Oncofusion

start / graph / drug-jdi-12

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

JDI-12 - published as a JMJD1C inhibitor that preferentially kills KMT2A-rearranged leukaemia and spares cord blood, then used as 'KDM3i' to CREATE KMT2A rearrangements in cord-blood progenitors, by two papers that do not cite each other

also called JDI-12 · JMJD1C Jumonji domain inhibitor 12 · KDM3i (as named in PMID 37788669)

JDI-12 is a laboratory compound found by computer screening against a DNA-tag-removing enzyme. One paper says it kills KMT2A-fusion leukaemia cells while leaving healthy cord blood stem cells alone; another paper uses the same compound to make those same cord blood stem cells start copying and breaking the KMT2A gene.

The same thing, for a scientist

JDI-12 is one of two structurally related hits (with JDI-4) from a virtual screen of 149,519 natural products and 33,765 Chinese-medicine components against the catalytic Jumonji domain of **JMJD1C/KDM3C**. Surface plasmon resonance showed both bind JMJD1C **and its family homolog KDM3B** with modest affinity; JDI-4 reversed KDM3B-conferred H3K9 demethylation in vitro; both induced a global increase of H3K9 methylation in cells; and both killed MLL-rearranged acute leukaemia and other malignant haematopoietic cells but **not** leukaemia cells resistant to JMJD1C depletion and **not cord blood cells** [E1]. PMID 37788669 then used it as 'KDM3i' - 25 nM to 1 microM - to produce site-specific *KMT2A* copy gains and break-apart events in RPE cells, KG1a, a primary AML line, a primary AML organoid and **primary cord-blood CD34+ HSPCs** [E2]. **The two readings of the same molecule are the core problem this entry exists to record**, and it is not resolvable from the published record, because the compound binds two paralogues whose KMT2A-selectivity profiles are opposite in this project's own DepMap scan - JMJD1C rank 23 of 17,728, KDM3B rank 9,382 [E3]. It is a virtual-screening hit with no selectivity panel, no PK, no animal data and no clinical development.

Go deeper
The same compound, two papers, opposite therapeutic readings
PMID 31271662 (Int J Cancer 2020)PMID 37788669 (Cell 2023)
what it calls the compoundJDI-12, a JMJD1C Jumonji domain inhibitorKDM3i, 'a KDM3 family inhibitor'
target assignedJMJD1C (with KDM3B binding noted)KDM3B throughout
effect on H3K9 methylationglobal increase in cellsconsistent - the mechanism requires H3K9me1/2 to rise
effect on MLL-rearranged leukaemia cellskills themnot tested
effect on cord blood cellsdoes not kill them - offered as the therapeutic windowproduces KMT2A copy gains and break-apart in cord-blood CD34+ HSPCs
proposed usetreatment of MLLr acute leukaemiatool to phenocopy KDM3B loss; the paper proposes a KDM3B agonist as the therapy

These are not contradictory observations. Viability and cytogenetic integrity are different endpoints, and a compound can spare a cell's life while rearranging its genome - that is the definition of a genotoxin. They are contradictory conclusions, because the first paper's safety argument rests on the endpoint that cannot see the second paper's hazard, and no study has measured both in the same cells. Neither paper cites the other for this; PMID 37788669 cites PMID 31271662 twice - once as the source of the compound, and once in its discussion for the proposition that 'KMT2A-rearranged cells have increased susceptibility to KDM3B inhibition', which is that paper's MLLr killing result reattributed from JMJD1C to KDM3B 12.

Why the target cannot be assigned, in four independent ways
  1. SPR shows binding to both. The discovery paper measured JDI-4 and JDI-12 binding to JMJD1C and KDM3B, both with 'modest affinity' - the authors' own word, with no Kd in the abstract 1.
  2. The dependency belongs to the paralogue. In this project's prior-free genome-wide DepMap scan, JMJD1C is the 23rd most KMT2A-selective dependency of 17,728 genes (-0.4697 in 11 KMT2A-rearranged leukaemia lines versus -0.0162 in 43 other leukaemia lines, pan-essentiality -0.0121), while KDM3B ranks 9,382nd with a difference of +0.0018 and a pan-essentiality of +0.0226, i.e. mildly favourable to lose 3. If the compound kills MLLr cells, the dependency data say it is doing so through JMJD1C.
  3. The paralogue's dependency is independently replicated in the index fusion. An shRNA screen in murine MLL-AF9 found Jmjd1c among the best candidates and reported its strongest human effect in SEM - a KMT2A::AFF1 B-lineage line - through apoptosis; a conditional mouse showed LSC-selective requirement; and CRISPR domain mapping localised it to the JmjC domain and zinc finger 4.
  4. The paralogue may not even be a histone demethylase. UniProt states outright that it is uncertain whether JMJD1C removes methyl groups from histone proteins as the other KDM3 proteins do, annotating instead non-histone demethylation of MDC1 for double-strand-break repair and of STAT3 5. So the compound's cellular H3K9 methylation increase is most simply explained by KDM3A/KDM3B inhibition while its anti-leukaemic effect is most simply explained by JMJD1C - the two halves may not share a target at all.

