The death of synapses, and the neurons that need them, marks the inevitable endgame of Alzheimer’s disease. Despite this eventual synaptic mayhem, a new work, led by Thomas Südhof and Marius Wernig at Stanford University, supports the idea that in its monomeric form, Aβ, the peptide behind amyloid plaques, spurs the growth of synapses. Writing in the October 19 Science Translational Medicine, the authors report that in cultured human neurons harboring the Aβ-boosting, Swedish mutation in amyloid precursor protein (APP), synapse numbers climbed by 20 percent, and synaptic transmission rose accordingly. In agreement with previous studies, neurons harboring this familial AD mutation also had swollen endosomes. Lowering Aβ with a BACE inhibitor, or by nixing APP expression, produced the opposite effect—lowering synapse numbers and shrinking endosomes. The findings support a physiological role for Aβ at the synapse, and the authors suggest this explains the cognitive impairment observed in BACE inhibitor trials.

  • In human stem-cell-derived neurons, the Swedish mutation boosted synapses.
  • BACE inhibition, or APP deletion, reduced synapse numbers.
  • Aβ monomer peptides fueled the synaptic boost.

“The results should not confuse or distract from the accelerating approach to neutralize and remove synaptotoxic Aβ aggregates from the brains of AD patients, which has just received a strong boost through the positive top-line results of the Phase 3 clinical trial of lecanemab,” commented Dennis Selkoe of Brigham and Women’s Hospital in Boston. 

While the literature on the synaptic effects of various forms of Aβ is complex, evidence indicates that at low, physiological concentrations, Aβ monomers can promote synaptic signaling, while at high concentrations, or in aggregated forms, the peptide is neurotoxic and extinguishes long-term potentiation, a critical component of functional neuronal circuitry (Puzzo et al., 2008; Nov 2009 news; Apr 2002 news). Previously, scientists found that in transgenic mice expressing the APP Swedish mutation synapses withered. Because the Swedish mutation accelerates Aβ production, this suggested that too much of the peptide is bad, and yet in hippocampal slice cultures and in human stem cell-derived neurons, other Aβ-boosting mutations stoked synaptic transmission (Fitzjohn et al., 2001; Feb 1999 news; Ghatak et al., 2019).  

To study the effects of Aβ at the synapse in human neurons, first author Bo Zhou and colleagues used a novel approach to generate stringently isogenic lines of human neurons expressing either wild-type APP or APP bearing the APP-Swe mutation. Previous studies have derived stem cells from carriers of this mutation, and then corrected the mutation using CRISPR to create isogenic controls. However, the authors noted that this approach requires selection and propagation of mutation-corrected clones, a process which promotes genetic changes. To avoid this, the researchers started with a single line of embryonic stem (ES) cells and, in a parallel approach, a line of induced pluripotent stem (tips) cells. Using homologous recombination, they inserted into the endogenous APP gene back-to-back copies of exon 16—one encoding the wild-type APP sequence and the other the Swedish mutation. These different versions of APP could be switched on using different recombinases, thus allowing the researchers to generate lines that expressed either normal or mutant forms of APP. As the approach only affected one copy of APP, the other, normal copy of APP was left intact, yielding ES and iPS lines that were either homozygous for normal APP, or heterozygous for APP-Swe. In neurons differentiated from these lines, the researchers detected more Aβ40 and Aβ42 in cells expressing the Swedish-mutant APP than in those expressing only normal APP. In keeping with previous work, endosomes swelled by 15 to 20 percent in the mutant APP neurons as well (Aug 2019 news).

Next came a finding the scientists had not expected. Relative to neurons expressing normal APP, those expressing a copy of the Swedish mutant boasted 20 percent more synapses, and they expressed substantially more synaptic proteins. Electrophysiological recordings suggested that the synaptic windfall stoked a proportional boost in synaptic transmission. The findings suggested that the Swedish mutation bolstered the growth and function of synapses. 

Aβ: Fuel for Synapses? Compared to neurons expressing only normal APP (left), those expressing a copy of APP with the Swedish mutation (right) boasted 20 percent more synapses (green). [Courtesy of Zhou et al., Science, 2022.]

