This article is mentioned in:

Introduction

At least 90% of the human genome is actively transcribed into non-coding RNAs (ncRNAs), whereas <2% of genome sequences encode proteins (1). Over the last 10 years, one type of ncRNA, namely microRNA (miRNA), has been increasingly investigated. miRNAs are ~21–23 nucleotides in length and are derived from long hairpin precursors. miRNAs are associated with Argonaute proteins, which post-transcriptionally regulate target genes, usually by binding to partially complementary sequences in the 3′-untranslated region (3′-UTR) of their mRNAs (2). The dysregulation of this process has been implicated in various biological disorders and human diseases.

Another category of ncRNAs, long non-coding RNAs (lncRNAs), has attracted considerable attention. The analysis of high-throughput RNA sequencing data has provided a robust platform for investigating transcriptomes and has led to the identification of a large number of lncRNAs (3). lncRNAs are defined as endogenous cellular RNA molecules >200 nucleotides in length, which resemble mRNAs but lack coding potential and show poor sequence conservation between species (4–6). Previous studies have indicated that lncRNAs function as signal, decoy, guide, or scaffold RNAs (7,8). However, the characteristics of lncRNAs require further investigation. For example, lncRNAs can guide cis- or trans-acting epigenetic-modifying complexes to distinct genomic loci and can control the formation and spread of heterochromatin domains, thereby activating or repressing transcriptional activity (8). Additionally, lncRNAs possibly function as adaptors that direct chromatin-remodeling complexes and transcription factors to specific chromatin loci. lncRNAs can also act as scaffolds that recruit multiple proteins simultaneously to coordinate their activities (9). Studies have shown that the aberrant expression of lncRNAs including BC200 (10,11), H19 (12–16), MALAT1 (17–20), UCA1/CUDR (21–23), HOTAIR (24–26) and GAS6 AS1 (27) can cause various types of human cancer. These lncRNAs modulate gene transcription and translation, and RNA processing and chromatin remodeling, and have considerable potential as biomarkers, targets and therapeutic agents (5).

One particular human pseudogene, PTENP1, has been reported to regulate its corresponding protein-coding mRNA, transcribed from the phosphatase and tensin homolog gene, by acting as a decoy for miRNAs that bind to similar sequences in their respective 3′-UTRs (28). Furthermore, our previous study showed that the human pseudogene ψPPM1K may generate endogenous small interfering (si)RNA to suppress oncogenic cell growth in hepatocellular carcinoma (29). These studies prompted the hypothesis that there are further lncRNAs that require identification. Consequently, the present study performed a genome-wide bioinformatics screen and subsequent biological validation experiments to test this hypothesis.

Materials and methods

Bioinformatics

A flowchart for identifying lncRNAs that potentially act as miRNA decoys is shown in Fig. 1. Over 20,000 human lncRNA transcripts were obtained from NONCODE v.4 (30). The sequences of miRNAs and 3′-UTRs of protein-coding genes were collected from miRBase 20 (31) and UCSC hg19 (32), respectively. Previous approaches were modified to assess whether lncRNAs behave as miRNA decoys (29). Briefly, the sequences of lncRNAs and 3′-UTRs of protein-coding genes were aligned, and those lncRNAs (the same orientation as coding genes) showing >90% identity, termed 3′-UTR-like lncRNAs, were collected. These sequences were then analyzed using miRNA target prediction tools, miRTar-Base 6.0 (33,34), TargetScan 7.0 (35–37), miRanda 3.3a (38), and RNAhybrid 2.1.2 (20). The miRNA was selected according to the score of four databases and the binding energy. The 3′-UTR-like lncRNAs were all transcribed in the same direction as their corresponding coding genes and thus had the potential to act as miRNA decoys through reversed expression between lncRNA and its corresponding coding gene. The expression profiles of these lncRNA/coding-gene pairs in 460 squamous cell carcinomas were obtained from The Cancer Genome Atlas (http://cancergenome.nih.gov/). To identify the functions of these corresponding coding genes, the Database for Annotation, Visualization and Integrated Discovery (DAVID) (39) annotation tool was used to assign Gene Ontology (GO) terms and to perform pathway analysis and promoter analysis.

Figure 1 Flowchart for the identification of lncRNAs that potentially act as miRNA decoys. (A) A set of 3′-UTR-related lncRNAs were identified by surveying the genome for sequence matches between the 3′-UTRs of protein-coding genes and lncRNAs. Subsequently, miRNA target sites in these matching regions were predicted using four miRNA target prediction tools. To determine that an lncRNA acts as an miRNA decoy, the reversed expression profiles of the lncRNA and its paired coding genes were collected. Finally, Gene Ontology, pathway analysis, and promoter analysis were applied to examine the functional annotation. Mapping results for 12 human lncRNA transcripts showing high matches with 3′-UTRs of protein-coding genes are shown. Most of these 12 lncRNAs were transcribed in the same direction as their corresponding genes and thus have a potential role as miRNA decoys. (B) Sense transcriptions of lncRNAs and their corresponding genes. (C) Antisense transcriptions of lncRNAs and their corresponding genes. lncRNAs, long non-coding RNAs; 3′-UTR, 3′-untranslated region; miRNA microRNA.

Clinical specimens

Resected primary tumor and adjacent non-tumor tissue samples were obtained from 83 patients with NSCLC, 30 patients with kidney cancer, 30 patients with colon cancer, 30 patients with oral cancer, and 30 patients with liver cancer at either China Medical University Hospital (Taichung, Taiwan; CMUH) or Changhua Christian Hospital (Changhua, Taiwan; CCH). The tumor tissues were composed of 90–100% cancer cells. They were frozen immediately following surgical resection and then stored in liquid nitrogen until extraction of either RNA or DNA. All participants provided their written informed consent to participate in the study. The institutional review boards of CMUH and CCH approved the study (CMUH102-RECJ-015 and CCH-IRB-080322) and the consent procedure. The baseline characteristics of the NSCLC population in the present study are summarized in Table I.

Table I Clinical variables and relative gene expression of the long non-coding RNA CR933609 in non-small cell lung cancer.

Table I

Clinical variables and relative gene expression of the long non-coding RNA CR933609 in non-small cell lung cancer.

