Hypothesis: KMT2A::AFF1 does not require phase separation — recruitment biochemistry is sufficient, and the condensate model does not transfer from NUP98/ENL
also called condensates dispensable hypothesis · AFF1 IDR not required
We predict that the KMT2A-AFF1 fusion works by simple recruitment — grabbing the right machinery at the right genes — and does not need to form liquid droplets, unlike some other fusion proteins that genuinely do. This is worth settling because a wrong answer would send the field chasing droplet drugs.
The same thing, for a scientist
We hypothesise that KMT2A::AFF1 function is fully accounted for by stoichiometric recruitment (CXXC/AT-hook targeting, menin/LEDGF tethering, PAF1c contact, constitutive SEC delivery, DOT1L recruitment) and that liquid-liquid phase separation is neither necessary nor sufficient for its oncogenic activity. The basis is architectural: NUP98 fusions require their FG-repeat IDR and are rescued by an unrelated FUS IDR, and ENL YEATS hotspot mutants require condensate formation — but both are oncoproteins whose gain of function IS multivalent self-association, whereas KMT2A fusions gain a defined enzymatic/elongation partner. CORRECTION 2026-09-19: the entry previously asserted that systematic PubMed search returns zero reports of AFF1, AFF4 or AF9 condensates. That is wrong. Guo et al. (Sci Adv 2020, PMID 32270036) report that ENL and its IDR initiate liquid-liquid phase separation of the super elongation complex, that AFF4 fluidises those droplets, and that fusion of ENL to MLL boosts SEC phase separation. The hypothesis survives as a null about KMT2A::AFF1 specifically - no separation-of-function condensate mutant of a KMT2A fusion exists - but the 'zero reports' premise is withdrawn. The hypothesis is stated as the null that the field should be required to reject.
Why state a null hypothesis at all1 deeper
Because the alternative is currently being assumed rather than tested, and because the assumption has a cost.
The verified position: no separation-of-function condensate mutant of a KMT2A fusion has been published. A previous version of this entry went further and claimed zero PubMed-indexed reports of AFF1/AF4, AFF4 or AF9/MLLT3 condensates; that claim was wrong and is withdrawn (2026-09-19) - Guo et al. report ENL-initiated LLPS of the super elongation complex with AFF4 fluidising the droplets, and MLL-ENL fusion boosting SEC phase separation [E3 counter-evidence], and no KMT2A-fusion condensate separation-of-function mutant exists (see mechanism-condensates-load-bearing-or-correlative). Meanwhile load-bearing condensate genetics is real for two other classes: NUP98 fusions, where the FG-repeat IDR is essential and can be replaced by an unrelated FUS IDR with full rescue of genome-wide binding and target activation 1, and ENL YEATS hotspot mutants, where condensate-disrupting mutagenesis impairs chromatin and oncogenic function in a knock-in mouse and an inhibitor displaces the condensates in vivo 2.
The architectural distinction is the substance of the hypothesis. In NUP98 fusions and mutant ENL, multivalent self-association is itself the gain of function — which is why an unrelated IDR substitutes. In KMT2A fusions the gain of function is a specific partner: constitutive SEC/AEP delivery and DOT1L recruitment, mediated by defined, structurally characterised, stoichiometric interactions 34. Nothing in the KMT2A mechanism currently requires a phase-separated compartment to explain it.
And one more constraint from the ENL work that cuts against casual condensate claims: expression beyond endogenous level produces non-functional condensates 2. Any positive KMT2A-fusion result obtained by overexpression should therefore be discounted in advance.
The decisive experiment, and why it is worth doing even though we predict a null
Endogenously tag KMT2A::AFF1 in a t(4;11) line, image it live at endogenous levels, and — the load-bearing part — make an AFF1 IDR-deletion separation-of-function allele that abolishes any observed self-association while preserving SEC assembly and menin/LEDGF binding (verified biochemically). Then ask whether transformation, target-gene output and fusion occupancy survive.
Four outcomes, all informative:
- No condensates at endogenous levels → hypothesis supported; the field can stop, and this is publishable as a mechanistic distinction between fusion classes.
- Condensates present, IDR deletion abolishes them, function preserved → condensates are correlative in KMT2A fusions. Strongest possible support.
