FGD5-AS1 Is a Hub lncRNA ceRNA in Hearts With Tetralogy of Fallot Which Regulates Congenital Heart Disease Genes Transcriptionally and Epigenetically

Abstract

Heart development requires robust gene regulation, and the related disruption could lead to congenital heart disease (CHD). To gain insights into the regulation of gene expression in CHD, we obtained the expression profiles of long non-coding RNAs (lncRNAs) and messenger RNAs (mRNAs) in 22 heart tissue samples with tetralogy of Fallot (TOF) through strand-specific transcriptomic analysis. Using a causal inference framework based on the expression correlations and validated microRNA (miRNA)-lncRNA-mRNA evidences, we constructed the competing endogenous RNA (ceRNA)-mediated network driven by lncRNAs. Four lncRNAs (FGD5-AS1, lnc-GNB4-1, lnc-PDK3-1, and lnc-SAMD5-1) were identified as hub lncRNAs in the network. FGD5-AS1 was selected for further study since all its targets were CHD-related genes (NRAS, PTEN, and SMAD4). Both FGD5-AS1 and SMAD4 could bind with hsa-miR-421, which has been validated using dual-luciferase reporter assays. Knockdown of FGD5-AS1 not only significantly reduced PTEN and SMAD4 expression in HEK 293 and the fetal heart cell line (CCC-HEH-2) but also increased the transcription of its interacted miRNAs in a cell-specific way. Besides ceRNA mechanism, RNAseq and ATACseq results showed that FGD5-AS1 might play repression roles in heart development by transcriptionally regulating CHD-related genes. In conclusion, we identified a ceRNA network driven by lncRNAs in heart tissues of TOF patients. Furthermore, we proved that FGD5-AS1, one hub lncRNA in the TOF heart ceRNA network, regulates multiple genes transcriptionally and epigenetically.

Keywords: FGD5-AS1; congenital heart disease; gene expression; lncRNA; tetralogy of Fallot.

FIGURE 1

Workflow of the entire study and the competing endogenous RNA (ceRNA) regulatory networks of each hub long non-coding RNA (lncRNA) in tetralogy of Fallot (TOF) hearts. (A) The workflow for the entire study. (B–E) The four networks all contain three types of genes, namely lncRNAs (purple), mRNAs (pink), and miRNAs (green). The edges connected between nodes indicate their co-expression relationship. The networks of the four hub lncRNAs (FGD5-AS1, lnc-GNB4-1, lnc-PDK3-1, and lnc-SAMD5-1) are presented, respectively.

FIGURE 2

FGD5-AS1 competing endogenous RNA (ceRNA) mechanism validation. The correlation coefficients between FGD5-AS1 and its ceRNA targets for our dataset (A) and the GEO: PRJNA156781 dataset (B) are shown. For lncRNA FGD5-AS1 and mRNA SMAD4, hsa-miR-421 was validated in the HEK 293 (C) and AC16 (D) cell lines, respectively. The correlation coefficients between hsa-miRNA-421 and FGD5-AS1/SMAD4 in our dataset (E) and the GEO: PRJNA156781 dataset (F) are also shown. n.s., not significant; The number of asterisks indicated the corresponding statistical significance (p-value). *p < 0.05; **p < 0.01; ***p < 0.001.

FIGURE 3

lncRNA FGD5-AS1 suppressed the apoptosis. (A) Quantitative reverse transcription PCR (RT-qPCR) assays were performed to detect the interference efficiency of FGD5-AS1 in the HEK 293 and CCC-HEH-2 cell lines. (B) RT-qPCR assays suggested a decreased transcriptional level of mRNA (SMAD4, PTEN, and RBSN) in two knockdowns (KD) of FGD5-AS1 cell lines (H, HEK 293 cell line; C, CCC-HEH-2 cell line). (C) RT-qPCR assays of four miRs for previous validation in the FGD5-AS1 KD HEK 293 and CCC-HEH-2 cell lines. (D) Western blotting was used to verify that interfered FGD5-AS1 can affect the expression of SMAD4, and the relative protein expression was analyzed. (E) Flow cytometry was performed to detect apoptosis in the FGD5-AS1 KD CCC-HEH-2 cell line. The positive controls were CCC-HEH-2 cells treated with apoptosis induction drugs. n.s., not significant; The number of asterisks indicated the corresponding statistical significance (p-value). *p < 0.05; **p < 0.01; ***p < 0.001.

FIGURE 4

RNA sequencing (RNAseq) analysis of FGD5-AS1 knockdown (KD) CCC-HEH-2 cells. (A) Principal component analysis (PCA) plot of RNAseq. Control and shRNA KD CCC-HEH-2 cell samples are shown in red and cyan, respectively. (B) Volcano plot of differential gene expression analysis between the control and the FGD5-AS1 KD CCC-HEH-2 cell line. Symbols of the top 20 significantly up/downregulated genes are labeled. (C) Gene Ontology (GO) and Disease Ontology (DO) enrichment analysis of the upregulated genes in RNAseq. (D) Heatmap of the 41 congenital heart disease (CHD)-related genes within differential expression in RNAseq. Nine of them were validated by RT-qPCR (WNT3, SOX9, PEX19, VIT, CDH11, IGFBP5, HAS2, ENO2, and EGR1), and the fold change values are labeled. n.s., not significant; The number of asterisks indicated the corresponding statistical significance (p-value). *p < 0.05; **p < 0.01; ***p < 0.001.

FIGURE 5

Assay for transposase-accessible chromatin with sequencing (ATACseq) analysis of FGD5-AS1 knockdown (KD) CCC-HEH-2 cells. (A) Principal component analysis (PCA) plot of ATACseq. Control and shRNA KD CCC-HEH-2 cell samples are shown in red and cyan, respectively. (B) Peak location comparison of the gene features. (C) Distance of peaks relative to the closest gene. (D) Peak distribution relative to the transcriptional start site. (D) Consensus peak number comparison of control of the KD samples. (E) Volcano plot of differential accessible region analysis. The top 20 significantly up/downregulated regions are labeled as the closest gene symbol. (F) Venn plot of differential gene expression analysis and differential accessible region analysis based on gene annotation. (G) Peak comparison of the differential accessible regions identified in ENPP2, HAS2, and VIT. (H) Enriched known motifs of the transcription factors within the differential accessible region sequences. Only the top 10 significant motifs are shown.