. The in-tissue molecular architecture of β-amyloid pathology in the mammalian brain. Nat Commun. 2023 May 17;14(1):2833. PubMed.


Please login to recommend the paper.


  1. This is a beautiful ultrastructural study from the Frank lab. As early as the 1960s, electron microscopy of Alzheimer’s disease brains (and later brains of mouse models) revealed dystrophic neurites, multilaminar vesicles, autophagic vacuoles, and other abnormal cellular structures intertwined with extracellular amyloid fibrils (e.g., Kidd, 1964; Yamaguchi et al., 1989; Masliah et al., 1996). However, these studies relied on heavy fixation and two-dimensional analysis. Now, Leistner et al. use cryo-electron tomography correlated with fluorescence microscopy to generate elegant three-dimensional maps of amyloid plaques in fresh Arctic mutant APP knock-in mouse brain tissue without fixation. The rich array of structures present in close apposition to Aβ fibrils (e.g. Figs 2 and 3) is impressive. It reminds us that the amyloid plaque is not simply an inert ball of fibrils—it is a dynamic structure with many pathologic cellular events happening within and nearby.

    The authors make special efforts to show that the amyloid fibrils are extracellular, not within the many subcellular organelles in the intimately associated cytoplasm, clarifying a recurring question about whether intracellular loci are common sites of Aβ accrual. Further, they show that isolated ex vivo preparations of fibrils (e.g., in sarkosyl extracts) are not identical to those they painstakingly visualize in fresh, intact brain.

    We disagree, at this juncture, with the conclusions reached from Figure 6, regarding the presence of branched, thin protofilaments. We reserve the term “protofilament” to have a strict definition in the context of AD: a single stack of misfolded Aβ in a β-pleated sheet conformation. In the case of Aβ, two protofilaments form one filament (or fibril). The 3-5 nm branched structures that Leistner et al. observe could be composed of Aβ, but they might instead be composed of any elongated macromolecules interspersed with bona fide Aβ fibrils: other proteins, carbohydrates, nucleic acids, or some complex of these.

    For instance, amyloid plaques have been shown to stain intensely for RNA (Ginsberg et al.,1997). Without further evidence, such as immunolabeling or higher-resolution structure determination, we cannot yet conclude that these thinner or branched structures, which are a minority among the abundant amyloid fibrils, are composed of Aβ. In aqueous extracts prepared by simple diffusion (without homogenization or sarkosyl) from typical AD brains, we have not observed any elongated 3-5 nm structures labeling with Aβ antibodies, but it is certainly possible these were lost during preparation (Stern et al., 2023). We look forward to further advances in cryoET and related techniques giving us clearer pictures of diverse Aβ fibrils and associated structures.

    The complexity and heterogeneity of the amyloid fibrils that accumulate in AD brain, and the many altered intracellular organelles and extracellular membranous structures documented by Leistner et al. in their cryoET reconstructions, underscore the malignancy of Aβ deposition and emphasize the importance of preventing its deposition in the first place. Prevention will likely be far more successful than attempting to arrest and repair this multicellular debacle in the brain.



    . Electron micrograph of diffuse plaques. Initial stage of senile plaque formation in the Alzheimer brain. Am J Pathol. 1989 Oct;135(4):593-7. PubMed.

    . Comparison of neurodegenerative pathology in transgenic mice overexpressing V717F beta-amyloid precursor protein and Alzheimer's disease. J Neurosci. 1996 Sep 15;16(18):5795-811. PubMed.

    . Sequestration of RNA in Alzheimer's disease neurofibrillary tangles and senile plaques. Ann Neurol. 1997 Feb;41(2):200-9. PubMed.

    . Abundant Aβ fibrils in ultracentrifugal supernatants of aqueous extracts from Alzheimer's disease brains. Neuron. 2023 Jul 5;111(13):2012-2020.e4. Epub 2023 May 10 PubMed.

