Experiment: does KMT2A::AFF1 form condensates at endogenous levels, and does it need them? The separation-of-function and IDR-swap test
also called AFF1 IDR deletion · IDR swap FUS · endogenous condensate imaging KMT2A fusion
Tag the fusion protein in a real leukemia cell, look at whether it forms droplets, then delete the floppy region that would make droplets while checking the protein still binds its partners. If the cell stays leukemic, droplets are not the mechanism.
The same thing, for a scientist
A three-arm test. Arm 1: endogenously tag KMT2A::AFF1 and image at endogenous levels with FRAP and 1,6-hexanediol, since supra-endogenous expression is known to generate non-functional condensates. Arm 2: an AFF1 IDR-deletion separation-of-function allele, with biochemical confirmation that SEC assembly and menin/LEDGF binding are preserved, scored for transformation, target output and occupancy. Arm 3: the symmetric IDR-swap, replacing the AFF1 IDR with the unrelated FUS IDR — the design that settled the NUP98 question. Arm 3 is what makes the experiment decisive rather than merely suggestive, because it distinguishes 'needs this specific sequence' from 'needs generic multivalent self-association'.
The three arms, and why arm 3 is non-negotiable1 deeper
Arm 1 — look, properly. Endogenous knock-in tag (HaloTag or mNeonGreen) on KMT2A::AFF1 in SEM or RS4;11. Live imaging, FRAP for internal dynamics, 1,6-hexanediol as a crude perturbation (interpreted cautiously, since it is the weakest tier of evidence), and co-imaging with menin and AFF4. Crucially at endogenous levels, because expression beyond endogenous level is documented to produce non-functional condensates 1.
Arm 2 — the separation-of-function allele. Delete or scramble the AFF1 low-complexity/IDR region in the endogenous fusion, then biochemically verify before phenotyping that SEC assembly (AFF4, ELL, CDK9 co-IP 2) and menin/LEDGF binding are intact. Only then score transformation, target-gene output and chromatin occupancy. Skipping the verification step produces the uninformative outcome in which both condensates and binding are lost — the most likely and least useful result.
Arm 3 — the IDR swap. Replace the AFF1 IDR with the unrelated FUS IDR. In NUP98-HOXA9 this swap fully rescued enhanced genome-wide binding and target activation, which is what established that the requirement was for generic LLPS capacity rather than for NUP98 sequence 3. Running the same design here is what turns a suggestive negative into a decisive one:
- If the FUS-IDR swap rescues a function lost by IDR deletion → KMT2A::AFF1 does need generic self-association, and
hyp-aff1-condensates-dispensable-for-kmt2a-fusion-functionis refuted. - If IDR deletion does not lose function in the first place → hypothesis supported, and arm 3 is moot.
- If IDR deletion loses function and the FUS swap does not rescue → the AFF1 IDR does something sequence-specific (most plausibly a protein-interaction surface), which is mechanistically interesting and is not a condensate result.
That three-way discrimination is unavailable from arms 1 and 2 alone, which is why proposals that stop at 'delete the IDR and see' generate arguments rather than answers.
Secondary arm worth including cheaply: the reciprocal product
The reciprocal AFF1::KMT2A product is IDR-rich and already has a node in this graph (mechanism-reciprocal-kmt2a-fusion-chromatin-opening). If condensate competence exists anywhere in t(4;11) it may sit there rather than in the forward product. Tagging it in the same clones costs little and nobody has asked the question.
A comparator that makes the whole experiment interpretable: run a NUP98-fusion-positive line in parallel as the positive control, since LLPS is established as critical for NUP98 fusion oncoproteins across four different partners 4. Without a positive control in the same hands, a negative imaging result in arm 1 is not publishable.
| design | Three arms in SEM and RS4;11 t(4;11) lines with endogenous knock-in alleles. Arm 1: HaloTag/mNeonGreen-tagged KMT2A::AFF1, live imaging with FRAP and 1,6-hexanediol at endogenous expression, co-imaged with menin and AFF4. Arm 2: AFF1 IDR deletion/scramble with prior biochemical confirmation that AFF4/ELL/CDK9 co-IP and menin/LEDGF binding are preserved. Arm 3: AFF1 IDR replaced by the unrelated FUS IDR. Secondary arm: tag the reciprocal AFF1::KMT2A product. Positive control: a NUP98-fusion-positive line imaged in parallel. |
|---|---|
| readout | Presence, size, number and FRAP dynamics of nuclear foci; co-localisation with target loci; fusion and menin occupancy by CUT&RUN; nascent transcription at fusion targets; serial replating and differentiation markers. |
| cost tier | med |
| timeline | 8-12 months |
decision table
[
{
"result": "no foci at endogenous levels",
"interpretation": "hypothesis supported; condensates not used by KMT2A::AFF1"
},
{
"result": "foci present; IDR deletion abolishes foci but preserves function",
"interpretation": "condensates correlative — strongest support for the hypothesis"
},
{
"result": "IDR deletion abolishes function AND FUS-IDR swap rescues",
"interpretation": "generic LLPS required; hypothesis refuted; new target class"
},
{
"result": "IDR deletion abolishes function and FUS-IDR swap does NOT rescue",
"interpretation": "sequence-specific AFF1 IDR function (likely a protein-interaction surface), not a condensate result"
},
{
"result": "IDR deletion abolishes both foci and partner binding",
"interpretation": "uninformative — the failure mode the biochemical pre-verification exists to avoid"
}
]controls
[ "untagged parental line", "NUP98-fusion-positive line as LLPS positive control", "tagged wild-type KMT2A allele", "biochemical verification of preserved SEC assembly and menin/LEDGF binding before phenotyping" ]
tests hypotheses
[ "hyp-aff1-condensates-dispensable-for-kmt2a-fusion-function", "hyp-cell-state-plasticity-limits-interface-disruption" ]
- The most likely outcome, IDR deletion abolishing both condensates and partner binding, is uninformative; without built-in biochemical pre-verification the experiment generates a result that will be over-interpreted in whichever direction the authors prefer.
