Long Noncoding RNA ELIT-1 Acts as a Smad3 Cofactor to Facilitate TGFβ/Smad Signaling and Promote Epithelial–Mesenchymal Transition

TGFβ family, including TGFβs, activins, and bone morphogenetic proteins, plays a pivotal role in diverse cellular processes, such as cell proliferation, differentiation, communication, adhesion, migration, metabolism, and cell death (1). Perturbation of signaling by TGFβ family members is often associated with a variety of disorders, such as malignancies, inflammatory conditions, and fibrotic conditions (2). During the process of tumorigenesis and malignant progression, TGFβ elicits both suppressive and promoting effects; at early stages of tumorigenesis, it acts as a tumor suppressor by inhibiting cell proliferation and stimulating apoptosis; at later stages, TGFβ becomes a tumor promoter via induction of epithelial–mesenchymal transition (EMT), which correlates with increased invasiveness, metastasis, and chemoresistance of tumor cells (2, 3). TGFβ can also facilitate tumor progression by suppressing the immune system and promoting angiogenesis (4). Considering the extensive role of TGFβ in late-stage tumor progression, several strategies to inhibit TGFβ signaling to combat malignant tumors have been proposed (e.g., small-molecule inhibitors of receptor kinases, TGFβ-neutralizing antibodies, and antisense compounds), based on a deep understanding of the TGFβ signaling pathway (5).

Binding of TGFβ induces assembly of a heterotetrameric complex composed of two TGFβ type I receptors (TβRI) and two TGFβ type II receptors (TβRII; ref. 6). In the complex, constitutively active TβRII phosphorylates and activates TβRI. Activated TβRI interacts with multiple proteins, thereby activating various downstream signaling pathways. In the canonical TGFβ signaling pathway, the activated TβRI phosphorylates receptor-activated Smad (R-Smad), Smad2 and Smad3, leading to heterotrimeric complex formation with Co-Smad, Smad4 (7, 8). This complex translocates into the nucleus, where it associates with regulatory regions of target genes through the consensus sequence containing the CAGA box (9) and regulates transcription through the recruitment of transcriptional coactivators and/or corepressors. The activated Smad complex usually cooperates with other DNA-binding protein factors, so-called Smad cofactors (e.g., FoxH1, p53, Ets1 etc.), to elicit specific transcriptional regulation. The collaboration between Smads and Smad cofactors is thought to mediate context-dependent signaling of TGFβ (5). However, lncRNA, which play a role as a Smad cofactor, has not been reported.

An increasing number of studies have demonstrated the importance of long noncoding RNAs (lncRNA), which are RNAs longer than 200 nucleotides that are not translated into protein. A recent study indicated that tens of thousands of lncRNAs are expressed in human cells. However, functional identifications have been reported with only a part of lncRNAs. LncRNAs are suggested to be involved in regulation of various biological processes, including development, differentiation, cell proliferation, cellular senescence, cell death, and cancer development (10, 11). Because lncRNAs are associated with both cause and progression in various diseases, they have received attention as useful targets for drug development (12). In particular, several lncRNAs related to the onset and progression of human cancers have been reported (13, 14). Moreover, lncRNAs are frequently dysregulated and associated with poor prognosis in multiple types of cancers. Therefore, investigations on lncRNA functions in cancer will lead to a better understanding of molecular mechanisms underlying cancer development and progression. Their reported molecular functions are very diverse, such as miRNA sponge, protein sponge, scaffolds, transcription regulator, and chromatin modulator (15). Some lncRNAs also work as recruiters for chromatin modifiers into DNA (10, 14).

Previous reports have shown that TGFβ induces several lncRNAs, including lincRNA-ATB, lincRNA-RoR, lncRNA-smad7, MEG3, HIT, and MALAT1, all of which modulate cellular responses elicited by TGFβ (16–24). Moreover, lincRNA-ATB, lincRNA-RoR, MEG3, HIT, H19, HOTAIR, ZEB1/2-AS1, and MALAT1 are reported to participate in EMT (17, 18). However, there has been no report about trans-acting lncRNA that promotes Smad-mediated transcription in TGFβ/Smad pathway as a Smad cofactor. Here, we identified a novel TGFβ-induced lncRNA ELIT-1 (EMT-associated lncRNA induced by TGFβ1), which assists Smad3-dependent transcriptional regulation. Mechanistically, ELIT-1 binds to Smad3, but not Smad2, and facilitates the binding of Smad3 to promoters of TGFβ target genes. Thus, ELIT-1 facilitates Smad-responsible promoter activities, in conjunction with Smad3. Our results suggest that ELIT-1 is a novel and crucial player in the canonical TGFβ signaling pathway, mediated by Smad3. Furthermore, ELIT-1 is the first trans-acting lncRNA that functions as a Smad3 cofactor and is involved in the promotion of EMT.

Huh7, A549, HepG2, HaCaT, and HEK293 cells were grown in DMEM (Sigma-Aldrich) supplemented with 10% FBS (Sigma-Aldrich) and penicillin/streptomycin. MDA-MB-231 cells were grown in Leibovitz L-15 medium (Gibco), supplemented with 10% FBS and penicillin/streptomycin. MCF10A cells were grown in mammary epithelium basal medium (Lonza) containing bovine pituitary extract, human EGF, human insulin, hydrocortisone, penicillin/streptomycin, and cholera toxin (Wako). All cells were maintained at 37°C in an atmosphere containing 5% CO₂. Huh7 cells were obtained from JCRB. A549, HEK293, MDA-MB-231, and MCF10A cells were obtained from ATCC. HepG2 cells were obtained from RCB. HaCaT cells were kindly provided by Dr. H. Hayashi (Nagoya City University, Nagoya, Japan).

The antibodies used in this study were as follows: anti-Smad2/3 (BD Biosciences, 610842), anti-phospho-Smad3 (Ser423/425; Cell Signaling Technology, #9520), anti-Smad3 (Abcam, ab28379 and Cell Signaling Technology, #9523), anti-Smad2 (Cell Signaling Technology, #5339), anti-Snail (Cell Signaling Technology, #4719), FITC mouse anti-E-cadherin (BD Biosciences), anti-HSP90 (BD Transduction Laboratories, 610418), anti-Lamin A/C (Santa Cruz Biotechnology, sc-7292), anti-FLAG (Sigma, M2), anti-c-myc (Roche, 9E10), normal mouse IgG (Cell Signaling Technology, #5415), normal rabbit IgG (Cell Signaling Technology, #2729), and anti-β-actin (Novus Biologicals, AC-15 and Wako, 017-24573). Phalloidin-TRITC was used to detect F-actin (Sigma-Aldrich).

