miRNA-337-3p suppresses neuroblastoma progression by repressing the transcription of matrix metalloproteinase 14

Recent evidence shows the emerging roles of endogenous microRNAs (miRNAs) in repressing gene transcription. However, the miRNAs inhibiting the transcription of matrix metalloproteinase 14 (MMP-14), a membrane-anchored MMP crucial for the tumorigenesis and aggressiveness, still remain largely unknown. In this study, through mining computational algorithm program and genome-wide Argonaute profiling dataset, we identified one binding site of miRNA-337-3p (miR-337-3p) within the MMP-14 promoter. We demonstrated that miR-337-3p was under-expressed and inversely correlated with MMP-14 expression in clinical specimens and cell lines of neuroblastoma (NB), the most common extracranial solid tumor in childhood. Patients with high miR-337-3p expression had greater survival probability. miR-337-3p suppressed the promoter activity, nascent transcription, and expression of MMP-14, resulting in decreased levels of vascular endothelial growth factor, in cultured NB cell lines. Mechanistically, miR-337-3p recognized its binding site and recruited Argonaute 2 to facilitate the enrichment of repressive epigenetic markers and decrease the binding of RNA polymerase II and specificity protein 1 on the MMP-14 promoter. Gain- and loss-of-function studies demonstrated that miR-337-3p suppressed the growth, invasion, metastasis, and angiogenesis of NB cells in vitro and in vivo. In addition, restoration of MMP-14 expression rescued the NB cells from changes in these biological features. Taken together, these data indicate that miR-337-3p directly binds the MMP-14 promoter to repress its transcription, thus suppressing the progression of NB.


INTRODUCTION
Neuroblastoma (NB), the most common extracranial solid tumor derived from neural crest, accounts for approximately 15% of all pediatric cancer deaths [1]. NB is characterized by heterogeneous clinical behaviors, ranging from spontaneous regression to rapid progression or resistance to multimodal treatment, such as surgery, chemoradiotherapy, stem cell transplantation, and immunotherapy [1]. Better understanding the mechanisms underlying the progression and aggressiveness of NB is needed for improving the therapeutic efficiency. Matrix metalloproteinase 14 , also known as membrane type-1 matrix metalloproteinase, is able to degrade various extracellular matrix (ECM) components and facilitate the tumor cells to remodel and penetrate the ECM [2]. Previous studies have indicated that MMP-14 promotes the tumor invasion by functioning as a pericellular collagenase and an activator of proMMP-2, and is directly linked to tumorigenesis and aggressiveness [3]. In addition, MMP-14 promotes tumor angiogenesis through facilitating the transcription of vascular endothelial growth factor (VEGF) via activating the Src-tyrosine kinase pathway and increasing the cell surface localization of vascular endothelial growth factor receptor 2 [4][5][6]. Our previous studies have shown that MMP-14 is highly expressed in NB tissues and cell lines, and is correlated the aggressiveness and poor outcome of NB patients [7]. However, the regulatory mechanisms of MMP-14 expression in NB still remain largely unknown.
MicroRNAs (miRNAs), a class of small noncoding RNAs with 22 to 25 nucleotides in length, are able to inhibit gene expression at the post-transcriptional or translational levels through forming base pairs with their targets, usually in the 3′-untranslated region (3′-UTR) [8]. It has been established that miRNAs can participate in the tumorigenesis and aggressiveness [8]. For example, miR-203 inhibits the proliferation, migration, and invasion of glioma cells by disrupting the roundabout homolog 1/ mitogen-activated protein kinase 1/matrix metalloproteinase 9 signaling axis [9]. miR-29b suppresses the growth, invasion, and metastasis of prostate cancer cells through repressing anti-apoptotic and pro-metastatic matrix molecules [10]. Previous studies have identified many aberrantly expressed miRNAs, which contribute to almost all aspects of tumor biology of NB, such as proliferation, apoptosis, differentiation, invasion, metastasis, and angiogenesis [11], suggesting the important roles of miRNAs in the progression and aggressiveness of NB.
In recent years, emerging studies show that endogenous miRNAs participate in the heterochromatin formation and regulation of gene transcription in human cells [12][13][14][15]. In the current study, through mining computational algorithm program and genome-wide Argonaute profiling dataset, we identified one binding site of miRNA-337-3p (miR-337-3p) within the MMP-14 promoter, implicating its potential roles in the transcriptional control of MMP-14 in NB. We demonstrate, for the first time, that miR-337-3p is under-expressed and anticorrelated with MMP-14 expression in clinical NB specimens, and directly binds the MMP-14 promoter to suppress its transcription via inducing chromatin remodeling, thus inhibiting the growth, invasion, metastasis, and angiogenesis of NB cells in vitro and in vivo, suggesting the tumor suppressive roles of miR-337-3p in the progression of NB.

