Oncofusion

start / graph / ana-q141-the-order-in-which-the-complex-leaves-fusion-occupied-promoters

Analysismoderate confidence 3 of 5: One strong study, or several smaller consistent ones.Q141-menin-at-fusion-targets

The order in which the complex leaves fusion-occupied promoters, all fifteen channels on one scale

Three hours after the fusion is destroyed, its partners come off its target genes in a definite order: DOT1L first, then menin, then the elongation machinery, then the histone marks - and one mark does not move at all. The fusion itself is essentially all gone, so everything else is only partly following it.

The same thing, for a scientist

Median dVHL/DMSO TSS signal ratio at the authors' MLL::AF9 target genes, 3 h after 500 nM dTAG-VHL, on one common (housekeeping-normalised) scale: | channel | ratio at target genes | ratio at non-occupied | gap (pts) | |---|---:|---:|---:| | HA (the fusion, RAW scale) | 0.010 | 0.611 | - | | DOT1L | **0.401** | 0.856 | 45.5 | | MLL1 N-terminal (fusion + WT) | 0.479 | 0.888 | 40.9 | | **MENIN** | **0.503** | 0.974 | **47.1** | | CDK9 (flagged) | 0.687 | 1.452 | 76.5 | | RNA Pol II gene body | 0.691 | 0.699 | 0.8 | | MLL1 C-terminal (WT only) | 0.703 | 0.625 | -7.9 | | H3K27ac | 0.772 | 0.924 | 15.2 | | RNA Pol II pS2 gene body | 0.775 | 0.785 | 1.0 | | H3K9ac | 0.776 | 0.921 | 14.5 | | RNA Pol II pS5 TSS | 0.779 | 0.967 | 18.8 | | RNA Pol II TSS | 0.801 | 0.906 | 10.5 | | AFF4 | 0.821 | 0.998 | 17.7 | | H3K79me2 | 0.844 | 0.960 | 11.6 | | H3K4me3 | 0.974 | 1.047 | 7.3 | At 24 h H3K79me2 falls to 0.536 and H3K27ac to 0.766 at the same genes, so the histone marks are the late events.

Go deeper
What the ordering does and does not license

It licenses a statement about the order of loss at fusion-occupied promoters in this system: catalytic and adapter partners before the elongation machinery before the marks. It does not license a causal chain, because n=1 per arm and the channels were sequenced in five different batches; two channels separated by 0.05 in this table are not distinguishable. The robust grouping is three tiers: DOT1L/MLL1-Nterm/MENIN around 0.40-0.50, the Pol II and acetylation channels around 0.69-0.82, and H3K4me3 alone at 0.97.

The two rows that are internal controls

MLL1 C-terminal (wild-type MLL1 only) has a negative gap, and H3K4me3 is flat - neither responds to fusion occupancy. The MLL1 N-terminal channel, which sees both proteins, sits at 0.479, i.e. roughly half its signal at target genes is fusion and half is wild-type, which is an internal consistency check on the antibody pair.

Scale and limits

No spike-in exists anywhere in GSE173599, so every number is a within-sample share, and the treated arms are scaled by a 21-gene housekeeping panel whose own raw 3 h H3K27ac ratio averages 0.809 and is forced to 1. A housekeeping normalisation DEFINES the global change to be zero, so only the DIFFERENCE between strata is measurable; a uniform genome-wide loss of any of these proteins would be invisible here. n = 1 per arm for every channel except HA, which has two independent 3 h batches. The cells are HCB1, transformed by KMT2A::MLLT3 (MLL::AF9) - NOT the index fusion of this project, KMT2A::AFF1 - and a prior beat in this corpus found that per-gene ordering transfers between MLLT3 and AFF1 lines while magnitude does not. The HA row is on a RAW scale and the rest are housekeeping-normalised, so the HA row is not comparable in level to the others and is shown only to fix what total removal looks like.