    View all comments by Dennis Selkoe
  2. This work represents a veritable tour de force. The authors used high-pressure freezing to preserve fresh brains from AppNL-G-F knock-in mice for fluorescence light microscopy and electron cryo-tomography (cryo-ET). These mice deposit abundant Aβ42 plaques that are made of human Aβ with the Arctic mutation E22G (Saito et al., 2014). Leistner et al. used the fluorescence of Aβ plaques labelled by the amyloid dye methoxy-X04 to know where to cut the frozen brain tissues into sections thin enough for transmission electron microscopy, and then performed cryo-ET on those sections.

    This work is pioneering the characterization of rodent brains with human disease pathologies at the structural molecular level. It describes the three-dimensional arrangement of Aβ filaments in situ. Ultimately, one would want to be able to perform this type of analysis in human brains from individuals with various neurodegenerative diseases. Knowledge of where amyloid filaments are in the brain and how they interact with other cellular components will be invaluable, not only for understanding fundamental aspects of human diseases, but also for examining the validity of model systems.

    The authors also extracted Aβ42 filaments from the brains of 11- to 13-month-old homozygous AppNL-G-F mice and performed standard helical reconstruction electron cryo-microscopy (cryo-EM) to obtain an atomic model. Filament structures were identical to those we determined previously from the brains of 22-month-old homozygous AppNL-G-F mice, demonstrating that the filament structures did not change between the ages of 11 and 22 months (Yang et al., 2023). 

    We also reported that the structures of Aβ42 filaments from the temporal cortex of an individual with the Arctic mutation differed from those of AppNL-G-F knock-in mice, raising doubts about the usefulness of this line. Homozygous mice from knock-in line AppNL-F deposit wild-type human Aβ42 (Saito et al., 2014) that is identical in structure to Type II Aβ42 filaments from the brains of individuals with Alzheimer’s disease or with a variety of other diseases with Aβ42 co-pathology (Yang et al., 2022). It will be interesting to study the molecular architecture of Aβ path biology in the brains from AppNL-F mice.


    . Single App knock-in mouse models of Alzheimer's disease. Nat Neurosci. 2014 May;17(5):661-3. Epub 2014 Apr 13 PubMed.

    . Cryo-EM structures of amyloid-β 42 filaments from human brains. Science. 2022 Jan 14;375(6577):167-172. Epub 2022 Jan 13 PubMed.

    . Cryo-EM structures of amyloid-β filaments with the Arctic mutation (E22G) from human and mouse brains. Acta Neuropathol. 2023 Mar;145(3):325-333. Epub 2023 Jan 7 PubMed.

    View all comments by Michel Goedert
  3. I find this study quite exciting to visualize the amyloid fibrils in the brains of APPNL-G-F knock-in mice using cryo-electron tomography (Cryo-ET). One of the striking findings is the identification of so many extracellular vesicles/exosomes (50-200 nm in diameter) in the plaque region, which are 100-fold more abundant than in wild-type mouse brain. This study demonstrates that EVs accumulate in the amyloid plaques. The EVs could originate from surrounding activated glia. Indeed, the ultrastructural analysis of the sub-cellular and membrane-bound compartments surrounding the MX04-stained plaques is consistent with those of microglia and astrocytes. This is in line with our recent study, showing that plaque-associated microglia are activated and exhibit enhanced biogenesis and secretion of EVs, which we visualized in single-vesicle resolution in the APPNL-G-F knock-in mouse brain (Clayton et al., 2021). 

    In accord, one of the well-accepted EV markers, CD9, is also a marker of disease-associated / neurodegenerative microglia. The authors also identified multilamellar bodies composed of vesicles wrapped with multiple concentric rings of membrane, which they discussed as intracellular intermediates of autophagy. A similar structure was also identified in the EVs isolated from human cerebrospinal fluid by Cryo-EM (Emelyanov et al., 2020), suggesting that autophagy intermediate-derived EVs may also be secreted from the surrounding activated glia.

    It would be of interest, if these findings can be reproduced in other APP mouse models or human AD brain tissues, to know how the depletion of microglia, or their inactivation, alters the EV deposition around plaques, and if these EVs also contain tau aggregates, as recently shown by the Duff and Ryskeldi-Falcon groups (Fowler et al., 2023). These EVs may also be the source of dystrophic neurites and other pathologies.


