# Chapter 58. Physical Principles and Evidence Standards for RNA Condensates

## Scope Note

RNA-containing condensates are compositionally enriched assemblies in which RNA, RNA-binding proteins, and other molecules interact without a surrounding lipid membrane. This chapter owns the portable physical principles needed to reason about those assemblies: phase coexistence, associative polymer networks, concentration and valency effects, RNA-dependent recruitment, material properties, aging, and standards of evidence. Named cellular bodies appear only as bounded examples of a physical principle. Their molecular inventories and biological programs belong to [Chapter 95](chapter1090.md) for nuclear RNA bodies, [Chapter 105](chapter1100.md) for cellular RNA granules, and [Chapter 118](chapter1112.md) for viral RNA condensation. RNA-protein recognition is developed in [Chapter 56](chapter1051.md), RNP machines in [Chapter 57](chapter1052.md), measurement methods in [Chapter 59](chapter1054.md), and general causal inference in [Chapter 5](chapter1005.md).

## Executive Summary

A biomolecular condensate is an enriched molecular region, not a mechanism. Liquid-liquid phase separation (LLPS) is one possible mechanism in which a dense liquid phase coexists with a dilute liquid phase. RNA-rich cellular assemblies can instead be polymer networks, gels, glasses, scaffold-bound clusters, actively maintained reaction zones, or mixtures of these states. The first discipline of condensate biology is therefore terminological: morphology should not be allowed to decide mechanism.

RNA is an unusually versatile regulator of condensation because it is a long, charged, sequence-specific polymer. Increasing RNA length can increase interaction valency; sequence and structure determine which contacts are available; chemical modifications can alter structure or recruit readers; and abundance controls stoichiometry. The result is often reentrant behavior. At one concentration RNA bridges proteins and promotes network formation, whereas excess RNA saturates binding sites and dissolves or redirects the same network. RNA may also partition to an interface and change condensate size, coalescence, or surface chemistry rather than residing uniformly in the interior.

Protein valency supplies the complementary interaction network. Folded RNA-binding domains provide affinity and selectivity, while intrinsically disordered or low-complexity regions often provide additional weak contacts. Whether a molecule acts as a scaffold or a client depends on concentration, partners, and condition; these are experimentally defined roles, not permanent molecular identities. ATP-dependent enzymes, transcription, translation, RNA decay, and transport continuously change those conditions. Cellular condensates are consequently nonequilibrium systems even when equilibrium phase diagrams provide a useful reference.

Material state governs molecular access and reaction kinetics. A condensate may flow over long times yet respond elastically over short times, and different components can have different mobilities in the same assembly. Aging can reflect stronger contacts, entanglement, structural conversion, chemical modification, or loss of active remodeling. A liquid-to-gel or liquid-to-solid transition is not automatically pathological, but kinetic arrest can become a physical failure mode when it prevents exchange, dissolution, or access to substrates.

Evidence must match the strength of the claim. Puncta and enrichment establish localization. Exchange and deformation constrain material behavior. Concentration-dependent coexistence and quantitative reconstitution test a proposed physical mechanism. Endogenous perturbation, separation-of-function mutants, rescue, and a defined RNA output test cellular causality. Fluorescence recovery after photobleaching (FRAP), fusion, hexanediol sensitivity, or droplets formed at high purified-protein concentration cannot carry that entire argument alone. The best studies triangulate across physical, molecular, and functional evidence rather than treating any single assay as a phase-separation detector.

