Esophageal squamous cell carcinoma (ESCC), the major histologic form of esophageal cancer, is a heterogeneous tumor displaying a complex variety of genetic and epigenetic changes. Aberrant RNA editing of adenosine-to-inosine (A-to-I), as it is catalyzed by adenosine deaminases acting on RNA (ADAR), represents a common posttranscriptional modification in certain human diseases. In this study, we investigated the status and role of ADARs and altered A-to-I RNA editing in ESCC tumorigenesis. Among the three ADAR enzymes expressed in human cells, only ADAR1 was overexpressed in primary ESCC tumors. ADAR1 overexpression was due to gene amplification. Patients with ESCC with tumoral overexpression of ADAR1 displayed a poor prognosis. In vitro and in vivo functional assays established that ADAR1 functions as an oncogene during ESCC progression. Differential expression of ADAR1 resulted in altered gene-specific editing activities, as reflected by hyperediting of FLNB and AZIN1 messages in primary ESCC. Notably, the edited form of AZIN1 conferred a gain-of-function phenotype associated with aggressive tumor behavior. Our findings reveal that altered gene-specific A-to-I editing events mediated by ADAR1 drive the development of ESCC, with potential implications in diagnosis, prognosis, and treatment of this disease. Cancer Res; 74(3); 840–51. ©2013 AACR.

Esophageal squamous cell carcinoma (ESCC), the leading cause of cancer death worldwide, is a heterogeneous tumor displaying a complex variety of genetic and epigenetic changes. ESCC is characterized by its remarkable geographic distribution and high-risk areas include Northern China, especially in Henan province, Northern Iran, and South Africa (1). Despite the recent advances in treatment, the benefit of all kind of treatment is not satisfactory. The prognosis of patients with ESCC is still poor and the 5-year overall survival rate is ranging from 20% to 30% (2). Therefore, it is very important to search for biologic markers, which can diagnose cancer as early as possible, estimate reaction to chemotherapy or radiotherapy in those patients with ESCC, and predict overall survival rate of patients undergoing treatment.

In human cancers, aberrant posttranscriptional modifications such as alternative splicing and RNA editing may lead to tumor-specific transcriptome diversity. The best-characterized type of RNA editing found in mammals converts cytosine (C) to uracil (U) and adenosine (A) to inosine (I). In humans, the most common type of RNA editing is conversion of A to I, which is catalyzed by the double-stranded RNA (dsRNA)-specific ADAR (adenosine deaminases acting on RNA) family of protein (3). As inosine preferentially base pairs with cytidine, inosines are interpreted as guanosines by the translation and splicing machineries. Thus, the ADARs may recode transcripts, which results in a proteome that is divergent from the genome (4–7) and therefore, modulates the protein sequence and function of gene products. Because of the diverse impacts of RNA editing on gene expression and function, it is possible that the misregulation of A-to-I RNA editing may play a role in tumorigenesis by either inactivating a tumor suppressor or activating genes that promote tumor progression.

The ADAR gene family includes three members, ADAR1 (also known as ADAR), ADAR2 (ADARB1), and ADAR3 (ADARB2). ADAR1 and ADAR2 are expressed in most tissues; ADAR3 is exclusively detected in brain tissue (8). There are two isoforms of ADAR1, a full-length ADAR1 p150 and an N-terminally truncated ADAR1 p110 (9). The p150 isoform is produced from an IFN-inducible promoter and the p110 form is constitutively expressed, which is initiated from a downstream methionine as the result of the skipping of the exon containing the upstream methionine. Because the p150 isoform of ADAR1 can be induced by IFNs and found in cytoplasm, possibly it exerts its functions to target viruses that replicate in the cytoplasm. However, the ADAR1 p110 isoform exerts its A-to-I modification in the nuclear pre-mRNA. Our recent study has reported that the recoding RNA editing of a gene AZIN1 (antizyme inhibitor 1) is specifically catalyzed by ADAR1, and the hyperediting pattern of AZIN1 gene predisposes to human hepatocellular carcinoma (10). Until now, the roles of RNA editing enzyme ADARs and the edited transcripts of target genes in the development of ESCC have not been studied. Here, we demonstrated that the RNA editing enzyme ADAR1, but not ADAR2 and ADAR3, was significantly overexpressed in primary ESCC tumor compared with their matched nontumor specimens. Moreover, the role of ADAR1 and the altered gene-specific editing pattern were further investigated in clinical specimen, cell models, and mice.

Cell lines

Six Japanese ESCC cell lines (KYSE140, KYSE410, KYSE180, KYSE30, KYSE510, and KYSE520) were obtained from DSMZ, the German Resource Centre for Biological Material (11). A Chinese ESCC cell line HKESC1 was kindly provided by professor Srivastava (Department of Pathology, The University of Hong Kong, Hong Kong, China), and two Chinese ESCC cell lines EC18 and EC109 were kindly provided by professor Tsao (Department of Anatomy, The University of Hong Kong). All nine ESCC cell lines were cultured in RPMI-1640 medium (Gibco BRL) supplemented with 10% FBS (Gibco BRL). All cell lines used in this study were regularly authenticated by morphologic observation and tested for absence of mycoplasma contamination (MycoAlert, Lonza Rockland). The cells were incubated at 37°C in a humidified chamber containing 5% CO2.

Clinical samples

Cohort 1.

A total of 69 paired primary ESCC tumor tissues and their matched nontumorous tissues that were surgically removed, snap-frozen in liquid nitrogen (for protein, RNA, and DNA extraction) were obtained from Linzhou Cancer Hospital (Henan, China) between 2010 and 2011.

Cohort 2.

A total of 180 paired primary ESCC tumor tissues and their matched nontumor tissues that were surgically removed and embedded in a paraffin block [for tissue microarray (TMA) construction] were obtained from Linzhou Cancer Hospital along with clinicopathological summaries between 2001 and 2005.

None of these patients received preoperative chemotherapy and radiotherapy. All clinical samples used in this study were approved by the committees for ethical review of research involving human subjects at Zhengzhou University (Zhengzhou, China), The University of Hong Kong, and National University of Singapore (Singapore).

Analysis of RNA editing

Direct sequencing was performed on PCR products, and the editing frequency was calculated using software ImageJ (http://rsb.info.nih.gov/ij/). The reliability of this method was further verified by cloning of individual sequences. As described previously (12), PCR products were subcloned into the T-easy vector (Promega), and approximately 50 individual plasmids were sequenced for each sample. For each sample, three independent real-time PCR (RT-PCR) reactions were performed.

