Most Aβ fibril structures described to date have come from the parenchyma or the brain vasculature of people with AD. A new study, posted December 22 on bioRxiv, resolved the structures of fibrils plucked from the cerebrovasculature of two people with cerebral amyloid angiopathy, a pathology that wreaks havoc on the blood vessels of the brain. Using cryo-electron microscopy and nuclear magnetic resonance, researchers led by Saikat Chowdury and Steven Smith of Stony Brook University in New York identified two predominant fibril types that shared a common protofilament structure featuring two cross-β segments. They differed at their N-termini, where one protofilament type snapped into a discernable structure and the other refused to conform.

The findings point to similarities and differences in the forms of amyloid that populate different regions of the brain. However, some researchers suggested that the structures may have been sullied by the in vitro amplification process that the authors used to boost their yield of fibrils.

Previously, researchers led by Marcus Fändrich of Ulm University in Germany used cryo-EM to resolve structures of Aβ40 fibrils extracted directly from the meningeal vasculature of people with AD (Nov 2019 news). That study, which involved no amplification step, found three structural polymorphs of Aβ fibril, each sharing a similar protofilament core structure comprising dimers of C-shaped Aβ40 peptides positioned back-to-back. Later, researchers led by Michel Goedert and Sjors Scheres of the MRC Laboratory of Molecular Biology, Cambridge, England, U.K., used cryo-EM to solve the structure of Aβ42 fibrils extracted from the brain parenchyma of people with AD (Jan 2022 news). That study identified two filament structures comprising distinct S-shaped protofilament cores.

To zero in on the structure of fibrils from CAA, first author Elliot Crooks and colleagues used laser capture microdissection to nab vascular amyloid deposits from brain slices of one person with sporadic CAA, and another with familial Dutch CAA, which is caused by a glutamate-to-glutamine mutation at position 22 within the Aβ peptide. They then amplified the extracted fibrils, using them to seed fibrillization of synthetic wild-type Aβ40 monomers. After three rounds of seeding, they analyzed the resulting fibrils via cryo-EM.

Two predominant fibril structures emerged, both of which appeared to varying degrees within both samples. Population A—a highly twisted, homogenous fibril—predominated in the person with sporadic CAA. Population B—a more heterogenous, less-twisted fibril— predominated in the person with familial CAA. The core protofilament structure of population A comprised two Aβ40 peptides, held together by hydrophobic interactions between two cross-β segments (see image below). Notably, the N-terminus was distinctly ordered, featuring two β-strands formed by several electrostatic contacts. The structure of population A was similar to the one Fändrich’s group found. “The independent observation of similar N-terminal conformations in separate individuals, despite different extraction and purification methods, argues that the population A fibrils are a common feature of vascular amyloid,” the authors wrote.

Two Flavors of Fibril in CAA. Both populations of fibril featured a protofilament core comprising two Aβ40 monomers, glued together via hydrophobic interactions. The N-terminus of population A (left) was ordered, while population B (right) was unstructured. [Courtesy of Crooks et al., bioRxiv, 2023.]

Population B featured a similar protofilament core, but it differed in having the N-terminus with no ordered structure. This unstructured N-terminus has also been described for Aβ40 fibrils amplified from the brain parenchyma (Ghosh et al., 2021).

Aβ42 peptides within parenchymal brain deposits of people with AD also have a wobbly N-terminus, and this disordered region is the target of several therapeutic antibodies, including aducanumab, gantenerumab, and lecanemab. Crooks and colleagues suggested that antibody binding to population B-type Aβ40 fibrils found within the vasculature of people with CAA might weaken blood vessels.

In a comment to Alzforum, Goedert and Scheres questioned whether the reported structures reflect those in human brain, noting that their conformations may have been altered during the in vitro amplification process. “We consider that the structures presented in this manuscript are themselves the result of ‘in vitro’ assembly, albeit with the use of brain-derived seeds,” they wrote (see comment below). “We would therefore recommend a clearer distinction in the wording used to refer to structures of filaments that are extracted directly from human tissues, structures of filaments of recombinant or synthetic proteins that are formed ‘in vitro’ by seeded aggregation with brain-derived seeds, and the structures of filaments of recombinant or synthetic proteins from spontaneous ‘in vitro’ aggregation.”—Jessica Shugart


