In plain terms

The claim being assessed: Endogenous transcriptional machinery forms liquid-liquid phase-separated condensates whose liquid state is required for transcription and is its primary organizing principle.

Where the evidence stands: Γ — Plausible mechanism. There is a believable biological story and partial support, but the causal step has not been demonstrated.

What this mechanism does: This makes the process worse or keeps it going, but something else starts it.

Strongest evidence for it: Weakest-link aggregation yields Γ. SC1 has clear physical capacity and native assemblies but incomplete phase diagnosis; SC2 has strong selected assembly/rescue evidence without cleanly isolating liquid state from affinity and composition; SC3 lacks representative native phase-specific primacy. Stronger levels for multivalency models are not transferred to the stricter LLPS conjunction.

What decisive evidence would look like

In plain terms, researchers would change the proposed cause, confirm it changed as intended, and measure a meaningful health outcome while checking other explanations. The result should be able to support or count against the claim.

Relative explanatory fit: Alternative ranked ahead

What this review means

Gene-regulating proteins form small assemblies in cells. This review asks whether those assemblies are liquids, whether the liquid state is specifically required for transcription and whether it is the main organizing principle.

Implications for health and treatment: Weakest-link aggregation yields Γ. SC1 has clear physical capacity and native assemblies but incomplete phase diagnosis; SC2 has strong selected assembly/rescue evidence without cleanly isolating liquid state from affinity and composition; SC3 lacks representative native phase-specific primacy. Stronger levels for multivalency models are not transferred to the stricter LLPS conjunction.

Limits for individual care: 64 primary reports with 10 selected partial-body examinations; density, heterogeneous controls and incomplete access set the confidence, not correctness. The grade applies to the exact registered claim; stronger findings for narrower subclaims remain separate. No complete-paper audit or blinded calibration is claimed. This educational review does not establish an individual treatment recommendation.

Medical information notice: This review provides general health information, not medical advice. It cannot diagnose a condition or replace an individual discussion with a qualified clinician.

Source checking used complete abstracts/status records and selected original partial-body passages. No complete-paper audit or independent scientific peer review is claimed. Detailed access counts and limitations appear in the review.

A low grade is not an accusation. It means the causal work has not been done yet, which is the ordinary state of most actively studied mechanisms in biology.

At a glance

Sub-claims assessed separately

Figures

Explanatory figures summarizing this review. They carry no evidential weight of their own; where a figure and the review differ, the review governs.

Full assessment

From here the review is written for a technical reader: caps and thresholds, competing models, the independence basis, and the citation ledger.

Sources: 64 primary reports; claims: 68; verified: 68; unverified: 0. Verification is bounded attribution. Access: 10 partial-body and 54 complete abstract/status examinations. Literature cutoff: 2026-10-08; source examination UTC date: 2026-10-09. No complete-paper audit or blinded calibration claimed. Adversarial review: Generated by Sol 6.1; adversarially reviewed by Claude Opus 5.5 (Anthropic) on 2026-10-14. 5 corrections applied (see ADVERSARIAL REVIEW AND APPROVAL). Status: Approved as revised.

AT A GLANCE

Native transcriptional hubs and functional multivalent assemblies are supported; liquid-state-specific necessity and broad organizing primacy remain unestablished.

Deciding experiment: Reversible native material-state tuning with independently matched binding and cofactor recruitment

Belief versus evidence: Weakest-link aggregation yields Γ. SC1 has clear physical capacity and native assemblies but incomplete phase diagnosis; SC2 has strong selected assembly/rescue evidence without cleanly isolating liquid state from affinity and composition; SC3 lacks representative native phase-specific primacy. Stronger levels for multivalency models are not transferred to the stricter LLPS conjunction.

Best-supported rival: Multivalent residence-time kinetic proofreading. At a native locus, tune state reversibly while keeping DNA occupancy, cofactor affinity, concentration and activation-domain chemistry matched; compare matched biochemical-only rescue and measure nascent bursts alongside phase diagnostics.

COMPETING MECHANISTIC MODELS

Rank 1: Multivalent residence-time kinetic proofreading (Ε). Weak cooperative interactions stabilize recruitment and reaction timing without requiring a bulk liquid phase. Synthetic matched activators and native EWS target search support residence/recruitment mechanisms (35537448,29930090,38548891). MYOCD unrelated IDR rescue and measured native assemblies could contain material-state effects not captured by simple affinity (38489358).

Rank 2: DNA-mediated dynamic hubs (Ε). DNA binding and cofactor networks nucleate finite dynamic assemblies. Native EWS hubs and CTD confinement support spatially restricted recruitment (29930090,42417395,38548891). RNA feedback and pH/IDR rescue show regulation beyond static DNA occupancy (33333019,40680734,33548202).

Rank 3: Phase separation coupled to percolation (Δ). Composition-dependent networks can cooperate with, or oppose, phase separation and transcription. Notch viscoelastic complexes and Gcn4 soluble/condensed comparisons support mixed assembly modes (39300145,40614721). Notch abundance and partner binding change, while Gcn4 cellular phase assignment is partly inferred; neither supplies generic endogenous phase necessity.

Rank 4: Chromatin bridging-induced clustering (Γ). Binding of multiple chromatin segments creates concentration and contact enrichment without bulk LLPS. Surface-condensation and original polymer model offer plausible clustering physics (34569155,33885786). Original model is theoretical; selected native IDR rescue suggests composition-specific function beyond generic bridging (38489358).

Rank 5: Classical liquid-liquid transcriptional condensates (Γ). Distinct liquid phases broadly organize and are required for active transcription. Native MYOCD and NELF controlled IDR replacements plus BRD4 pH reversibility are the strongest supportive cases (38489358,33548202,40680734). Condensation without MYOCD function, limited MYC phase effects, soluble-complex alternatives and inhibitory droplets defeat broad phase-specific necessity (38489358,38811792,40614721,35537448).

The tournament does not alter the nominal level. Compatible pathways and flexible alternatives are not automatically proved by explaining observations.

FULL REVIEW

Registered claim and scope

Endogenous transcriptional machinery forms liquid-liquid phase-separated condensates whose liquid state is required for transcription and is its primary organizing principle.

Use the index LLPS-in-transcription conjunction. Puncta, recruitment, FRAP and sensitivity to hexanediol are not by themselves a phase-state diagnosis; assembly, biochemical interaction and LLPS-specific necessity remain separate subclaims.

Selection: public-interest rank 44 from the editorial ranking compiled 2026-09-23; categorical prior verdicts were visible. Private fresh review; earlier documents are preserved.

Evidence and causal interpretation

Native assemblies are real; identifying their state and mechanism requires more than puncta. Live PolII clusters are transient, and CTD-dependent yeast confinement occurs without apparent phase separation (23828889,27138339,38548891). Native coactivator, MYOCD and EWS imaging supports spatial organization, not automatically a distinct liquid phase (29930094,29930091,38489358,29930090).

Sabari original sections use endogenous BRD4/MED1 tags and find lower coactivator/PolII occupancy after hexanediol (29930091). MED1 optoIDR and purified/nuclear-extract droplets establish capacity. The serine-to-alanine droplet test is in vitro; extracted droplets can sequester machinery and suppress transcription. No native interaction-matched phase-only rescue follows from those assays.

Hexanediol inhibits kinases/phosphatases at concentrations commonly used for phase dissolution (33814344). Both 1,6- and 2,5-alcohols suppress coalescing and noncoalescing yeast transcription and alter histone occupancy (35963432). Alcohol sensitivity therefore cannot carry liquid-state causality alone.

MYOCD is a strong positive case, not dismissed as a reporter alone. Endogenous mEGFP knock-in hiPSC differentiation yields estimated nuclear concentrations of 197.5±71.9 nM in smooth muscle cells and 131.6±33.6 nM in cardiomyocytes (38489358). FUS/TAF15 reporter rescues extend to matched-expression FUS/TAF15/DDX4/NPHS1 reprogramming rescue in mouse fibroblasts. These systems remain distinct: native concentration was measured in hiPSC derivatives, whereas the IDR replacement causality tests were ectopic reprogramming.

MYOCD-CDT1 is the crucial negative rescue (38489358). It restores condensates at matched protein expression and has high DNA occupancy, but fails ACTA2/reprogramming, p300/PolII recruitment and acetylation. Aromatic-IDR chemistry selectively restores those functions. This narrows generic DNA-binding failure while showing that condensate existence itself is insufficient; whether material state contributes independently of compatible chemistry is unresolved.

NELF stress experiments preserve nuclear localization, complex incorporation and basal promoter binding after IDR deletion (33548202). Stress-dependent promoter enrichment and transcriptional downregulation fail, while unrelated FUS/EWS replacements form constitutive condensates, repress tested genes without stress and improve survival. This supports a context-dependent assembly function with better controls than protein loss alone, but phase, recruitment and phosphorylation/SUMO chemistry remain linked.

MYC SparkDrop holds nuclear abundance near 200 nM while switching droplets through an engineered multivalent tag (38811792). Original Results report 99 genes changed at least 1.5-fold between diffuse and condensed states, with 88 up and 11 down; HOTag-free drug controls yield 10 genes without overlap. Most MYC target responses occur in the diffuse state. Phase effects are real and restricted in this engineered system, including oncogenic functional consequences, not general transcription necessity.

Trojanowski original reporter experiments distinguish multivalence from visible droplets (35537448). Multivalent activation domains enhance residence/cofactor recruitment below the phase threshold; additional droplets are neutral or can inhibit. In the same system, p300 or some inactive BLInCR constructs enrich acetylation/BRD4 without output, and JQ1 leaves much nascent reporter activation intact. Synthetic promoter context limits transfer to native global transcription.

Chong 2018 uses endogenous Halo-tagged EWS-FLI1 and controlled rescue, finding native GGAA hubs with longer residence and LCD-dependent target-gene activation/transformation (29930090). Tagging preserves reporter activation but somewhat reduces soft-agar efficiency. Later interaction-strength variants show that too much self-association suppresses native targets (35483357). These studies share an EWS programme and are not wholly independent lines.

The 2026 EWS source distinguishes approximately 60 nm pair-correlation diameter from mean 94 nm reconstructed diameter; both use explicit blinking correction (42417395). A fitted classical-nucleation model suggests arrested supersaturated growth and estimates about 34 dissolutions per second, rather than directly observing that rate. Parental EWSR1/FLI1 also form smaller puncta. RNA inhibition alters residence but not dimensions/bound fraction; authors retain transcription-dependent accessibility/cofactor alternatives.

BRD4 pH control adds useful specificity (40680734). Native protein abundance remains stable; isolated nuclei and reversible acidification narrow cytoplasmic explanations. Matched-expression histidine segment mutants rescue protein function while changing pH response. Nonetheless protonation changes interaction affinity as well as phase, and FRAP/alcohol do not establish a state-only perturbation. The result is a selective context-dependent response, not global transcriptional collapse.

Notch assemblies combine liquid-like exchange with less mobile networks and are framed as phase separation coupled to percolation (39300145). Native ligand/GSI responses change nuclear abundance; OptoNotch deletion mutants affect abundance or RBPJ/partner recruitment. MYC contact rescue and RNA output support Notch function, but hexanediol and domain deletion do not isolate liquid-state necessity.

