Purpose: Emerging studies demonstrate that long noncoding RNAs (lncRNA) participate in the regulation of various cancers. In the current study, a novel lncRNA-TTN-AS1 has been identified and explored in esophageal squamous cell carcinoma (ESCC).
Experimental Design: To discover a new regulatory circuitry in which RNAs crosstalk with each other, the transcriptome of lncRNA-miRNA-mRNA from ESCC and adjacent nonmalignant specimens were analyzed using multiple microarrays and diverse bioinformatics platforms. The functional role and mechanism of a novel lncRNA-TTN-AS1 were further investigated by gain-of-function and loss-of-function assays in vivo and in vitro. An ESCC biomarker panel, consisting of lncRNA-TTN-AS1, miR-133b, and FSCN1, was validated by qRT-PCR and in situ hybridization using samples from 148 patients.
Results: lncRNA-TTN-AS1 as an oncogene is highly expressed in ESCC tissues and cell lines, and promotes ESCC cell proliferation and metastasis. Mechanistically, lncRNA-TTN-AS1 promotes expression of transcription factor Snail1 by competitively binding miR-133b, resulting in the epithelial–mesenchymal transition (EMT) cascade. Moreover, lncRNA-TTN-AS1 also induces FSCN1 expression by sponging miR-133b and upregulation of mRNA-stabilizing protein HuR, which further promotes ESCC invasion cascades. We also discovered and validated a clinically applicable ESCC biomarker panel, consisting of lncRNA-TTN-AS1, miR-133b, and FSCN1, that is significantly associated with overall survival and provides additional prognostic evidence for ESCC patients.
Conclusions: As a novel regulator, lncRNA-TTN-AS1 plays an important role in ESCC cell proliferation and metastasis. The lncRNA-TTN-AS1/miR133b/FSCN1 regulatory axis provides bona fide targets for anti-ESCC therapies. Clin Cancer Res; 24(2); 486–98. ©2017 AACR.
Long noncoding RNAs (lncRNA) play pivotal roles in esophageal squamous cell carcinoma (ESCC) proliferation, metastasis, diagnosis, and prognosis. We identified a novel ESCC-related lncRNA-TTN-AS1 as a vital regulator of ESCC progression. lncRNA-TTN-AS1 promoted snail1 and FSCN1 expression by competitively binding miR-133b and interacting with mRNA to stabilize protein HuR, resulting in activation of a metastasis cascade. The biomarker panel of lncRNA-TTN-AS1-miR-133b-FSCN1 correlates with overall survival and provides accurate prognostic evidence.
Esophageal carcinoma is the sixth leading cause of tumor-related mortality worldwide (1). There are two main esophageal carcinoma types: esophageal squamous cell carcinoma (ESCC) and adenocarcinoma (EAC). ESCC is the predominant subtype of esophageal carcinoma in Asia. Although multimodal therapies have improved treatment and prognosis of esophageal carcinoma, the overall 5-year survival rate is still poor (2). The poor outcomes of ESCC are associated with diagnosis at advanced stages and the propensity for metastasis (3). Therefore, diagnosis of ESCC in the early stages is crucial.
With advances in high throughput analysis, increasing tumor-related noncoding RNAs have been identified (4). Particularly, accumulating esophageal carcinoma-related long noncoding RNAs (lncRNAs) have been verified to exert diverse functions through various biological processes. For example, HNF1A-AS1 induces H19 expression and modulates chromatin and nucleosome assembly, resulting in gene imprinting (5). HOX transcript antisense RNA (HOTAIR) suppresses WIF-1 expression by inducing histone H3K27 methylation in the promoter region (6). Numerous esophageal carcinoma-related lncRNAs have been found, but the precise molecular mechanisms of most lncRNAs in ESCC are still not fully understood.
In this study, we identified a novel lncRNA-TTN-AS1 (ENST00000589434) and demonstrated that high levels of lncRNA-TTN-AS1 were correlated with poor ESCC prognosis, tumor growth, and invasion cascades. Further mechanistic studies revealed that lncRNA-TTN-AS1 upregulates Snail1 and actin-binding protein fascin homolog 1 (FSCN1) by competitive regulation of miR-133b, resulting in ESCC cell metastasis. In addition, lncRNA-TTN-AS1 also facilitates and combines directly with the HuR to stabilize FSCN1 mRNA. Taken together, the study unveils a novel biomarker panel, consisting of lncRNA-TTN-AS1/miR133b/FSCN1, which plays a pivotal role in ESCC progression and metastasis.