The Swedish mutation promotes Aβ production by facilitating the cleavage of APP by BACE1. Therefore, the researchers reasoned that blocking BACE1 should reverse the synaptic boost. Indeed, treatment with the BACE inhibitor LY2886721 completely prevented the synaptic enhancement. It also decreased synaptic numbers and transmission in wild-type neurons. In agreement with these findings, the researchers detected a similar decrease in synapse numbers and transmission in neurons when they deleted APP. “Thus, the deletion and Swedish mutation of APP induce mirror-image phenotypes: The APP deletion suppresses synapse numbers and synaptic function, but the APP-Swedish mutation enhances them,” the authors wrote.

What is responsible for the synaptogenic effect? Besides producing a cadre of Aβ peptides, the cleavage of APP by β- and γ-secretases yields several other proteolytic fragments. Through a series of experiments where they expressed different APP fragments in neurons and/or added them exogenously, the researchers zeroed in on the synaptic booster—none other than Aβ. They did not distinguish between Aβ40 and Aβ42 in this regard, but their findings suggest that Aβ monomers—as opposed to oligomers or fibrils—promoted synapses.

“This is excellent work, and the experiments are very well controlled,” wrote Bart De Strooper of Dementia Research Institute, UK. “It shows in a convincing way that the APP-Swe mutation promotes synaptogenesis and that the increase in Aβ peptide is responsible for this effect.” 

De Strooper also agreed with the authors that the observations help to interpret the BACE1 inhibitor side effects. However, he noted that, as familial AD mutations go, APP-Swe is unique. While other ADAD mutations in APP and the presenilins skew Aβ production to larger, more aggregation-prone forms, APP-Swe is the lone familial AD mutation that boosts overall production of Aβ. Therefore, De Strooper believes that the synaptogenic effect of the mutation is unlikely a disease-promoting effect. Rather, the findings suggest that loss of Aβ, due to its aggregation, could be pathologically relevant.

“The big question that remains is, what is the best therapeutic target?” asked De Strooper. “The relative loss of Aβ peptide might reduce synaptic function but the gain of toxic effects, due to amyloid peptide aggregates and amyloid plaques causing glial pathology, might also drive the neurons also into dysfunction and cell death,” he added.

“This is a convincing and exciting study providing intriguing new evidence for Aβ-dependent effects on synapses and endosomes,” commented Gunnar Gouras of Lund University in Sweden. “Nevertheless, how does this enhancement in synapse number and endosome size from the APP Swedish mutation relate to what occurs in AD where synapses are lost?” Gouras asked. He noted that human stem-cell-derived neurons cannot account for the effects of aging that occur in the human brain, making it difficult to know if Aβ might influence synapses differently with age (see full comment below).

How to explain the findings in light of other studies that have deemed Aβ toxic to synapses? For one, the authors noted that many animal and cell culture studies were performed with high concentrations of Aβ. Furthermore, they acknowledged decades of work indicating oligomeric forms of Aβ are neurotoxic, while monomeric forms are beneficial. They agreed that oligomeric Aβ may only cause harm after it accumulates over decades.

Of course, the latter statement is a key tenant of the amyloid cascade hypothesis formulated by Selkoe and others. Selkoe noted that Zhou’s findings are quite consistent with previous studies from his lab and others, in which soluble Aβ dimers/oligomers isolated from the AD brain unleashed a bevy of pathogenic events in cultured neurons, whereas Aβ monomers isolated in the same experiments had no deleterious effects (Jin et al., 2011). “The new data from Zhou et al. rigorously extend this concept in a physiologically relevant system,” Selkoe wrote.

