KDM4C (JMJD2C/GASC1) — recruited by the retained KMT2A N-terminus, drugged in a KMT2A::AFF1 line, and invisible to CRISPR

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.

• Essential-function toxicity: the family this drug hits is required for HSC maintenance in vivo (PMID 31434704). A child on a KDM4 inhibitor is a child whose stem-cell compartment is being suppressed at the same time as their marrow is being asked to recover, and the single-gene safety data cited in favour of KDM4C do not cover the family inhibitor that would actually be given.
• Compensation is not a hypothetical here, it is the default reading of our own data: KDM4C knockout does nothing in four KMT2A::AFF1 lines. Any KDM4C-selective agent, if one were made, should be expected to be buffered by KDM4A/B.
• Genome instability: H3K9me3 is the heterochromatin mark, and the cited literature notes that Suv39h loss reduces H3K9me3 and accelerates tumour development. Pharmacologically stripping H3K9me3 in an infant is a second-malignancy question nobody in this chain has asked.
• The strike risk runs the other way for this entry. If KDM4C is added to the target list on a SEM viability panel and the mechanism turns out to be family-level, the project will have spent a medicinal-chemistry cycle chasing a selectivity that the biology forbids.
• Delivery and exposure: SD70 is a tool compound dosed intraperitoneally in mice. No clinical KDM4 inhibitor exists, so every statement here is at least two development steps from a patient.
• ESSENTIAL-FUNCTION TOXICITY — The most on-point safety result in this tranche. KDM4 activity is required for HSC maintenance in vivo, shown in a study explicitly designed to probe the side effects of continuous KDM4 inhibition in MLL-translocated AML. A pan-KDM4 inhibitor given continuously is therefore predicted to exhaust normal HSCs over time, and a short in vitro CD34-sparing assay is not evidence against that.

1. Does SD70 sensitivity in SEM survive rescue by re-expression of KDM4A or KDM4B? That single experiment separates 'KDM4C is the target' from 'the family is the target' and nothing in the cited literature does it.
2. Is HOXA9 suppressed in a KMT2A::AFF1 line by SD70, or only viability? Every KDM4C mechanism claim in the cited work rests on HOXA9 in MLL-GAS7, MLL-AF9 and MOZ-TIF2 cells; the SEM result is a viability panel with no transcriptional readout.
3. Does the SD70/MI-503 synergy reported in MLL::AF9 reproduce in KMT2A::AFF1 infant B-ALL, where the cell of origin, the lineage and the menin dependency are all different?
4. If the KDM4 family requirement in normal HSC is real and paralogue-selective chemistry is not achievable, is there any exposure schedule that separates the leukemia from the stem cell, or is this a target that can only ever be a brief adjunct?
5. The normal-HSC toxicity result needed a Kdm4a/b/c triple knockout, and SD70 is a KDM4C inhibitor of unstated family selectivity. Does SD70 at efficacious concentrations inhibit KDM4A and KDM4B too? If it does, the in vitro CD34-sparing claim and the in vivo triple-knockout toxicity are measuring the same perturbation at different durations and the safety margin is much narrower than the combination paper implies.
6. Every KDM4C cell-context result is myeloid - AML LSC, MLL::AF9, murine HSC. KMT2A::AFF1 infant B-ALL has no KDM4C experiment of any kind. Does the KDM4C-ALKBH5-AXL axis even exist in a B-lineage blast, where AXL expression is not established?

1. E1strong confidence 4 of 5: Replicated in peer-reviewed work by more than one group.paper

   MLL fusions recruit KDM4C through an interaction domain mapped to the MLL 5' moiety that is present in all MLL fusions; KDM4C is required for initiation and maintenance of MLL-fusion leukemia in vivo; the KDM4C inhibitor SD70 preferentially suppressed human MLL-fusion leukemia cell lines including SEM; and KDM4C knockdown in normal HSPCs did not impair colony formation or multilineage reconstitution.

