Abstract Body (Do not enter title and authors here): Background: The complex pathophysiology of heart failure with preserved ejection fraction (HFpEF) is difficult to replicate in vitro and predict the effect of interventions in disease animals based on the in vitro data. To address this, we developed an innovative in vivo pooled mosaic screening that simultaneously evaluates hundreds of therapeutic targets in a single diseased animal.
Hypothesis: Mosaic screening reliably predicts therapeutic impact of novel targets on HFpEF physiological phenotypes in vivo.
Methods: A library of 151 therapeutic interventions which includes interventions known to modulate pathways implicated in HFpEF or shown to improve the phenotype were packaged into a pool of adeno-associated virus (AAV) vectors. This AAV pool was administered at a low-dose, so that each cardiomyocyte (CM) in diseased animals received at most one intervention. Subsequent single nuclei RNA sequencing (snRNA–seq) captured the transcriptional changes resulting from each intervention. Mosaic screening was performed in female ZSF1 obese rats, a preclinical HFpEF model that mirrors the clinical cardiometabolic endotype. The screen targeted key HFpEF characteristics by evaluating transcriptome–wide gene sets. Hits were chosen based on favorable transcriptional shifts and assessed using a complete transduction saturation AAV dose of CMs in female ZSF1 obese rats (n=11) or in a second model high-fat diet + L–NAME-treated C57BL/6N male mice (n=14) using echocardiography and invasive hemodynamics.
Results: snRNA–seq data confirmed that positive controls, such as overexpression of NPPA and constitutively active PPP1R1A, and knockdown of HDAC6, induced expected transcriptional shifts in mosaic screens. To assess novel targets, we used custom gene sets measuring a range of biology, and found candidates modulating metabolism, hypertrophy, calcium handling, cardiac stress, and cell damage. Screen predictions were validated by administering therapies individually to HFpEF animals, improving key diastolic function parameters, including E/e’ (reduction by 15%) and IVRT (reduction by 24.2%) in echocardiography and LVEDP (reduction by 55.4%) in invasive hemodynamics.
Conclusions: Mosaic screen snRNA–seq transcriptome data from HFpEF CMs allow the simultaneous prediction of organism–level therapeutic effects for hundreds of targets, as validated by in vivo studies in HFpEF preclinical models.