Oncofusion

start / graph / genome-kmt2a-exon-11-ctcf-site

Non-codingmoderate confidence 3 of 5: One strong study, or several smaller consistent ones.Q102-h3k9-rearrangement-brake

The CTCF site in KMT2A exon 11 - a conserved insulator element sitting exactly on the functional cut point of the breakpoint cluster region, whose occupancy decides whether the locus rearranges

also called KMT2A exon 11 CTCF site · the BCR CTCF site · KMT2A intron 10/exon 11 CTCF element

Inside the fragile stretch of the KMT2A gene there is one spot where the clamp protein CTCF sits, in every cell type anyone has looked at. That spot happens to be exactly the boundary that decides whether a break produces a cancer-causing fusion protein or a harmless one.

The same thing, for a scientist

A single CTCF binding site in **KMT2A exon 11**, inside the 8.3 kb breakpoint cluster region, with occupancy conserved across multiple ENCODE cell lines and tissues, directly overlapping the strongest KDM3B ChIP-seq peak in the gene [E1]. Three independent facts converge on this one element and no published work connects them. **(1)** Its occupancy is the terminal step of the H3K9me1/2 rearrangement brake: KDM3B loss or doxorubicin raises H3K9me1 here - among the largest H3K9me1 increases at any CTCF site genome-wide - CTCF occupancy falls, and the locus amplifies and breaks; depleting CTCF alone is sufficient and CTCF over-expression prevents the doxorubicin-induced gain [E1]. **(2)** It coincides with the **functional cut point** of the BCR: `breakpoint-kmt2a-major-bcr` names the intron-10/exon-11 boundary as the point that decides whether the fusion retains the PHD1-PHD2-ePHD3 cluster, whose inclusion is incompatible with transformation, and as the B/C subregion boundary with 39.5% of all patient breaks falling in C [E2]. **(3)** In KMT2A-rearranged AML patients' own **normal remission cells**, the KMT2A intron-10/exon-11 region (hg38 chr11:118,488,500-800) carries 2- to 10-fold more DNA breaks than in non-fusion patients, a 3.2-fold group average, and low-dose etoposide raises breakage there in the fusion patients only [E3]. **Whether these are one element or three nearby elements is unestablished** and is a coordinate-overlap check nobody has run.

Go deeper
Three papers, one address, no citation between them on this point
sourcewhat it says about this addresswhat it calls it
Gray et al. 2023 1conserved CTCF occupancy in exon 11; overlaps the KDM3B peak; occupancy falls on KDM3B loss or Dox; CTCF depletion alone causes rearrangement; CTCF over-expression prevents it'the CTCF site within the BCR of KMT2A'
Meyer et al., via breakpoint-kmt2a-major-bcr 2the intron 10 / exon 11 boundary is the functional cut point; subregion B (exon 10 + intron 10, 679 bp) / subregion C (exon 11 - exon 13, 5,026 bp, 39.5% of breaks)'the functional cut point'
Atkin et al. 2021 3hg38 chr11:118,488,500-800, intron 10/exon 11, 2-10 fold more break-prone in KMT2A-rearranged patients' normal cells; etoposide-inducible in those patients only'the KMT2A intron 10/exon 11 region'

Gray et al. cite Atkin et al. (their ref 56) and do not make the connection. gene-ctcf holds the Atkin result and does not mention the CTCF site in exon 11 because the paper reporting it had not reached this graph. breakpoint-kmt2a-major-bcr holds the functional cut point and contains no CTCF content.

Why the convergence is interesting rather than merely tidy. The functional cut point is defined by protein consequence - which domains survive into the chimera - and there is no obvious reason a CTCF motif should sit there. If the element and the cut point are the same base pairs, then the boundary that determines whether a break is oncogenic is also the boundary whose chromatin state determines whether a break happens. That would be a coincidence worth a paper. If they are merely close, the convergence is an artefact of a 5 kb subregion and three different coordinate conventions, and this entry should shrink to a footnote. This beat cannot tell which, because the three sources use exon numbering (three different conventions - see data.numbering_warning), a functional description, and hg38 base positions respectively, and none gives the CTCF motif's own coordinates.