Bremer compares Gcn4 variants with purified affinity and native promoter readouts (40614721). DNA-bound complexes can oppose phase separation, and soluble Med15 affinity strongly covaries with condensation propensity. High-affinity variants underperform despite promoter occupancy not accounting for the deficit, suggesting condensation-associated attenuation. Cellular phase assignment partly follows predictive relationships, and ChIP antibody accessibility remains a limitation; the paper does not reduce the mechanism to either an all-soluble or an all-droplet model.

PSPC1 acute loss, enhancer RNA readers, TRIM28/ERV relocation, SREBP2 IDR rescue and BMAL1 mouse/cell rescues extend positive multivalent regulation (34916619,34375583,35864192,40394324,42062246). The currently examined depth for most is abstract-level. IDR loss or RNA/protein removal also alters recruitment chemistry, so these do not automatically supply native phase-only necessity.

Other systems guard against a universal direction. PolI condensation can mark transcriptional inactivation; endogenous PER2 assemblies differ from overexpression liquids, whereas BMAL1 is a distinct positive clock limb (33055158,38127982,42062246). Viral replication compartment accumulation can arise through accessible-DNA rebinding with unrestricted boundary diffusion (31038454). These are scope tests and alternative mechanisms, not claims all normal enhancers behave identically.

The bounded conclusion is Γ for the strict LLPS conjunction, with stronger support for finite cooperative hubs and context-dependent multivalent recruitment. Tag-dependent transcriptomes, plasmid-backbone RNA and DNA-induced TCF4 dissolution make quantitative native controls essential (38799575,40119729,42106138). Research gaps are independently reconciled; provider-missed originals are not treated as absent evidence.

HYPOTHESIS STATE EVALUATION FRAMEWORK (HSEF v5.3)

Subclaim matrix and aggregation

Preregistered aggregation: weakest-link. Weakest-link aggregation yields Γ. SC1 has clear physical capacity and native assemblies but incomplete phase diagnosis; SC2 has strong selected assembly/rescue evidence without cleanly isolating liquid state from affinity and composition; SC3 lacks representative native phase-specific primacy. Stronger levels for multivalency models are not transferred to the stricter LLPS conjunction. Sensitivities: weakest-link Γ; primary-subclaim Γ; scope-matched Γ.

Decisive-perturbation inventory and causal tally

[5] Acidic pH disrupts BRD4/MED1 assemblies and changes inflammatory output through histidine-rich regions. Stable protein abundance, isolated nuclei, reversible pH and functional segment mutants narrow generic collapse; protonation affects molecular affinities and phase together. FRAP/alcohol are not specific state diagnoses.

[6] Nearby condensate enhances burst size/frequency while basal output persists when distant; local DNA/cohesin perturbations affect enhancement. Assembly position is not itself liquid diagnosis; DNA/protein perturbations affect other functions. Erratum linked, body audit pending.

[7] Matched N-myc levels show phase separation further modulates fewer than three percent of regulated genes with proliferation effects. Engineered state/overexpression and small affected set constrain broad primacy; correction linked, body audit pending.

[8] Multivalent interactions enhance occupancy/residence/cofactor recruitment; droplets are neutral or inhibitory for output. Reporter and synthetic platform cannot refute all endogenous contexts; it separates assembly benefits from droplet state.

[14] Endogenous MYOCD knock-in differentiation shows hubs at measured native concentrations; matched-expression aromatic IDR replacements rescue reprogramming and preserve DNA occupancy. CDT1 replacement restores condensates and high DNA occupancy but not p300 recruitment, acetylation or output; chemistry and composition matter, while liquid state alone remains unisolated.

[15] NELF IDR replacement with FUS/EWS supports stress-dependent promoter recruitment and downregulation. Nuclear localization, complex incorporation and basal promoter binding are preserved after IDR deletion; stress-dependent occupancy changes. FUS/EWS replacements constitutively repress and improve survival, but chemical recruitment and material state remain coupled.

[21] BRD4/MED1 puncta at super-enhancers, chemical sensitivity and IDR droplets support condensation capacity. Puncta, FRAP, hexanediol and extract compartmentalization cannot isolate endogenous liquid-state necessity.

[24] PSPC1 links RNA/Pol II engagement; acute degradation impairs polymerase binding and nascent transcription. PSPC1 RNA binding and polymerase chemistry change with loss; binding-matched phase rescue must be demonstrated.

[25] Pol II enters Xist territory; apparent depletion reflects loss of chromatin-bound fraction. Inactive-X initiation context differs from active loci; disproves a proposed diffusion barrier, not every phase assembly.

[27] Patterned charge blocks determine functional partitioning and gene activation during transition. Charge sequence tunes affinity and RNA/protein binding as well as phase; selective liquid property not fully disentangled.

[29] NICD forms transcription assemblies and enhancer loops, with phase separation coupled to percolation model. TAD changes affect activation surfaces; viscoelastic network and classical liquid demixing are distinct propositions. Final replaces preprint 39131356.

[30] Dynamic LCD hubs stabilize DNA binding and activate transcription without detectable phase separation at physiological abundance. Native hub interactions are real; lack of detected bulk LLPS cannot exclude subresolution or coupled transitions.

[31] Post-initiation CTD ablation preserves pause maintenance and productive release. Stage-specific test does not refute CTD initiation function or all phase pathways.

[35] Distinct TF classes depend on local enhancer context and respond differently to bromodomain inhibition. Causal enhancer syntax is shown; liquid-state necessity is not measured and condensates remain one possible implementation.

[36] Both 1,6- and 2,5-hexanediol suppress coalescing and noncoalescing gene transcription and alter histone occupancy. Alcohol is pleiotropic and 2,5-hexanediol is not a reliable inert negative control; result does not exclude specific condensate functions.

[39] Native circadian PER2 assemblies differ from overexpression-induced liquid condensates and resist several liquid-state probes. PER2 negative-feedback assemblies are not the same as positive-limb BMAL1 hubs; this is not a universal clock-protein negative result.

[40] Cryptic plasmid-backbone transcripts generate misleading liquid-like structures and paraspeckle protein recruitment. A method-specific artifact does not explain every native transcription hub; RNA scaffold controls are essential.

[42] Condensation-associated inactive caps exclude initiation factors and suppress transcription. Pol I differs from Pol II; condensates may activate or inhibit depending on composition and reaction stage.

[44] Excess self-association toward macroscopic LLPS represses target transcription and transformation. Variants change localization and interaction chemistry; does not eliminate a role for small native hubs.

[48] Endogenous EWS-FLI1 forms sub-diffraction hubs larger than parental puncta; a fitted nucleation/dissolution model explains the stable size distribution without macroscopic droplets. Estimated dissolution rate and supersaturation are model-dependent, not directly observed thermodynamic equilibrium; drugs also alter signaling, DNA binding or localization and RNA effects have alternatives.

[49] An N-terminal IDR supports circadian complexes and rescue of cellular and behavioral rhythms. IDR deletion changes partner recruitment as well as condensation; phase-only necessity and universal relevance remain unresolved.

[50] FG-rich IDRs support fusion-driven binding, looping and transformation; FUS IDR replacement rescues. Ectopic oncogenic fusion and multivalent binding are coupled; native broad primacy is not tested.

[52] Aromatic-IDR mutation reduces output and feeding-induced cholesterol responses; FUS replacement rescues cellular function. Activation affinity and phase propensity covary; sex and metabolic context limit generalization and the examined abstract does not isolate phase alone.

[53] Accessible DNA rebinding explains Pol II accumulation with unrestricted diffusion across compartment boundaries. A viral compartment is not a normal enhancer; demonstrates a viable non-LLPS mechanism rather than excluding all native LLPS.

[57] Hexanediol directly inhibits kinases and phosphatases at concentrations used in phase experiments. In-vitro enzyme tests expose nonspecificity but do not directly measure every cellular phase mechanism.

[59] Tags cause larger transcriptome changes than CTD alterations; global bursting is largely robust to CTD changes. Gene-specific CTD dependence remains; tagging artifacts require explicit controls.

[61] CTD-dependent confinement improves target search and bursting without apparent phase-separated condensates. CTD truncation also changes protein interactions; absence of visible droplets does not exclude nanoscale cooperativity.

[62] An optimal interaction range supports enhancer assembly and output; too little or too much is detrimental. Heterotypic partner binding covaries with condensation; this does not show more LLPS causes more transcription.

[64] Soluble-complex affinity and phase propensity covary; high-affinity variants show condensation-associated attenuation. Cellular phase assignment partly rests on predictive relationships and ChIP access may differ; neither all-soluble nor all-condensate models are established.

Tally: 29 discriminating contrasts. Null and negative contrasts are discriminators, not positive threshold crossings. Reports and replications are not independent causal lines. Discriminator flags include positive and negative contrasts, not independent positive LLPS threshold proofs. One original theoretical-model report is marked Weak and nonexperimental; not counted as a perturbation. The strongest MYOCD/NELF controls narrow alternatives but do not isolate liquid state across the registered conjunction.

Cap audit, precedence and Β/Γ/Δ boundary

Puncta/FRAP, in-vitro phase and reporter-only behavior cannot establish native liquid-state necessity. Positive native rescue sets are assessed jointly rather than dismissed one by one: they narrow abundance, localization and occupancy errors, but do not remove the shared affinity/composition alternative. Scoped assembly-model thresholds therefore do not lift the stricter phase-state claim.

Gamma hard-rule audit: Puncta, FRAP and hexanediol sensitivity are not a specific liquid-state diagnosis. In-vitro droplets and optogenetic concentration thresholds establish capacity, not native necessity. IDR deletion changes cofactor binding and catalytic recruitment as well as phase propensity. DNA-mediated hubs and surface condensation need not be bulk LLPS.

SC1 cap: “Mechanistic plausibility plus in vitro reconstitution or simplified cellular support caps at Γ.” Purified proteins and engineered systems phase separate, and native MYOCD, EWS and coactivators form measurable assemblies. Surface condensation, small DNA-bound networks and fitted nonequilibrium nucleation remain distinct alternatives to bona fide endogenous LLPS.

SC2 cap: “Mechanistic plausibility plus in vitro reconstitution or simplified cellular support caps at Γ.” MYOCD and NELF preserve important occupancy/localization controls and use unrelated IDR rescue; these establish meaningful assembly contributions. CDT1 condensation without function and coupled cofactor affinity show that liquid state alone is not isolated. MYC phase switching demonstrates restricted engineered effects, not native global necessity.

SC3 cap: “Mechanistic plausibility plus in vitro reconstitution or simplified cellular support caps at Γ.” Native hubs without macroscopic LLPS, yeast CTD confinement, largely phase-insensitive MYC output and inhibitory droplets constrain universal primacy. Selected positive contexts do not dominate all competent biochemical recruitment models.

SC1 field-valid definition: “ - Endogenous degron of a condensate component with matched rescue, AND evidence that the specific condensate state (not generic protein loss) drives the phenotype - Locus-specific disruption of looping with locus-specific effect - Combinatorial histone-mark perturbation showing direct output dependence - Critically: specificity against cell-cycle collapse, global transcriptional collapse, and replication stress must be demonstrated” Positive full-scope result: no. Purified proteins and engineered systems phase separate, and native MYOCD, EWS and coactivators form measurable assemblies. Surface condensation, small DNA-bound networks and fitted nonequilibrium nucleation remain distinct alternatives to bona fide endogenous LLPS.

