Hypermethylation‐mediated down‐regulation of lncRNA TBX5‐AS1:2 in Tetralogy of Fallot inhibits cell proliferation by reducing TBX5 expression

Abstract Tetralogy of Fallot (TOF) is the most common complex congenital heart disease (CHD) with uncertain cause. Although long non‐coding RNAs (lncRNAs) have been implicated in heart development and several CHDs, their role in TOF is not well understood. This study aimed to investigate how dysregulated lncRNAs contribute to TOF. Using Gene Expression Omnibus data mining, bioinformatics analysis and clinical heart tissue sample detecting, we identified a novel antisense lncRNA TBX5‐AS1:2 with unknown function that was significantly down‐regulated in injured cardiac tissues from TOF patients. LncRNA TBX5‐AS1:2 was mainly located in the nucleus of the human embryonic kidney 293 (HEK293T) cells and formed an RNA‐RNA double‐stranded structure in the overlapping region with its sense mRNA T‐box transcription factor 5 (TBX5), which is an important regulator in heart development. Knock‐down of lncRNA TBX5‐AS1:2 via promoter hypermethylation reduced TBX5 expression at both the mRNA and protein levels by affecting its mRNA stability through RNA‐RNA interaction. Moreover, lncRNA TBX5‐AS1:2 knock‐down inhibited the proliferation of HEK293T cells. In conclusion, these results indicated that lncRNA TBX5‐AS1:2 may be involved in TOF by affecting cell proliferation by targeting TBX5.


| INTRODUC TI ON
Congenital heart disease (CHD) is one of the commonest human congenital anomalies, with a morbidity of six to eight per 1000 live births, and nearly, a third of all major congenital malformations are accompanied by cardiac abnormalities. 1 CHD comprises a group of structural heart and great vessel disorders caused by cardiovascular dysplasia during the embryonic period.
Tetralogy of Fallot (TOF) is the most frequent complex CHD accounting for 7%-10% of all CHDs, 2,3 with an estimated incidence among live births of three per 10,000. 3 TOF was one of the first-described CHDs and was named after the French physician Dr Etienne-Louis Arthur Fallot, 4 6 TOF was also one of the first CHDs to be successfully repaired by surgery. 5 Although the post-operative survival rate among TOF patients has been greatly improved due to advances in surgical techniques, the incidence of late cardiac death in long-term survivors continues to increase. 7 There is thus an urgent need to explore the aetiology and pathogenesis of TOF. The precise cause of TOF is currently unclear. As a multifactorial disease involving genetic-environmental interactions, 8 TOF may be related to chromosome aneuploidy (eg trisomy of chromosomes 21, 18, or 13) or to mutations of several genes (eg NKX2.5, GATA4, TBX5, HAND2, JAG1, NOTCH and VEGF) encoding transcription factors or components of signalling pathways. [9][10][11][12] However, single-gene mutations occur in only a small minority of TOF patients, and more complex dysregulation of multiple genes is more common. Epigenetic modification plays an important role in gene expression and provides a bridge and mechanism for genetic and environmental interactions. There are three main types of epigenetic modifications: DNA methylation, histone modification and non-coding RNAs. However, the role of epigenetic factors in CHDs (including TOF) remains largely unclear. Abnormal gene expression during cardiac development leading to CHD may due to changes in the epigenetic landscape surrounding the genes' regulatory regions, 2 and more complex causes of TOF may therefore be associated with epigenetic variation. A few studies have investigated the role of epigenetic modifications in TOF. 1,[13][14][15][16] We previously found that methylation abnormalities in multiple genes were involved in the pathogenesis of TOF 9,17-19 and detected abnormal expression of some microRNAs (miRNAs) and significant changes in histone modification in injured heart tissues from TOF patients. [20][21][22] As the largest class and most important component of non-coding RNAs, long non-coding RNAs (lncRNAs) of over 200 nucleotides are numerous in eukaryotes and function as transcriptional regulators in many cell processes and diseases. 23 Dysregulation of numerous lncRNAs has been shown to participate in mammalian cardiogenesis and in the pathogenesis of related diseases. Differential expression of lncRNAs in heart tissue of CHD can regulate gene expression in several ways. Enhancer lncRNAs specifically regulate chromatin state transition during cardiac development, and participate in the differentiation of embryonic stem cells into heart muscle and in cardiac remodelling, whereas decoy lncRNAs, guide lncRNAs and scaffold lncRNAs affect the activity of cardiac transcription factors by binding protein factors. LncRNAs can also modulate cardiac development via the lncRNA-miRNA-mRNA co-expression network. 24 However, there has been only one previous report on the role of lncRNAs in TOF, 16 and their function in TOF therefore remains largely unknown.
In the current study, we characterized the lncRNA and mRNA profiles in human foetal and adult heart tissues by Gene Expression Omnibus (GEO) data mining and bioinformatics analysis, and focused on a previously unreported antisense lncRNA TBX5-AS1:2 with unknown function. LncRNA TBX5-AS1:2 was significantly up-regulated in foetal heart and was predicted to adjust the expression of its sense gene TBX5, which is one of the vital transcription factors related to cardiac development. We further demonstrated that lncRNA TBX5-AS1:2 expression was markedly decreased in injured heart tissue from patients with TOF. In vitro, lncRNA TBX5-AS1:2 down-regulation, mediated by DNA hypermethylation in the promoter region, significantly suppressed cell proliferation by reducing the expression of TBX5 at both the mRNA and protein levels, the reason of which was that lncRNA TBX5-AS1:2 affected the stability of TBX5 mRNA through the formation of an RNA-RNA duplex.

