Oncofusion

start / graph / risk-the-h3k9me3-exoneration-of-g9a-inhibition-is-epithelial-only

Riskmoderate confidence 3 of 5: One strong study, or several smaller consistent ones.Q108-h3k9me3-at-the-bcr

The H3K9me3 null that appears to exonerate prophylactic G9a inhibition was measured in retinal pigment epithelium, at 1 kb resolution, with n = 2 - and the warning it appears to lift lives in a leukaemic blast

also called do not let an RPE null soften a blast-level warning · the exoneration that does not transfer · what this beat's own result must not be used for

This work found that one silencing tag does not change at the KMT2A gene, which makes a proposed preventive drug look safer. But the measurement was in eye cells, and the safety worry is about blood cancer cells. The reassurance does not reach the place the worry lives.

The same thing, for a scientist

This beat measured H3K9me3 as flat at the *KMT2A* exon-11 anchor across all four knockdown arms while H3K9me1 rose 3.7-3.9x [E1]. The natural inference is that a G9a/GLP inhibitor used to prevent therapy-related *KMT2A* rearrangement would not disturb H3K9me3, which softens `drug-unc0642`'s recorded contraindication and `mark-h3k9me3`'s warning that lowering H3K9 methylation could worsen the fusion programme. **That inference does not follow, and this entry exists to block it.** Four independent reasons, each sufficient on its own. **(1) Wrong cell.** hTERT-RPE is retinal pigment epithelium; `mark-h3k9me3`'s argument concerns SETDB1-deposited promoter H3K9me3 repressing *HOXA9* and *MEIS1* in *KMT2A*-rearranged leukaemia stem cells, which carry **less** H3K9me3 than their normal counterparts [E2]. **(2) Wrong DNA.** The null is at one CTCF anchor inside the *KMT2A* gene body; the warning is at fusion-target promoters elsewhere in the genome. **(3) Wrong perturbation.** siG9A knocks down EHMT2 only; UNC0642 inhibits G9a **and GLP/EHMT1**, and the drug would be given systemically for weeks, not as a 72 h knockdown. **(4) Wrong resolution and power.** 1 kb bins, n = 2, and a control-replicate spread larger than any arm difference - 'no change detectable' is not 'no change'. **The net effect of this beat on the safety question is zero.** It removed one explanation for a conflict; it supplied no evidence about the conflict's dangerous limb.

Go deeper
The specific inference to refuse, and the one that is allowed

Refuse: 'H3K9me3 does not move, so a G9a inhibitor is me3-sparing and mark-h3k9me3's contraindication can be relaxed.'

Allowed: 'At the KMT2A exon-11 anchor in RPE cells, the KDM3B-G9a balance moves H3K9me1 and not H3K9me3, so mark identity is available as an explanation for why this graph's two H3K9 entries point in opposite directions - and that explanation is about the mechanism at this locus, not about the drug's safety at fusion targets.'

The difference between those two sentences is the whole content of this entry. The first is the sentence a reader will write; the second is what the data support.

And note which way the other new findings push. hyp-the-kmt2a-anchor-is-hypersensitive-not-uniquely-targeted proposes that the enzyme axis is global and the locus specificity lives in the element's sensitivity. If that is right, then a systemic G9a inhibitor alters H3K9 methylation across thousands of loci in order to protect one - which makes the off-target question worse, not better. mechanism-kdm3b-loss-shifts-the-kmt2a-anchor-from-h3k9me2-to-h3k9me1 adds that the two methyl states move in opposite directions at this element, so 'lowering H3K9 methylation' is not even a single intervention here. So the honest net movement of this cycle on the therapeutic question is: the mechanistic picture got more specific and the safety picture got slightly worse.

What would actually settle it, and what it would cost

The measurement that would license the reassurance is H3K9me3 ChIP at HOXA9 and MEIS1 promoters, and at the KMT2A anchor, in a KMT2A-rearranged leukaemic blast or leukaemia stem cell, with and without a G9a/GLP inhibitor at the concentration and duration a prophylactic regimen would use. That is a different experiment from the one this beat ran and from exp-h3k9me3-and-ctcf-at-the-kmt2a-anchor-in-a-cd34-progenitor, which addresses the genesis question in a normal progenitor.

Cost: low to medium. KMT2A-rearranged lines are standard (SEM, MV4-11, RS4;11), UNC0642 is a published in vivo chemical probe, and the readout is four ChIP-qPCRs per arm. There is no material or ethical obstacle. It has simply never been done in the form that answers this question, which is why this graph currently holds a contraindication and an apparent exoneration side by side with nothing joining them.

