Partner-protein occupancy is not fusion occupancy: every KMT2A::AFF1 chromatin claim resting on an AFF1, AFF4 or SEC-subunit ChIP is an upper bound that includes free complex
Many statements about where the fusion protein sits are actually statements about where one of its partner proteins sits - and that partner is also present, doing a normal job, on many of the same genes. The two are not the same measurement.
The same thing, for a scientist
**Statement.** In a KMT2A::AFF1 cell, an antibody to AFF1 (or AFF4, ELL, CDK9, ENL) reports the union of fusion-bound complex and free super-elongation complex. Because the fusion's own contribution is to make an existing recruitment constitutive rather than to create it [E3], the free-complex component is not small and is not confined to non-target genes. **Therefore any occupancy claim in this corpus sourced from a partner-protein ChIP is an upper bound on fusion occupancy with an unmeasured free-complex term, and should be read as such.** **Direct measurement supporting it.** In three primary infant KMT2A::AFF1 samples, AFF1's enrichment at N/C-defined fusion promoters over wild-type KMT2A promoters is 1.03-1.45 with one antibody and 0.56-0.76 with another, against ENL's 1.90-2.78 and an H3K4me3 negative control at 0.97-1.06 [E1]. The source paper reaches the same conclusion and states it plainly [E2]. **Scope of the problem in this corpus, which is not yet measured.** This graph contains hundreds of occupancy statements about the index fusion. How many are sourced from a partner-protein readout rather than from an N/C ratio, an epitope tag or a fusion-junction reagent has not been counted. That census is the first analysis this hypothesis needs, and it is cheap. **Prior probability 0.8** - the mechanism is established biochemistry, the direct measurement exists in the right cell type, and the only question is how much of this corpus it touches.
What the hypothesis does NOT say
It does not say partner ChIPs are worthless. ENL - a partner-family protein - is the best fusion-site discriminator in this deposit, at 1.9-2.8x. The claim is specific to AFF1-family proteins in an AFF1-partnered fusion, where the antibody cannot distinguish the fusion's partner moiety from the free wild-type protein, and it generalises to any partner whose free pool reaches the same chromatin.
Why ENL escapes and AFF1 does not, as a testable sub-claim
One explanation is stoichiometric: the fusion binds ENL through the AFF1 moiety's ANC1-homology domain, so a fusion complex concentrates ENL locally above whatever free ENL is present, while AFF1 is itself the moiety and cannot be concentrated relative to itself. If that is right, the discriminating power of a partner antibody should scale with how far downstream of the fusion junction the partner sits in the assembly. CDK9 and ELL would then discriminate better than AFF1 and worse than ENL. All three are measurable in the same assay and none is in this deposit.
| statement | Partner-protein occupancy measured with a total-protein antibody is an upper bound on fusion occupancy that includes the free complex, so corpus claims sourced that way overstate what is known about where KMT2A::AFF1 sits. |
|---|---|
| origin | data-pattern |
| prior probability | 0.8 |
predictions
[ "A fusion-junction-specific or tagged-fusion readout in the same cells will give a strictly smaller target set than a total-AFF1 antibody.", "Partner antibodies will discriminate fusion sites in proportion to their distance from the fusion junction in the assembly: ENL > CDK9/ELL > AFF1.", "A census of this corpus will find a substantial minority of index-fusion occupancy claims sourced from a partner-protein readout." ]
experiments
[ "exp-q142-a-fusion-junction-readout-against-a-total-partner-antibody-in-the-same-nuclei" ]
analyses
[ "ana-q142-aff1-occupancy-does-not-mark-fusion-sites-in-primary-infant-kmt2a-aff1", "ana-q142-the-kmt2a-n-over-c-control-works-and-the-c-terminal-arm-is-the-half-that-carries-it" ]
falsifiers
[ "A total-AFF1 antibody in a KMT2A::AFF1 line or patient sample reaching >= 1.5-fold enrichment at N/C-defined fusion promoters over wild-type KMT2A promoters, with BOTH of two independent reagents, in at least two primary samples. Scoreable against the table in ana-q142-aff1-occupancy-does-not-mark-fusion-sites-in-primary-infant-kmt2a-aff1, where the observed values are 1.03-1.45 and 0.56-0.76.", "A degron or knock-in experiment showing that essentially all AFF1 in a KMT2A::AFF1 cell is in the fusion complex - i.e. AFF1 signal collapsing genome-wide, not just at fusion targets, when the fusion is degraded - which would mean there is no free pool to confound the readout.", "A corpus census finding that fewer than 5% of index-fusion occupancy claims are sourced from a partner-protein readout, which would make the hypothesis true and unimportant." ]
- If this is right, target lists built from partner ChIPs are inflated, and a drug screened against such a list will have been screened against genes the fusion never occupied.