SC2 field-valid definition: “ - Endogenous degron of a condensate component with matched rescue, AND evidence that the specific condensate state (not generic protein loss) drives the phenotype - Locus-specific disruption of looping with locus-specific effect - Combinatorial histone-mark perturbation showing direct output dependence - Critically: specificity against cell-cycle collapse, global transcriptional collapse, and replication stress must be demonstrated” Positive full-scope result: no. MYOCD and NELF preserve important occupancy/localization controls and use unrelated IDR rescue; these establish meaningful assembly contributions. CDT1 condensation without function and coupled cofactor affinity show that liquid state alone is not isolated. MYC phase switching demonstrates restricted engineered effects, not native global necessity.

SC3 field-valid definition: “ - Endogenous degron of a condensate component with matched rescue, AND evidence that the specific condensate state (not generic protein loss) drives the phenotype - Locus-specific disruption of looping with locus-specific effect - Combinatorial histone-mark perturbation showing direct output dependence - Critically: specificity against cell-cycle collapse, global transcriptional collapse, and replication stress must be demonstrated” Positive full-scope result: no. Native hubs without macroscopic LLPS, yeast CTD confinement, largely phase-insensitive MYC output and inhibitory droplets constrain universal primacy. Selected positive contexts do not dominate all competent biochemical recruitment models.

Upward threshold ladder

SC1 Δ: “Requires at least one endogenous perturbation showing the proposed feature contributes causally to the claimed output.” Satisfied: no. Native concentration titration and orthogonal quantitative state measurements distinguishing LLPS, chromatin binding, surface condensation and arrested network growth.

SC2 Δ: “Requires at least one endogenous perturbation showing the proposed feature contributes causally to the claimed output.” Satisfied: no. Reversible endogenous state-selective perturbation with matched DNA occupancy, protein abundance and cofactor affinity, plus native nascent-output rescue.

SC3 Δ: “Requires at least one endogenous perturbation showing the proposed feature contributes causally to the claimed output.” Satisfied: no. Representative prespecified loci/lineages with matched biochemical-only and material-state-only perturbations establishing dominant necessity and sufficiency.

Upper ladder: Upper levels remain possible. Native IDR replacements, measured abundance, basal promoter controls, acute protein loss and mouse functional phenotypes were considered. These lift bounded assembly models above mere correlation but leave liquid-state specificity unresolved. No impossible clinical RCT is required.

Independence: Physical reconstitution, native imaging, genetic rescue and engineered phase switching fail differently. Together they establish capacity and context-dependent multivalent function. They do not all assay endogenous liquid-state necessity, so their convergence is credited to assembly alternatives rather than automatically to the LLPS conjunction. EWS followups share A673 labels, and early PolII imaging shares a platform.

Native MYOCD concentration and matched-expression IDR swaps [14]: Preserved occupancy and nontranscriptional aromatic-IDR rescue narrow generic binding loss; CDT1 nonfunctional condensation exposes composition dependence.

NELF stress loss and constitutive heterologous rescue [15]: Preserved nuclear targeting, complex assembly and basal promoter binding narrow generic protein failure; stress occupancy and recruitment still covary.

Matched abundance phase switching and native-hub imaging [7, 30, 48]: Drug/HOTag controls test phase effects without abundance increase; native super-resolution has distinct localization/model errors.

Falsification and pitfalls

More liquid condensation fails to enhance output. Trojanowski reporter droplets neutral/inhibitory; Chong high self-association inhibits native targets; Bremer high-affinity variants underperform (35537448,35483357,40614721). Additional downgrade: False.

Condensates and occupancy recover but function does not. MYOCD-CDT1 forms condensates and binds DNA more strongly but fails acetylation/reprogramming (38489358). Additional downgrade: False.

Native function without visible LLPS. Endogenous EWS hubs and yeast CTD confinement; engineered diffuse MYC drives most target-gene responses (29930090,42417395,38548891,38811792). Additional downgrade: False.

FRAP or puncta treated as phase diagnosis. Native hubs, surfaces and diffusion/rebinding are separately interpreted.

Hexanediol treated as selective LLPS loss. Enzyme inhibition and yeast 1,6/2,5 controls retained (33814344,35963432).

IDR dependence treated as liquid-state dependence. MYOCD occupancy/CDT1 control and NELF basal binding plus rescue are weighed against cofactor chemistry.

All condensates assumed transcription activating. NELF repression, PolI caps and inhibitory Gcn4/EWS states retained.

Tag or cryptic RNA artifact ignored. PolII tagging and vector-backbone controls included (38799575,40119729).

Small hubs confused with absent molecular assembly. Negative macroscopic LLPS results do not deny cooperative sub-diffraction networks.

Provider retrieval counts and substituted references treated as validation. Only two exact original-aspect comparisons credited; canonical dates restored.

Adversarial steelman: above, then below

One rung above Γ is Δ: MYOCD occupancy-matched heterologous rescue, NELF basal-binding controls and BRD4 pH-sensor mutants could be interpreted as field-valid causal liquid-state interventions.

One rung below Γ is Β: native puncta and cofactor colocalization could be entirely associative while engineered droplets remain outside the endogenous claim.

Δ is rejected for the registered phase-specific conjunction because successful rescue also restores selective partner chemistry and no matched state-only endogenous contrast is established. Β understates demonstrated phase capacity, controlled assembly interventions and measured native concentration. Those contributions earn stronger alternative-model scores without upgrading the LLPS necessity/primacy claims.

Final score, causal direction and M/S/D

Weakest-link aggregation yields Γ. SC1 has clear physical capacity and native assemblies but incomplete phase diagnosis; SC2 has strong selected assembly/rescue evidence without cleanly isolating liquid state from affinity and composition; SC3 lacks representative native phase-specific primacy. Stronger levels for multivalency models are not transferred to the stricter LLPS conjunction.

Step 1 is not universal because selected assemblies can precede and alter output. Step 2 is exceeded by controlled genetic/engineered contrasts. Step 3 fails broad necessity, sufficiency and rate-limiting primary status. Step 4 may fit individual assemblies but no general phase-state necessity is established. Step 5 describes bounded multivalent stabilization/amplification and composition-dependent repression, assigned before model ranking; it is not proof that bona fide LLPS is universally necessary.

Momentum: Mechanistic only. The most recent primary perturbation work in this corpus tests mechanism or scope rather than the boundary check: reversible pH control of BRD4 and MED1 assemblies (40680734), the Gcn4 soluble-versus-condensed comparison (40614721) and the 2026 single-molecule measurement of endogenous EWS::FLI1 hubs (42417395). None of them delivers the matched state-only endogenous contrast - reversible native material-state tuning with DNA occupancy, protein abundance and cofactor affinity held equal - that the Delta threshold requires here. No presently recruiting trial or named pending result is asserted.

Stability: Stable. No named pending result is asserted to be likely to move the registered score imminently.

Divergence: Contested. Required artifact: the two framings carry different implied levels, so no single implied level and no arithmetic gap can be stated. Review and press framing of transcriptional condensates as the organizing principle of gene control implies a level well above the assessed Γ, while a substantial part of the primary literature cited here argues for multivalent hubs and soluble complexes and against liquid-state necessity. No manufactured field consensus is asserted. These dimensions do not change level or confidence.

PROPOSED DECISIVE EXPERIMENTS

Reversible native material-state tuning with independently matched binding and cofactor recruitment

Readout: Quantitative endogenous state diagnostics, molecule abundance/occupancy, nascent burst rates, elongation and viability in the same prespecified loci/cells.. Controls: Biochemical-only versus state-only mutants with orthogonal matched rescue; tag-free confirmation and alcohol-free perturbation.. Supporting: State-only loss suppresses nascent output despite unchanged chemistry/occupancy and state restoration rescues across independent lineages.. Falsifying: Output follows binding or cofactor chemistry while state changes are neutral or inhibitory, or rescue works without liquid-phase restoration..

RESEARCH TIMELINE

KNOWLEDGE GAPS

Native quantitative discrimination of bulk LLPS, surface condensation, finite hubs and percolated networks.

Interaction-matched reversible material-state perturbations preserving protein abundance, DNA occupancy and cofactor affinity.

Independent replication of MYOCD/NELF/SREBP2 IDR replacement controls across laboratories and native loci.

Same-cell time ordering of hub nucleation, occupancy, cofactor recruitment and nascent transcription.

Representative locus panels measuring both positive and inhibitory phase effects without responder-selected scope.

Nonlinear concentration and affinity maps reconciling soluble complexes with condensation.

Native tag controls, cryptic scaffold-RNA controls and orthogonal state readouts.

Phase effects during activation, pausing, elongation, stress repression and cell-cycle transitions separately.

Prespecified in vivo sex, lineage and disease-fusion versus normal-TF generalization.

Original correction bodies and supplements, unavailable provider target texts and source-substitution audit.

METHODS AND CONSENSUS CONTRIBUTION

Consensus was used through the signed-in browser for discovery, counterevidence, research-gap and timeline suggestions, and attempted checks of specific indexed full-text passages. Provider statements about Methods/Results and any returned passages were compared with original-paper passages where available; failed retrievals, source substitutions and unverified aspects are recorded in the case reconciliation. Successful original-passage comparisons are distinguished from provider summaries and assertions of access. Provider screening counts, citation counts and meters were not used as primary counts or HSEF scores. This source log records access depth; it does not claim complete-paper or supplement review. The monthly API allowance was conserved: zero Consensus API calls in this batch.

REVIEWER NOTES

64 original primary reports examined at complete abstract/status level; one is an original theoretical model, not an empirical perturbation. Ten selected original bodies partially examined; no complete-paper/supplement audit. Exact archived v5.3, not mislabeled 5.4. SciSpace final/preprint resolution and Consensus 2 Pro/0 Deep/0 API with two correctly attributed original-aspect comparisons. Corrections remain metadata-level. Unblinded coordinator appraisal; canonical cutoff 2026-10-08 with examination 9 October UTC.

Search leads, reviews, protocols and correction notices do not pad the primary count. Downloaded XML does not earn examination credit. Linked notices remain attached; unexamined correction details limit inference. This is a coordinator source/claim consistency assessment, not independent blinded calibration.

ADVERSARIAL REVIEW AND APPROVAL

Authoring model: Sol 6.1. Adversarial reviewer: Claude Opus 5.5 (Anthropic). Date: 2026-10-14.

