ER-associated degradation regulates Alzheimer's amyloid pathology and memory function by modulating γ-secretase activity

Abstract

Endoplasmic-reticulum-associated degradation (ERAD) is an important protein quality control system which maintains protein homeostasis. Constituents of the ERAD complex and its role in neurodegeneration are not yet fully understood. Here, using proteomic and FRET analyses, we demonstrate that the ER protein membralin is an ERAD component, which mediates degradation of ER luminal and membrane substrates. Interestingly, we identify nicastrin, a key component of the γ-secretase complex, as a membralin binding protein and membralin-associated ERAD substrate. We demonstrate a reduction of membralin mRNA and protein levels in Alzheimer's disease (AD) brain, the latter of which inversely correlates with nicastrin abundance. Furthermore, membralin deficiency enhances γ-secretase activity and neuronal degeneration. In a mouse AD model, downregulating membralin results in β-amyloid pathology, neuronal death, and exacerbates synaptic/memory deficits. Our results identify membralin as an ERAD component and demonstrate a critical role for ERAD in AD pathogenesis.

Fig. 1

Proteomic analysis of membralin-interacting proteins. a Venn diagram showing the number of interacting proteins identified from cell lysates prepared in NP40 (red), Digitonin (yellow), and TritonX-100 (blue). b Among 180 proteins identified in the membralin interactome (see Supplementary Fig. 1), a heatmap of 14 known ERAD components is presented. The color intensity represents protein intensity (summed peptide area under the curve) (c) Interaction network of the membralin interactome by IPA. Three main ERAD subnetworks were highlighted in green (SYVN1 subnetwork), blue (AMFR subnetwork) and yellow (EMC subnetwork). Nicastrin was highlighted in orange color. Solid and dashed lines represent direct interactions and indirect interactions, respectively

Fig. 2

Confirmation of membralin-interacting protein components by FRET and co-IP. a FRET images of cells co-transfected with vectors expressing membralin-mCherry and either SYVN1-EGFP, AMFR-EGFP, EMC3-EGFP or Ubxd8-EGFP. Scale bar, 10 µm. b Quantification of FRET efficiency between membralin and SYVN1, AMFR, EMC3, Ubxd8, Derlin1, EMC1, Ubac2, FAM8A1, respectively. mCherry-7AA-EGFP was used as a positive control (n = 10 cells). Data represent mean ± s.e.m. from three independent experiments. c Co-IP analysis of endogenous membralin interactions with endogenous SYVN1, or AMFR in N2a cells under normal and ER stress conditions induced by tunicamycin (5 μg/ml) or thapsigargin (500 nM) for 4 hrs. The fold effect shown was calculated using the ratio from AMFR or SYVN1 (IP/Input) normalized to the membralin IP (IP/β-actin). Non-treated cell samples were set to 1.0

Fig. 3

NHK (ERAD-L) and CD3δ (ERAD-M) degradation in HEK293T cells transfected with membralin siRNA. a Representative images depicting reconstituted NHK-GFP fluorescence were shown from cells transfected with control or membralin siRNA and subsequently co-transfected with S1-10 and SP-S11-NHK-HA without proteasome inhibitor MG132 treatment. b Representative images of NHK-GFP-positive cells transfected with control or membralin siRNA and exposed to MG132 (10 µM) for 4 h. c Quantification of NHK-GFP-positive cells with control and membralin siRNA transfection with or without MG132 treatment (n = 14 random field images per group). d Quantified GFP intensity in NHK-GFP-positive cells transfected with control or membralin siRNAs, with or without MG132 treatment (n = 14 random field images per group). e Representative images of reconstituted CD3δ-GFP fluorescence were shown from cells transfected with control or membralin siRNA and subsequently co-transfected with S1-10 and SP-S11-CD3δ-HA without proteasome inhibitor MG132 treatment. f Representative images of CD3δ-GFP-positive cells transfected with control or membralin siRNA and exposed to MG132 (10 µM) for 4 h. g Quantification of CD3δ-GFP-positive cells with control or membralin siRNA transfection with or without MG132 treatment (n = 14 random field images per group). h Quantified GFP intensity in CD3δ-GFP-positive cells transfected with control or membralin siRNAs, with or without MG132 treatment (n = 14 random field images per group). Data represent mean ± s.e.m. from three independent experiments. ***P < 0.001, unpaired t-test. Scale bar, 100 µm

