The endogenous MLL-AF4 localisation experiment was unattributable, not negative: an anti-AF4 polyclonal in a t(4;11) cell images mostly wild-type AF4
also called Li Frestedt Kersey 1998 · endogenous MLL-AF4 confocal · 240 kDa MLL-AF4 · punctate nuclear staining AF4
In 1998 a group raised antibodies against AF4 and looked at real t(4;11) leukemia cells rather than cells forced to make extra fusion protein. They found the fusion in the same speckled nuclear pattern as normal AF4, indistinguishable from leukemia cells without the translocation — and they found there is less fusion protein than normal AF4, not more.
The same thing, for a scientist
**Re-read 2026-09-20 and the verdict is reversed in kind.** This entry previously treated PMID 9808577 as the best-designed endogenous experiment on fusion localisation and read its result as a **negative** - no mislocalisation. It is not a negative; it is **unattributable**. The reagent is an anti-AF4 polyclonal, which in a t(4;11) cell binds wild-type AF4 and the fusion alike, and the readout is immunofluorescence, which carries no size dimension to separate them. The same paper states the fusion proteins were present *in lesser quantity than the wild-type AF4*, so the image is dominated by the species that is not the fusion. **The experiment could not have detected a difference in either direction**, which is a different and weaker statement than the one this entry was making.
Why this is the strongest KMT2A-fusion localisation dataset in existence1 deeper
Grade the published KMT2A-fusion imaging by material and construct:
| study | protein | cells | construct |
|---|---|---|---|
| Li 1998 1 | endogenous MLL-AF4, 240 kDa | t(4;11) patient-derived leukemia lines and non-t(4;11) controls | none — antibody against native protein |
| Caslini 2000 | MLL N-terminal fragment, MLL-AF9, MLL-AF4 | U937, HeLa | tet-inducible overexpression |
| Ayton 2004 | MLL fusion constructs | COS (monkey kidney) | transfected |
| Siemund 2022 | MLL-AF4, murinised variant | transfected mammalian cells | transfected |
Only the first row uses the real protein in the real disease context. Its answer is 'the fusion looks like wild-type AF4, and leukemia cells with and without the translocation look the same'.
That should temper the enthusiasm of the entire beat, including the parts of this graph that treat nucleolar or speckle localisation as a live mechanism. It is not a refutation — see the limits below — but it is the closest thing to a controlled negative that exists, and it has been quietly ignored by a literature that prefers the overexpression images.
The abundance result, which nobody builds on
Fusion protein was present in lesser quantity than wild-type AF4 1. Every condensate and every avidity argument in this field implicitly assumes the fusion reaches a concentration where multivalent interactions dominate. A protein present at sub-wild-type-partner levels is a poor candidate for a concentration-driven demixing mechanism, and the in-vitro droplet concentrations used in the SEC literature are far above anything implied here.
This also bears on the degron and occupancy work in the graph: if the fusion is the scarce species, then its partners (AFF1/AFF4, menin, ENL) are in excess, and the rate-limiting quantity is fusion copy number. That is a measurable parameter nobody has reported in absolute units for a t(4;11) blast.
The limits of the negative — why this is a reason to repeat, not to close
- The antibody cannot separate the two species in the image. The polyclonals were raised against AF4 and recognise both the 125/145-kDa wild-type protein and the 240-kDa fusion. In immunofluorescence, the signal is the sum. Wild-type AF4 is abundant and the fusion is less abundant, so the fusion's own distribution is a minority contribution to the image. A distinct fusion distribution could be entirely hidden. This alone is enough to stop the paper being read as 'the fusion is not mislocalised'.
- 1998 confocal resolution. Diffraction-limited, no deconvolution, no quantitation of puncta number, size or intensity reported in the abstract. A 'similar pattern' at that resolution covers a wide range of real differences.
- Cell lines, not primary blasts. t(4;11) lines (SEM, RS4;11, MV4;11 class) are the appropriate material but are not patient marrow.
- No compartment markers. Nothing is co-stained — not fibrillarin, not SC35/SRSF2, not NEAT1, not PML, not lamin B1. The paper establishes 'punctate', not which puncta.
The modern repeat is obvious and cheap: knock a HaloTag or split-fluorophore into the fusion junction at the endogenous locus in SEM, which labels the fusion and nothing else, and co-stain the six compartment markers this beat catalogues. That single experiment would either revive or retire the entire KMT2A mislocalisation hypothesis.
