Cryo-EM structure of ex vivo fibrils associated with extreme AA amyloidosis prevalence in a cat shelter

Amyloidosis involves extracellular deposition of proteinaceous aggregates identifiable by Congo Red birefringence. Over 50 distinct amyloidogenic proteins are linked to specific diseases. Serum amyloid A (SAA), an acute-phase apolipoprotein that can reach up to 1000-fold elevated serum levels during chronic inflammation, underlies systemic AA amyloidosis. Misfolded SAA can transition under lysosomal conditions into highly ordered, proteolysis-resistant fibrils that damage organs, commonly the kidney. Prior cryo-EM studies showed that ex vivo AA amyloid from human and mouse share cross-β architecture but adopt distinct folds despite ~76% sequence identity. Domestic animals, including cats, also develop AA amyloidosis; certain breeds have familial predisposition. Notably, captive cheetahs and shelter-kept domestic short hair (DSH) cats show extreme AA amyloidosis prevalence (≈70% in cheetah; 57–73% in shelter cats), with evidence suggesting prion-like transmission, potentially via fecal-oral routes (SAA detected in bile). The present study aims to determine the high-resolution cryo-EM structure of ex vivo AA fibrils extracted from the kidney of a DSH cat from a shelter with extreme disease prevalence and to compare the feline fold with human and mouse structures, and with cheetah SAA sequences, to explore structural bases for stability and potential transmissibility.

Key prior findings include: (1) Amyloids are diverse; cryo-EM has enabled structure-based disease classification (e.g., tauopathies). (2) Ex vivo human and mouse AA amyloid fibrils exhibit distinct folds despite high sequence identity, highlighting polymorphism. (3) SAA biology: a conserved acute-phase apolipoprotein involved in lipid handling and retinol delivery; under chronic inflammation elevated SAA and lysosomal conditions promote amyloid formation. (4) High prevalence of AA amyloidosis in captive cheetahs and in shelter cats; experimental work indicates prion-like transmissibility of AA amyloid via intravenous and oral routes, with fecal transmission implicated in cheetahs. (5) In vitro vs ex vivo fibrils can differ; disease-associated amyloids tend to be more stable than functional reversible amyloids. These studies establish the context for examining feline AA fibril structure and its potential link to transmission and prevalence.

Case and sampling: A sterilized female DSH cat from a Northern Italy shelter developed systemic disease and was euthanized at 6 years. Organs were collected within 5 hours post-mortem. Histology and immunofluorescence: Formalin-fixed, paraffin-embedded 4–5 µm sections were stained with H&E and Congo Red; amyloids and nuclei were labeled with Thioflavin S and DAPI. Immunofluorescence used mouse sera raised against feline SAA-derived peptides (residues 41–50, 63–74, 109–122) and appropriate secondary detection. Fibril extraction: From frozen kidney tissue (~0.5 g), samples were washed in Tris-calcium buffer, digested with 5 mg Clostridium histolyticum collagenase, and subjected to multiple homogenization cycles (Tris-EDTA buffer, then water). Water extracts were analyzed by SDS-PAGE; water extract #2 was used for cryo-EM and proteomics. LC-MS/MS: Solubilized fibril proteins (8 M urea/0.1 M DTT) were alkylated (150 mM iodoacetamide), diluted (final 1.3 M urea), digested with trypsin (1:20 w/w, 16 h, 37 °C), C18-purified, and analyzed by LC-MS/MS. Top Uniprot hits from Felis catus proteome included Q1T770; peptides mapped predominantly to residues 19–111 of the feline SAA precursor (residues 1–93 of mature protein). Cryo-EM data collection: 4 µl fibril sample was applied to glow-discharged C-flat 1.2/1.3 300-mesh Cu grids, blotted, and plunge-frozen. Data (2,652 movies) were collected on a Talos Arctica 200 kV with a Falcon 3 detector in counting mode. Helical reconstruction: In RELION 3.1, fibrils were manually picked. An initial 65,131 segments (binned) informed model generation (estimated crossover ~700 Å). A larger set of 381,233 segments (250-pixel boxes) underwent 3D auto-refinement (initial C1) yielding ~4 Å; imposing C2 symmetry improved to 3.8 Å. After classification, polishing, and CTF refinement, 65,122 particles were refined to a final map at 3.3 Å resolution with left-handed twist, helical twist 1.3° and rise 4.9 Å. Model building and validation: Auto-sharpened maps (Phenix) guided de novo building starting from a distinctive bulge identified as P66GGAW70. Five 76-residue chains per proto-filament were built and refined with NCS restraints in Coot, ISOLDE, and Phenix (with/without Amber). Pro-66 was modeled as cis to fit density; additional nearby densities (near Gly-65, Ser-82) may correspond to ions/unknown ligands. Validation metrics: CCmask 0.74, EMringer 5.1, MolProbity 1.4. Data deposition: EMPIAR-11001, EMD-14726, PDB 7ZH7; proteomics PXD035851.