   Targeting Aberrant Epigenetic Networks Mediated by PRMT1 and KDM4C in Acute Myeloid Leukemia · Cheung N, Fung TK, Zeisig BB, et al. · Cancer Cell 2016
   PMID 26766589DOI
   ▸What this evidence cannot show

   Full text read (PMC4712026), not abstract only. The two verbatim sentences load-bearing for this entry are 'Biochemical mapping revealed the interaction domain located at MLL 5′, which is present in all MLL fusions' and 'we observed a specific and preferential suppression on human leukemia cell lines carrying MLL fusions (such as SEM and THP1) over the non-MLL leukemia cell lines'. Note the limits: the in vivo mouse work is MLL-GAS7, MLL-AF9 and MOZ-TIF2; the primary human xenograft is 'MLL3', an MLL-fusion AML sample whose partner is not specified in the text. No KMT2A::AFF1 in vivo experiment is reported. The SEM result is a cell-viability panel, not a HOXA9 readout, and the normal-HSPC reconstitution was assessed at 6 weeks, which is short for a self-renewal claim. **Resolved, and half the deposition is not a KMT2A fusion.** E-MTAB-3322 is 16 of 16 Mus musculus; by the sweep's reading of the sample files the fusions are **MOZ-TIF2 (8), MLL-GAS7 (6) and MLL-AF9 (2)**, with the study's own `clinical history` field splitting 8 mixed-lineage against 8 non-mixed-lineage. The coordinator confirmed Mus musculus and the presence of MOZ-TIF2 in the study metadata on 2026-09-19; the 8/6/2 split comes from the sweep's reading of the per-file names. Separately, the human 'MLL3' 90-day-survival xenograft this item leans on **deposited nothing**, and its fusion partner is never named even in the full text - and 'MLL3' there is a **patient-sample label, unrelated to the gene KMT2C**. Resolved 2026-09-19 by the Q18b unstated-systems sweep from the study's deposition, not from its text; bounded as a statement about the deposited data rather than about the whole study.

   ▸Evidence that argues against this (2)

   • The same paper is the reason PRMT1 should NOT be treated as a general KMT2A-fusion dependency, and by extension a reason to read its KDM4C claim carefully rather than enthusiastically. It reports that 'while transformation mediated by most of the other MLL fusions were not affected by Prmt1 knockdown, MLL-GAS7 and MOZ-TIF2 exhibited high degrees of Prmt1 dependence', and separately that 'MLL-AF9 recruits only Kdm4c but not Prmt1'. A screen across more than ten fusions in which most were unaffected is a screen whose positive hits are fusion-specific, and none of the hits is KMT2A::AFF1. Cheung N et al. 2016
   • No caveat anywhere in this graph anticipated that an 'MLL fusion' citation would resolve to a fusion that is not a KMT2A fusion at all. MOZ-TIF2 is a KAT6A fusion, and MLL-GAS7 is the partner that transforms without requiring Hoxa7 or Hoxa9 - so a dependency measured across this series is averaging over fusions with different downstream requirements, and cannot be read as a KMT2A-fusion dependency. EMBL-EBI BioStudies 2026
2. E2established confidence 5 of 5: Multiple independent lines of evidence; treated as settled in the field.database

   The SEM cell line, named in the SD70 sensitivity panel, is a childhood B-ALL line carrying t(4;11) with the KMT2A-AFF1 fusion, established from a 5-year-old female.

   Cellosaurus entry CVCL_0095 (SEM) · Bairoch A; SIB Swiss Institute of Bioinformatics · Cellosaurus 2026
   ▸What this evidence cannot show

   Cell-line provenance database, fetched and read. SEM is a relapse-derived line from a 5-year-old, not an infant line, so it is the right fusion in approximately the right disease but not the index population of this project. RS4;11 and MV4-11 are the other KMT2A::AFF1 lines used below; MV4-11 (CVCL_0064) is an acute monocytic leukemia line from a 10-year-old male carrying t(4;11)(q21;q23) with KMT2A-AFF1 and an additional FLT3-ITD, so it is a KMT2A::AFF1 line with a confounding driver.

   ▸Evidence that argues against this (1)

   • SEM's identity is not the variable that decides whether the SD70 result reproduces here. Ben-David et al. found extensive clonal diversity across 106 lines grown in two laboratories, rapid genetic diversification across 27 strains of one line, and - across 321 anti-cancer compounds - at least 75% of compounds that strongly inhibited some strains completely inactive in others, with barcoding showing culture-condition-sensitive positive selection as the cause. The SD70 sensitivity that makes this entry's KMT2A::AFF1 case is a viability panel in one laboratory's SEM. Before that panel is used to justify a programme, it needs repeating in an independently sourced SEM, and the entry should record that this has not been done. Ben-David U et al. 2018
3. E3strong confidence 4 of 5: Replicated in peer-reviewed work by more than one group.database

   KDM4C is not a DepMap dependency in any KMT2A::AFF1 line: SEM +0.009, SEMK2 +0.006, RS4;11 -0.157, MV4-11 +0.002, with a mean Chronos gene effect of -0.017 across 1,258 lines, 0.8% of lines depleted below -0.5, and -0.056 across 97 haematopoietic lines.

