Oncofusion

start / graph / gene-myc

Genemoderate confidence 3 of 5: One strong study, or several smaller consistent ones.

MYC - named in a hundred entries of this graph and, until now, written in none

also called c-Myc · MYC proto-oncogene · BHLHE39 · P01106

MYC is one of the most famous cancer genes there is: a master switch that tells a cell to grow and divide, and it sits downstream of the broken KMT2A machinery this graph is about. It has been mentioned in around a hundred entries here without ever getting one of its own, and the reason it stayed invisible is embarrassing - the tool meant to find such gaps was ranking a web address and a survival statistic above it.

The same thing, for a scientist

MYC (P01106, 454 aa) is a downstream effector of the KMT2A-fusion transcriptional programme rather than a direct fusion partner. Two recent primary results anchor it here: a CRISPR-tiling screen in an MLL-r AML model shows the **SETD1B SET domain** sustains H3K4me3 breadth and MYC expression, and **exogenous MYC rescues the growth defect** of SETD1B SET-domain loss - genetic epistasis placing MYC as the essential effector; and **HDAC8** is upregulated in KMT2A-r AML, where its inhibition downregulates a **STAT3-MYC axis independent of TP53 status** and synergises with venetoclax. **This entry is deliberately narrow**: MYC's general oncology is not this graph's business, and only KMT2A-relevant evidence is carried.

Go deeper
Why this entry did not exist

MYC is named in 100 node files. It became the top of the named-but-unwritten queue only on 2026-09-20, when data/symbol-collisions.json blocked three artefacts that had been ranked above it since the queue was built: REST (217 files, the web API), EFS (136, event-free survival) and PDF (17, the file format). All three are genuine HGNC symbols.

This is the sixth instance of the project's recurring shape - the mechanism gets written and the proteins named inside it do not - and the first where the instrument built to catch that shape was itself the reason the gap persisted.

What the primary literature supports, and what it does not

MYC is downstream, not a partner. No KMT2A::MYC fusion is claimed here and none is implied; MYC's relevance is as an effector of the fusion's transcriptional output.

The cleanest result is an epistasis one 1. A CRISPR-tiling screen against known H3K4 methylation modifiers in an MLL-r AML model found that disrupting the SETD1B catalytic SET domain depleted FLT3-ITD or NRAS-G12D-expressing AML cells and downregulated the MYC pathway, with a significant decrease in H3K4me3 breadth. Critically, exogenous MYC expression restored the growth defect - so MYC is not merely correlated with the programme, it is the limiting output of it in that system. Disrupting the H3K4 demethylase KDM5C also rescued, which is a coherent second route to the same mark.

A therapeutic axis 2. HDAC8 is upregulated in KMT2A-r AML and high HDAC8 associates with poor overall survival. In a KMT2A::MLLT3 mouse model, genetic knockout and pharmacological inhibition both delayed progression and prolonged survival; mechanistically HDAC8 binds STAT3 directly, promoting its deacetylation and stabilisation, and HDAC8 inhibition downregulates the STAT3-MYC axis independent of TP53 status. It synergises with venetoclax.

What this entry does NOT claim. It does not claim MYC is a direct KMT2A-fusion target gene in the sense HOXA9 and MEIS1 are - that is a separate question this entry has not established, and the two results here reach MYC through SETD1B and through STAT3 respectively, which are different routes in different systems. One of them is a mouse KMT2A::MLLT3 model and the other an MLL-r AML model driven by FLT3-ITD or NRAS-G12D, so neither is the index infant KMT2A::AFF1 B-lineage disease.

Identifier note. gene_exact:MYC returns exactly one reviewed human record, P01106, 454 aa, sole synonym BHLHE39 - unusually clean for this graph. The trap here is the opposite of rule 66's: not ambiguity, but the fact that MYC is so ubiquitous that a co-occurrence search returns 333,979 records, so relevance cannot be inferred from a hit count and every claim must come from a read paper.