FISH

The detailed procedure for normal karyotype preparation was described in “Supplementary Materials and Methods.” A BAC clone at 1q21.3 containing the ADAR1 gene (RP11-61L14) was labeled with SpectrumRed (Vysis). The chromosome 1 centromere probe was labeled with SpectrumGreen (Vysis). FISH reaction was performed according to the method described previously (13).

cDNA synthesis and quantitative real-time PCR

As described previously (10), the total RNA was isolated and the equal amounts of cDNA were synthesized using the Advantage RT-for-PCR kit (Clontech) and used for quantitative PCR (qPCR) analysis. The sequences of primers are included in “Supplementary Materials and Methods.”

Cell proliferation assay

Cells were seeded in 96-well plate at a density of 0.5 to 1 × 103 cells per well. The cell growth rate was measured using Cell Proliferation XTT Kit (Roche Diagnostic) according to the manufacturer's instruction. Three independent experiments done in triplicate were performed.

Focus formation and colony formation in soft agar

Briefly, 1 × 103 cells were seeded in a 6-well plate. After culture for 12 days, surviving colonies (>50 cells per colony) were counted and stained with crystal violet (Sigma-Aldrich). Triplicate independent experiments were performed and the data were expressed as the mean ± SD of triplicate wells within the same experiment. For soft agar assay, 2 × 103 cells in 0.4% low-melting agarose (Lonza Rockland) were placed on the top of the bottom layer of 0.6% low-melting agarose in a 6-well plate. After 2 to 4 weeks, surviving colonies (>80 cells/colony) were counted.

Cell migration assay

The Transwell cell migration assay was performed using Bio-coat cell migration chambers (BD Biosciences) containing polyethylene terephthalate membranes (PET) of 8 μm pore size according to the manufacturer's instructions. Briefly, 1 to 2 × 105 cells in FBS-free RPMI were added. RPMI supplemented with 10% FBS was added to the bottom chamber as a chemoattractant. After 24 to 36 hours, the number of cells that had migrated through the filter pores was stained with crystal violet (Sigma-Aldrich), counted, and imaged using SPOT imaging software (Nikon).

Matrigel invasion assay

We performed invasion assays using 24-well BioCoat Matrigel Invasion Chambers (BD Biosciences) according to the manufacturer's instructions. Briefly, 1 to 2 × 105 cells in FBS-free RPMI were added to the top chamber, and 10% FBS in RPMI was added to the bottom chamber as a chemoattractant. After 36 to 48 hours of incubation, cells that invaded the Matrigel were fixed and stained with crystal violet (Sigma-Aldrich). The number of cells was counted and imaged using SPOT imaging software (Nikon).

In vivo tumorigenicity assay

We subcutaneously injected approximately 2 × 106 of EC109-CTL or EC109-AR1 cells into the left or right flank of 4- to 5-week-old male severe combined immunodeficient (SCID) mice (n = 5), respectively. To compare the tumorigenic abilities of the wild-type or edited AZIN1 gene, we subcutaneously injected approximately 2 × 106 of EC109-LacZ, EC109-wt/AZI, or EC109-edt/AZI cells into the right flank of SCID mice (n = 6). We monitored tumor formation in the SCID mice over a 4-week period and calculated the tumor volume weekly by the formula V (volume) = 0.5 × L (length) × W (width) × W. All animal experiments were approved by and performed in accordance with the Institutional Animal Care and Use Committees of National University of Singapore.

Antibodies and Western blot analysis

Mouse anti-ADAR1, anti-AZIN1, and anti-β-actin antibodies were purchased from Abcam. The mouse anti-ADAR2 and anti-GAPDH antibodies were purchased from Sigma-Aldrich and Santa Cruz Biotechnology. Protein lysates were quantified and resolved on a SDS-PAGE gel, transferred onto a polyvinylidene difluoride membrane (Millipore), and immunoblotted with a primary antibody, followed by incubation with a secondary antibody. The blots were visualized by enhanced chemiluminescence (GE Healthcare).

Immunohistochemical staining

The TMA blocks were sectioned (5 mm thick) for immunohistochemical (IHC) staining. Briefly, sections were deparaffinized and rehydrated. The endogenous peroxidase activity was blocked with 3% hydrogen peroxide for 10 minutes. For antigen retrieval, the slides were immersed in 10 mmol/L citrate buffer (pH 6.0) and boiled for 15 minutes in a microwave oven. Nonspecific binding was blocked with 5% normal goat serum for 10 minutes. The slides were incubated in a 1:100 dilution of anti-ADAR1 (Abcam) at 4°C overnight in a humidified chamber. The slides were then sequentially incubated with biotinylated goat anti-mouse immunoglobulin G (IgG; 1:100 dilution, Santa Cruz Biotechnology) for 30 minutes at room temperature and streptavidin-peroxidase conjugate for 30 minutes at room temperature. Isotope-matched human IgG was used in each case as a negative control. Finally, the 3,3′-diaminobenzidine (DAB) Substrate Kit (Dako Ltd.) was used for color development followed by Mayer's hematoxylin counterstaining. On the basis of staining intensities, the ADAR1 immunoreactivity was scored as negative (0; total absence of staining), weak expression (1; faint staining in <50%, or moderate staining in <25% of tumor cells), moderate expression (2; moderate staining in > = 25% to <75%, or strong staining in <25% of tumor cells), and strong expression (3; moderate staining in > = 75%, or strong staining in > = 25% of tumor cells).

Statistical analysis

Unless otherwise indicated, the data are presented as the mean ± SD of three independent experiments. The SPSS statistical package for Windows (version 16; SPSS) was used to perform the data analyses. The ADAR1 or ADAR2 expression levels in any two groups of clinical samples (e.g., ESCC tumors and matched nontumor tissues) were compared using a Wilcoxon signed rank test. Kaplan–Meier plots and log-rank tests were used for overall survival analysis. For the TMA analysis, which was based on the scores of IHC staining, ADAR1 expression levels in the primary ESCC tissues and their matched nontumor tissues were compared using the Wilcoxon signed rank test. A paired Student t test was used to compare the editing levels as well as the total mRNA levels of FLNB or AZIN1 in ESCC and their matched nontumor specimens of patients. Spearman correlation coefficients were used to evaluate the positive correlation between the expression of ADAR1 and the editing level of AZIN1 or FLNB in clinical samples. An unpaired two-tailed Student t test was used to compare the number of colonies, the number of migrative and invasive cells, and tumor volume between any two preselected groups. P < 0.05 was considered to be statistically significant.

