Oncofusion

start / graph / concept-ctcf-at-kmt2a-fragile-and-guarded-is-one-element

Explainermoderate confidence 3 of 5: One strong study, or several smaller consistent ones.Q105-ctcf-adjudication

Adjudicated: CTCF does not do opposite things at KMT2A - it does one thing, and this graph's contradiction is an untested premise about whether a break becomes a rearrangement

also called the CTCF adjudication · fragile and guarded are not incompatible · the formation-versus-fate triad at the KMT2A anchor · Q105 adjudication

One model says the spot where CTCF sits in the KMT2A gene is where the DNA snaps; another says CTCF sitting there is what stops the gene rearranging. Both can be true at once, because nobody has ever checked whether a snap at that spot is more likely to turn into a cancer-causing rearrangement when the protein is gone.

The same thing, for a scientist

This graph holds two published models that appear to make CTCF opposite things at the *KMT2A* locus. The adjudication has three parts. **First, they are about the same DNA.** In genomic coordinates, Gray et al.'s protective exon-11 CTCF site and Atkin et al.'s break-prone patient window are one element - a CTCF+RAD21+SMC3 co-occupied loop anchor at GRCh38 chr11:118,488,159-118,489,206, spanning Ensembl canonical exon 11 - so the dispute cannot be dissolved by saying the two sides mean different places [E1]. **Second, they answer different questions and share no readout.** The fragility side measures where double-strand breaks form (END-seq, sBLISS, DSBCapture) and never measures an inherited rearrangement at this element; Gray et al. measure FISH copy gain and probe break-apart and never measure a double-strand break, mentioning neither DSBs nor cohesin anywhere in the paper [E2][E3][E4]. **Third - and this is the result - the two published propositions are not formally contradictory. They become contradictory only when joined to a third, which everybody assumes and nobody has tested at this element:** (i) occupancy up, DSBs up [E3]; (ii) occupancy up, rearrangement down [E4]; (iii) DSBs up, rearrangement up. **All three cannot hold, and (iii) is the weakest.** It is not a law - it has already been falsified at this very locus for a different factor: TDP2 loss *increases* TOP2-induced translocations at *MLL* without increasing breakage, because TDP2 governs how a break is rejoined rather than whether it forms, and TDP2-dependent end-joining even *creates* a specific therapy-related translocation subclass [E5]. **So the leading reconciliation, and the beat's substantive claim, is that CTCF/cohesin occupancy at this anchor is both the cause of the break and the chaperone of its correct repair.** Lose it and you get fewer breaks, each far likelier to end as an inherited rearrangement - net more rearrangements from less breakage, which is precisely what both sides measured, each seeing only its own half. **The settling measurement is a single experiment** (`exp-ctcf-degron-breaks-and-rearrangements-same-cells-kmt2a`): degron-grade CTCF removal versus cohesin removal in one human cell system, with DSB mapping, copy-gain/break-apart and junction sequencing at the same element in the same cells, reported as rearrangements *per break*.

Go deeper
What each side actually measured
Canela 2017 [E3]Canela 2019 [E3]Gothe 2019 [E6]Atkin 2021 [E2]Gray 2023 [E4]
systemactivated mouse B cells, mouse T cells, human lines; M. musculus vs M. spretusMEFs (SMC3 CRE), human HCT-116 (RAD21 degron)human leukaemia lineshuman AML patients' normal, non-blast remission cells, n = 3 vs 3, adulthuman RPE (immortalised, near-diploid), extended to KG1a, primary CD34+ HSPCs, mice
perturbationetoposide; natural CTCF-motif variation between speciesetoposide; acute cohesin depletionetoposide; TOP2A/TOP2B loss; elongation inhibitionlow-dose etoposide ex vivo; no CTCF perturbationsiKDM3B, siCTCF, siG9a, JDI-12, UNC0642, doxorubicin, etoposide, overexpression rescues
readoutEND-seq DSB maps, Hi-C loop anchors, CTCF/RAD21/TOP2B ChIPEND-seq; TOP2cc trapping and conversionTOP2 DSB maps; MLL fusion formationlow-input DSBCapture DSB mapsDNA FISH copy gain and break-apart; locus qPCR; CTCF and H3K9me1/2 ChIP
question answeredwhere do breaks form, and does occupancy predict itdoes break formation require the architecturedoes break formation require transcription, and do fusions followis break-proneness at this locus a constitutional trait of people who developed a fusiondoes the chromatin set point control whether the locus amplifies and rearranges
about DSB formation?yesyesyesyesno - not measured at all
about rearrangement outcome?nonopartly (fusion formation, not at this element)noyes - copy gain and break-apart, inherited at 72 h

