Background
Nasopharyngeal carcinoma (NPC) is an epithelial malignancy with striking racial and geographic distribution differences. It is particularly prevalent among populations from southern China, Southeast Asia, northern Africa and Alaska. These incidence rates are approximately 100-fold higher than in the Caucasian populations [
1]. Several environmental factors, including infection with the Epstein-Barr virus (EBV), long-term cigarette smoking, occupational exposure to formaldehyde, and various dietary factors, have been reported to confer the risk of developing NPC [
2]. Furthermore, numerous genetic linkage and association studies have reported a few genes contributing to the risk of this malignancy [
3,
4]. The identification of susceptibility genes contributing to NPC would assist in predicting individual and population risk of NPC development and would help to clarify the pathogenesis relevant to this disorder.
Argonaute 2 (AGO2), a member of the Argonaute protein family, can bind microRNAs (miRNAs) or short interfering RNAs (siRNAs) and mediate the repression of specific target RNAs either by degrading RNA or inhibiting translation [
5]. Of the four human AGO proteins, AGO2 is the only member with intrinsic endoribonuclease activity and essential non-redundant slicer-independent function within the mammalian miRNA pathway [
6]. Recently, alterations (genomic amplifications and/or over-expression) of the
AGO2 gene have been extensively described in a variety of cancers [
7‐
11], and these alterations have been shown to be linked with an increased metastasis [
11‐
13] and poorer prognosis [
12]. Further studies indicated that the AGO2 is involved in several steps of cancer development, including cell proliferation, differentiation, apoptosis, migration and invasion [
8‐
10,
14,
15]. With regard to NPC, the elevated mRNA expressions of
AGO2 were observed in tumor tissues and latent membrane protein 1 (LMP1)-positive tumors compared with normal adjacent nasopharyngeal epithelium tissues and LMP1-negative tumors, respectively [
16]. Taken together, these studies suggest that the AGO2 may play crucial roles in the cancer development and progression.
Theoretically, the genetic variants such as single nucleotide polymorphisms (SNPs) within the
AGO2 gene, which may alter the expression of
AGO2 and then influence the miRNAs processing and function, could result in genotype-dependent differences in risk of cancers. Indeed, there is increasing evidence that the genetic variants in the
AGO2 gene are associated with the risk or development of several cancers, including breast cancer [
17,
18] and malignant peripheral nerve sheath tumor [
19]. The role of genetic variants within
AGO2 in NPC, however, has never been specifically investigated. In the present study, we examined whether the genetic variation in the
AGO2 gene affect the risk or severity of NPC in the Chinese populations. We also evaluated the biologic roles of
AGO2 in the development of NPC by functional assays.
Methods
Study population
This study consisted of two populations of patients with NPC and control subjects resided in Guangxi and Guangdong province, respectively, both of which were well-known high-risk regions for NPC located in southern China (Additional file
1: Table S1). The Guangxi population, which contained 855 incident patients with NPC and 1036 controls, has been described in detail previously [
20]. Briefly, all subjects were unrelated ethnic adult Chinese and residents in Nanning city (Nanning, China) and the surrounding regions. All patients were newly diagnosed and pathologically confirmed, and were consecutively recruited between September 2003 and January 2008 at the Guangxi Cancer Hospital (Nanning, China). Patients that received chemotherapy or radiotherapy prior to surgery or had other type of cancer were excluded from this study. Tumor staging was performed according to the tumor-node-metastasis (TNM) classification by the 1997 American Joint Committee on Cancer (AJCC) system. All TNM classifications were determined by senior pathologists of the hospital on the basis of the postoperative histopathologic examination. All the controls were recruited in the same regions during the same time that the NPC cases were collected. The selection criteria for the controls included no individual history of cancer and frequency matching to the cases on sex and age (± 5 years). The Guangdong population, which contained 997 NPC patients and 972 controls, has been described in detail previously [
3]. Briefly, all subjects were unrelated ethnic adult Chinese and residents in Guangdong province. All patients were consecutively recruited between October 2005 and October 2007 from Sun Yat-sen University Cancer Center (SYSUCC) in Guangzhou city (Guangzhou, China). All patients were histopathologically diagnosed by at least two pathologists according to the World Health Organization (WHO) classification. During the same period, control subjects were recruited from physical examination centers of several large comprehensive hospitals in local communities in Guangdong and were frequency matched to the cases by age (± 5 years), sex, geographic region and ethnicity. All the subjects were interviewed for collection of personal information on demographic factors, medical history, cigarette smoking and alcohol use via structured questionnaire. This study was approved by the Ethics Committee of Beijing Institute of Radiation Medicine (Beijing, China). At recruitment, the written informed consent was obtained from all the participants involved in this study.