    . Plaque associated microglia hyper-secrete extracellular vesicles and accelerate tau propagation in a humanized APP mouse model. Mol Neurodegener. 2021 Mar 22;16(1):18. PubMed. Correction.

    . Cryo-electron microscopy of extracellular vesicles from cerebrospinal fluid. PLoS One. 2020;15(1):e0227949. Epub 2020 Jan 30 PubMed.

    . Tau filaments are tethered within brain extracellular vesicles in Alzheimer's disease. bioRxiv. 2023 Apr 30; PubMed.

    View all comments by Tsuneya Ikezu
  4. This interesting and rigorous study from the Frank and Ranson labs provides additional structural and molecular information on the diversity of amyloid plaques isolated from fresh App KINL-G-F brains. The author's integrated cryogenic correlated light electron microscopy and cryo-electron tomography-based analysis of hydrated plaques isolated from brain tissues nicely complements and extends previously reported amyloid structures. Maybe most interesting is their finding that the NL-G-F structures did not show an enrichment in short protofibrils as one might have expected based on results using synthetic or recombinant peptides.

    Furthermore, these results provide additional evidence that Arctic Aβ fibrils differ significantly from Aβ fibrils formed in sporadic AD brain and provide some suggestions for how these structures may physically interact with additional macromolecular complexes within the extracellular space.

    View all comments by Jeffrey Savas
  5. Fresh eyes reveal diversity of Aβ fibrillar structures in amyloid plaques.

    It’s always good to see fresh eyes looking at the major AD unsolved questions: What constitutes an Aβ plaque, its central core, peripheral halo, and surrounding tau-positive reactive neurites? When the experts claim that the Aβ-PET signals lowered by anti-amyloid therapies are the result of “plaque-lowering” effects, how do we know that all constituents of the plaque are captured by the PET-ligand in the 30 minutes after infusion? Does extra-cellular Aβ ever come into direct contact with intracellular tau? Does interaction (direct or indirect) of tau and Aβ involve liquid-liquid phase condensations, which result in various structural forms of globular/fibrillar aggregates?

    These and other questions are beginning to be addressed by studies, such as this one from Leistner and colleagues in Leeds, which once again demonstrates the structural diversity of Aβ fibrillar aggregates, which are associated here with the "Arctic" Aβ /APP mutation found in familial forms of Alzheimer's disease. Notably, this mutation has led to the development of highly promising therapeutics for AD treatment, such as the FDA-approved antibody lecanemab. The cryo-EM structure of “ex vivo” purified fibrils in this study differs from previously reported structures, highlighting the notable impact of the Arctic mutation. Furthermore, these structural findings have unveiled additional fibrillar species, such as thin protofilament-like rods and branched fibrils. Nevertheless, all of these species share a common structural core element with an S-shape (Yang et al., 2022), encompassing residues 20–36 of the Aβ 42 sequence.

    While cryogenic correlated light and electron microscopy (cryo-CLEM) and cryo-electron tomography (cryo-ET) have revealed that “in-tissue” Aβ fibrils are organized in a lattice or parallel bundles, and are interdigitated by subcellular compartments, extracellular vesicles, extracellular droplets, and extracellular multilamellar bodies, no information has been provided regarding the potential presence of neurofibrillary tangles (NFT) formed by tau filaments or the presence of tau filaments in sarkosyl extracts. In this study, it was assumed that AD pathologic changes include only intracellular tangles of tau. However, it has been reported that tau can be secreted into the extracellular compartment through vesicular pathways, such as exosomes, and via direct crossing of the plasma membrane. Furthermore, its interaction with lipid membranes may play a role in the formation and spreading of these pathological aggregates (El Mammeri et al., 2023). Additionally, Aβ and tau may form heterogeneous globular aggregates (Mukherjee et al., 2023), possibly through biomolecular condensation pathways and extracellular droplets, which were noted in the reported cryo-ET survey.

