Type 2 diabetes mellitus approximately doubles the risk of late-onset Alzheimer's disease (Ott et al., 1999), yet a molecular mechanism connecting peripheral metabolic dysfunction to central tau pathology has lacked a molecularly specific, anatomically localized account. Six independent research programs studying dietary fat physiology, neuropeptide signaling, adipokine-driven neuroinflammation, tanycyte gating biology, chemokine neuroimmune crosstalk, and stress kinase signaling have converged on a common molecular node: impaired IRS-1/PI3K coupling, primarily through inhibitory phosphorylation of insulin receptor substrate-1 (IRS-1) at serine 312 (human; serine 307 in rat IRS-1), demonstrated in hypothalamic arcuate nucleus neurons and proposed to reach cortical and hippocampal neurons through the conserved inflammatory kinases JNK and IKKbeta, which are broadly expressed in those regions. IRS-1 Ser312/Ser307 phosphorylation blocks PI3K/AKT signaling, disinhibiting glycogen synthase kinase-3 beta (GSK-3beta), which hyperphosphorylates tau. IRS-1 serine phosphorylation is specifically elevated across Alzheimer's disease, Pick's disease, corticobasal degeneration, and progressive supranuclear palsy in 157 human autopsy brains, while remaining normal in alpha-synucleinopathies and TDP-43 proteinopathies. Neuronal-enriched extracellular vesicles from plasma in the Baltimore Longitudinal Study of Aging show elevated pSer312-IRS-1 about 3.5 years before clinical AD onset in cognitively normal participants (Kapogiannis et al., 2019). Functional MRI demonstrates that type 2 diabetic patients show the predicted hypothalamic signaling failure, which normalizes within four days of dietary intervention (Teeuwisse et al., 2012). As mechanistic proof of concept, intranasal insulin augments hypothalamic BOLD responses to glucose in healthy subjects. A common IRS-1 polymorphism (Gly972Arg) produces genetically determined cerebrocortical insulin resistance in metabolically matched human subjects. We present this convergence as a mechanistic hypothesis with eight falsifiable predictions.
Arthur Stewart (Tue,) studied this question.