Introduction
Currently, knockout animal models, microRNA antisense inhibitor oligonucleotides (1–3) and miRNA sponges (4) are the three dominant approaches to study the effect of microRNA (miRNA)-induced loss of function in a specific cell type of interest. However, inhibitor oligonucleotides are only effective in short-term (24–72 h) experiments due to their unpredictable degradation. miRNA-knockout animal models are an alternative for functional studies in vivo; however, the time-consuming procedures and costs required to generate genetic knockout animals often preclude the extensive use of the method. The use of an miRNA sponge, which contains multiple target sites that are complementary to an miRNA of interest, is a dominant-negative method. These transcripts can specifically sequester target miRNAs and thereby prevent the miRNAs from binding to their endogenous target mRNAs (4,5). Indeed, the experimental application of miRNA sponge technology has attracted increasing interest for in vitro and in vivo applications, suggesting that this approach can greatly aid in the understanding of miRNA functioning (6–9).
Glioblastoma multiforme (GBM), World Health Organization (WHO) grade IV, is the most common and biologically aggressive malignant glioma. Diffuse infiltration, robust angiogenesis, uncontrolled cellular proliferation, intense resistance to apoptosis and rampant genomic instability are the hallmark features that make this disease incurable (10). miR-23b has been extensively studied in several pathologies and has been found to be upregulated in human GBM regardless of the subtype being studied, compared to adjacent non-neoplastic brain tissues (11,12). However, the expression of miR-23b varies between cancer types, which adds to the complexity of understanding the role of this miRNA in etiopathology (8,11,13–15). It was reported that miR-23b was downregulated in human colon cancer samples and potently mediated multiple steps of metastasis in vivo, including tumor growth, invasion and angiogenesis (8). However, in contrast to colon cancer, GBM showed a relatively high expression level of miR-23b, and our previous study demonstrated that the downregulation of miR-23b suppressed tumor growth by targeting VHL (11). Therefore, we aimed to examine the pivotal role of miR-23b downregulation in tumor angiogenesis, invasion and migration in GBM. Although we previously used chemically modified antisense oligonucleotides to knock down the expression of miR-23b (11), in the present study, we designed an miRNA sponge as a decoy target to effectively and specifically inhibit miR-23b. However, many cells both in vitro and in vivo are resistant to the uptake of oligonucleotides. Thus, to mimic the downregulation of miR-23b in GBM, we employed a lentivirus-mediated miR-23b sponge to execute miR-23b downregulation in glioma cells in Petri dishes and in an orthotopic model.
In the present study, we showed that the stable lentivirus-mediated introduction of an miR-23b sponge both in vitro and in vivo downregulated miR-23b, resulting in reduced glioma angiogenesis, invasion and migration. These findings facilitate a better understanding of the role of miR-23b in the molecular pathogenesis of cancer and suggest that miR-23b might be a candidate for the treatment of GBM patients.
Materials and methods
Cells and cell culture
Human U87, LN229 and U251 GBM cells and human umbilical vein endothelial cells (HUVECs) were obtained from the China Academia Sinica Cell Repository (Shanghai, China). The GBM cells were maintained in Dulbecco’s modified Eagle’s medium (DMEM; Gibco) supplemented with 10% fetal bovine serum (FBS; HyClone) and incubated at 37°C in a 5% CO2 atmosphere. The HUVECs were grown in endothelial growth medium, EGM-2 (Gibco, EBM-2 supplemented with growth factors and other supplements) with 2% FBS.
Lentivirus preparation and in vitro infection
Lentiviral vectors expressing a scrambled control, an miR-23b sponge, or luciferase were all generated by GenePharma (Shanghai, China). Cell infections were conducted according to GenePharma’s recommendations.
RNA extraction and quantitative real-time analysis
Total RNA was extracted using TRIzol (Invitrogen). Real-time quantification for miRNA measurement was performed with the the Hairpin-it™ miRNAs qPCR Quantitation kit (GenePharma) and the DNA Engine Opticon 2 Two-Color Real-Time PCR Detection system (Bio-Rad Laboratories). miR-23b expression was normalized to the expression of U6 small nucleolar RNA (snRU6). Each reaction was performed in triplicate. Quantification was performed using the 2−ΔΔCt method.