Facts and figures
uniprotP01106
length AA454
gene primaryMYC
named in entries when written100
gene symbolMYC
gene symbol notepromoted from gene_primary 2026-09-20: verify-accessions.py reads data.gene_symbol, so a gene_primary-only entry was being length-checked or id-inferred rather than symbol-asserted
subnuclear localizationnucleoplasm plus a distinct NUCLEOLAR pool that drives RNA polymerase I on rDNA, independently of Pol II output (PMID 15723053). This is the one place in the graph where a fusion-relevant protein and a KMT2A fusion have been assigned the SAME subnuclear compartment by independent literatures - localization-nucleolus-kmt2a-fusion holds the only published images of a KMT2A fusion inside a compartment - and nobody has co-stained them.
shown for fusionNO. MYC has never been localised in a KMT2A-rearranged cell. Measured 2026-09-21 with NCBI esearch: (MYC[tiab] OR c-Myc[tiab]) AND (MLL[tiab] OR KMT2A[tiab] OR MLL-AF4[tiab] OR MLL-AF9[tiab] OR mixed lineage leukemia[tiab]) AND (localiz*[tiab] OR localis*[tiab] OR subcellular[tiab] OR nuclear body[tiab] OR nuclear bodies[tiab] OR speckle*[tiab] OR condensat*[tiab] OR nucleolar[tiab] OR immunofluorescence[tiab] OR compartment*[tiab]) returns 11 records, ALL title-checked off-topic for MYC localisation - they are menin/c-Myc transcription, WDR5 targeting, Musashi2, MOZ, ribavirin/eIF4E, LEDGF interactome, expression profiling and two case reports. The eIF4E paper (PMID 30478448) is the near miss: it concerns eIF4E-dependent nuclear export of transcripts in infant ALL, which is a localisation claim about mRNA and not about MYC protein. Operands: gene clause 51,560, gene AND fusion 248, so the intersection is large and the zero is specifically at the protein-localisation step. Positive control through the identical shape with (MLLT1[tiab] OR ENL[tiab]) returns 23. Zero in titles and abstracts; supporting figures not searched (brief rule 80).
localization
{
  "subcellular": "Nucleus, with an experimentally annotated cytoplasmic pool. UniProt P01106 lists Nucleus nucleoplasm, Nucleus nucleolus, Nucleus, Cytoplasm and Chromosome, all at ECO:0000269. For a transcription factor the informative split is not nucleus-versus-cytoplasm but nucleoplasm-versus-nucleolus, because the two pools drive different polymerases.",
  "subnuclear": "NUCLEOPLASM AND NUCLEOLUS, and the nucleolar pool is a separate job. c-Myc and Max interact in nucleoli and associate with ribosomal DNA; the association rises when quiescent cells are activated and is followed by TRRAP recruitment, histone acetylation, Pol I recruitment and rDNA transcription - and c-Myc activates Pol I transcription even when Pol II transcription is blocked. So MYC in the nucleolus is doing ribosome biogenesis, not the Pol II programme the menin-MLL axis regulates.",
  "how_shown": "immunofluorescence under proteasome inhibition, in situ proximity ligation for the c-Myc/Max interaction in nucleoli, ChIP at rDNA, and a MycER system separating Pol I from Pol II output (PMID 15723053); UniProt P01106 SUBCELLULAR LOCATION comment.",
  "systems": "cultured human non-haematopoietic cells; no leukaemia and no KMT2A-rearranged cell has ever been used to localise MYC",
  "node": "localization-nucleolus-kmt2a-fusion"
}
cell type
{
  "expressed_in": "Deregulated or amplified in most AMLs [E_q66_cell1]. No cell-type-resolved expression measurement within normal human haematopoiesis is recorded in this entry, and the graph should not infer one.",
  "required_in": "Leukaemia stem cells of murine MLL-AF9 AML, and in a specific sense: the Miz1-interaction-defective MycV394D mutant reduces LSC frequency on serial transplantation and reduces leukaemogenic capacity in vivo [E_q66_cell1]. Separately, in a KMT2A-rearranged AML setting HDAC8 inhibition targets a STAT3-MYC axis and synergises with venetoclax [E2], and disrupting the SETD1B catalytic SET domain in an MLL-rearranged AML model depletes FLT3-ITD or NRAS-G12D leukaemic cells with an effect on MYC expression and H3K4me3 breadth [E1].",
  "not_required_in": "Nothing. The relevant caution is the opposite one - almost all AML cells depend on Myc for proliferation and survival [E_q66_cell1], so there is no cell type in this disease where MYC is dispensable, and therefore no selectivity.",
  "how_shown": "separation-of-function point mutant (MycV394D) with intrinsic Myc deleted, colony formation, serial transplantation and LSC frequency in murine MLL-AF9 AML (PMID 32040550)",
  "nodes": [
    "cell-type-gmp",
    "cell-type-hsc-bone-marrow"
  ]
}
cell state
{
