Blocking where DOT1L binds instead of what it does - the modality that removes the mark only at fusion targets
also called AF9/ENL-DOT1L PPI inhibitors · DOT1L peptide mimetics · DOT1L recruitment inhibition
Instead of switching DOT1L off everywhere, you can block the handshake that brings it to the leukaemia's genes. One such compound left the rest of the cell's H3K79 tagging untouched and removed it only from the genes the fusion controls - which is the selectivity a whole-cell enzyme inhibitor can never have.
The same thing, for a scientist
A third modality sits between catalytic inhibition and degradation: disrupting the **protein-protein interaction** that recruits DOT1L. The decisive published result is a selectivity one. A DOT1L peptide mimetic (compound 12) binds AF9 and ENL in cell-free assays and in leukaemia cells, selectively inhibits MLL-fusion-containing lines with cellular activity comparable to EPZ-5676, and - crucially - **did not affect global H3K79 methylation while effectively suppressing H3K79 methylation at MLL-fusion target genes** and their expression [E1]. A separate small-molecule series against the AF9/ENL AHD pocket (compound-1, IC50 0.9-3.5 microM) reduces both SEC assembly and DOT1L-mediated H3K79 methylation, suppresses the onco-MLL, DOT1L, HOXA9 and MYC signatures, and has antitumour activity in a mouse MLL-rearranged model [E2]. Genetics supports the therapeutic-index argument from the other side: a DOT1L mutant defective in AF9 binding loses recruitment and leukaemic growth, while **PPI-deficient DOT1L still rescues normal haematopoiesis** [E3]. The hard limit is partner specificity - AF9 and ENL have an AHD; **KMT2A::AFF1 does not**, and the endogenous DOT1L complex is directed by MLLT10 through a different interface entirely [E4].
The selectivity result, and why it is the strongest thing in this beat
Every pharmacodynamic argument in the DOT1L field has been about depth - how far H3K79 methylation was reduced. Compound 12 reframes it as location: global H3K79 methylation unchanged, fusion-target H3K79 methylation down, target gene expression down, anticancer activity comparable to EPZ-5676 1.
That is precisely the profile the CRISPR-ChIP partition predicts should be achievable. If H3K79 methylation is two regulatory units - an oncogenic one directed by the fusion and an endogenous one directed by MLLT10 4 - then a recruitment-interface inhibitor should deplete one and leave the other, and a catalytic inhibitor cannot distinguish them at all. Compound 12's readout is the pharmacological confirmation of the genetic partition.
It also explains a puzzle in the clinical data. Pinometostat's pharmacodynamic endpoint was total H3K79 methylation 5. On this model, total H3K79 methylation is the wrong quantity: it is dominated by the endogenous unit, which is the one you do not want to remove, and it is the one whose depletion should cause the cytopenias.
The partner problem that bounds it
The AHD (ANC1 homology domain) pocket that these compounds target belongs to AF9 (MLLT3) and ENL (MLLT1). Roughly a quarter of KMT2A translocations involve AF9 and a tenth ENL 2; AF4/AFF1, at around a third and the dominant fusion in infant ALL, has no AHD. AF10/MLLT10 reaches DOT1L through a coiled-coil/OM-LZ interface that is structurally unrelated.
So a claim of the form "PPI inhibition is the better modality for KMT2A-rearranged leukaemia" is not transferable across partners and this graph should police it. Compound-1 is described as an inhibitor of AF9/ENL interactions with DOT1L and with AF4/AFF4 2, which would give it reach into AFF1-containing complexes by a different route - blocking AF9/ENL-AF4 rather than AF9/ENL-DOT1L - but that is a SEC-disruption mechanism, not a DOT1L-recruitment mechanism, and the two should not be conflated.