Materials and Methods
Patients and specimens
Two independent cohorts comprising 148 ESCC patients were enrolled for this study. In cohort 1, ESCC and adjacent nontumor specimens were gathered from 58 patients who were diagnosed with ESCC between December 2014 and November 2015 at Nanjing General Hospital of Nanjing Military Command (Jiangsu, China). The study was approved by Nanjing General Hospital of Nanjing Military Command Review Board, and written informed consent was obtained from all participants. In cohort 2, paraffin-embedded tissue samples were collected from archival material stored in the Biobank Center at the National Engineering Center for Biochip at Shanghai (Shanghai Outdo Biotech Co., Ltd.). Specimens from ESCC and adjacent nonmalignant tissues were collected from 90 ESCC patients who underwent surgical resection between 2006 and 2008, and were followed up for 7.8 years. The clinical characteristics of all patients are listed in Supplementary Tables S4 and S7. All patients were diagnosed according to the guidelines of the American Joint Commission on Cancer and the guidelines of the International Union Against Cancer (IUAC). The study was conducted in accordance with Ethics Committee of China Pharmaceutical University.
Seven paired samples of ESCC versus adjacent noncancerous tissues were selected for microarrays analysis (Outdo Biotech Co., Ltd.; Supporting Information Table S1). Briefly, fluorescent (Cy5 and Cy3-dCTP) labeled cDNA was synthesized and hybridized to the 4 × 180 K Agilent human lncRNA+mRNA Array v4.0 (Agilent). After hybridization and washing, the slides were scanned using an Agilent G2565CA Microarray Scanner. Quantile normalization and differential analysis were performed with the Agilent GeneSpring software v.11.5 (Agilent Technologies Inc.). To select the differentially expressed genes, we used threshold values of ≥2 and ≤−2 fold change and a Benjamini–Hochberg corrected P value of 0.05. The data were log2 transformed and median centered by genes using the Adjust Data function of CLUSTER 3.0 software (University of Tokyo, Human Genome Center, Tokyo, Japan) and then further analyzed with hierarchical clustering with average linkage. Finally, we performed tree visualization by using Java TreeView (Stanford University School of Medicine, Stanford, CA).
miRNA microarray analysis was undertaken using the above seven paired samples of ESCC versus adjacent noncancerous tissues (Outdo Biotech Co., Ltd.). Briefly, miRNAs were extracted and purified from total RNA using a mirVana miRNA Isolation Kit (Ambion), then a poly A tail was added in the 3′ end of miRNA using poly A polymerase and labeled with biotin labeling 3DNA dendrimer (FlashTag Biotin RNA Labeling). Biotin-labeled miRNA hybridized to the Affymetrix GeneChip miRNA Array 4.0. Array (Affymetrix) was scanned with a GeneChip Scanner 3000, and the images processed using the AGCC software (Affymetrix GeneChip Command Console Software). Signals were normalized by the median center tool for genes in the Cluster 3.0 software and analyzed by significance analysis of microarrays (SAM), with the FDR threshold set at 0 and fold change set (fold change ≥2; P ≤ 0.05 or change ≤ −2; P ≤ 0.05).
Cell culture and reagents
ESCC cell lines Eca-109, KYSE 30, KYSE 150, KYSE180, KYSE410, KYSE450, KYSE510, TE-10, and TE-13 and normal esophageal epithelial cell lines HECC kindly supplied by Professor Yifeng Zhou, Suzhou University in 2015. The cell lines have been tested and authenticated by the company. All cells were tested for mycoplasma every 3 months and were negative. Antibodies against cleaved-caspase-3, cleaved-caspase-9, HuR, β-catenin, Lamin B1, and Ki-67 were purchased from Cell Signaling Technology. Antibodies against pro-caspase-3, pro-caspase 9, β-actin, Bcl-2, Bax, E-cadherin, ZO-1, N-cadherin, Vimentin, Snail1, Argonaute2 (Ago2), and FSCN1 were purchased from Abcam. Flag antibody was purchased from Sigma-Aldrich. Cell viability assay was performed using CCK8 (Dojindo) kits. Propidium Iodide and Annexin V were purchased from Biolegend.
Cell viability, colony formation, and in vivo xenograft assays
The cell viability, colony formation, and in vivo tumor growth assays were performed as described previously (7), and as detailed in the Supplementary Materials and Methods section.
Quantitative real-time PCR, Western blot analysis, and flow cytometry
The procedures for performing qRT-PCR and Western blot analysis have been described previously (7). Flow cytometry was conducted according to the manufacturer's standard protocol.
Female Balb/c nude mice (aged, 4–5 weeks; Cavens Lab Animal Co.) were cared for according to Provisions and General Recommendation of Chinese Experimental Animals Administration Legislation. The procedure of all animal experiments complied with Institutional Animal Care and Use Committee (IACUC) regulations. All animal experiments were approved by the Ethics Committee of China Pharmaceutical University Permit Number: SYXK2012-0035.