The negative result is partner-specific, not a law (added 2026-09-18)
This entry's negative finding is about MLL-AF4 in t(4;11) lines, and it should not be generalised to every KMT2A fusion. The opposite result exists for a different partner. Rogaia et al. reported that HRX/ALL1 itself is found in cytoplasm and nucleus with diffuse staining, perinuclear accumulation and nuclear bodies, while the HRX/ALL1-eps15 fusion from t(1;11)(p32;q23) is exclusively nuclear, in bodies smaller and more numerous than wild-type 2. Their conclusion was the reverse of Li et al.'s: that fusion with eps15 alters cellular compartmentalisation and that this could be a mechanism of activation.
Which to believe depends on what you are asking. Li et al. 1 is the stronger endogenous experiment - real fusion protein in patient-derived t(4;11) lines - and it is negative. Rogaia et al. is largely a transfected-construct experiment in COS/HeLa, which is exactly the setup that inflates apparent puncta, but it is positive and it concerns a cytoplasmic-partner fusion where a change in compartment is mechanistically expected.
The defensible synthesis: there is no single 'KMT2A fusion localisation'. For the dominant nuclear-partner fusions the one endogenous test found nothing; for a cytoplasmic-partner fusion a construct-level test found relocalisation. Any hypothesis in this graph that treats subnuclear repositioning as a general property of KMT2A fusions is asserting more than either paper supports.
| compartment | unassigned punctate nucleoplasm |
|---|---|
| evidence class for kmt2a | endogenous protein in t(4;11) patient-derived lines — the only study of this grade |
| result direction | negative for mislocalisation |
| fusion molecular weight kda | 240 |
| fusion abundance vs wildtype partner | lower |
| graph correction | the claim that no endogenous-protein imaging of a KMT2A fusion exists is false; it exists, from 1998, and it is negative |
wildtype af4 molecular weight kda
[ 125, 145 ]
compartment markers costained
[]
methods used
[ "western blot", "immunoprecipitation", "confocal immunofluorescence with rabbit polyclonal anti-AF4" ]
- If the fusion is present at lower abundance than its wild-type partner, a stoichiometric inhibitor needs less occupancy than usually assumed to be effective — but a degrader faces the opposite problem, because removing a scarce protein from an excess of partners may simply let wild-type AFF1 take its place in the same complexes.
- A negative localisation result that the field has ignored for 28 years is a warning about publication incentives in this beat: the overexpression images are prettier and are the ones that get cited. Any new positive finding here will face, and should face, the same scepticism.
- Repeat the localisation question with a readout that keeps a size dimension - subcellular fractionation followed by anti-MLL-N and anti-AF4 blots - which needs no new reagent and would convert an unattributable image into an attributable measurement.
- Repeat with a fusion-specific label: a HaloTag or split-fluorophore knocked into the KMT2A::AFF1 junction at the endogenous locus in SEM labels the fusion and not wild-type AFF1 or wild-type KMT2A. This is the experiment the 1998 paper could not do and it settles the beat.
- What is the absolute copy number of KMT2A::AFF1 per nucleus in a t(4;11) blast? 'Less than wild-type AF4' is the only quantitative statement in the literature, and every condensate argument needs the number.
- Do the puncta seen in both t(4;11) and non-t(4;11) cells correspond to any named compartment? Nothing was co-stained in 1998 and nothing has been co-stained since.
- If the fusion is the scarce species and its partners are in excess, is fusion copy number the rate-limiting variable for target-gene output? A titratable degron series reading transcription against measured protein copies would test it directly.
- E1moderate confidence 3 of 5: One strong study, or several smaller consistent ones.paper
Rabbit polyclonal antibodies against AF4 detect endogenous AF4 at 125 and 145 kDa and the predicted 240-kDa MLL-AF4 fusion protein in all cell lines carrying t(4;11); the fusion protein is present in lesser quantity than wild-type AF4, has a similar subcellular distribution to AF4, and both t(4;11) and non-t(4;11) leukemic cells show a similar pattern of punctate nuclear staining in all cell lines tested by confocal immunofluorescence.
AF4 encodes a ubiquitous protein that in both native and MLL-AF4 fusion types localizes to subnuclear compartments · Li Q, Frestedt JL, Kersey JH · Blood 1998What this evidence cannot show
The polyclonal antibodies recognise both wild-type AF4 and the fusion, so the immunofluorescence signal is a mixture dominated by the more abundant wild-type protein and cannot isolate the fusion's own distribution. 1998 confocal resolution with no quantitation of puncta and no co-staining for any compartment marker. Leukemia cell lines rather than primary patient marrow. The negative result is therefore an absence of detected difference, not a demonstration of no difference. **Correction 2026-09-20: unattributable, not negative.** An anti-AF4 polyclonal cannot separate wild-type AF4 from MLL-AF4 in an immunofluorescence image, and the paper reports the fusion is the minority species. Read as a measured negative this supports a claim the data cannot carry.