A per-partner map of which DOT1L-recruitment interface each fusion uses is a gap in this graph and is written as an open question here rather than asserted.
| mechanism class | target modality / protein-protein interaction inhibition |
|---|---|
| selectivity evidence | compound 12 left global H3K79 methylation unchanged while suppressing it at MLL-fusion target genes (PMID 35612819) |
| potency gap | AHD PPI inhibitors reported at 0.9-3.5 microM IC50 against a catalytic-inhibitor benchmark Ki of 80 pM |
| genetic support | PPI-deficient DOT1L rescues normal haematopoiesis while abolishing MLL-AF9 recruitment and growth (PMID 33562706) |
interfaces targeted
[
{
"interface": "AF9/ENL AHD - DOT1L",
"partners_reached": [
"gene-mllt3",
"gene-mllt1"
],
"compounds": [
"compound 12 (PMID 35612819)",
"compound-1 (PMID 34373735)"
]
},
{
"interface": "MLLT10 OM-LZ - DOT1L coiled coil",
"partners_reached": [
"gene-mllt10"
],
"compounds": []
}
]partners not reached by ahd inhibitors
[ "gene-aff1 (no AHD)" ]
- An interface inhibitor competes against a tethered partner at very high local concentration, so the occupancy needed for a cellular effect is far above what a Kd measured in dilute solution suggests.
- Blocking the AF9/ENL AHD also displaces AF4/AFF4 and therefore the super elongation complex, so the selectivity for the oncogenic H3K79 unit is not as clean as the DOT1L framing implies, and the toxicity profile would be that of a SEC inhibitor.
- In KMT2A::MLLT10, the oncogenic and endogenous DOT1L routes run through the same protein, so the locus-selectivity argument that justifies this modality does not apply to that fusion.
- Which DOT1L-recruitment interface does each KMT2A fusion partner actually use? AF9 and ENL use the AHD, MLLT10 uses the OM-LZ coiled coil, and the route in KMT2A::AFF1 - the commonest infant fusion - is not established in this graph.
- Can the compound 12 selectivity result be reproduced with a locus-resolved, calibrated ChIP readout rather than a global western blot, and does it hold at concentrations that kill?
- Is there any chemical matter against the MLLT10 OM-LZ/DOT1L coiled-coil interface? An inhibitory peptide has been structurally characterised, but nothing drug-like appears in this beat's reading.
- E1moderate confidence 3 of 5: One strong study, or several smaller consistent ones.paper
DOT1L peptide mimetics bind AF9 and ENL in cell-free binding assays and in leukaemia cells and selectively inhibit growth of MLL-fusion-containing leukaemia lines; the most potent, compound 12, had anticancer cellular activity comparable to EPZ-5676 in several MLL-fusion lines, did not affect global H3K79 methylation catalysed by DOT1L, but effectively suppressed H3K79 methylation at MLL-fusion target genes and inhibited their expression.
Design, Synthesis, and Biological Evaluations of DOT1L Peptide Mimetics Targeting the Protein-Protein Interactions between DOT1L and MLL-AF9/MLL-ENL. · Yuan Y, Du L, Tan R, et al. · J Med Chem 2022What this evidence cannot show
Abstract read from the eutils record; full text not read. A peptide mimetic series with no binding affinity, no in vivo data and no pharmacokinetics stated in the abstract; peptide mimetics are a notoriously difficult starting point for a systemic drug. 'Global H3K79 methylation unaffected' is presumably a bulk western blot, which has limited sensitivity to a partial global change, so the selectivity claim is bounded by that assay's dynamic range. MLL-AF9 and MLL-ENL only.
