Cardiac biopsies reveal differences in transcriptomics between left and right ventricle in patients with or without diagnostic signs of heart failure

Abstract

New or mild heart failure (HF) is mainly caused by left ventricular dysfunction. We hypothesised that gene expression differ between the left (LV) and right ventricle (RV) and secondly by type of LV dysfunction. We compared gene expression through myocardial biopsies from LV and RV of patients undergoing elective coronary bypass surgery (CABG). Patients were categorised based on LV ejection fraction (EF), diastolic function and NT-proBNP into pEF (preserved; LVEF ≥ 45%), rEF (reduced; LVEF < 45%) or normal LV function. Principal component analysis of gene expression displayed two clusters corresponding to LV and RV. Up-regulated genes in LV included natriuretic peptides NPPA and NPPB, transcription factors/coactivators STAT4 and VGLL2, ion channel related HCN2 and LRRC38 associated with cardiac muscle contraction, cytoskeleton, and cellular component movement. Patients with pEF phenotype versus normal differed in gene expression predominantly in LV, supporting that diastolic dysfunction and structural changes reflect early LV disease in pEF. DKK2 was overexpressed in LV of HFpEF phenotype, potentially leading to lower expression levels of β-catenin, α-SMA (smooth muscle actin), and enhanced apoptosis, and could be a possible factor in the development of HFpEF. CXCL14 was down-regulated in both pEF and rEF, and may play a role to promote development of HF.

Keywords: Cardiac biopsy; Gene expression; Heart failure; Ischemic heart disease; Left ventricular dysfunction.

Figure 1

Gene expression profile discriminates left and right ventricles. (a) Number of tissue samples and conditions from left and right ventricles. (b) Distribution of tissue samples in sequencing batches. (c) Principal component analysis (PCA) score plot with the two principal components (PC1 and PC2) plotted on the x- and y-axis, respectively. Each data point represents one sample, which is color-coded according to the tissue and shaped according to the sequencing batch. (d) Same plot as shown in “c” but colour-coded based on conditions and shape of the points indicates tissue.

Figure 2

Differential expression and functional analysis between left ventricle (LV) and right ventricle (RV) tissues. (a) Volcano plot of the differentially expressed genes. The x-axis represents fold change in log2 scale of LV versus RV while the y-axis indicates the p-values in − log10 scale. Each point represents a gene, and significantly expressed genes are highlighted in green. Top 10 significantly differentially expressed (FDR-adjusted p-value < 0.05) genes each from up and down regulated are labelled with gene symbols. (b) Heatmap showing normalised expression values of the differentially expressed genes in all the tissue samples and clustered using unsupervised hierarchical clustering. Rows and columns indicate tissue samples and differentially expressed genes respectively. The expression values are standardised sample-wise for each gene. Gene ontology (GO) annotation of the differentially expressed genes in term of biological process. (c) Up-regulated and (d) Down-regulated genes. The horizontal bars show percentage of the genes with the corresponding GO annotations (scale at bottom x-axis), The orange lines represent significance of the corresponding GO annotations (scale at top x-axis) as calculated by Enrichr.

Figure 3

Functional analysis and predicted regulatory effect of the differentially expressed genes between LV and RV. (a) Shows analysis of upstream regulators using Ingenuity Pathway Analysis (IPA) analysis on the 2383 DEGs. The bar plot shows 47 with activation z-score of ≥  + 2.5 or ≤  − 2.5, i.e. 25 predicted to be inhibited and 22 predicted activated. (b) Shows the network with the highest consistency score, having a predicted effect of activating muscle function, cell viability of tumour cell lines and organismal death. This network has 3 regulators (BMP31B, mir-26 and NKX2-5) operating on 17 DEGs including NPPA, RYR2 and DIO2.

Figure 4

Differential expression analysis between left ventricle (LV) and right ventricle (RV) tissues in different conditions. Volcano plots showing fold change distribution of the differentially expressed genes in different HF phenotypes. The x-axis represents fold change in log2 scale of LV versus RV while the y-axis indicates the p-values in − log10 scale. Each point represents a gene, and significantly expressed genes are highlighted in green. Top 10 significantly differentially expressed (FDR-adjusted p-value < 0.05) genes each from up and down regulated are labelled with gene symbols. HF phenotypes: (a) normal, (b) preserved ejection fraction (pEF), (c) reduced ejection fraction (rEF). Heatmaps showing normalised expression values of the differentially expressed genes in all the tissue samples and clustered using unsupervised hierarchical clustering. Rows and columns indicate tissue samples and differentially expressed genes respectively. The expression values are standardised sample-wise for each gene. HF phenotypes: (d) Normal, (e) pEF, (f) rEF.

Figure 5

Venn-diagram of differentially expressed genes in different HF phenotypes. Top 20 differentially expressed genes filtering on log₂ fold change 1 in each HF phenotype. (a,b) Show up- and down-regulated, respectively. Genes are displayed in descending order of log₂ fold change.

Figure 6

Functional classification of the differentially expressed genes in different HF phenotypes. (a) Shows GO annotations of biological process of up and down-regulated genes in LV versus RV in Normal. The horizontal bars display percentage of the genes with the corresponding GO annotations (scale at bottom x-axis), the orange lines represent significance of the corresponding GO annotations (scale at top x-axis) as calculated by Enrichr. (b) Displays pEF and (c) rEF.

Figure 7

Predicted upstream regulator and regulatory effect network of the differentially expressed genes between LV and RV in pEF. (a) Shows upstream regulators predicted utilising the Ingenuity Pathways Analysis (IPA). Out of the upstream regulators 22 have an activation score above + 2 or below − 2. In (b), calculated activation z-scores are shown for 11 functional groups based upon enrichment of 96 DEGs involved in cardiovascular system development and functionally separated into 10 networks with significant p-values ≤ 0.05. (c) Shows a network with a predicted activation of cardiac muscle function and cardiac contractility through the upstream regulators CSRP3, CACNA2D1, ACE2, DIO2, BDNF and CLU.