| Data mining in GEO database and bioinformatics analysis
Online data mining was performed in the GEO database (https:// www.ncbi.nlm.nih.gov/geo/) using the keywords lncRNA, human, heart development or CHD. Differentially expressed lncRNAs were analysed using the DESeq package, and WikiPathways database was applied to screen mRNAs related to heart development or CHD with differential expression. A coding and non-coding co-expression (CNC) network was established followed by these procedures: (a) data pre-processing: for same gene, median value of different transcripts for same genes represents gene expression value; (b) data screening: assessing differential expression of lncRNA and mRNA; (c) calculation and removal of subset of data based on Pearson's correlation coefficient (PCC) and calculation of correlation coefficient of PCC between lncRNA coding genes using R values; (d) screening with a standard of PCC ≥0.9 or ≤−0.9 and P ≤ .05 as meaningful subset and constructing CNC network using Cytoscape. Meanwhile, mRNAs adjacent to lncRNAs in the CNC network (≤10 kbp on the genome) were annotated. Target lncRNA was selected based on the following requirements: (a) in the CNC network; (b) adjacent mRNA was one of the differentially expressed in relation to heart development or CHD; and (c) adjacent mRNA co-expressed with the lncRNA.

| RNA extraction and quantitative polymerase chain reaction
Total RNA was extracted from cardiac tissues and cells using TRIzol reagent (Invitrogen) according to the manufacturer's instructions.
The quantity and quality of the extracted RNA were assayed using a

| Construction of eukaryotic overexpression vector and transient transfection
Human lncRNA TBX5-AS1:2-pcDNA 3.1 and human TBX5-pcDNA 3.1 were obtained from GeneRay. The plasmids were sequenced and shown to be consistent with the sequence in the National Center for Biotechnology Information database (https://www.ncbi.nlm.nih. gov/). Empty vector (pcDNA 3.1) was used as a negative control (NC).
When the cells plated in 6-cm dishes reached about 75% confluence, they were transfected with plasmids using Lipofectamine 3000 (Invitrogen). The cells were harvested 48 hours after transfection.

| Cell proliferation assays
Cell proliferation was determined using a Cell Counting Kit-8 (CCK8; Dojindo) according to the manufacturer's protocol. A total of 1 × 10 4 HEK293T cells per well were seeded into 96-well plates for adherence. CCK8 reagent (10 µL) was added to each well followed by incubation for 3 hours at 37°C. Cell viability was used to represent for proliferation and evaluated by the absorbance at 450 nm. All samples were prepared in triplicate and normalized to blank controls.
Three time-points were set at 40, 56 and 72 hours after transfection or incubation. The experiments were repeated three times.

