RUNX2 - a direct KMT2A-chimera target and dependency in T-ALL, a patient-specific enhancer target in KMT2A::AFF1 B-ALL, and the gene a menin inhibitor represses to restore steroid sensitivity in t(4;11) infants

No three-dimensional model is available for this protein, so the map above is the whole picture. That is itself informative: AlphaFold does not model chains this long, and the parts that have been crystallised are held under the entries for the individual domains.

• RUNX2 is the osteoblast master gene AND a gene whose repression contributes to menin-inhibitor benefit in infant t(4;11). The same on-target axis therefore carries the class's documented paediatric skeletal signal - irreversible growth-plate closure in juvenile rats at ~2x human AUC, alkaline phosphatase elevation in a third to two-fifths of patients - so this is essential-function toxicity, not off-target toxicity, and it cannot be engineered away by a more selective menin inhibitor.
• RUNX2 drives CXCR4 and homing in KMT2A-R T-ALL. Suppressing RUNX2 could mobilise blasts out of the marrow niche rather than kill them, which in a patient looks like a rising peripheral count during a response - a failure mode with a known precedent in CXCR4-axis agents and one that a marrow-only MRD assessment would misread.
• The KMT2A::AFF1 RUNX2 signal is patient-specific enhancer usage, not a subtype property. A RUNX2-directed strategy enrolled on the fusion alone would dilute responders with non-responders by construction, and the biomarker that fixes it - per-patient enhancer activity - currently requires Micro Capture-C on primary material.

1. Which RUNX2 isoform does a menin inhibitor repress in t(4;11) blasts - P1/p57 or P2/p56? The leukaemia ChIP bin is at P2 and the osteoblast COMPASS literature is at P1. If the two are separable, the drug class's leukaemia benefit and its growth-plate liability may be separable too, and if they are not, they are the same on-target effect and no selectivity gain can split them.
2. Is the RUNX2 enhancer that Micro Capture-C finds active in some KMT2A::AFF1 patients the same element as, or contacting, the P2 promoter the RS4;11 bin sits at? The published Micro Capture-C data and the binned RS4;11 data are both coordinate-resolved and could be intersected without new material.
3. Does RUNX2 have the same dependency status in KMT2A::AFF1 B-ALL as in KMT2A-R T-ALL? The functional dependency, the CXCR4 axis and the CBFB-interface vulnerability are all T-lineage results; the index fusion has expression heterogeneity and a drug-response correlate and no loss-of-function experiment.
4. Where is RUNX2 in the nucleus in a leukaemic blast? This entry deliberately asserts nothing about subnuclear localisation because nothing was fetched and read. For a project whose standing priority names subnuclear localisation, a RUNT-family factor with no subnuclear record is a gap, not a nuance.
5. Does the RUNX2-CBFB interface inhibitor that is effective in immature and KMT2A-R T-ALL do anything in KMT2A::AFF1 B-ALL? It is an existing chemical probe against an interface this graph already reasons about through fusion-cbfb-myh11.

1. E1established confidence 5 of 5: Multiple independent lines of evidence; treated as settled in the field.database

   RUNX2 is Q13950 (RUNX2_HUMAN), 521 aa, with synonyms AML3, CBFA1, OSF2 and PEBP2A; the feature table gives a Runt domain at 101-229, a KAT6A-interaction region at 336-439, a KAT6B-interaction region at 374-468 and a FOXO1-interaction region at 242-258; it is localised to nucleus and cytoplasm, is recorded as specifically expressed in osteoblasts, is essential for osteoblast maturation and both intramembranous and endochondral ossification, and germline loss causes cleidocranial dysplasia 1.

   UniProtKB Q13950 (RUNX2_HUMAN), Runt-related transcription factor 2 · UniProt Consortium · UniProt 2026
   ▸What this evidence cannot show

   Fetched live 2026-09-23 via gene_exact:RUNX2 AND organism_id:9606 AND reviewed:true, which returned exactly ONE reviewed record, so rule 62 does not fire. data/symbol-collisions.json (208 rows) word-boundary grepped for RUNX2 before the first query: no row. 'Specifically expressed in osteoblasts' is UniProt's tissue-specificity summary and is FALSE as a general statement in this project's setting - RUNX2 is expressed and functionally required in KMT2A-R T-ALL blasts (E3) and is highly expressed in a subset of KMT2A::AFF1 B-ALL patients (E4). The localisation comment is not resolved to a fragment or isoform and, per rule 91, its cited experiments were not opened.

