hsa_circ_0072389, hsa_circ_0072386, hsa_circ_0008621, hsa_circ_0072387, and hsa_circ_0072391 aggravate glioma via miR-338-5p/IKBIP

Glioma is a primary intracranial tumor with high morbidity and mortality. We acquired miR-338-5p, which suppresses the development of glioma, from the GEO and CGGA databases. In addition, we predicted that hsa_circ_0072389, hsa_circ_0072386, hsa_circ_0008621, hsa_circ_0072387, and hsa_circ_0072391 could relieve the silencing of IKBIP by miR-338-5p. By analyzing genes related to IKBIP expression, possible pathways affecting glioma were identified. This study provides new ideas for investigating multiple circRNAs in ceRNAs.


INTRODUCTION
Glioma is the most common primary malignant brain tumor and accounts for approximately half of all intracranial primary tumors [1]. The current effective treatment methods are surgery, radiotherapy, and chemotherapy, the survival of patients posttreatment is still suboptimal [2]. Fortunately, the development of molecular biotechnology provides new ideas for the diagnosis and treatment of glioma. Circular RNA is a type of noncoding RNA lacking both a 5′ end cap and a 3′ end polytail that induces a circular RNA structure via covalent bonds [3]. Due to their closed circular structure, circRNAs are seldom degraded by exonuclease, which makes them more stable than linear RNAs [4]. An increasing number of studies show that circRNAs participate in the occurrence and development of a variety of cancers [5]. By regulating gene transcription, interacting with proteins, or translating into polypeptides, circRNAs can participate in the regulation of cancer [6,7]. However, current research mainly regards circRNA as a competing endogenous RNA [8]. As miRNAs can not only silence downstream genes by targeting mRNAs but also bind sites located in circRNAs (miRNA response elements, MREs), circRNAs can prevent the suppression of target genes by competitively binding to miRNAs [9]. For example, Cen J et al. discovered that circular RNA circSDHC sponges adsorb miR-127-3p, which upregulates the expression of CDKN3 and E2F1 and promotes the proliferation and metastasis of renal cell carcinoma [10]. Chen LY et al. revealed that the circular RNA circ-ERBIN promoted colorectal cancer by sponging miR-125a-5p and miR-138-5p, thereby alleviating the silencing of 4EBP-1 [11]. According to multiple studies, we found that a miRNA could be sponged by multiple circRNAs in different cancers. For example, miR-942-5p could be adsorbed by hsa_circ_0015756 in ovarian cancer, circRNA-AKT1 in cervical cancer, and circ-CEP85L in gastric cancer [12][13][14]. However, AGING studies of the mechanism by which multiple circRNAs regulate miRNAs in the same cancer have not been performed.

Public database collection
High-throughput data on circRNAs, miRNAs and mRNAs in glioma were obtained from the GEO database. The screening standard for circRNA was |logFC| ≥ 1.5 p <0.05 and |logFC| ≥ 2 p < 0.05 for miRNA and mRNA. The whole screening process was implemented with the limma R package [15].

Prediction of proteins bind to circRNA
The circinteractome database (https://circinteractome.nia.nih.gov/) was employed to reveal proteins that interact with circRNA. In addition, the TargetScan prediction tool enables the prediction and binding sites for RBPs on reported circRNAs that have yet to be mapped [18].

KEGG and GO Enrichment Analysis
KEGG was used to analyze the pathways involved in mRNAs, and GO was used to analyze molecular function (MF), biological process (BP), and cellular component. All operations were implemented by using the Clusterprofiler R package [19].

circRNA secondary structure and MFE structure
The sequences of circRNAs were obtained from circbase (http://circrna.org/) and further entered into the RNAfold database (http://rna.tbi.univie.ac.at/) to acquire the secondary structures and MFE structures of circRNAs.

Analysis of survival rate
The overall survival time and gene expression of glioma patients were obtained from the GEPIA database (http://gepia.cancer-pku.cn/). Kaplan-Meier plots were used to analyze the relationship between the survival time and gene expression of glioma patients. The hazard ratio [21] and 95% confidence intervals were acquired by the statistical software SPSS 19.0.

The correlation of gene expression
Pathway-related genes were obtained from the Pathcards database (https://pathcards.genecards.org/), and the correlation between IKBIP and gene expression in glioma was obtained through the GEPIA database.

Cell culture
The glioma cell Line U251 was cultured in MEM containing 10% fetal bovine serum and 1% glutamine, AGING and the cell culture environment was 5% carbon dioxide in a 37°C incubator.

