Proteogenomic analysis of cerebrospinal fluid reveals causal role of proteins from the autophagy‐lysosome pathway in Parkinson’s disease

Abstract Background Integrating genetic variants associated with protein levels (protein quantitative trait loci; pQTLs) with variants from genome‐wide association studies (GWAS) using Mendelian randomization approaches has uncovered a causal role of previously unsuspected proteins in several diseases. We have successfully applied a combination of targeted proteomic and genomic approaches to the cerebrospinal fluid (CSF) of Alzheimer Disease (AD) patients to identify multiple novel genetic associations. However, there are currently no proteogenomic studies in Parkinson’s Disease (PD). Methods We used the SOMAscan 1.3k Assay (∼1,305 SOMAmers) in ∼1015 samples from the ADRC‐WUSTL. SOMAscan data from the Parkinson's Progression Markers Initiative cohort and Sasayama D, et al 2017 were used as replication cohorts. After stringent QC steps were applied to SOMA data, each SOMAmer was log 10 transformed and standardized to zero. We used linear additive genetic regression models adjusted for age, sex, genotyping array and the first two principal components using Plink 1.9. We used the R package MendelianRandomization to integrate our CSF pQTL variants with PD‐related SNPs from the 2019 METAPD GWAS. Results We found significant cis ‐pQTLs for Cathepsin B (p = 3.6 × 10 −27 ), Alpha‐L‐iduronidase (p = 3.4 × 10 −41 ) and Galactin‐3 (p = 2.7 × 10 −41 ). MR found significant causal associations of PD risk with pQTLs for IDUA (b = 1.2, p = 5.0 × 10 −6 ), CSTB (b = ‐1.4, p = 2.7 × 10 −4 ) and Galactin‐3 (b = ‐1, p = 4.0 × 10 −2 ). Most importantly, we found an association with progranulin (PGRN) trans ‐pQTLs (p = 2.5 × 10 −9 ). This association was replicated in the PPMI cohort (p = 1.9 × 10 −9 ), and MR confirmed the significant causal link of PGRN (b = 4.6, p = 1.0 × 10 −4 ) with PD. The PGRN pQTL in CSF is located in the LRRK2 gene locus. This SNP is also associated with the CSF levels of proteins from two additional PD‐associated genes CTSB (p = 1.6 × 10 −3 ) and glycoprotein nmb ( GPNMB , p = 1.5 × 10 −4 ). Conclusion Our proteogenomic approach identified LRRK2 as a genetic modifier of three additional PD‐associated proteins in CSF. MR analyses confirmed the causal link of proteins from the autophagy‐lysosome pathway with PD. Our data provide new biological insights into how LRRK2 modifies PD risk and novel markers for screening of LRRK2 kinase activity.