   ▸Evidence that argues against this (1)

   • UniProt's 'specifically expressed in osteoblasts' would, taken at face value, exclude every observation in this entry. The leukaemia literature contradicts it directly, and the RUNX2 protein required in KMT2A-R T-ALL is the same gene product. A tissue-specificity line in a database is a summary of the annotated literature at the time of curation, not a bound on expression. J Clin Invest 2021
2. E2established confidence 5 of 5: Multiple independent lines of evidence; treated as settled in the field.database

   HGNC assigns RUNX2 (HGNC:10472, 'RUNX family transcription factor 2') the primary Ensembl gene id ENSG00000124813, cytoband 6p21.1 and UniProt Q13950, with PREVIOUS official symbols CCD, CBFA1 and CCD1 and aliases AML3, PEBP2A1 and PEBP2aA1; Ensembl GRCh37 places the gene at chr6:45,295,894-45,632,086 on the plus strand and GRCh38 at chr6:45,328,157-45,664,349.

   HGNC record HGNC:10472 (RUNX2) and Ensembl gene ENSG00000124813 on GRCh37 and GRCh38 · HUGO Gene Nomenclature Committee; Ensembl · genenames.org / Ensembl 2026
   ▸What this evidence cannot show

   Fetched live 2026-09-23, resolved into a note before any sentence using them was written. THREE previous official symbols means rule 69 applies and a symbol-only literature search undercounts; the retired-symbol arm is in E7. Note also that AML3 is a contaminated token in a leukaemia context - it is also the FAB class M3 shorthand and part of 'AML3' cell-line naming - which is why the retired-symbol arm below is reported per-symbol and not unioned into a single count.

   ▸Evidence that argues against this (1)

   • Nomenclature is not the only thing that shifts: the GRCh37 and GRCh38 spans for RUNX2 differ by ~32 kb in start coordinate, so any interval arithmetic that mixes builds is wrong by more than a gene body. The deposited data this entry reads is hg19 and the current primary assembly is GRCh38, and this entry carries both spans for that reason rather than converting one into the other. Ensembl 2026
3. E3strong confidence 4 of 5: Replicated in peer-reviewed work by more than one group.paper

   RUNX2 is upregulated in high-risk T-ALL carrying KMT2A rearrangements or an immature immunophenotype; in KMT2A-R cells RUNX2 is a direct target of the KMT2A chimeras and reciprocally binds the KMT2A promoter, forming a feed-forward loop; RUNX2 is required for survival of immature and KMT2A-R T-ALL cells in vitro and in vivo, directly drives CXCR4 transcription and thereby chemotaxis, adhesion and homing to medullary and extramedullary sites, increases glycolysis, oxidative phosphorylation and mitochondrial biogenesis, and immature and KMT2A-R T-ALL cells are vulnerable to pharmacological disruption of the RUNX2-CBFB interaction.

   RUNX2 regulates leukemic cell metabolism and chemotaxis in high-risk T cell acute lymphoblastic leukemia · Matthijssens F, Sharma ND, Nysus M, Nickl CK, Kang H, Perez DR, et al. · J Clin Invest 2021
   PMID 33555272DOI
   ▸What this evidence cannot show

   Abstract-level read; the full text was not opened this session. T-LINEAGE throughout, and the abstract says 'KMT2A rearrangements' without naming the 3' partners, so this is NOT a KMT2A::AFF1 result and the partner composition of the KMT2A-R T-ALL panel is unknown from the abstract (rule 86: the fusion is named as KMT2A-R T-ALL, not as a bare YES). 'Direct target' and 'reciprocally bound the KMT2A promoter' are the abstract's words; the underlying ChIP was not inspected. The CBFB-interaction vulnerability is a proof-of-concept in the paper's own framing.

   ▸Evidence that argues against this (1)

   • An earlier transcriptomic study of the same disease class found RUNX2 only as one of 330 discriminating probe sets separating KMT2A-R (n = 12) and MLLT10-R (n = 9) T-ALL out of 100 patients - i.e. as a correlate in a small panel, alongside TCF4 and MYO6. The dependency claim rests on the 2021 functional work, not on the expression signal, and the n in the discovery cohort is small. Kang H et al. 2018
4. E4strong confidence 4 of 5: Replicated in peer-reviewed work by more than one group.paper

   In KMT2A::AFF1 ALL, single-cell multiome analysis and Micro Capture-C in primary patient samples identify patient-specific enhancer activity at MEIS1 and RUNX2 that drives high expression of both oncogenes in an individual-specific manner, and novel enhancer usage is a major driver of transcriptional heterogeneity in this low-mutational-burden leukaemia.

