Oncofusion

start / graph / gene-kdm3c

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

JMJD1C / KDM3C - named in three entries of this graph as one of the only loci that loses fusion occupancy under a menin inhibitor, ranked 23rd of 17,728 for KMT2A-selective dependency in this project's own scan, with the strongest published shRNA effect in SEM - and no entry until now

also called JMJD1C · KDM3C · JHDM2C · TRIP8 · KIAA1380 · s-JMJD1C (isoform 3) · jumonji domain-containing protein 1C · jumonji domain containing 1C

JMJD1C is the close cousin of the enzyme that guards the KMT2A gene from breaking, and the leukaemia needs it: switching it off kills KMT2A-fusion leukaemia cells hardest of all, and the strongest effect ever published was in SEM, the exact cell line this project's index leukaemia is studied in. The awkward part is that the drug used to study the guard enzyme also hits this cousin, so nobody yet knows which of the two any published result belongs to.

The same thing, for a scientist

JMJD1C (Q15652, 2,540 aa, HGNC:12313; approved alias KDM3C) is the third KDM3/JHDM2-family member, sharing KDM3B's domain architecture - JmjC at 2274-2498, a C6-type zinc finger at 1846-1871 and an LXXLL nuclear-receptor motif at 2066-2070 [E1]. **Three things make it load-bearing in this graph and it had no entry.** First, it is already named in `mechanism-the-menin-inhibitor-loss-set-is-small-and-named` as one of only three loci - with MEIS1 and MEF2C - that lose more than 2-fold fusion occupancy under VTP-50469 or MEN1 deletion, while HOXA and MYB do not. Second, this project's own genome-wide DepMap scan ranks it **23rd of 17,728** for KMT2A-selective dependency: -0.4697 in 11 KMT2A-rearranged leukaemia lines against -0.0162 in 43 other leukaemia lines, with a pan-essentiality of -0.0121, i.e. required in the rearranged lines and close to dispensable everywhere else [E2]. Third, the founding shRNA screen's **strongest effect was in SEM, a KMT2A-rearranged ALL line** - this project's index-fusion cell line - via apoptosis [E3]. **And it is the reason the KDM3B locus story cannot yet be attributed to KDM3B**, because the compound used to phenocopy KDM3B loss, JDI-12, was discovered and published as a JMJD1C modulator and binds both paralogues [E4]. **The enzymology is worse than unclear: UniProt states outright that it is uncertain whether JMJD1C removes methyl groups from histone proteins at all**, its canonical annotated function being demethylation of the non-histone substrates MDC1 - enabling RNF8-dependent MDC1 ubiquitination essential for double-strand-break repair - and STAT3 [E1].

What it looks like
JMJD1C2,540 aa
JmjC (catalytic demethylase domain)JmjC (catalytic demethylase domain) · residues 2274–2498Zinc finger, C6-type · residues 1846–1871LXXLL motif (nuclear-receptor box) · residues 2066–207005001,0001,5002,0002,500
loading AF-Q15652-F1…
AlphaFold AF-Q15652-F1 ↗
colourshape

This is a prediction, not a measurement. No one has seen this protein in this shape. The colours say how much the software trusts itself, and that is worth reading before anything else — the orange stretches are places the model is effectively saying this part has no fixed shape.

very high (90+)confident (70–90)low (50–70)very low (<50)
I have never read one of these pictures before

A protein is one long chain of amino acids that folds up. The ribbon traces that chain from one end to the other; it is not a wire, it is a simplification of where the backbone runs. Drag to rotate, scroll to zoom.

Domains are the stretches of chain that do a particular job, and they are what the coloured blocks above the picture mark out. Click one and the view flies to it. In this project the important question about any domain is usually whether the fusion still has it — the labels say lost where it does not.

Anything in yellow sticks is a bound small molecule, which is usually the drug.

Go deeper
Why this entry exists: three facts already in this graph that had nowhere to point
existing contentwhat it says about JMJD1C
mechanism-the-menin-inhibitor-loss-set-is-small-and-namedunder VTP-50469, MLL1n occupancy falls >2-fold at MEIS1, MEF2C and JMJD1C and not at HOXA, MYB or GAPDH; MEN1 CRISPR deletion plus CUT&RUN reproduces the same three
the same entry's open question'What makes MEIS1, MEF2C and JMJD1C menin-dependent when HOXA and MYB are not?'
analyses/depmap-kmt2a-selective-scan-fullJMJD1C is the 23rd most KMT2A-selective dependency in the genome, and nothing in the corpus said so
concept-eviction-names-menin-not-the-fusionnames JMJD1C in the small loss set while arguing about what eviction means

So the graph has been carrying JMJD1C as a target of the fusion for several cycles, in three entries, with a rank-23 dependency signal sitting unquoted in its own analysis directory. This beat arrived at it from the opposite end - the KDM3 family's role in whether the KMT2A locus rearranges at all - and the two ends are the same protein family.

The shape that results is worth stating plainly. KDM3B gates whether KMT2A becomes a fusion; JMJD1C is a dependency of the fusion once it exists and a locus the fusion sits on. One family, two disease stages, opposite therapeutic directions, and one compound that cannot tell them apart. That is the content of hyp-kdm3-family-sits-on-both-ends-of-this-disease.