Read the bottom two rows together. The fragility literature is a literature about formation. Gray et al. is a paper about outcome. The single overlapping quantity - breaks at this element under a CTCF perturbation - has never been measured by anybody, which is why five papers can disagree without a single conflicting number between them.

One asymmetry worth naming, because it cuts against the index disease. Atkin's measurement is the only one made in human patients with this disease, and it is in myeloid adult AML remission cells. Gray's chromatin measurements are all in epithelium. Canela's directional experiments are in mouse B cells, MEFs and a colorectal line. No CTCF occupancy or breakage measurement at this element exists in a B-lineage cell, in an infant, or in a KMT2A::AFF1 context - and this graph's index disease is de novo infant t(4;11) with an in-utero origin and no prior anthracycline.

Compatibility, tested rather than asserted

This graph already listed three ways both could be true. With the coordinates and the readouts in hand, two of them collapse and one survives and sharpens.

Route 1 - 'they mean different places'. DEAD. Refuted in coordinates: Gray's site and Atkin's window are 0 bp apart and the window lies wholly inside the CTCF cluster 1. This was worth killing: it was the cheapest exit and it is not available.

Route 2 - 'occupancy versus the motif'. WEAKENED but alive in a specific form. The worry was that a ChIP-based hotspot map cannot distinguish a CTCF site from a CTCF-occupied site. ENCODE clusters this element at maximum score for CTCF, RAD21 and SMC3 in two assemblies 1, so in the cells ENCODE sampled it is occupied, and Canela's fragility class is CTCF+RAD21 co-occupied sites at 48-fold DSB enrichment 3. The form that survives is cell-type-specific: nobody has shown the element is occupied in a haematopoietic progenitor, and Lehman et al. show that CTCF's chromatin binding can be globally stable while a distinct, stress-labile, speckle-localised pool is destroyed 7. So 'occupied' is condition-dependent in a way neither side controls for.

Route 3 - 'different quantities'. THIS IS THE ANSWER, and it is stronger than it looked. It was offered as a possibility. It is now the only reading consistent with all five papers, and it has an existing precedent at this exact locus 5.

Route 4, which was not on the list and should have been - 'one of them is measuring cohesin'. Canela's acute directional experiment depletes SMC3 or RAD21, not CTCF 3. CTCF depletion does not remove cohesin from chromatin so much as release it from boundary positions. So there is no published head-to-head: nobody has removed CTCF and measured breaks at an anchor, and nobody has removed cohesin and measured rearrangement. The two arms of the apparent contradiction perturb two different proteins.

The genuine incompatibility, as a falsifiable triad

State it as three propositions about the element at GRCh38 chr11:118,488,159-118,489,206. Any two are compatible. All three cannot be true.

propositionstrongest supportstrongest weakness
(i)At this anchor, TOP2-mediated DSB rate is an increasing function of CTCF/cohesin occupancy. Remove occupancy and breaks fall.acute SMC3/RAD21 depletion drops END-seq signal genome-wide, proportionally - 27% RAD21 loss, 26% END-seq loss - and CTCF/RAD21 co-occupied sites are 48-fold DSB-enriched 3the acute arm is cohesin, not CTCF; the CTCF arm is a two-species natural-variant comparison; nothing was measured at this human element
(ii)At this anchor, the rate of inherited focal amplification and rearrangement is a decreasing function of CTCF occupancy. Remove CTCF and events rise.siCTCF alone is sufficient, KDM3B co-depletion adds nothing, CTCF overexpression rescues the Dox-induced gain 4one siRNA arm in one epithelial line against a near-pan-essential protein (this project's DepMap scan: -1.827 across all lines); no DSB measured; no haematopoietic occupancy data
(iii)At this anchor, rearrangement rate is an increasing function of DSB rate. Fewer breaks means fewer rearrangements.the field's default premise; Atkin's whole design rests on it, and it justifies using break-proneness as a susceptibility biomarker 2never tested at this element, and already falsified at this locus for another factor: TDP2 loss raises TOP2-induced translocations at MLL by changing how breaks are rejoined, not how many form 5

