LncRNA DGCR5 plays a tumor-suppressive role in glioma via the miR-21/Smad7 and miR-23a/PTEN axes

Selection of lncRNAs related to glioma overall survival

To investigate lncRNAs that might affect glioma EMT and progression, we first analyzed the correlation between deregulated lncRNAs and overall survival in glioma patients. Figure 1A shows the analysis and selection of candidate lncRNAs according to the TCGA GBMLGG dataset, REMBRANDT dataset, and microarray expression profiles from the GEO database. Based on the TCGA GBMLGG dataset, we selected 144 lncRNAs that were positively correlated with better overall survival in glioma patients when they were highly expressed (Figure 1B). Analysis of the REMBRANDT database, showed that 32 of the 144 lncRNAs had a positive correlation with better overall survival in patients with glioma when they were highly expressed (Figure 1C). Based on the online microarray expression profileGSE4271, higher lncRNA DGCR5 expression was significantly positively correlated with better overall survival in glioma patients (Figure 1D). Based on the TCGA Brain Lower Grade Glioma (LGG) dataset, higher lncRNA DGCR5 expression was significantly positively correlated with better overall survival in glioma patients (Figure 1E). According to all these online data, we hypothesized that lncRNA DGCR5 might play a protective role against glioma progression.

Figure 1. Selection of lncRNAs related to glioma overall survival (A) A schematic diagram showing the process of lncRNA selection. (B) Based on TCGA GBMLGG dataset, we selected 144 lncRNAs that are correlated with better overall survival in glioma patients when they are highly expressed. (C) Based on REMBRANDT dataset, 32 of the 144 lncRNAs are correlated with better overall survival in glioma patients when they are highly expressed. (D) Based on online microarray expression profile (GSE4271), higher lncRNA DGCR5 expression is significantly correlated with better overall survival in glioma patients. (E) Based on TCGA Brain Lower Grade Glioma (LGG) dataset, higher lncRNA DGCR5 expression is significantly correlated with better overall survival in glioma patients.

Expression of lncRNA DGCR5 according to online data and experimental results

To confirm this hypothesis, we examined lncRNA DGCR5 expression under different grouping conditions. According to online HPA RNA-seq data, DGCR5 expression was specifically upregulated in brain tissues among 27 types of tissues from 95 persons (Figure 2A). According to TCGA data, the expression of lncRNA DGCR5 was reduced in proneural glioma, neural glioma, mesenchymal glioma, and classical glioma, compared with noncancerous (normal) tissue samples (Figure 2B). According to REMBRANDT data, lncRNA DGCR5 expression was dramatically downregulated in oligodendroglioma, astrocytoma, and glioblastoma, compared with normal (noncancerous) tissues (Figure 2C). According to GSE4290, lncRNA DGCR5 expression showed to be remarkably downregulated within oligodendroglioma, astrocytoma, and glioblastoma, in comparison with that in normal (non-cancerous) tissues (Figure 2D). According to Ivy Glioblastoma Atlas Project data, lncRNA DGCR5 expression was analyzed in different parts of glioma, including leading-edge, infiltrating carcinoma, perinecrotic zone, cellular tumor, pseudoproliferative cells surrounding necrosis, proliferative blood vessels in cell tumor, and microvascular proliferation glioma. Figure 2E shows that lncRNA DGCR5 expression was the highest in the leading edge, whereas it was the lowest in microvascular proliferation glioma. Since microvascular proliferation glioma is the most malignant and invasive, these data indicated that lncRNA DGCR5 expression continued to be downregulated with tumor progression.

