Background
Nasopharyngeal carcinoma (NPC) is a head and neck epithelial malignancy that occurs frequently in Southern China [
1‐
5]. Epstein-Barr virus exposure [
6‐
15], diet, and genetic factors [
16] are implicated in its etiology. Although radiotherapy is an effective treatment for NPC patients in the early stages of the disease [
17‐
20], the majority (75–90%) of NPC cases are predisposed to metastasis at initial diagnosis [
21‐
24], which hampers efficacious treatment and poses a high risk of disease recurrence. Enhancing our understanding of the molecular mechanisms underlying NPC may promote the development of effective metastasis-targeted therapy and improve the overall prognosis of patients with this disease.
Noncoding RNAs, including microRNAs (miRNAs), long noncoding RNAs (lncRNAs) [
25‐
33] and circular RNAs (circRNAs) [
34‐
37], have important roles in the regulation of gene expression and are involved in the development of a variety of human diseases, including cancer. The lncRNA actin filament associated protein 1 antisense RNA 1 (
AFAP1-AS1) maps to the antisense DNA strand of
AFAP1 and was first identified as highly expressed in esophageal cancer [
38].
AFAP1-AS1 is also highly expressed in lung cancer and pancreas cancer, and hepatocellular carcinoma [
39‐
43]. We previously discovered that
AFAP1-AS1 was significantly highly expressed in NPC and associated with metastasis and poor prognosis [
44,
45]. Numerous cytoskeleton proteins, including proteins in the small GTPase Rho/Rac signaling pathway, were identified and significantly regulated using the liquid chromatography-tandem mass spectrometry (LC-MS/MS) by si
AFAP1-AS, suggesting that
AFAP1-AS1 may regulate NPC migration and invasion through this pathway; however, as a newly identified lncRNA, the mechanism by which
AFAP1-AS1 regulates the Rho/Rac signaling pathway remains unknown.
A new model of lncRNA involvement in gene regulation has recently been proposed, called competing endogenous RNA (ceRNA). This model proposes that lncRNAs and mRNA transcripts affect one another by competitively combining with miRNA response elements (MREs) to influence posttranscriptional regulation [
46‐
48]. In this way, lncRNAs bind with miRNAs to modulate regulation of protein-coding gene expression and participate in the regulation of cell behavior. These interactions between mRNAs, miRNAs, and lncRNAs generate complex regulatory networks [
49]. In this study, we found that
AFAP1-AS1 acts as a ceRNA, competitively binding with miR-423-5p and directly regulating
RAB11B and
LASP1 genes in the Rho/Rac pathway.
AFAP1-AS1 can also regulate the expression of
FOSL2 and its downstream genes by competing with
FOSL2 to bind miR-423-5p, resulting in increased migration and invasion of NPC through the Rho/Rac signaling pathway.
Methods
Tissues and cell lines, chemicals, plasmids, and transfection
Thirty-two nasopharyngeal carcinoma samples and 13 nontumor nasopharyngeal epithelium tissues were obtained from patients. All samples were collected with the consent of patients and the experiments were approved by the ethics committee of Hunan Cancer Hospital, Changsha, China. All fresh tissues were immersed in RNALater (Ambion, Austin, TX, USA). The tissue samples were stored in a − 80 °C laboratory freezer. RNA was isolated using TRIzol reagent (Invitrogen, Carlsbad, CA, USA) according to the manufacturer’s protocol. The NPC cell lines 5-8F and HNE2 were maintained in our laboratory in RPMI 1640 medium supplemented with 10% fetal bovine serum (FBS; Invitrogen), penicillin (100 U/ml, Sigma-Aldrich, St. Louis, MO, USA), and streptomycin (100 mg/ml, Sigma-Aldrich). All cell lines had been recently tested for mycoplasma contamination and identified no cross-contaminated cell lines or mycoplasma contamination.
The synthetic miR-423-5p mimics and inhibitors were supplied by Ruibo Co. (Guangzhou, China). siRNAs and nontargeting scrambled control siRNAs were provided by Invitrogen. The sequences of the
AFAP1-AS1- and
FOSL2-targeting siRNAs are listed in Additional file
1: Table S1. To construct the
AFAP1-AS1 expression vector, the entire
AFAP1-AS1 sequence was amplified by reverse transcription PCR (RT-PCR) and cloned into the pcDNA3.1 vector. The open reading frame of
FOSL2 (NM_005253) was cloned into the pENTR vector and tagged with the C-terminal Flag and His peptide sequences was purchased from the Vigene Co. (Rockville, MD, USA). Transfection of cells with plasmids, siRNAs, and miRNAs was performed using Lipofectamine 3000 (Life Technologies, Grand Island, NY, USA) or HiPerFect (Qiagen, Hilden, Germany) transfection reagents, according to the manufacturer’s instructions.