From the Guangxi population of 855 patients with incident NPC, 37 patients who had undergone resection before receiving any further treatment at Guangxi Cancer Hospital were selected, and primary NPC biopsies were collected from them (Additional file
1: Table S2). The histological type of all tumor tissues was poorly differentiated squamous cell carcinoma (SCC). Histological non-cancerous nasopharyngeal epithelium tissues were collected from 18 of the 1036 control subjects (Additional file
1: Table S2). All the tissues were fixed in paraformaldehyde, embedded in paraffin wax, and prepared for subsequent immunohistochemical staining.
SNP selection and genotyping
Twenty-five SNPs in the
AGO2 gene were selected for genotyping in our study (Table
1). These SNPs were chosen on the basis of previous reports of their association with the risk for cancer, or with potential function, and a comprehensive tag SNP approach. Two SNPs (rs2292779 and rs7005286) significantly associated with cancer in previous studies [
18,
19] were directly selected in our study. Two SNPs (rs4961280 and rs11996715) located in the promoter region (−1809A/C and −1686 A/C) of
AGO2, were selected as they could alter the expression of
AGO2. Tag SNPs of the
AGO2 gene were selected from the Han Chinese in Beijing, China (CHB) HapMap database (HapMap release 27, February 2009) using HaploView (pairwise
r2 > 0.8) [
21]. To ensure enough statistical power, a value of 0.10 of minor allele frequency (MAF) was set as the threshold value of inclusion in this study. Finally, 25 tag SNPs were selected to capture the
AGO2 gene.
Table 1
Positions and frequencies of SNPs within AGO2 gene
1 | rs4961280 | C/A | 0.111 | −1809 | Promoter |
2 | rs11996715 | C/A | 0.430 | −1686 | Promoter |
3 | rs4961226 | G/T | 0.400 | 3050 | Intron 1 |
4 | rs10088596 | C/T | 0.453 | 5311 | Intron 1 |
5 |
rs883596
| G/C | 0.105 | 9957 | Intron 1 |
6 | rs13261055 | A/G | 0.386 | 16208 | Intron 1 |
7 | rs7001653 | G/A | 0.444 | 24567 | Intron 1 |
8 | rs7819727 | A/G | 0.452 | 48394 | Intron 1 |
9 | rs7009635 | T/C | 0.367 | 50724 | Intron 2 |
10 | rs7005286 | C/T | 0.337 | 51145 | Intron 2 |
11 | rs11776034 | C/G | 0.211 | 53712 | Intron 2 |
12 | rs3735805 | C/T | 0.356 | 68083 | Intron 3 |
13 |
rs2977490
| A/G | 0.453 | 72247 | Intron 3 |
14 | rs2292773 | C/T | 0.278 | 73107 | Intron 4 |
15 | rs3928672 | G/A | 0.222 | 73438 | Intron 4 |
16 | rs2271735 | G/T | 0.178 | 76295 | Intron 6 |
17 |
rs2977481
| G/C | 0.183 | 83748 | Intron 10 |
18 | rs2292779 | G/C | 0.244 | 84394 | Intron 11 |
19 | rs2944764 | C/T | 0.239 | 84576 | Intron 11 |
20 | rs11166983 | G/A | 0.278 | 87519 | Intron 11 |
21 | rs12542354 | C/T | 0.239 | 89462 | Intron 13 |
22 | rs13252337 | A/G | 0.186 | 96222 | Intron 16 |
23 | rs2977469 | G/A | 0.465 | 96314 | Intron 16 |
24 | rs2977464 | C/T | 0.333 | 100412 | Intron 16 |
25 | rs2977462 | T/C | 0.467 | 102467 | Intron 16 |
All SNPs were genotyped using GenomeLab SNPstream Genotyping System according to the manufacturer’s instructions (Beckman Coulter, USA) as previously described [
22]. This platform uses a single base pair extension reaction to incorporate two-color fluorescence terminal nucleotides which are detected by a specialized imager. Details of the primer sequences are listed in Additional file
1: Table S3. The genotype data were analyzed by GenomeLab SNPstream version 2.2 software (Beckman Coulter, USA). To ensure genotyping quality, we genotyped 20 randomly selected duplicate samples and 4 blanks in each 384-well plate and obtained a concordance rate of >99 %.