Transwell migration/invasion and wound healing assays and in vivo imaging assays
The migration/invasion assays, wound healing assays and in vivo tumor growth and imaging assay were performed as described previously (7), and as detailed in the Supplementary Materials and Methods section.
Dual luciferase reporter assays
Luciferase activities were performed using the Dual Luciferase Assay Kit (Promega) according to the manufacturer's instructions.
RNA pull-down and RNA immunoprecipitation assays
RNA pull down and RNA immunoprecipitation (RIP) were performed as described previously (7, 8). For RNA pull down assay, RNAs were biotin-labeled and in vitro transcribed with Biotin RNA Labeling Mix (Roche) and T7/SP6 RNA polymerase (Roche). Cell lysates were mixed and incubated with biotinylated RNAs. Streptavidin agarose beads were added to each binding reaction, followed by 1-hour incubation at room temperature. Beads were washed and boiled in SDS buffer, and the retrieved proteins detected by Western blot analysis.
The Magna RIP Kit (Millipore) was used in RIP experiments according to the manufacturer's instructions. The coprecipitated RNAs were detected by qRT-PCR.
RNA-LNA in situ hybridization
In situ hybridization (ISH) of lncRNA/mRNA and miRNA with ESCC tissue microarrays (TMA) were performed by Shanghai Outdo Biotech Co., Ltd. (catalog no. HEso-Squ180Sur-04). For the TMAs, there were 90 ESCC patient samples complete with survival times and related clinicopathologic characteristics. The tissue array was stained with hematoxylin and eosin (H&E) to verify the presence of tumor cells. Detailed descriptions can be found in the Supplementary Materials and Methods section.
Statistical analyses were performed using SPSS statistics 22.0. The paired t test was performed to detect the differential expression of lncRNA-TTNF-AS1 in ESCC cancer tissues compared with adjacent normal tissues. The relationship between lncRNA-TTN-AS1 and clinicopathologic characteristics was evaluated using χ2 test. Survival curves were calculated using Kaplan–Meier and log-rank tests. The effects of variables on survival were analyzed by univariate and multivariate Cox proportional hazards modeling. For two group comparison, multiple group comparison and correlation analyses were calculated with a paired two-tailed Student t test, two-way ANOVA test, linear regression test, and Pearson test using GraphPad Prism 5 software (Graph Pad software Inc.), respectively. P values less than 0.05 were considered statistically different (*, P < 0.05; **, P < 0.01; ***, P < 0.001).
Upregulation of lncRNA-TTN-AS1 in ESCC tissues and cell lines
To understand the regulatory circuitries by which RNAs can crosstalk with each other, the expression profiles of lncRNA, mRNA, and miRNA in ESCC tissues and adjacent normal tissues from seven ESCC patients (Supplementary Table S1) were detected by multiple microarrays. Comparison of differently expressed miRNAs was calculated by single channel chip and normalized by Lowess (Fig. 1A). lncRNA and mRNA expression profiles were detected by dual channel chip (Fig. 1B and C). All hierarchical clustering results showed systematic variation in transcript levels between ESCC tissues and adjacent normal tissues. The microarray data have been deposited in NCBI Gene Expression Omnibus and are accessible through GEO Series Accession Number GSE97051 (https://www.ncbi.nlm.nih.gov/geo/query/acc.cgi?acc=GSE97051). To validate our microarray data, 14 transcripts with significantly different expression were examined by qRT-PCR (Supplementary Fig. S1). The results showed that the levels of 11 transcripts were consistent with the microarray data (Supplementary Fig. S4A). Furthermore, potential lncRNAs were screened using coexpression network (Fig. 1D) and diverse bioinformatics tools (Fig. 1E). Finally, the number of miRNA–lncRNA pairs with significant correlation was narrowed to three (Supplementary Table S3). Microarray data showed ENST00000589434 was notably downregulated in ESCC tissues and the correlation score between ENST00000589434 and hsa-miR-133b was the highest. Therefore, we focused on the function and mechanism of lncRNA-TTN-AS1 (ENST00000589434) in ESCC. Notably, lncRNA-TTN-AS1 is derived from the opposite strand of Titin (TTN) gene, which encodes a large abundant protein of striated muscle, meanwhile, miR-133b is derived from linc-MD1, which modulates early phases of muscle differentiation. Hence, we also detected the expression of TTN and linc-MD1 in ESCC specimens, as shown in Supplementary Fig. S4B and S4C, TTN and linc-MD1 levels between ESCC and adjacent normal tissues were significantly different.
