Two surfaces, two jobs: the AHD picks the scaffold and the YEATS pocket reads the histone — so a YEATS ligand cannot ask which of the three SEC-family complexes it is disabling
also called scaffold sets the address, reader sets the mark · YEATS is scaffold-blind
ENL and AF9 each have two separate grips: one holds onto the scaffold that decides which genes the machine visits, the other grips a chemically tagged histone. A drug that plugs the histone grip leaves the scaffold grip untouched — so it disables the reading function in all three versions of the machine at once, including the two that have nothing to do with the leukaemia.
The same thing, for a scientist
ENL/MLLT1 and AF9/MLLT3 carry two structurally and functionally separate interaction surfaces. The **ANC1 homology domain (AHD)**, at the C terminus, is intrinsically disordered and recruits AFF-family scaffolds and other partners by coupled folding and binding; the AF9 AHD–AF4 complex was solved by NMR and its conserved aliphatic residues form the hydrophobic core of the interface [E1]. The **YEATS domain**, at the N terminus, is an acyl-lysine reader that binds acetylated histone H3 and co-localises with H3K27ac and H3K9ac at the promoters of actively transcribed genes; structure-based mutagenesis that disrupts the YEATS–acetylation contact reduces Pol II recruitment to ENL target genes and suppresses the oncogenic programme, without removing the protein [E2]. Because the two surfaces are separate, **which** scaffold a reader is attached to is decided at the AHD, and a small molecule occupying the YEATS pocket does not and cannot discriminate on that basis. ENL and AF9 are subunits of SEC, SEC-L2 and SEC-L3 alike [E3], so a YEATS-pocket ligand is *compositionally* scaffold-blind across a family whose three members control largely different gene sets — 1,539 genes in union, 27.5% of them invisible to AFF4 depletion (`ana-q187-sec-family-target-set-overlap-measured`). **That is a statement about composition, not about effect**, and this entry is careful to stop there: the strongest published counterweight, a therapeutic window for SGC-iMLLT in KMT2A::AFF1 with no detectable HSPC toxicity [E4], is handled in `ana-q187-the-scaffold-blindness-argument-does-not-survive-the-published-therapeutic-window`.
The two surfaces, and why the separation is the whole argument
| surface | location in ENL/AF9 | binds | what it determines |
|---|---|---|---|
| YEATS domain | N-terminal | acylated histone H3 (H3K27ac, H3K9ac and other acylations) | whether the reader is anchored to chromatin at a mark-bearing nucleosome 2 |
| AHD (ANC1 homology domain) | C-terminal, intrinsically disordered | AFF1, AFF4, AFF2, AFF3, DOT1L, BCOR, CBX8 and others, by coupled folding and binding | which complex the reader is a subunit of, and therefore which promoters it is delivered to 1 |
These are different residues, different folds, and different ligand classes. A competitive YEATS-pocket ligand is by construction a chromatin-anchoring inhibitor, not a complex-disassembling agent: it does not release ENL from AFF1, AFF2, AFF3 or AFF4. This corpus already records the AHD side of that separation for the index-adjacent case in mechanism-kmt2a-mllt1-engages-ptefb-through-the-enl-ahd-494-559 and interface-af9-ahd-dot1l-three-sites-2mv7.
The consequence for selectivity is a one-line inference and it is worth writing out so it can be attacked:
Target specificity of a SEC-family particle is set by its scaffold, at the AHD. A YEATS ligand acts at the other surface. Therefore no YEATS ligand can be selective between SEC, SEC-L2 and SEC-L3 on the basis of which scaffold the reader is holding.
Every word of that is about composition and binding surfaces. It says nothing about whether the three particles depend on the reader equally, and that is where it can fail — see the last section.
What the reader-disruption experiment actually showed, which is more interesting than 'the drug works'
Wan and colleagues did the clean version of this in AML 2:
- ENL, but not its paralogue AF9, is required for disease maintenance — CRISPR-Cas9 depletion of ENL gave anti-leukaemic effects, terminal myeloid differentiation and suppression of growth in vitro and in vivo.
- ENL binds acetylated H3 and co-localises with H3K27ac and H3K9ac at promoters of actively transcribed leukaemia-essential genes.
- Disrupting the YEATS–acetylation interaction by structure-based mutagenesis reduced Pol II recruitment to ENL target genes and suppressed the oncogenic programme.