| Cell apoptosis assays
Cell apoptosis was detected by harvesting 1 × 10 6 HEK293T cells and staining using an Annexin V-fluorescein isothiocyanate (FITC)/propidium iodide (PI) Kit (Dojindo). Cells were stained successively with 5 µL FITC reagent and 5 µL PI reagent, and incubated in the dark for 15 minutes at room temperature each time. Flow cytometry analysis was performed using a FACSCalibur (BD Biosciences) and apoptosis was analysed using FlowJo software.

| Nuclear-cytoplasmic separation
Nuclear and cytoplasmic fractions were isolated from HEK293T cells using a PARIS kit (Thermo Fisher Scientific) according to the manufacturer's instructions. Actin was used as a cytoplasmic control, and U1 was used as a nuclear control. RNA levels of lncRNA TBX5-AS1:2, Actin and U1 in the cytoplasm and nuclear components were measured by qPCR after RNA extraction and reverse transcription.

| RNA fluorescence in situ hybridization (FISH) assay
Cells were grown in a 4-chamber slides for 24 hours, fixed with 4% paraformaldehyde for 20 minutes and dehydrated in an ascending series of ethanol solutions. Cells were hybridized overnight at 42°C with probe. Non-specific probe was removed using 0.5× saline sodium citrate containing 50% formamide at 37°C. Biotinlabelled lncRNA TBX5-AS1:2 was detected using anti-biotin monoclonal antibody and secondary antibody. Finally, the slides were stained with DAPI (Cell Signaling Technology) and subjected to fluorescence signal detection under Leica TCS SP8 laser confocal microscope (Leica).The probe of lncRNA TBX5-AS1:2 used was as follow: CY3-GGUUUCGAUUAAGAUACACCAUAGGCUCUACACGAUC GGC.

| Western blot and antibodies
Human embryonic kidney 293 cells were collected and lysed with RIPA buffer (Yeasen) containing a protease inhibitor cocktail (Sigma-Aldrich). Proteins were determined using a BCA Protein Assay Kit (Pierce) and equivalent amounts of proteins were separated by 10% sodium dodecyl sulphate-polyacrylamide gel electrophoresis and transferred to nitrocellulose membranes (Pall).
The membranes were blocked with 5% milk in Tris-buffered saline Tween for 1 hour at room temperature and then incubated with specific primary antibodies at 4°C overnight, followed by secondary antibodies at room temperature for 1 hour. The primary specific antibodies used were anti-TBX5 (1:1000, Novus) and anti-Actin (1:8000, Proteintech), and the secondary horseradish peroxidaseconjugated antibody was antimouse IgG (1:5000, Kangwei). The proteins were then visualized on X-ray film using Clinx ChemiScope (Clinx Science Instruments).

| Ribonuclease protection assay
Each RNA sample from HEK293T cells was incubated for 1 hour at 37°C and then treated with RNAse A+T cocktail (Ambion) to digest single-stranded but not duplex RNAs. After incubation for 30 min-

| Measurement of RNA stability
RNA stability was measured by plating HEK293T cells at 2.5 × 10 5 cells per well in 6-well plates and culturing overnight, followed by incubation with actinomycin D (Sigma-Aldrich) for 4, 6, 8 and 10 hours, respectively. Total RNA was extracted after treatment at each time-point and subjected to qPCR for TBX5 mRNA quantification.

| Construction of dual-luciferase reporter plasmids
A lncRNA TBX5-AS1:2 DNA fragment containing CpG island 2 was amplified and cloned into the pGL3-Basic-firefly vector (Promega). (Y = C or T, R = A or G).

| Statistical analysis
All experiments were repeated three times. All statistical analyses were performed using paired two-tailed Student's t tests with GraphPad Software. Data were shown as mean ± standard error (SEM), and a P value < .05 was considered statistically significant.

