BackgroundLate-onset Alzheimer’s disease (LOAD) is the most common form of dementia; it disproportionally affects women in terms of both incidence rates and severity of progression. The cellular and molecular mechanisms underlying this clinical phenomenon remain elusive and ill-defined.MethodsIn-depth analyses were performed with multiple human LOAD single-nucleus transcriptome datasets to thoroughly characterize cell populations in the cerebral cortex. ROSMAP bulk human brain tissue transcriptome and DNA methylome datasets were also included for validation. Detailed assessments of microglial cell subpopulations and their relevance to sex-biased changes at the tissue level were performed. Clinical trait associations, cell evolutionary trajectories, and transcription regulon analyses were conducted.ResultsThe relative numbers of functionally defective microglia were aberrantly increased uniquely among affected females. Substratification of the microglia into different subtypes according to their transcriptomic signatures identified a group of female-enriched and disease-associated microglia (FDAMic), the numbers of which were positively associated with disease severity. Phenotypically, these cells exhibit transcriptomic signatures that support active proliferation, MHC class II autoantigen presentation and amyloid-β binding, but they are also likely defective in phagocytosis. FDAMic are likely evolved from female activated response microglia (ARMic) with an APOE4 background and compromised estrogen receptor (ER) signaling that is deemed to be active among most subtypes of microglia.ConclusionThis study offered important insights at both the cellular and molecular levels into how ER signaling affects microglial heterogeneity and function. FDAMic are associated with more advanced pathologies and severe trends of cognitive decline. Their emergence could, at least in part, explain the phenomenon of greater penetrance of the APOE4 genotype found in females. The biases of FDAMic emergence toward female sex and APOE4 status may also explain why hormone replacement therapy is more effective in APOE4 carriers. The pathologic nature of FDAMic suggests that selective modulations of these cells may help to regain brain neuroimmune homeostasis, serving as a new target for future drug development.