## Concept Inventory

- **Biomolecular condensate:** a region enriched in selected biomolecules without a surrounding lipid bilayer. The term is deliberately mechanism-neutral.
- **Liquid-liquid phase separation:** demixing into coexisting dense and dilute liquid phases. A liquid state requires evidence about relaxation, exchange, and flow; a punctum is not sufficient.
- **Saturation concentration:** the dilute-phase concentration above which an equilibrium system enters a two-phase regime under specified conditions. Cellular thresholds may be shifted by reactions and active flux.
- **Associative network:** a connected set of multivalent interactions among polymers or proteins. Network formation can accompany phase separation, gelation, or both.
- **Multivalency:** the capacity of one molecule to make multiple contacts. Valency depends on accessible sites, not merely sequence length or domain count.
- **Scaffold and client:** condition-specific roles. Removing or weakening a scaffold changes assembly; a client partitions into an assembly without being required to build it.
- **Partition coefficient:** the ratio of a molecule's concentration in two regions or phases. Enrichment may reflect specific binding, generic compatibility, trapping, or active delivery.
- **Viscoelasticity:** combined viscous and elastic response whose apparent balance depends on the measurement timescale.
- **Aging and kinetic arrest:** time-dependent evolution toward altered composition or slower relaxation; arrest denotes failure to equilibrate on the relevant timescale.
- **Reentrant RNA effect:** promotion of condensation in one RNA concentration regime and inhibition or dissolution in another.

## What to Know Before Reading This Chapter

RNA is a negatively charged, directional polymer. Its bases carry sequence information, its backbone supplies repeated electrostatic contacts, and intramolecular base pairing changes which sites are physically accessible. [Chapter 2](chapter1002.md) and [Chapter 3](chapter1003.md) provide the chemical and structural background. Proteins bind RNA through folded domains, disordered segments, electrostatic surfaces, and cooperative assemblies; [Chapter 56](chapter1051.md) develops those interactions in detail.

Three distinctions prevent most conceptual errors. First, concentration is not the same as amount: local concentration can rise because a volume shrinks even if molecule number stays constant. Second, binding is not phase separation: a high-affinity complex can remain stoichiometric and dispersed, whereas a multivalent network may demix through many weak contacts. Third, dynamics are component- and timescale-specific. Rapid exchange of one fluorescently labeled client does not prove that the underlying scaffold is liquid.

Named bodies are useful only when the example teaches a transferable principle. Ongoing transcription can nucleate a nuclear condensate; release of untranslated messenger RNPs can change the cytoplasmic pool available for assembly; and RNA synthesis can drive a viral protein across a concentration threshold. The body-specific mechanisms, RNA fates, and organismal consequences are treated in [Chapter 95](chapter1090.md), [Chapter 105](chapter1100.md), and [Chapter 118](chapter1112.md), respectively.

## 58.1. Physical principles of RNA-containing condensates

The simplest equilibrium model asks whether mixing is energetically favored. Mixing increases configurational entropy, but attractive interactions can lower free energy when selected molecules associate. If the balance favors demixing, a mixture separates into a dense phase and a dilute phase. At equilibrium, the chemical potential of each exchanging component is equal across the phases. Adding more material above the saturation concentration increases the amount of dense phase more strongly than it increases the dilute-phase concentration. This buffering behavior is a useful signature, although cellular production and degradation can obscure it.

![Figure 58.1. Evidence Ladder for RNA Condensate Claims](../assets/figures/chapter1053_figure1.png)

**Figure 58.1. Evidence Ladder for RNA Condensate Claims.** A vertical evidence ladder begins with localization and enrichment, then adds quantitative composition and dynamics, phase-boundary or material-state measurements, controlled reconstitution, endogenous perturbation, separation-of-function rescue, and a defined RNA-fate output. Side arrows show that each rung excludes only some alternatives. The figure should make clear that morphology, FRAP, or in vitro droplets cannot independently establish cellular LLPS and function.

RNA-protein condensation is usually associative: unlike oil and water, oppositely charged or motif-matched components enter the dense phase together. A long RNA can bind several proteins, and each protein can contact multiple RNA sites or other proteins. When enough bonds connect into a system-spanning network, the mixture percolates. Percolation and phase separation are related but distinct. A mixture can form a connected gel without macroscopic demixing, or demix before its dense phase becomes mechanically arrested. This distinction explains why network connectivity and fluidity must be measured separately.