In an email to Alzforum, Südhof was reticent to speculate about what the findings imply about AD pathogenesis. “Ours is a basic cell biology study that unbiasedly examines the effect of an APP mutation and Aβ on human synapses at physiological levels, but does not intend to construe any disease mechanisms,” he wrote. “Of course, observing such dramatic effects of a familial Alzheimer's disease mutation on synapses is relevant for Alzheimer's disease, but understanding what the results mean for the disease will require additional work.”—Jessica Shugart

Comments

  1. Zhou et al. provide an elegantly designed and executed set of experiments supporting a role for positive synaptic effects of Aβ released from the Swedish mutant form of APP, which causes an aggressive, early onset form of familial AD. The authors use a clever homologous recombination approach to express Swedish-mutant, or wild-type APP, conditionally in human ES- and iPS-derived neuronal lines, thereby obtaining the full isogenically controlled effect of what they describe as a modest elevation of wild-type Aβ40 and 42 (plus other lengths not examined). This recombinant approach in isolated, defined neurons is interfaced with the pharmacological effects of a clinical BACE1 inhibitor. Collectively, the results document positive effects of the mildly elevated wt Aβ monomers from Swedish APP on synapse formation and synaptic transmission in these glutamatergic neurons. The data are detailed and clear.

    The main point I will comment on is the sense of paradox that the results and the title seem to convey. How does this evidence of positive effects on synapse number and function relate to the extensively documented negative role of Aβ accumulation in AD? The authors address this key question in the second paragraph of the discussion: “[I]t is possible that oligomeric Aβ is neurotoxic but not normally produced by neurons … whereas monomeric Aβ may be synaptogenic.” They go on to point out correctly that “this would agree with the delayed development of AD in patients, suggesting that secreted endogenous Aβ may only become oligomeric as it accumulates during a person’s lifetime.” The latter statement, of course, is a key tenant of the “amyloid (Aβ) hypothesis” of AD pathogenesis. Therefore, the sense of paradox that the authors convey for their “unexpected” results could have been clarified by highlighting their key discussion point already in the title, abstract, or introduction. 

    Early in the study of synthetic Aβ peptides in culture—more than three decades ago—a few papers described positive synaptic effects of monomeric, synthetic Aβ. Later, we observed that soluble Aβ dimers/oligomers isolated by size-exclusion chromatography from AD brain induced neuritic dystrophy and AD-type tau hyperphosphorylation in cultured wild-type neurons, whereas Aβ monomers isolated in the same experiments had no deleterious effects (Jin et al., 2011) rigorously extend this concept in a physiologically relevant system. But the results should not confuse or distract from the accelerating approach to neutralize and remove synaptotoxic Aβ aggregates from the brains of AD patients, which has just received a strong boost through the positive top-line results of the Phase 3 clinical trial of lecanemab. 

    References:

    . Soluble amyloid beta-protein dimers isolated from Alzheimer cortex directly induce Tau hyperphosphorylation and neuritic degeneration. Proc Natl Acad Sci U S A. 2011 Apr 5;108(14):5819-24. PubMed.

  2. This is excellent work, and the experiments are very well controlled. The work shows in a convincing way that the APP/SW mutation promotes synaptogenesis and that the increase in Aβ peptide (not the other APP cleavage products) are responsible for this effect. I agree with the authors that the observations help to interpret BACE1 inhibitor side effects. However, the implications for the understanding of pathogenesis of AD are less clear.

    APPSW is a rather exceptional mutation as it is the only one which increases Aβ peptide production. All other APP mutations (and presenilin mutations) alter Aβ production to generate longer, more aggregation-prone species, but do not increase the total Aβ. Thus, the increased synaptogenesis seen with APP/SW is likely not the disease-causing effect of this mutation, at least I wonder whether this is also seen with other mutations. 

    More likely is that the loss of Aβ (as seen in APP knockouts, but which would also theoretically be the consequence of the Aβ aggregation) is pathologically relevant. This correlates with the clinical observation that one of the first biomarkers of incipient AD is lowering of Aβ peptide in the CSF.

    Still, the true situation is likely more complicated. The common denominator of APP and presenilin mutations is longer Aβ peptides and their aggregation, thus not only is depletion of physiological Aβ a problem, but also the generation of soluble neurotoxic aggregates. Moreover, increased aggregation of Aβ peptides also leads to increased amyloid plaque formation, which induces strong microglial and astroglial responses that likely play a role in the pathogenesis, given the GWAS data indicating genetic risk associated with these cells. 