  "maintains_state": "SELF-RENEWAL AND THE UNDIFFERENTIATED STATE, by repression rather than activation. In murine MLL-AF9 AML, Myc represses Miz1-mediated expression of Cebpa and Cebpd, and the transrepression-dead mutant yields partially DIFFERENTIATED AML cells with reduced self-renewal [E_q66_cell1]. The mechanism matters for druggability: a Myc-Max-directed agent would not touch the arm this experiment shows to be required.",
  "regulates_state": "Ribosome-biogenesis capacity, through the nucleolar pool: c-Myc drives Pol I transcription of rDNA independently of Pol II output [E_q66_loc1]. That is a growth-state lever sitting in a different compartment from the transcriptional one, and it is untested in this disease.",
  "expressed_in_state": "not established in KMT2A-rearranged disease. No measurement of MYC across quiescent, cycling, diagnosis-versus-relapse or menin-inhibitor-treated states of a KMT2A-rearranged blast is recorded here.",
  "how_shown": "MycV394D serial transplantation and differentiation marker readouts in murine MLL-AF9 AML (PMID 32040550); rDNA ChIP and MycER Pol I activation (PMID 15723053)",
  "nodes": [
    "cell-state-stem-like-self-renewal",
    "cell-state-differentiation-block"
  ]
}
How this could bite us back
  • HDAC8 inhibition plus venetoclax is a mouse result in a KMT2A::MLLT3 model. Venetoclax already carries real tumour-lysis and myelosuppression risk in children, and adding a second epigenetic agent on the strength of a preclinical synergy claim is exactly the step this project has repeatedly found to be under-evidenced.
  • MYC is not druggable directly in any approved sense. Nothing in this entry should be read as proposing a MYC-targeted intervention in infants; the two tractable handles described here are SETD1B and HDAC8, and both are preclinical.
  • MYC is the single most over-claimed gene in oncology. Its presence in a hundred entries of this graph is not evidence that it is important here - it is what a ubiquitous gene looks like in any corpus, and this entry is deliberately narrow for that reason.
  • Essential-function toxicity is the whole story with MYC and it is stated inside the supporting evidence: almost all AML cells depend on Myc for proliferation and survival (PMID 32040550), and so does normal proliferating tissue. There is no KMT2A-selective window in any experiment cited in this entry, and a differential would have to be demonstrated against a non-KMT2A comparator arm rather than against a Myc-null control.
  • Arm mismatch as a resistance route: the LSC requirement demonstrated here is the Myc-MIZ1 TRANSREPRESSION arm, not the Myc-MAX transactivation arm (PMID 32040550). Most MYC-directed chemical matter targets the Max dimerisation interface or MYC abundance, so an agent could suppress transactivation fully and leave the arm that maintains self-renewal intact.
  • Two compartments, one target: the nucleolar MYC pool drives Pol I on rDNA even when Pol II transcription is blocked (PMID 15723053). A drug that shuts down the Pol II programme leaves the ribosome-biogenesis arm running, and no MYC-directed programme in this disease measures rDNA output as a pharmacodynamic readout.
  • Do not import the condensate rationale. The rigorous phase-separation work in this family is on N-myc/MYCN, not c-MYC, and it found over 97 percent of N-myc-regulated genes unaffected by phase separation, with soluble complexes sufficient for activation (PMID 38811792). A condensate-disrupting strategy justified by 'MYC forms condensates' would be justified by the wrong protein and by the minority of its effect.
Still open
  1. Is MYC a direct KMT2A-fusion target in the sense HOXA9 and MEIS1 are, or only an indirect output? The two results here reach it by different routes in different systems.
  2. Has anything been measured about MYC in infant KMT2A::AFF1 B-lineage disease specifically, as opposed to MLL-r AML models?
  3. Does menin inhibition move MYC? 231 records name both a menin inhibitor and MYC and none was read for this entry.
  4. Does MYC occupy the nucleolus in a KMT2A-rearranged blast, and does it share that compartment with the fusion? localization-nucleolus-kmt2a-fusion holds the only published images of a KMT2A fusion inside any compartment, and MYC has an independently established nucleolar pool. A single co-stain in an index line would either create the first shared-compartment observation in this graph or rule it out.
  5. Is the Myc-MIZ1 transrepression requirement reproducible in a human B-lineage KMT2A::AFF1 system? The result is murine MLL-AF9 and the index disease is neither murine nor myeloid.
  6. Does menin inhibition change the nucleoplasmic-to-nucleolar ratio of MYC? Menin inhibitors lower MYC, but no one has asked whether they lower both pools or only the Pol II-associated one.
Every source behind this entry (6)
  1. E1strong confidence 4 of 5: Replicated in peer-reviewed work by more than one group.paper