Scope of checks. All 64 REFERENCES PMIDs were verified against PubMed metadata for title, first author, journal, year and DOI: 64/64 resolve to the stated record, 0 failures, no retraction or expression-of-concern flag. Twelve load-bearing claims were spot-checked against the primary abstracts, chosen because the headline turns on them: that 1,6-hexanediol directly impairs kinases and phosphatases at the concentrations used to dissolve condensates (PMID 33814344); that both 1,6- and 2,5-hexanediol suppress transcription of coalescing and non-coalescing genes and raise histone H3 density (PMID 35963432); that multivalent activation-domain interactions enhance activation while actual droplet formation is neutral or inhibitory (PMID 35537448); that fewer than 3% of N-myc-regulated genes are further modulated by phase separation, that those include key oncogenes and tumour suppressors, that proliferation rises, and that more than 97% are unaffected (PMID 38811792); that MYOCD condensate formation is concentration-dependent and switch-like, that the C-terminal disordered region is necessary and sufficient, and that unrelated disordered regions rescue gene activation and reprogramming (PMID 38489358); that endogenous low-complexity-domain hubs stabilize DNA binding and recruit RNA polymerase II without detectable phase separation at physiological concentrations (PMID 29930090); that EWS::FLI1 hubs are sub-diffraction with dissolution mechanisms that prevent macroscopic phase separation, are neomorphic relative to the parental proteins, and that nascent RNA destabilizes them without changing their dimensions (PMID 42417395); that NELF condensation requires stress-induced dephosphorylation and SUMOylation, that the NELFA disordered region can be functionally replaced by that of FUS or EWSR1, and that condensation is required for viability under stress (PMID 33548202); that Gcn4 variants with the highest Med15 affinity underperform because condensation tempers their activity, and that transcription factors can work as soluble complexes as well as condensates (PMID 40614721); and that polymerase I condensation into a nucleolar cap excludes an initiation factor and silences ribosomal DNA (PMID 33055158). All matched the review, including its careful separation of what each paper's authors conclude from what the design isolates. Scoring logic was re-derived independently of the recorded letter: the preregistered weakest-link aggregation and its two reported alternatives, the five Beta/Gamma/Delta boundary clauses per subclaim, the Delta requirements quoted per subclaim, the independence discussion, and the Momentum, Stability and Divergence artifacts. Both the validator of record (version 2.3) and the newer 2.4 build were re-run after editing.

Findings.

Rejected findings (checked, no change).

1. Gamma rather than Delta for SC2. This is the live question for this review, because the MYOCD paper's own abstract concludes that "MYOCD condensate formation is required for gene activation during cardiovascular differentiation" and uses unrelated-disordered-region rescue to say so. The review does not misreport that; it argues that the same paper's CDT1 replacement forms condensates and binds DNA more strongly while failing p300 recruitment, acetylation and reprogramming, so condensate presence is not sufficient and the successful rescues also restore partner chemistry. That is an adjudication of what the design isolates, not a contradiction of the source, and the framework's boundary check asks exactly that question. No change.
2. Weakest-link aggregation. Declared before scoring, applied as written across three subclaims that all sit at Gamma, with both alternatives reported as Gamma. The round-2 aggregation misread is not present.
3. Steelman order and content. The one-rung-above case (Delta) is written before the one-rung-below case (Beta), and the rejection names the blocker - no matched state-only endogenous contrast. Compliant.
4. Role of mechanism "reinforcing / stabilizing factor", derived through the numbered direction-of-causation steps in the prose. Defensible.
5. Stability "Stable" - correct where no pending result is named.
6. Reference style. This review prints bare PMID numbers inline and full "PMID nnnnn" entries in REFERENCES. House style, not an error.
7. The v5.claim_form warning that the entry form is a phenomenon. The registered claim is a full causal proposition with a necessity-plus-primacy consequence, so this is a rejected-finding convention for this round.
8. The v5.lay_summary FAIL under validator 2.4. This package is a declared v5.3 review and the lay summary became mandatory only in v5.4; the same FAIL is produced by running 2.4 against the pre-edit original, and the package's recorded report is from validator 2.3, which passes. Not a defect of this edition and not a regression.

9. Revision regression check. Compared against the earlier edition of this review: the revision added only the dated Consensus provenance-reconciliation block, which asserts no score change, and altered nothing else. No error was introduced by the revision.

10. Whether the confidence band should move to High once access depth is removed as a determinant. Re-tested against the framework text: the load-bearing results are single-laboratory engineered or knock-in systems whose interpretation is actively contested, and the review records that the EWS follow-ups share A673 labels and that the early polymerase imaging shares a platform; that is the Moderate text, not the High text. The band holds at Moderate. The superseded wording is preserved in the dated audit snapshots in this folder by design.

11. The coordinator-flagged revision artifact in which the spliced Consensus block contradicts preserved reviewer notes asserting that no Consensus work was done. Not present here: this review carries an affirmative "METHODS AND CONSENSUS CONTRIBUTION" section and reviewer notes that record the provider query counts. The two blocks agree.
12. The "Best-supported rival" renderer defect reported elsewhere in this round. Checked against the score card: models_evaluated[0] here is M2, is_nominal: false, rank_in_tournament: 1, and the nominal sits at rank 5. The prose names the rival correctly. No change.

Residual limitations. Access is ten partial-body and 54 complete abstract or status examinations. Body-level figures that the abstracts do not reproduce - the 197.5 and 131.6 nM native MYOCD concentrations, the 99 differentially expressed genes with 88 up and 11 down in the MYC phase-switching experiment, the 60 nm pair-correlation and 94 nm reconstructed hub diameters, and the fitted estimate of about 34 dissolutions per second - were checked for internal consistency and against the abstract-level summaries that bound them, but could not be confirmed against a complete paper in this pass. Two linked errata or corrections remain at metadata level by the review's own disclosure. No independent blinded calibration or scientific peer review is claimed by this adversarial pass.

Approval: Following adversarial review and the corrections listed above, the revised document is approved. - Claude Opus 5.5 (Anthropic), 2026-10-14.

REFERENCES

1. Henninger JE, Oksuz O, Shrinivas K et al. RNA-Mediated Feedback Control of Transcriptional Condensates. Cell. 2021. PMID 33333019. DOI: 10.1016/j.cell.2020.11.030. Access: abstract_metadata. Study: PMID33333019.

2. Basu S, Mackowiak SD, Niskanen H et al. Unblending of Transcriptional Condensates in Human Repeat Expansion Disease. Cell. 2020. PMID 32386547. DOI: 10.1016/j.cell.2020.04.018. Access: abstract_metadata. Study: PMID32386547.

3. Cho WK, Spille JH, Hecht M et al. Mediator and RNA polymerase II clusters associate in transcription-dependent condensates. Science (New York, N.Y.). 2018. PMID 29930094. DOI: 10.1126/science.aar4199. Access: abstract_metadata. Study: PMID29930094.

4. Lee JH, Wang R, Xiong F et al. Enhancer RNA m6A methylation facilitates transcriptional condensate formation and gene activation. Molecular cell. 2021. PMID 34375583. DOI: 10.1016/j.molcel.2021.07.024. Access: abstract_metadata. Study: PMID34375583.

5. Wu Z, Pope SD, Ahmed NS et al. Regulation of inflammatory responses by pH-dependent transcriptional condensates. Cell. 2025. PMID 40680734. DOI: 10.1016/j.cell.2025.06.033. Access: full_text_partial. Study: PMID40680734.

6. Du M, Stitzinger SH, Spille JH et al. Direct observation of a condensate effect on super-enhancer controlled gene bursting. Cell. 2024. PMID 38194964. DOI: 10.1016/j.cell.2023.12.005. Access: abstract_metadata. Study: PMID38194964.

7. Yang J, Chung CI, Koach J et al. MYC phase separation selectively modulates the transcriptome. Nature structural & molecular biology. 2024. PMID 38811792. DOI: 10.1038/s41594-024-01322-6. Access: full_text_partial. Study: PMID38811792.

8. Trojanowski J, Frank L, Rademacher A et al. Transcription activation is enhanced by multivalent interactions independent of phase separation. Molecular cell. 2022. PMID 35537448. DOI: 10.1016/j.molcel.2022.04.017. Access: full_text_partial. Study: PMID35537448.

9. Wei MT, Chang YC, Shimobayashi SF et al. Nucleated transcriptional condensates amplify gene expression. Nature cell biology. 2020. PMID 32929202. DOI: 10.1038/s41556-020-00578-6. Access: abstract_metadata. Study: PMID32929202.

10. Asimi V, Sampath Kumar A, Niskanen H et al. Hijacking of transcriptional condensates by endogenous retroviruses. Nature genetics. 2022. PMID 35864192. DOI: 10.1038/s41588-022-01132-w. Access: abstract_metadata. Study: PMID35864192.

11. Shrinivas K, Sabari BR, Coffey EL et al. Enhancer Features that Drive Formation of Transcriptional Condensates. Molecular cell. 2019. PMID 31398323. DOI: 10.1016/j.molcel.2019.07.009. Access: abstract_metadata. Study: PMID31398323.

12. Kent S, Brown K, Yang CH et al. Phase-Separated Transcriptional Condensates Accelerate Target-Search Process Revealed by Live-Cell Single-Molecule Imaging. Cell reports. 2020. PMID 33053359. DOI: 10.1016/j.celrep.2020.108248. Access: abstract_metadata. Study: PMID33053359.

13. Huoh YS, Zhang Q, Törner R et al. Mechanism for controlled assembly of transcriptional condensates by Aire. Nature immunology. 2024. PMID 39169234. DOI: 10.1038/s41590-024-01922-w. Access: abstract_metadata. Study: PMID39169234.

14. Gan P, Eppert M, De La Cruz N et al. Coactivator condensation drives cardiovascular cell lineage specification. Science advances. 2024. PMID 38489358. DOI: 10.1126/sciadv.adk7160. Access: full_text_partial. Study: PMID38489358.

15. Rawat P, Boehning M, Hummel B et al. Stress-induced nuclear condensation of NELF drives transcriptional downregulation. Molecular cell. 2021. PMID 33548202. DOI: 10.1016/j.molcel.2021.01.016. Access: full_text_partial. Study: PMID33548202.

16. Gibson BA, Doolittle LK, Schneider MWG et al. Organization of Chromatin by Intrinsic and Regulated Phase Separation. Cell. 2019. PMID 31543265. DOI: 10.1016/j.cell.2019.08.037. Access: abstract_metadata. Study: PMID31543265.

17. Basu S, Martínez-Cristóbal P, Frigolé-Vivas M et al. Rational optimization of a transcription factor activation domain inhibitor. Nature structural & molecular biology. 2023. PMID 38049566. DOI: 10.1038/s41594-023-01159-5. Access: abstract_metadata. Study: PMID38049566.

18. Zamudio AV, Dall'Agnese A, Henninger JE et al. Mediator Condensates Localize Signaling Factors to Key Cell Identity Genes. Molecular cell. 2019. PMID 31563432. DOI: 10.1016/j.molcel.2019.08.016. Access: abstract_metadata. Study: PMID31563432.

19. Chowdhary S, Kainth AS, Paracha S et al. Inducible transcriptional condensates drive 3D genome reorganization in the heat shock response. Molecular cell. 2022. PMID 36327976. DOI: 10.1016/j.molcel.2022.10.013. Access: abstract_metadata. Study: PMID36327976.

20. Yamamoto T, Sakaue T, Schiessel H. Slow chromatin dynamics enhances promoter accessibility to transcriptional condensates. Nucleic acids research. 2021. PMID 33885786. DOI: 10.1093/nar/gkab275. Access: abstract_metadata. Study: PMID33885786.

21. Sabari BR, Dall'Agnese A, Boija A et al. Coactivator condensation at super-enhancers links phase separation and gene control. Science (New York, N.Y.). 2018. PMID 29930091. DOI: 10.1126/science.aar3958. Access: full_text_partial. Study: PMID29930091.