- 1,6-hexanediol is non-specific and disrupts many nuclear structures, so any conclusion resting on it alone is weak evidence and should be presented as such.
- A positive condensate result would not by itself justify a therapeutic programme, since condensates can be necessary without being sufficient, as shown for NUP98-NSD1.
- Which AFF1 region is the right deletion target? AFF1 has an ALF domain, a pSER domain and extensive low-complexity sequence; a principled choice needs a disorder/LLPS-propensity prediction pass first, which is a cheap computational precursor.
- Does the endogenous tag disturb fusion function? The tagged-untreated control is mandatory, and a negative imaging result is uninterpretable without it.
- Would a negative result be publishable? This Queen's view is yes, as a mechanistic distinction between fusion classes — but only with the NUP98 positive control run in the same hands.
- E1strong confidence 4 of 5: Replicated in peer-reviewed work by more than one group.paper
ENL YEATS hotspot mutants form condensates at native genomic targets requiring two disordered regions of opposing charges and extrinsic elongation factors, with extensive mutagenesis establishing condensate formation as a driver of oncogenic activation; critically, expression of ENL mutants beyond the endogenous level leads to 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, not a KMT2A fusion; cited for the endogenous-expression requirement that constrains this experiment's design and for the mutagenesis strategy to emulate.
Evidence that argues against this (1)
- Condensate formation as a driver rests on assays that cannot distinguish phase separation from clustered binding to a template. DNA-mediated nuclear compartmentalisation produces puncta, FRAP recovery and hexanediol sensitivity without liquid-liquid phase separation, and this project's standing rule is that only half-bleach internal mixing separates the two. The requirement for extrinsic elongation factors is equally read as cooperative assembly on chromatin. eLife 2019
- E2established confidence 5 of 5: Multiple independent lines of evidence; treated as settled in the field.paper
KMT2A chimeras co-purify a super elongation complex containing ELL, P-TEFb and AFF4, with AFF4 required for complex stability — defining the co-immunoprecipitation panel that must be shown intact in an IDR-deletion separation-of-function allele.
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 does not report stoichiometry; cited as the assay specification for verifying preserved complex assembly.
Evidence that argues against this (1)
- Co-purification of a super elongation complex defines what travels together through a purification, not what is required in cells. Using the co-IP panel as the readout for an IDR-deletion separation-of-function allele therefore builds the experiment's conclusion into its assay: an allele could preserve every co-IP and still fail, or lose a band and function normally. Nature 2012
- E3strong confidence 4 of 5: Replicated in peer-reviewed work by more than one group.paper
In NUP98-HOXA9 the FG-repeat IDRs are essential for LLPS puncta and transformation, and replacing them with the unrelated LLPS-forming FUS IDR recapitulated enhanced genome-wide binding and target activation — the IDR-swap design that distinguishes a generic LLPS requirement from a sequence-specific one.
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 as the experimental-design template and evidentiary standard, not as evidence about KMT2A fusions.
Evidence that argues against this (1)
- The FUS-IDR swap result argues the opposite way from how it is used here. If an UNRELATED IDR recapitulates enhanced binding and target activation, then what the FG repeats contribute is a generic biophysical property rather than a sequence-specific one — which makes an IDR swap a poor discriminator, not a good one, and means a KMT2A::AFF1 allele that survives such a swap would have shown very little. eLife 2019
- E4strong confidence 4 of 5: Replicated in peer-reviewed work by more than one group.paper
Both homotypic and heterotypic interactions are required for NUP98-HOXA9 nuclear puncta associated with aberrant transcriptional activity and HSPC transformation, and three further NUP98 fusion oncoproteins also form nuclear puncta and transform, establishing NUP98 fusions as a validated positive control for condensate imaging in leukemia cells.
Phase Separation Mediates NUP98 Fusion Oncoprotein Leukemic Transformation · Chandra B, Michmerhuizen NL, Shirnekhi HK, et al. (Kriwacki RW, senior author) · Cancer Discovery 12(4):1152-1169 2022What this evidence cannot show
NUP98 class only; used here solely to justify the choice of positive control.
Evidence that argues against this (1)
- Even in the NUP98 class, condensate formation can be necessary without being sufficient: SMARCA5 inhibition abolished NUP98-NSD1 function without affecting the condensates, so a positive condensate result in this experiment would still require a functional separation-of-function test. Jevtic Z et al. 2022
raises the value of the experiment: it would be the first AFF1 phase-behaviour measurement of any kind