H3K9ac - the mark KMT2A fusions pile up on their own targets, the one the AF9 reader actually binds, and the one a DOT1L inhibitor strips off first
also called H3K9ac · H3K9Ac · H3K9 acetylation · histone H3 lysine 9 acetylation · acetyl-H3K9 · H3K9acetyl
H3K9ac is an acetyl tag at the start of active genes. In KMT2A-fusion leukaemia it is piled abnormally high on exactly the genes the fusion switches on - HOXA and MEIS1 are the two biggest patches in the genome - and when a DOT1L inhibitor is given, the cell strips that tag off those genes faster than off any others. It is also the tag the fusion partner AF9 physically grips.
The same thing, for a scientist
H3K9ac is the acetyl mark at active transcription start sites, written by the SAGA acetyltransferases KAT2A/GCN5 and KAT2B/PCAF and erased by SIRT1 and class I HDACs. **In KMT2A-rearranged leukaemia it is the best-characterised acetyl mark by some distance, and every part of the writer-reader-eraser cycle has been touched.** MLL-AF9 target genes carry elevated H3K9ac relative to other expressed genes in both mouse and human MLL-rearranged cells; the enlarged 'H3K9ac-high' domains overlap the MLL-AF9-occupied peaks, and the HOXA cluster and MEIS1 are the two most enlarged of them, while normal LSK progenitors show a scattered distribution at the same loci [E1]. DOT1L inhibition recruits a SIRT1-SUV39H1 repressive complex to those TSSs and removes H3K9ac **more strongly at MLL-AF9 targets than at other active genes**, an effect completely blocked by Sirt1 knockdown; a SIRT1 agonist deepens both the H3K9ac loss and the antiproliferative effect [E1]. The **reader is inside the fusion partner**: the AF9 YEATS domain binds H3K9ac with high affinity, colocalises with it genome-wide, and uses it to recruit DOT1L [E2] - and this graph has already established that this domain is discarded in ~92% of KMT2A::MLLT3 patients while its ENL counterpart is retained in 84.1% of KMT2A::MLLT1 patients. **Two 2026 studies make the writer druggable**: a degron-independent CRBN glue degrades KAT2A selectively and ablates H3K9ac with in vivo PDX efficacy [E3], and a KAT2A/B PROTAC removes H3K9ac asymmetrically at MYC, MYB and the HOXA cluster while displacing ENL from chromatin, including in an isogenic KMT2A-rearranged model [E4]. **What is still missing is the index disease**: no H3K9ac measurement exists in KMT2A::AFF1 B-lineage cells, in infant material, or under a menin inhibitor.
The measurement, and whether it is on the fusion's targets or genome-wide
This is the mark where the graph's recurring question has the cleanest published answer, because the study that made it reports both scales in the same figure.
- Genome-wide: H3K9ac ChIP-seq across all genes at TSS +/- 2 kb in mouse MLL-AF9 leukaemia cells, and again in human MOLM-13, NOMO-1 and MonoMac6 1.
- Fusion targets: a defined 129-gene MLL-AF9 target set scored separately against 4,560 active genes, 4,560 silent genes and the whole 18,240-gene background. The target set sits above the active-gene set at baseline, and falls further than it under DOT1L inhibition 1.
- Cell of origin control: in normal Lin-Sca1+cKit+ cells the same gene set shows a scattered rather than elevated H3K9ac distribution 1. That control is what makes the enrichment a property of the leukaemia rather than of the gene set.
- Occupancy overlap: the genomic regions with exaggerated H3K9ac-high status overlap the previously defined MLL-AF9-occupied peaks, and the HOXA cluster and MEIS1 are the two most enlarged H3K9ac-high domains 1.
The 2026 degrader work adds the reverse direction at locus resolution: KAT2A/B degradation produces H3K9ac loss concentrated asymmetrically at core AML oncogene loci - MYC, MYB and the HOXA cluster - rather than uniformly 4.
The reader is the fusion partner, and it is present or absent depending on the translocation
The AF9 YEATS domain binds H3K9ac strongly - more strongly than H3K27ac or H3K18ac - through a serine-lined aromatic sandwiching cage, colocalises with H3K9ac genome-wide by ChIP-seq, and that binding is important for chromatin recruitment of DOT1L 2. The authors of the MLL-AF9 chromatin study explicitly connect the two results, writing that AF9's high-affinity H3K9ac binding provides "a direct link between H3K9ac, H3K79me2, and active gene transcription" 1.
