Proteomic and metabolomic analyses of the human adult myocardium reveal ventricle-specific regulation in end-stage cardiomyopathies

Figures

• 

  Fig. 1. Schematic summary. a Schematic outlining sample acquisition, omic platforms applied and the downstream analysis workflow. b Patient and sample summary of the proteomics experiment. Top, from left to right: disease condition distribution in all measured hearts; sex counts in donor, DCM and ICM hearts; age distribution of donor, DCM and ICM hearts, where boxplots indicate the medians (the middle line), first and third quartiles (the box) and the whiskers show the 1.5× IQR above and below the box; and sample location distribution. Bottom: pie charts counting the number of donor, DCM or ICM hearts that have both LV and RV samples measured; only LV samples measured or only RV samples measured in proteomics. c Patient and sample summary of the metabolomics experiment. Top, from left to right: disease condition distribution in all measured hearts; sex counts of donor, DCM and ICM hearts; age distribution of donor, DCM and ICM hearts by boxplots; and sample location distribution. Bottom: pie charts counting the number of donor, DCM or ICM hearts that have both LV and RV samples measured; only LV samples measured or only RV samples measured in metabolomics. d Counts of individual hearts…

• 

  Fig. 2. Differential analysis in protein expression levels between left and right ventricles and sex differences at each ventricle in donor hearts. a MDS plot of proteomic data, a dimension reduction tool to visualise high dimensional data, where samples from donor hearts are highlighted in green on the right panel. b Summary of DE test results in proteins to compare LV with RV samples from donors. Estimates are derived using a linear regression model adjusted for sex and log2-transformed age, with nDonor,LV = 27 and nDonor,RV = 13. Direction of regulation is in reference to RV, that is, up-regulated proteins are higher expressed in LV than in RV samples. c Volcano plot of LV versus RV proteins in donors. d KEGG pathway analysis of proteins up-regulated or down-regulated in LV compared to RV samples in donors. Enriched KEGG pathways at P-value ≤ 0.05 are shown. e Heatmap with dendrograms of top 100 DE proteins in donor LV and RV samples. Samples are annotated by sex and location. Proteins are annotated by direction of DE. f Plot of log2 FCs of proteins in female versus male donors in LV (x-axis) and RV (y-axis) samples. Direction of regulation is in reference to males, meaning that…

• 

  Fig. 3. Differential analysis in metabolite abundance levels between left and right ventricles and sex differences at each ventricle in donor hearts. a MDS plot of metabolomic data, a dimension reduction tool to visualise high dimensional data, with samples from donor hearts highlighted in green on the right panel. b Summary of DA test results in metabolites to compare LV with RV samples in donors. Estimates are derived using a linear regression model adjusted for sex and log2-transformed age, with nDonor,LV = 24 and nDonor,RV = 13. Direction of regulation is in reference to RV, that is, up-regulated metabolites are more abundant in LV than in RV samples. c Volcano plot of LV versus RV metabolites in donors. d KEGG pathway analysis of metabolites up-regulated or down-regulated in LV compared to RV samples in donors. Enriched KEGG pathways at P-value ≤ 0.05 are shown. e Heatmap with dendrograms of DA metabolites in donor LV and RV samples. Samples are annotated by sex and location. Metabolites are annotated by direction of DA. f Plot of log2 FCs of metabolites in female versus male donors in LV (x-axis) and RV (y-axis) samples. Direction of regulation is in reference to males, meaning…

• 

  Fig. 4. Differential analysis in protein expression levels between ventricles in heart failures (HF), DCM or ICM compared with donors. a Volcano plot of LV versus RV proteins in heart failures. Estimates are derived using a linear regression model adjusted for sex and age, log2-transformed, with nHF,LV = 32 and nHF,RV = 16. Direction of regulation is in reference to RV, where up-regulated proteins are higher expressed in LV than in RV samples. b Venn diagram illustrating overlaps in DE proteins between ventricles in donors and in HF. c Plot of log2 FCs of DE proteins between ventricles in both donors and HF. d–g Differential analysis in protein expression levels between left and right ventricles in DCM compared with donors, where nDCM,LV = 13 and nDCM,RV = 5, adjusted for sex and age, log2-transformed. Direction of regulation is in reference to RV. d MDS plot of proteomics data with DCM samples highlighted in brown. e Volcano plot of LV versus RV proteins in DCM. f Plot of log2 FCs of proteins in LV versus RV samples in donors against those of DCM. g Venn diagram counting the number of DE proteins between ventricles shared by/unique to donors and/or DCM. h–l Differential analysis in…

• 

  Fig. 5. Differential analysis in metabolomics abundance levels between ventricles in heart failures, DCM and ICM hearts compared with donors. a Volcano plot of LV versus RV metabolites in heart failures. Estimates are derived using a linear regression model adjusted for sex and age, log2-transformed, with nHF,LV = 29 and nHF,RV = 16. Direction of regulation is in reference to RV, that is, up-regulated metabolites are more abundant in LV than in RV samples. b Venn diagram illustrating overlaps in DA metabolites between ventricles in donors and in HF, where the only DA metabolite common to donor and HF hearts is adenosine. c Boxplot of adenosine abundances in donors and HFs stratified by sample location. d–g Differential analysis in metabolite abundance levels between left and right ventricles in DCM hearts compared with donors, where nDCM,LV = 12 and nDCM,RV = 5, adjusted for sex and age, log2-transformed. Direction of regulation is in reference to RV. d MDS plot of metabolomics data with DCM samples highlighted in brown. e Volcano plot of LV versus RV metabolites in DCM hearts. f Plot of log2 FCs of metabolites in LV versus RV samples in donors against those of DCM hearts. g Venn diagram…

• 

  Fig. 6. DE protein (circles) correlation networks overlaid by potential biological links to DA metabolites (squares) in donor, DCM and ICM. a Donor network plot. b DCM network plot. c ICM network plot. Proteomic data of top 200 DE proteins between ventricles are used to construct networks for donor and ICM. Protein nodes are coloured according to community detection results by the short random walk algorithm. Weights of edges between proteins (solid) are proportional to the absolute value of the pairwise Pearson correlation coefficient between proteins. Edges connecting metabolites to protein clusters (dashed) are annotated by hand, where for metabolite nodes, DA metabolites between ventricles for donor (BH-adjusted P-value ≤ 0.05), DCM or ICM (BH-adjusted P-value ≤ 0.1) are considered. Protein clusters are annotated by enriched Gene Ontology Molecular Function terms via gene set tests. Source data are provided in Supplementary Data 3, 9, and 12. (BH Benjamini-Hochberg; DA differentially abundant; DCM dilated cardiomyopathy; DE differential expression; ICM ischaemic cardiomyopathy).