The consequence for this graph, stated as a rule. A compound whose name carries a gene family's name is not a target assignment. 'KDM3i' reads as 'the KDM3B tool' and the molecule is a JMJD1C hit that binds KDM3B. Every phenotype in PMID 37788669 that comes from the compound rather than from siRNA is currently unassigned between two paralogues, and the siRNA arms - which are KDM3B-specific and use two independent sequences - are the parts of that paper that survive this objection intact.

Facts and figures
targetgene-kdm3c
target notePUBLISHED TARGET IS JMJD1C/KDM3C - the compound was found in a virtual screen against JMJD1C's Jumonji domain. It also binds KDM3B (gene-kdm3b) by SPR with modest affinity, and PMID 37788669 uses it as a KDM3B tool. THE TARGET OF ANY GIVEN PHENOTYPE IS UNASSIGNED. Both ids are the honest answer and the schema takes one.
second targetgene-kdm3b
modalitysmall molecule (virtual-screening hit from natural-product and traditional-medicine libraries; NOT a characterised chemical probe)
statusRESEARCH COMPOUND - no chemical-probe designation, no selectivity panel, no PK, no animal data, no IND, no approval, no clinical trial. Sibling compounds from the same series: JDI-4 (co-discovered) and JDI-16 (reported as more potent, inducing apoptosis and differentiation and synergising with ATRA).
chembl noteNo ChEMBL identifier was located for JDI-12 by this beat; the compound is named only by its series code in the literature and no structure or SMILES appears in the abstracts read. A ChEMBL id is therefore recorded as absent rather than guessed.
the identity trapA COMPOUND NAMED FOR A GENE FAMILY IS NOT A TARGET ASSIGNMENT. 'KDM3i' reads as the KDM3B tool; JDI-12 is a JMJD1C hit that also binds KDM3B. The KMT2A-selective dependency belongs to JMJD1C (DepMap rank 23 of 17,728) and not to KDM3B (rank 9,382), and UniProt doubts JMJD1C is a histone demethylase at all. The siRNA arms of PMID 37788669 - two independent KDM3B sequences - are the parts of that paper that survive this objection intact; the compound arms do not.
absence classnever-measured
trials
[]
doses used
{
  "in_vitro_enzyme": "1 uM against recombinant human KDM3B/JMJD1B (abcam ab271569) with calf-thymus histones or a 3.3 uM H3K9me2 peptide; readout was western blot for H3K9me1 and H3K9me3",
  "rpe_cells": "25 nM (no growth effect) and 1 uM (modest growth suppression); 12 h for the transient phenotype, 72 h for the inherited one",
  "leukaemic_and_primary_cells": "1 uM for 72 h in KG1a, a primary AML line, an AML organoid and cord-blood CD34+ HSPCs",
  "with_ehmti": "25 nM JDI-12 with 2.5 uM UNC0642, 12 h - complete rescue"
}
pubmed position
{
  "\"JDI-12\"[tiab]": 3,
  "note": "three records total, run 2026-09-22 with querytranslation read back. A compound with three papers is a research reagent, and this graph should never write about it as though it were an agent."
}
key ae
[
  "GENOTOXICITY IS THE HEADLINE LIABILITY, and it is not listed as an adverse effect anywhere in its own literature: at 1 uM for 72 h it produced KMT2A copy gains and break-apart events in primary human cord-blood CD34+ HSPCs, and also in TCF3/E2A and AFF3/LAF4 - all three lymphoid-rearrangement loci.",
  "no other toxicology exists; there has been no animal exposure."
]
resistance
[
  "leukaemia cells resistant to JMJD1C depletion are reported not to be killed by it, which is the discovery paper's own target-validation control and simultaneously a named resistance mechanism",
  "MLL-rearranged AML bearing activating RAS mutations is published as resistant to JMJD1C loss, via IL-3 receptor upregulation with RAS/MAPK and JAK-STAT activation - a pre-existing genotype, common in KMT2A-rearranged disease",
  "redundancy among JmjC-domain family members is the explanation given for Jmjd1c dispensability in the JAK2V617F MPN model, and applies to any single-paralogue agent in a three-paralogue family"
]
How this could bite us back
