Oncofusion

start / graph / drug-unc0642

Drugmoderate confidence 3 of 5: One strong study, or several smaller consistent ones.Q102-h3k9-rearrangement-brake

UNC0642 - the G9a/GLP chemical probe that abolished doxorubicin-induced Kmt2a copy gain in mice, and the one intervention in this beat that has been given to an animal

also called UNC0642 · UNC-0642 · UNC 0642 · GTPL7017 · EHMTi (as named in PMID 37788669)

UNC0642 is a laboratory compound that switches off the enzyme which adds the silencing tag to the KMT2A gene, and it is good enough to give to a mouse. Mice given it before doxorubicin did not develop the extra copies of Kmt2a that doxorubicin otherwise produces.

The same thing, for a scientist

UNC0642 is the first in vivo chemical probe of the lysine methyltransferases **G9a/EHMT2 and GLP/EHMT1**, developed from the cellular probe UNC0638 to retain potency, low cell toxicity and selectivity while gaining the pharmacokinetics needed for animal work [E1]. ChEMBL CHEMBL2441082, C29H44F2N6O2, MW 546.71 [E2]. In this beat it is 'EHMTi': co-treatment with the KDM3-family inhibitor JDI-12 at 2.5 microM completely rescued *KMT2A* amplification and rearrangement in RPE cells with no growth effect; 1.5 microM prevented the CTCF-depletion-induced and the doxorubicin-induced alterations; and **B6129SF1/J mice given 5 mg/kg i.p. daily for four days, with doxorubicin 1.5 mg/kg i.v. daily for three days from day 2, did not generate *Kmt2a* DNA copy gains in spleen** - the paper's only in vivo intervention result, at n = 3 in that arm, scored double-blinded [E3]. **It is a probe, not a drug**: no clinical trial, no approval, no NCT. And it is the molecule on which two of this graph's arguments collide, because `mark-h3k9me3` records that UNC0642's direct predecessor UNC0638 lowered both H3K9me2 **and** H3K9me3 in mouse lineage-negative bone marrow, and argues on that basis that lowering H3K9 methylation is the wrong direction in KMT2A-rearranged disease [E4].

Go deeper
Why the compound's identity matters here, in three corrections

1. It is UNC0642, not UNC0638. This beat's dispatch brief said 'the widely used tool compound UNC0638 lowers me2 and me3 together, so no experiment using it separates them'. The compound in PMID 37788669 is UNC0642, named in its STAR Methods 3. The two are different strings in PubMed - UNC0642[tiab] returns 58 records and UNC0638[tiab] 57, and the union is smaller than the sum because the medicinal-chemistry papers name both 5. The observation about me3 is real but belongs to UNC0638 in a different experiment; it has never been made for UNC0642.

2. It is a DUAL inhibitor and the genetic result is single. In the same paper, depleting G9a/EHMT2 prevents the KMT2A alterations and depleting EHMT1/GLP does not 3. So the genetic requirement is G9a-specific while the pharmacological rescue is delivered by a compound designed against both. That is not a flaw in the experiment - the rescue is real - but it means the compound is a broader perturbation than the genetics licenses, and UniProt annotates GLP/EHMT1, unlike G9a, as also writing H3K9me3.

3. So the me3 question is not closed by mark identity alone. 'G9a writes me1/me2, not me3' resolves the tension with mark-h3k9me3 at the level of enzymology. It does not resolve it at the level of this molecule, because (a) the molecule also inhibits an annotated me3 writer, (b) me1 is the substrate pool from which me3 is built, and (c) the one published blot of this compound class in a haematopoietic cell shows me3 falling 4. The experiment that would settle it was run and never reported: PMID 37788669 performed H3K9me3 ChIP-seq in control, siKDM3B, siG9A and double-knockdown RPE cells - eight samples, deposited in GSE210480 - and reported no H3K9me3 result anywhere 3.

