Supporting anti‐tau antibody trials using computer simulation of spatial tau propagation

Abstract

AbstractBackgroundAlthough tau pathology is a major driver of AD pathology, the mere complexity of tau biology significantly hampers the identification and validation of the most impactful target. Recent clinical trials using antibodies against various forms of tau have come up short in improving cognitive readout.MethodA Quantitative Systems Pharmacology computer platform of tau biology has been developed including tau species diversity, neuronal activity‐dependent tau secretion, tau binding to membrane surface receptors, subsequent internalization at the synaptic cleft and along the axonal membrane following a diffusion gradient, slow motor‐mediated axonal transport, intracellular degradation and templating of monomeric tau by seed‐competent tau to form higher‐order aggregates. Polysynaptic progression is implemented by modeling secretion from the first synapse and uptake in the second axon.ResultThe platform was calibrated to reproduce preclinical experiments from in‐vitro cellular assays and after injection of brain extract in Tg models. The model is humanized using published data on seed‐competent tau from postmortem AD brain at different Braak stages and spatio‐temporal progression over neuronal projections. By exploring the effect of target mediated drug disposal of antibodies, we show that depending upon the disease state only between 15 and 30% of misfolded tau protein can be captured by antibodies currently in clinical trials. Introducing PSP specific uptake parameters from preclinical experiments, the simulations suggest an even lower efficacy due to the higher affinity of PSP‐tau for membrane surface receptors. We further identify conditions under which specificity (i.e. monomeric vs misfolded tau) of antibodies outweigths sensitivity (affinity). Finally, the simulations suggest that reduced oligomerization and increased degradation have a greater impact on the accumulation of seed‐competent tau in distal regions, likely because they act during the long intracellular residence of misfolded tau protein.ConclusionAdvanced Quantitative Systems Pharmacology computer modeling combining drug exposure with mechanistic modeling allows to ‘humanize’ preclinical experiments and aims to support tauopathy R&D projects.