Attribution audit 3 of 5 - where KMT2A::AFF1 sits in the nucleus: the endogenous image does not survive, the overexpression images do and say something different
also called MLL-AF4 localisation audit · punctate nuclear staining attribution · speckle attribution
Pictures of where the fusion protein sits inside the nucleus were taken with antibodies that also light up the normal protein, and in a microscope image there is no way to tell the two apart. The pictures that can be trusted were taken in cells given an extra, artificially high copy of the fusion - which is a different question.
The same thing, for a scientist
Two reagent designs carry the localisation literature and they audit in opposite directions. **Endogenous imaging fails.** The only immunofluorescence of a KMT2A fusion in patient-derived t(4;11) cells used rabbit polyclonals raised against GST-AF4. On a blot those reagents separate the species cleanly - fusion at about 240 kDa, wild-type AF4 at 125 and 145 kDa, and the fusion is the **less abundant** of the two. In the microscope that separation is gone. The reported result is that *both t(4;11) and non-t(4;11) leukaemic cells showed a similar pattern of punctate nuclear staining* [E1]. That is exactly the image a cell would give if the minor fusion species contributed nothing visible on top of the abundant wild-type AF4 pool, and the experiment cannot distinguish 'the fusion goes where AF4 goes' from 'the fusion is invisible in this image'. This graph currently reads that paper as a **negative result for mislocalisation**; the audit finding is that it is not a negative result, it is an unattributable measurement (`localization-endogenous-mll-af4-punctate-negative`). **Overexpression imaging survives, and it reports a positive effect.** Chimeric ALL-1 proteins expressed in transfected cells showed punctate speckles, and deletion mapping in the same system assigned that distribution to two short trithorax-conserved elements in the N-terminal ~1,100 residues [E2]. More pointedly, HRX/ALL1-eps15 from t(1;11) localised **exclusively to the nucleus in bodies that were smaller and more numerous** than wild-type HRX/ALL1 bodies, and wild-type HRX/ALL1 was found in both cytoplasm and nucleus [E3]. Those are attributable statements, because the fusion was introduced into a cell that did not otherwise contain it, and they say that fusing a partner **does** change compartmentalisation. **Verdict. This class survives only in overexpression, and the surviving result contradicts the endogenous non-result the graph has been leaning on.** The honest position is that the best-attributed experiments show a localisation change in an artificial system, and the only endogenous experiment could not have detected one either way.
Why overexpression is a particularly bad system for THIS readout
For most readouts overexpression is a quantitative worry. For punctate appearance it is a qualitative one, because the number of copies per nucleus is an input to whether puncta form at all. This graph already records that at physiological expression, low-complexity-domain interactions that control transcription form dynamic hubs without detectable phase separation, and that 1,6-hexanediol and classical FRAP cannot separate phase separation from binding to spatially clustered sites (localization-condensate-evidence-standards).
So the localisation literature for KMT2A fusions sits in a bad corner: the attributable experiments are the ones whose expression level most distorts the readout, and the physiological experiment is the one that cannot attribute. The SEC condensate work that the field's condensate framing rests on has the same shape 4.
What survives, stated precisely
- Survives: any imaging of an exogenously introduced, tagged or otherwise identifiable fusion in a cell that does not carry the rearrangement 23. Attribution is by construction. Interpretation is limited by copy number.
- Survives: subcellular or chromatin fractionation read out on a Western blot, because the ~240 kDa fusion band is resolved from wild-type AF4 and from wild-type MLL-N320. This gives compartment-level localisation (soluble nuclear / chromatin-bound / cytoplasmic) with real attribution and needs only reagents the field already has. This beat found no published example for KMT2A::AFF1.
- Does not survive: immunofluorescence in a rearranged cell with any shared-epitope antibody, including every anti-MLL-N, anti-AF4 and anti-ENL reagent 1.
- Does not survive: any claim that the fusion is or is not mislocalised relative to wild-type, at endogenous level, in a human cell.
The minimal experiment
HaloTag or HA knocked in at the endogenous fusion allele in SEM or RS4;11, imaged live. One knock-in settles the localisation class, the composition class and the turnover class at once, which is why it is the highest-leverage single experiment this audit identifies (exp-endogenous-tag-and-degron-knock-in-kmt2a-aff1). The tag must be C-terminal on the AFF1 side or internal: Olsen et al. found that an N-terminal tag on MLL::AF9 changed leukaemogenesis kinetics, which they attribute to interference with the N-terminal DNA-binding and menin-binding elements 5 - and those are the same elements Yano et al. mapped the punctate distribution to 2.