    I disagree with the interpretation that we do not know yet how presenilin mutations cause AD. They have very similar effects as the APP mutations in the sense that they cause a destabilization of the APP-γ-secretase enzyme substrate complex, which causes the release of longer Aβ peptides (Szaruga et al., 2017). Full loss-of-function mutations do not cause such a shift in Aβ peptides, and do not cause Alzheimer’s disease. In that sense, the APP mutations at the γ-secretase site and the presenilin mutations in the γ-secretase complex act in very similar ways.

    The big question that remains is, what is the best therapeutic target? the relative loss of Aβ peptide might reduce synaptic function, but the gain of toxic effects due to amyloid peptide aggregates and amyloid plaques causing glial pathology might drive the neurons also into dysfunction and cell death.

    References:

    . Alzheimer's-Causing Mutations Shift Aβ Length by Destabilizing γ-Secretase-Aβn Interactions. Cell. 2017 Jul 27;170(3):443-456.e14. PubMed. Correction.

  3. This is a convincing and exciting study providing intriguing new evidence for Aβ-dependent effects on synapses and endosomes; importantly, these cellular studies conducted on human stem-cell-derived neurons are done at more physiological concentrations of Aβ than many prior studies. It is remarkable that analogous to the AD-risk gene APOE4, a familial AD mutation (the Swedish APP mutation) also paradoxically enhances synapse number. The methods used to confirm that specifically Aβ is involved are excellent.

    Nevertheless, how does this enhancement in synapse number and endosome size from the APP Swedish mutation relate to what occurs in AD, where synapses are lost? I completely agree that we need a more thorough understanding of the normal biology of Aβ, APP, ApoE, etc., and since seeing Aβ accumulate specifically in endosomes near synapses two decades ago (Takahashi et al., 2002), my lab has focused on synapses and endosomes in AD. How might some of our work, much on mouse primary neurons, provide input to the current study? The arrival of human neuron models has been a highly important development; however, maturity and age-related abnormalities remain an issue with these powerful new models, which might explain why Zhou et al. see increases rather than AD-like loss of synapses in their model.

    While fully acknowledging the weakness of mice as AD models, remarkably, we and others have observed in numerous studies over the years that there are “aging-like” changes in mature primary neurons with time in culture that may not yet be achievable with cultured human induced-neuron models. In culture, mature, primary mouse neurons carrying the Swedish mutation only develop AD-like reductions of synaptic receptors, with time; intriguingly, such primary neurons also show a gradual decline in secretion of Aβ with time in culture.

    Zhou et al.’s focus on excitatory neurons also fits with our recent study showing that addition of picomolar Aβ led to activation of excitatory neurons, specifically; interestingly, CAMKII-negative neurons did not show such increased activity when we used calcium imaging as a proxy for activity (Martinsson et al., 2022). It will be interesting to analyze inhibitory neurons, to explore Aβ40 versus 42, and to see if further maturation might reveal more AD-like synapse changes in the Swedish APP mutation stem cell-derived model Zhou et al. used here.

    The interface of endosomes and synapses is increasingly seen as a major area of interest for basic AD research that will likely be important for the development of more effective future therapy. While the authors highlight worsening cognition seen in clinical studies of Aβ antibodies and BACE1 inhibitors, they could also keep in mind that reduced Aβ via the Icelandic APP variant protects against AD (Jonsson et al., 2012) and that increasingly Aβ antibodies are showing benefits in clinical trials.

    References:

    . Intraneuronal Alzheimer abeta42 accumulates in multivesicular bodies and is associated with synaptic pathology. Am J Pathol. 2002 Nov;161(5):1869-79. PubMed.

    . Aβ/Amyloid Precursor Protein-Induced Hyperexcitability and Dysregulation of Homeostatic Synaptic Plasticity in Neuron Models of Alzheimer's Disease. Front Aging Neurosci. 2022;14:946297. Epub 2022 Jul 6 PubMed.

    . A mutation in APP protects against Alzheimer's disease and age-related cognitive decline. Nature. 2012 Aug 2;488(7409):96-9. PubMed.