The potential coding capability of lncRNA-TTN-AS1 was also evaluated as follows: although two short reading frames (ORF3 and ORF5) with more than 200 nt were predicted using ORF Finder from the National Center for Biotechnology Information (Supplementary Fig. S2A), neither of their AUGs showed the Kozak consensus, nor were homologous protein sequences found using a BLAST search; PhyloCSF value of all the exons of lncRNA-TTN-AS1 were less than zero and their sequences were less conserved, which further suggested that it was unlikely to encode any protein (Supplementary Fig. S2B); the online bioinformatics analysis (coding potential calculator) also confirmed lncRNA-TTN-AS1 has no coding capability (coding potential score: −1.11259; http://cpc.cbi.pku.edu.cn/programs/run_cpc.jsp) in agreement with our finding that lncRNA-TTN-AS1 has no coding capability (Supplementary Fig. S2C).
lncRNA-TTN-AS1 regulates miR-133b as a ceRNA
It is well known that noncoding RNAs as competing endogenous RNA (ceRNA) bind to miRNAs and protect their target RNAs from repression or degradation. The downregulation of miR-133b that was found in human ESCC (9) prompted us to see whether lncRNA-TTN-AS1 was negatively correlated with miR-133b in ESCC tissues. As expected, lncRNA-TTN-AS1 was robustly upregulated in ESCC tissues in cohort 1 (P = 0.000; Supplementary Fig. S3A.). Conversely, miR-133b expression was significantly downregulated in ESCC tissues (P = 0.000; Supplementary Fig. S3B). Moreover, ISH studies also confirmed that lncRNA-TTN-AS1 significantly increased in ESCC (Fig. 6A). In addition, compared to normal esophageal epithelial cells (HEEC) cells, lncRNA-TTN-AS1 expression was notably higher in ESCC cell lines, whereas miR-133b level was lower (Supplementary Fig. S3C and S3D). Consistently, a strong negative correlation was found between lncRNA-TTN-AS1 and miR-133b in ESCC tissues (r = −0.8704, P < 0.001; Supplementary Fig. S3E) and cell lines (r = −0.8500, P < 0.001; Supplementary Fig. S3F). Although the qRT-PCR results were not fully consistent with the microarray data, lncRNA-TTN-AS1 has been validated as a potential oncogene.
Next we performed luciferase reporter assays and RNA pull-down assays to test the direct binding between lncRNA-TTN-AS1 and miR-133b. The luciferase intensity was decreased by cotransfected miR-133b mimics and lncRNA-TTN-AS1-WT but not in the mutant reporter vector lacking the putative miR-133b binding site (Fig. 2A). Consistently, lncRNA-TTN-AS1 was pulled down by biotin-labeled miR-133b, whereas miR-133b mutant could not pull down lncRNA-TTN-AS1. In a reciprocal manner, miR-133b was also precipitated by wild-type lncRNA-TTN-AS1 but not the lncRNA-TTN-AS1 mutant (Fig. 2B). These data suggest that lncRNA-TTN-AS1 is a bona fide miR-133b–targeting lncRNA.
Intriguingly, miR-133b was downregulated by lncRNA-TTN-AS1 overexpression, as well as upregulated by lncRNA-TTN-AS1 knockdown (Supplementary Fig. S3I and S3J). However, no significant difference of lncRNA-TTN-AS1 expression was detected after ectopic expression or deficiency of miR-133b (Supplementary Fig. S4D and S4E). Moreover, overexpression of lncRNA-TTN-AS1 significantly attenuated miR-133b level in contrast to the mutant lacking the miR-133b targeting site (Fig. 2C). These data confirm that miR-133b binds lncRNA-TTN-AS1 without degradation of lncRNA-TTN-AS1.
lncRNA-TTN-AS1 was mainly found in the cytoplasm of ESCC cell lines (Fig. 2D), which suggested that lncRNA-TTN-AS1 may bind to miR-133b through the Ago2-dependent RNAi pathway. As expected, RIP assay showed levels of lncRNA-TTN-AS1 and miR-133b precipitated by anti-Ago2 antibody were markedly increased, with a 2- to 3-fold enrichment compared with IgG (Fig. 2E). Meanwhile, endogenous lncRNA-TTN-AS1 pull-down by Ago2 was specifically enriched upon overexpression of miR-133b (Fig. 2F). These data confirmed that lncRNA-TTN-AS1 was bound to miR-133b in the cytoplasm in an Ago2-dependent manner. To further confirm lncRNA-TTN-AS1 as a ceRNA, we compared the abundance of lncRNA-TTN-AS1 and miR-133b. The exact copy numbers of lncRNA-TTN-AS1 (∼1.45 copies per cell) was higher than that of miR-133b (∼0.37 copies per cell) in TE-13 cells (Supplementary Fig. S4F). Moreover, lncRNA-TTN-AS1 overexpression decreased the copy numbers of miR-133b (Supplementary Fig. S4G). Taken together, these data indicate that lncRNA-TTN-AS1 physically interacts with miR-133b as a ceRNA.