   DepMap CRISPR (Chronos) gene effect for KDM4C (ENSG00000107077), retrieved per cell line via the Open Targets Platform GraphQL API depMapEssentiality field · Broad Institute DepMap; Open Targets Platform · Open Targets 2026
   ▸What this evidence cannot show

   Queried by this agent this cycle; values are as returned and the API does not state the DepMap release, so the release is unverified. The haematopoietic subset is defined by a tissue-name match and is not a curated leukemia panel. Chronos measures proliferation in culture over roughly three weeks and is a poor detector of paralogue-buffered phenotypes, of differentiation phenotypes, and of anything requiring reconstitution stress — all three of which are in play here.

   ▸Evidence that argues against this (1)

   • A near-zero Chronos value cannot be read as 'not a target' when the gene has close paralogues, and the KDM4 family is the textbook case. Agger et al. could only expose the requirement in normal HSC using a conditional TRIPLE knockout of KDM4A/B/C, which is direct evidence that single-gene KDM4 loss is buffered in haematopoietic cells. The DepMap null is therefore consistent with a real family-level dependency and does not by itself refute the SD70 pharmacology. Agger K et al. 2019
4. E4strong confidence 4 of 5: Replicated in peer-reviewed work by more than one group.database

   KDM4C is not constrained against loss of function in humans: gnomAD reports pLI 3.8e-16 and LOEUF 0.754 from 81 observed against 129.3 expected loss-of-function variants, and unlike most candidates in this graph the estimate is adequately powered.

   gnomAD v4 loss-of-function constraint for KDM4C (ENSG00000107077) via the gnomAD GraphQL API · Genome Aggregation Database, Broad Institute · gnomAD 2026
   ▸What this evidence cannot show

   Queried by this agent this cycle. Constraint measures selection against lifelong heterozygous loss of one copy and is a weak model for months of partial pharmacological inhibition; it is also gene-specific and therefore silent about a family inhibitor, which is the only kind of KDM4 compound that has shown activity here. Under the power rule in ana-target-safety-triage-v2 (30 or more expected LoF is usable) this value is usable, but the asymmetry rule in the same entry says a HIGH LOEUF is the weaker direction to act on.

   ▸Evidence that argues against this (1)

   • An unconstrained gene is not a safe target, and the metric's authors are explicit about what it does measure. Karczewski et al. aggregated 125,748 exomes and 15,708 genomes, called 443,769 high-confidence predicted loss-of-function variants after filtering annotation and sequencing artefacts, and placed genes on a spectrum of tolerance to INACTIVATION validated against model-organism and cellular essentiality data. LOEUF 0.754 for KDM4C therefore says heterozygous loss of this one paralogue has not been selected against - which is exactly what paralogue redundancy predicts and exactly what the entry argues elsewhere. Constraint computed per gene is structurally blind to a family requirement, so reading it as reassurance about a family inhibitor such as SD70 gets the inference backwards. Karczewski KJ et al. 2020
5. E5strong confidence 4 of 5: Replicated in peer-reviewed work by more than one group.paper

   KDM4 activity is required for haematopoietic stem cell maintenance in vivo: conditional triple-knockout mice show loss of HSC maintenance, with H3K9me3 accumulating at transcription start sites and downregulation of genes essential for maintaining haematopoietic cells.

   The KDM4/JMJD2 histone demethylases are required for hematopoietic stem cell maintenance · Agger K, Nishimura K, Miyagi S, Messling JE, Rasmussen KD, Helin K · Blood 2019
   PMID 31434704DOI
   ▸What this evidence cannot show

   Abstract read, not full text. Mouse; the paper's framing is explicitly that it probes 'potential side effects of continuous inhibition' of KDM4 enzymes proposed as AML targets, so it is designed as a therapeutic-window objection and should be read as one. It measures the consequence of removing the whole family, which is the right comparator for a family inhibitor such as SD70 and the wrong comparator for a hypothetical KDM4C-selective agent that does not exist.