  • This compound is simultaneously proposed as a treatment for KMT2A-rearranged leukaemia and demonstrated to induce KMT2A rearrangement in normal cord-blood progenitors, by two papers that do not cite each other on the point. A therapy that can cause the disease it treats needs its therapeutic index expressed in cytogenetic terms, and none exists.
  • The safety claim in its own discovery paper - 'does not kill cord blood cells' - is measured with an instrument blind to the hazard. Viability assays cannot see a copy gain or a translocation. Any future development of this series needs karyotype or FISH as a safety endpoint from the first experiment.
  • The target is unassigned between two paralogues with opposite KMT2A-dependency profiles, so both the efficacy and the toxicity could belong to different proteins. That is the worst possible configuration for a development programme: you cannot engineer out the liability without knowing which protein carries it.
  • Resistance is pre-specified: cells resistant to JMJD1C depletion are not killed, and RAS-mutant MLL-rearranged AML is published as resistant to JMJD1C loss. RAS co-mutation is common in KMT2A-rearranged disease, so a predictable fraction of patients is outside reach before a molecule exists.
  • If JMJD1C's annotated function - demethylating MDC1 to enable RNF8-dependent ubiquitination for double-strand-break repair - is engaged by this compound, then it impairs break repair AND raises H3K9me1/2, pushing both halves of the rearrangement mechanism in the same direction. That is a plausible amplifying interaction and nobody has tested it.
Still open
  1. Does JDI-12 still produce KMT2A copy gains in a KDM3B-null cell? If it does, the cytogenetic phenotype belongs to JMJD1C or to something else, and the single most-cited claim about this compound in the KMT2A field is misattributed. This is one isogenic pair and a FISH assay.
  2. Does JDI-12 rearrange KMT2A in the very leukaemia cells it is proposed to treat? The discovery paper measured killing in MLLr lines and the mechanism paper measured rearrangement in normal cells. Nobody has measured rearrangement in an MLLr line under the compound, which is the experiment that decides whether the therapeutic proposal is safe on its own terms.
  3. What is the compound's structure and selectivity? No structure, SMILES, Kd, IC50 or selectivity panel was located in the abstracts available to this beat, and no ChEMBL entry was found. Without these it cannot be compared to anything or reproduced outside the originating laboratory.
  4. Do JDI-4 and JDI-16 behave the same way on the cytogenetic endpoint? JDI-16 is reported as more potent at killing and was not used in the mechanism paper; if potency at killing tracks potency at rearranging, the series has no safe member.
Every source behind this entry (5)
  1. E1strong confidence 4 of 5: Replicated in peer-reviewed work by more than one group.paper

    JDI-4 and JDI-12 were identified among the top 15 compounds from a virtual screen of 149,519 natural products and 33,765 Chinese-medicine components against the catalytic Jumonji domain of the H3K9 demethylase JMJD1C, and share a common structural backbone; surface plasmon resonance showed both bind JMJD1C and its family homolog KDM3B with modest affinity; in vitro demethylation assays showed JDI-4 can reverse the H3K9 demethylation conferred by KDM3B; in vivo demethylation assays indicated both induce a global increase of H3K9 methylation; and cell proliferation and colony formation assays showed both kill MLL-rearranged acute leukaemia and other malignant haematopoietic cells but NOT leukaemia cells resistant to JMJD1C depletion and NOT cord blood cells - with a related compound JDI-16 showing superior killing, inducing apoptosis and differentiation of MLLr cells, downregulating lactic acid and NADP+, and collaborating with all-trans retinoic acid to repress MLLr AML cells.

