Not one phosphorylation site has ever been mapped on a KMT2A fusion protein itself - every site in this beat is an inference across the junction
also called fusion phosphosite absence · no phospho-mapping on MLL fusions · retained-site inference
Every phosphate spot discussed in this beat was found on a normal, unfused protein - nobody has ever taken the actual cancer fusion protein out of a cell and asked where its phosphates are. So every statement that the fusion 'keeps' a switch is an argument from where the chromosome broke, not a measurement.
The same thing, for a scientist
The central claim of this beat - that certain kinase-gated contacts lie on retained partner sequence - is in every case an inference of the form 'the site is at residue X, the breakpoint distribution puts the junction at residue Y, therefore the fusion contains X'. **No study has mapped a phosphorylation site on a KMT2A fusion protein.** No immunoprecipitation of a fusion followed by phospho-MS, no phospho-specific antibody raised against a fusion, no phospho-dead knock-in of a partner residue in a fusion model appears anywhere in this beat's searches. That is a load-bearing absence and it applies equally to the previous beat's CK2/KMT2A S136-S142 result, whose sites are on the KMT2A side of the junction. It is an **absence of testing**: the assay that could see the thing has not been run, so nothing here is a negative result about fusion phosphorylation.
The queries
| query | records |
|---|---|
("fusion protein"[tiab]) AND (KMT2A[tiab] OR MLL[tiab]) AND ("phosphorylation site"[tiab] OR phosphosite[tiab] OR "phospho-specific"[tiab]) | 0 |
(KMT2A[tiab] OR MLL[tiab]) AND fusion[tiab] AND ("phospho-specific antibody"[tiab] OR "phosphospecific antibody"[tiab] OR "phospho antibody"[tiab]) | 0 |
("MLL-ENL"[tiab] OR "MLL-AF9"[tiab] OR "MLL-AF10"[tiab] OR "MLL-ELL"[tiab]) AND (phosphosite[tiab] OR "phosphorylation site"[tiab] OR phosphoproteomic*[tiab]) | 1 |
The single record in the third query is the 74-patient KMT2A-rearranged AML phosphoproteomic study, which measures the proteome of patients carrying fusions, not the fusion protein. Its site labels are gene-level (MLLT1(S359);), so a peptide from the retained portion of a partner is indistinguishable from the same peptide on the wild-type allele.
That last point is not a quibble, it is the whole methodological obstacle. A tryptic peptide from retained partner sequence is identical in the fusion and in the wild-type protein. Shotgun phosphoproteomics of a patient sample therefore cannot attribute a retained-region site to the fusion, by construction, no matter how deep it goes. Only the peptide spanning the junction itself is fusion-specific, and no junction phosphopeptide is reported anywhere.
What follows for every retention claim in this beat
Each of the following is a sequence inference and is labelled as such in its own entry:
- AF9 S395 retained, because 92% of breakpoints fall at or after AF9 residue 376.
- ENL SQ sites retained, because 84.1% of breakpoints fall at or before ENL residue 5.
- ELL T551/S561/S589 and K355 retained, because MLL-ELL contains ELL 46-621.
- AF10 S658/S689 retained, because the discarded region is the N-terminal PZP module.
All four are strong sequence arguments. None is a measurement, and three of the four rest on pooled breakpoint percentages from reviews rather than on residue-resolved primary series.
There is also a second-order problem the sequence argument cannot address: a residue present in a fusion need not be phosphorylated in it. The fusion has a different N-terminus, different folding neighbours and different subcellular partners; a kinase docking site that is on the partner's own N-terminus would be gone even where the phospho-acceptor residue remains. This graph already records the general form of that objection - the retained ELL elongation-stimulatory region is present in MLL-ELL and dispensable for leukaemogenesis.