Fig. 4

Nicastrin is a membralin-dependent ERAD substrate. a Endogenous co-IP of nicastrin with membralin. Mouse brain organelle-enriched (nuclear and cytosol-free) lysates were precipitated with an anti-membralin antibody and immunoblotted with anti-nicastrin and membralin antibodies as indicated. A lower molecular weight band (~ 75 kDa) in the nicastrin blot was the brain organelle-enriched lysates treated with PNGase F to remove the glycosylation of nicastrin, serving as unglycosylated control. Untreated sample was used for endogenous co-IP, and shown in the input. b Western blot analysis and quantification of nicastrin protein levels detected in neuronal tissue from Mem +/+(n = 4 mice) and Mem−/− (n = 5 mice) animals. Data represent mean ± s.e.m. unpaired t-test, **P < 0.01. c Western blot analysis of nicastrin protein levels measured in Mem+/+or Mem−/− mouse NPCs treated with 10 µg/ml CHX for the indicated time. Data represent mean ± s.e.m. from three independent experiments, two-way ANOVA, **P < 0.01. d Native PAGE western blot analysis and quantification of γ-secretase subunits (nicastrin, PEN2 and PS1-NTF) in brains from Mem+/+(n = 4 mice) and Mem−/− (n = 5 mice) animals normalized to β-actin levels from an equivalent quantity of denatured lysate. Data represent mean ± s.e.m. unpaired t-test, *P < 0.05. e Western blot analysis and quantification of NICD protein levels in Mem+/+ and Mem−/− mouse NPCs. Data represent mean ± s.e.m. from 3 independent experiments, unpaired t-test, * P < 0.05. f γ-secretase activity was analyzed by measuring Aβ40 and Aβ42 generation in purified membrane fractions derived from Mem+/+ (n = 4 mice) and Mem−/− (n = 5 mice) mouse brains. Data represent mean ± s.e.m. unpaired t-test, * P < 0.05, ** P < 0.01. g Golgi staining of the hippocampus of Mem+/+ and Mem−/− mice (postnatal 3-day-old, n = 3 mice per genotype). Scale bars, 600 µm (left), and 60 µm in the magnified images (right). Number of Golgi-staining⁺ neurons per hippocampus (n = 9 sections per genotype) and number of primary dendrites per cell were quantified (n = 30-32 neurons per genotype). Data represent mean ± s.e.m. unpaired t-test, ***P < 0.001

Fig. 5

Membralin expression levels are dysregulated in AD brain. a Intron/exon distribution of human membralin long and short isoforms (top), and schematic depiction of membralin long (620aa) and short (408aa) protein isoforms (bottom). Primers used to detect total and membralin long isoforms were shown. b Quantification of total and long membralin transcripts in non-AD and AD brain normalized to the 18 s ribosomal RNA control (n = 10 for non-AD samples; n = 12 for AD samples). Data represent mean ± s.e.m. unpaired t-test, * P < 0.05, **P < 0.01. c Membralin expression detected by immunohistochemistry staining (brown) with hematoxylin counter-staining (purple) in human postmortem tissues (three random field images from each sample were quantified using Leica Aperio analysis software in a blinded manner from a total of n = 7 samples per group). Data generated from individual samples are indicated with a sample ID. Scale bars, 15 μm. Data represent mean ± s.e.m. unpaired t-test, **P < 0.01. d, e Western blot analysis and quantification of d membralin and e nicastrin protein levels in non-AD (n = 7) and AD (n = 8) samples. Representative images from the same blots are shown. Asterisks indicate membralin d and nicastrin e. Data represent mean ± s.e.m. unpaired t-test, **P < 0.01, n.s. no significance. f Correlation analysis of membralin (X axis) with nicastrin protein levels (Y axis) in non-AD and AD samples. Y = -0.3401 * X + 0.8727, R ² = 0.3982, P = 0.0116 (n = 7 for non-AD samples, n = 8 for AD samples); the slope was found to be significantly deviant from non-zero by linear regression analysis

Fig. 6

Modulation of membralin expression in TgCRND8 mice. a Membralin protein levels from TgCRND8 mouse hippocampus injected with scrambled control or membralin shRNA constructs. Protein levels were determined by immunoprecipitation and immunoblot using an anti-membralin antibody. Protein levels were normalized relative to β-actin (n = 3 mice per group). b XBP1-s and CHOP protein levels from TgCRND8 mouse hippocampus injected with scrambled control or membralin shRNA constructs. c Nicastrin protein levels normalized to β-actin in the hippocampus of scrambled control (n = 6) and membralin shRNA (n = 7) injected mice. d Aβ plaque staining and quantification was determined by staining with the MOAB2 Aβ antibody in the DG region (n = 7 for scrambled control group; n = 4 for membralin shRNA injected animals). e Human Aβ40 and Aβ42 levels in hippocampal lysates from 6-8-month-old TgCRND8 mice injected with scrambled control or membralin shRNA lentiviruses were determined by ELISA (n = 6 mice per group). g, h Histological analysis of 6-8-month-old TgCRND8 mice injected with scrambled control or membralin shRNA lentiviruses. Presence of active (cleaved) caspase 3 and apoptotic TUNEL staining in the dentate gyrus (DG) region and quantification is presented in g and h, respectively. Data represent mean ± s.e.m. unpaired t-test, *P < 0.05, **P < 0.01, and ***P < 0.001. Scale bars, 300 µm (left) and 200 µm in magnified images (right)

Fig. 7

AD-associated behaviors in TgCRND8 mice injected with membralin shRNA. a PSD95 levels in scrambled control or membralin shRNA-injected hippocampal tissue as determined by immunoblot, and normalized relative to β-actin (n = 6 mice per group). b, c Barnes maze analysis: b Comparison of primary latency to target in control or membralin shRNA-injected groups in five training trials (n = 9 mice per group). c Comparison of primary latency to target in control or membralin shRNA-injected groups in probe tests (n = 9 mice per group). d, e Fear conditioning analysis: d Comparison of freezing time (% time) in shRNA-injected experimental cohorts (n = 9 mice per group) during the pre-conditioning trial. e Comparison of freezing time (% time) following electrical foot-shock in shRNA-injected experimental cohorts (n = 9 mice per group) during the probe test. Data represent mean ± s.e.m. unpaired t-test, *P < 0.05, **P < 0.01, and ***P < 0.001

Fig. 8

A schematic model of the ERAD component membralin in regulating nicastrin and Aβ generation. Membralin interacts with several known ERAD components and mediates nicastrin degradation. Membralin deficiency increases nicastrin levels through inhibiting nicastrin degradation, thereby increasing Aβ generation