Characteristic	CR933609		P-value	INO80D		P-value
	n	ΔCqa		n	ΔCqa
Age (years)
<65	33	5.35±1.56	0.469	33	5.91±1.31	0.129
≥65	50	5.07±1.73		50	5.38±1.65
Sex
Men	48	5.43±1.79	0.105	48	5.68±1.52	0.554
Women	35	4.84±1.42		35	5.47±1.56
Smoking
Never	46	4.96±1.45	0.197	46	5.51±1.59	0.607
Current/ex-smoker	37	5.45±1.87		37	5.69±1.48
ECOG PS
0	38	4.99±1.50	0.336	38	5.36±1.59	0.213
1	45	5.34±1.78		45	5.78±1.48
Histology
Adenocarcinoma	51	4.66±1.37	0.001	51	5.40±1.42	0.165
Squamous cell carcinoma	32	6.01±1.77		32	5.89±1.68
Grade
Well/moderate	72	5.19±1.66	0.925	72	5.53±1.58	0.351
Poor differentiation	11	5.14±1.73		11	5.99±1.18
Stage
I+II	65	5.19±1.67	0.973	65	5.51±1.59	0.350
III+IV	18	5.17±1.67		18	5.89±1.31

a Data shown as mean ± standard deviation. ECOG PS, Eastern Cooperative Oncology Group Performance Status; INO80D, INO80 complex subunit D.

Reverse transcription quantitative polymerase chain reaction (RT qPCR) analysis

Total RNA was extracted using the REzol reagent (Protech Technology Enterprise Co., Ltd., Taipei, Taiwan) according to the manufacturer's protocol and treated with RQ1 RNase-Free DNase (Promega Corporation, Madison, WI, USA). cDNA was generated from total RNA using the High-Capacity cDNA Reverse Transcription kit (Applied Biosystems; Thermo Fisher Scientific, Inc., Waltham, MA, USA). The RT-qPCR analysis was performed in a total reaction volume of 10 μl, including 1 μl of cDNA, 5 μl of LightCycler FastStart DNA Master HybProbe (Roche Diagnostics, Indianapolis, IN, USA), 0.6 μl of mixed primers (Table II), 1.2 μl of universal probes (Table II) (Roche Diagnostics), and 2.2 μl of double-distilled water. A LightCycler 480 (Roche Diagnostics) was used with the following program: Preincubation for 10 min at 95°C; followed by 50 cycles of 10 sec at 95°C, 30 sec at 60°C, and 1 sec at 72°C; and finally cooling to 4°C. Glyceraldehyde-3- phosphate dehydrogenase (GAPDH) served as an internal reference. The relative RNA levels were calculated using the comparative quantification cycle method (39). The RT-qPCR product was verified through Sanger sequencing to ensure the primers specificity. Different target sites were verified to ensure primer specificity. INO80 Complex Subunit D (INO80D): Primer target on exon 5-exon 6 (ΔCq, 9.92±0.33), primer target on exon 11 used in the present study (ΔCq, 9.45±0.36), primer target on exon 11 (ΔCq, 9.41±0.27). CR933609: primer used in the present study (ΔCq, 7.38±0.16), primer target on other site (ΔCq, 7.86±0.61).

Table II Primer sequences.

Table II

Primer sequences.

Name	Forward primer (5′-3′)	Reverse primer (5′-3′)	Probe no.
NONHSAT143029	ggccaggtctgggagataag	atagcagacaggtggggtgt	38
CMTM4	tgcggcatatgcagtgaa	gctcggatgtagtcattggtg	72
NONHSAT095816	cctgcaaatgtccagcacta	aaacttggctgtgttttccttc	21
CCDC149	ctcacctggactccttcgag	atccctttgccgtcttcc	66
NONHSAT085150	ccaatggaaaatctgggttc	caggtaagtgcagctcagga	84
RAC2	cccagcctcttatgagaacg	gtgcccaccaggatgatg	43
NONHSAT026159	accccaccatcctcttcatt	aacactgtgcctccctgaac	75
LEPREL2	cagggtgctaagctgcttct	caggtgggtgaaggacagat	7
NONHSAT118794	gcggttcctctgttgcag	cagagccagagagcacagg	4
EIF3B	ggtggacactgacgagctg	gacgaagaactcaatggtctcc	48
NONHSAT105574	tggaagcctttgggaagag	cttcagcggaagacaggtg	33
ADAMTS2	cctatgactgcctgctggat	attgctcgttcatggagtagtg	43
NONHSAT028817	gctgacctatgataaggttgagtattt	caataaataatcctccccaaactg	79
IL23A	cttctccgcttcaaaatcctt	tgctccatgggcaaagac	5
NONHSAT138672	ctggcaaccaaatcgaatct	aaaggcacccatcagagaga	2
FAM127B	gaaggtgacgttcctcatcac	cccggtaatcattgagcag	12
NONHSAT055977	caccatgtcctcctccctta	ggtcacctgggacttcagg	43
EXOC7	tgcacaagcagacggaga	gcaggacagcgtcttctca	2
NONHSAT098146	aagagctccctgctgtgtg	tggcatacaaatgcagctaaag	8
ANXA5	tctgtttggcagggatcttc	tcataaagccgagagggtttc	21
NONHSAT125456	gagcagtgaatgggatcgtc	ctgggagttttgcaacagttt	4
FAM160B2	aggactgctcccacgatg	agctcctcggtctcagca	5
CR933609	gacaaaacaaactagtgaagcacct	tatacaccttgacacggcaga	47
INO80D	cctgatgacttacaagattttgattt	ctcctcagcctcttcggtag	52
GAPDH	agccacatcgctcagacac	gcccaatacgaccaaatcc	60

[i] Probe numbers were obtained from Universal Probe Library (Roche Diagnostics, Indianapolis, IN, USA).

Northern blots

Total RNA was extracted from cells by using REzol C&T RNA extraction reagent (Protech Technology Enterprise Co., Ltd.). Following extraction, 15 μg of total RNA was dissolved in a loading buffer containing 10 mM EDTA (pH 8.0), 96% (v/v) formamide, 0.01% xylene cyanol, and 0.01% bromophenol blue, heated at 95°C for 5 min, loaded onto a 1% MOPS 3-(N-morpholino) propanesulfonic acid gel with 2% formamide, separated for 2 h at 150 V, and then transferred onto nitrocellulose membranes (Pall Corporation, East Hills, NY, USA). The subsequently underwent cross-linking with UV irradiation for 5 min. The membranes with RNA blots were prehybridized at 42°C for 3 h using digoxigenin (DIG) Easy Hyb Granules (Roche Diagnostics) and subjected to hybridization with a DIG probe for the lncRNA NONHSAT098146 (0.1 μg/ml) overnight at 42°C. Following hybridization, the membranes were rinsed and then washed sequentially with 2X SSC/0.1% sodium dodecyl sulfate (SDS) and 0.1X SSC/0.1% SDS at 42°C. Detection was performed using the DIG Northern Starter kit (Roche Diagnostics) according to the manufacturer's protocol. The blots were detected with a chemiluminescent substrate (CDP-Star) and visualized using the ChemiDoc-It imaging system (UVP, Inc., Upland, CA, USA).

