In our 2008 paper (Shankar, G. M. et al. Nat. Med. 14, 837–842; 2008), we moved beyond cellular and animal models to isolate pathogenic forms of amyloid-β directly from human brains and apply them to reproduce key features of AD: synaptic dysfunction, loss of dendritic spines, and impaired memory per se. We identified stable dimers of amyloid-β in Alzheimer’s cerebral cortex and their dose-dependent impairment of synaptic plasticity. Separating dimers from other amyloid-β assemblies in AD brain extracts enabled us to attribute synaptic deficits directly to these smallest building blocks of oligomers and fibrils.

Perhaps most relevant to the human condition, we found that microinjection of dimer-rich isolates from the brains of deceased patients into the cerebral ventricles of healthy adult rats decreased the animals’ ability to remember a learned behavior (a passive avoidance task). The timing of this amyloid-β-induced failure of recall mirrored the known temporal pattern of transcriptional regulation of synapse remodeling that accompanies such learning.