What JMJD1C is required for, and where the therapeutic window claim comes from

The dependency is well replicated and it is LSC-selective.

  • An epigenetic shRNA library screen in murine MLL-AF9 AML, counter-screened against normal mouse bone marrow, identified Jmjd1c as one of the best candidates; depletion impaired growth, colony formation and leukaemia establishment after transplant, and in human lines the strongest effect was in the MLL-rearranged ALL line SEM, primarily through increased apoptosis 3.
  • An in vivo shRNA screen plus a conditional mouse showed loss of JMJD1C substantially decreased LSC frequency and drove differentiation of MLL-AF9- and HOXA9-driven leukaemias; JMJD1C directly interacts with HOXA9 and modulates a HOXA9-controlled expression programme; loss caused only minor defects in blood homeostasis and modest effects on HSC self-renewal 5.
  • CRISPR negative-selection screening localised the requirement to the catalytic JmjC domain and the zinc finger; H3K36me - not H3K9me - was the marker of JMJD1C activity at gene loci; and MLLr AML bearing activating RAS mutations was resistant to JMJD1C loss, with IL-3 receptor upregulation and RAS/MAPK and JAK-STAT activation on JmjC disruption 6.
  • Small-molecule KDM3-family modulators kill MLL-rearranged acute leukaemia cells and not cells resistant to JMJD1C depletion, and not cord blood cells 4.

And the window claim has a documented hole. Jmjd1c knockout mice show no significant change in peripheral blood composition and JAK2V617F-driven MPN initiates normally in them, which the authors read as redundancy among JmjC-domain family members in that context 7. So 'dispensable for normal haematopoiesis' is measured as blood counts and disease initiation, endpoints that cannot see a cytogenetic event - and the same compound class, in primary cord-blood CD34+ HSPCs, produces KMT2A copy gains and break-apart 8. The safety claim and the hazard are measured with different instruments and nobody has run both.

The enzymology is contested in the database itself, and that matters for the compound

UniProt's canonical FUNCTION comment for Q15652 says, in its own words, that 'It is uncertain whether JMJD1C removes methyl groups from histone proteins as the other JMJD1/KDM3 proteins do', citing three separate papers, and offers two alternatives: that JMJD1C is a context-specific histone demethylase whose activity is regulated post-translationally, or that it acts as a scaffold recruiting or stabilising other demethylase complexes 1. What the record does annotate experimentally is non-histone demethylation: of MDC1, promoting MDC1-RNF8 interaction and the RNF8-dependent MDC1 ubiquitination that is essential for double-strand-break repair, and of STAT3 at Lys-140 1. Independent support for the doubt: the CRISPR domain-mapping study found H3K36me, not H3K9me, marking JMJD1C activity at gene loci 6.

Three consequences.

  1. A compound described as a KDM3/H3K9 demethylase inhibitor may not be acting on H3K9 at all when its principal target is JMJD1C. The global H3K9 methylation increase reported for JDI-4 and JDI-12 in cells 4 is consistent with KDM3B or KDM3A inhibition and does not establish which paralogue carries the anti-leukaemic effect.
  2. There is a second, entirely separate route by which inhibiting JMJD1C could raise rearrangement frequency: through MDC1 and DSB repair. Gray et al.'s mechanism requires a break to be repaired for the alteration to become inherited, and it cites NHEJ and TDP2 work for that step 8. A compound that impairs MDC1-dependent repair while also raising H3K9me1/2 would push both halves of the mechanism in the same direction, and nobody has separated them.
  3. The isoform trap is live and this beat fell into it. Q15652 carries three isoforms, and the molecule-scoped FUNCTION comment for Isoform 3 (s-JMJD1C) reads simply 'Coactivator of androgen receptor'. A first pass at this record in this beat printed that comment as the protein's function, because the reader took the last FUNCTION block rather than the unscoped one. The isoform-scoped comment is not the protein's function, and reading it as such would have made a 2,540 aa DNA-damage-response demethylase into an androgen-receptor coactivator - the same failure class this graph already records for the KMT2A N320/C180 fragment comments.
The identifier position: the approved symbol is the minority spelling

Measured 2026-09-22 on live esearch, full ID sets retrieved and compared as sets 9:

operandrecords
JMJD1C[tiab]141
KDM3C[tiab]13
union149
reachable only via the approved alias KDM3C8

HGNC's approved symbol is JMJD1C and KDM3C is listed as a synonym, so unlike the EHMT2/G9a row already in the registry the approved symbol is here also the productive one - but 8 records (5.4% of the union) are KDM3C-only, and this graph's own node id is gene-kdm3c because the graph's demethylase ids follow the KDM naming (gene-kdm3b, gene-kdm4c). Both strings must be run. JMJD1C[tiab] AND (MLL[tiab] OR KMT2A[tiab]) returns 7 records, of which one is a mangrove-rivulus fish Kdm/Kmt gene-family survey - the same off-topic record that contaminates the EHMT2 x KMT2A set recorded in gene-ehmt2 E5, which is a useful signature that a small intersection is being padded by a comparative-genomics paper.