Immunohistochemistry in NPC tissues and non-cancerous nasopharyngeal tissues
The paraformaldehyde-fixed and paraffin-embedded NPC tissues (n = 37) and non-cancerous nasopharyngeal tissues (n = 18) (Additional file
1: Table S2) were analyzed for protein expression of AGO2. Two slides from each biopsy were stained with hematoxylin and eosin for routine histological evaluation. Histologic slides with tissue sections were subjected to immunohistochemistry (IHC) as previously described [
20,
22], using the primary antibody raised against AGO2 (diluted 1:200; ab57113, Abcam, UK). Negative controls and positive controls were performed at the same time. Photographs were taken (BX51 microscopic/Digital Camera System; Olympus) for study comparison. The IHC signals were scored as previously described [
20,
22] by two pathologists (Wu J and Huang W) who did not have knowledge of ligand-binding assay results or patient outcome. Briefly, a proportion score was assigned representing the estimated proportion of positive staining tumor cells (0, none; 1, < 1/100; 2, 1/100 to < 1/10; 3, 1/10 to < 1/3; 4, 1/3 - 2/3; 5, > 2/3). Average estimated intensity of staining in positive cells was assigned an intensity score (0, none; 1, weak; 2, intermediate; 3, strong). The two parameters were combined and resulting in an overall score (0 or 2–8). The scores were classified into three groups: group 1 (score 0, negative expression), group 2 (score 2–4, low expression) and group 3 (score 5–8, high expression). A total of five fields per slide were selected, counted, and averaged.
AGO2 knockdown in NPC cell line
The human nasopharyngeal carcinoma cell line CNE2Z was obtained from the Peking Union Medical College (Chinese Academy of Medical Science, China) and maintained in RPMI 1640 (Gibco-BRL, USA) supplemented with 10 % fetal bovine serum (HyClone, USA) and 1 % penicillin/streptomycin (Gibco-BRL, USA) at 37 °C under a 5 % CO2 atmosphere. The AGO2 specific shRNA expression vectors and the scrambled ineffective shRNA cassette (as the negative control) in the pGPU6/GFP/Neo plasmid were purchased from GenePharma (China). The sequences of the three shRNAs directed against AGO2 are as follows: sh1: 5’-GCAAGGATCGCATCTTCAAGG-3’; sh2: 5’-GGTCTAAAGGTGGAGATAAGG-3’; and sh3: 5’-GCCTGAAGATCAACGTCAAGC-3’. The sequence of control shRNA is 5’-GTTCTCCGAACGTGTCACGT-3’. A mixture of constructs including three AGO2 shRNAs or the control shRNA construct was transfected into CNE2Z cells using Lipofectamine 2000 (Invitrogen, USA). The efficiency of RNA interfering was assessed by western blot analysis.
