Abstract Body: Background: Cardiometabolic heart failure with preserved ejection fraction (HFpEF) is driven by comorbidities like obesity and hypertension, inducing widespread pathophysiology poorly recapitulated in vitro. Multi-omic datasets yield extensive disease-associated genes, but scalable in vivo validation is required to determine their causal roles.
Hypothesis: We hypothesized that a pooled in vivo mosaic screening platform, combining multiplexed AAV-mediated genetic perturbations with single-nucleus RNA-sequencing (snRNA-seq), predicts the physiological impact of novel targets on HFpEF phenotypes.
Methods: We generated an AAV library of >200 genetic interventions targeting HFpEF-implicated pathways. The pooled library was delivered to ZSF1 obese rats (a preclinical cardiometabolic HFpEF model) at a low dose so that most cardiomyocytes received ≤1 perturbation. Post-treatment, snRNA-seq captured transcriptional responses. Candidate hits were identified via favorable shifts in transcriptomic gene sets representing key HFpEF biology. Selected targets were validated using full cardiomyocyte transduction AAV doses in ZSF1 obese rats and an independent 2-hit mouse model (high-fat diet + L-NAME-treated C57BL/6N mice). Phenotypes were evaluated via biomarkers, echocardiography, and invasive hemodynamics.
Results: snRNA-seq data confirmed clinical trial targets induced expected transcriptional shifts. For example, Nppa overexpression attenuated hypertrophy-related gene sets, and constitutively active Ppp1r1a activated a calcium kinetics-related gene set. Combined Pde5a/Pde9a knockdown activated cGMP-PKG signaling. Utilizing custom gene sets, we identified novel candidates modulating metabolism, hypertrophy, calcium handling, stress, and cell damage. Individual in vivo validation of multiple hits improved key diastolic and heart failure parameters. Compared to AAV controls, Herald1 significantly reduced E/e’ (15.0%) and IVRT (24.2%) in ZSF1 obese rats. In the 2-hit mouse model, Herald7 significantly reduced LVEDP (55.2%), Herald11 improved NT-proBNP (23.0%), tau (28.9%), and dp/dt min (28.1%), and Herald13 reduced tau (20.6%).
Conclusions: In vivo mosaic screening enables simultaneous, high-throughput evaluation of hundreds of targets directly within diseased animals. Transcriptome-based predictions translated to robust physiological improvements in preclinical models, highlighting the platform's power to accelerate therapeutic discovery for complex cardiometabolic HFpEF.