Introduction

Glioblastoma multiforme (GBM), as the most prevalent and malignant category of central nervous system (CNS) tumours, has a very poor prognosis [1]. Every year, 3.19 cases of GBM are diagnosed per 100,000 people on average, and the median age at diagnosis is 64 [2]. To date, it is expected that the median overall survival for GBM is only 15 months after diagnosis [3]. Currently, surgical resection, adjuvant chemotherapy using temozolomide (TMZ) and radiation therapy make up the conventional treatment for GBM; however, surgery fails to cure GBM patients, and TMZ therapy always complies with the emergence of acquired resistance [4, 5]. Therefore, it is worthwhile to explore a better method for the diagnosis and therapeutic improvement of GBM. The 2021 World Health Organization Classification of Tumours of the Central Nervous System (2021 CNS WHO) advances the importance of molecular diagnostics in classifying CNS tumours [6]. However, based on our knowledge, few biomarkers have been discovered as effective treatment for GBM. Thus, it is urgent for us to digest GBM-related biomarkers and explore their biological functions in the tumorigenesis of GBM.

LncRNAs refer to RNA transcripts longer than 200 nucleotides that are not directly translated into proteins following transcription [7]. LncRNAs appear to play major roles in various cellular and biological functions of carcinogenesis and metastasis. Probably resulting from their critical roles in promoting or suppressing tumor growth, metastasis, differentiation, and different phrases of cell cycles [8, 9]. Interesting, lncRNAs are also considered competing endogenous RNAs (ceRNAs) to regulate RNA transcription by competing for the binding of microRNAs (miRNAs), which have attracted the attention of researchers [10]. Several studies have demonstrated the involvement of aberrant lncRNA levels in glioma progression via miRNA sponging, such as Linc01094, PCBP1-AS1 and MNX1-AS1 [11,12,13]. These molecular players provide novel insights into GBM and offer clues for developing anti-GBM therapies. However, the biological values and potential mechanisms of many lncRNAs in GBM have not been fully explored; finding and identifying new relevant lncRNAs and their potential mechanisms is crucial to improving the clinical outcome of GBM patients.

This study aims to identify the potential lncRNA signatures as well as their potential mechanisms for the oncogenesis of GBM. Initially, 174 GBM patients’ RNA-seq data acquired from the TCGA database were used in this work (containing 5 normal brain samples and 169 GBM samples). UNC5B-AS1 was expressed at lower levels in GBM samples and cells than in normal groups based on our bioinformatic and experimental results. We also discovered that miR-24-3p, miR-654-5p and miR-580-3p can specifically target UNC5B-AS1 and involved in the prognosis of LGG. Those miRNAs and their target genes might regulate GBM tumorigenesis by regulating pathways in cancers, mitogen activated protein kinase (MAPK) signaling pathway and cell cycle progression. Moreover, the UNC5B-AS1-miR-24-3p regulatory interaction was verified in vitro. Finally, through mechanistic exploration, we found that lncRNA UNC5B-AS1 underlies the oncogenic function by interacting with miR-24-3p. The procedure for this study is displayed in A. Figure 1 in the Additional file 1.

Materials and methods

GBM dataset preparation

In this work, GBM sample RNA-seq data were collected from the publicly available TCGA database (https://www.cancer.gov/tcga), which contains 169 GBM samples and 5 normal brain samples.

Identification of DELs in GBM

GENCODE database accesses users to annotate protein-coding and noncoding regions of human and mouse reference genomes, which draws on primary data as well as bioinformatic tools and analysis to enable the development of transcript structures and the identification of their functions [14]. We retained only the nonprotein-coding genes from the transcript sequences that were annotated in GENCODE. Then, we selected the lncRNA expression profiles across 174 GBM samples with a mean number of RPKM (reads per kilobase of exon model per million mapped reads) ≥ 0.1 [15]. Finally, lncRNAs that were expressed in at least 50% of the GBM samples were chosen for additional study. A bioconductor package called "edgeR" was used to examine differentially expressed lncRNAs between GBM and normal samples [16]. The thresholds were fixed at FDR < 0.001 and |log2FC|> 2. A volcano map and a heatmap plot were used to describe the results. All data was analyzed by R software.

Identification of potential downstream miRNAs of UNC5B-AS1

The lncRNASNP2 database offers thorough resources on single nucleotide polymorphisms (SNPs) across both human and mouse lncRNAs and provides miRNA and lncRNA interaction outcomes integrated by the Pita, TargetScan and miRanda databases [17]. We used the lncRNASNP2 web tool to discover potential downstream miRNAs of UNC5B-AS1. Cystoscope, an open-source bioinformatics software platform, is far more effective for working with large networks and provides more flexibility in terms of networking, importing, and visualizing [18, 19]. By using Cytoscape software (version 3.7.0), a coexpression network of lncRNA-miRNAs was constructed.

