A rare mutation in UNC5C predisposes to late-onset Alzheimer's disease and increases neuronal cell death

Abstract

We have identified a rare coding mutation, T835M (rs137875858), in the UNC5C netrin receptor gene that segregated with disease in an autosomal dominant pattern in two families enriched for late-onset Alzheimer's disease and that was associated with disease across four large case-control cohorts (odds ratio = 2.15, Pmeta = 0.0095). T835M alters a conserved residue in the hinge region of UNC5C, and in vitro studies demonstrate that this mutation leads to increased cell death in human HEK293T cells and in rodent neurons. Furthermore, neurons expressing T835M UNC5C are more susceptible to cell death from multiple neurotoxic stimuli, including β-amyloid (Aβ), glutamate and staurosporine. On the basis of these data and the enriched hippocampal expression of UNC5C in the adult nervous system, we propose that one possible mechanism in which T835M UNC5C contributes to the risk of Alzheimer's disease is by increasing susceptibility to neuronal cell death, particularly in vulnerable regions of the Alzheimer's disease brain.

Figure 1. Identification of T835M UNC5C as a candidate variant for LOAD

(a) A large family with a propensity for developing LOAD was selected for study. Linkage arrays were run on the indicated sixteen individuals with available DNA. An affected individual (indicated by *) was whole genome sequenced and a second case was exome sequenced (indicated by **). (b) Linkage results from the pedigree in (a) using a parametric (rare dominant) model. Five regions segregated with LOAD with a LOD score >1.0. (c) A second pedigree in which T835M segregates with LOAD. Heterozygous carriers of T835M are indicated. (d–f) Additional LOAD enriched pedigrees with at least one member heterozygous for UNC5C T835M (rs137875858). Hollow and filled diamonds indicate healthy family members and those diagnosed with AD respectively. A Taqman assay was used to confirm genotype at T835M (rs137875858) and heterozygous carriers are indicated by “+” and non-carriers a “−”. Individuals with an age at last visit of <70 are shaded in blue.

Figure 2. Association of T835M UNC5C (rs137875858) in AD case and control cohorts

The association of T835M UNC5C (rs137875858) is shown in four independent case and control cohorts totaling 8,050 LOAD cases and 98,194 control individuals. The ADGC cohort includes 12,789 additional controls as described in the supplementary information.

Figure 3. T835M UNC5C increases cell death beyond that induced by WT UNC5C

Representative growth curves of (a) GFP+ cells (transfected only) and (b) total cells (transfected and non-transfected) (c) Growth of βgal-, T835M-, or WT-expressing cells. GFP+ specific growth was normalized to total growth, and slopes of the best-fit lines were calculated. Data represents four independent transient transfection experiments with n=6 randomly plated technical replicates per condition for each experiment. ***p=0.007, ** p=0.0011, *p=0.0259. (d) Representative flow cytometry plots of βgal-, WT- and T835M UNC5C-transfected cells analyzed for GFP and AnnexinV-APC; untransfected cell population (orange), GFP+ AnnexinV- cell population (green), and GFP+ AnnexinV+ population (black). (e) Percentage of AnnexinV+ cells within the GFP+ population representing four independent transient transfection experiments with 100,000 cells analyzed per condition. ***p< 0.0001, ** p=0.0014, *p=0.0111. (f) Representative overlay of fluorescent and bright field images from time-lapse imaging experiments; transfected cells (red), caspase-3/7 activity (green); Scale bar = 50 μM (g) Average lifetime of cells expressing βgal (n=919), WT UNC5C (n=687), or T835M UNC5C (n=590) at movie start or that initiated expression during first 20 h of imaging. (h) Kaplan-Meier survival curves of cells expressing βgal (grey, n=1696), WT (blue, n=1085), or T835M UNC5C (red, n=751) analyzed by Wald test for p-values. (g–h) Data represent three independent transient transfection experiments with n=6 randomly plated technical replicates per experiment and 24 movies per genotype, ***p< 0.0001. Data was analyzed with experimenter blinded to samples, represents the mean ± SEM, and was analyzed by two-tailed unpaired t-test.

Figure 4. UNC5C is expressed in the adult hippocampus and T835M variant generally sensitizes hippocampal neurons to neurodegenerative cues

(a) Unc5c in situ hybridization (red) Unc5c^+/+ and Unc5c^−/− adult mouse hippocampal coronal sections counterstained with hematoxylin (blue) (b) Unc5c quantitative PCR (qPCR) of Unc5c^+/+ and Unc5c^−/− mouse hippocampus (Hipp), cortex (Ctx), and cerebellum (Cbm), fold change normalized to Unc5c^−/−, n=3 mice per genotype. (c, d) qPCR of adult Unc5c^+/+ mouse hippocampi flow cytometry sorted for: neurons (NeuN), astrocytes (GFAP), microglia/ macrophages (CD11b) or a non-enriched negative population (c) Unc5c (d) Trem2, fold change normalized to negative population, n=3 Unc5c^+/+ mice. (e) Representative images of cultured hippocampal neuron expression of βgal-, WT- and T835M UNC5C-IRES-GFP vectors at plateau of expression (48 h) and terminal time point (116 h). (f) Representative plot of GFP+ expression over time βgal (gray), WT (blue) and T835M UNC5C (red) (g) n=5 independent experiments with 60 images per condition, WT vs. T835M UNC5C by two-tailed unpaired t-test *p=0.0394, or two-tailed one sample t-test with βgal (theoretical mean) vs. WT *p=0.0319, or vs. T835M **p=0.002, respectively. (h–j) βgal gray), WT (blue) and T835M (red) UNC5C-IRES-GFP vector expressing hippocampal neurons were challenged with (h) Aβ_1–42 or scrambled Aβ_1–42 (i) glutamate (j) staurosporine. Surviving neurotoxin-treated cells normalized to their respective untreated vector controls (h) n=4 independent neuronal culture experiments, βgal (n=120 images) vs. WT (n=120 images) p=0.0609, WT vs. T835M UNC5C (n=120 images) ****p=1.77×10⁻⁹(j) n=3 independent experiments, βgal (n=90 images) vs. WT (n=90 images) p=0.6032, WT vs. T835M UNC5C (n=90 images) **p=0.001446 (i) n=3 independent experiments, βgal (n=90 images) vs. WT (n=90 images) p=0.323, WT vs. T835M UNC5C (n=90 images) **p=0.0005545. (b–d, g–j) data presented as mean ± SEM, (h–j) P value is from a likelihood-ratio test comparing two linear mixed models for growth curve analysis (fit using function lmer in the R package lme4): 1. model with the Dose: Treatment interaction term and 2. model without that term.