Analysis of genetic variant associated with heart failure mortality implicates thymic stromal lymphopoietin as mediator of strain-induced myocardial fibroblast-mast cell crosstalk and fibrosis.

Heart failure (HF) is a leading cause of death and disability globally. Heritable factors and the extent and pattern of myocardial fibrosis are important determinants of outcomes in patients with HF. In a genome-wide association study of mortality in HF, we recently identified a genetic polymorphism on chromosome 5q22 associated with HF mortality. Here, we sought to study the mechanisms by which this variant may influence myocardial disease processes. We find that the risk allele is located in an enhancer motif upstream of the TSLP gene (encoding thymic stromal lymphopoietin), conferring increased binding of the transcription factor nescient helix-loop helix 1 (NHLH1) and increased TSLP expression in human heart. Further, we find that increased strain of primary human myocardial fibroblasts results in increased TSLP expression and that the TSLP receptor is expressed in myocardial mast cells in human single nuclei RNA sequence data. Finally, we show that TSLP overexpression induces increased transforming growth factor ß expression in myocardial mast cells and tissue fibrosis. Collectively, our findings based on follow-up of a human genetic finding implicate a novel pathway in myocardial tissue homeostasis and remodeling.

Methods for isolation and transcriptional profiling of individual cells from the human heart.

BACKGROUND: Global transcriptional profiling of individual cells represents a powerful approach to systematically survey contributions from cell-specific molecular phenotypes to human disease states but requires tissue-specific protocols. Here we sought to comprehensively evaluate protocols for single cell isolation and transcriptional profiling from heart tissue, focusing particularly on frozen tissue which is necessary for study of human hearts at scale. METHODS AND RESULTS: Using flow cytometry and high-content screening, we found that enzymatic dissociation of fresh murine heart tissue resulted in a sufficient yield of intact cells while for frozen murine or human heart resulted in low-quality cell suspensions across a range of protocols. These findings were consistent across enzymatic digestion protocols and whether samples were snap-frozen or treated with RNA-stabilizing agents before freezing. In contrast, we show that isolation of cardiac nuclei from frozen hearts results in a high yield of intact nuclei, and leverage expression arrays to show that nuclear transcriptomes reliably represent the cytoplasmic and whole-cell transcriptomes of the major cardiac cell types. Furthermore, coupling of nuclear isolation to PCM1-gated flow cytometry facilitated specific cardiomyocyte depletion, expanding resolution of the cardiac transcriptome beyond bulk tissue transcriptomes which were most strongly correlated with PCM1+ transcriptomes (r = 0.8). We applied these methods to generate a transcriptional catalogue of human cardiac cells by droplet-based RNA-sequencing of 8,460 nuclei from which cellular identities were inferred. Reproducibility of identified clusters was confirmed in an independent biopsy (4,760 additional PCM1- nuclei) from the same human heart. CONCLUSION: Our results confirm the validity of single-nucleus but not single-cell isolation for transcriptional profiling of individual cells from frozen heart tissue, and establishes PCM1-gating as an efficient tool for cardiomyocyte depletion. In addition, our results provide a perspective of cell types inferred from single-nucleus transcriptomes that are present in an adult human heart.