| LncRNA TBX5-AS1:2 was selected by data mining and bioinformatics analysis
To gain an insight into the role of lncRNAs in heart development or CHD, we mined data in the GEO database and a result of transcriptome sequencing of two human foetal hearts (https://www.ncbi. nlm.nih.gov/geo/query /acc.cgi?acc=GSE68279) was filtered out, which was from an article related to the lncRNA profile in human foetal and adult hearts. 25 Two other human foetal heart RNA-Seq data sets (GSM1059494, 17 weeks and GSM1059495, 13 weeks) and three normal adult heart RNA-Seq datasets (GSM1101970, GSM1698563, and GSM1698564) were also analysed in this article. Further bioinformatics analysis of the RNA-Seq data for the seven samples identified 277 lncRNAs and 47 mRNAs related to heart development or CHD that were differentially expressed between foetal and adult heart tissues (Table S1 and Figure 1A). Among the 19 lncRNAs, lncRNA TBX5-AS1:2 was highly expressed in foetal heart tissues ( Figure 1B) and was selected for further analysis based on the inclusion requirements we set. Its neighbour gene was TBX5 ( Figure 1C), which encodes a transcription factor involved in the control of cardiogenesis and related to CHD. LncRNA TBX5-AS1:2 was transcribed from the negative strand of the sense gene TBX5 in a head-to-head orientation. Their overlapping region was 92 bp long and covered exon1 of lncRNA TBX5-AS1:2 and part of the 5′ untranslated region of TBX5 ( Figure 1C). The CNC network of lncRNA TBX5-AS1:2 indicated that it may also act on TBX5 ( Figure 1D).

| LncRNA TBX5-AS1:2 and TBX5 were downregulated in TOF heart tissues
To validate the role of lncRNA TBX5-AS1:2 in heart development or TOF and to determine whether lncRNA TBX5-AS1:2 affected the

| LncRNA TBX5-AS1:2 knock-down inhibited cell proliferation in vitro
Abnormal cell proliferation and apoptosis is a key feature of TOF. To investigate the effects of lncRNA TBX5-AS1:2 on cell proliferation and apoptosis, lncRNA TBX5-AS1:2 was successfully down-regulated by stable transfection of shRNA2 lentivirus with the most interference efficiency and up-regulated separately in HEK293T cells (Figure 2A,B).

| LncRNA TBX5-AS1:2 was mainly located in the nucleus of HEK293T cells
To further explore the underlying mechanism of lncRNA TBX5-AS1:2 affecting cell proliferation involved in TOF, subcellular localization of lncRNA TBX5-AS1:2, which determines its action mode, was confirmed firstly. Nucleus cytoplasm separation indicated that lncRNA TBX5-AS1:2 was mainly distributed in the nucleus in HEK293T cells ( Figure 3A), and this was verified by RNA-FISH assay ( Figure 3B).

| LncRNA TBX5-AS1:2 knock-down reduced TBX5 mRNA and protein levels
Previous bioinformatics prediction indicated that lncRNA TBX5-AS1:2 may be co-expressed with its sense gene TBX5. We therefore detected TBX5 mRNA and protein expression levels by qPCR and Western blot (WB) in HEK293T cells with down-regulated and up-regulated lncRNA TBX5-AS1:2. TBX5 mRNA and F I G U R E 3 LncRNA TBX5-AS1:2 regulated expression of TBX5 in mRNA and protein levels by forming an RNA duplex with TBX5 to increase its stability. A, Nucleus cytoplasm separation indicated that lncRNA TBX5-AS1:2 was mainly located in the nucleus of HEK293T cells, similar to U1. B, The result of a RNA-FISH assay also showed that lncRNA TBX5-AS1:2 was almost nuclear in HEK293T cells. Centre DAPI was used to stain nuclei (blue); left red fluorescence was from the biotin fusions; right the merged image. C and E, Dysregulation of lncRNA TBX5-AS1:2 positively regulated TBX5 protein levels according