Facts and figures
uniprotQ15652
gene symbolJMJD1C
approved symbol noteHGNC:12313 approves JMJD1C; KDM3C is a synonym. This node's id is `gene-kdm3c` to match this graph's KDM naming for demethylases (gene-kdm3b, gene-kdm4c). Both strings must be used when searching.
length AA2540
hgncHGNC:12313
entrez gene ID221037
chembl targetCHEMBL3792271
alphafold IDAF-Q15652-F1
accession resolutiongene_exact:JMJD1C and gene_exact:KDM3C (each AND organism_id:9606 AND reviewed:true) both return the SAME single row Q15652 on 2026-09-22; primary gene name JMJD1C and length 2,540 read back from that row.
familyKDM3/JHDM2 - KDM3A (Q9Y4C1, 1,321 aa), KDM3B (Q7LBC6, 1,761 aa), JMJD1C/KDM3C (Q15652, 2,540 aa)
domain architecture notethe same three elements in the same order as KDM3B - JmjC, C6-type zinc finger, LXXLL - and CRISPR negative selection in MLL-rearranged leukaemia requires BOTH the JmjC and the zinc finger, so the shared architecture is functionally shared and not only sequence-level
isoformsthree (Q15652-1 displayed, -2, -3). Isoform 3 is s-JMJD1C and carries its OWN molecule-scoped FUNCTION comment, 'Coactivator of androgen receptor'. That comment is NOT the protein's function and this beat misread it once before reading the unscoped comment.
localizationNUCLEUS, and here the annotation IS experimental - ECO:0000269 on two independent citations (PubMed:23593242 and PubMed:32034158), with Isoform 3 separately annotated as nuclear on ECO:0000269 PubMed:17353003. That is a stronger localisation record than its paralogue KDM3B, whose single 'Nucleus' value rests on sequence similarity alone. Neither underlying paper was retrieved for this entry, so rule 91 applies to which experiment the curator saw.
subnuclear localizationNEVER MEASURED. No nucleolar, speckle, lamina, PML-body, chromocentre or repair-focus placement for JMJD1C exists in UniProt or in the literature reached by this beat, in any cell type. The one inferential lead worth naming, and it is a lead only: JMJD1C's annotated experimental function is demethylating MDC1 to enable RNF8-dependent ubiquitination in double-strand-break repair, and MDC1 is a canonical component of IONISING-RADIATION-INDUCED FOCI - so if JMJD1C acts on MDC1 in situ it should be recruitable to damage foci, which is a one-experiment imaging question nobody has asked. No imaging of JMJD1C in any leukaemic cell exists. Absence class: never-measured.
localization shown for fusionNO for the protein's own compartment, in any KMT2A fusion, in any lineage - JMJD1C has never been imaged in a KMT2A-rearranged cell. What HAS been shown in the index fusion is a dependency and a chromatin address, not a compartment: shRNA depletion has its strongest reported effect in SEM (KMT2A::AFF1, B-lineage) through apoptosis, and the JMJD1C LOCUS loses more than 2-fold fusion occupancy under VTP-50469 and under MEN1 deletion. Those are a phenotype and a DNA address respectively, and neither is localisation.
epigenetic readoutH3K36 methylation was the mark found to mark JMJD1C activity at gene loci in CRISPR domain mapping (PMID 31076406); H3K9me2 demethylation at hepatic lipogenic promoters is a reported but hedged route in the UniProt record; and the JMJD1C LOCUS itself is a chromatin readout in this graph - MLL1n occupancy at JMJD1C falls more than 2-fold under VTP-50469 and under MEN1 deletion, one of only three such loci.
shown for fusionYES, and in the index fusion: strongest shRNA effect of the human panel in SEM (KMT2A::AFF1, B-lineage) via apoptosis (PMID 24501218, abstract only); conditional loss reduces LSC frequency in murine MLL-AF9 (PMID 26878175); JmjC and zinc finger both required in MLLr leukaemia (PMID 31076406); 23rd most KMT2A-selective dependency in this project's DepMap scan. NOTHING in an infant or primary human KMT2A-rearranged sample.
why this entry existsJMJD1C was named in mechanism-the-menin-inhibitor-loss-set-is-small-and-named, concept-eviction-names-menin-not-the-fusion and this project's own DepMap scan without ever having a node, while its rank-23 selectivity sat unquoted in analyses/. This beat reached it from the locus-genesis end because the compound used to phenocopy KDM3B loss is a published JMJD1C modulator.
absence classnever-measured
hgnc approved symbolJMJD1C
symbol authority noteThis entry's id uses KDM3C, which HGNC does not approve; the approved symbol is JMJD1C (jumonji domain containing 1C). The id is deliberately NOT renamed - entry ids are referenced by more than 10,800 edges. The approved symbol is carried in aliases instead, so that a symbol-keyed search finds this entry. Recorded 2026-09-22 by analyses/entries-written-under-a-non-approved-symbol/.
