Oncofusion

start / graph / exp-yeats-backbone-swap-matrix-mllt1-mllt3

Experimentspeculative confidence 1 of 5: Our own inference or analogy. Not shown to be true.Q29-mllt3-yeats-loss

The 2x2 paralogue swap: separate 'is the reader there' from 'which backbone is it on', in one background with matched expression

also called YEATS backbone swap · MLL-AF9 vs MLL-ENL matched comparison

Build four versions of the cancer protein in the same cells - AF9 with and without its reader, ENL with and without its reader - and measure them the same way. Nobody has ever run these two fusions side by side in one experiment.

The same thing, for a scientist

Every claim in this beat about what the YEATS asymmetry means rests on comparing results produced in different laboratories, in different years, with different vectors, in different progenitor pools. **This beat found no published experiment in which MLL-AF9 and MLL-ENL were expressed in one background at matched levels and scored on the same readouts.** The design is a 2x2: backbone (AF9 or ENL) crossed with reader (present or absent), plus an aromatic-cage-dead reader arm on each YEATS-containing backbone as the discriminating control, all in one retroviral vector series in one murine progenitor pool, expression matched by immunoblot before any phenotype is scored. Readouts: serial replating, transplantation latency, Hoxa9 and Meis1 transcripts, CUT&RUN for fusion occupancy and for H3K79me2/me3 and H3K9ac co-occupancy, and SGC-iMLLT sensitivity. The design isolates the variable that the existing in-cis add-back proposal (`exp-yeats-cis-resupply-kmt2a-mllt3`) cannot: **whether a reader helps because it is a reader, or because of what it is attached to.**

Go deeper
The eight constructs
#constructrole
1MLL-AF9 (patient-like, AF9 ~376-568)the field's default; YEATS absent
2MLL-YEATS(AF9)-AF9reader restored on its own backbone
3MLL-YEATS(AF9, cage-dead)-AF9same fold, no acyl-lysine binding
4MLL-linker-AF9length-matched inert insert
5MLL-ENL (patient-like, full ENL from aa 5)YEATS present
6MLL-ENL(deltaYEATS)reader removed on its own backbone
7MLL-ENL(cage-dead)same fold, no acyl-lysine binding
8empty vectorbaseline

Constructs 5-7 reproduce the published MLL-ENL arms 1; constructs 1-4 reproduce the proposed MLL-AF9 arms. The point of the matrix is that 1-4 and 5-7 have never been run together.

The cage-dead arms are what make it interpretable. Comparing 2 against 1 confounds 'reader added' with '138 folded residues added'; comparing 2 against 3 does not, and the same logic runs on the ENL side between 5 and 7.

The four questions it answers that no existing experiment does
  1. Is the reader's contribution backbone-dependent? If 2-minus-3 and 5-minus-7 are similar, YEATS does the same job on either backbone and the asymmetry is purely about which patients exist. If they differ, the backbone matters and the two fusions are mechanistically distinct beyond the reader.
  2. Is MLL-AF9 handicapped or merely different? Construct 2 against 1 says whether the field's default assay has been run on a weakened molecule for thirty years.
  3. Does H3K9ac co-occupancy follow the reader? CUT&RUN for fusion occupancy alongside H3K9ac across the four YEATS states gives the chromatin-level answer directly, rather than inferring it from target-gene transcripts.
  4. Is SGC-iMLLT sensitivity a fusion property or a wild-type property? Construct 1 has no fusion YEATS and constructs 5 and 2 do; if 1 is as sensitive as 5, the inhibitor is acting on wild-type ENL and AF9, which is what the affinity data predict 3.
What it cannot settle, and what to run alongside

It cannot settle whether patients' breakpoints are selected or geographic. That is a DNA question and this beat's companion analysis argues the geography alone predicts the direction of the asymmetry 5. The cheap orthogonal experiment is a topoisomerase-II cleavage map over MLLT1 and MLLT3 in etoposide-treated human haematopoietic progenitors, asking whether cleavage hotspots coincide with the observed patient breakpoint zones - which would explain the distribution without any appeal to selection.