- Condensates present and required, with binding preserved → hypothesis refuted, and a new target class opens. We would be wrong in a useful direction.
- IDR deletion abolishes both condensates and binding → uninformative, and the most likely practical outcome, which is exactly why the biochemical verification of preserved SEC assembly must be built into the design rather than checked afterwards.
A secondary question worth carrying: the reciprocal AFF1::KMT2A product is also IDR-rich and already has a node in this graph (mechanism-reciprocal-kmt2a-fusion-chromatin-opening). If condensate competence sits anywhere in t(4;11), it may sit there rather than in the forward product, and nobody has asked.
Separation of function: compartment address is not transformation (added 2026-09-18)
Counter-evidence added 2026-09-18: the compartment address and the transforming function are separable. Ayton, Chen & Cleary mutagenised the retained MLL N-terminus and found three regions, including CXXC/Basic, that are essential for myeloid transformation by mechanisms that do not compromise subnuclear localisation; CXXC point mutants that kill transformation also kill nonmethylated-CpG binding and transactivation 5.
This is the cleanest separation-of-function result available on this question, and it cuts a specific way. A fusion can hold its subnuclear position and still be transformation-dead. So being in the right compartment is not sufficient, and any hypothesis that makes compartment occupancy - condensate, speckle, body - the driver of transformation has to explain why the address can be kept while the function is lost. What the same experiment does not do is exonerate localisation generally: the CXXC mutants lose the locus address (CpG binding) at the same time. The result separates compartment from transformation; it leaves locus and transformation firmly coupled.
Correction and what is left of the hypothesis (2026-09-19)
Withdrawn: the premise that no AFF-family condensate has been reported. Guo et al. (Sci Adv 2020, PMID 32270036) report that ENL or its isolated intrinsically disordered region is sufficient to initiate liquid droplet formation of the super elongation complex, that AFF4 acts with ENL to fluidise SEC droplets, and that fusing ENL to MLL boosts SEC phase separation. That is a direct positive report about an AFF-family SEC subunit and about a KMT2A fusion.
What survives: the hypothesis as stated is about KMT2A::AFF1 and about necessity. No separation-of-function mutant exists that removes condensate formation from a KMT2A fusion while leaving its stoichiometric interactions intact, so necessity is untested rather than disproved. The Guo result raises the prior on condensate involvement substantially and lowers this hypothesis's prior probability; it does not settle it.
What the methodological literature adds on the other side: LLPS claims in cells are hard to diagnose (McSwiggen 2019; Alberti 2019), fixation can manufacture or erase puncta (Irgen-Gioro 2022), and 1,6-hexanediol inactivates CDK9/P-TEFb and phosphatases at the concentrations used to dissolve droplets (Duster 2021) - which means the very assay most likely to be run on SEC condensates reports a confounded readout.
| statement | Liquid-liquid phase separation is neither necessary nor sufficient for KMT2A::AFF1 oncogenic function; the fusion operates by stoichiometric recruitment, and the condensate models validated for NUP98 fusions and ENL YEATS mutants do not transfer because in those cases multivalent self-association is itself the gain of function. |
|---|---|
| origin | literature-gap |
| prior probability | 0.45 |
| prior probability note | Lowered from 0.6 to 0.45 on 2026-09-19. Guo et al. (PMID 32270036) report ENL-initiated LLPS of the super elongation complex, AFF4 fluidising those droplets, and MLL-ENL fusion boosting SEC phase separation - a direct positive result on an AFF-family SEC subunit and on a KMT2A fusion. |
predictions
