Background
Glioma is the most common and lethal primary brain tumor in adults. Many treatment approaches, including surgery and radiochemotherapy are not ideal, and the average survival time of patients is less than 15 months [
1,
2]. Glioma is comprised of heterogeneous cell populations, including a subpopulation of glioma stem cells (GSCs) showing tumor initiation, self-renewal, and multi-lineage differentiation abilities [
3]. GSCs have been shown to be responsible for glioma proliferation, therapeutic resistance, and recurrence [
4]. There is therefore a great need to identify the molecular mechanisms responsible for GSCs proliferation and progression, as well as to identify novel molecular targets for treatment of glioma.
ISL2 is a LIM/homeodomain-type transcription factor of the Islet-1 family, and is mainly expressed in the primary sensory and motor neurons [
5]. It has been reported that
ISL2 is essential for acquisition of motor neuron identity, and it contributes to the restriction of motor neurons within the neural tube via slit and semaphorin signaling [
6,
7], while
ISL2 inhibition impairs peripheral axonal outgrowth in embryonic zebrafish [
5]. In addition,
ISL2 participates in the formation of topographic maps in the visual system [
8,
9].
ISL1 is a member of the Islet-1 family and shares 72% protein sequence identity with
ISL2 [
6].
ISL1 participates in the development and functional regulation of sympathetic neurons, motor neurons, and retinal ganglion cells [
10‐
12].
ISL1 also acts as an oncogene in breast cancer, gastric cancer, and neuroendocrine carcinoma [
13‐
15]. However, there has been no study of the possible effects of
ISL2 on cancers, including glioma. As a transcription factor involved in development of the nervous system, it is doubtful whether
ISL2 affects the development and progress of glioma.
Circular RNAs (circRNAs) have emerged as a new class of noncoding RNAs that form single-stranded closed loop structures by forming covalent bonds without the 5′ caps and 3′ poly(A) tails [
16]. The circular structure of circRNAs facilitates their stable existence in different tissues, and their ability to play vital roles in multiple biological functions [
17]. Moreover, studies have shown that circRNAs are dysregulated in cancers and can either promote or inhibit the proliferation, metastasis, apoptosis, and angiogenesis of cancers [
18]. Mechanistically, circRNAs can either mediate transcription, interact with RNA-binding proteins, or function as competitive endogenous RNAs (ceRNAs) to regulate the expression of genes involved in tumorigenesis and progression [
19‐
21]. Although several circRNAs have been reported in glioma, few GSCs-related circRNAs and their functions and molecular mechanisms have been clearly elucidated.
In the present study, we first identified
ISL2 as a novel oncogene in glioma, which was overexpressed and mainly involved in glioma angiogenesis via
VEGFA-mediated ERK signaling, according to both bioinformatics analyses and molecular experiments. Moreover, we found a novel and overexpressed circRNA, cARF1 (circBase ID: hsa_circ_0016767), in GSCs, which regulated
ISL2 via sponging miR-342–3p. The cARF1 was back-spliced from the
ARF1 gene located at chr1: 228082660–228,099,212, and finally formed a sense-overlapping circular transcript of 1597 nucleotides with three exons from the
ARF1 mRNA transcript 1. MiR-342–3p is reported to exert tumor inhibiting effects in several cancers [
22]. Finally, as a transcription factor,
ISL2 directly transcribed the expression of
U2AF2, which is a type of RNA binding protein (RBP), contains a sequence-specific RNA-binding region for splicing, and promotes the stability of cARF1. Our study therefore identified a
U2AF2/cARF1/miR-342–3p/
ISL2 feedback loop in GSCs, which promoted glioma angiogenesis, and which may provide novel targets for glioma therapy.
Methods
Patient samples and ethical approval
Seventy clinical samples from glioma patients were collected from January 2007 to December 2012 at the First Affiliated Hospital of China Medical University. There were 20 samples of grade II, 25 samples of grade III, and 25 samples of grade IV glioma. During the same period, 10 more acute brain injury patient samples were collected as a control group. Clinical information for these samples is outlined in Table
S1. This study was approved by the Ethics Committee of the First Affiliated Hospital of China Medical University, and written informed consent was obtained from each patient.
