Parkinson’s disease (PD) is the second most common neurodegenerative disorder, afflicting more than 4 million people worldwide. Diagnosis is typically based on clinical symptoms related to impaired motor function, at which point most dopaminergic neurons have been lost. PD symptoms vary across individuals, and a prodromal phase may precede clinical symptoms by several years. Hence there is an urgent need for robust PD biomarkers to facilitate early detection and all aspects of disease management.
We utilized fully automated, online 3D peptide fractionation with TMT isotope labels for data-dependent quantification of protein, protein phosphorylation, and protein glycosylation across BioFIND CSF samples (BioFIND cohort), comprising 118 PD patients and 88 healthy control (HC) subjects. We used targeted mass spectrometry and biochemical assays to validate candidates in an independent cohort of PD/HC CSF samples from the Harvard Neurodiscovery Center (HNDC cohort).
We identified over 6,000 proteins across CSF samples in the BioFIND cohort. Requiring zero missing values, we quantified over 1,200 proteins in all BioFIND CSF samples. Regression analysis revealed more than 60 proteins whose abundance in the CSF was associated with PD (Padj < 0.05). We used biochemical and targeted mass spectrometry assays to validate more than 15 candidates, including several not previously associated with PD, in the HNDC cohort comprising 40 PD and 40 HC CSF samples.
We generated the most detailed view of the CSF proteome to date. The robustness of our approach provided quantification of more than 1,200 proteins (with no missing values) across all samples in our discovery cohort, including glycosylated and phosphorylated proteins. Our validation assays nominate several proteins as candidates for further interrogation across longitudinal and other PD patient cohorts. Our study further provides a roadmap for generating comprehensive CSF spectral libraries for use in new, data independent quantitative proteomic assays.