Severity if the reassurance is wrong. The population is children receiving curative anthracycline- and epipodophyllotoxin-containing regimens, and the proposed intervention is a prophylactic epigenetic inhibitor given alongside curative therapy. A drug that reduced therapy-related rearrangement risk while de-repressing HOXA9/MEIS1 in any residual KMT2A-rearranged clone would be trading a low-probability late risk for a higher-probability early one. This is why the entry is graded as a serious risk on a speculative evidence base rather than being left as a caveat inside another entry.

Facts and figures
kindessential-function
severityserious if the reassurance is acted on, because the population is children receiving curative therapy and the proposed use is prophylactic co-administration; the probability that the reassurance is wrong is genuinely unknown
mitigationdo not transfer the RPE H3K9me3 null to any statement about fusion-target promoters, leukaemic cells, or pharmacological G9a/GLP inhibition. The measurement that would license such a statement is H3K9me3 ChIP at HOXA9 and MEIS1 promoters in a KMT2A-rearranged line under UNC0642 at prophylactic-equivalent exposure, and it has never been done in that form.
the inference to refuse'H3K9me3 does not move, so a G9a inhibitor is me3-sparing and mark-h3k9me3's contraindication can be relaxed.'
the inference that is allowed'At the KMT2A exon-11 anchor in RPE cells the KDM3B-G9a balance moves H3K9me1 and not H3K9me3, so mark identity is available as an explanation for why two entries in this graph point opposite ways - an explanation about the mechanism at this locus, not about drug safety at fusion targets.'
net effect of this cycle on the safety questionzero on the reassuring side, slightly negative on the whole: one explanation for a conflict was removed, no evidence about the conflict's dangerous limb was added, and two of this beat's other findings (a possibly global enzyme axis, and me1 and me2 moving in opposite directions at the element) make the off-target picture worse rather than better
localizationthe risk concerns H3K9me3 at fusion-target gene promoters (HOXA9, MEIS1) in the nucleus of a leukaemic cell - a location at which this beat measured nothing
subnuclear localizationnot measured, and the gap is real: SETDB1-dependent H3K9me3 domains have a described nuclear-periphery association that no measurement in this beat or in the founding paper addresses
cell typethe evidence is from hTERT-immortalised human retinal pigment epithelial (RPE) cells; the population at risk is children with KMT2A-rearranged acute leukaemia and children receiving topoisomerase-II-poison-containing curative regimens - a cell type and a clinical population with nothing in common with the evidence
cell statethe measurement is a 72 h siRNA knockdown in proliferating epithelium; the relevant state is a leukaemia stem cell in a child under chemotherapy, possibly as measurable residual disease
epigenetic readoutH3K9me3 at a CTCF anchor (measured, flat) versus H3K9me3 at fusion-target promoters (not measured, and where the warning lives); H3K9me1 and H3K9me2 move in opposite directions at the measured element
shown for fusionThe measurement was NOT shown for any fusion - it is in unrearranged RPE cells. The risk it concerns is specifically about KMT2A::AFF1 and KMT2A::MLLT3-class fusions, where H3K9 methylation at HOXA9 and MEIS1 promoters is argued to restrain the leukaemic programme. That mismatch between the evidence and the fusion is the entire content of this entry.
affects
[
  "drug-unc0642"
]
four independent reasons the transfer fails
[
  "wrong cell - hTERT-RPE is retinal pigment epithelium; the warning concerns KMT2A-rearranged leukaemia stem cells",
  "wrong DNA - the null is at one CTCF anchor inside the KMT2A gene body; the warning is at fusion-target promoters elsewhere in the genome",
  "wrong perturbation - siG9A depletes EHMT2 only for 72 h; UNC0642 inhibits G9a AND GLP/EHMT1 and would be given systemically for weeks",
  "wrong resolution and power - 1 kb bins, n = 2, control-replicate spread larger than any arm difference"
]
How this could bite us back
  • This entry could be used to block a genuinely useful line of work by making any epithelial measurement look worthless. That is not the argument: the RPE measurement is good and it answered the question it was designed for. The argument is narrowly about one transfer step, and the entry should be read as naming that step rather than as a general objection to model systems.
  • There is a real cost to over-caution here. Therapy-related leukaemia after topoisomerase II poisons is a known, quantified harm in children, and a prophylactic intervention that reduced it would be valuable. An entry that makes the safety bar look unreachable could delay the one experiment that would clear it - which is why that experiment is named, costed as low, and put in the open questions rather than left implicit.
Still open
  1. What does a G9a/GLP inhibitor at prophylactic-equivalent exposure do to H3K9me3 at HOXA9 and MEIS1 promoters in a KMT2A-rearranged line? Standard lines, a published probe, four ChIP-qPCRs per arm. Low cost, never done in the form that answers this, and it is the measurement that would license or kill the reassurance.
  2. Is the restraining mark at fusion targets me2 rather than me3? mark-h3k9me3's own open questions say this is unresolved. If it is me2, then this beat's finding that me2 falls at the KMT2A element while me1 rises becomes directly relevant to safety, in the unfavourable direction.
  3. Would a G9a-SELECTIVE inhibitor be safer than UNC0642's dual G9a/GLP activity, and does one exist as an in vivo probe? The founding paper's own genetics point at EHMT2 and not EHMT1 - siG9a prevented the event, siEHMT1 did not - so the GLP half of the dual probe is surplus to the mechanism and is pure added exposure.
Every source behind this entry (2)
  1. E1strong confidence 4 of 5: Replicated in peer-reviewed work by more than one group.dataset