22. Boija A, Klein IA, Sabari BR et al. Transcription Factors Activate Genes through the Phase-Separation Capacity of Their Activation Domains. Cell. 2018. PMID 30449618. DOI: 10.1016/j.cell.2018.10.042. Access: abstract_metadata. Study: PMID30449618.

23. Selig EE, Sohn EJ, Stoja A et al. Phase separation of the oncogenic fusion protein EWS::FLI1 is modulated by its DNA-binding domain. Proceedings of the National Academy of Sciences of the United States of America. 2025. PMID 40377985. DOI: 10.1073/pnas.2221823122. Access: abstract_metadata. Study: PMID40377985.

24. Shao W, Bi X, Pan Y et al. Phase separation of RNA-binding protein promotes polymerase binding and transcription. Nature chemical biology. 2022. PMID 34916619. DOI: 10.1038/s41589-021-00904-5. Access: abstract_metadata. Study: PMID34916619.

25. Collombet S, Rall I, Dugast-Darzacq C et al. RNA polymerase II depletion from the inactive X chromosome territory is not mediated by physical compartmentalization. Nature structural & molecular biology. 2023. PMID 37291424. DOI: 10.1038/s41594-023-01008-5. Access: abstract_metadata. Study: PMID37291424.

26. Chandra B, Michmerhuizen NL, Shirnekhi HK et al. Phase Separation Mediates NUP98 Fusion Oncoprotein Leukemic Transformation. Cancer discovery. 2022. PMID 34903620. DOI: 10.1158/2159-8290.CD-21-0674. Access: abstract_metadata. Study: PMID34903620.

27. Lyons H, Veettil RT, Pradhan P et al. Functional partitioning of transcriptional regulators by patterned charge blocks. Cell. 2023. PMID 36603581. DOI: 10.1016/j.cell.2022.12.013. Access: abstract_metadata. Study: PMID36603581.

28. Cai D, Feliciano D, Dong P et al. Phase separation of YAP reorganizes genome topology for long-term YAP target gene expression. Nature cell biology. 2019. PMID 31792379. DOI: 10.1038/s41556-019-0433-z. Access: abstract_metadata. Study: PMID31792379.

29. Foran G, Hallam RD, Megaly M et al. Notch1 Phase Separation Coupled Percolation facilitates target gene expression and enhancer looping. Scientific reports. 2024. PMID 39300145. DOI: 10.1038/s41598-024-71634-6. Access: full_text_partial. Study: PMID39300145.

30. Chong S, Dugast-Darzacq C, Liu Z et al. Imaging dynamic and selective low-complexity domain interactions that control gene transcription. Science (New York, N.Y.). 2018. PMID 29930090. DOI: 10.1126/science.aar2555. Access: full_text_partial. Study: EWS-native-hub-programme.

31. Gerber A, Roeder RG. The CTD Is Not Essential for the Post-Initiation Control of RNA Polymerase II Activity. Journal of molecular biology. 2020. PMID 32707132. DOI: 10.1016/j.jmb.2020.07.010. Access: abstract_metadata. Study: PMID32707132.

32. Zhu T, Li H, Mao S et al. Endogenous YAP/TAZ partitioning in TEAD condensates orchestrates the Hippo response. Molecular cell. 2025. PMID 40914169. DOI: 10.1016/j.molcel.2025.08.014. Access: abstract_metadata. Study: PMID40914169.

33. Gao L, Gao Q, Hai N et al. Phase separation of DUX family proteins drives totipotent-like state via 3D genome reorganization and retrotransposon activation. Protein & cell. 2026. PMID 41832971. DOI: 10.1093/procel/pwag014. Access: abstract_metadata. Study: PMID41832971.

34. Hao S, Lee YJ, Benhamou Goldfajn N et al. YAP condensates are highly organized hubs. iScience. 2024. PMID 38784009. DOI: 10.1016/j.isci.2024.109927. Access: abstract_metadata. Study: PMID38784009.

35. Kribelbauer-Swietek JF, Pushkarev O, Gardeux V et al. Context transcription factors establish cooperative environments and mediate enhancer communication. Nature genetics. 2024. PMID 39363017. DOI: 10.1038/s41588-024-01892-7. Access: abstract_metadata. Study: PMID39363017.

36. Meduri R, Rubio LS, Mohajan S et al. Phase-separation antagonists potently inhibit transcription and broadly increase nucleosome density. The Journal of biological chemistry. 2022. PMID 35963432. DOI: 10.1016/j.jbc.2022.102365. Access: abstract_metadata. Study: PMID35963432.

37. Li M, Gong H, Wen C et al. GATA1 Drives Human Erythropoiesis via Liquid-Liquid Phase Separation. FASEB journal : official publication of the Federation of American Societies for Experimental Biology. 2026. PMID 41693617. DOI: 10.1096/fj.202503628R. Access: abstract_metadata. Study: PMID41693617.

38. Imada T, Shimi T, Kaiho A et al. RNA polymerase II condensate formation and association with Cajal and histone locus bodies in living human cells. Genes to cells : devoted to molecular & cellular mechanisms. 2021. PMID 33608942. DOI: 10.1111/gtc.12840. Access: abstract_metadata. Study: PMID33608942.

39. Xie P, Xie X, Ye C et al. Mammalian circadian clock proteins form dynamic interacting microbodies distinct from phase separation. Proceedings of the National Academy of Sciences of the United States of America. 2023. PMID 38127982. DOI: 10.1073/pnas.2318274120. Access: abstract_metadata. Study: PMID38127982.

40. Mao S, Wu R, Luo W et al. Spuriously transcribed RNAs from CRISPR-sgRNA expression plasmids scaffold biomolecular condensate formation and hamper accurate genomic imaging. Nucleic acids research. 2025. PMID 40119729. DOI: 10.1093/nar/gkaf192. Access: abstract_metadata. Study: PMID40119729.

41. Shin Y, Chang YC, Lee DSW et al. Liquid Nuclear Condensates Mechanically Sense and Restructure the Genome. Cell. 2018. PMID 30500535. DOI: 10.1016/j.cell.2018.10.057. Access: abstract_metadata. Study: PMID30500535.

42. Ide S, Imai R, Ochi H et al. Transcriptional suppression of ribosomal DNA with phase separation. Science advances. 2020. PMID 33055158. DOI: 10.1126/sciadv.abb5953. Access: abstract_metadata. Study: PMID33055158.

43. Li Z, Burgos-Bravo F, Xu K et al. Phase-separated NDF-FACT condensates facilitate transcription elongation on chromatin. Nature cell biology. 2025. PMID 41028835. DOI: 10.1038/s41556-025-01778-8. Access: abstract_metadata. Study: PMID41028835.

44. Chong S, Graham TGW, Dugast-Darzacq C et al. Tuning levels of low-complexity domain interactions to modulate endogenous oncogenic transcription. Molecular cell. 2022. PMID 35483357. DOI: 10.1016/j.molcel.2022.04.007. Access: abstract_metadata. Study: EWS-native-hub-programme-followup.

45. Zuo L, Zhang G, Massett M et al. Loci-specific phase separation of FET fusion oncoproteins promotes gene transcription. Nature communications. 2021. PMID 33674598. DOI: 10.1038/s41467-021-21690-7. Access: abstract_metadata. Study: PMID33674598.

46. Davis RB, Kaur T, Moosa MM et al. FUS oncofusion protein condensates recruit mSWI/SNF chromatin remodeler via heterotypic interactions between prion-like domains. Protein science : a publication of the Protein Society. 2021. PMID 34018649. DOI: 10.1002/pro.4127. Access: abstract_metadata. Study: PMID34018649.

47. Schneider N, Wieland FG, Kong D et al. Liquid-liquid phase separation of light-inducible transcription factors increases transcription activation in mammalian cells and mice. Science advances. 2021. PMID 33523844. DOI: 10.1126/sciadv.abd3568. Access: abstract_metadata. Study: PMID33523844.

48. Yoshida S, Zhong Y, Kumagai A et al. Dynamic Regulation of Endogenous Transcription Factor Hubs at Single-Molecule Resolution. Advanced science (Weinheim, Baden-Wurttemberg, Germany). 2026. PMID 42417395. DOI: 10.1002/advs.202600048. Access: full_text_partial. Study: EWS-native-hub-programme-2026.

49. Gao W, Zhu L, Wei Y et al. BMAL1 regulates circadian rhythms via phase separation-mediated transcriptional hub formation. Signal transduction and targeted therapy. 2026. PMID 42062246. DOI: 10.1038/s41392-026-02711-7. Access: abstract_metadata. Study: PMID42062246.

50. Ahn JH, Davis ES, Daugird TA et al. Phase separation drives aberrant chromatin looping and cancer development. Nature. 2021. PMID 34163069. DOI: 10.1038/s41586-021-03662-5. Access: abstract_metadata. Study: PMID34163069.

51. Stortz M, Oses C, Vázquez Echegaray C et al. SOX2 Modulates the Nuclear Organization and Transcriptional Activity of the Glucocorticoid Receptor. Journal of molecular biology. 2022. PMID 36309135. DOI: 10.1016/j.jmb.2022.167869. Access: abstract_metadata. Study: PMID36309135.

52. Xu M, Jiang SY, Tang S et al. Nuclear SREBP2 condensates regulate the transcriptional activation of lipogenic genes and cholesterol homeostasis. Nature metabolism. 2025. PMID 40394324. DOI: 10.1038/s42255-025-01291-0. Access: abstract_metadata. Study: PMID40394324.

53. McSwiggen DT, Hansen AS, Teves SS et al. Evidence for DNA-mediated nuclear compartmentalization distinct from phase separation. eLife. 2019. PMID 31038454. DOI: 10.7554/eLife.47098. Access: abstract_metadata. Study: PMID31038454.

54. Boehning M, Dugast-Darzacq C, Rankovic M et al. RNA polymerase II clustering through carboxy-terminal domain phase separation. Nature structural & molecular biology. 2018. PMID 30127355. DOI: 10.1038/s41594-018-0112-y. Access: abstract_metadata. Study: PMID30127355.

55. Erdogdu NU, Guhathakurta S, Oellers R et al. Histone acetylation-dependent clustering of BRD2 instructs transcription dynamics. Nature genetics. 2026. PMID 41957181. DOI: 10.1038/s41588-026-02533-x. Access: abstract_metadata. Study: PMID41957181.

56. Cho WK, Jayanth N, English BP et al. RNA Polymerase II cluster dynamics predict mRNA output in living cells. eLife. 2016. PMID 27138339. DOI: 10.7554/eLife.13617. Access: abstract_metadata. Study: PolII-live-cluster-platform-beta-actin.

57. Düster R, Kaltheuner IH, Schmitz M et al. 1,6-Hexanediol, commonly used to dissolve liquid-liquid phase separated condensates, directly impairs kinase and phosphatase activities. The Journal of biological chemistry. 2021. PMID 33814344. DOI: 10.1016/j.jbc.2021.100260. Access: abstract_metadata. Study: PMID33814344.

58. Pancholi A, Klingberg T, Zhang W et al. RNA polymerase II clusters form in line with surface condensation on regulatory chromatin. Molecular systems biology. 2021. PMID 34569155. DOI: 10.15252/msb.202110272. Access: abstract_metadata. Study: PMID34569155.