miR-337-3p inhibits the MMP-14 expression through transcriptional repression
To explore the effects of miR-337-3p on MMP-14 expression in NB cell lines, we performed the miRNA overexpression experiments. Stable transfection of miR-337-3p precursor into SH-SY5Y and SK-N-BE(2) cells increased the cytoplasmic and nuclear miR-337-3p levels ( Figure 2A). Western blot, real-time quantitative RT-PCR, and nuclear runon assays demonstrated that stable over-expression of miR-337-3p decreased the protein and nascent transcript levels of MMP-14 in NB cells, than those stably transfected with empty vector (mock) ( Figure 2B, Figure 2C, and Figure 2D). The expression levels of VEGF, the MMP-14 downstream target gene in NB [7], were significantly decreased in miR-337-3p over-expressing NB cells, consistent with the MMP-14 reduction ( Figure 2B and Figure 2C). Since the analysis of microPIR database revealed no potential binding site of miR-337-3p within the VEGF promoter, we ruled out the possibility that miR-337-3p may directly suppress the transcription of VEGF. To further examine the suppressive roles of miR-337-3p in MMP-14 expression, we performed the miRNA knockdown experiments by transfection of anti-miR-337-3p or negative control (anti-NC) inhibitors into SH-SY5Y and SK-N-SH cells. Transfection of anti-miR-337-3p inhibitor obviously decreased the cytoplasmic and nuclear expression of miR-337-3p ( Figure 2E), and increased the MMP-14 and VEGF protein levels than those of anti-NC-transfected cells ( Figure 2F). Real-time quantitative RT-PCR and nuclear run-on assays indicated the enhanced transcriptional levels of MMP-14 and VEGF in NB cells transfected with anti-miR-337-3p inhibitor, than those transfected with anti-NC ( Figure 2G and Figure 2H). In addition, over-expression or knockdown of miR-337-3p in NB cells did not change the 3′-UTR luciferase activity of MMP-14 and VEGF (Supplemenatry Figure S3A and Supplemenatry Figure S3B

miR-337-3p recognizes the binding site and recruits AGO2 on MMP-14 promoter
To determine whether or not miR-337-3p could repress the MMP-14 expression by recognizing its binding site, the MMP-14 promoter-luciferase reporter and a mutant of miRNA recognition site ( Figure 3A) were transfected into NB cells stably transfected with empty vector (mock) or miR-337-3p precursor. The firefly luciferase activity normalized to that of Renilla was significantly reduced in the tumor cells stably transfected with miR-337-3p precursor ( Figure 3B), and the effects were abolished by mutation of miR-337-3p binding site within the MMP-14 promoter ( Figure 3B). In addition, transfection of anti-miR-337-3p inhibitor increased the luciferase activity in SH-SY5Y and SK-N-SH cells ( Figure 3C), while mutation of miR-337-3p recognition site abolished these effects ( Figure 3C). Since previous studies have revealed the potential involvement of AGO1 and AGO2 in miRNA-induced transcriptional repression [12,14,18], the MMP-14 promoter regions were assayed for enrichment of AGO1 and AGO2 by chromatin immunoprecipitation (ChIP) and real-time quantitative PCR (qPCR). In cultured SH-SY5Y and SK-N-BE(2) cells, enrichment of AGO2, but not of AGO1, was observed around the binding site (-122/+69) of miR-337-3p ( Figure 3D). As controls, no MMP-14 promoter regions were immunoprecipitated with unspecific antibody (isotype IgG) or detected by qPCR with primer set (-326/-157) distal to the binding site of miR-337-3p ( Figure 3D). In addition, over-expression or knockdown of miR-337-3p increased and decreased the enrichment of AGO2 on MMP-14 promoter, respectively ( Figure 3D). These results indicated that miR-337-3p recognized the binding site and recruited AGO2 on the MMP-14 promoter in NB cells.