Cell culture and GSCs isolation
Human brain microvessel endothelial cells (hBMECs) were purchased from ScienCell Research Laboratories (San Diego, CA, USA). The hBMECs were maintained in endothelial cell medium (ECM; ScienCell Research Laboratories). Six patient-derived primary glioma stem cells from WHO grade II to IV (grade II: GSC205 and GSC207; grade III: GSC306 and GSC307; grade IV: GSC406 and GSC408) were isolated, and neurosphere cultures were obtained as previously described [
23]. The detailed clinicopathological information is presented in Table
S2. Briefly, freshly resected glioma samples were dissociated into single cells and grown in serum-free DMEM/F12 with 2% B27, 20 ng/mL rh-bFGF, and rh-EGF (Gibco, Gaithersburg, MD, USA). The stem cell markers of GSCs were detected by immunofluorescence using anti-CD133 (Abcam Technology, Cambridge, UK) and nestin antibodies (Abcam). The immunofluorescence staining of glial fibrillary acidic protein (GFAP; Abcam) and β-III tubulin (Abcam) was used to evaluate the multi-lineage differentiation capacity of GSCs.
Preparation of the glioma conditioned medium (GCM)
The preparation of GCM has been previously described [
24]. Briefly, we used serum-free DMEM/F12 to wash GSCs three times, followed by culturing the GSCs for 24 h. The medium was then collected and centrifuged at 3000×
g for 15 min at 4 °C to remove GSCs and debris. The GCM was prepared and used immediately for the treatment of hBMECs, followed by subsequent experiments, or stored at − 80 °C for no more than 1 week.
Lentiviral vector construction and transfection
The lentivirus-based vectors for
ISL2 overexpression,
U2AF2 overexpression, cARF1 overexpression, RNAi-mediated knockdown of
ISL2,
U2AF2 and cARF1, and their negative controls were all constructed by Gene-Chem (GV358, Shanghai, China). The detailed sequence of the lentivirus-based vectors can be obtained on the GeneChem website (
http://www.genechem.com.cn/index/supports/zaiti_info.html?id=50). The miR-342–3p mimic, inhibitor, and their negative controls were obtained from Thermo Fisher Scientific (Assay ID: MH12328 and MC12328; Thermo Fisher Scientific, Waltham, MA, USA). The sequences of all siRNAs are listed in Table
S3. The lentivirus transfection and efficacy measurements were performed as previously described [
23].
qRT-PCR (real-time quantitative reverse transcription PCR)
Real-time PCR was performed as previously described [
23]. The Mini-BEST Universal RNA Extraction kit (TaKaRa, Kyoto, Japan) was used to extract the total RNA of GSCs. For circRNA and mRNA, the RNA was reverse transcribed into cDNA using a Prime Script RT Master Mix reagent kit (TaKaRa). The qPCR assays were detected using the SYBR Green Master Mix (TaKaRa) with PCR LightCycler480 (Roche Diagnostics, Basel, Switzerland). Furthermore, RNase R (Epicentre Technologies, Madison, WI, USA) was used to confirm the existence of cARF1 and eliminate the effect of linear ARF1 RNA. The β-actin was used as an endogenous control. For miRNA, cDNA was synthesized using the PrimeScript™ RT reagent kit (TaKaRa, Shiga, Japan). The expression levels of miR-342–3p were detected using the TaqMan Universal Master Mix II (Assay ID: 002260; Applied Biosystems, Foster City, CA, USA). The U6 housekeeping gene was used as an endogenous control (Assay ID: 001973, Applied Biosystems). Primers used in this study are listed in Table
S4.
