Oncofusion

start / graph / interface-kmt2a-lef1-declined-a-t-lineage-factor-in-a-t-lineage-host

Interfaceweak confidence 2 of 5: Preliminary, small, indirect, or not yet peer reviewed.Q180-the-af4-specific-interactors

KMT2A::AFF1–LEF1 is one AP-MS row (p 0.0261, 10|8 spectra) in a Jurkat T-ALL host, where LEF1 is a resident lineage factor; it is the only one of the six AF4-specific candidates with non-zero negative-control signal, and no Wnt co-factor was co-recovered — declined

also called MLL-AF4-LEF1 · KMT2A-LEF1 · LEF1 as an AF4-specific interactor

LEF1 was the most promising of six proteins said to bind only the MLL-AF4 fusion. But the experiment was done in T-cells, LEF1 is a T-cell protein, and LEF1 was the only one of the six that also showed up in the negative control. We are declining it as evidence of a real partnership while keeping the Wnt question open.

The same thing, for a scientist

**Verdict: DECLINED as evidence for a KMT2A::AFF1–LEF1 contact.** The whole of the evidence is one prey row in Skucha 2018 Supplementary Table 5: MLL-AF4 p = 0.0261, spectral counts 10|8 (18 summed), rank 62 of 291, `n_fusions` = 1, neg_ctrl **1|3** [E1]. Four things count against it and one for it. **Against:** (i) the host is Jurkat T-ALL and LEF1 is UniProt's *"T cell-specific transcription factor 1-alpha"*, so an abundant resident nuclear factor is being scored against a baseline estimated in the same cells [E2]; (ii) it is the **only** one of the six p<0.05 AF4-specific candidates with non-zero control signal (RGPD8, ACADM, RPA2, CEP112 are 0|0 and GLUD1 is 0|1) [E1]; (iii) β-catenin and every other TCF/LEF member are absent from all seven baits' lists, so the bridge the mechanism needs was not co-recovered [E3]; (iv) the class this row belongs to — MLL-AF4 singletons — has a median p of 0.114 and its only credible members elsewhere in the table sit at p ≤ 0.0017 on ≥24 spectra. **For:** there is a published mechanism, β-catenin recruiting MLL-family H3K4 activity to Wnt targets [E4], and 18 spectra is the highest count of the six. That is not enough, and the decline is on the evidence, not on the biology.

Go deeper
The six candidates side by side, with the control column

Read back from the sheet with the column map asserted 1:

preyUniProtMLL-AF4 pspectra (a|b, sum)neg_ctrlrank/291
RGPD8O147150.01472|1 = 30|050
ACADMP113100.02085|3 = 80|053
RPA2P159270.02213|3 = 60|055
LEF1Q9UJU20.026110|8 = 181|3 = 462
GLUD1P003670.03995|6 = 110|1 = 188
CEP112Q8N8E30.04364|3 = 70|095

LEF1 has the most bait signal and the most control signal. A control-to-bait ratio of 4/18 = 0.22 is eight times the AF4 singleton-class average of 0.027 computed in this beat's denominator control. The thing that made LEF1 the best candidate — abundance — is the same thing that makes its control signal expected, and the two cannot be separated in a dataset with one control column and no spike-in.

This beat also measured that the negative-control column cannot discriminate prey classes at all once its denominator is put on a per-bait footing (ana-q180-the-core-is-cleaner-per-unit-signal-figure-is-a-sevenfold-denominator-artefact), so a reader should not over-weight the 1|3 either. It is a flag, not a verdict; the verdict rests on the lineage confound.

The brief's premise, and how it inverts

The dispatch brief for this beat said LEF1 was the one to spend most effort on, being "a Wnt-pathway transcription factor with a real independent ALL literature". The independent ALL literature is real and it is T-ALL — 20 PubMed records against 2 for B-cell precursor — and the host cell of the experiment is a T-ALL line 2. So the fact offered as external corroboration is, on inspection, the same fact as the confound: LEF1 matters in T-cells, and this was done in T-cells.

That is not a criticism of the brief; it is what checking a premise is for. The corrected statement is: LEF1's independent leukaemia literature corroborates that LEF1 is abundant and functional in the host cell used, which is the leading alternative explanation for the row, not support for it.

What would overturn this decline

In increasing cost:

  1. CTNNB1 recovered under MLL-AF4 below the rank cut in PXD009338. A re-analysis of the deposit, no wet work. Would revive the β-catenin bridge.
  2. LEF1 expressed in infant KMT2A::AFF1 blasts at a level comparable to Jurkat. A lookup in published expression data. If LEF1 is not expressed in the index disease, the row is closed outright and no further work is warranted.
  3. KMT2A::AFF1 occupancy at canonical Wnt-responsive elements in a B-lineage KMT2A-rearranged line, from deposited ChIP data this corpus already uses. Tests the mechanism rather than the co-purification.
  4. A co-IP of endogenous KMT2A::AFF1 and LEF1 in SEM or RS4;11. The decisive experiment, and the only one that would make this an interface rather than a co-residency.

If 2 returns "not expressed", 1, 3 and 4 should not be run.