domains
Zinc finger, C6-type1,846Zinc finger, C6-type: 1846LXXLL motif (nuclear-rece…2,066LXXLL motif (nuclear-receptor box): 2066JmjC (catalytic demethyla…2,274JmjC (catalytic demethylase domain): 2274
domains — the numbers behind the chart
[
  {
    "name": "JmjC (catalytic demethylase domain)",
    "start": 2274,
    "end": 2498,
    "pfam": "PF02373"
  },
  {
    "name": "Zinc finger, C6-type",
    "start": 1846,
    "end": 1871,
    "pfam": null
  },
  {
    "name": "LXXLL motif (nuclear-receptor box)",
    "start": 2066,
    "end": 2070,
    "pfam": null
  }
]
PDB IDS
[]
substrates
{
  "non_histone_annotated_experimentally": "MDC1 - demethylation promotes MDC1-RNF8 interaction and RNF8-dependent MDC1 ubiquitination essential for DOUBLE-STRAND BREAK REPAIR (ECO:0000269 PubMed:24240613); STAT3 Lys-140 (by similarity)",
  "histone_status": "CONTESTED IN THE DATABASE ITSELF. UniProt: 'It is uncertain whether JMJD1C removes methyl groups from histone proteins as the other JMJD1/KDM3 proteins do', citing three papers, with context-specific demethylation or a scaffold role offered as alternatives. A reported route is USF1-recruited JMJD1C at hepatic lipogenic promoters potentially promoting H3K9me2 demethylation.",
  "mark_named_by_the_leukaemia_work": "H3K36me, not H3K9me, was the marker of JMJD1C activity at gene loci in the CRISPR domain-mapping study"
}
cell type
{
  "required_in": "leukaemia stem cells of MLL-AF9- and HOXA9-driven murine AML (LSC frequency falls, cells differentiate); human leukaemic lines, STRONGEST in SEM, a KMT2A-rearranged B-lineage ALL line; 11 KMT2A-rearranged DepMap leukaemia lines (mean gene effect -0.4697)",
  "largely_dispensable_in": "murine haematopoietic stem cells (minor blood-homeostasis defects, modest HSC self-renewal effects); Jmjd1c-knockout mouse peripheral blood composition is unchanged; JAK2V617F MPN initiates normally in Jmjd1c-null mice",
  "the_cell_type_that_matters_here": "SEM. This graph's recurring finding is that its load-bearing evidence is imported from murine myeloid MLL-AF9; the JMJD1C dependency has a datum in the index fusion's own B-lineage line, and it is the strongest datum in that panel. It is one line, one technology, abstract-only, and it should be read in full before being leaned on.",
  "not_measured": "no JMJD1C measurement in an infant primary sample, in fetal liver, or in any human primary KMT2A-rearranged material"
}
cell state
{
  "state": "STEM/SELF-RENEWAL-SCOPED, and that is the whole point of the target: loss reduces LSC frequency and drives differentiation while leaving bulk blood homeostasis largely intact. So 'is JMJD1C required' has different answers in a leukaemia stem cell and a normal HSC.",
  "genotype_dependent_state": "MLL-rearranged AML bearing ACTIVATING RAS MUTATIONS is resistant to JMJD1C loss, with IL-3 receptor upregulation driving RAS/MAPK and JAK-STAT activation on JmjC disruption - so resistance is a pre-existing genotype rather than an acquired state.",
  "redundancy_state": "In the JAK2V617F MPN context the authors attribute dispensability to redundancy among JmjC-domain family members, which is the same buffering argument that applies to any single-paralogue intervention in this family.",
  "bounded": "no JMJD1C cell-state measurement exists in a B-lineage KMT2A-rearranged primary cell"
}
depmap kmt2a selectivity
{
  "gene": "JMJD1C",
  "difference": -0.4535,
  "mean_kmt2a_rearranged": -0.4697,
  "mean_other_leukaemia": -0.0162,
  "pan_essentiality_all_lines": -0.0121,
  "n_all_lines": 1208,
  "selectivity_rank": 23,
  "rank_denominator": 17728,
  "percentile": 0.13,
  "empirical_p_vs_scan": 0.001636,
  "paralogue_for_contrast": "KDM3B, rank 9,382 of 17,728, difference +0.0018, pan-essentiality +0.0226",
  "provenance": "Read 2026-09-22 from analyses/depmap-kmt2a-selective-scan-full/results.json (generated 2026-09-19T10:12:06Z, seed 20260918, 11 KMT2A-rearranged and 43 other leukaemia lines, none infant-derived). Not re-fetched; DepMap release not stated by the API (rule 83).",
  "how_to_read": "A large negative difference with a near-zero pan-essentiality is the shape of a selective dependency rather than a housekeeping gene. It is still a cell-line difference, not a mechanism, and RAS-mutant MLLr AML is published as resistant."
}
identifier trap
{
  "token": "JMJD1C / KDM3C -- the approved symbol is the majority spelling here, which is the OPPOSITE of the EHMT2/G9a row",
  "measured_2026_09_22": {