    In an MLL-rearranged AML model, disrupting the SETD1B catalytic SET domain depleted FLT3-ITD or NRAS-G12D-expressing AML cells, downregulated gene expression particularly in the MYC pathway, and significantly decreased H3K4me3 breadth; exogenous MYC expression, or disruption of the H3K4 demethylase KDM5C, significantly restored the growth defect caused by SET-domain loss.

    PMID 40341256DOI
    What this evidence cannot show

    Abstract fetched and read back on 2026-09-20; full text not read. **Strong as a result within the reported system, which is an MLL-r AML model driven by FLT3-ITD or NRAS G12D - not the index infant KMT2A::AFF1 B-lineage disease**, and the specific KMT2A fusion is not named in the abstract. The rescue is the strong part: it establishes MYC as a limiting output rather than a correlate. 'H3K4me3 breadth' is a measure whose biological interpretation is itself contested.

    Evidence that argues against this (1)
    • A rescue by exogenous MYC shows MYC is sufficient to restore growth in that assay; it does not show that the SETD1B effect is mediated solely through MYC, since forced expression of a potent growth driver can compensate for defects it had no part in causing. The KDM5C result is the better evidence that the mark, not the gene, is the axis. Zhang L et al. 2026
  2. E2moderate confidence 3 of 5: One strong study, or several smaller consistent ones.paper

    HDAC8 is upregulated in KMT2A-rearranged AML and high HDAC8 is associated with poor overall survival; in a KMT2A::MLLT3 mouse model both genetic knockout and pharmacological inhibition of HDAC8 delayed leukaemia progression, prolonged survival and reduced recurrence, and HDAC8 binds STAT3 directly, promoting its deacetylation and stabilisation, so that HDAC8 inhibition downregulates the STAT3-MYC axis independent of TP53 status and synergises with venetoclax.

    PMID 42014858DOI
    What this evidence cannot show

    Abstract fetched and read back on 2026-09-20; full text not read. Moderate rather than strong: the survival association is correlative, the causal work is a **mouse KMT2A::MLLT3 model** rather than the index B-lineage fusion, and 'synergises with venetoclax' is a claim this graph has seen fail to survive formal synergy analysis elsewhere. **HDAC8-selective inhibitors have not been tested in infants**, and this project has separately recorded that the HDAC erasers, not the writers, carry the infant evidence.

