Prediction of single-cell mechanisms for disease progression in hypertrophic remodelling by a trans-omics approach

Abstract

Heart failure is a heterogeneous disease with multiple risk factors and various pathophysiological types, which makes it difficult to understand the molecular mechanisms involved. In this study, we proposed a trans-omics approach for predicting molecular pathological mechanisms of heart failure and identifying marker genes to distinguish heterogeneous phenotypes, by integrating multiple omics data including single-cell RNA-seq, ChIP-seq, and gene interactome data. We detected a significant increase in the expression level of natriuretic peptide A (Nppa), after stress loading with transverse aortic constriction (TAC), and showed that cardiomyocytes with high Nppa expression displayed specific gene expression patterns. Multiple NADH ubiquinone complex family, which are associated with the mitochondrial electron transport system, were negatively correlated with Nppa expression during the early stages of cardiac hypertrophy. Large-scale ChIP-seq data analysis showed that Nkx2-5 and Gtf2b were transcription factors characteristic of high-Nppa-expressing cardiomyocytes. Nppa expression levels may, therefore, represent a useful diagnostic marker for heart failure.

Figure 1

Overview of the analysis of single-cardiomyocyte transcriptome data by a trans-omics approach. (a) Single-cell RNA-seq data were acquired from mice exposed to pressure overload by transverse aortic constriction (TAC) or sham operation. Day 3 (D3), week 1 (W1), week 2 (W2), week 4 (W4) and week 8 (W8). Marker genes were identified by time-series gene expression profiles at the single-cell level. (b) Biological functions of the marker genes are estimated by a network analysis of interactome data, and regulators of the marker genes are detected by a large-scale analysis of ChIP-seq data.

Figure 2

Identification of genes with expression change in single cardiomyocytes in the process of heart failure. (a) The top panel shows a dendrogram of hierarchical clustering of genes with a temporal change of expression identified by ANOVA. Each of the bottom panels shows the average gene expression levels in each of the six clusters. The horizontal axis of each panel indicates the time points (sham, D3, W1, W2, W4, W8) and the vertical axis indicates the average of gene expression levels. (b) Bar plot of top 10 genes with the greatest fold change among the upregulated genes in cluster 1.

Figure 3

Nppa and Nppb expression levels in single cardiomyocytes and non-cardiomyocytes after TAC. (a) Violin plot on the left shows the distribution of Nppa expression levels at each time point. Violin plot on the right shows the distribution of Nppb expression levels at each time point. (b) Dot plot on the left shows the distribution of Nppa expression levels at each time point. Dot plot on the right shows the distribution of Nppb expression levels at each time point.

Figure 4

Distribution of cell-to-cell transcriptional variation in single cardiomyocytes. (a) Scatter-plot shows t-distributed stochastic neighbour embedding (t-SNE) visualization of cardiomyocytes. Cells represented by dots are coloured to reflect the time points when they were extracted. (b–f) Scatter-plots show t-SNE visualization of cardiomyocytes. Cells represented by dots are coloured to reflect the expression levels of Nppa (b), Nppb (c), Atf3 (d), Ctgf (e) and Tgfb2 (f). Red dots indicate cardiomyocytes in which the expression levels of Nppa (b), Nppb (c), Atf3 (d), Ctgf (e) and Tgfb2 (f) were higher than average, while blue dots indicate those in which the expression levels of Nppa (b), Nppb (c), Atf3 (d), Ctgf (e) and Tgfb2 (f) were lower than the average.

Figure 5

The difference of gene expression patterns in cardiomyocytes with high and low Nppa expression. (a) Hierarchical clustering of both genes and samples based on their gene expression profiles in the gene expression matrix. Cardiomyocytes were divided into two subgroups based on Nppa expression level; high-Nppa group and low-Nppa group. The average gene expression levels were calculated for samples from each time point in the high-Nppa group and low-Nppa group. (b) Bar graph shows the number of DEGs detected in high-Nppa, low-Nppa and common group, compared with sham group. “Common genes” indicate DEGs present in both high-Nppa expressing cardiomyocytes and low-Nppa expressing cardiomyocytes.

Figure 6

Nppa expression and Hdac class I expression in cardiomyocytes. (a–e) Box plots show the expression levels of Nppa (a), Hdac1 (b), Hdac2 (c), Hdac3 (d) and Hdac8 (e) at day 3 after TAC. Blue columns show the expression levels in cardiomyocytes with low Nppa expression and red columns show the expression levels in cardiomyocytes with high Nppa expression.

Figure 7

Detection of genes correlated with Nppa and visualization of gene–gene association network involving genes negatively correlated with Nppa. (a) Correlation coefficient between the expression levels of Nppa and those of other genes at day 3 after TAC in cardiomyocytes. (b) The network shows a graphical representation of gene–gene associations involving 500 negatively correlated genes, where circles indicate genes. Genes involved in oxidative phosphorylation are represented by yellow nodes.

Figure 8

Prediction of transcription factors that regulate genes correlated with Nppa. High scoring transcription factors (TFs) for genes correlated with Nppa are shown. The horizontal axis in each panel indicates -log₁₀P-values. TFs with statistical significance (P < 0.05) are represented by red bars, and TFs with non-statistical significance (P > 0.05) are represented by blue bars.

Figure 9

A summary of the molecular mechanisms in the process of heart failure, as inferred by this study. Nppa expression level was notably induced by Hdac2 after stress loading, and exhibited cell-to-cell heterogeneity. Under the regulation of Nkx2-5 and Gtf2b, activation of components of muscle and mitochondrial dysfunction were induced, resulting in heart failure.