And this graph already knows what happens to that reader in the oncoprotein. subtraction-mllt3-yeats records that approximately 92% of KMT2A::MLLT3 patients have breakpoints at or after AF9 residue 376, excluding the YEATS domain (residues 1-138) while retaining the AHD that docks DOT1L. subtraction-mllt1-yeats-retained records the opposite for ENL: 84.1% of KMT2A::MLLT1 patients retain YEATS, and deleting it impairs chromatin binding and lengthens leukaemia latency.
So the H3K9ac-reading module is usually missing from the MLLT3 fusion and usually present in the MLLT1 fusion, which makes this mark's functional significance partner-dependent in a way no single-fusion study can see. The 2026 KAT2A/B degrader result is consistent with the ENL half of that: removing H3K9ac displaces ENL from target loci 4.
The eraser arm is the only complete drug story on any acetyl mark in this disease
DOT1L inhibition does not simply fail to maintain H3K9ac; it actively recruits the machinery that removes it. SIRT1 occupancy rises at active TSSs after DOT1L inhibitor treatment, H3K9ac falls at those loci, and Sirt1 knockdown completely blocks the loss while barely affecting baseline H3K9ac. Concomitantly H3K9me2 and H3K27me3 rise at MLL-AF9-bound TSSs, and Suv39h1 is the relevant H3K9 methyltransferase 1.
The therapeutic consequence is on record and is the kind this graph is for: a SIRT1 agonist (SRT1720) combined with a DOT1L inhibitor further reduces H3K9ac at the Hoxa7 and Meis1 TSSs and enhances antiproliferative activity 1. That is a combination whose pharmacodynamic readout is this mark, at named loci, and it has never been taken further in the fifteen years since.
The writer arm became druggable in 2026 from two directions - a selective KAT2A degrader built on degron-independent CRBN recruitment 3 and a KAT2A/B PROTAC 4 - so for the first time both ends of the cycle for this mark have chemical matter in hand.
What has never been measured
Nothing on this mark exists in KMT2A::AFF1. Every locus-resolved H3K9ac dataset in this disease is myeloid: murine MLL-AF9, human MOLM-13, NOMO-1, MonoMac6, and the 2026 degrader work's 'isogenic KMT2A-rearranged model' 14. No B-lineage line, no SEM, no RS4;11, no primary infant sample.
No H3K9ac measurement exists under a menin inhibitor. The drug-name-bounded query returns one irrelevant record and three menin-inhibitor chromatin papers read in full text contain the string zero times (see mark-h3k27ac E6 for the counts and the exact name list).
The KAT2A/B condensate claim needs this project's condensate standard applied. 4 reports that removing acetylation "dissolves ENL-anchored transcriptional condensates". This graph's standing rule (localization-condensate-evidence-standards, brief section on condensate claims) is that 1,6-hexanediol sensitivity and classical FRAP cannot distinguish phase separation from binding to spatially clustered sites, and that 1,6-hexanediol independently condenses chromatin in living cells. Displacement of a reader from acetylated chromatin is fully explained by loss of its ligand; no condensate model is required, and the preprint is cited here for the H3K9ac and ENL-displacement observations rather than for the phase-separation interpretation.