Electrostatics are important but not sufficient. Salt screens charge-charge interactions, yet bases and amino-acid side chains also participate in hydrogen bonding, cation-pi, pi-pi, hydrophobic, and stereospecific contacts. Counterion release can favor association between oppositely charged polymers. Sequence patterning matters because a chain with clustered positive residues presents a different interaction geometry from a chain with the same net charge distributed evenly. Temperature, pH, ion identity, crowding, and post-translational modification therefore reshape phase boundaries rather than serving as generic on/off switches.

RNA concentration produces a characteristic stoichiometric problem. At low RNA abundance, there may be too few bridges to connect proteins. At intermediate abundance, RNA can increase effective protein valency and favor condensation. At high abundance, separate RNA molecules can saturate protein-binding sites, preventing protein molecules from sharing a common RNA scaffold. The dense phase then shrinks or dissolves. This reentrant regime means that a statement such as “RNA promotes LLPS” is incomplete unless the RNA:protein ratio is specified.

![Figure 58.2. Concentration-Dependent Physical Roles of RNA](../assets/figures/chapter1053_figure2.png)

**Figure 58.2. Concentration-Dependent Physical Roles of RNA.** RNA can shift from ineffective bridging to network promotion, surface enrichment, binding-site saturation, and dissolution as RNA:protein stoichiometry changes; scaffold, client, buffer, surface regulator, and dissolver are condition-dependent roles rather than fixed RNA classes.

Interfaces add another layer. RNA can be concentrated in the interior, excluded, or enriched at the dense-dilute boundary. Interfacial RNA can behave like a biological surfactant: by lowering or modifying interfacial tension and creating steric or electrostatic barriers, it can limit droplet coalescence and alter size distributions. Reconstituted systems show that RNA surface localization can change condensate size and number, but the prevalence and regulation of this mechanism in cells remain system-dependent.

Cells continuously consume energy and exchange material. Transcription injects nascent RNA locally; translation loads and unloads ribosomes; helicases remodel RNA-protein contacts; enzymes alter covalent state; degradation removes polymers; and motors transport complexes. These fluxes can create steady states that resemble phase coexistence without satisfying equilibrium assumptions. Equilibrium thermodynamics remains useful as a null model, but a cellular mechanism must also account for rates of production, remodeling, transport, and clearance.

## 58.2. RNA length, sequence, structure, and modification effects

RNA length changes the maximum number and spacing of contacts. A longer transcript can recruit more copies of an RNA-binding protein and can bridge interaction networks over larger distances. Length alone is not predictive, however. Compaction by secondary structure may hide motifs; intramolecular contacts can compete with intermolecular contacts; and a long RNA with few accessible sites may have lower effective valency than a short RNA carrying repeated exposed motifs. The physically relevant variable is accessible interaction architecture, not nucleotide count by itself.

Sequence controls both specific and generic interactions. Recognition motifs recruit cognate RNA-binding proteins, repeated sequences amplify local valency, and base composition alters stacking, structure, and charge presentation. Sequence patterning can also determine whether two RNAs form intermolecular duplexes or remain intramolecularly folded. Consequently, scrambling a sequence while preserving length and composition is often a more informative control than deleting the RNA entirely: it separates motif grammar from bulk polymer effects.

Structure supplies geometry. Stems can rigidify an RNA and space motifs; loops and single-stranded regions expose contacts; tertiary folds can juxtapose distant sites. Structure may promote condensation by presenting a multivalent surface or inhibit it by sequestering sites. Computational prediction suggests candidate folds, but physical conclusions require experimental probing or structure-sensitive perturbations. A compensatory mutation that restores base pairing while changing primary sequence is especially useful because it tests structural rather than motif-based causality.

