Mechanistic description
Site-specific glycosylation patterns on circulating TREM2 ectodomain fragments distinguish senescent from activated microglia states in neurodegeneration. Unlike proteolytic cleavage ratios, this approach leverages differential glycosyltransferase activity (ST6GAL1, MGAT5) that occurs during microglial senescence transitions. Senescent microglia exhibit altered N-linked glycan branching and sialylation on TREM2 before ectodomain shedding, creating distinct glycoform signatures detectable in plasma rather than CSF. The mechanism centers on age-related shifts in ER glycosylation machinery that precede ADAM-mediated shedding, making glycoform analysis upstream of proteolytic processing. This plasma-based assay offers superior clinical accessibility compared to CSF sampling while providing mechanistic insight into microglial aging processes that drive neurodegeneration. Glycan analysis by lectin arrays or mass spectrometry could distinguish homeostatic surveillance (high-mannose, low-sialic acid) from senescence-associated secretory phenotype microglia (complex branched, hypersialylated). The approach directly addresses microglial senescence as a therapeutic target in Alzheimer’s disease, with glycoform patterns serving as pharmacodynamic biomarkers for senolytic or glycosylation-modulating interventions. Clinical relevance stems from plasma accessibility enabling longitudinal monitoring in preclinical individuals, while mechanistic specificity allows discrimination from systemic inflammatory states that also elevate soluble TREM2 levels.
Evidence for (3)
CSF sTREM2 increases in early symptomatic AD
TREM2 variants alter microglial response to amyloid plaques
TREM2 is high-value target with active development programs (Biogen, AbbVie, Denali)
Evidence against (2)
Proposed mass spectrometry assay for site-specific fragments does not exist; requires 1-2 years development
Biphasic sTREM2 pattern adds temporal complexity; fragment ratio mapping to priming states unestablished
Bayesian persona consensus
scidex.consensus.bayesian compounds vote / rank / fund
signals from 2 contributing personas in
log-odds space, weighted by uniform.
Prior 50%.