The cheap interim experiment is fractionation plus blot. It uses existing antibodies, resolves the species by size, and would convert 'we do not know where the fusion is' into 'the fusion is X% chromatin-bound against Y% for wild-type MLL in the same lysate'.
| claim class | subcellular and subnuclear localisation of the KMT2A::AFF1 protein |
|---|---|
| verdict | does not survive at endogenous level; survives in overexpression, where the surviving result is that fusion DOES alter compartmentalisation |
| correction owed in graph | localization-endogenous-mll-af4-punctate-negative currently reads PMID 9808577 as a negative result for mislocalisation; it is better described as an unattributable measurement, because an anti-AF4 image in a t(4;11) cell is dominated by the more abundant wild-type species |
reagent classes found
[
{
"reagent": "anti-AF4 rabbit polyclonal, immunofluorescence",
"system": "patient-derived t(4;11) lines",
"fusion_attributable": "no",
"why": "no size dimension in an image; the abundant wild-type AF4 pool dominates the signal"
},
{
"reagent": "anti-ALL-1 antibodies on transfected chimeras",
"system": "transfected cultured cells",
"fusion_attributable": "yes",
"why": "the chimera is present only because it was introduced; but copy number is uncontrolled"
},
{
"reagent": "anti-AF4 rabbit polyclonal, Western blot",
"system": "patient-derived t(4;11) lines",
"fusion_attributable": "yes",
"why": "~240 kDa fusion resolved from 125/145 kDa wild-type AF4"
}
]minimal fix
[ "HaloTag or HA knock-in at the endogenous fusion allele in SEM or RS4;11, imaged live, tag placed C-terminally or internally", "subcellular and chromatin fractionation read out on a size-resolving Western blot, using existing antibodies" ]
- A therapeutic strategy premised on disrupting a fusion-specific nuclear body or condensate is premised on imaging that cannot attribute the body to the fusion at endogenous level, so the target may be a wild-type AF4 or wild-type MLL structure present in every cell.
- 1,6-hexanediol, the reagent most often used to argue a body is a condensate, rapidly immobilises and condenses chromatin in living human cells, so a chemical-biology programme that screens for 'condensate dissolution' in this disease may be selecting for a chromatin artefact rather than for target engagement.
- If a knock-in tag is used to fix the attribution problem and is placed N-terminally, it will sit on the menin-binding and CXXC elements that are themselves the drug target, and the tagged line will then be the wrong model for testing menin inhibitors.
- What fraction of KMT2A::AFF1 is chromatin-bound at steady state, and how does that compare with wild-type MLL in the same lysate? Fractionation plus a size-resolving blot answers it with reagents that already exist, and no published number exists.
- Does the eps15 result generalise? KMT2A::EPS15 changes compartmentalisation in an overexpression system, but EPS15 is a cytoplasmic endocytic protein and AFF1 is already nuclear, so the expected effect size for KMT2A::AFF1 is smaller and may be below what any published assay could resolve.
- E1strong confidence 4 of 5: Replicated in peer-reviewed work by more than one group.paper
The only immunofluorescence of a KMT2A fusion in patient-derived t(4;11) cells used rabbit polyclonals against GST-AF4 and found that both t(4;11) and non-t(4;11) leukaemic cells showed a similar pattern of punctate nuclear staining, while the same reagents on a blot resolved the fusion at about 240 kDa from wild-type AF4 at 125 and 145 kDa, with the fusion the less abundant species.
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
Abstract read in full from the Europe PMC core record; no PMCID exists and the full text was not reached, so the imaging conditions, the number of lines and any quantification of the staining pattern could not be checked. 1998 confocal resolution. The audit conclusion does not depend on resolution: it depends on the absence of a size dimension in an image, which is a property of the method.
Evidence that argues against this (1)
- The paper's own blot is the counter-argument to its own image, and it is the stronger half: because the fusion is separately visible at 240 kDa, an immunoprecipitation-then-fractionation-then-blot design using the very same antibodies would have been attributable. The limitation is in the choice of readout, not in the reagent. Rogaia D et al. 1997
- E2strong confidence 4 of 5: Replicated in peer-reviewed work by more than one group.paper
ALL-1/MLL distributes in a nuclear punctate pattern; several chimeric ALL-1 proteins encoded by translocation products and expressed in transfected cells showed similar speckles, and dissection assigned the punctate distribution to two main domains within the roughly 1,100 N-terminal residues, spanning two short sequences conserved with trithorax.
Nuclear punctate distribution of ALL-1 is conferred by distinct elements at the N terminus of the protein. · Yano T, Nakamura T, Blechman J, et al. · Proc Natl Acad Sci U S A 1997What this evidence cannot show
Transfected-cell overexpression, in cultured cells rather than leukaemia blasts, and the chimeras are a panel rather than KMT2A::AFF1 specifically. The attribution is sound - the chimera is present only because it was introduced - but copy number is uncontrolled and punctate appearance is copy-number-sensitive.