**Table 58.2. RNA Features That Change Condensate Behavior.** RNA length, sequence, structure, concentration, modification, and protein occupancy can change partitioning, network connectivity, and material state; an observed condensate effect is conditional rather than a universal property of an RNA class.

| RNA feature | Physical route | Expected measurement | Key boundary case |
| --- | --- | --- | --- |
| **Length** | Changes maximum valency and chain connectivity | Length series at matched molarity and mass | Folding may hide sites, so length is not effective valency |
| **Sequence motif** | Recruits a cognate RBP | Motif mutation with expression and structure controls | Scrambling may alter structure as well as recognition |
| **Repeat pattern** | Creates repeated contacts or intermolecular pairing | Repeat-number series and binding stoichiometry | Overexpression can create nonphysiological thresholds |
| **Secondary structure** | Exposes, masks, or spaces sites | Structure probing and compensatory mutations | Predicted folding is not physical evidence |
| **Chemical modification** | Alters structure, reader recruitment, or RNA metabolism | Site and stoichiometry mapping plus reader perturbation | Modification effects are not transcript-independent |
| **Abundance** | Changes RNA:protein stoichiometry | Endogenous calibration and broad dose response | One tested dose can conceal reentrant dissolution |
| **Interfacial localization** | Alters surface chemistry and coalescence | Spatial partitioning and size-distribution measurements | Reconstituted surface effects may not dominate in cells |

Chemical modifications can change partitioning through at least three routes. A modification can alter local base pairing, create or eliminate a binding site for a reader protein, or change the kinetics of RNA metabolism and thereby RNA abundance. N6-methyladenosine (m6A) illustrates why those routes must be separated. Modification-dependent recruitment in one defined system does not establish a general tendency of all m6A-containing RNAs to condense. Mapping artifacts, incomplete stoichiometry, reader abundance, transcript context, and cell state all affect interpretation. Body-specific consequences are owned by the corresponding nuclear or cellular-granule chapter.

RNA abundance is therefore both a mechanistic variable and a confounder. Overexpression can cross a phase boundary, create repeat-dependent aggregates, or titrate an RBP away from its normal partners. Endogenous concentration calibration should accompany claims based on ectopic RNA. Dose-response experiments should span both promotion and inhibition regimes, because testing only one concentration can conceal reentrance.

## 58.3. Protein valency, RBPs, and client recruitment

Condensate-associated RNA-binding proteins often combine folded RNA-binding domains with intrinsically disordered regions (IDRs) or low-complexity domains. Folded domains can recognize sequence or structure with high specificity. IDRs can contribute distributed electrostatic, aromatic, hydrophobic, and cation-pi contacts. This division of labor is common but not universal: folded domains can drive protein-protein association, and disordered regions can encode selective sequence patterns. “Disordered” is a structural description, not a synonym for nonspecific or phase separating.

Multivalency must be measured functionally. Counting domains gives a first approximation, but sites can be masked, mutually exclusive, or too close together to bridge partners. Phosphorylation, methylation, acetylation, and other modifications can change charge or interaction geometry. Oligomerization can increase valency, whereas ATP-dependent remodeling can reduce bond lifetime. A useful perturbation series changes one physical parameter—RNA affinity, protein self-association, oligomeric state, or enzymatic activity—while preserving expression and localization as much as possible.

Scaffold and client roles are defined by perturbation. A candidate scaffold should affect the phase boundary, abundance, or persistence of the assembly when its relevant interaction is weakened. A client may be highly enriched yet dispensable for assembly. Partitioning of a client is described by its concentration ratio between dense and dilute regions, but that ratio is not a direct measure of binding affinity. Size exclusion, charge compatibility, transient binding, and active import or export can all contribute.

RNA-binding proteins can also act as modifiers rather than static building blocks. Helicases bind RNA, hydrolyze ATP, displace proteins, and remodel secondary structure. These reactions can fluidize a network, dissolve it, or expose new sites that promote assembly. DDX3X and DDX3Y provide a bounded example in which homologous helicases show different condensation-associated RNA-metabolism effects, but the physical contribution must be distinguished from expression, localization, and catalytic differences. Comparative treatment of helicase mechanisms belongs to [Chapter 54](chapter1050.md).

