, in human oral cancer cells, DEPDC1B enhances anchorage-independent cell growth by promoting the expression of p-ERK. The activation of the gene encoding the p110α catalytic subunit of phosphatidylinositide-3 kinase (PIK3CA)/epidermal growth factor receptor (EGFR)/ERK pathway has also been shown to promote the growth of human basal-like breast cancer cells. In ovarian cancer, miR-497 inhibits angiogenesis by suppressing the mitogen-activated protein kinase (MAPK)/ERK pathway.

Based on predictions using the miRanda bioinformatics software, dual-luciferase reporter assays were performed to indicate that miR-212-3p and miR-429 specifically targeted and bound to the 3′ UTR region of ALDH1A1. Our study also found that the overexpression of miR-212-3p inhibited the proliferation, migration, invasion, and VM capacity of glioma cells and decreased the expression of VM-associated proteins. The overexpression of ALDH1A1 promoted the proliferation, migration, invasion, and VM capacity of glioma cells and increased the expression of VM-associated proteins. At the same time, the overexpression of miR-212-3p and ALDH1A1 reversed the inhibition of the proliferation, migration, invasion, and VM capacity of glioma cells induced by the overexpression of miR-212-3p. miR-429 had similar functions. These results suggest that the overexpression of miR-212-3p and miR-429 affected the biological behavior of the glioma cells through increasing the negative regulation of the target gene ALDH1A1. This study used bioinformatics to examine the MREs present within SNHG16 and applied the starBase v2.0 bioinformatics system to find a binding site between SNHG16 and miR-212-3. Furthermore, the MRE sequence for their binding (5′-GACUGUU-3′) was the same as that between miR-212-3p and the ALDH1A1 mRNA-3′ UTR (426–432). In addition, LINS00667 and miR-429 had the same MRE sequence (5′-AGUAUU-3′) as that for the binding of miR-429 and ALDH1A1 mRNA-3′ UTRs (127–132). The simultaneous inhibition of SNHG16 and the overexpression of miR-212-3p, or the simultaneous inhibition of linc00667 and overexpression of miR-429, inhibited the VM capacity of the gliomas. Double inhibition of SNHG16 and miR-212-3p expression or double inhibition of linc00667 and miR-429 expression reversed this effect. These results indicate that SNHG16 bound to miR-212-3p as its ceRNA, and linc00667 bound to miR-429 as its ceRNA, which in turn regulated the proliferation, migration, invasion, and VM capacity of the gliomas.

Finally, by performing nude mouse xenograft experiments, this study showed that the mice in the USF1(−) group, SNHG16(−) group, linc00667(−) group, and USF1(−)+SNHG16(−)+linc00667(−) group could significantly inhibit the growth of the glioma xenografts and prolong survival. Compared with the individual USF1(−), SNHG16(−), and linc00667(−) groups, the combined use of all three resulted in the smallest tumor xenografts and the longest survival. These results suggest that the combined use of USF1(−), SNHG16(−), and linc00667(−) may be clinically valuable.

In summary, this study demonstrated that USF1 regulated the biological behavior of glioma cells by regulating ALDH1A1 through SNHG16/miR-212-3p and linc00667/miR-429. The results provided a new experimental basis supporting research on the mechanism of VM in gliomas, as well as a novel direction for the treatment of gliomas.

Materials and Methods

Cell Culture and Human Tissue Samples

Human glioma tissues and NBTs were collected from patients at the Department of Neurosurgery of Shengjing Hospital of China Medical University (n = 5). All of the tissue samples were immediately frozen in liquid nitrogen after surgical resection and stored at −80°C until use. Informed consent was obtained from all patients, and the study was approved by the Ethics Committee of Shengjing Hospital of China Medical University. Glioma tissue samples were divided into two groups, low-grade glioma tissues (LGGTs; n = 16) and HGGTs (n = 10), by experienced neuropathologists according to the 2007 World Health Organization (WHO) classification of tumors in the CNS. HA cells were obtained from Scien-Cell Research Laboratories (Carlsbad, CA, USA) and grown in RPMI-1640 culture medium (GIBCO, Grand Island, NY, USA) with 10% fetal bovine serum (FBS; GIBCO, Carlsbad, CA, USA). Human glioma cell lines (U87 and U251) and HEK293T cells were purchased from Shanghai Genechem and grown in DMEM/high glucose with 10% FBS. All cells were maintained in a humidified incubator at 37°C with 5% CO2.