It also cannot speak to lineage. All eight constructs go into myeloid progenitors, and 42% of KMT2A::MLLT3 patients have ALL. A second matrix in a lymphoid-permissive system would be the natural follow-on and is out of scope for a first pass.

And it inherits the species import: this is murine retroviral transduction, the assay whose limits this graph documents in import-class-murine-mll-af9-myeloid. The design improves the internal comparison, not the external one.

Facts and figures
design2x2 factorial - backbone (AF9 or ENL) x reader (wild-type YEATS, aromatic-cage-dead YEATS, or absent) - as eight retroviral constructs in one vector series, transduced into one pool of murine haematopoietic progenitors, with expression matched by immunoblot before any phenotype is scored.
readoutserial replating over four weeks; transplantation latency and leukaemic stem-cell frequency by limiting dilution; Hoxa9 and Meis1 transcripts by qRT-PCR; CUT&RUN for fusion occupancy, H3K79me2, H3K79me3 and H3K9ac; co-immunoprecipitation of DOT1L and AFF4 from each arm; SGC-iMLLT dose-response.
controlsAromatic-cage-dead YEATS on each backbone (separates reader function from added fold); an inert length-matched linker on the AF9 backbone (separates added fold from added length); empty vector; expression matching by immunoblot with a shared epitope tag. NOT controlled: endogenous wild-type Mllt1 and Mllt3 in the recipient progenitors, which are asymmetric between the arms because ENL is required for AML maintenance and AF9 is not.
cost tiermed
timelinecloning and expression matching 2-3 months; replating 6 weeks; transplantation cohorts 6-9 months; CUT&RUN and co-IP in parallel.
primary comparison(construct 2 - construct 3) versus (construct 5 - construct 7): is the reader's contribution the same size on both backbones?
orthogonal experimenttopoisomerase-II cleavage mapping across MLLT1 and MLLT3 in etoposide-treated human haematopoietic progenitors, to test the geography model directly
what it cannot settle
[
  "whether patient breakpoints are selected or geographic",
  "anything about lymphoid lineage, where 42% of KMT2A::MLLT3 patients sit",
  "species transfer"
]
tests hypotheses
[
  "hyp-the-yeats-asymmetry-is-breakage-geography-not-selection"
]
How this could bite us back
  • A positive result - the reader helps on both backbones - would argue for developing YEATS inhibitors, which in this graph's own analysis have no fusion target in 99.5% of t(9;11) patients; the experiment could therefore justify a drug for a population it does not serve.
  • A negative result in murine myeloid progenitors would be read as closing the question, when 42% of this partner's patients have lymphoid disease and the matrix does not touch them.
Still open
  1. Should the matrix be run on Mllt1/Mllt3 double-knockout progenitors to remove the asymmetric endogenous background, at the cost of a much harder experiment?
  2. Is there a human t(9;11) or t(11;19) line in which the endogenous fusion allele could be edited to swap YEATS status, which would avoid the retroviral import entirely?
  3. What is the minimum expression-matching tolerance at which a negative result would be believable? The published MLL-ENL and MLL-AF9 literatures use different vectors and never report cross-comparable levels.
Every source behind this entry (5)
  1. E1strong confidence 4 of 5: Replicated in peer-reviewed work by more than one group.paper

    Deleting the YEATS domain from MLL-ENL impairs fusion-protein binding and decreases expression of pro-leukemic genes such as Eya1 and Meis1; point mutations that disrupt YEATS binding to acetylated histones decrease stem-cell frequency and increase MLL-ENL-mediated leukemia latency; and YEATS-containing MLL-ENL leukemic cells show increased sensitivity to the YEATS inhibitor SGC-iMLLT compared with control AML cells.

