Stacked binding of a PET ligand to Alzheimer's tau paired helical filaments

Abstract

Accumulation of filamentous aggregates of tau protein in the brain is a pathological hallmark of Alzheimer's disease (AD) and many other neurodegenerative tauopathies. The filaments adopt disease-specific cross-β amyloid conformations that self-propagate and are implicated in neuronal loss. Development of molecular diagnostics and therapeutics is of critical importance. However, mechanisms of small molecule binding to the amyloid core is poorly understood. We used cryo-electron microscopy to determine a 2.7 Å structure of AD patient-derived tau paired-helical filaments bound to the PET ligand GTP-1. The compound is bound stoichiometrically at a single site along an exposed cleft of each protofilament in a stacked arrangement matching the fibril symmetry. Multiscale modeling reveals pi-pi aromatic interactions that pair favorably with the small molecule-protein contacts, supporting high specificity and affinity for the AD tau conformation. This binding mode offers critical insight into designing compounds to target different amyloid folds found across neurodegenerative diseases.

Fig. 1. Cryo-EM map of AD tau PHF with density for bound GTP-1.

a Chemical structure of GTP-1. b X-Y-slice view of the cryo-EM map of AD PHFs incubated with GTP-1. Extra density corresponding to GTP-1 is indicated by white triangles. c Cryo-EM map of tau PHF:GTP-1. The density corresponding to GTP-1 is colored in green. d Difference map (salmon density) between (C) and a previously determined apo-AD PHF map (EMDB: 0259), low-pass filtered to 3.5 Å. The density for the apo PHF protofilament (grey) is shown as a reference. e Side view of tau PHF:GTP-1 structure showing the ligand density (green) in a stacked arrangement with one molecule spanning across multiple rungs of the tau protofilament.

Fig. 2. Atomic model of tau PHF and bound GTP-1.

a Refined tau PHF atomic model fit into the PHF:GTP-1 density. b Map and model of the GTP-1 binding site with GTP-1 modeled into the density using a combination of molecule mechanics and density functional theory (DFT) approaches. c Side view of tau PHF:GTP-1 model, showing individual GTP-1 molecules fit at an angle relative to the backbone and making contact across 3 rungs of tau. d GTP-1 electrostatic (Coulomb) potential surface representation showing complementarity to the GTP-1 binding pocket. e Close contacts (<3.5 Å) of GTP-1 with sites in the binding pocket.

Fig. 3. Favorable ligand-ligand interactions support stacked arrangement.

a Comparison of the structure of GTP-1 monomer from an unconstrained DFT optimization (yellow) with the final modeled structure optimized in the context of amyloid-imposed constraints (coral). b Energy decomposition of the GTP-1 stacking interaction in a dimer using an HFLD calculation. c Illustration of the stacked GTP-1 interactions demonstrating the slipped nature of the stack, the retention of the amyloid displacement vector, and the distance of the pi-pi interactions. d Abstracted depiction of how the crossing angle between the plane of the amyloid backbone and the plane of the heterocycle is determined by the amyloid displacement vector and the optimal dimer interaction distance. e The RMSD of the GTP-1 heavy atoms throughout the 100 ns MD simulation, showing the stability of the GTP-1 binding pose. f Representative final frames of a 100-ns MD simulation of tau PHF:GTP-1 (left) and unliganded tau PHF (right) demonstrate both the stability of the GTP-1 binding pose and the complete occlusion of water from the GTP-1 binding site throughout the trajectory.

Fig. 4. Comparison of the GTP-1 binding pocket residues in other tau filament structures.

a The ligand binding pocket of GTP-1 is highlighted in gold, and the specific residues forming the binding pocket (Gln351, Lys353, Asp358, and Ile360) are shown. This binding pocket is unique to AD filaments compared to existing filament structures, thus indicating GTP-1 binding may be specific to the AD conformation. b Residues 351–360 in AD (purple) and CTE (pink) filament structures, with the rotamer of Lys353 matching that in our AD + GTP-1 structure. The change in concavity of the pocket would move Ile360 away from GTP-1 and would prevent a productive apolar interaction with C7 on the GTP-1 heterocycle.