Client recruitment should be linked to an outcome. Enrichment can accelerate a reaction by concentrating enzyme and substrate, inhibit it by separating partners, or have little effect if the client remains inaccessible. Measuring total enrichment without measuring the accessible or reactive fraction can therefore mislead. The mechanistic chain should specify recruitment, local mobility or binding state, and the resulting RNA reaction.

## 58.4. Material properties, aging transitions, and physical failure modes

Material properties describe response to force and time. A Newtonian liquid flows with a constant viscosity, whereas a viscoelastic material stores some deformation elastically and dissipates some through flow. Biomolecular condensates often display multiple relaxation times because RNA entanglement, transient bonds, and substructures relax at different rates. Calling an assembly “liquid” without stating the observation timescale discards this information.

FRAP measures return of fluorescence to a bleached region. Recovery can reflect molecular exchange with the surroundings, diffusion within the assembly, binding turnover, chemical conversion, or movement of unbleached material. The recovery fraction and timescale are useful comparative observables, but neither is viscosity by itself. Fusion-relaxation, particle tracking, optical manipulation, and microrheology interrogate complementary properties. Agreement across methods is stronger than any one readout.

Aging is a time-dependent change in composition or dynamics. Weak interactions can reorganize into longer-lived contacts; RNA can anneal into intermolecular structures; proteins can undergo conformational conversion; and loss of ATP-dependent remodeling can lengthen bond lifetimes. The resulting material may become more elastic, gel-like, glassy, or solid-like. Some transitions are functional maturation. Others create kinetic arrest, meaning that the assembly cannot exchange or dissolve on the timescale required by the cell.

![Figure 58.4. Material-State Aging and Physical Failure](../assets/figures/chapter1053_figure4.png)

**Figure 58.4. Material-State Aging and Physical Failure.** A time axis connects a dynamic viscoelastic network to longer-lived contacts, entanglement or structural conversion, gel-like behavior, and kinetic arrest. Parallel arrows show active remodeling and reversal. A physical failure is labeled only when an operation is lost—for example exchange, dissolution, selective permeability, or substrate access—not merely when an assembly appears less liquid.

Physical failure should be defined by a failed operation, not by appearance. Examples include failure to dissolve after the initiating signal ends, exclusion of a required enzyme, trapping of an RNA substrate, loss of selective permeability, or inability of a remodeling reaction to restore exchange. This definition separates a mechanistic failure mode from a mere correlation between visible condensates and stress. Disease-specific consequences and organelle-specific quality control are developed in [Chapter 121](chapter1148.md), [Chapter 95](chapter1090.md), and [Chapter 105](chapter1100.md).

Hysteresis is an important boundary case. An assembly may form at one concentration but persist when concentration falls below that value because nucleation barriers or structural maturation make the return path different. A single endpoint therefore cannot distinguish equilibrium buffering, slow dissolution, and irreversible conversion. Time courses that vary both the direction and rate of a perturbation are needed to reveal path dependence.

## 58.5. Evidence standards and phase-separation controversies

The evidence ladder begins with observation. Imaging can establish that a component is locally enriched and can measure shape, number, size, position, and lifetime. Colocalization identifies shared space at optical resolution, not direct interaction. Super-resolution imaging and single-molecule measurements improve spatial information but still require biochemical or genetic tests to assign molecular contacts.

The next level constrains material behavior. Fusion followed by rounding supports surface-tension-driven relaxation, but clustered objects can also merge. FRAP supports mobility, but mobile clients can recover within an immobile scaffold. Concentration calibration, dilute-phase buffering, partition coefficients, deformation response, and multiple timescales provide a more quantitative physical description. A phase diagram is especially informative because it records the boundary across composition, salt, temperature, or another controlled variable.

Reconstitution tests sufficiency under defined conditions. Purified components can reveal minimal interaction rules and allow phase diagrams to be measured. It does not establish physiological necessity if concentrations, tags, crowding agents, or ionic conditions differ greatly from the cell. Extracts preserve more partners but introduce unknown composition. The strongest reconstitution studies therefore calibrate components to endogenous ranges and predict perturbations that are then tested in cells.

