Extracellular Forms of Aβ and Tau from iPSC Models of Alzheimer's Disease Disrupt Synaptic Plasticity

Abstract

The early stages of Alzheimer's disease are associated with synaptic dysfunction prior to overt loss of neurons. To identify extracellular molecules that impair synaptic plasticity in the brain, we studied the secretomes of human iPSC-derived neuronal models of Alzheimer's disease. When introduced into the rat brain, secretomes from human neurons with either a presenilin-1 mutation, amyloid precursor protein duplication, or trisomy of chromosome 21 all strongly inhibit hippocampal long-term potentiation. Synaptic dysfunction caused by presenilin-1 mutant and amyloid precusor protein duplication secretomes is mediated by Aβ peptides, whereas trisomy of chromosome 21 (trisomy 21) neuronal secretomes induce dysfunction through extracellular tau. In all cases, synaptotoxicity is relieved by antibody blockade of cellular prion protein. These data indicate that human models of Alzheimer's disease generate distinct proteins that converge at the level of cellular prion protein to induce synaptic dysfunction in vivo.

Keywords: Alzheimer’s disease; Down syndrome; amyloid β-protein; dementia; extracellular tau; induced pluripotent stem-cell-derived cortical neurons; prion protein; secretome; trisomy 21.

Graphical abstract

Figure 1

Inhibition of LTP In Vivo by Secretomes of Human Stem Cell Models of Alzheimer’s Disease (A) Scheme outlining the generation of cortical cultures and harvesting of secretomes from iPSCs. (B–E) Representative confocal images confirming the differentiation of cortical neurons from NDC (B), PS1 Int4 (C), APP^Dp (D), and Ts21 (E) iPSCs of each genotype by the expression of TBR1 (red), a transcription factor expressed in layer 6 glutamatergic neurons, and neuron-specific MAP2 (green) in dendrites at day 80 post-neural induction. Scale bar, 100 μm. (F) The application of high-frequency conditioning stimulation (HFS, arrow) in the hippocampal CA1 area of the anesthetized rat induced a robust and persistent LTP after an intracerebroventricular injection (# inj) of PBS vehicle or NDC secretome (mean ± SEM % pre-HFS baseline at 3 hr: Veh, 129.7 ± 1.8%; NDC, 123.2 ± 2.3%). (G) Injection of secretomes from APP^Dp (103.3 ± 2.6%), PS1 Int4 (105.1 ± 3.4%), or Ts21 (95.7 ± 3.0%) neurons completely inhibited LTP at 3 hr post-HFS. The y axis is as shown in (F). (H) Values represent the strength of synaptic transmission before (Pre) and 3 hr after the application of HFS for data in (F) and (G). ^∗p < 0.05, one-way ANOVA-Sidak and paired t test. n, number of rats. Calibration bars for EPSP traces: vertical, 2 mV; horizontal, 10 ms. See also Figures S1 and S4.

Figure 2

LTP Inhibition by PS1 Int4 or APP^Dp Secretomes Is Prevented by Immunodepletion with a Pan-Aβ Polyclonal Antibody (A and B) Neuronal secretomes treated with pre-immune serum (Mock) contain Aβ40 (A) and Aβ42 (B) at levels readily detectable with immunoassays, whereas AW7 immunodepleted secretomes do not contain quantifiable levels of Aβ. (C) Immunodepletion of Aβ peptides with AW7 rescued the inhibition of LTP by APP^Dp (APP^Dp/AW7: 123.4 ± 3.5%) and PS1 Int4 (PS1 Int4/AW7: 131.8 ± 2.8%), but not Ts21 (Ts21/AW7: 105.1 ± 3.7%). Calibration bars for EPSP traces: vertical, 2 mV; horizontal, 10 ms. (D) Values before (Pre) and 3 hr post-HFS. The y axis is as shown in (C). ^∗p < 0.05, one-way ANOVA-Sidak and paired t test. See also Figures S2 and S4.