Western blotting
Western blotting was performed as previously described [
23]. Briefly, the total proteins of GSCs or tissues were isolated using a total cell protein extraction kit (KeyGen Biotechnology, Nanjing, China). Protein lysates were prepared, subjected to SDS-PAGE, transferred onto polyvinylidene difluoride membranes and blocked with 2% bovine serum albumin (KeyGen Biotechnology). The primary antibodies against ISL2 (1:1000; Abcam), VEGFA (1:1000; Abcam), VEGFR2 (1:1000; Abcam), p-VEGFR2 (1:1000; Abcam), MEK1/2 (1:1000; Abcam), p-MEK1/2 (1:1000; Abcam), ERK1/2 (1:500; Abcam), p-ERK1/2 (1:500; Abcam), and β-actin (1:2000; Proteintech, Rosemont, IL, USA) were incubated at 4 °C overnight. After secondary antibody (Proteintech) incubation, the bands were detected using a chemiluminescence ECL kit (Beyotime Biotechnology, Beijing, China) and quantified by ImageJ software (National Institutes of Health, Bethesda, MD, USA).
Immunohistochemistry (IHC)
IHC was performed and the results were semi-quantified as previously described [
23]. Briefly, the paraffin-embedded tissue sections were labeled with primary antibody against ISL2 (1:100; Abcam), U2AF2 (1:100; Abcam), VEGFA (1:100; Abcam), and CD31 (1:100; Abcam). The sections were then treated with an immunohistochemical labeling kit (MaxVision Biotechnology, Fuzhou, China) and photographed with a light microscope (Olympus, Tokyo, Japan). The German immunohistochemical score was used to evaluate the staining intensity and expression levels [
25].
Immunofluorescence
Immunofluorescence staining was performed as previously described [
24]. Briefly, the GSCs were fixed with 4% paraformaldehyde, permeabilized with 0.5% Triton X-100, blocked with 5% bovine serum albumin, and detected with primary antibodies against CD133, nestin, GFAP, and β-IIItubulin (1:100; Abcam) at 4 °C overnight. The samples were stained by fluorescein isothiocyanate- or rhodamine-conjugated secondary antibodies. Finally, GSCs were counterstained using 4′,6-diamidino-2-phenylindole (Sigma-Aldrich, St. Louis, MO, USA) and were visualized using a laser scanning confocal microscope (Olympus).
Cell viability assay
The hBMECs were plated in 96-well plates at a density of 1000 cells/well and incubated in GCM for 0, 24, 48, 72, 96, and 120 h. Cell viability was determined using the CellTiter 96® Aqueous Non-Radioactive Cell Proliferation Assay Kit (Promega, Madison, WI, USA) according to the manufacturer’s instructions.
EDU assay
The EDU assay was conducted to examine the proliferation of cells using an EDU assay kit (Beyotime, Biotechnology) according to the manufacturer’s protocol. Briefly, the hBMECs were treated with GCM and seeded into 24-well plates at 1 × 105 cells/well for 24 h, then 10 μM EDU reagent was added to the medium and incubated for 2 h. After being fixed and permeabilized, the hBMECs were counterstained. The percentage of EDU positive cells was calculated using a laser scanning confocal microscope (Olympus).
Transwell invasion assay
For the Transwell invasion assay, approximately 1 × 105 hBMECs under different conditions were plated in the upper chamber (Corning, Corning, NY, USA) with a Matrigel filter (BD Biosciences, San Jose, CA, USA) and ECM medium with 10% fetal bovine serum was added to the lower chamber. After incubation for 24 h, the invaded cells were fixed with 4% paraformaldehyde and stained with Crystal Violet (Beyotime, Biotechnology). The stained cells were photographed and counted using a light microscope (Olympus).
The tube formation assay was performed as previously described [
24]. Briefly, pre-chilled 96-well plates were coated with 70 μL Matrigel filter reagent (BD Biosciences) per well at 37 °C for 30 min. The hBMECs under different conditions were seeded on the surface of the Matrigel at 2 × 10
4 cells/well at 37 °C for 4 h. A microscope (Olympus) was used to visualize the images for each well, and Image J software was used to calculate the total number of branches and tubule lengths.
