Oncofusion

start / graph / exp-h3k9me3-and-ctcf-at-the-kmt2a-anchor-in-a-cd34-progenitor

Experimentmoderate confidence 3 of 5: One strong study, or several smaller consistent ones.Q108-h3k9me3-at-the-bcr

The experiment the public data cannot supply: H3K9me1/2/3 and CTCF at the KMT2A exon-11 anchor in a cord-blood CD34+ progenitor, with and without KDM3B depletion

also called the haematopoietic arm of the H3K9 brake · put the brake in a blood cell

Everything known about this mechanism was measured in eye cells. The experiment that matters is the same measurement in the blood stem cells where infant leukaemia actually starts - and those cells are already the ones the original paper used for its one blood-cell result.

The same thing, for a scientist

The entire chromatin arm of the *KMT2A* rearrangement brake - baseline H3K9me1 depletion at the BCR, the H3K9me1 rise on KDM3B depletion, the CTCF loss at exon 11, and now the H3K9me3 null at that element - is measured in **hTERT-RPE cells only** [E1, E2]. The cytogenetic arm reaches into **cord-blood CD34+ HSPCs**, U937, KG1a, a primary AML line and mouse spleen, but no chromatin was measured in any of them [E2]. **So the whole mechanism rests on a chromatin measurement in epithelium and a phenotype measurement in blood, with nothing joining them.** The experiment: cord-blood CD34+ HSPCs, control and KDM3B siRNA or shRNA, ChIP-seq or targeted ChIP-qPCR for H3K9me1, H3K9me2, H3K9me3 and CTCF at the *KMT2A* exon-11 anchor (GRCh38 chr11:118,488,159-118,489,206, summit chr11:118,488,694) with the *ZNF274* body and the *GAPDH* promoter as mark-identity controls and matched *MLLT3* as a locus-level negative. **Readouts and their meanings are pre-committed below.** The decisive point is that a null H3K9me3 result in RPE - which this beat measured - licenses nothing about a blood cell, and the entry that motivated the whole beat says so in its own bite-back.

Go deeper
Design, with the numbers the RPE data supply as the comparison

Cells. Cord-blood CD34+ HSPCs, the same source the founding paper used for its FISH break-apart result, so the phenotype and the chromatin would finally be in one cell type. A B-cell precursor (CD19+ CD34+, or a KMT2A-rearrangement-relevant fetal liver progenitor) is the more disease-relevant alternative and is harder to obtain; CD34+ cord blood is the version that is actually doable.

Arms. Control siRNA; KDM3B knockdown with two independent reagents (the founding study's si43 and si58 equivalents); optionally G9a knockdown and the double, which is what makes the balance signature testable. Note a gap this beat found: GSE210480 contains no siG9A CTCF arm at all, so the balance shape of the CTCF step has never been measured in any cell type. Including siG9A + CTCF here would close that too.

Readout. ChIP-qPCR is sufficient for the primary question and is the right tier for precious primary cells: amplicons at the summit (chr11:118,488,694), at 118,488,486 and 118,488,886 to bracket it at the 200 bp scale the deposited RPE CTCF data support, at the DCAL cut point (chr11:118,484,862) as an internal negative 3.8 kb away, at the ZNF274 body as an H3K9me3 positive control and at the GAPDH promoter as a negative. ChIP-seq if material allows, because it would also give the anchor-ranking comparison that hyp-the-kmt2a-anchor-is-hypersensitive-not-uniquely-targeted needs.

Comparison values from RPE, to be beaten or contradicted 1: H3K9me1 siKDM3B/control 3.85x at exon 11; H3K9me2 0.44x; H3K9me3 1.04x; CTCF 0.30x.