   ▸Evidence that argues against this (1)

   • The window may still be real if the leukemia requires more KDM4 activity than the stem cell does. Cheung et al. report that Kdm4c-knockdown HSPCs reconstituted mice across all lineages indistinguishably from controls at 6 weeks, and that Kdm4c-knockout mice are phenotypically normal, while the same knockdown abolished MLL-GAS7 leukemogenesis. A dose that spares the normal compartment is not excluded by the triple-knockout result, which removes all three paralogues completely and permanently. Cheung N et al. 2016
6. E6moderate confidence 3 of 5: One strong study, or several smaller consistent ones.paper

   A combinatorial drug screen identified the menin-MLL inhibitor MI-503 as a synergistic partner for the KDM4C inhibitor SD70: the combination synergistically enhanced apoptosis in MLL::AF9 leukaemia cells without affecting normal CD34+ cells, prolonged survival in AML xenograft models, and downregulated MYC target genes most significantly.

   Synergistic effects of the KDM4C inhibitor SD70 and the menin inhibitor MI-503 against MLL::AF9-driven acute myeloid leukaemia · Zhu W, Ding Y, et al. · British Journal of Haematology 2024
   PMID 38877874DOI
   ▸What this evidence cannot show

   Abstract read, not full text; the author list is abbreviated because the full list was not transcribed. MLL::AF9 AML only — no KMT2A::AFF1 arm, and infant B-ALL is not modelled. 'Without affecting normal CD34+ cells' is a short-term in vitro cytotoxicity statement and is not a reconstitution assay, so it does not speak to the stem-cell phenotype in E5. Synergy calls in combinatorial screens are sensitive to the reference model used and none is named in the abstract.

   ▸Evidence that argues against this (1)

   • A combination built on a menin inhibitor inherits that inhibitor's resistance problem, and the problem arrives earlier than exposure. Mahdavi et al. tested menin inhibition in patient-derived xenografts of KMT2A-rearranged leukaemias and found that serial samples from two patients with multiply relapsed ALL were much less sensitive than earlier samples from the same patients - resistance acquired with no menin-inhibitor exposure at any point - while transcriptomic analysis showed sustained on-target activity against the canonical menin-inhibitor targets in the resistant cells. Their samples included KMT2A::AFF1 ALL. Synergy demonstrated in treatment-naive MLL::AF9 models therefore says nothing about the relapsed, heavily pretreated population where a second agent would actually be deployed, and on-target pharmacodynamics will not report the failure. Mahdavi L et al. 2025
7. E7established confidence 5 of 5: Multiple independent lines of evidence; treated as settled in the field.database

   The accession used here is Q9H3R0, whose UniProt primary gene name reads back as KDM4C at 1,056 residues, with JmjN 16-58, JmjC 144-310 and tandem Tudor domains at 877-934 and 935-991.

   UniProtKB Q9H3R0 (KDM4C_HUMAN), lysine-specific demethylase 4C · UniProt Consortium · UniProt 2026
   ▸What this evidence cannot show

   Fetched and the primary gene name read back rather than assumed, per the standing accession rule; GASC1, JMJD2C and JHDM3C are listed as synonyms, not as the primary name. Domain boundaries are UniProt feature annotations, which are curated predictions anchored on structures rather than experimentally determined limits for this protein.

   ▸Evidence that argues against this (1)

   • NO PUBLISHED OPPOSITION to the accession itself, but the symbol choice has a measurable retrieval cost that bears on every counted-absence claim in this entry. PubMed searches run this session: KDM4C[tiab] = 178 records, JMJD2C[tiab] = 81, GASC1[tiab] = 30 (raw counts, not de-duplicated and not individually title-checked). The HGNC-approved symbol is the most productive here, which is the opposite of the septin case, but the two synonyms still carry roughly a hundred records between them, so any claim in this graph of the form 'KDM4C has never been tested in X' must be run over all three symbols before it is believed. What would falsify the accession item: a UniProt revision of the primary gene name or of the 1,056-aa canonical length, or an experimental structure placing the JmjN, JmjC or tandem Tudor boundaries outside the annotated ranges - these are curated feature predictions, not measured limits for this protein. No citation is attached because there is nothing to cite: this note is a bounded absence plus a falsifier list, resting on the UniProt record already cited by the parent item, not a sourced fact. Uncited and unciteable look identical in this field, so it is said here explicitly (Q42 citation-integrity repair, 2026-09-20).
8. E8moderate confidence 3 of 5: One strong study, or several smaller consistent ones.paper

   Nuclear localisation of JMJD2C/KDM4C is contributed to by the JmjN domain together with the PHD and Tudor domains, and the KDM4 paralogues are NOT interchangeable on this or any other axis measured: JMJD2A and C demethylate tri- and dimethylated H3K9 and H3K36 while JMJD2D - which lacks PHD and Tudor - demethylates tri-, di- and monomethylated H3K9 only, and JMJD2A and C hetero-oligomerise with each other while JMJD2D does not.