Enzyme-linked immunosorbent assay (ELISA)
The ELISA was performed using a commercial kit (Cusabio, Stratech, UK) to detect the concentration of VEGFA in the supernatant of the GSCs medium, as previously described [
25]. All results were normalized to the protein concentration in the control group.
Luciferase reporter assay
Luciferase reporter assays were performed as previously described [
24]. Briefly, the luciferase reporter plasmids (
VEGFA-wt and
VEGFA-mt,
ISL2–3′-UTR-wt and
ISL2–3′-UTR-mt, cARF1-wt and cARF1-mt, and
U2AF2-wt and
U2AF2-mt) were constructed by Gene-Chem (GV102). The detailed sequence can be obtained on the GeneChem website (
http://www.genechem.com.cn/index/supports/zaiti_info.html?id=). The luciferase reporter plasmids were co-transfected into GSCs. After 48 h, the luciferase activities were detected using a Dual-Luciferase Reporter Assay System (Promega). Relative luciferase activity was calculated as the ratio of firefly luciferase activity to Renilla luciferase activity.
Chromatin immunoprecipitation (ChIP) assays
ChIP assays were performed using the ChIP Assay Kit (Beyotime Biotechnology) according to the manufacturer’s instructions. The chromatin complexes were immunoprecipitated using anti-ISL2 antibody or normal rabbit IgG, and the purified DNA samples were analyzed by qPCR. The primers for ChIP qPCR are listed in Table
S4.
RNA immunoprecipitation (RIP) assay
The RIP assay was performed using the EZ-magna RIP RNA-binding Protein Immunoprecipitation kit (Millipore, Darmstadt, Germany) according to the manufacturer’s protocols. GSCs under different conditions were lysed in RIP buffer including magnetic beads conjugated with negative control IgG, anti-AgO2, or anti-U2AF2 antibodies (Millipore). After incubation with proteinase K, the immunoprecipitated RNAs were isolated. Finally, qRT-PCR was used to examine the precipitants.
RNA pull-down assay
The Pierce Magnetic RNA Protein pull-down Kit (Thermo Fisher Scientific) was used to detect the interaction between cARF1 and U2AF2 according to the manufacturer’s suggestions. Briefly, biotinylated RNA probes were used to label purified RNA, and then the positive control (input), negative control (antisense RNA), and biotinylated RNA were mixed and co-incubated with GSCs proteins at room temperature. The RNA-protein complex was added with magnetic beads to prepare a probe-magnetic bead complex. After being washed and boiled, the complexes were detected by western blotting, using β-actin as a control.
RNA stability measurement
GSCs were cultured in the medium containing 2 μg/ml actinomycin D (Act D, NobleRyder, China) to block the de novo RNA synthesis. Then total RNA was collected at indicated times and cARF1 expression was detected by qRT-PCR. The half-life of cARF1 was determined as the time required to reach 50% of the RNA levels before actinomycin D treatment.
Xenograft experiments
Xenograft experiments were performed as previously described [
24]. Under specific pathogenic conditions, 6-week-old female BALB/c nude mice (Beijing Vital River Laboratory Animal Technology, Beijing, China) were raised at the Laboratory Animal Center of China Medical University. GSCs under different conditions were injected (5 × 10
4 cells per mouse) orthotopically into the mouse brains, 2 mm lateral and 2 mm anterior to the bregma using a stereotaxic instrument (
n = 5, per group). The tumor volume was measured according to the following formula: V = (D × d
2)/2, where D was the longest diameter and d was the shortest diameter of the tumor. All animal experiments were performed in accordance with the Animal Care Committee of China Medical University.
