Spatial Proteomics Maps Alzheimer's-Specific Microglial States in Human Brain

Microglia, the brain's resident immune cells, are increasingly recognized as central players in Alzheimer's disease (AD), yet their precise states in human tissue have remained poorly characterized. Prior work relied heavily on rodent models, low-plex immunohistochemistry, or transcriptomics that sacrifice spatial context. This study addresses that gap with a high-dimensional, spatially resolved approach applied directly to post-mortem human brain tissue.

The research team developed a 38–40 plex Multiplexed Ion Beam Imaging (MIBI) panel targeting neuronal structures, astrocytes, vasculature, metabolic and oxidative stress markers, AD hallmark proteins (amyloid-β, tau), and 17 microglial proteins including Iba1, P2RY12, TREM2, HLA-DR, CD33, CD44, and ApoE. They imaged five brain regions—hippocampus (HIP), cerebellum, substantia nigra (SN), caudate nucleus, and middle frontal gyrus (MFG)—from a cognitively normal donor, segmenting 93,679 bona fide microglia. Rather than falling into discrete clusters, microglia arranged along a continuous 'microglial state continuum' (MSC) from low to high immune activation, with distinct anatomical skewing: MFG and caudate trended low, cerebellum was mid-range, and HIP and SN showed higher activation profiles.

Pixel-level clustering defined 20 tissue microenvironments (synaptic fields, white matter, vasculature, astrocyte bodies, etc.), and low-MSC microglia preferentially localized near synapse-dense zones while high-MSC cells clustered near myelin tracts or vasculature. The continuum was reproducible across nine additional cognitively normal older adults (17,455 microglia in paired CA1 and caudate), with CA1 consistently showing higher P2RY12 and HLA-DR and caudate showing higher ApoE. Orthogonal validation using single-nucleus ATAC-seq (snATAC-seq) linked the proteomic MSC to an epigenetic axis spanning synapse-supporting to immune-effector gene programs.

Critically, 24,266 microglia from 12 AD cases matched to the same brain regions revealed a disease-specific shift: AD microglia were enriched at the high end of the MSC, with elevated CD33 and CD44 and markedly reduced HLA-DR, P2RY12, and ApoE expression—particularly in hippocampal CA1. This pattern suggests that while AD microglia appear more 'activated,' they simultaneously lose key homeostatic and antigen-presentation functions, potentially reflecting a dysfunctional rather than simply hyperactive state. Reductions in engulfed synaptic protein markers were also observed, pointing to impaired synaptic pruning capacity.

The study provides the most spatially and proteomically comprehensive single-cell characterization of human microglia in AD to date, offering a quantitative framework that moves beyond binary M1/M2 categories. Identified proteins such as CD33 and CD44—both druggable targets—emerge as potential therapeutic entry points. Limitations include the cross-sectional, post-mortem nature of the tissue, relatively small cohort sizes for the disease comparison, and the inability to directly infer causality from proteomic snapshots.