Mutation-induced loss of APP function causes GABAergic depletion in recessive familial Alzheimer's disease: analysis of Osaka mutation-knockin mice

Abstract

The E693Δ (Osaka) mutation in APP is linked to familial Alzheimer's disease. While this mutation accelerates amyloid β (Aβ) oligomerization, only patient homozygotes suffer from dementia, implying that this mutation is recessive and causes loss-of-function of amyloid precursor protein (APP). To investigate the recessive trait, we generated a new mouse model by knocking-in the Osaka mutation into endogenous mouse APP. The produced homozygous, heterozygous, and non-knockin littermates were compared for memory, neuropathology, and synaptic plasticity. Homozygotes showed memory impairment at 4 months, whereas heterozygotes did not, even at 8 months. Immunohistochemical and biochemical analyses revealed that only homozygotes displayed intraneuronal accumulation of Aβ oligomers at 8 months, followed by abnormal tau phosphorylation, synapse loss, glial activation, and neuron loss. These pathologies were not observed at younger ages, suggesting that a certain mechanism other than Aβ accumulation underlies the memory disturbance at 4 months. For the electrophysiology studies at 4 months, high-frequency stimulation evoked long-term potentiation in all mice in the presence of picrotoxin, but in the absence of picrotoxin, such potentiation was observed only in homozygotes, suggesting their GABAergic deficit. In support of this, the levels of GABA-related proteins and the number of dentate GABAergic interneurons were decreased in 4-month-old homozygotes. Since APP has been shown to play a role in dentate GABAergic synapse formation, the observed GABAergic depletion is likely associated with an impairment of the APP function presumably caused by the Osaka mutation. Oral administration of diazepam to homozygotes from 6 months improved memory at 8 months, and furthermore, prevented Aβ oligomer accumulation, indicating that GABAergic deficiency is a cause of memory impairment and also a driving force of Aβ accumulation. Our findings suggest that the Osaka mutation causes loss of APP function, leading to GABAergic depletion and memory disorder when wild-type APP is absent, providing a mechanism of the recessive heredity.

Keywords: Alzheimer’s disease; GABA; Knockin mouse; Loss of function; Recessive mutation.

Fig. 1

Generation of OSK-KI mice. (a) Mice were generated by knocking-in the Osaka mutation (deletion of codon 693) into endogenous mouse APP by homologous recombination in embryonic stem cells. (i) Mouse APP contains 18 exons (black boxes), and Aβ is coded in exons 16 and 17. (ii) The targeting vector contains mouse APP exon 16, mutant exon 17 with the deletion (white box), and the neomycin-resistance gene driven by the phosphoglycerate kinase 1 promoter (PGK-neo). (iii) Homologous recombinants were determined by Southern blotting using the 5' and 3' probes. (b) Expression levels of APP in homozygous, heterozygous, and non-KI mice. Brain homogenates at 24 months were subjected to Western blot with antibodies to APP C-terminus (C40) and actin. Each bar represents the mean ± SEM (n = 4 for each group). AU, arbitrary unit

Fig. 2

Memory impairment in OSK-KI mice. Spatial reference memory of mice was examined at 4 (a), 6 (b), and 8 months (c) using the Morris water maze. Each point represents the mean latency of five trials per day ± SEM. (a) *p < 0.05 versus hetero-KI when means of day 4 were separately compared using the Tukey-Kramer test. (b) *p < 0.05 versus non-KI and <0.05 versus hetero-KI. (c) *p < 0.05 versus non-KI and <0.05 versus hetero-KI

Fig. 3

Aβ accumulation in OSK-KI mice. Brain sections were stained with anti-Aβ42 Ter42 (a), anti-Aβ N-terminus β001 (b) and Aβ oligomer-specific 11A1 antibodies (c). Photographs were taken from the posterior parietal association area (PPtA) of the cerebral cortex (CTX), hippocampal CA3 region (CA3), dentate gyrus (DG), and entorhinal cortex (EC). Arrowheads indicate Aβ accumulated within neurons. Scale bar = 30 μm. (d) Brain homogenates at 4 and 8 months were separated into TBS-soluble and insoluble (SDS-soluble) fractions and subjected to Aβ42 sandwich ELISA. Each bar represents the mean ± SEM (n = 4 for each group). *p = 0.0391 versus hetero-KI, **p = 0.0255 versus non-KI and = 0.0441 versus hetero-KI. (e) For Aβ oligomers, brain homogenates were subjected to direct ELISA with 11A1 antibody. Each bar represents the mean ± SEM (n = 4 for each group). †p = 0.0001 versus non-KI and = 0.0002 versus hetero-KI. (f) HEK293 cells were transfected with human (h) or mouse (m) APP_SW and APP_SW/OSK constructs. Three days after transfection, the cells were homogenized and subjected to Western blot to measure APP expression (C40) and actin. Intracellular Aβ were immunoprecipitated using anti-Aβ antibody β001 and subjected to Western blot with the same antibody

