Context-dependent effects of rosmarinic acid on SH-SY5Y neurite morphology using a label-free, Parkinson’s disease-oriented image analysis workflow

Parkinson’s disease (PD) pathophysiology includes axonal and neuritic abnormalities, manifested in vitro as changes in neurite phenotype. Consequently, the discovery of novel PD drug candidates could be accelerated by accessible, explainable, semi-automated in vitro screening assays that prioritize compounds affecting neurite morphology. The aim of the present study was to develop an image analysis pipeline and use it to study the effects of rosmarinic acid (RosAc), a known neuroactive compound, on neurites and compare them with those of the reference compound, retinoic acid (RetAc). To this end, phase-contrast images of SH-SY5Y cells treated with RosAc or RetAc under varying conditions were obtained and analyzed using a newly developed software pipeline based on pixel classification, followed by operator-curated object filtering based on cutoff-profile optimization and downstream region-of-interest (ROI) and skeleton measurements, implemented in the “PhaseContrastNeuriteAnalyzer” tool. The dependence of neurite changes on proliferation rate, low vs. high fetal bovine serum (FBS) concentrations, and phorbol 12-myristate 13-acetate (PMA) post-treatment was tested. RetAc treatment at concentrations of up to 10 µM did not inhibit SH-SY5Y cell proliferation, whereas RosAc significantly reduced it, with an IC 50 of 26.27 µM. RetAc caused significant dose-dependent increases in neurite length only at 1% FBS and not at 10% FBS. This finding both confirmed that low-serum conditions are appropriate for stimulating neuritogenesis and demonstrated that the developed pipeline is capable of quantifying such changes. RosAc induced neurite elongation at 1 µM, but only after PMA post-treatment. Stratification based on the median confluence value indicated that the observed changes were not solely confluence-related artifacts. In conclusion, the clearest RosAc-associated increase in neurite length occurred at a low concentration after PMA post-treatment, although the effect should be interpreted cautiously because it was modest and partly limited by concentration-dependent reductions in viability and proliferation signals. The developed image analysis pipeline provides an auditable, operator-curated approach for detecting neurite-associated morphology in noisy phase-contrast images and is organized to support reproducibility and FAIR (Findable, Accessible, Interoperable, and Reusable) reuse.