    "JMJD1C[tiab]": 141,
    "KDM3C[tiab]": 13,
    "union": 149,
    "kdm3c_only": 8,
    "kdm3c_only_pct": 5.4,
    "JMJD1C AND (MLL OR KMT2A)": 7,
    "JMJD1C AND leukemia": 12
  },
  "id_sets_compared_not_counts": true,
  "registry_row_existed_before": false,
  "registry_grepped_first": true,
  "contaminated_alias": "TRIP8 is a UniProt-listed synonym and must NOT be run bare - it is also used for the HCN-channel accessory subunit PEX5L",
  "shared_off_topic_record": "PMID 30458291, a mangrove-rivulus fish Kdm/Kmt gene-family survey, contaminates both the JMJD1C x MLL set and the EHMT2 x KMT2A set recorded in gene-ehmt2 E5 - a useful signature that a small intersection is padded by comparative genomics",
  "the_trap_that_matters": "A COMPOUND NAMED FOR A GENE FAMILY IS NOT A TARGET ASSIGNMENT. 'KDM3i' in PMID 37788669 is JDI-12, published as a JMJD1C modulator, binding both JMJD1C and KDM3B; the KMT2A-selective dependency belongs to JMJD1C (rank 23) and not to KDM3B (rank 9,382)."
}
How this could bite us back
  • A JMJD1C inhibitor's safety case rests on endpoints that cannot see this beat's hazard. 'Dispensable for healthy adult haematopoiesis' was measured as peripheral blood composition in a knockout mouse; the hazard is a KMT2A copy gain and break-apart in a primary human cord-blood progenitor, detected by DNA FISH. Any development programme here needs cytogenetics as a safety readout, not counts.
  • Resistance is pre-specified and common: MLL-rearranged AML with activating RAS mutations is published as resistant to JMJD1C loss, via IL-3 receptor upregulation and RAS/MAPK and JAK-STAT activation. This graph already holds RAS co-mutation entries, so a predictable fraction of patients is outside the target's reach before a molecule exists.
  • The family is redundant where it has been tested for redundancy: the MPN study attributes Jmjd1c dispensability to compensation among JmjC-domain family members. A single-paralogue inhibitor in a three-paralogue family is the intervention most likely to fail quietly.
  • If the JmjC domain acts as a scaffold rather than a catalyst - one of UniProt's two stated alternatives - then an active-site inhibitor could be the wrong modality entirely, and the CRISPR result showing the JmjC domain is required would have been equally consistent with that.
  • The same compound class is simultaneously a proposed treatment for KMT2A-rearranged leukaemia and a demonstrated inducer of KMT2A rearrangement in normal cord-blood progenitors. A drug that treats the disease it can also cause needs its therapeutic index expressed in cytogenetic terms, and nobody has measured one.
Still open
  1. Which paralogue does JDI-12 act through in a KMT2A-rearranged cell? It binds JMJD1C and KDM3B, the published selectivity is JMJD1C's, and this project's DepMap scan assigns the KMT2A-selective dependency to JMJD1C - so every phenotype attributed to 'KDM3B inhibition' in a leukaemia cell is currently unassigned. A JMJD1C-null and a KDM3B-null cell treated with the compound would settle it.
  2. Does JMJD1C actually demethylate a histone in a KMT2A-rearranged cell? UniProt says it is uncertain in general, the domain-mapping study points at H3K36me rather than H3K9me, and nobody has run a histone-mark panel in SEM or RS4;11 after JMJD1C loss.
  3. Is the MDC1/DSB-repair function relevant to KMT2A rearrangement? JMJD1C's only experimentally annotated substrate enables RNF8-dependent MDC1 ubiquitination for double-strand-break repair, and the locus mechanism in this beat requires a break to be repaired for the alteration to be inherited. A compound that raises H3K9me1/2 AND impairs MDC1-dependent repair pushes both halves the same way, and the two have never been tested together.
  4. Why is the JMJD1C LOCUS one of only three that lose fusion occupancy under menin inhibition? That question is already open in mechanism-the-menin-inhibitor-loss-set-is-small-and-named; this entry adds that the gene at that locus is also the 23rd most KMT2A-selective dependency in the genome, which makes it the best-motivated of the three to answer first.
  5. Does the SEM dependency replicate? It is the only B-lineage, index-fusion datum in this entry, it is 2014 shRNA, and it is known to this graph only from an abstract. A CRISPR re-test in SEM and RS4;11 is a week of work and would either promote or retire the most important sentence here.
Every source behind this entry (9)
  1. E1established confidence 5 of 5: Multiple independent lines of evidence; treated as settled in the field.database

