lncRNA LINC00665 Stabilized by TAF15 Impeded the Malignant Biological Behaviors of Glioma Cells via STAU1-Mediated mRNA Degradation.

Glioma is a brain cancer characterized by strong invasiveness with limited treatment options and poor prognosis. Recently, dysregulation of long non-coding RNAs (lncRNAs) has emerged as an important component in cellular processes and tumorigenesis. In this study, we demonstrated that TATA-box binding protein associated factor 15 (TAF15) and long intergenic non-protein coding RNA 665 (LINC00665) were both downregulated in glioma tissues and cells. TAF15 overexpression enhanced the stability of LINC00665, inhibiting malignant biological behaviors of glioma cells. Both metal regulatory transcription factor 1 (MTF1) and YY2 transcription factor (YY2) showed high expression levels in glioma tissues and cells, and their knockdown inhibited malignant progression. Mechanistically, overexpression of LINC00665 was confirmed to destabilize MTF1 and YY2 mRNA by interacting with STAU1, and knockdown of STAU1 could rescue the MTF1 and YY2 mRNA degradation caused by LINC00665 overexpression. G2 and S-phase expressed 1 (GTSE1) was identified as an oncogene in glioma, and knockdown of MTF1 or YY2 decreased the mRNA and protein expression levels of GTSE1 through direct binding to the GTSE1 promoter region. Our study highlights a key role of the TAF15/LINC00665/MTF1(YY2)/GTSE1 axis in modulating the malignant biological behaviors of glioma cells, suggesting novel mechanisms by which lncRNAs affect STAU1-mediated mRNA stability, which can inform new molecular therapies for glioma.

Introduction

Glioma is one of the most common and lethal primary malignant tumors of the brain, and it is classified into four categories: low-grade glioma (including categories I and II) and high-grade glioma (including categories III and IV), according to World Health Organization grading criteria., High-grade glioma cells easily infiltrate the extracellular matrix of human brain cells, which is a major challenge for achieving a radical cure, especially using traditional methods such as surgery, radiotherapy, and chemotherapy., Therefore, identifying tumor biomarkers and developing molecular targeted therapies for glioma have become important research goals for improving the prognosis of patients with glioma. In the present study, we focused on the expression of long non-coding RNAs (lncRNAs) and potential interacting factors in glioma to highlight new potential targets for treatment or diagnosis, as well as to gain new insight into the mechanisms of glioma malignant transformation.

The important roles of RNAs and their regulatory mechanisms in physiological and pathological processes, including tumorigenesis, are increasingly being revealed and recognized as candidate targets for biomarkers and cancer treatments. lncRNAs are a class of endogenous RNA molecules with transcripts longer than 200 nt and limited protein-coding potential. Several lncRNAs have been identified to act as either oncogenes or tumor suppressors, regulating the biological process of the tumor cell cycle and tumorigenesis. Accumulating evidence indicates that RNA-binding proteins (RBPs) directly bind to and stabilize lncRNAs to regulate tumor biological processes. For instance, HuR was found to stabilize lncRNA HOTAIR to promote its expression in head and neck squamous cell carcinoma. HuR also functions as a tumor facilitator in ovarian cancer by enhancing the stability of lncRNA NEAT1 to promote malignant progression. The lncRNA CRNDE could be stabilized by hnRNPUL2, thereby accelerating cell proliferation and migration in colorectal carcinoma.

The dynamic complex of RNAs and RBPs mediates virtually all stages of the RNA life cycle, and dysfunction in RBPs can cause changes to the transcriptome and proteome of the cell, leading to tumor growth and aberrant biological behaviors., TATA-box binding protein associated factor 15 (TAF15), a member of the FET family, plays an important role in regulating mRNA transcription, RNA splicing, and transportation, and it shows a high translocation rate in certain malignant tumors such as liposarcoma and Ewing’s sarcoma.13, 14, 15 TAF15 was also shown to promote the proliferation of human neuroblastoma cells via altering gene expression through a microRNA-involved pathway. Therefore, we examined the expression of TAF15 in glioma tissues and cells and used microarray analysis to identify differentially expressed lncRNAs in TAF15-overexpressed glioma cells. Among the candidates, the downregulated long intergenic non-protein coding RNA 665 (LINC00665) in glioma sparked our interest. Studies suggest that LINC00665 contributes to a high risk of oral premalignant lesions.

STAU1-mediated mRNA decay (SMD) is a common mRNA degradation process in mammalian cells, which regulates numerous biological processes, including myogenesis, keratinocyte-mediated wound healing, and adipogenesis., This pathway involves STAU1, a double-stranded RBP, that recognizes the STAU1-binding site (SBS) as a complementary double-stranded RNA structure formed either by intramolecular or intermolecular base pairing of the target mRNA 3′ untranslated region (UTR) and lncRNA. Upon binding of STAU1 to the SBS, the complex recruits UPF1 to the 3′ UTR of the target mRNA, thereby inducing mRNA degradation and shortening the half-life of target mRNA. Therefore, we examined the potential influence of LINC00665 downregulation in glioma with STAU1 expression as a potential mechanism by which RNA dysfunction mediates malignancy.

Metal regulatory transcription factor 1 (MTF1) is a highly conserved Cys2His2 zinc finger protein with six zinc finger domains, and it also plays a key role in transcriptional regulation by binding to DNA., MTF1 can be activated by growth factors, redox stress, and cytokines, thus playing a crucial role in both embryogenesis and hematopoiesis. Moreover, MTF1 has been shown to be overexpressed in breast cancer, cervical cancer, colorectal cancer, and lung cancer, suggesting a potential role in malignancy., We further focused on the potential role of the YY2 transcription factor (YY2) in glioma, as a multi-functional transcription factor of the YY family. As a paralog of YY1, YY2 has different modes of regulation on multiple tissues and cells, and it participates in the self-renewal and differentiation of mouse embryonic stem cells into cardiovascular lineages.,

Finally, we explored the influence of LINC00665 downregulation in glioma by examining its influence on microtubule-localized protein G2 and S-phase expressed 1 (GTSE1), which is expressed in the S and G2 phases of the cell cycle and is thus directly associated with carcinogenesis. Indeed, GTSE1 expression is upregulated in acral melanoma and gastric cancer., When facing DNA damage, GTSE1 accumulates and binds to p53, inhibiting its ability to induce apoptosis, suggesting its involvement in the p53 signaling pathway.