Figure 3

Extracellular Tau Mediates the Blockade of LTP by the Ts21 Secretome (A) Immunodepletion of Ts21 secretome with Tau5 reduced the levels of mid-region containing tau to less than 20% of the original concentration. The monoclonal antibody 46-4 was used as an isotype control. (B and C) Tau immunodepletion with Tau5 prevented the inhibition of LTP by the Ts21 secretome that had previously been immunodepleted with AW7, while mock ID with 46-4 did not (B). Data at 3 hr (Ts21/AW7/Tau5: 134.3 ± 7.0%; Ts21/AW7/46-4: 105.2 ± 4.9%) are summarized in (C). ^∗p < 0.05, two-way ANOVA RM-Sidak and paired t test. (D and E) Ts21 secretome that had previously been immunodepleted with AW7 inhibited LTP (Ts21/AW7: 106.7 ± 4.9%). Co-injection of Tau5 prevented inhibition of LTP (Ts21/AW7+Tau5: 120.8 ± 2.6%), while an isotype control antibody (6E10, 2.5 μg, i.c.v. injection) did not (Ts21/AW7+6E10: 102.9 ± 1.7%), as summarized in (D) and (E). The y axis is as shown in (D). ^∗p < 0.05, one-way ANOVA-Sidak and paired t test. (F and G) Stable LTP was induced by HFS 30 min after the i.c.v. injection of NDC secretome. Tau5 (2.5 μg, i.c.v. injection) was co-administered with APP^Dp secretome but did not affect the inhibition of LTP (APP^Dp+Tau5: 107.4 ± 5.4%). LTP was inhibited to a similar extent in animals treated with PS1 Int4 secretome (99.2 ± 2.5%) or co-treated with Tau5 (PS1 Int4+Tau5: 102.4 ± 6.2%), as summarized in (F) and (G). The y axis is as shown in (F). ^∗p < 0.05, one-way ANOVA-Sidak and paired t test. Calibration bars for EPSP traces: vertical, 2 mV; horizontal, 10 ms. See also Figures S3 and S4.

Figure 4

Synaptic Dysfunction Induced by PS1 Int4, APP^Dp, and Ts21 Secretomes Can Be Relieved by Targeting Cellular PrP (A and B) The inhibition of LTP mediated by the PS1 Int4 secretome (104.6 ± 3.3%) was restored to the level of vehicle controls (132.4 ± 7.2%) by i.c.v. pre-injection (triangle) of 6D11 (6D11+PS1: 129.6 ± 5.3%) (A). Data at 3 hr post-HFS are summarized in (B). The y axis is as shown in (A). ^∗p < 0.05, one-way ANOVA-Sidak and paired t test. (C and D) Pre-injection of 6D11 also prevented the inhibition of LTP by the APP^Dp secretome (C). Data at 3 hr (Veh+APP^Dp: 105.7 ± 1.1%; 6D11+APP^Dp: 128.0 ± 3.8%) are summarized in (D). The y axis is as shown in (C). ^∗p < 0.05, two-way ANOVA RM-Sidak and paired t test. (E and F) The effect of Ts21 secretome (106.7 ± 3.1%) was blocked by 6D11 pre-injection (6D11+Ts21: 127.3 ± 2.9%), but not by an isotype control antibody (IgG2a+Ts21: 105.8 ± 2.1%), as summarized in (E) and (F). The y axis is as shown in (E). ^∗p < 0.05, one-way ANOVA-Sidak and paired t test. (G and H) A weak conditioning stimulation protocol (wHFS, arrow) induced a decremental LTP (Veh: 99.8 ± 7.2%), and 6D11 alone (96.4 ± 3.2%) did not facilitate this decremental LTP, as summarized in (G) and (H). The y axis is as shown in (G). ^∗p < 0.05, two-way ANOVA RM-Sidak and paired t test. Calibration bars for EPSP traces: vertical, 2 mV; horizontal, 10 ms.