Evidence that argues against this (3)
- Conditionally expressed MLL-AF9 and MLL-AF4 chimeras in U937 and HeLa were reported at the nucleolar periphery and in the nuclear matrix, a compartment assignment this endogenous study neither made nor excluded. Caslini C et al. 2000
- A later transfection study reported human MLL-AF4 as predominantly nucleolar with activation of the nucleolar stress pathway, directly at odds with 'similar distribution to wild-type AF4'. Siemund AL et al. 2022
- **The reagent decides what this experiment could show, and it could not show this.** Immunofluorescence has no size dimension; the audit of attribution in this graph finds that every surviving endogenous claim in this literature keeps a size or mass dimension in its readout, and this one does not. Fractionate-and-blot would make the same question answerable with the same antibody. Blood 1998
- E2moderate confidence 3 of 5: One strong study, or several smaller consistent ones.paper
For a different KMT2A fusion, the imaging result is positive rather than negative: HRX/ALL1 localises to both cytoplasm and nucleus with diffuse staining, perinuclear accumulation and nuclear bodies, whereas the HRX/ALL1-eps15 fusion product of t(1;11)(p32;q23) localises exclusively to the nucleus in bodies that are smaller and more numerous than the wild-type HRX/ALL1 nuclear bodies.
The localization of the HRX/ALL1 protein to specific nuclear subdomains is altered by fusion with its eps15 translocation partner · Rogaia D, Grignani F, Carbone R, Riganelli D, LoCoco F, Nakamura T, Croce CM, Di Fiore PP, Pelicci PG · Cancer Research 57(5):799-802 1997What this evidence cannot show
ABSTRACT ONLY - full text not obtained. This is a DIFFERENT fusion (KMT2A::EPS15, a rare partner) and, unlike the Li/Kersey t(4;11) work, it rests substantially on transfected constructs in COS and HeLa cells rather than on endogenous protein in patient-derived lines; the abstract does report that leukemic blasts with 11q23 abnormalities and an unknown partner showed similar morphological features. 1997 indirect immunofluorescence. Added 2026-09-18 so that this entry's negative result is not read as a general statement about all KMT2A fusions. **Correction 2026-09-20: unattributable, not negative.** An anti-AF4 polyclonal cannot separate wild-type AF4 from MLL-AF4 in an immunofluorescence image, and the paper reports the fusion is the minority species. Read as a measured negative this supports a claim the data cannot carry.
Evidence that argues against this (2)
- The positive imaging result rests on fixed transfected cells, and fixation is now known to change punctate appearance in both directions. Irgen-Gioro et al. imaged the same cells before and after paraformaldehyde fixation and found that fixation can both enhance and diminish apparent droplet-like behaviour, that for specific proteins puncta APPEAR in fixed cells that show none when live, and that fixing in the presence of glycine reverses the direction of the artefact; their kinetic model attributes the effect to the balance between protein-protein interaction dynamics and the relative fixation rates of the proteins involved. 'Smaller and more numerous nuclear bodies' in a 1997 indirect-immunofluorescence comparison of transfected wild-type and fusion constructs is precisely a measurement of punctum number and size in fixed cells, and the fusion and wild-type proteins have different interaction partners and therefore plausibly different fixation kinetics. Irgen-Gioro S et al. 2022
- **The reagent decides what this experiment could show, and it could not show this.** Immunofluorescence has no size dimension; the audit of attribution in this graph finds that every surviving endogenous claim in this literature keeps a size or mass dimension in its readout, and this one does not. Fractionate-and-blot would make the same question answerable with the same antibody. Blood 1998
that entry reads PMID 9808577 as a negative result; an anti-AF4 image in a t(4;11) cell cannot attribute, so it is an unattributable measurement rather than a negative
- related to →KMT2A::AFF1 (MLL-AF4), t(4;11)(q21;q23)
240 kDa fusion detected in all t(4;11) lines, punctate, indistinguishable from wild-type AF4
the only endogenous test returned no detectable mislocalisation
- refutes →The nucleolus — the one compartment with direct published images of a KMT2A fusion inside it
endogenous fusion distribution matched wild-type AF4, not a nucleolar pattern
wild-type AFF1/AF4 at 125 and 145 kDa, more abundant than the fusion
the fusion-specific label the 1998 experiment lacked
the equivalent endogenous-fusion imaging experiment has been done for KMT2A::AFF1 and was negative; it has never been done for KMT2A::MLLT1
the only endogenous AF4/MLL-AF4 image is unattributable by construction
the MLL-side twin of the same unattributable-image shape, one year earlier
the AT-hook-imposed-nuclear-address claim rests on transfected 1997 constructs; the only endogenous fusion-protein comparison found no relocation (Li/Kersey 1998)