Human Left Ventricle circRNA-miRNA-mRNA Network Analyses Reveal a Novel Proangiogenic Role for circNPHP1 Under Ischemic Conditions

Figure 1. Identification and Quantification of circRNAs (A) A bioinformatics framework to build circular RNA (circRNA)-microRNA (miRNA)-messenger RNA (mRNA) networks: after extracting RNAs (circRNAs, miRNAs, and mRNAs) from RNA sequencing (RNA-seq) data on left ventricle (LV) biopsies of cardiac surgery patients (ischemic heart disease [IHD] [n = 12], IHD+ type 2 diabetes mellitus [T2DM] [n = 11], non-IHD [n = 12]), differential expression analyses were performed. Next, the circRNA-miRNA sponging interactions were predicted using databases: Tissue-Specific circRNA Database (TSCD) (http://gb.whu.edu.cn/TSCD/), cancer-Specific circRNA Database (CSCD) (http://gb.whu.edu.cn/CSCD), and Circular RNA Interactome (CRI) (https://circinteractome.irp.nia.nih.gov/). The miRNA-mRNA interactions were predicted using miRWalk (http://mirwalk.umm.uni-heidelberg.de/) and miRcode (https://bio.tools/miRcode). Furthermore, only those interactions were kept where the connecting nodes (circRNAs, miRNA, and mRNAs) were expressed in the endothelial cells (ECs) (using published EC data sets from GEO). Gene Ontology and Kyoto Encyclopedia of Genes and Genomes (KEGG) pathway analyses were subsequently performed on…

Figure 2. Differential Expression Analyses (A to C) CircRNAs, miRNAs and mRNAs were identified following the pipeline outlined in the Methods section: IHD (n = 12), IHD+T2DM (n = 11), non-IHD (n = 12). Differential expression analyses were performed on normalized and log-transformed counts using the R package Limma. Three comparisons were performed: 1) IHD+T2DM vs IHD; 2) IHD vs non-IHD; and 3) IHD+T2DM vs non-IHD. The corresponding differentially expressed circRNAs, miRNAs, and mRNAs were plotted in a volcano plot with log2 fold change (FC) on the x-axis and −log10 P value on the y-axis. CircRNAs, miRNAs, and mRNAs that passed the log2 FC threshold of ±0.58 and P value <0.05 are shown in red, those that just pass the FC threshold are in green, and those that only pass the P value threshold are in blue. (D) Number of significant differentially expressed circRNAs, miRNAs, and mRNAs in LV biopsies (log2 FC threshold of ±0.58 and P value <0.05). Abbreviations as in Figure 1.

Figure 3. Predicted circRNA-miRNA-mRNA Networks in the Heart ECs (A to C) The endothelial circRNA-miRNA-mRNA networks were constructed in Cytoscape using predicted interactions among circRNAs, miRNAs, and mRNAs in IHD (± T2DM) (using in-house LV biopsies and heart ECs [GEO data]). All nodes (circRNAs, miRNA, and mRNAs) shown in the network are expressed in LV biopsies and ECs. The networks show circRNAs in red, miRNAs in cyan, and mRNAs in yellow. Following Gene Ontology (GO) and KEGG pathway analyses, all significant GO biological processes and KEGG pathways (passing threshold of FDR <0.05) were plotted as a bar graph with −log10(FDR) on the x-axis. Moreover, pathways relevant to EC biology are highlighted in the graph as solid colors. The corresponding genes (downstream targets of circRNAs and miRNAs) are outlined with the color of the relevant pathways. Genes that correspond to more than 1 pathway are outlined with multiple colors. (A) Network constructed from IHD vs non-IHD comparison. CircMCU emerged as the top significantly up-regulated circRNA (enlarged and labelled), which together with its miRNA and mRNA partners, mapped to EC-relevant pathways including apoptosis, angiogenesis,…

Figure 4. CircNPHP1 Expression in the Human LV and Cultured ECs Cardiac CircNPHP1 levels are higher in patients with IHD with/without associated T2DM and in cultured endothelial cells (human cardiac microvascular endothelial cells and human umbilical vein endothelial cells) exposed to disease-mimicking conditions, such as hypoxia and hypoxia, combined with increased D-glucose level (High glu). (A) Preliminary screening of circRNAs (identified from predicted networks) in endothelial cells: human cardiac microvascular endothelial cells were cultured in normal conditions, hypoxia (1% Oxygen), and hypoxia-high glucose (High Glu) (25 mmol/L D-glucose) conditions, respectively. After 48 hours, cells were harvested for quantitative real-time polymerase chain reaction (qRT-PCR) for the analysis and validation of indicated circRNAs. RNA expression is relative to normoxia. 18S is used as housekeeping gene; n = 3. Data are expressed as mean ± SEM and were assessed by 1-way analysis of variance with Dunnett's post hoc test. ∗P < 0.05, ∗∗P < 0.01, ∗∗∗P < 0.001, and ns = not significant. CircSTX17 was also tested but the levels were undetectable; hence, it is not included in the plot. CircNPHP1 emerges…

Figure 5. CircNPHP1 and Linear NPHP1 Expression in Different Cell Types (A) Single-cell and single-nuclei RNAseq data from healthy donor hearts (human heart atlas42) of NPHP1 gene as log2 transformed counts per million (CPM). Each dot indicates the data point as 1 donor (source: human heart atlas42). (B) RNAseq: A preliminary screen was carried out in different cells (sample size = 2 each) to determine absolute levels of circNPHP1 and linear NPHP1 in human umbilical vein endothelial cells (HUVECs or HU), human cardiac microvascular endothelial cells (HCMECs or HC), human AC16 cardiomyocytes (ACs), and cardiac fibroblasts (CFs). (C) HCMEC (left) and HUVEC cells (right) were cultured in either normal conditions, hypoxia (1% oxygen), or hypoxia combined with high glucose (25 Mm D-glucose). After 48 hours, cells were harvested for quantitative real-time polymerase chain reaction for the analysis of circNPHP1 (n = 3). Fold change in RNA expression is relative to normoxia; 18S is used as housekeeping gene. Data are expressed as mean ± SEM and were assessed by 1-way analysis of variance with Dunnett's post hoc test. ∗P < 0.05, ∗∗P < 0.01, ∗∗∗P < 0.001, and ns = not significant. (D) 2 μg of total…