What the element would have to be for each model to hold

For the brake model 1: an insulator whose bound CTCF constrains the locus - by holding a loop, by blocking a replication or transcription conflict, or by excluding whatever machinery produces the copy gain. The prediction is that CTCF occupancy here is protective and higher occupancy means fewer events. Supported by: CTCF depletion alone is sufficient; CTCF over-expression prevents the Dox-induced gain.

For the fragility model [E4, and mechanism-loop-anchor-dsb-hotspots]: a loop anchor, i.e. one of the CTCF/cohesin-bound bases where TOP2-mediated DSBs concentrate and where break frequency scales with transcriptional output. The prediction is that CTCF occupancy here marks breakage, and the Atkin result - CTCF sites preferentially break-prone in the patients who went on to have a KMT2A fusion - is read that way.

A reconciliation that fits all three datasets and has not been tested. A CTCF-bound loop anchor could be both the commonest place for a TOP2 break to occur and the place whose occupancy determines whether that break resolves harmlessly or into an inherited amplification. In that reading, Atkin's break-prone CTCF sites and Gray's protective CTCF occupancy are measuring the numerator and the denominator of the same event, and the apparent contradiction in this graph dissolves. It is testable: map CTCF occupancy and DSBCapture signal at this element in the same cells, with and without KDM3B depletion.