The role of lncRNA-TTN-AS1 in ESCC cell proliferation, cell apoptosis, and cell-cycle progression
To dissect the effect of lncRNA-TTN-AS1 in ESCC progression, gain- and loss-of-function assays were performed using ESCC cell lines. lncRNA-TTN-AS1 was inhibited in KYSE-410 cells and then we stably overexpressed miR-133b in lncRNA-TTN-AS1–overexpressing clones (Supplementary Fig. S3G and S3I). Meanwhile, lncRNA-TTN-AS1 was stably silenced in TE-13 cells, followed by knockdown of miR-133b in lncRNA-TTN-AS1 deleted clones (Supplementary Fig. S3H and S3J).
To explore the influence of lncRNA-TTN-AS1 on ESCC proliferation, CCK-8 and colony formation assays were performed. Ectopic expression of lncRNA-TTN-AS1 induced cell proliferation and colony formation, whereas overexpression of miR-133b abolished this increase (Fig. 3B; Supplementary Fig. S6A). In an in vivo assay, tumor growth in xenografts with lncRNA-TTN-AS1–overexpressing clones was increased compared with that of a negative control, whereas ectopic expression of miR-133b eliminated the lncRNA-TTN-AS1–induced tumor growth (Fig. 3D). In addition, overexpression of lncRNA-TTN-AS1 augmented the proportion of proliferating (Ki67+) cancer cells (Supplementary Fig. S6C).
To further investigate the effect of lncRNA-TTN-AS1 and miR133b on cell proliferation, apoptosis-related experiments were performed. As shown in Fig. 3A, the percentage of Annexin V-Light 650-positive cells decreased upon overexpression of lncRNA-TTN-AS1, whereas ectopic expression of miR-133b abrogated the decrease. Consistently, overexpression of lncRNA-TTN-AS1 resulted in reduction of well-known apoptotic proteins, including cleaved caspase-3, cleaved caspase-9, and Bax and increase of antiapoptosis protein Bcl-2 in ESCC cell lines that was overcome by ectopic expression of miR-133b (Fig. 3C). Furthermore, cell-cycle analysis confirmed ESCC cells with overexpressing lncRNA-TTN-AS1 had a significantly reduced G1 population and a markedly increased S-phase, and ectopic expression of miR-133b reversed the above phenomena (Supplementary Fig. S6B). Collectively, these results indicate that lncRNA-TTN-AS1 induces cell proliferation by inactivation of apoptosis-related signaling pathway and facilitates cell-cycle progression.
lncRNA-TTN-AS1 promotes Snail1 expression
Because miR-133b targets Snail1 (10) and lncRNA-TTN-AS1 shares a miR-133b response element with Snail1, we reasoned that lncRNA-TTN-AS1 could induce EMT-transcription factor Snail1 and promote invasion of ESCC cells. We found that lncRNA-TTN-AS1 significantly increased Snail1 expression, which could be abrogated by ectopic expression of miR-133b (Supplementary Fig. S5A and S5B). Furthermore, dual luciferase reporter assays validated ectopic expression of lncRNA-TTN-AS1, but not the mutant, increased the luciferase intensity, which could be abolished by overexpression of miR-133b (Supplementary Fig. S5D). In addition, lncRNA-TTN-AS1 was positively correlated with Snail1 mRNA level (Supplementary Fig. S5E). These data illustrate that lncRNA-TTN-AS1 modulates Snail1 by competitively binding miR-133b, which may further promote ESCC metastasis.
lncRNA-TTN-AS1 induces ESCC cell metastasis in vitro and in vivo
To further examine the effect of lncRNA-TTN-AS1 on ESCC cell metastasis and EMT cascades, cell migration and invasion and wound healing assays were conducted. Ectopic expression of lncRNA-TTN-AS1 promoted cell migration, cell invasion, and scratch closure rate, whereas overexpression of miR-133b attenuated lncRNA-TTN-AS1-induced cell metastasis (Fig. 4A; Supplementary Fig. S6D and S6E). Meanwhile, lncRNA-TTN-AS1 significantly induced mesenchymal markers N-cadherin and Vimentin and decreased expression of epithelial markers E-cadherin and ZO-1, which were rescued by ectopic expression of miR-133b (Fig. 4B and C). In addition, lncRNA-TTN-AS1 abolished the repression of Snail1 and EMT induced by miR-133b (Supplementary Fig. S7B and S7C). In summary, our data demonstrate that lncRNA-TTN-AS1 plays a pivotal role in activating EMT by the competitive binding of miR-133b.