Online expression profiles analysis
NPC data sets and the corresponding clinical data were downloaded from publicly available GEO databases. These included 62 patients with NPC and 6 non-NPC nasopharyngeal epithelial tissues from GSE73460 [
50], 6 patients with NPC and 12 non-NPC epithelial tissues from GSE32906 [
50] and 122 patients with NPC with survival status from GSE70970 [
51].
Five software tools, RNAhybrid, RNA22, PITA, TargetScan, and RegRNA, were used for miRNA target prediction in this study. For miRNA target validation assays, 58 bp fragments of the
AFAP1-AS1 sequence, or the
RAB11B,
LASP1,
RAC1, and
FOSL2 3′-UTRs containing the wild-type (WT) and mutant (MT) binding sites for miR-423-5p, were cloned into the pMIR-REPORT vector. The sequences of these synthetic oligonucleotides are listed in Additional file
1: Table S1. Cells (5-8F and HNE2) were cotransfected with pMIR-REPORT vectors containing the WT sequences of the
RAB11B,
RAC1,
LASP1, or
FOSL2 3′-UTRs or their corresponding mutants, along with the pRL-TK control vector and miR-423-5p mimics, inhibitors, or negative control.
For the
LASP1 promoter activity analysis, the Promoter 2.0 Prediction Server (
http://www.cbs.dtu.dk/services/Promoter/) was used to predict potential AP1-binding sites in the
LASP1 promoter sequence. Three tandem the wild type or mutant AP1-binding sites were inserted into the TA-luciferase vector as synthetic oligonucleotides encoding the wild type or mutant AP1-binding sites (Additional file
1: Table S1).
The AP-1 cis-element reporter plasmid (the AP-1 reporter), which contains multiple conserved AP1-binding sites for sensitive detection of AP-1 transcriptional activity, was purchased from Promega (pAP1-TA-luc; Promega, Fitchburg, WI, USA) to measure the AP-1 transcriptional activity. Plasmid transfection and luciferase assays were performed as described previously [
52]. Cells were harvested 48 h posttransfection and assayed using a Dual Luciferase Assay (Promega) according to the manufacturer’s instructions. All transfection assays were carried out in triplicate.
Wound healing and transwell assays
For wound healing assays, 5-8F and HNE2 cells were seeded and grown to 90% confluence in 6-well culture plates, and a 200 μl pipet tip was used to create a scratch in the cell monolayer. Images were captured and measured with an ocular ruler to ensure that all wounds were the same width at the beginning of each experiment.
For transwell assays, 5-8F and HNE2 cells in 200 μl serum-free media were placed into upper transwell chambers (8.0 μl pore size, BD Biosciences, Franklin Lakes, NJ, USA) for invasion assays after transfection, according to the manufacturer’s protocol. Invaded cells were fixed and stained with methanol and 0.1% crystal violet, counted under a microscope, and imaged. Independent experiments were carried out in triplicate.
Western blotting
Proteins were extracted and separated according to previously described methods. Membranes were incubated overnight at 4 °C with primary anti-RHOA, anti-RHOC, anti-RAC2, anti-RAB10, anti-RAB11B, anti-RAB11A, anti-RHOGDI, anti-PFN1, anti-LASP1, or anti-FOSL2 antibodies (Proteintech, Wuhan, China). Blots were visualized by exposure to X-ray films using an enhanced chemiluminescence detection system (EMD Millipore, Burlington, MA, USA). Blots were probed with anti-GAPDH antibody (Cell Signaling Technology, Danvers, MA, USA) as an endogenous control for equal loading.
Real-time PCR
Total RNA was isolated with TRIzol reagent (Invitrogen) according to the manufacturer’s protocol. For miRNA real-time PCR, cDNA was synthesized using the miScript system (Ruibo) according to the manufacturer’s instructions. miR-423-5p expression levels were measured using a Ruibo miRNA primer assay (Ruibo) and a miDETECT Track miRNA qRT-PCR kit (Ruibo) according to the manufacturer’s instructions. Data were normalized to the small nuclear RNA RNU6B (U6 snRNA) expression level. For mRNAs and lncRNAs, real-time PCR expression analysis was carried out using a SYBR green real-time PCR kit (Applied Biological Materials, Inc., Richmond, BC, Canada). Data were normalized to the
GAPDH expression level and further normalized to the negative control, unless otherwise indicated. The primers used for PCR are listed in Additional file
1: Table S1. Fold changes in expression were calculated using a relative quantification (2
-ΔΔCt) method. All reactions were performed in triplicate and repeated in three independent experiments.