Figure 2. Expression of lncRNA DGCR5 according to online data and experimental results. (A) LncRNA DGCR5 expression in 27 types of tissues from 95 persons according to online HPA RNA-seq data. (B) LncRNA DGCR5 expression in normal (noncancerous), proneural glioma, neural glioma, mesenchymal glioma, and classical glioma according to TCGA dataset. (C) LncRNA DGCR5 expression in oligodendroglioma, astrocytoma, and glioblastoma according to REMBRANDT dataset. (D) LncRNA DGCR5 expression in oligodendroglioma, astrocytoma, and glioblastoma according to GSE4290. (E) LncRNA DGCR5 expression in different parts of glioma, including leading edge, infiltrating tumor, perinecrotic zone, cellular tumor, pseudopalisading cells around necrosis, hyperplastic blood vessels in cellular tumor, and microvascular proliferation according to Ivy Glioblastoma Atlas Project data. (F) LncRNA DGCR5 expression in normal (noncancerous, n=28) and glioma tissues (stage II n=11, stage III n=9, stage IV n=8) determined by real-time PCR. (G) LncRNA DGCR5 expression in normal human astrocytes, NHA, and glioma cell lines, LN229, U251-MG, SHG44, and T98G was determined by real-time PCR. **P<0.01, ***P<0.005, compared to control group. ###P<0.005, compared to stage II group. $$P<0.01, compared to stage III group.

To further confirm this hypothesis, we examined lncRNA DGCR5 expression in tissue samples and cells. Figure 2F shows that the expression of lncRNA DGCR5 was significantly reduced in glioma tissue samples (stage II, n=11; stage III, n=9; stage IV, n=8), in comparison with normal (noncancerous, n=28) tissues. Moreover, DGCR5 expression was lower in stage III tissue samples and the lowest in stage IV tissues (Figure 2F), further suggesting that lncRNA DGCR5 expression continued to be downregulated with tumor progression. In cell lines, lncRNA DGCR5 expression was markedly downregulated in the four glioma cell lines, LN229, U251-MG, SHG44, and T98G compared to that in normal human astrocytes (NHA) (Figure 2G).

NF-κB1 inhibits the expression of lncRNA DGCR5 by targeting its promoter region

As we have mentioned, NF-κB is considered a common transcriptional factor and exerts a significant effect on the mediation of various major characteristics related to mesenchymal differentiation [29]. According to TCGA data (Supplementary Figure 1A), GSE4290 (Supplementary Figure 1B), REMBRANDT data (Supplementary Figure 1C), CGGA data (Supplementary Figure 1D), and IVY data (Supplementary Figure 1E), NF-κB1 (p50) expression was negatively correlated with DGCR5 expression. Moreover, we performed Kyoto Encyclopedia of Genes and Genomes (KEGG) and Gene Set Enrichment Analysis (GSEA) on DGCR5 and found that DGCR5 was correlated with NF-κB1 signaling, including interferon α response and interferon γ response (Supplementary Figure 2A, 2B). More importantly, by using the online tool JASPER (http://jaspar.genereg.net/) according to the ensemble binding database, we found that the DGCR5 promoter region possesses NF-κB1 response element (RE) (Supplementary Table 1). Thus, we hypothesized that NF-κB1 can bind to the DGCR5 promoter region to inhibit its expression.

To verify the hypothesis, we overexpressed or silenced NF-κB1 in U251-MG and SHG44 cells by transfection of an NF-κB1-overexpressing vector or si-NF-κB1 and performed immunoblotting and real-time PCR to verify the transfection efficiency (Figure 3A, 3B). Then, DGCR5 expression was determined in NF-κB1 or si-NF-κB1 transfected glioma cells. As expected, NF-κB1 overexpression significantly inhibited DGCR5 expression, whereas NF-κB1 silencing promoted DGCR5 expression (Figure 3C). To verify the predicted binding between NF-κB1 and the NF-κB1 RE in the DGCR5 promoter region, two different types of psiCHECK-2-proDGCR5 luciferase reporter vectors, wild-type and mutant, were constructed as described (Figure 3D). We cotransfected these vectors into 293T cells with pcDNA3.1/NF-κB1 and examined luciferase activity. Figure 3D shows that psiCHECK-2-proDGCR5 luciferase activity was significantly decreased via the overexpression of NF-κB1. Mutating the predicted NF-κB1 RE in the DGCR5 promoter region abolished the alterations in luciferase activity (Figure 3D). Therefore, considering that NF-κB1 exerts an essential effect on the mediation of various major characteristics related to mesenchymal differentiation [29], we hypothesize that DGCR5 exerts an effect on mesenchymal differentiation] in glioma.