Pre-committed readings.

result in CD34+meaning
CTCF falls ~0.3x and me1 rises, me3 flatthe mechanism transfers to blood and mark identity holds. The strongest possible support for the published chain and for this beat's null.
CTCF falls, me1 rises, me3 also risesmark identity fails in the cell type that matters, and this beat's RPE null becomes a cell-type-specific curiosity. drug-unc0642's contraindication hardens.
me1 rises but CTCF does not fallthe middle of the chain is broken in blood. This is what Lehman et al. found for CTCF genomic binding under stress in primary mammary epithelium, and it is the single most likely negative.
nothing movesKDM3B is not the relevant demethylase in a progenitor, and the paralogue question (JMJD1C, which ranks 23rd of 17,728 for KMT2A-selective dependency in this project's scan against KDM3B's 9,382nd) becomes primary.
Cost, and why it is worth the primary material

Tier: medium. Cord-blood CD34+ cells, nucleofection or lentiviral knockdown, and four ChIPs per arm. ChIP in primary CD34+ cells at 10^5-10^6 cells is routine in the field with low-input protocols (the founding study's own libraries used 1 ng of input DNA and 14 PCR cycles, so low input is not a novel obstacle here). Weeks, not months. No animals, no cohort, no permissions beyond a cord-blood source.

Why it is worth it. Three entries in this graph and one published mechanism currently join an epithelial chromatin measurement to a haematopoietic phenotype by assumption. This beat added a fourth measurement to the epithelial pile and the pile is now well characterised: the mark rises, the anchor loses CTCF, the trimethyl state does not move, and the resolution floor is 1 kb. Nothing further about the epithelial system is the rate-limiting unknown. The rate-limiting unknown is one cell type away, and it is the cell type the disease starts in.

What would make it cheaper. If any public ChIP-seq exists for H3K9me1 or H3K9me2 in CD34+ cells - ENCODE and BLUEPRINT both carry CD34+ histone panels - then the baseline half of the question (is H3K9me1 depleted at the KMT2A BCR in a progenitor as it is in RPE?) is answerable from public data for the cost of an afternoon, with no wet work at all. That check was not run by this beat and it should be the first thing anyone does, because a baseline that does not reproduce would change the design before any primary material is spent.

