RUNX1 — activated by KMT2A::AFF1 and then bound by the reciprocal fusion, in the one published two-oncoprotein circuit in t(4;11)
also called RUNX1 · AML1 · CBFA2 · Runt-related transcription factor 1 · PEBP2alphaB
The KMT2A fusion switches the RUNX1 gene on, and the RUNX1 protein that results then sticks to the other, mirror-image fusion protein made by the same translocation. It is the only published case where the two halves of the t(4;11) break work on each other through a shared target.
The same thing, for a scientist
RUNX1 (Q01196, 453 aa) is directly activated by MLL-AF4/KMT2A::AFF1, and the RUNX1 protein interacts with the product of the reciprocal AF4-MLL translocation, giving a circuit in which one fusion protein induces a gene whose product then modulates the other fusion protein. The interaction is therefore **not** a contact on KMT2A::AFF1 itself — it is a contact on the der(4) product — while the *transcriptional* relationship to KMT2A::AFF1 is direct and well established. RUNX1 is separately reported as a core component of the enhancer programme that KMT2A fusions target in 11q23 acute leukemia. The DepMap data qualify the picture in a direction the literature does not lead you to expect: KMT2A-rearranged lines are **less** RUNX1-dependent than other leukemia lines (mean -0.552 vs -0.741, difference **+0.189**), while both groups are more dependent than cancer lines at large (-0.013).
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.
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.
The circuit as published
The mechanism reported is two-step 1:
- RUNX1 is directly activated by MLL-AF4 (KMT2A::AFF1).
- The RUNX1 protein interacts with the AF4-MLL reciprocal fusion protein.
The authors frame this as cooperation between two oncogenic fusion proteins: one activates a target gene, the other modulates the function of its product. That is a genuinely distinctive claim — most fusion-partner biology is a single protein with a single complex — and it is the mechanistic reason t(4;11) is sometimes argued to be a two-oncoprotein disease rather than a one-oncoprotein disease.
Method: ChIP for direct target status, plus protein-interaction assays for the AF4-MLL contact, in human t(4;11) cell lines and primary material 1. Shown for: KMT2A::AFF1 for the transcriptional half; the reciprocal AFF1::KMT2A for the protein-interaction half. Which part of the fusion: the interaction reported is with AF4-MLL, so the contact is not on the der(11) product this project targets.
RUNX1 as part of the enhancer programme, which is a different claim
Separately from the interaction, KMT2A fusions in 11q23 acute leukemia bind a distinct enhancer repertoire and target the RUNX1 programme, with MLL-AF9 and MLL-AF4 binding overlapping but non-identical enhancer sets 2. That is a statement about where the fusion goes on chromatin and which transcription-factor network it co-opts, not about physical contact, and it should be kept separate in the reader's mind from 1.
Taken together the two give RUNX1 three distinct relationships to this biology — target gene, enhancer-network partner, and reciprocal-fusion binder — which is why it recurs in the t(4;11) literature so often and why the graph needs one entry that separates them rather than a sentence that runs them together.
The dependency number points the other way, and should be stated
If RUNX1 were a central KMT2A::AFF1 dependency, KMT2A-rearranged lines should need it more than other leukemia lines. They need it less: mean gene effect -0.552 in 11 KMT2A-rearranged lines against -0.741 in 43 other leukemia lines, a difference of +0.189 3. Both are well below the all-lines mean of -0.013, so RUNX1 is a leukemia dependency — it is just not a KMT2A-selective one, and the sign is positive.
Three readings are available and this entry does not choose between them. (i) The RUNX1 requirement is a general leukemia requirement that KMT2A-rearranged cells partly bypass. (ii) Eleven lines is too few and the difference is noise. (iii) RUNX1's role in t(4;11) runs through the reciprocal product, which most DepMap KMT2A-rearranged lines — dominated by AML fusions such as KMT2A::MLLT3 — do not have in the same form.
Reading (iii) is the one that would preserve the published model, and it is testable: restrict the comparison to t(4;11) lines only. That comparison is not made here because the number of KMT2A::AFF1 lines in DepMap is very small.