miR-337-3p suppresses the growth, invasion, metastasis, and angiogenesis of NB cells through repressing MMP-14 in vitro
Since previous studies indicate that MMP-14 promotes the growth, migration, invasion, and angiogenesis of tumor cells [7], we further investigated the effects of miR-337-3p over-expression and MMP-14 restoration on cultured NB cells. Western blot indicated that transfection of MMP-14 rescued the miR-337-3p-repressed MMP-14 expression ( Figure 5A). In MTT colorimetric assay, tumor cells stably transfected with miR-337-3p precursor possessed the decreased cell viability capability, when compared with those stably transfected with empty vector (mock; Supplemenatry Figure S6A). In scratch migration assay, miR-337-3p over-expression attenuated the migration capability of SH-SY5Y and SK-N-BE(2) cells ( Figure 5B and Supplemenatry Figure S6B). Matrigel invasion assay revealed that NB cells stably transfected with miR-337-3p precursor presented an impaired invasion capacity than mock cells ( Figure 5C). The tube formation of endothelial cells was suppressed by treatment with the medium preconditioned by stable transfection of NB cells with miR-337-3p precursor ( Figure 5D). In addition, transfection of MMP-14 into SH-SY5Y and SK-N-BE(2) cells restored the decrease in growth, migration, invasion, and angiogenesis induced by stable over-expression of miR-337-3p ( Figure  5B, Figure 5C, Figure 5D

Over-expression of miR-337-3p attenuates the growth, metastasis, and angiogenesis of NB cells in vivo
We next investigated the efficacy of miR-337-3p against tumor growth and metastasis in vivo. Stable transfection of miR-337-3p precursor into SH-SY5Y cells resulted in decreased growth and tumor weight of subcutaneous xenograft tumors in athymic nude mice, when compared to those stably transfected with empty vector (mock) ( Figure 6A and Figure 6B). In addition, the miR-337-3p levels within tumors were increased, and the intratumoral expression of MMP-14 and VEGF was also reduced by stable transfection of miR-337-3p precursor ( Figure 6C). Moreover, stable transfection of miR-337-3p precursor resulted in decrease in CD31-positive microvessels and mean vessel density within tumors ( Figure 6C). In the experimental metastasis studies, SH-SY5Y cells stably transfected with miR-337-3p precursor established statistically fewer lung metastatic colonies than mock group ( Figure 6D). These results suggested that miR-337-3p could inhibit the growth, metastasis, and angiogenesis of NB cells in vivo.