Evidence that argues against this (1)
- Cellular activity comparable to EPZ-5676 is a double-edged comparison. EPZ-5676 is the compound whose clinical activity was 2 complete remissions in 51 adults, so matching it in vitro is matching the agent that failed, and the selectivity advantage would have to translate into a dose escalation that the catalytic inhibitor could not achieve - which is untested, since pinometostat's maximum tolerated dose was never reached. Stein EM et al. 2018
- E2moderate confidence 3 of 5: One strong study, or several smaller consistent ones.paper
The AHD domain of AF9/ENL binds AF4, AFF4 or DOT1L; compound-1, found by screening and medicinal chemistry, inhibits AF9/ENL-DOT1L/AF4/AFF4 interactions with IC50s of 0.9-3.5 micromolar, reduces SEC assembly and DOT1L-mediated H3K79 methylation in leukaemia cells, suppresses the onco-MLL, DOT1L, HOXA9 and MYC gene signatures, selectively inhibits proliferation of onco-MLL- or MYC-driven cancer cells with differentiation and apoptosis, and shows strong antitumour activity in a mouse model of MLL-rearranged leukaemia. MLL is fused with AF4 in about 35% of cases, AF9 in 25% and ENL in 10%.
Small-molecule inhibitor of AF9/ENL-DOT1L/AF4/AFF4 interactions suppresses malignant gene expression and tumor growth. · Wu F, Nie S, Yao Y, et al. · Theranostics 2021What this evidence cannot show
Abstract read from the eutils record; full text not read. Micromolar potency against a protein-protein interaction, which the authors themselves describe as an in vivo probe and a pharmacological lead rather than a candidate. Because compound-1 blocks AF9/ENL binding to DOT1L AND to AF4/AFF4, its cellular effects cannot be attributed to DOT1L displacement specifically - SEC disruption is a sufficient explanation for the HOXA9 and MYC signature loss. The fusion-partner frequencies quoted are the paper's and are not split by age group or lineage.
Evidence that argues against this (1)
- The same pocket is the target of ENL YEATS-domain and AHD programmes aimed at the SEC rather than at DOT1L, and this graph already holds the alternative interpretation that disrupting AF9/ENL primarily disables elongation. Attributing a phenotype from a dual-interface inhibitor to the DOT1L arm is exactly the inference the compound cannot support. Grigsby SM et al. 2021
- E3strong confidence 4 of 5: Replicated in peer-reviewed work by more than one group.paper
In MLL-AF9 cells, a DOT1L mutant with a defective AF9 interaction site shows complete disruption of recruitment to critical target genes and inhibition of leukaemic cell growth, whereas in adult mouse marrow wild-type and PPI-deficient DOT1L mutants - but not an enzymatically inactive mutant - rescue sustained haematopoiesis, showing the AF9-DOT1L interaction is dispensable in non-leukaemic haematopoiesis.
Elucidating the Importance of DOT1L Recruitment in MLL-AF9 Leukemia and Hematopoiesis. · Grigsby SM, Friedman A, Chase J, et al. · Cancers (Basel) 2021What this evidence cannot show
Abstract read from the eutils record; full text not read. The genetic separation is clean but the chemical equivalent is not: a mutant that cannot bind AF9 is a permanent, complete loss of that interface, whereas a competitive small molecule gives partial, reversible occupancy against a high local concentration of a tethered partner. MLL-AF9 only.
Evidence that argues against this (1)
- Targeting a protein-protein interface is harder pharmacologically than targeting an enzyme active site, and the best published AF9/ENL AHD compounds sit at micromolar IC50 against a picomolar-Ki benchmark set by the catalytic inhibitor. A superior theoretical therapeutic index is worth little if the achievable occupancy is three or four orders of magnitude lower. Wu F et al. 2021
- E4strong confidence 4 of 5: Replicated in peer-reviewed work by more than one group.paper
CRISPR-ChIP partitions H3K79 methylation into an oncogenic DOT1L complex directed by the MLL fusion protein in a menin-dependent manner and a separate endogenous DOT1L complex whose catalytic activity is directed by MLLT10.