The data of mRNA expression, WHO grades, isocitrate dehydrogenase (
IDH) status (
IDH 1/2) of
ISL2 and
U2AF2, the survival times, and status of glioma patients were obtained from the Chinese Glioma Genome Atlas (CGGA,
http://www.cgga.org.cn) using the mRNA seq-693 dataset and The Cancer Genome Atlas (TCGA,
http://cancergenome.nih.gov) in the HG-U133A platform. Gene set enrichment analysis (GSEA,
http://www.broadinstitute.org/gsea/index.jsp) was used to analyze enrichment of a biological process or signal pathway with high versus low
ISL2 expressions. Four online databases, Starbase (
http://starbase.sysu.edu.cn), TargetScan (www.
targetscan.org), microRNA (
http://www.microrna.org/microrna/home.do), and miRDB (
http://mirdb.org) were used to predict possible miRNAs targeting
ISL2. Starbase and circBase (
http://www.circbase.org/) databases were used to predict potential circRNAs as sponges of miRNA. The Starbase database was also used to predict the proteins binding to circRNAs.
Statistical analysis
Results are reported as the mean ± SD of at least three independent experiments. The chi-square test, two-tailed Student’s t-test, and one-way analysis of variance were used to compare the statistical significance among different groups. Pearson’s correlation analysis was used to assess the correlation between two groups. The survival difference was evaluated using a log-rank test and Kaplan-Meier analysis. SPSS statistical software for Windows, version 23.0 (IBM, Armonk, N. Y, USA) was performed for statistical analysis, and two-tailed P values < 0.05 were considered significant.
Discussion
In this study, we first showed that
ISL2 was overexpressed in glioma and correlated with poor patient survival using bioinformatics analysis and our clinical specimens. As a transcriptional component mainly involved in the development and function of motor and sensory neurons, it is reasonable that
ISL2 may exert possible effects on the central nervous system and glioma [
28]. Our study showed that
ISL2 shared a similar oncogenic role as its family member,
ISL1, in other cancers [
29,
30]. Due to its rapid and infiltrating growth properties, glioma shows active metabolism and uses an abundant blood supply in tumor tissues [
27,
31]. These properties also lead to complete surgical resection and tumor recurrence [
32]. Active angiogenesis is frequently observed in glioma, which can further promote its proliferation and aggressiveness [
33]. Our study revealed that
ISL2 transcriptionally regulated
VEGFA expression and promoted
VEGFA secretion in GSCs, and that
ISL2-mediated GCM promoted the proliferation, invasion, and angiogenesis of hBMECs via ERK signaling. We therefore conclude that the oncogenic effects of
ISL2 in glioma involved promotion of angiogenesis. Moreover, because anti-angiogenic treatment, represented by anti-VEGF therapy, such as bevacizumab, is one of the most important strategies for glioma treatment [
34],
ISL2 may act as a possible therapeutic target.
Accumulating evidence has recently indicated that there are numerous circRNAs expressed in neuronal tissues, and that dysregulation of circRNAs can lead to diseases of the nervous system, including glioma [
35]. The detailed regulatory molecular mechanisms of circRNAs include direct transcription and translation into functional proteins, transcriptional, and splicing regulation as well as miRNAs and RBP sponges [
21,
27,
36]. For example, circ-FBXW7 encodes a novel 21 kDa protein called FBXW7-185aa in glioma, which inhibits proliferation and cell cycle acceleration [
21]. Circular RNA MAPK4 (circ-MAPK4) inhibits glioma cell apoptosis via the MAPK signaling pathway by sponging miR-125a-3p in glioma [
36]. Circ_002136 can bind to a RBP,
FUS, and this regulates angiogenesis via the miR-138-5p/SOX13 axis in glioma [
27]. Among these mechanisms, circRNA-mediated miRNA and RBP sponges are currently the most extensively studied. Our study therefore focused on
ISL2 regulation by circRNAs via miRNA and RBP sponges.