DISCUSSION
Canonically, miRNAs inhibit gene expression at the post-transcriptional levels, and this process is executed through recognition of miRNA-mRNA pairing by RNAinduced silencing complex machinery, with AGO proteins as the main effectors [11]. Recent evidence shows that a subset of miRNAs predominantly localizing in the nucleus of human cells can recognize complementary genomic sites within gene promoters, and participate in the regulation of gene expression [12,14]. At promoter complementary sites, miRNAs can function as an interface between chromatin remodeling complexes and genome [20,21]. For example, endogenous miR-709 is able to repress the transcription of early growth response 2 through forming epigenetic silencing complexes with H3K27me3 and AGO1 on its promoter [18]. miR-320 www.impactjournals.com/oncotarget  induces the transcriptional repression of RNA polymerase III polypeptide D (POLR3D) through directing the association of AGO1, EZH2, and H3K27me3 with POLR3D promoter [14]. Meanwhile, miR-423-5p-induced transcriptional repression of progesterone receptor (PR) is associated with the recruitment of AGO2 to an antisense non-coding RNA transcribed from PR promoter, and accompanied by an increase in H3K9me2 enrichment at PR promoter [12]. However, studies on the roles of miRNAs in regulating gene transcription are still in its infancy, and it is currently urgent to investigate the expression and functions of these miRNAs in the tumor biology of NB.
miR-337-3p is a novel identified miRNA involved in the tumorigenesis. Previous studies show that miR-337-3p is under-expressed in endometrial cancer [22], gastric cancer [23], colon cancer [24], and medulloblastomas [25]. Loss of miR-337-3p expression is associated with the lymph node metastasis of gastric cancer [23]. miR-337-3p promotes the senescence of colorectal cancer cells by repressing the expression of casein kinase II subunit alpha [26], and sensitizes the lung cancer cell lines to paclitaxel by directly repressing the expression of signal transducer and activator of transcription 3 and Ras-related Protein 1A [27]. Since miR-337-3p host gene resides within the chromosome 14q32.2 [28], an imprinted region essential for development [17], and aberrant regulation of miRNAs located on chromosome 14q32 is involved in tumorigenesis [24,[28][29][30], we suspect that miR-337-3p may also participate in the progression of NB. In this study, we found the under-expression of miR-337-3p in NB tissues, which was associated with unfavorable outcome of patients. Meanwhile, in line with previous studies [31], we noticed the copy number loss of chromosome 14q32.2 in NB tissues, implying that under-expression of miR-337-3p may be due to allelic loss, which warrants our further investigation. Through an integrative approach to analyze the public datasets, we identified miR-337-3p as a novel repressor of MMP-14 transcription. Our evidence showed that miR-337-3p suppressed the expression of MMP-14 through binding its promoter and inducing transcriptional repression in NB cells. The fact that restoration of MMP-14 was sufficient to prevent the NB cells from miR-337-3p-inhibited biological behaviors indicates that miR-337-3p exerts its tumor suppressive functions, at least in part, through repressing the MMP-14 expression in NB. In addition, we noted the different expression profiles of miR-337-3p in publicly available clinical tumor datasets, and found that miR-337-3p attenuated the MMP-14 expression in cervical cancer cells, but not in hepatocellular cancer, prostate cancer, or renal cell carcinoma cells. Our evidence shows that miR-337-3p suppresses the aggressiveness of cervical cancer cells, suggesting the tissue-specific expression patterns and functions of miR-337-3p in human tumors.
Human MMP-14 gene, consisting of 10 exons, is localized at chromosome 14q11 [19]. Previous studies show that Sp1 is essential for maintaining the MMP-14 transcription [19]. Transcriptionally inactive MMP-14 gene promoter is characterized by deposition of H3K27me3 and hypermethylation of CpG islands [32], while low levels of H3K27me3 and under-methylation are observed at the MMP-14 promoter region in invasive and migratory cancer cells [32], suggesting that histone modification and DNA methylation are crucial factors impacting the transcriptional efficiency of MMP-14 promoter. Since our preliminary data revealed that administration of DNA methyltransferase inhibitor resulted in no significant changes in miR-337-3pinduced transcriptional repression of MMP-14 in NB cells, we ruled out the possible involvement of aberrant promoter hypermethylation in this process. In this study, we found that in miR-337-3p over-expressing NB cells, the recruitment of repressive epigenetic markers EZH2, H3K27me3, and H3K9me2 was increased, accompanied by decreased enrichment of RNA Pol II and Sp1 on MMP-14 promoters. These findings indicate that miR-337-3p may induce epigenetic inactivation by deposition of repressive chromatin marks, or through a process that prevents RNA Pol II stalling at gene promoter.
The mechanisms underlying the recruitment of miRNAs to accessible promoter sequences include direct binding to promoter [18] and indirect association with promoter-derived non-coding transcripts [12,13]. Since RNase H treatment (specifically degrades the RNA present in RNA-DNA hybrid) [12] attenuated the miR-337-3pinduced enrichment of AGO2, EZH2, H3K27me3, and H3K9me2, these findings indicated the direct binding of miR-337-3p on MMP-14 promoter. Previous studies have implicated the participation of AGO1 and AGO2 in small RNA-inhibited gene expression [33]. AGO1, but not AGO2, was selectively enriched in vicinity to the miR-744 complementary site on cyclin B1 promoter [34]. In this study, our evidence indicated that AGO2, but not AGO1, was enriched at the MMP-14 promoter in miR-337-3p over-expressing NB cells. In addition, knockdown of AGO2 abolished the miR-337-3p-induced binding of repressive epigenetic markers on MMP-14 promoter. We believe that miR-337-3p/AGO2 complexes may bring in co-repressors such as histone methyltransferases to repress gene expression, which warrants our further investigation.
In summary, we have shown that miR-337-3p is under-expressed in human NB, and over-expression of miR-337-3p inhibits the growth, invasion, metastasis, and angiogenesis of NB cells in vitro and in vivo. Furthermore, miR-337-3p suppresses the transcription of MMP-14 via epigenetic repression of its promoter activity in NB cell lines. This study extends our knowledge about the regulation of MMP-14 at transcriptional level by miRNAs, and suggests that miR-337-3p may be of potential values as a novel therapeutic target for human NB. www.impactjournals.com/oncotarget