Facts and figures
designCord-blood CD34+ HSPCs; arms = control siRNA, KDM3B knockdown with two independent reagents, and (to close a gap this beat found) G9a knockdown and the double. ChIP-qPCR for H3K9me1, H3K9me2, H3K9me3 and CTCF at amplicons tiling the KMT2A exon-11 anchor, plus internal controls. ChIP-seq if material allows.
readoutper-amplicon ChIP enrichment, arm versus control, at: the measured CTCF summit GRCh38 chr11:118,488,694; brackets at chr11:118,488,486 and chr11:118,488,886 (the 200 bp scale the RPE CTCF tracks support); the DCAL functional cut point chr11:118,484,862 as an internal negative 3.8 kb away; the ZNF274 body as an H3K9me3 positive control; the GAPDH promoter as a negative; and MLLT3 as a locus-level negative
controlsZNF274 body (H3K9me3 must be enriched, 3.76x chr11 mean in the RPE tracks) and GAPDH promoter (must not be, 0.31x) as mark-identity controls; MLLT3 as the clean locus-level negative on every channel in the RPE data; the DCAL cut point as a within-locus negative; KDM3B and G9a protein knockdown confirmed by western before any ChIP is believed
cost tiermed
timelineweeks - cord-blood CD34+ isolation, nucleofection or lentiviral knockdown, four ChIPs per arm, qPCR. No animals, no cohort.
cheaper first stepPUBLIC DATA, NOT RUN BY THIS BEAT: ENCODE and BLUEPRINT both carry CD34+ histone panels. If H3K9me1 or H3K9me2 tracks exist for CD34+ cells, the BASELINE half of the question - is H3K9me1 depleted at the KMT2A BCR in a progenitor as it is in RPE? - is answerable in an afternoon with no wet work. This should be the first thing anyone does, because a baseline that does not reproduce changes the design before primary material is spent.
gap it also closesGSE210480 contains no siG9A CTCF arm and no double-knockdown CTCF arm, so the balance signature of the CTCF step - the step that carries the locus specificity - has never been measured in any cell type. Including those arms here would close that.
localizationchromatin at the KMT2A exon-11 CTCF anchor, chr11q23.3, in a haematopoietic progenitor nucleus
subnuclear localizationnot proposed to be measured. If material allowed, a CTCF/cohesin anchor's nuclear position (lamina-associated versus interior) would be the natural addition and is currently unmeasured at this element in any cell type
cell typecord-blood CD34+ haematopoietic stem and progenitor cells - primary, human, haematopoietic; the cell type in which the founding paper showed the cytogenetic phenotype and in which no chromatin has ever been measured at this locus
cell statecytokine-supported ex vivo culture, proliferating, 48-96 h after knockdown; no differentiation, drug or damage arm in the primary design, with a doxorubicin or etoposide arm as the obvious extension
epigenetic readoutH3K9me1, H3K9me2 and H3K9me3 ChIP enrichment plus CTCF ChIP enrichment at the KMT2A exon-11 anchor and five control amplicons
shown for fusionNot shown for any fusion, and deliberately so: the experiment interrogates the intact KMT2A locus in a normal progenitor BEFORE any rearrangement exists, which is the disease stage the founding mechanism addresses. A fusion-bearing arm would answer a different question.
pre committed readings
{
  "ctcf_falls_me1_rises_me3_flat": "the mechanism transfers to blood and mark identity holds; strongest support for the published chain and for this beat's RPE null",
  "ctcf_falls_me1_rises_me3_ALSO_rises": "mark identity fails in the cell type that matters; this beat's RPE null becomes cell-type-specific and drug-unc0642's contraindication hardens",
  "me1_rises_but_ctcf_does_not_fall": "the middle of the chain is broken in blood - the most likely negative, and what Lehman et al. found for CTCF genomic binding under stress in primary mammary epithelium",
  "nothing_moves": "KDM3B is not the relevant demethylase in a progenitor and the JMJD1C paralogue question becomes primary"
}
How this could bite us back
  • A positive result here would strengthen the case for prophylactic G9a inhibition alongside anthracyclines in children - an intervention whose safety rests on n = 3-4 mice per arm with a cytogenetic endpoint and no leukaemia-incidence data. This experiment would make that proposal more credible without making it any better supported at the level that matters.
  • A negative result is the more likely outcome and would be read as 'the mechanism does not apply', when the honest reading of a negative in cord blood would be narrower: CD34+ cord blood is a proxy for a fetal progenitor, and a failure in the proxy does not settle the fetal case.
  • Spending primary human progenitor material to confirm a chain whose output phenotype is already demonstrated in those same cells is arguably the wrong allocation. The counter-argument, and the reason this entry exists, is that the chain's middle step is the one an intervention would act on, and it has never been seen in a blood cell.
Still open
  1. Do ENCODE or BLUEPRINT already carry H3K9me1 or H3K9me2 in CD34+ cells? If so the baseline half of this experiment is free, and this beat did not check. It is the cheapest unrun step in this whole beat.
  2. Is a cord-blood CD34+ cell the right cell, or is a fetal liver progenitor? The index disease is de novo infant t(4;11) with an in-utero origin, and cord blood is the closest routinely obtainable proxy rather than the right cell. Naming the proxy as a proxy is part of the design.
  3. Would a doxorubicin or etoposide arm in the same cells be worth adding? The founding mechanism's clinical relevance runs through topoisomerase II poisons, etoposide is in Interfant-06, and the chromatin consequence of etoposide at this anchor has never been measured in a blood cell either.
Every source behind this entry (2)
  1. E1strong confidence 4 of 5: Replicated in peer-reviewed work by more than one group.dataset

    The RPE comparison values this experiment would be measured against were read directly from the founding paper's deposited data: at Ensembl canonical KMT2A exon 11, siKDM3B/control is 3.85x for H3K9me1, 0.44x for H3K9me2, 1.04x for H3K9me3 and 0.30x for CTCF; and GSE210480 contains no siG9A or double-knockdown CTCF arm, so the balance shape of the CTCF step is unmeasured in any cell type.