MiR-342–3p was the only candidate miRNA that we predicted could target the 3′-UTR of
ISL2, based on four datasets including microRNA, miRDB, TargetScan, and Starbase. Although there has been no previous study on the regulation between miR-342–3p and
ISL2, miR-342–3p has been reported to play an anti-tumor role in several cancers including glioma. For example, miR-342–3p expression levels have been negatively correlated with advanced WHO grades and inhibit the progression of glioma by directly targeting
PAK4 [
22]. MiR-342–3p can also inhibit the malignant biological behaviors of glioblastoma cells via
Zic4 [
37]. Our study further showed that miR-342–3p exerted anti-glioma effects by inhibiting GSC-GCM-mediated angiogenesis in hBMECs. Moreover, we also showed that miR-342–3p downregulated
ISL2 expression in GSCs and inhibited the angiogenesis mediated by
ISL2.
ADP ribosylation factor 1 (
ARF1) is a GTPase that is involved in vesicle trafficking and the Golgi apparatus [
38,
39]. It was reported that
ARF1 gene promoter methylation is associated with
EGFR gene amplification and can promote the distinct tumor infiltration in glioblastoma [
38].
ARF1 promotes cancer stem cell viability via lipid metabolism, and its ablation induces anti-tumor immune responses in mice [
40]. Our study found a novel circRNA, cARF1 (hsa_circ_0016767), which was back-spliced from transcript one of
ARF1 mRNA and comprised its second, third, and fourth exons. cARF1 was overexpressed in our glioma specimens, was positively correlated with poor patient survival, and also promoted proliferation, invasion, and angiogenesis of hBMECs via
VEGFA signaling. Moreover, as a circRNA, we also showed cARF1 had strong miRNA sponging ability toward miR-342–3p. All these results showed that cARF1 upregulated
ISL2 expression in GSCs via sponging miR-342–3p.
The RBPs are a group of more than 800 proteins, which have been identified and mainly involved in post-transcriptional regulation of RNAs, gene transcription, and translation, and participate in both physiological and pathological processes and diseases [
41]. Studies of circRNA biogenesis have shown that RBP participates in circRNA splicing and expression [
42]. For example, RBP binds to the introns of circRNAs linear genes near splice sites and promotes the production of circRNAs [
43]. RBPs can therefore serve as an essential element underlying the functions of circRNAs, especially circRNA-mediated gene transcriptional regulation [
44]. In our study, we assessed the possible effects of RBPs on the regulation of cARF1 via bioinformatics predictions, which showed that
U2AF2 was an appropriate candidate for experimental molecular validation.
U2AF2 is a spliceosome factor and a non-snRNP protein required for the binding of U2 snRNP to the pre-mRNA branch site [
45]. Our study showed that
U2AF2 binds to and promotes the stability and expression of cARF1 in GSCs, while there was no effect on the expression of its ARF1 linear form.
U2AF2 has been reported to act as an oncogene in several cancers. Bioinformatics analysis suggested that
U2AF2 was upregulated in
IDH-mutated glioma with malignant transformation [
46].
U2AF2 expression was significantly upregulated in primary non-small cell lung cancer and was associated with metastasis, advanced tumor stages, poor survival, and recurrence [
47]. In addition,
U2AF2 is significantly increased in melanoma progression and participates in brain metastasis [
48]. Our study also focused on the relationship between
U2AF2 and glioma, and we showed that
U2AF2 was also a novel oncogene in glioma, because it was expressed at higher levels in glioma correlated with poor patient survival. Furthermore,
U2AF2 can also lead to the proliferation, invasion, and angiogenesis of hBMECs via upregulating cARF1 in GSCs. As a transcription factor, we showed that
ISL2 transcribed the expression of
U2AF2, thus establishing a feedback loop among
U2AF2, cARF1, miR-342–3p, and
ISL2 in GSCs. This feedback loop may not only promote glioma angiogenesis, but may also promote the tumorigenesis, aggressiveness, and malignant transformation, which all need to be investigated in further studies.
Publisher’s Note
Springer Nature remains neutral with regard to jurisdictional claims in published maps and institutional affiliations.