To ascertain the correlation between lncRNA-TTN-AS1 and EMT markers, we examined the levels of EMT markers in four ESCC cell lines with different expression of lncRNA-TTN-AS1. High levels of Snail1, N-cadherin, and Vimentin and low levels of E-cadherin and ZO-1 were observed in lncRNA-TTN-AS1 high expression cells (Supplementary Fig. S7D and S7E). Consistently, the lncRNA-TTN-AS1 transcript was negatively correlated with E-cadherin mRNA levels in ESCC specimens (Supplementary Fig. S5F).
To evaluate the effect of lncRNA-TTN-AS1 on tumor metastasis in vivo, we then intravenously injected the indicated ESCC cells into nude mice to establish a tumor metastasis model. In vivo imaging indicated that the different clones labeled with GFP mainly distributed in the livers and lungs of nude mice (Fig. 4D), overexpression of lncRNA-TTN-AS1 augmented the fluorescent intensities of liver and lung, which were attenuated by ectopic expression of miR-133b (Fig. 4E and F). Similarly, metastatic tumor cells in liver and lung were significantly increased with ectopic expression of lncRNA-TTN-AS1, whereas miR-133b abrogated the increase (Fig. 4G and H; Supplementary Fig. S7F and S7G). All above data verify that lncRNA-TTN-AS1 promotes ESCC metastasis.
lncRNA-TTN-AS1 enhances FSCN1 expression via its sponge activity and interaction with HuR
miR-133b targets and modulates FSCN1 (actin-binding protein, Fascin homolog1) expression that is associated with ESCC cell metastasis (9). Because lncRNA-TTN-AS1 harbors an identical miR-133b–binding site with FSCN1, we questioned whether lncRNA-TTN-AS1 may regulate FSCN1 by miR-133b and further enhances the EMT signaling pathway in ESCC cells. Consistently, overexpression of lncRNA-TTN-AS1 augmented FSCN1 level while ectopic expression of miR-133b abrogated the increase. (Fig. 5A and B). To further test whether the above effect correlated to modulation of the FSCN1 3′UTR, a luciferase plasmid (pmirGLO or pmirGLO-FSCN1) was transfected into TE13 cells. Overexpression of lncRNA-TTN-AS1 increased the luciferase intensity of pmirGLO-FSCN1. Ectopic expression of miR-133b overcame this upregulation (Fig. 5C). In addition, lncRNA-TTN-AS1 transcript levels were significantly positively correlated with FSCN1 mRNA levels in ESCC specimens (Spearman correlation = 0.9231, P < 0.001) and cell lines (Spearman correlation = 0.9167, P < 0.001; Supplementary Fig. S8A and S8B). Collectively, these data indicated that the underlying mechanism by which lncRNA-TTN-AS1 induced the EMT cascade is also associated with promotion of FSCN1 expression via its sponge activity.
Subcellular location of lncRNAs determines its underlying mechanism. Cytoplasmic lncRNAs are well known for regulating gene transcription through interaction with RNA-binding proteins (RBP; ref. 11). Because miR-133b targets and modulates HuR mRNA, on the contrary, HuR also represses miR-133b release from linc-MD1 (12). Thus, we inferred that lncRNA-TTN-AS1 may also promote HuR via its sponge activity, which further induces FSCN1 mRNA expression and stability. As expected, first, ectopic expression of miR-133b reduced HuR levels (Supplementary Fig. S8C), conversely, depletion of HuR increased miR-133b levels in TE13 cells (Supplementary Fig. S8D). Second, overexpression of lncRNA-TTN-AS1 upregulated HuR levels, which was abolished by ectopic expression of miR-133b (Fig. 5D). Moreover, ectopic expression of miR-133b abolished the increase in luciferase intensity of pmirGLO-HuR induced by overexpression of lncRNA-TTN-AS1 (Supplementary Fig. S8I), indicating that lncRNA-TTN-AS1 enhances HuR expression via sponging miRNA-133b. Third, HuR directly interacted with both lncRNA-TTN-AS1 and FSCN1 mRNA in RIP assays (Fig. 5E). Furthermore, the 5′-end (768–518 nt) of lncRNA-TTN-AS1 is indispensable for the interaction between lncRNA-TTN-AS1 and HuR (Fig. 5F). Fourth, lncRNA-TTN-AS1 overexpression increased the level of HuR protein in the cytoplasm (Supplementary Fig. S8E), suggesting that lncRNA-TTN-AS1 may induce HuR translocation to stabilize FSCN1 mRNA. As expected, overexpression of lncRNA-TTN-AS1 elongated the half-life of FSCN1 mRNA, which was overcome upon depletion of HuR (Fig. 5G). In addition, miR-133b inhibited the increase in FSCN1 mRNA stability upon overexpression of lncRNA-TTN-AS1 (Fig. 5H). Taken together, on one hand, lncRNA-TTN-AS1 induces HuR expression via competitive binding of miR-133b, which further enhances FSCN1 mRNA stability through binding of HuR. On the other hand, lncRNA-TTN-AS1 also stabilizes FSCN1 mRNA via its sponge activity.
