Liquid-liquid phase separation (LLPS) is the process by which a uniform mixture of molecules spontaneously splits into two distinct liquid phases: a dense phase enriched with certain molecules and a surrounding dilute phase depleted of them. Think of it like oil separating from vinegar in a salad dressing. In cells, this process creates concentrated droplets of proteins and RNA that function as tiny compartments, organizing biochemistry without needing a membrane to hold everything in place.
How Phase Separation Works
At its core, LLPS is governed by thermodynamics. Every mixture has a balance between two competing forces: the tendency of molecules to mix evenly (driven by entropy, or disorder) and the tendency of certain molecules to stick together or repel each other (driven by the energy of their interactions). When conditions tip the balance, such as a change in temperature, concentration, or salt levels, the mixture becomes unstable and splits into two phases.
A phase diagram maps out where this happens. The key boundary is called the binodal curve, which marks the concentrations and temperatures where separation begins. Inside this boundary sits a second line called the spinodal curve. Between the two curves, the system is metastable: separation can happen, but it needs a little push, like a seed crystal forming first. This is called nucleation. Inside the spinodal curve, the system is fully unstable and separation happens spontaneously, with droplets appearing everywhere at once in a process called spinodal decomposition.
The result is always the same: dense droplets enriched with specific molecules, surrounded by a dilute background. The concentrations of the two phases are set by thermodynamics, not by how much total material you started with. Add more of the separating molecule, and you get more droplets rather than denser ones.
What Drives It Inside Cells
In biology, LLPS is driven primarily by proteins and RNA forming weak, reversible contacts with each other. The critical ingredient is multivalency: each molecule has many small interaction sites that can grab onto partners simultaneously, creating a web-like network that condenses into a droplet.
Many of the proteins involved contain intrinsically disordered regions (IDRs), stretches of amino acids that don’t fold into a fixed 3D shape. Instead, they remain flexible, constantly shifting between conformations. This flexibility lets them make many simultaneous contacts with neighboring molecules. Researchers describe IDR-driven phase separation using a “stickers and spacers” framework: certain amino acids or short motifs act as sticky patches that drive the interactions, connected by non-sticky spacer regions. Increasing the number of sticky patches, or making each one stickier, lowers the concentration needed to trigger separation and widens the range of conditions where droplets form.
Not all disordered regions drive phase separation equally. The specific sequence matters. Proteins like FUS, TDP-43, and the DDX4 RNA helicase contain IDRs that are potent drivers of condensation, while other disordered proteins barely phase-separate at all. The pattern and chemistry of the sticky residues, not just the overall disorder, determines the behavior.
Membraneless Organelles in the Cell
Cells are full of structures built by LLPS. These are called membraneless organelles or biomolecular condensates, and they perform specialized jobs without the lipid membranes that define traditional compartments like the nucleus or mitochondria.
In the nucleus, examples include the nucleolus (where ribosomal RNA is made), Cajal bodies (which help assemble the machinery for gene splicing), nuclear speckles, paraspeckles, and PML bodies. In the cytoplasm, stress granules form when cells are under duress, sequestering messenger RNA to pause protein production. P-bodies help degrade or store unused RNA. Germ granules in reproductive cells package RNA for the next generation. Even structures like centrosomes, which organize cell division, and signaling clusters on cell membranes show liquid-like properties consistent with phase separation.
These condensates share telltale liquid characteristics. They’re roughly spherical, they fuse together when they touch (like merging soap bubbles), and their internal molecules are constantly moving, exchanging with the surrounding environment on timescales of seconds to minutes.
Why Cells Use Phase Separation
Phase separation gives cells a fast, reversible way to organize their interior. Because condensates form and dissolve based on local conditions, they can appear in seconds when needed and vanish just as quickly, no membrane assembly or disassembly required.
One major benefit is concentrating reactants. By gathering specific enzymes and their substrates into a small droplet, cells can dramatically speed up biochemical reactions that would be sluggish if those same molecules were diluted across the entire cell volume. LLPS can also act as a molecular switch, channeling raw materials between competing metabolic pathways depending on what the cell needs. Under stress, for instance, condensates can reroute metabolism toward survival.
Condensates don’t just speed things up. They can also slow things down or block them entirely, by sequestering molecules away from their normal partners. The ability to tune local concentrations, control which molecules are present or absent, adjust the timescale of interactions, and even alter solvent conditions inside the droplet makes LLPS a versatile regulatory tool.
How Cells Control Phase Separation
Cells don’t leave phase separation to chance. They fine-tune it through chemical modifications to the proteins involved, particularly phosphorylation, the addition of a phosphate group to specific amino acids. Phosphorylation adds negative charge, which can either promote or inhibit phase separation depending on the context. Adding a phosphate to an uncharged amino acid can strengthen electrostatic attraction with positively charged partners, pushing separation forward. Or it can introduce repulsion between similarly charged molecules, dissolving condensates.