| mark | H3K9ac |
|---|---|
| histone | H3 |
| residue | K9, N-terminal tail |
| chemistry | acetylation (+42 Da); mutually exclusive with H3K9me1/2/3 on the same residue, which is why the DOT1L-inhibitor switch at MLL-AF9 targets runs from H3K9ac to H3K9me2 |
| writer complex | SAGA |
| eraser partner | SUV39H1 (no gene entry in this graph as of 2026-09-20) deposits H3K9me2 behind SIRT1 at the same TSSs |
| measured in kmt2a r | true |
| measured in kmt2a aff1 | false |
| measured in b lineage | false |
| measured in infant material | false |
| measurement scope | BOTH, in one figure - genome-wide TSS +/- 2 kb across 18,240 genes AND a 129-gene MLL-AF9 target set scored separately, with a normal-LSK cell-of-origin control (PMID 25822366) |
| dot1l inhibitor effect | loss at active TSSs, stronger at MLL-AF9 targets than at other active genes; SIRT1-dependent and abolished by Sirt1 knockdown (PMID 25822366) |
| menin inhibitor effect | never measured |
| kat2a degrader effect | ablation of H3K9ac; locus-resolved loss concentrated at MYC, MYB and the HOXA cluster (PMIDs 42424456, 42523331) |
| named in entries when written | 15 |
| mentions when written | 55 |
| count method | regex /\bH3K9ac/i over all 1,701 node JSON files, run 2026-09-20 |
| abstract query count 2026 09 20 | 4 |
| combination with a pharmacodynamic readout on this mark | SRT1720 (SIRT1 agonist) + EPZ4777 (DOT1L inhibitor); readout is H3K9ac at Hoxa7 and Meis1 TSSs |
writers
[ "gene-kat2a", "gene-kat2b" ]
erasers
[ "gene-sirt1" ]
readers
[ "gene-mllt3 AF9 YEATS - high affinity, and discarded in ~92% of KMT2A::MLLT3 patients", "gene-mllt1 ENL YEATS - retained in 84.1% of KMT2A::MLLT1 patients" ]
named loci with data
[ "HOXA cluster", "MEIS1", "Hoxa7", "MYC", "MYB" ]
antibodies named in sources
[ "Abcam ab4441 (anti-H3K9ac, PMID 25822366)", "Abcam ab1220 (anti-H3K9me2, same study)" ]
cell systems with data
[ "mouse MLL-AF9 leukaemia (LSK-derived)", "MOLM-13 (human, KMT2A::MLLT3)", "NOMO-1 (human, KMT2A::MLLT3)", "MonoMac6 (human, KMT2A::MLLT3)", "normal mouse LSK (control)", "isogenic KMT2A-rearranged model, partner unspecified (preprint)" ]
- A SIRT1 agonist given to strip H3K9ac from fusion targets activates a deacetylase with hundreds of non-histone substrates including TP53, FOXO and PGC-1alpha; SRT1720's pharmacology has been contested since publication, and a combination built on it inherits that dispute.
- KAT2A/KAT2B degradation removes acetylation from SEC components ENL, AFF1 and AFF3 as well as from histones, so a clinical effect cannot be attributed to the histone mark - and any resistance mechanism arising against it may run through the non-histone arm entirely.
- H3K9ac and H3K9me are mutually exclusive on the same residue, so lowering the acetyl mark at fusion targets hands those loci to SUV39H1 and SETDB1 - the same heterochromatin machinery whose activity at HOXA9 and MEIS1 this graph records as tumour-suppressive. The intended and unintended effects share one residue.
- The DOT1L inhibitor's own action on this mark is SIRT1-dependent, so any co-medication that inhibits SIRT1 - and sirtuin inhibitors are in development - would be expected to blunt a DOT1L inhibitor's chromatin effect while leaving its target engagement intact, producing a drug that hits its target and does nothing.
- Is H3K9ac elevated at KMT2A::AFF1-bound loci in a B-lineage cell? Every dataset in this disease is myeloid and the index fusion has never been assayed, so the graph's central fusion is absent from its best-characterised acetyl mark.
- Does a menin inhibitor move H3K9ac at fusion targets, and in the same direction as a DOT1L inhibitor? The two drugs already diverge on H2AK119ub geometry; H3K9ac would be an independent test of whether the divergence is general.
- Does the SRT1720-plus-DOT1L-inhibitor combination, whose pharmacodynamic readout is this mark at named loci, work in a KMT2A::AFF1 B-lineage model? It has never been tried outside murine MLL-AF9, and the DOT1L-inhibitor clinical failure in that disease makes a sensitising combination worth revisiting.
- Does H3K9ac occupancy at fusion targets differ between t(9;11) patients (YEATS reader discarded) and t(11;19) patients (YEATS reader retained)? If the reader shapes the mark rather than only reading it, the two should differ, and nobody has looked.
- E1strong confidence 4 of 5: Replicated in peer-reviewed work by more than one group.paper
In mouse MLL-AF9 leukaemia cells and in human MLL-rearranged lines (MOLM-13, NOMO-1, MonoMac6), a 129-gene MLL-AF9 target set carries elevated H3K9ac relative to other expressed genes, the enlarged H3K9ac-high domains overlap MLL-AF9-occupied peaks with the HOXA cluster and MEIS1 the two most enlarged, normal LSK progenitors show a scattered rather than elevated distribution at the same loci, and DOT1L inhibition (EPZ4777) recruits SIRT1 to active TSSs and removes H3K9ac more strongly at MLL-AF9 targets than at other active genes - a loss completely blocked by Sirt1 knockdown and deepened, with enhanced antiproliferative activity, by the SIRT1 agonist SRT1720.
