The Secretory Shunt in Tauopathy: Proteomics-validated biomarker candidates and stage-aware secretory dynamics

Tauopathies including Alzheimer's disease (AD) and progressive supranuclear palsy (PSP) lack effective disease-modifying treatments. This report tests a "Secretory Shunt" framework: when autophagy becomes constrained, cells recruit vesicular export that offloads proteotoxic burden at the cost of altered propagation. We integrate proteomics and transcriptomics to identify stage-linked biomarker candidates and specify falsifiable wet-lab experiments. In Mount Sinai Brain Bank (GSE84422; n=1,053 with Braak staging), a locked 7-gene composite correlates with Braak stage (r=0.169, p=3.43e-08; modest effect size consistent with bulk tissue). CAT increases and VAMP2 decreases with staging; VAMP2's decline survives neuron-proxy adjustment. Proteomics context—LAMP2–MAPT decoupling in PSP CSF (p≈2.7e-07) and ATN-staged EV marker shifts—supports a routing/fusion-capacity narrative. We provide a staging-aware biomarker panel and ARF6 perturbation plan for wet-lab validation. Key Findings (v5) Human staging score: 7-gene composite correlates with Braak (r=0.169); weighted and unweighted variants consistent Gene-level: CAT↑ and VAMP2↓ with Braak; VAMP2 association survives neuron-proxy adjustment Proteomics context: LAMP2–MAPT decoupling and ATN-linked EV shifts support routing hypothesis (interpreted conservatively) Reproducibility: Scripts regenerate all artifacts from included GEO/PRIDE data Datasets Transcriptomics (GEO): GSE84422 (MSBB): bulk microarray, n=2,004; Braak subset n=1,053 GSE62935 (acute neuronal model), GSE56452 (chronic comparator) GSE174367 (snRNA-seq replication), GSE255902 (neuron seeding proxy) Proteomics (PRIDE): PXD041417, PXD045736, PXD058777, PXD047282, PXD029212 Contents Manuscripts: secretory_shunt_master.pdf and secretory_shunt_v5.pdf (+ Markdown/LaTeX sources) Scripts: Full pipeline in scripts/, orchestration via run_analysis.py Figures/Results: Publication figures, generated tables/plots Synthesis: Wet-lab plans and falsifiability grid Reproducibility python -m pip install -r requirements.txt python run_analysis.py --skip-perturbation Full pipeline (includes perturbation scan): python run_analysis.py Limitations Bulk tissue heterogeneity; modest effect sizes expected Repeated measures may inflate significance; effect directions emphasized Cross-model comparisons involve species/cell-type differences (treated as context) All results hypothesis-generating; wet-lab validation required Research use only; not medical advice IP Notice U.S. Provisional Patent Application No. 63/964,355 (filed Jan 20, 2026): "Proteomics Validated Biomarkers and ARF6 Therapeutic Methods for Treating Tauopathies" License: CC BY 4.0Correspondence: aaron@lumenais.com