Oncofusion

start / graph / mechanism-atm-enl-sq-sites-prc1-switch

Mechanismmoderate confidence 3 of 5: One strong study, or several smaller consistent ones.Q25-partner-phospho

ATM phosphorylates ENL at conserved SQ sites and flips it from elongation to Polycomb repression - on sequence every KMT2A::MLLT1 fusion keeps

also called ENL SQ sites · MLLT1 ATM phosphorylation · ENL BMI1 PRC1 switch · ENL DSB repression

ENL normally helps genes get read out, but when DNA nearby breaks the alarm kinase ATM tags it and the tagged ENL recruits a silencing complex instead - the same protein switching from 'on' to 'off'. Because KMT2A::MLLT1 fusions keep almost all of ENL, the fusion should carry this switch too.

The same thing, for a scientist

In response to DNA double-strand breaks ENL (MLLT1) is phosphorylated by ATM at conserved SQ sites. Phospho-ENL binds the PRC1 E3-ubiquitin-ligase complex more strongly via BMI1, enriching PRC1 at transcription-elongation sites near breaks, ubiquitinating H2A and repressing transcription; the ENL SQ sites and BMI1 are required for KU70 accumulation at breaks near active transcription and for cellular DSB resistance. The authors frame it as a switch from elongation to Polycomb-mediated repression. **Roughly 84.1% of MLL-ENL patients break at or before ENL residue 5**, so essentially the whole of ENL - and therefore this switch - is on retained sequence. This is the most complete transfer of a partner's regulatory apparatus into a fusion anywhere in this graph.

Go deeper
What the primary report establishes

ATM phosphorylates ENL at conserved SQ sites after DSB induction; the phosphorylation increases ENL-PRC1 interaction through BMI1; PRC1 is enriched at elongation sites near breaks; H2A is ubiquitinated; transcription is repressed. ENL SQ sites and BMI1 are both necessary for KU70 accumulation at breaks near active transcription and for cellular resistance to DSBs 1.

The residue numbers are not stated here, deliberately. The primary paper is not retrievable in full text: Europe PMC returns no PMCID for PMID 25921070, the NCBI ID converter returns 'Identifier not found in PMC', and elink dbfrom=pubmed db=pmc returns nothing, so routes two, three and four of this project's unreadable test cannot even be attempted. The abstract says 'conserved SQ sites' and that is all this entry asserts about which residues.

Which residues they must be, as a separate and weaker claim

A scan of the ENL canonical sequence performed in this session finds exactly two S/T-Q motifs in all 559 residues, at S394 and S398 2. The companion review frames the mechanism throughout as acting on 'ENL/AF9' at 'evolutionarily well-conserved SQ-sites' 3, and the AF9 site subsequently mapped by a different laboratory is S395, AF9's only S/T-Q motif, which aligns with ENL S394: the two proteins share the context ...P-S-Q...G-P-L-R-S... at that position 2.

So the conserved SQ site of the ENL/AF9 pair is almost certainly ENL S394 / AF9 S395, with ENL carrying an extra site at S398 that AF9 lacks. This is an inference from sequence plus a second paper, not a reading of the primary source, and this project's standing rule is that a motif match means almost nothing on its own. It is recorded because the motif count is two rather than dozens and because an independently mapped functional site in the paralogue lands on one of them.

Neither S394 nor S398 appears in the PeptideAtlas set for Q03111. That is expected for a damage-induced site and is also exactly the pattern seen for AF9 S395.

Retained or lost: retained, essentially in full

About 84.1% of MLL-ENL patients break at or before ENL residue 5 4. Both candidate SQ sites, and all 23 observed phosphosites on ENL, are far C-terminal of that and are retained.

The consequence is specific and awkward: KMT2A::MLLT1 should carry an ATM-controlled PRC1-recruiting switch. A fusion protein that can be flipped into recruiting Polycomb is a different object from one that only drives elongation, and this graph already holds localization-polycomb-bodies-kmt2a-r, mechanism-polycomb-prc1-prc2-cbx8 and mechanism-dot1l-prc1-antagonism without any of them connecting to a kinase input on the fusion itself.

