For a decade, biomolecular condensates have been modeled as liquid droplets — amorphous blobs formed by liquid-liquid phase separation, where function emerges from concentration effects. Proteins and RNA partition into dense phases, reactions accelerate, and the thermodynamics of mixing explains the rest.
Lasker and colleagues (Nature Structural & Molecular Biology 2026) used cryo-electron tomography to look inside a condensate. It was not a liquid.
PopZ condensates — essential for bacterial cell division and DNA segregation — contain an intricate network of protein filaments. The droplet has internal architecture. When the filaments are engineered away, the condensate still forms (it phase-separates normally), but the cells stop dividing and fail to segregate their chromosomes. The liquid is present. The function is not.
This splits what the field had merged. Phase separation produces the compartment. The filament scaffold produces the function. These are different things. Treating the compartment as the explanation was like treating the building as the company — the walls create the space, but the work happens on the infrastructure inside.
The wrong-physics error is specific: liquid thermodynamics predicts properties like viscosity, surface tension, and mixing kinetics. Polymer network mechanics predicts properties like elasticity, strain response, and load bearing. The condensates look liquid from outside. They are structural from inside. The measurement method — light microscopy, fluorescence recovery — reports liquid-like behavior because it averages over the filament network. Cryo-ET, which freezes and sections, reveals the architecture.
The implications extend beyond bacteria. Analogous filamentous condensates regulate protein clearance in human neurons (failure mode: ALS) and growth suppression in tumor cells (failure mode: cancer). If the function lives in the scaffold rather than the phase, then disrupting phase separation is the wrong therapeutic target. The drug needs to reach the filaments.