Recombinant hTGFβ1 was purchased from R&D Systems (240-B). The inhibitors used in this study were as follows: SB431542 (Selleck Chemicals, S1067), SB203580 (Abcam, ab120162), and Smad3 inhibitor, SIS3 (Merck, 566405).

Human ELIT-1 cDNA was cloned into pcDNA3.1 (Invitrogen). A reporter construct containing -2526/+61 of the human ELIT-1 promoter linked to a luciferase reporter gene (pGL4-ELIT-1) was generated using PCR, then cloned into pGL4.10 (Promega). Substitutions of Ser⁴²³ and Ser⁴²⁵ to aspartic acid in Smad3 (Smad3-2SD) were generated using PCR-based mutagenesis and then cloned into pCMV5-FLAG. These plasmids were constructed using standard DNA techniques. pCMV-β-galactosidase (β-gal) was kindly provided by Dr. H. Hayashi (Nagoya City University, Nagoya, Japan).

Plasmids were transfected into HEK293, Huh7 or A549, HepG2, or HaCaT cells using the calcium phosphate method, Fugene6 (Promega, E2692), or ViaFect (Promega, E4981), or HilyMax (Dojindo, H357).

Hepatitis B virus (HBV)–replicating cells were prepared as follows: Huh7 cells were transfected with the episomal HBV expression vector, pEB-HBCe, where the 1.24-fold HBV genome derived from the viral genotype Ce (25) was inserted into pEBMulti-Puro (Wako), followed by maintaining in the medium in the presence of 5 μg/mL puromycin for 14 days. Total RNAs were prepared from the HBV-replicating cells and the parental control cells where the empty vector pEBMulti-Puro was transfected and subjected to microarray analysis. Microarray analyses with SurePrint G3 Human Gene Expression 8 × 60K v2 Microarrays (Agilent Technologies) were performed by Cell Innovator Co., Ltd. Our microarray dataset is deposited in NCBI Gene Expression Omnibus (GEO; GSE128965).

Huh7 cells were transfected with siRNAs (siCtrl, siELIT-1 #2, or siELIT-1 #3). At 24 hours after the first transfection, cells were transfected with siRNAs, again. At 6 hours after the second transfection, cells were treated with or without 10 ng/mL TGFβ for 48 hours. Total RNA was isolated from cultured cells using RNeasy Mini Kit (Qiagen, #74104), following the manufacturer's instructions. Gene expression was analyzed using the SurePrint G3 Human GE Microarray 8 × 60K v2 (Agilent Technologies), following the manufacturer's protocol by Cell Innovator Co., Ltd. Our microarray dataset is deposited in NCBI Gene Expression Omnibus (GEO; GSE129008). We classified target genes from microarray data. First, upregulated genes that were common to both TGFβ and ELIT-1 were classified. We selected genes that were increased by >2-fold by TGFβ, compared with the unstimulated condition; among those genes, we selected genes that were suppressed <0.5-fold by ELIT-1 depletion, compared with TGFβ stimulation. Next, genes that were increased by TGFβ were classified. We selected genes that were increased by >2-fold by TGFβ, compared with the unstimulated condition; among those genes, we selected genes that were increased >0.5-fold by ELIT-1 depletion, compared with TGFβ stimulation. Moreover, genes that were increased by ELIT-1 were classified. After depletion of control siRNA and ELIT-1 siRNA, we compared genes changed by TGFβ stimulation and selected genes whose expression decreased by <0.5-fold; among those genes, we selected genes that increased from 1-fold to <2-fold, compared with TGFβ stimulation and nonstimulation, after depletion of control siRNA.

In addition, suppressed genes that were common to both TGFβ and ELIT-1 were classified. We selected genes that were decreased by <0.5-fold by TGFβ, compared with the unstimulated condition; among those genes, we selected genes that were increased by >2-fold by ELIT-1-depletion, compared with TGFβ stimulation. Next, genes that were suppressed by TGFβ were classified. We selected genes that were decreased by <0.5-fold by TGFβ, compared with the unstimulated condition; among those genes, we selected genes that were increased by <2-fold by ELIT-1 depletion, compared with TGFβ stimulation. Finally, genes that were suppressed by ELIT-1 were classified. After depletion of control siRNA and ELIT-1 siRNA, we compared genes changed by TGFβ stimulation and selected genes whose expression increased by >2-fold; among those genes, we selected genes that were increased from 0.5-fold to <1-fold, compared with TGFβ stimulation and nonstimulation after depletion of control siRNA. We classified by gene symbol, excluding those without a gene symbol. Then, genes with the same gene symbol were counted as a single gene.

Cells were transfected with human ELIT-1 siRNA or control siRNA oligonucleotides using Lipofectamine RNAiMAX (Invitrogen), following the manufacturer's protocols. At 24 hours after transfection, cells were transfected again with ELIT-1 siRNA or control siRNA. At 6 hours after transfection, cells were stimulated with or without 10 (Huh7) or 5 (A549) ng/mL TGFβ for 72 or 48 hours, respectively. ELIT-1 siRNA and Negative Control siRNA (#1027310) or Smads siRNA were purchased from Qiagen or Sigma, respectively. The nucleotide sequence of siELIT-1-#2 was 5′-GGC CUA AUC CCG UCA UGA A-3′ (SI03681041), siELIT-1-#3 was 5′-CCG UUG GUU AGG AAU UCA A-3′ (SI3681048), siSmad2 was 5′-AAC AGG CCU UUA CAG CUU CUC-3′, and siSmad3 was 5′-AAG GCC AUC ACC ACG CAG AAC-3′ (Sigma Genosys) with a 3′-dTdT overhang.

Cells were lysed in lysis buffer (0.3% Triton X-100, 300 mmol/L NaCl, 50 mmol/L Tris-HCl, pH 7.5). Cell lysate was denatured by treatment with SDS sample buffer at 95°C for 8 minutes. Cell lysate was separated by SDS-PAGE and transferred onto a polyvinylidene difluoride (PVDF) membrane (Millipore), followed by immunoblotting with the indicated primary antibodies and corresponding secondary antibodies conjugated with horseradish peroxidase. Proteins were visualized using an enhanced chemiluminescence system (Perkin Elmer).

Total RNA was isolated from cultured cells using RNeasy Mini Kit. Reverse transcription was performed with random hexanucleotide primers and reverse transcriptase SuperScript II (Invitrogen, #18064014). PCR was performed using TaKaRa Ex Taq HS (TaKaRa, RR006Q). Primer sequences are shown in Supplementary Table S1. The PCR program was set with an initial melting step at 94°C for 3 minutes, then: (for ELIT-1 in Huh7 cells) 32 cycles of 94°C for 30 seconds, 56°C for 30 seconds, and 72°C for 15 seconds; (for ELIT-1 in A549 cells) 26 cycles of 94°C for 30 seconds, 56°C for 30 seconds, and 72°C for 15 seconds; (for GAPDH in Huh7 cells) 26 cycles of 94°C for 30 seconds, 56°C for 30 seconds, and 72°C for 15 seconds; (for GAPDH in A549 cells) 24 cycles of 94°C for 30 seconds, 56°C for 30 seconds, and 72°C for 15 seconds. PCR products were then analyzed on 3% agarose gel by electrophoresis.