Evidence that argues against this (1)
- Punctate appearance in an overexpression system is weak evidence about the endogenous protein: at physiological expression, low-complexity-domain interactions that control transcription form dynamic hubs without detectable phase separation, and the two commonest assays for droplet behaviour cannot distinguish phase separation from binding to spatially clustered sites. Guo C et al. 2020
- E3moderate confidence 3 of 5: One strong study, or several smaller consistent ones.paper
Wild-type HRX/ALL1 localised to both cytoplasm and nucleus with diffuse staining, perinuclear accumulation and nuclear bodies of variable size, whereas the HRX/ALL1-eps15 fusion from t(1;11) localised exclusively to the nucleus in bodies that were smaller and more numerous - so fusion with a partner alters compartmentalisation.
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, et al. · Cancer research 1997What this evidence cannot show
Abstract read in full from the Europe PMC core record; Cancer Research 1997 has no PMCID and the full text was not reached, so the expression system for the fusion could not be confirmed from the methods, although the comparative design implies introduction of the fusion. The fusion is KMT2A::EPS15, not KMT2A::AFF1, and EPS15 is a cytoplasmic endocytic protein, so the result may be partner-specific rather than general. The reported cytoplasmic pool of wild-type HRX/ALL1 is not seen in most later work.
Evidence that argues against this (1)
- The one endogenous experiment in a t(4;11) line reports no difference in staining pattern between rearranged and non-rearranged cells, which is the opposite conclusion - though that experiment could not have detected a difference, so the two do not actually conflict as evidence. Li Q et al. 1998
- E4moderate confidence 3 of 5: One strong study, or several smaller consistent ones.paper
The condensate framing applied to KMT2A fusions derives from super elongation complex work in which ENL initiates multivalent phase separation of the SEC, not from imaging of a KMT2A fusion at endogenous level.
ENL initiates multivalent phase separation of the super elongation complex (SEC) in controlling rapid transcriptional activation. · Guo C, Che Z, Yue J, et al. · Science advances 2020What this evidence cannot show
Cited here only to identify the provenance of the condensate framing, not to evaluate the SEC result itself. That evaluation lives in localization-condensate-evidence-standards, which records that 1,6-hexanediol sensitivity and classical FRAP both fail to distinguish phase separation from binding to spatially clustered sites.
Evidence that argues against this (1)
- 'Binding to spatially clustered sites' is a substantive rival model rather than an artefact, and in a chromatin context it is arguably the more interesting one, because clustered binding predicts sequence-specific targeting where phase separation predicts bulk enrichment. Nothing in this audit argues that the puncta are not real. Yano T et al. 1997
- E5strong confidence 4 of 5: Replicated in peer-reviewed work by more than one group.paper
An N-terminal tag on MLL::AF9 did not reproduce the leukaemogenesis kinetics of the untagged construct, which the authors attribute to interference with chromatin localisation because the DNA-binding motifs and the menin-binding domain are N-terminal; only the C-terminally tagged construct behaved like the parent.
MLL::AF9 degradation induces rapid changes in transcriptional elongation and subsequent loss of an active chromatin landscape. · Olsen SN, Godfrey L, Healy JP, et al. · Molecular cell 2022What this evidence cannot show
Read verbatim from the PMC body (PMC9044330, efetch db=pmc, <body> present). MLL::AF9 in a murine retroviral system, not KMT2A::AFF1 knock-in. Cited here as a design constraint on where a knock-in tag may be placed, which is how the paper itself uses it.
Evidence that argues against this (1)
- A C-terminal tag on KMT2A::AFF1 sits on the AFF1 portion, which carries the AFF4/ENL/P-TEFb contacts, so the constraint that rules out an N-terminal tag does not automatically make a C-terminal one safe for this fusion. Tag position has to be validated per fusion, by transformation assay, not assumed from MLL::AF9. Yano T et al. 1997
audits every published localisation claim about this protein by reagent class
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
copy-number sensitivity of punctate readouts is what makes the surviving overexpression images hard to interpret
- related to →Where KMT2A and its fusions sit inside the nucleus — SNL1/SNL2, nuclear speckles and the condensate question
audits the reagent basis of the SNL-1/SNL-2 punctate mapping that entry rests on
both classes lose the size dimension that rescues the composition and turnover classes
the same size-resolved blot that rescues turnover would rescue localisation if applied to fractionated lysate
live imaging of a HaloTag knock-in is the only attributable endogenous localisation design
verdict 3 of 5
adds a second, MLL-side instance of the unattributable-immunofluorescence shape and the fractionation route out of it
the audit's conclusion stated by the original authors in 1997
the hypothesis is currently supported only by evidence the audit found cannot discriminate it