RNA editing enzyme ADAR1 is significantly upregulated in primary ESCCs and its clinical implications

A-to-I RNA editing is a posttranscriptional modification in stem-loop structures within precursor mRNA, which is catalyzed by dsRNA-specific ADAR enzymes (14). To investigate the expression profiles of the three RNA editing enzymes in ESCC specimens, we examined the expression levels of ADAR1, ADAR2, and ADAR3 in 69 pairs of primary ESCC tumor and their corresponding nontumor tissues that were obtained from Linzhou Cancer Hospital (cohort 1). As detected by the qRT-PCR, only ADAR1 was significantly upregulated in ESCC samples (P = 0.0045) and approximately 48 of 69 (69.57%) ESCC specimens demonstrated the higher expression of ADAR1 than their matched nontumor specimens (Fig. 1A, left). There was no significant difference in ADAR2 expression between primary ESCCs and their matched nontumor specimens (P = 0.1465; Fig. 1A, right). In addition, ADAR3 was undetectable in all of the samples (data not shown). To confirm our findings, Western blot analysis of ADAR1 and ADAR2 expression was performed in the paired ESCC and nontumor specimens of 35 randomly selected ESCC cases from the cohort 1. Consistently, the upregulation of ADAR1 protein (particularly the p110 isoform) in tumors was observed in 19 of 35 (54.29%) ESCCs, and there was no obvious difference in ADAR2 protein expression between ESCC and their matched nontumor specimens (Fig. 1B).

Figure 1.

ADAR1 is significantly overexpressed in primary ESCC samples and its clinical implication. A, box plots represent the relative ADAR1 (left) and ADAR2 (right) expression levels in 69 matched pairs of ESCC and nontumor specimens in cohort 1. The data are presented as box plots with the median (horizontal line), 25% to 75% (box), and 5% to 95% (error bar) percentiles for each group. B, Western blot analyses of ADAR1 and ADAR2 expression levels in six paired ESCC and nontumor specimens in cohort 1. GAPDH was used as a loading control. C, example of the ADAR1 expression level detected in a matched pair of primary ESCC and nontumor tissue in cohort 2. The boxed regions are magnified and displayed in the bottom panels. Scale bar, 200 μm. D, Kaplan–Meier plots for the overall survival rate of patients with (n = 90, red line) or without (n = 46, blue line) the tumoral overexpression of ADAR1. E, FISH analysis of the ADAR1 gene (red signal) and the control chromosome 1 centromere probe (green signal) specifically hybridized to the chromosome 1 (indicated by white arrow) of normal human karyotype (left). A representative example of ADAR1 gene amplification (red signals) in a matched pair of primary ESCC and nontumor tissue (middle and right panels). Scale bar, 500 μm.

Figure 1.

ADAR1 is significantly overexpressed in primary ESCC samples and its clinical implication. A, box plots represent the relative ADAR1 (left) and ADAR2 (right) expression levels in 69 matched pairs of ESCC and nontumor specimens in cohort 1. The data are presented as box plots with the median (horizontal line), 25% to 75% (box), and 5% to 95% (error bar) percentiles for each group. B, Western blot analyses of ADAR1 and ADAR2 expression levels in six paired ESCC and nontumor specimens in cohort 1. GAPDH was used as a loading control. C, example of the ADAR1 expression level detected in a matched pair of primary ESCC and nontumor tissue in cohort 2. The boxed regions are magnified and displayed in the bottom panels. Scale bar, 200 μm. D, Kaplan–Meier plots for the overall survival rate of patients with (n = 90, red line) or without (n = 46, blue line) the tumoral overexpression of ADAR1. E, FISH analysis of the ADAR1 gene (red signal) and the control chromosome 1 centromere probe (green signal) specifically hybridized to the chromosome 1 (indicated by white arrow) of normal human karyotype (left). A representative example of ADAR1 gene amplification (red signals) in a matched pair of primary ESCC and nontumor tissue (middle and right panels). Scale bar, 500 μm.

Close modal

To investigate the clinical implication of ADAR1 during the development of ESCC, we constructed a panel of TMAs consisting of 180 matched pairs of primary ESCC and nontumor specimens that were collected from Linzhou Cancer Hospital (cohort 2). Informative results of IHC staining were observed in 136 matched pairs of ESCC and nontumor specimens. Noninformative samples included lost samples, inappropriately stained samples, and samples with too few cells; such were not used as valid data. By performing IHC staining, we observed the differential nuclear expression of ADAR1 between the primary ESCC tumor and their matched nontumor tissues (Fig. 1C). A detailed analysis of the IHC data revealed that the ADAR1 was overexpressed in 66.18% (90/136) of the informative ESCC tumor tissues (P < 0.001, Wilcoxon signed rank test; Fig. 1C and Supplementary Table S1). On the basis of the TMA analysis of ADAR1, the patients with tumoral overexpression of ADAR1 demonstrated the shorter overall survival time than the patients without overexpression (log rank: 5.738, P = 0.017; Fig. 1D). In the univariate Cox analyses, the statistically significant predictors for a patient's overall survival were cell differentiation (P = 0.036), tumor invasion (P = 0.003), clinical stage (P = 0.001), and the tumoral overexpression of ADAR1 (P = 0.020; Table 1). In the multivariate Cox analyses, cell differentiation (P = 0.029), clinical stage (P = 0.004), and the tumoral overexpression of ADAR1 (P = 0.005) were shown to be the independent prognostic factors for the overall survival (Table 1). Altogether, we conclude that the tumoral overexpression of ADAR1 predicts a poor prognosis.

Table 1.

Cox proportional hazard regression analyses for overall survival

Univariate analysisMultivariate analysis
Clinical featuresHR (95% CI)PHR (95% CI)P
Age 1.160 (0.716–1.878) 0.547 — — 
Gender 1.364 (0.857–2.170) 0.190 — — 
LN metastasis 1.141 (0.601–2.164) 0.688 — — 
Differentiation 0.759 (0.587–0.982) 0.036a 0.749 (0.577–0.971) 0.029a 
Clinical stage 10.227 (2.506–41.733) 0.001a 8.534 (1.949–37.358) 0.004a 
Tumor invasion 2.995 (1.435–6.248) 0.003a 1.599 (0.731–3.501) 0.240 
ADAR1 overexpression overregulation 0.577 (0.364–0.916) 0.020a 0.513 (0.321–0.822) 0.005a 
Univariate analysisMultivariate analysis
Clinical featuresHR (95% CI)PHR (95% CI)P
Age 1.160 (0.716–1.878) 0.547 — — 
Gender 1.364 (0.857–2.170) 0.190 — — 
LN metastasis 1.141 (0.601–2.164) 0.688 — — 
Differentiation 0.759 (0.587–0.982) 0.036a 0.749 (0.577–0.971) 0.029a 
Clinical stage 10.227 (2.506–41.733) 0.001a 8.534 (1.949–37.358) 0.004a 
Tumor invasion 2.995 (1.435–6.248) 0.003a 1.599 (0.731–3.501) 0.240 
ADAR1 overexpression overregulation 0.577 (0.364–0.916) 0.020a 0.513 (0.321–0.822) 0.005a 

Abbreviation: CI, confidence interval.

aStatistical significance (P < 0.05) is shown in bold.