Figure 3. NF-κB1 binds to the promoter region of lncRNA DGCR5 to inhibit its expression. (A, B) NF-κB1 overexpression or silencing generated in U251-MG and SHG44 cells by transfection of NF-κB1-overexpressing vector or si-NF-κB1, as confirmed by immunoblotting and real-time PCR. (C) DGCR5 expression in response to NF-κB1 overexpression or silencing determined by real-time PCR. (D) Wild-type and mutant-type psicheck-2-proDGCR5 luciferase reporter vectors were constructed as described in the Materials and methods section. These vectors were cotransfected in 293T cells with pcDNA3.1/NF-κB1 and the luciferase activity was determined. *P<0.05, **P<0.01, compared to the si-NC group. ##P<0.01, compared to vector group.

Effects of lncRNA DGCR5 overexpression on glioma

To validate the specific effect of lncRNA DGCR5 on glioma cells, we first constructed the coexpression network consisting of DGCR5 and its negatively-correlated genes according to TCGA data, REMBRANDT data, and GSE4290. As shown in Supplementary Figure 3A and Supplementary Table 2, a total of 18 genes were significantly negatively correlated with DGCR5, including GNAI3 [30], VIM [31], ITGB1 [32, 33], HIF1A [34], and HDAC1 [35], all of which are critical factors affecting glioma development. The expression of these five genes was significantly upregulated in U251-MG and SHG44 cells (Supplementary Figure 3B–3F). Thus, further experiments were performed using U251-MG and SHG44 cells.

Next, we overexpressed and silenced lncRNA DGCR5 in U251-MG and SHG44 cells by transfection of the lncRNA DGCR5-overexpressing vector or short hairpin RNA (sh-DGCR5) and performed real-time PCR to verify the transfection efficiency (Figure 4A). Next, we examined the specific effects on glioma cells. LncRNA DGCR5 overexpression significantly inhibited the capacity of DNA synthesis (Figure 4B), invasion (Figure 4D) and migration (Figure 4E) in glioma cells, whereas it enhanced the apoptosis rate of glioma cells (Figure 4C). Regarding the signaling involved, the overexpression of lncRNA DGCR5 dramatically enhanced the protein levels of E-cadherin but inhibited the protein levels of VIM, Snai2, and TWIST (Figure 4F); in contrast, DGCR5 silencing significantly decreased E-cadherin protein but increased VIM, Snai2, and TWIST proteins (Figure 4F). In summary, lncRNA DGCR5 might inhibit the capacity of glioma cells to proliferate, migrate, and invade, thus exerting a tumor-suppressive effect.

Figure 4. Effects of lncRNA DGCR5 overexpression on glioma cells. (A) LncRNA DGCR5 overexpression or silencing was generated in U251-MG and SHG44 cells by transfection of lncRNA DGCR5-overexpressing vector (lncDGCR5) or small hairpin RNA (sh-DGCR5), as confirmed by real-time PCR. Next, DNA synthesis capacity was determined by EdU assay (B); cell apoptosis was determined by Flow cytometry assay (C); cell invasive capacity was determined by Transwell assay (D); cell migratory capacity was determined by Wound healing assay (E). (F) U251-MG and SHG44 cells were transfected with lncDGCR5 or sh-DGCR5 and the protein levels of E-cadherin, VIM, Snai2, and TWIST were determined by immunoblotting. *P<0.05, **P<0.01, ***P<0.005 compared to vector group; ## P<0.01 compared to sh-NC group.

LncRNA DGCR5 directly binds to miR-21-3p and miR-23a-5p

Commonly, lncRNAs exert an effect on the expression of miRNA-targeted mRNA via interaction with miRNAs. To further investigate the suppressive effects of lncRNA DGCR5 on the proliferation, migration, and invasion of glioma cells, we analyzed TCGA data to identify miRNAs that were negatively correlated with DGCR5. As shown in Supplementary Figure 4, miR-21 and miR-23a were negatively correlated with DGCR5. Reportedly, miR-21 is regarded as promising treatment target for glioma [36]. In addition, miR-23a is also implicated in promoting glioma cell proliferation [37]. In tissue samples, the expression of miR-21/miR-23a was significantly upregulated in glioma tissue samples (stage II, n=11; stage III, n=9; stage IV, n=8) compared with normal (noncancerous, n=28) tissues. Moreover, miR-21/miR-23a expression was higher in stage III tissue samples and highest in stage IV tissues (Figure 5A). Accordingly, the expression of miR-21/miR-23a was significantly upregulated in glioma cell lines compared with NHAs (Figure 5B). Moreover, online tools predicted that miR-21 and miR-23a might both be the downstream targets of DGCR5. To confirm that lncRNA DGCR5 negatively regulates miR-21/miR-23a, we transfected U251-MG and SHG44 cells with lncDGCR5 or sh-DGCR5 and examined the expression levels of miR-21/miR-23a. As shown in Figure 5C, lncRNA DGCR5 overexpression significantly downregulated, whereas lncRNA DGCR5 silencing significantly upregulated the expression levels of miR-21/miR-23a. Thus, we hypothesized that DGCR5 exerts its tumor-suppressive effects through miR-21/miR-23a.

