Oncofusion

start / graph / hyp-genotoxic-therapy-toggles-the-retained-partner-switches

Hypothesisspeculative confidence 1 of 5: Our own inference or analogy. Not shown to be true.Q25-partner-phospho

Induction chemotherapy phospho-inactivates KMT2A fusions through PIKK sites the partner retains - and a DNA-PK or ATM inhibitor would block that

also called chemotherapy phospho-inactivation hypothesis · PIKK fusion off-switch · DNA-PK inhibitor antagonism KMT2A

Three of the four partner proteins carry an off-switch operated by the kinases that respond to broken DNA, and in each case the switch sits on the part of the protein the cancer fusion keeps. Since standard chemotherapy breaks DNA, it may be turning the fusion off for a while without anyone realising - and a class of drug now being trialled alongside chemotherapy would stop that from happening.

The same thing, for a scientist

Three independent kinase-substrate relations, from three laboratories, share a pattern this beat did not expect. ENL is phosphorylated by ATM at conserved SQ sites and switches from elongation to PRC1-mediated repression; AF9 is phosphorylated by DNA-PKcs at S395 and loses SEC binding; ELL is phosphorylated by ATM at T551/S561/S589 and leaves the SEC via enhanced EAF1 association. **All three sites lie on partner sequence the fusion retains**, in three of the four commonest 3' partners, and all three consequences point the same way: the protein leaves or is expelled from the elongation machinery. The hypothesis is that KMT2A fusions inherit this PIKK-controlled off-switch, so that DNA-damaging induction chemotherapy transiently disengages the fusion from the SEC on a timescale of hours - and, correspondingly, that a DNA-PK or ATM inhibitor given as a chemosensitiser would preserve fusion-SEC engagement and could antagonise rather than potentiate chemotherapy in KMT2A-rearranged disease specifically.

Go deeper
The pattern
partnerkinasesite(s)consequenceretained?source
MLLT1 (ENL)ATMconserved SQ sites (candidates S394/S398)leaves elongation, recruits PRC1 via BMI1, H2A ubiquitination, repressionyes, ~84% of patients keep ENL from residue 5PMID 25921070
MLLT3 (AF9)DNA-PKcsS395, the only S/T-Q motif in AF9loses SEC binding (CDK9, ELL)yes, ~92% of patients break at or after 376PMID 38750015
ELLATMT551, S561, S589enhanced EAF1 binding, self-association, leaves SECyes, MLL-ELL contains ELL 46-621PMID 36305813
MLLT10 (AF10)none known--12 S/T-Q motifs, never testedthis beat's bounded absence

The convergence is not trivially explained by one laboratory: the ENL result is from Yasui's group in 2015, the ELL results from Biswas's group from 2022, and the AF9 result from Biswas's group in 2024 - so two of the three are related and one is independent. Attribute it as two laboratories, three papers, not three.

AF10 is the test case the hypothesis makes a prediction about: it has twelve PIKK motifs, is the one partner with no kinase-site relation at all, and if the pattern is real it should have one.

Why this has never been seen

Three reasons, each of which is its own entry in this beat.

  1. No one has mapped a phosphosite on a KMT2A fusion protein. Every site is from wild-type protein in a non-haematopoietic cell line.
  2. Retained-region peptides are identical between fusion and wild-type allele, so even a deep phosphoproteomic run on patient blasts cannot attribute a site to the fusion.
  3. The AF9 literature is unreachable from the obvious query because the fusion's name contains the partner's symbol; 106 records for AF9-plus-phosphorylation contain no site mapping in the first 40 titles.

So the hypothesis is not contrarian - it is what falls out of putting three papers on one page, which nothing in this field has done.