| LncRNA TBX5-AS1:2 influenced TBX5 mRNA stability by RNA duplex formation
Antisense lncRNA can hybridize with its sense mRNA to form an RNA duplex that protects the mRNA from RNase degradation.
LncRNA can influence mRNA stability via this RNA-RNA interaction to modulate sense mRNA expression. We therefore verified the formation of a protective lncRNA TBX5-AS1:2 and TBX5 mRNA duplex in HEK293T cells by RPA, specifically at their overlapping region. RT-PCR revealed that this overlapping portion was at least partially protected from RNase degradation ( Figure 3H).
The combination of lncRNA TBX5-AS1:2 and TBX5 mRNA was further validated by RNA-RNA pull-down assay ( Figure 3I). We then blocked RNA synthesis in HEK293T cells using actinomycin D during a 10h period and measured subsequent levels of TBX5 mRNA to determine if its stability was augmented by lncRNA TBX5-AS1:2. The stability of TBX5 mRNA increased in HEK293T cells overexpressing lncRNA TBX5-AS1:2 compared with NC ( Figure 3J).

| Hypermethylation of lncRNA TBX5-AS1:2 in injured heart tissues from TOF patients
DNA methylation is an important factor regulating the expression of lncRNAs. We further explored the mechanism by which lncRNA TBX5-AS1:2 was down-regulated in injured heart tissues from TOF patients by predicting the distribution of CpG islands in the lncRNA TBX5-AS1:2 regulatory sequence using MethPrimer online. Four CpG islands were identified ( Figure 4A), of which islands 2 and 3 located in the basal core promoter were selected for investigation of their methylation status in human heart samples. BSP for clones revealed that CpG island 2 was remarkably hypermethylated in the injured heart tissues of TOF patients compared with normal heart tissue (P = .001) ( Figure 4B and Table S3), but there was no significant difference in CpG island 3 (Table S4).

| Hypermethylation of lncRNA TBX5-AS1:2 caused its down-regulation
Hypermethylation of CpG islands can inhibit the transcriptional activity of lncRNAs. We therefore transfected methylated and unmethylated lncRNA TBX5-AS1:2 reporter plasmids containing CpG island 2 into HEK293T cells. The unmethylated construct was digested by MRSEs whereas the methylated construct was not ( Figure 4C). BSP revealed that the methylation efficiency of CpG island 2 was increased in HEK293T cells when the construct was methylated ( Figure 4D,E). Dual-luciferase reporter assay revealed that the methylated construct decreased the transcriptional activity of lncRNA TBX5-AS1:2 in HEK293T cells compared with the unmethylated construct ( Figure 4F). showed that the influence of lncRNA TBX5-AS1:2 on cell proliferation may be mediated by its interaction with TBX5. TBX5 is a member of the T-box transcription factor family primarily known for its role in cardiac development. [49][50][51][52] Mutations or abnormal expression of TBX5 can increase the risk of CHD, including TOF. 12,53 Furthermore, non-coding transcripts may influence heart development by targeting TBX5. 54,55 As a complex developmental process, cardiogenesis includes cell proliferation, which contributes to cardiac growth and regeneration, 56,57 and decreased cell proliferation comprises part of the clinical phenotype of TOF. 58,59 Moreover, TBX5 plays a critical regulatory role in cell proliferation during cardiogenesis. 60,61 As an epigenetic mechanism, DNA methylation may play important roles in gene expression and regulation. Furthermore, abnormal promoter methylation of lncRNAs was shown to be connected to their dysregulation yy. 62,63 Hypermethylation of the lncRNA TBX5-AS1:2 promoter was accordingly detected in injured heart tissue from patients with TOF, associated with lncRNA TBX5-AS1:2 down-regulation in vitro.

| D ISCUSS I ON
We therefore suggest that the previously unknown non-cod-

CO N FLI C T O F I NTE R E S T S
The authors declare no conflict of interest.

DATA AVA I L A B I L I T Y S TAT E M E N T
The data sets used and/or analysed during the current study are available from the corresponding author on reasonable request.