   Enhancer heterogeneity in acute lymphoblastic leukemia drives differential gene expression in patients · Smith AL, Denny N, Chahrour C, Sharp K, Arachi M, Dopico-Fernandez AM, et al. · Blood 2025
   PMID 40729681DOI
   ▸What this evidence cannot show

   Abstract-level read; the full text was not opened this session, so the number of patients with a RUNX2 enhancer, the enhancer coordinates and whether they overlap the P2 promoter region are all unknown here. This is the INDEX FUSION in primary human material, which makes it the most directly relevant citation in this entry - and the claim it supports is heterogeneity, not universality: patient-specific means some patients and not others.

   ▸Evidence that argues against this (1)

   • Patient-specific enhancer usage is the opposite of a uniform fusion-driven target. If RUNX2's high expression in KMT2A::AFF1 is enhancer-driven and individual, then a cell-line measurement in RS4;11 - including the one in E6 - may represent one patient-like configuration and not the disease, and any RUNX2-directed strategy in this fusion needs a patient-level biomarker rather than a subtype label. Blood 2025
5. E5strong confidence 4 of 5: Replicated in peer-reviewed work by more than one group.paper

   In t(4;11) infant B-ALL, HDAC7 is epigenetically silenced by EZH2 and the KMT2A::AFF1 fusion protein; the menin inhibitor MI-538 restores HDAC7 expression, an effect enhanced by the class I HDAC inhibitor chidamide; the combination drives leukaemic pro-B cells toward a more differentiated state, and increases glucocorticoid sensitivity of patient-derived xenograft cells ex vivo by repressing the RUNX2 transcription factor; adding MI-538 and chidamide to standard chemotherapy reduces PDX engraftment in vivo and delays relapse.

   HDAC7 induction combined with standard-of-care chemotherapy provides a therapeutic advantage in t(4;11) infant B-cell acute lymphoblastic leukemia · de Barrios O, Ocón-Gabarró I, Gusi-Vives M, Collazo O, Meler A, Romecín PA, et al. · Biomark Res 2025
   PMID 40722046DOI
   ▸What this evidence cannot show

   Abstract-level read; the full text was not opened this session. The menin inhibitor is MI-538, a tool compound, not revumenib or ziftomenib - the clinical agents are not what was tested, and this project has already been burned once by publishing a claim that was really a fact about a weak tool compound. 'By repressing RUNX2' is the abstract's own causal wording for a partially-mediated effect ('an effect partially mediated by HDAC7 induction'); which RUNX2 isoform or promoter moves is not stated. Glucocorticoid sensitivity was measured ex vivo in PDX cells.

   ▸Evidence that argues against this (1)

   • RUNX2 repression here is one arm of a multi-agent combination (menin inhibitor + class I HDAC inhibitor + glucocorticoid), and the paper's own framing attributes the benefit partly to HDAC7 induction. Attributing the steroid resensitisation to RUNX2 specifically is an over-read of the abstract, and the alternative - that HDAC7 restoration is the driver and RUNX2 falls downstream - is equally consistent with what is stated. Biomark Res 2025
6. E6moderate confidence 3 of 5: One strong study, or several smaller consistent ones.dataset

   In RS4;11 (KMT2A::AFF1) the hg19 2 kb bin chr6:45,390,000 reaches 76.0x the MLL1-N track's own genome median and is in the top thirty bins genome-wide; under VTP-50469 its MLL1-N within-sample share ratio is 0.286 and its MENIN share ratio is 0.030.

   Q118 re-analysis of GSE127507, instrument B: 2 kb genome-wide bins from the deposited .deduped.tdf files (RS4;11 MLL1n GSM3635695/96, MENIN GSM3635693/94, input GSM3635699); analyses/q118-menin-at-the-human-index-fusion/results-instrumentB.json, run 2026-09-23 · Iqbal Lab oncofusion knowledge graph 2026
   ▸What this evidence cannot show

   THE MEASUREMENT THAT NAMED THIS GENE IS NOT FUSION-SPECIFIC. The track is MLL1n, Bethyl A300-086A, an MLL1 N-TERMINUS antibody whose epitope is retained in KMT2A::AFF1, so it reads fusion and wild-type MLL1 together at a masking factor the same beat measured as 3.54-4.00x. Every occupancy statement from it is about MLL1-N-reactive material, never about the fusion. n = 1 per antibody per condition - GSE127507 has no replicate - and the depositors measured a Drosophila spike-in and deliberately did not apply it, so all cross-condition figures are within-sample shares, not fold changes. Gene assignment in the source entry is by overlap with the deposited series' own RefGene TSS windows; the promoter-level assignment to P2 is this entry's own Ensembl coordinate resolution (E8).