Western blot
Cells were lysed in Laemmli sample buffer (Bio-Rad, USA) with an EDTA-free, protease inhibitor cocktail (Roche, USA). Proteins at the same amount were separated by SDS-PAGE and transferred onto polyvinylidene difluoride (PVDF) membranes (Millipore, USA). After probing with primary and secondary antibody, antigen-antibody complex was visualized by enhanced chemiluminescence-plus reagent (Pierce Biotechnologies, USA). For AGO2, mouse anti-AGO2 monoclonal antibody (diluted 1:1000; ab57113, Abcam, UK) was used. As an internal control, mouse anti-β-actin monoclonal antibody (diluted 1:1000; sc-130301, Snata cruz, USA) was used.
Cell proliferation, apoptosis and migration assays
Cell proliferation was evaluated by measuring cell viability with Cell Counting Kit-8 (CCK-8) assay (Beyotime Inst Biotech, China) according to the manufacturer’s instructions. Cells (2 × 103 cells per well) were plated in 96-well plates in triplicate. CCK-8 was added to each well at a final concentration of 10 % at different time points (24, 48, 72 and 96 h) and incubation continued at 37 °C for 50 min. Subsequently, the absorbance of the samples was measured at 450 nm using a Multiskan MK3 microplate reader (Thermo Labsystems, USA) to calculate the numbers of viable cells in each well.
Apoptosis was detected by flow cytometry using the Annexin V-APC/7-AAD Apoptosis Detection Kit (KeyGEN, China). Briefly, cells were harvested, washed, resuspended in the staining buffer, and doubly stained with annexin-V and 7-amino-actinomycin D (7-AAD). For each experiment, 5 × 104 cells were analyzed using FACSCalibur and CellQuest software (BD Biosciences, USA). The Annexin V-positive cells were regarded as apoptotic cells.
Ability of cell migration was evaluated by Transwell assay (Corning Inc, USA) according to the manufacturer’s instructions. Cells (1.5 × 105 cells per well) in 300 μl serum-free medium were placed in the upper chamber of the transwell, whereas the lower chamber was loaded with 500 μl medium containing 20 % fetal bovine serum. After 16 h of incubation, cells that migrated to the lower chamber were fixed and stained with crystal violet. The number of cells was counted in five random microscopic fields (magnification 400 ×).
Genome-wide expression microarray and pathway analysis
Total RNA was isolated for microarray analysis from three biological replicates of cells transfected with
AGO2 shRNAs or control shRNA using TRIzol reagent (Invitrogen, USA) according to the manufacturer’s instructions. All samples with an RNA integrity number (RIN) ≥ 8.0 and A260/A280 ratio 1.8-2.1 were considered suitable for microarray analysis. The Affymetrix GeneChip® Human Gene U133 Plus 2.0 Arrays were used for genome-wide expression profiling assay, which was performed by CapitalBio Corporation (Beijing, China) according to the manufacturer’s instructions (Affymetrix, USA). Raw feature data from microarrays were subsequently corrected for background and normalized, and the log
2 intensity expression summary values for each probe set were calculated using Robust Multiplechip Average (RMA). Significantly altered genes following
AGO2 knockdown were analyzed by significance analysis of microarray (SAM, version 3.01), and the
P values of the
t test were calculated for each gene. Multiple hypothesis testing was performed to calculate the false discovery rates (FDRs) through the Bioconductor package Qvalue (
http://www.bioconductor.org). Significantly altered genes following
AGO2 knockdown were investigated for biological processes and signaling pathways using the cytoscape plug-in of Reactome FI [
23].
