Single nucleus RNA sequencing identifies PLXNA4 upregulation in end-stage arrhythmogenic right ventricular cardiomyopathy

Figures

• 

  Figure 1. Cell types present in the right ventricle of ARVC and NF hearts (A) Uniform manifold approximation and projection for dimension reduction (UMAP) representation of single-nucleus RNA-sequencing data from right ventricles of 11 non-failing (NF) and 8 patients with arrhythmogenic right ventricular cardiomyopathy (ARVC) (nnuclei = 139,347). (B) Stacked bar plots display the relative contribution of each cell type in each patient, with cell types colored as in A. (C) Boxplots display the proportion of each cell type by disease status (nARVC = 8, nNF = 11). Rare cell types are shown in the inset for viewability. Statistically credible changes based on a false discovery rate of 0.05 are denoted with an ∗. Center line, median; Box limits, upper and lower quartiles; Whiskers, 1.5× interquartile range; Points, outliers.

• 

  Figure 2. Differential expression between ARVC and NF by cell type (A) Volcano plots depict the log fold-change (logFC) and p value for the differential expression test comparing patients with ARVC (n = 6–8) versus patients with NF (n = 10–11), by cell type. Significant genes are highlighted in red and were identified as those with FDR-adjusted two-sided p value <0.05 based on both CellRanger and CellBender, a concordant direction of effect between CellRanger and CellBender, and a low probability of being derived from background expression. The top 5 up-/down-regulated protein-coding genes are labeled. (B) (left) Dot plot shows the expression of 16 genes with a previously determined genetic link to ARVC. The size of each dot represents the percentage of cells of the given cell type that express the gene at non-zero levels (%Expr >0), and the shade represents the average log-normalized expression, normalized to the max across cell types (Avg. norm expression). (right) Heatmap depicts the logFC estimate for ARVC compared to NF for each gene in each cell type. Genes that are significantly differentially expressed at FDR<0.05 in the limma-voom test based on CellBender and CellRanger counts are…

• 

  Figure 3. Combined dataset of ARVC, HCM, DCM, and ICM (A) Uniform manifold approximation and projection for dimension reduction (UMAP) representation of single-nucleus RNA-sequencing (nnuclei = 831,720) from right ventricles of 11 patients with non-failing (NF) hearts and 8 patients with arrhythmogenic right ventricular cardiomyopathy (ARVC), and left ventricles of 24 patients with NF, 7 patients with ischemic cardiomyopathy (ICM), 11 patients with dilated cardiomyopathy (DCM), and 15 patients with hypertrophic cardiomyopathy (HCM). (B) Principal component analysis of patients based on a pseudo-bulk representation of single-nucleus RNA sequencing across all nuclei. Each patient is colored by their disease status, with a different shape for males and females. The amount of variance explained by each principal component (PC) is displayed on the respective axis. (C) Boxplots display the proportion of each cell type by disease status. Each cardiomyopathy group and its respective non-failing controls are separated by the background shading. Composition of each cardiomyopathy type compared to its respective control group was tested with scCODA, and statistically credible changes based on a false…

• 

  Figure 4. Comparing the ARVC versus NF effect to other cardiomyopathies (A) Venn diagram of ARVC versus NF differentially expressed genes. Genes significantly differentially expressed (FDR <0.05 based on both CellRanger and CellBender) between patients with ARVC and NF (nARVC = 8, nNF = 11) in the combined map across ARVC, ICM, DCM, HCM, and NF (nnuclei = 831,720) by cell type were filtered to those with a low probability of being background contamination in all groups and a consistent estimate of log fold-change based on CellRanger and CellBender (see STAR Methods). The effect between patients with ARVC and NF was then compared to the effect between each other cardiomyopathy (DCM: n = 11, HCM: n = 15, ICM: n = 7) and NF (NFDCM/HCM = 12, NFICM = 7), respectively, and genes with a significant difference in effect (FDR <0.05 based on both CellRanger and CellBender) are depicted in the Venn diagram. The gray circle depicts genes where the effect of ARVC versus NF was not significantly different from the effect of each other cardiomyopathy compared to NF, respectively. The number of genes in each section is labeled by cell type. (B) Forest plot of the log fold-change estimates comparing…

• 

  Figure 5. Representative images of Plexin A4 staining in human cardiac tissue (A and B) Representative staining of Plexin A4 and a-sarcomeric actinin (A) and Plexin A4 and Desmoplakin 1/2 (B) carried out on right ventricle sections taken from two non-failing patients and two patients with ARVC. Four images were taken for each patient sample. Sections stained for: nuclei (Hoechst, blue), cardiomyocytes (a-sarcomeric actinin, green) or intercalated disk (desmoplakin 1/2, green), Plexin A4 (red), and fibroblasts (vimentin, gray). Scale bars, 100 μm.

• 

  Figure 6. Expression and electrophysiology changes in hiPSC-derived cardiomyocytes with overexpression and knockdown of PLXNA4 (A and B) Volcano plots show changes in mRNA expression in hiPSC-derived cardiomyocytes with overexpression (OE, nOE = 3 and ncontrol = 3) (A) and knockdown (KD, nKD = 3 and ncontrol = 3) (B) of PLXNA4. Genes with false discovery rate (FDR) < 0.05 are highlighted in red, and the top up- and down-expressed genes are labeled. (C and D) UMAP embedding of significantly enriched Gene Ontology Biological Processes in the PLXNA4 OE model. Ontologies driven by significantly down-regulated genes with PLXNA4 OE (log fold-change < −1, FDR-adjusted p < 0.05) are shown in C, and ontologies driven by significantly up-regulated genes with PLXNA4 OE (log fold-change >1, FDR-adjusted p < 0.05) are shown in D. UMAP embeddings are constructed by calculating the Jaccard similarity between the genes driving the enrichment of each ontology (see STAR Methods). Labels were manually assigned to clusters of ontologies. (E and F) Dendrograms show the relationship between ontologies within the cardiac contraction/cardiac conduction cluster (E) and the immune response cluster (F). Dendrograms…