As described previously (15), chromosome 1q is one of the frequent regions with chromosomal abnormalities during the development of ESCC. To investigate whether the tumoral overexpression of ADAR1 is due to the genomic amplification of the ADAR1 gene, FISH was used to detect DNA copy number of ADAR1 in the same TMA as described above. Using a bacterial artificial chromosome (BAC) probe containing ADAR1 gene, the gain of ADAR1 copy number was detected in 39 out of 55 (70.9%) ESCC cases (Fig. 1E), and there was a significantly positive correlation between the overexpression of ADAR1 and the genomic amplification of ADAR1 gene in ESCC specimens (P < 0.001, Spearman r = 0.582; Supplementary Table S2).

ADAR1 functions as an oncogene during ESCC progression

As upstream manipulators of the A-to-I RNA editing of the nuclear pre-mRNA, we are particularly interested in the functional role of ADAR1 (the p110 isoform) during ESCC progression. For this purpose, we introduced ADAR1 (p110) construct into two ESCC cell lines (KYSE180 and EC109) expressing the relative low endogenous ADAR1 among nine ESCC cell lines using a lentiviral system (Fig. 2A and Supplementary Fig. S1). As detected by in vitro functional assays, cells transduced with the ADAR1 lentivirus (180-AR1 and EC109-AR1) had the accelerated growth rates and higher frequencies of focus formation and colony formation in soft agar than cells transduced with the control LacZ lentivirus (180-CTL and EC109-CTL; Fig. 2B–D). Moreover, 180-AR1 and EC109-AR1 cells demonstrated the increased migrative and invasive capabilities compared with 180-CTL and EC109-CTL cells, respectively (Fig. 2E). Xenograft studies in mice demonstrated that the growth rate of tumors derived from EC109-AR1 cells was markedly higher than those derived from EC109-CTL cells (Fig. 2F).

Figure 2.

ADAR1 functions as an oncogene during ESCC progression. A, Western blot analyses showing expression of ADAR1 and ADAR2 proteins in the indicated cell lines. β-actin was the loading control. B, cell growth rates of the indicated cell lines were compared by XTT assays. The results are expressed as the mean ± SD of triplicate wells within the same experiment (*, P < 0.05; **, P < 0.01; ***, P < 0.001, unpaired, two-tailed Student t test). C, quantification of foci formation induced by the indicated stable cell lines. Triplicate independent experiments were performed and the data were expressed as the mean ± SD of triplicate wells within the same experiment (*, P < 0.05; ***, P < 0.001, unpaired, two-tailed Student t test). Scale bar,0.5 cm. D, quantification of colonies (formed in soft agar) that were induced by the indicated cell lines. Triplicate independent experiments were performed and the data were expressed as the mean ± SD of triplicate wells within the same experiment (**, P < 0.01, unpaired, two-tailed Student t test). Scale bar, 200 μm. E, quantification of cells from the indicated cells that migrated through the PET membrane or invaded through the Matrigel-coated membrane. (*, P < 0.05; ***, P < 0.001, unpaired, two-tailed Student t test). Scale bar, 200 μm. F, representative images of the tumors derived from the indicated cell lines at 4 weeks postinjection (left). Right, growth curves of tumors over a period of 4 weeks. Data are presented as the mean ± SD (n = 5, *, P < 0.05, unpaired, two-tailed Student t test). Scale bar, 1 cm.

Figure 2.

ADAR1 functions as an oncogene during ESCC progression. A, Western blot analyses showing expression of ADAR1 and ADAR2 proteins in the indicated cell lines. β-actin was the loading control. B, cell growth rates of the indicated cell lines were compared by XTT assays. The results are expressed as the mean ± SD of triplicate wells within the same experiment (*, P < 0.05; **, P < 0.01; ***, P < 0.001, unpaired, two-tailed Student t test). C, quantification of foci formation induced by the indicated stable cell lines. Triplicate independent experiments were performed and the data were expressed as the mean ± SD of triplicate wells within the same experiment (*, P < 0.05; ***, P < 0.001, unpaired, two-tailed Student t test). Scale bar,0.5 cm. D, quantification of colonies (formed in soft agar) that were induced by the indicated cell lines. Triplicate independent experiments were performed and the data were expressed as the mean ± SD of triplicate wells within the same experiment (**, P < 0.01, unpaired, two-tailed Student t test). Scale bar, 200 μm. E, quantification of cells from the indicated cells that migrated through the PET membrane or invaded through the Matrigel-coated membrane. (*, P < 0.05; ***, P < 0.001, unpaired, two-tailed Student t test). Scale bar, 200 μm. F, representative images of the tumors derived from the indicated cell lines at 4 weeks postinjection (left). Right, growth curves of tumors over a period of 4 weeks. Data are presented as the mean ± SD (n = 5, *, P < 0.05, unpaired, two-tailed Student t test). Scale bar, 1 cm.

Close modal

To further confirm our findings, RNA interference (RNAi) was used to silence the expression of ADAR1 in an ESCC cell line KYSE510 with the highest endogenous ADAR1 expression (Supplementary Fig. S1). Western blot analysis of ADAR1 demonstrated that the expression of ADAR1 could be effectively silenced by the introduction of two specific short hairpin RNAs (shRNA) against ADAR1 gene (Fig. 3A). Compared with cells transfected with the control scramble shRNA (510-CTL), cells transfected with two shRNAs targeting ADAR1 (510-shAR1 #5 and 510-shAR1 #7) were found to be less tumorigenic, as manifested by the decreased cell growth rates, the decreased frequencies of focus formation, and the reduced migrative and invasive capabilities (Fig. 3B–D). All these data suggest that ADAR1 functions as an oncogene during ESCC progression.

Figure 3.

Silencing ADAR1 by RNAi inhibits its tumorigenicity. A, Western blot analyses showing expression of ADAR1 and ADAR2 proteins in the indicated cell lines. β-actin was the loading control. B, cell growth rates of the indicated cell lines were compared by XTT assays. The results are expressed as the mean ± SD of triplicate wells within the same experiment (*, P < 0.05; **, P < 0.01; ***, P < 0.001, unpaired, two-tailed Student t test). C, quantification of foci formation induced by the indicated stable cell lines. Triplicate independent experiments were performed and the data were expressed as described above (**, P < 0.01; ***, P < 0.001, unpaired, two-tailed Student t test). Scale bar, 0.5 cm. D, quantification of cells from the indicated cells that migrated through the PET-membrane or invaded through the Matrigel-coated membrane. (***, P < 0.001, unpaired, two-tailed Student t test). Scale bar, 100 μm.

Figure 3.