| uniprot | Q01196 |
|---|---|
| gene symbol | RUNX1 |
| length AA | 453 |
| protein name | Runt-related transcription factor 1 |
| alphafold ID | AF-Q01196-F1 |
| accession resolution | UniProt query gene_exact:RUNX1 AND organism_id:9606 AND reviewed:true; primary gene name read back as RUNX1 and length 453 confirmed (the canonical isoform; RUNX1 has multiple isoforms of differing length) |
| interaction evidence class | ChIP for direct target status; protein-interaction assays for the AF4-MLL contact |
| shown for fusion | yes for KMT2A::AFF1 transcriptionally; the protein interaction is with the reciprocal product |
| contact on kmt2a aff1 | none reported |
| druggability | RUNX1 is a transcription factor and a germline tumour-suppressor gene (familial platelet disorder with predisposition to AML). Not a straightforward target, and the direction of the dependency here argues against it for this subtype. |
| domain source | UniProt Q01196 feature annotation (Runt domain 50-178; KAT6A-interaction region 291-371) and Zeng 1997 PNAS PMID 9192636 for the NMTS at 351-381 (PMC full text read for the residue numbers) |
| subnuclear localization | NUCLEAR MATRIX, and this is one of the best-defined subnuclear addresses in the whole graph. The transcriptionally ACTIVE isoform AML-1B binds the nuclear matrix and the inactive AML-1 does not; association is independent of DNA binding and requires a 31-amino-acid nuclear matrix targeting signal at residues 351-381, distinct from the NLS, which is necessary and sufficient to direct a heterologous GAL4 fusion to the matrix - see this entry's existing evidence item E4 (PMID 9192636). Loss of the C-terminal domain carrying the NMTS is a frequent consequence of t(8;21) and t(3;21). [EXPLICIT NEGATIVE] NOT MEASURED IN A KMT2A-REARRANGED CELL. The NMTS work is in non-haematopoietic transfection systems and in the t(8;21) context, not in t(4;11) or any KMT2A-rearranged line. UNPUBLISHED rather than not-applicable. |
| epigenetic readout | IT IS HOW KMT2A FINDS ITS DNA - a sequence-specific transcription factor that recruits and retains H3K4me3, and the measurement is on KMT2A itself rather than on a fusion. RUNX1 (AML1) and MLL physically and functionally interact, and BOTH ARE REQUIRED to maintain the H3K4me3 mark at two critical regulatory regions of the RUNX1 target gene PU.1 (SPI1); MLL binding also abrogates RUNX1's proteasome-dependent degradation, as CBFB does; and a subset of previously uncharacterised frame-shift and missense mutations at the AML1 N-terminus found in MDS and AML patients IMPAIRS the MLL interaction, with the consequence that H3K4me3 is LOST within the PU.1 regulatory regions and PU.1 expression falls [E_q97_epi1]. The authors' own framing is the load-bearing part for this graph: THE INTERACTION PROVIDES A MECHANISM FOR THE SEQUENCE-SPECIFIC BINDING OF MLL TO DNA, and RUNX1 target genes become candidate effectors of MLL function. KMT2A's CXXC domain reads unmethylated CpG, which is an address class rather than an address; RUNX1 supplies an actual motif. WHY THIS IS NOT JUST ANOTHER INTERACTOR. This entry's existing content is about the RECIPROCAL product: RUNX1 is directly activated by KMT2A::AFF1 and the RUNX1 protein then binds AF4-MLL, the der(4) product - a two-oncoprotein circuit with no contact on KMT2A::AFF1 itself. The H3K4me3 result adds the third leg: RUNX1 also binds WILD-TYPE MLL and is required for that enzyme's mark at a myeloid master-regulator locus. So in a t(4;11) blast RUNX1 is simultaneously a fusion target gene, a binder of the reciprocal fusion, and a DNA-targeting cofactor of the retained wild-type allele - and the three roles predict different things. A SECOND, WEAKER ARM WORTH RECORDING BECAUSE IT TOUCHES A KNOWN TRAP: RUNX1 associates with the protein arginine methyltransferase PRMT6, which deposits H3R2me2a at RUNX1 target genes; H3R2me2a antagonises the positive H3K4me3 mark and so acts repressively, the association affects BIVALENT histone marks, and on differentiation a RUNX1/PRMT6 corepressor complex is exchanged for a coactivator complex [E_q97_epi2]. Note that THIS is the genuine chromatin sense of 'bivalent' - co-occurring H3K4me3 and H3K27me3 - and not the two-motif protein-interaction sense that data/symbol-collisions.json records for the menin literature. BOUNDED: RUNX1[tiab] AND H3K79[tiab] returns 0, against RUNX1[tiab] AND H3K4me3[tiab] at 19 as the control through the identical shape. The mark this graph's DOT1L arm turns on has never been measured together with RUNX1 in any system, so RUNX1's relationship to the fusion's own signature mark is unknown. |
relationships to the fusion
[
{
"kind": "direct transcriptional target",
"of": "KMT2A::AFF1",
"evidence": "E1"
},
{
"kind": "protein interaction",
"of": "reciprocal AFF1::KMT2A (AF4-MLL)",
"evidence": "E1"
},
{
"kind": "enhancer-programme co-occupancy",
"of": "KMT2A fusions in 11q23 AML",
"evidence": "E2"
}
]essentiality
{
"depmap_common_essential": false,
"chronos_mean": -0.0125,
"n_lines": 1208,
"fraction_lines_depleted": 0.0488,
"chronos_mean_kmt2a_rearranged": -0.5523,
"chronos_mean_other_leukemia": -0.7408,
"kmt2a_selectivity_difference": 0.1885,
"n_kmt2a_rearranged_lines": 11,
"n_other_leukemia_lines": 43,
"benchmark_men1_difference": -0.8656,
"interpretation": "A leukemia dependency but an anti-selective one: KMT2A-rearranged lines need RUNX1 less than other leukemia lines."