Our study first validated the expression of TAF15, LINC00665, MTF1, YY2, and GTSE1 in glioma. We further explored the potential mechanisms by which these factors interact and influence glioma malignant progression, focusing on the SMD process. These results can lay a new foundation for obtaining insights into the roles of lncRNAs and the underlying molecular mechanisms in glioma malignancy, and guide the development of innovative approaches for glioma treatment.

Results

LINC00665 Stabilized by TAF15 Inhibited Tumorigenesis in Glioma

As shown in Figures 1A and 1B, TAF15 expression was downregulated in both glioma tissues and cells. Moreover, high expression of TAF15 was negatively associated with poor prognosis in patients with glioma (Figure S1A). The microarray analysis results showed that the expression level of LINC00665 was decreased in U87 and U251 glioma cells compared to that in human astrocytes (Figure S2A). Similarly, LINC00665 expression was downregulated in glioma tissues compared with that in normal brain tissues (NBTs) and was negatively correlated with the pathological grade of glioma (Figure 1C). An RNA fluorescence in situ hybridization (FISH) assay was used to determine the subcellular location and expression of LINC00665, confirming decreased expression in U87 and U251 glioma cells compared with that in human astrocytes (Figures 1D and 1E).

Figure 1

TAF15 And LINC00665 Served as Tumor Suppressors in Glioma Cells

(A) TAF15 protein levels in normal brain tissues (NBTs), low-grade glioma tissues (LGGTs) (grade I, n = 5; grade II, n = 5), and high-grade glioma tissues (HGGTs) (grade III. n = 5; grade IV, n = 5) (∗p < 0.05, ∗∗p < 0.01 versus NBTs group; #p < 0.05 versus LGGTs group). (B) TAF15 protein levels in human astrocytes (HAs) and the U251 and U87 groups (n = 3, each group; ∗p < 0.05, ∗∗p < 0.01 versus HAs group). (C) LINC00665 expression level in glioma tissues (∗∗p < 0.01 versus NBTs group). (D) RNA FISH assay to confirm subcellular location of LINC00665 in HA, U87, and U251 cells. Scale bars represent 20 μm. (E) LINC00665 expression level in normal HAs and glioma cell lines (n = 3, each group; ∗∗p < 0.01 versus HA group). (F) A CCK-8 assay was performed to test the effect of TAF15 and LINC00665 overexpression on proliferation in U87 and U251 cells. (G) Flow cytometry analysis of U87 and U251 cells with TAF15 and LINC00665 overexpression. (H) Quantification number of migration and invasion cells treated with upregulated TAF15 and LINC00665 (n = 3, each group; ∗p < 0.05 versus TAF15+-NC group; #p < 0.05 versus LINC00665+-NC group; ▲p < 0.05 versus TAF15+ group; ▪p < 0.05 versus LINC00665+ group). Scale bars represent 200 μm.

To verify the functions of TAF15 and LINC00665 in glioma cells, the impact on cell proliferation was assessed with the Cell Counting Kit-8 (CCK-8) assay, apoptosis was assessed with flow cytometry, and migration/invasion potential was assessed with transwell assays. As expected, upregulation of TAF15 and LINC00665 expression, respectively, inhibited the proliferation, migration, and invasion of glioma cells and promoted their apoptosis (Figures 1F–1H). Quantitative real-time PCR and microarray analysis showed that LINC00665 expression was upregulated in glioma cells with TAF15 overexpression (Figure 2A; Figure S2B). Furthermore, simultaneous overexpression of TAF15 and LINC00665 resulted in weaker proliferation, migration, and invasion capacity, as well as stronger induction of apoptosis, compared with overexpression of TAF15 or LINC00665 alone (Figures 1F–1H).

Figure 2

TAF15 Stabilized LINC00665 and MTF1 Played an Oncogenic Role in Glioma Cells

(A) Relative expression of LINC00665 in glioma cells treated with TAF15 overexpression (n = 3, each group; ∗p < 0.05 versus TAF15+-NC group). (B and C) An RNA-IP assay (B) and RNA pull-down assay (C) were used to identify LINC00665 in the TAF15 complex. LINC00665 enrichment was measured using quantitative real-time PCR (n = 3, each group; ∗∗p < 0.01 versus anti-IgG group). (D) Expression level of nascent LINC00665 was measured by quantitative real-time PCR (n = 3, each group; p > 0.05 versus TAF15+-NC group). (E) The half-life of LINC00665 in the U87 glioma cells (left) and U251 glioma cells (right) treated with TAF15 overexpression. (F) MTF1 expression levels in NBTs, LGGTs, and HGGTs are shown (∗∗p < 0.01 versus NBTs group; ##p < 0.01 versus LGGTs group). (G) MTF1 expression levels in HA, U87, and U251 cell lines are shown (n = 3, each group; ∗∗p < 0.01 versus HA group). (H) A CCK-8 assay was used to measure the effect of MTF1 on the proliferation of glioma cells. (I) The apoptotic percentages of glioma cells were detected with MTF1 upregulation or downregulation. (J) A transwell assay was used to measure the effect of MTF1 on cell migration and invasion of U87 and U251 glioma cells (n = 3, each group; ∗p < 0.05 versus MTF1+-NC group; #p < 0.05 versus MTF1−-NC group). Scale bars represent 200 μm.

starBase was used to predict the existence of the binding site between LINC00665 and TAF15 (Figure S1B), and their interaction was confirmed with RNA immunoprecipitation (RNA-IP) and RNA pull-down assays. As shown in Figure 2B, the enrichment of LINC00665 was significantly higher in the anti-TAF15 group compared with that in the anti-immunoglobulin G (IgG) group. The RNA pull-down assay also revealed that TAF15 could bind to LINC00665 (Figure 2C). Therefore, we hypothesized that LINC00665 was involved in the TAF15-mediated regulation of glioma cells. In exploring the mechanism in this regulation network, we found no significant difference of the nascent LINC00665 expression level between the TAF15 overexpression group and the NC group (Figure 2D). However, the half-life of LINC00665 was found to be significantly prolonged in the TAF15+ group compared with that in the TAF15+-negative control (NC) group (Figure 2E). These results confirmed that LINC00665 stabilized by TAF15 could inhibit the malignant progression of glioma cells.