   Diversity within the JMJD2 histone demethylase family. · Shin S, Janknecht R · Biochem Biophys Res Commun 2007
   PMID 17207460DOI
   ▸What this evidence cannot show

   2007, transfected mammalian cells, epitope-tagged constructs - a domain-mapping study rather than an endogenous-protein localisation study, and overexpression can mask a weak localisation determinant. The reason it is recorded despite that is brief rule 62-65 discipline: it is the one source located that separates KDM4C from KDM4A, B and D on a localisation axis, and this graph's collision register flags exactly that confusion.

   ▸Evidence that argues against this (1)

   • A different and conflicting subnuclear placement is published: KDM3A and KDM4C drive condensin-mediated HETEROCHROMATIN reorganisation in mesenchymal stromal cell senescence and bone ageing. That is the opposite compartment to the active HOXA/MEIS1 chromatin where this graph's mechanism puts KDM4C. Both can be true of a protein in different cell types, but this graph asserts the active-chromatin placement without ever having measured it in a leukaemic cell, and the heterochromatin literature is the reason that assertion needs a citation it does not have. Huang B et al. 2019
9. E9strong confidence 4 of 5: Replicated in peer-reviewed work by more than one group.paper

   KDM4 demethylase activity is required for haematopoietic stem cell maintenance in vivo: conditional Kdm4a/Kdm4b/Kdm4c triple-knockout mice show loss of HSC maintenance, accumulation of H3K9me3 at transcription start sites and corresponding downregulation of several genes in HSCs, two of which - Taf1b and Nom1 - are themselves essential for maintaining haematopoietic cells. The study was explicitly framed as probing the side effects of continuous KDM4 inhibition in AML harbouring MLL translocations.

   The KDM4/JMJD2 histone demethylases are required for hematopoietic stem cell maintenance. · Agger K, Nishimura K, Miyagi S, Messling JE, Rasmussen KD, Helin K · Blood 2019
   PMID 31434704DOI
   ▸What this evidence cannot show

   Mouse conditional TRIPLE knockout - this is a family-level result and nothing in it attributes the HSC requirement to KDM4C alone, so it must not be quoted as 'KDM4C is required for HSC maintenance'. Normal haematopoiesis, not leukaemia. Read: abstract only.

   ▸Evidence that argues against this (1)

   • The therapeutic arm reports the opposite margin: the KDM4C inhibitor SD70 combined with the menin inhibitor MI-503 induced synergistic apoptosis in MLL::AF9 leukaemia cells WITHOUT affecting normal CD34-positive cells. A short in vitro exposure of human CD34-positive cells and a lifelong murine triple knockout are not the same experiment and the disagreement may be entirely about exposure - but a normal-cell-sparing claim from a cell-culture assay cannot answer a maintenance defect that only appears over time in vivo. Zhu W et al. 2024
10. E10strong confidence 4 of 5: Replicated in peer-reviewed work by more than one group.paper

    KDM4C maintains the AML leukaemia stem cell state through a chromatin-to-m6A axis: KDM4C regulates ALKBH5 expression by increasing chromatin accessibility at the ALKBH5 locus, reducing H3K9me3 and promoting recruitment of MYB and Pol II, and ALKBH5 is then required for maintaining LSC function while being dispensable for normal haematopoiesis, acting by m6A-dependent stabilisation of AXL mRNA.

    Leukemogenic Chromatin Alterations Promote AML Leukemia Stem Cells via a KDM4C-ALKBH5-AXL Signaling Axis. · Wang J, Li Y, Wang P, Han G, Zhang T, Chang J, Yin R, Shan Y, Wen J, Xie X, Feng M, Wang Q, Hu J, Cheng Y, Zhang T, Li Y, Gao Z, Guo C, Wang J, Liang J, Cui M, Gao K, Chai J, Liu W, Cheng H, Li L, Zhou F, Liu L, Luo Y, Li S, Zhang H · Cell Stem Cell 2020
    PMID 32402251DOI
    ▸What this evidence cannot show

    Human AML; the abstract does not state which fusions or genotypes the AML models carry, so this is not a KMT2A-rearranged-specific result. The causal chain is long - KDM4C to chromatin accessibility to ALKBH5 to m6A to AXL to LSC function - and the LSC-sparing-of-normal-haematopoiesis claim is made for ALKBH5, the downstream node, not for KDM4C. Read: abstract only.