The experiment that would close it
One experiment closes the whole class: immunoprecipitate the fusion protein from a KMT2A-rearranged cell line with a tag or an N-terminal KMT2A antibody, digest, enrich phosphopeptides, and sequence. SEM (KMT2A::AFF1), THP-1 or MOLM-13 (KMT2A::MLLT3) and KOPN-8 are standard. The readout is a list of sites on the actual oncoprotein, which this field has never had, and it would immediately test all four retention claims above plus the CK2/S136-S142 result from the previous beat. It is a one-sample experiment on a standard platform.
The junction peptide caveat cuts the other way here and helpfully: because the fusion is pulled down first, every peptide recovered is from the fusion, so the attribution problem disappears.
| status | run |
|---|---|
| fusion phosphosites mapped | 0 |
| absence class | measured-not-searched |
| methodological obstacle | retained-region tryptic peptides are identical between fusion and wild-type allele; only a junction-spanning peptide, or an immunoprecipitation of the fusion, is fusion-specific |
| closing experiment | exp-fusion-immunoprecipitation-phosphoproteomics |
| result summary | Zero phosphosites mapped on any KMT2A fusion protein. Every retention claim in this beat, and the previous beat's CK2/S136-S142 result, is a sequence inference from a breakpoint distribution. |
| absence class previous freetext | absence of testing - the assay that could detect fusion phosphorylation has not been run |
| absence class basis | PXD009338 is a nuclear-lysate pulldown of the fusion itself; a fusion-specific phosphopeptide is unassignable against a standard database and would not be reported as absent. |
| absence class assigned | 2026-09-21 (cycle 47, king) |
queries — the numbers behind the chart
[
{
"query": "(\"fusion protein\"[tiab]) AND (KMT2A[tiab] OR MLL[tiab]) AND (\"phosphorylation site\"[tiab] OR phosphosite[tiab] OR \"phospho-specific\"[tiab])",
"records": 0
},
{
"query": "(KMT2A[tiab] OR MLL[tiab]) AND fusion[tiab] AND (\"phospho-specific antibody\"[tiab] OR \"phosphospecific antibody\"[tiab] OR \"phospho antibody\"[tiab])",
"records": 0
},
{
"query": "(\"MLL-ENL\" OR \"MLL-AF9\" OR \"MLL-AF10\" OR \"MLL-ELL\")[tiab] AND (phosphosite OR \"phosphorylation site\" OR phosphoproteomic*)[tiab]",
"records": 1,
"note": "the one record is a patient-cohort study, not a study of the fusion protein"
}
]claims in this beat that depend on it
[ "mechanism-dnapk-af9-s395-sec-release", "mechanism-atm-enl-sq-sites-prc1-switch", "mechanism-atm-ell-t551-s561-s589-eaf1-switch", "mechanism-akt-p300-ell-k355-lec-sec-partition", "ana-mllt10-patient-phosphosite-isoform-resolution" ]
- Every therapeutic proposition in this beat, and the previous beat's central CK2/LEDGF claim, rests on sites measured on wild-type proteins. If the fusion turns out not to be phosphorylated at those positions - because a docking site was lost with the discarded half, or because the fusion is in a different complex - then the interventions built on them target nothing in the cancer cell while still acting on the wild-type protein in every other cell. That failure mode is silent: the drug would engage its target and do nothing to the fusion.
- Immunoprecipitate the fusion from SEM, THP-1 or MOLM-13 and run phospho-MS. Which of the inferred retained sites are actually phosphorylated on the oncoprotein?
- Is the fusion junction peptide itself ever phosphorylated? A novel site created by the junction would be the most fusion-selective drug target imaginable and nobody has looked.
- Do any of the retained sites differ in occupancy between the fusion and the wild-type partner allele in the same cell? That needs the immunoprecipitation, not a deeper shotgun run.
- Does the previous beat's CK2-phosphorylated KMT2A S136/S142 carry phosphate on the fusion protein specifically? The same experiment answers this.
- E1strong confidence 4 of 5: Replicated in peer-reviewed work by more than one group.database
PubMed returns zero records for queries combining KMT2A or MLL fusion terms with 'phosphorylation site', 'phosphosite', 'phospho-specific' or phospho-specific antibody terms, and one record for MLL-partner fusion names combined with phosphosite or phosphoproteomic terms, that record being a patient-cohort phosphoproteomic study rather than a study of the fusion protein.