   ▸Evidence that argues against this (1)

   • RUNX2 loses MLL1-N-reactive signal (0.286) as well as menin (0.030) under the drug, which is the ordinary behaviour of the held set and the opposite of BAHCC1's 4.036. So the single most interesting gene in this beat by literature is the least interesting by measurement, and nothing in the RS4;11 data marks RUNX2 out. Iqbal Lab oncofusion knowledge graph 2026
7. E7moderate confidence 3 of 5: One strong study, or several smaller consistent ones.database

   Bounded literature counts, 2026-09-23: RUNX2[tiab] 12,959; RUNX2[tiab] AND KMT2A[tiab] 4; RUNX2[tiab] AND MLL[tiab] 6; RUNX2[tiab] AND leukemia[tiab] 112; RUNX2[tiab] AND "T-ALL"[tiab] 6; RUNX2[tiab] AND CXCR4[tiab] 59; RUNX2[tiab] AND menin[tiab] 10; RUNX2[tiab] AND revumenib[tiab] 0; RUNX2[tiab] AND "MLL-AF4"[tiab] 0. On the retired official symbol: CBFA1[tiab] 1,021; CBFA1[tiab] AND leukemia[tiab] 15; CBFA1[tiab] AND MLL[tiab] 0; AML3[tiab] AND MLL[tiab] 5.

   PubMed esearch via scripts/pubmed_safe.py, 2026-09-23 · Q119-genes-the-measurement-named · NCBI PubMed 2026
   ▸What this evidence cannot show

   Bounded by TITLES AND ABSTRACTS only (rule 80). One tag per term, no parenthesised group (rule 94); every querytranslation was read back and each expanded to "TERM"[Title/Abstract]. Counts on RETIRED symbols are reported SEPARATELY and are deliberately not unioned with the RUNX2 counts, because AML3 is a contaminated token in a leukaemia setting (FAB M3 shorthand and cell-line naming) and a union would import that contamination into a number about this gene. PEBP2A1[tiab] AND leukemia[tiab] returned 0 but its querytranslation came back as the literal query string rather than a field-tagged expansion, so that number is NOT reported as a bounded absence. The only queries whose records were read are those cited as E3, E4 and E5.

   ▸Evidence that argues against this (1)

   • The RUNX2-plus-KMT2A count of 4 understates the real intersection badly, and its own membership proves it: the four records include the KMT2A::AFF1 enhancer paper and the t(4;11) HDAC7/menin-inhibitor paper, neither of which would be found by a naive reader expecting four marginal hits. A small count here means a narrow indexing surface, not a thin literature - rule 80's point about supporting figures applies with force to a gene this well studied. NCBI PubMed 2026
8. E8strong confidence 4 of 5: Replicated in peer-reviewed work by more than one group.database

   RUNX2 has two promoters ~93 kb apart on GRCh37: the distal P1 at chr6:45,296,464 (ENST00000371436, 499 aa, protein beginning MASNSLFSTVTPCQQNFFWDPSTSR - the type II / Runx2/p57 bone isoform) and the proximal P2 at chr6:45,390,222 and chr6:45,389,914 (ENST00000359524, 507 aa, and ENST00000371432, 485 aa, both beginning MRIPVDPSTSRRFSPPSSSLQPGKM - the type I / Runx2/p56 isoform). The Q118 bin centre chr6:45,390,000 lies 86 bp from the nearest P2 transcription start and 93.5 kb from P1.

   Ensembl GRCh37 REST: lookup/id/ENSG00000124813?expand=1 for per-transcript starts, and sequence/id/<transcript>?type=protein for the encoded N-termini · Ensembl · Ensembl GRCh37 archive 2026
   ▸What this evidence cannot show

   Fetched live 2026-09-23. The P1/P2 and type I/type II assignment is made from the FETCHED protein sequences (MASNS versus MRIPV), not from recall of the nomenclature, because the naming of RUNX2's promoters is inconsistent across the literature and 'distal' and 'proximal' are used in both directions. Transcript-level TSS positions are Ensembl gene models, not experimentally mapped start sites in a leukaemic cell, and a 2 kb bin cannot resolve which of several transcripts within 500 bp is being transcribed.