in vivo nude mouse models
Male BALB/C nude mice (5–6 weeks of age, 16–18 g) were maintained in the experimental animal center of Central South University in a pathogen-free facility. The Institutional Animal Ethics Committee of Central South University approved this animal study. Mice were divided into four groups, including the over-AFAP1-AS1, negative control (NC), miR-423-5p, and over-FOSL2 groups. For metastasis assay, 2 × 106 5-8F cells in a volume of 200 ml with corresponding treatment were tail-vein injected into five-week-old male BALB/c nude mice which were randomly divided into four groups (n = 10 for each group). 60 days later, the mice were killed, and all the lungs are surgically removed and the number of macroscopically visible pulmonary metastases nodules per mouse was counted by two experienced pathologists. Then, the lung tissues were fixed in 10% neutral phosphate-buffered formalin, followed by HE staining.
Statistical analysis
All continuous variables are presented as the means ± standard deviation (SD), and categorical data are expressed as percentages. GraphPad Prism 5.0 (GraphPad Software, Inc., La Jolla, CA, USA) was used for statistical analysis. Student’s t-test was performed to assess significant differences between two groups using p < 0.05 as the significance threshold. Survival analysis was performed by Kaplan-Meier curves and log-rank test for significance. A two-sided p-value of < 0.05 was considered statistically significant.
Discussion
A ceRNA hypothesis was proposed by Salmena et al. in 2011 [
55]. The theory explains why mRNA, pseudogenes, lncRNA, and other molecules share miRNA response elements (MREs) and bind to identical miRNAs competitively to affect cell status [
56]. miRNAs and MREs are two key elements in this hypothesis; miRNAs bind not only to MREs in mRNAs but also to pseudogenes and lncRNAs, influencing mRNA and lncRNA expression levels posttranscriptionally. For example, the tumor suppressor gene phosphatase and tensin homolog (
PTEN), a negative regulator of the oncogenic phosphoinositide 3-kinase/Akt signaling pathway, provides the best evidence for disease-associated ceRNAs [
57].
PTEN pseudogene 1 (
PTENP1) can increase PTEN expression by binding to several miRNAs to suppress proliferation of cancer cells [
58,
59]. Conversely, when
PTENP1 is downregulated, more miRNAs are available to inhibit
PTEN expression, facilitating tumor growth. Both
PTEN and
PTENP1 are controlled by ceRNA circuitry in many cancers [
60]. LncRNAs interact with miRNAs and mRNAs as components of the cellular RNA language, which comprises a complex, finely controlled network.
AFAP1-AS1 is highly expressed and tightly associated with tumor progression in many cancers including NPC [
43]. We previously showed that
AFAP1-AS1 was associated with poor prognosis in NPC and it promoted cell migration and invasion via regulation of actin filament integrity and the Rho/Rac signaling [
44]. In this study, we explored the detailed molecular mechanism of
AFAP1-AS1 as a ceRNA in regulation of NPC cell migration and invasion. We hypothesized that
AFAP1-AS1 may act as a ceRNA to regulate the Rho/Rac pathway; therefore, we employed bioinformatics methods to predict potential miRNAs regulated by
AFAP1-AS1. miR-423-5p was the only miRNA predicted by all four tools used for these analyses and had the lowest minimum free energy, suggesting that
AFAP1-AS1 is likely a direct target of miR-423-5p. Many reports have identified miR-423-5p as an oncogene in various cancers, including gastric cancer and hepatocellular carcinoma [
61,
62]. In this study, we found that miR-423-5p was low-expressed in NPC and its expression was negatively correlated with the clinical TNM stages in NPC tumors. Overexpression of miR-423-5p inhibited NPC cells migration and invasion, and inhibition of miR-423-5p produced the opposite phenotype. The function of miR423-5p had an opposite effect to those of
AFAP1-AS1 on migration and invasion. Luciferase assays confirmed that miR-423-5p regulated the 3’-UTR of
AFAP1-AS1. These results suggested that
AFAP1-AS1 may regulate downstream genes through binding of miR-423-5p, and that miR-423-5p may have a role as a tumor suppressor gene through its effects on different downstream genes in NPC.