Western blot analysis and immunohistochemistry (IHC)
Western blotting and IHC assays were performed as previously described (16). The antibodies used were anti-VHL, anti-VEGF, anti-MMP2, anti-CD31, anti-MMP9 and anti-HIF-1α from Santa Cruz Biotechnology; anti-E-cadherin and anti-ZEB1 from Cell Signaling Technology; and anti-β-catenin from Signalway Antibody. Western blot bands were subjected to densitometric analyses using ImageJ (National Institutes of Health). The mean relative intensities of the relevant protein bands were quantified and normalized to GAPDH levels.
Tubule formation assay
A Matrigel (BD Biosciences) assay was performed to assess endothelial tubule formation in vitro. Ninety-six-well plates were coated with 75 μl of Matrigel/well and incubated at 37°C for 35 min. HUVECs were harvested and suspended in EBM supplemented with an equal volume of the conditioned medium (CM) from glioma cells treated with the scrambled control or the miR-23b sponge. In each well, 1×104 HUVECs were plated and maintained at 37°C. The experiments were performed in triplicate, repeated at least once and analyzed in a double blind fashion by two observers.
In vitro invasion assays
Transwell membranes coated with Matrigel (BD Biosciences) were used to quantify glioma cell invasion in vitro. Transfected cells were plated at 2×104/well in the upper chamber in serum-free medium; medium containing 10% FBS was added to the lower chamber. After an 18-h incubation, the non-invading cells were removed from the top well with a cotton swab and the cells on the bottom were fixed in methanol, stained with 0.1% crystal violet, and photographed in three independent ×10 fields/well. The data represent the mean ± standard error (SE) of three independent experiments.
In vitro scratch assay
At 72 h after infection with a scrambled control-expressing or an miR-23b sponge-expressing lentivirus, an artificial wound was created using a 200-μl pipette tip, and the cells were further incubated. To analyze cell migration into the wound, images were obtained at 0 and 18 h using a digital camera system coupled to a microscope. ImageJ was used to determine the relative migration distance as the reduction of the width of the open area. A statistical analysis of the results of three experiments was performed using the Student’s t-test.
Nude mouse tumor intracranial model
Before implantation, U87 cells that were co-infected with a luciferase-expressing lentivirus and an miR-23b sponge-expressing or a scrambled control-expressing lentivirus were serially diluted in a 24-well plate prior to luminescence imaging. The U87 cells were then injected intracranially into 5-week-old BALB/c-nu mice. After 15 days, the tumors were measured by luminescence imaging of whole mice using an IVIS Lumina Imaging System (Xenogen). Paraffin-embedded sections (8 μm) were stained with hematoxylin and eosin (H&E) and used for IHC analysis. When experimental animals were used, the research followed the internationally recognized guidelines on animal welfare and local and national regulations.
Statistical analysis
Statistical analysis was performed using the SPSS Graduate Pack, version 11.0, statistical software (SPSS). Descriptive statistics, including the mean and the SE and a one-way analysis of variance (ANOVA), were used to determine statistically significant differences. The overall survival curves were plotted according to the Kaplan-Meier method, with the log-rank test being applied for comparison. P<0.05 was considered to indicate a statistically significant result.
Results
Knockdown of miR-23b expression in glioma cells via the lentivirus-mediated miR-23b sponge
In order to knock down miR-23b expression, we constructed an miRNA sponge consisting of a decoy vector expressing tandem repeated miRNA binding sites (Fig. 1A). The binding sites for miR-23b were perfectly complementary in the seed region, with a bulge at positions 9–12 to prevent RNA interference-type cleavage and degradation of the sponge RNA (Fig. 1B). As a control, we constructed another vector with repeated scrambled binding sites. Due to the toxicity and low efficiency of plasmid transfection, we utilized a lentivirus as the gene delivery vector, a system that can deliver an miR-23b sponge with relatively low toxicity and high efficiency (17,18). We engineered two lentiviral systems to express the miR-23b sponge or scrambled control and then infected three glioma cell lines (U87, LN229 and U251). To verify the efficacy of endogenous miR-23b inhibition using the lentivirus-mediated miR-23b sponge, we performed qRT-PCR analysis of the glioma cell lines after infection. The results showed that miR-23b expression in all three cell lines was reduced by ~50% (Fig. 1C).
