Synaptogenic effect of APP-Swedish mutation in familial Alzheimer's disease

Abstract

Mutations in β-amyloid (Aβ) precursor protein (APP) cause familial Alzheimer's disease (AD) probably by enhancing Aβ peptides production from APP. An antibody targeting Aβ (aducanumab) was approved as an AD treatment; however, some Aβ antibodies have been reported to accelerate, instead of ameliorating, cognitive decline in individuals with AD. Using conditional APP mutations in human neurons for perfect isogenic controls and translational relevance, we found that the APP-Swedish mutation in familial AD increased synapse numbers and synaptic transmission, whereas the APP deletion decreased synapse numbers and synaptic transmission. Inhibition of BACE1, the protease that initiates Aβ production from APP, lowered synapse numbers, suppressed synaptic transmission in wild-type neurons, and occluded the phenotype of APP-Swedish-mutant neurons. Modest elevations of Aβ, conversely, elevated synapse numbers and synaptic transmission. Thus, the familial AD-linked APP-Swedish mutation under physiologically relevant conditions increased synaptic connectivity in human neurons via a modestly enhanced production of Aβ. These data are consistent with the relative inefficacy of BACE1 and anti-Aβ treatments in AD and the chronic nature of AD pathogenesis, suggesting that AD pathogenesis is not simply caused by overproduction of toxic Aβ but rather by a long-term effect of elevated Aβ concentrations.

Fig. 1.. Generation of conditional Swedish-mutant APP^Swe/+ human neurons with APP expression and Aβ secretion measurements.

(A) Gene-targeting strategy. The mutant allele, inserted by homologous recombination into hES and iPS cells, contains a duplicated exon 16 (E16), with one of the copies including the Swedish mutation (asterisks = Swedish point mutation). The duplicated exons are flanked by FRT and loxP sites, such that Flp and Cre recombination induces retention of either only the wild-type (APP^+/+) or the Swedish-mutant exon (APP^Swe/+), respectively. See fig. S1 for details. (B) Immunoblotting analysis of APP expression on wild-type APP^+/+ and Swedish-mutant APP^Swe/+ human neurons derived from H1.2 stem cells. For representative blots, see fig. S3E. (C) Assessment of secreted Aβ peptides by enzyme-linked immunosorbent assay (ELISA) on supernatants harvested from APP^+/+ and APP^Swe/+ neuron cultures at day 11 (D11), day 17 (D17), or day 23 (D23) after induction of independent stem cell clones (top, H1.2; bottom, 5d1.1). All data are from human neurons analyzed 5 weeks after neuronal induction unless noted. Analyses of additional clones, representative immunoblots, and further images are shown in fig. S3. All numerical data are means ± SEM; numbers of biological replicates are indicated in the graphs. Statistical significance was assessed by two-way ANOVA with post hoc corrections (C) or Student’s t test (B), with *P < 0.05, **P < 0.01, ***P < 0.001, and ****P < 0.0001. Nonsignificant comparisons are not indicated.

Fig. 2.. Characterization of dendrite development, endosome sizes, and synapse densities in human conditional Swedish-mutant APP^Swe/+ neurons compared to their perfect isogenic controls.

(A and B) Dendritic arborization of wild-type APP^+/+ and Swedish-mutant APP^Swe/+ human neurons [(A) representative images of human neurons sparsely transfected with GFP; (B) quantification of selected dendritic parameters of neurons derived from H1.2 stem cells). (C and D) Rab5-positive endosome sizes in wild-type APP^+/+ and Swedish-mutant APP^Swe/+ human neurons [(C) representative images; insets show higher magnification views of endosomes; (D) summary graphs of endosome sizes in neurons obtained from two independent stem cell clones (left, H1.2; right, 5d1.1)]. (E to G) Quantification of synapsin- and PSD95-positive synaptic puncta in isogenic APP^+/+ and APP^Swe/+ human neurons [(E) representative images of wild-type APP^+/+ and Swedish-mutant APP^Swe/+ human neurons immunolabeled for the presynaptic marker synapsin (SYN; green) and the dendritic marker MAP2 (red); left: overviews; right: higher-magnification images (from clone H1.2); (F) summary graphs of the synapsin-puncta density determined in neurons derived from four independent conditionally Swedish-mutant stem cell clones; (G) summary graph of the synapse density measured by PSD95 immunocytochemistry in wild-type and Swedish-mutant human neurons derived from clone H1.2]. For further representative images, see fig. S4. DAPI, 4′,6-diamidino-2-phenylindole. (H) Immunoblotting analysis of the synaptic proteins expressed in human wild-type APP^+/+ and Swedish-mutant APP^Swe/+ neurons (for representative immunoblots, see fig. S4D). All neurons were analyzed 5 weeks after neuronal induction; additional data are shown in figs. S3 and S4. All numerical data are means ± SEM; numbers of images/experiments (B, D, F, and G) or of experiments (H) are indicated in the bars. Statistical significance was assessed by two-way ANOVA with post hoc corrections (H) or Student’s t test (all other bar graphs), with*P < 0.05, **P < 0.01, and ***P < 0.001. Nonsignificant comparisons are not indicated. GLUA1, glutamate receptor 1; SYB2, synaptobrevin 2; STX1, syntaxin 1; GLUA2, glutamate receptor 2; SNAP25, synaptosomal-associated protein 25.

