Myelin dysfunction drives amyloid-β deposition in models of Alzheimer's disease

Alzheimer’s disease features extracellular amyloid-β plaques and intracellular neurofibrillary tangles primarily in neocortex and hippocampus. Although aging is the major risk factor, the causal link between brain aging and amyloid deposition is unclear. Myelin, produced by oligodendrocytes, supports axonal conduction and metabolism but turns over slowly and deteriorates structurally with age. The authors hypothesized that age-related myelin dysfunction is an upstream driver of Aβ deposition by promoting axonal distress and altering microglial responses, thereby linking aging-associated myelin decline to core AD pathology.

Prior macroscopic neuroimaging suggests cortical myelin damage occurs in preclinical AD, but causal microscopy-level evidence has been limited. Human transcriptomic studies identified myelin-related gene alterations in AD autopsy samples, likely reflecting downstream effects of established pathology. Single-cell RNA-seq in AD mouse models showed oligodendrocyte signature changes near plaques. These correlative data prompted experimental tests of whether myelin defects precede and drive amyloidosis rather than result from it.

- Human tissue analysis: Immunofluorescence on medial temporal lobe (transentorhinal cortex) from AD and non-AD autopsy samples to assess intracortical myelin (CNP, PLP), microglia (IBA1), and Aβ plaques (Me-04). - Mouse models: Two AD models (5xFAD overexpression; APP NL-G-F knock-in). Myelin dysfunction models: Cnp knockout (Cnp−/−) and Plp knockout (Plp−/−) with subtle myelin disintegration; forebrain-specific shiverer (Emx-Mbp, largely unmyelinated cortical axons). Crossbreeding to combine AD and myelin mutant genotypes. - Whole-brain amyloid load quantification: Congo red in toto staining and tissue clearing based on iDISCO, imaged by light-sheet microscopy (LSM) for unbiased 3D quantification of plaque burden across cortex, hippocampus, and alveus. - Acute demyelination paradigms: (1) Cuprizone feeding (4 weeks) followed by 4-week recovery in young adult 5xFAD mice; (2) Experimental autoimmune encephalomyelitis (EAE) induction via MOG peptide immunization in 5xFAD to generate demyelinating lesions (notably in spinal cord). Amyloid aggregates assessed with anti-Aβ antibodies and β-sheet dye methoxy-04 (Me-04). - Histology and EM: Immunostaining for APP, BACE1 (β-secretase), PSEN2 (γ-secretase component), various Aβ antibodies, and microglia markers; high-pressure freezing electron microscopy of optic nerve to characterize axonal swellings and endo-lysosomal accumulations. - Biochemistry: Western blots of cortex and white matter for BACE1 and APP processing (full-length APP and carboxy-terminal fragments, including phospho-C99), normalized to total protein or fAPP. - Microglia functional assays: In vitro phagocytosis assays using bone-marrow-derived macrophages pre-exposed to myelin debris to test amyloid uptake. - Transcriptomics: Microglia isolation by magnetic-activated cell sorting (MACS) followed by bulk RNA-seq; single-nucleus RNA-seq (10x Genomics) from brain hemispheres of WT, Cnp−/−, 5xFAD, and Cnp−/−;5xFAD to define microglial subpopulations (UMAP clustering, DEG analyses, GO enrichment). In situ hybridization for Cst7 to localize amyloid-DAM relative to plaques. - Behavioral testing: Y-maze and elevated plus maze to assess activity and anxiety-related behavior across genotypes. - Statistics: Primarily two-sided unpaired Student’s t-tests; sample sizes indicated per experiment (e.g., LSM quantifications with n=8 controls and n=7 mutants; immunostaining quantifications with n=3–5 per group).