}domains — the numbers behind the chart
[
{
"name": "RHD (Runt homology domain)",
"start": 50,
"end": 178,
"pfam": "PF00853",
"note": "DNA-binding and CBFB-heterodimerisation domain; boundaries from UniProt Q01196 feature annotation"
},
{
"name": "NMTS (nuclear matrix targeting signal)",
"start": 351,
"end": 381,
"pfam": null,
"note": "31-residue segment necessary and sufficient to direct AML-1B to the nuclear matrix; lost in the t(8;21) and t(3;21) translocation products [E4]"
},
{
"name": "TAD (transactivation domain)",
"start": 291,
"end": 371,
"pfam": null,
"note": "UniProt Q01196 annotates 291-371 as the KAT6A-interaction region, which corresponds to the classically defined RUNX1 transactivation domain; the exact TAD boundaries are defined differently across papers and this range is the one this entry can defend"
}
]localization
{
"subcellular": "NUCLEUS. UniProt Q01196 annotates Nucleus and nothing else (fetched 2026-09-21); note the annotation carries no ECO:0000269 evidence code in the returned record, so the compartment is well established in the literature but is not experimentally evidenced in the UniProt entry itself.",
"subnuclear": "NUCLEAR MATRIX, and this is one of the best-defined subnuclear addresses in the whole graph. The transcriptionally ACTIVE isoform AML-1B binds the nuclear matrix and the inactive AML-1 does not; association is independent of DNA binding and requires a 31-amino-acid nuclear matrix targeting signal at residues 351-381, distinct from the NLS, which is necessary and sufficient to direct a heterologous GAL4 fusion to the matrix - see this entry's existing evidence item E4 (PMID 9192636). Loss of the C-terminal domain carrying the NMTS is a frequent consequence of t(8;21) and t(3;21).",
"why_this_matters_here": "This graph already carries localization-nuclear-matrix-mitotic-scaffold-kmt2a, recording that KMT2A itself sits on the nuclear matrix and mitotic scaffold beside topoisomerase II. RUNX1 and KMT2A are therefore two proteins with independently measured nuclear-matrix association, and the RUNX1 case is the cleaner one because a 31-residue signal has been isolated and transplanted. Nobody has asked whether they occupy the same matrix sites.",
"explicit_no": "NOT MEASURED IN A KMT2A-REARRANGED CELL. The NMTS work is in non-haematopoietic transfection systems and in the t(8;21) context, not in t(4;11) or any KMT2A-rearranged line. UNPUBLISHED rather than not-applicable.",
"how_shown": "nuclear matrix fractionation, deletion mapping and GAL4 heterologous targeting (PMID 9192636, this entry's E4); UniProt Q01196 SUBCELLULAR LOCATION comment, fetched 2026-09-21",
"systems": "non-haematopoietic transfection systems for the NMTS; murine embryos for the cell-type work",
"identifier_warning": "THE GENE ENTRY IS NOT THE FUSION ENTRY. This graph also contains fusion-runx1-runx1t1 and fusion-etv6-runx1, and the NMTS result is partly ABOUT a fusion - t(8;21) removes the C-terminal domain that carries the signal. A localisation claim about RUNX1 the protein must not be read as a localisation claim about either fusion product.",
"node": "localization-nuclear-matrix-mitotic-scaffold-kmt2a"
}cell type
{
"evidence_class_key": "concept-expressed-required-maintains-three-evidence-classes",
"identifier_note": "RUNX1 (Q01196, 453 aa, synonyms AML1 and CBFA2) is not RUNX2 or RUNX3, and the NMTS paper explicitly reports a similar signal in AML-2 and the bone-related AML-3, so a claim about 'the AML/CBF-alpha NMTS' is a family claim and a claim about residues 351-381 is a RUNX1 claim. The 1990s literature calls this gene AML1 and CBFA2; an absence claim keyed to RUNX1 alone would miss most of it (brief: search the retired symbol too).",
"expressed_in": {
"class": 1,
"summary": "haematopoietic and osteogenic lineages; a master transcription factor of definitive haematopoiesis rather than a ubiquitous housekeeping protein."
},
"required_in": {
"class": 2,
"claim": "In MICE, Runx1 is required at the EMERGENCE of definitive haematopoiesis: Aml1-null embryos lack fetal liver haematopoiesis and have no definitive myeloid or erythroid progenitors, while primitive yolk-sac erythropoiesis is normal, and the defect is cell-intrinsic by chimera analysis [E_q71_ct1]. The cell types this names are the haemogenic endothelium-to-HSC transition and the fetal liver HSC.",
"pmids": [
"8565077"
],
"system_named_in_the_claim": "murine germline Aml1 knockout embryos, normal development - no fusion present",
"nodes": [
"cell-type-haemogenic-endothelium",
"cell-type-fetal-liver-hsc"
]
},
"not_required_in": {
"claim": "ANTI-SELECTIVE, AND NEARLY AT THE BOTTOM. Chronos -0.5523 in KMT2A-rearranged against -0.7408 in other leukaemias, difference +0.1885, rank 17,571 of 17,728. Both arms are genuine dependencies - this is not floor compression and not a non-dependency - so the reading is clean: RUNX1 is REQUIRED in leukaemia lines generally and MEASURABLY LESS SO where the fusion is.",
"what_that_does_to_this_entry": "This entry describes a two-oncoprotein circuit in which KMT2A::AFF1 activates RUNX1 and the reciprocal AFF1::KMT2A product binds it. The circuit can be exactly as published and RUNX1 still not be a fusion-selective vulnerability - activation is not dependency. The entry should be read as describing a wiring diagram, not proposing a target."