Figure 5. LncRNA DGCR5 directly binds to miR-21-3p and miR-23a-5p (A) miR-21-3p and miR-23a-5p expression was determined in normal (non-cancerous, n=28) and glioma tissues (stage II n=11, stage III n=9, stage IV n=8) by real-time PCR. (B) miR-21-3p and miR-23a-5p expression in normal human astrocytes, NHA, and glioma cell lines, LN229, U251-MG, SHG44, and T98G was determined by real-time PCR. (C) U251-MG and SHG44 cells were transfected with lncDGCR5 or sh-DGCR5 and the expression of miR-21-3p and miR-23a-5p was determined by real-time PCR. (D) miR-21-3p and miR-23a-5p overexpression or inhibition generated in U251-MG and SHG44 cells by transfection of miR-21-3p/miR-23a-5p or anti-miR-21-3p/anti-miR-23a-5p, as confirmed by real-time PCR. (E) Wild-type and mutant-type lncRNA DGCR5 luciferase reporter vectors were constructed as described in Materials and methods section. These vectors were co-transfected in 293T cells with miR-21-3p/miR-23a-5p or anti-miR-21-3p/anti-miR-23a-5p, and the luciferase activity was determined. **P<0.01, ***P<0.005, compared to NHA, vector or miR-NC group; #P<0.05, ##P<0.01, ###P<0.005, compared to sh-NC or anti-NC group.

To verify this hypothesis, we transfected U251-MG and SHG44 cells with miR-21-3p/miR-23a-5p or anti-miR-21-3p/anti-miR-23a-5p to achieve overexpression or inhibition of miR-21-3p and miR-23a-5p and performed real-time PCR to verify the transfection efficiency (Figure 5D). Second, we conducted a luciferase reporter assay and constructed two different types of lncRNA DGCR5 luciferase reporter vectors, wild-type and mutant, as described, to verify the putative binding sites. We cotransfected these vectors into 293T cells with miR-21-3p/miR-23a-5p or anti-miR-21-3p/anti-miR-23a-5p, and examined luciferase activity. Figure 5E shows that wild-type DGCR5 luciferase activity was suppressed via the overexpression of miR-21/miR-23a, but enhanced via the inhibition of miR-21/miR-23a. Mutating the putative DGCR5 bindings sites abolished alterations in the luciferase activity (Figure 5E). These data indicate that miR-21 and miR-23a are downstream targets of lncRNA DGCR5.

The DGCR5/miR-21 axis affects glioma cell migration and invasion through Smad7

The effect of the miR-21/Smad7 axis has been found in many diseases [38–40]. Smad7 is a TGFβ1 signaling inhibitor and has been reported to inhibit TGFβ1-induced cancer cell epithelial-mesenchymal transition (EMT) and metastasis [41–43]. In the present study, Smad7 was identified as a direct downstream target of miR-21; thus, the cellular functions of miR-21 might be related to cancer cell migration and invasion. Moreover, online data also showed that DGCR5 expression was significantly positively correlated with Smad7 expression (Supplementary Figure 5A–5E). Thus, we investigated whether the DGCR5/miR-21 axis affects Smad7 to act on glioma cell migration and invasion. We transfected U251-MG and SHG44 cells with lncRNA DGCR5-overexpressing vector and examined Smad7 protein levels. Figure 6A shows that the overexpression of lncRNA DGCR5 remarkably increased Smad7 protein levels in both U251-MG and SHG44 cell lines. Second, we cotransfected U251-MG and SHG44 cells with DGCR5 and miR-21 mimic and evaluated their dynamic effects on glioma cells. In contrast to DGCR5 overexpression, the overexpression of miR-21 dramatically promoted the capacity of glioma cells to migrate and invade (Figure 6B, 6C). Moreover, the overexpression of miR-21 significantly attenuated the effects of DGCR5 overexpression (Figure 6B, 6C). Regarding the signaling involved, miR-21 overexpression significantly decreased Smad7 and E-cadherin proteins, but increased VIM, Snai2, and TWIST proteins (Figure 6D). Accordingly, the overexpression of miR-21 significantly attenuated the effects of DGCR5 overexpression on these protein levels (Figure 6D). In summary, the DGCR5/miR-21 axis affects glioma cell migration, invasion, and EMT through Smad7.