What was actually given, to what, at what dose
usedosesystemresult
with JDI-122.5 microM, 12 hRPEcomplete rescue of KMT2A amplification and rearrangement; no growth effect
with siCTCF1.5 microM, 48 hRPEprevented CTCF-depletion-induced alterations
with doxorubicin1.5 microM, 72 h total, Dox 1 pg/ul for the final 24 hRPEsignificantly reduced Dox-induced copy gains; prevented rearrangements
in vivo5 mg/kg i.p. daily x4, solubilised in DMSO at 100 mg/ml then diluted in saline; Dox 1.5 mg/kg i.v. daily x3 from day 2B6129SF1/J mice, 4 arms: vehicle n=4, EHMTi n=4, Dox n=4, both n=3no Kmt2a DNA copy gain in spleen, double-blinded FISH scoring

All from the STAR Methods, read in full 3.

The honest weight of the in vivo arm. Fifteen mice across four arms, three in the combination arm, one endpoint (a FISH copy-number score in splenocytes), no leukaemia incidence, no survival, no latency. It is the right first experiment and it is not a preclinical efficacy package.

Facts and figures
targetgene-ehmt2
target noteDUAL inhibitor of G9a/EHMT2 AND GLP/EHMT1. `gene-ehmt1` does not exist in this graph as of 2026-09-22, which is why only one target id is given; the GLP half is not a secondary or off-target activity but half of the designed mechanism.
chembl IDCHEMBL2441082
modalitysmall molecule (quinazoline-series SAM-competitive-site G9a/GLP inhibitor lineage; chemical probe)
molecular formulaC29H44F2N6O2
molecular weight546.71
statusCHEMICAL PROBE - not an investigational drug. No NCT, no IND, no approval, no clinical trial in any indication. The first in vivo G9a/GLP probe by its own discovery paper's title.
predecessorUNC0638 (CHEMBL1231795, C30H47N5O2, MW 509.74) - the cellular probe, unsuitable for animal work on PK grounds. The two names are distinct PubMed operands (58 vs 57 records).
the brief correctionThis beat's dispatch brief named UNC0638 as the compound whose me2/me3 non-selectivity prevents the arms being separated. PMID 37788669 used UNC0642. The me2+me3 observation is real and belongs to UNC0638 in mouse lineage-negative bone marrow (PMID 33054052, recorded in mark-h3k9me3 E2); it has never been made for UNC0642. The correction does not close the question - UNC0642 inhibits GLP/EHMT1, which UniProt annotates as an H3K9me3 writer, and the H3K9me3 arm of the KMT2A-locus experiment was deposited (GSE210480, 8 samples) and never reported.
absence classnever-measured
trials
[]
doses used in this beat
{
  "with_kdm3i_in_cells": "2.5 uM, 12 h, RPE",
  "with_sictcf_in_cells": "1.5 uM, 48 h, RPE",
  "with_dox_in_cells": "1.5 uM, 72 h total; Dox 1 pg/ul for the final 24 h, RPE",
  "in_vivo": "5 mg/kg i.p. daily x4, DMSO stock 100 mg/ml diluted in saline; B6129SF1/J (JAX 101043); Dox 1.5 mg/kg i.v. daily x3 from day 2",
  "in_vivo_n": "4 arms: vehicle 4, EHMTi 4, Dox 4, both 3 - 15 mice total, double-blinded FISH scoring",
  "in_vivo_endpoint": "splenocyte Kmt2a DNA copy gain by FISH. NOT leukaemia incidence, survival or latency."
}
key ae
[
  "none characterised - no clinical exposure. The only tolerability datum in this beat is 'no impact on cell growth' in RPE at 1.5-2.5 uM and no reported toxicity in 4 mice at 5 mg/kg i.p. x4 doses."
]
resistance
[
  "not applicable - no clinical use. Mechanistically, the rescue does not work on alterations that have already become inherited: G9a depletion rescued transient KMT2A gains and did not rescue inherited ones, so the window is pre- or co-exposure only."
]
How this could bite us back
  • This graph contains a published argument that this compound class is pointed the WRONG way in KMT2A-rearranged disease: `mark-h3k9me3` records that H3K9 methylation restrains the fusion's target programme and that UNC0638 lowers both H3K9me2 and H3K9me3 in mouse marrow. A prophylactic G9a inhibitor given to prevent a rearrangement in a patient who already has a KMT2A fusion is acting on both arguments simultaneously, in opposite directions.
  • G9a/EHMT2 is essential in oocyte meiosis for homologous chromosome segregation via Cyclin B3 transcription. A prophylactic G9a inhibitor co-administered with curative chemotherapy in a child or young adult carries a germline liability at the life stage where fertility preservation is already a concern in this population.
  • The protective window closes before the clinically relevant moment. G9a depletion rescued transient KMT2A copy gains and did not rescue gains already inherited, so this is a pre- or co-exposure intervention only, and a patient presenting with therapy-related disease is years past it.
  • Adding any agent to a curative leukaemia backbone risks the backbone. Anthracyclines and epipodophyllotoxins cure children; a co-administered epigenetic inhibitor that blunted their anti-leukaemic effect would trade a small long-term risk for a large immediate one, and nothing in this beat measures the interaction on leukaemia kill at all.
  • The entire in vivo case is 15 mice, 3 in the combination arm, one cytogenetic endpoint, no pharmacodynamic confirmation and no incidence data. It is a first experiment, and quoting it as preclinical proof of a prevention strategy would be the kind of overreach this graph exists to prevent.
Still open
  1. Does UNC0642 lower H3K9me3 in a haematopoietic cell? Its predecessor UNC0638 does, it inhibits GLP which is annotated as an me3 writer, and the answer decides whether this compound is a clean test of the me1/me2 hypothesis or a general H3K9 methylation reducer.
  2. Was H3K9 methylation actually reduced in the treated mice? No pharmacodynamic measurement accompanies the in vivo result, so 'EHMTi prevented Kmt2a copy gain' is currently an intervention with an unverified proximal effect.
  3. Does a post-exposure schedule do anything? Mice were treated before and during doxorubicin. Since G9a depletion does not rescue already-inherited alterations, the therapeutically interesting question - can anything be done after chemotherapy - has not been asked.
  4. Would a GLP-sparing G9a inhibitor rescue? The genetics say G9a and not EHMT1 is required. A selective G9a compound would both test that and remove the GLP-mediated me3 liability, and would be the single most informative chemical experiment in this beat.
Every source behind this entry (6)
  1. E1established confidence 5 of 5: Multiple independent lines of evidence; treated as settled in the field.paper