},
"context": "murine embryonic haematopoiesis for the requirement; this project's DepMap scan for the selectivity"
}cell state
{
"evidence_class_key": "concept-expressed-required-maintains-three-evidence-classes",
"maintains_state": {
"class": 3,
"states": [
"cell-state-stem-like-self-renewal"
],
"claim": "In MICE, Runx1 is required for the EMERGENCE of the definitive haematopoietic stem cell state rather than for maintaining an existing one: primitive erythropoiesis proceeds normally in its absence and it is the definitive programme that fails [E_q71_ct1]. This is a state-ESTABLISHING requirement at a developmental transition, which is a different evidence class from the maintenance claims most of this graph's cell-state entries carry.",
"pmids": [
"8565077"
],
"system_named_in_the_claim": "murine embryonic definitive haematopoiesis"
},
"confidence": "established",
"caveats": "Murine germline knockout in embryonic development. Nothing here is human, leukaemic, KMT2A-rearranged or infant B-lineage. The requirement is developmental, and an established leukaemia is a maintenance problem - a distinction this graph has already recorded biting for KAT8/MOF, where the transformation and maintenance answers differ.",
"counter_evidence": {
"note": "This is squarely a LINEAGE property, not a fusion property, and it is one of the clearest examples in this beat. RUNX1's requirement is documented in a mouse embryo with an intact Kmt2a locus and no fusion of any kind, and the requirement is at the very origin of the blood system. The same shape - every state found here exists in an animal with no fusion - is what led cycle 39's review to conclude the lineage explanation is currently better supported. The DepMap placement at rank 17,571 makes the point quantitatively: if the state were fusion-driven, the gene setting it should be MORE required where the fusion is.",
"citation": {
"title": "AML1, the target of multiple chromosomal translocations in human leukemia, is essential for normal fetal liver hematopoiesis.",
"authors": "Okuda T, van Deursen J, Hiebert SW, Grosveld G, Downing JR",
"year": 1996,
"venue": "Cell",
"pmid": "8565077",
"doi": "10.1016/s0092-8674(00)80986-1",
"type": "paper"
}
}
}depmap kmt2a selectivity
{
"chronos_mean_kmt2a_rearranged": -0.5523,
"chronos_mean_other_leukaemia": -0.7408,
"difference": 0.1885,
"pan_essentiality_all_lines": -0.0125,
"selectivity_rank": 17571,
"rank_denominator": 17728,
"n_all_lines": 1208,
"open_targets_dataversion": "26.06",
"category": "anti-selective, and a genuine one - both arms are negative, there is no floor compression, and the gene is not a tumour suppressor artefact on this scale",
"provenance": "Joined 2026-09-21 by Q71 from analyses/depmap-kmt2a-selective-scan-full (generated 2026-09-19T10:12:06Z), over 11 KMT2A-rearranged and 43 other leukaemia DepMap lines, 17,728 genes scanned. Open Targets dataVersion fetched live 2026-09-21 as year 26 month 06, per brief rule 83. NOT independent of that scan: every caveat the scan carries applies here, including that the 11-line index arm is myeloid-enriched and contains no infant-derived line. This entry carried NO depmap block before - the Q71 dispatch brief asserted that every gene in its set carried one, and three did not.",
"how_to_read": "A negative difference means MORE required in KMT2A-rearranged lines. A rank near 17,728 is ANTI-selective. Check pan_essentiality_all_lines and BOTH arm means first: if both arms sit past -2 the rank is floor-compressed and uninterpretable, and if both arms are POSITIVE the gene is not a dependency at all and the scan's rank does not apply to it (the CDKN2A category error Q69 named in cycle 39, and the ATM case Q71 added in cycle 40)."
}epigenetic bounded absence
{
"claim": "RUNX1 has never been measured together with H3K79 methylation, the mark this graph's DOT1L arm turns on, in any system.",
"date_measured": "2026-09-21",
"tool": "NCBI esearch db=pubmed (eutils), querytranslation read back from every response",
"queries": [
{
"query": "RUNX1[tiab] AND H3K79[tiab]",
"querytranslation": "\"RUNX1\"[Title/Abstract] AND \"H3K79\"[Title/Abstract]",
"count": 0,
"errorlist_note": "warninglist outputmessages ['No items found.']; quotedphrasesnotfound empty, so both terms are indexed"
},
{
"query": "RUNX1[tiab] AND H3K4me3[tiab]",
"querytranslation": "\"RUNX1\"[Title/Abstract] AND \"H3K4me3\"[Title/Abstract]",
"count": 19,
"role": "POSITIVE CONTROL through the identical shape - one gene token, one compact mark token - and it returns 19, so the zero above is not an instrument failure"
},
{
"query": "AML1[tiab] AND MLL[tiab] AND H3K4[tiab]",
"querytranslation": "\"AML1\"[Title/Abstract] AND \"MLL\"[Title/Abstract] AND \"H3K4\"[Title/Abstract]",
"count": 1,
"title_checked": "PMID 22012064, the paper cited at E_q97_epi1. Run under the RETIRED symbol AML1 because the 2011 literature uses it; the approved-symbol form RUNX1 AND MLL AND H3K4 was not the route that found this paper."