Total RNA was isolated from cultured cells using RNeasy Mini Kit. Reverse transcription was performed with random hexanucleotide primers and reverse transcriptase SuperScript II. The resulting cDNA was subjected to real-time PCR using the StepOnePlus system (Applied Biosystems) and a SYBR Green Realtime PCR Master Mix (Toyobo Co., #QPS-201). Primer sequences are shown in Supplementary Table S1. Transcripts were normalized to GAPDH mRNA.

Huh7 or A549 cells cultured in 24-well plates (2.5 × 10⁴ cells per well) were transfected with the luciferase reporter plasmid, expression plasmids, β-gal expression plasmid, and empty vector using Fugene6 reagent or ViaFect reagent in Opti-MEM (Invitrogen). The total amount of transfected DNA was the same in each experiment. Cells were lysed at 48 hours after transfection and assayed for luciferase and β-gal activities. Luciferase activities were normalized by β-gal activities.

A549 or Huh7 cells (1.0 × 10⁷ cells) were fixed by 1% formaldehyde, and crosslinking was terminated by addition of 0.125 mol/L glycine. After washing cells twice with PBS, cells were lysed with lysis buffers LB1 and LB2 (see next sentence) for nuclear isolation. The formulas of LB1 and LB2 are as follows: LB1 [50 mmol/L HEPES/pH 7.5, 0.5% NP-40, 140 mmol/L NaCl, 1 mmol/L EDTA, 10% Glycerol, 0.25% Triton X-100, and protease inhibitor cocktail (Roche)]; LB2 (10 mmol/L Tris/pH 8, 200 mmol/L NaCl, 1.0 mmol/L EDTA, 0.5 mmol/L EGTA, and protease inhibitor cocktail). After collecting the isolated nuclei, chromatin was digested by Microccocal Nuclease (Cell Signaling Technology, #10011) with an optimized number of enzyme units; then, the nuclear membrane was destroyed by sonication (Bioruptor, COSMO-BIO). Chromatin was incubated with antibodies as follows: anti-Smad3 antibody (Cell Signaling Technology, #9523), anti-FLAG antibody (Sigma, M2), normal mouse IgG (Cell Signaling Technology, #5415), and normal rabbit IgG (CST, #2729). The antibody–chromatin complex was immunoprecipitated by Dynabeads protein G (Thermo Fisher Scientific, #10004D) and the precipitated chromatin was washed in accordance with standard chromatin immunoprecipitation (ChIP) procedure. The washed antibody–chromatin complex was incubated at 65°C for decrosslinking and ProteinaseK treatment; then, DNA was purified using Labo Pass PCR purification kit (Hokkaido System Science Co., Ltd. CMR0112). The purified DNA was subjected to ChIP-qPCR. Primer sequences are shown in Supplementary Table S2.

HEK293 cells (2.5 × 10⁶ cells per 100-mm dish) were transiently transfected with ELIT-1, FLAG-Smad3, FLAG-Smad2, and ALK5-T204D-HA. At 48 hours after transfection, cells were harvested by trypsinization and resuspended in lysis buffer [0.3% Triton X-100, 300 mmol/L NaCl, 50 mmol/L Tris-HCl, pH 7.5, protease inhibitor cocktail, RNase inhibitor (Toyobo)] on ice for 30 minutes. After centrifugation, the supernatant was collected as a whole-cell extract. A549 cells (2.5 × 10⁶ cells per 100-mm dish) were plated. At 48 hours after incubation, the cells were treated with or without 5 ng/mL TGFβ for 24 hours. Then, the cells were harvested by trypsinization and resuspended in lysis buffer (0.3% Triton X-100, 300 mmol/L NaCl, 50 mmol/L Tris-HCl, pH 7.5, protease inhibitor cocktail, RNase inhibitor) on ice for 30 minutes. After centrifugation, the supernatant was collected as whole-cell extract.

HEK293 cells (2.5 × 10⁶ cells) were transiently transfected with ELIT-1, FLAG-Smad3, FLAG-Smad2, and ALK5-T204D-HA. At 48 hours after transfection, cells were harvested by trypsinization and resuspended in 2 mL PBS with protease inhibitor cocktail and RNase inhibitor, 2 mL nuclear isolation buffer (1.28 mol/L sucrose, 40 mmol/L Tris/pH 7.5, 20 mmol/L MgCl₂, 4% Triton X-100) and 6 mL distilled water on ice for 20 minutes. After centrifugation, the supernatant was collected as cytoplasmic extract and the nuclear pellet was resuspended in RIP buffer (150 mmol/L KCl, 25 mmol/L Tris/pH7.4, 0.5 mmol/L DTT, 0.5% NP40, protease inhibitor cocktail, RNase inhibitor) on ice. Resuspended nuclei were mechanically sheared using a Dounce homogenizer with 30 strokes. After centrifugation, the supernatant was collected as nuclear extract. Immunoprecipitation was performed with an antibody specific to anti-Smad3 antibody (Abcam, ab28379), anti-Smad2 antibody (Cell Signaling Technology, #5339), anti-FLAG antibody (Sigma, M2); normal mouse IgG (Cell Signaling Technology, #5415) or normal rabbit IgG (Cell Signaling Technology, #2729) was used as a control for overnight at 4°C with gentle rotation. After immunoprecipitation, 40 μL of Dynabeads protein G (Thermo Fisher Scientific, #10004D) was added and incubated for 1.5 hours at 4°C with gentle rotation. Then, the supernatant was removed using a magnetic stand; the beads were washed three times with 500 μL of RIP buffer and then washed once with 1 mL of PBS. Purification of coprecipitated RNA from beads was performed using ISOGEN (Nippon Gene Co., LTD; #311-02501), following the manufacturer's protocols. Reverse transcription was performed with random hexanucleotide primers and reverse transcriptase SuperScript IV (Invitrogen, 18090050). The resulting cDNA was subjected to real-time PCR using the StepOnePlus system and a SYBR Green Realtime PCR Master Mix. Fold enrichment was calculated as a ratio to IgG control.

