Many large molecules in cells can separate into distinct phases, forming biomolecular condensates that organize key biochemical reactions. While these condensates are normally dynamic and liquid-like, under certain conditions they can undergo a transition to a more solid-like state, a process increasingly implicated in neurodegenerative diseases such as ALS and frontotemporal dementia. Yet the mechanisms driving this pathological transformation in otherwise healthy condensates remain poorly understood. In our research, we combined microfluidic, optical techniques and computational simulation to study this transition in the disease-relevant protein FUS. We discovered that FUS condensates do not solidify uniformly; instead, they exhibit coexisting liquid and solid regions, producing structural heterogeneity. Strikingly, this change originates at the condensate boundary and progresses inward, a feature that may underlie the initial nucleation of harmful aggregates in cells. Furthermore, we found that nucleic acids significantly modulate the phase behavior, echoing the complex cellular environment where DNA/RNA are abundant. To capture these local dynamics, we developed spatial dynamic mapping microscopy, which revealed that the liquid-to-solid transition begins specifically at the interface between dense and dilute regions. Importantly, through another system we also demonstrated that LLPS can act as an alternative pathway to fibril formation, highlighting the dual role of condensates in either suppressing or redirecting aggregation. Together, these findings emphasize not only how the location and timing of condensate solidification are critical, but also how edge-initiated transitions may seed pathological protein clumps that drive neurodegeneration.
Shen et al. (2026) studied this question.