Intracellular droplets suppress fluctuations in protein concentrations

Every cell is different, even those that have identical DNA. Natural variability in gene expression and protein turnover rates leads to differences in protein concentration, both between cells and over time within a single cell. But if concentrations become too large or too small, processes such as biosynthesis and cell replication are impeded.

Scientists have hypothesized that cells might use membraneless droplets formed through liquid–liquid phase separation (LLPS) to keep protein concentrations at just the right levels. It works in equilibrium systems: If a solute, such as a protein, is evenly distributed in a fluid, adding or removing it changes the overall concentration. But when it phase-separates into solute-rich droplets and a dilute background, the concentrations in those two phases are fixed by thermodynamic constraints. The size and number of droplets—not the concentrations—vary to accommodate fluctuations in the amount of protein present.

Cells operate away from equilibrium, so it’s not a given that intracellular LLPS should reliably regulate protein concentrations. However, that’s what Adam Klosin, Florian Oltsch, and coworkers at the Max Planck Institute of Molecular Cell Biology and Genetics in Dresden, Germany, have now found. Their work confirms that LLPS in living cells can minimize fluctuations in protein concentration, particularly outside the droplets where other cellular processes occur.

The researchers first developed a model that combines stochastic protein production and turnover with protein phase separation, and then they characterized fluctuations in the protein’s concentration through noise strength—the ratio of the concentration’s variance to its squared mean. The model predicted that LLPS, which occurs above a threshold protein concentration, would substantially reduce noise outside the droplets.

To test their model, the researchers engineered a protein (2NT-DDX4^YFP) that phase separates in vitro and expressed it in living HeLa cells; the above image of the fluorescently labeled protein shows its expression and subsequent droplet formation. Fluorescence images of more than 10 000 cells revealed that when the concentration exceeded the system’s threshold, the noise in the dilute background fell by nearly a factor of 10, as shown in the graph. The drop is not as deep as the theory predicts, which could be due to factors not captured by the simple model.

Now that they have a starting point, the researchers plan to extend their theory to more complex systems, such as multicomponent phase-separating liquids. (A. Klosin et al., Science 367, 464, 2020 .)