    UNC0642 is the first in vivo chemical probe of the lysine methyltransferases G9a and GLP - which catalyse mono- and dimethylation of histone H3 lysine 9 and of non-histone proteins - developed because the previously discovered cellular probe UNC0638, while showing excellent separation of functional potency from cell toxicity, was unsuitable for animal studies due to poor pharmacokinetic properties; UNC0642 maintains high in vitro and cellular potency, low cell toxicity and excellent selectivity with improved in vivo PK.

    Discovery of an in vivo chemical probe of the lysine methyltransferases G9a and GLP. · Liu F; Barsyte-Lovejoy D; Li F; Xiong Y; Korboukh V; Huang XP; Allali-Hassani A; Janzen WP; Roth BL; Frye SV; et al. · J Med Chem 2013
    PMID 24102134DOI
    What this evidence cannot show

    **Abstract only; full text not retrieved**, so the selectivity panel behind 'excellent selectivity', the cellular and biochemical IC50 values, and the PK parameters are not characterised here. The abstract describes the target as a G9a/GLP pair throughout and does not claim selectivity between them - this is a dual inhibitor by design. It also does not state anything about H3K9me3, in either direction. No clinical development: this is a chemical-probe paper.

    Evidence that argues against this (1)
    • 'Excellent selectivity' in a probe paper is selectivity against a panel chosen by the authors, and the relevant off-target question here is not another methyltransferase but the mark: the published cellular effect of this compound's direct predecessor in mouse lineage-negative bone marrow was a reduction in both H3K9me2 AND H3K9me3, which no selectivity panel would have predicted or excluded. Ropa J; Saha N; Hu H; Peterson LF; Talpaz M; Muntean AG 2020
  2. E2established confidence 5 of 5: Multiple independent lines of evidence; treated as settled in the field.database

    UNC0642 is ChEMBL CHEMBL2441082 with registered synonyms UNC0642, UNC-0642, UNC 0642 and GTPL7017, molecular formula C29H44F2N6O2 and full molecular weight 546.71; its predecessor UNC0638 is CHEMBL1231795, C30H47N5O2, MW 509.74.

