Integratie van artificiële intelligentie gestuurde histopathologie en een nieuw muismodel voor multipele systeem atrofie om onderzoek naar synucleïnopathieën te bevorderen

Synucleinopathies, such as Parkinson's disease (PD) and multiple system atrophy (MSA), are fatal neurodegenerative disorders characterized by widespread alpha‑synuclein (αSyn) aggregation, neurodegeneration and neuroinflammation. Currently, there are no disease-modifying therapies available, leaving patients without options to halt the relentless autonomic and motor decline. Investigating disease mechanisms and developing effective treatments requires robust preclinical models and quantitative readouts, yet current approaches are limited by incomplete recapitulation of human pathology and by labor‑intensive, operator‑dependent analyses. This thesis directly addresses these limitations through two integrated studies. First, to resolve the analytical bottleneck, we developed an advanced deep learning framework utilizing convolutional neural networks (CNNs). This automated pipeline enables high-throughput, unbiased brain-wide mapping and quantification of neuronal populations, neuroinflammation and pathological inclusions. It provides a scalable solution that outperforms the spatial resolution and speed of traditional histopathology analysis and can be efficiently applied to screen mouse models of synucleinopathies. Second, leveraging our CNN-based pipeline, we characterized a novel MSA mouse model. Preclinical MSA models have failed to capture the disease's complex, dual cell pathology, defined by αSyn aggregates in both oligodendrocytes (as glial cytoplasmic inclusions, GCIs) and neurons. Furthermore, standard overexpression models struggle to reproduce the amyloid architecture of these inclusions. To overcome this, we combined viral vector-mediated oligodendroglial αSyn overexpression with αSyn preformed fibrils (PFFs). This combinatorial paradigm induced the formation of amyloid-rich GCI-like and neuronal inclusions, and captured several core pathological features of MSA, including progressive motor deficits, nigrostriatal dopaminergic degeneration, severe demyelination and chronic neuroinflammation. Importantly, by comparing this combination model with a PFF-only cohort that developed exclusively neuronal aggregates, we demonstrated in vivo that identical αSyn fibrillar seeds drive divergent synucleinopathies dictated by the permissive host cell milieu. Together, the studies included in this thesis establish a rigorous and complementary preclinical platform for synucleinopathy research. The CNN pipeline provides a robust and scalable framework for high-throughput spatial pathology quantification, while the combination mouse model offers a valuable, biologically relevant platform to study MSA. Ultimately, these tools can help decode cell-type-specific features that contribute to MSA pathogenesis and accelerate the preclinical screening of therapies.