Three miRNAs’ targets enrichment analysis in KEGG pathways

The starBase database (https://starbase.sysu.edu.cn/index.php) is a free, accessible online tool that contains multidimensional sequencing data from 32 different kinds of cancer and enables investigation of RNA‒RNA interactions [20]. In this study, we examined the prognostic value of those miRNAs based on the miRNA's median value by using the starBase web tool. The GSEA web tool enable users to detect hallmark gene, chemical and genetic perturbations, KEGG and other gene sets analysis of mRNAs flexibility [21, 22]. We conducted enrichment analysis to explore the potential functions and mechanisms of the miRNAs’ targets based on GESA databases and visualized by the package ‘ggplot2’ built in R, we also conducted the KEGG pathway associated with GBM by Cytoscape software.

Cell culture

All cells were purchased from BeNa Culture Collection (BNCC, China). The U87MG (BNCC; 338,150; China) and HEB cell lines (BNCC; 338,123; China) were cultured in 10-cm petri dishes containing penicillin‒streptomycin solution (Gibco; 2,257,223; Thermo Fisher Scientific) and 10% foetal bovine serum (FBS; 10,099–141; Gibco; Thermo Fisher Scientific) in Dulbecco's Modified Eagle Medium (DMEM; 8,121,356; Gibco; Thermo Fisher Scientific), which was maintained at 37 °C with 5% CO2, and the cells were then passaged every 2–3 days until they reached the logarithmic growth phase [23, 24].

Quantitative real-time PCR (qRT-PCR)

We identified significant lncRNAs from those DELs for qRT‒PCR analysis. Total RNA was extracted from cells using a TRIzol kit (1 ml; QIAGEN; DP424; China) in line with the manufacturer’s guideline. Complementary DNA (cDNA) was synthesized by reverse transcription reagent containing gDNA Eraser (Cat# RR047A, TaKaRa) [25]. TB Green Premix (Cat# RR820A, TaKaRa) was used for qRT‒PCR, and the LightCycler® 96 Instrument (Roche, Switzerland) was used for analysis [26]. Based on the 2−ΔΔCt method, we normalized the results and conducted our data analysis. The internal criteria for lncRNAs and miRNAs were β-actin and U6, respectively.

Cell transfection

In this study, after 1 × 106/well U87MG cells were plated onto 6-well plates, UNC5B-AS1 overexpression Lentivirus (sh-UNC5B-AS1) and negative control (sh-NC) were transfected into the U87MG cell line at an MOI = 5 for 72 h. Furthermore, the cells were incubated for 48 h in medium containing 2 µg/mL puromycin and then cultured in fresh DMEM. The Lentivirus vector GV367 was designed by Shanghai GeneChem Company (Shanghai, China). MiR-24-3p mimics and NC mimics were acquired from RiboBio Company (Guangzhou, China), and with the use of Lipofectamine 3000 (Invitrogen, L3000150, USA), cell transfection was accomplished [27]. Twenty-four hours before transfection, U87MG cells were seeded in 24-well plates until the density reached 50%, and the transfection concentration was 50 nM based on the manufacturer's guidelines. Then, the cells in each group were harvested for subsequent experiments.

Cell proliferation assay

To assess cell proliferative capacity, the CCK-8 test was used. After transfection, U87MG cells were plated into 96‐well plates with 5 × 103 cells per well and incubated for 24 h or 48 h in an incubator at 37 °C with 5% CO2 [4]. Subsequently, the cells received 10 µL of CCK-8 solution (Beyotime, C0039, China) and were then incubated for 2 h [28]. Finally, using a microplate reader, the optical density was calculated at a wavelength of 450 nm.

Dual-luciferase assay

To identify probable binding sites between UNC5B-AS1 and miR-24-3p, the RNAhybrid (https://bibiserv.cebitec.uni-bielefeld.de/rnahybrid) web tool was utilized [29]. The artificially synthesized reporter plasmids were provided by GenePharma Company (GenePharma, Suzhou, China). A total of 5 × 104 U87MG cells per well were plated into 96-well plates. UNC5B-AS1 WNT and MUT plasmids were separately transfected with miR-24-3p mimics or negative control by utilizing Lipofectamine 3000, and a dual luciferase assay solution (Promega, E1910, USA) was used to determine the luciferase activity after 48 h: 10ul LARII regent were added to detect the fluorescence value of firefly luciferase and 10ul stop&Go solution was added to test the Renilla fluorescence. The detected relative luciferase activity was normalized to the Renilla luciferase activity.

