Single-cell and spatial transcriptomic insights into neural substrates of Behavioural dysfunction in Alzheimer's disease

The transition from bulk-tissue profiling to single-cell and spatially resolved transcriptomics has reshaped our understanding of Alzheimer's disease (AD) from a passive proteinopathy to a coordinated multicellular process spanning the neurovascular unit. However, translating these molecular catalogues into disease-modifying therapies requires a framework that links cellular transitions to neural circuit dysfunction and behavioural abnormalities. This review synthesizes the multicellular landscape of AD, tracing how cell-state-specific remodelling drives network collapse and cognitive decline. Neuronally, selective vulnerability targets projection networks and subclass-specific inhibitory interneurons. Their transcriptional silencing compromises projection pathways, including the entorhinal-hippocampal perforant path, disrupts cortical microcircuits, and underlies episodic memory impairment, hyperexcitability, and executive control failure. This neuronal degeneration is dynamically coupled to reactive glial and vascular remodelling across distinct spatial niches. We examine the non-linear transition of homeostatic astrocytes and microglia into specialized, plaque-associated reactive sub-states, as well as paracrine signalling decay at the neurovascular interface. Compromised capillary tight junctions and the shedding of pericytic survival receptors are proposed to promote blood-brain barrier disintegrity, triggering a feed-forward neuroinflammatory cascade of complement-mediated synaptic pruning. Crucially, we establish a three-tiered hierarchy of evidence, rigorously distinguishing cross-sectional, associative human postmortem findings from inferred functional trajectories in human cellular platforms, and causal, perturbation-based validations in transgenic rodent models. We review how contemporary multi-scale integration-combining single-nucleus transcriptomics with functional electrophysiological platforms and biophysical connectomic modelling-bridges the gap between terminal postmortem signatures and living brain dynamics. Finally, we outline a coordinated translational roadmap illustrating how spatially resolved cell-state networks can guide the development of fluid and imaging biomarkers, support biologically informed patient stratification, enrich clinical trials, and accelerate the development of circuit-restoring precision therapies, thereby providing a mechanistic framework that links cell-state transitions to neural circuit dysfunction, behavioural impairment, and precision medicine in Alzheimer's disease.