    Evidence that argues against this (1)
    • This graph's acetylation beat found that HDAC-directed claims in KMT2A-r frequently rest on tool compounds whose selectivity is weaker than the paper's naming implies - the same problem logged for GSK-J4 and for CX-4945. An 'HDAC8-selective inhibitor' phenotype attributed to HDAC8 is naming until the genetic knockout and the compound are shown to phenocopy in the same cells, which this abstract does assert but does not quantify. Izumi S et al. 2025
  3. E3established confidence 5 of 5: Multiple independent lines of evidence; treated as settled in the field.database

    The reviewed human UniProt record for MYC is P01106, 454 amino acids, primary gene name MYC with the sole listed synonym BHLHE39, and gene_exact:MYC returns exactly one reviewed human record.

    What this evidence cannot show

    Fetched from the UniProt REST API on 2026-09-20 with gene_primary and length read back. Established for the database state on that date. Recorded because this graph asserts accession and length in gene entries specifically so that `verify-accessions.py` fails a swap - and because MYC, unlike the RAS family or KMT2B/KMT2D, is unambiguous, which is worth stating rather than assuming.

    Evidence that argues against this (1)
    • Cleanliness at the symbol level does not make MYC claims easy to verify: MYC returns 333,979 records in a bare literature search, so a co-occurrence hit count carries essentially no information about relevance to KMT2A-rearranged disease and cannot be used to bound an absence here. Izumi S et al. 2025
  4. E_q66_loc1strong confidence 4 of 5: Replicated in peer-reviewed work by more than one group.paper

    c-MYC has a functional NUCLEOLAR pool, not merely a nucleoplasmic one: proteasome inhibition causes c-Myc to accumulate within nucleoli, c-Myc and Max interact in nucleoli and are associated with ribosomal DNA, the association increases on activation of quiescent cells and is followed by recruitment of TRRAP, enhanced histone acetylation, recruitment of RNA polymerase I and activation of rDNA transcription - and c-Myc can activate Pol I transcription in the absence of Pol II transcription.

    c-Myc associates with ribosomal DNA and activates RNA polymerase I transcription. · Arabi A; Wu S; Ridderstråle K; Bierhoff H; Shiue C; Fatyol K; Fahlén S; Hydbring P; Söderberg O; Grummt I; Larsson LG; Wright AP · Nat Cell Biol 2005
    PMID 15723053DOI
    What this evidence cannot show

    Cultured human cells, not haematopoietic and not leukaemic; the quiescent-to-activated transition is a serum-stimulation model, not a stem-cell state. Part of the nucleolar accumulation is seen UNDER PROTEASOME INHIBITION, which is a perturbation, so the resting nucleolar fraction is not quantified by this work. The rDNA association is measured by ChIP and by in situ proximity ligation, which report co-occupancy rather than a compartment boundary. Read: abstract only. Identifiers copied field by field from the efetch record fetched 2026-09-21, ArticleId anchored per brief rule 81.

    Evidence that argues against this (1)
    • The only rigorous phase-separation study in this family is about the WRONG FAMILY MEMBER and it argues AGAINST condensates mattering much. It characterises N-myc (MYCN), not c-MYC, and concludes that only a small percentage - under 3 percent - of N-myc-regulated genes is further modulated by phase separation, with over 97 percent unaffected, demonstrating that soluble complexes of the transcription factor with the transcriptional machinery are sufficient to activate transcription. Any statement that 'MYC forms transcriptional condensates' in this graph has to name which MYC and has to carry that 97 percent. Yang J; Chung CI; Koach J; Liu H; Navalkar A; He H; Ma Z; Zhao Q; Yang X; He L; Mittag T; Shen Y; Weiss WA; Shu X 2024
  5. E_q66_cell1strong confidence 4 of 5: Replicated in peer-reviewed work by more than one group.paper

    In a murine MLL-AF9 AML model, Myc maintains the leukaemia stem cell through TRANSREPRESSION rather than transactivation: AML cells expressing the Miz1-interaction-defective MycV394D mutant, with intrinsic Myc deleted, are PARTIALLY DIFFERENTIATED and show reduced colony formation in vitro and reduced leukaemogenic capacity in vivo, with a reduced frequency of leukaemia stem cells on serial transplantation, and Myc represses Miz1-mediated expression of Cebpa and Cebpd to maintain the undifferentiated state and the self-renewal capacity of LSCs.