59. Callan-Sidat A, Zewdu E, Cavallaro M et al. N-terminal tagging of RNA Polymerase II shapes transcriptomes more than C-terminal alterations. iScience. 2024. PMID 38799575. DOI: 10.1016/j.isci.2024.109914. Access: abstract_metadata. Study: PMID38799575.

60. Cisse II, Izeddin I, Causse SZ et al. Real-time dynamics of RNA polymerase II clustering in live human cells. Science (New York, N.Y.). 2013. PMID 23828889. DOI: 10.1126/science.1239053. Access: abstract_metadata. Study: PolII-live-cluster-platform.

61. Ling YH, Ye Z, Liang C et al. Disordered C-terminal domain drives spatiotemporal confinement of RNAPII to enhance search for chromatin targets. Nature cell biology. 2024. PMID 38548891. DOI: 10.1038/s41556-024-01382-2. Access: abstract_metadata. Study: PMID38548891.

62. Chen L, Zhang Z, Han Q et al. Hormone-induced enhancer assembly requires an optimal level of hormone receptor multivalent interactions. Molecular cell. 2023. PMID 37738977. DOI: 10.1016/j.molcel.2023.08.027. Access: abstract_metadata. Study: PMID37738977.

63. Sozańska N, Klepka BP, Niedzwiecka A et al. Intrinsically disordered bHLH family member TCF4 drives phase separation in a DNA-dependent manner. Archives of biochemistry and biophysics. 2026. PMID 42106138. DOI: 10.1016/j.abb.2026.110829. Access: abstract_metadata. Study: PMID42106138.

64. Bremer A, Lang WH, Kempen RP et al. Reconciling competing models on the roles of condensates and soluble complexes in transcription factor function. Molecular cell. 2025. PMID 40614721. DOI: 10.1016/j.molcel.2025.06.008. Access: full_text_partial. Study: PMID40614721.

Source-specific evidence cards

[1] RNA-Mediated Feedback Control of Transcriptional Condensates.

Shows: Low RNA promotes and high RNA dissolves electrostatic condensates with transcription feedback.

Does not show: RNA changes binding and transcription chemistry; endogenous phase-selective necessity is not isolated.

Access/status: Complete primary abstract and linked-status metadata examined. No complete-paper or supplement audit claimed. none_found. Subclaims: SC1, SC2, SC3.

[2] Unblending of Transcriptional Condensates in Human Repeat Expansion Disease.

Shows: HOXD13 repeat expansion changes co-condensation and cell-specific developmental transcription.

Does not show: Repeat changes also alter interactions/aggregation; human variant effects do not uniquely attribute liquid state.

Access/status: Complete primary abstract and linked-status metadata examined. No complete-paper or supplement audit claimed. none_found. Subclaims: SC1, SC2.

[3] Mediator and RNA polymerase II clusters associate in transcription-dependent condensates.

Shows: Mediator and Pol II form dynamic clusters that colocalize and respond to transcription inhibitors.

Does not show: Puncta, dynamics and inhibitor sensitivity do not establish phase boundary or phase-state necessity.

Access/status: Complete primary abstract and linked-status metadata examined. No complete-paper or supplement audit claimed. none_found. Subclaims: SC1, SC2.

[4] Enhancer RNA m6A methylation facilitates transcriptional condensate formation and gene activation.

Shows: m6A-eRNAs recruit YTHDC1/BRD4 and depletion reduces enhancer activation.

Does not show: Reader depletion and IDR/arginine changes also affect RNA/cofactor binding; state specificity remains.

Access/status: Complete primary abstract and linked-status metadata examined. No complete-paper or supplement audit claimed. none_found. Subclaims: SC1, SC2.

[5] Regulation of inflammatory responses by pH-dependent transcriptional condensates.

Shows: Acidic pH disrupts BRD4/MED1 assemblies and changes inflammatory output through histidine-rich regions.

Does not show: Stable protein abundance, isolated nuclei, reversible pH and functional segment mutants narrow generic collapse; protonation affects molecular affinities and phase together. FRAP/alcohol are not specific state diagnoses.

Access/status: Complete primary abstract and linked-status metadata examined. Selected body sections: 1:Results:paragraph7, 1:Results:paragraph12, 1:Results:paragraph13, 1:Results:paragraph14, 1:Results:paragraph15. No complete-paper or supplement audit claimed. none_found. Subclaims: SC1, SC2.

[6] Direct observation of a condensate effect on super-enhancer controlled gene bursting.

Shows: Nearby condensate enhances burst size/frequency while basal output persists when distant; local DNA/cohesin perturbations affect enhancement.

Does not show: Assembly position is not itself liquid diagnosis; DNA/protein perturbations affect other functions. Erratum linked, body audit pending.

Access/status: Complete primary abstract and linked-status metadata examined. No complete-paper or supplement audit claimed. ErratumIn: PMID38614098. Subclaims: SC1, SC2, SC3.

[7] MYC phase separation selectively modulates the transcriptome.

Shows: Matched N-myc levels show phase separation further modulates fewer than three percent of regulated genes with proliferation effects.

Does not show: Engineered state/overexpression and small affected set constrain broad primacy; correction linked, body audit pending.

Access/status: Complete primary abstract and linked-status metadata examined. Selected body sections: 1:RESULTS:paragraph14, 1:RESULTS:paragraph18, 1:RESULTS:paragraph20, 1:RESULTS:paragraph21, 1:RESULTS:paragraph22, 1:RESULTS:paragraph23. No complete-paper or supplement audit claimed. ErratumIn: PMID38867115. Subclaims: SC1, SC2, SC3.

[8] Transcription activation is enhanced by multivalent interactions independent of phase separation.

Shows: Multivalent interactions enhance occupancy/residence/cofactor recruitment; droplets are neutral or inhibitory for output.

Does not show: Reporter and synthetic platform cannot refute all endogenous contexts; it separates assembly benefits from droplet state.

Access/status: Complete primary abstract and linked-status metadata examined. Selected body sections: Scite body characters 0-8000, Scite body characters 16000-24000, Scite body characters 32000-40000. No complete-paper or supplement audit claimed. none_found. Subclaims: SC1, SC2, SC3.

[9] Nucleated transcriptional condensates amplify gene expression.

Shows: CTD clusters nucleate engineered TAF15 condensates and amplify transcription.

Does not show: Optogenetic concentration and FET interactions establish capacity, not endogenous universal need.

Access/status: Complete primary abstract and linked-status metadata examined. No complete-paper or supplement audit claimed. none_found. Subclaims: SC1, SC2.

[10] Hijacking of transcriptional condensates by endogenous retroviruses.

Shows: TRIM28 degradation shifts assemblies to ERVs; RNA knockdown or TF overexpression rescues localization.

Does not show: TRIM28/RNA/TF changes alter gene regulation directly; condensate relocalization is not selective liquid-state intervention.

Access/status: Complete primary abstract and linked-status metadata examined. No complete-paper or supplement audit claimed. none_found. Subclaims: SC1, SC2.

[11] Enhancer Features that Drive Formation of Transcriptional Condensates.

Shows: TF site density/affinity thresholds support cooperative condensation and enhancer output.

Does not show: DNA binding and recruitment share thresholds; phase-specific endogenous output not isolated.

Access/status: Complete primary abstract and linked-status metadata examined. No complete-paper or supplement audit claimed. none_found. Subclaims: SC1, SC2.

[12] Phase-Separated Transcriptional Condensates Accelerate Target-Search Process Revealed by Live-Cell Single-Molecule Imaging.

Shows: CBX2-DNA interactions nucleate assemblies and accelerate repeated target search.

Does not show: DNA binding and rebinding can produce hubs; repression differs from broad active-locus primacy.

Access/status: Complete primary abstract and linked-status metadata examined. No complete-paper or supplement audit claimed. none_found. Subclaims: SC1, SC2.

[13] Mechanism for controlled assembly of transcriptional condensates by Aire.

Shows: Aire polymerization, histone binding and coactivator recruitment balance enhancer assemblies and transcription.

Does not show: Controlled polymerization is a network assembly mechanism; liquid demixing and functional specificity are separate.

Access/status: Complete primary abstract and linked-status metadata examined. No complete-paper or supplement audit claimed. none_found. Subclaims: SC1, SC2.

[14] Coactivator condensation drives cardiovascular cell lineage specification.

Shows: Endogenous MYOCD knock-in differentiation shows hubs at measured native concentrations; matched-expression aromatic IDR replacements rescue reprogramming and preserve DNA occupancy.

Does not show: CDT1 replacement restores condensates and high DNA occupancy but not p300 recruitment, acetylation or output; chemistry and composition matter, while liquid state alone remains unisolated.

Access/status: Complete primary abstract and linked-status metadata examined. Selected body sections: 1:RESULTS:paragraph4, 1:RESULTS:paragraph6, 1:RESULTS:paragraph14, 1:RESULTS:paragraph18, 1:RESULTS:paragraph19, 1:RESULTS:paragraph22, 1:RESULTS:paragraph25. No complete-paper or supplement audit claimed. none_found. Subclaims: SC1, SC2.

[15] Stress-induced nuclear condensation of NELF drives transcriptional downregulation.

Shows: NELF IDR replacement with FUS/EWS supports stress-dependent promoter recruitment and downregulation.

Does not show: Nuclear localization, complex incorporation and basal promoter binding are preserved after IDR deletion; stress-dependent occupancy changes. FUS/EWS replacements constitutively repress and improve survival, but chemical recruitment and material state remain coupled.

Access/status: Complete primary abstract and linked-status metadata examined. Selected body sections: 1:Results:paragraph58, 1:Results:paragraph71. No complete-paper or supplement audit claimed. none_found. Subclaims: SC1, SC2.

[16] Organization of Chromatin by Intrinsic and Regulated Phase Separation.

Shows: Histone tails, H1/linker length and acetylation tune chromatin droplets; acetylated chromatin co-condenses with BRD4.

Does not show: Microinjected/reconstituted polymer capacity is not endogenous active transcription liquid-state necessity.

Access/status: Complete primary abstract and linked-status metadata examined. No complete-paper or supplement audit claimed. none_found. Subclaims: SC1.

[17] Rational optimization of a transcription factor activation domain inhibitor.

Shows: AD-directed optimized compounds reduce AR transcription and tumour effects with target affinity.

Does not show: AD interaction/nuclear entry effects accompany condensation; drug efficacy is not a phase-only perturbation.

Access/status: Complete primary abstract and linked-status metadata examined. No complete-paper or supplement audit claimed. none_found. Subclaims: SC1, SC2.

[18] Mediator Condensates Localize Signaling Factors to Key Cell Identity Genes.

Shows: Signalling-factor IDRs partition with Mediator and beta-catenin occupies selected enhancers through multiple interactions.

Does not show: DNA-binding and cofactor interactions can localize signals without unique phase necessity.

Access/status: Complete primary abstract and linked-status metadata examined. No complete-paper or supplement audit claimed. none_found. Subclaims: SC1, SC2.

[19] Inducible transcriptional condensates drive 3D genome reorganization in the heat shock response.

Shows: Hsf1/Hsp70/IDR mutations alter clusters, intergenic coalescence and fitness.

Does not show: Stress and protein binding changes remain coupled; networked clustering need not diagnose classical isotropic LLPS.

