Abstract Body: Hypoplastic Left Heart Syndrome (HLHS) is a congenital heart disease with high mortality, but its molecular underpinnings remain poorly understood. Previous findings in patient-derived induced pluripotent stem cell cardiomyocytes (hiPSC-CMs) revealed conserved defects (reduced proliferation, disrupted sarcomere organization). We hypothesize these phenotypes result from shared transcriptional and epigenetic regulators.
We performed a functional siRNA screen with high-content morphological profiling to identify these regulators. From single-cell RNA-seq data, we selected 112 transcriptional/epigenetic regulators enriched in hiPSC-CMs for a custom siRNA library. Day 25 hiPSC-CMs from an unaffected father were treated and compared to control siRNA-treated father- and HLHS proband-derived hiPSC-CMs (4 replicate wells/siRNA, ~137 cells/well). Morphological profiling used high-content microscopy of ACTN2, DAPI, phalloidin, and MitoTracker-stained cells, measuring ~150 features (sarcomere organization, nuclear morphology, cell size, cytoskeleton integrity…). This workflow will be applied to additional HLHS trios to validate conserved regulators.
Using 85 unbiased morphological features, we developed a k-nearest neighbors classifier, trained on untreated father- and proband-derived cells, that achieved 85% accuracy (AUC = 0.92) in distinguishing healthy from disease cells. Applied to siRNA-treated cells, it identified eight candidate genes whose knockdown caused ≥2-fold shift from healthy to disease-like phenotypes including known cardiac transcription factors (GATA6, PRDM16, PBX1) and additional regulators (TEAD4, NR6A1, PPP1R13B, NR4A3, ASB3) implicated in proliferation, cell structure, and extracellular matrix regulation—processes linked to HLHS.
By integrating siRNA screening, morphological profiling, and machine learning, our study reveals a core regulatory network driving HLHS-associated cardiomyocyte defects. These findings advance our understanding of HLHS pathology and could inform therapeutic strategies. Our classification model may also generalize to other HLHS variants and congenital heart diseases, enabling broader applications in personalized disease modeling and drug discovery.