• Determined a 3.3 Å cryo-EM structure of ex vivo AA amyloid fibrils from a DSH cat kidney from a shelter with extreme disease prevalence. • Fibrils are homogeneous, straight, with crossover distances of ~650–700 Å; left-handed helical twist 1.3°, helical rise 4.9 Å. • Architecture: two identical proto-filaments; each proto-filament subunit comprises 76 ordered residues forming 11 β-strands spanning residues 19–94 of the SAA precursor; residues 95–111 were not resolved. • Distinctive features: a central β-arch (Asp50–Arg64); an unusual backbone bulge in segment P66GGAW70 with Pro-66 modeled as cis (first reported cis-Proline in amyloid; requires higher resolution for definitive assignment). N-terminal tails are surface-exposed at edges; C-terminal tails are buried facing each other; tails tilted ~15° (edge) and ~10° (face) relative to the β-arch. • Stabilization: extensive staggered intra- and inter-protofilament interactions, including multiple hydrophobic clusters and ionic locks across four rung layers, contributing to fibril stability. • Fold comparison: Despite >70% sequence identity with human and mouse SAA, the feline fibril adopts a distinct fold and longer core (by 22 residues vs human). Some segments (human 24–54) superpose well (rmsd ~2.5 Å), consistent with type-2 polymorphism. • A feline-specific eight-residue insert (around precursor position 86) is buried at the inter-protomer interface, extending the interface. • Stability estimates: ΔGsol ≈ −42 kcal/mol per molecule (similar to mouse; 5–10 kcal/mol more stable than human); dissociation energy cost (ΔGdiss) is ~4 kcal/mol higher than mouse and ~8 kcal/mol higher than human. The insert increases buried surface area by ~2000 Ų and core mass; it may also destabilize native SAA, increasing misfolding propensity. • Cheetah relevance: AA amyloid from captive cheetah is 99% identical in sequence to feline (only N93S difference) and is structurally compatible with the presented fold; cheetahs exhibit ~70% prevalence with reported prion-like transmission.

The high-resolution structure answers the central question of how feline SAA assembles ex vivo in a setting of extreme AA amyloidosis prevalence. The unique fold, extensive interfacial interactions, and incorporation of a feline-specific eight-residue insert rationalize higher stability and larger buried interfaces relative to human AA fibrils. Such stability may enhance environmental persistence and resistance to clearance, offering a structural rationale for observed prion-like features and fecal-oral transmissibility in felids. The structure clarifies why sequence-homologous SAA proteins can yield distinct fibril architectures (type-2 polymorphism), influencing exposure/burial of segments and interaction networks. The near-identity with cheetah SAA suggests that a similar fold could underlie the high prevalence in captive cheetahs. The observation of a cis-Proline at position 66, if confirmed, expands known conformational motifs in amyloids and may contribute to local structural constraints and stability. Overall, the findings bridge sequence variation to fibril architecture and biophysical robustness, providing a mechanistic basis for epidemiological observations in shelters and zoos.

This study reports the first ex vivo cryo-EM structure of a spontaneously occurring animal AA amyloid fibril, obtained from a shelter cat with systemic AA amyloidosis. The feline fibril exhibits a distinct two-protofilament fold with 76-residue cores, stabilized by staggered ionic and hydrophobic interactions and a unique eight-residue insert that increases core mass and buried surface area, yielding higher stability than human AA fibrils. The structure is compatible with the nearly identical cheetah sequence, suggesting a shared structural basis that may facilitate prion-like transmission and contribute to extreme prevalence in captive felids. Future work should include direct structural determination of cheetah AA fibrils, higher-resolution mapping to definitively resolve proline isomerization and unmodeled densities, biochemical tests of environmental persistence and transmissibility linked to stability, and broader sampling across individuals and tissues to assess structural variability and polymorphism in feline AA amyloidosis.

• Map did not resolve residues 95–111 of the SAA precursor; the ordered core spans residues 19–94 only. • Pro-66 was modeled as cis based on density, but higher resolution is needed to conclusively establish proline isomerization. • Additional nearby densities (e.g., near Gly-65 and Ser-82) could not be assigned to specific ions or ligands. • The structural analysis is based on fibrils from a single cat sample; generalizability across animals and tissues is untested. • Histology/IF and LC-MS/MS were performed on the same extract and representative sections (n=1 per final dataset), limiting statistical robustness.