[ "Endogenously tagged KMT2A::AFF1 will not form discrete, dynamic nuclear condensates at endogenous expression levels in a live t(4;11) cell. NOTE (2026-09-19): 1,6-hexanediol was removed from this prediction as a diagnostic. Duster et al. (PMID 33814344) show it inactivates kinases and phosphatases including CDK9/P-TEFb at 5-10%, so a hexanediol response in a SEC-linked system is uninterpretable; and Irgen-Gioro et al. (PMID 36444977) show fixation alone can create or erase puncta, so the readout must be live-cell.", "An AFF1 IDR-deletion allele that preserves SEC assembly and menin/LEDGF binding will retain transformation capacity and target-gene output.", "Fusion chromatin occupancy will scale with the concentration of its binding partners in a stoichiometric, non-cooperative manner, without the sharp concentration threshold characteristic of phase separation.", "If condensate competence exists anywhere in t(4;11), it will be found in the IDR-rich reciprocal AFF1::KMT2A product rather than in KMT2A::AFF1." ]
experiments
[ "exp-aff1-idr-condensate-separation-of-function" ]
analyses
[ "ana-menin-resistance-cell-state-signature" ]
falsifiers
[ "Endogenously tagged KMT2A::AFF1 forms dynamic condensates at its target loci at endogenous expression levels.", "An AFF1 IDR-deletion allele that demonstrably preserves SEC assembly and menin/LEDGF binding nonetheless loses transformation capacity.", "Replacing the AFF1 IDR with an unrelated LLPS-forming IDR (e.g. the FUS IDR) rescues function, which would be the NUP98-style result and would refute the architectural argument outright.", "Fusion occupancy shows a sharp concentration threshold consistent with a phase transition rather than stoichiometric binding.", "AFF1 or AFF4 is shown to concentrate KMT2A::AFF1 at target loci in a live cell at endogenous levels, extending the Guo 2020 SEC phase-separation result (PMID 32270036) from ENL/MLL-ENL to the AFF1 fusion." ]
- If this hypothesis is wrong, the field will have under-invested in a genuinely new target class, and condensate-modulating chemistry would be years behind.
- If it is right, effort spent on condensate-dissolving approaches in KMT2A-r is wasted and, worse, such agents are intrinsically non-selective because the same physical chemistry underlies nucleoli, speckles and Pol II clusters.
- The most likely experimental outcome — IDR deletion abolishing both condensates and binding — is uninformative, so a poorly designed version of this experiment produces an ambiguous result that will be over-interpreted in whichever direction the authors prefer.
- Has anyone endogenously tagged a KMT2A fusion and simply looked? The absence of this experiment in a field that has published extensively on condensates in other fusions is itself notable.
- Is the IDR-swap test (replace AFF1 IDR with FUS IDR) the right decisive experiment? It is the design that settled the NUP98 question and would settle this one symmetrically.
- If KMT2A fusions genuinely do not use condensates while NUP98 fusions do, is that difference a general rule separating 'reader-plus-recruiter' fusions from 'IDR-plus-DNA-binding-domain' fusions? That would be a useful classification for the Q11 comparative beat.
- Does KMT2A::AFF1 itself, at endogenous levels, concentrate SEC in a hexanediol-independent, live-cell-verified condensate - and can a separation-of-function AFF1 IDR mutant remove that without removing SEC binding?
- E1strong confidence 4 of 5: Replicated in peer-reviewed work by more than one group.paper
The NUP98 FG-repeat IDRs in NUP98-HOXA9 are essential for LLPS puncta and for leukemic transformation, LLPS is required for the broad super-enhancer-like binding pattern, and replacing the FG repeats with the unrelated LLPS-forming FUS IDR recapitulates enhanced genome-wide binding and target activation, with CTCF-independent loops at proto-oncogenes.
Phase separation drives aberrant chromatin looping and cancer development · Ahn JH, Davis ES, Daugird TA, et al. (Wang GG, senior author) · Nature 595(7868):591-595 2021What this evidence cannot show
A different oncoprotein class. Cited here to define the evidentiary standard a KMT2A-fusion condensate claim must meet, and to make the architectural contrast, not as evidence about KMT2A fusions.
