Presenilin1 familial Alzheimer disease mutants inactivate EFNB1- and BDNF-dependent neuroprotection against excitotoxicity by affecting neuroprotective complexes of N-methyl-d-aspartate receptor

Abstract

Excitotoxicity is thought to play key roles in brain neurodegeneration and stroke. Here we show that neuroprotection against excitotoxicity by trophic factors EFNB1 and brain-derived neurotrophic factor (called here factors) requires de novo formation of 'survival complexes' which are factor-stimulated complexes of N-methyl-d-aspartate receptor with factor receptor and presenilin 1. Absence of presenilin 1 reduces the formation of survival complexes and abolishes neuroprotection. EPH receptor B2- and N-methyl-d-aspartate receptor-derived peptides designed to disrupt formation of survival complexes also decrease the factor-stimulated neuroprotection. Strikingly, factor-dependent neuroprotection and levels of the de novo factor-stimulated survival complexes decrease dramatically in neurons expressing presenilin 1 familial Alzheimer disease mutants. Mouse neurons and brains expressing presenilin 1 familial Alzheimer disease mutants contain increased amounts of constitutive presenilin 1-N-methyl-d-aspartate receptor complexes unresponsive to factors. Interestingly, the stability of the familial Alzheimer disease presenilin 1-N-methyl-d-aspartate receptor complexes differs from that of wild type complexes and neurons of mutant-expressing brains are more vulnerable to cerebral ischaemia than neurons of wild type brains. Furthermore, N-methyl-d-aspartate receptor-mediated excitatory post-synaptic currents at CA1 synapses are altered by presenilin 1 familial Alzheimer disease mutants. Importantly, high levels of presenilin 1-N-methyl-d-aspartate receptor complexes are also found in post-mortem brains of Alzheimer disease patients expressing presenilin 1 familial Alzheimer disease mutants. Together, our data identify a novel presenilin 1-dependent neuroprotective mechanism against excitotoxicity and indicate a pathway by which presenilin 1 familial Alzheimer disease mutants decrease factor-depended neuroprotection against excitotoxicity and ischaemia in the absence of Alzheimer disease neuropathological hallmarks which may form downstream of neuronal damage. These findings have implications for the pathogenic effects of familial Alzheimer disease mutants and therapeutic strategies.

Keywords: excitotoxicity; neuronal survival; post-synaptic currents; survival complexes; trophic factors.

Graphical Abstract

Figure 1

Role of PS1 on EFNB1-stimulated EPHB2-GLUN1 association and neuroprotection. Primary cortical neuronal cultures prepared from WT (PS1 +/+), PS1 hemizygous (PS1 +/−) r PS1 homozygous knock out (PS1 −/−) mouse embryonic brains, were stimulated at 10 DIV with EFNB1-Fc (eB1, 2 µg/ml) or Fc for 60 min, lysed and immunoprecipitated (IPed) with anti-EPHB2 antibody. (A) Obtained immunoprecipitates (IPs) were then probed on WBs with anti-GLUN1 or anti-EPHB2 antibodies (upper panel). Input: neuronal lysates used in this IP experiment were immunoblotted with antibodies against GLUN1, EPHB2, PS1-CTF and ACTIN as shown in figure (lower panel). NIS = non-immune serum. (B) Graphs show quantification of EPHB2-GLUN1 complexes IPed as above. Data are mean ± SEM. *P < 0.05 versus Fc, two-tailed paired t-test, n = 3. (C) Down-regulation of PS1 using PS1 siRNA decreases the EFNB1-dependent complex between EPHB2 and GLUN1. Scrambled siRNA was used as control. (D) ExEPHB2 inhibits the EFNB1-induced association of EPHB2 with GLUN1. Neuronal cultures were pretreated for 1 h with either exEPHB2 or BSA (50 nM) followed by EFNB1 (eB1) stimulation for 1 h. Neurons were lysed afterwards, and extracts were IPed with anti-GLUN1 followed by WB with anti-EPHB2 and anti-GLUN1 antibodies. All detected antigens of co-IPs and Inputs are indicated at left of figures. (E) Extracellular peptide, exEPHB2, abolishes EFNB1 mediated neuroprotection from glutamate excitotoxicity. WT primary cortical neuronal cultures were pretreated for 1 h as indicated in the figure, and then challenged with glutamate (G) at 50 μM for 3 h. Neuronal survival was then determined as described in Materials and methods section. Percent survival is normalized to the survival of the culture treated with glutamate (Glu) alone (dotted horizontal line). EFNB1 rescues neurons from excitotoxicity in the absence (third bar from left) but not in the presence of peptide ex-EPHB2 (50 nM; 4th bar). BSA = bovine serum albumin. Data are mean ± SEM. *P < 0.05, **P < 0.01; one-way ANOVA with Tukey’s post hoc test, n = 3–5. Cell survival in the presence of glutamate is set at 100%.