    Small molecular modulators of JMJD1C preferentially inhibit growth of leukemia cells. · Xu X; Wang L; Hu L; Dirks WG; Zhao Y; Wei Z; Chen D; Li Z; Wang Z; Han Y; et al. · Int J Cancer 2020
    PMID 31271662DOI
    What this evidence cannot show

    **Abstract only; full text not retrieved.** No Kd, no IC50, no selectivity panel against KDM3A, the KDM4 family or other 2-oxoglutarate oxygenases, and no statement of which H3K9 methylation state the in vitro assay read out - so 'KDM3 family inhibitor' describes two measured family members, not a selectivity profile. These are virtual-screening hits from natural-product libraries, not optimised chemical probes; there is no PK, no animal work and no clinical development in the abstract. The 'resistant to JMJD1C depletion' comparator is an internal control whose identity is not given. 'Cord blood cells' is not further specified in the abstract - whether CD34-enriched, whether colony-forming assays were used, and at what concentration and duration are all unknown, which matters because the contradicting result (E2) used 1 microM for 72 h on immunomagnetically enriched CD34+ cells.

    Evidence that argues against this (1)
    • The safety claim rests on an endpoint that cannot see the hazard. The same compound at 1 microM for 72 h produced KMT2A copy gains and break-apart events in primary CD34+ HSPCs enriched from human umbilical cord blood - a cytogenetic readout, in the cell type this paper reports as spared. Sparing viability and inducing rearrangement are compatible observations and incompatible conclusions. Gray ZH; Chakraborty D; Duttweiler RR; Alekbaeva GD; Murphy SE; et al. 2023
  2. E2strong confidence 4 of 5: Replicated in peer-reviewed work by more than one group.paper

    PMID 37788669 uses JDI-12 under the name 'KDM3i': at 1 microM it suppressed KDM3B enzymatic activity in vitro and modestly slowed RPE growth while 25 nM did not; both doses produced significant KMT2A copy gains and rearrangements in RPE without altering KDM3B protein levels; 1 microM for 72 h significantly increased KMT2A copy gains and rearrangements in KG1a, a primary patient-derived AML line, a primary AML organoid on irradiated OP9 stroma, and primary CD34+ HSPCs immunomagnetically enriched from fresh human umbilical cord blood; 12 h of exposure gave alterations that vanished on washout while 72 h gave alterations inherited into later passages; and in RPE lines already carrying inherited KMT2A gains, 3-6 h re-exposure raised copy gains further while 12 h returned them to baseline with increased Annexin V, which the authors read as negative selection of aberrant cells alongside emergence of a new aberrant population.

    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 HTTP 500; <body> confirmed, 80,638 stripped characters). The compound identity, doses and cell sources are from the STAR Methods and are not in the abstract.** The readout in every arm is two-colour DNA FISH copy gain and probe break-apart - a cytogenetic proxy with no sequenced junction anywhere in the paper. The paper attributes every compound effect to KDM3B and does not report which paralogue is engaged in any cell. KG1a and HL60 carry KDM3B loss of heterozygosity. The in vitro enzyme validation used recombinant human KDM3B/JMJD1B (abcam ab271569) with calf-thymus histones or an H3K9me2 peptide and a western readout for H3K9me1 and H3K9me3 - so the enzymatic validation is against KDM3B only and JMJD1C was not assayed. No dose-response, no washout kinetics in the primary cells, and the primary-cell arms are single time points.

    Evidence that argues against this (1)
    • The compound's own discovery paper reports that it does NOT kill cord blood cells and proposes it as a treatment for MLL-rearranged acute leukaemia, and PMID 37788669's discussion cites that paper approvingly for the proposition that KMT2A-rearranged cells are more susceptible to KDM3B inhibition - so the paper using it as a genotoxin also uses it as evidence of a therapeutic opportunity, in the same discussion section. Xu X; Wang L; Hu L; Dirks WG; Zhao Y; Wei Z; Chen D; Li Z; Wang Z; Han Y; et al. 2020
  3. E3strong confidence 4 of 5: Replicated in peer-reviewed work by more than one group.dataset

    In this project's prior-free genome-wide DepMap scan over 17,728 protein-coding genes, 11 Cellosaurus-annotated KMT2A-rearranged leukaemia lines and 43 other leukaemia lines, the two paralogues this compound binds have opposite profiles: JMJD1C ranks 23rd (percentile 0.13, difference -0.4535, KMT2A-rearranged mean -0.4697, other leukaemia -0.0162, pan-essentiality -0.0121, empirical p versus scan 0.0016) while KDM3B ranks 9,382nd (percentile 52.9, difference +0.0018, pan-essentiality +0.0226).

