Oncofusion

start / graph / mechanism-q181-aff2-and-aff3-carry-the-same-yeats-readers-as-the-index-fusion-partner

Mechanismstrong confidence 4 of 5: Replicated in peer-reviewed work by more than one group.Q181-epigenetics-dimension

SEC-L2 and SEC-L3 contain ENL and AF9 exactly as SEC does, so AFF2 and AFF3 sit in complexes built around the same acyl-lysine readers as the index fusion's partner - and both entries carry no epigenetics annotation at all

also called SEC-L2 SEC-L3 YEATS · q181 AFF2 AFF3

The protein fused to KMT2A in the commonest infant leukaemia, AFF1, has two little-studied relatives, AFF2 and AFF3. Those relatives build their own elongation machines - and those machines contain the very same two histone-mark-reading proteins, ENL and AF9, that make the fusion dangerous.

The same thing, for a scientist

gene-aff2 and gene-aff3 are two of the eight entries in this beat's sampling-free real-gap floor: no epigenetics field, no mark, no chromatin concept. That is a striking absence, because the one biochemical paper that defines them gives them a direct route into the chromatin vocabulary this corpus reasons about constantly. Luo and colleagues isolated SEC-like 2 and SEC-like 3, containing AFF2 and AFF3 respectively, in association with P-TEFb, ENL/MLLT1 and AF9/MLLT3 - the two YEATS-domain acyl-lysine readers that are themselves recurrent KMT2A fusion partners and whose YEATS-H3K27ac structure this corpus already holds. All SEC family members show high Pol II CTD kinase activity, but only SEC is required for proper HSP70 induction on stress, and SEC-L2 and SEC-L3 control different gene subsets from SEC and from each other. So AFF2 and AFF3 are not redundant copies of AFF1: they are paralogue scaffolds with distinct outputs, assembled around the same reader modules, and this corpus annotates neither on the epigenetics dimension.

Go deeper
What the composition result says

From the biochemical isolation 1:

complexAFF scaffoldalso contains
SECAFF1, AFF4P-TEFb, ENL/MLLT1, AF9/MLLT3
SEC-L2AFF2P-TEFb, ENL/MLLT1, AF9/MLLT3
SEC-L3AFF3P-TEFb, ENL/MLLT1, AF9/MLLT3

ENL and AF9 are in all three. Both are YEATS-domain readers of acylated lysines and both are recurrent KMT2A fusion partners in their own right (fusion-kmt2a-mllt1, fusion-kmt2a-mllt3). The corpus already holds the YEATS-H3K27ac structure 5J9S under gene-mllt1.

So the chromatin-reading module is shared across the whole SEC family, and the scaffold is what differs.

The functional non-redundancy, which is the part that makes this a gap and not a footnote

If SEC-L2 and SEC-L3 were redundant with SEC, a missing annotation would be cheap. They are not 1:

  • All SEC family members show high levels of Pol II CTD kinase activity -- so the shared module works in all three.
  • Only SEC is required for proper HSP70 induction upon stress.
  • Genome-wide mRNA-Seq shows SEC-L2 and SEC-L3 control different subsets of genes, and AFF4/SEC plays the dominant role in rapid transcriptional induction.
  • MYC is a direct AFF4/SEC target, and SEC recruitment to MYC regulates its expression in acute myeloid and lymphoid leukaemia cells.

Different scaffolds, same readers, different target sets. That is a specific, testable statement about which loci each paralogue's reader module is delivered to, and it is exactly the kind of content the epigenetics dimension is supposed to capture.