Facts and figures
partner agene-kmt2a
partner a uniprotQ03164
partner a length AA3969
partner bgene-lef1
partner b uniprotQ9UJU2
partner b length AA399
verdictDECLINED
verdict scopeDeclined as evidence of a physical KMT2A::AFF1-LEF1 contact. NOT a claim that LEF1 is irrelevant to leukaemia, and not a closure of the Wnt question - see open_questions for the four things that would reopen it, in cost order.
adjudicated byQ180-the-af4-specific-interactors, cycle 159, 2026-09-24
source publication pmid29777171
source datasetPXD009338, Skucha 2018 seven-bait MLL-fusion AP-MS
detection methodsingle-step StrepTactin affinity purification of a Strep-HA-tagged bait from nuclear extract, LC-MS/MS, spectral counting with a GLM background model
experimental role partnerprey
experimental role kmt2abait (as the MLL-AF4 fusion, not wild-type KMT2A)
bait exists in datasettrue
n publications1
n methods1
orthogonal method existsfalse
intact curatedfalse
intact noteIntAct has never curated PMID 29777171, so this row is invisible to every IntAct-derived interactome coverage figure this project quotes.
af4 p0.0261
af4 spectral10|8
af4 spectral total18
neg ctrl1|3
neg ctrl total4
rank in af4 by p62
n fusions1
druggabilityNot assessable and not pursued. There is no interface, no residue, no affinity and no second method. A Wnt-pathway intervention would target beta-catenin or the beta-catenin/TCF contact, neither of which is evidenced here.
absence classmeasured-not-searched
absence class noteThe decisive missing datum is whether CTNNB1 was detected under MLL-AF4 below the top-300 rank cut. It is in PXD009338, public since 2018-05-25, and has never been reopened.
shown for fusionMLL-AF4 only, as one prey row in one AP-MS experiment, in a Jurkat T-ALL host with a Strep-HA-tagged single-copy transgene. Nobody has looked in a B-lineage KMT2A-rearranged cell, in a primary sample, or with any second method.
epigenetic readoutnone for this pair. The nearest epigenetic measurement in the literature is H3K4 methylation at Wnt target genes under beta-catenin (PMID 16510874), which concerns beta-catenin and wild-type MLL-family activity, not this fusion and not LEF1.
host cell lines
[
  "Jurkat (human T-ALL), isogenic FRT/Flp single-copy Dox-inducible transgene"
]
measurements
[
  {
    "quantity": "control_to_bait_spectral_ratio",
    "subject": "LEF1|MLL-AF4",
    "value": 0.222,
    "units": null,
    "n": 2,
    "universe": "Skucha 2018 Suppl Table 5, neg_ctrl summed over two replicates divided by MLL-AF4 spectra summed over two replicates",
    "null": 0.027,
    "note": "null column is the same ratio pooled over all 110 MLL-AF4 singletons (10 control counts / 371 AF4 counts); LEF1 is 8.2x that"
  }
]
interface residues a
[]
structures
[]
ligands
[]
How this could bite us back
  • If this row were accepted and a Wnt/beta-catenin inhibitor proposed for KMT2A-rearranged disease on the strength of it, the recurrent LEF1 lesion in the best-studied lineage is LOSS of function (PMID 40857670) - so pathway inhibition could phenocopy an oncogenic event rather than oppose one.
  • LEF1 is required for normal T- and early-B-lymphopoiesis. Any intervention at this node in an infant carries a developmental-toxicity liability independent of the fusion, and the evidence base here does not come close to justifying it.
Still open
  1. Is LEF1 expressed in infant KMT2A::AFF1 B-ALL blasts, and at what level relative to Jurkat? Cheapest of the four reopening tests and the one that can close the question outright in the negative.
  2. Was CTNNB1 detected under MLL-AF4 below the rank cut in PXD009338? Second cheapest; would revive the beta-catenin bridge.
  3. Does KMT2A::AFF1 occupy canonical Wnt-responsive elements in a B-lineage KMT2A-rearranged line, in deposited ChIP data this corpus already uses? Tests the mechanism rather than the co-purification.
  4. Would a co-IP of endogenous KMT2A::AFF1 and LEF1 in SEM or RS4;11 be positive? The decisive test, and the only one that would convert this to an interface.
Every source behind this entry (4)
  1. E1established confidence 5 of 5: Multiple independent lines of evidence; treated as settled in the field.paper

    In Skucha 2018 Supplementary Table 5, LEF1 (Q9UJU2) carries MLL-AF4 p = 0.0261 on spectral counts 10|8, n_fusions = 1, rank 62 of 291 in the MLL-AF4 column, and neg_ctrl 1|3 — the only non-zero negative control among the six AF4-specific preys below p 0.05, the others being 0|0 for RGPD8, ACADM, RPA2 and CEP112 and 0|1 for GLUD1.