    JMJD1C is UniProt Q15652 (JHD2C_HUMAN), 2,540 aa, HGNC:12313, with approved gene name JMJD1C and synonyms JHDM2C, KDM3C, KIAA1380 and TRIP8; it carries a JmjC domain at 2274-2498, a C6-type zinc finger at 1846-1871 and an LXXLL motif at 2066-2070; its canonical FUNCTION comment records 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, states explicitly that 'It is uncertain whether JMJD1C removes methyl groups from histone proteins as the other JMJD1/KDM3 proteins do', and offers context-specific histone demethylation or a scaffold role as alternatives; its subcellular location is Nucleus on ECO:0000269; and a molecule-scoped FUNCTION comment for Isoform 3 (s-JMJD1C) reads only 'Coactivator of androgen receptor'.

    What this evidence cannot show

    **Accession resolved live 2026-09-22: both `gene_exact:JMJD1C` and `gene_exact:KDM3C` (AND organism_id:9606 AND reviewed:true) return the SAME single row, Q15652, and the primary gene name (JMJD1C) and length (2,540) were read back from it.** The MDC1 demethylation claim is ECO:0000269 on PubMed:24240613, which was not retrieved here - rule 91 applies. The STAT3 demethylation and the spermatogenesis role are 'By similarity' (ECO:0000250 from the mouse orthologue Q69ZK6), not human measurements. **Reading discipline note recorded because this beat failed it once:** Q15652 has three isoforms and a molecule-scoped FUNCTION comment on Isoform 3; a script that takes the last FUNCTION block returns 'Coactivator of androgen receptor' as the protein's function, which is wrong, and that is what happened on the first pass here before the unscoped comment was read.

    Evidence that argues against this (1)
    • The histone-demethylase doubt is corroborated from the functional side rather than only asserted: CRISPR domain mapping in MLL-rearranged AML found H3K36 methylation, not H3K9 methylation, to be the marker of JMJD1C activity at gene loci - so even where the JmjC domain is demonstrably required for leukaemia-cell survival, the mark it is acting on may not be H3K9. Izaguirre-Carbonell J; Christiansen L; Burns R; Schmitz J; Li C; Mokry RL; et al. 2019
  2. E2strong confidence 4 of 5: Replicated in peer-reviewed work by more than one group.dataset

    In this project's own prior-free genome-wide DepMap scan over 17,728 protein-coding genes, 11 Cellosaurus-annotated KMT2A-rearranged leukaemia lines and 43 other leukaemia lines, JMJD1C ranks 23rd (percentile 0.13) with a difference of -0.4535 (mean -0.4697 in KMT2A-rearranged lines versus -0.0162 in other leukaemia lines), a pan-essentiality across all 1,208 lines of -0.0121, and an empirical p versus the scan of 0.0016 - while its paralogue KDM3B ranks 9,382nd.

    What this evidence cannot show

    Read directly from the stored results file on 2026-09-22, not quoted from a memo or a ledger. The scan's own caveats apply in full: **11 KMT2A-rearranged lines, none infant-derived**; Cellosaurus fusion annotation may be incomplete, biasing toward the null; the empirical p is the fraction of all scanned genes with an equally extreme difference, i.e. the genome-wide background rather than a separate random draw; and **the DepMap release is not stated by the API** (rule 83). A near-zero pan-essentiality with a large negative difference is the shape a selective dependency should have, but a cell-line dependency difference is not a mechanism, and none of these lines is a primary infant sample.

    Evidence that argues against this (1)
    • A rank this high in a 54-line scan is not by itself an actionable dependency: MLL-rearranged AML bearing activating RAS mutations is resistant to JMJD1C loss, with IL-3 receptor upregulation and RAS/MAPK and JAK-STAT activation on JmjC disruption. RAS co-mutation is common in KMT2A-rearranged disease, and this graph already holds RAS/FLT3 co-mutation entries - so a fraction of patients is predicted resistant before any drug exists. Izaguirre-Carbonell J; Christiansen L; Burns R; Schmitz J; Li C; Mokry RL; et al. 2019
  3. E3strong confidence 4 of 5: Replicated in peer-reviewed work by more than one group.paper

    An epigenetic shRNA library screen in a mouse MLL-AF9 AML model, counter-screened against normal mouse bone marrow cells for general shRNA toxicity, identified Jmjd1c as one of the best candidate drug targets; Jmjd1c depletion impaired growth and colony formation of mouse MLL-AF9 cells in vitro and leukaemia establishment after transplantation, JMJD1C depletion impaired expansion and colony formation of human leukaemic cell lines with THE STRONGEST EFFECT OBSERVED IN THE MLL-REARRANGED ALL CELL LINE SEM, and in both mouse and human cells the growth defect appeared to be primarily due to 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**, so the SEM effect size, the number of hairpins, the knockdown efficiency and whether an off-target control or rescue was performed are uncharacterised - and the SEM result is the part that matters most to this graph, because SEM is a KMT2A::AFF1 B-lineage line and this project has repeatedly found that its load-bearing evidence is partner-imported from murine myeloid systems. Read it before quoting the magnitude. 'Strongest effect' is a comparison across the cell lines that happened to be in the panel, not a systematic lineage comparison. The screen is shRNA, so seed-based off-target effects are a live concern in a 2014 dataset.

