Oncofusion

start / graph / mechanism-baf-acts-apart-from-the-menin-kmt2a-axis

Mechanismstrong confidence 4 of 5: Replicated in peer-reviewed work by more than one group.Q47-swi-snf

BAF does not work where the fusion works: most direct MLL-AF9 targets do not respond to Brg1 loss, and the SWI/SNF address is a distal MYC enhancer cluster 1.7 Mb away

also called SWI/SNF versus menin-KMT2A loci · E1-E5 Myc enhancers

You might expect the SWI/SNF machine to be helping the MLL fusion protein at the genes the fusion controls. When somebody checked, most of those genes did not care whether SWI/SNF was there - the machine was working somewhere else entirely, on a cluster of switches more than a million bases away from the MYC gene.

The same thing, for a scientist

This is the entry that answers the beat's central question - does BAF act at the same loci as the menin-KMT2A axis - and the answer is no. Shi et al. report that *"the majority of direct MLL-AF9 target genes, however, were not down-regulated upon Brg1 knockdown ... suggesting that MLL-AF9 and Brg1 display overlapping functions at select genes yet are largely distinct from one another in their regulatory activities"* [E1]. What Brg1 does instead is hold open a cluster of five lineage-specific enhancers (**E1-E5**) located **1.7 Mb downstream of Myc**, occupied by Brg1, Brd4, BAF250a and BAF60a; Brg1 is required there for transcription-factor occupancy and for long-range looping to the Myc promoter, and 3C/4C shows contact between E3-E5 and MYC falls on Brg1 loss [E2]. The enhancers are leukaemia-specific: Brg1 occupies the Myc promoter in every cell type examined, but E1-E5 only in RN2, B-ALL and NOMO-1, and H3K27ac is absent there in most normal mouse tissues [E2]. The region coincides with a focal amplification in ~3% of AML [E2]. **Hoxa9, the canonical menin-KMT2A output, was down-regulated by Brg1 loss but significantly less than Myc [E1].** So the two axes touch at a few genes and are otherwise separate - which makes BAF a candidate complement to menin inhibition rather than a parallel route to the same programme.

Go deeper
The address: E1-E5

Five elements 1.7 Mb downstream of Myc, in a region the paper describes as otherwise a gene desert plus Gsdmc, Fam49b and Asap1 - none of which is the regulatory target. 4C-seq anchored near E3 shows preferential contact with Myc while effectively ignoring the nearer genes 2.

Occupancy at E1-E5: Brg1, Brd4, H3K27ac, plus BAF250a and BAF60a (so a SWI/SNF complex, not a lone ATPase), plus haematopoietic transcription factors Lmo2, PU.1, Erg, Cebpa and Cebpb by ChIP-qPCR, with Erg, Fli1, Gfi1b, Meis1 and Runx1 predicted from HPC-7 data 2. Flag-PU.1 immunoprecipitates Brg1 2.

So the mechanism is: haematopoietic TFs + SWI/SNF + BRD4 hold a distal enhancer cluster open, and the cluster loops to MYC. That is a leukaemia-stem-cell transcription-factor circuit, not a KMT2A-fusion circuit.

Why this matters for the menin-interface hypothesis

This graph's central hypothesis is that the menin-KMT2A interface is the load-bearing point of the disease. The SWI/SNF evidence does not threaten it - it sits beside it:

  • Brg1 loss suppresses Myc strongly and Hoxa9 weakly 1. Menin inhibition suppresses HOXA9/MEIS1 as its defining pharmacodynamic readout 3.
  • Most direct MLL-AF9 targets are Brg1-independent 1.
  • The two are therefore non-redundant, which is the precondition for combination rather than for substitution - and that is exactly what the pharmacology later found, with FHD-286 plus a menin inhibitor synergising 34.

The threat to the hypothesis is a different one and is worth naming: if MYC can be sustained from a distal enhancer cluster that the fusion does not control, then a programme that suppresses HOXA9/MEIS1 perfectly may still leave the MYC arm intact. That is a mechanism for menin-inhibitor escape that does not require a MEN1 mutation - and a menin-inhibitor-resistant state with SMARCA4 and BRD4 among its CRISPR co-dependencies has been derived and published 4.

Facts and figures
mechanism classdistal enhancer maintenance and enhancer-promoter looping, largely disjoint from the fusion's own target set
relation to menin axisnon-redundant: menin inhibition suppresses HOXA9/MEIS1, Brg1 loss suppresses MYC via distal enhancers; combination synergises preclinically
clinical correlatethe region coincides with focal copy-number amplification in about 3% of AML
locus
{
  "gene": "MYC",
  "elements": "E1-E5",
  "distance": "1.7 Mb downstream of Myc",
  "assembly_note": "mouse mm8 in the source figures; the human amplification comparison is on hg19"
}
occupancy at elements
[
  "Brg1 (SMARCA4)",
  "Brd4",
  "BAF250a (ARID1A)",
  "BAF60a (SMARCD1)",
  "H3K27ac",
  "Lmo2",
  "PU.1 (SPI1)",
  "Erg",
  "Cebpa",
  "Cebpb"
]
response to brg1 loss
{
  "Myc": "among the most down-regulated genes, p<0.01",
  "Hoxa9": "down-regulated, p<0.01, but less than Myc",
  "direct MLL-AF9 targets": "majority NOT down-regulated",
  "pro-apoptotic genes": "Bid, Btg1, Bmf up, p<0.01 each"
}
How this could bite us back
  • If MYC is sustained from an enhancer cluster the fusion does not control, then perfect menin-axis suppression still leaves a proliferation arm intact - a resistance route that no MEN1 sequencing would detect.
  • Conversely a BAF-directed agent that suppresses MYC without touching HOXA9/MEIS1 leaves the fusion programme intact, so neither agent alone is a complete answer and the combination inherits both toxicity profiles.
  • The E1-E5 enhancers are occupied by normal haematopoietic transcription factors including PU.1 and CEBPA; an intervention that closes them is acting on the normal myeloid circuit, not on a neomorphic structure.
Still open
  1. Do the human orthologues of E1-E5 carry SWI/SNF and BRD4 in infant KMT2A::AFF1 B-ALL? All the enhancer mapping is mouse myeloid plus one human myeloid 3C experiment.
  2. Does menin inhibition change E1-E5 occupancy or MYC looping? If the MYC arm is fusion-independent, it should not - and that would make it a ready-made escape route.
  3. Is the ~3% AML focal amplification of this region associated with SWI/SNF-inhibitor sensitivity, or with menin-inhibitor resistance? The cohorts to ask exist.
Every source behind this entry (4)
  1. E1strong confidence 4 of 5: Replicated in peer-reviewed work by more than one group.paper