    MLL-fusion-driven leukemia requires SETD2 to safeguard genomic integrity. · Skucha A, Ebner J, Schmöllerl J, Roth M, et al. (20 authors; Grebien F senior author) · Nature Communications 2018
    PMID 29777171DOI
    What this evidence cannot show

    All values read from the cached Supplementary Table 5 workbook with the bait column map and the neg_ctrl column asserted against the sheet's own header rows before any value was read (analyses/q180-is-af4-specificity-a-depth-artefact/run.py). None of these numbers appears in the paper's abstract or body. The neg_ctrl column's composition - how many control purifications, of what cells - is stated in neither the paper's Methods nor the PRIDE record, so 1|3 is two numbers of unknown exposure.

    Evidence that argues against this (1)
  2. E2established confidence 5 of 5: Multiple independent lines of evidence; treated as settled in the field.paper

    The AP-MS host is Jurkat, a human T-ALL line, and LEF1 is a T-lineage factor — UniProt's protein name for Q9UJU2 is 'T cell-specific transcription factor 1-alpha' — so LEF1 is an abundant resident nuclear protein of the cell in which the pulldown was performed.

    MLL-fusion-driven leukemia requires SETD2 to safeguard genomic integrity. · Skucha A, Ebner J, Schmöllerl J, Roth M, et al. (20 authors; Grebien F senior author) · Nature Communications 2018
    PMID 29777171DOI
    What this evidence cannot show

    The Jurkat host is read from the paper's Results ('isogenic Jurkat leukemia cell lines') and the PRIDE sample-processing protocol; the protein name is read from UniProtKB Q9UJU2 on 2026-09-24. What is NOT measured here is LEF1's actual abundance in Jurkat relative to other cell lines - no CCLE, DepMap or HPA expression value was retrieved, and the 'abundant resident factor' step is an inference from the lineage and the protein's name, not a number. That is the weakest link in this evidence item and it is cheap to close.

    Evidence that argues against this (1)
    • Being abundant in the host does not make a prey background - MEN1, KMT2A's best-established partner, is also abundant in Jurkat nuclei and is recovered under all seven baits at p 1e-07 to 1e-22. Abundance raises the prior on background; it does not establish it. The reason LEF1 is declined and MEN1 is not is that MEN1 is recovered by every bait at four to twenty orders of magnitude lower p, with a directed structure and a co-crystal behind it, and LEF1 has one row. Nature Communications 2018
  3. E3strong confidence 4 of 5: Replicated in peer-reviewed work by more than one group.database

    No Wnt co-factor was co-recovered with LEF1: CTNNB1, TCF7, TCF7L1, TCF7L2, APC, GSK3B, TLE1 and TLE3 are all absent from the 960-protein seven-bait network, so the β-catenin bridge that would make a KMT2A–LEF1 association mechanistic is unsupported within the dataset.

    What this evidence cannot show

    This is an absence within the top-300-per-bait selection, not within the raw data - the study recovered 4,600 proteins and published 960. The absence is therefore measured-not-searched and is answerable from PXD009338. Symbols were matched exactly against the sheet's gene_name column; a co-factor listed under a synonym or in a multi-gene group string would have been missed, and the sheet does use multi-gene group strings (e.g. 'HIST1H2AG; HIST1H2AI; ...').

    Evidence that argues against this (1)
    • Beta-catenin is a poor AP-MS prey in a nuclear extract: the bulk cellular pool is junctional and never enters the sample, the free nuclear pool is small, and the beta-catenin-TCF contact is salt-sensitive at the 150 mM used here. Its absence is therefore weak evidence against the bridge. Symmetrically, though, that same argument means the LEF1 row cannot be read as evidence FOR the bridge either - if the bridge protein cannot survive the preparation, a bridged LEF1 should not have survived it. ProteomeXchange / PRIDE, EMBL-EBI 2018
  4. E4moderate confidence 3 of 5: One strong study, or several smaller consistent ones.paper

    A published mechanism does exist that would connect a TCF/LEF factor to KMT2A: β-catenin recruits MLL-family H3K4 methyltransferase activity to Wnt target genes, with APC counteracting both β-catenin occupancy and H3K4 methylation there — indirect, in colon-cancer cells, and not about LEF1 itself.

    The APC tumor suppressor counteracts beta-catenin activation and H3K4 methylation at Wnt target genes. · Sierra J, Yoshida T, Joazeiro CA, Jones KA · Genes & Development 2006
    PMID 16510874DOI
    What this evidence cannot show

    Cited from the PubMed record and title; the full text was not read, so no figure, antibody or cell line is claimed here. The system is colon-cancer cells and the recruitment shown is to beta-catenin; LEF1's role is inferred from its being beta-catenin's DNA anchor at those elements, and that inference is this entry's, not the paper's. 'MLL' in a 2006 paper is also ambiguous between KMT2A and its paralogues under rule 66.

    Evidence that argues against this (1)
    • This mechanism predicts that a TCF/LEF-KMT2A association should be a property of WILD-TYPE KMT2A at Wnt target genes, present in any Wnt-active cell, and therefore NOT specific to MLL-AF4 - which is the opposite of what the AP-MS row asserts. If the Sierra mechanism is the explanation for the LEF1 row, the row should have appeared under several or all seven baits, since all seven retain the same KMT2A N-terminus. It appeared under one. The mechanism and the observation are in tension, not in agreement. Nature Communications 2018
Connected to