Cell culture, and siRNA and short hairpin (sh)RNA transfec tion

Cells from the A549, H441, and H520 NSCLC lines [Bioresource Collection and Research Center (BCRC), Hsinchu, Taiwan] were cultured in Dulbecco's modified Eagle's medium (DMEM; Gibco; Thermo Fisher Scientific,) supplemented with 10% fetal bovine serum (FBS; Gibco; Thermo Fisher Scientific, Inc.) at 37°C in a 5% CO2 atmosphere. The oligonucleotides used for three siRNA and INO80D/annexin A5 (ANXA5) shRNA constructs are shown in Table III. The INO80D and ANXA5 shRNAs were obtained from the National Research Program for Biopharmaceuticals (Taipei, Taiwan). The A549 cells were transfected with the shRNAs (6 μg of INO80D shRNA in 2×10 A549 cells/ml) using Lipofectamine 2000 (Thermo Fisher Scientific, Inc.). The efficiency of transfection was determined using RT-qPCR analysis.

Table III Oligonucleotides for siRNA and shRNA constructs.

Table III

Oligonucleotides for siRNA and shRNA constructs.

Symbol	Ref ID	Target sequence (5′-3′)	Note
siRNA1	FR273090	GGGCACGTGTGTGAGAGATATGAAT	Common region with CR933609 and INO80D
siRNA2	FR273090	CAGACATGAGCTGTGTTTATCTATT	Common region with CR933609 and INO80D
siRNA3	FR273090	GGGATCATTTCACCGTGCATATTTC	Common region with CR933609 and INO80D
shRNA1	NM_017759	GCACTTACTTTCAGCAGAAAT	INO80D-specific
shRNA2	NM_017759	CCATTTGCTTTCAATGAGGAA	INO80D-specific
shRNA3	NM_017759	CAACATCGTACTCTGGTGATA	INO80D-specific
shRNA4	NM_017759	CTATTGAATGGGCGTATAGTA	INO80D-specific
shRNA5	NM_017759	GAGCTATTGAATGGGCGTATA	INO80D-specific
NONHSAT098146	NONHSAT098146	GUGUACUUAAUGUUACUAAdTdT	Common region with NONHSAT098146 and ANXA5
ANXA5-homo-531	NM_001154	GAGCCAUCAAACAAGUUUATT	ANXA5-specific
ANXA5-homo-993	NM_001154	GUGAGAUUGAUCUGUUUAATT	ANXA5-specific
ANXA5-homo-844	NM_001154	CGAGACUUCUGGCAAUUUATT	ANXA5-specific

[i] siRNA, small interfering RNA; shRNA, short hairpin RNA; INO80D, INO80 complex subunit D; ANXA5, annexin A5.

Establishment of CR933609 overexpressing and NONHSAT098146 overexpressing stable cell lines

The cDNA from the lncRNAs CR933609 and NONHSAT098146 was cloned in pCDNA3.0 (Invitrogen; Thermo Fisher Scientific, Inc.). The A549 cells (2×106) were transfected with CR933609-overexpressing, NONHSAT098146-overexpressing, and control vectors using Lipofectamine 2000 (Thermo Fisher Scientific, Inc.). Following 48 h of recovery, the transfected cells were cultured in a medium containing 500 μg/ml G418 for 2–3 weeks. Clones representing stable cell lines were then selected and expanded to large-scale cultures. The RNA extracts prepared from these clones were analyzed for target RNA levels using RT-qPCR analysis.

Cell proliferation assay

To examine whether lncRNA knockdown altered the viability of NSCLC cells, cell proliferation assays were performed. Briefly, 10,000 cells from the NSCLC cell lines, stable transfectants subjected to lncRNA knockdown or parental NSCLC cell lines, were cultured in DMEM with 10% FBS (according to the BCRC). After 72 h, cell proliferation and viability were assessed using the 3-(4,5-dimethylthiazol-2-yl)-2,5-diphenyl-tetrazolium bromide assay. All experiments were performed in triplicate.

Wound healing assay

A wound-healing assay was conducted to assess cell migration. The basic steps involved creating a 'wound' in a cell monolayer, capturing images at the beginning and at regular intervals as cells migrated into the wound, and comparing the images to quantify the cell migration rate. For each assay, 2×105 cells were plated in 48-well plates (Corning Costar, Schiphol-Rijk, The Netherlands) with serum-free medium. A 200-μl plastic pipette tip was drawn across the center of the well to produce a clean wound of ~1 mm in width in the cell monolayer.

miRNA mediated knockdown of CR933609 and INO80D

A stable negative control (siCon: 5′-FAM-UUCUCCGAACGUGUCACGUTT-3′) and has-miR-5096 were purchased from GeneDireX, Inc. (Las Vegas, NV, USA) and transfected into cells using Lipofectamine RNAiMax (Invitrogen; Thermo Fisher Scientific, Inc.) according to the manufacturer's protocol. The efficacy of mRNA knockdown following miRNA transfection for 48 h was determined using RT-qPCR analysis.

Luciferase reporter assay

Fragments of DNA from the lncRNA CR933609 promoter region were obtained through PCR amplification and cloned into the pGL3-basic reporter vector (Promega Corporation). The constructs were transfected into A549 cells (2×106) using Lipofectamine 2000 (Thermo Fisher Scientific, Inc.). The cells were lysed and assayed for luciferase activity using the Steady-Glo Luciferase assay system (Promega Corporation) according to the manufacturer's protocol at 24 h post-transfection.

Statistical analysis

All data are presented as the mean ± standard deviation or as percentages. Statistical analyses, including the χ2 test, independent t-test, one-way analysis of variance followed by Fisher's least significant difference post hoc test, and paired t-test, were performed using SPSS version 17.0 software (SPSS, Inc., Chicago, IL, USA. P<0.05 was considered to indicate a statistically significant difference.