The minority ENL architecture that breaks at exon 6-7 excludes the YEATS domain; whether it also excludes S394/S398 is not determinable from anything this beat found, because that architecture's junction has not been mapped to residue.

Facts and figures
kinasegene-atm
kinase nameATM (with ATR and DNA-PKcs able to use the same motif)
substrategene-mllt1
uniprotQ03111
siteconserved SQ sites; residue numbers NOT stated in the accessible record
candidate site basisthe only two [ST]Q motifs in Q03111; S394 aligns with the independently mapped AF9 S395
gated contactPRC1 binding via BMI1, leading to H2A ubiquitination and transcriptional repression
directionphosphorylation INCREASES PRC1 binding and switches elongation to repression
mapped inwild-type ENL
mapped in fusionfalse
retained in fusiontrue
retained basisabout 84.1% of MLL-ENL breakpoints fall at or before ENL residue 5
observed in peptideatlasfalse
primary paper readablefalse
gene symbolMLLT1
candidate sites from sequence
[
  394,
  398
]
unreadable routes tried
[
  "EPMC search EXT_ID core -> no pmcid",
  "NCBI PMC ID converter -> Identifier not found in PMC",
  "elink dbfrom=pubmed db=pmc -> no linked uid",
  "fullTextXML, efetch db=pmc and PMC HTML all unreachable without a PMCID"
]
How this could bite us back
  • An ATM inhibitor combined with chemotherapy in a KMT2A::MLLT1 leukaemia would, on this mechanism, prevent the fusion from switching into its repressive mode after damage - leaving it driving elongation through a genotoxic insult. That is a plausible and unmeasured antagonism.
  • If the fusion can be flipped into recruiting PRC1, then a therapy that raises ENL phosphorylation would silence genes indiscriminately near breaks, not selectively at fusion target loci. Repression is not the same as removing the oncoprotein.
  • BMI1 and PRC1 are required for normal haematopoietic stem-cell self-renewal, so any intervention aimed at this interface has a stem-cell toxicity floor independent of the leukaemia.
  • The residue identity is an inference. Designing a phospho-specific antibody or a phospho-dead knock-in against S394/S398 on this basis risks building reagents against the wrong residues, and the primary paper that would settle it cannot currently be read.
Still open
  1. Which residues are the ENL SQ sites? The primary paper states them and is unreadable; a single email to the authors, or a reprint, would convert a moderate-confidence inference into a fact.
  2. Does KMT2A::MLLT1 recruit PRC1 after DNA damage? If it does, the fusion is a kinase-switchable repressor as well as an activator, which no model of KMT2A-rearranged leukaemia contains.
  3. This graph holds localization-polycomb-bodies-kmt2a-r and mechanism-dot1l-prc1-antagonism with no kinase input. Is ATM the missing input?
  4. Does the minority ENL exon 6-7 architecture, which excludes the YEATS domain, also exclude S394/S398? That junction has never been mapped to residue.
Every source behind this entry (4)
  1. E1strong confidence 4 of 5: Replicated in peer-reviewed work by more than one group.paper

    In response to DNA double-strand breaks ENL (MLLT1) is phosphorylated by ATM at conserved SQ sites; this increases ENL interaction with the PRC1 E3-ubiquitin-ligase complex via BMI1, promoting PRC1 enrichment at transcription elongation sites near breaks, H2A ubiquitination and transcriptional repression, with ENL SQ sites and BMI1 required for KU70 accumulation and cellular DSB resistance.

    PMID 25921070DOI
    What this evidence cannot show

    ABSTRACT ONLY - the full text could not be obtained. Routes tried 2026-09-19: Europe PMC core record for EXT_ID:25921070 returns no pmcid; the NCBI PMC ID converter returns 'Identifier not found in PMC'; elink dbfrom=pubmed db=pmc returns no linked uid, so fullTextXML, efetch db=pmc and the PMC HTML page are all unreachable for want of a PMCID. Consequently the specific residue numbers, the cell system, the number of SQ sites mutated and the magnitude of every effect are unknown to this entry. Nothing about the experiments is asserted beyond what the abstract states.

