Coding mutations in SORL1 and Alzheimer disease

Abstract

Objective: Common single nucleotide polymorphisms in the SORL1 gene have been associated with late onset Alzheimer disease (LOAD), but causal variants have not been fully characterized nor has the mechanism been established. The study was undertaken to identify functional SORL1 mutations in patients with LOAD.

Methods: This was a family- and cohort-based genetic association study. Caribbean Hispanics with familial and sporadic LOAD and similarly aged controls were recruited from the United States and the Dominican Republic, and patients with sporadic disease of Northern European origin were recruited from Canada. Prioritized coding variants in SORL1 were detected by targeted resequencing and validated by genotyping in additional family members and unrelated healthy controls. Variants transfected into human embryonic kidney 293 cell lines were tested for Aβ40 and Aβ42 secretion, and the amount of the amyloid precursor protein (APP) secreted at the cell surface was determined.

Results: Seventeen coding exonic variants were significantly associated with disease. Two rare variants (rs117260922-E270K and rs143571823-T947M) with minor allele frequency (MAF) < 1% and 1 common variant (rs2298813-A528T) with MAF = 14.9% segregated within families and were deemed deleterious to the coding protein. Transfected cell lines showed increased Aβ40 and Aβ42 secretion for the rare variants (E270K and T947M) and increased Aβ42 secretion for the common variant (A528T). All mutants increased the amount of APP at the cell surface, although in slightly different ways, thereby failing to direct full-length APP into the retromer-recycling endosome pathway.

Interpretation: Common and rare variants in SORL1 elevate the risk of LOAD by directly affecting APP processing, which in turn can result in increased Aβ40 and Aβ42 secretion.

FIGURE 1

Histogram of −log10 of the probability values obtained from SNP-set Kernel Association Test (SKAT) analysis of 1,000 data sets created by randomly choosing 1 subject from each of the 87 families and 498 controls. The SKAT analysis was conducted assuming for the unadjusted model: Alzheimer disease (AD) ~ single nucleotide polymorphism (SNP) burden; and for the model with age, sex, and apolipoprotein E (APOE) ε4 status as covariates: AD ~ SNP burden + age + sex + APOE ε4 yes/no. [Color figure can be viewed in the online issue, which is available at www.annalsofneurology.org.]

FIGURE 2

Overexpression of SORL1 mutants leads to elevated Aβ secretion. (A–C) Measurement of secreted Aβ40, Aβ42 and sAPPβ from culture medium in stable HEK293 cells expressing the APP Swedish mutant (HEKsw) together with either wild-type (wt) SORL1 or mutant (mut) SORL1. Aβ levels were normalized to the protein levels of the cell lysates. Error bars = standard error of the mean. *p < 0.05, **p < 0.01, ***p < 0.001, ns = not significant after Bonferroni correction; n = 3 independent replications. (D) Cultured media from cells were collected and subjected to Western blot and probed with 6E10 antibody to detect sAPPα. Bar graphs were normalized to control. **p < 0.01, ***p < 0.001, after Bonferroni correction; n = 3 independent replications. (E) Cell lysates were harvested to perform Western blot of full-length amyloid precursor protein (FL-APP) and PS1. β-Actin was used as loading control; n = 3 independent replications.

FIGURE 3

The expression of SORL1 mutants (mut) leads to changes of cell surface amyloid precursor protein (APP) and SORL1 levels. Cell surface proteins were biotinylated and precipitated. Surface levels of APP and SORL1 were analyzed by Western blot. (A) APP levels at the cell surface are elevated in all 3 mutants. *p < 0.05, **p < 0.01, after Bonferroni correction, n = 3 replications. (B) SORL1 surface levels are decreased in the T947M mutant. **p < 0.01, ns = not significant, after Bonferroni correction, n = 3 replications. FL = full length; IP = immunoprecipitated; wt = wild type.

FIGURE 4

All 3 SORL1 mutants (mut) have a reduced binding affinity to amyloid precursor protein (APP). SORL1 was pulled down from cell lysates with a c-MYC antibody and the amount of coprecipitated full-length APP (FL-APP) was measured. *p < 0.05, **p < 0.01, after Bonferroni correction, n = 3 replications. IgG = immunoglobulin G; IP = immunoprecipitated; wt = wild type.

FIGURE 5

Position of the coding mutations relative to the single nucleotide polymorphisms (SNPs) significantly associated with Alzheimer disease (Rogaeva et al).