Endogenous perturbation tests necessity. Deletion of an entire protein or RNA may alter many functions besides condensation, so separation-of-function designs are preferred. A mutation can weaken self-association while preserving RNA binding, or alter motif accessibility while preserving RNA abundance. Expression, localization, stability, catalytic activity, and partner binding must be checked. Rescue with a wild-type molecule and failure of a physically defective variant make the causal inference stronger.

Function is the final required link. The experiment should name an RNA fate—translation, decay, processing, localization, modification, or accessibility—and measure it directly. A cellular phenotype without an intermediate molecular readout leaves open whether the condensate acts through RNA metabolism or another pathway. Body-specific functional chains are evaluated in [Chapter 95](chapter1090.md), [Chapter 105](chapter1100.md), and [Chapter 118](chapter1112.md).

> **Box 58.1. What Does Not Prove LLPS?**
>
> - Round puncta do not establish a liquid state.
> - Fusion does not by itself establish equilibrium phase separation.
> - FRAP recovery does not directly measure viscosity.
> - In vitro droplets at high concentration prove possibility, not physiological necessity.
> - Hexanediol sensitivity is nonspecific.
> - Low-complexity sequence suggests interaction potential, not a cellular mechanism.
> - Enrichment does not establish recruitment mechanism or RNA regulation.
>
> A careful claim names the component, concentration, condition, phase or material criterion, perturbation, alternative mechanism, rescue, and molecular output.

Artifacts recur across the evidence ladder. Fluorescent tags can change valency, fixation can create or erase assemblies, overexpression can cross thresholds, lysis can mix compartments, and 1,6-hexanediol perturbs membranes and proteins as well as weak hydrophobic contacts. Circular reasoning is another artifact: selecting “phase-separation proteins” by puncta formation and then using puncta as proof of phase separation. Orthogonal assays and explicit alternative models are the remedy.

> **Box 58.3. Terminology Discipline**
>
> - Use *condensate* for an enriched assembly without implying a formation mechanism.
> - Use *LLPS* only when dense-dilute phase coexistence and a liquid material state are supported.
> - Use *associative network* when multivalent connectivity is demonstrated but macroscopic demixing is not.
> - Use *scaffold* and *client* as condition-specific, perturbation-tested roles.
> - Use *material state* with a measurement timescale.
> - Use *physical failure* only when a required operation is impaired.

## Experimental Foundations and Evidence

Quantitative concentration is the bridge between cell biology and physical chemistry. Fluorescence intensity must be calibrated against standards or molecule counts if it is used to infer a phase boundary. Dense-phase and dilute-phase concentrations should be measured separately, and the imaging point-spread function must be considered when assemblies approach the diffraction limit. Partition coefficients should include uncertainty and should not be compared across incompatible acquisition settings.

Time-resolved perturbation distinguishes cause from consequence. Acute induction or depletion can reveal whether condensation precedes an RNA-fate change. Reversing the perturbation tests dissolution and hysteresis. Combining imaging with sequencing, biochemical fractionation, or single-molecule RNA measurements connects material state to composition and function. Each method answers a different question: imaging maps space and dynamics, reconstitution tests physical sufficiency, genetics tests necessity, and RNA assays test consequence.

Negative evidence is informative when assay sensitivity is stated. Absence of rounded droplets does not exclude nanoscale clustering or gelation. Absence of FRAP recovery does not distinguish an immobile network from photodamage. Failure to reconstitute may mean a missing cofactor rather than absence of the mechanism. A rigorous conclusion therefore bounds what the experiment excludes and leaves alternative models explicit.

## Biological Boundaries and Cross-Chapter Handoffs

The physical questions in this chapter travel across biological systems: What raises effective valency? Is a dense phase buffered? Which molecules are scaffolds or clients? What is the material response timescale? Which active flux maintains the assembly? What RNA fate changes when the physical driver is perturbed?