- Human tissue: Reduced intracortical myelin density in AD medial temporal lobe not restricted to plaque vicinity; increased IBA1+ microglia at myelin loss sites. - Crossbred myelin mutants with AD models: Cnp−/−;5xFAD and Plp−/−;5xFAD mice exhibited markedly increased amyloid plaque load in cortex and especially hippocampal alveus at 6 months versus 5xFAD controls (3D LSM). Effects not present at 3 months, paralleling progressive myelin defects. - In alveus, Aβ appeared as numerous small aggregates suggestive of enhanced amyloid seeding. Control 5xFAD and single myelin mutants lacked alveus plaques at this age. - Findings replicated in APP NL-G-F knock-in crossed with Cnp−/−. - Acute demyelination: - Cuprizone-treated 5xFAD: Global copper chelation reduced compact plaque cores, but in strongly demyelinated tracts (alveus) there was a large increase in small amyloid aggregates detected by anti-Aβ immunostaining (n=4 control, n=5 cuprizone). - EAE in 5xFAD: No change in brain plaque load (little brain demyelination), but demyelinated spinal cord lesions showed small, atypical Me-04+ amyloid aggregates absent in controls (n=5 per group); WT EAE spinal cord lacked Me-04+ material. - Absence of myelin (forebrain shiverer;5xFAD): Strong protection from hippocampal and cortical amyloid at 3 months; effect largely lost by 6 months. Alveus plaque burden increased at 6 months but with different morphology/distribution versus Cnp/Plp models. Suggests healthy myelin initially inhibits plaque formation, but later demyelination promotes it. - APP metabolism and axonal pathology: - Myelin-damage-associated axonal swellings abundant in Cnp−/− and Cnp−/−;5xFAD, enriched in endosomal/lysosomal vesicles (primary sites for Aβ generation). - Swellings accumulated β- and γ-secretase components (BACE1, PSEN2), APP, and Aβ epitopes. Quantified increases in APP- and BACE1-positive swellings (e.g., APP swellings P=0.0001; BACE1 swellings P=0.0038), with variable PSEN2. - Western blots: Increased BACE1 (significant in white matter) and elevated cortical APP C-terminal fragments (CTFs) without change in full-length APP or α/β CTF ratio (e.g., cortical CTF increase P≈0.0456). - Microglia responses: - In vitro, prior exposure to myelin debris suppressed amyloid phagocytosis by macrophages. - In vivo, despite increased microglial numbers, plaque-corralling by microglia was lost in Cnp−/−;5xFAD and Plp−/−;5xFAD cortices; automated IBA1 coverage per plaque was reduced (P=0.0062; 2,017 plaques in 5xFAD and 2,190 in Cnp−/−;5xFAD analyzed; n=5 mice per group). - Bulk RNA-seq of MACS-isolated microglia: Progressive loss of homeostatic markers and increased activation/inflammatory genes from WT → 5xFAD → Cnp−/− → Cnp−/−;5xFAD. DAM genes (e.g., Clec7a, Gpnmb, Apoe, Spp1, Axl, Itgax; Ms4a cluster) upregulated, with strong lipid-metabolism gene induction (Apoe, Apoc1/4, Lpl). TREM2 and TYROBP induction not altered, and TREM2 protein/cleavage not changed. - APOE elevated in microglia and plaques of Cnp−/−;5xFAD; mean APOE intensity per plaque increased (P=0.0031). - snRNA-seq identified distinct microglial states: amyloid-DAM (mostly from 5xFAD) and myelin-DAM (from Cnp background). Myelin-DAM showed higher expression of lipid/metabolism genes (Apoe, Abca1, Apobec1) and Ms4a cluster; amyloid-DAM had Cst3, Ctnna3, Gpc5, and Cst7 enrichment. In Cnp−/−;5xFAD, both states were induced but microglia were frequently engaged in myelin clearance distal to plaques (Cst7+ cells located away from plaques), supporting a ‘distraction’ model that reduces plaque corralling and clearance. - Behavior: Myelin mutants were hyperactive; Cnp−/−;5xFAD showed supra-additive hyperactivity and reduced anxiety-like behavior in elevated plus maze, suggesting synergistic effects of myelin defects and amyloid pathology on disinhibition-related phenotypes. - Controls: Neither Plp−/− (14 months) nor conditional forebrain Plp knockout (22 months) developed amyloid without human APP transgene, consistent with rodent Aβ resistance to aggregation.

The study addresses whether myelin dysfunction is an upstream driver of amyloidosis. Combining AD mouse models with genetic and acute demyelination paradigms shows that myelin defects markedly enhance Aβ deposition, especially as small aggregates/seeds in white matter tracts (alveus), and increase cortical plaque burden. Mechanistically, myelin damage leads to axonal transport disturbances and swellings enriched in endosomal/lysosomal compartments where APP processing occurs, increasing local interactions between APP and secretases (BACE1/γ-secretase) and shifting cortical APP toward higher CTF levels, consistent with elevated Aβ production. Concurrently, microglia exhibit a DAM-like activation but are diverted toward clearing defective myelin (myelin-DAM), losing their plaque-corralling function around Aβ deposits; lipid metabolism genes and APOE are upregulated, potentially further promoting amyloid seeding. snRNA-seq delineates distinct yet related DAM states for amyloid and myelin injury, both induced in the combined pathology but spatially misallocated. Together, these findings provide a causal link between aging-associated myelin deterioration and increased amyloid deposition, aligning with the amyloid hypothesis and highlighting neuroinflammatory mechanisms, specifically microglial engagement with myelin, as modulators of plaque dynamics.

This work reframes oligodendrocyte/myelin pathology as an active, upstream contributor to amyloid-β deposition in AD models. Genetic myelin defects and acute demyelination increase amyloid load via (1) enhanced APP processing within axonal swellings and (2) microglial diversion toward myelin clearance at the expense of plaque corralling. Conversely, a near-complete absence of forebrain myelin initially delays plaque formation. The data suggest that maintaining oligodendrocyte health and myelin integrity could delay or slow AD progression. Future research should: (i) test therapeutics that promote myelin maintenance/remyelination or modulate microglial lipid handling; (ii) dissect the relative contributions of axonal transport defects versus microglial diversion; (iii) validate these mechanisms in human longitudinal cohorts and tissues; and (iv) investigate comorbidity and shared mechanisms between demyelinating diseases (e.g., MS) and AD.

- Species/model limitations: Mouse models overexpress or humanize APP and may not fully recapitulate human AD; rodent Aβ is resistant to aggregation, limiting generalizability. - Confounding factors: Cuprizone’s copper-chelating effect can reduce compact plaque cores, complicating interpretation of global plaque changes; analyses focused on demyelinated tracts mitigate but do not eliminate this confound. - Causality among downstream events: Axonal distress and microglial activation are interlinked consequences of myelin injury and difficult to isolate experimentally. - Spatial constraints: Some primary demyelination models had extensive microglia/macrophage infiltration precluding equivalent cortical analyses. - Human data: Human findings are correlative; direct causal validation in humans is pending. Epidemiological indications of MS–AD comorbidity are preliminary and require larger studies.