Figure 2

Effects of PS1 FAD mutants on factor-dependent neuroprotection, and EPHB2-GLUN1 association. (A) Cortical neuronal cultures prepared from E15 mouse embryos wild-type (WT), heterozygous (M146V/WT and I213T/WT) or homozygous (M146V/M146V and I213T/I213T) for PS1 FAD alleles M146V or I123T were cultured in 24-well plates and 10–12 days later were treated with either EFNB1-Fc (eB1, 2 µg/ml) or BDNF (50 ng/mL) for 30 min. Cultures were then treated with 50 μM glutamate (Glu) for 3 h. Neuronal viability was quantified by counting healthy nuclei stained with Hoechst kit 33342 as described (Barthet et al., 2013). Percent survival of each neuronal genotype is normalized to the survival of the same culture treated with glutamate alone (dotted horizontal line). Data are mean ± SEM. *P < 0.05, two-tailed paired t-test, n = 4. Cell survival in the presence of glutamate is set at 100%. (B) Cortical neuronal cultures from WT, heterozygous (KI/WT) and homozygous (KI/KI) for PS1 FAD KI mutant M146V or I213T mouse embryonic brains were treated with eB1 or Fc for 60 min, lysed and IPed with anti-EPHB2 antibody, then immunoblotted with anti-GLUN1 and anti-EPHB2 antibodies, respectively. The input of the IP experiment is also shown in the lower panel. (C) Graphs illustrate fold change of GLUN1 IPed with EPHB2 following eB1 stimulation. Experiments were repeated four times. Data are mean ± SEM. *P < 0.05, eB1 versus Fc in WT; ^#P < 0.05 and ^##P < 0.01 FAD Fc control versus WT Fc control, two-tailed paired t-test, n = 4.

Figure 3

PS1 FAD mutants modulate EPHB2 co-clustering with GLUN1 on neuronal surface. (A) Cortical neuronal cultures from WT and PS1 FAD (M146V) mouse embryonic brains either heterozygous (M146V KI/WT) or homozygous (M146V KI/KI) were stimulated with eB1 or Fc for 60 min, followed by live-labelling with anti-EPHB2 and anti-GLUN1 antibodies to detect clusters of the receptors on cell surface and their co-localization. In all cases, green indicates EPHB2 staining, red indicates GLUN1 staining, and blue is DAPI nuclear staining. White arrow indicates both single channel immunostaining and co-localized clusters. Scale bar, 5 µm. (B) Graphs showing percent change of co-localization of EPHB2 and GLUN1 over Fc treated WT group by following Mander’s co-localization coefficients. Data are mean ± SEM. **P < 0.01, eB1 versus Fc in WT; ^###P < 0.001 FAD Fc control versus WT Fc control, Mander’s co-localization coefficients, n = 9–7.

