Integrated computational and in vivo evidence prioritizes MYH6 as a candidate node in PFOS-associated HCM-like cardiac remodeling.

Per- and polyfluoroalkyl substances (PFAS), particularly perfluorooctane sulfonate (PFOS), persist in humans and have been associated with cardiovascular outcomes, but cardiac-specific molecular responses remain incompletely characterized. We integrated transcriptomic screening, ensemble machine learning, molecular docking and 100-ns molecular dynamics, cross-species cardiac single-cell atlases with virtual knockout, and a time-stratified mouse exposure study (approximately 5 μg/kg/d for 4-8 weeks) to prioritize candidate molecular nodes associated with PFOS-related cardiac remodeling. Robustness analyses supported the stability of the machine-learning prioritization, and MYH6 was repeatedly selected and enriched in cardiomyocytes across human and mouse atlases. Docking and simulation supported a structurally plausible modeled PFOS-MYH6 complex. In exposed mice, Myh6 expression declined over time as septal asymmetry and interstitial fibrosis increased. Batch docking revealed 11 of 26 PFAS with scores at least as favorable as PFOS, indicating compound-level variation in modeled MYH6 compatibility. Together, the data prioritize MYH6 as a candidate molecular node associated with PFOS-related HCM-like remodeling and integrate these multiscale observations into a testable putative AOP-like hypothesis.