},
{
"query": "RUNX1[tiab] AND \"bivalent\"[tiab]",
"querytranslation": "\"RUNX1\"[Title/Abstract] AND \"bivalent\"[Title/Abstract]",
"count": 5,
"role": "run against the 'bivalent' registry row; here the chromatin sense is the correct one - PMID 23879542 uses it for co-occurring H3K4me3/H3K27me3 - so the menin two-motif contamination did not fire. The other 4 were not read."
}
],
"symbols_searched": [
"RUNX1",
"AML1",
"H3K79",
"H3K4me3",
"H3K4",
"MLL",
"bivalent"
],
"symbols_not_searched": [
"CBFA2",
"AMLCR1",
"PEBP2A",
"PEBP2aB",
"H3-K79",
"lysine 79",
"DOT1L"
],
"absence_class": "never-measured",
"absence_class_basis": "Bounded by titles and abstracts only; supporting figures not searched (rule 80), and an H3K79me2 ChIP track at a RUNX1 motif is precisely the supporting panel rule 80 says this index cannot see. The retired symbol CBFA2 and the spellings 'H3-K79' and 'lysine 79' were not run, so the bound is on the name list as well as on the index (rules 70-72).",
"reopening_condition": "Any H3K79me2 track examined at RUNX1-motif-containing regions, or any report of RUNX1 occupancy overlapping DOT1L or H3K79me2 domains in a KMT2A-rearranged cell, reopens this. The graph already holds several H3K79 mechanism entries that would be the place to link it."
}- RUNX1 loss of function causes familial platelet disorder with predisposition to myeloid malignancy; lowering RUNX1 activity therapeutically risks the exact phenotype the germline disease demonstrates.
- KMT2A-rearranged lines are less RUNX1-dependent than other leukemia lines, so a RUNX1-directed agent would have its weakest effect in the subtype it was reasoned into.
- If the mechanism depends on the reciprocal AF4-MLL product, patients whose der(4) is lost or not expressed would be intrinsically resistant, and der(4) status is not routinely measured.
- RUNX1 is required for the emergence of the entire definitive blood system in mice (PMID 8565077). There is no therapeutic window for removing it from a child, and any RUNX1-directed strategy has to be a state- or complex-specific one rather than a loss-of-function one.
- The scan points away from RUNX1 as a target in this disease: rank 17,571 of 17,728, LESS required in KMT2A-rearranged lines than in other leukaemias (difference +0.1885), with both arms genuinely depleted. Whatever the two-oncoprotein circuit in this entry means, it is not a selectivity argument.
- Is the RUNX1 dependency different in KMT2A::AFF1 lines specifically, as opposed to the AML-dominated KMT2A-rearranged group? The published mechanism is t(4;11)-specific and the DepMap comparison is not.
- Which surfaces mediate the RUNX1-AF4::KMT2A interaction? Without mapping it is impossible to know whether it is druggable or even whether it is direct.
- RUNX1 is a haploinsufficient tumour suppressor in the germline and a required factor in these leukemia lines. Which of those two faces dominates in an infant t(4;11) blast?
- Does the positive selectivity (+0.189) reflect genuine bypass, or the absence of reciprocal-product biology in the DepMap KMT2A-rearranged panel?
- Does the RUNX1 NMTS at 351-381 matter to the reported interaction with the reciprocal AF4::KMT2A product? The interaction is unmapped on both sides, and a nuclear-matrix-tethered transcription factor meeting a chromatin-bound fusion is at least a testable spatial model.
- Do RUNX1 and KMT2A occupy the same nuclear-matrix sites? Both have independently measured matrix association, RUNX1 through a transplantable 31-residue signal and KMT2A alongside topoisomerase II, and nobody has asked whether the two co-reside - which would be directly relevant to how the t(4;11) break is made.
- Why is RUNX1 LESS required in KMT2A-rearranged lines than in other leukaemias? An activated target that becomes less essential is an unusual pattern and may mean the fusion supplies RUNX1's output by another route.
- Does KMT2A::AFF1 retain the region that binds RUNX1? PMID 22012064 maps the interface on the AML1 side only. If the fusion keeps it, RUNX1 supplies the sequence-specific DNA targeting that CXXC cannot; if it does not, the fusion loses an address the wild-type allele has - and the two predictions are opposite.
- Is H3K4me3 at the PU.1/SPI1 regulatory regions altered in KMT2A-rearranged B-ALL? The mark there is jointly maintained by RUNX1 and MLL, and PU.1 is a myeloid master regulator in a disease defined by a differentiation block.
- Does RUNX1 occupancy overlap H3K79me2 domains in a KMT2A-rearranged cell? RUNX1[tiab] AND H3K79[tiab] returns 0 corpus-wide against a control of 19 for H3K4me3, so the relationship between this graph's central mark and its best-evidenced sequence-specific cofactor is entirely unmeasured.