A549 cells (4 × 10⁵ cells per well in 12-well plates) were transfected with siRNAs (siCtrl or siELIT-1 #2). At 24 hours after transfection, cell monolayers were wounded by scratching with sterile plastic 200 μL micropipette tips and stimulated with 5 ng/mL TGFβ. Cells were photographed using phase-contrast microscopy: immediately, 12 hours, and 24 hours after wounding. The assay was independently performed in triplicate. The migration area was measured by graphic software Adobe Photoshop (Adobe).

A549 cells were transfected with siRNAs (siCtrl or siELIT-1 #2). At 24 hours after transfection, cells were stimulated with 5 ng/mL TGFβ for 48 hours. Then the cells (5 × 10⁴ cells per well) were seeded on Matrigel-coated Transwell (Corning), following manufacturer's instructions. After 18 hours, invaded cells were fixed by 4% paraformaldehyde (PFA), and stained with 0.1% crystal violet and photographed using phase-contrast microscopy.

Huh7 and A549 cells were transfected with siRNAs (siCtrl or siELIT-1 #2) in 4-well Lab Tek glass Chamber Slides (Nunc International Corp.). At 24 hours after transfection, Huh7 or A549 cells were treated with or without 10 ng/mL or 5 ng/mL TGFβ for 72 or 48 hours, respectively. Cells were fixed in 4% PFA for 10 minutes at room temperature. Then, cells were treated with ice-cold acetone for 2 minutes on ice and permeabilized with 0.5% Triton X-100 in PBS for 10 minutes at room temperature. Finally, cells were treated with 5% BSA and 0.1% Triton X-100 in PBS for 30 minutes at room temperature, and subsequently stained with FITC Mouse anti-E-cadherin (BD Biosciences, 1:250) and 0.23 mmol/L tetramethylrhodamine B isocyanate (TRITC)-conjugated phalloidin (Sigma-Aldrich). The cover glasses were mounted with VECTASHIELD Hard Set Mounting with DAPI (Vector Laboratories). Fluorescence was examined by confocal laser scanning microscopy (Leica).

Data are presented as mean ± SD or SEM. Comparisons between two groups were analyzed by Student t test, where P < 0.05 was considered as statistically significant.

We have studied regulatory mechanisms of HBV replication in host cells, and are interested in whether cellular lncRNAs participate in this process. Therefore, to identify lncRNAs induced by HBV replication, microarray analysis was performed, comparing human hepatoma Huh7-derived cells constitutively replicating HBV genome with its control cells expressing empty vector. Fourteen lncRNAs were upregulated in HBV-replicated Huh7 cells compared with its control (Supplementary Table S3). Importantly, the second highest lncRNA in the list was UCA1, which is reported to be upregulated by HBx and is involved in hepatocarcinogenesis (26), indicating that our system is well-optimized for screening of HBV-upregulated lncRNAs. Given that TGFβ signaling is positively regulated by HBx (27), we investigated lncRNAs associated with TGFβ signaling. In the upstream promoter region of LINC00842 (LOC643650), a lncRNA increased its expression by HBV replication, we found five putative CAGA boxes, which are target sites for the activated Smad complex (Fig. 1A). Indeed, LINC00842 was induced by TGFβ treatment in Huh7 cells and in human lung adenocarcinoma A549 cells (Fig. 1B). LINC00842 is a 54 kb gene located between ANXA8L1 [NC_000010.11 (46,375,590–46,391,784)] and NPY4R [NC_000010.11 (46,458,548-46,470,694, complement)] on human chromosome 10 (Fig. 1A). It consists of four exons and the length of transcript is a 2,872 nt. LINC00842 is expressed in liver, pancreas, lung, intestine, skin, placenta, embryonic tissue, heart, and uterus, according to the NCBI UniGENE database (https://www.ncbi.nlm.nih.gov/unigene). However, there is no information regarding its function. At 12 hours after TGFβ treatment, obvious LINC00842 induction was observed in Huh7 cells (Fig. 1C, left). In contrast, LINC00842 was induced earlier, beginning at 2 hours after TGFβ treatment in A549 cells (Fig. 1C, right). Although there were some differences in induction time points and induction levels in the other TGFβ-responsive cell lines such as HepG2, MDA-MB-231, HaCaT, and MCF10A, LINC00842 was also induced by TGFβ in all of the cell lines we analyzed (Fig. 1D).

HBV and HCV infections, obesity, and alcohol abuse are known to be major cause of chronic liver diseases, in which hepatocytes are damaged, resulting in healing and cellular remodeling. During these processes, hepatocytes progress to fibrosis, cirrhosis, and hepatocellular carcinoma. Moreover, TGFβ is required for liver fibrosis progression and EMT, which is related to liver fibrosis (28). Because LINC00842 was induced by TGFβ, we investigated whether LINC00842-depletion affected TGFβ-mediated EMT. Huh7 and A549 cells changed their original cobblestone-like cell morphology to a spindle-like shape cell upon TGFβ treatment; this is a typical phenotype of mesenchymal cells (Fig. 2A). Depletion of LINC00842 prominently inhibited spindle-like formation in both cell lines (Fig. 2A; Supplementary Fig. S1A and S1B). In TGFβ-induced EMT, localization of E-cadherin on the plasma membrane is perturbed and actin stress fiber formation is promoted as a typical mesenchymal characteristic. Depletion of LINC00842 inhibited perturbation of E-cadherin localization and actin stress fiber formation induced by TGFβ (Fig. 2B). TGFβ promoted suppression of E-cadherin mRNA and induction of N-cadherin and fibronectin mRNA, but these TGFβ-mediated EMT characters were canceled by LINC00842 depletion (Supplementary Fig. S2). In addition, cell migration of A549 cells in the presence of TGFβ was significantly inhibited by LINC00842 depletion (Fig. 2C). Moreover, invasion assays revealed that TGFβ-promoted invasive activity was inhibited by LINC00842 depletion in A549 cells (Fig. 2D). Taken together, LINC00842 may positively contribute to TGFβ-induced EMT, and we renamed LINC00842 as ELIT-1.

Given that Snail (Snail1/SNAI1) is a transcription factor important for the early step of EMT and early induced by TGFβ via the Smad pathway (29, 30), we investigated whether expression of Snail is affected by ELIT-1 depletion. In A549 cells, phospho-Smad3 (pSmad3) and Snail mRNA expression were observed at 1 hour after TGFβ stimulation (Supplementary Fig. S3A and S3B). Depletion of intrinsic ELIT-1 apparently inhibited induction of Snail mRNA in both A549 (Fig. 3A; Supplementary Fig. S4A) and MDA-MB-231 cells (Supplementary Fig. S4B). Moreover, ELIT-1 depletion also inhibited expression of Snail protein (Fig. 3B). Snail negatively regulates expression of E-cadherin during TGFβ-induced EMT (31). The decrease of E-cadherin expression during TGFβ-induced EMT was restored by ELIT-1 depletion (Fig. 3A), suggesting that ELIT-1 may participate in Snail-mediated E-cadherin suppression during EMT progression. Furthermore, we investigated a correlation between Snail and ELIT-1 expression using available online Gene Expression Omnibus (GEO) datasets (GSE22377) in NCBI (32). Snail mRNA expression is significantly correlated with ELIT-1 expression in human gastric adenocarcinomas (Fig. 3C). These results suggested that ELIT-1 participates in Snail expression in gastric cancers.

