Coronary Artery Disease Risk Variant Dampens the Expression of CALCRL by Reducing HSF Binding to Shear Stress Responsive Enhancer in Endothelial Cells In Vitro

Abstract

Background: CALCRL (calcitonin receptor-like) protein is an important mediator of the endothelial fluid shear stress response, which is associated with the genetic risk of coronary artery disease. In this study, we functionally characterized the noncoding regulatory elements carrying coronary artery disease that risks single-nucleotide polymorphisms and studied their role in the regulation of CALCRL expression in endothelial cells.

Methods: To functionally characterize the coronary artery disease single-nucleotide polymorphisms harbored around the gene CALCRL, we applied an integrative approach encompassing statistical, transcriptional (RNA-seq), and epigenetic (ATAC-seq [transposase-accessible chromatin with sequencing], chromatin immunoprecipitation assay-quantitative polymerase chain reaction, and electromobility shift assay) analyses, alongside luciferase reporter assays, and targeted gene and enhancer perturbations (siRNA and clustered regularly interspaced short palindromic repeats/clustered regularly interspaced short palindromic repeat-associated 9) in human aortic endothelial cells.

Results: We demonstrate that the regulatory element harboring rs880890 exhibits high enhancer activity and shows significant allelic bias. The A allele was favored over the G allele, particularly under shear stress conditions, mediated through alterations in the HSF1 (heat shock factor 1) motif and binding. CRISPR deletion of rs880890 enhancer resulted in downregulation of CALCRL expression, whereas HSF1 knockdown resulted in a significant decrease in rs880890-enhancer activity and CALCRL expression. A significant decrease in HSF1 binding to the enhancer region in endothelial cells was observed under disturbed flow compared with unidirectional flow. CALCRL knockdown and variant perturbation experiments indicated the role of CALCRL in mediating eNOS (endothelial nitric oxide synthase), APLN (apelin), angiopoietin, prostaglandins, and EDN1 (endothelin-1) signaling pathways leading to a decrease in cell proliferation, tube formation, and NO production.

Conclusions: Overall, our results demonstrate the existence of an endothelial-specific HSF (heat shock factor)-regulated transcriptional enhancer that mediates CALCRL expression. A better understanding of CALCRL gene regulation and the role of single-nucleotide polymorphisms in the modulation of CALCRL expression could provide important steps toward understanding the genetic regulation of shear stress signaling responses.

Keywords: coronary artery disease; coronary vessels; gene expression; genome-wide association study; polymorphism, single nucleotide.

Figure 1:

A) Association of rs880890 with CVD related traits from Open Targets Genetics-portal. rs880890 was significantly associated with hypertension and CAD. B) WashU browser shot of CAD loci with CALCRL region showing the regulatory regions (1-12) included in the study. A close-up look of the enhancer-8 harboring CAD associated SNPs. DNAse hypersensitivity shows that region with rs880890/rs840585 is accessible compared to nearby regions C) Complex heatmap showing the eRNA expression of each enhancer included in the study, CALCRL mRNA expression, R2 value of enhancers and the CAD SNPs annotations. D) Results of CRISPR-Cas9–mediated regulatory region deletion in TeloHAECs (ΔEnh clone). qPCR was performed from three independent experiments. The statistical significance was evaluated using a two-tailed Studenťs t-test (unpaired). For the bar plot, significance is denoted with an asterisk. *** p=0.003.

Figure 2:

A) Single cell RNA-Seq data from Tabula muris representing data accross 20 organs and tissues from mice shows that CALCRL is highly expressed in endothelial cells, especially in the lung. B) Single cell RNA-Seq data from human coronary arteries demonstrating that CALCRL is predominantly expressed in endothelial cells followed by smooth muscle cells and fibroblasts. C) Pseudobulk coverage track visualization of single nucleus ATAC-Seq signal at the CALCRL loci showing endothelial cell type-specific activity of the rs880890 containing regulatory element. D) The activity of the rs880890-containing enhancer was investigated in MOVAS, HEPG2, RAW 264.7, 3T3-L1 and TeloHAECs. Luciferase assay showing a significant increase in enhancer activity in region carrying rs880890 “A” allele compared to “G” allele in TeloHAEC followed by RAW 264.7 while other cell types did not show enhancer activity. E) STARR-Seq results from TeloHAEC and HepG2 showing the enhancer activity in the presence of rs880890 or rs840585. Results show that difference in enhancer activity was observed only in the presence of rs880890 in TeloHAEC and no difference in enhancer activity in HepG2 was seen (n=3). The statistical significance was evaluated using a two-tailed Studenťs t-test (unpaired). For the bar plot, significance is denoted with an asterisk. ** p<0.01.

