Objectives: The aim of the conducted study was the generation of a spatial and cell-type resolved proteomic map of the healthy human heart. A global protein expression “footprint” of the healthy heart can be used as a reference library to compare against diseased hearts in the search for biomarkers, therapeutic targets or disease signatures.
Methods: The human heart proteome was measured in 16 anatomical regions from 3 different healthy subjects and in three major cardiac cell types (endothelial cells, fibroblasts and smooth muscle cells) by high-resolution mass spectrometry-based proteomics.
Results: The MaxLFQ algorithm quantified a total of 11,163 proteins, 10,751 in the 16 heart regions and 10,447 in the non-cardiomyocyte cell types. For overall assessment of proteomics similarities and differences of the 16 heart regions, we employed principal component analysis (PCA). The major groups cavities, vessels and valves clearly clustered separately. Each of the 3 main clusters exhibited further sub-groupings. Heart cavities divided into atrial (RA, LA, SepA) and ventricular (RV, LV, SepV) parts and within them atrial and ventricular septa were separated from atria and ventricles, respectively. There was a moderate but clear distinction of the left and right side of the heart. The vessel group subdivided into large arteries and large veins. Coronary arteries formed a subcluster, demonstrating differences between coronary and large arteries at the proteomic level. Finally, both atrioventricular valves clustered together whereas semilunar valves formed a separate group. Additionally, we performed unsupervised hierarchical clustering of the 6,807 proteins with statistically different expression across the heart regions (FDR< 0.05). This again clustered cavities, vessels and valves with their subdivisions. Of a total of 11,236 different proteins, 7,965 were identified in all three cell types, indicating that the majority of the cardiac cell proteome is expressed in its major cell types. PCA analysis clearly separated the non-cardiomyocyte cell populations reflecting a distinct protein expression profile.
Conclusion: PCA and unsupervised clustering showed distinct subgroups reflecting molecular defined differences between anatomical regions on the proteomic level (e.g., atria versus ventricles). This high resolution proteomic map is an indispensable prerequisite to define molecular differences between healthy and diseased human hearts.