Silencing ADAR1 by RNAi inhibits its tumorigenicity. A, Western blot analyses showing expression of ADAR1 and ADAR2 proteins in the indicated cell lines. β-actin was the loading control. B, cell growth rates of the indicated cell lines were compared by XTT assays. The results are expressed as the mean ± SD of triplicate wells within the same experiment (*, P < 0.05; **, P < 0.01; ***, P < 0.001, unpaired, two-tailed Student t test). C, quantification of foci formation induced by the indicated stable cell lines. Triplicate independent experiments were performed and the data were expressed as described above (**, P < 0.01; ***, P < 0.001, unpaired, two-tailed Student t test). Scale bar, 0.5 cm. D, quantification of cells from the indicated cells that migrated through the PET-membrane or invaded through the Matrigel-coated membrane. (***, P < 0.001, unpaired, two-tailed Student t test). Scale bar, 100 μm.

Close modal

Tumoral overexpression of ADAR1 contributes to the hyperediting patterns of editing targets

As reported previously, ADAR1 protein (the p110 isoform) has a number of well-known target transcripts, such as serotonin receptor subunit 2C (5-HT2CR), filamin B (FLNB), antizyme inhibitor 1 (AZIN1), and bladder cancer-associated protein (BLCAP; refs. 10, 12, 16–18).

In this study, we examined the editing levels of two representative targets, AZIN1 and FLNB, in primary ESCC tumor and their matched nontumor specimens of 69 patients with ESCC (cohort 1) as well as nine ESCC cell lines. Among nine ESCC cell lines, there was a positive correlation between the relative quantification value of ADAR1 and the editing frequency of AZIN1 or FLNB (AZIN1: Spearman r = 0.9139, P = 0.0013; FLNB: spearman r = 0.75, P = 0.025; Fig. 4A and Supplementary Fig. S2). In clinical samples, the editing levels of AZIN1 and FLNB in ESCC tumors were significantly higher than those in nontumor specimens (AZIN1: P < 0.0001; FLNB: P < 0.0001, paired Student t test; Fig. 4B), suggesting that the hyperediting patterns of AZIN1 and FLNB might play a role in ESCC progression. We also determined the total mRNA levels of AZIN1 and FLNB in the same clinical cases as described above. The mRNA levels of AZIN1 and FLNB were significantly increased in ESCC tumors than their matched nontumor specimens (AZIN1: P = 0.043; FLNB: P = 0.023, paired Student t test; Supplementary Fig. S3), indicating that the hyperediting phenotypes of AZIN1 and FLNB in tumors would not be affected by the low endogenous transcript levels. Furthermore, we defined the fold-change of ADAR1 in ESCC cases with the ratio of the relative quantification values of tumors to that of the corresponding nontumor specimens. In addition, the fold-change of AZIN1 or FLNB editing level in ESCC cases was characterized by the ratio of the editing frequency of tumors to that of their matched nontumor specimens. As expected, there was a positive correlation between the editing level of AZIN1 or FLNB and the expression level of ADAR1 in ESCC cases (AZIN1: Spearman r = 0.3689, P = 0.0018; FLNB: Spearman r = 0.4186, P = 0.0003; Fig. 4C). To provide direct evidence that ADAR1 regulates AZIN1 and FLNB editing, the introduction of the ADAR1 p110 expression constructs into EC109 and KYSE180 cell lines led to the increased editing frequencies of AZIN1 and FLNB transcripts (Fig. 4D and E). Consistently, silencing ADAR1 by two shRNAs targeting ADAR1 gene (shAR1 #5 and #7) in KYSE510 cells resulted in the reduced editing levels of AZIN1 and FLNB (Fig. 4F). Together, these results suggest that the upregulation of ADAR1 in ESCC tumors contributes to the gene-specific hyperediting pattern.

Figure 4.

The hyperediting patterns of AZIN1 and FLNB transcripts induced by the upregulation of ADAR1 in ESCC tumors. A, line-bar chart showing the editing levels of AZIN1 and FLNB as well as the expression level of ADAR1 in nine ESCC cell lines. B, the AZIN1 (left) and FLNB (right) editing levels in 69 paired ESCC and matched nontumor specimens in cohort 1 (paired Student t test). C, correlation between the expression level of ADAR1 and the editing level of AZIN1 (left) or FLNB (right) in 69 paired ESCC and matched nontumor specimens in cohort 1. D and E, sequence chromatograms of the AZIN1 and FLNB transcripts in EC109 (D) or KYSE180 (E) cells that transiently transduced with an ADAR1 p110 lentivirus (+ADAR1 p110) or LacZ control lentivirus (+control). The percentages of edited AZIN1 and FLNB transcripts were detected as described in Materials and Methods. Arrow, the editing position. F, sequence chromatograms of the AZIN1 and FLNB transcripts in KYSE510 cells (E) that were transiently transfected with two shRNAs against ADAR1 gene (shAR1 #5 and shAR1 #7) or the control shRNA (control). Arrow, the editing position.

Figure 4.

The hyperediting patterns of AZIN1 and FLNB transcripts induced by the upregulation of ADAR1 in ESCC tumors. A, line-bar chart showing the editing levels of AZIN1 and FLNB as well as the expression level of ADAR1 in nine ESCC cell lines. B, the AZIN1 (left) and FLNB (right) editing levels in 69 paired ESCC and matched nontumor specimens in cohort 1 (paired Student t test). C, correlation between the expression level of ADAR1 and the editing level of AZIN1 (left) or FLNB (right) in 69 paired ESCC and matched nontumor specimens in cohort 1. D and E, sequence chromatograms of the AZIN1 and FLNB transcripts in EC109 (D) or KYSE180 (E) cells that transiently transduced with an ADAR1 p110 lentivirus (+ADAR1 p110) or LacZ control lentivirus (+control). The percentages of edited AZIN1 and FLNB transcripts were detected as described in Materials and Methods. Arrow, the editing position. F, sequence chromatograms of the AZIN1 and FLNB transcripts in KYSE510 cells (E) that were transiently transfected with two shRNAs against ADAR1 gene (shAR1 #5 and shAR1 #7) or the control shRNA (control). Arrow, the editing position.