Facts and figures
element classCTCF binding site / candidate insulator, intragenic
geneKMT2A
cytoband11q23.3
numbering warningTHIS GRAPH HOLDS THREE EXON CONVENTIONS FOR THE SAME 8.3 kb INTERVAL: intron 7 - exon 13 (DCAL, breakpoint-kmt2a-major-bcr), exons 5-11 (historic, breakpoint-kmt2a-bcr-architecture), and exons 8-14 (PMID 37788669). 'Exon 11' therefore does NOT unambiguously identify a position across sources. Any coordinate work on this element must start from hg38 base positions. Do not reconcile the conventions silently.
conservationCTCF occupancy at this site was observed across multiple cell lines and tissues in ENCODE (PMID 37788669, Figure 5A). No orthologue-level sequence conservation analysis was performed by that paper or by this beat.
why it matters if it is the cut pointCOUNTERFACTUAL AS OF CYCLE 57 - IT IS NOT THE CUT POINT, so this field no longer states a live rationale and is kept to record what was believed. The argument ran: the cut point is defined by protein consequence, so an element that both gates breakage and sits at the boundary deciding whether the chimera is oncogenic would be the most important 300 bp in this graph's genesis model. Measured, it is 3,752 bp centromeric of the element. What replaces it is a weaker but still live argument: the element is the SAME 1 kb as the patient break window, is cohesin-bound at maximum score, and sits in the t-AML-skewed telomeric half - so it is a candidate for where breaks START, not for what makes the resulting chimera oncogenic.
the check that settles itA coordinate overlap: lift the CTCF ChIP-seq peak from GSE210480 (deposited, 4 CTCF samples plus inputs) onto hg38, place the intron-10/exon-11 boundary and the chr11:118,488,500-800 window on the same axis, and read off whether they overlap. See ana-kmt2a-bcr-ctcf-site-coordinate-overlap.
absence classmeasured-not-analysed
position as stated by each source
{
  "PMID 37788669": "exon 11, within the 8.3 kb breakpoint cluster region (exons 8-14 in that paper's numbering). NO COORDINATES GIVEN.",
  "breakpoint-kmt2a-major-bcr": "the intron 10 / exon 11 boundary, the B/C subregion boundary of BCR1 (intron 7 - exon 13, DCAL numbering)",
  "PMID 34405474 via gene-ctcf": "hg38 chr11:118,488,500-800, described as the intron 10 / exon 11 region - a 300 bp scored window, not a motif position"
}
occupancy dependencies
{
  "raises_H3K9me1_here_and_lowers_CTCF": [
    "KDM3B depletion (siRNA x2)",
    "KDM3-family chemical inhibition (JDI-12)",
    "doxorubicin",
    "G9a/EHMT2 over-expression (inferred - it causes the phenotype; the mark was not measured in that arm)"
  ],
  "resets_H3K9me1_here": [
    "G9a/EHMT2 co-depletion (complete reset)",
    "dual G9a/GLP inhibition with UNC0642"
  ],
  "prevents_the_downstream_event": [
    "CTCF over-expression (copy gain only - break-apart unassessable because CTCF over-expression itself separated the FISH probes)",
    "KDM3B over-expression",
    "siG9a",
    "UNC0642"
  ]
}
what is NOT known about this element
[
  "its CTCF motif ORIENTATION - still unknown, and it matters if this is a loop anchor",
  "the peak SUMMIT - ENCODE gives cluster intervals, and GSE210480 deposits coverage bigWigs with no peak calls, so no summit is obtainable from either without peak calling raw signal",
  "what loop, TAD or boundary it forms, and with which partner anchor",
  "whether it is occupied in ANY haematopoietic cell - every occupancy statement here is from epithelial or aggregated ENCODE data",
  "ANSWERED cycle 57: its GRCh38 coordinates (chr11:118,488,159-118,489,206)",
  "ANSWERED cycle 57: cohesin occupancy - RAD21 and SMC3 both at cluster score 1000",
  "ANSWERED cycle 57: whether it is the same element as the patient break window - yes, 0 bp",
  "ANSWERED cycle 57: whether it is at the functional cut point - NO, 3,752 bp away"
]
rival readings
{
  "protective_insulator": "PMID 37788669 - CTCF occupancy prevents copy gain and rearrangement",
  "fragility_marker": "PMID 31202576 and PMID 34405474 - CTCF/cohesin-bound loop anchors are where TOP2 breaks concentrate, and CTCF sites are preferentially break-prone in patients who developed a KMT2A fusion",
  "reconciliation_not_yet_tested": "a CTCF-bound loop anchor could be both the commonest place for a break AND the element whose occupancy decides whether that break resolves into an inherited amplification - numerator and denominator of one event",
  "adjudicated_cycle_57": "The protective-insulator and fragility-marker readings are about the SAME element and share no readout: the fragility side maps DSBs and never reports a rearrangement here, the protective side reports FISH rearrangement and never measures a DSB. See concept-ctcf-at-kmt2a-fragile-and-guarded-is-one-element and hyp-ctcf-occupancy-splits-break-formation-from-rearrangement-fate."
}
coordinates measured cycle 57
{
  "the element": "GRCh38 chr11:118,488,159-118,489,206, an ENCODE CTCF+RAD21+SMC3 cluster, 1,048 bp",
  "canonical exon 11 within it": "chr11:118,488,614-118,488,760 (ENST00000534358), entirely inside",
  "cohesin": "BOUND, and at maximum cluster score: CTCF 1000, RAD21 1000, SMC3 1000, reproduced in GRCh37. This answers 'whether cohesin (RAD21/SMC3) is bound' - which this entry listed as not known - and it puts the element in Canela's CTCF+RAD21 class, the one reported as 48-fold enriched for DSBs.",
  "same element as the patient break window": "YES. Atkin's chr11:118,488,500-800 (PMID 34405474) contains all 147 bp of canonical exon 11. 0 bp apart.",
  "NOT at the functional cut point": "The DCAL cut point is canonical intron 9/exon 10 at chr11:118,484,861|118,484,862, which is 3,752 bp centromeric of canonical exon 11's start.",
  "position within the BCR": "70.4% along the Broeker BCR - the telomeric, t-AML-skewed half; inside DCAL subregion C",
  "a second, weaker CTCF site nobody mentions": "score 514 with RAD21 196, 147 bp from canonical exon 10 - which is where a DCAL reading of 'exon 11' would land. Any future coordinate claim here must say which of the two sites it means.",
  "caveat_on_all_of_it": "ENCODE cluster scores are aggregated across cell lines and are NOT a measurement in any haematopoietic cell, and a cluster interval is not a summit. The occupancy call is about the element in general; no infant or B-precursor CTCF ChIP exists at it."
}
How this could bite us back
  • Any intervention aimed at restoring CTCF occupancy at this element would, under the rival published model, be increasing occupancy at a TOP2B-mediated break hotspot. The two models prescribe opposite manipulations of the same protein at the same address, and neither has been tested in the cell type that matters.
  • CTCF is close to pan-essential - this project's DepMap scan gives -1.827 across all lines - so every CTCF perturbation in this literature, including the depletion that is 'sufficient' to cause the phenotype, moves a great deal besides this element. A locus-specific conclusion from a genome-wide knockdown is the weakest link in the brake model.
  • This element is inside a coding exon of a gene that must be expressed. Any attempt to stabilise it - a dCas9 block, a small molecule, an oligonucleotide - is being placed inside the transcription unit of KMT2A itself, where it risks interfering with the very gene it is protecting.
Still open
  1. What are this element's hg38 coordinates? No source gives them. Until they exist, three of this entry's four evidence items cannot be placed relative to each other and the central claim of the entry is a coincidence of exon labels.
  2. Is CTCF bound here in a fetal or cord-blood haematopoietic progenitor? Every occupancy observation is in RPE or ENCODE lines. If CTCF does not occupy this site in the cell of origin, the brake does not exist there.
  3. Is cohesin bound here? No RAD21 or SMC3 ChIP was performed at this element, and the distinction between a CTCF-only site and a CTCF/cohesin loop anchor is exactly what separates the protective reading from the fragility reading.
  4. What does the loop look like? Nobody has run Hi-C, Micro-C or a 4C anchored on this element in any cell type, so 'insulator' is an inference from a CTCF peak.
Every source behind this entry (4)
  1. E1strong confidence 4 of 5: Replicated in peer-reviewed work by more than one group.paper

