The assumption that microglial activity in Alzheimer's disease is either harmful or protective may be fundamentally incomplete. Emerging evidence points to a third possibility: that the Alzheimer's brain actively generates its own suppressive immune programs near amyloid plaques — a finding with significant implications for how therapeutics targeting neuroinflammation should be designed.
This review, published in Cells, synthesizes genetic, single-cell transcriptomic, and spatial biology data to advance a specific mechanistic hypothesis: that partial reduction of PU.1 — the myeloid transcription factor encoded by SPI1 — near amyloid plaques enables a distinct microglial subpopulation characterized by CD28 surface expression. CD28 is canonically associated with T-cell co-stimulation in adaptive immunity, making its appearance on microglia conceptually striking. The proposed PU.1-low/CD28-positive microglial state appears to operate as an endogenous immunoregulatory checkpoint, potentially restraining both neuroinflammation and amyloid burden rather than amplifying them. Human genetic data suggesting that lower SPI1 expression correlates with reduced Alzheimer's risk lends epidemiological weight to this axis, though the review also notes that excessive PU.1 suppression impairs core microglial housekeeping functions, highlighting a narrow therapeutic window.
This framework builds on the well-established disease-associated microglia (DAM) literature but diverges from the dominant TREM2-APOE signaling narrative by foregrounding a lymphoid-mimicry program. The concept that brain-resident innate immune cells can adopt adaptive immune regulatory signatures — analogous to regulatory T-cells — is relatively novel in neurodegeneration research and, if confirmed causally, could reframe therapeutic strategies. Currently, most anti-neuroinflammatory approaches aim at broad microglial suppression; a PU.1/CD28-informed strategy might instead seek to selectively amplify this endogenous suppressive subpopulation. The primary limitation here is that this remains a review synthesizing correlative and experimental data rather than a primary causal human study. Replication in large longitudinal cohorts with single-cell resolution, combined with functional perturbation experiments, will be essential before clinical translation is warranted. Still, as a conceptual contribution, this represents a genuinely paradigm-adjacent hypothesis in Alzheimer's neuroimmunology.