MTF1 Played an Oncogenic Role in Glioma Cells

MTF1 expression was upregulated in both the glioma tissues and cells, and RNA FISH assays confirmed the high expression level of MTF1 in glioma cell lines (Figures 2F and 2G; Figure S2C). Moreover, high expression of MTF1 was positively associated with a poor prognosis in patients with glioma (Figure S1C). Compared with the NC group, the proliferation rate of glioma cells in the MTF1 overexpression group, MTF1+, was significantly higher, whereas that in the MTF1 knockdown group, MTF1−, was significantly decreased (Figure 2H). In addition, glioma cells in the MTF1+ group showed significantly lower rates of apoptosis, along with increased numbers of migrating and invading cells, whereas knockdown of MTF1 significantly promoted apoptosis with reduced migration and invasion capacity (Figures 2I and 2J).

LINC00665 Impaired the Association of MTF1 with STAU1, Thus Regulating Malignant Progression of Glioma Cells

Interestingly, overexpression of LINC00665 depleted the expression of MTF1 at both the mRNA and protein levels, whereas knockdown of STAU1 significantly increased MTF1 expression and could partially rescue the LINC00665-induced reduction of MTF1 levels (Figures 3A and 3B). To explore the mechanism of this LINC00665/MTF1 axis, an RNA-IP assay was used to verify the association among LINC00665, MTF1, and STAU1. The relative enrichment levels of LINC00665 and MTF1 were both significantly increased in the anti-STAU1 group compared to those in the anti-IgG group (Figure 3C). Furthermore, luciferase gene reporter assays were performed to clarify the association between LINC00665 and the MTF1 3′ UTR, and to identify the SBS. As shown in Figure 3D, the relative luciferase activity in the MTF1-3′ UTR-wild-type (WT) + LINC00665 group was significantly decreased compared with that in the NC group, whereas there was no such difference between the MTF1-3′ UTR-mutant-type (Mut) + LINC00665 and control groups. After treatment with actinomycin D, quantitative real-time PCR was used to compare the half-life of MTF1 mRNA in the two groups, showing a significantly shortened half-life in the LINC00665+ group and a prolonged half-life in the STAU1− group (Figures 3E and 3F). Considering the potential involvement of UPF1 in the SMD pathway, we also measured the half-life along with the mRNA and protein expression levels of MTF1 in glioma cells with UPF1 knockdown, demonstrating the expected results (Figures S3A–S3C).

Figure 3

LINC00665 Destabilized MTF1 mRNA by Interacting with STAU1, thus Regulating Glioma Cell Malignant Progression

(A and B) Quantitative real-time PCR (A) and western blot (B) were applied to test expression levels of MTF1 (∗p < 0.05 versus LINC00665+-NC group; #p < 0.05 versus STAU1−-NC group; ▲p < 0.05 versus LINC00665+ group). (C) LINC00665 and MTF1 could bind to STAU1 protein, respectively. Both relative enrichment levels of LINC00665 and MTF1 were measured using quantitative real-time PCR (n = 3, each group; ∗∗p < 0.01 versus anti-IgG group). (D) The predicted LINC00665 binding site in MTF1 (MTF1-3′ UTR-WT) and the designed mutant sequence (MTF1-3′ UTR-Mut) are shown (left). Luciferase reporter gene assays of HEK293T cells are shown (right) (n = 3, each group; ∗p < 0.05 versus MTF1-3′ UTR-WT + LINC00665-NC group). (E) Remaining MTF1 mRNA (%) at the different actinomycin D treatment times in control group, LINC00665+-NC group, and LINC00665+ group. (F) Remaining MTF1 mRNA (%) in control group, STAU1−-NC group, and STAU1− group. (G–I) CCK-8 (G), flow cytometry (H), and transwell assays (I) were used to assess the proliferation, apoptosis, migration, and invasion capacity of U87 and U251 glioma cells (n = 3, each group; ∗p < 0.05 versus control group; #p < 0.05 versus LINC00665+ + MTF1+ group; ▲p < 0.05 versus LINC00665− + MTF1− group). Scale bars represent 200 μm.

These results confirmed that LINC00665 destabilized MTF1 mRNA by forming a complex with STAU1 to recruit UPF1. Thus, we next investigated whether MTF1 could rescue the LINC00665-induced inhibition of the malignant progression of glioma cells. The LINC00665+ + MTF1− group showed a decreased proliferation rate, increased apoptosis rate, and reduced migration and invasion capacity compared with those of the LINC00665+ + MTF1+ group. In contrast, the LINC00665− + MTF1+ group showed significantly enhanced proliferation, migration, and invasion, with reduced apoptosis compared with those of the LINC00665− + MTF1− group (Figures 3G–3I). These results indicated that LINC00665 could inhibit the malignant behaviors of glioma cells via STAU1-mediated MTF1 mRNA degradation.

YY2 Was a Target of LINC00665 via STAU1-Mediated mRNA Decay to Modulate Malignant Biological Behaviors

Western blot and RNA FISH assays demonstrated that YY2 was upregulated in both glioma tissues and cells (Figures 4A and 4B; Figure S2D). Moreover, high expression of YY2 was positively associated with poor prognosis in patients with glioma (Figure S1D). Glioma cells with YY2 overexpression also showed a significantly increased proliferation rate compared with that of the cells transfected with the NC vector (Figure 4C). Moreover, overexpression of YY2 significantly inhibited the apoptosis of glioma cells, whereas knockdown of YY2 promoted apoptosis (Figure 4D). Similarly, overexpression of YY2 significantly increased the numbers of migrating and invading glioma cells, which were reduced with YY2 knockdown (Figure 4E).