Figure 6. DGCR5/miR-21 axis affect glioma cell migration and invasion through Smad7. (A) U251-MG and SHG44 cells were transfected with lncRNA DGCR5-overexpressing vector and examined for the protein levels of Smad7 by immunoblotting. Next, U251-MG and SHG44 cells were co-transfected with DGCR5 and miR-21 mimic and examined for (B) migratory capacity by wound healing assay; (C) invasive capacity by Transwell assay; (D) the protein levels of Smad7, E-cadherin, VIM, Snai2, and TWIST were determined by immunoblotting. *P<0.05, **P<0.01, compared to the control group; #P<0.05, ##P<0.01, compared to the vector + miR-21 mimic group.

The DGCR5/miR-23a axis affects glioma cell proliferation and apoptosis through PTEN

PTEN is often knocked down in glioma cells and has been identified as a miR-23a downstream target in glioma [44]. Similarly, online data demonstrated a significant positive correlation between DGCR5 expression and PTEN expression (Supplementary Figure 6A–6D). The tumor-suppressor activity of PTEN depends largely on its lipid phosphatase activity, which inhibits PI3K/AKT activation. Activated AKT has many downstream effects, including the promotion of cell proliferation, increase in cellular glycolytic flux, and inhibition of apoptosis. PTEN antagonizes this survival/proliferative paradigm [45, 46]. In the present study, PTEN was identified as a direct downstream target of miR-23a; thus, the cellular functions of miR-23a might be related to cancer cell survival/proliferation/apoptosis. Therefore, we speculated that the DGCR5/miR-23a axis might affect PTEN to act on glioma cell proliferation and apoptosis. To verify this hypothesis, we transfected U251-MG and SHG44 cells with the lncRNA DGCR5-overexpressing vector and evaluated PTEN proteins. As shown in Figure 7A, DGCR5 overexpression markedly enhanced PTEN protein levels in the two glioma cell lines. Second, we cotransfected U251-MG and SHG44 cells with DGCR5 and miR-23a mimic and evaluated the dynamic effects of DGCR5 and miR-23a overexpression. miR-23a overexpression significantly increased glioma cell proliferation, and enhanced glioma cell apoptosis (Figure 7B, 7C). Moreover, the overexpression of miR-23a dramatically attenuated the effects of DGCR5 overexpression (Figure 7B, 7C). Regarding the signaling involved, miR-23a overexpression significantly decreased the protein levels of PTEN, increased AKT and PI3K phosphorylation and increased Cyclin D1 protein levels (Figure 7D). Similarly, the overexpression of miR-23a significantly attenuated the effects of DGCR5 overexpression on these protein levels (Figure 7D). In summary, the DGCR5/miR-23a axis affects glioma cell proliferation and apoptosis through PTEN/PI3K/AKT.

Figure 7. DGCR5/miR-23a axis affect glioma cell proliferation and apoptosis through PTEN. (A) U251-MG and SHG44 cells were transfected with lncRNA DGCR5-overexpressing vector and examined for the protein levels of PTEN by immunoblotting. Next, U251-MG and SHG44 cells were co-transfected with DGCR5 and miR-23a mimic, (B) DNA synthesis capacity was determined by EdU assay (C); cell apoptosis was determined by Flow cytometry assay; (D) the protein levels of PTEN, p-AKT, AKT, p-PI3K, PI3K, and Cyclin D1 were determined by immunoblotting. *P<0.05, **P<0.01, compared to the control group; #P<0.05, ##P<0.01, compared to the vector + miR-23a mimic group.