Evidence that argues against this (3)
- The sceptical arm of the phase-separation literature holds that the observations used to call an in-cell condensate rarely discriminate LLPS from other forms of clustering. McSwiggen, Mir, Darzacq and Tjian reviewed the evidence for LLPS across cellular processes and concluded that it is often phenomenological and inadequate to distinguish phase separation from alternative mechanisms, and that causality and function are more elusive still; they urge measurement of endogenous proteins at physiological abundance. The NUP98-HOXA9 FG/FUS-IDR swap is consistent with LLPS but is equally consistent with multivalent IDR-driven avidity that is not a separate liquid phase. McSwiggen DT et al. 2019
- Alberti, Gladfelter and Mittag set out explicit guidelines for claiming LLPS in vitro and in cells and catalogue the caveats of the common assays; by those criteria droplet formation at non-physiological concentrations in vitro does not establish an in-vivo condensate. Musacchio argues further that validating the phase-separation paradigm's assumptions is much harder than appreciated and that highly specific interactions, not unspecific ones, appear to be the main driver of membraneless-compartment biogenesis - which is the architecture this hypothesis proposes for KMT2A fusions. Alberti S et al. 2019
- Musacchio's analysis of membraneless-compartment biogenesis concludes that specific interactions rather than weak unspecific ones dominate, with phase separation harnessed locally in selected instances to tune material properties. That is a general argument that a stoichiometric-recruitment null is the right default, and it weakens the inference from any single IDR-swap rescue. Musacchio A 2022
- E2strong confidence 4 of 5: Replicated in peer-reviewed work by more than one group.paper
ENL YEATS hotspot mutations trigger aberrant condensates at native genomic targets requiring two disordered regions of opposing charges plus extrinsic elongation factors, with extensive mutagenesis establishing condensate formation as a driver of oncogenic gene activation; expression of ENL mutants beyond the endogenous level produces non-functional condensates.
Hotspot mutations in the structured ENL YEATS domain link aberrant transcriptional condensates and cancer · Song L, Yao X, Li H, et al. (Wan L, senior author) · Molecular Cell 82(21):4080-4098.e12 2022What this evidence cannot show
ENL point mutants in Wilms tumour and AML, not KMT2A fusions. The supra-endogenous-expression caveat is the authors' own and is cited here as a methodological constraint on testing this hypothesis.
Evidence that argues against this (2)
- Two of the standard experimental supports for in-cell condensates are confounded. Irgen-Gioro, Yoshida, Walling and Chong showed that paraformaldehyde fixation can both enhance and diminish apparent LLPS, and for some proteins creates droplet-like puncta in cells that have none when alive - so fixed-cell imaging of condensates at native loci needs a live-cell control. Irgen-Gioro S et al. 2022
- 1,6-hexanediol, the usual reagent for showing a focus is 'liquid', is not inert: Duster et al. found it strongly impairs kinases and phosphatases at 1% and leaves both classes virtually inactive at the 5-10% used to dissolve droplets, with CDK9/P-TEFb among the affected kinases. Any SEC- or elongation-linked readout taken after hexanediol treatment is therefore confounded in exactly the pathway this hypothesis is about. Duster R et al. 2021
- E3established confidence 5 of 5: Multiple independent lines of evidence; treated as settled in the field.paper
KMT2A chimeras co-purify a defined super elongation complex containing ELL, P-TEFb and AFF4, with AFF4 required for complex stability and for transcription by poised Pol II — a stoichiometric recruitment mechanism that accounts for fusion function without invoking phase separation.
AFF4, a component of the ELL/P-TEFb elongation complex and a shared subunit of MLL chimeras, can link transcription elongation to leukemia · Lin C, Smith ER, Takahashi H, et al. (Shilatifard A, senior author) · Molecular Cell 37(3):429-437 2010What this evidence cannot show
Affinity purification cannot report stoichiometry or spatial organisation, so it does not positively exclude a condensate; it establishes that a stoichiometric model is sufficient to explain the known biology. CORRECTED 2026-09-19: this entry previously treated affinity-purification stoichiometry as the only published account of SEC. Guo et al. (PMID 32270036) report SEC phase separation with AFF4 fluidising ENL-initiated droplets and MLL-ENL boosting the behaviour, so the stoichiometric model is one of two live accounts rather than the uncontested default.
Evidence that argues against this (1)
- This is the strongest published counter to the hypothesis and it was missed by the entry's original literature search. Guo et al. report that SEC components compartmentalise and concentrate P-TEFb by liquid-liquid phase separation away from the inactive HEXIM1-P-TEFb pool; ENL or its isolated IDR is sufficient to initiate SEC droplet formation, AFF4 acts with ENL to fluidise those droplets, and fusion of ENL to MLL boosts SEC phase separation. So an AFF-family SEC subunit is reported in a condensate, and a KMT2A fusion is reported to increase that behaviour. The entry's claim of 'zero reports' was withdrawn on 2026-09-19. Guo C et al. 2020
- E4strong confidence 4 of 5: Replicated in peer-reviewed work by more than one group.paper
Reconstitution of defined minimal complexes shows AF4/AFF4 mediate P-TEFb association with AF9 through direct interactions with AF9 and cyclin T1, P-TEFb links ELL and EAF1/EAF2 to AF4, and DOT1 forms a stable complex with AF9 that does not interact with AF9-AF4-P-TEFb complexes — demonstrating that the relevant assemblies are discrete and stoichiometric rather than a single mixed compartment.