Figure 4

Knockdown of YY2 Inhibited Malignant Progression of Glioma Cells

(A) YY2 expression levels in NBTs and glioma tissues are shown (∗∗p < 0.01 versus NBTs group; #p < 0.05 versus LGGTs group). (B) YY2 expression levels in HA, U87, and U251 cell lines are shown (n = 3, each group; ∗p < 0.05, ∗∗p < 0.01 versus HA group). (C) A CCK-8 assay was used to measure the effect of YY2 on the proliferation of glioma cells. (D) The apoptotic percentages of glioma cells were detected with YY2 upregulation or downregulation. (E) A transwell assay was used to measure the effect of YY2 on cell migration and invasion of U87 and U251 glioma cells (n = 3, each group; ∗p < 0.05 versus YY2+-NC group; #p < 0.05 versus YY2−-NC group). Scale bars represent 200 μm. (F and G) Quantitative real-time PCR (F) and western blot (G) were performed to observe YY2 expression variation (n = 3, each group; ∗p < 0.05 versus LINC00665+-NC group; #p < 0.05 versus STAU1−-NC group; ▲p < 0.05 versus LINC00665+ group).

Overexpression of LINC00665 in glioma cells resulted in downregulated expression of the transcription factor YY2. Furthermore, when STAU1 was knocked down in U87 and U251 glioma cells, YY2 expression was upregulated compared with that in the STAU1− NC group. Moreover, LINC00665 overexpression combined with STAU1 inhibition could rescue the reduction in MTF1 mRNA and protein expression levels compared with that in the LINC00665+ group (Figures 4F and 4G).

Using the bioinformatics software RepeatMasker and IntaRNA, we analyzed the putative SBS between LINC00665 and YY2, and luciferase gene reporter assays were used to precisely identify the SBS to clarify the interaction between LINC00665 and the YY2 mRNA 3′ UTR (Figure 5A). An RNA-IP assay was used to further investigate the effect of the combination of LINC00665, YY2, and STAU1, showing that the relative enrichment of both LINC00665 and YY2 was significantly increased in the anti-STAU1 group compared to those in the anti-IgG group (Figure 5B). Moreover, compared with the LINC00665+-NC group, overexpression of LINC00665 could crucially reduce the half-life of YY2 (Figure 5E), whereas STAU1 or UPF1 inhibition in glioma cells could substantially extend the half-life of YY2 mRNA (Figure 5D; Figures S3D–S3F).

Figure 5

YY2 Was Destabilized by LINC00665 via the SMD Pathway Regulating Malignant Progression of Glioma

(A) The predicted LINC00665 binding site in YY2 (YY2-3′ UTR-WT) and the designed mutant sequence (YY2-3′ UTR-Mut) are shown (left). Luciferase reporter gene assays of HEK293T cells are shown (right) (n = 3, each group; ∗p < 0.05 versus YY2-3′ UTR-WT + LINC00665-NC group). (B) RNA-IP assay was conducted to determine the association among LINC00665, YY2, and STAU1 (n = 3, each group; ∗∗p < 0.01 versus anti-IgG group). (C) The half-life of YY2 in the control group, LINC00665+-NC group, and LINC00665+ group. (D) The half-life of YY2 in the control group, STAU1−-NC group, and STAU1− group. (E–G) CCK-8 (E), flow cytometry (F), and transwell assays (G) were conducted to determine the effect of co-transfection of LINC00665 and YY2 on the proliferation, apoptosis, migration, and invasion of glioma cells (n = 3, each group; ∗p < 0.05 versus control group; #p < 0.05 versus LINC00665+ + YY2+ group; ▲p < 0.05 versus LINC00665− + YY2− group). Scale bars represent 200 μm.

Consistent with our hypothesis, the proliferation of glioma cells in the LINC00665+ + YY2− group was significantly decreased compared with that in the LINC00665+ + YY2+ group (Figure 5E). However, there was a higher ratio of apoptotic cells in the LINC00665+ + YY2− group compared with that in the LINC00665+ + YY2+ group (Figure 5F). Moreover, glioma cells in LINC00665− + YY2+ group showed increased migration and invasion rates compared with those in the LINC00665− + YY2− group (Figure 5G). These results clarified that LINC00665 also inhibited YY2 mRNA degradation through the SMD pathway to regulate the malignant behaviors of glioma cells.

Knockdown of GTSE1 Inhibited Cell Proliferation, Migration, and Invasion while Promoting Apoptosis of Glioma Cells

Compared with NBTs and human astrocytes, the expression level of GTSE1 in both glioma tissues and cells, respectively, was significantly increased (Figures 6A and 6B). The Gene Expression Profiling Interactive Analaysis (GEPIA) database shows that GTSE1 is upregulated in glioblastoma and high expression of GTSE1 was positively associated with poor prognosis in patients with glioma (Figures S1E and S1F). Knockdown of GTSE1 inhibited glioma cell proliferation and promoted apoptosis compared with those in the GTSE1−-NC group (Figures 6C and 6D). Moreover, the numbers of migrating and invading cells was reduced in the GTSE1− group compared with those in the GTSE1-NC group (Figure 6E).

Figure 6

GTSE1 Was Upregulated in Glioma Cells, and Knockdown of GTSE1 Inhibited Malignant Biological Behaviors

(A) GTSE1 expression levels in NBTs, LGGTs, and HGGTs are shown (∗∗p < 0.01 versus NBTs group; ##p < 0.01 versus LGGTs group). (B) GTSE1 expression levels in HA, U87, and U251 cell lines are shown (n = 3, each group; ∗∗p < 0.01 versus HA group). (C) CCK-8 assay was performed to assess the effect of GTSE1 knockdown on the proliferation of glioma cells. (D) The ratio of apoptotic glioma cells were detected with GTSE1 downregulation. (E) Number of migrated and invaded glioma cells was counted to represent the effect of GTSE1 inhibition on cell migration and invasion (n = 3, each group; ∗p < 0.05 versus GTSE1−-NC group). Scale bars represent 200 μm. (F) The expression level of GTSE1 mRNA in glioma cells treated with MTF1 variation. (G) The expression level of GTSE1 protein in glioma cells are shown (n = 3, each group; ∗p < 0.05 versus MTF1+-NC group; #p < 0.05 versus MTF1−-NC group). (H) The expression level of GTSE1 mRNA in glioma cells treated with YY2 variation. (I) The expression level of GTSE1 protein in glioma cells treated with YY2 variation (n = 3, each group; ∗p < 0.05 versus YY2+-NC group; #p < 0.05 versus YY2−-NC group).