Close modal

The functional alteration of AZIN1 transcript as a result of the A-to-I RNA editing during ESCC progression

As a specific RNA editing target of ADAR1 protein, AZIN1 gene encodes a protein that undergoes an amino acid substitution from serine (Ser) to glycine (Gly) at residue 367 (10). In this study, the functional difference between the wild-type and the edited AZIN1 protein was further studied. For this purpose, we introduced V5-tagged wild-type AZIN1 (wt/AZI) or edited AZIN1 (edt/AZI) expression constructs into two ESCC cell lines (KYSE180 and EC109; Supplementary Fig. S4A). Cells transduced with the edt/AZI lentivirus (180-edt/AZI and EC109-edt/AZI) with 100% of their AZIN1 transcripts edited (Fig. 5A) had accelerated growth rates and higher frequencies of focus and colony formation in soft agar than cells transduced with the wt/AZI (180-wt/AZI and EC109-wt/AZI) or control LacZ lentivirus (180-LacZ and EC109-LacZ; Fig. 5B–D). Moreover, 180-edt/AZI and EC109-edt/AZI cells had the increased migrative and invasive capabilities (Fig. 5E and Supplementary Fig. S4B). Xenograft studies in mice demonstrated that tumors derived from EC109-edt/AZI cells grew significantly faster than those derived from EC109-wt/AZI and EC109-LacZ cells (Fig. 5F).

Figure 5.

The edited AZIN1 confers more aggressive tumorigenic phenotypes during ESCC progression. A, sequence chromatograms of the AZIN1 transcript in the indicated cell lines. Arrow, the editing position. B, cell growth rates of the indicated cell lines were compared by XTT assays. The results are expressed as described above (*, P < 0.05; ***, P < 0.001). C, quantification of foci formation induced by the indicated stable cell lines. Triplicate independent experiments were performed and the data were expressed as described above (**, P < 0.01; ***, P < 0.001). Scale bar, 0.5 cm. D, quantification of colonies (formed in soft agar) that were induced by the indicated cell lines. Triplicate independent experiments were performed and the data were expressed as described above (*, P < 0.05; **, P < 0.01). Scale bar, 200 μm. E, quantification of cells from the indicated cells that migrated through the PET-membrane or invaded through the Matrigel-coated membrane. (*, P < 0.05; **, P < 0.01; ***, P < 0.001, unpaired, two-tailed Student t test). Scale bar, 200 μm. F, growth curves of tumors derived from the indicated cell lines over a period of 4 weeks. Data are presented as the mean ± SD (n = 6, *, P < 0.05; ***, P < 0.001; unpaired, two-tailed Student t test).

Figure 5.

The edited AZIN1 confers more aggressive tumorigenic phenotypes during ESCC progression. A, sequence chromatograms of the AZIN1 transcript in the indicated cell lines. Arrow, the editing position. B, cell growth rates of the indicated cell lines were compared by XTT assays. The results are expressed as described above (*, P < 0.05; ***, P < 0.001). C, quantification of foci formation induced by the indicated stable cell lines. Triplicate independent experiments were performed and the data were expressed as described above (**, P < 0.01; ***, P < 0.001). Scale bar, 0.5 cm. D, quantification of colonies (formed in soft agar) that were induced by the indicated cell lines. Triplicate independent experiments were performed and the data were expressed as described above (*, P < 0.05; **, P < 0.01). Scale bar, 200 μm. E, quantification of cells from the indicated cells that migrated through the PET-membrane or invaded through the Matrigel-coated membrane. (*, P < 0.05; **, P < 0.01; ***, P < 0.001, unpaired, two-tailed Student t test). Scale bar, 200 μm. F, growth curves of tumors derived from the indicated cell lines over a period of 4 weeks. Data are presented as the mean ± SD (n = 6, *, P < 0.05; ***, P < 0.001; unpaired, two-tailed Student t test).

Close modal

All these data indicate that the edited AZIN1 confers “gain-of-function” phenotypes and promotes the tumorigenicity during ESCC progression, suggesting that AZIN1 is likely to be one of the downstream editing targets that are responsible for the oncogenic function of ADAR1.

RNA editing is broadly defined as the posttranscriptional process that changes the sequence of primary RNA transcripts. It is becoming apparent that A-to-I editing is not a rare phenotype, but instead it is the most common type of RNA editing in mammalian transcriptome. ADARs, the enzymes responsible for the conversion of A-to-I in dsRNA, were first noticed as cellular RNA unwinding activity (19, 20), as they lead to destabilization of RNA duplexes by introducing I.U mismatch. ADARs exhibit strict tissue-specific and environment-dependent expression patterns (21, 22). There are three orthologs: ADAR1 and ADAR2 occur in most animals, whereas ADAR3 is vertebrate and brain specific. The ADAR1 p150 isoform is presumably responsible for the A-to-I editing of viral RNAs produced by viruses (14, 23), but not of the nuclear pre-mRNAs (24). It has been reported all the three editing enzymes ADAR1, ADAR2, and ADAR3 were downregualted in brain tumors. Overepxression of ADAR1 and ADAR2 in the U87 glioblastoma multiforme cell line resulted in a decreased proliferation rate, suggesting that the reduced A-to-I editing in brain tumors is involved in the pathogenesis of cancer (18). On the contrary, ADAR1 or/and ADAR2 were also found to be upregulated in tumor tissues, such as prostate cancer and breast cancer (25). In this study, we investigated the expression profiles of all three RNA editing enzymes in ESCC clinical samples. ADAR1 and ADAR2 were abundantly expressed in ESCC samples, but ADAR3 was undetectable in all samples. Because of the genomic amplification of the ADAR1 gene, which is mapped to chromosome 1q21, ADAR1 was the only RNA editing enzyme found to be significantly upregulated in primary ESCC tumors compared with their matched nontumor specimens in two individual cohorts (cohort 1 and 2). Clinically, the tumoral overexpression of ADAR1 was correlated with the shorter overall survival time of patients with ESCC, which could be used independently for the prediction of a poor prognosis. Consistently, our functional assays have indicated that ADAR1 could induce tumorigenic phenotypes in cell models and animals.