Falsify (iii) and everything reconciles. If CTCF/cohesin occupancy is simultaneously what invites TOP2 to cut and what holds the cut ends in register for faithful religation, then removing it lowers the numerator and raises the conversion ratio, and the net direction is an empirical question with no a priori sign. Canela 2019 already separates the two steps mechanistically - trapping of the cleavage complex is one process, and its conversion to an irreversible break is another, requiring proteasomal degradation and TDP2 phosphodiesterase activity and running about two-fold faster at transcribed loci 3. Gómez-Herreros et al. then show at MLL specifically that the rejoining step, not the cutting step, is what decides whether a translocation appears 5. The machinery for 'more rearrangements from fewer breaks' is published, at this gene, and neither side of this dispute invokes it.

Falsify (i) instead and the reconciliation is duller but real: if CTCF removal redistributes rather than removes cohesin at this anchor, breaks may not fall at all, and (i) simply does not apply to a CTCF perturbation.

Falsify (ii) and Gray is wrong, on the strength of a single knockdown of a pan-essential protein. Nobody should be embarrassed to hold this, and it is the cheapest of the three to test - a second, independent CTCF perturbation with a rescue.

The measurement that settles it

One experiment, three arms, one element, one cell system, both readouts in the same cells. Written out in exp-ctcf-degron-breaks-and-rearrangements-same-cells-kmt2a; the logic here.

System. A human cell in which the anchor is occupied and the locus is competent to rearrange. The honest first choice is a degron-tagged line for tractability plus a primary CD34+ arm for relevance, because the lineage gap is this graph's commonest documented synthesis failure and every chromatin measurement on the protective side is currently in epithelium.

Perturbation. Not siRNA. Auxin-inducible degron on CTCF, and separately on RAD21, with matched depletion kinetics - the same system Canela 2019 used for RAD21 3, so the two literatures share an instrument. Add a doxorubicin or etoposide arm, and a CTCF re-expression rescue.

Readouts, all at chr11:118,488,159-118,489,206, all at the same timepoints.

  1. DSBs: END-seq or low-input DSBCapture, so the number is on the same scale as Canela's and Atkin's.
  2. CTCF, RAD21, SMC3 and TOP2B occupancy by ChIP-qPCR at the anchor, to confirm what was actually removed - the step that separates route 4 from route 1.
  3. Rearrangement outcome: FISH break-apart and locus copy number, as Gray measured, plus junction sequencing so that an event is a sequence and not a probe separation.
  4. The derived quantity nobody has ever reported: rearrangements per break at this element.

What each outcome means.

CTCF degron resultreading
breaks down, rearrangements up, ratio up(iii) is false. Occupancy causes the break and guards its repair. Both literatures are right and this graph should hold one mechanism, not two.
breaks unchanged, rearrangements up(i) does not apply to CTCF. The fragility model is a cohesin model; CTCF's role here is repair-side only. Check the RAD21 arm to confirm breaks fall there.
breaks up, rearrangements up(i) is false as written - occupancy is protective at both steps at this element, and the 48-fold class statement does not generalise to it. Gray wins outright.
breaks down, rearrangements down(ii) is false. Gray's siCTCF result does not survive a degron, and the likeliest explanation is off-target or pan-essentiality effects of knocking down CTCF.
nothing moves because the anchor is unoccupied in this cellthe whole dispute is epithelium-specific and the first thing to publish is the occupancy map in a haematopoietic progenitor.

Cost and tractability. Degron lines exist; END-seq, DSBCapture, FISH break-apart and junction sequencing are all established; the element is 1 kb and needs no genome-wide analysis to read out. This is a medium-cost, single-lab experiment that would decide a question five papers in Cell, Molecular Cell and Genes Chromosomes Cancer have left open, and the reason it has not been done is that no group owns both readouts.

What this changes in the graph, stated as requests to the coordinator

This beat owns none of the entries below and has edited none of them.