MTF1 and YY2 Bound to the GTSE1 Promoter Region and Induced Its Expression

We next evaluated the potential regulatory effect of MTF1 and YY2 on GTSE1. Western blot and quantitative real-time PCR showed that both MTF1 and YY2 overexpression increased the mRNA and protein expression levels of GTSE1, whereas simultaneous knockdown of MTF1 and YY2 reduced GTSE1 levels (Figures 6F–6I). The bioinformatics databases DataBase of Transcriptional Start Sites (DBTSS) and JASPAR were used to identify the putative binding site in the region 2,000 bp upstream of the GTSE1 transcription start site, indicating that MTF1 could directly bind to the GTSE1 promoter region through putative binding site 1, and this interaction was confirmed with a chromatin immunoprecipitation (ChIP) assay (Figure 7A). Since LINC00665 and MTF1 showed opposite functions in regulating the malignant progression of glioma, we hypothesized that downregulation of MTF1 expression might rescue the LINC00665-mediated reduction in GTSE1 expression. In line with this expectation, the LINC00665+ + MTF1− group showed stronger reductions in GTSE1 mRNA and protein expression levels compared with the LINC00665+ + MTF1+ group, and their expression levels were significantly higher in the LINC00665− + MTF1+ group than those in the LINC00665− + MTF1− group (Figures 7B and 7C).

Figure 7

MTF1 and YY2 Bound to the GTSE1 Promoter and Regulated the Expression Level of GTSE1

(A) Schematic representation of the human GTSE1 promoter region 2,000 bp upstream of the transcription start site (TSS), which was designated as +1. Putative MTF1 binding site was indicated. PCR was conducted with the resulting precipitated DNA. (B) The change of GTSE1 mRNA expression level caused by LINC00665 and MTF1 variation. (C) The change of GTSE1 protein expression level caused by LINC00665 and MTF1 variation (n = 3, each group; ∗p < 0.05 versus control group; #p < 0.05 versus LINC00665+ + MTF1+ group; ▲p < 0.05 versus LINC00665− + MTF1− group). (D) YY2 bound to the promoter of GTSE1 in glioma cells. Three putative YY2 binding sites were indicated. PCR was conducted with the resulting precipitated DNA. (E) The change of GTSE1 mRNA expression level caused by LINC00665 and YY2 variation. (F) The change of GTSE1 protein expression level caused by LINC00665 and YY2 variation (n = 3, each group; ∗p < 0.05 versus control group; #p < 0.05 versus LINC00665+ + YY2+ group; ▲p < 0.05 versus LINC00665− + YY2− group).

DBTSS and JASPAR indicated that the GTSE1 promoter contains three putative binding sites for YY2, and the ChIP assay showed that YY2 could directly bind to the GTSE1 promoter at putative binding sites 2 and 3, whereas there were no interactions observed in the NC or at putative binding site 1 (Figure 7D). Similarly, both GTSE1 mRNA and protein levels were decreased in the LINC00665+ + YY2− group compared with those of the LINC00665+ + YY2+ group, and the expression levels were higher in the latter group (Figures 7E and 7F). These results indicated that both MTF1 and YY2 participate in the transcriptional promotion of GTSE1.

Furthermore, we verified that GTSE1 overexpression might rescue the LINC00665-mediated reduction in GTSE1 expression. In line with this expectation, the LINC00665+ + GTSE1+ group increased GTSE1 expression compared with the LINC00665+ + GTSE1+-NC group (Figure S4A). Additionally, the LINC00665+ + GTSE1+ group showed an increased proliferation rate, decreased apoptosis rate, and enhanced migration and invasion capacity compared with those of the LINC00665+ + GTSE1+-NC group (Figures S4B–S4D).

MTF1 or YY2 Knockdown Combined with TAF15 and LINC00665 Overexpression Reduced Tumor Growth In Vivo

Finally, we evaluated the influence of these four factors in vivo using a nude mice glioma xenograft model. As shown in Figure 8A, nude mice in the groups injected with TAF15+, LINC00665+, MTF1−, and YY2− glioma cells developed smaller tumors compared with those of the control group. Furthermore, the TAF15+ + LINC00665+ + MTF1− and TAF15+ + LINC00665++YY2− groups had the smallest tumor volumes overall (Figure 8B). The survival analysis indicated that mice in all of the treatment groups achieved longer survival than did those in the control group, although the nude mice in the TAF15+ + LINC00665+ + MTF1− and TAF15+ + LINC00665++YY2− groups showed the longest survival time (Figure 8C). Furthermore, we evaluated the influence of GTSE1 knockdown in vivo and found that nude mice in the groups injected with GTSE1− glioma cells developed smaller tumors and gained longer survival time than did those in the control group (Figures S5A–S5C).

Figure 8

In Vivo Study and the Schematic Picture of the Mechanisms in Our Study

(A) Nude mice carrying tumors from respective groups are shown. The sample tumors from respective groups are shown. (B) Tumor volume was calculated every 4 days after injection, and tumors were excised after 44 days (∗p < 0.05 versus control group; #p < 0.05 versus TAF15+ group; ▲p < 0.05 versus LINC00665+ group; ▪p < 0.05 versus MTF1+ group; ●p < 0.05 versus YY2+ group). (C) The survival curves of nude mice injected into the right striatum. (D) The schematic picture of the mechanism in which TAF15 combined with LINC00665, thus enhancing the degradation of MTF1 and YY2 mRNA via the SMD pathway in glioma cells. Ter, translation termination codon.

