Rare Genetic Risk in Progressive Supranuclear Palsy

Background: Genetic risk in common variants have been identified for Progressive Supranuclear Palsy (PSP), but common variants only account for a portion of disease heritability. A proportion of the remaining disease heritability is likely contributed by rare variants identifiable via sequencing. Leveraging whole-genome sequencing of 1719 PSP cases and 2940 controls from the Alzheimer’s Disease Sequencing Project, we detected single nucleotide variants (SNVs), insertion/deletions (INDELs), and copy number variants (CNVs). Note that all samples are non-Hispanic whites with identity by descent (IBD) >0.25, and no outlier is >6 standard deviation in the principal component analysis. We conducted association analysis for variants, genes and gene sets to identify rare genetic risk for PSP. Method: After quality control (variants and genotypes notated PASS by the calling algorithms), we detected 85,228,234 rare SNVs/INDELs and 163,987 rare CNVs (minor allele frequency (MAF) < 1%). For rare SNVs/INDELs, the Sequence Kernel Association Test Optimized was performed to identify rare genetic PSP risk from protein truncating variants (PTV) and damaging missense variants (MAF<0.01%) in both protein coding genes and gene sets that were previously associated with PSP based on protein co-expression networks. For rare CNVs, a permutation-based association test implemented in Plink was performed to identify CNVs associated with PSP. Result: Combining protein truncating variants (PTV) and damaging missense variants, we identified one genome-wide significant gene, FBXO38 (FDR = 0.0003, OR = 1.78). We also found enrichment of PTVs and damaging missense variants in genes comprising the C2 module, a neuronal module previously identified in post mortem PSP brain (FDR = 0.0016). For CNV analysis, there were 17 significant deletions and 2 significant duplications that were associated with PSP (P_adjust < 0.05). The deletion (18:58488230-58488489) in ALPK2, which has been identified as an Alzheimer’s Disease risk gene, was the only CNV that conferred a higher-risk for PSP. Particularly, one deletion (17:46009357-46009595) was in linkage disequilibrium with MAPT H2 haplotype which has been associated various neurodegenerative diseases including PSP. Conclusion: We identified several new genes and a protein co-expression network in which risk variants were associated with PSP. These genes and pathways provide potential candidate drug targets for future investigation.