RNA extraction, RT-qPCR
RNA was extracted by TRIzol reagent, and RT-qPCR was performed by a 7500 real-time PCR system (Thermo Fisher Scientific). The miR-338-5p inhibitor was obtained from Boshang Biotechnology (Shanghai). Taking GAPDH as a reference, the primer sequences were as follows:

Western blot
A BCA Protein Detection Kit (Thermo Scientific ™ , Shanghai, China) was used to detect the protein concentration. SDS (5×) was added to the total protein, and the mixture was further heated at 100°C for 10 minutes. The protein was isolated by SDS-PAGE and transferred to polyvinylidene fluoride (PVDF) film. Five percent skim milk was sealed at room temperature for 2 hours and further incubated with primary antibody in a shaking bottle at 4°C for 8-12 hours. Then, we washed the film with Tris-buffered saline and Tween 20 (TBST) 3 times, and each wash lasted 10 minutes. After that, the secondary antibody was incubated with film at room temperature for 1 hour and washed with TBST once more 3 times (10 min each time). Proteins were observed by enhanced chemiluminescence.

Statistical analyses
Student's t-test was used to determine statistically significant differences. When P < 0.05, the difference was statistically significant. The statistical analysis software used in this study was SPSS 19.0.

Data availability statement
The datasets used in the project are available from the corresponding author. The data that support the findings of this study are openly available in GEO at https://www.ncbi.nlm.nih.gov/geo/.

miR-338-5p inhibited the progression of glioma
A number of studies have shown that miRNAs can promote or prevent the occurrence and development of glioma. In this study, we selected GEO datasets (GSE139031, GSE25632, GSE103228, and GSE65626) with large sample sizes for analysis. When comparing gene expression in tumor tissues with gene expression in normal tissues, there were 3556 differential genes in GSE139031, 1145 differential genes in GSE25632, 6658 differential genes in GSE103228, and 3556 differential genes in GSE65626 ( Figure 1A-1D). After screening (|logFC| ≥ 2, p < 0.05), we found 1426 differential genes in GSE139031 (1312 upregulated genes and 114 downregulated genes), 32 differential genes in GSE25632 (21 upregulated genes and 11 downregulated genes), 26 differential genes in GSE103228 (4 upregulated genes and 22 downregulated genes), and 30 differential genes in GSE65626 (20 upregulated genes and 10 downregulated genes) ( Figure  1E-1H). To identify miRNAs stably expressed in glioma, we intersected these differential genes in four datasets and found that miR-338-5p was the only intersection of the four datasets ( Figure 1I). By analyzing the CGGA database data, we found that patients with glioma in the miR-338-5p high expression group had a longer survival time, and the expression of miR-338-5p in patients with WHO2 was significantly higher than those with WHO3-4 ( Figure 1J-1K); therefore, we speculate that miR-338-5p can inhibit the progression of glioma.
hsa_circ_0072389, hsa_circ_0072386, hsa_circ_0008621, hsa_circ_0072387, and hsa_circ_0072391 bind to miR-338-5p in glioma circRNA can target miRNA to prevent the degradation of mRNA induced by miRNA [22]. To explore circRNAs binding to miR-338-5p, 4872 circRNAs were screened out by using circBANK. In addition, we revealed that 472 circRNAs in GSE146463 (|logFC| ≥ 1.5, p < 0.05) were differentially expressed between glioma tissues and normal tissues, which included 351 upregulated circRNAs and 121 downregulated circRNAs (Figure 2A and 2B). We found that hsa_circ_0072389, hsa_circ_0072386, hsa_circ_0008621, hsa_circ_0072387, hsa_circ_0072391, hsa_circ_0044234, and hsa_circ_0001685 were not only differentially expressed in GSE146463 but also had separate binding sites for miR-338-5p ( Figure 2C). In addition, binding AGING sites for miR-338-5p located in these 7 circRNAs were analyzed in the CircInteractome database (Supplementary Figure 1A-1G). AGO2 is the core component of RISC [21], which connects target sites of miRNA and mRNA [23]. The formation of a ternary complex of circRNA-miRNA-AGO2 can also prove that circRNA interacts with miRNA [24]. We predicted the proteins that may bind to hsa_circ_0072389, hsa_circ_0072386, hsa_circ_0008621, hsa_circ_0072387, hsa_circ_0072391, hsa_circ_0044234, and hsa_circ_0001685 by searching the CircInteractome database and obtained the number Figure 1. (A-D) Cluster analysis showed the differential genes in GSE139031, GSE25632, GSE103228, and GSE65626 in turn .red dots indicated upregulated differential genes and blue dots indicated downregulated differential genes. (E-H) volcano map showed the differential genes in GSE139031, GSE25632, GSE103228, and GSE65626, Red dots represent differential genes with logFC ≥ 2, p < 0.05, blue dots represent differential genes with logFC ≤ 2, p < 0.05. (I)The Venn diagram shows the intersection of differential genes between GSE139031, GSE25632, GSE103228, and GSE65626. (J) Survival analysis demonstrated the effect of miR-338-5p on glioma. (K) The box diagram showed that the expression level of miR-338-5p in WHO II was significantly higher than that in WHO III and WHO IV.

