Haploinsufficiency of ITSN1 is associated with a substantial increased risk of Parkinson's disease

Abstract

Despite its significant heritability, the genetic basis of Parkinson's disease (PD) remains incompletely understood. Here, in analyzing whole-genome sequence data from 3,809 PD cases and 247,101 controls in the UK Biobank, we discover that protein-truncating variants in ITSN1 confer a substantially increased risk of PD (p = 6.1 × 10^-7; odds ratio [95% confidence interval] = 10.5 [5.2, 21.3]). We replicate this association in three independent datasets totaling 8,407 cases and 413,432 controls (combined p = 4.5 × 10^-12). Notably, ITSN1 haploinsufficiency has also been associated with autism spectrum disorder, suggesting variable penetrance/expressivity. In Drosophila, we find that loss of the ITSN1 ortholog Dap160 exacerbates α-synuclein-induced neuronal toxicity and motor deficits, and in vitro assays further suggest a physical interaction between ITSN1 and α-synuclein. These results firmly establish ITSN1 as a PD risk gene with an effect size exceeding previously established loci, implicate vesicular trafficking dysfunction in PD pathogenesis, and potentially open new avenues for therapeutic development.

Keywords: CP: Genomics; ITSN1; Intersectin 1; Parkinson's disease; SNCA; autism; endocytosis; neurodegeneration; rare variants; synaptic transmission; synuclein.

Figure 1.. Genetic association study design

(A) Schematic of the genetic association analyses. The discovery cohort included whole-genome sequenced participants from the UKB. The pan-ancestry analysis included individuals of European (EUR), African, East Asian, and South Asian genetic ancestries. The dotted box indicates the three additional EUR cohorts used to replicate the association between ITSN1 and PD. ^†The deCODE replication was performed using summary statistics from a recently published PD collapsing analysis. (B) Schematic of gene-level collapsing models. Qualifying variants were defined based on three major criteria: minor allele frequency (MAF) upper limit, variant impact, and missense scores (see Table S3). The ptvraredmg model, which combines the ptv and raredmg models, is not depicted. ExWAS, exome-wide association study; AMP-PD, Accelerating Medicines Partnership program for Parkinson’s Disease; AoU, All of Us.

Figure 2.. Variant- and gene-level associations with PD

(A and B) Manhattan plots of variant-level (A) and gene-level (B) associations with PD in UKB Europeans. If the same association was detected in multiple models, we depicted the most significant association. Horizontal dashed lines indicate the significance (p <1 × 10⁻⁸) and suggestive (p <1 × 10⁻⁶) thresholds. Orange points indicate marginally significant (p < 1 × 10⁻⁴) associations in established PD genes. (C) Odds ratio versus MAFs for ITSN1 and other established PD loci that achieved p < 1 × 10⁻⁴ in the ExWAS or collapsing analysis. (D) Lollipop plot showing the location of ITSN1 protein-truncating variants in cases against the MANE transcript (ENST00000381318) and PFam domains of the corresponding protein (Q15811); bars below the transcript indicate the long and short isoforms. (E) Association between ITSN1 PTVs and PD across the discovery and replication cohorts. Data are represented as odds ratios and 95% confidence intervals. p values in individual cohorts were generated via a two-tailed Fisher’s exact test; the combined p value was calculated via an exact Cochran-Mantel-Haenszel (CMH) test. EUR, European-ancestry cohort; AMP-PD, Accelerating Medicines Partnership program for Parkinson’s Disease; AoU, All of Us; 100kGP = 100,000 Genomes Project. See also Tables S3–S5; Figures S1–S4.

Figure 3.. Interrogation of ITSN1’s function and connection to PD

(A) Schematic of vesicular trafficking pathway. Previously established PD-associated genes are listed in green. (B) PPI map of ITSN1. Line width depicts confidence of the interaction. (C) Disease phenotypes ranked highly against ITSN1 by the machine-learning algorithm Mantis-ML. All shown are well-established clinical signs of parkinsonism and were ranked by Mantis-ML in the top 1.25% of predicted associations. See also Table S13.

Figure 4.. Dap160(ITSN1) modulates α-synuclein-induced neurodegeneration in Drosophila

(A–F) Representative images of Drosophila eyes among the following genotypes: wild type (A), Dap160^+/− (B), α-Syn/no modifier (C), α-Syn/Dap160^+/− (D), and two independent Dap160 overexpression alleles in the α-Syn background. Scale bar, 100 μm in the top row and 10μm in the bottom row. (G) Quantification of the disorganized eye area as a percentage of total eye area for the genotypes shown in (A)–(F). Bar represents the mean across three flies per genotype; error bars represent standard error of the mean. p values were generated via ANOVA followed by Dunnett’s test using α-Syn/no modifier as control (***p < 0.001). (H and I) Motor performance as a function of age measured via speed (H) and number of stumbling events (I). Each point represents the average speed for a given replicate (n = 10 flies per replicate; see STAR Methods). Lines represent spline fits, and shaded areas represent 95% confidence intervals. p values were obtained using a nonlinear random mixed-effect model ANOVA applied to spline regressions of α-Syn/no modifier vs. α-Syn/Dap160^+/−. α-Syn, α-synuclein; OE, overexpression. See Table S14 for full genotype information.

Figure 5.. Molecular interaction between ITSN1 and α-synuclein

(A) Immunoblots of lysates (input) and FLAG immunoprecipitates (IPs) from HEK293T cells overexpressing FLAG-tagged ITSN1 and α-Syn. Data are representative of n = 3 independent experiments. (B–D) Immunofluorescence against α-Syn (red) and Dap160 (green) in the synaptic boutons of Drosophila neuromuscular junctions from wild-type larvae (B) or larvae expressing α-Syn (C and D). Arrowheads indicate colocalization events. Scale bar, 2 μm. Data are representative of n = 245 neuromuscular junctions across 18 α-Syn and 14 wild-type larvae.