    The identification of Aβ fibrillar deposits in the tissue was accomplished using Methoxy-X04, a Congo Red derivative known to bind to any β-pleated sheet, and which is also utilized for NFT detection (Kuchibhotla et al., 2014). Uncertainties regarding the observed fibrillar/filamentous aggregates could be addressed through immunolabeling EM, particularly by employing lecanemab. It remains unclear whether the observed “in-tissue” molecular architecture includes exclusively Aβ fibrillar forms or if other β-sheet-rich species (such as tau) or globular (co)-aggregates contribute to the lesions of AD. Finally, it might be more fruitful to shift the focus away from the use of end-stage AD brain models and place greater emphasis on investigating the earliest potential sites of Aβ/tau interaction in the precuneus in the default mode network, where the process of AD usually starts (Ruwanpathirana et al., 2022).


    . Membrane-induced tau amyloid fibrils. Commun Biol. 2023 Apr 28;6(1):467. PubMed.

    . Neurofibrillary tangle-bearing neurons are functionally integrated in cortical circuits in vivo. Proc Natl Acad Sci U S A. 2014 Jan 7;111(1):510-4. Epub 2013 Dec 24 PubMed.

    . Quantitative proteomics of tau and Aβ in detergent fractions from Alzheimer's disease brains. J Neurochem. 2023 Feb;164(4):529-552. Epub 2022 Nov 22 PubMed.

    . Mapping the association between tau-PET and Aβ-amyloid-PET using deep learning. Sci Rep. 2022 Aug 30;12(1):14797. PubMed.

    . Cryo-EM structures of amyloid-β 42 filaments from human brains. Science. 2022 Jan 14;375(6577):167-172. Epub 2022 Jan 13 PubMed.

    View all comments by Colin Masters
  6. I had initially shied away from commenting mainly because the EM images with CLEM are more difficult for me to interpret. However, this is an interesting paper, and the news story and comments have added to this. I’m also prompted to write because of the suggestion in a comment to do immuno-EM with lecanemab. I agree with this suggestion and point out that we had done immuno-EM with an antibody against high-molecular-weight oligomers in AD transgenic and human biopsy brain tissue quite some time ago (Takahashi et al., 2004). Interestingly, we did see organelles that seem to be in a transition from intra- to extracellular that associated with pathological Aβ oligomer accumulations within neuritic dystrophies and with fibrillar extracellular amyloid.

    Neuritic dystrophies contain the most abundant autophagic vesicles associated with plaques and could therefore be the source of the extracellular vesicles within plaques shown in this paper. However, I wonder how certain the authors are about the plasma membrane demarcations; might some amyloid fibrils intermingling with the "extracellular vesicles” possibly even be within microglia?


    . Oligomerization of Alzheimer's beta-amyloid within processes and synapses of cultured neurons and brain. J Neurosci. 2004 Apr 7;24(14):3592-9. PubMed.

    View all comments by Gunnar Gouras
  7. It is amazing to be able to visualize the organization of Aβ42 fibrils in the brain of an AD model. Others have imaged amyloid fibrils in cultured cells before (Bäuerlein et al., 2017); however, this is the first time I know of that amyloid fibrils were imaged in a brain at this resolution. I am hopeful that one day it will be possible to acquire analogous images from human AD patients.

    Leistner et al. showed us that the atomic-level structure of the Aβ42 fibrils purified from the mouse AD model differ from those purified from human AD patients (i.e., distinct fibril polymorphs). I am curious whether these atomic-level differences would lead to structural differences on a larger scale. That is, would the pattern of distribution of fibrils in the brains of humans with AD differ from the pattern revealed in mice by Leistner et al.? If different, it may suggest a mechanism by which fibril polymorphs define disease (viewing mouse and human AD as similar, but non-identical diseases). If the fibril distributions in human and mouse are the same, then the result would further validate the mouse system as a model for human AD.

    It might be helpful, as a reference for amyloid researchers, to include cryoET density in the Amyloid Atlas [added by editors: see May 2023 news].


    . In Situ Architecture and Cellular Interactions of PolyQ Inclusions. Cell. 2017 Sep 21;171(1):179-187.e10. Epub 2017 Sep 7 PubMed.

    View all comments by Michael Sawaya

Make a Comment

To make a comment you must login or register.

This paper appears in the following:


  1. Amyloid Jungle: Plaque Fibrils Mesh With All Manner of Vesicles, Membranes


  1. APP E693G (Arctic)