Composition, secondary structure and sequence information of circRNAs
Most circRNAs are composed of exons, followed by introns and exons, and few are composed of introns [25]. We obtained the composition and sequence of   Figure 1H-1L, Supplementary  Figure 2). In addition, the secondary structure and minimum free energy (MFE) structure of circRNAs were acquired from the RNAfold database (Supplementary Figure 3A-3E). Interestingly, we found that hsa_circ_0072389, hsa_circ_0072386, hsa_circ_0008621, hsa_circ_0072387, and hsa_circ_0072391 were all transcribed by the same host gene HMGCS1 on chromosome 5, and they all contained EXON4 and EXON5 (Table 1, Supplementary Figure  1H-1L). In addition, 5 circRNAs were composed of single-stranded structures, stems, hairpin loops, convex loops, inner loops, and multibranched loops, which provide directions for further exploration of the genes regulating transcription, interaction with proteins, or translation into polypeptides.

IKBIP binds to miR-338-5p and aggravates glioma
We further predicted mRNAs that might bind miR-338-5p. In GSE4290, we found 1319 differentially expressed genes (|logFC| ≥ 2, p < 0.05), which involved 451 upregulated genes and 868 downregulated genes ( Figure 3A and 3B). In addition, we obtained 879, 1328, and 4812 mRNAs that may bind to miR-338-5p from the miRwalk, miRDB, and TargetScan databases, respectively. By comparing differential mRNAs in the GSE4290, miRwalk, miRDB, and TargetScan databases, we found that 100 differential mRNAs may not only be correlated with glioma but also bind to miR-338-5p ( Figure 3C). Further analysis of these 100 genes by KEGG and GO revealed that these 100 genes tend to regulate signal transduction in pathways, biological processes, cell components, and molecular functions (Supplementary Figure 4A-4D). According to the results described above, we plotted the circRNA-miRNA-mRNA network (Supplementary Figure 5). In addition, using the STRING database, we analyzed the interaction relationships among proteins originating from 100 mRNAs (Supplementary Figure 6). Furthermore, to screen glioma-related mRNAs, we analyzed the effects of these 100 mRNAs on the prognosis of glioma patients in the GEPIA database. We found that patients with low expression of CCBE1, IKBIP, NRG1, and RGS4 had longer survival times ( Figure 3D-3G). In addition, we separately acquired the binding sites of CCBE1, IKBIP, NRG1, and RGS4 with miR-338-5p from the TargetScan database (Supplementary Figure 4E-4H). In the GEPIA database, we found that IKBIP was upregulated in glioma relative to normal tissue, while CCBE1, NRG1, and RGS4 were downregulated in glioma ( Figure 3H). Since we found that patients with low expression of CCBE1, IKBIP, NRG1, and RGS4 had a better prognosis than patients with high expression, there was a contradiction between the expression of CCBE1, NRG1, and RGS4 and their effect on prognosis. Therefore, we hypothesized that IKBIP was regulated by miR-338-5p in glioma. In summary, we constructed the hsa_circ_0072389, hsa_circ_0072386, hsa_circ_0008621, hsa_circ_0072387, hsa_circ_0072391, miR-338-5p, and IKBIP networks ( Figure 3I). hsa_circ_0072389, hsa_circ_0072386, hsa_circ_0008621, hsa_circ_0072387, and hsa_circ_0072391 promote the expression of IKBIA, while miR-338-5p has the opposite effect To verify the relationship among hsa_circ_0072389, hsa_circ_0072386, hsa_circ_0008621, hsa_circ_0072387, and hsa_circ_0072391 with IKBIP, AGING we knocked down hsa_circ_0072389, hsa_circ_0072386, hsa_circ_0008621, hsa_circ_0072387, and hsa_circ_0072391 in U251 glioma cells separately ( Figure 4A-4E). Then, Western blotting was used to determine the expression of IKBIP in U251 glioma cells and found that the expression of IKBIP protein was downregulated relative to that in the normal control group ( Figure 4G). In addition, we inhibited the expression of miR-338-5p with a miR-338-5p inhibitor ( Figure 4F) and revealed that the expression of IKBIP was upregulated relative to that in the normal control group.
Diagram of circRNA-miRNA-mRNA pathways hsa_circ_0072386, hsa_circ_0008621, hsa_circ_0072387, and hsa_circ_0072391 can bind to miR-338-5p, and miR-338-5p can target IKBIP, which is coexpressed with genes in the NF-κB signaling pathway and JAK/STAT signaling pathway. Therefore, we speculate that after being transcribed from the host gene HMGCS1, pre-RNAs are cleaved into mRNAs and circRNAs through splicing and modification. In these circRNAs, hsa_circ_0072389, hsa_circ_0072386, hsa_circ_0008621, hsa_circ_0072387, and hsa_circ_0072391 are transported between or within cells, which bind to miR-338-5p and hinder the silencing of IKBIP mRNA. Then, the upregulation of IKBIP results in the activation of the NF-κB signaling pathway, JAK/STAT signaling pathway and TGFβ/SMAD signaling pathway, which worsen the deterioration of glioma ( Figure 6).