    A highly conserved CTCF occupancy site lies within exon 11 of KMT2A, inside the 8.3 kb breakpoint cluster region, observed across multiple cell lines and tissues in ENCODE and directly overlapping the strongest KDM3B ChIP-seq peak in the gene; KDM3B depletion raises H3K9me1 at this site - among the strongest H3K9me1 changes of all CTCF sites genome-wide - and significantly reduces CTCF occupancy there by ChIP-seq and ChIP-qPCR without changing total CTCF protein; doxorubicin reproduces both changes; depleting CTCF alone is sufficient to generate KMT2A copy gains and break-apart events and is not enhanced by KDM3B co-depletion; G9a co-depletion completely resets H3K9me1 at this site; and CTCF over-expression prevents doxorubicin-induced KMT2A copy gains.

    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; et al. · Cell 2023
    PMID 37788669DOI
    What this evidence cannot show

    **Full text read 2026-09-22 via efetch db=pmc on PMC10591855 (Europe PMC fullTextXML HTTP 500; <body> confirmed, 80,638 stripped characters).** All CTCF and H3K9 measurements are in RPE, an immortalised near-diploid epithelial line - **not a haematopoietic cell**. **The paper gives no coordinates for the CTCF site**, only 'exon 11 within the BCR', which is why the overlap with the Atkin coordinates cannot be settled from the text; it also gives no motif orientation, no cohesin (RAD21/SMC3) ChIP at the site, and no loop or Hi-C data showing what the site anchors. CTCF antibody clone D31H2 (CST #3418). The ChIP-qPCR primers given in the methods are for the KMT2A CTCF FLANKING site, not the site itself. CTCF over-expression 'appears to alter local chromosomal organization at the locus, with increased separation observed between red and green probe', so the break-apart readout could not be assessed in that arm - the over-expression rescue is a copy-gain result only. CTCF is close to pan-essential (this project's DepMap scan: -1.827 across all lines), so a CTCF knockdown perturbs a great deal besides this element.

    Evidence that argues against this (1)
    • The rival published model makes CTCF/cohesin-bound loop anchors the sites where TOP2-mediated DSBs concentrate, with break frequency scaling with transcriptional output, and places the recurrently translocating leukaemia genes including KMT2A at loop anchors. On that account this element is a break hotspot marked by CTCF rather than an element protected by it. Gothe HJ; Bouwman BAM; Gusmao EG; Piccinno R; Petrosino G; Sayols S; et al. 2019
  2. E2established confidence 5 of 5: Multiple independent lines of evidence; treated as settled in the field.database

    The KMT2A major breakpoint cluster region spans intron 7 to exon 13 (~8.3 kb, DCAL exon numbering) and carries 3,336 of 3,401 (98%) of genomically characterised KMT2A rearrangements; within it, breaks partition into subregion A (exon 9 + intron 9, 1,761 bp, 37.2%), B (exon 10 + intron 10, 679 bp, 19.8%) and C (exon 11 through exon 13, 5,026 bp, 39.5%); and the FUNCTIONAL cut point is the intron 10 / exon 11 boundary, because exons 11-16 encode the PHD1-PHD2-ePHD3 cluster whose inclusion is incompatible with transformation.