    Evidence that argues against this (1)
    • Dependency in a cell line does not establish a therapeutic window: 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 context and redundant in another within the haematopoietic system. Staehle HF; Heinemann J; Gruender A; Omlor AM; Pahl HL; Jutzi JS 2020
  4. E4strong confidence 4 of 5: Replicated in peer-reviewed work by more than one group.paper

    JDI-4 and JDI-12, the compounds identified by virtual screening against the catalytic Jumonji domain of JMJD1C, bind JMJD1C AND its family homolog KDM3B with modest affinity by surface plasmon resonance; JDI-4 reverses the H3K9 demethylation conferred by KDM3B in vitro; both induce a global increase of H3K9 methylation in cells; and both kill MLL-rearranged acute leukaemia and other malignant haematopoietic cells but not leukaemia cells resistant to JMJD1C depletion and not cord blood cells - so the compound that PMID 37788669 calls 'KDM3i' is a published JMJD1C modulator that binds both paralogues.

    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.** 'Modest affinity' is the authors' own word with no Kd in the abstract, and there is no reported selectivity panel against KDM3A, the KDM4 family or other 2-oxoglutarate oxygenases - so 'KDM3 family inhibitor' describes two measured family members and not a selectivity profile. These are virtual-screening hits from natural-product and Chinese-medicine libraries, not optimised chemical probes; the abstract does not state which H3K9 methylation state the in vitro demethylation assay read out. The 'resistant to JMJD1C depletion' comparator is the paper's own internal control and its identity is not given in the abstract.

    Evidence that argues against this (1)
    • The compound's target assignment is undermined by its own target's annotation: UniProt states it is uncertain whether JMJD1C removes methyl groups from histone proteins at all, and its canonical annotated activity is non-histone demethylation of MDC1 for double-strand-break repair. A compound selected against the JmjC domain of a protein whose histone activity is doubted cannot be read as an H3K9-pathway tool without independent evidence. UniProt Consortium 2026
  5. E5strong confidence 4 of 5: Replicated in peer-reviewed work by more than one group.paper

    An in vivo shRNA screen identified JMJD1C as an important driver of MLL-AF9 leukaemia; in a conditional mouse, loss of JMJD1C substantially decreased leukaemia-stem-cell frequency and caused differentiation of MLL-AF9- and HOXA9-driven leukaemias; JMJD1C directly interacts with HOXA9 and modulates a HOXA9-controlled gene-expression programme; and loss of JMJD1C led to only minor defects in blood homeostasis and modest effects on HSC self-renewal.

    MLL-AF9- and HOXA9-mediated acute myeloid leukemia stem cell self-renewal requires JMJD1C. · Zhu N; Chen M; Eng R; DeJong J; Sinha AU; Rahnamay NF; et al. · J Clin Invest 2016
    PMID 26878175DOI
    What this evidence cannot show

    **Abstract only; full text not retrieved.** Murine, myeloid, MLL-AF9 and HOXA9 - no KMT2A::AFF1 and no B-lineage arm, which matters because the one B-lineage datum in this entry (SEM, E3) comes from a different screen with a different technology. The direct HOXA9 interaction is asserted in the abstract without the method stated. 'Minor defects in blood homeostasis' is the therapeutic-window claim and is an endpoint that cannot detect a cytogenetic event; see E7 and E8.

    Evidence that argues against this (1)
    • The HOXA9 link cuts against the simplest reading of this graph's own menin-inhibitor data: JMJD1C is one of the three loci that LOSE fusion occupancy under menin inhibition, while HOXA retains occupancy even as HOXA transcription falls. If JMJD1C both needs HOXA9 and is a menin-dependent fusion target, the regulatory relationship is circular and its direction is not established by either dataset. Krivtsov AV; Evans K; Gadrey JY; Eschle BK; Hatton C; Uckelmann HJ; et al. 2019
  6. E6strong confidence 4 of 5: Replicated in peer-reviewed work by more than one group.paper

    JMJD1C is aberrantly expressed in MLL-rearranged leukaemias; CRISPR/Cas9 negative-selection screening identified a requirement for the catalytic JmjC domain AND the zinc finger domain for leukaemia cell survival in vitro and in vivo; histone H3 lysine 36 methylation - not H3K9 methylation - was found to be the marker of JMJD1C activity at gene loci; single-cell transcriptomics of cells carrying a JmjC-domain sgRNA showed increased RAS/MAPK and JAK-STAT activation via upregulation of interleukin-3 receptor genes; and MLL-rearranged AML bearing activating RAS mutations was resistant to JMJD1C loss.

    Critical role of Jumonji domain of JMJD1C in MLL-rearranged leukemia. · Izaguirre-Carbonell J; Christiansen L; Burns R; Schmitz J; Li C; Mokry RL; et al. · Blood Adv 2019
    PMID 31076406DOI
    What this evidence cannot show

    **Abstract only; full text not retrieved**, so the H3K36me finding - which is the substantive challenge to calling this an H3K9 enzyme - is uncharacterised as to antibody (rule 74), which H3K36 methylation state, and whether the association is causal or correlative. Mouse leukaemia cells for the single-cell arm; the lineage of the in vivo work is myeloid MLLr, with no B-lineage or KMT2A::AFF1 arm. The RAS resistance result is the clinically important one and is stated as a suggestion in the abstract rather than a quantified comparison.