Two things follow that complicate the scaffold-blindness picture in opposite directions.
Against blindness mattering: ENL and AF9 are a paralogous pair sitting in the same three complexes, and one is required while the other is not. So co-membership in a complex clearly does not equalise functional importance, and by the same logic co-membership of SEC-L2 and SEC-L3 does not establish that disabling the reader matters there.
For blindness mattering: the mutagenesis result is the proof that pocket occupancy is sufficient to break the function. The mutant ENL is still in its complex and still holds its scaffold; only the histone contact is gone, and Pol II recruitment falls. A ligand in that pocket is therefore a genuine functional knock-out of the reading step wherever the reading step is load-bearing — which, on composition grounds, is three complexes.
The three ways this mechanism could be wrong, stated before anyone else states them
- The reader may not be load-bearing in SEC-L2 and SEC-L3. Nothing in any paper reports ENL or AF9 occupancy at a single AFF2- or AFF3-dependent gene, or a YEATS mutant in those complexes. Membership is not requirement. This is the largest hole and it is an absence, recorded as
absence-q187-no-yeats-inhibitor-has-been-tested-against-sec-l2-or-sec-l3. - The pool may be lopsided. If 95% of cellular ENL and AF9 is in AFF4/SEC, then pocket occupancy is overwhelmingly an anti-SEC action whatever the composition table says, and the sibling complexes are a rounding error. No experiment in any system partitions the reader pool across SEC, SEC-L2, SEC-L3, DotCom and free protein. The two scaffolds in question are about 10-fold lower expressed than AFF4 in the one cell type where this was measured 3, which argues for lopsidedness.
- Selectivity is not obliged to come from composition. A therapeutic index can arise from differential dependence — how close a cell is to the cliff — with no molecular selectivity at all. Menin inhibitors are the in-house example: menin is expressed everywhere and the window comes from dependence, not from binding selectivity. If that is what is happening with YEATS antagonists, then scaffold-blindness is a true statement with no clinical consequence. The published in vivo window 4 is consistent with exactly that.
| mechanism class | binding-surface separation with a selectivity consequence |
|---|---|
| scaffold blind | true |
| scaffold blind scope | COMPOSITIONAL ONLY. Established: a YEATS-pocket ligand cannot discriminate between SEC, SEC-L2 and SEC-L3 by which scaffold the reader holds. NOT established: that this produces a detectable difference in effect, because the reader-pool partition is unmeasured and reader dependence in SEC-L2 and SEC-L3 is untested. |
| epigenetic readout | the YEATS domain reads acyl-lysine on histone H3, principally H3K27ac and H3K9ac, at promoters of actively transcribed genes; that readout is carried identically into all three AFF-scaffolded complexes |
| shown for fusion | PARTIAL and it matters which part. The YEATS-to-acetylation requirement was shown by mutagenesis in myeloid AML (PMID 28241141), not in KMT2A::AFF1. A YEATS probe has an in vivo efficacy and HSPC-sparing result in KMT2A::AFF1 (PMID 42242459), but KMT2A::AFF1 carries no YEATS domain, so that is wild-type-ENL pharmacology. The scaffold-blindness claim itself has been shown in NO fusion and in no system at all — it is an inference from two composition results. |
two surfaces
{
"YEATS": {
"location": "N-terminal",
"ligand": "acylated histone H3 (H3K27ac, H3K9ac and other acylations)",
"determines": "chromatin anchoring at a mark-bearing nucleosome",
"druggable": true,
"probe": "SGC-iMLLT (drug-sgc-imllt), PDB 6HT1"
},
"AHD": {
"location": "C-terminal, intrinsically disordered",
"ligand": "AFF1, AFF4 (solved with AF4/AFF1); AFF2 and AFF3 assumed from ALF homology, never measured; also DOT1L, BCOR, CBX8",
"determines": "which SEC-family complex the reader belongs to, and therefore which promoters it reaches",
"druggable": "no reported ligand",
"structure": "NMR, AF9 AHD with AF4 (PMID 23260655)"
}
}marks named
[ "H3K27ac", "H3K9ac" ]
measurements
[
{
"quantity": "sec_family_complexes_sharing_the_targeted_surface",
"subject": "SEC, SEC-L2, SEC-L3 | ENL/AF9 YEATS domain membership",
"value": 3,
"units": "complexes",
"n": 3,
"universe": "the three AFF-scaffolded P-TEFb complexes characterised in PMID 22547686 (HEK293 Flp-In-TRex / HEK293T), with SEC composition from PMID 20159561",
"null": 1,
"note": "Null is the fusion-selective ideal: the targeted surface present in one complex only. The comparator, ELL module membership, is 1 of 3 and is the slot nobody is drugging."