- E1strong confidence 4 of 5: Replicated in peer-reviewed work by more than one group.paper
RUNX1 is directly activated by MLL-AF4 in t(4;11) leukemia, and the RUNX1 protein interacts with the product of the reciprocal AF4-MLL translocation, supporting a mechanism in which two oncogenic fusion proteins cooperate by activating a target gene and then modulating the function of its downstream product.
RUNX1 is a key target in t(4;11) leukemias that contributes to gene activation through an AF4-MLL complex interaction. · Wilkinson AC, Ballabio E, Geng H, North P, Tapia M, Kerry J, Biswas D, Roeder RG, Allis CD, Melnick A, de Bruijn MFTR, Milne TA · Cell Reports 2013What this evidence cannot show
Abstract read in full; not full text. The abstract states the interaction is with the AF4-MLL complex and does not give residue-level mapping or affinity, so this entry cannot say which surface of RUNX1 or of AF4-MLL is involved. Work is in human t(4;11) cell lines; the two-oncoprotein interpretation depends on the reciprocal product being expressed and functional, which is itself contested (see gene-siah1).
Evidence that argues against this (1)
- The dependency data do not support RUNX1 being a heightened requirement in KMT2A-rearranged cells. Across DepMap, KMT2A-rearranged leukemia lines are LESS RUNX1-dependent than other leukemia lines (-0.552 against -0.741, difference +0.189). That does not contradict the published interaction, which is a statement about binding rather than requirement, but it does mean the interaction cannot be assumed to translate into a KMT2A-selective vulnerability. Broad Institute DepMap; Cellosaurus; Open Targets Platform; analysis run by this project 2026
- E2moderate confidence 3 of 5: One strong study, or several smaller consistent ones.paper
MLL-AF9 and MLL-AF4 oncofusion proteins bind a distinct enhancer repertoire and target the RUNX1 programme in 11q23 acute myeloid leukemia.
MLL-AF9 and MLL-AF4 oncofusion proteins bind a distinct enhancer repertoire and target the RUNX1 program in 11q23 acute myeloid leukemia. · Prange KHM, Mandoli A, Kuznetsova T, Wang SY, Sotoca AM, Marneth AE, van der Reijden BA, Stunnenberg HG, Martens JHA · Oncogene 2017What this evidence cannot show
PMID fetched and verified against the NCBI record for title, authors, year and venue; abstract read, not full text. An earlier draft of this entry carried PMID 28369024, which is an unrelated nursing care-management survey; the error was caught by fetching the PMID before the entry was finished. This is AML (11q23), not infant B-ALL, and the enhancer claim is about genomic co-occupancy, not physical interaction — it must not be read as evidence that RUNX1 binds KMT2A::AFF1.
Evidence that argues against this (1)
- Binding a distinct enhancer repertoire and targeting the RUNX1 programme is occupancy plus expression correlation in 11q23 AML. It does not establish that RUNX1 is required, and this graph's own lineage-split work shows myeloid transcription factors are the class most prone to appearing selective for lineage reasons. Oncofusion knowledge graph (internal analysis) 2026
- E3strong confidence 4 of 5: Replicated in peer-reviewed work by more than one group.dataset
RUNX1 has DepMap mean gene effect -0.0125 across 1,208 lines with 4.9% depleted; mean -0.5523 in 11 KMT2A-rearranged leukemia lines against -0.7408 in 43 other leukemia lines, a difference of +0.189; MEN1 in the same lines is -0.866.
DepMap CRISPR (Chronos) gene effect via the Open Targets Platform GraphQL API, with KMT2A-rearrangement status from Cellosaurus; per-gene recomputation for this beat · Broad Institute DepMap; Cellosaurus; Open Targets Platform; analysis run by this project · Database and derived analysis 2026What this evidence cannot show
DepMap release version not stated by the API. The KMT2A-rearranged group is dominated by AML fusions rather than t(4;11), and the published RUNX1 mechanism is specifically a t(4;11) reciprocal-product mechanism, so this comparison may be testing the wrong lines. Eleven lines; not permutation-tested here.
Evidence that argues against this (1)
- The numbers argue against the entry: −0.5523 in index lines against −0.7408 in comparators means RUNX1 is LESS required in KMT2A-rearranged lines than in other leukaemias. That is a negative selectivity result and should be stated as one. Oncofusion knowledge graph (internal analysis) 2026
- E4established confidence 5 of 5: Multiple independent lines of evidence; treated as settled in the field.paper
The transcriptionally active AML-1B (RUNX1) binds the nuclear matrix through a 31-amino-acid nuclear matrix targeting signal at residues 351-381, which is distinct from nuclear localisation signals and is necessary and sufficient to direct a heterologous protein to the nuclear matrix; loss of the AML-1B C-terminal domain is a frequent consequence of the t(8;21) and t(3;21) translocations.