Given that EMT participates in metastasis, fibrosis, and drug resistance, which are involved in poor prognosis of human cancers, we investigated whether there is a correlation between ELIT-1 expression and prognosis in human gastric cancers using the available online database, Kaplan–Meier Plotter (http://kmplot.com/analysis/; ref. 33). In gastric cancers, the prognosis of patients with high expression of ELIT-1 was significantly poor (P = 0.012; Fig. 3D). In human lung cancers (34), the prognosis of patients with adenocarcinoma with high expression of ELIT-1 was also poor, compared with patients with low expression of ELIT-1 (P = 0.00014), whereas the prognosis of squamous cell carcinomas was not correlated with expression of ELIT-1 (Fig. 3E). These data suggest that high expression of ELIT-1 may be involved in malignancy of some types of human cancers.

Because TGFβ inhibits epithelial cell proliferation via p15^INK4B induction (35), we investigated whether ELIT-1 contributes to TGFβ-mediated inhibition of cell proliferation. We confirmed TGFβ-mediated growth inhibition in Huh7 cells, but ELIT-1 depletion scarcely affected it (Supplementary Fig. S5A). Microarray data indicated that ELIT-1 depletion inhibited TGFβ-induced expression of p15^INK4B, while TGFβ-mediated cell growth inhibition was not rescued by ELIT depletion (Supplementary Fig. S5B). Because it is suggested that TGFβ promotes apoptosis induction via non-Smad pathway (36), we sought to investigate the exact mechanism and found that TGFβ-induced apoptosis in Huh7 cells, while ELIT-1 depletion had no effect (Supplementary Fig. S5C). These data suggest that TGFβ inhibits cell proliferation in Huh7 cells through ELIT-1–independent apoptosis via non-Smad pathway.

Because ELIT-1 was involved in TGFβ-induced EMT, we further investigated the function of ELIT-1 in TGFβ-mediated gene regulation. We performed microarray analysis to analyze the effect of ELIT-1 depletion on gene expression profiles in Huh7 cells, with or without TGFβ treatment. We examined expression profiles of 1,711 probes and created a list of genes that were commonly or individually increased or suppressed by TGFβ and ELIT-1 depletion (Supplementary Tables S4–S6). Transcription of 640 genes (represented by 1,037 probes) was increased (magenta), and transcription of 394 genes (represented by 674 probes) was decreased (green), upon TGFβ treatment (Fig. 4A). A total of 20.9% of increased genes were suppressed by ELIT-1 depletion; 12.4% of decreased genes were restored by ELIT-1 depletion. Regarding TGFβ-induced genes, 134 were ELIT-1 target genes among 327 TGFβ target genes; 193 TGFβ target genes were not affected by ELIT-1 depletion (Fig. 4A). Conversely, regarding TGFβ-suppressed genes, 49 genes were ELIT-1 target genes among 138 TGFβ target genes; 89 TGFβ target genes were not affected by ELIT-1 depletion (Fig. 4A). These results suggested that ELIT-1 is involved in regulation of a subset of TGFβ target genes.

In addition, we analyzed the relationship between the TGFβ signaling pathway and ELIT-1 using the above microarray data for Kyoto Encyclopedia of Genes and Genomes (KEGG) Pathway analysis via the Database for Annotation, Visualization and Integrated Discovery (DAVID) version 6.8 (https://david.ncifcrf.gov/home.jsp). This analysis suggested that ELIT-1 is involved in the TGFβ signaling pathway (Fig. 4A; Supplementary Table S7). Moreover, it suggested that pathways involved in TGFβ signaling are also associated with ELIT-1 function (light purple).

We confirmed whether ELIT-1 contributes to regulate TGFβ target gene expression using qRT-PCR (Fig. 4B; Supplementary Fig. S2). Expression of E-cadherin, an epithelial marker, was suppressed by TGFβ; this suppression was recovered by ELIT-1 depletion. The expression of mesenchymal markers, such as vimentin, α1(I) collagen (COL1A) and N-cadherin, were increased by TGFβ; this increased expression was suppressed by ELIT-1 depletion. Moreover, the expression of other known TGFβ target genes, such as fibronectin, PAI-1, AFAP1L2, ANGPTL2, BHLHE40, ATF3, and COL4A1, were increased by TGFβ; this increased expression was suppressed by ELIT-1 depletion (Fig. 4B; Supplementary Fig. S6). These results suggested that ELIT-1 is closely associated with TGFβ signaling.

To clarify the mechanism of ELIT-1 induction by TGFβ, we examined the effects of chemical inhibitors of the TGFβ pathway on ELIT-1 induction. SB431542, a TβRI (ALK5) inhibitor, inhibited TGFβ-mediated ELIT-1 induction (Fig. 5A). Moreover, SIS3, a selective inhibitor of Smad3 but not Smad2 (37), inhibited ELIT-1 induction partially. This suggests that ELIT-1 induction is promoted not only via Smad3 but also via other factors such as Smad2. In fact, depletion of Smad3 as well as Smad2 inhibited ELIT-1 induction (Fig. 5B). We found that five putative Smad-binding CAGA boxes were located in the upstream promoter region of ELIT-1 (Figs. 1A and 5C), which indicated that Smads transcriptionally mediate ELIT-1 induction by TGFβ. We cloned the ELIT-1 promoter region and constructed a luciferase reporter plasmid, pGL4-ELIT-1 (Fig. 5C). ELIT-1 promoter activity was also inhibited by both SB431542 and SIS3, but not by SB203580, a p38 inhibitor (Fig. 5D). Alternatively, forced expression of Smad3, but weakly Smad2, resulted in enhanced ELIT-1 promoter activity (Fig. 5E). Coexpression of Smad4 enhanced both Smad2- and Smad3-mediated transcription. Furthermore, Smad3 bound to positive locus with CAGA box but not negative locus without CAGA box, in the PAI-1 promoter (Fig. 5F). Moreover, we found that Smad3 bound to each CAGA box in the ELIT-1 promoter. Interestingly, at 1.5 h after TGFβ stimulation, Smad3 binding was also found in the ELIT-1 promoter, as well as in known Smad target genes (Supplementary Fig. S7). These results suggested that ELIT-1 is transcriptionally induced by the canonical TGFβ/Smad pathway via CAGA boxes.