Statistical analysis

GraphPad Prism 7.0 software was utilized to perform statistical analysis. The data are displayed as the mean ± S.D. To determine the statistical significance of a difference between two divisions, Student’s t test was used. We considered p values < 0.05 to be significant.

Results

Exploration of significant DELs and downstream miRNAs of UNC5B-AS1

A total of 622 differentially expressed lncRNAs were discovered (169 GBM tumour samples compared with 5 normal samples) with the thresholds | log 2 FC |> 2 and FDR < 0.001, including 252 upregulated lncRNAs and 370 downregulated lncRNAs (eTable 1 and eTable 2 in the Supplementary 1). The volcano map in Fig. 1A and heatmap in Fig. 1B display the distribution of the top 10 significant DELs and UNC5B-AS1.

Fig. 1
figure 1

DELs identifying process and descriptions. A DEL volcano plot. LncRNAs that are upregulated are indicated by red dots, while lncRNAs that are downregulated are indicated by green dots. B Heatmap of DELs and UNC5B-AS1. Samples are represented on the horizontal axis, and DELs are represented on the vertical axis. C UNC5B-AS1 was expressed at lower levels in GBM. D A ceRNA regulation network of UNC5B-AS1 and downstream miRNAs. The pink and yellow circles indicate UNC5B-AS1 and miRNAs, respectively. Unpaired Student’s t test was utilized in A, ****p < 0.0001

Full size image

In addition, qRT‒PCR was applied to estimate the expression of DELs. Considering that downregulated lncRNAs might serve as therapeutic targets or prognostic biomarkers in tumorigenesis, we focused on the identification of downregulated lncRNAs that orchestrate GBM progression and their phenotypic targets. Among them, lncRNA UNC5B-AS1, which was one of the prominently downregulated lncRNAs in GBM, was selected for further evaluation. As illustrated in Fig. 1C, the mRNA expression of UNC5B-AS1 in the U87MG cell line was downregulated compared to that in the HEB cell line, which is consistent with our bioinformatics analysis results above and had never been reported in GBM; therefore, we focused on UNC5B-AS1 for subsequent analysis.

We further explored the potential miRNAs that bind to UNC5B-AS1 using the lncRNASNP2 web tool. A total of 47 miRNAs were identified and visualized by Cytoscape software (Fig. 1D). The starBase web tool was utilized to explore the prognosis of the downstream miRNAs of UNC5B-AS1. A total of 523 LGG patients were divided into two groups, including a high group (N = 261 patients) and a low group (N = 262 patients), based on the median miRNA expression value. Furthermore, the overall survival of these miRNAs was analyzed by plotting KM survival curves. Our results indicated that high expression of miR-24-3p, miR-654-5p and miR-580-3p was significantly associated with LGG patients’ prognosis, respectively (Fig. 2A, B and C). P < 0.05 was considered significant.

Fig. 2
figure 2

Survival analysis of downstream miRNAs of UNC5B-AS1 in LGG. A Overall survival of miR-24-3p. B Overall survival of miR-654-5p. C Overall survival of miR-580-3p. The horizontal axis represents time (months), and the vertical axis represents percent survival. The green colour represents the high expression group (N = 261), and the brown colour represents the low expression group (N = 262). P < 0.05

Full size image

KEGG enrichment analysis of the miRNAs’ targets

KEGG enrichment analysis was carried out to determine the probable roles of miR-24-3p, miR-654-5p and miR-580-3p by the starBase database, respectively. Our findings showed that the three miRNAs' target genes were significantly correlated with the Pathways in Cancer, Regulation of Actin Cytoskeleton, Endocytosis and Glioma, suggesting potential roles for its coding proteins in various diseases (Fig. 3A-C and Tables 1, 2 and 3) and we visualized the regulatory network based on Cytoscape software (Fig. 3D). The results demonstrated that the three miRNAs might play pivotal role in administrating glioma tumorigenesis.