- It also cuts the other way for safety: a partner-directed therapeutic will hit the free complex in normal cells at all the wild-type KMT2A sites this analysis shows the partner occupies.
- How many occupancy claims in this corpus are partner-sourced? Nobody has counted.
- Is the free AFF1 pool in a KMT2A::AFF1 blast large or small? No stoichiometry measurement exists in primary material.
- E1strong confidence 4 of 5: Replicated in peer-reviewed work by more than one group.dataset
AFF1 does not separate fusion-occupied from wild-type KMT2A promoters in any of three primary infant KMT2A::AFF1 states, while ENL does by 1.9-2.8-fold.
GSE252378 - AutoCUT&RUN and AutoCUT&Tag profiles of 36 MLL-rearranged leukemia samples (171 GSMs, hg38) · Janssens DH, Henikoff S (Fred Hutchinson Cancer Center / Van Andel Institute) · NCBI Gene Expression Omnibus 2024What this evidence cannot show
Two patients, promoter windows, within-sample shares, two disagreeing AFF1 antibodies. See ana-q142-aff1-occupancy-does-not-mark-fusion-sites-in-primary-infant-kmt2a-aff1 for the full table and its limits.
Evidence that argues against this (1)
- If the N/C-defined fusion set systematically excludes promoters that wild-type KMT2A also occupies strongly - which it does by construction - then a genuinely fusion-bound AFF1 at those shared promoters is invisible, and the hypothesis would be resting on a site definition that is biased against it. Janssens DH et al. 2024
- E2established confidence 5 of 5: Multiple independent lines of evidence; treated as settled in the field.paper
The source paper concludes that co-occupancy with AF4 cannot validate oncoprotein-target site calls whereas co-occupancy with DOT1L and ENL can.
MLL oncoprotein levels influence leukemia lineage identities · Janssens DH, Duran M, Otto DJ, Wu W, Xu Y, Kirkey D, Mullighan CG, Yi JS, Meshinchi S, Sarthy JF, Ahmad K, Henikoff S · Nature Communications 15:9341 2024What this evidence cannot show
Four samples, the group's own interval definitions, two AF4 antibodies pooled.
Evidence that argues against this (1)
- A single group's reagent panel is the whole evidence base for this; if both their AF4 antibodies underperform, the conclusion is about the reagents and not about AFF1. Janssens DH et al. 2024
- E3established confidence 5 of 5: Multiple independent lines of evidence; treated as settled in the field.paper
AF4-family and ENL-family proteins with P-TEFb form a complex normally recruited to wild-type MLL-target chromatin, which MLL fusions cause to be constitutively present.
A higher-order complex containing AF4 and ENL family proteins with P-TEFb facilitates oncogenic and physiologic MLL-dependent transcription · Yokoyama A, Lin M, Naresh A, Kitabayashi I, Cleary ML · Cancer Cell 17:198-212 2010What this evidence cannot show
Biochemical and mouse transformation work, not primary human infant material.
Evidence that argues against this (1)
- The complex composition at a given locus in a patient's cells has never been measured; the free-versus-fusion partition this hypothesis rests on is inferred from purified biochemistry. Janssens DH et al. 2024
its title claims 'this corpus has been treating it as such'; the corpus's attribution audit explicitly ruled the partner-intersection heuristic invalid and named the same paper q142 cites. The measurement stands, the claim about the corpus does not - and the title also generalises from AFF1 to partner proteins while its own data show ENL discriminating at 1.90-2.78
supplies the missing link between two independent derivations of one verdict: the audit argued in an earlier cycle from PMID 20153263 that partner intersection cannot attribute occupancy, and this hypothesis measured it in primary infant material without citing the audit; no edge connected them before
the direct measurement
free SEC pool confounds the readout
the same confound applies to AFF4 readouts
separates fusion-bound from free AFF1 in the same nuclei
both concern what level of description the target set belongs to