    What this evidence cannot show

    A database identity check, not a pharmacology claim: it establishes that the string UNC0642 resolves to a distinct registered compound with a distinct formula from UNC0638, which is the whole point of recording it. ChEMBL lists no `pref_name` for either compound, so there is no INN and no development name - consistent with both being chemical probes rather than clinical candidates. Note that a free-text ChEMBL search for 'UNC0638' returns two molecules, so the string alone is not unambiguous and the ChEMBL id must be carried.

    Evidence that argues against this (1)
    • A distinct ChEMBL entry does not mean a pharmacologically distinguishable agent: UNC0642 is a close analogue of UNC0638 developed specifically to keep the same activity with better PK, so findings about the mark are more likely to transfer between them than the separate registrations suggest. The separation matters for citation accuracy, not for assuming different biology. Liu F; Barsyte-Lovejoy D; Li F; Xiong Y; Korboukh V; et al. 2013
  3. E3strong confidence 4 of 5: Replicated in peer-reviewed work by more than one group.paper

    In PMID 37788669, UNC0642 ('EHMTi') at 2.5 microM co-treated with KDM3i for 12 h completely rescued KMT2A amplification and rearrangement in RPE cells with no impact on cell growth; at 1.5 microM for 48 h it prevented siCTCF-induced alterations and at 1.5 microM for 72 h with 1 pg/ul doxorubicin for the final 24 h it significantly reduced Dox-induced copy gains and prevented rearrangements; and B6129SF1/J mice (JAX 101043, 7 male and 8 female across four arms - vehicle n=4, EHMTi 5 mg/kg i.p. daily x4 n=4, doxorubicin 1.5 mg/kg i.v. daily x3 n=4, and both n=3) given EHMTi before doxorubicin did not generate Kmt2a DNA copy gains in spleen, scored double-blinded by DNA FISH; in the same paper depletion of G9a/EHMT2 but NOT of EHMT1 prevented the alterations, and H3K9me3 ChIP-seq was performed in all four knockdown arms without any H3K9me3 result being reported.

    Epigenetic balance ensures mechanistic control of MLL amplification and rearrangement. · Gray ZH; Chakraborty D; Duttweiler RR; Alekbaeva GD; Murphy SE; Chetal K; Ji F; Ferman BI; Honer MA; Wang Z; et al. · Cell 2023
    PMID 37788669DOI
    What this evidence cannot show

    **Full text and STAR Methods read 2026-09-22 via efetch db=pmc on PMC10591855 (Europe PMC fullTextXML returned HTTP 500; <body> confirmed, 80,638 stripped characters); the compound identity, doses, vehicle and mouse strain all come from the STAR Methods and are not in the abstract.** The in vivo arm is **n = 3-4 per group with a single cytogenetic endpoint** - splenocyte Kmt2a copy gain by FISH - with no leukaemia incidence, no survival, no latency arm and no pharmacodynamic confirmation that H3K9 methylation actually fell in the treated mice. The cell-based rescues are in RPE, an epithelial line. Dox doses in cells are given as pg/ul, a concentration unit that cannot be compared with a clinical mg/m2 exposure without the conversion the paper does not supply (rule 76). The mice were treated with EHMTi BEFORE and during Dox; a post-hoc schedule was not tested, and the same paper shows G9a depletion does not rescue already-inherited alterations.

    Evidence that argues against this (1)
    • The genetic and pharmacological arms are not matched: G9a depletion rescued and EHMT1 depletion did not, yet the rescuing compound inhibits both G9a and GLP. The pharmacological result is therefore consistent with the genetics but does not confirm it, and a GLP-dependent contribution cannot be excluded from the drug arm. Gray ZH; Chakraborty D; Duttweiler RR; Alekbaeva GD; Murphy SE; et al. 2023
  4. E4strong confidence 4 of 5: Replicated in peer-reviewed work by more than one group.paper

    The G9a/GLP inhibitor UNC0638 reduced both H3K9me2 AND H3K9me3 in mouse lineage-negative bone marrow, and in the same body of work SETDB1-mediated H3K9 methylation suppresses MLL-fusion target expression and leukaemic transformation - so in KMT2A-rearranged disease the published direction is that H3K9 methylation restrains the fusion's programme, and an agent lowering it is pointed the wrong way.