    PMID 36435883DOI
    What this evidence cannot show

    Read verbatim from the PMC HTML body (PMC11246743) on 2026-09-20. Murine retroviral MLL-ENL with transplantation; the SGC-iMLLT comparison is against control AML cells rather than an isogenic YEATS-deleted line, which is precisely the gap this design closes.

    Evidence that argues against this (1)
    • The comparison that would make the SGC-iMLLT result fusion-attributable - YEATS-containing against YEATS-deleted MLL-ENL in the same background - is not reported, so the published sensitivity difference is confounded by everything that differs between the MLL-ENL cells and the control AML cells. Hu H et al. 2023
  2. E2strong confidence 4 of 5: Replicated in peer-reviewed work by more than one group.paper

    Approximately 98.4% of 442 t(9;11) patients break downstream of MLLT3 exon 4, so the patient-like MLL-AF9 construct is the YEATS-absent architecture, while 84.1% of 302 t(11;19) patients retain the full ENL YEATS domain, making the YEATS-containing MLL-ENL construct the patient-like one for that partner.

    PMID 36435883DOI
    What this evidence cannot show

    Genomic breakpoint distributions from a referral archive; no patient fusion protein has been measured. The 'patient-like' label on each construct is therefore an inference from DNA.

    Evidence that argues against this (1)
    • The residue at which the AF9 moiety should begin is uncertain: the exon-4 boundary and the reviewed residue-376 boundary are different statements, and constructs in the literature begin at various points, so 'patient-like MLL-AF9' is not a single defined molecule. Hu H et al. 2023
  3. E3strong confidence 4 of 5: Replicated in peer-reviewed work by more than one group.paper

    DOT1L peptides bind ENL with affinities similar to those measured for AF9, and structure-guided AF9 mutants give graded rather than binary effects on DOT1L recruitment, H3K79 methylation and transformation.

    PMID 25921540DOI
    What this evidence cannot show

    In vitro peptide binding on isolated tagged domains plus murine serial replating; the ENL comparison is a supplementary panel.

    Evidence that argues against this (1)
    • Equal peptide affinity does not predict equal behaviour in cells, where ENL's DOT1L and AF4-family contacts are reported to be mutually exclusive; the matrix design should therefore measure DOT1L and AFF4 co-IP in each arm rather than assuming parity. Wan L et al. 2017
  4. E4strong confidence 4 of 5: Replicated in peer-reviewed work by more than one group.paper

    ENL, but not its paralogue AF9, is required for disease maintenance in acute myeloid leukaemia, so any construct series expressed on a wild-type background carries a partner-asymmetric endogenous contribution.

    PMID 28241141DOI
    What this evidence cannot show

    CRISPR depletion in AML lines and mouse models. Relevant to the design because the wild-type paralogues are present in the recipient progenitors and are not controlled by the construct series.

    Evidence that argues against this (1)
    • MLLT3 is required for normal human HSC self-renewal, so the endogenous background is not inert on the AF9 side either, and a clean comparison would need Mllt1 and Mllt3 knockout progenitor pools - which doubles the size of the experiment. Calvanese V et al. 2019
  5. E5moderate confidence 3 of 5: One strong study, or several smaller consistent ones.database

    In hg38 the two genes have identical exon-to-residue mapping over the YEATS domain but MLLT1 intron 1 is 9,013 bp against MLLT3's 1,410 bp, so uniform intronic breakage alone predicts a 26.9-fold difference in YEATS retention, in the observed direction.

    What this evidence cannot show

    Computed in this beat from one canonical transcript per gene; the same computation fails badly on the partial-YEATS class, so it is a reason for caution rather than a finding. Full treatment in `ana-yeats-asymmetry-tested-against-intron-geometry`.

    Evidence that argues against this (1)
    • If geography rather than selection sets the distribution, the entire matrix measures the behaviour of molecules that were never selected for or against, which changes what a positive result would mean. Meyer C et al. 2018
Connected to