    Evidence that argues against this (1)
    • The authors' own companion review states that ATR and DNA-PKcs phosphorylate the same SQ sites as ATM, and explicitly raises that the ENL/AF9 phosphorylation 'may arise from ATR and DNA-PKcs' in response to events other than DSBs. Attributing this switch to ATM specifically is therefore narrower than the authors themselves are willing to be. Ui A et al. 2016
  2. E2strong confidence 4 of 5: Replicated in peer-reviewed work by more than one group.database

    ENL (Q03111) contains exactly two S/T-Q motifs in 559 residues, at S394 and S398; AF9 (P42568) contains exactly one, at S395; and the two align, both sitting in a shared P-S-Q ... G-P-L-R-S context (ENL 380-410 SSSSDSSSDSDFEPSQNHSQGPLRSMVEDLQ, AF9 380-410 SPASSSSSSSSSFTPSQTRQQGPLRSIMKDL).

    What this evidence cannot show

    A sequence computation done in this session, not a published alignment and not a structural superposition. It identifies candidate residues; it does not show that either is phosphorylated. The comparison is a direct index-for-index comparison of two sequences of similar length, not the output of an alignment algorithm, so the claim of correspondence is a visual one over a 31-residue window.

    Evidence that argues against this (1)
    • Neither ENL S394 nor S398 is reported as phosphorylated in the PeptideAtlas/PTMeXchange re-analysis of Q03111, which finds 23 sites and none in that region. This project has already retracted one motif-based inference when the phosphosite atlas showed the candidate residue is not phosphorylated at all, and the same objection applies here with the same force. EMBL-EBI 2026
  3. E3moderate confidence 3 of 5: One strong study, or several smaller consistent ones.paper

    The companion review by the same authors describes the mechanism as ATM phosphorylating ENL/AF9 at evolutionarily well-conserved SQ-sites, with the phosphorylation recruiting PRC1 through BMI1 to ubiquitinate H2A and repress transcription, and notes that ATR and DNA-PKcs phosphorylate the same SQ sites as ATM.

    PMID 27310306DOI
    What this evidence cannot show

    A review, read in full via the PMC HTML page (PMC4916901) after fullTextXML returned HTTP 500 and efetch db=pmc returned a document with no body element - route four of the unreadable test. It is the authors' own commentary on their own primary paper, so it is not independent of E1, and it gives no residue numbers either.

    Evidence that argues against this (1)
    • A structural review of MLLT1 and MLLT3 from a different laboratory describes both proteins' regulatory repertoire without mentioning ATM phosphorylation or a PRC1 switch at all, listing the AHD partners AF4, DOT1L, BCOR and CBX8 and ENL's phase separation as the functional axes. The DSB switch has had little uptake in the structural literature on these proteins in the decade since. Kabra A et al. 2022
  4. E4strong confidence 4 of 5: Replicated in peer-reviewed work by more than one group.paper

    Approximately 84.1% of MLL-ENL patients break at or before ENL residue 5, so essentially the whole ENL protein, including any site C-terminal of residue 5, is retained in the fusion.

    PMID 36435883DOI
    What this evidence cannot show

    Read at abstract level and through this graph's subtraction-mllt1-yeats-retained entry which cites the same paper for the same figure. A minority exon 6-7 architecture excludes the YEATS domain and its junction has not been mapped to residue here, so the fate of S394/S398 in that subgroup is unknown.

    Evidence that argues against this (1)
    • Retention of sequence is not retention of function. The same review that supplies the AF9 breakpoint figure treats ENL and AF9 as a functional pair despite their opposite retention patterns, which implies the retained ENL sequence is not doing something categorically different in the fusion - an implication that contradicts the reading offered here. Hu H et al. 2023
Connected to