PubMed esearch, three stated queries run 2026-09-19; counts 0, 0 and 1 · NCBI E-utilities 2026What this evidence cannot show
A counted absence over three stated queries, all [tiab]. It cannot see a phosphosite reported only in a supplementary table, a figure panel or a thesis, nor a paper that describes the work without using any of these terms in title or abstract. Hyphenated fusion spellings were used because PubMed's parser handles them as phrases; the double-colon spellings were also searched elsewhere in this beat and add nothing.
Evidence that argues against this (1)
- The one record that does exist demonstrates the limitation of the count rather than confirming the absence: it reports MLLT1, MLLT3, MLLT10 and ELL phosphosites in 74 KMT2A-rearranged AML patients in a supplementary matrix whose contents are not described in the abstract. If that study had not been reachable by another route, this absence claim would have looked cleaner and been less true. Casado P et al. 2023
- E2established confidence 5 of 5: Multiple independent lines of evidence; treated as settled in the field.database
A tryptic peptide from a retained region of a fusion partner is sequence-identical to the same peptide from the wild-type allele, so shotgun phosphoproteomics of patient material cannot attribute a retained-region phosphosite to the fusion protein rather than to wild-type protein.
Sequence reasoning from the UniProt canonical sequences of Q03111, P42568, P55197 and P55199, which are the sequences from which both the wild-type protein and the retained fusion portion derive · UniProt Consortium 2026What this evidence cannot show
This is a logical consequence of sequence identity, not an experimental result. It has one exception that is also the solution: a peptide spanning the fusion junction is unique to the fusion, and so is any peptide recovered after immunoprecipitating the fusion first.
Evidence that argues against this (1)
- The objection is softened where the partner's wild-type allele is poorly expressed relative to the fusion, since abundance then favours fusion-derived peptides; but that is a quantitative argument that no study in this beat makes, and KMT2A-rearranged cells retain a wild-type partner allele. Kabra A et al. 2022
- E3strong confidence 4 of 5: Replicated in peer-reviewed work by more than one group.paper
A residue retained in a fusion is not necessarily functional in it: the ELL elongation-stimulatory region is present in MLL-ELL and is dispensable for leukaemogenesis, while the retained C-terminal EAF1-interaction domain is critical.
The elongation domain of ELL is dispensable but its ELL-associated factor 1 interaction domain is essential for MLL-ELL-induced leukemogenesis. · Luo RT, et al. · Mol Cell Biol 2001What this evidence cannot show
Cited through this graph's subtraction-ell-n-terminal-eaf-site entry, which read the full text in a previous cycle. A 2001 retroviral immortalisation assay in mouse myeloid progenitors.
Evidence that argues against this (1)
- The inverse is also documented in this family and is the reason retention arguments are worth making at all: the ENL YEATS domain is retained and its deletion from MLL-ENL impairs chromatin binding, reduces stem-cell frequency and lengthens latency, so a retained module can be load-bearing. Hu H et al. 2023
quantifies the obstacle that entry names: retained-region peptides are shared, and the junction peptide needs a missed cleavage
the same gap one layer down: no fusion polypeptide has ever been identified as a species, so no modification on one can have been
adding phospho-forms of the junction peptides to the transition list would give the first attributed fusion phosphosite
supplies the mechanism for that absence: peptide-level identification cannot break a sequence tie
narrows the ambiguous set to the five KMT2A sites on retained sequence
retention is inferred, not measured
retention is inferred, not measured
applies equally to the previous beat CK2/S136-S142 result
patient shotgun data cannot attribute retained-region sites to the fusion
the experiment that closes the absence
counted absence 2
INSTRUMENT UNAVAILABLE: PhosphoSitePlus 403s on its own homepage, so the census's named instrument mismatch could not be closed and no zero was written