Figure 6. Identification and Validation of the CircNPHP1 Proangiogenic Interactome in Endothelial Cells CircNPHP1 is physically associated with the antiangiogenic miR-221-3p to putatively command a proangiogenic subnetwork. (A) CircNPHP1 subnetwork extracted from T2DM network (left). CircNPHP1 is connected to many miRNAs and mRNAs (circled in black) that mapped to EC-relevant pathways (VEGF signaling and angiogenesis; right, top). In particular, miR-221-3p bound exclusively to circNPHP1 via its back-splice junction sequence (right, bottom). The arc consisting of circNPHP1, its sponged miRNA partners (miR-221-3p, miR-222-3p, miR-299-3p, miR-141-3p, and miR-139-5p) and downstream targets (VEGFA, BCL2, BCL2L11) were selected for further validation. (B) CircNPHP1 ASO (anti-sense oligonucleotide) pulldown assay: HUVEC cell extracts were incubated with biotinylated ASO targeting the back-splice junction of circNPHP1 (probes 1 and 2) or control probe (nontargeting sequence) (see Supplemental Figure 1A for the details). Subsequently, the pulldown of endogenous circNPHP1 was carried out using streptavidin beads (schematic representation; see details in Methods section). The precipitated RNA was…

Figure 7. Impact of Endogenous CircNPHP1 on Endothelial Cell Proliferation, Apoptosis, and Cord Formation Endogenous CircNPHP1 promotes the proliferative and networking capacities of HUVECs and HCMECs. (A) HCMECs (left) and HUVECs (right) were transfected with 30 nmol/L of circNPHP1 siRNA or control siRNA, and cultured in normal conditions, hypoxia (1% oxygen), and hypoxia-high glucose (25 mmol/L D-glucose) conditions, respectively. 48 hours post-transfection, cells were harvested for proliferation assay (BrDU incorporation, 6 hours) (n = 3), (B) HCMEC: cells were counted, and cell numbers were determined using hemocytometer (n = 3) and (C) HCMEC: apoptosis assay (Annexin V assay) (n = 3). (D) HUVECs were transfected with 30 nmol/L of circNPHP1 siRNA or control siRNA, and cultured in normal conditions, hypoxia (1% oxygen), or hypoxia combined with high-glucose (25 mmol/L D-glucose) conditions. 48 hours post-transfection, the cells were seeded on a 96-well plate containing Growth Factor Reduced Matrigel and cultured in similar conditions for 8 hours to determine their capacity to network in a cord formation assay, which is a classical angiogenesis assay. Representative images of the cord…

Figure 8. Linear NPHP1 Is Involved in Proliferation and Angiogenesis in the ECs (A) HUVECs were transfected with 30 nmol/L of linear NPHP1 siRNA or control siRNA. 48 hours post-transfection, cells were harvested for quantitative real-time polymerase chain reaction for the analysis of circNPHP1 and linear NPHP1 expression (n = 3). Fold change in mRNA expression is relative to control siRNA; 18S is used as housekeeping gene. Data are expressed as mean ± SEM and were assessed by unpaired Student's t-test. ∗P < 0.05, ∗∗P < 0.01, ∗∗∗P < 0.001, and ns = not significant. The table shows the average Ct values of linear NPHP1. Ct values >35 were considered undetectable. (B) HUVECs were transfected with 30 nmol/L of circNPHP1 siRNA2, linear NPHP1 siRNA, or control siRNA, and cultured in normal conditions, hypoxia (1% oxygen), and hypoxia-high glucose (25 mmol/L D-glucose) conditions, respectively. 48 hours post-transfection, cells were harvested for proliferation assay (BrDU incorporation, 6 hours) (n = 3) and (C) 48 hours post-transfection, cells were seeded on a 96-well plate containing Growth Factor Reduced Matrigel for 8 hours to determine cord formation. Representative images of the cord…

Figure 9. CircNPHP1 Regulation of miR-221-3p Target Genes’ Expression (A) HUVECs were transfected with 30 nmol/L of circNPHP1 siRNA (siRNA1: top or siRNA2: bottom) or control siRNA, and cultured in normal conditions, hypoxia (1% oxygen), and hypoxia-high glucose (25 mmol/L D-glucose) conditions, respectively. 48 hours post-transfection, cells were harvested for quantitative real-time polymerase chain reaction for the analysis of VEGFA, BCL2, and BIM (n = 3). Fold change in mRNA expression is relative to normoxia; 18S is used as housekeeping gene. (B) HUVECs were transfected with 30 nmol/L of circNPHP1 siRNA (siRNA1) or control siRNA and cultured as mentioned earlier. Cells were harvested for Western blot analysis of VEGF-A and BCL2 (n = 3). Images shown are representative (left). Histograms showing Western blot quantification are expressed as fold change against normoxia and normalized to Lamin B1 (loading control) (right). (C) HUVECs were transfected with 30nmol/L of circNPHP1 siRNA (siRNA1: left or siRNA2: right), anti-miR-221-3p, or in combination. Combination of control siRNA and anti-miRNA control is used as negative control. (D) 30 nmol/L of circNPHP1 siRNA (siRNA1), miR-221-3p…