Discussion

In this study, we identified the role of the TAF15/LINC00665/MTF1(YY2)/GTSE1 axis in the malignant progression of glioma, highlighting new targets for research and molecular therapy. TAF15 and LINC00665 were downregulated in glioma tissues and cells; moreover, TAF15 overexpression enhanced the stability of LINC00665, thus inhibiting the malignant progression of glioma cells. In contrast, the transcription factors MTF1 and YY2 were upregulated in glioma cells, and their knockdown, respectively, inhibited malignant progression. Mechanistically, overexpression of LINC00665 was confirmed to destabilize MTF1 and YY2 mRNAs by interacting with STAU1. In line with this mechanism, depletion of STAU1 could rescue the MTF1 and YY2 mRNA degradation caused by LINC00665 overexpression. GTSE1 was upregulated in glioma. However, knockdown of MTF1 or YY2 decreased the mRNA and protein expression levels of GTSE1 through direct binding to its promoter region, thus repressing proliferation, migration, and invasion and promoting the apoptosis of glioma cells. A schematic of the proposed mechanisms driven by this axis is provided in Figure 8D.

TAF15 belongs to the FET family of conserved RBPs that play essential roles in the tumorigenesis of various cancers such as liposarcoma and neuroectodermal tumor., Our study verified that TAF15 was downregulated in glioma tissues and cells, and overexpression of TAF15 inhibited the malignant progression of glioma cells. TAF15 was also previously shown to be downregulated in human differentiating embryonic stem cells.

Our study revealed that LINC00665 is downregulated in glioma tissues and cells, and overexpression of LINC00665 inhibited the malignant biological behaviors of glioma cells. In previous studies, LINC00665 functioned as an oncogene in non-small-cell lung cancer and lung adenoma and was associated with poor prognosis in hepatocellular carcinoma.33, 34, 35 However, a recent study indicated that LINC00665 knockdown promoted migration and invasion in triple-negative breast cancer MDA-MB-231 cells, which is consistent with our experimental results. Likewise, lncRNA GAS5 is downregulated in glioma, breast cancer, and hepatocellular carcinoma, but it promotes proliferation, migration, and invasion by regulation of miR-301a in esophageal cancer.37, 38, 39, 40 These results suggest that the same lncRNA may show differential gene expression and contribute to inverse functions in different cancers. At present, dysregulation of lncRNAs has emerged as an important component in cellular processes and tumorigenesis. Our previous studies provided some insight into the regulatory mechanisms of various lncRNAs on the biological behaviors of glioma cells. PVT1, which is overexpressed in glioma, was found to promote glioma cell survival by sponging miR-190a-5p and miR-488-3p; HOXA-AS2 increased the malignant behaviors of glioma via the miR-373/EGFR axis; and SOX2OT combined with miR-194-5p and miR-122, respectively, facilitated glioma cell growth.41, 42, 43

lncRNAs regulate downstream gene functions, including the well-established regulation of the competing endogenous RNA (ceRNA) pathway. Our study revealed that overexpressed LINC00665 significantly decreased the mRNA expression levels of MTF1 and YY2, while STAU1 knockdown rescued their depletion induced by LINC00665 overexpression. Further investigation of the underlying mechanism demonstrated that LINC00665 might facilitate the degradation of MTF1 and YY2 mRNA by interacting with STAU1. The SMD pathway is a translation-dependent mechanism that ubiquitously exists in mammalian cells. Alu elements on lncRNAs were suggested to directly bind to Alu elements on the 3′ UTR of the target mRNA in the complementary base-pairing pattern via a STAU1-binding site. When translation is terminated sufficiently upstream of the SBS, STAU1 bound to the SBS might recruit and bind to UPF1, inducing the mRNA degradation process and therefore shortening the half-life of the target mRNA and activating the SMD pathway.46, 47, 48 Using the bioinformatics software RepeatMasker and IntaRNA, we predicted the putative SBS between the Alu element on LINC00665 and the Alu element on either the MTF1 or YY2 mRNA 3′ UTR, which was confirmed by luciferase gene reporter assays. The RNA-IP assay confirmed the associations among LINC00665, STAU1, and MTF1(YY2), and both LINC00665 overexpression and STAU1 inhibition influenced MTF1 and YY2 mRNA degradation based on half-life assays. Regulation networks of the SMD pathway involving lncRNAs have attracted increasing research attention. SNHG5 was reported to be upregulated in colorectal cancer cells and to participate in STAU1-mediated mRNA destabilization. In addition, lncRNAs participating in the SMD pathway were suggested to play a role in the pathogenesis of ischemic stroke.

A previous immunohistochemistry study showed that MTF1 expression is relatively high in human cancers such as lung, breast, and cervical carcinomas compared to that of adjacent normal tissues., MTF1 facilitates the proliferation of murine fibrosarcoma cells, and knockdown of MTF1 suppressed the malignant progression of advanced-stage head-and-neck squamous cell carcinoma., YY2, a transcription factor with activation and repression domains, plays transactivation or transcriptional inhibition roles in the regulation of target genes. Elevated expression of WT YY2 was reported to promote neuronal cell death and change the direction of neuronal differentiation in vitro. Moreover, analysis of 698 glioma tissue samples from The Cancer Genome Atlas database indicated that YY2 is overexpressed in glioma tissues, in accordance with the results from the present study.

Since the promoter region of GTSE1 harbors diverse putative binding sites, a ChIP assay was used to confirm that MTF1 and YY2, respectively, directly bind to the GTSE1 promoter in a sequence-specific manner. We further showed that either MTF1 or YY2 overexpression increased the mRNA and protein expression levels of GTSE1. In line with the present study, high expression levels of GTSE1 have been observed in lung cancer and bladder cancer., Moreover, upregulation of GTSE1 expression was positively related to advanced tumor stages, and knockdown of GTSE1 inhibited the proliferation, migration, and invasion of bladder cancer cells, exerting a similar effect on malignant progression as found in our study with glioma cells. Thus, MTF1 and YY2 can act as transactivators in regulating GTSE1 expression. MTF1 and YY2 are well established to function as transcription factors. MTF1 specifically binds to the promoter region on the Ins sequence and activates insulin gene transcription. Knockdown of YY2 decreased the level of Myc mRNA in line with the function of YY1 as an initiator-binding transcription factor on the Myc promoter.