ADARs can change the structure of RNA by changing an Adenosine Uracil (AU) base pair to an Inosine Uracil (IU) mismatch. Conceivably, ADARs could affect any biologic process that is associated with the sequence- or structure-specific interaction of RNA (14). So far, ADARs have been shown to alter the protein coding, create or delete splice sites, and modulate transcript stability. The A-to-I editing can be very specific, leading to deamination of select adenosine residues, or it can be almost random and lead to nonselective conversion of many adenosines. For long dsRNA (<100 bp) within 3′ untranslational region (3′-UTR), many adenosine residues are edited promiscuously, leading to approximately 50% of adenosines being converted to inosines. However, in terms of A-to-I editing within protein-coding sequences, it is highly selective, and an imperfect fold-back dsRNA structure is formed between the exon sequence surrounding the editing site(s) and a downstream intronic complementary sequence termed editing-site complementary sequence (8). Until now, only a few recoding RNA editing events (e.g., Q/R site editing in the glutamate receptor) have been verified. The editing events within 3′UTRs can affect transcript stability via affecting microRNA targeting or the nuclear retention of transcripts (26–29). However, the editing targets within coding region will cause amino acid change and affect protein function rather than protein level, suggesting that the recoding editing events are of high potentials to play a role in tumorigenesis by either inactivating a tumor suppressor or activating genes that promote tumor progression. As reported previously, two recoding editing events at codon 2269 (Met → Val) of the FLNB gene and codon 367 (Ser → Gly) of the AZIN1 gene are catalyzed by ADAR1 protein (10, 12). In this study, we proved that ADAR1 was responsible for the editing of FLNB and AZIN1 in ESCC cells and clinical samples, suggesting that as a result of the differential expression of ADAR1 in ESCC tumors, the editing alterations in the specific genes may be responsible for the ADAR1-induced malignant phenotypes during ESCC progression. Therefore, we selected one of the recoding editing targets, AZIN1, for further study. Intriguingly, we found the editing frequency of AZIN1 was significantly increased in primary ESCC tumors compared with their matched nontumor specimen, suggesting there is a hyperediting phenotype of the AZIN1 editing event during ESCC progression. AZIN1, termed antizyme inhibitor 1, blocks the effects of antizyme on omithine decarboxylase (ODC). This protein has substantial similarity to ODC itself but has no ODC activity (30). It binds to antizyme more tightly than ODC and thus releases ODC from the antizyme–ODC complex (30, 31). Although the physiologic importance of AZIN1 is not yet fully understood, a strong case can be made for it to be included as a component of the polyamine pathway. Downregulation of antizyme inhibitor in lung cancer cells reduced ODC levels and led to growth inhibition (32). Also, it has been reported that overexpression of AZIN1 in NIH3T3 fibroblasts provides growth advantage through neutralization of antizyme functions (33). In the present study, we demonstrated that although the wild-type AZIN1 could promote the tumorigenicity, the edited AZIN1 conferred “gain-of-function” phenotypes that were manifested by more aggressive behaviors during ESCC progression, which has also been found in human hepatocellular carcinoma as reported recently (10). All these findings suggest that the ADAR1-induced editing alteration may play a pivotal role in human solid tumors and the precise regulation of the expression level of ADARs is essential for accurate editing and the altered expression of ADARs could be at the origin of cell transformation.

The characterization of editing events is a necessary step toward fully understanding the function and regulation of transcriptome, and investigating the connection between RNA editing and cancer progression is only the initial step in this research. The existence of RNA editing, as opposed to hard-wired genomic mutation, indicates that it can be spatiotemporally controlled. Indeed, there is a great excitement that we could develop potential therapeutic approaches for targeting this epigenetic process. Here, we propose two feasible approaches: (i) modulating the expression of RNA editing enzyme ADARs by overexpressing or silencing ADARs and (ii) reinstating a specific hyperedited or hype-edited transcript (34). In this study, we reported that one of the RNA editing enzyme ADAR1 was overexpressed in ESCC tumors, which can be used as an independent factor for prognosis prediction. As a result of the tumoral overexpression of ADAR1, the edited AZIN1 transcript is more abundant in tumors, leading to “gain-of-function” phenotypes during ESCC progression. Therefore, we speculate that monitoring the expression level of ADAR1 or the editing activities of key editing targets represent a useful biomarker for the detection of disorders in ESCC and our study may also provide the a key biologic process for rationale drug development.

No potential conflicts of interest were disclosed.

Conception and design: Y.-R. Qin, J.-J. Qiao, T.H.M. Chan, X.-Y. Guan, L. Chen

Development of methodology: L. Chen

Acquisition of data (provided animals, acquired and managed patients, provided facilities, etc.): Y.-R. Qin, J.-J. Qiao, T.H.M. Chan, Y.-H. Zhu, F.-F. Li, H. Liu, Y. Li, L. Chen

Analysis and interpretation of data (e.g., statistical analysis, biostatistics, computational analysis): Y.-R. Qin, J.-J. Qiao, T.H.M. Chan, L. Chen

Writing, review, and/or revision of the manuscript: Y.-R. Qin, J.-J. Qiao, T.H.M. Chan, L. Chen

Administrative, technical, or material support (i.e., reporting or organizing data, constructing databases): Y.-R. Qin, J.-J. Qiao, J. Fei, L. Chen

Study supervision: Y.-R. Qin, J.-J. Qiao, X.-Y. Guan, L. Chen

This work was supported by the Singapore Translational Research (STaR) Award (NMRC/STaR/0001/2008), the “Hundred Talents Program” at Sun Yat-Sen University (Guangzhou, China; 85000-3171311), the National Natural Science Foundation of China (81150010 and 81272227), and the National Key Sci-Tech Special Project of Infectious Diseases (2012ZX10002-013).

The costs of publication of this article were defrayed in part by the payment of page charges. This article must therefore be hereby marked advertisement in accordance with 18 U.S.C. Section 1734 solely to indicate this fact.