DOT1L inhibits SIRT1-mediated epigenetic silencing to maintain leukemic gene expression in MLL-rearranged leukemia. · Chen CW; Koche RP; Sinha AU; Deshpande AJ; Zhu N; Eng R; et al. · Nat Med 2015What this evidence cannot show
Read in full text (PMC4390532). Murine MLL-AF9 myeloid leukaemia for every perturbation; the human data are baseline H3K9ac ChIP-seq in three AML lines with no drug time course, and one of those datasets was sequenced on a Heliscope platform while the mouse work used HiSeq, so cross-platform comparison of signal is not straightforward. The H3K9ac antibody is a polyclonal (Abcam ab4441). No spike-in normalisation. The MLL-AF9 target set is 129 genes defined from a prior occupancy study, so the enrichment claim inherits that definition. **This is myeloid, mouse-first biology and the index disease of this project is infant B-lineage KMT2A::AFF1, in which none of this has been repeated.**
Evidence that argues against this (1)
- MLL-AF9 target genes are among the most highly transcribed loci in the cell, and H3K9ac tracks transcriptional activity, so elevated H3K9ac at those loci may be a consequence of high transcription rather than a fusion-specific chromatin feature. The LSK control weakens but does not remove this, because those genes are not comparably transcribed in LSK cells either - the control changes both the chromatin and the expression at once. Deshpande A; Chiang CY; Perales M; Niranjan N; Sinha N; Finlay D; et al. 2026
- E2established confidence 5 of 5: Multiple independent lines of evidence; treated as settled in the field.paper
The AF9 (MLLT3) YEATS domain binds histone H3K9 acetylation with high affinity through a serine-lined aromatic sandwiching cage, binds H3K27ac and H3K18ac more weakly, colocalises strongly with H3K9 acetylation genome-wide by ChIP-seq, and that binding is important for chromatin recruitment of the H3K79 methyltransferase DOT1L.
AF9 YEATS domain links histone acetylation to DOT1L-mediated H3K79 methylation. · Li Y; Wen H; Xi Y; Tanaka K; Wang H; Peng D; et al. · Cell 2014What this evidence cannot show
Peptide binding and crystallography in vitro; the ChIP-seq colocalisation is in a human cell system and not in a KMT2A-rearranged cell. The finding is about wild-type AF9. In the oncoprotein the YEATS domain is discarded in about 92% of KMT2A::MLLT3 patients, so for most patients with this translocation the H3K9ac reader described here is not part of the fusion at all - a qualifier the paper predates and does not make.
Evidence that argues against this (1)
- If AF9 YEATS reading of H3K9ac were the dominant route to DOT1L recruitment in KMT2A::MLLT3 leukaemia, discarding the domain in ~92% of patients should abolish the H3K79me2 phenotype, and it does not - the AHD-mediated protein-protein recruitment of DOT1L is sufficient. The reader is therefore an input to targeting rather than the mechanism of it. Chen CW; Koche RP; Sinha AU; Deshpande AJ; Zhu N; Eng R; et al. 2015
- E3strong confidence 4 of 5: Replicated in peer-reviewed work by more than one group.paper
A cereblon-based molecular glue that recruits KAT2A independently of a degron - engaging a surface-exposed tyrosine in an antibody-like recognition mode resolved by cryo-EM - produces selective KAT2A degradation, potent ablation of H3K9 acetylation, antiproliferative effects in acute myeloid leukaemia cell lines and in vivo efficacy in a patient-derived xenograft.
Degron-independent recruitment of KAT2A expands the target space of CRBN molecular glues. · Ojeda S; Wang M; Baek K; Bourgeois W; Sommerschield A; Yue H; et al. · Science 2026What this evidence cannot show
Read at abstract level only. The paper's framing is chemical - extending the CRBN-targetable proteome - and the leukaemia data are AML cell lines plus one PDX with no fusion stratification given in the abstract; whether the PDX is KMT2A-rearranged is not stated there. H3K9ac ablation is reported as a pharmacodynamic consequence of degradation, not as a locus-resolved map, so this establishes that the mark is chemically removable and not where its removal matters.