Given that FSCN1 is a downstream target of β-catenin (13) and β-catenin interacts with HuR (14), we first examined the effect of modulating lncRNA-TTN-AS1 on the HuR and β-catenin levels in TE-13 cells. Ectopic expression of lncRNA-TTN-AS1 induced HuR expression (Fig. 5D) and a concomitant upregulation and nuclear accumulation of β-catenin (Supplementary Figs. S8F and S5I). Second, silencing of HuR decreased β-catenin expression (Fig. 5J), and knockdown of β-catenin reduced FSCN1 (Fig. 5K), suggesting the existence of a concerted lncRNA-TTN-AS1 >HuR > β-catenin > FSCN1 axis. Notably, HuR deficiency overcame the effect of lncRNA-TTN-AS1 in promoting FSCN1 expression (Supplementary Fig. S8G), cell proliferation (Fig. 5L), and migration (Supplementary Fig. S8H) in ESCC cells, suggesting HuR also predominately modulates the function of lncRNA-TTN-AS1. In summary, lncRNA-TTN-AS1 enhances HuR expression via sponging miR-133b, and upregulation of HuR further decreases miR-133b level and increases β-catenin expression, resulting in upregulation of FSCN1.
lncRNA-TTN-AS1/miR-133b/FSCN1 as a biomarker panel in ESCC
To investigate the association between lncRNA-TTN-AS1/miR-133b/FSCN1 levels and ESCC progression, we measured the expression levels of lncRNA-TTN-AS1/miR-133b/FSCN1 in cohorts 1 and 2 by qRT-PCR and ISH assays, respectively. The results showed that lncRNA-TTN-AS1 and FSCN1 were mainly expressed in the ESCC tissues compared with adjacent normal tissues (Supplementary Fig. S3A and S8J, Fig. 6A and 6C), whereas miR-133b expression was more abundant in the normal tissues (Fig. 6B; Supplementary Fig. S3B).
In cohort 1, lncRNA-TTN-AS1-high group was notably correlated with high advanced TNM stage (N stage, P = 0.033) and clinical stage (P = 0.013; Supplementary Table S4). The data were also examined by analysis of samples from cohort 2 (Supplementary Table S7). In contrast, in cohort 1, miR-133b-low expression was significantly associated with pathologic grade (P = 0.075), high advanced TNM stage (N stage, P = 0.033), and clinical stage (P = 0.001; Supplementary Table S5). These correlations were also validated by analysis of samples from cohort 2 (Supplementary Table S8). There was also significant association between FSCN1-high expression and pathologic grade (P = 0.004), TNM stage (T stage, P = 0.002), N stage (P = 0.000), and clinical stage (P = 0.004; Supplementary Table S6) in cohort 1. The data were also validated by analysis of specimens from cohort 2 (Supplementary Table S9). In addition, Kaplan–Meier and log-rank test analyses verified that patients with high expression of lncRNA-TTN-AS1/FSCN1 or low expression of miR-133b were positively correlated with a reduced overall survival (OS; P < 0.001; Fig. 6D–F). Moreover, the OS of ESCC patients with lncRNA-TTN-AS1-high/miR-133b-low was shorter than patients in the other three groups (Fig. 6G). Similarly, the lncRNA-TTN-AS1-high/FSCN1-high group has a significantly lower survival rate compared with the other groups (Fig. 6H). In addition, multivariate analysis indicated that stage (HR, 2.39; 95% CI, 1.14–4.96; P = 0.001), lncRNA-TTN-AS1 expression (HR, 2.73; 95% CI, 1.27–4.58; P = 0.004), miR-133b expression (HR, 2.13; 95% CI, 1.36–5.78; P = 0.023), and FSCN1 expression (HR, 3.68; 95% CI, 1.35–7.02; P =0.000) were independent prognostic factors for OS in ESCC patients (Supplementary Table S10). These data suggested that combination of lncRNA-TTN-AS1 and miR-133b or FSCN1 could separate patients into distinct prognostic groups (P < 0.01).
The key finding of this study was the discovery of a novel lncRNA-TTN-AS1 that plays a vital role in ESCC progression and metastasis. We discovered that lncRNA-TTN-AS1 enhances Snail1 and FSCN1 levels by competitively binding to miR-133b, resulting in the promotion of EMT cascades. Meanwhile, lncRNA-TTN-AS1 also induces HuR via its sponge activity, which further activates the FSCN1/β-catenin signaling pathway (Fig. 6l). In addition, lncRNA-TTN-AS1/miR-133b/FSCN1 is a potential prognostic biomarker panel in ESCC carcinogenesis.
