Sign in

Abstract

Filament assembly of amyloid-{beta} peptides ending at residue 42 (A{beta}42) is a central event in Alzheimers disease. We report the cryo-EM structures of A{beta}42 filaments from brain. Two structurally related S-shaped protofilament folds give rise to two types of filaments. Type I filaments were found mostly in the brains of individuals with sporadic Alzheimers disease and Type II filaments in individuals with familial Alzheimers disease and other conditions. The structures of A{beta}42 filaments from brain differ from those of filaments assembled in vitro. By contrast, in AppNL-F knock-in mice, A{beta}42 deposits were made of Type II filaments. Knowledge of A{beta}42 filament structures from human brain may lead to the development of inhibitors of assembly and improved imaging agents.

PDFDOI
Full-text digest ↓

02 /Specification & Classification

Click any tag to browse matching papers

In-depth summary

✦ Multi-agent verified full-text synthesis

Claude Opus 5.5 · xhigh effort · 8 Nov 2026 · from the full text of version 1 (19 Oct 2021) · CC BY 4.0

Cryo-EM shows brain-derived Aβ42 filaments adopt two related S-shaped protofilament folds: Type I dominates sporadic Alzheimer's, Type II familial Alzheimer's, other disorders and App^NL-F knock-in mice.

Aβ42 filament assembly is central to Alzheimer's disease, yet the structures of brain-derived Aβ42 filaments were unknown; earlier brain-derived cryo-EM work described Aβ40 (meningeal aggregates or seeded filaments). Solving the native folds shows that in vitro assemblies only partly match them and identifies a knock-in mouse line that does, which can guide assembly inhibitors and imaging agents.

Key results

  1. ShownType I filaments, the majority form in sporadic Alzheimer's disease, contain two identical S-shaped protofilaments that wrap each other with extended N-terminal arms across a largely hydrophobic interface. A putative bound metal ion sits next to the E22 and D23 carboxylates.

    Atomic model built in a 2.5 Å map from sporadic Alzheimer's case 1 (Krios G4 dataset); ordered core G9-A42 with S-shaped domain residues 19 to 42 and N-terminal arm 9 to 18; pseudo-21 symmetry; interface from L34, V36, V39, I41 and Y10, V12, Q15, L17; minority Type Ib (paired Type I filaments, 3.5 Å map) in sporadic cases 1 and 3. · Type I Aβ42 Filaments from Human Brain; Helical reconstruction; Model building and refinement

  2. ShownType II filaments share the S-shaped fold over residues F20-A42 but differ by two flipped peptide groups. They pack with C2 symmetry through a smaller interface on the opposite face and leave hydrophobic patches exposed on the filament surface.

    Built in a 2.8 Å map from the pathological aging (PA) case; ordered core V12-A42; G25-S26 and V36-G37 peptides flipped by approximately 180°, the latter shifting the C-terminal segment by about one peptide along the helical axis; interface held by reciprocal K28 amino to A42 carboxyl salt bridges; putative metal density at E22/D23 retained. · Type II Aβ42 Filaments from Human Brain; Model building and refinement

  3. SuggestedFilament type tracked disease category. A Type I majority occurred only in sporadic Alzheimer's cases, which had the most plaque cores. Type II dominated familial Alzheimer's and non-Alzheimer conditions rich in diffuse Aβ deposits, and some brains held both types.

    Ten individuals: three sporadic and two familial (APP V717F, PSEN1 F105L) Alzheimer's cases plus one case each of age-related tau astrogliopathy, Parkinson's disease dementia, dementia with Lewy bodies, GRN-mutant frontotemporal dementia and PA; Type II was 17% in sporadic case 3, Type I 24% in familial case 2; Type I cases older at death; no APOE genotype correlation. · Main Text; Type II Aβ42 Filaments from Human Brain; Discussion; Clinical history and neuropathology

  4. ShownNeither in vitro assembled Aβ42 nor meningeal Aβ40 aggregates reproduce the brain folds. In vitro Aβ42 structures share an S-shaped domain but differ in side-chain contacts and protofilament packing, and NMR models of Osaka-mutant Aβ40 most resemble Type I.

    Brain Type I/II filaments are left-handed, unlike right-handed C-shaped meningeal Aβ40; seeded Aβ40 (PDB 6W0O) shares a near-identical G25-G37 conformation; none of the in vitro Aβ42 cryo-EM/NMR structures (PDB 5OQV, 2NAO, 2MXU, 5KK3) matched at residue level; Aβ40ΔE22 (Osaka) NMR models closest to Type I. · Comparison with Known Structures; Figure legends

  5. ShownAβ42 filaments from homozygous App^NL-F knock-in mice were identical to human Type II. This makes the line the first experimental system reported to reproduce a brain-derived Aβ42 filament structure, which the authors link to a raised Aβ42 to Aβ40 ratio.