For the histone H1 protein, phosphorylation of just three sites was enough to inhibit its phase separation. A change of only 10% in the protein’s surface charge significantly disrupted its ability to bind DNA and form condensates. Phosphorylation of the FUS protein’s disordered region introduces charges that weaken the multivalent interactions driving condensation, reducing the risk of harmful aggregation. Similarly, phosphorylation of TDP-43 reduces aggregate formation in cells.
Other modifications matter too. Acetylation (adding an acetyl group) can neutralize positively charged amino acids, blocking key interactions. For the influenza nucleoprotein, acetylation of a single lysine residue impedes its condensation with viral RNA. Methylation of arginine residues in FUS by a specific enzyme reduces its phase separation in lab experiments. Cells use these modifications as dials, continuously adjusting which condensates exist, how large they are, and how long they persist.
When Phase Separation Goes Wrong
The same liquid-like properties that make condensates useful also make them vulnerable to a dangerous transition: solidification. Over time or under pathological conditions, the dynamic liquid droplets can harden into gel-like or solid aggregates that the cell cannot dissolve. This liquid-to-solid transition is now recognized as a central event in several neurodegenerative diseases.
In Alzheimer’s disease, the tau protein forms cytosolic aggregates. In Parkinson’s disease, alpha-synuclein accumulates into Lewy bodies. In ALS and frontotemporal dementia, TDP-43 and FUS form solid inclusions in the cytoplasm. In Huntington’s disease, proteins with expanded polyglutamine stretches aggregate. In each case, proteins that normally exist in soluble form or in dynamic condensates become trapped in insoluble deposits.
RNA plays a protective role in this process. In the nucleus, high RNA concentrations keep RNA-binding proteins soluble by preventing excessive protein-protein interactions. When these proteins are mislocalized to the cytoplasm, where RNA concentrations are lower, they become prone to solidifying. Pathogenic mutations in genes like FUS and TDP-43 accelerate this transition, while also disrupting normal RNA processing. The toxicity comes from both the loss of the protein’s normal function and the direct damage caused by the solid aggregates themselves.
Aberrant phase separation is also increasingly linked to cancer. Many oncogenic proteins contain intrinsically disordered regions regulated by LLPS, and disruptions to condensate formation can interfere with chromatin remodeling, gene transcription, and other processes that keep cell growth in check.
How Scientists Identify Phase Separation
Proving that a biological structure forms through LLPS requires specific experimental evidence. The commonly accepted criteria are that the structure should be spherical, that droplets should fuse together and relax back into a round shape, and that molecules inside should be mobile.
Mobility is often tested using a technique called FRAP (fluorescence recovery after photobleaching). Researchers use a laser to bleach the fluorescent label inside a droplet, then watch how quickly unbleached molecules flow in from the surroundings. Fast recovery suggests a liquid interior. However, FRAP alone isn’t definitive because recovery rates depend on droplet size, the specific molecule being tracked, and how large an area was bleached. A more rigorous measurement is the inverse capillary velocity, determined by filming two droplets fuse and measuring how long it takes them to relax into one sphere. This gives a direct readout of viscosity relative to surface tension.
Researchers also compare wild-type proteins with mutant versions (for example, with sticky residues removed) to confirm that specific molecular interactions are required. In cells, the same criteria apply: the structure should be spherical, show fusion events, and recover after photobleaching, with careful controls to rule out other explanations.
Engineered Phase Separation in Biotechnology
Scientists are now building synthetic condensates for practical purposes. By designing artificial disordered proteins or repurposing natural ones, researchers can create custom compartments inside living cells that concentrate specific enzymes and substrates.
In one application, synthetic condensates were used to recruit enzymes for producing a human milk sugar (2′-fucosyllactose) in bacteria, achieving yields of nearly 682 mg/L. Another system used light-controlled condensates: in the dark, enzymes clustered together inside droplets to produce one product, while blue light triggered the droplets to dissolve, redirecting the pathway toward a different product. Artificial organelles that concentrate specialized translation machinery have improved the efficiency of incorporating non-standard amino acids into proteins, boosting output of a target molecule by 75%.
LLPS is also being used to solve a long-standing problem in biotechnology: expressing proteins that are toxic or prone to aggregation. Condensates create a protective microenvironment that reduces premature aggregation and shields cells from toxic products. A phase-separation-based tag called BEAK-tag enabled production of aggregation-prone peptides in bacteria, and similar strategies have allowed expression of antimicrobial peptides while reducing their toxicity to the host cell. Because condensates can be triggered to form and dissolve on demand, they also offer a simple route for purifying recombinant proteins without expensive chromatography steps. Synthetic condensates are being explored as drug delivery platforms, taking advantage of their ability to encapsulate and release cargo in response to environmental changes.