Arzeda (United States)

Mixture-of-Experts (MoE) showcases tremendous potential to scale large language models (LLMs) with enhanced performance and reduced computational complexity. However, its sparsely activated architecture shifts feed-forward networks (FFNs) from being compute-intensive to memory-intensive during inference, leading to substantially lower GPU utilization and increased operational costs.

Transthyretin (TTR) is a natively tetrameric thyroxine transporter in blood and cerebrospinal fluid whose misfolding and aggregation causes TTR amyloidosis. A rational drug design campaign identified the small molecule tafamidis (Vyndamax) as a stabilizer of the native TTR fold, and this aggregation inhibitor is regulatory agency approved for the treatment of TTR amyloidosis. Here we used cryo-EM to investigate the conformational landscape of this 55 kDa tetramer in the absence and presence of one or two ligands, revealing inherent asymmetries in the tetrameric architecture and previously unobserved conformational states. These findings provide critical mechanistic insights into negatively cooperative ligand binding and the structural pathways responsible for TTR amyloidogenesis, underscoring the capacity of cryo-EM to identify pharmacological targets suppressed by the confines of the crystal lattice, opening uncharted territory in structure-based drug design.