CD34 Endothelial Marker Periodic Acid-Schiff Dual Staining

CD34-PAS was examined for the existence of VM. Human glioma tissue samples were fixed with 4% formaldehyde, embedded in paraffin, and sectioned into 5-μm tumor slides. These were deparaffinized in xylene, hydrated, and boiled in EDTA antigen-unmasking solution. When cooled to room temperature, slides were incubated in peroxide at room temperature for endogenous peroxidase ablation, blocked with goat serum, and stained with a rabbit anti-human CD34 primary monoclonal antibody (1:100; Beijing Zhongshan Golden bridge, China) overnight at 4°C. After washing with PBS thrice and incubating with goat anti-rabbit secondary antibody at room temperature for 10 min, the slides were treated with a Dako REAL EnVision Detection System, Peroxidase/DAB+, Rabbit/Mouse (DAB) kit (Fuzhou MaiXin Biotech, China). Then, the slides were exposed to periodic acid solution for 10 min, incubated with Schiff solution for 10 min in the dark, and counterstained with Mayer’s hematoxylin (Zhuhai Baso, Guangdong, China). Lastly, the slides were viewed under a light microscope to detect CD34 and PAS signals. Immunohistochemical staining was performed with the help of the UltraSensitive S-P kit (Fuzhou MaiXin Biotech, China).

Real-Time qPCR Analysis

Total RNA was extracted from tissues and cells (U87 and U251) with Trizol reagent (Life Technologies Corporation, Carlsbad, CA, USA). RNA concentration and quality were determined via 260/280 nm absorbance with NanoDrop Spectrophotometer (ND-100; Thermo, USA). Using the 7500 Fast RTPCR System, we used the One-Step SYBR Prime-Script RT-PCR Kit (TakaraBio, Japan) to detect the expression of USF1, SNHG16, and linc00667; meanwhile, glyceraldehyde-3-phsophate dehydrogenase (GAPDH) was used as the endogenous control. The TaqMan MicroRNA Reverse Transcription kit (Applied Biosystems, Foster City, CA, USA) was used for the reverse transcription of miR-212-3p and miR-429; then the expression of miR-212-3p and miR-429 were detected with TaqMan Universal Master Mix II, and U6 was used as the endogenous control.

RIP Assay

RIP was determined using the Magna RNA Binding Protein Immunoprecipitation Kit according to the manufacturer’s instructions. Whole-cell lysate was incubated with RIP buffer containing magnetic beads conjugated with human anti-Ago2 antibody or NC normal mouse IgG. Samples were incubated with Proteinase K, and immunoprecipitated RNA was isolated. The RNA concentration was measured by a spectrophotometer (NanoDrop; Thermo Scientific, Waltham, MA, USA), and the RNA quality was assessed using a bioanalyzer (Agilent, Santa Clara, CA, USA). Furthermore, purified RNAs were extracted and analyzed by real-time qPCR to demonstrate the presence of the binding targets.

Transfection and Generation of Stably Transfected Cell Lines

Short hairpin RNA directed against human USF1, SNHG16, linc00667, and ALDH1A1 gene were constructed in pGPU6/GFP/Neo vector (Gene-Pharama, Shanghai, China). The full-length ALDH1A1 gene was constructed in pIRES2-EGFP (GenScript, Piscataway, NJ, USA). Plasmid carrying a non-targeting sequence was used as an NC. When cells reached 70%–80% confluence, the transaction was performed by Lipofectamine 3000 Reagents (Life Technologies, Carlsbad, CA, USA). G418 and puromycin (Sigma-Aldrich, St. Louis, MO, USA) were used to select the stably transfected cells. G418-resistant (or puromycin-resistant) cell clones were established around approximately 4 weeks. Transfected efficiencies of stable cell lines were analyzed by qRT-PCR. To evaluate the effect of USF1 on Glioma U87 and U251 cells, we divided experiments into three groups: control, USF1(−)NC, and USF1(−) groups. To evaluate the effect of SNHG16 on Glioma U87 and U251 cells, we divided experiments into three groups: control, SNHG16(−)NC, and SNHG16(−) groups. To evaluate the effect of linc00667 on Glioma U87 and U251 cells, we divided experiments into three groups: control, linc00667NC(−), and linc00667(−) groups. To evaluate the effect of ALDH1A1 on Glioma U87 and U251 cells, we divided experiments into five groups: control, ALDH1A1(+)NC, ALDH1A1(+), ALDH1A1(−)NC, and ALDH1A1(−) groups.