Figure 4

PS1 FAD mutants affect EPHB2 association with GLUN1 in mouse brains. (A) Four-week-old WT, heterozygous (KI/WT) and homozygous (KI/KI) for PS1 FAD mutant M146V or I213T mouse brain hippocampi were homogenized in Hepes buffer containing 1% Triton-X 100. After centrifugation, total lysates were IPed with anti-EPHB2 antibody, then immunoblotted with anti-GLUN1 or anti-EPHB2 antibodies, respectively. The neuronal lysates used in the IP experiment were immunoblotted with antibodies against GLUN1, EPHB2 and ACTIN (input, lower panel). (B) Graphs display fold of GLUN1 immunoprecipitated with EPHB2. Data are mean ± SEM. *P < 0.05 versus WT, one-way ANOVA with Tukey’s post hoc test, n = 3.

Figure 5

Effects of EFNB1 and PS1 FAD mutants on NMDAR interactions with PS1. (A) Cortical neuronal cultures were stimulated with EFNB1 (eB1) or Fc for 60 min. After stimulation, cells were lysed and IPed with anti-PS1 NTF antibody, then immunoblotted with anti-GLUN1 and anti-GLUN2B antibodies. PS1 NTF (PS1) was detected under the same condition with unboiled samples. Graphs show fold change of GLUN1 immunoprecipitated with PS1 following eB1 stimulation (right panel). Experiments were repeated at least three times. Data are mean ± SEM. *P < 0.05 versus Fc, two-tailed paired t-test, n = 3. The inputs are shown in the lower panel. (B) Cortical neuronal cultures from WT, heterozygous (KI/WT) and homozygous (KI/KI) for PS1 FAD KI mutant M146V or I213T mouse embryonic brains were treated with eB1 or Fc for 60 min, lysed and IPed with anti-PS1 NTF antibody, then immunoblotted with anti-GLUN1. PS1 was detected under the same condition with unboiled samples. PIS = pre-immune serum. (C) Three-month-old WT and PS1 FAD (M146V and I213T) KI heterozygous (KI/WT) or homozygous (KI/KI) mouse brain cortices were homogenized in Hepes buffer containing 1% Triton-X 100. After centrifugation, total lysates were IPed with anti-PS1 NTF antibody, then immunoblotted with anti-GLUN1 and anti-GLUN2B antibodies. PS1 was detected under the same condition with unboiled samples.

Figure 6

PS1 FAD mutants affect GLUN1 co-localization with PS1, and modulate PS1-GLUN1 complex stability. (A, B) Co-localization of PS1 and GLUN1 in mouse cortex was detected using double fluorescence staining with antibodies against PS1 (green) and GLUN1 (red), followed by analysis with metamorph. Increased overlay of PS1 and GLUN1 staining (yellow) were detected in PS1 FAD mutant M146V heterozygous (KI/WT) and homozygous (KI/KI) mice. White arrow indicates both single channel immunostaining and co-localization. Scale bar 1 µm. One-way ANOVA followed by Tukey’s multiple comparison test was used for statistical analyses. **P < 0.01, ***P < 0.001 versus WT, n = 7–9. (C, D) Cortical neuronal cultures from WT, and heterozygous for PS1 M146V mutant (M146V/WT) mouse embryonic brains were treated with eB1 for 1 h in neurobasal media followed by withdrawal of the ligand, and replacing the media with fresh neurobasal. Cells were thereafter collected and lysed at post-eB1 withdrawal hours as indicated. Neuronal lysates were used for IP experiments as depicted in the figure. No withdrawal means eB1 ligand remained in the media the entire experimental period. (E) Frontal cortices from normal (control) and FAD human brains carrying PS1 S170F mutant (PS1 FAD) were used to isolate crude synaptosomal membrane fractions. The fractions were IPed with anti-PS1 NTF antibody. WB experiments showed constitutively higher amount of PS1-GLUN1 complexes in FAD brains than control brains.