Access/status: Complete primary abstract and linked-status metadata examined. No complete-paper or supplement audit claimed. none_found. Subclaims: SC1, SC2.

[20] Slow chromatin dynamics enhances promoter accessibility to transcriptional condensates.

Shows: Model predicts extrusion-dependent promoter contact probabilities from linker relaxation.

Does not show: Original theoretical research, not empirical phase-state or transcription necessity; assumptions require tests.

Access/status: Complete primary abstract and linked-status metadata examined. No complete-paper or supplement audit claimed. none_found. Subclaims: SC1, SC2.

[21] Coactivator condensation at super-enhancers links phase separation and gene control.

Shows: BRD4/MED1 puncta at super-enhancers, chemical sensitivity and IDR droplets support condensation capacity.

Does not show: Puncta, FRAP, hexanediol and extract compartmentalization cannot isolate endogenous liquid-state necessity.

Access/status: Complete primary abstract and linked-status metadata examined. Selected body sections: 3:Coactivator puncta and super-enhancer occupancy are sensitive to condensate perturbation, 5:MED1-IDR participates in liquid-liquid phase separation in cells, 6:Conserved serine bias in the MED1-IDR is necessary for phase separation, 7:MED1-IDR droplets can incorporate proteins necessary for transcription. No complete-paper or supplement audit claimed. none_found. Subclaims: SC1, SC2.

[22] Transcription Factors Activate Genes through the Phase-Separation Capacity of Their Activation Domains.

Shows: AD residues jointly control droplet propensity and transcriptional activity.

Does not show: Same residues mediate cofactor affinity; correlated losses do not separate phase from binding.

Access/status: Complete primary abstract and linked-status metadata examined. No complete-paper or supplement audit claimed. none_found. Subclaims: SC1, SC2.

[23] Phase separation of the oncogenic fusion protein EWS::FLI1 is modulated by its DNA-binding domain.

Shows: FLI1/ETS DNA-binding domains enhance EWS assembly, whereas DNA binding suppresses these interactions and condensation.

Does not show: Biophysical reconstitution, not native transcription necessity; shows DBD and LCD functions are intertwined.

Access/status: Complete primary abstract and linked-status metadata examined. No complete-paper or supplement audit claimed. none_found. Subclaims: SC1, SC2.

[24] Phase separation of RNA-binding protein promotes polymerase binding and transcription.

Shows: PSPC1 links RNA/Pol II engagement; acute degradation impairs polymerase binding and nascent transcription.

Does not show: PSPC1 RNA binding and polymerase chemistry change with loss; binding-matched phase rescue must be demonstrated.

Access/status: Complete primary abstract and linked-status metadata examined. No complete-paper or supplement audit claimed. none_found. Subclaims: SC1, SC2.

[25] RNA polymerase II depletion from the inactive X chromosome territory is not mediated by physical compartmentalization.

Shows: Pol II enters Xist territory; apparent depletion reflects loss of chromatin-bound fraction.

Does not show: Inactive-X initiation context differs from active loci; disproves a proposed diffusion barrier, not every phase assembly.

Access/status: Complete primary abstract and linked-status metadata examined. No complete-paper or supplement audit claimed. none_found. Subclaims: SC1, SC2, SC3.

[26] Phase Separation Mediates NUP98 Fusion Oncoprotein Leukemic Transformation.

Shows: Homotypic/heterotypic interaction changes affect puncta, aberrant transcription and transformation.

Does not show: Oncofusion and ectopic construct scope; interacting residues and chromatin binding must remain matched for LLPS attribution.

Access/status: Complete primary abstract and linked-status metadata examined. No complete-paper or supplement audit claimed. none_found. Subclaims: SC1, SC2.

[27] Functional partitioning of transcriptional regulators by patterned charge blocks.

Shows: Patterned charge blocks determine functional partitioning and gene activation during transition.

Does not show: Charge sequence tunes affinity and RNA/protein binding as well as phase; selective liquid property not fully disentangled.

Access/status: Complete primary abstract and linked-status metadata examined. No complete-paper or supplement audit claimed. none_found. Subclaims: SC1, SC2.

[28] Phase separation of YAP reorganizes genome topology for long-term YAP target gene expression.

Shows: YAP assemblies recruit TEAD/TAZ and eventually Pol II; AD deletion suppresses assembly and signalling.

Does not show: Osmotic crowding and activation-domain chemistry accompany assembly; physiological unstressed scope differs.

Access/status: Complete primary abstract and linked-status metadata examined. No complete-paper or supplement audit claimed. none_found. Subclaims: SC1, SC2.

[29] Notch1 Phase Separation Coupled Percolation facilitates target gene expression and enhancer looping.

Shows: NICD forms transcription assemblies and enhancer loops, with phase separation coupled to percolation model.

Does not show: TAD changes affect activation surfaces; viscoelastic network and classical liquid demixing are distinct propositions. Final replaces preprint 39131356.

Access/status: Complete primary abstract and linked-status metadata examined. Selected body sections: 1:Results:paragraph3, 1:Results:paragraph19, 1:Results:paragraph20, 1:Results:paragraph23, 1:Results:paragraph25. No complete-paper or supplement audit claimed. UpdateOf: PMID39131356. Subclaims: SC1, SC2.

[30] Imaging dynamic and selective low-complexity domain interactions that control gene transcription.

Shows: Dynamic LCD hubs stabilize DNA binding and activate transcription without detectable phase separation at physiological abundance.

Does not show: Native hub interactions are real; lack of detected bulk LLPS cannot exclude subresolution or coupled transitions.

Access/status: Complete primary abstract and linked-status metadata examined. Selected body sections: 4:EWS/FLI1 forms hubs at endogenous GGAA microsatellites, 5:Dynamic EWS LCD-LCD interactions mediate formation of EWS/FLI1 hubs, 6:EWS LCD-LCD interactions are essential for transcription and transformation functions of EWS/FLI1. No complete-paper or supplement audit claimed. none_found. Subclaims: SC1, SC2, SC3.

[31] The CTD Is Not Essential for the Post-Initiation Control of RNA Polymerase II Activity.

Shows: Post-initiation CTD ablation preserves pause maintenance and productive release.

Does not show: Stage-specific test does not refute CTD initiation function or all phase pathways.

Access/status: Complete primary abstract and linked-status metadata examined. No complete-paper or supplement audit claimed. none_found. Subclaims: SC2, SC3.

[32] Endogenous YAP/TAZ partitioning in TEAD condensates orchestrates the Hippo response.

Shows: YAP/TAZ are TEAD-recruited clients and inhibitor effects depend on VGLL4.

Does not show: Interaction loss explains disappearance; endogenous morphology does not independently isolate phase state.

Access/status: Complete primary abstract and linked-status metadata examined. No complete-paper or supplement audit claimed. none_found. Subclaims: SC1, SC2.

[33] Phase separation of DUX family proteins drives totipotent-like state via 3D genome reorganization and retrotransposon activation.

Shows: Homeobox-IDR interactions assemble coactivators and influence target transcription and developmental phenotypes.

Does not show: R70A and homeobox changes may alter DNA or partner binding; phase-specific necessity at native concentrations is unresolved.

Access/status: Complete primary abstract and linked-status metadata examined. No complete-paper or supplement audit claimed. none_found. Subclaims: SC1, SC2.

[34] YAP condensates are highly organized hubs.

Shows: Hippo signaling and nuclear mechanics regulate YAP hubs containing TEAD and BRD4.

Does not show: Puncta and slow diffusion do not diagnose LLPS; upstream mechanics and binding change together; linked erratum body not examined.

Access/status: Complete primary abstract and linked-status metadata examined. No complete-paper or supplement audit claimed. ErratumIn: PMID41550760. Subclaims: SC1, SC2.

[35] Context transcription factors establish cooperative environments and mediate enhancer communication.

Shows: Distinct TF classes depend on local enhancer context and respond differently to bromodomain inhibition.

Does not show: Causal enhancer syntax is shown; liquid-state necessity is not measured and condensates remain one possible implementation.

Access/status: Complete primary abstract and linked-status metadata examined. No complete-paper or supplement audit claimed. none_found. Subclaims: SC2, SC3.

[36] Phase-separation antagonists potently inhibit transcription and broadly increase nucleosome density.

Shows: Both 1,6- and 2,5-hexanediol suppress coalescing and noncoalescing gene transcription and alter histone occupancy.

Does not show: Alcohol is pleiotropic and 2,5-hexanediol is not a reliable inert negative control; result does not exclude specific condensate functions.

Access/status: Complete primary abstract and linked-status metadata examined. No complete-paper or supplement audit claimed. none_found. Subclaims: SC1, SC2.

[37] GATA1 Drives Human Erythropoiesis via Liquid-Liquid Phase Separation.

Shows: IDR2 variants impair puncta, transcription and erythroid differentiation.

Does not show: The same variants disrupt chromatin occupancy and FOG1/LMO2 assembly, precluding a phase-only inference.

Access/status: Complete primary abstract and linked-status metadata examined. No complete-paper or supplement audit claimed. none_found. Subclaims: SC1, SC2.

[38] RNA polymerase II condensate formation and association with Cajal and histone locus bodies in living human cells.

Shows: Pol II clusters associate with histone locus bodies during S phase.

Does not show: Coilin loss preserves key clustering kinetics; FRAP and alcohol sensitivity do not establish LLPS or its causal necessity.

Access/status: Complete primary abstract and linked-status metadata examined. No complete-paper or supplement audit claimed. none_found. Subclaims: SC1.

[39] Mammalian circadian clock proteins form dynamic interacting microbodies distinct from phase separation.

Shows: Native circadian PER2 assemblies differ from overexpression-induced liquid condensates and resist several liquid-state probes.

Does not show: PER2 negative-feedback assemblies are not the same as positive-limb BMAL1 hubs; this is not a universal clock-protein negative result.

Access/status: Complete primary abstract and linked-status metadata examined. No complete-paper or supplement audit claimed. UpdateOf: PMID37961341. Subclaims: SC1, SC3.

[40] Spuriously transcribed RNAs from CRISPR-sgRNA expression plasmids scaffold biomolecular condensate formation and hamper accurate genomic imaging.

Shows: Cryptic plasmid-backbone transcripts generate misleading liquid-like structures and paraspeckle protein recruitment.

Does not show: A method-specific artifact does not explain every native transcription hub; RNA scaffold controls are essential.

Access/status: Complete primary abstract and linked-status metadata examined. No complete-paper or supplement audit claimed. none_found. Subclaims: SC1.

[41] Liquid Nuclear Condensates Mechanically Sense and Restructure the Genome.

Shows: Light-induced IDP droplets restructure chromatin and connect genomic targets.

Does not show: Engineered concentration and scaffolding establish capacity rather than endogenous regulatory necessity; correction linked but not body-audited.

Access/status: Complete primary abstract and linked-status metadata examined. No complete-paper or supplement audit claimed. ErratumIn: PMID30849377. Subclaims: SC1, SC2.

[42] Transcriptional suppression of ribosomal DNA with phase separation.

Shows: Condensation-associated inactive caps exclude initiation factors and suppress transcription.

Does not show: Pol I differs from Pol II; condensates may activate or inhibit depending on composition and reaction stage.