The functions of lncRNAs are closely related to their subcellular localization. Nuclear lncRNAs mainly affect chromatin structure and gene transcription (15, 16). Cytosolic lncRNAs modulate protein subcellular localization (17) and mRNA translation (18, 19). Herein, lncRNA-TTN-AS1 has been shown to target miR-133b using multiple bioinformatics platforms. Intriguingly, ectopic expression of miR-133b decreased the luciferase intensities of the lncRNA-TTN-AS1 WT reporter. However, there was no significant difference in lncRNA-TTN-AS1 expression upon overexpression of miR-133b. Moreover, endogenous lncRNA-TTN-AS1 and miR-133b were pulled down by anti-Ago2. Taken together, all these data suggest that miR-133b recognizes and binds with lncRNA-TTN-AS1 without triggering degradation of lncRNA-TTN-AS1. The reciprocal modulation between miRNAs and lncRNAs is still elusive miR-21 and lncRNA-GAS5 suppress expression of each other in an Ago2-dependent manner (20), whereas miR-200 only bind to lncRNA-ATB via the RISC complex but does not affect the expression of lncRNA-ATB (21). Further studies are still required to fully understand the miRNA regulatory network.
The miR-133 family has tumor-suppressive genes (22, 23) that are involved in ESCC progression (9). Because miR-133b mediated Snail1 repression was shown to improve cardiac reprogramming (10), thus, we reasoned that miR-133b may regulate Snail1, and then induce an invasion cascade in ESCC. Snail1 as an EMT transcription factor coordinates the repression of epithelial phenotype and the induction of mesenchymal phenotype (24, 25). lncRNA-TTN-AS1 abolishes the suppression of Snail1 mediated by miR-133b, and the upregulation of Snail1 contributes to the repression of epithelial markers E-cadherin and ZO-1 and activation of mesenchymal markers N-cadherin and vimentin, promoting the EMT cascade. Meanwhile, miR-133b binds and inhibits FSCN1 expression (9), which is significantly correlated with poor prognosis in ESCC (26). Therefore, we assumed that lncRNA-TTN-AS1 could modulate FSCN1 mRNA level by competitively sponging miR-133b, thereby enhancing ESCC metastasis. As expected, lncRNA-TTN-AS1 increased FSCN1 expression in ESCC cells, which was reversed by miR-133b. In addition, lncRNA-TTN-AS1 facilitated HuR expression via sponging miR-133b, the upregulation of HuR further decreased miR-133b level and enhanced β-catenin expression, which induced FSCN1 mRNA level and stability. Notably, the combination between lncRNA-TTN-AS1 and miR-133b/FSCN1 could be a potential prognostic biomarker panel of ESCC. All the above data suggested that the lncRNA–miR-133b–FSCN1 axis plays a critical role in the metastasis-invasion cascade in ESCC progression.
Taken together, our research revealed that lncRNA-TTN-AS1 promotes ESCC cell proliferation and invasion metastasis, which induces competitive binding to miR-133b, resulting in upregulation of Snail1, HuR, and FSCN1. In addition, HuR-induced by lncRNA-TTN-AS1 increases β-catenin expression, enhancing FSCN1 mRNA expression and stability by interaction with HuR. The pleiotropic effect of lncRNA-TTN-AS1 on ESCC progression has provided important new insights into our comprehension of ESCC carcinogenesis.
Disclosure of Potential Conflicts of Interest
No potential conflicts of interest were disclosed.
Conception and design: C. Lin, E. Nice
Development of methodology: C. Lin
Acquisition of data (provided animals, acquired and managed patients, provided facilities, etc.): C. Lin, S. Zhang, M. Li, C. Liu, J. Hao, W. Qi
Analysis and interpretation of data (e.g., statistical analysis, biostatistics, computational analysis): C. Lin, Ying Wang, Yuanshu Wang, L. Yu, L. Hu
Writing, review, and/or revision of the manuscript: C. Lin, S. Zhang, C. Guo, E. Zhang
Administrative, technical, or material support (i.e., reporting or organizing data, constructing databases): H. Xu
The authors are very grateful to Nanjing General Hospital for providing the ESCC tissues and Prof. Yifeng Zhou (Suzhou University), who provided the ESCC cell lines; Project Program of State Key Laboratory of Natural Medicines in China (no. SKLNMBZ201403); National Science and Technology Major Projects of New Drugs in China (2012ZX09103301-004 and 2014ZX09508007); and The Priority Academic Program Development of Jiangsu Higher Education Institutions (PAPD). This work was supported by the National Natural Science Foundation of China (30873073).
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.