What is NOT claimed, and the honest weakness
  • No mark is claimed for AFF2 or AFF3 themselves. Neither is a writer, reader or eraser. Their epigenetic readout is complex-level: they scaffold a particle containing two YEATS readers.
  • No KMT2A-fusion experiment. Luo 2012 is HEK293 and Jurkat biochemistry with mRNA-Seq; the leukaemia relevance in that paper runs through AFF4/SEC and MYC, not through AFF2 or AFF3.
  • No locus-level occupancy for the readers within SEC-L2 or SEC-L3. The 'different gene subsets' result is expression, not ChIP. Whether ENL's YEATS domain reads the same acylation when delivered by AFF2 as when delivered by AFF1 is unmeasured, and it is the question this entry most wants answered.
  • This is one paper. The searches behind this entry returned a single record under the natural query, which per Rule 72 is a warning about the name list rather than a statement about the literature.
Facts and figures
epigenetic readoutComplex-level, not intrinsic. AFF2 scaffolds SEC-L2 and AFF3 scaffolds SEC-L3, each containing the YEATS-domain acyl-lysine readers ENL/MLLT1 and AF9/MLLT3 alongside P-TEFb (PMID 22547686). Neither AFF2 nor AFF3 is itself a writer, reader or eraser.
epigenetic bounded absenceNo histone-mark measurement, ChIP or CUT&RUN for AFF2 or AFF3 in any system was found, and nothing in any KMT2A-rearranged cell. Whether the ENL YEATS domain engages the same acylation when delivered by AFF2 or AFF3 as when delivered by AFF1 has never been tested.
epigenetics not applicablefalse
shown for fusionNO for AFF2 and AFF3 themselves - nobody has looked in any KMT2A fusion. The shared reader subunits ENL and AF9 ARE index-adjacent, being recurrent KMT2A fusion partners in their own right, but that is a fact about them and not about the AFF paralogues.
functional non redundancyAll SEC family members carry high Pol II CTD kinase activity, but only SEC is required for HSP70 induction on stress, and SEC-L2 and SEC-L3 control different gene subsets (mRNA-Seq, PMID 22547686).
marks named
[]
complexes
{
  "AFF2": "SEC-L2 (with P-TEFb, ENL/MLLT1, AF9/MLLT3)",
  "AFF3": "SEC-L3 (with P-TEFb, ENL/MLLT1, AF9/MLLT3)"
}
uniprot
{
  "AFF2": "P51816",
  "AFF3": "P51826"
}
measurements
[
  {
    "quantity": "node files naming AFF3 with no entry-level epigenetics record",
    "value": 40,
    "units": "node files",
    "subject": "word-boundary naming count for AFF3 across the corpus against gene-aff3 carrying no epigenetics field",
    "universe": "3,032 node files at commit ac2aeab, via scripts/beat-inventory.py 2026-09-24",
    "null": null,
    "note": "AFF2 is named in 23; both are in the eight-entry real-gap floor despite being paralogues of the index fusion partner"
  },
  {
    "quantity": "YEATS-domain acyl-lysine readers shared between SEC, SEC-L2 and SEC-L3",
    "value": 2,
    "units": "proteins (ENL/MLLT1, AF9/MLLT3)",
    "subject": "the chromatin-reading module common to all three SEC-family complexes",
    "universe": "the three complexes biochemically isolated in PMID 22547686",
    "null": null,
    "note": "both are themselves recurrent KMT2A fusion partners with entries in this corpus"
  }
]
How this could bite us back
  • ENL and AF9 are being pursued as YEATS-domain targets in KMT2A-rearranged leukaemia. If those readers are also stoichiometric members of SEC-L2 and SEC-L3 in normal tissue, a YEATS inhibitor is not a fusion-selective agent - it disables the reader module of three elongation complexes at once, including two whose target gene sets are distinct and largely uncharacterised. The selectivity argument for YEATS inhibition rests on where the readers are, and this composition result says they are in more places than the fusion. CORRECTED 2026-09-24 (Q187): the closing inference is refuted where it has been tested. Rajhansa and colleagues treated KMT2A::AFF1 leukaemia with the MLLT1 YEATS probe SGC-iMLLT and found it strongly inhibited growth in vitro and in vivo while NOT altering the colony-forming potential of human HSPCs or the long-term in vivo function of mouse HSPCs - a measured therapeutic window, in the index fusion, for the exact agent class this bite_back argues cannot be selective (PMID 42242459, doi:10.1016/j.exphem.2026.105458). The composition observation above still stands, and so does the reasoning that shared readers put the target in more places than the fusion; what fails is the step from that to 'therefore no window'. NOTE THE LIMIT OF THE REFUTATION: the window demonstrated is haematopoietic. SEC-L2 and SEC-L3 biology is neural - FRAXE and KINSSHIP are neurodevelopmental syndromes - and an HSPC colony assay cannot see that compartment. bite_back[1] is therefore NOT addressed by this paper and remains live, particularly for infant patients.
  • AFF2 is FMR2 and its loss causes FRAXE intellectual disability; AFF3 mutation causes the KINSSHIP syndrome. Any strategy that degrades or destabilises an AFF-family scaffold as a class, rather than the fusion specifically, has a developmental-toxicity profile that the AFF1-focused literature does not surface.
  • This entry's own inference - complex membership transfers a chromatin readout to the scaffold - if adopted as an annotation standard would credit large parts of the corpus on the epigenetics dimension without any new measurement. That would inflate the very figure this beat is trying to make honest, and the standard should be argued explicitly rather than absorbed.
Still open
  1. Does the ENL YEATS domain engage the same acylated lysines when delivered by AFF2 or AFF3 as when delivered by AFF1? If the scaffold sets the locus and the reader sets the mark, the paralogues are a natural experiment in separating the two, and nobody has run it.
  2. Are AFF2 and AFF3 expressed in KMT2A::AFF1 blasts, and does either scaffold compete with the fusion for ENL and AF9? A competing scaffold that sequesters the readers would be a resistance or modifier node; gene-aff2's own title records it as 'the AFF-family member with no dependency anywhere', which argues against but does not settle it.
  3. Run the bare-operand controls on AFF2 and AFF3 per Rule 72. The query behind this entry named the authors and returned one record; the literature on these two paralogues has not been surveyed.
  4. Is ENL/AF9 membership in SEC-L2 and SEC-L3 stoichiometric or substoichiometric? The composition claim is from a 2012 purification and the answer changes how much of the reader module each paralogue carries.
Every source behind this entry (1)
  1. E1strong confidence 4 of 5: Replicated in peer-reviewed work by more than one group.paper