We previously determined by the proteomic analysis that expression levels of at least 209 proteins were significantly altered after
AFAP1-AS1 knockdown in NPC cells. Among these, the expression levels of many proteins involved in regulation of cytoskeletal dynamics were confirmed as significantly altered using liquid chromatography-tandem mass spectrometry (LC-MS/MS) after
AFAP1-AS1 knockdown, including those in the small GTPase Rho/Rac signaling pathway:
RAB10, RAB11A, RAC2, PFN1, RHOA, RAC1, RHOC, LASP1, and
RAB11B [
44]. The Rho/Rac signaling pathway is a well-established regulator of the actin cytoskeleton, as highlighted by the RAC1 dependence of membrane ruffling and RHOA-induced stress-fiber formation [
63]. In this study, we focused on the role of
AFAP1-AS1 as a noncoding RNA in regulation of the expression of these proteins. We investigated whether
AFAP1-AS1 acted as a ceRNA through miR-423-5p to regulate the Rho/Rac pathway. miR-423-5p inhibited the expression of
RAC1,
RHOC,
LASP1, and
RAB11B and increased that of other molecules, including
RAB10,
RAB11A,
RAC2,
PFN1, and
RHOA, consistently with the results after
AFAP1-AS1 inhibition or overexpression [
26]. These results implied that
AFAP1-AS1 regulated the Rho/Rac signaling pathway through miR-423-5p. As a microRNA, miR-423-5p may directly regulate expression or translation of its target genes; therefore, we first examined whether some molecules including
RAC1,
RAB11B,
RHOC, and
LASP1 in the Rho/Rac pathway were direct target genes of miR-423-5p, since their expression levels were reduced by the specific miR-423-5p mimics. Bioinformatics analysis predicted the
RAB11B and
LASP1 genes as targets of miR-423-5p. Further experiments confirmed that miR-423-5p directly targeted these two genes and that
AFAP1-AS1 acted as a ceRNA of
RAB11B and
LASP1, regulating their expression by competitively binding miR-423-5p, thus affecting the migration and invasion of NPC. As we know,
LASP1 is an actin-binding phosphoprotein that is overexpressed in several cancers [
64‐
66] and it is a cAMP- and cGMP-dependent signaling protein which binds to the actin cytoskeleton at extensions of the cell membrane [
67,
68].
RAB11B belongs to the Rab family and regulates exocytotic and endocytotic pathways [
69]. Our experiments confirmed the hypothesis that
AFAP1-AS1 acts as a ceRNA to regulate the expression of key genes, such as
LASP1 and
RAB11B, by competitively binding miR-423-5p, resulting in dysregulation of the Rho/Rac1 signaling pathway and migration and invasion of human NPC cells [
70‐
72].
The expression of
RAB10,
RAB11A,
RAC2,
PFN1, and
RHOA were upregulated in 5-8F and HNE2 cells transfected with miR-423-5p mimics, which is consistent with the results of experiments using si
AFAP1-AS1 and suggests that
AFAP1-AS1 and miR-423-5p are tightly associated with the Rho/Rac signaling pathway. miR-423-5p may also indirectly regulate the expression of these proteins through its effects on other molecules. To identify other direct targets of miR-423-5p, we screened genes whose expression levels were highly correlated with those of
AFAP1-AS1 in NPC tissues based on the results of previous microarray analyses and the ceRNA theory.
FOSL2, a member of the Fos family [
73], was analyzed using various prediction tools and found to contain several potential miR-423-5p binding sites in its 3′-UTR, while the expression of
AFAP1-AS1 and
FOSL2 was significantly correlated in NPC GEO datasets [
53].
FOSL2 is a member of AP-1 transcription factor complexes. FOS and Jun protein family members can form homo- or heterodimers and bind to the promoters of specific genes to promote or inhibit their transcription in combination with other transcription and regulatory factors. In this study, we found that FOSL2 could regulate the
LASP1 transcription by binding its promoter, suggesting that
AFAP1-AS1 acts as a ceRNA in regulation of
FOSL2 and the AP-1 transcription factor activation, thereby affecting the expression of small GTPase molecules. Whether
AFAP1-AS1 also regulates the expression of other downstream genes through
FOSL2 is worthy of further investigation.
In this study, we confirmed that
AFAP1-AS1 functions as a ceRNA by adsorption of miR-423-5p and verified that
LASP1,
RAB11B, and
FOSL2 are downstream target genes of miR-423-5p (Additional file
2: Figure S5). However, miRNAs can target multiple genes, and there may be other protein-encoding genes regulated by miR-423-5p among the 209 proteins regulated by
AFAP1-AS1 in NPC cells using our previous proteomic approach; this is also worthy of further investigation.