The answers are system-specific. Nuclear bodies couple condensation to transcription, chromatin, RNA processing, and ribosome biogenesis; their catalog and mechanisms belong to [Chapter 95](chapter1090.md). Cytoplasmic stress granules, processing bodies, germline granules, transport granules, and other cellular RNP bodies have distinct triggers and RNA fates; those comparisons belong to [Chapter 105](chapter1100.md). Viral condensates combine viral proteins, genomic or subgenomic RNAs, replication kinetics, and host responses; those mechanisms belong to [Chapter 118](chapter1112.md). These chapters should import the evidence standards defined here rather than duplicating the physical primer.

## Technology, Computational, and Engineering Links

Computational models can connect molecular features to phase behavior at several resolutions. Mean-field models estimate phase boundaries from interaction parameters; polymer models represent chain length and valency; coarse-grained simulations explore sequence patterning and interfacial organization; and machine-learning classifiers identify candidate condensate-associated proteins. Predictions are hypotheses, not evidence of cellular LLPS. Training sets can inherit circular annotations, and a model may learn low-complexity composition without learning mechanism.

Engineering condensates requires control over threshold, selectivity, reversibility, and function. Synthetic multivalent modules can tune assembly, while designed RNA motifs can recruit clients or alter stoichiometry. A useful design must work near physiological concentrations, avoid uncontrolled aging, and demonstrate the intended reaction rather than merely produce puncta. Measurement methods and their limitations are developed in [Chapter 59](chapter1054.md), while specific synthetic RNA systems are treated in later engineering chapters.

## Recent Consensus

RNA is an active regulator of phase behavior, not passive cargo. Its concentration, length, structure, sequence, and modification can promote, inhibit, redirect, or interfacially regulate an assembly. Condensates occupy a continuum of material states, and active cellular reactions can maintain states that equilibrium descriptions alone do not capture.

The term *condensate* is broader than LLPS. No single morphology or dynamics assay proves liquid-liquid phase separation. Strong mechanistic claims combine quantitative physical evidence with endogenous perturbation, rescue, and a defined molecular output. Named bodies may share these principles without sharing composition, function, or causal architecture.

## Open Questions, Controversies, Deprecated Models, and Common Misconceptions

Open questions:

- Which sequence grammars predict whether an RNA will scaffold, enter, coat, or dissolve a condensate at endogenous concentration?
- How do active reactions reshape phase boundaries and material states in living cells?
- Which measurements can resolve nanoscale heterogeneous networks that do not form optically visible droplets?
- When does condensate aging support maturation, and when does it become kinetic failure?
- How can in vitro interaction parameters be transferred quantitatively to crowded, nonequilibrium cellular environments?

Controversies:

- Whether a given cellular assembly should be described as LLPS often remains disputed because available data support enrichment and exchange but not a complete phase diagram or liquid-state measurement.
- The relative importance of equilibrium demixing, percolation, gelation, scaffold binding, and active flux differs among systems and may change over time within one assembly.

Common misconceptions:

- "All round membraneless bodies are liquid droplets." Roundness can arise from several mechanisms and does not establish phase coexistence.
- "FRAP measures condensate viscosity." FRAP measures fluorescence recovery produced by exchange, diffusion, binding, and other processes; viscosity requires additional physical modeling or measurement.
- "RNA promotes phase separation." RNA can promote, inhibit, dissolve, redirect, or coat condensates depending on concentration and accessible interaction sites.
- "A low-complexity domain proves LLPS." Low complexity suggests possible interaction modes but does not establish a cellular mechanism.
- "Reconstitution proves physiological function." Reconstitution establishes possibility or sufficiency under defined conditions; necessity and function require cellular tests.

Deprecated or weakened claims:

- Generic “droplet equals function” diagrams and irreversible use of scaffold/client labels are inadequate. Modern models specify concentration, condition, alternative mechanisms, material timescale, perturbation, and molecular output.