Statistical analysis
Genotype and allele frequencies of
AGO2 polymorphisms were determined by direct counting, and departures from Hardy-Weinberg equilibrium (HWE) were tested using the random-permutation procedure implemented in the Arlequin package (
http://cmpg.unibe.ch/software/arlequin3/). Multivariate logistic regression analysis was done to evaluate whether there were genetic associations between the
AGO2 polymorphisms and risk and severity of NPC. The
P values, odds ratios (ORs), and 95 % confidence intervals (CIs) were calculated and adjusted for sex, age, status of smoking and drinking, smoking level, family history and nationality where it was appropriate. To assess the probability of a spurious association due to multiple comparisons, the online software SNPSpD (
http://gump.qimr.edu.au/general/daleN/SNPSpD) was used to calculate the correction factor for multiple testing in gene unit [
24]. A
P value of 0.0036 (0.05/14.01; correction factor = 14.01) was used as the criterion of statistical significance in Guangxi population. Association analysis was performed by SNPStats (
http://bioinfo.iconcologia.net/snpstats/start.htm). The power for our genetic association study was calculated using the Power for Genetic Association Analyses (PGA) [
25]. Differences of the protein expression levels detected by IHC between the NPC tissues and non-cancerous nasopharyngeal tissues, and between the rs3928672 GG genotype and A allele (GA + AA genotype) carriers were assessed by a Wilcoxon signed-ranks test and logistic regression analysis, respectively. Unpaired
t test was used to analyze the results of cell proliferation, apoptosis and migration assays. A
P value of 0.05 was used as the criterion of statistical significance, and all statistical tests were two sided. These analyses were performed using SPSS software (Version 17.0; SPSS Inc., USA).
Discussion
In the present study, we systematically evaluated the effect of SNPs in the AGO2 gene on the risk of occurrence or progression of NPC in two case–control populations of Chinese ancestry. We found that the AGO2 rs3928672 was significantly associated with the advanced lymph node metastasis of NPC. The functional experiments demonstrated that the genotype of rs3928672 was significantly associated with the expression of AGO2 in NPC tissues, and AGO2 can play an oncogenic role in the development of NPC by regulating genes related to the tumorigenesis and metastasis. To the best of our knowledge, this is the first report that the genetic polymorphisms of AGO2 may be risk factor for the progression of NPC, and AGO2 acts as an oncogene in NPC.
The polymorphisms in the
AGO2 gene have been used to search for susceptibility alleles of a wide spectrum of cancers (Additional file
1: Table S11). For instance, three SNPs, i.e. rs3864659, rs2292779 and rs11786030, in the
AGO2 gene were shown to be associated with breast cancer in Korea population [
17,
18]. Unfortunately, on the basis of analysis of 855 NPC cases and 1036 controls in the Guangxi population, we did not find a statistically significant association between these SNPs and the susceptibility to NPC. The inconsistent results may be attributed to different molecular mechanisms of carcinogenesis among cancers, small sample size, marginal statistical significance and different ethnicities of study populations. However, by using the tag SNP approach, we identified a novel tag SNP (i.e. rs3928672) that was associated with the risk of lymph node metastasis of NPC in two Chinese subpopulations. Together, our results support the
AGO2 gene as a susceptibility gene for cancers.
In both study populations, the subjects bearing the rs3928672 A allele (GA + AA genotype) had a significantly increased frequency of involvement of lymph node metastasis of NPC compared with ones bearing GG genotype. Moreover, the rs3928672 A allele carriers having higher AGO2 protein expression than the GG genotype carriers in NPC tissues. Given the role of AGO2 in the development of NPC, one might expect individuals who carry the rs3928672 A allele, and thus have increased expression of AGO2 over a lifetime, may have a higher risk of developing lymph node metastasis of NPC after establishment of this malignancy. However, the rs3928672, which is a polymorphism in intron 4 of
AGO2, may not have a functional consequence. It is plausible that some other functional polymorphisms in linkage disequilibrium (LD) with the rs3928672 should be responsible for the allele-specific expression. However, the rs3928672 is in low linkage disequilibrium with other SNPs in both the patients and comparison subjects in the Guangxi population as well as in the HapMap CHB (r
2 < 0.80), indicating that the association of rs3928672 with NPC severity is likely independent. Alternatively, we cannot exclude the possibility that the rs3928672 itself is a functional variant directly affecting the AGO2 production. Introns in eukaryotes fulfill a broad spectrum of functions, such as acting as transposable elements, and are involved in virtually every step of mRNA processing [
29]. Indeed, by computer analysis (F-SNP database,
http://compbio.cs.queensu.ca/F-SNP) we found that the polymorphism rs3928672 maybe influence binding of transcription factor RUNX1 (runt-related transcription factor 1, also called AML-1) and CAP1 [CAP, adenylate cyclase-associated protein 1 (yeast)], since the sequences flanking rs3928672 A allele (5’-AG
TGGT-3’) as a potential binding site for the RUNX1, and the sequences flanking G allele (5’-TCACC
GCT-3’) as a potential binding site for the CAP1, respectively. Therefore, one mechanism by which this could occur is if the risky rs3928672 A allele can influence binding of transcription factor in the intron 4 of
AGO2. Further studies are needed to clarify which polymorphism(s) may possess functional consequence(s) for AGO2, and in turn to provide mechanistic plausibility for the observed association between rs3928672 and involvement of lymph node metastasis of NPC.