    Evidence that argues against this (1)
    • That the JmjC domain is required does not establish that its catalytic activity is required on histones, and UniProt's own alternative reading is that JMJD1C may act as a scaffold recruiting or stabilising other active demethylase complexes - in which case a JmjC-domain sgRNA would disrupt a protein-protein surface rather than a catalytic function. UniProt Consortium 2026
  7. E7strong confidence 4 of 5: Replicated in peer-reviewed work by more than one group.paper

    Jmjd1c knockout mice are dispensable for healthy adult haematopoiesis - showing no significant changes in peripheral blood composition - and Cre-inducible JAK2V617F introduced into Jmjd1c-null mice produced a myeloproliferative phenotype similar to JAK2V617F in a wild-type background, so Jmjd1c is also dispensable for JAK2V617F-driven MPN disease initiation; the authors conclude there is a difference between JMJD1C's role in leukaemic stem cells and in MPN, and that JmjC-domain family members may serve redundant roles compensating for the loss of individual proteins.

    Jmjd1c is dispensable for healthy adult hematopoiesis and Jak2V617F-driven myeloproliferative disease initiation in mice. · Staehle HF; Heinemann J; Gruender A; Omlor AM; Pahl HL; Jutzi JS · PLoS One 2020
    PMID 32017785DOI
    What this evidence cannot show

    **Abstract only; full text not retrieved.** Murine germline knockout, so lifelong compensation is available in a way it is not to an acute inhibitor - the reverse of the usual caveat direction, and it means 'dispensable' here is a statement about a compensated animal. The endpoints are **peripheral blood composition and MPN disease initiation**: neither can detect a locus-level cytogenetic event, which is precisely the hazard this beat is about. The disease model is JAK2V617F MPN, not a KMT2A fusion.

    Evidence that argues against this (1)
    • The same family's inhibition in primary human cord-blood CD34+ HSPCs produced KMT2A copy gains and break-apart events by DNA FISH. Normal blood counts in a knockout mouse and a cytogenetic alteration in a primary human progenitor are not in conflict - they are different assays - and the safety inference drawn from the first is not licensed against the second. Gray ZH; Chakraborty D; Duttweiler RR; Alekbaeva GD; Murphy SE; et al. 2023
  8. E8strong confidence 4 of 5: Replicated in peer-reviewed work by more than one group.paper

    The KDM3-family inhibitor JDI-12, at 1 microM for 72 hours, significantly increased KMT2A DNA copy gains and genomic rearrangements by two-colour DNA FISH in KG1a, a primary patient-derived AML line, a primary patient-derived AML organoid on irradiated OP9 stroma, and primary CD34+ HSPCs immunomagnetically enriched from fresh human umbilical cord blood; in RPE cells 25 nM for 12 hours was sufficient to produce transient alterations reversible on washout, and 72 hours produced alterations inherited into later passages.

    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 read 2026-09-22 via efetch db=pmc on PMC10591855 (Europe PMC fullTextXML returned HTTP 500; <body> confirmed, 80,638 stripped characters), including the STAR Methods, which is where the compound's identity, concentrations and cell sources come from.** The readout in every primary and leukaemic arm is DNA FISH copy gain and probe break-apart - a cytogenetic proxy, with no sequenced junction anywhere in the paper. The paper attributes the effect to KDM3B throughout and does not report which paralogue the compound is acting through in any of these cells; JDI-12 binds both. KG1a and HL60 carry KDM3B loss of heterozygosity, so their baseline is abnormal. Cord-blood HSPCs are a proxy for the fetal progenitor of infant t(4;11), not that cell.

    Evidence that argues against this (1)
  9. E9established confidence 5 of 5: Multiple independent lines of evidence; treated as settled in the field.database

    As of 2026-09-22, PubMed returns 141 records for JMJD1C[tiab] and 13 for KDM3C[tiab], with a union of 149 and 8 records reachable only through KDM3C; JMJD1C[tiab] AND (MLL[tiab] OR KMT2A[tiab]) returns 7 records, one of which is an off-topic mangrove-rivulus fish Kdm/Kmt gene-family survey - the same record that contaminates the EHMT2 x KMT2A intersection recorded in gene-ehmt2 E5 - and JMJD1C[tiab] AND leukemia[tiab] returns 12.

    What this evidence cannot show

    Titles and abstracts only. `data/symbol-collisions.json` was grepped for JMJD1C, KDM3C, JHDM2C and TRIP8 before querying and returned no row, so this is a new observation rather than a rediscovery (rule 88). The alias list run was JMJD1C and KDM3C; JHDM2C, KIAA1380 and TRIP8 were not run and would widen the union - TRIP8 in particular is a contaminated token, being also used for the HCN-channel accessory subunit PEX5L, so it should not be run bare.

    Evidence that argues against this (1)
    • A 5.4% alias-only fraction is small enough that the practical retrieval risk here is the opposite of the EHMT2 case: the danger is not missing literature but writing the node under the wrong id. This graph's demethylase ids follow KDM naming, so a reader searching the corpus for 'JMJD1C' would find three entries naming it and no node until this one. Iqbal Lab oncofusion knowledge graph symbol-collision registry 2026
Connected to