Fig. 3
figure 3

KEGG enrichment analysis of miRNAs’ targets. A Top 40 KEGG enrichment of the miR-24-3p target genes. B Top 40 KEGG enrichment of the miR-654-5p target genes. C Top 40 KEGG enrichment of the miR-580-3p target genes. The horizontal axis indicates GeneRatio (%), and the vertical axis indicates enrichment pathways. Count indicates gene numbers of each pathway. P < 0.05. D KEGG enrichment analysis of miRNAs’ target gene sets. The three yellow circles represent three miRNAs (miR-24-3p, miR-654-5p and miR-580-3p), respectively. The orange circles represent common enrichment pathways between miR-24-3p and miR-654-5p; The grey circles represent common enrichment pathways between miR-654-5p and miR-580-3p; The purple circles represent common enrichment pathways between miR-24-3p and miR-580-3p; The green, pink and blue circles represent independent enrichment pathways of miR-24-3p, miR-654-5p and miR-580-3p, respectively. The red circle represents glioma related pathways

Full size image
Table 1 The enrichment analysis of hsa-miR-24-3p targets in KEGG pathways
Full size table
Table 2 The enrichment analysis of hsa-miR-654-5p targets in KEGG pathways
Full size table
Table 3 The enrichment analysis of hsa-miR-580-3p targets in KEGG pathways
Full size table

To verify our bioinformatic analysis, we conducted in vitro analysis to explore the molecular biological roles of UNC5B-AS1 in GBM cells. The CCK-8 test was used to assess UNC5B-AS1's capacity for cell proliferation. Our results indicated that in contrast to LV-Vector group, the expression of UNC5B-AS1 in the LV-UNC5B-AS1 group was especially higher (Fig. 4A). The results demonstrated that cell proliferation in the LV-UNC5B-AS1 group was inhibited in contrast to that in the LV-Vector group (Fig. 4B). Overall, based on these findings, the function of UNC5B-AS1 could regulate the proliferative ability of GBM cells by sponging miR-24-3p.

Fig. 4
figure 4

UNC5B-AS1 enhanced the proliferative ability of GBM cells. A Efficiency of UNC5B-AS1 overexpression in U87MG cells. B UNC5B-AS1 inhibited cell proliferation. Time is represented on the horizontal axis, and cell viability is represented on the vertical axis. The black and grey pillars indicate LV-Vector and LV-UNC5B-AS1, respectively. Unpaired Student’s t test was utilized in a, and two-way ANOVA with Sidak’s multiple comparisons test was utilized in b. C Overexpression of miR-24-3p attenuated the effects of UNC5B-AS1 on GBM cell proliferation. Multiple t test was utilized in B, Two-way ANOVA with Tukey’s multiple comparisons test was utilized in C, **p < 0.01, ***p < 0.001 and ****p < 0.0001

Full size image

UNC5B-AS1 regulate cell proliferation through the ceRNA sponging of miR-24-3p

To detect the potential binding sites of UNC5B-AS1, qRT‒PCR analysis was utilized, and the results indicated that increased expression of UNC5B-AS1 repressed the expression of miR-24-3p (Fig. 5A). Figure 5B shows that miR-24-3p expression was significantly higher in U87MG cells than in HEB cells. Moreover, miR-24-3p overexpression inhibited the expression of UNC5B-AS1 in U87MG cells (Fig. 5C). Figure 5D demonstrates the miR-24-3p mimic transfection efficiency.

Fig. 5
figure 5

MiR-24-3p is sponged by UNC5B-AS1 in GBM. A UNC5B-AS1 overexpression inhibited miR-24-3p expression in the U87MG cell line. B The expression of miR-24-3p was significantly high in GBM. C Enhanced expression of miR-24-3p suppressed UNC5B-AS1 expression in the U87MG cell line. D Efficiency of the miR-24-3p mimics. Unpaired Student’s t test was utilized in A-D, *p < 0.05, **p < 0.01 and ****p < 0.0001. E Possible binding locations between UNC5B-AS1 and miR-24-3p. F The relative luciferase activity after transfection of both the miR-24-3p mimic and the 3’UTR of UNC5B-AS1 mRNA into U87MG cells

Full size image

Additionally, the probable binding regions between UNC5B-AS1 and miR-24-3p were estimated using RNAhybrid (Fig. 5E). The dual luciferase reporter assay revealed that, in comparison to wild-type UNC5B-AS1 mimics of the NC group, wild-type UNC5B-AS1 transfected with miR-24-3p mimics had suppressed luciferase activity, whereas the luciferase activity of the mutant-type UNC5B-AS1 mimic NC group and mutant-type UNC5B-AS1 miR-24-3p mimics were similar (Fig. 5F). These findings predicted that UNC5B-AS1 may target miR-24-3p, thereby affect the biological process of GBM.

We further evaluated the association of UNC5B-AS1 and miR-24-3p in U87MG cells by CCK-8 experiments. The findings revealed that UNC5B-AS1 overexpression suppressed GBM cell proliferation, whereas cotransfection of miR-24-3p mimics and LV-UNC5B-AS1 effectively diminished the inhibitory effects of UNC5B-AS1 overexpression on U87MG cells (Fig. 4C).