CRISPR-ChIP reveals selective regulation of H3K79me2 by Menin in MLL leukemia. · Gilan O, Talarmain L, Bell CC, et al. · Nat Struct Mol Biol 2023What this evidence cannot show
Abstract read from the eutils record; full text not read. The pooled screen runs on an integrated reporter containing a cis-regulatory element and a single guide RNA rather than the endogenous HOXA locus, but the partition is not reporter-only — GSE192555 holds genome-wide H3K79me2 ChIP-seq for six arms. **The fusion is now resolved and it is not the index fusion: murine MLL-AF9.** It is not named in the abstract, and the full text is unreachable, but the study's GEO depositions name it per sample. In GSE192555 all eight H3K79me2 ChIP-seq samples are Mus musculus with `cell type: MLLAF9`, including both DOT1L-KO guides, both MLLT10-KO guides and **the menin-inhibitor arm** (GSM5751590). GSE192561 is 24/24 murine. Across all 65 samples in the three series, **none mentions AFF1, AF4 or any AFF-family gene**. The human arm is K562 — a CML line with no KMT2A rearrangement — carrying H3K4me3, Pol II, and five H3K79me2 samples used as a mouse/human library-mix species-spike normalisation under a DOT1L inhibitor, not as a fusion system. So this result reaches KMT2A::AFF1 only by import across fusion, species and lineage at once. Verified by the coordinator directly against the GEO SOFT records on 2026-09-19. Cited here for the existence of two regulatory units, which is what makes locus-selective depletion conceptually possible.
Evidence that argues against this (2)
- If the endogenous unit is MLLT10-directed, then in KMT2A::MLLT10 leukaemia the oncogenic and endogenous routes converge on the same protein, and the selectivity argument for recruitment-interface inhibition collapses for that fusion specifically. Deshpande AJ et al. 2014
- The fusion behind this result is murine MLL-AF9, not KMT2A::AFF1 — established from the study's own GEO depositions, since the full text is unreachable. Every H3K79me2 sample in GSE192555 is Mus musculus, cell type MLLAF9, and no sample in any of the three series mentions an AFF-family gene. The menin-dependence of the oncogenic unit, which is the part this graph leans on hardest, was measured in mouse myeloid MLL-AF9 cells. KMT2A::AFF1 engages DOT1L through a different architecture — AFF1 carries neither the AHD that MLLT3 and MLLT1 use nor the AF10 OM-LZ — so the partition may not transfer, and `hyp-index-fusion-h3k79-is-entirely-endogenous` is the sharper alternative. NCBI Gene Expression Omnibus 2022
- E5established confidence 5 of 5: Multiple independent lines of evidence; treated as settled in the field.paper
The pinometostat adult phase 1 used reduction of H3K79 methylation as its pharmacodynamic endpoint and reported 2 complete remissions in 51 adults.
The DOT1L inhibitor pinometostat reduces H3K79 methylation and has modest clinical activity in adult acute leukemia. · Stein EM, Garcia-Manero G, Rizzieri DA, et al. · Blood 2018What this evidence cannot show
Abstract read from the eutils record; full text not read. The abstract does not state the tissue, assay or methylation state used for the pharmacodynamic measurement, so the argument in this entry that a total-H3K79me endpoint is the wrong quantity is an inference from the two-unit model rather than a criticism of a stated assay.
Evidence that argues against this (1)
- A locus-resolved pharmacodynamic endpoint was not available as a clinical assay in 2018 and arguably still is not, so criticising the trial for using a bulk measure asks it to have used a method that did not exist. The actionable version of this criticism applies to the next trial, not the last one. Yuan Y et al. 2022
blocks recruitment rather than catalysis
AF9 AHD pocket is the drugged interface
ENL AHD pocket is the drugged interface
removes the mark at fusion targets while leaving global H3K79 methylation intact
- related to →Pinometostat (EPZ-5676) — the DOT1L inhibitor, and the most instructive failure in this field
the alternative modality to catalytic inhibition
- ← supportsNormal blood needs DOT1L too - lose it in an adult mouse and the marrow fails within weeks
PPI-deficient DOT1L rescues normal blood; AF9 interface is dispensable there
a third modality: remove an activator of catalysis rather than the enzyme or its recruitment
if DOT1L recruitment is convergent rather than interface-shared, a PPI inhibitor against the OM-LZ contact should fail here while a catalytic inhibitor should not