What it would change

If true:

  • Scheduling matters. A menin inhibitor given during or shortly after genotoxic induction would be acting on a fusion already partly disengaged from the SEC, which could read as synergy or as antagonism depending on the readout.
  • DNA-PK and ATM inhibitors become subtype-specific liabilities. Both classes are in oncology trials as chemo- and radio-sensitisers. In KMT2A-rearranged disease they would block the fusion's chemo-induced off-switch.
  • Part of chemotherapy's benefit in this disease is target inactivation, not cytotoxicity. That is a different mechanism from the one the field assumes and would be measurable.

If false, the finding is still useful: it would mean retained partner sequence is not phosphorylated in the fusion context, which is the single most important thing to know about every claim in this beat.

Facts and figures
statementKMT2A fusion proteins retain PIKK-phosphorylated sites on their 3' partners, and DNA-damaging chemotherapy transiently phosphorylates those sites and disengages the fusion from the super elongation complex; a DNA-PK or ATM inhibitor would block this and could antagonise chemotherapy in KMT2A-rearranged disease specifically.
origindata-pattern
prior probability0.3
laboratories behind the pattern2
papers behind the pattern3
untested partnergene-mllt10
predictions
[
  "Immunoprecipitated KMT2A::MLLT3 from an irradiated or anthracycline-treated cell line carries phospho-S395, detectable with a phospho-S/T-Q antibody.",
  "Fusion-SEC co-immunoprecipitation (fusion with CDK9 or AFF1) falls within 1-2 hours of genotoxic treatment and recovers by 6-8 hours.",
  "NU7441 or an ATM inhibitor prevents both the phosphorylation and the loss of fusion-SEC association.",
  "An S395A knock-in KMT2A::MLLT3 model shows reduced sensitivity to genotoxic chemotherapy relative to wild-type, with no difference in untreated proliferation.",
  "MLLT10, the untested partner, is phosphorylated at one or more of its twelve S/T-Q motifs after damage.",
  "In KMT2A::MLLT1 cells, damage increases fusion-BMI1 association, which is not predicted for any other partner."
]
experiments
[
  "exp-fusion-immunoprecipitation-phosphoproteomics",
  "exp-s395a-knock-in-genotoxic-sensitivity"
]
analyses
[
  "ana-partner-phosphosite-atlas-vs-uniprot",
  "ana-af9-regulatory-elements-lost-and-kept"
]
falsifiers
[
  "Immunoprecipitated fusion protein shows no phospho-S/T-Q signal after irradiation at a dose that clearly phosphorylates wild-type AF9 or ELL in the same lysate. That kills the whole hypothesis.",
  "Fusion-SEC co-immunoprecipitation is unchanged by genotoxic treatment even where the phosphorylation occurs - meaning the KMT2A half holds the complex together and the partner switch is irrelevant in the fusion.",
  "An S395A knock-in KMT2A::MLLT3 model shows no difference in genotoxic sensitivity from wild-type.",
  "Breakpoint mapping at residue resolution shows that a substantial fraction of KMT2A::MLLT3 patients break C-terminal of 395, removing the site from the fusion in those patients.",
  "A DNA-PK inhibitor combined with chemotherapy in a KMT2A-rearranged model shows the same or better efficacy than chemotherapy alone, which would refute the antagonism prediction that is the clinically actionable half."
]
How this could bite us back
  • The actionable half of this hypothesis is a warning, not a therapy: it predicts that a drug class currently developed as a chemosensitiser would be subtype-specifically harmful. If the prediction is right and nobody tests it, KMT2A-rearranged patients could be enrolled in exactly the combination that antagonises their treatment.
  • The therapeutic direction the hypothesis implies - more PIKK activation - is more DNA damage in children whose principal long-term risk is second malignancy. Any intervention built on it must find a way to phosphorylate the switch without breaking DNA, and no such agent exists.
  • If the switch is real, it also means the fusion is a repressor under damage, at least in KMT2A::MLLT1 where the consequence is PRC1 recruitment. A therapy that pushed the switch would be silencing genes near breaks genome-wide, not selectively at fusion targets.
  • Testing this requires a phospho-specific reagent for ENL SQ sites, and the paper that names those residues cannot be read. Building the reagent from the sequence inference risks targeting the wrong residues.
Still open
  1. Does the fusion's KMT2A half - menin, LEDGF, PAF1 - hold the SEC in place regardless of partner phosphorylation? That is the most likely way the hypothesis fails and it is the first thing to test.
  2. Is the effect large enough to matter clinically, or is it a transient hour-scale dip with no consequence for leukaemic burden?
  3. Does any existing clinical dataset stratify KMT2A-rearranged outcomes by concurrent use of a DNA-damage-response inhibitor? Probably not, but ATM/DNA-PK inhibitor trials are running.
  4. If the switch is real, could it be pumped deliberately without genotoxicity - for instance by an activator of PIKK signalling that does not require DNA breaks?
Every source behind this entry (4)
  1. E1strong confidence 4 of 5: Replicated in peer-reviewed work by more than one group.paper