In summary, our study verified that TAF15 and LINC00665 are downregulated in glioma tissues and cells, whereas MTF1, YY2, and GTSE1 are upregulated. Overexpression of TAF15 stabilized LINC00665, thus increasing its expression level to reduce the STAU1-mediated mRNA degradation of both MTF1 and YY2, further restraining the transcription of GTSE1, and ultimately disrupting the malignant progression of glioma. Moreover, we found that TAF15 overexpression, LINC00665 overexpression, MTF1(YY2) inhibition, and the combination of these three conditions all significantly reduced xenograft tumor growth in nude mice and prolonged their survival, with the combination group producing the smallest tumor size and longest survival. These results highlight the potential clinical application values of TAF15, LINC00665, MTF1, and YY2 respectively and cooperatively, which can provide novel strategies for molecular therapies for glioma.

Materials and Methods

Human lncRNA Microarrays

Samples including normal human astrocytes and U251 and U87 glioma cells, which were prepared for lncRNA analysis, were sent to KangChen Bio-tech (Shanghai, China) to perform the microarray hybridization.

Cell Culture and Human Tissue Samples

We purchased human U251 and U87 glioma cells and the HEK293T cell line from the China Academy of Chinese Medical Sciences and cultured these cells in DMEM high-glucose medium with 10% fetal bovine serum (FBS) (Gibco, USA) added. Normal human astrocytes were purchased from the ScienCell Research Laboratories (Carlsbad, CA, USA) and grown in RPMI 1640 culture medium (Gibco, USA) supplemented with 10% FBS. Stable 37°C humidified incubators with 5% CO2 were applied to culture the above cells. In this study, five samples in each grades of glioma tissues were applied, and the glioma tissues were classified according to the 2016 World Health Organization (WHO) classification by neuropathologists. Normal brain tissues and glioma tissues were obtained from patients who signed informed consents while hospitalized in the Department of Neurosurgery of Shengjing Hospital of China Medical University; moreover, our study was approved by the Ethics Committee of China Medical University. All assays on nude mice were performed strictly according to the protocol approved by the Administrative Panel on Laboratory Animal Care of the Shengjing Hospital.

RNA Extraction and Quantitative Real-Time PCR

Our study used TRIzol reagent (Life Technologies, USA) to extract total RNA from cells and tissues. A NanoDrop spectrophotometer (Thermo Fisher Scientific, USA) was used to determine the concentration and quality of extracted RNA with absorbance at 260/280 nm. We utilized a One-Step SYBR PrimeScript RT-PCR kit (Takara Bio, Japan) to measure the relative expression levels of TAF15, LINC00665, STAU1, MTF1, YY2, and GTSE1, as well as the expression of GAPDH as an endogenous control, with an Applied Biosystems 7500 Fast RT-PCR system. Fold change was calculated in the relative quantification (2−ΔΔCt) manner.

Western Blot

Total protein of cells was extracted with a radioimmunoprecipitation assay (RIPA) buffer containing protease inhibitor under the condition of an ice bath. The extracted protein and loading buffer were added and transferred onto 0.22-mm methanol-excited polyvinylidene fluoride membranes after SDS-PAGE electrophoresis, and then membranes were blocked with 5% nonfat milk. Next, we incubated membranes with the following primary antibodies at 4°C overnight: TAF15 (1:1,000) (Cell Signaling Technology, USA), MTF1 (1:1,000) (Proteintech, USA), YY2 (1:200) (Santa Cruz Biotechnology, USA), GTSE1 (1:1,000) (Proteintech, USA), and GAPDH (1:10,000) (Proteintech, USA), followed by incubation with related horseradish peroxidase (HRP)-conjugated secondary antibodies, including goat anti-rabbit or goat anti-mouse secondary antibody (1:10,000) (Proteintech, USA) at room temperature for 1.5 h. The relative integrated density values (IDVs) were observed based on GAPDH as a control.

Cell Transfection

TAF15 full-length plasmid TAF15+, MTF1 full-length plasmid MTF1+, YY2 full-length plasmid YY2+, and their respective expression vectors (NCs) were designed and synthesized in GenScript (Nanjing, China), while LINC00665 full-length plasmid LINC00665+ and its NC were designed and synthesized in GeneChem (Shanghai, China). Short-hairpin LINC00665 (sh-LINC00665), MTF1 (sh-MTF1), YY2 (sh-YY2), GTSE1 (sh-GTSE1), STAU1 (sh-STAU1), and UPF1 (sh-UPF1) plasmids and their respective NCs (sh-NCs) were designed and synthesized in GenePharma (China) (Table 1). Cell transfections were performed as previously described. Briefly, cells were seeded into 24-well plates (Corning Life Sciences) and then transfected with the reagent Lipofectamine 3000 (Life Technologies, USA). G418 and puromycin were used to screen stable transfected cells.

Table 1

shRNA Sequences

Name	Target Sequence
sh-NC	5′-GTTCTCCGAACGTGTCACGTC-3′
sh-LINC00665	5′-GCGTCCACGGGTGGGAAATTG-3′
sh-MTF1	5′-GGATACAAATCACTCACTTTG-3′
sh-YY2	5′-GCCTCCAACGAAGATTTCTCC-3′
sh-GTSE1	5′-GCCTACTCCTACAAATCAATT-3′
sh-STAU1	5′-GCCGCAGGGAGTTTGTGATGC-3′
sh-UPF1	5′-GCGAGAAGGACTTCATCATCC-3′

RNA FISH Assays

Alexa Fluor 555-labeled LINC00665 probes and Alexa Fluor 488-labeled MTF1 and YY2 probes were designed and synthesized by RiboBio (Guangzhou, China). The probe signals were determined with a FISH kit (RiboBio, Guangzhou, China) according to the manufacturer’s instruction.