Results

Overview of lncRNAs

The bioinformatics analysis (Fig. 1A) showed that 12 human lncRNA transcripts contained sequences with significant matches to the 3′-UTRs of protein-coding genes (Table IV). These lncRNAs were all transcribed in the same direction as their corresponding coding genes and thus have the potential to act as miRNA decoys (Fig. 1B and C). To identify the functions of these protein-coding genes, the annotation tool DAVID (40) was used to assign GO terms. Certain GO biological process terms that were significantly associated with the coding genes, including regulation of translational initiation [eukaryotic translation initiation factor (EIF)4G2 and EIF3B; P<0.03] and response to chemical stimulus [ras-related C3 botulinum toxin substrate 2, transporter 2, ATP-binding cassette, subfamily B (TAP2), RNA binding motif protein 4, ANXA5 and CKLF-like MARVEL trans-membrane domain containing 4 (CMTM4); P<0.03], and molecular-function terms, including receptor binding [interleukin 23 α subunit p19 (IL23A), TAP2, relaxin 2, ANXA5 and CMTM4; P<0.01] (data not shown).

Table IV Basic data of 12 human lncRNA transcripts with high homology with 3′-UTRs of protein-coding genes.

Table IV

Basic data of 12 human lncRNA transcripts with high homology with 3′-UTRs of protein-coding genes.

No.	Transcript ID	Type	Chromosome	Start site	End site	Strand	Exon	Length	3′-UTR similarity	ncRNA similarity
1	CR933609	lncRNA	2	206,858,445	206,867,214	−	1	8,811	0.99	1
	INO80D	Gene	2	206,858,445	206,950,906	−	11	92,462
2	NONHSAT143029	lncRNA	16	66,648,652	66,655,678	−	1	7,026	0.96	1
	CMTM4	Gene	16	66,648,653	66,730,610	−	5	81,958
3	NONHSAT105574	ncRNA	5	178,537,853	178,540,736	−	1	1,492	0.96	1
	ADAMTS2	Gene	5	178,537,852	178,772,431	−	22	234,580
4	NONHSAT095816	lncRNA	4	24,807,739	24,809,957	−	1	2,218	0.98	1
	CCDC149	Gene	4	24,807,739	24,981,826	−	13	106,862
5	NONHSAT125456	lncRNA	8	21,960,467	21,961,890	+	1	1,423	0.98	1
	FAM160B2	Gene	8	21,958,951	21,961,891	+	4	15,227
6	NONHSAT085150	lncRNA	22	37,621,303	37,622,035	−	1	732	0.94	1
	RAC2	Gene	22	37,621,310	37,640,305	−	7	19,039
7	NONHSAT118794	lncRNA	7	2,419,855	2,420,380	+	1	525	0.94	0.99
	EIF3B	Gene	7	2,394,474	2,420,377	+	19	25,940
8	NONHSAT098146	lncRNA	4	122,589,152	122,589,620	−	1	468	0.99	1
	ANXA5	Gene	4	122,589,152	122,618,147	−	13	28,996
9	NONHSAT138672	lncRNA	X	134,184,962	134,185,301	−	1	339	0.9	1
	FAM127B	Gene	X	134,184,963	134,186,221	−	2	1,259
10	NONHSAT028817	lncRNA	12	56,733,915	56,734,193	+	1	278	0.91	1
	IL23A	Gene	12	56,732,663	56,734,194	+	4	1,532
11	NONHSAT026159	lncRNA	12	6,948,604	6,949,010	+	1	408	0.99	1
	LEPREL2	Gene	12	6,937,538	6,949,018	−	5	11,482
12	NONHSAT055977	lncRNA	17	74,097,193	74,099,845	+	1	2,652	0.99	1
	EXOC7	Gene	17	74,077,086	74,099,868	−	20	40,584

[i] 3′-UTR, 3′-untranslated region; lncRNA, long non-coding RNA.

Expression profiles of lncRNAs and parental coding genes in NSCLC

The average (Fig. 2) expression levels of lncRNAs and parental genes were quantitatively analyzed through RT-qPCR analysis. The results showed that the expression levels of lncRNAs and parental genes were lower in cancer tissues than in non-tumor tissues. These results were summarized from observations at either CMUH or CCH. The expression levels of lncRNAs in a small number of samples were undetectable, which is the reason for the difference in the number of samples. The expression levels of INO80D and the 3′-UTR-like lncRNA CR933609 were significantly lower in the cancerous tissues (P<0.01). The expression levels of ANXA5 and the 3′-UTR-like lncRNA NONHSAT098146 showed the most marked change. INO80D is involved in the transcription-coupled nucleotide excision repair and DNA double-strand break (DSB) repair pathways. In addition, DNA DSBs contribute to the genomic instability driving cancer development. Therefore, INO80D and CR933609 were selected for the subsequent investigations. The INO80D and the 3′-UTR-like lncRNA CR933609 pair, and the ANXA5 and the 3′-UTR-like lncRNA NONHSAT098146 pair were selected for further examination.

Figure 2 Expression levels of lncRNAs and parental genes in non-small cell lung cancer. The expression levels of lncRNAs and parental genes were quantitatively analyzed using reverse transcription-quantitative polymerase chain reaction analysis. The internal control was GAPDH. The white bar denotes the normal region, and the black bar denotes the tumor region. Error bars represent the standard deviation. Student's t-test was performed to evaluate differences between the groups. *P<0.05, **P<0.01 and ***P<0.001, compared with normal region. The results are summarized from observations at either CMUH or CCH. lncRNAs, long non-coding RNAs; GAPDH, glyceraldehyde-3-phosphate dehydrogenase; CMUH, China Medical University Hospital; CCH, Changhua Christian Hospital.

Association between lncRNAs and parental coding genes

INO80 Complex Subunit D (INO80D), a protein-coding gene located on chromosome 2q33.3, is involved in transcriptional regulation and possibly DNA repair (41). The computational results showed that the sequence of the lncRNA CR933609, which maps to the same region of chromosome 2, is 99% identical to the region of 8,762 nucleotides in the 3′-UTR of INO80D (Fig. 3A, top). This result suggested that the CR933609 transcriptional unit is embedded in the INO80D gene.