Conclusion

Long noncoding RNAs (lncRNAs), a recent innovation in molecular biology, are crucial for controlling gene expression at the stages of the epigenome, transcription and posttranscription [30]. LncRNAs, which appear to act as miRNA sponges, can negatively attenuate protein-coding gene expression and then regulate cell differentiation and progression in various diseases, including GBM and other cancers. Numerous lncRNAs and their sponged targets have been identified as both diagnostic and therapeutic targets in the development of GBM tumours, such as LINC02015, H19, KIAA0495, AC068888.1 and MALAT1 [31,32,33,34]. In spite of the ceRNAs theory has made considerable progress in the development of diagnostic markers, little is known about the role of lncRNAs and its combined targets in GBM development. Moreover, the lack of sensitive and specific GBM biomarkers has resulted in overdiagnosis and overtreatment [35]. Hence, for the effective diagnosis and treatment of GBM, there is an urgent need to discover GBM-related lncRNAs and uncover their potential mechanisms.

UNC5B-AS1 is a novel carcinogenic lncRNA that has been found in thyroid papillary carcinoma, ovarian cancer, colon cancer and prostate cancer. For example, UNC5B-AS1 was considered to exert a crucial role in cell growth and metastasis of papillary thyroid cancer (PTC) and may be a therapeutic target for PTC [36]. In addition, UNC5B-AS1 accelerates the proliferation, migration, and epithelial-mesenchymal transition (EMT) of hepatocellular carcinoma (HCC) cells via sponging miR-4306 [37]. However, its exact biological roles and direct targets in the progression of GBM are still unknown. In this study, we aimed to discover the roles of UNC5B-AS1 in glioma and investigate its potential mechanisms in the development of GBM. Additionally, many therapeutic targets of UNC5B-AS1 are needed to be explored for GBM treatment.

MiRNAs act as post-transcriptional regulators of gene expression, with 22 ~ 25 nucleotides in length. MiRNAs can be significantly overexpressed in many cancer cells and exert biological roles via interacting with target mRNAs. For example, miR-22-3p can target Wnt-signaling pathway-associated mRNAs and promote the carcinogenesis of colon cancer cells [38]. In glioma, miR-24-3p can specially target the 3’ UTR of MAX interactor 1 (Maxi 1) to promote the cell proliferation [39]. Based on the background above, in this study, we performed bioinformatic analysis and survival analysis to screen potential miRNAs of UNC5B-AS1 and explore its biological functions in vitro. We assumed that miR-24-3p might act as a target of UNC5B-AS1 and be involved in regulating tumor development, which will provide a deeper understanding of GBM development.

To validate our assumptions, we aimed to determine abnormally expressed lncRNAs and miRNAs, as well as their biological regulatory roles in GBM cells. In this study, we firstly discovered the roles of UNC5B-AS1 was downregulated in GBM cells, which is consistent with the bioinformatic analysis. In addition, we found that overexpressed UNC5B-AS1 inhibited cell proliferation, indicates that UNC5B-AS1 might act as tumor suppressor in GBM. We also found that miR-24-3p can sponge UNC5B-AS1 to reverses the effect of cell proliferation, demonstrates the potential biological mechanisms of UNC5B-AS1 in GBM. Moreover, miR-24-3p was significantly associated with LGG patients’ prognosis, indicates that UNC5B-AS1 may interact with miR-24-3p to influence patents’ survival. However, more experiments are needed to further test these hypotheses. Based on our results, a novel UNC5B-AS1/miR-24-3p regulation axis was constructed, which may act a key pathway in the GBM development and become a new therapeutic application of GBM in the foreseeable future.

The current work also has a few shortcomings. First, in the TCGA dataset, we solely examined and verified the biological roles of lncRNA biomarkers, and no other database of lncRNA expression data associated with GBM was used for further validation. Second, the interaction between UNC5B-AS1 and miR-24-3p were only explored by the qRT-PCR analysis and luciferase assay. More experiments, including RNA pull down and immunoprecipitation (RIP). However, we will process more intensive research on anther experiments in the future. Finally, this study used a database mining design without validation in fresh samples due to our limited conditions, such as patients’ tissues and animal experimental studies. We hope to perfect our work in the future.

In conclusion, our study suggested the potential functions of the UNC5B-AS1-miR-24-3p regulatory link, which may serve as cutting-edge potential biomarkers for GBM diagnosis and prognosis.