Evidence that argues against this (1)
- KAT2A and KAT2B are paralogous SAGA catalytic subunits and a KAT2A-selective degrader leaves KAT2B intact; the companion work degrading both reports the antileukaemic effect, so selectivity for KAT2A may be a chemical achievement that costs biological potency. Which of the two is the relevant writer at fusion targets is not settled. Deshpande A; Chiang CY; Perales M; Niranjan N; Sinha N; Finlay D; et al. 2026
- E4moderate confidence 3 of 5: One strong study, or several smaller consistent ones.preprint
A cereblon-recruiting PROTAC degrading the SAGA catalytic subunits KAT2A and KAT2B removes acetylation of both H3K9 and of SEC components including ENL, AFF1 and AFF3, displaces the chromatin reader ENL from target loci, and produces H3K9ac loss concentrated asymmetrically at core AML oncogene loci - MYC, MYB and the HOXA cluster - with antileukaemic activity across genetically diverse AML lines, primary patient samples and an isogenic KMT2A-rearranged model.
Disrupting a Convergent Acetylation Circuit Collapses Leukemic Identity Across AML Subtypes. · Deshpande A; Chiang CY; Perales M; Niranjan N; Sinha N; Finlay D; et al. · bioRxiv 2026What this evidence cannot show
**A bioRxiv preprint, not peer reviewed**, read at abstract level only. The DOI prefix 10.1101 is shared between bioRxiv and Cold Spring Harbor's peer-reviewed journals (see data/symbol-collisions.json); this record's DOI is 10.64898/2026.07.13.737885 and its venue is stated as bioRxiv, so no such confusion arises here. The degrader removes histone AND non-histone acetylation simultaneously, so no experiment in the abstract separates the H3K9ac contribution from acetylation of ENL, AFF1 and AFF3 themselves - the entry deliberately does not attribute the ENL displacement to the histone mark. The 'isogenic KMT2A-rearranged model' is not specified by fusion partner or lineage in the abstract.
Evidence that argues against this (1)
- This project's standing condensate rule holds that 1,6-hexanediol sensitivity and classical FRAP cannot distinguish phase separation from binding to spatially clustered sites, and that 1,6-hexanediol itself condenses chromatin in living cells. Displacement of an acetyl-reader from chromatin whose acetylation has been removed is fully explained by loss of ligand, so 'dissolves ENL-anchored condensates' is an interpretation the data in the abstract do not force. Li Y; Wen H; Xi Y; Tanaka K; Wang H; Peng D; et al. 2014
READ BY THE RETAINED ENL YEATS DOMAIN, alongside H3K18ac and H3K27ac.
follows the same elevated pattern at MLL-AF9 target genes (Supplementary Fig. 8)
SAGA catalytic subunit; selective degradation ablates H3K9ac
paralogous SAGA catalytic subunit; KAT2A/B degradation removes H3K9ac at MYC, MYB and HOXA
deacetylates H3K9 at MLL-AF9 target TSSs after DOT1L inhibition; Sirt1 knockdown blocks the loss
AF9 YEATS binds H3K9ac with high affinity and recruits DOT1L - but is discarded in ~92% of KMT2A::MLLT3 patients
H3K9ac reading by AF9 YEATS is important for DOT1L chromatin recruitment
- ← causesPinometostat (EPZ-5676) — the DOT1L inhibitor, and the most instructive failure in this field
DOT1L inhibition removes H3K9ac preferentially at MLL-AF9 targets, via SIRT1
the HOXA cluster is one of the two most enlarged H3K9ac-high domains in MLL-r leukaemia
MEIS1 is the other most enlarged H3K9ac-high domain
same residue, mutually exclusive; the DOT1L-inhibitor switch runs acetyl-to-methyl at fusion targets
ENL YEATS also reads H3K9ac and is displaced from target loci when acetylation is removed
the only mark in this disease with writer, reader and eraser all measured
- part of →Plain-English: chromatin and epigenetics — the sticky notes on the DNA, and which ones matter here
layman explainer for the mark layer
all locus-resolved data are in myeloid KMT2A-r systems
- co-occurs with →H3K79me2 - the di-methyl state the leukaemia actually runs on, and the one menin turns out to govern only half of
H3K79me2-high and H3K9ac-high domains coincide at MLL-AF9 targets
supplies the counted bound; the abstract index was adequate here
the highest-affinity ligand of the YEATS reader that entry dissects
mutually exclusive modifications of the same K9 residue
mutually exclusive modifications of the same K9 residue
H3K9/K14 acetylation peaks at the SATB1-binding site and spreads ~10 kb; in Satb1-null thymocytes the same site carries H3K9 methylation instead