    What this evidence cannot show

    Read directly from the stored results file 2026-09-22. The scan's caveats apply: 11 KMT2A-rearranged lines, none infant-derived; Cellosaurus fusion annotation may be incomplete, biasing toward the null; the empirical p is the genome-wide background fraction, not a separate random draw; the DepMap release is not stated by the API (rule 83). Rule 84 applies to the KDM3B number - a rank at the median is what half of all genes have and carries no information on its own; the informative quantity is the contrast. **And a genetic dependency does not establish which protein a promiscuous compound acts through** - it establishes which protein the cells need, which is a different question and is offered here as the strongest available prior, not as a target-engagement measurement.

    Evidence that argues against this (1)
    • A pooled viability screen cannot see the phenotype this compound is used to produce. DepMap scores proliferation; the effect in question is a cytogenetic alteration in a surviving cell. So the dependency contrast tells you which paralogue explains the KILLING and says nothing about which one explains the REARRANGEMENT - and those could be different paralogues in the same experiment. 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

    An epigenetic shRNA library screen in murine MLL-AF9 AML, counter-screened against normal mouse bone marrow, identified Jmjd1c as one of the best candidate drug targets; depletion impaired growth, colony formation and leukaemia establishment after transplantation, and in human leukaemic cell lines the STRONGEST effect was observed in the MLL-rearranged ALL cell line SEM, primarily through increased apoptosis.

    shRNA screening identifies JMJD1C as being required for leukemia maintenance. · Sroczynska P; Cruickshank VA; Bukowski JP; Miyagi S; Bagger FO; Walfridsson J; et al. · Blood 2014
    PMID 24501218DOI
    What this evidence cannot show

    **Abstract only; full text not retrieved** - see `gene-kdm3c` E3, where the same limitations are recorded. The SEM result is the datum that matters to this graph (SEM is KMT2A::AFF1, B-lineage) and its effect size, hairpin number and controls are uncharacterised. 2014 shRNA, so seed-based off-target effects are a live concern. 'Strongest effect' is a comparison within the panel that happened to be tested.

    Evidence that argues against this (1)
    • Jmjd1c knockout mice show no significant change in peripheral blood composition and initiate JAK2V617F-driven myeloproliferative disease normally, which the authors attribute to redundancy among JmjC-domain family members - so the same gene is essential in one haematopoietic context and redundant in another, and a compound hitting it would inherit that context dependence. Staehle HF; Heinemann J; Gruender A; Omlor AM; Pahl HL; Jutzi JS 2020
  5. E5established confidence 5 of 5: Multiple independent lines of evidence; treated as settled in the field.database

    UniProt's canonical FUNCTION comment for JMJD1C (Q15652) states that 'It is uncertain whether JMJD1C removes methyl groups from histone proteins as the other JMJD1/KDM3 proteins do', citing three papers, and annotates instead demethylation of the non-histone substrates MDC1 - promoting MDC1-RNF8 interaction and the RNF8-dependent MDC1 ubiquitination essential for double-strand break repair - and STAT3 at Lys-140, offering context-specific histone demethylation or a scaffold role recruiting other demethylase complexes as alternatives.

    What this evidence cannot show

    A database record, and the three papers behind the uncertainty were not retrieved - rule 91. Note the isoform trap on this record, which this beat fell into once: a molecule-scoped FUNCTION comment for Isoform 3 (s-JMJD1C) reads only 'Coactivator of androgen receptor', and a reader taking the last FUNCTION block gets that instead of the protein's function. The MDC1/DSB-repair annotation is ECO:0000269 on a single citation (PubMed:24240613).

    Evidence that argues against this (1)
    • The functional genetics say the JmjC domain matters whatever it does: CRISPR negative-selection screening in MLL-rearranged leukaemia found both the catalytic JmjC domain and the zinc finger required for cell survival in vitro and in vivo - which is equally consistent with catalysis and with the scaffold reading, and so does not settle the enzymology but does establish that the domain the compound was screened against is functionally required. Izaguirre-Carbonell J; Christiansen L; Burns R; Schmitz J; Li C; Mokry RL; et al. 2019
Connected to