    SETDB1 mediated histone H3 lysine 9 methylation suppresses MLL-fusion target expression and leukemic transformation. · Ropa J; Saha N; Hu H; Peterson LF; Talpaz M; Muntean AG · Haematologica 2020
    PMID 33054052DOI
    What this evidence cannot show

    Imported from `mark-h3k9me3` E2, where it is recorded as read in full text (PMC7556517); it was not re-read for this entry. The compound there is **UNC0638, not UNC0642**, the system is mouse lineage-negative bone marrow rather than a leukaemia, and the me2/me3 measurement is a western blot of bulk histones rather than a locus-resolved map. Critically, the paper uses UNC0638 as a proxy because no selective SETDB1 inhibitor exists, so its pharmacological arm addresses H3K9 methyltransferases as a class.

    Evidence that argues against this (1)
    • G9a inhibition points the SAME way at both disease stages in the other half of the literature, which is unusual and is the cleanest part of the therapeutic picture: G9a is a selective regulator of fast-proliferating myeloid progenitors with no discernible HSC function, its deletion delays MLL-AF9 and Hoxa9/Meis1 AML and raises survival, UNC0638 drove myeloid differentiation, and primary human AML cells were sensitive to G9a inhibition. Lehnertz B; Pabst C; Su L; Miller M; Liu F; Yi L; Zhang R; Krosl J; Yung E; Kirschner J; et al. 2014
  5. E5established confidence 5 of 5: Multiple independent lines of evidence; treated as settled in the field.database

    As of 2026-09-22, UNC0642[tiab] returns 58 PubMed records and UNC0638[tiab] returns 57; both sets contain PMID 24102134, the UNC0642 discovery paper, and PMID 33054052, the SETDB1/H3K9 leukaemia paper appears in the UNC0638 set only - so the two compound names are distinct operands and a query built on one does not retrieve the other.

    What this evidence cannot show

    Titles and abstracts only. Hyphenated and spaced forms (`UNC-0642`, `UNC 0642`) were not run separately; PubMed generally normalises these but rule 72 records that hyphenated and unhyphenated compound codes are different strings and this was not verified for this pair. The counts are recorded to establish that the two names are distinct operands, not as a bibliometric claim about either compound.

    Evidence that argues against this (1)
    • Distinct operands do not make distinct biology, and a beat that corrects a compound name should not imply it has corrected a conclusion. Substituting UNC0642 for UNC0638 in this graph's reasoning changes which paper is being cited and leaves the me3 question exactly as open as it was. Liu F; Barsyte-Lovejoy D; Li F; Xiong Y; Korboukh V; et al. 2013
  6. E6moderate confidence 3 of 5: One strong study, or several smaller consistent ones.paper

    Maternal EHMT2/G9a is essential for homologous chromosome segregation in oocyte meiosis, acting by regulating Cyclin B3 transcription - so the target of this compound has an essential germline function in a tissue no leukaemia protocol intends to treat.

    PMID 35864958DOI
    What this evidence cannot show

    **Title, journal, year, authors and DOI verified by PubMed esummary 2026-09-22; neither the abstract nor the full text was read**, so the species (very likely mouse), the conditional-knockout design and the penetrance are uncharacterised, and confidence is set accordingly. It is recorded only to give the germline bite-back a resolvable source rather than to characterise a toxicity. No study has given a G9a inhibitor to a juvenile animal and assessed gonadal endpoints, so this is a mechanistic liability inferred from a genetic requirement, not an observed drug toxicity.

    Evidence that argues against this (1)
    • A germline genetic requirement in the oocyte does not predict a drug toxicity in a treated child: G9a-null mice are born at normal frequency with no overt haematological abnormality, and G9a is reported to have no discernible function in haematopoietic stem cells, so the tissue selectivity of the enzyme's requirement is real and the gonadal risk is a hypothesis to be tested rather than an expected effect. Lehnertz B; Pabst C; Su L; Miller M; Liu F; Yi L; Zhang R; Krosl J; Yung E; Kirschner J; et al. 2014
Connected to