    The majority of direct MLL-AF9 target genes were not down-regulated upon Brg1 knockdown; Myc was among the most down-regulated genes and Hoxa9 was down-regulated significantly less; the authors conclude MLL-AF9 and Brg1 overlap at select genes and are largely distinct in their regulatory activities.

    Role of SWI/SNF in acute leukemia maintenance and enhancer-mediated Myc regulation. · Shi J, Whyte WA, Zepeda-Mendoza CJ, Milazzo JP, et al. · Genes Dev 2013
    PMID 24285714DOI
    What this evidence cannot show

    Microarray, three independent Brg1 shRNAs, 96 h dox in mouse RN2 cells; the 'majority of direct MLL-AF9 target genes' statement is reported as data not shown against the Bernt et al. target list. Read in full via PMC3877755. The negative is therefore an author assertion inside a primary paper rather than a displayed figure.

    Evidence that argues against this (1)
    • A 96-hour knockdown timepoint after which cells are already differentiating cannot cleanly separate direct transcriptional effects from the consequences of differentiation, so 'not down-regulated' at that timepoint is a weaker statement than it appears. Cruickshank VA et al. 2015
  2. E2strong confidence 4 of 5: Replicated in peer-reviewed work by more than one group.paper

    Brg1 and Brd4 occupy a cluster of five enhancers (E1-E5) 1.7 Mb downstream of Myc that is occupied only in leukaemia lines; BAF250a and BAF60a are also present; Brg1 is required for transcription-factor occupancy and for looping to the Myc promoter, and E3-E5-MYC contact falls on Brg1 knockdown; the region coincides with focal amplification in ~3% of AML.

    Role of SWI/SNF in acute leukemia maintenance and enhancer-mediated Myc regulation. · Shi J, Whyte WA, Zepeda-Mendoza CJ, Milazzo JP, et al. · Genes Dev 2013
    PMID 24285714DOI
    What this evidence cannot show

    ChIP-seq and ChIP-qPCR in mouse RN2 with 3C-qPCR in human NOMO-1; the enhancer reporter assay was done in K-562 and HEK293T because those were the transfectable lines, so the reporter result is not from an MLL-rearranged cell.

    Evidence that argues against this (1)
    • K-562 carries H3K27ac and Brg1 at E3 but not E1-E2 or E4-E5, so the element set is not uniform even across leukaemia lines and the 'leukaemia-specific cluster' is really a partially shared set. Xiu Y et al. 2018
  3. E3strong confidence 4 of 5: Replicated in peer-reviewed work by more than one group.paper

    Menin inhibitors act by disrupting menin-MLL1 binding and repressing MLL1-fusion target genes including HOXA9 and MEIS1; FHD-286 combined with a menin inhibitor synergistically kills MLL1-rearranged AML cells and improves xenograft survival.

    PMID 38437498DOI
    What this evidence cannot show

    Preclinical combination data from one laboratory; synergy is an in vitro lethality metric plus xenograft burden and survival.

    Evidence that argues against this (1)
    • Synergy between two agents does not establish that their targets act at different loci - it is equally consistent with two ways of hitting one pathway harder. The loci claim rests on the ChIP and expression data, not on the pharmacology. Shi J et al. 2013
  4. E4strong confidence 4 of 5: Replicated in peer-reviewed work by more than one group.paper

    Menin-inhibitor-resistant AML cells derived by repeated SNDX-50469 shocks have an altered epigenome, transcriptome and proteome without menin mutations, and a CRISPR screen identifies BRD4, SMARCA4 and CREBBP among druggable co-enrichments; menin inhibitor plus FHD-286 or plus OTX015 kills those cells and patient-derived menin-inhibitor-resistant cells.

    PMID 41734382DOI
    What this evidence cannot show

    Derived resistance in MV4-11 and OCI-AML3, not samples from patients who progressed on a menin inhibitor; the patient-derived cells tested were ex vivo resistant rather than clinically progressed.

    Evidence that argues against this (1)
    • A CRISPR co-dependency in a resistant line is not proof that the resistant state runs through the E1-E5 MYC enhancers; the link between this entry's loci mechanism and this resistance phenotype is inference, not measurement. Shi J et al. 2013
Connected to