Transient Transfection of miRNAs and Grouping

Pre-miR-212-3p, anti-miR-212-3p, and their respective NCs were synthesized (Gene Pharma, Shanghai, China). Glioma U87 and U251 cells were respectively transfected with pre-miR-212-3p, anti-miR-212-3p, and their respective NCs by Lipofectamine 3000 reagent. The transfected efficacy was analyzed by qRT-PCR, and the high transfected efficacy occurred at 24 h after transfection. Experiments were divided into five groups: control, pre-NC, pre-miR-212-3p, anti-NC, and anti-miR-212-3p groups. The processing and grouping of miR-429 was the same as miR-212-3p. To determine whether SNHG16-mediated regulation of miR-212-3p expression could affect the behaviors of glioma cells, we divided experiments into five groups: control, SNHG16(−)NC+pre-miR-212-3p-NC, SNHG16(−)+pre-miR-212-3p, SNHG16(−)NC+anti-miR-212-3p-NC, and SNHG16(−)+anti-miR-212-3p groups. Meanwhile, to determine whether linc00667-mediated regulation of miR-429 expression could affect the behaviors of glioma cells, we divided experiments into five groups: control, linc00667(−)NC+pre-miR-429-NC, linc00667(−)+pre-miR-429, linc00667(−)NC+anti-miR-429-NC, and linc00667(−)+anti-miR-429 groups. To determine whether miR-212-3p-mediated regulation of ALDH1A1 expression could affect the behaviors of glioma cells, we divided experiments into four groups: control, ALDH1A1(+)NC+pre-miR-212-3p-NC, ALDH1A1(+)+pre-miR-212-3p-NC, and ALDH1A1(+)+pre-miR-212-3p groups. To determine whether miR-429-mediated regulation of ALDH1A1 expression could affect the behaviors of glioma cells, we divided experiments into four groups: control, ALDH1A1(+)NC+pre-miR-429-NC, ALDH1A1(+)+pre-miR-429-NC, and ALDH1A1(+)+pre-miR-429 groups.

In Vitro VM Tube Formation Assay

Each hole in the 96-well culture plate was covered with 60 μL of Matrigel Basement Membrane Matrix (BD Biosciences, Bedford, MA, USA). Meanwhile, the bubbles were avoided. The 96-well culture plate was incubated for 30 min in the incubator at 37°C. The cells were resuspended in 100 μL of serum-free medium and seeded onto the surface of Matrigel at a density 6 × 105 cells/mL and incubated for 8 h. The cell vascular structures were observed and photographed under an inverted microscope (Olympus, Tokyo, Japan). An independent observer counted the total number of tube-like structures per image.

Cell Proliferation Assay

Cells were seeded in 96-well plates at a density of 2,000 cells per well, and 20 μL of Cell Counting Kit-8 (Beyotime Institute of Biotechnology, Jiangsu, China) was added to each well over 48 h. Cells were incubated for 2 h at 37°C, and the absorbance was recorded at 450 nm.

Western Blot Analysis

Harvested cells (U87 and U251) were lysed using RIPA (Beyotime Institute of Biotechnology) buffer on ice for 30 min and were centrifuged at 17,000 × g for 45 min at 4°C. The protein concentrations were measured by the BCA protein assay kit (Beyotime Institute of Biotechnology, Jiangsu, China). The proteins were processed by SDS-PAGE electrophoretically transferring to polyvinylidene fluoride (PVDF) membranes. The membranes were blocked by Tween-Tris-buffered saline (TTBS) containing 5% non-fat milk for 2 h at room temperature and then incubated with primary antibodies as follows: USF1 (Santa Cruz Biotechnology), ALDH1A1 (Proteintech, USA), MMP-14 (Proteintech, USA), MMP-2 (Proteintech, USA), VE-cadherin (Abcam, UK), EphA2 (Abcam, UK), ERK (Abcam, UK), p-ERK (Abcam, UK), and GAPDH (Proteintech, USA) overnight at 4°C. After washing three times with TTBS, membranes were incubated with horseradish peroxidase-conjugated secondary antibody for 2 h at room temperature and then developed with enhanced chemiluminescence (ECL) kit (Santa Cruz Biotechnology) and scanned by ChemImager 5500 V2.03 software according to the manufacturer’s protocol. The relative integrated density values (IDVs) were calculated using Fluor Chen 2.0 software based on GAPDH as an internal control.

Cell Migration and Invasion Assay

The cell migration and invasion were assessed by penetrating a 8-μm pore size polycarbonate membrane (Costar, Corning, NY, USA). The cells were resuspended in serum-free medium at a density of 2 × 105 cells/mL and seeded into the upper chamber of a polycarbonate membrane (pre-coated with Matrigel and incubated at 37°C for 30 min before the invasion assay started). Then, 600 μL of 10% FBS medium was placed into the lower chamber. After incubation at 37°C for 36 h, the cells migrated or invaded from the upper chamber to the lower surface of the membrane, and the cells were fixed with methanol and glacial acetic acid at a ratio of 3:1, before staining with 20% Giemsa. Five random fields were chosen to count and take photos under a microscope.