    ▸Evidence that argues against this (1)

    • A genome-scale CRISPR screen finds no requirement for KDM4C in any leukaemia line and places it in the anti-selective half for this disease: 81.84th percentile of 17,728, -0.063 in 11 KMT2A-rearranged lines against -0.110 in 43 other leukaemia lines, pan-line mean -0.018. Paralogue redundancy is the likely explanation, and the same literature supports it - the normal-HSC phenotype required a triple knockout - but that reasoning also means the screen gives this entry no independent support at all. Oncofusion knowledge graph, analyses/depmap-kmt2a-selective-scan-full; DepMap Chronos via Open Targets; 11 KMT2A-rearranged versus 43 other leukaemia lines, 17,728 genes 2026
11. E11moderate confidence 3 of 5: One strong study, or several smaller consistent ones.paper

    In MLL::AF9-driven AML specifically, the KDM4C inhibitor SD70 combined with the menin inhibitor MI-503 induced synergistically enhanced apoptosis in leukaemia cells without affecting normal CD34-positive cells, significantly prolonged survival in AML xenograft models, and most strongly downregulated MYC target genes.

    Synergistic effects of the KDM4C inhibitor SD70 and the menin inhibitor MI-503 against MLL::AF9-driven acute myeloid leukaemia. · Zhu W, Ding Y, Huang W, Guo N, Ren Q, Wang N, Ma X · Br J Haematol 2024
    PMID 38877874DOI
    ▸What this evidence cannot show

    MLL::AF9, i.e. KMT2A::MLLT3 in myeloid disease - not KMT2A::AFF1, not B-lineage and not infant material. A combinatorial drug screen followed by a two-agent combination, so the KDM4C-alone contribution to the phenotype is bounded only by the single-agent arm reported in the paper and not resolvable from the abstract. SD70 selectivity across the KDM4 family is not addressed here. Read: abstract only.

    ▸Evidence that argues against this (1)

    • Sparing of normal CD34-positive cells in vitro sits directly against the family-level in vivo result that KDM4 activity is required for HSC maintenance, in a study designed to ask exactly that question about continuous KDM4 inhibition in MLL-translocated AML. Short exposure in culture is the weaker instrument for a maintenance phenotype. Agger K et al. 2019

• ← analogous toKAT6A / MOZ — the acetyltransferase that is a fusion oncogene in its own right
• ← related toParalogue redundancy does NOT explain anti-selective DepMap results here: clearly selective genes carry paralogues at 30.6% against a 32.7% base rate, with no depletion

  THE OTHER: its HSC requirement is a Kdm4a/b/c triple knockout. Same untested explanation.

• ← related toH3K9me3 - measured in KMT2A::AFF1 cells, in a supplementary figure, and it rises GENERALLY where H3K27ac falls SPECIFICALLY in the same experiment

  KDM4-family demethylases erase H3K9me2/3; none assayed at fusion-bound loci

• regulates →H3K9me2 - named sixty times in this graph with no entry of its own, the mark that defines the lamina-associated domains the fusion's targets sit beside, and the only H3K9 methylation state ever measured at a KMT2A fusion's own binding sites

  KDM4 family erases me2 and me3, making me2 the only H3K9 state with two eraser routes

• ← testsThe deposit this graph filed as 'four SUV39H1 ChIP-seqs' is actually 77 samples in MCF-7 carrying the ENTIRE H3K9 axis in one cell type and one pipeline - writer SUV39H1, writer G9a/EHMT2, reader HP1-alpha/CBX5 (wild-type AND two phospho-mutants), co-repressor KAP1/TRIM28, erasers KDM4B and KDM4C, the mark H3K9me3, and an shKDM4B+shKDM4C eraser-knockdown arm

  KDM4C occupancy and a double KDM4B/KDM4C knockdown H3K9me3 arm

• ← related toKDM4B (JMJD2B) - the weakest of the four H3K9me3 demethylases, one of the two erasers knocked down in the only KDM4-perturbation arm in any archive that touches this mark

  paralogue; the two are knocked down TOGETHER in the only public H3K9me3-eraser perturbation, so neither can be attributed an effect separately