Figure 7

Role of PS1 on BDNF-stimulated TRKB-GLUN1 association and neuroprotection; and effects of PS1 FAD mutants on GLUN1 interactions with TRKB and PS1. Primary cortical neurons prepared from WT (PS1 +/+), PS1 hemizygous (PS1 +/−) or PS1 homozygous knock out (PS1 −/−) mouse embryonic brains, were stimulated with BDNF (50 ng/ml) or no treatment (NT) for 30 min, lysed and IPed with anti-TRKB antibody. (A) Obtained IPs were then probed on WB with anti-GLUN1 or anti-TRKB antibodies (upper panel). Graphs show quantification of TRKB-GLUN1 complexes IPed as above. Data are mean ± SEM. **P < 0.01, one-way ANOVA with Tukey’s post hoc test, n = 3 (right panel). Input: Neuronal lysates used in this IP experiment were immunoblotted with antibodies against GLUN1, TRKB, PS1-CTF and ACTIN as shown in figure (lower panel). NI IgG = non-immune IgG. (B) Down-regulation of PS1 using PS1 siRNA decreases the BDNF-stimulated complex between TRKB and GLUN1. Scrambled siRNA was used as control. (C) (upper panel) Extracellular peptide, exGLUN1, abolishes BDNF mediated neuroprotection from glutamate excitotoxicity. Primary cortical neuronal cultures were pretreated for one hour as indicated in the figure, and then challenged with glutamate (Glu) at 50 μM for 3 h. Neuronal survival was then determined as described in Materials and methods section. BDNF rescues neurons from excitotoxicity in the absence (3rd bar from left), but not in the presence of peptide ex-GLUN1 (50 nM; 4th bar). Peptide exEPHB2, however, does not abolish BDNF neuroprotection (5th bar). Data are mean ± SEM. **P < 0.01; one-way ANOVA with Tukey’s post hoc test, n = 3–5. Cell survival in the presence of glutamate is set at 100%. (upper panel). ExGLUN1 reduces the BDNF-stimulated association of TRKB with GLUN1. Neuronal cultures were pretreated for 1 h with either exGLUN1 or BSA (50 nM) followed by BDNF stimulation for 30 min. Neurons were lysed afterwards, and extracts were IPed with anti-TRKB followed by WB with anti-GLUN1 or anti-TRKB antibodies. (D) Cortical neuronal cultures from WT, heterozygous (KI/WT) and homozygous (KI/KI) for PS1 FAD (M146V) KI mouse embryonic brains were treated with BDNF for 30 min, lysed and IPed with anti-TRKB antibody, then immunoblotted with anti-GLUN1 and anti-TRKB antibodies, respectively. The input of the IP experiment is also shown in the lower panel. (E) Cortical neuronal cultures from WT, heterozygous (KI/WT) and homozygous (KI/KI) for PS1 FAD (I213T) KI mouse embryonic brains were treated with BDNF for 30 min, lysed and IPed with anti-PS1 NTF (PS1) antibody, then immunoblotted with anti-GluN1. PS1 was detected under the same condition with unboiled samples.

Figure 8

PS1 FAD mutant decreases NMDAR EPSCs in the CA1 region of mouse hippocampal slices and increases neuronal vulnerability after ischaemia in vivo. (A) Representative EPSC traces from WT (black) and heterozygous for PS1 FAD mutant M146V (M146V/WT) (red) CA1 neurons. Traces are averages of 5 NMDAR EPSCs recorded in the presence of 10 µM NBQX and 10 µM GBZ. (B) Input–output plot showing significantly reduced amplitudes of NMDAR EPSCs in PS1 FAD mutant-expressing neurons in comparison with WT neurons at higher stimulation intensities. The values in parentheses indicate the number of neurons/mice used in the analysis. Data are expressed as mean + SEM; P = 0.013, F = 6.36, two-way ANOVA. (C) WT (WT/WT), M146V (WT/M146V) and I213T (WT/I213T) heterozygous mice were subjected to MCAO. Brains were isolated and sectioned 30 days later, followed by counting viable neurons with stereo investigator software. Number of live neurons in the MCAO lesioned cortical area was normalized to number of neurons in the contralateral side of each section. *P < 0.05, two-tailed unpaired t-test versus WT, n = 3 (WT), n = 6 (WT/M146V), n = 6 (WT/I213T).