Downregulation of miR-23b impairs the glioma cell functional capacity for angiogenesis, invasion and migration
To evaluate the functional consequences of expression of the miR-23b sponge in glioma cells, three glioma cell lines (U87, LN229 and U251) were infected with a lentivirus expressing either the miR-23b sponge or scrambled control. We performed western blot analysis to assay the expression of the miRNA target protein (VHL) (19) and proteins associated with tumor angiogenesis [(HIF-1α) (19) and VEGF (19)], invasion [(β-catenin) (20), ZEB1 (21) and E-cadherin (20)] and migration [(MMP2 (22) and MMP9 (22)]. As shown in Fig. 1D, the inhibition of miR-23b elevated the expression of tumor-suppressive proteins (VHL and E-cadherin) and abrogated the expression of oncogenic proteins (HIF-1α, VEGF, MMP2, MMP9, β-catenin and ZEB1), which is consistent with the qRT-PCR results previously reported by our laboratory (11).
Tubule formation, Transwell and scratch assays were then performed to further examine the effects of miR-23b suppression on glioma cell phenotypes of angiogenesis, invasion and migration. The culture supernatant from miR-23b sponge-treated glioma cells (U87, LN229 and U251) was notably less potent than the supernatant from the scrambled-control lentivirus-infected glioma cells in inducing tubule formation by HUVECs, further supporting the decreased angiogenic protein production revealed by our western blot analysis (Fig. 2A). The Transwell assay showed a significant reduction in invasive cells, which indicated an inhibitory effect of reduced miR-23b expression on the invasive ability of glioma cells in vitro (Fig. 2B). To examine the cell migration in vitro, the scratch assay was employed, and decreased mobility was observed for all of the sponge-treated cell lines (Fig. 2C). Collectively, these results further support that miR-23b downregulation has a significantly suppressive impact on tumor angiogenic, invasive and migratory phenotypes in vitro.
Sponge-mediated miR-23b knockdown impedes U87 tumor growth in vivo
Given the advantages mentioned above, lentiviral vectors are superior tools for gene introduction. To investigate the in vivo antitumor effects of miR-23b knockdown, we co-infected U87 cells with a lentivirus-expressing luciferase and the scrambled control or miR-23b sponge. As miR-23b downregulation was found to inhibit angiogenesis, invasion, and migration in vitro, we predicted the inhibitio of cell growth in an orthotopic graft after treatment with the miR-23b sponge. Prior to implantation, we conducted a serial-dilution validation using U87 cells infected with the luciferase-expressing lentivirus (Fig. 3A). An excellent linear correlation was observed between the number of live cells and luciferase activity. Fig. 3B shows a visualization of a U87 tumor expressing GFP in situ. In vivo imaging analysis of the mice every 15 days for 45 days revealed the suppressed growth rate of the miR-23b sponge-treated U87 cells (Fig. 3C and D). Furthermore, treatment with the miR-23b sponge was associated with a significantly longer survival rate of the mice (Fig. 3E).
miR-23b sponge-treated U87 cells have reduced angiogenic, invasive and migratory potential in vivo
Consistent with our data from the in vitro assays, an IHC analysis showed decreased expression of HIF-1α, β-catenin, MMP2, MMP9, VEGF and ZEB1 and increased expression of VHL and E-cadherin after treatment with the mR-23b sponge (Fig. 4A). To further investigate angiogenesis within the U87 orthotopic graft, we stained tumor sections with an anti-CD31 antibody, which demonstrated a reduction in CD31-positive vessels after miR-23b sponge treatment, further supporting the anti-angiogenic effect of miR-23b inhibition in vivo (Fig. 4B). H&E staining of the miR-23b-downregulated tumors revealed less peripheral invasion, which may have resulted from reduced invasive and migratory abilities (Fig. 4C). In summary, our results demonstrate anti-angiogenic, anti-invasive and anti-migratory implications of the introduction of the miR-23b sponge.