    Biochemical isolation of SEC-like 2 (SEC-L2) and SEC-like 3 (SEC-L3), containing AFF2 and AFF3 respectively, in association with P-TEFb, ENL/MLLT1 and AF9/MLLT3; all SEC family members demonstrate high levels of RNA polymerase II C-terminal domain kinase activity, but only SEC is required for proper induction of the HSP70 gene upon stress, and genome-wide mRNA-Seq shows SEC-L2 and SEC-L3 control different subsets of genes while AFF4/SEC plays the dominant role in rapid transcriptional induction, with MYC a direct AFF4/SEC target whose expression SEC recruitment regulates in acute myeloid and lymphoid leukaemia cells.

    The super elongation complex family of RNA polymerase II elongation factors: gene target specificity and transcriptional output · Luo Z, Lin C, Guest E, Garrett AS, Mohaghegh N, Swanson S, Marshall S, Florens L, Washburn MP, Shilatifard A · Molecular and Cellular Biology 32(13):2608-17 2012
    PMID 22547686DOI
    What this evidence cannot show

    Abstract-level read; the full text was not retrieved. HEK293 and Jurkat cells. expected_absent_terms names the acylation marks deliberately: this paper establishes that ENL and AF9 are IN these complexes, it does NOT report any histone-mark measurement, so the chromatin-reading inference is mine from the composition and from the separate YEATS literature. The mRNA-Seq 'different gene subsets' result is expression, not occupancy, and does not establish that the complexes are delivered to different loci.

    Evidence that argues against this (2)
    • The inference this entry rests on - AFF2/AFF3 have a chromatin readout because their complexes contain YEATS readers - is precisely the imported-boilerplate failure the parent triage penalised in gene-lmo2, where a sentence about KDM5A was counted as LMO2's chromatin content. A property of ENL is not a property of AFF3. The defence is that complex membership is a stronger relation than co-mention, but it is a defence and not a disproof, and an author applying this standard consistently would credit every subunit of every chromatin complex in the corpus.
    • This is a 2012 paper and no search was run for later work that might have revised SEC-L2/SEC-L3 composition - in particular whether ENL and AF9 are stoichiometric members or substoichiometric co-purifying proteins, which would change how much of the reader module each paralogue actually carries. The PubMed query used returned exactly 1 record because it named the authors; the bare-operand controls on AFF2 and AFF3 were not run, so this literature has not been surveyed.
Connected to