Reporter Vector Construction and Luciferase Reporter Gene Assays

The WT or Mut luciferase expression vector of the putative binding site in MTF1 and YY2 mRNA 3′ UTR was constructed and co-transfected into HEK293T cells with LINC00665 and its NC plasmid, respectively, and luciferase reporter gene system analysis was performed. These cells were divided into five groups, respectively: control, MTF1-3′ UTR-WT + LINC00665-NC, MTF1-3′ UTR-WT + LINC00665, MTF1-3′ UTR-Mut + LINC00665-NC, and MTF1-3′ UTR-Mut + LINC00665 groups; and control, YY2-3′ UTR-WT + LINC00665-NC, YY21-3′ UTR-WT + LINC00665, YY2-3′ UTR-Mut + LINC00665-NC, and YY2-3′ UTR-Mut + LINC00665 groups.

RNA-IP Assay

Briefly, glioma cells were transfected with WT LINC00665 or Mut LINC00665 plasmid with the putative binding site mutation. Cell lysates were obtained in which magnetic bead-conjugated TAF15 and STAU1 antibodies (IgG as a control) were added, respectively. Protein complexes were obtained after incubation, centrifugation, and rinsing. Proteinase K digestion and extraction of RNA was performed with TRIzol reagent. The expression level of LINC00665 was detected by quantitative real-time PCR.

RNA Pull-Down Assay

The biotin-labeled wild-type LINC00665 probe or a mutant LINC665 probe with a putative binding site mutation, which is coated with streptavidin-labeled magnetic beads, was added into cell lysate. Protein complexes were enriched that bind to LINC00665. The protein expression level of TAF15 was detected by western blot.

Nascent RNA Capture Assay

A Click-iT nascent RNA capture kit and quantitative real-time PCR were used to determine the relative expression of nascent LINC00665 according to the manufacturer’s instructions.

Cell Proliferation

The CCK-8 method was used to examine cell cytotoxicity and proliferation ability. We prepared cell suspension counts, inoculated them into 96-well plates, and put them back in the cell incubator to wait for cells to adhere. In accordance with the experiment group, different kinds of plasmid were transfected into U87 and U251 cells, maintaining a total volume of 100 μL. 10 μL of CCK-8 was added and incubated for 1–2 h. The wavelength of the plate reader was selected at 450 nm to determine the optical density (OD) value.

Apoptosis

The cells were digested with trypsin without EDTA. We used serum-containing medium to stop the digestion. After centrifugation at 1,000 rpm for 3 min, the supernatant was discarded and gently washed with cold PBS. Using preconstituted 1× binding buffer, we made a cell suspension with a final concentration of 1 × 107 cells/mL, added 10 μL of annexin V-PE (phycoerythrin), and cultured it for 15 min on ice in a dark room. Then, we added 10 μL of 7-aminoactinomycin D (7-AAD) with 380 μL of cold binding buffer. Cell apoptosis was detected within 1 h using flow cytometry.

Transwell Assay

Digested glioma cells, about 20,000 cells per well, were seeded into the polycarbonate membrane chambers. When the cell migration ability was measured, the upper chamber contained DMEM high-glucose medium without FBS, whereas the lower chamber contained high-glucose medium with 10% FBS. When the cell-invasive ability was detected, Matrigel was additionally added to the polycarbonate membrane chambers. After 24 h of culture, the cells were removed and cell migration could be observed under a microscope.

ChIP Assay

MTF1 and YY2 antibodies were used to bind to DNA-binding proteins in glioma cells, respectively, and the complex was precipitated and separated, with reverse crosslinks releasing the DNA. The oligonucleotide sequence of the 5′ flanking sequence of the GTSE1 promoter containing the putative MTF1 and YY2 binding sites was separately amplified by PCR to confirm whether it coprecipitated with MTF1 and YY2 antibodies (Table 2).

Table 2

ChIP Primers

Name	Forward Primers (5′→3′)	Reverse Primers (5′→3′)	Product Size (bp)
Control	CTCCTGACCTCGTGATCTGC	TGCAGTGAGCCAAGATGTCG	150
MTF1 binding site	AAGCTGTTTGACCACACCCA	AGACTGGGGGCTCAAAAAGG	126
YY2 binding site 1	GCTCACAGTTCTGCAGGCTT	GTTCAGCACCATCCCCTTGG	215
YY2 binding site 2	AAAACCAGGGACATGGGACC	GGACCCGCTCTGAAAAGGA	172
YY2 binding site 3	TGTTAGTAACAGCTCGCGCG	GTCACTAGGTATCCGCGTCA	153

Tumor Xenografts in Nude Mice

Lentivirus encoding TAF15 and LINC00665 was generated with a pLenti6.3/V5eDEST Gateway vector kit (Life Technologies). sh-MTF1 and sh-YY2 were ligated into the LV3-CMV-GFP-Puro vector (GenePharma), respectively. Thus, target lentivirus vectors were generated to obtain the stable expressing cells of precursor (pre-)TAF15, pre-LINC00665, and sh-MTF1 or sh-YY2. Four-week-old BALB/c athymic nude mice were purchased from Beijing Vital River Laboratory Animal Technology (Beijing, China). These nude mice were divided into seven groups: control, TAF15+, LINC00665+, MTF1−, YY2−, TAF15+ + LINC00665+ + MTF1−, and TAF15+ + LINC00665+ + YY2− groups. 3 × 105 (100 μL) cells were injected subcutaneously into the right limb of nude mice. We observed and recorded the tumor formation time and tumor formation of nude mice in each group, calculating the tumor formation rate and detecting the weight and volume of transplanted tumor to draw the tumor growth curve [tumor volume (mm3) = (longest diameter × shortest diameter)2 × 0.5]. The number of survived nude mice was recorded, and survival analysis was performed using a Kaplan-Meier survival curve.

Statistical Analysis

All data in our study are presented as mean ± SD. SPSS 22.0 statistical software (IBM, New York, NY, USA) was used to perform the statistical analyses with a t test or ANOVA. Once p <0.05, differences in data were considered to be significant.

Author Contributions

Y.X. contributed to the experiment design, manuscript draft, and data analysis. X.R. contributed to the experiment implementation, manuscript draft, and data analysis. Y.L. and J.Z. participated in designing the experiment. X.L., Q.H., C.Y., and D.W. performed the experiment. L.L., J.M., H.C., Z.L., and J.L. analyzed the data. X.R. 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.