    What this evidence cannot show

    **Imported from this graph's own entry rather than re-read from the primary source**, which is the correct provenance to state and the reason this item is not stronger. The exon numbering is DCAL convention; this graph holds two other conventions for the same 8.3 kb (see data.numbering_warning), so 'exon 11' in this item and 'exon 11' in E1 are not guaranteed to be the same exon. The URL points at PMID 37019990, 'The KMT2A recombinome of acute leukemias in 2023', which is the primary source breakpoint-kmt2a-major-bcr E1 cites for the interval and the subregion percentages; the subregion percentages should be checked against that paper before being quoted alongside the CTCF result, because the coincidence this entry is about depends entirely on the numbering matching.

    Evidence that argues against this (1)
    • Subregion C is 5,026 bp and carries 39.5% of breaks, so 'the CTCF site is at the functional cut point' may be a statement about being somewhere in a 5 kb window rather than about coincidence at base-pair resolution. Until both elements are placed on hg38 coordinates the convergence is suggestive and not evidential. Iqbal Lab oncofusion knowledge graph 2026
  3. E3strong confidence 4 of 5: Replicated in peer-reviewed work by more than one group.paper

    Low-input DSBCapture on NORMAL (non-blast) cells from AML patients in remission found breakage at the KMT2A intron 10 / exon 11 region, hg38 chr11:118,488,500-800, to be 2- to 10-fold higher in each KMT2A-rearranged patient than in each non-fusion patient with a significant 3.2-fold group average, while genome-wide median breakage did not differ between any samples; the same study identified 404 CTCF, 517 RAD21, 256 SMC3 and 304 TOP2-sensitive sites preferentially break-prone in the fusion-driven patients, and low-dose etoposide ex vivo raised breakage at these regions in the KMT2A-rearranged patients only.

    PMID 34405474DOI
    What this evidence cannot show

    **Imported from `gene-ctcf` E2, where it is recorded in detail; not re-read for this entry.** n = 3 patients per arm, adult AML, myeloid lineage, normal remission cells. The 300 bp window given is the region scored, not a CTCF motif position, and the study does not report whether a CTCF peak falls inside that window in those cells. The comparison is between patients who did and did not develop a KMT2A fusion, so an elevated break rate there is a correlate of having had the disease and not necessarily a cause. This item is cited here for a coordinate coincidence, which is a lead and not a finding.

    Evidence that argues against this (1)
    • The same study contains a partner-side negative that this graph already records: none of the three patients' own partner genes appeared among the 34 shared break-prone regions, and only the 3' end of AFDN - not its breakpoint region - carried a break-prone CTCF site. So CTCF-site breakage is not a general explanation for where KMT2A fuses. Atkin ND; Raimer HM; Wang Z; Zang C; Wang YH 2021
  4. E4strong confidence 4 of 5: Replicated in peer-reviewed work by more than one group.paper

    TOP2-mediated double-strand breaks are enriched at chromatin loop anchors with high transcriptional activity, occurring recurrently at CTCF- and cohesin-bound sites at the bases of loops with frequency correlating positively with transcriptional output and directionality, and the genes that recurrently translocate to drive leukaemias are highly transcribed and enriched at loop anchors.

    Spatial Chromosome Folding and Active Transcription Drive DNA Fragility and Formation of Oncogenic MLL Translocations. · Gothe HJ; Bouwman BAM; Gusmao EG; Piccinno R; Petrosino G; Sayols S; Drechsel O; Minneker V; Josipovic N; Mizi A; et al. · Mol Cell 2019
    PMID 31202576DOI
    What this evidence cannot show

    **Abstract only for this entry**; this paper is cited in 12 entries of this graph and read there. Recorded here as the rival reading of what this element is, not as an account of its own content. Its cells and lineage are not stated in this item and should be taken from the entries that read it.

    Evidence that argues against this (1)
Connected to