To address the molecular mechanism of ELIT-1–mediated transcription of target genes, we investigated whether ELIT-1 facilitated promoter activities of target genes using several reporter plasmids. Because TGFβ target genes, such as PAI-1 and ELIT-1, have putative CAGA boxes in their promoters, we investigated the effect of ELIT-1 depletion on the promoter activities of 3TP-Lux, which consisted of three TREs and the CAGA box-containing PAI-1 promoter, in A549 cells. Promoter activities of 3TP-Lux were enhanced by TGFβ and significantly suppressed by ELIT-1 depletion (Fig. 6A). Moreover, the same result was observed using (CAGA)₁₂-MLP-Luc, which was an artificial promoter containing twelve CAGA boxes (Fig. 6B), and 4xSBE-MLP-Luc, which was another artificial promoter containing four SBE (Smad-binding element) motifs (Fig. 6C). Next, we investigated the effects of forced expression of ELIT-1 on promoter activities using Huh7 cells, which scarcely express ELIT-1 without TGFβ. The promoter activities of both 3TP-Lux and (CAGA)₁₂-MLP-Luc were enhanced in Huh7 cells by transfection with Smad3-WT or Smad3-2SD (a constitutively active Smad3), but not Smad3-3SA (an inactive Smad3; Fig. 6D and E). These upregulated promoter activities were significantly enhanced by coexpression of ELIT-1. These results suggested that ELIT-1 facilitates activities of Smad-responsive promoters in conjunction with Smad3. Furthermore, we investigated the effect of ELIT-1 overexpression on the expression of PAI-1, which is a putative Smad target gene associated with EMT (38). We found that overexpression of ELIT-1 enhanced the PAI-1 promoter activity (Supplementary Fig. S8A), and increased its mRNA expression in Huh7 (Supplementary Fig. S8B), HepG2 (Supplementary Fig. S8C), and HaCaT (Supplementary Fig. S8D) cells. Obvious EMT-related phenotypes such as spindle-like shape formation were not observed by forced expression of ELIT-1. Facilitation of Smad3 pathway activation by ELIT-1 may not be sufficient for EMT induction. However, because overexpression of ELIT-1 enhanced PAI-1 expression in different cell lines, ELIT-1 alone is functionally involved in the regulation of endogenous Smad target gene expression to some extent.

The above results suggested that ELIT-1 helped Smad3 activate CAGA-containing promoters. To investigate how ELIT-1 aids in Smad3 function, we investigated whether ELIT-1 binds to Smad using an RNA immunoprecipitation assay (RIP assay). ELIT-1, FLAG-Smads, and ALK5-T204D-HA were transfected into HEK293 cells. Two days after transfection, cells were harvested and cell lysate was subjected to immunoprecipitation with anti-FLAG antibody or control IgG. Then, RNA was extracted from the immunoprecipitates and analyzed by qRT-PCR. We found that Smad3 bound to ELIT-1, but Smad2 did not (Fig. 6F). In addition, the Smad3–ELIT-1 complex was enriched in the nuclear fraction (Fig. 6G). Moreover, we confirmed that endogenous ELIT-1 bound to endogenous Smad3, but not Smad2, in TGFβ-stimulated A549 cells (Fig. 6H). These results suggested that ELIT-1 is involved in the TGFβ/Smad pathway via binding to Smad3.

To clarify the functional meaning of ELIT-1 binding to Smad3, we determine whether ELIT-1 is involved in heterodimer formation between activated Smad3 and Smad4 using immunoprecipitation following immunoblotting (IP-IB) assays. As shown in Supplementary Fig. S9A, ELIT-1 scarcely promoted heterodimer formation between Smad3 and Smad4. Subsequently, we investigated the effects of ELIT-1 depletion on the subcellular distribution and the phosphorylation of Smad3 in A549 cells (Supplementary Fig. S9B). pSmad3 was increased in both cytoplasmic and nuclear cellular extracts by TGFβ stimulation without ELIT-1 depletion, whereas pSmad3 was not increased in either extract by TGFβ stimulation combined with ELIT-1 depletion. This is due to the reduced Smad3 levels resulting from ELIT-1 depletion, suggesting that ELIT-1 affects neither the subcellular distribution nor the phosphorylation of Smad3. Next, we investigated the distribution of ELIT-1 and found that it was increased in both cytoplasm and nucleus after TGFβ stimulation, with the nucleus being the predominant location (Supplementary Fig. S9C). Moreover, RIP assays indicated that Smad3–ELIT-1 complexes, but not Smad2–ELIT-1 complexes, were predominantly increased in the nuclei by cotransfection of constitutively active TGFβ receptor (Fig. 6G), whereas ELIT-1 was present in both cytoplasm and nucleus. These results suggest that Smad3/4 complexes activated by TGFβ signaling translocate to nucleus and then ELIT-1 binds to Smad3/4 and facilitates the target gene transcription.

To clarify how ELIT-1 facilitates Smad3-dependent transcription, we investigated the roles of ELIT-1 in Smad3 binding to target gene promoters using ChIP-qPCR. PAI-1 Luc plasmid and FLAG-Smad3 were transfected into Huh7 cells, with or without ELIT-1. Then, specific binding of FLAG-Smad3 to endogenous and exogenous PAI-1 promoter was quantified by ChIP-qPCR, comparing immunoprecipitation between anti-FLAG and anti-IgG (control). We found that binding of overexpressed FLAG-Smad3 to the PAI-1 promoter was enhanced by coexpression of ELIT-1 (Fig. 7A, left). Importantly, ELIT-1 also enhanced FLAG-Smad3 binding to endogenous ELIT-1 and vimentin promoters in the same experiment (Fig. 7A, middle and right). Similar results were reproducibly confirmed in another experiment (Supplementary Fig. S10). Next, we investigated whether endogenous ELIT-1 participates in the recruitment of Smad3 to the target gene promoters using ChIP-qPCR. TGFβ stimulation enhanced Smad3 binding to the PAI-1 promoter, whereas ELIT-1 depletion significantly suppressed this binding (Fig. 7B). The same results were observed in different promoters, such as ELIT-1, vimentin, and Snail (Fig. 7C–E); these results strongly suggested that ELIT-1 participates in TGFβ-induced recruitment of Smad3 to the target gene promoters.

Taken together, our results suggest that ELIT-1 binds to Smad3 in nucleus upon TGFβ-stimulation, promoting the transcription of Smad-target genes by facilitating Smad3 recruitment to their respective promoters as a Smad3 cofactor (Fig. 7F). Because expression of ELIT-1 is promoted by ELIT-1 itself via the TGFβ/Smad3 pathway, ELIT-1 may contribute to TGFβ signaling through formation of an autostimulating loop.