    GEO GSE210480 H3K9me1/2/3 and CTCF bigWigs read at interval resolution with pybigtools 0.3.0, 2026-09-22; scripts and full output at analyses/q108-gse210480-h3k9me3-kmt2a-bcr/ · Gray ZH, Chakraborty D, Duttweiler RR, Alekbaeva GD, Murphy SE, Chetal K, Ji F, Ferman BI, Honer MA, Wang Z, Myers C, Sun R, Kaniskan HU, Toma MM, Bondarenko EA, Santoro JN, Miranda C, Dillingham ME, Tang R, Gozani O, Jin J, Skorski T, Duy C, Lee H, Sadreyev RI, Whetstine JR · NCBI GEO (data accompanying Cell 186:4528-4545.e18) 2023
    PMID 37788669DOI
    What this evidence cannot show

    n = 2 per arm, no spike-in, histone tracks 1 kb binned so the 147 bp exon-11 figure is a 1 kb value, CTCF tracks 200 bp, and the CTCF arm is four unique files under eight accessions. These are input-ratio tracks, so the magnitudes are compressed relative to raw coverage and a CD34+ ChIP-qPCR would not be numerically comparable - only the DIRECTION and the relative ordering across amplicons would be. RPE cells.

    Evidence that argues against this (1)
    • Comparing a primary-cell ChIP-qPCR against fold-over-chr11-mean values from an input-ratio bigWig is a comparison across two normalisations and two assay types, and the experiment could produce a real effect that looks numerically unlike these values. The design should therefore be read as pre-committing DIRECTIONS and the ordering of amplicons, not thresholds. Iqbal Lab oncofusion knowledge graph 2026
  2. E2strong confidence 4 of 5: Replicated in peer-reviewed work by more than one group.paper

    In PMID 37788669 every chromatin measurement is in hTERT-RPE cells while the cytogenetic break-apart and copy-gain phenotype is additionally shown in U937, KG1a, a primary AML line, a primary AML organoid, cord-blood CD34+ HSPCs and mouse spleen - with no chromatin measurement in any haematopoietic system, and no sequenced rearrangement junction anywhere in the paper.

    Epigenetic balance ensures mechanistic control of MLL amplification and rearrangement. · Gray ZH, Chakraborty D, Duttweiler RR, Alekbaeva GD, Murphy SE, Chetal K, Ji F, Ferman BI, Honer MA, Wang Z, Myers C, Sun R, Kaniskan HU, Toma MM, Bondarenko EA, Santoro JN, Miranda C, Dillingham ME, Tang R, Gozani O, Jin J, Skorski T, Duy C, Lee H, Sadreyev RI, Whetstine JR · Cell 2023
    PMID 37788669DOI
    What this evidence cannot show

    The system-by-system inventory is taken from this graph's `mechanism-h3k9me1-2-ctcf-balance-gates-kmt2a-rearrangement`, which tabulates it from the paper's figures; this beat independently confirmed only that all 60 deposited sequencing samples are RPE. If a chromatin panel in a haematopoietic cell exists in the paper's supplement without a deposited sample, the inventory would be wrong - but there would then be no public data for it either, which is the operative point for this experiment.

    Evidence that argues against this (1)
    • The cytogenetic result DOES transfer to blood cells - cord-blood CD34+ HSPCs, U937 and KG1a all show the copy gain - so the mechanism's output is demonstrably haematopoietic even if its chromatin input is not. That weakens the case that the cell type is the binding constraint: the phenotype travels, so the chain plausibly travels with it, and someone could reasonably argue this experiment would confirm what is already implied. Iqbal Lab oncofusion knowledge graph 2026
Connected to