Machine Learning-Assisted Label-Free SERS Decoding of Mitochondrial Molecular Dynamics in Ovarian Granulosa Cells during Aging

Ovarian aging is a pivotal determinant of female reproductive decline, of which ovarian granulosa cell senescence is the core executive link and amplifier, while mitochondrial dysfunction is considered an initiator of granulosa cell senescence. Consequently, nondestructive and in situ dynamic measurements of dynamic molecular events within mitochondria during granulosa cell senescence are crucial for deciphering this complex biological process. Herein, we developed mitochondria-targeting gold nanobipyramids (Au NBs) as a SERS substrate for monitoring the associated molecular stress responses within mitochondria during the oxidative aging process of an ovarian granulosa cell at the single-cell level. The label-free SERS spectra revealed key aging-induced biomolecular events and signatures within mitochondria, including DNA damage, protein conformational changes, and lipid peroxidation. Notably, accurate discrimination between senescence cells and normal cells can be achieved using the machine learning-based SERS spectra of mitochondria, achieving 100% identification accuracy. This study established a machine learning-assisted label-free SERS sensing platform for decoding mitochondrial molecular dynamics, offering profound insights into the mechanisms of ovarian aging at the cellular level and important evidence for clinical evaluation of ovarian reserve in the future.