Long Non-Coding RNAs Might Regulate Phenotypic Switch of Vascular Smooth Muscle Cells Acting as ceRNA: Implications for In-Stent Restenosis

Abstract

Coronary in-stent restenosis is a late complication of angioplasty. It is a multifactorial process that involves vascular smooth muscle cells (VSMCs), endothelial cells, and inflammatory and genetic factors. In this study, the transcriptomic landscape of VSMCs' phenotypic switch process was assessed under stimuli resembling stent injury. Co-cultured contractile VSMCs and endothelial cells were exposed to a bare metal stent and platelet-derived growth factor (PDGF-BB) 20 ng/mL. Migratory capacity (wound healing assay), proliferative capacity, and cell cycle analysis of the VSMCs were performed. RNAseq analysis of contractile vs. proliferative VSMCs was performed. Gene differential expression (DE), identification of new long non-coding RNA candidates (lncRNAs), gene ontology (GO), and pathway enrichment (KEGG) were analyzed. A competing endogenous RNA network was constructed, and significant lncRNA-miRNA-mRNA axes were selected. VSMCs exposed to "stent injury" conditions showed morphologic changes, with proliferative and migratory capacities progressing from G0-G1 cell cycle phase to S and G2-M. RNAseq analysis showed DE of 1099, 509 and 64 differentially expressed mRNAs, lncRNAs, and miRNAs, respectively. GO analysis of DE genes showed significant enrichment in collagen and extracellular matrix organization, regulation of smooth muscle cell proliferation, and collagen biosynthetic process. The main upregulated nodes in the lncRNA-mediated ceRNA network were PVT1 and HIF1-AS2, with downregulation of ACTA2-AS1 and MIR663AHG. The PVT1 ceRNA axis appears to be an attractive target for in-stent restenosis diagnosis and treatment.

Keywords: competing endogenous RNA; epigenetics; in vitro cellular model; in-stent restenosis; long non-coding RNA; post transcriptional regulation; transcriptomics.

Figure 1

Phenotypic appearance of contractile HUASMC differs from HUASMC under stent injury model conditions. Panel (A) shows selected images from both conditions. Stent injury model cells proliferate at a significantly faster rate compared to contractile (B). While contractile cells show cell cycle arrest in G0-G1 phase (C), HUASMC exposed to stent injury model have a significantly higher proportion of cells in the S and G2-M phases. * p < 0.05; *** p< 0.001.

Figure 2

A wound healing assay was performed to compare the migratory capacity in both conditions; selected images of both conditions are depicted in (A). Panel (B) shows the quantitative expression of this experiment. Contractile cells were mostly quiescent, while stent-induced proliferative cells covered almost 60% of the scratch at 36 h. Panel (C) shows the expression of selected genes known to be markers of phenotypic switch; blue and red color intensity accounts for down- and upregulation, respectively. In contractile cells there was an overexpression of ACTA2, CNN1, and COL4A1; the same genes were downregulated in the stent-induced proliferative phenotype, while FN1, MMP1, and SPP1 were uniformly upregulated. Cont_ = Replicates with contractile HUASMC; Stent_ = Replicates with stent-injured HUASMC; ** p < 0.01; *** p < 0.001.

Figure 3

MA scatter plots of sequencing data assessing overall distribution of the two datasets. Expression pattern of all transcripts (A) and non-coding transcripts (ncRNA) (B). Labeled red dots indicate differential expression (≥two-fold change and FDR ≤ 0.05); volcano plots reflect number, significance, and reliability of differentially expressed transcripts; red dots; and green dots indicate upregulation and downregulation, respectively. The x-axis represents the value of log2 Fold change and the y-axis represents the adjusted FDR.

Figure 4

Hierarchical clustering using stringent criteria for DE mRNA (A), DE lncRNA (B), and DE microRNA (C); sample clusters are included above the heatmap and clusters of DE transcripts are noted on the left of each heatmap. Red and blue represent upregulated genes and downregulated genes, respectively. The lower boxplots depict the distribution of each sample gene’s expression. Cont_ = Replicates with contractile HUASMC; Stent_ = Replicates with stent-injured HUASMC; DE: differentially expressed.

Figure 5

Gene ontology analysis and functional enrichment of KEGG terms for differentially expressed mRNA (panels (A,B)) and mRNA cis-target of differentially expressed lncRNAs (panels (C,D)); color intensity depicts significance, while ellipse size represents the number of genes.

Figure 6

The lncRNA–miRNA–mRNA ceRNA network representation of upregulated (panel (A)) and downregulated (panel (B)) lncRNAs. Diamonds, triangles, and circles represent miRNAs, lncRNAs, and mRNAs, respectively. Red and green represent upregulated and downregulated RNAs, respectively, in the stent-induced injury cellular model. Red and grey lines indicate miRNA–lncRNA and lncRNA–mRNA interactions, respectively. ceRNA, competitive endogenous RNA; lncRNA, long non-coding RNA; miRNA, microRNA; mRNA, messenger RNA.

Figure 7

PPI network representation of 150 mRNAs involved in the ceRNA network (panel (A)); The network was created by STRING and visualized in Cytoscape. Hub protein network (panel (B)) according to CytoHubba MCC coefficient. Red and green represent upregulated and downregulated proteins, respectively; circle size represents MCC score. PPI, protein–protein interaction; ceRNA, competing endogenous RNA.

Figure 8

Intersection of ceRNA and hub PPI network in Cytoscape. The network (panel (A)) includes 54 nodes (9 mRNA, 8 miRNA, and 27 lncRNA). Four main lncRNAs (PVT1, HIF1A-AS2, ACTA2-AS1, and MIR663AHG) were selected for ceRNA regulatory subnetworks (panel (B)). Pearson correlation coefficient for lnrRNA–mRNA interaction (r > 0.5). Red and green represent upregulated and downregulated RNAs, respectively, grey lines indicate interactions. ceRNA, competitive endogenous RNA; lncRNA, long non-coding RNA; miRNA, microRNA; mRNA, messenger RNA.

Figure 9

Co-culture system of human umbilical artery smooth muscle cells (HAUSMCs) and endothelial cells with M231 medium plus Smooth Muscle Differentiation Supplement (SMDS). Endothelial cells were seeded on inserts of 0.4 mm pore size in a no-touch manner. For the stent injury model, a piece of bare metal stent was set gently over the HUASMCs. Platelet-derived growth factor (PDGF-BB) 20ng/mL was added to conditioning media.