    DNA-PKcs phosphorylates AF9 at S395 during genotoxic stress and phospho-S395 AF9 loses interaction with SEC components including CDK9 and ELL; the S395A mutant retains SEC interaction after irradiation and supports enhanced global transcription.

    PMID 38750015DOI
    What this evidence cannot show

    HEK293T, wild-type AF9, no fusion. Several supporting panels are n=1 or n=2.

    Evidence that argues against this (1)
    • The same paper's acetylation arm shows the AF9 K339R mutant retains chromatin binding after irradiation by a mechanism the model does not explain, so the authors themselves report that damage-induced regulation of AF9 is incompletely accounted for by the phospho and acetyl switches together. Kabra A et al. 2022
  2. E2strong confidence 4 of 5: Replicated in peer-reviewed work by more than one group.paper

    ATM phosphorylates ELL at T551, S561 and S589 during genotoxic stress, enhancing EAF1 association and ELL self-association and reducing ELL interaction with other SEC components, producing global transcriptional inhibition.

    PMID 36305813DOI
    What this evidence cannot show

    Full text read; human cell lines, doxorubicin, no fusion and no haematopoietic system. Only the triple mutant was tested, so individual residue contributions are unresolved. A published correction applies.

    Evidence that argues against this (1)
  3. E3moderate confidence 3 of 5: One strong study, or several smaller consistent ones.paper

    ATM phosphorylates ENL at conserved SQ sites in response to double-strand breaks, switching it from transcription elongation to PRC1-mediated repression through increased BMI1 binding.

    PMID 25921070DOI
    What this evidence cannot show

    Abstract only; the full text is not retrievable by any of this project's four routes because no PMCID exists. Residue numbers, cell system and effect sizes are unknown to this entry.

    Evidence that argues against this (1)
    • The authors' own review states ATR and DNA-PKcs use the same SQ sites, so the specific kinase in any context is undetermined - which weakens 'ATM' in the hypothesis but strengthens the hypothesis itself, since it means any PIKK activated by chemotherapy could do it. Ui A et al. 2016
  4. E4strong confidence 4 of 5: Replicated in peer-reviewed work by more than one group.paper

    The breakpoint distributions place all three sites on retained partner sequence: about 84.1% of MLL-ENL patients break at or before ENL residue 5, about 92% of MLL-AF9 patients break at or after AF9 residue 376 with S395 C-terminal of it, and MLL-ELL contains ELL residues 46-621.

    PMID 36435883DOI
    What this evidence cannot show

    Pooled distributions from reviews plus a 2001 construct boundary, not residue-resolved primary breakpoint series. Retention of a residue is not evidence that the residue is phosphorylated in the fusion.

    Evidence that argues against this (1)
    • Retention does not imply function: the retained ELL elongation-stimulatory region is dispensable for MLL-ELL leukaemogenesis while a different retained module is essential, so a retained residue can be present and inert. Luo RT et al. 2001
Connected to