Discussion
As there is increasing evidence that miRNAs participate in carcinogenesis, the identification of physiologically relevant and/or therapeutically promising miRNAs has become an essential task (6,23,24). Our previous study showed that miR-23b is overexpressed in glioma samples and cell lines (11). However, miR-23b has been found to be downregulated in human colon cancer samples and to potently mediate the multiple steps of metastasis in vivo, including tumor growth, invasion and angiogenesis (8). However, there has been no systematic report to date on the effects of miR-23b on GBM pathological processes (11,25). Therefore, we investigated the role of miR-23b in glioma angiogenesis, invasion and migration, characteristics that make GBM incurable.
As miR-23b is overexpressed in gliomas, we first knocked down miR-23b expression in glioma cell lines. There are three general methods for miRNA loss-of-function studies: genetic knockout, antisense oligonucleotide inhibitors (1–3) and sponges (4). Although these oligonucleotide-based methods have been successfully applied, the techniques do elicit off-target side-effects and unwanted toxicity. In contrast, specific miRNA sponges are superior for substantially inhibiting target miRNAs, monitoring the transfection efficacy and specifically selecting a particular subpopulation of transfected cells (4,5). We, therefore, replaced antisense oligonucleotides with an miR-23b sponge, which was engineered in a lentiviral vector to allow more effective gene delivery. In the present study, the lentiviral vector encoding the miR-23b sponge was found to sufficiently silence miR-23b activity, resulting in the detection of broad changes in the angiogenesis, invasion and migration of glioma cell lines (U87, L229 and U251). After replacing antisense nucleotide with the miR-23b sponge, we performed western blot analysis to detect alterations in the expression of proteins associated with angiogenesis, invasion and migration. In our previous study (11), we demonstrated that miR-23b directly targeted VHL and consequently inhibited the β-catenin/Tcf-4 and HIF-1α/VEGF signaling pathways (11). HIF-1α and VEGF are well-established pro-angiogenic proteins among cell lines and are negatively regulated by VHL (26). ZEB1 and E-cadherin are known for their implications in epithelial-mesenchymal transition (EMT), which stabilizes the invasive mesenchymal phenotype of epithelial tumor cells, and β-catenin is involved by interacting closely with E-cadherin (27). Matrix metalloproteinases (MMPs) are a family of enzymes responsible for the proteolytic processing of extracellular matrix structural proteins, which regulate endothelial cell migration (22). Therefore, the reduced expression of HIF-1α, VEGF, ZEB1, β-catenin and MMP2/9 and the elevation of VHL and E-cadherin levels indicated that miR-23b inhibition exerted an anticancer effect on all of the glioma cell lines tested. Consistent with the western blotting results, the suppression of angiogenic, invasive, and migratory capabilities was confirmed in glioma cell lines by tubule formation, Transwell and scratch assays, respectively. We speculated that similar tumor inhibition would be observed in vivo, and experiments using an orthotopic glioma mouse model affirmed this assumption. Importantly, the miR-23b sponge-treated mice demonstrated a more favorable outcome when compared to the mice treated with the scrambled control. We proposed that the impairment of angiogenesis, invasion, and migration that decreased the growth of the orthotopic graft was responsible for this difference in survival. Consistent with the in vitro results, histological staining demonstrated a reduction in vessels and less peripheral invasion in the miR-23b sponge-treated orthotopic grafts.
Based on these observations, miR-23b-based anticancer therapy (such as an miR-23b sponge) is promising and can be developed, either alone or in combination, with current targeted therapies, with the goal of improving disease responses and increasing cure rates. This therapy may be particularly essential for GBM, which has a uniformly poor prognosis due to diffuse infiltration and robust angiogenesis. However, tissue-specific delivery, potential off-target effects, and biological safety must be addressed before miRNA-based therapeutics can be adopted.