    Filaments extracted from brains of 18-month-old homozygous mice (C57BL/6 background; three brains pooled) carrying humanized Aβ with Swedish KM670/671NL and Iberian I716F App mutations; cryo-EM structure matched human Type II; authors note I716F, like human APP V717F, increases the Aβ42 to Aβ40 ratio. · Type II Aβ42 Filaments from AppNL-F Mouse Brain; Discussion; Mice

Methods, data and reproducibility

Sarkosyl-insoluble Aβ filaments were extracted from human brain regions and App^NL-F mouse brain, imaged by cryo-EM and solved by helical reconstruction in RELION, with atomic models built in COOT. Composition was checked by immunoblotting, MALDI-TOF and LC-MS/MS, immunohistochemistry and immunogold negative-stain EM.

  • Frontal or parietal cortex from sporadic Alzheimer's disease cases; sarkosyl-insoluble filaments for cryo-EM, immunoblotting and mass spectrometry · human · n = three sporadic cases
  • Frontal cortex from familial Alzheimer's disease cases carrying APP V717F or PSEN1 F105L · human · n = two familial cases
  • Non-Alzheimer cases with Aβ copathology: tau astrogliopathy, Parkinson's disease dementia, Lewy body dementia, GRN-mutant frontotemporal dementia, pathological aging · human · n = five individuals, one per condition
  • Brains of 18-month-old homozygous App^NL-F knock-in mice (C57BL/6 background), pooled for sarkosyl-insoluble filament extraction · mouse · n = three mice (pooled)
  • Exome sequencing of human cases to check APP, PSEN1 and PSEN2 (wild-type except the two familial cases) · human
  • Published Aβ filament structures used for comparison: seeded Aβ40 and in vitro assembled Aβ42 (cryo-EM and solid-state NMR) · PDB 6W0O, 5OQV, 2NAO, 2MXU, 5KK3
Methods & tools
RELION (helical reconstruction; built-in MotionCor2 implementation), CTFFIND-4.1, COOT, REFMAC and PHENIX (model refinement), MolProbity, ChimeraX, EPU acquisition software, Thermo Fisher Titan Krios G2/G3/G4 with Gatan K2/K3 or Falcon-4/Falcon-4i detectors, MALDI-TOF (Bruker Ultraflex III) and LC-MS/MS (Q Exactive Plus), searched with Mascot v 2.4, Scaffold 4.8.4, Agilent SureSelectTX human all-exon V6 sequencing on Illumina HiSeq4,000, Western blotting, immunohistochemistry and immunogold EM (Aβ40/Aβ42 antibodies, D54D2, 1E11)
Reagents & environment
CTFFIND-4.1; Mascot v 2.4; Scaffold version 4.8.4; RELION (incl. its MotionCor2 implementation), COOT, REFMAC, PHENIX, MolProbity and ChimeraX used without stated versions.
Reproducibility gaps
The availability statement says data are unrestricted and deposited, but the accession codes are cut off in the supplied text; no code repository is given; acquisition and model statistics are deferred to Tables S2 and S3.
Limitations (6)
  • No filaments made mainly of Aβ40 were found, but they may have been present at low abundance or may not have extracted as dispersed filaments suited to cryo-EM. Small amounts of Aβ40 or truncated peptides could also be built into both filament types. · stated by the authors
  • The biological relevance of the Type I versus Type II difference is unknown. Type I cases were older at death, but filament type did not correlate with APOE genotype. · stated by the authors
  • Because App^NL-F mice form only Type II filaments, the authors suggest the line may model some familial Alzheimer's cases but not necessarily sporadic disease. · stated by the authors
  • Contralateral immunohistochemistry showed abundant Aβ40 deposits in the frontotemporal dementia case, which disagrees with immunoblotting and mass spectrometry of the extracted tissue. The authors suggest this may reflect hemispheric asymmetry in Aβ deposition. · stated by the authors
  • The links between filament type and disease rest on ten individuals, with one case per non-Alzheimer condition and one region analysed by cryo-EM per brain. The age-at-death and diagnostic associations were not tested statistically. · noted in this summary
  • The modified sarkosyl extraction and the pronase treatment before cryo-EM could enrich particular filament populations. The study does not directly test how well the solved structures represent all Aβ42 aggregates in tissue, for example diffuse versus cored deposits. · noted in this summary

Who should read:Read this if you study Aβ aggregation, amyloid imaging ligands or Alzheimer's mouse models and need atomic-level brain-derived Aβ42 filament folds to benchmark in vitro or animal-model assemblies.

Reproducibility ledger

Nobody has reported trying to reproduce this paper yet. Did you replicate an experiment or run the code? Your report saves the next lab a week.

Sign in to file a reproducibility report.

Similarity Map
SelectedSame fieldCross-field
THISYang '23Yang '23Aubrey '23Zielinski '23Kostylev '26Frieg '24
84% semantic matchNeuroscience
Yang Yang, Alexey G. Murzin, Sew‐Yeu Peak‐Chew+6 more·Acta Neuropathologica Communications·Dec 4, 2023· 56 citations