APOE4 Triggers Brain Blood Vessel Scarring — and TGF-β Inhibition May Reverse It

APOE4 is the strongest common genetic risk factor for late-onset Alzheimer's disease. Carriers show blood vessel damage, blood-brain barrier breakdown, and vascular amyloid buildup, even before classic Alzheimer's pathology appears. Anti-amyloid antibodies also carry higher risks of brain swelling and bleeding (ARIA) in carriers, so therapies that protect the vasculature are urgently needed. This Cell study asks how APOE4 actually damages brain vessels, and identifies a specific cellular transition and a druggable signal behind it.

The team integrated three published single-nucleus RNA-sequencing datasets of postmortem human brain into a vascular atlas of roughly 66,000 nuclei from 220 individuals. They used propensity score matching for age, sex, AD status, and cell yield. Differential abundance analysis (MiloR) found 250 cell neighborhoods that differed significantly between APOE4 carriers and APOE3/3 individuals (spatial FDR < 0.05), with fewer pericytes and more smooth muscle-like cells in carriers. NG2 staining of human hippocampus confirmed reduced pericyte coverage. αSMA staining showed positive cells scattered through the tissue rather than confined to large vessels.

Subclustering showed that one smooth muscle population, SMC_2, had 61 dysregulated pathways, mostly contraction and extracellular matrix production. Cells co-expressing both programs resemble myofibroblasts, the fibrosis-driving cells known from peripheral organs, and unbiased annotation named myofibroblasts as the top match. In a covariate-adjusted model, APOE4 was the strongest predictor of myofibroblast abundance, independent of AD status, sex, and dataset, and leave-one-dataset-out checks held. The same phenotype appeared in aged APOE4/4 knock-in mice and in isogenic iPSC-derived 3D human "miBrain" tissue: less pericyte coverage and more nonvascular αSMA and Tagln.

Mixing experiments in miBrains showed that APOE4 mural cells are both necessary and sufficient. Swapping APOE3/3 mural cells into APOE4/4 miBrains restored pericyte coverage. Adding APOE4/4 mural cells to APOE3/3 miBrains reproduced the phenotype. Pseudotime analysis revealed a pericyte-to-myofibroblast transition through an intermediate CSPG4+/ACTA2+ state, which was more common in APOE4 carriers, including those without AD. In APOE-TR mice, pericyte signatures declined and myofibroblast signatures rose at 12 and 24 months but not at 3 months. APOE4 iPSC mural cells shifted toward myofibroblast states by passage 5.

Myofibroblasts were the largest source of fibronectin (FN1) in the atlas, exceeding perivascular fibroblasts. APOE4 mice and miBrains showed more vascular fibronectin, collagen I and III, and amyloid. In human data, mural cell FN1 expression was higher in people with greater neuropathological burden and inversely correlated with age at AD diagnosis. FN1 knockdown in APOE4 mural cells reduced vascular amyloid, and fibronectin increased amyloid accumulation in endothelial cells. NicheNet predicted TGF-β as the top upstream regulator, and pSMAD2 staining was elevated in APOE4 miBrains and mice. The authors report that TGF-β inhibition restored pericyte coverage and reduced fibrosis and amyloid to APOE3 levels in iPSC-derived tissue and mice.

This work offers a mechanistic link between APOE4 and vascular fibrosis, and a candidate target that could complement anti-amyloid therapy. Caveats are substantial. The human data are cross-sectional postmortem transcriptomics, trajectories are inferred computationally, and the interventions were tested only in iPSC models and mice. TGF-β has broad roles in immunity and tissue repair, so the safety of inhibiting it in aging humans is untested.