   ▸Evidence that argues against this (1)

   • A bin 86 bp from the P2 TSS is also 93.5 kb from P1 inside the same gene body, and MLL1-family complexes spread across gene bodies. Concluding 'the signal is at P2' from a single 2 kb bin ignores the rest of the locus; the binned data on disk covers the whole 336 kb gene and has not been profiled. Iqbal Lab oncofusion knowledge graph 2026
9. E9moderate confidence 3 of 5: One strong study, or several smaller consistent ones.paper

   In differentiating MC3T3 mouse osteoblasts, activating histone marks at the Runx2/p57 P1 promoter accompany simultaneous Wdr5 and Utx occupancy, knockdown of either confirms an activating role at P1, Mll2/COMPASS and Mll3/COMPASS-like complexes bind the P1 promoter to modulate the H3K4me1-to-H3K4me3 transition, and the chromatin modifiers that regulate Runx2/p57 in mesenchymal precursors (Ezh2, Prmt5, Jarid1b) do not contribute in fully committed osteoblasts.

   Mll-COMPASS complexes mediate H3K4me3 enrichment and transcription of the osteoblast master gene Runx2/p57 in osteoblasts · Rojas A, Sepulveda H, Henriquez B, Aguilar R, Opazo T, Nardocci G, et al. · J Cell Physiol 2019
   PMID 30256410DOI
   ▸What this evidence cannot show

   Abstract-level read; the full text was not opened this session. MOUSE MC3T3 OSTEOBLASTS, not leukaemia and not human. The complexes named are Mll2/COMPASS and Mll3/COMPASS-like - i.e. KMT2B/KMT2C-family, NOT KMT2A - so this is not a KMT2A result and rule 66's MLL2/MLL4 naming ambiguity applies to reading it. It is cited here only to establish that the osteoblast COMPASS literature is about the P1 promoter, which is the promoter the leukaemia bin is NOT at.

   ▸Evidence that argues against this (1)

   • Because the complexes here are Mll2/Mll3-family rather than KMT2A, this paper cannot be used to argue that KMT2A or a KMT2A fusion acts at Runx2 in any lineage. The stage-dependence it reports - Ezh2, Prmt5 and Jarid1b matter in precursors and not in committed osteoblasts - also means a single RUNX2 promoter result is only valid for the differentiation stage it was measured at, which is the same sign-flipping problem mechanism-menin-has-opposite-jobs-at-two-stages-of-the-osteoblast already records for menin. J Biol Chem 2005

• related to →Menin builds bone early and blocks it late: two opposite jobs in the same lineage

  the osteoblast master gene of that mechanism; this entry adds that the leukaemia bin is at P2 and the osteoblast COMPASS literature is at P1

• ← regulatesKMT2A::AFF1 (MLL-AF4), t(4;11)(q21;q23)

  patient-specific enhancer activity at RUNX2 in KMT2A::AFF1 ALL (Micro Capture-C, primary patients); RUNX2 repressed under MI-538 + chidamide in t(4;11) infant PDX

• regulates →KMT2A / MLL1 — the gene that breaks

  RUNX2 reciprocally binds the KMT2A promoter in KMT2A-R T-ALL, a feed-forward loop with the chimera

• regulates →CXCR4 - named in eighteen entries of this graph with no entry of its own, and the one niche molecule with a drug, a paediatric trial and a direct KMT2A-r experiment

  direct transcriptional regulation of CXCR4, driving chemotaxis, adhesion and homing in KMT2A-R T-ALL

• related to →HDAC7 - the eraser the index fusion silences, where the proposed therapy is to INDUCE the HDAC, and two 2024-2025 papers point in opposite directions

  menin-inhibitor restoration of HDAC7 in t(4;11) infant B-ALL is accompanied by RUNX2 repression and glucocorticoid resensitisation

• expressed in →Osteoblast — the bone-building niche cell whose expansion suppresses CML and makes MLL-AF9 AML worse

  UniProt: specifically expressed in osteoblasts; master regulator of osteoblast maturation and ossification

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

  the axis joining this drug class's leukaemia benefit to its paediatric growth-plate liability; RUNX2[tiab] AND revumenib[tiab] returns 0

• supports →Instrument check: the top thirty RS4;11 MLL1-N 2 kb bins recovered from a deposited .tdf are JMJD1C (259x the genome median), ZMYND11, BAHCC1, CDK6, MYB, RUNX2, IGF2BP2, FOXP1, SATB1 and ARID1B - canonical KMT2A::AFF1 targets, annotated only against the series' own RefGene windows

  the entry that named this gene; this entry resolves its bin to the RUNX2 P2 promoter

• analogous to →CBFB::MYH11 (CBFbeta-SMMHC), inv(16)(p13;q22)

  the RUNT-CBFB heterodimerisation interface targeted pharmacologically in KMT2A-R T-ALL is the interface that CBFB::MYH11 subverts