}
]- This entry's inference is the kind that is easy to over-quote. 'A YEATS inhibitor is not fusion-selective' would be a stronger claim than the evidence carries: what is established is that it cannot be selective BY SCAFFOLD COMPOSITION. Selectivity by differential dependence is untouched by anything here, and a published in vivo window in the index fusion with no HSPC toxicity already exists.
- If scaffold-blindness does have consequences, the consequences are in tissues nobody is measuring. SEC-L3's documented physiological function is allele-specific regulation at imprinted control regions and the XIST DMR, and SEC-L2's scaffold AFF2 is the FRAXE gene. Haematopoietic safety endpoints — colony assays, transplant arms — cannot see either, so a clean safety package is compatible with the liability being real and invisible.
- The mirror risk is over-caution. Raising an imprinting objection with no measurement behind it can stall a programme that a single ChIP experiment would clear. The honest position is that this is an untested liability with a cheap test, not a known one, and `absence-q187-no-yeats-inhibitor-has-been-tested-against-sec-l2-or-sec-l3` states it in those terms.
- The ELL slot is the family's one compositional asymmetry and it is unexploited — but `protein-ell-eaf-module` records all three ELL/EAF members scoring positively or neutrally on selectivity in KMT2A-rearranged lines. The compositionally selective handle points at proteins the dependency data say are not dependencies, so 'target the asymmetry instead' is not an available recommendation.
- What fraction of cellular ENL and AF9 sits in SEC, in SEC-L2, in SEC-L3, in DotCom and free, in a KMT2A::AFF1 blast? Quantitative immunoprecipitation from one cell lysate answers it, no new reagent is needed, and it converts this entry from an inference into a measurement or kills it.
- Does the ENL or AF9 AHD actually bind AFF2 and AFF3? The interaction is assumed from ALF homology across the family and has never been measured for either paralogue. If the affinity is 100-fold weaker than for AFF1/AFF4, the family sharing is much thinner than the purification implies.
- Is the YEATS contact load-bearing in SEC-L2 or SEC-L3? A YEATS aromatic-cage mutant rescue at an AFF2- or AFF3-dependent gene is the direct test and nobody has run it in any system.
- Is there any ligandable pocket on the AHD side? If the AHD could be drugged, it would be the scaffold-SELECTIVE intervention this analysis says does not currently exist, and the AF9 AHD-AF4 NMR structure is the starting point. The domain is intrinsically disordered, which makes this hard rather than impossible.
- related to →SEC — the AFF1/AFF4-scaffolded super elongation complex, the particle KMT2A::AFF1 is a permanent subunit of
the targeted surface (YEATS) is shared across the family; the scaffold surface (AHD) is what differs
- related to →SEC-L2 — the AFF2-scaffolded super-elongation-like complex: P-TEFb plus ENL and AF9, and NO ELL
SEC-L2 carries the same YEATS readers and cannot be distinguished by a pocket ligand
SEC-L3 carries the same YEATS readers and cannot be distinguished by a pocket ligand
the AHD is the scaffold-binding surface, which is what makes the YEATS pocket scaffold-blind
adds a selectivity axis the dossier does not carry: between complexes containing the same YEATS protein, not between YEATS paralogues
the probe whose pocket occupancy is scaffold-blind by construction
ENL carries both surfaces; only one of them is druggable today
- depends on →MLLT3 / AF9 — a famous fusion partner that is not a dependency, and a hub domain that is
AF9 AHD-AF4 is the solved instance of the scaffold-binding surface
supplies the 27.5% non-AFF4 territory the scaffold-blindness argument quantifies over
upholds the composition clauses and refutes the fusion-selectivity conclusion drawn from them
names the measurement that would confirm or kill the scaffold-blindness mechanism, and the public data to do it with
scaffold-blindness is the mechanism by which a reader-directed drug could reach the imprinting complex
the surface-separation mechanism this hypothesis makes quantitative