TGFβ, a multifunctional cytokine, orchestrates an intricate signaling network to modulate tumorigenesis and cancer progression (39). When activated, TGFβ receptors phosphorylate Smad2/3 and transfer the activated Smad complex into the nucleus to regulate target gene transcription. Thus, TGFβ signaling through Smads is, conceptually, a simple and linear signaling pathway. Nevertheless, TGFβ induces highly complex programs of gene expression responses that are extensively regulated and dependent on physiologic context (40). Increasing evidence has revealed that the diversity of the TGFβ signaling response is determined by combinatorial usage of core pathway components, including ligands, receptors, Smads, and Smad-interacting proteins (e.g., transcriptional coactivators, corepressors, and Smad cofactors) (5). In this study, we identified a novel lncRNA, ELIT-1, which assists in Smad3-dependent transcriptional regulation. ELIT-1 binds to Smad3, but not Smad2, and facilitates CAGA box-containing promoter activities, in conjunction with Smad3. Smad3/4 complexes activated by TGFβ signaling translocate to nucleus and then ELIT-1 binds to Smad3/4 to facilitate recruitment of the complex to Smad-binding element of the target gene promoter. Therefore, ELIT-1 is a novel and crucial player in the canonical TGFβ signaling pathway, mediated by Smad3. Smad cofactors are a group of Smad-binding transcription factors that play a crucial role in recruiting Smad complexes to the specific target gene promoters, but a lncRNA that behaves as a Smad cofactor has thus far not been identified. This is the first report that ELIT-1 functions as a novel trans-acting lncRNA cofactor of Smads.

Moreover, ELIT-1 contributes to execution of TGFβ-mediated EMT. Depletion of ELIT-1 inhibited TGFβ-induced perturbation of E-cadherin location and actin stress fiber formation (Fig. 2B). In addition, ELIT-1 positively controls mRNA expression of TGFβ-induced mesenchymal marker genes, such as Snail, vimentin, N-cadherin, fibronectin, and PAI-1 (Fig. 4B; Supplementary Fig. S2). The promoter region of ELIT-1 has a CAGA box to which Smad3 binds; ELIT-1 helps Smad3 bind to its own promoter (Figs. 5C and F and 7A). ELIT-1 is ubiquitously expressed in human tissues, according to the NCBI UniGENE database. Therefore, ELIT-1 may be the core molecule of the canonical TGFβ signaling pathway in multiple human tissues; ELIT-1 may thus be involved in various pathologies associated with TGFβ. Moreover, ELIT-1 is transcriptionally induced by TGFβ/Smad pathway, the TGFβ/Smad–ELIT-1 axis forms an autostimulatory positive feedback system, which enhances the action of TGFβ. Thus, the TGFβ/Smad–ELIT-1 axis is considered to be a sophisticated system for both amplification and persistence of TGFβ signaling.

TGFβ-responsive cell lines, such as Huh7, A549, HepG2, MDA-MB-231, HaCaT, and MCF10A, all of which we tested, expressed ELIT-1. However, expression levels of ELIT-1 varied among these cells (Fig. 1D). Expression of ELIT-1 was low without TGFβ stimulation in Huh7 cells, was induced at 12 hours after TGFβ stimulation, and continued to increase for at least 72 hours. In A549 cells, in contrast, ELIT-1 expression was high prior to TGFβ stimulation, was early induced at 2 hours after TGFβ stimulation, and continued to increase for at least 24 hours (Fig. 1C). We here demonstrate that ELIT-1 promotes its self-transcription in a TGFβ/Smad pathway–dependent manner. Because of low intrinsic ELIT-1 in Huh7 cells, ELIT-1 scarcely contributes to its own Smad3-dependent transcription in the early phase. Later, when ELIT-1 has been expressed and accumulates in the cells, ELIT-1 may contribute to its self-transcription. In contrast, in A549 cells, intrinsic ELIT-1 may effectively participate in binding of the activated Smad3 complex to CAGA boxes in the ELIT-1 promoter, resulting in rapid induction of ELIT-1. The high level of intrinsic ELIT-1 may also contribute to early expression of Snail, the key transcription factor for EMT, because ELIT-1 depletion suppressed Snail expression, beginning at 1 hour after TGFβ stimulation, in both A549 and MDA-MB-231 cells (Fig. 3A; Supplementary Fig. S4A and S4B). It is known that expression of E-cadherin is suppressed by Snail (31), but ELIT-1 was considered to be involved in this regulation. Actually, expression of E-cadherin recovered when ELIT-1 was depleted in A549 cells (Fig. 3A). Therefore, cells expressing a high level of ELIT-1 may easily acquire mesenchymal characteristics through induction of Snail by TGFβ/Smad signaling. It has been reported ATF3 and BHLHE40 were involved in EMT of breast cancer (41) and pancreatic cancer (42), respectively. Depletion of ELIT-1 also suppressed expression of these genes, which were induced by 72 hours of TGFβ stimulation (Supplementary Fig. S6). We suspect that intrinsic ELIT-1 and accumulated ELIT-1 both contribute to the regulation of EMT-related genes, such as Snail, ATF3, and BHLHE40, via the TGFβ/Smad3–ELIT-1 axis; the positive-feedback-loop may orchestrate persistent EMT progression.

Some RNAs, such as miRNA and lncRNA, are involved in the TGFβ/Smad pathway. These RNAs are classified into three types by their functions, which include regulation of target genes in the TGFβ/Smad pathway. First, some RNAs regulate expression of signal transduction components, such as TβRI, TβRII, Smad2, and Smad3; these include miR-21, miR-148a, miR-99s, and MEG3 (16, 43). MALAT1 and lncRNA-ATB, involved in TGFβ1 and TGFβ2 production, respectively (44, 45). Second, some RNAs act as effector molecules; these include lincRNA-ATB, linc-RoR, MALAT1, and H19 are examples of miRNA sponges that act to prevent repression of ZEB expression by miRNAs, such as miR-200s and miR-205 (17). HIT, H19, UCA1, and HOTAIR, are known to repress E-cadherin expression and contribute to EMT promotion (17). Third, some RNAs act as functional RNA, which binds to Smad and regulates TGFβ/Smad signaling, includes ELIT-1 (Fig. 7F) and, possibly, DEANR1 and NORAD. It has been reported that the lncRNA DEANR1 promotes endoderm differentiation by transcriptional activation of FOXA2 (46). DEANR1 gene (LINC00261) is located downstream of FOXA2 gene; DEANR1 transcript binds to Smad2/3 and this complex associates with FOXA2 promoter to facilitate transcription cis-actingly. ELIT-1 and DEANR1 bind to Smad to facilitate the binding of the Smad complex to target gene promoters. However, there are important differences between DEANR1 and ELIT-1: DEANR1 is a definitive endoderm-specific gene that is involved in endoderm differentiation; DEANR1-Smad2/3 complexes exert the limited function as a cis-acting lncRNA to promote expression of FOXA2 gene nearby DEANR1 gene. Conversely, ELIT-1 binds to Smad3 and recruits Smad3 to Smad-binding motifs, such as the CAGA box, in multiple TGFβ target genes as a trans-acting lncRNA. Thus, ELIT-1 facilitates transcription of various TGFβ target genes as a Smad3 cofactor responding to TGFβ stimulation. Moreover, ELIT-1 itself is transcriptionally activated by the TGFβ/Smad3–ELIT-1 axis and forms a positive-feedback-loop to accelerate TGFβ signaling. A very recent study showed that NORAD, which had been reported as a PUMILIO-binding RNA involved in cancer metastasis, is involved in nuclear translocation of Smad3 (47). NORAD does not play as a Smad cofactor, whereas it binds to Smad3. Therefore ELIT-1, DEANR1, and NORAD comprise a third type of functional RNA, which binds to Smad3 and regulates TGFβ/Smad signaling. Here we identified a novel lncRNA ELIT-1 that is a trans-acting lncRNA that binds with Smad3 and behaves as a Smad3 cofactor. Notably, the TGFβ/Smad–ELIT-1 axis is considered very important for facilitating TGFβ signaling.