In vivo effects of lncRNA DGCR5 on tumor growth

To further confirm these in vitro findings, we established an orthotopic implanted xenograft mouse model. We examined the effects of lncRNA DGCR5 on tumor growth, proliferation markers, cell cycle regulators, and EMT markers. As shown in Figure 8A–8B, lncRNA DGCR5 overexpression significantly reduced the tumor volume and weight of both U251-MG-derived and SHG44-derived tumors. Moreover, DGCR5 overexpression significantly decreased the protein levels of proliferation marker ki-67, cell cycle regulator Cyclin D1, and mesenchymal marker TWIST, but increased the protein level of the epithelial marker E-cadherin in tumor tissues (Figure 8C). These data indicate that lncRNA DGCR5 acts as a tumor-suppressor to inhibit glioma tumor growth in vivo.

Figure 8. In vivo effects of lncRNA DGCR5 on tumor growth An xenograft model was established in mice as described in the Materials and methods section. (A, B) The tumor volume and weight were examined after the mice were sacrificed. (C) Tumor tissues were collected and the protein levels of ki-67, Cyclin D1, TWIST, and E-cadherin were determined by immunoblotting. *P<0.05, **P<0.01.

Clinical tissue sampling

A total of 28 noncancerous peritumoral brain edema (PTBE) tissues and 28 glioma tissues (11 stage II, 9 stage III, 8 stage IV) were collected as described in our previous studies [24, 25] with the approval of Sichuan Provincial People’s Hospital. Sampling was obtained with singed informed consent from all participants. Once collected, samples were stored in liquid nitrogen or fixed in 4% paraformaldehyde.

Cell lines and cell transfection

The human glioma cell lines LN229 (ATCC® CRL-2611) and T98G (ATCC® CRL-1690) were obtained from the American Type Culture Collection (ATCC, USA). SHG44 cells were obtained from the Cell Resource Center, Shanghai Institute of Life Sciences (Shanghai, China). U251-MG cells were purchased from the China Center for Type Culture Collection (CCTCC, Wuhan, China). All glioma cells were cultured in supplemented RPMI-1640 medium (Invitrogen, USA) supplemented with 10% fetal bovine serum (FBS; Gibco, Waltham, MA, USA) at 37 °C with 5% CO₂. Normal human astrocytes (NHAs) were obtained from Lonza (Basel, Switzerland) and cultured in astrocyte basal medium (Lonza) supplemented with 10% FBS at 37 °C with 5% CO₂.

The lncDGCR5-overexpressing vector (lncDGCR5) was used to generate DGCR5-overexpressing cells (GenePharma, Shanghai, China). The overexpression and inhibition of miR-21-3p and miR-23a-5p was achieved by transfection of miR-21-3p/miR-23a-5p or anti-miR-21-3p/anti-miR-23a-5p vectors (GenePharma). All transfections were performed with the help of Lipofectamine 3000 (Invitrogen, Waltham, MA, USA).

PCR-based analyses

After extracting total RNA from target tissues or cells, we examined the expression levels of miRNAs and mRNAs using an SYBR Green PCR kit (Qiagen, Germany‎). The expression of RNU6B and GAPDH served as endogenous controls, for the determination of miRNA and mRNA expression, respectively. The data were processed and calculated using the 2^-ΔΔCT method.

DNA synthesis capacity detected by the EdU assay

DNA synthesis was detected using EdU reagent following previously described methods [26]. Apollo staining and DAPI staining were used to examine the EdU positive cells with a fluorescence microscope following previously described methods [26].

Cell apoptosis detected by flow cytometry

For apoptosis analysis, the quantification of apoptotic cells was performed with an Annexin V-FITC apoptosis detection kit (Keygen, Nanjing, China) following the previously described methods [27]. Propidium iodide (PI) was used for nuclear staining. The excitation wavelength and the emission wavelength were 488 nm and 530 nm, respectively.