Figure 3

A) EMSA for the rs880890 SNP showing (arrow pointing) that the “A” allele significantly gains binding affinity the “G” allele. B) Allele specific ChIP-qPCR showing 1.63-fold higher enrichment of HSF1 binding at A allele compared to G allele (Donor 1: p = 0.004 and Donor 2: p=0.007). C) Luciferase assay showing a significant decrease in enhancer-8 activity during HSF1 knockdown D) Luciferase assay showing a further decrease in enhancer-8 activity under disturbed flow (DF) compared to unidirectional flow (UF) in the presence of “G” allele. E) ATAC-seq demonstrating higher accessibility of the rs880890 cis-regulatory element (enhancer-8) in HAECs under UF compared to DF. F) Increased ATAC-seq and ChIP-Seq reads in rs880890-containing region in HAECs under UF compared to DF. Bar plot shows increased ATAC/ChIP-Seq reads in cells under UF and lower reads from the G allele-containing chromosome under DF compared to A allele. G) A significant decrease (P<0.01) in the expression of CALCRL was detected under DF compared to UF. H) Knockdown of HSF1 under shear stress resulted in significant downregulation of CALCRL expression (P=0.02). I) HSF1 ChIP-PCR showed significant changes in binding activity in HAEC under DF when compared to UF. The condition was true for both SNP centric (binding detection with rs880890 at the center) and peak centric (enhancer-8 peak’s center). The statistical significance for the experiments in the figure was evaluated using a two-tailed Studenťs t-test (unpaired). For the bar plot, significance is denoted with an asterisk. ** p<0.01 and * p<0.05

Figure 4:

A) Network generated by Cytoscape app geneMANIA. Black nodes with blue outline represent known CALCRL co-expressed genes while black nodes with green border are novel co-expressed genes. Purple connections denotes co-expression, cyan denotes pathway, pink denotes physical interaction and purple denotes co-localization. B) Functional enrichment plot of CALCRL siRNA and rs880890-enhancer deleted samples using IPA. z-score indicates a predicted activation or inhibition of a pathway (i.e. Erythropoietin, Angiopoietin, Relaxin, Nitric Oxide, Renin-Angiotensin, Apelin Endothelial, Endothelin-1, Adrenomedullin, Thrombopoietin and eNOS signaling pathways are predicted to be repressed). C-D) Volcano plot highlighting selected candidates differentially expressed in RNA-Seq data from CALCRL siRNA and rs880890-enhancer deleted samples.

Figure 5

A-B) Effect of CALCRL downregulation on proliferation of TeloHAECs. Normalized cell index (xCELLigence system) upon siRNA knockdown of CALCRL (si-CALCRL) and enhancer deleted clone (ΔEnh clone) in TeloHAECs shows significant decrease in cell proliferation. A two-way repeated measure ANOVA test with Šídák's multiple comparisons test for CALCRL siRNA treatment vs scrambled negative controls and ΔEnh clone vs. control at each time point: *P < 0.05; **P < 0.01; ***P < 0.001; Data points are average values of three biological replicates. C) Effect of si-CALCRL and ΔEnh clone on tube formation. Heat map of the averaged total tube lengths 4–16 h after plating TeloHAECs on Matrigel (n=3). D) Fold change in tube length formation. Each bar represents the average + SEM of total tube lengths obtained from each well image. Tube length values comparing siCALCRL to siNegative (P=0.026) and ΔEnh clone (P=0.023) to control was used analyzed by 2-tailed Student t test. *P<0.05. E) Representative of 9 spot well output images from Incucyte after 24 h siCALCRL treatment (n=3) and ΔEnh clones (n=3). F) OxLDL (200ug/ml) and LDL (200ug/ml)) induced cell death in enhancer deleted clone (ΔEnh clone) and control cells (TeloHAEC with similar passage as ΔEnh clone). oxLDL induced cytotoxicity showed significantly higher cytotoxicity in ΔEnh clones (P=0.026). G) NO production in ΔEnh clone and control cells (TeloHAEC with similar passage as ΔEnh clone). NO production was detected by Griess assay, n = 3, Total nitrate/nitrite concentration was significantly downregulated in DF compared to UF (P = 0.035) in clones. For the bar plots, the statistical significance was evaluated using a two-tailed Studenťs t-test (unpaired). Significance is denoted with an asterisk. * p<0.05