Function of leukemogenic mixed lineage leukemia 1 (MLL) fusion proteins through distinct partner protein complexes · Biswas D, Milne TA, Basrur V, Kim J, Elenitoba-Johnson KSJ, Allis CD, Roeder RG · Proceedings of the National Academy of Sciences USA 108(38):15751-15756 2011What this evidence cannot show
In-vitro reconstitution with recombinant proteins; mutual exclusivity in vitro does not prove spatial separation in a nucleus, and a condensate could in principle host exchanging discrete complexes. This is supportive reasoning, not proof.
Evidence that argues against this (1)
- Reconstitution of discrete stoichiometric complexes does not exclude a condensate, and Guo et al. report the same components behaving as a phase-separating assembly in cells, with MLL-ENL fusion boosting SEC phase separation. Mutual exclusivity of purified complexes in vitro and coexistence inside a droplet in vivo are compatible: a condensate can host exchanging discrete complexes. Guo C et al. 2020
- E5strong confidence 4 of 5: Replicated in peer-reviewed work by more than one group.paper
High-resolution mutagenesis of the retained MLL N-terminus identified three domains - the CXXC/Basic domain and two domains of unknown function - that are essential for in vitro myeloid transformation **via mechanisms that do not compromise subnuclear localization**; CXXC point mutants that abolish transformation also abolish binding to nonmethylated CpG DNA and transactivation in vivo.
Binding to nonmethylated CpG DNA is essential for target recognition, transactivation, and myeloid transformation by an MLL oncoprotein · Ayton PM, Chen EH, Cleary ML · Molecular and Cellular Biology 24(23):10470-8 2004What this evidence cannot show
ABSTRACT ONLY - the PMC full text was not read, so the specific microscopy controls behind 'do not compromise subnuclear localization' were not inspected; the imaging in that era was fluorescence microscopy of transfected constructs in COS cells, which is a coarse readout of compartment and cannot exclude changes in condensate number, size or dynamics. In-vitro myeloid immortalisation of mouse progenitors with an MLL fusion construct. The result separates gross subnuclear position from transformation; it does NOT separate chromatin occupancy from transformation - on the contrary, CXXC mutants lose CpG binding, so locus-level address and transformation stay coupled.
Evidence that argues against this (1)
- The mutagenesis was scored on in vitro myeloid transformation, which is the assay this hypothesis should be most careful with: a domain dispensable for colony formation in murine progenitors may be required in a human B-lymphoid context, and "does not compromise subnuclear localization" is a statement about where a tagged protein appears, not about what it is doing there. Oncofusion knowledge graph (internal analysis) 2026
factory / Pol II cluster / condensate treated as one phenomenon after the 2026-09-19 correction
Counter-evidence (Xiang 2026, PMID 42748909): elongation-coupled splicing in MLL-r leukaemia runs through ECAS, an AFF4-ELL2-EAF1 module joined to a U2 snRNP-associated intron binding complex; AQR/XAB2 perturbation increases intron retention with a long-gene preference and cripples MLL-r leukaemia stem cell function in mouse cells - AFF4 and gene length, not AFF1 G-quadruplex binding
artefact classes make the dispensability hypothesis the parsimonious prior
Counter-evidence (Guo 2020, PMID 32270036): AFF4 fluidises ENL-initiated SEC droplets and MLL-ENL fusion boosts SEC phase separation. The entry's 'zero reports of AFF-family condensates' premise is withdrawn.
Methodological counter (Duster 2021, PMID 33814344): 1,6-hexanediol at droplet-dissolving concentrations inactivates CDK9/P-TEFb, so hexanediol-based condensate tests on SEC read out a poisoned kinase rather than a dissolved phase.
bounds the hypothesis: there is no positive AFF1 condensate claim for it to dispute