Fig. 4

Aβ-related neuropathology in OSK-KI mice. Abnormal tau phosphorylation, synapse loss, glial activation, and neuron loss were examined. (a) Brain sections were stained with PHF-1 antibody specific to pSer396/Ser404-tau. Photographs were taken from the hippocampal CA2/3 region. Scale bar = 30 μm. (b) Brain homogenates at 8 months were subjected to pSer396-tau sandwich ELISA. Each bar represents the mean ± SEM (n = 4 for each group). (c) Brain sections were stained with anti-synaptophysin antibody. Photographs were taken from the hippocampal CA3 region. Scale bar = 30 μm. Fluorescence intensities in a constant area were quantified as described previously [28]. Each bar represents the mean ± SEM (n = 3 for each group). AU, arbitrary unit. *p = 0.0019 versus non-KI and = 0.0093 versus hetero-KI, †p = 0.0052 versus non-KI and = 0.0424 versus hetero-KI, ‡p < 0.0001 versus non-KI and = 0.0278 versus homo-KI, ‡‡p = 0.0008 versus non-KI. (d) Hippocampal homogenates at 4 and 8 months were subjected to Western blot with antibodies to synaptophysin and actin. The signal densities were quantified. Each bar represents the mean ± SEM (n = 3 for each group). (e) Brain sections were stained with antibodies to Iba-1 (microglia) and GFAP (astrocytes). Photographs were taken from the hippocampus. Scale bar = 30 μm. (f) Brain sections at 24 months were stained with anti-NeuN antibody. Photographs were taken from the hippocampal CA2/3 region (HC) and entorhinal cortex (EC). Scale bar = 30 μm. Neu-N-positive neurons in an area within 300 μm along the pyramidal cell layer of the hippocampal CA3 region and in an area within 1000 μm along the layer II (the region between the two broken lines) of the entorhinal cortex were counted, essentially as described previously [26]. *p = 0.0448 versus non-KI and = 0.0245 versus hetero-KI, †p = 0.0285 versus non-KI and = 0.0207 versus hetero-KI (n = 4 for each group)

Fig. 5

Aberrant synaptic activity in OSK-KI mice. Synaptic plasticity was examined by electrophysiology using hippocampal slices at 4 (a, c) and 8 months (b, d). HFS (100 Hz, 100 pulses) was delivered to the dentate gyrus and fEPSP was recorded for 60 min in the same region in the presence (a, b) and absence (c, d) of 40 μM picrotoxin. Right panels, the levels of fEPSP slope at 60 min were compared. Each bar represents the mean ± SEM. *p < 0.05 versus non-KI and <0.05 versus hetero-KI

Fig. 6

GABAergic neuron loss in OSK-KI mice. (a) Brain sections at 4 months were stained with anti-parvalbumin antibody. Photographs were taken from the dentate gyrus (DG) and entorhinal cortex (EC). Scale bar = 30 μm. Parvalbumin-positive GABAergic neurons in the granular cell layer (GCL) of the dentate gyrus within 1000 μm and in an area (700 × 600 μm) of the entorhinal cortex were counted. Each bar represents the mean ± SEM (n = 7 for each group). (b) Hippocampal homogenates at 4 months were subjected to Western blot with antibodies to GAD67/65 (GABA-synthetic enzyme), VGAT (GABA transporter), VGLUT1/2 (glutamate transporter), and actin. The signal densities were quantified. Each bar represents the mean ± SEM (n = 3 for each group)

Fig. 7

Effects of diazepam treatment on memory and Aβ pathology in OSK-KI mice. Diazepam (DZP) was orally administered to 6-month-old homo-KI mice at 2 μg/day for 2 months. (a) Spatial reference memory was examined at 8 months using the Morris water maze. Each point represents the mean latency of five trials per day ± SEM. *p < 0.05 versus non-KI. (b) Brain sections were stained with Aβ oligomer-specific 11A1 antibody. Photographs were taken from the PPtA of the cerebral cortex (CTX), hippocampal CA3 region (CA3), dentate gyrus (DG), and entorhinal cortex (EC). Arrowheads indicate Aβ accumulated within neurons. Scale bar = 30 μm. (c) Brain sections were stained with anti-synaptophysin antibody. Photographs were taken from the hippocampal CA3 region. Scale bar = 30 μm. Fluorescence intensities in a constant area were quantified. Each bar represents the mean ± SEM (n = 5 for each group). AU, arbitrary unit. (d) Brain sections were stained with anti-parvalbumin antibody. Photographs were taken from the dentate gyrus. Scale bar = 30 μm. Parvalbumin-positive GABAergic neurons in a constant area in the granular cell layer (GCL) of the dentate gyrus were counted. Each bar represents the mean ± SEM (n = 5 for each group)

Fig. 8

A proposed mechanism for the recessive heredity of the Osaka mutation. The Osaka mutation has dual effects; a loss-of-function of APP leading to GABAergic depletion and gain-of-toxic-function of Aβ to accelerate its oligomerization. While the former effect accounts for the recessive inheritance of this mutation, the latter seems to be a dominant effect to cause dementia even in heterozygotes. But practically, Aβ oligomer accumulation is not detected until GABAergic depletion proceeds. This is probably because Aβ production is negatively regulated by GABAergic inhibitory input. Once Aβ oligomers accumulate, it triggers pathological cascade of AD, including abnormal tau phosphorylation, synapse loss, glial activation, and eventual neuron loss