• ← testsRUN, with its positive control: knocking down KDM4B and KDM4C together raises H3K9me3 at the +1 nucleosome of 2,000 active promoters by 35.7 percentile points, moves the KMT2A promoter's +1 nucleosome by the same 35.3, and moves the 90 kb KMT2A gene body 5.05 points the WRONG way - so the methyl-poor domain is not an erasure product

  the other eraser removed; removing both does not move H3K9me3 at the KMT2A locus, while it raises it 35.7 percentile points at active-promoter +1 nucleosomes

• ← related toRUN: KDM4B and KDM4C both read the 76.6th-79.7th percentile of MCF-7 CTCF anchors over the whole 90 kb KMT2A gene body - above the 25-track panel median of 61.39, LOCUS-WIDE ELEVATED - and both are ORDINARY at the exon-11 anchor with replicate spreads of 92.3 and 61.5 percentile points

  first KDM4C occupancy measurement at the KMT2A locus: 76.62nd percentile median over the 90 kb, ORDINARY at the anchor

• related to →Chromocentres and pericentromeric satellite heterochromatin — no KMT2A evidence, but the AT-hook argument says there should be

  KDM4C's only published subnuclear placement is heterochromatin, which conflicts with this graph's active-target-gene model

• depends on →Bone marrow haematopoietic stem cell — the postnatal comparator, and why the same fusion behaves differently here

  KDM4 activity is required for HSC maintenance in vivo - but only shown in a Kdm4a/b/c TRIPLE knockout, so this is family-level and not attributable to KDM4C; a toxicity result for continuous KDM4 inhibition

• regulates →The stem-like self-renewal state — a limited stem-cell module grafted onto a committed cell

  KDM4C opens the ALKBH5 locus by lowering H3K9me3, and ALKBH5 maintains AML LSC function via m6A-dependent AXL stabilisation; human AML, fusion context unstated

• related to →Revumenib (SNDX-5613, Revuforj) — first approved menin–KMT2A interface inhibitor

  the SD70 + MI-503 synergy argument inherits menin-inhibitor resistance that arises without menin-inhibitor exposure in pretreated KMT2A::AFF1 samples (Mahdavi 2025)

• binds →KMT2A::AFF1 (MLL-AF4), t(4;11)(q21;q23)

  recruited via the MLL 5' moiety retained in every KMT2A fusion

• binds →KMT2A / MLL1 — the gene that breaks

  interaction domain mapped by deletion to the KMT2A 5' region

• regulates →HOXA9 — required for survival of established MLL-rearranged leukemia, and dispensable for starting it

  removes repressive H3K9me3 at HOXA9; knockdown restores the mark and silences the gene

• ← co-occurs withPRMT1 — an MLL-fusion cofactor whose first-in-human inhibitor was terminated for lack of efficacy and toxicity

  co-recruited two-enzyme circuit; PRMT1 necessary but not sufficient without KDM4C

• related to →MEN1 / menin — the adaptor the fusion cannot work without

  KDM4C inhibitor SD70 synergises with the menin inhibitor MI-503 in MLL::AF9 AML

• ← related toTarget safety triage v2: constraint with its uncertainty attached (run)

  KDM4C is the rare candidate with well-powered permissive constraint (LOEUF 0.754, 129.3 expected LoF)

• ← proposed byAuditing the seven strikes before applying them: nine of ten decisions stand, four for different reasons than were given, and one clinical argument fails

  the audit's full-text read of PMID 26766589 is what established the KMT2A::AFF1 evidence for KDM4C

• ← testsThe dependencies this project keeps failing to see are the buffered ones: paralogue redundancy and guide placement, not absence of biology

  explains why KDM4C scores zero in four KMT2A::AFF1 lines while a KDM4 inhibitor kills one of them

• ← testsOne experiment that decides whether KDM4C is a target or the KDM4 family is: paralogue rescue of SD70 sensitivity in SEM

  the gate KDM4C must clear before anything is spent on it; a null result strikes the entry

• ← related toHalf the KDM4C deposition is not a KMT2A fusion at all: MOZ-TIF2 8, MLL-GAS7 6, MLL-AF9 2, and the human xenograft deposited nothing

  the deposition is half MOZ-TIF2 and its KMT2A half is dominated by MLL-GAS7