Fig. 3.. Synaptic transmission in human wild-type APP^+/+ and Swedish-mutant APP^Swe/+ neurons.

(A to F) Assessment of the frequency and amplitude of spontaneous miniature excitatory postsynaptic currents (mEPSCs), monitored in the presence of tetrodotoxin in APP^+/+ and APP^Swe/+ human neurons derived from H1.2 (A to C) and 5d1.1 clones (D to F) [(A and D) representative traces; (B and E) cumulative probability plots of the interevent times and summary graph of the mEPSC frequency (inset); (C and F) cumulative probability plots and summary graphs (inset) of mEPSC amplitudes]. (G to J) Assessment of the amplitude and coefficient of variation of the evoked EPSCs measured on isogenic APP^+/+ and APP^Swe/+ human neurons derived from H1.2 (G and I) and 5d1.1 clones (H and J) [(G and H) representative traces; (I and J) summary graphs of EPSC amplitudes (left) and coefficients of variation (c.v.) (right)]. All neurons were analyzed 5 weeks after neuronal induction; additional data are shown in fig. S4. All numerical data are means ± SEM; numbers of cells/experiments analyzed are indicated in bars. Statistical significance was assessed by Kolmogorov-Smirnov (KS) test (all cumulative probability plots) or Student’s t test (all bar graphs), with*P < 0.05 and ****P < 0.0001. Nonsignificant comparisons are not indicated.

Fig. 4.. The effect of pharmacological BACE1 inhibition on synapse formation and synaptic transmission in human wild-type APP^+/+ and Swedish-mutant APP^Swe/+ neurons.

(A to C) Quantification of synapse density in WT APP^+/+ and Swedish-mutant APP^Swe/+ neurons with or without BACE1 inhibition measured by immunocytochemistry for synapsin in human neurons derived from two independent stem cell clones with conditional Swedish mutations [(A) representative images of APP^+/+ and APP^Swe/+ human neurons with or without BACE1 inhibitor (BACEi) treatment; neurons were stained for synapsin (SYN; green) and MAP2 (red; bottom: high magnification of dendrites); (B and C) summary graphs of the density (left) and size (right) of synapsin-positive synaptic puncta in neurons derived from H1.2 (B) or 5d1.1 cells (C)]. Neurons were treated with the BACE1 inhibitor LY2886721 at 2 μM starting 1 week after neuronal induction and analyzed 4 weeks later. (D to I), The frequency and amplitude of mEPSCs in isogenic APP^+/+ and APP^Swe/+ neurons with or without BACE1 inhibition monitored by whole-cell patch-clamp recordings in human neurons derived from two independent stem cell clones [H1.2 (D to F) and 5d1.1 (G to I)] with conditional APP-Swedish mutations [(D and G) representative mEPSC traces; (E and H) cumulative probability plots of the mEPSC interevent interval and summary graph of the mEPSC frequency (inset); (F and I) cumulative probability plot and summary graph (inset) of the mEPSC amplitude]. All numerical data are means ± SEM (cells/experiments are indicated in all bars); statistical significance was assessed by two-way ANOVA with post hoc corrections (all bar graphs) or KS test (all cumulative probability plots), with*P < 0.05, **P < 0.01, ***P < 0.001, and ****P < 0.0001. Nonsignificant comparisons are not indicated.

Fig. 5.. Generation and characterization of isogenic human APP^+/+ and APP^−/− neurons derived from the same conditional APP knockout stem cell subclones.

(A) Design of the conditional APP knockout allele. (B) Immunoblotting analysis of APP protein to validate the removal of APP alleles in human neurons with APP conditionally deleted. The remaining APP protein is derived, at least in part, from the cocultured mouse glia that also express APP, albeit at lower amounts. (C) Validation of the APP deletion in human APP^−/− neurons measured by Aβ peptides ELISA on supernatants harvested at 17 days after neuronal induction with Ngn2. (D and E) Dendritic arborization of human APP^+/+ and APP^−/− neurons derived from the same APP conditional knockout stem cell clones. [(D) representative images of sparsely transfected human APP^+/+ and APP^−/− neurons expressing tdTomato; (E) quantification of the total dendrite length and number of branches of APP^+/+ and APP^−/− neurons from two independent mutant hES cells clones (cKO2 and cKO3)]. (F and G) Assessment of the Rab5-positive endosomes in APP^−/− neurons and their isogenic APP^+/+ controls. [(F) representative images of APP^+/+ and APP^−/− neurons immunolabeled for Rab5 (green) and MAP2 (red); (G) quantification of the size of Rab5-positive endosomes (insets, expanded views of endosomes) in human APP^+/+ and APP^−/− neurons generated from two independent mutant hES cells clones (cKO2 and cKO3)]. All summary graphs show means ± SEM [number of experiments (B and C) or cells/experiments (E and G) are indicated in bars]; statistical significance was assessed by Student’s t test, with *P < 0.05, **P < 0.01, and ***P < 0.001. Nonsignificant comparisons are not indicated. Additional data are shown in fig. S7.