Access/status: Complete primary abstract and linked-status metadata examined. No complete-paper or supplement audit claimed. none_found. Subclaims: SC1, SC3.

[43] Phase-separated NDF-FACT condensates facilitate transcription elongation on chromatin.

Shows: NDF-FACT promotes rapid nucleosomal elongation and loss of NDF impairs transcription and differentiation.

Does not show: Protein loss also removes biochemical chaperone interactions; a phase-specific matched rescue is not established by the examined abstract.

Access/status: Complete primary abstract and linked-status metadata examined. No complete-paper or supplement audit claimed. none_found. Subclaims: SC1, SC2.

[44] Tuning levels of low-complexity domain interactions to modulate endogenous oncogenic transcription.

Shows: Excess self-association toward macroscopic LLPS represses target transcription and transformation.

Does not show: Variants change localization and interaction chemistry; does not eliminate a role for small native hubs.

Access/status: Complete primary abstract and linked-status metadata examined. No complete-paper or supplement audit claimed. none_found. Subclaims: SC1, SC2, SC3.

[45] Loci-specific phase separation of FET fusion oncoproteins promotes gene transcription.

Shows: DNA density and fusion domains support condensate formation and Pol II recruitment.

Does not show: Fusion and DNA scaffold conditions do not establish native phase-only necessity.

Access/status: Complete primary abstract and linked-status metadata examined. No complete-paper or supplement audit claimed. none_found. Subclaims: SC1, SC2.

[46] FUS oncofusion protein condensates recruit mSWI/SNF chromatin remodeler via heterotypic interactions between prion-like domains.

Shows: Fusion assemblies recruit SWI/SNF through the disordered domain.

Does not show: Partner binding and assembly are coupled; oncogenic fusion capacity is not a general transcription principle.

Access/status: Complete primary abstract and linked-status metadata examined. No complete-paper or supplement audit claimed. none_found. Subclaims: SC1, SC2.

[47] Liquid-liquid phase separation of light-inducible transcription factors increases transcription activation in mammalian cells and mice.

Shows: Engineered condensate-forming TFs increase reporter output.

Does not show: Synthetic delivery and expression establish engineered sufficiency, not endogenous necessity.

Access/status: Complete primary abstract and linked-status metadata examined. No complete-paper or supplement audit claimed. none_found. Subclaims: SC1, SC2.

[48] Dynamic Regulation of Endogenous Transcription Factor Hubs at Single-Molecule Resolution.

Shows: Endogenous EWS-FLI1 forms sub-diffraction hubs larger than parental puncta; a fitted nucleation/dissolution model explains the stable size distribution without macroscopic droplets.

Does not show: Estimated dissolution rate and supersaturation are model-dependent, not directly observed thermodynamic equilibrium; drugs also alter signaling, DNA binding or localization and RNA effects have alternatives.

Access/status: Complete primary abstract and linked-status metadata examined. Selected body sections: 1:Results:paragraph0, 1:Results:paragraph2, 1:Results:paragraph8, 1:Results:paragraph9, 1:Results:paragraph12, 1:Results:paragraph20, 1:Results:paragraph22, 1:Results:paragraph29. No complete-paper or supplement audit claimed. UpdateOf: PMID41659683. Subclaims: SC1, SC2, SC3.

[49] BMAL1 regulates circadian rhythms via phase separation-mediated transcriptional hub formation.

Shows: An N-terminal IDR supports circadian complexes and rescue of cellular and behavioral rhythms.

Does not show: IDR deletion changes partner recruitment as well as condensation; phase-only necessity and universal relevance remain unresolved.

Access/status: Complete primary abstract and linked-status metadata examined. No complete-paper or supplement audit claimed. none_found. Subclaims: SC1, SC2.

[50] Phase separation drives aberrant chromatin looping and cancer development.

Shows: FG-rich IDRs support fusion-driven binding, looping and transformation; FUS IDR replacement rescues.

Does not show: Ectopic oncogenic fusion and multivalent binding are coupled; native broad primacy is not tested.

Access/status: Complete primary abstract and linked-status metadata examined. No complete-paper or supplement audit claimed. none_found. Subclaims: SC1, SC2.

[51] SOX2 Modulates the Nuclear Organization and Transcriptional Activity of the Glucocorticoid Receptor.

Shows: SOX2 changes GR condensates and DNA occupancy through indirect partner effects.

Does not show: Expression perturbation and DNA binding change simultaneously; GR N-terminal domain is dispensable in this setting.

Access/status: Complete primary abstract and linked-status metadata examined. No complete-paper or supplement audit claimed. none_found. Subclaims: SC1, SC2.

[52] Nuclear SREBP2 condensates regulate the transcriptional activation of lipogenic genes and cholesterol homeostasis.

Shows: Aromatic-IDR mutation reduces output and feeding-induced cholesterol responses; FUS replacement rescues cellular function.

Does not show: Activation affinity and phase propensity covary; sex and metabolic context limit generalization and the examined abstract does not isolate phase alone.

Access/status: Complete primary abstract and linked-status metadata examined. No complete-paper or supplement audit claimed. none_found. Subclaims: SC1, SC2.

[53] Evidence for DNA-mediated nuclear compartmentalization distinct from phase separation.

Shows: Accessible DNA rebinding explains Pol II accumulation with unrestricted diffusion across compartment boundaries.

Does not show: A viral compartment is not a normal enhancer; demonstrates a viable non-LLPS mechanism rather than excluding all native LLPS.

Access/status: Complete primary abstract and linked-status metadata examined. No complete-paper or supplement audit claimed. none_found. Subclaims: SC1, SC3.

[54] RNA polymerase II clustering through carboxy-terminal domain phase separation.

Shows: CTD length changes clustering and chromatin association; phosphorylation dissolves in-vitro assemblies.

Does not show: CTD repeats also recruit essential transcription proteins; clustering and chemistry are not independently varied.

Access/status: Complete primary abstract and linked-status metadata examined. No complete-paper or supplement audit claimed. none_found. Subclaims: SC1, SC2.

[55] Histone acetylation-dependent clustering of BRD2 instructs transcription dynamics.

Shows: BRD2 and its IDR support TFIID-associated Pol II recruitment and transcription.

Does not show: Chromatin binding, acetylation and direct recruitment change together; a phase-specific matched rescue is not established.

Access/status: Complete primary abstract and linked-status metadata examined. No complete-paper or supplement audit claimed. none_found. Subclaims: SC1, SC2.

[56] RNA Polymerase II cluster dynamics predict mRNA output in living cells.

Shows: Brief Pol II cluster lifetimes correlate with stimulated transcription.

Does not show: Temporal correlation and crowding do not diagnose LLPS or prove phase-state necessity.

Access/status: Complete primary abstract and linked-status metadata examined. No complete-paper or supplement audit claimed. none_found. Subclaims: SC1, SC2.

[57] 1,6-Hexanediol, commonly used to dissolve liquid-liquid phase separated condensates, directly impairs kinase and phosphatase activities.

Shows: Hexanediol directly inhibits kinases and phosphatases at concentrations used in phase experiments.

Does not show: In-vitro enzyme tests expose nonspecificity but do not directly measure every cellular phase mechanism.

Access/status: Complete primary abstract and linked-status metadata examined. No complete-paper or supplement audit claimed. none_found. Subclaims: SC1, SC2.

[58] RNA polymerase II clusters form in line with surface condensation on regulatory chromatin.

Shows: Regulatory chromatin surfaces explain Pol II cluster shapes below bulk LLPS thresholds.

Does not show: Surface condensation differs from bulk liquid droplets; chemical perturbation and fitted models do not isolate native phase necessity.

Access/status: Complete primary abstract and linked-status metadata examined. No complete-paper or supplement audit claimed. none_found. Subclaims: SC1, SC2.

[59] N-terminal tagging of RNA Polymerase II shapes transcriptomes more than C-terminal alterations.

Shows: Tags cause larger transcriptome changes than CTD alterations; global bursting is largely robust to CTD changes.

Does not show: Gene-specific CTD dependence remains; tagging artifacts require explicit controls.

Access/status: Complete primary abstract and linked-status metadata examined. No complete-paper or supplement audit claimed. none_found. Subclaims: SC1, SC2.

[60] Real-time dynamics of RNA polymerase II clustering in live human cells.

Shows: Clusters are transient and stimulus-responsive rather than static factories.

Does not show: Crowding and dynamics alone do not establish liquid-state identity or causal transcription dependence.

Access/status: Complete primary abstract and linked-status metadata examined. No complete-paper or supplement audit claimed. none_found. Subclaims: SC1.

[61] Disordered C-terminal domain drives spatiotemporal confinement of RNAPII to enhance search for chromatin targets.

Shows: CTD-dependent confinement improves target search and bursting without apparent phase-separated condensates.

Does not show: CTD truncation also changes protein interactions; absence of visible droplets does not exclude nanoscale cooperativity.

Access/status: Complete primary abstract and linked-status metadata examined. No complete-paper or supplement audit claimed. UpdateOf: PMID37577667. Subclaims: SC1, SC2, SC3.

[62] Hormone-induced enhancer assembly requires an optimal level of hormone receptor multivalent interactions.

Shows: An optimal interaction range supports enhancer assembly and output; too little or too much is detrimental.

Does not show: Heterotypic partner binding covaries with condensation; this does not show more LLPS causes more transcription.

Access/status: Complete primary abstract and linked-status metadata examined. No complete-paper or supplement audit claimed. none_found. Subclaims: SC1, SC2.

[63] Intrinsically disordered bHLH family member TCF4 drives phase separation in a DNA-dependent manner.

Shows: TCF4 phase separates in vitro but specific and nonspecific DNA dissolve the droplets.

Does not show: No endogenous transcription necessity test; isoforms and crowding conditions limit interpretation.

Access/status: Complete primary abstract and linked-status metadata examined. No complete-paper or supplement audit claimed. none_found. Subclaims: SC1.

[64] Reconciling competing models on the roles of condensates and soluble complexes in transcription factor function.

Shows: Soluble-complex affinity and phase propensity covary; high-affinity variants show condensation-associated attenuation.

Does not show: Cellular phase assignment partly rests on predictive relationships and ChIP access may differ; neither all-soluble nor all-condensate models are established.

Access/status: Complete primary abstract and linked-status metadata examined. Selected body sections: 1:RESULTS:paragraph9, 1:RESULTS:paragraph11, 1:RESULTS:paragraph19, 1:RESULTS:paragraph23, 1:RESULTS:paragraph26, 1:RESULTS:paragraph27. No complete-paper or supplement audit claimed. UpdateOf: PMID39605529. Subclaims: SC1, SC2, SC3.

Documentary record

Every grade on this site is reproducible from its record. The files below are the posted edition exactly as assessed.

• ADVERSARIAL_REVIEW.md
• SHA256.txt
• adversarial-findings.json
• codex-correction-provenance.json
• consensus-addendum.md
• consensus-validation.json
• edition-approval-audit.json
• reader-summary.json
• transcriptional-condensates.review.md
• transcriptional-condensates.scorecard.json
• validation-report.json
• validation-round2-2026-10-14.json
• validation.txt

Sources: 64 · load-bearing claims: 68 · verified: 68 · unverified: 0 · contradicted: 0 · schema 5.3