Patient tissue samples
Approval to conduct this study was obtained from the Institutional Review Board of Tongji Medical College (approval number: 2011-S085). Fresh tumor specimens from 30 well-established primary NB cases were collected, and stored at −80°C until use. Based on the Shimada classification system, including the mitosis karyorrhexis index, degree of neuroblastic differentiation and stromal maturation, and patient's age, 14 patients were classified as having favorable histology and 16 as having unfavorable histology. According to the INSS, 4 patients were classified as stage 1, 9 as stage 2, 9 as stage 3, 4 as stage 4, and 4 as stage 4S. Total RNA of normal human dorsal ganglia, pooling from 21 male/female Caucasians, was obtained from Clontech (Mountain View, CA).

miRNA prediction and expression detection
miRNA binding sites within the MMP-14 promoter were analyzed using the algorithm microPIR [16]. Cytoplasmic and nuclear fractions were prepared by using the NE-PER Nuclear and Cytoplasmic Extraction Reagents (Thermo Fisher Scientific, Inc., Waltham, MA). The mature miR-337-3p levels were determined using Bulge-Loop TM miRNAs qPCR Primer Set (RiboBio Co. Ltd, Guangzhou, China). After cDNA was synthesized with a miRNA-specific stem-loop primer, the quantitative PCR was performed with the specific primers (Supplementary Table S2). The miRNA levels were normalized as to those of U6 snRNA.

Nuclear run-on assay
Nuclear run-on assay was performed based on the incorporation of biotin-16-uridine-5′-triphosphate (biotin-16-UTP) into nascent transcripts as previously described [33]. Briefly, nuclei of 5 × 10 6 tumor cells were isolated and consequently incubated in reaction buffer containing rNTPs and biotin-16-UTP (Roche, Indianapolis, IN) at 30°C for 45 min. The reaction was stopped by adding RNase-free DNase I, and nuclei were lysed and treated with proteinase K. Total RNA was extracted using Trizol (Invitrogen), and biotinylated nascent RNA was purified using agarose-conjugated streptavidin beads (Invitrogen) for real-time quantitative RT-PCR assay.