Fig. 6.. Synapse density and synaptic transmission in human APP^+/+ and APP^−/− neurons.

(A to C) Quantification of synapsin- and PSD95-positive synaptic puncta in human APP^+/+ and APP^−/− neurons derived from the same APP conditional knockout stem cell clones [(A) representative images of human APP^+/+ and APP^−/− neurons immunolabeled for synapsin (green) and MAP2 (red); (B) quantification of the density of synapsin-positive puncta in APP^+/+ and APP^−/− neurons derived from two independent mutant hES cells clones (cKO2 and cKO3); (C) quantification of the density of PSD95-positive puncta in APP^+/+ and APP^−/− neurons generated from cKO2 hES cells]. For representative images, see fig. S7 (G to I). (D to I) Assessment of the frequency and amplitude of mEPSCs in APP^+/+ and APP^−/− neurons derived from cKO2 (D to F) and cKO3 clones (G to I) [(D and G) representative traces; (E and H) cumulative probability plots of the interevent times, and summary graph of the mEPSC frequency (inset); (F and I) cumulative probability plots and summary graphs (inset) of mEPSC amplitudes]. (J to M) Assessment of the amplitude and coefficient of variation of the evoked EPSCs measured on isogenic APP^+/+ and APP^−/− human neurons derived from cKO2 (J and K) and cKO3 clones (L and M) [(J and L) representative traces; (K and M) summary graphs of EPSC amplitudes (left) and coefficients of variation (right)]. All data were obtained from human neurons at 5 weeks after neuronal induction with Ngn2; additional data are shown in fig. S7. All summary graphs show means ± SEM (cells per experiments are indicated in bars); statistical significance was assessed by Student’s t test (bar graphs) or KS test (cumulative probability plots), with *P < 0.05, **P < 0.01, ***P < 0.001, and ****P < 0.0001. Nonsignificant comparisons are not indicated.

Fig. 7.. The effect of modest elevations of Aβ on synapse formation in human neurons.

(A) Experimental strategy for (B) to (F). The supernatant of HEK293T cells expressing EGFP (control), full-length APP (flAPP), or the C99 and sAPPβ fragments of APP was added to human APP^+/+ and APP^−/− neurons 7 days after neuronal Ngn2 induction; neurons were analyzed 4 weeks afterwards. (B) Total Aβ peptides measured by ELISA in supernatants of HEK293T cells expressing EGFP, full-length APP, C99, and sAPPβ. (C) Quantification of synapse density measured by synapsin immunocytochemistry on human APP^+/+ and APP^−/− neurons treated with indicated supernatants from (A). For representative images, see fig. S8F. (D to F) Assessment of the mEPSC frequency in human APP^+/+ and APP^−/− neurons treated with indicated supernatants from (A). [(D) Representative mEPSC traces; (E and F) cumulative probability plots of mEPSC interevent intervals (insets, summary graphs of the mEPSC frequency) from APP^+/+ (E) and APP^−/− (F) neurons]. (G) Experimental strategy for (H) to (K). Human APP^−/− neurons were infected with lentiviruses encoding mOrange (control), full-length APP (flAPP), or the C99 and sAPPβ fragments of APP 4 days after neuronal Ngn2 induction; neurons were analyzed 5 weeks after neuronal induction. Uninfected APP^+/+ and APP^−/− neurons were used as further controls in (H) to (K). (H) Total Aβ peptides in supernatants of human APP^−/− neurons expressing EGFP, full-length APP, C99, and sAPPβ. Values were measured by ELISA at 17 days after neuronal Ngn2 induction. (I) Quantification of synapse density by synapsin immunocytochemistry on human APP^−/− neurons expressing EGFP, full-length APP, C99, and sAPPβ. For representative images, see fig. S9F. (J and K) Assessment of the mEPSC frequency on human APP^−/− neurons expressing EGFP, full-length APP, C99, and sAPPβ. [(J) Representative mEPSC traces; (K) cumulative probability plot of mEPSC interevent intervals and summary graph of the mEPSC frequency(inset)]. Additional data are shown in figs. S8 and S9. All summary graphs show means ± SEM [number of experiments (B and H) or of images or cells/experiments (all other graphs) are indicated in bars]; statistical significance was assessed by one-way ANOVA (B, H, I, and K) or two-way ANOVA (C, E, and F) with post hoc corrections, or by KS test (cumulative probability plots), with *P < 0.05, **P < 0.01, and ***P < 0.001. Nonsignificant comparisons are not indicated.