Figure 3 Associations between INO80D and CR933609, and between ANXA5 and NONHSAT098146. (A) INO80D is a protein-coding gene located on chromosome 2q33.3. Computational results showed that the lncRNA CR933609 matches the 3′-UTR of INO80D with 99% identity. To examine whether CR933609 acts as an miRNA decoy, four miRNA target prediction tools were used to investigate potential miRNA interactions with INO80D and CR933609. miRNA-5096 was found to be a likely candidate for interaction with the common regions of CR933609 and INO80D. (B) ANXA5, a protein-coding gene located on chromosome 4q27, is implicated in membrane-related events along exocytotic and endocytotic pathways. The computational results showed that the sequence of the lncRNA NONHSAT098146, which maps to the same region of chromosome 4, was 100% identical to a 468-nucleotide region in the 3′-UTR of ANXA5. To ascertain whether NONHSAT098146 acts as an miRNA decoy, four miRNA target prediction tools were used to investigate potential miRNA interactions with ANXA5 and NONHSAT098146. miRNA-4694-3p was found to be a likely candidate for interaction with the common regions of ANXA5 and NONHSAT098146. miRNA/miR, microRNA; lncRNA, long non-coding RNA; INO80D, INO80 complex subunit D; ANXA5, annexin A5.

ANXA5, a protein-coding gene located on chromosome 4q27, is implicated in membrane-related events along exocytotic and endocytotic pathways (42). The computational results showed that the sequence of the lncRNA NONHSAT098146, which maps to the same region of chromosome 4, is 100% identical to the region of 468 nucleotides in the 3′-UTR of ANXA5 (Fig. 3B, top). This result suggested that the NONHSAT098146 transcriptional unit is embedded in the ANXA5 gene.

To examine whether CR933609 and NONHSAT098146 act as miRNA decoys, and thereby modulate miRNA regulation in INO80D and ANXA5, four miRNA target prediction tools were used to investigate possible target sequences within the INO80D and CR933609 transcripts and the ANXA5 and NONHSAT098146 transcripts. The results indicated that miRNA-5096 interacts with a sequence within the common regions of CR933609 and INO80D (Fig. 3A, bottom), and miRNA-4694-3p potentially interacts with a sequence within the common regions of NONHSAT098146 and ANXA5 (Fig. 3B, bottom). The microRNA, hsa-mir-5096, was selected as a candidate according to the score of four databases and the lowest binding energy (−16.34 kCal/mol), and so was miRNA-4694-3p (-34.49 kCal/mol).

Expression levels of selected lncRNAs and parental coding genes in five types of solid tumor

To determine the expression levels of the selected lncRNAs and parental genes in cancer, RT-qPCR analysis was used to measure the mRNA expression levels of CR933609 and INO80D in five types of solid tumor. The CR933609 and INO80D pair had significantly lower mRNA expression levels in the tumors of lung, colon and kidney cancer, compared with their adjacent non-tumor tissues (Fig. 4A). The NONHSAT098146 and ANXA5 pair exhibited significantly lower mRNA expression levels in the tumors of lung, colon, and kidney cancer, compared with their adjacent non-tumor tissues (Fig. 4B). Notably, the NONHSAT098146 and ANXA5 pair had significantly higher mRNA expression levels in mucosal tumors. It was also found that, in individual tumors and in A549 NSCLC cells, the expression levels of CR933609 were higher than the expression levels of INO80D (Fig. 4C). However, the expression levels of NONHSAT098146 were lower than the expression levels of ANXA5 (Fig. 4D). As lung cancer is the leading cause of cancer-associated mortality worldwide, with NSCLC being the most prominent subgroup accounting for >80% of all lung cancer cases (43), the following experiments focused on NSCLC. Among types of NSCLC, the expression levels of CR933609 were significantly lower in squamous cell carcinoma than in adenocarcinoma (P=0.001; Table I).

Figure 4 Gene expression levels of CR933609 and INO80D, and NONHSAT098146 and ANXA5 are similar between various solid tumors and A549 cells. (A) Expression levels of CR933609 and INO80D genes were quantitatively analyzed by using RT-qPCR analysis. The internal control was GAPDH. The white bar denotes the normal region, the black bar denotes the tumor region. Error bars represent the standard deviation. Student's t-test was used to evaluate differences between the groups. *P<0.05 and ***P<0.001, compared with normal region. The results are summarized from observations at CMUH. (B) Gene expression levels of NONHSAT098146 and ANXA5 were quantitatively analyzed using RT-qPCR analysis. The internal control was GAPDH. The white bar denotes the normal region and the black bar denotes the tumor region. Error bars represent the standard deviation. Student's t-test was used to evaluate differences between the groups. *P<0.05, **P<0.01 and ***P<0.001, compared with normal region. Results are summarized from observations at CMUH. (C) Expression levels of CR933609 and INO80D genes were quantitatively analyzed using RT-qPCR analysis. The internal control was GAPDH. Error bars represent the standard deviation. Student's t-test was used to evaluate differences between the groups. *P<0.05, compared with the INO80D group. The results are summarized from observations of three independent experiments. (D) Gene expression levels of NONHSAT098146 and ANXA5 were quantitatively analyzed using RT-qPCR analysis. The internal control was GAPDH. Error bars represent the standard deviation. Student's t-test was used to evaluate differences between the groups. *P<0.05, compared with the ANXA5 group. The results are summarized from observations of three independent experiments. RT-qPCR, reverse transcription-quantitative polymerase chain reaction analysis; INO80D, INO80 complex subunit D; ANXA5, annexin A5; GAPDH, glyceraldehyde-3-phosphate dehydrogenase; CMUH, China Medical University Hospital; CCH, Changhua Christian Hospital.

Associations between the expression of CR933609 and INO80D, and expression of NONHSAT098146 and ANXA5 in NSCLC cells

To examine the association between CR933609 and INO80D, cells from three NSCLC lines, namely A549, H441 and H520 cells, were transfected with siRNAs (representing their common regions) and shRNAs (specifically targeted to INO80D). The siRNAs downregulated CR933609 and INO80D in NSCLC cells (Fig. 5A), whereas the cell proliferation assay revealed that the siRNA-transfected cells showed a marginally higher proliferation rate, compared with that in the control (Fig. 5B). Notably, the downregulation of INO80D resulted in decreased expression levels only in INO80D (Fig. 5A). These results suggested that CR933609 and INO80D have distinct transcripts. To determine whether CR933609 and INO80D have different promoters, a luciferase reporter assay was used to identify the promoter of CR933609. As shown in Fig. 6A, a significant increase in luciferase activity was detected in constructs containing DNA fragments extending between −3602 and 0, −3602 and −3207, and −3602 and −3049 nucleotides upstream of the CR933609 transcription initiation site, indicating a possible location of the CR933609 promoter in the region between −3602 and −3360 nucleotides (Fig. 6B). Although the CR933609 sequence is embedded within the INO80D gene, it is transcribed from a different promoter.