Identification of a nuclear matrix targeting signal in the leukemia and bone-related AML/CBF-alpha transcription factors. · Zeng C, van Wijnen AJ, Stein JL, Meyers S, Sun W, Shopland L, Lawrence JB, Penman S, Lian JB, Stein GS, Hiebert SW · Proceedings of the National Academy of Sciences USA 1997What this evidence cannot show
Abstract read, plus PMC full text consulted for the residue range 351-381 ('amino acids between 351 and 381 contain a unique NMTS'). Numbering is for AML-1B, the 453-aa isoform that is UniProt Q01196's canonical sequence, so the coordinates are consistent with this entry's length_aa. The domains added here are for the 3D viewer's benefit and are not part of this entry's interactome argument.
Evidence that argues against this (1)
- A nuclear-matrix targeting signal was defined by deletion mapping with biochemical matrix preparations, a fractionation whose operational definition of the matrix has been contested for three decades. The localisation claim is method-dependent. eLife 2019
- E_q71_ct1established confidence 5 of 5: Multiple independent lines of evidence; treated as settled in the field.paper
In MICE, Runx1 (AML1) is essential for DEFINITIVE haematopoiesis of all lineages and is dispensable for primitive erythropoiesis: Aml1-null embryos show normal morphogenesis and normal yolk-sac-derived erythropoiesis but LACK FETAL LIVER HAEMATOPOIESIS and die around E12.5; no myeloid or erythroid progenitors of definitive origin are detectable in yolk sac or fetal liver, sequentially targeted AML1-/- ES cells still make primitive erythroid cells in vitro, and the defect is intrinsic to the stem cells because AML1-/- ES cells fail to contribute to haematopoiesis in chimeric animals.
AML1, the target of multiple chromosomal translocations in human leukemia, is essential for normal fetal liver hematopoiesis. · Okuda T, van Deursen J, Hiebert SW, Grosveld G, Downing JR · Cell 1996What this evidence cannot show
Abstract read in full; full text not read. MURINE germline knockout, normal embryonic development - not leukaemia, not human, not KMT2A-rearranged (brief rule 73). This is a DEVELOPMENTAL requirement measured at the emergence of definitive haematopoiesis; it says nothing directly about whether an established leukaemic cell needs RUNX1, which is a maintenance question. The chimera experiment is the part that makes the defect cell-intrinsic rather than niche-derived.
Evidence that argues against this (1)
- This project's own genome-wide scan places RUNX1 at rank 17,571 of 17,728 - almost the bottom - with Chronos -0.5523 in the KMT2A-rearranged arm against -0.7408 in other leukaemias, difference +0.1885. RUNX1 is a real dependency in BOTH arms and is LESS required where the fusion is. A developmental requirement at the origin of the lineage therefore does not transfer to a fusion-selective vulnerability, and the fusion appears to relieve rather than deepen the requirement. The comparison shape - a lineage gene whose requirement is documented in a fusion-free animal - is the same one that led this graph's cycle-39 review to conclude that the lineage explanation is currently better supported than the fusion one. Ceccaldi R et al. 2012
- E_q97_epi1strong confidence 4 of 5: Replicated in peer-reviewed work by more than one group.paper
RUNX1 (AML1) and MLL physically and functionally interact and both are required to maintain H3K4me3 at two critical regulatory regions of the AML1 target gene PU.1; MLL binding abrogates AML1's proteasome-dependent degradation; and a subset of frame-shift and missense mutations at the AML1 N-terminus found in MDS and AML patients impairs the MLL interaction, resulting in loss of H3K4me3 within the PU.1 regulatory regions and decreased PU.1 expression - providing a mechanism for the sequence-specific binding of MLL to DNA.
What this evidence cannot show
Abstract fetched and read in full 2026-09-21; full text not read. The MLL here is WILD-TYPE KMT2A, not a fusion - which is what makes it relevant (the fusion keeps the N-terminus) and also what limits it: no KMT2A::AFF1 or B-lineage arm is reported, and the patient mutations are MDS/AML, i.e. myeloid (rule 73). The system is not stated in the abstract beyond the patient-mutation source, so the cell context of the H3K4me3 ChIP is not established here. The interaction is mapped to the AML1 N-terminus (the RUNT region side); the reciprocal interface on KMT2A is not given, so no residue range is added to this entry's domains.
Evidence that argues against this (1)
- The graph should not read this as 'RUNX1 targets the FUSION to DNA'. The interaction surface on the MLL side is unmapped in this abstract, and this entry's own existing content establishes that the demonstrated RUNX1 contact in t(4;11) is on AF4-MLL, the RECIPROCAL der(4) product, not on KMT2A::AFF1. Whether KMT2A::AFF1 retains whatever region binds RUNX1 is undetermined, so the attractive inference - that RUNX1 supplies the fusion's missing sequence specificity - is exactly the kind of partner-imported claim this project has had to strike repeatedly. Crit Rev Eukaryot Gene Expr 2013
- E_q97_epi2moderate confidence 3 of 5: One strong study, or several smaller consistent ones.review
RUNX1 recruits epigenetic cofactors to its target loci, and is associated with the protein arginine methyltransferase PRMT6: PRMT6 triggers H3R2me2a at RUNX1 target genes, that modification negatively influences the positive H3K4me3 mark and so acts repressively, the RUNX1/PRMT6 association has an impact on bivalent histone marks, and on differentiation a RUNX1 corepressor complex with PRMT6 is exchanged for a RUNX1 coactivator complex.