Reporter Vectors Construction and Luciferase Assays

The sequence of SNHG16 (or linc00667) was amplified by PCR and cloned into pmirGLO Dual-luciferase miRNA Target Expression Vectors along with its mutant sequence of miR-212-3p (or mir-429) binding sites (GenePharama, Shanghai, China). HEK293T cells were seeded in a 96-well plate (Corning) and co-transfected with WT pmirGLO-SNHG16 (or SNHG16 mutant) reporter plasmid and agomir-212-3p or agomir-212-3p-NC, respectively. linc00667 and miR-429 were the same as above. The luciferase activities were performed with the Dual-Lucifer Reporter Assay System (Promega, Madison, WI, USA) after 48 h according to the instruction manual operation procedure. The relative luciferase activity was calculated by normalizing to renilla luciferase activity. The 3′ UTR sequences of ALDH1A1 and its mutant sequence of miR-212-3p (or miR-429) binding sites were cloned into pmirGLO Dual-luciferase miRNA Target Expression Vectors (GenePharama, Shanghai, China). The transfection procedure and calculating method of Lucifer’s activities were the same as above.

Chromatin Immunoprecipitation Assay

ChIP assay was carried out with Simple ChIP Enzymatic Chromatin IP Kit (Cell Signaling Technology, Danvers, MA, USA) according to the manufacturer’s instructions. In brief, cells (U87 and U251) were cross-linked with 1% formaldehyde in culture medium for 10 min and quenched with glycine for 5 min at room temperature. These cells were then collected in lysis buffer. 2% lysates were used as an input reference, and other lysates were incubated with normal rabbit IgG or anti-USF1 antibody with rotation. DNA cross-links were reversed by NaCl and Proteinase K and purified. DNA was amplified by PCR with the following primers: SNHG16, PCR1 (F) 5′-GGTTGAACGATTCCCTCGGA-3′, (R) 5′-GGGCTTGGCCGAGTATCTTT-3′; PCR2 (F) 5′-AGAGCAGTGGGTGTTACAGTATGC-3′, (R) 5′-TGTTTTTAAGTTTGGGGCGAG-3′; PCR3 (F) 5′-CCTCTTCACTAACCCAGCGA-3′, (R) 5′-CAGTTTGTCCCTGGTCGAGA-3′; PCR4 (F) 5′-CCAGCTTCGTTTTCTGACAGC-3′, (R) 5′-TAGTTCAAAACTACGGCGGCT-3′; linc00667, PCR1 (F) 5′-GTCAGTGGTAGGCACATCCT-3′, (R) 5′-GGAAAGCCATCTACCGAGTCAG-3′; PCR2 (F) 5′-ATAATTTGCGCTGTGAGGCG-3′, (R) 5′-CTTCCACCCGGACAGTATGC-3′.

Tumor Xenograft Implantation in Nude Mice

For the in vivo study, the stably transfected cell lines (U87 and U251) were used. The mice were divided into five groups: control; USF1(−); SNHG16(−); linc00667(−); and USF1(−)+SNHG16(−)+linc00667(−) groups. Cells stably transfected with USF1(−), SNHG16(−), and linc00667(−) were selected as described before. After infection, the stable expressing cells were picked. Four-week-old athymic nude mice (BALB/c) were purchased from the Cancer Institute of the Chinese Academy of Medical Science. Experiments with mice were conducted strictly in accordance with a protocol approved by the Administrate Panel on Laboratory Animal Care of China Medical University. Each nude mouse was subcutaneously injected with 6 × 105 cells in the right flank area for subcutaneous implantation. Tumors were measured every 5 days and calculated according to the formula: volume (mm3) = length × width2/2. The mice were sacrificed and the tumors were isolated after 50 days. For orthotopic inoculations (cells were injected into the right striatum of nude mice), the number of surviving nude mice was registered, and survival analysis was performed using the Kaplan-Meier survival curve. Lastly, we detected a VM in vivo mouse xenograft model by CD34-PAS. Samples were collected from patients undergoing surgery at the Department of Neurosurgery, Shengjing Hospital of China Medical of University from January 2016 to December 2016, after they provided written informed consent. Samples were processed following the standard operation and storage procedures with appropriate ethical approval by the Research Ethics Committee of Shengjing Hospital. The Ethics Committee of China Medical University and Ethical Committee for Animal Experimentation approved this study.

Statistical Analysis

Experimental data were presented as means ± SD. All differences were analyzed by SPSS 22.0 statistical software with the Student’s t test (two-tailed) or one-way ANOVA. p < 0.05 was considered as statistically significant.

Author Contributions

Y.L. contributed to the experimental design, manuscript draft, and data analysis. D.W. contributed to the experimental implementation, manuscript draft, and data analysis. Y.X. designed the experiments. J.Z., X.L., J.M., and Q.H. performed the experiments. L.L., H.C., and Z.L. analyzed the data. D.W. conceived or designed the experiments, performed the experiments, and wrote the manuscript. All authors read and approved the final manuscript.

Conflicts of Interest

The authors declare no competing interests.