Figure 5 Expression levels of CR933609 and INO80D decrease in CR933609-KD NSCLC cells but not in INO80D-KD cells. (A) Total RNA was isolated and subjected to reverse transcription-quantitative polymerase chain reaction analysis to detect RNA expression levels of CR933609 (white) and INO80D (black) in NSCLC cells. The fold change was standardized using mRNA expression levels of GAPDH, relative to the mock group. (B) Cell proliferation was detected using the MTT assay between 24 and 72 h. The optical density of the experiment was standardized using the optical density of empty wells. Error bars represent the standard deviation. A Student's t-test was used to evaluate differences between the groups. *P<0.05, compared with the mock group. The results are summarized from observations of three independent experiments. NSCLC, non-small cell lung cancer; INO80D, INO80 complex subunit D; GAPDH, glyceraldehyde-3-phosphate dehydrogenase; KD, knockdown.

Figure 6 lncRNA CR933609 has different promoters to the INO80D gene. (A) Activity of promoters determined using a luciferase assay upstream of CR933609. The human cytomegalovirus promoter was a positive control. Mock cells were transfected with luciferase only. Error bars represent the standard deviation. Student's t-test was used to evaluate differences with respect to the mock group. *P<0.05, compared with the mock group. The results are summarized from observations of three independent experiments. (B) Diagram showing the location of the CR933609 promoter. lncRNA, long non-coding RNA; UTR, untranslated region; INO80D, INO80 complex subunit D.

To examine the association between NONHSAT098146 and ANXA5, the A549 cells were transfected with artificial siRNAs (representing their common regions) and shRNAs (specifically targeted to ANXA5). NONHSAT098146 and ANXA5 were underexpressed in NONHSAT098146 knockdown and ANXA5 knockdown cells, respectively (Fig. 7A). Cell proliferation was stable in the NONHSAT098146-knockdown cells (Fig. 7B). These results indicated that NONHSAT098146 and ANXA5 have the same transcripts. The RNA expression of the lncRNA NONHSAT098146 was identified through northern blotting. The results showed that NONHSAT098146 hybridization signals [468 base pairs (bp)] were invisible, whereas ANXA5 hybridization signals (1,762 bp) were present in the cells (Fig. 8). Therefore, the lncRNA NONHSAT098146 may be an artificial gene.

Figure 7 Expression levels of NONHSAT098146 and ANXA5 decrease in KD A549 cells. (A) Total RNA was isolated and subjected to reverse transcription-quantitative polymerase chain reaction analysis to detect RNA expression levels of NONHSAT098146 (white) and ANXA5 (black) in A549 cells. The fold change was standardized using mRNA expression levels of GAPDH, relative to the mock group. (B) Cell proliferation was detected using the MTT assay between 24 and 48 h. The optical density of the experiment was standardized using the optical density of empty wells. Error bars represent the standard deviation. Student's t-test was used to evaluate differences between the groups. *P<0.05, compared with the mock group. The results are summarized from observations of three independent experiments. ANXA5, annexin A5; GAPDH, glyceraldehyde-3-phosphate dehydrogenase; KD, knockdown.

Figure 8 Long non-coding RNA NONHSAT098146 cannot be detected through northern blotting. Northern blotting revealed that NONHSAT098146 hybridization signals (468 bp) were not visible in the examined cell lines. ANXA5 hybridization signals (1,762 bp) were present in the cell lines. The marker was DIG-labeled RNA with 40 μg (left) and 20 μg (right). The positive control was DIG-labeled NONHSAT098146 RNA produced through in vitro transcription. The marker size is labeled on the right side. ANXA5, annexin A5.

Established lncRNA overexpressing stable cell lines

To examine the effect of the overexpression of lncRNA CR933609, the A549 cells were transfected with a CR933609 recombinant plasmid vector, and stable and control clones were isolated. The gene expression of CR933609 increased by >2.5-fold in the CR933609-overexpressing cells, compared the vector control cells, but no change in INO80D was observed (Fig. 9A). No significant change in cell proliferation was observed in the CR933609-overexpressing cells (Fig. 9B). It was found that cell migration was significantly decreased in the CR933609-overexpressing cells (Fig. 9C and D). Therefore, CR933609 may act as a tumor suppressor gene.

Figure 9 Established CR933609-overexpressing stable clone. (A) Total RNA was isolated and subjected to reverse transcription-quantitative polymerase chain reaction analysis to detect RNA expression levels of CR933609 (white) and INO80D (black) in the A549 cell line and stable clone. The fold change was standardized using mRNA expression levels of glyceraldehyde-3-phosphate dehydrogenase, relative to A549. (B) Cell proliferation was detected using the MTT assay between 24 and 72 h. The optical density of the samples was standardized using the optical density of empty wells. (C) Wound-healing assay for A549 and stable clone. (D) Bar figure showing the quantitative results of the wound-healing assay. Error bars represent the standard deviation. Student's t-test was used to evaluate differences between groups. *P<0.05, compared with the A549 group or 0-h group. The results are summarized from observations of three independent experiments. INO80D, INO80 complex subunit D; OV, overexpression.

NONHSAT098146-overexpressing stable cell lines were established to ascertain the effect of overexpression. The expression of NONHSAT098146 increased by 2-fold in NONHSAT098146-overexpressing cells, compared with the vector control cells, but no change in ANXA5 was observed (Fig. 10A). No significant differences in proliferation or migration were observed in the NONHSAT098146-overexpressing cells (Fig. 10Bx–D). These established cell lines were used in the subsequent experiment.