  4. This is an important paper that unbiasedly adds to the growing literature demonstrating the important neuronal functions of the monomeric Aβ peptide. We agree with the previous comment by Dr. Bart De Strooper that these results likely indicate “… that the loss of Aβ (as seen in APP knockouts, but which would also theoretically be the consequence of the Aβ aggregation) is pathologically relevant. This correlates with the clinical observation that one of the first biomarkers of incipient AD is lowering of Aβ peptide in the CSF ...” and that “… the observations help to interpret the BACE1 inhibitors' side effects ...” in terms of the cognitive worsening this class of drugs induces via lowering Aβ levels in the CSF. These results are also in line with our recent findings where we demonstrated that both in sporadic and familial Alzheimer’s disease (AD), high levels of soluble monomeric Aβ42 in the CSF preserve cognition irrespective of laque load (Sturchio et al, 2021; Sturchio et al, 2022). Furthermore, this is in line with large literature supporting the role of monomeric Aβ in synaptic plasticity and memory, mainly via α7 nicotinic acetylcholine receptor signaling, as compiled in the supplementary table 9 of Sturchio et al., 2022

    The pathophysiological role of Aβ loss of function (LOF) is also in line with the fact that all familial AD mutations lead to lower levels of CSF Aβ42 in patients, and that includes APP duplications in Down’s syndrome and the APP-Swedish mutation studied by the authors (Fagan et al., 2021; Lannfelt et al., 1995). Moreover, some mutations, such as the Osaka mutation and the Arctic mutation, lead to lower CSF Aβ42 levels and AD symptoms in the absence of detectable plaques (Tomiyama and Shimada, 2020; Schöll et al., 2012). In cell culture, the vast majority of presenilin 1 (PSEN1) mutations reduce the production of Aβ peptides (Sun et al., 2017). Additionally, low CSF Aβ42 levels are present in animal models harboring the APP-Swedish mutation and in other animal models harboring combinations of APP and PSEN1 mutations (Maia et al., 2013; Andersson et al., 2021; Maia et al., 2015). Finally, Aβ42 LOF is also in line with the recent lecanemab report of cognitive enhancement given that this antibody treatment was shown to be associated with a significant increase in CSF Aβ42 in the Phase 2B trial, which contrasts with the worsening of cognition associated with lowering CSF Aβ42 in the BACE-1 inhibitor trials, as mentioned earlier (Swanson et al., 2021; Egan et al., 2019). 

    Taken together, there is substantial evidence for the physiological importance of Aβ42 and the role of its LOF in AD pathogenesis, which can also explain the lack of correlation between the plaque load and disease phenotype and the high correlation between low CSF Aβ42 levels and dementia. It fits with the clinical and genetic data, where all familial AD mutations (including APP-Swedish and APP duplications) lead to the same clinical outcome: low CSF Aβ42. Such depletion can take place either via enzymatic dysfunction, as in the case of PSEN1 mutations, or monomer consumption due to enhanced aggregation as a result of mutations leading to structural instability or overexpression.

    Additionally, the clinical nature of Aβ42 depletion makes it more pathophysiologically and mechanistically relevant compared to other postulated mediators of toxicity, such as oligomers/protofibrils, which are of undefined nature, heavily manipulated during preparation, and tested in cell culture or in animals at supraphysiologic concentrations. Finally, acknowledging the importance of LOF in AD pathogenesis opens the door for testing a new class of therapeutics aiming to stabilize or supplement the Aβ42 monomer levels.

    References:

    . CSF Aβ42 and Aβ40 and their relation to brain soluble and insoluble Aβ in the 5xFAD mouse model of Alzheimer’s disease. Alzheimer’s & Dementia : The Journal of the Alzheimer’s Association, 17, e055684. Alzheimer's & Dementia

    . Randomized Trial of Verubecestat for Prodromal Alzheimer's Disease. N Engl J Med. 2019 Apr 11;380(15):1408-1420. PubMed.

    . Comparison of CSF biomarkers in Down syndrome and autosomal dominant Alzheimer's disease: a cross-sectional study. Lancet Neurol. 2021 Aug;20(8):615-626. PubMed.