Facts and figures
scriptanalyses/menin-retention-by-ha-occupancy/analyze_tss.py
statusrun
result summaryFifteen channels on one scale at fusion-occupied promoters, 3 h. Three tiers: DOT1L/MLL1-Nterm/MENIN 0.40-0.50; Pol II and acetylation 0.69-0.82; H3K4me3 0.97. HA itself 0.010.
peak width ratio noteSignal instrument. Companion peak width ratios where a peak file exists: MEN1 0.93, DOT1L 1.13, Pol II 1.01, pS2 1.04, pS5 0.93, H3K79me2 0.96, AFF4 1.05, CDK9 2.46 (FLAGGED).
inputs
[
  "GSE173599_TSS_gene_body_signal_Figure3.xlsx",
  "GSE173599_ChIPseq_TSS_Figure3_revisions.xlsx",
  "GSE173599_TSS_signal_HA_ChIPs.xlsx"
]
How this could bite us back
  • The mark this corpus treats as the DOT1L-inhibitor pharmacodynamic readout, H3K79me2, is the slowest responder here (0.844 at 3 h, 0.536 at 24 h), so an early PD timepoint would report a working DOT1L inhibitor as inactive.
  • If menin leaves after DOT1L rather than before it, menin displacement is not the first event in the collapse, and a menin inhibitor is acting upstream of an already-slow chain.
Still open
  1. Does the DOT1L-before-MENIN ordering hold in KMT2A::AFF1, where AFF1 rather than MLLT3 recruits the elongation machinery? A prior beat found ordering transfers between fusion partners while magnitude does not, which predicts it should.
  2. Why does AFF4 barely move (0.821) when CDK9, its partner in the same complex, moves to 0.687? Either the SEC partially disassembles or one of the two antibodies is reporting badly.
Every source behind this entry (3)
  1. E1strong confidence 4 of 5: Replicated in peer-reviewed work by more than one group.dataset

    Median dVHL/DMSO TSS ratios at the authors' MLL::AF9 target genes, 3 h: DOT1L 0.401, MLL1 N-term 0.479, MENIN 0.503, CDK9 0.687, Pol II gene body 0.691, MLL1 C-term 0.703, H3K27ac 0.772, pS2 0.775, H3K9ac 0.776, pS5 0.779, Pol II TSS 0.801, AFF4 0.821, H3K79me2 0.844, H3K4me3 0.974.

    What this evidence cannot show

    n=1 per arm; five sequencing batches; housekeeping-normalised so levels are relative. CDK9's row carries a separate width and normalisation caveat. Differences under ~0.05 are not resolvable.

    Evidence that argues against this (1)
    • The authors report the same ordering qualitatively but by ChIP-qPCR at selected loci for the fast kinetics (DOT1L ~50% lost at 15 min, MENIN only after 3 h), which is a different assay and a different gene set from the genome-wide panel. Olsen SN et al. 2022
  2. E2strong confidence 4 of 5: Replicated in peer-reviewed work by more than one group.dataset

    By 24 h H3K79me2 falls to 0.536 and H3K27ac to 0.766 at the same target genes, confirming the marks as late events.

    What this evidence cannot show

    24 h arms, n=1, and the 24 h libraries are from a different batch (190814) than several 3 h channels.

    Evidence that argues against this (1)
    • The 24 h libraries come from batch 190814 and several 3 h channels do not, so an apparent kinetic ordering across timepoints is partly an ordering across batches. Olsen SN et al. 2022
  3. E3moderate confidence 3 of 5: One strong study, or several smaller consistent ones.dataset

    Wild-type MLL1 and H3K4me3 do not respond to fusion occupancy, which is the internal negative control for the whole ordering.

    What this evidence cannot show

    MLL1 C-terminal sheet holds 1,249 genes and only 69 target genes.

    Evidence that argues against this (1)
    • With only 69 target genes surviving into the MLL1 C-terminal sheet, that control is underpowered, and a negative result from an underpowered control is weak evidence of no effect. Olsen SN et al. 2022
Connected to