Figure 10 Established NONHSAT098146-overexpressing stable clone. (A) Total RNA was isolated and subjected to reverse transcription-quantitative polymerase chain reaction analysis to detect the RNA expression levels of NONHSAT098146 (white) and ANXA5 (black) in the A549 cell line and stable clone. The fold change was standardized using mRNA expression levels of glyceraldehyde-3-phosphate dehydrogenase, relative to A549. (B) Cell proliferation was detected using the MTT assay between 24 and 48 h. The optical density of the samples was standardized using the optical density of empty wells. (C) Wound-healing assay for A549 cell line and stable clone. (D) Bar figure showing the quantitative results of the wound-healing assay. Error bars represent the standard deviation. Student's t-test was used to evaluate differences between the groups. *P<0.05, compared with the A549 group. The results are summarized from observations of three independent experiments. ANXA5, annexin A5; OV, overexpression.

lncRNA CR933609 acts as a repressor decoy to prevent downregulation of expression levels of INO80D

As shown in Fig. 11A (top), the expression levels of CR933609 and INO80D were decreased in the A549 and mock (A549 transfected entry vector) cells transfected with siRNAs targeting CR936609, in accordance with the results in Fig. 5A. Only the expression levels of CR933609 were decreased in the CR933609-overexpressing cells transfected with siRNAs (Fig. 11A, bottom). The overexpression of CR933609 may counteract the inhibitory effects of siRNAs on the expression levels of INO80D. There were five target sites for INO80D knockdown, which decreased the expression of INO80D only; the expression of CR933609 remained stable (Fig. 11B). The expression levels of NONHSAT098146 and ANXA5 were decreased in the NONHSAT098146-knockdown (Fig. 12A) and in the ANXA5-knockdown (Fig. 12B) cells overexpressing NONHSAT098146. These results indicated that the lncRNA CR933609 acts as a repressor decoy to prevent downregulation of the expression of INO80D, whereas NONHSAT098146 was not a repressor decoy.

Figure 11 lncRNA CR933609 acts as a decoy to prevent downregulation of the expression levels of INO80D. (A) Three target sites for CR933609 KD were in common with those for INO80D. (B) Five target sites for INO80D KD were independent from CR933609. Total RNA was isolated and subjected to reverse transcription-quantitative polymerase chain reaction analysis to detect the RNA expression levels of CR933609 (white) and INO80D (black) in the A549 cell line and stable clone. The fold change was standardized using mRNA expression levels of glyceraldehyde-3-phosphate dehydrogenase, relative to the mock group. Error bars represent the standard deviation. Student's t-test was used to evaluate differences between the groups. *P<0.05, compared with the mock group. The results are summarized from observations of three independent experiments. INO80D, INO80 complex subunit D; KD, knockdown; OV, overexpression.

Figure 12 Expression levels of NONHSAT098146 and ANXA5 do not between A549 and NONHSAT098146-overexpressing stable cell lines. (A) One target site for NONHSAT098146 KD (1 and 5 nM) was in common with those for ANXA5. (B) Three target sites for ANXA5 KD were independent from NONHSAT098146. Total RNA was isolated and subjected to reverse transcription-quantitative polymerase chain reaction analysis to detect RNA expression levels of NONHSAT098146 (white) and ANXA5 (black) in the A549 cell line and stable clone. The fold change was standardized using mRNA expression levels of glyceraldehyde-3-phosphate dehydrogenase, relative to the mock group. Error bars represent the standard deviation. Student's t-test was used to evaluate differences between the groups. *P<0.05, compared with the mock group. The results are summarized from observations of three independent experiments. ANXA5, annexin A5; KD, knockdown; OV, overexpression.

Overexpression of lncR NA CR933609 prevents downregulation of the expression levels of INO80D in endog enous miRNA 5096 treatment

As the computational results suggested that miRNA-5096 may target the INO80D and CR933609 transcripts at their common sequences (Fig. 3A, bottom), the A549 and CR933609-overexpressing A549 cells were transfected with miRNA-5096. The results indicated that miRNA-5096 at a concentration of 100 nM downregulated the expression levels of INO80D and CR933609 in the A549 cells (Fig. 13, left), but not in the CR933609-overexpressing A549 cells (Fig. 13, middle and right). However, miRNA-5096 at a concentration of 200 nM downregulated the expression levels of INO80D and CR933609 in the A549 cells (Fig. 13, left), but decreased only the expression levels of INO80D in the CR933609-overexpressing A549 cells (Fig. 13, middle and right). Accordingly, these results suggested that CR933609 acts as a repressor decoy to protect INO80D from downregulation by miRNAs.

Figure 13 Overexpression of CR933609 prevents downregulation of expression levels of INO80D in endogenous miRNA-5096 treatment. miRNA-5096 was transfected into the A549 stable clone. The dose of miRNA was 20-200 nM. Total RNA was isolated and subjected to reverse transcription-quantitative polymerase chain reaction analysis to detect RNA expression levels of CR933609 (white) and INO80D (black) in the A549 stable clone. The fold change was standardized using mRNA expression levels of GAPDH and expressed relative to the mock group. Error bars represent the standard deviation. Student's t-test was used to evaluate differences between the groups. *P<0.05, compared with the mock group. The results were summarized from observations of three independent experiments. miRNA, microRNA; INO80D, INO80 complex subunit D; OV, overexpression.

Discussion

In the present study, a bioinformatics pipeline was constructed to identify lncRNA sequences that were similar to the 3′-UTRs of protein-coding genes, and to confirm that these lncRNA and coding gene pairs contain the same miRNA target sites. The expression levels of CR933609 and INO80D, and those of NONHSAT098146 and ANXA5 were significantly decreased in NSCLC. In addition, the expression levels of CR933609 and INO80D were significantly decreased in colon and kidney cancer, and the expression levels of NONHSAT098146 and ANXA5 were significantly decreased in colon, mucosal, and kidney cancer. The expression levels of CR933609 and INO80D were decreased in CR933609-knockdown NSCLC cells, whereas only the expression levels of INO80D were decreased in INO80D-knockdown cells. The expression levels of NONHSAT098146 and ANXA5 were decreased in NONHSAT098146- and ANXA5-knockdown NSCLC cells. It was found that there were independent promoters in CR933609 and INO80D, however, NONHSAT098146 hybridization signals were undetectable. Furthermore, it was found that INO80D was downregulated by endogenous miRNA-5096 in A549 cells but not in CR933609-overexpressing A549 cells. Therefore, the lncRNA CR933609 may act as a decoy to protect INO80D from downregulation by miRNA-5096 in NSCLC cells. Thus, a protocol was established to identify novel lncRNAs in 3′-UTRs.

