Biomolecular condensates form by liquid-liquid phase separation — proteins and nucleic acids concentrate into droplets without membranes, like oil in water. The field modeled these droplets as disordered liquids: molecules diffuse freely inside, mixing and remixing. The boundary was the interesting part. The interior was assumed to be homogeneous.
Researchers at Scripps (Nature Structural & Molecular Biology 2026) used cryo-electron tomography to image the inside of bacterial condensates formed by the protein PopZ. They found filaments. Not diffuse protein soup but an intricate network of thread-like structures assembled through a stepwise process. Single-molecule FRET showed that PopZ adopts distinct conformations inside versus outside the condensate — the protein doesn't just concentrate, it reorganizes. The condensate's physical properties depend on the filament scaffold, not on the liquid phase alone.
In human cells, similar filament-based condensates perform two functions: clearing damaged proteins (failure leads to ALS-type accumulation) and suppressing uncontrolled growth (failure leads to cancers of the prostate, breast, and endometrium). The filament architecture is required for both. Disrupting the scaffold disrupts function, even if the condensate still forms as a droplet. The boundary can survive when the interior fails.
The general principle: correctly identifying the mechanism that creates a boundary — phase separation, in this case — does not determine what happens inside that boundary. A system can form by one set of rules and organize its interior by another. The droplet model captured the thermodynamics of formation but missed the structural biology of function. When a model explains how something comes into existence, the temptation is to assume it also explains how the thing works. Formation and function can be governed by different physics, and the interior can be invisible to any method that only sees the boundary.