    In GSE210480's RPE cells H3K9me3 at the KMT2A exon-11 CTCF anchor is 1.01x of control on KDM3B depletion, 0.98x on G9a depletion and 1.19x in the double knockdown, with every arm value inside the control replicate range, while H3K9me1 at the same interval is 3.74x - and the histone tracks are 1 kb binned with two replicates per arm and no spike-in.

    GEO GSE210480 H3K9me1/2/3 bigWigs read at interval resolution with pybigtools 0.3.0, 2026-09-22; scripts, pre-registration and full output at analyses/q108-gse210480-h3k9me3-kmt2a-bcr/ · Gray ZH, Chakraborty D, Duttweiler RR, Alekbaeva GD, Murphy SE, Chetal K, Ji F, Ferman BI, Honer MA, Wang Z, Myers C, Sun R, Kaniskan HU, Toma MM, Bondarenko EA, Santoro JN, Miranda C, Dillingham ME, Tang R, Gozani O, Jin J, Skorski T, Duy C, Lee H, Sadreyev RI, Whetstine JR · NCBI GEO (data accompanying Cell 186:4528-4545.e18) 2023
    PMID 37788669DOI
    What this evidence cannot show

    This is the measurement whose over-reading the entry exists to prevent, so its limits are the entry's subject: RPE cells, one anchor inside the KMT2A gene body, 1 kb bins, n = 2 per arm, control-replicate spread at the element larger than any arm difference, no spike-in, and a 72 h siRNA knockdown of EHMT2 alone rather than pharmacological dual G9a/GLP inhibition. A null under these conditions is 'no change detectable here', which is a much weaker statement than 'H3K9me3 is unaffected'.

    Evidence that argues against this (1)
    • The most likely truth is still that H3K9me3 genuinely does not move at this element, and a reader could reasonably say this entry is over-cautious about its own beat's cleanest result. The reply is that the entry does not dispute the null; it disputes the transfer of the null to a different cell, a different locus and a different drug - and the transfer is the step nobody writes down. Iqbal Lab oncofusion knowledge graph 2026
  2. E2strong confidence 4 of 5: Replicated in peer-reviewed work by more than one group.database

    The warning this null appears to lift concerns a different system entirely: SETDB1-deposited promoter H3K9me3 represses Hoxa9 and Meis1, and MLL-rearranged leukaemia stem cells carry less H3K9me3 than their normal counterparts, so H3K9 methylation restrains the fusion programme and a G9a or SETDB1 inhibitor is a candidate for making the leukaemia worse.

    What this evidence cannot show

    This item summarises an existing entry rather than re-reading its five primary sources, which is the correct provenance to state and is the same provenance the absence entry that launched this beat used for the same material. If mark-h3k9me3 has mis-stated any of those papers, this entry inherits the error - but the structural point (the warning concerns fusion-target promoters in leukaemic cells, not a CTCF anchor in epithelium) would survive any plausible correction to the details.

    Evidence that argues against this (1)
    • mark-h3k9me3's own open questions name 'which H3K9 methylation state actually matters at fusion targets - me2 or me3' as unresolved, so the warning it carries is itself not firmly attached to me3. If the restraining mark at fusion targets turns out to be me2 rather than me3, then this beat's finding that me2 FALLS at the KMT2A element while me1 rises becomes directly relevant to the safety question - and in the unfavourable direction. Iqbal Lab oncofusion knowledge graph 2026
Connected to