Our results suggest that Smad3/4 complexes activated by TGFβ signaling translocate to nucleus and then ELIT-1 binds to Smad3/4 to facilitate recruitment of the complex to Smad-binding elements of the target gene promoter. Further analysis is required to clarify the molecular mechanism(s) how ELIT-1 participates in the recruitment of Smad3 complex to the target gene promoters. There are at least three possibilities to address the question, as follows: first, ELIT-1 may change the conformation of Smad complexes and makes it easier to bind to the CAGA motif in the target gene promoter; second, ELIT-1 may enhance stability of Smad complexes in the nucleus; third, the intramolecular CAGA sequence in the RNA molecule of ELIT-1 may promote recruitment of Smad complexes to the CAGA motif in the target gene promoter. In addition, it has been reported that several lncRNAs exert their function via molecular interactions with RNA-binding proteins (RBP; ref. 48). RBPs contribute to Smad-mediated transcriptional activation (49, 50). Our current data strongly suggests that ELIT-1 promoted TGFβ signaling via interaction with Smad3, whereas another protein, possibly an RBP, may participate in ELIT-1–mediated facilitation of TGFβ/Smad3 signaling. To clarify this, further investigations may be required.

Pathway analysis data indicated that ten pathways, including the TGFβ signaling pathway itself, were significantly upregulated by TGFβ stimulation (Fig. 4A). As expected, the TGFβ signaling pathway was a target of both TGFβ and ELIT-1. Crosstalk has been reported between the TGFβ signaling pathway and the following pathways: MAPK, TNF, and FoxO (51–53). Associations of TGFβ with hypertrophic cardiomyopathy, dilated cardiomyopathy, proteoglycans in cancer, focal adhesion, HIF1 signaling pathway, and Rap1 signaling have also been reported (54–59). Therefore, the results of pathway analysis of TGFβ targets are appropriate, as these are similar to previous reports. Nine of ten pathways, except the HIF1 signaling pathway, were significantly upregulated by ELIT-1, suggesting that ELIT-1 function is closely related with TGFβ signaling. In addition, in a comparison of individual gene expression data, there were many target genes common to both TGFβ and ELIT-1. These results suggest ELIT-1 functions as a Smad cofactor. However, different target genes were also found. The following possibilities may contribute to this result: TGFβ target genes induced by TGFβ are mediated through the TGFβ/Smad pathway or non-Smad pathways. Because this experiment was a result of 48 hours of stimulation, secondary effects may be present. Because we could not rule out the possibility that ELIT-1 may have functions other than Smad3 recruiting, further studies may be required.

Using Kaplan–Meier Plotter, we found that there is a correlation between ELIT-1 expression and prognosis in gastric cancers and lung adenocarcinomas (Fig. 3D and E). The prognosis of both patients with gastric cancer and lung adenocarcinomas with high expression of ELIT-1 were poor, compared with patients with low expression of ELIT-1. Here, we showed that ELIT-1 depletion suppressed EMT, which is closely associated with tumor malignancy, including metastasis and chemoresistance. We also showed that ELIT-1 depletion suppressed cell migration and invasion. Therefore, poor prognosis of lung and gastric cancers with high expression of ELIT-1 may be caused by accelerated malignancy via ELIT-1–mediated EMT. Further studies are required; however, ELIT-1 may be useful to serve as a prognostic marker (i.e., a novel cancer-associated lncRNA). ELIT-1 may be also a novel therapeutic target for malignant cancers.

No potential conflicts of interest were disclosed.

Conception and design: S. Sakai, T. Ohhata, K. Miyazawa, M. Kitagawa

Development of methodology: C. Uchida, T. Suzuki

Acquisition of data (provided animals, acquired and managed patients, provided facilities, etc.): S. Sakai, T. Ohhata, T. Aoshima

Analysis and interpretation of data (e.g., statistical analysis, biostatistics, computational analysis): S. Sakai, T. Ohhata, K. Kitagawa

Writing, review, and/or revision of the manuscript: S. Sakai, T. Ohhata, K. Kitagawa, C. Uchida, H. Niida, T. Suzuki, Y. Inoue, K. Miyazawa, M. Kitagawa

Administrative, technical, or material support (i.e., reporting or organizing data, constructing databases): T. Suzuki, Y. Inoue, K. Miyazawa

Study supervision: T. Suzuki, Y. Inoue, K. Miyazawa, M. Kitagawa

We thank Dr. Masahiko Ito and Dr. Kenji Nakashima for HBV-related experiments; Dr. Haruhiko Sugimura, Dr. Tomoaki Kahyo, Dr. Masato Karayama, and Dr. Kosuke Shimizu for useful discussion; Dr. Ryo Horiguchi, Mr. Kosuke Ota, Ms. Kiyomi Kimpara, Ms. Mika Yoshida, and Ms. Hazuki Yokota for technical support. We also thank Ryan Chastain-Gross, Ph.D., from Edanz Group (www.edanzediting.com/ac) for editing a draft of this manuscript. This research was supported in part by grants from the Ministry of Education, Culture, Sports, Science and Technology of Japan. This research was partially supported by the Research Program on Hepatitis B from Japan Agency for Medical Research and Development (AMED), Princess Takamatsu Cancer Research Fund, and Hamamatsu University School of Medicine Grant-in-Aid.

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.