  1. The manuscript by Crooks et al. describes cryo-EM structures and provides ssNMR data for filaments of synthetic wild-type Aβ40 peptides that were produced by seeded aggregation using seeds of vascular Aβ deposits extracted from brain sections of individuals with sporadic (wild-type Aβ Alzheimer’s disease) or inherited (E22Q Aβ, Dutch disease) cerebral amyloid angiopathy (CAA) by laser capture microdissection. Only one case of sporadic and one case of inherited CAA were used. The structures presented have some interesting similarities to, and some differences with, the Aβ40 structures that were previously described for non-amplified filaments from the leptomeninges of individuals with sporadic AD (Kollmer et al., 2019).

    Crooks et al. interpret these differences with the (implicit) assumption that the filaments formed by seeded aggregation have the same structures as the initial seeds. As was previously shown for seeded aggregation of recombinant α-synuclein with seeds from the putamina of individuals with multiple system atrophy (MSA), seeded aggregation does not necessarily replicate the structures of the seeds (Lövestam et al., 2021). We recently also raised this point in regard to the manuscript by Fan et al. on seeded aggregation of recombinant α-synuclein with seeds from cerebrospinal fluid of individuals with Parkinson’s disease (Dec 2022 news). Because Crooks et al. provide no evidence to support their assumption that seeded aggregation replicated the seed structures, it is possible that the differences between the structures from seeds of sporadic and familial CAA, and the differences between these structures and those obtained by Kollmer et al. are due to artefacts of the seeded aggregation process.

    This issue is ignored in the manuscript, which pitches structures of “human-derived fibrils” against structures obtained “in vitro.” However, we consider that the structures presented in this manuscript are themselves the result of in vitro assembly, albeit with the use of brain-derived seeds. We would therefore recommend a clearer distinction in the wording used to refer to structures of filaments that are extracted directly from human tissues (like the ones described by Kollmer et al.), structures of filaments of recombinant or synthetic proteins that are formed in vitro by seeded aggregation with brain-derived seeds (like the ones described by Crooks et al.), and the structures of filaments of recombinant or synthetic proteins from spontaneous in vitro aggregation.


    . Cryo-EM structure and polymorphism of Aβ amyloid fibrils purified from Alzheimer's brain tissue. Nat Commun. 2019 Oct 29;10(1):4760. PubMed.

    . Seeded assembly in vitro does not replicate the structures of α-synuclein filaments from multiple system atrophy. FEBS Open Bio. 2021 Apr;11(4):999-1013. Epub 2021 Feb 24 PubMed.

  2. The preprint at bioRXiv by Crooks et al. reports on the different structural properties of Aβ40 fibrils between an individual with sporadic CAA and an individual with familial CAA using cryo-EM and NMR. Although both structures shared a common fibrillar core, they differed in the fold of the N-terminal sequence. While one structure exhibited an ordered N-terminal fold comprised of two β-strands, the other structure showed a disordered N-terminus. The data may have relevance for potential pharmacological targets that distinguish amyloid aggregates in CAA from those within plaques. The authors argue that the binding properties of therapeutic anti-Aβ antibodies to vascular amyloid may explain the observed side effects such as hemorrhages in some of the treated AD patients.

    On the other hand, soluble low-molecular weight oligomers and soluble protofibrils may prove better targets to prevent AD due to their higher toxicity. Therefore, it would make sense, in my view, to develop therapeutic antibodies or vaccines reacting against soluble aggregates of the amyloid cascade with a well-defined structure to avoid binding to CAA and plaques (see Mar 2022 news).

Make a Comment

To make a comment you must login or register.


News Citations

  1. Right Turn: Aβ Fibril Structure from Alzheimer’s Brain Reveals Surprising Twist
  2. Cryo-EM Unveils Distinct Aβ42 Fibril Structures for Sporadic, Familial AD

Mutations Citations

  1. APP E693Q (Dutch)

Paper Citations

  1. . Molecular structure of a prevalent amyloid-β fibril polymorph from Alzheimer's disease brain tissue. Proc Natl Acad Sci U S A. 2021 Jan 26;118(4) PubMed.

Further Reading

No Available Further Reading

Primary Papers

  1. . An electrostatic cluster guides Aβ40 fibril formation in cerebral amyloid angiopathy. bioRxiv 2022.12.22 bioRxiv