Oncofusion

start / graph / concept-anti-mllt1-antibodies-the-cleanest-partner-reagent-and-the-worst-confound

Explainermoderate confidence 3 of 5: One strong study, or several smaller consistent ones.Q27-fusion-attribution

Anti-ENL antibodies are the cleanest partner reagent in this disease and sit on top of the worst confound

also called anti-MLLT1 attribution · ENL antibody specificity · ENL-C ChIP in KMT2A::MLLT1

In KMT2A::MLLT1 leukemia an antibody against ENL sees only two proteins - the normal one and the fusion - which is one fewer than in the AF9 case. The trouble is that normal ENL is itself a cancer-driving protein sitting on exactly the same genes, so mixing the two signals is worse here than anywhere else.

The same thing, for a scientist

On a species count, anti-ENL-C is the cleanest partner reagent in the KMT2A field. Three reasons: (i) about 50% of KMT2A::MLLT1 fusions (241 of 460) break upstream of MLLT1 exon 1 as spliced fusions expressing **no reciprocal transcript** [E1], so in that half there is no reciprocal product to confound an MLL-C measurement and no MLLT1-N-bearing species at all; (ii) the internal promoter that produces a YEATS-devoid truncated isoform in MLLT3 exists in MLLT1 but is **inactive in the haematopoietic lineage and in AML** [E2], so the third-species problem that afflicts anti-AF9 reagents does not arise here; (iii) the retained ENL region is the AHD, a compact and well-characterised epitope. On an interpretive count, it is the worst place to pool signals. Wild-type ENL is not a passive bystander in KMT2A-rearranged leukaemia: it is a YEATS-domain acetyl-reader that is itself a validated dependency and a live drug target, with selective degraders now published. So an ENL-C immunoprecipitate or ChIP in a KMT2A::MLLT1 cell mixes the fusion with a wild-type protein that is independently required, at the same loci, for the same programme. **Fewer species does not mean easier attribution when the extra species is itself causal.**

Go deeper
The comparator arm this creates

The reciprocal-negative half of KMT2A::MLLT1 is the only place in this disease where the partner gene is held constant across a reciprocal-positive and a reciprocal-negative population of comparable size - roughly 241 against 219. This graph already identifies that as the untested test of the reciprocal-product hypothesis. The attribution argument adds a second use for the same population: an anti-MLL-C experiment run in both arms bounds the reciprocal product's contribution to MLL-C signal empirically, without needing any new reagent.

The two uses are independent and both are free, in the sense that the samples exist in a diagnostic archive.

What is not readable here

The principal MLL-ENL chromatin study - 165 direct target genes split into two clades by chromatin state and inhibitor sensitivity - could not be read. Europe PMC returns no PMCID for it (isOpenAccess: N), so routes 2, 3 and 4 of the unreadable test are unavailable by construction. Its antibody or tagging strategy is therefore unknown to this beat and no claim is made about whether its occupancy assignment is attributable 3. It is recorded here rather than guessed at.

Facts and figures
reagent classantibodies against the ENL/MLLT1 C-terminal AHD (retained) region
species count vs mllt3two rather than three - the MLLT1 internal-promoter isoform is inactive in haematopoiesis
why pooling is worse herewild-type ENL is an independently validated dependency and drug target in the same cells and at the same loci, so the confounding species is causal rather than inert
free comparator armreciprocal-positive versus reciprocal-negative KMT2A::MLLT1, partner held constant
species recognised
[
  "wild-type ENL",
  "KMT2A::MLLT1 fusion"
]
reciprocal negative fraction
{
  "n": 241,
  "total": 460,
  "mechanism": "breakpoint upstream of MLLT1 exon 1"
}
unreadable
[
  "27050521"
]
How this could bite us back
  • ENL degraders and YEATS inhibitors remove wild-type ENL as well as the fusion's ENL portion, and wild-type ENL is required in normal haematopoiesis - so a therapeutic index argument here cannot be built on fusion selectivity, because no reagent distinguishes them.
  • Choosing KMT2A::MLLT1 as the clean system for attribution work imports a different disease: MLLT1 fusions differ from t(4;11) in age distribution and lineage, so a mechanism established there needs an explicit transfer argument before it is applied to infant ALL.
Still open
  1. Does anti-MLL-C signal differ between reciprocal-positive and reciprocal-negative KMT2A::MLLT1 samples? A null bounds the reciprocal product's contribution across the whole field; a difference measures it. Nobody has run it and the samples exist.
  2. In an ENL-C immunoprecipitate from a KMT2A::MLLT1 cell, what fraction of the recovered protein is fusion? Wild-type ENL is abundant and functionally engaged, so the ratio is the number that decides how to read every ENL co-IP in this disease.
Every source behind this entry (3)
  1. E1strong confidence 4 of 5: Replicated in peer-reviewed work by more than one group.paper

    About half of KMT2A::MLLT1 fusions (241 of 460) break upstream of MLLT1 exon 1 and are spliced fusions expressing only the direct fusion transcript and no reciprocal one.

    The KMT2A recombinome of acute leukemias in 2023. · Meyer C, Larghero P, Almeida Lopes B, et al. · Leukemia 2023
    PMID 37019990DOI
    What this evidence cannot show

    Read verbatim from PMC10169636 on 2026-09-20. A transcript-level inference from breakpoint position in a diagnostic archive, not a per-case measurement of reciprocal transcript. It says nothing about KMT2A*, which the truncated 3'-KMT2A allele could still produce.

    Evidence that argues against this (1)
    • The spliced-fusion architecture removes the reciprocal fusion by leaving a truncated 3'-KMT2A allele, which is exactly the substrate for the gene-internal promoter that produces KMT2A*, so the reciprocal-negative arm may not be as clean as the transcript count suggests. Scharf S et al. 2007
  2. E2strong confidence 4 of 5: Replicated in peer-reviewed work by more than one group.paper

    The MLLT1 locus carries an internal promoter producing a YEATS-domain-devoid ENL isoform, but it is inactive in the haematopoietic lineage and in AML, being functional instead in pituitary, brain, thyroid, uterus and lung.

    PMID 42121261DOI
    What this evidence cannot show

    Abstract read verbatim. 'Inactive in AML' is a statement about the cohorts profiled; KMT2A::MLLT1 leukaemias were not singled out. The contrast with MLLT3 is the load-bearing part and it is stated directly by the authors.

    Evidence that argues against this (1)
    • The companion paper shows the equivalent MLLT3 promoter is massively active in HSCs and the most lethal AML, so the architecture is present in both genes and only the tissue pattern differs - which means a haematopoietic context in which the MLLT1 promoter switches on cannot be excluded on current data. Bessière C et al. 2025
  3. E3weak confidence 2 of 5: Preliminary, small, indirect, or not yet peer reviewed.paper

    The principal MLL-ENL occupancy study reports 165 direct target genes in two clades distinguished by chromatin state and by sensitivity to DOT1L versus P-TEFb inhibition; its antibody or tagging strategy could not be determined because the full text is unreadable by all four routes.

    PMID 27050521DOI
    What this evidence cannot show

    Recorded as known-but-unreadable. Europe PMC core record gives no pmcid and isOpenAccess is N, so the fullTextXML, efetch db=pmc and PMC HTML routes are all unavailable. Only the abstract was read. **No inference is drawn about what the paper's occupancy assignment rests on.**

    Evidence that argues against this (1)
    • Murine MLL-ENL systems are typically retrovirally expressed rather than endogenous, which changes both the attribution question and the stoichiometry; but whether that applies to this study was not verifiable here. Sykes DB et al. 2003
Connected to