Contributions of the histone arginine methyltransferase PRMT6 to the epigenetic function of RUNX1. · Crit Rev Eukaryot Gene Expr 2013What this evidence cannot show
A REVIEW, included as a map and not as a primary source, per the standing quality bar. Abstract fetched and read in full 2026-09-21; full text not read, and the primary PRMT6/RUNX1 papers it reviews were NOT retrieved for this amendment, so no specific target locus, cell type or effect size is claimed. H3R2me2a is an ARGININE mark and is matched by none of the mark regexes in scripts/check-pi-dimensions.py, which is worth noting as a measurement gap in the reporter rather than in the biology. Cited principally because it is the one place in this tranche where 'bivalent' is used in its genuine chromatin sense.
Evidence that argues against this (1)
- A review cannot settle whether the PRMT6 arm operates in this graph's disease, and the H3R2me2a-versus-H3K4me3 antagonism it describes would, if it holds at KMT2A fusion targets, mean that a PRMT6 inhibitor RAISES H3K4me3 at RUNX1 target genes - the opposite direction from every H3K4/H3K79-lowering agent this graph considers. Nothing here tests that and this entry does not claim it. Blood 2011
- ← regulatesKMT2A::AFF1 (MLL-AF4), t(4;11)(q21;q23)
RUNX1 is a direct transcriptional target of MLL-AF4 in t(4;11)
RUNX1 protein interacts with the reciprocal AF4-MLL complex, not with the der(11) product
- related to →SIAH1 — the E3 ligase that degrades AFF1, and that the reciprocal AF4::KMT2A fusion escapes
both bear on whether the reciprocal der(4) product is functionally important in t(4;11)
- related to →RUNX1::RUNX1T1 (AML1-ETO), t(8;21)(q22;q22)
same protein, opposite role: 5' partner of a different fusion, and a target-programme node in KMT2A-r
RUNX1 is one of the TFs recruited by BENC modules, and the drug-induced RUNX1 peak in inv(16) is at element E3
RUNX1/PRMT6 corepressor complex deposits H3R2me2a at RUNX1 targets
PRMT6 triggers H3R2me2a at RUNX1 target genes - the mark's only haematopoietic foothold, and it is not KMT2A
CRISPR deletion in T-ALL deregulates RUNX1 target genes - the second arginine methyltransferase to land on RUNX1, after PRMT6/H3R2me2a
- ← co-occurs withELF2 / NERF - an ETS factor that works with RUNX1 on a B-cell kinase promoter, and one of eleven TF baits recovering KMT2A
isoform 1 synergises with RUNX1 to transactivate the BLK promoter; isoform 2 may repress RUNX1-mediated transactivation
CRG contacts fall on RUNX1 knockdown; CRGs are enriched in RUNX1 targets
Forkhead domain to Runt domain; FOXC1 stabilises RUNX1/HDAC1/TLE3 at enhancers
direct protein-protein contact among RUNX1, GATA2, SCL and ERG stabilises heptad binding to DNA in blood stem/progenitor cells
RUNX1 has the cleanest nuclear-matrix address in this graph: the active AML-1B isoform binds the matrix through a transplantable 31-residue NMTS at 351-381 that the inactive isoform lacks, and KMT2A is independently matrix-associated - nobody has asked whether they share sites (non-haematopoietic transfection systems; PMID 9192636)
REQUIRED IN: Aml1-null mouse embryos lack definitive haematopoiesis entirely while primitive yolk-sac erythropoiesis is normal, and the defect is cell-intrinsic by chimera analysis - the requirement sits at the endothelial-to-haematopoietic transition (murine germline knockout; PMID 8565077)
REQUIRED IN: Aml1-null embryos lack fetal liver haematopoiesis and die around E12.5, with no definitive myeloid or erythroid progenitors in yolk sac or fetal liver (murine; PMID 8565077)
- regulates →The stem-like self-renewal state — a limited stem-cell module grafted onto a committed cell
Runx1 is required to ESTABLISH the definitive haematopoietic stem cell state rather than to maintain an existing one - a state-establishing evidence class, in an animal with no fusion (murine embryos; PMID 8565077)
PU.1's only retrievable subnuclear connection is RUNX1's: a tyrosine in RUNX1's nuclear matrix targeting signal is required for subnuclear targeting and for myeloid-promoter activation with PU.1 as the cooperating factor (PMID 16149049). PU.1's own compartment has never been measured (Q91 2026-09-22)
RUNX1 and MLL are jointly required to maintain H3K4me3 at two PU.1 regulatory regions; MDS/AML RUNX1 N-terminal mutations that break the MLL contact lose the mark
PU.1/SPI1 is the RUNX1 target gene at which the RUNX1-MLL-dependent H3K4me3 was measured
METTL3 depletion disrupts CTCF binding and reduces chromatin accessibility at MYB and RUNX1 in MLL-rearranged AML
RUNX1 protein interacts with the reciprocal product, per an independent laboratory; the species assignment rests on apparent molecular weight.