Chromatin immunoprecipitation
ChIP assay was performed according to the instructions of EZ-ChIP kit (Upstate Biotechnology, Temacula, CA) [33,36,40,43], with antibodies for AGO1, AGO2, EZH2, H3K27me3, H3K9me2, RNA Pol II, and Sp1 (Upstate Biotechnology, Temacula, CA). Lysates were treated with either RNase H (10 U) or RNase A (20 μg) prior to immunoprecipitation. DNA was sonicated into fragments of an average size of 200 bp. Real-time qPCR was performed with SYBR Green PCR Master Mix and primer sets indicated in Supplemenatry Table S2. The amount of immunoprecipitated DNA was calculated in reference to a standard curve and normalized to input DNA.

Scratch migration assay
Tumor cells were cultured in 24-well plates and scraped with the fine end of 1-ml pipette tips (time 0). Plates were washed twice with phosphate buffered saline to remove detached cells, and incubated with the complete growth medium. Cell migration was photographed using 10 high-power fields, at 0, 24 hr post-induction of injury. Remodeling was measured as diminishing distance across the induced injury, normalized to the 0 hr control, and expressed as outgrowth (μm) [7,35,36,43].

Cell invasion assay
Matrigel invasion assay was performed using membranes coated with Matrigel matrix (BD Science, Sparks, MD). Homogeneous single cell suspensions (1 × 10 5 cells/well) were added to the upper chambers and allowed to invade for 24 hrs at 37°C in a CO 2 incubator. Invaded cells were stained with 0.1% crystal violet for 10 min at room temperature. Quantification of invaded cells was performed according to published criteria [7,35,37,38,43,44].

Tube formation assay
Fifty microliters of growth factor-reduced matrigel were polymerized on 96-well plates. HUVECs were serum starved in RPMI1640 medium for 24 hrs, suspended in RPMI1640 medium preconditioned with tumor cells, added to the matrigel-coated wells at the density of 5 × 10 4 cells/well, and incubated at 37°C for 18 hrs. Quantification of anti-angiogenic activity was calculated as previously described [35][36][37].

In vivo growth and metastasis assay
All animal experiments were approved by the Animal Care Committee of Tongji Medical College (approval number: Y20080290). For the in vivo tumor growth studies, 2-month-old male nude mice (n = 5 per group) were injected subcutaneously in the upper back with 1 × 10 6 tumor cells stably transfected with empty or miR-337-3p precursor vectors. One month later, mice were sacrificed and examined for tumor weight and gene expression. The experimental metastasis (0.4 × 10 6 tumor cells per mouse, n = 5 per group) studies were performed with 2-month-old male nude mice as previously described [7,35,36].

Immunohistochemistry
Immunohistochemical staining was performed as previously described [7,35,36,45], with antibodies specific for MMP-14 (Abcam Inc; Santa Cruz Biotechnology; 1:200 dilution), VEGF, and CD31 (Santa Cruz Biotechnology; 1:200 dilutions). The negative controls included parallel sections treated without primary antibody or with rabbit polyclonal IgG (Abcam Inc.). The immunoreactivity in each tissue section was assessed by at least two pathologists. The degree of positivity was determined according to the percentage of positive tumor cells.

Statistical analysis
Unless otherwise stated, all data were shown as mean ± standard error of the mean (SEM). The χ 2 analysis and Fisher exact probability analysis were applied to compare the gene expression in tumor tissues with www.impactjournals.com/oncotarget different clinicopathological features. Pearson's coefficient correlation was applied for analyzing the relationship among gene expression. The Kaplan-Meier method was used to estimate survival rates, and the log-rank test was used to assess survival difference. Difference of tumor cells was determined by t test or analysis of variance (ANOVA).