    . Amyloid beta-peptide in cerebrospinal fluid in individuals with the Swedish Alzheimer amyloid precursor protein mutation. Neurosci Lett. 1995 Oct 27;199(3):203-6. PubMed.

    . Changes in amyloid-β and Tau in the cerebrospinal fluid of transgenic mice overexpressing amyloid precursor protein. Sci Transl Med. 2013 Jul 17;5(194):194re2. PubMed.

    . Increased CSF Aβ during the very early phase of cerebral Aβ deposition in mouse models. EMBO Mol Med. 2015 May 15;7(7):895-903. PubMed.

    . Low PiB PET retention in presence of pathologic CSF biomarkers in Arctic APP mutation carriers. Neurology. 2012 Jul 17;79(3):229-36. PubMed.

    . High Soluble Amyloid-β42 Predicts Normal Cognition in Amyloid-Positive Individuals with Alzheimer's Disease-Causing Mutations. J Alzheimers Dis. 2022;90(1):333-348. PubMed.

    . High cerebrospinal amyloid-β 42 is associated with normal cognition in individuals with brain amyloidosis. EClinicalMedicine. 2021 Aug;38:100988. Epub 2021 Jun 28 PubMed.

    . Analysis of 138 pathogenic mutations in presenilin-1 on the in vitro production of Aβ42 and Aβ40 peptides by γ-secretase. Proc Natl Acad Sci U S A. 2017 Jan 24;114(4):E476-E485. Epub 2016 Dec 5 PubMed.

    . A randomized, double-blind, phase 2b proof-of-concept clinical trial in early Alzheimer's disease with lecanemab, an anti-Aβ protofibril antibody. Alzheimers Res Ther. 2021 Apr 17;13(1):80. PubMed. Correction.

    . APP Osaka Mutation in Familial Alzheimer's Disease-Its Discovery, Phenotypes, and Mechanism of Recessive Inheritance. Int J Mol Sci. 2020 Feb 19;21(4) PubMed.

  5. This is great data showing a role on synaptogenesis of Aβ released from the Swedish mutant form of APP. Aβ in APPSwe mice has a propensity to aggregate. I am wondering about other mice such as APP/PS1, APPswe, PSEN1dE9, and 5XFAD in which Aβ aggregates even more. Do the authors assume that the increase in synaptogenesis seen with APP/SW is due to the disease-causing effect of this mutation? It will be interesting to investigate which particular conformational aggregates lead to this phenomenon and also to validate the same hypothesis in other APP and presenilin mutations.

Make a Comment

To make a comment you must login or register.

References

News Citations

  1. What’s My Line?—Fishing for the True Role of Aβ
  2. Earliest Amyloid Aggregates Fingered As Culprits, Disrupt Synapse Function in Rats
  3. LTP Deficits Seen in Hsiao Mice
  4. Familial AD Mutations, β-CTF, Spell Trouble for Endosomes

Mutations Citations

  1. APP KM670/671NL (Swedish)

Paper Citations

  1. . Picomolar amyloid-beta positively modulates synaptic plasticity and memory in hippocampus. J Neurosci. 2008 Dec 31;28(53):14537-45. PubMed.
  2. . Age-related impairment of synaptic transmission but normal long-term potentiation in transgenic mice that overexpress the human APP695SWE mutant form of amyloid precursor protein. J Neurosci. 2001 Jul 1;21(13):4691-8. PubMed.
  3. . Mechanisms of hyperexcitability in Alzheimer's disease hiPSC-derived neurons and cerebral organoids vs isogenic controls. Elife. 2019 Nov 29;8 PubMed.
  4. . Soluble amyloid beta-protein dimers isolated from Alzheimer cortex directly induce Tau hyperphosphorylation and neuritic degeneration. Proc Natl Acad Sci U S A. 2011 Apr 5;108(14):5819-24. PubMed.

Further Reading

Primary Papers

  1. . Synaptogenic effect of APP-Swedish mutation in familial Alzheimer's disease. Sci Transl Med. 2022 Oct 19;14(667):eabn9380. PubMed.