    Myc-Miz1 signaling promotes self-renewal of leukemia stem cells by repressing Cebpα and Cebpδ. · Zhang L; Li J; Xu H; Shao X; Fu L; Hou Y; Hao C; Li W; Joshi K; Wei W; Xu Y; Zhang F; Dai S; Breslin P; Zhang J; Zhang J · Blood 2020
    PMID 32040550DOI
    What this evidence cannot show

    Murine MLL-AF9 myeloid leukaemia, not human and not B-lineage: this is precisely the generalisation brief rule 73 names as this project's commonest synthesis failure, so the claim is written with the system in its first clause. MycV394D is a separation-of-function point mutant and the result is that the transrepression arm is required; it does not show that the transactivation arm is dispensable. MycV394D-AML cells were also more chemosensitive, which is a drug-response readout mixed into a state readout. Read: abstract only.

    Evidence that argues against this (1)
    • The same paper opens by stating that almost all AML cells are dependent upon Myc for proliferation and survival, which is the problem with MYC as a target in this disease rather than the argument for it: a dependency present in almost every AML, and in normal proliferating tissue, gives no KMT2A-selective window. This graph's own DepMap practice would call that floor compression, and the LSC-frequency result does not escape it because it is measured against Myc-expressing controls rather than against a non-KMT2A comparator arm. Oncofusion knowledge graph standing methodology, docs/QUEEN_BRIEF.md; DepMap Chronos 2026
  6. E_q66_depmapstrong confidence 4 of 5: Replicated in peer-reviewed work by more than one group.dataset

    MYC's apparent KMT2A-selectivity is a floor-compression artefact and must not be quoted as selectivity: on this project's genome-scale scan MYC ranks 168 of 17,728 (0.9th percentile) with a difference of -0.241 - but the underlying means are -2.683 in the 11 KMT2A-rearranged lines and -2.442 in 43 other leukaemia lines. Both arms are among the most deeply depleted genes in the screen, so there is almost no dynamic range left in which selectivity could appear.

    Without a prior, which genes are most selectively required in KMT2A-rearranged leukemia lines? Genome-scale scan, 11 KMT2A-rearranged versus 43 other leukaemia lines, 17,728 genes · Oncofusion knowledge graph, analyses/depmap-kmt2a-selective-scan-full/results.json; DepMap Chronos via Open Targets 2026
    What this evidence cannot show

    This is the HCFC1/KAT8 case that the standing brief names, at the opposite end of the effect scale: there the arms sat at the floor with a tiny difference, here they sit far past it with a larger one. The index arm is myeloid-enriched and this figure is not lineage-split, so the -0.241 mixes fusion status with lineage. A difference of -0.241 between two arms at -2.7 and -2.4 is a difference between 'essential' and 'essential'.

    Evidence that argues against this (1)
    • The rank is nonetheless real and worth keeping in view: 168 of 17,728 is inside the top 1 percent of the same scan that puts MEN1 first, and a dismissal on floor-compression grounds alone would be as lazy as an endorsement on rank alone. The honest reading is that the screen cannot discriminate here, in either direction, and the question has to be settled by a system with a differentiation or self-renewal readout rather than by a fitness screen - which is what the murine MLL-AF9 MycV394D experiment in this entry provides. Zhang L; Li J; Xu H; Shao X; Fu L; Hou Y; Hao C; Li W; Joshi K; Wei W; Xu Y; Zhang F; Dai S; Breslin P; Zhang J; Zhang J 2020
Connected to