The interaction of RNA-binding proteins and miRNAs with the 3′-UTRs of mRNAs is known to affect the expression of eukaryotic genes by regulating mRNA translation, stability, and subcellular localization (44,45). The present study tested the hypothesis that, in addition to miRNAs, lncRNAs are involved in posttranscriptional gene regulation. A genome-wide computational pipeline identified 12 human lncRNA transcripts that showed high sequence matches with the 3′-UTRs of protein-coding mRNAs, suggesting that lncRNAs may be miRNA decoys (Fig. 1B and C). The protein-coding genes matching this set of 12 lncRNAs were involved in regulation of translational initiation and receptor binding, and their promoters contained similar transcription factor binding sites. Two of these lncRNA/coding-gene pairs, CR933609 and INO80D, and NONHSAT098146 and ANXA5 were investigated. Notably, the transcription unit of the lncRNA CR933609 was found to be embedded within the 3′-UTR of the INO80D gene (Fig. 3A), whereas CR933609 had a distinct promoter region (Fig. 6). Furthermore, the lncRNA NONHSAT098146 was identified as an artificial gene (Figs. 7 and 8). The results indicated that certain lncRNAs embedded in coding genes may not be transcripts, and a distinct promoter region requires identification.

To examine the biological functions of CR933609 and INO80D, RNA interference experiments were performed using artificial siRNAs (targeting sequences common to both transcripts) and shRNAs (specific for INO80D). The lncRNA CR933609 was expressed at significantly lower levels in three types of tumor tissue, compared with adjacent normal tissues (Fig. 4A). In addition, cell migration decreased in A549 clonal lines overexpressing CR933609 (Fig. 9). Therefore, CR933609 may act as a tumor suppressor gene in A549 cells. Notably, the expression levels of INO80D were downregulated in the CR933609- and INO80D-knockdown cells, whereas the lncRNA CR933609 was downregulated in CR933609-knockdown NSCLC cells (Fig. 5A). However, no significant changes were observed in the expression level of INO80D following CR933609 knockdown in CR933609-overexpressing cells (Fig. 11A). Additionally, endogenous miRNA-5096 downregulated CR933609 and INO80D in A549 cells, but a different pattern was observed in CR933609-overexpressing A549 cells (Fig. 13). Previous data have shown that the expression level of miRNA-5096 was reduced, whereas the expression level of miRNA-4694-3p was not altered significantly in lung cancer (46). CR933609 acted as a decoy to prevent the downregulation of INO80D.

Chromatin remodeling is the dynamic modification of chromatin architecture to allow condensed genomic DNA access to the regulatory transcription machinery, thereby controlling gene expression (47). It also provides fine-tuning at crucial cell growth and division steps to suppress tumor development. Targeting of the chromatin remodeling pathways is currently emerging as a major therapeutic strategy for the treatment of several types of cancer. INO80D encodes subunit D of the INO80 chromatin remodeling complex and is involved in transcriptional regulation, DNA replication and DNA repair, and may also be crucial in tumorigenesis (48). The data in the present study suggested that the lncRNA CR933609 regulated the mRNA stability of INO80D by competing for transcriptional enhancers or repressors. Therefore, the lncRNA CR933609 is likely to be involved in chromatin remodeling through the regulation of INO80D. Furthermore, the present study found that the lncRNA CR933609 may act as a tumor suppressor by regulating NSCLC cell migration (Fig. 9), and may serve as a repressor decoy to prevent the downregulation of INO80D (Fig. 11). Chromatin remodeling is one of the most important mechanisms for DNA repair, particularly in DNA DSBs. DNA DSBs result in cell cycle checkpoint arrest and apoptosis. This response can suspend tumorigenesis. The appropriate regulation of the mechanisms involved in DNA DSBs, including chromatin remodeling, is critical (49). Therefore, the lncRNA CR933609, which acts as a decoy to protect INO80D, a chromatin remodeling regulator, is essential for cell survival.

3′-UTRs can be independently expressed as developmentally regulated lncRNAs, which further blurs the distinction between coding and non-coding RNAs (50). This serves as a reminder that the traditional concept of the gene is becoming increasingly outmoded (51,52). However, compared with coding RNA and miRNA, significant gaps in the understanding of lncRNA function remain. To examine whether other lncRNAs, which may likewise be embedded within protein-coding genes, also function as repressor decoys, similar experiments on the remaining 11 lncRNA/coding-gene pairs examined in in silico analyses can be performed. When the NONCODE database updated to the second edition, NONHSAT143029 and NONHSAT055977 had been removed from the database (30). This indicated that the novel lncRNA from the database required confirmation. The present study established a protocol to identify novel lncRNAs in the 3′-UTR and confirmed the existence of novel lncRNAs using promoter activity or northern blotting.

As there are defects in all of the bioinformatics analytical methods, a protocol was established in the present study to verify the results of the analysis. It was also found that the lncRNAs published in databases may not exist. It is necessary to detect the existence of an lncRNA prior to investigating its function. This may be the reason that certain lncRNAs were removed from the database a number of years following publication.

In conclusion, the present first constructed a bioinformatics pipeline to investigate genome-wide 3′-UTR lncRNAs and then identified their clinical significance. The results indicated that the lncRNA CR933609 may act as a repressor decoy to protect INO80D from downregulation in NSCLC.

Abbreviations:

Acknowledgments

Not applicable.

Funding

This study was supported by the Ministry of Science and Technology of the Republic of China (grant nos. NSC-99-2320-B-037-006-MY3, MOST-105-2632-E-468-002 and MOST-106-2221-E-468-018), the National Health Research Institutes (grant no. NHRI-EX103-10326BI) and the Asia University and China Medical University Hospital (grant nos. DMR-107-206, ASIA104-CMUH-22 and ASIA105-CMUH-15).

Availability of data and materials

The datasets used and/or analyzed during the current study are available from the corresponding author on reasonable request.

Authors' contributions

CCC, TYL, YTL, YSC, YLC, YCC, CCL and PCL detected the biological function of cells, performed the RT-qPCR assays, performed the cell proliferation and luciferase reporter assays, performed the statistical analysis, and assisted in drafting the manuscript. WLC conceived the study, participated in the bioinformatics analysis, and coordinated and drafted the manuscript. KTY provided the tissue samples and the clinical data. YSC, WLC, JGC and TCL conceived the study, participated in its design and coordination, and assisted in drafting the manuscript. All authors read and approved the final manuscript.

Ethics approval and consent to participate

All participants provided their written informed consent to participate in the present study. The institutional review boards of CMUH and CCH approved the study (CMUH102-RECJ-015 and CCH-IRB-080322) and the consent procedure.

Consent for publication

All participants provided their written informed consent to participate in the present study. The publication of such data does not compromise anonymity.

Competing interests

The authors declare that they have no competing interests.

References