Migratory capacity detected by the wound healing assay

The cells were digested, and the cell concentration was adjusted to 5 × 10⁵ cells/ml. Cell suspensions (100 μl) were plated in 96-well plates coated with Matrigel, routinely cultured until the cell monolayer was established. Then the cell scratch test was performed. The cells were cultured in RPMI-1640 supplemented with 10 g/l bovine serum albumin (BSA) and 1% FCS, and the scratch area was measured under a microscope. Twenty-four hours later, the cells continued to culture for another 24 h in RPMI-1640 supplemented with 10% FCS. Then the relative distance of cell migration to the injury area was also measured under a microscope.

Invasive capacity detected by the transwell assay

The invasive capacity was determined using polycarbonate Transwell filters following previously described methods [28]. Medium without serum was used in the upper chambers coated with Matrigel, and medium with serum was used in the bottom chambers. After discarding the noninvasive cells in the top chambers, the invasive cells on the lower membrane surface were fixed, stained with DAPI (Beyotime Institute of Biotechnology, Haimen, China) for nuclear staining, and the cell number was counted under a microscope.

Protein expression levels detected by immunoblotting

Protein levels were determined following previously described methods [28] with the following antibodies: anti-E-cadherin (ab1416), anti-Vimentin (ab92547), anti-Snai2 (AV33302, Sigma-Aldrich, St. Louis, MO, USA), anti-TWIST (ab50887), anti-Smad7 (ab90086), anti-PTEN (ab32199), anti-AKT (ab179463), anti-p-AKT (ab81283), anti-PI3K (ab151549), anti-p-PI3K (ab182651), anti-Cyclin D1 (ab16663), anti-NF-κB1 (ZRB2082, Sigma-Aldrich), anti-GAPDH (ab8245), and Horseradish peroxidase-conjugated secondary antibody IgG (ab6721 and ab205719). Enhanced Chemiluminescent (ECL) Substrates (Millipore, MA, USA) were used for signals’ visualization taking Tubulin as an endogenous normalization control.

Luciferase reporter assay

The fragment of lncRNA DGCR5 was amplified by PCR, cloned downstream of the Renilla gene in the psiCHECK-2 vector (Promega, Madison, WI, USA), and the resulting construct was named wt-DGCR5. To generate the DGCR5 mutant reporter, we mutated the seed region of DGCR5 to remove all complementarity to miR-21-3p and miR-23a-5p, and the resulting construct was named mut-DGCR5. These vectors were cotransfected into 293T cells with miR-21-3p/miR-23a-5p or anti-miR-21-3p/anti-miR-23a-5p. Luciferase assays were performed 48 h after transfection using the Dual-Luciferase Reporter Assay System (Promega). Renilla luciferase activity was normalized to firefly luciferase activity for each transfected well.

To investigate the binding of NF-κB1 to the DGCR5 promoter harboring the predicted NF-κB1 binding site, we cloned the DGCR5 promoter into psiCHECK-2 (Promega) to construct a DGCR5 promoter-containing luciferase reporter vector (psiCHECK-2-proDGCR5). Cells were cotransfected with these vectors and pcDNA3.1-NF-κB1 or blank pcDNA3.1. Luciferase assays were performed 48 h after transfection to determine the activities of firefly luciferase and Renilla luciferase.

Xenograft model

To establish a human glioma cell xenograft nude mouse model, 24 female nude mice (BALB/c-nu/nu) at 6-week-old and weighing 16-20 g were obtained from the SLAC Laboratory Animal Center. A total of 24 mice were randomly divided into 4 groups: U251-MG +vector group, U251-MG +lncDGCR5 group, SHG44 +vector group, SHG44 +lncDGCR5 group. A total of 1x10⁶ transfected U251-MG or SHG44 cells were suspended in 100 μl PBS and subcutaneously injected into the right anterior armpits of the nude mice. Thirty-five days later, the animals were sacrificed and tumor weight and volume were examined and calculated. The tumor tissues were collected for protein level examination using immunoblotting. All experiments were repeated at least in triplicate. All animal experiments were performed with the approval of the ethics committee of the Sichuan Provincial People’s Hospital.

Statistical analysis

The results from at least three independent experiments were processed using GraphPad and then expressed as the means ± SD. Data were statistically analyzed by one-way analysis of variance (ANOVA) followed by Tukey's multiple comparison test or independent sample t-test. A P-value of < 0.05 was considered significantly different.