The mechanism of how
AGO2 SNPs regulates human susceptibility to cancer is unknown. However,
AGO2 has significant roles controlling the tumorigenesis and progression of several cancers, including tumor invasion and metastasis [
8‐
10,
14,
15]. This processes appears to be through two different molecular mechanisms, RNAi-dependent gene silencing [
8,
30] and RNAi-independent ways such as stabilizing insulator-independent looping [
31], facilitating DNA double-strand break repair [
32], targeting the intragenic long interspersed nuclear element-1 (LINE-1) transcription complexes [
33], and directly regulating downstream gene expression [
9]. Consistent with the previous study, we showed that the AGO2 protein was significantly over-expressed in NPC tissues compared with non-cancerous nasopharyngeal epithelium tissues in the present study. Moreover, functional experiments illustrate that
AGO2 knockdown reduced cell proliferation, induced apoptosis, and inhibited migration of NPC cells. These effects of
AGO2 knockdown were further corroborated by genes with altered expression following
AGO2 knockdown, which were functionally clustered in biological processes related with cell cycle, apoptosis and cell migration. Furthermore, several tumorigenesis and metastasis associated genes with altered expression following
AGO2 knockdown were targets of miRNAs (e.g.,
TPBG [
34],
JUP [
35],
CDKN1A [
36] and
S100A11 [
37] were experimentally validated targets of miR-155 [
38,
39]). Therefore,
AGO2 may promote tumorigenesis and metastasis by regulating miRNAs and their targets in NPC. Taken together, our findings indicated that AGO2 over-expression can contribute to NPC malignant behaviors.
In reviewing the results of this study, one must also keep several potential limits in mind. First, as a hospital-based study, our NPC cases were recruited from the hospital, while the controls were selected from the community population; inherent selection bias cannot be completely excluded. However, by further adjustment and stratification in data analyses, the potential confounding effects might have been minimized. Second, a number of association studies have addressed identifying the genes that may relate to the susceptibility to NPC [
40‐
42]. Most of the results, however, could not be replicated in subsequent studies in other populations. Although the highly significant association between the
AGO2 and lymph node metastasis of NPC was strengthened by our two independent case–control studies, our initial findings should be independently verified in other populations with high incidence rate of NPC, such as other Southern Chinese, Singaporeans, and Taiwanese.
Competing interests
The authors have declared that no competing interests exist.
Authors’ contributions
PL and JM conceived and designed the experiments. PL performed the genotyping, analyzed the data, and drafted the manuscript. JM performed the functional experiments, and analyzed the data. PC and XC analyzed the microarray data. HZ, LY, ZW, XZ, YZ, HC, YL and YL helped to the sample preparation and genotyping. YZ and YH helped to perform functional experiments. YC was responsible for recruitment of Guangxi subjects, phenotype collection and biological sample collection. JB and YZ were responsible for recruitment of Guangdong subjects, phenotype collection and biological sample collection. GZ and FH conceived and designed the experiments, and drafted and revised the manuscript. All authors read and approved the final manuscript.