DISCUSSION
In GSE139031, GSE25632, GSE103228, and GSE65626, we found that miR-338-5p was expressed at low levels in glioma. In the CGGA database, we further verified that glioma patients with high miR-338-5p expression had a better prognosis. Meanwhile, a great number of studies have also shown that miR-338-5p is able to inhibit the proliferation and invasion of glioma cells [28][29][30]. Therefore, miR-338-5p was selected as a candidate gene for subsequent study. By searching for the presence of conserved 8-mer, 7-mer, and 6-mer sites that match the seed region of miR-338-5p, we predicted circRNAs and mRNA to bind with miR-338-5p [16]. After analyzing the GSE146463, circBANK and circInteractome databases, we found that hsa_circ_0072389, hsa_circ_0072386, hsa_circ_0008621, hsa_circ_0072387, and hsa_circ_0072391 in glioma were most likely to target miR-338-5p.
In most studies of ceRNAs in tumors, a single circRNA or mRNA could be screened from circRNA chips or mRNA chips. Then, the targeted miRNA could be predicted easily [31]. However, it was not determined whether a circRNA could only bind to a miRNA invariably [22]. Therefore, this study selected one miRNA, miR-338-5p, as the initial research object and analyzed relationships among circRNAs, miRNAs and mRNAs by exploring the circRNAs to which miR-338-5p may bind. In addition, all genes in 8 common pathways were screened to predict pathways that might be affected by IKBIP. We found that circRNAs (hsa_circ_0072389, hsa_circ_0072386, hsa_circ_0008621, hsa_circ_0072387, and hsa_circ_007239) that bind to miR-338-5p in glioma were all transcribed from the same host gene HMGCS1. In addition, these circRNAs all contained exons 4 and 5. Through the GEPIA database, we found that the change in HMGCS1 mRNA expression in glioma was not statistically significant when compared with the control group. Therefore, we speculated that pre-RNA splicing of hsa_circ_0072389, hsa_circ_0072386, hsa_circ_0008621, hsa_circ_0072387, and hsa_circ_007239 increased with the progression of glioma. As a large-scale research study based on bioinformatics analysis, this work inevitably lacks sufficient experimental verification, so we need to confirm its authenticity in future research. In addition, due to the lack of glioma data in the GEO and TCGA databases, follow-up data of patients are insufficient. Furthermore, although we revealed that hsa_circ_0072389, hsa_circ_0072386, hsa_circ_0008621, hsa_circ_0072387, and hsa_circ_007239 all originate from the same host gene HMGCS1, limited expression data of circRNAs in glioma prevents us from further investigating differences among these five circRNAs. We suspect that the five circRNAs may be differentially expressed not only in diverse grades or prognostic stages of glioma but also in exosomes, cytoplasm or blood, which play significant roles in different times or spaces. In summary, our study provides a novel idea for the study of the complex network among circRNAs-miRNAs-mRNAs, and we also hope to verify the interaction mechanisms among multiple circRNAs, multiple miRNAs, and multiple mRNAs in our future research.

AUTHOR CONTRIBUTIONS
JL, XL collected and analyzed data, drew plots by R language and accomplished the manuscript. BZ performed the tests. JL drafted the manuscript. All authors read and approved the final manuscript.

ACKNOWLEDGMENTS
This work is supported by grants from Science and Technology Program of Shenzhen AGING (GJHZ20190821161601670). Thanks Jinan University and Shenzhen People's Hospital for their education and support. Thanks all members for their efforts in this research.