1.
Parkin
DM
,
Bray
F
,
Ferlay
J
,
Pisani
P
. 
Global cancer statistics, 2002
.
CA Cancer J Clin
2005
;
55
:
74
108
.
2.
Kamangar
F
,
Dores
GM
,
Anderson
WF
. 
Patterns of cancer incidence, mortality, and prevalence across five continents: defining priorities to reduce cancer disparities in different geographic regions of the world
.
J Clin Oncol
2006
;
24
:
2137
50
.
3.
Valente
L
,
Nishikura
K
. 
ADAR gene family and A-to-I RNA editing: diverse roles in posttranscriptional gene regulation
.
Prog Nucleic Acid Res Mol Biol
2005
;
79
:
299
338
.
4.
Athanasiadis
A
,
Rich
A
,
Maas
S
. 
Widespread A-to-I RNA editing of Alu-containing mRNAs in the human transcriptome
.
PLoS Biol
2004
;
2
:
e391
.
5.
Blow
M
,
Futreal
PA
,
Wooster
R
,
Stratton
MR
. 
A survey of RNA editing in human brain
.
Genome Res
2004
;
14
:
2379
87
.
6.
Li
JB
,
Levanon
EY
,
Yoon
JK
,
Aach
J
,
Xie
B
,
Leproust
E
, et al
Genome-wide identification of human RNA editing sites by parallel DNA capturing and sequencing
.
Science
2009
;
324
:
1210
3
.
7.
Maas
S
,
Kawahara
Y
,
Tamburro
KM
,
Nishikura
K
. 
A-to-I RNA editing and human disease
.
RNA Biol
2006
;
3
:
1
9
.
8.
Nishikura
K
. 
Functions and regulation of RNA editing by ADAR deaminases
.
Annu Rev Biochem
2010
;
79
:
321
49
.
9.
Patterson
JB
,
Samuel
CE
. 
Expression and regulation by interferon of a double-stranded-RNA-specific adenosine deaminase from human cells: evidence for two forms of the deaminase
.
Mol Cell Biol
1995
;
15
:
5376
88
.
10.
Chen
L
,
Li
Y
,
Lin
CH
,
Chan
TH
,
Chow
RK
,
Song
Y
, et al
Recoding RNA editing of AZIN1 predisposes to hepatocellular carcinoma
.
Nat Med
2013
;
19
:
209
16
.
11.
Shimada
Y
,
Imamura
M
,
Wagata
T
,
Yamaguchi
N
,
Tobe
T
. 
Characterization of 21 newly established esophageal cancer cell lines
.
Cancer
1992
;
69
:
277
84
.
12.
Chan
TH
,
Lin
CH
,
Qi
L
,
Fei
J
,
Li
Y
,
Yong
KJ
, et al
A disrupted RNA editing balance mediated by ADARs (Adenosine DeAminases that act on RNA) in human hepatocellular carcinoma
.
Gut
2013
Jun 13. [Epub ahead of print]
.
13.
Guan
XY
,
Sham
JS
,
Tang
TC
,
Fang
Y
,
Huo
KK
,
Yang
JM
. 
Isolation of a novel candidate oncogene within a frequently amplified region at 3q26 in ovarian cancer
.
Cancer Res
2001
;
61
:
3806
9
.
14.
Bass
BL
. 
RNA editing by adenosine deaminases that act on RNA
.
Annu Rev Biochem
2002
;
71
:
817
46
.
15.
Wang
LD
,
Qin
YR
,
Fan
ZM
,
Kwong
D
,
Guan
XY
,
Tsao
GS
, et al
Comparative genomic hybridization: comparison between esophageal squamous cell carcinoma and gastric cardia adenocarcinoma from a high-incidence area for both cancers in Henan, northern China
.
Dis Esophagus
2006
;
19
:
459
67
.
16.
Burns
CM
,
Chu
H
,
Rueter
SM
,
Hutchinson
LK
,
Canton
H
,
Sanders-Bush
E
, et al
Regulation of serotonin-2C receptor G-protein coupling by RNA editing
.
Nature
1997
;
387
:
303
8
.
17.
Higuchi
M
,
Single
FN
,
Kohler
M
,
Sommer
B
,
Sprengel
R
,
Seeburg
PH
. 
RNA editing of AMPA receptor subunit GluR-B: a base-paired intron-exon structure determines position and efficiency
.
Cell
1993
;
75
:
1361
70
.
18.
Paz
N
,
Levanon
EY
,
Amariglio
N
,
Heimberger
AB
,
Ram
Z
,
Constantini
S
, et al
Altered adenosine-to-inosine RNA editing in human cancer
.
Genome Res
2007
;
17
:
1586
95
.
19.
Bass
BL
,
Weintraub
H
. 
An unwinding activity that covalently modifies its double-stranded RNA substrate
.
Cell
1988
;
55
:
1089
98
.
20.
O'Connell
MA
,
Krause
S
,
Higuchi
M
,
Hsuan
JJ
,
Totty
NF
,
Jenny
A
, et al
Cloning of cDNAs encoding mammalian double-stranded RNA-specific adenosine deaminase
.
Mol Cell Biol
1995
;
15
:
1389
97
.
21.
Paupard
MC
,
O'Connell
MA
,
Gerber
AP
,
Zukin
RS
. 
Patterns of developmental expression of the RNA editing enzyme rADAR2
.
Neuroscience
2000
;
95
:
869
79
.
22.
Sansam
CL
,
Wells
KS
,
Emeson
RB
. 
Modulation of RNA editing by functional nucleolar sequestration of ADAR2
.
Proc Natl Acad Sci U S A
2003
;
100
:
14018
23
.
23.
Farajollahi
S
,
Maas
S
. 
Molecular diversity through RNA editing: a balancing act
.
Trends Genet
2010
;
26
:
221
30
.
24.
Samuel
CE
. 
Antiviral actions of interferons
.
Clin Microbiol Rev
2001
;
14
:
778
809
.
25.
Shah
SP
,
Morin
RD
,
Khattra
J
,
Prentice
L
,
Pugh
T
,
Burleigh
A
, et al
Mutational evolution in a lobular breast tumour profiled at single nucleotide resolution
.
Nature
2009
;
461
:
809
13
.
26.
DeCerbo
J
,
Carmichael
GG
. 
Retention and repression: fates of hyperedited RNAs in the nucleus
.
Curr Opin Cell Biol
2005
;
17
:
302
8
.
27.
Gu
T
,
Buaas
FW
,
Simons
AK
,
Ackert-Bicknell
CL
,
Braun
RE
,
Hibbs
MA
. 
Canonical A-to-I and C-to-U RNA editing is enriched at 3′UTRs and microRNA target sites in multiple mouse tissues
.
PloS ONE
2012
;
7
:
e33720
.
28.
Prasanth
KV
,
Prasanth
SG
,
Xuan
Z
,
Hearn
S
,
Freier
SM
,
Bennett
CF
, et al
Regulating gene expression through RNA nuclear retention
.
Cell
2005
;
123
:
249
63
.
29.
Zhang
Z
,
Carmichael
GG
. 
The fate of dsRNA in the nucleus: a p54(nrb)-containing complex mediates the nuclear retention of promiscuously A-to-I edited RNAs
.
Cell
2001
;
106
:
465
75
.
30.
Murakami
Y
,
Ichiba
T
,
Matsufuji
S
,
Hayashi
S
. 
Cloning of antizyme inhibitor, a highly homologous protein to ornithine decarboxylase
.
J Biol Chem
1996
;
271
:
3340
2
.
31.
Nilsson
J
,
Grahn
B
,
Heby
O
. 
Antizyme inhibitor is rapidly induced in growth-stimulated mouse fibroblasts and releases ornithine decarboxylase from antizyme suppression
.
Biochem J
2000
;
346
Pt 3
:
699
704
.
32.
Choi
KS
,
Suh
YH
,
Kim
WH
,
Lee
TH
,
Jung
MH
. 
Stable siRNA-mediated silencing of antizyme inhibitor: regulation of ornithine decarboxylase activity
.
Biochem Biophys Res Commun
2005
;
328
:
206
12
.
33.
Keren-Paz
A
,
Bercovich
Z
,
Porat
Z
,
Erez
O
,
Brener
O
,
Kahana
C
. 
Overexpression of antizyme-inhibitor in NIH3T3 fibroblasts provides growth advantage through neutralization of antizyme functions
.
Oncogene
2006
;
25
:
5163
72
.
34.
Li
Y
,
Chen
L
,
Chan
TH
,
Guan
XY
. 
Hepatocellular carcinoma: transcriptome diversity regulated by RNA editing
.
Int J Biochem Cell Biol
2013
;
45
:
1843
8
.

Supplementary data