  1. mechanism-h3k9me1-2-ctcf-balance-gates-kmt2a-rearrangement - the section 'The contradiction with the graph's existing CTCF layer, which is real and unresolved' should be retitled and rewritten around the triad. Its reconciliation route 1 ('different quantities') is now the answer rather than one of three, and route 3 ('one of them is wrong about direction') should name which proposition. The sentence 'THE CTCF SITE IS AT THE FUNCTIONAL CUT POINT OF THE BCR' is false and should read '3,752 bp telomeric of the functional cut point, inside subregion C and inside the telomeric t-AML-skewed half' 1. The sentence 'PMID 37788669 cites both Gothe and Atkin (its refs 57 and 56) and addresses neither conflict' has the numbers right but should become 'cites both in one direction-free sentence, declares the direction unresolved, and does not cite either Canela paper' 8.
  2. genome-kmt2a-exon-11-ctcf-site - four of the seven items in data.what_is_NOT_known_about_this_element are now known: the hg38 coordinates, cohesin occupancy, identity with the break-prone window, and which half of the BCR 1. data.rival_readings.reconciliation_not_yet_tested should be upgraded to the triad, and data.why_it_matters_if_it_is_the_cut_point should be rewritten because it is not the cut point.
  3. ana-kmt2a-bcr-ctcf-site-coordinate-overlap - data.status 'planned' → 'run'; its pre-registered most-likely outcome did not occur; data.secondary_test is answered affirmatively.
  4. gene-ctcf - data.subnuclear_localization records two measured compartments and should record a third: the SC-35 nuclear-speckle pool, stress-labile, proteasome-dependent, present in primary human mammary epithelium and at a specific stage of neuronal commitment but not in differentiated neurons, PMID 33411704 7. The entry already cites the paper nowhere and the field's own bounded-absence block counts seven speckle records. Also data.patient_break_mapping.kmt2a_region should gain the element's identity: this window is a CTCF+RAD21+SMC3 anchor spanning canonical exon 11.
  5. breakpoint-kmt2a-major-bcr - data.exon_convention is null; it should record DCAL = Ensembl canonical + 1, with the three base-pair identities that establish it, so no future beat re-derives it or reconciles it by eye 1.
  6. mechanism-loop-anchor-dsb-hotspots - this entry already holds the cohesin-dependence result as E3 and is the graph's best statement of the fragility side. It should gain the pointer that its model has a directional rival at one specific element, and the note that Canela's mouse Kmt2a exon labels are a fourth numbering frame.
  7. A cross-cutting rule candidate: when two entries disagree about the direction of an effect, check whether they share a readout before writing the contradiction down. Here they do not, and the contradiction was recorded for a cycle as though they did.
Facts and figures
leading reconciliationCTCF/cohesin occupancy at this anchor is both the cause of the TOP2 break and the chaperone of its faithful rejoining. Removing it lowers the number of breaks and raises the fraction that end as inherited rearrangements; the net direction is empirical and has no a priori sign. This predicts that a CTCF degron gives FEWER breaks and MORE rearrangements at the same element - a rise in rearrangements per break - and it is the only reading consistent with all five papers.
settling measurementexp-ctcf-degron-breaks-and-rearrangements-same-cells-kmt2a
lineage warningNo CTCF occupancy or breakage measurement at this element exists in a B-lineage cell, in an infant, or in a KMT2A::AFF1 context. Atkin's is adult myeloid patient material; Gray's chromatin work is epithelial; Canela's directional work is mouse B cells, MEFs and human colorectal. The index disease of this graph is de novo infant t(4;11) with an in-utero origin and no prior anthracycline, and the mechanism under dispute is a therapy-induced one. Stating that in the claim and not only the caveat is the point.
adjudication
{
  "question": "Does CTCF occupancy at the KMT2A breakpoint cluster region promote breakage or prevent rearrangement?",
  "verdict": "Both published claims can be true. They are not about the same quantity, they share no readout, and they conflict only through a third premise - that rearrangement rate follows break rate at this element - which has never been tested here. **CORRECTED CYCLE 58: it has NOT been falsified at this locus for TDP2, and this field previously said it had.** PMID 28794467's abstract, re-fetched by the king, states that TDP2 *rejoins* TOP2-induced DSBs at the *MLL* hotspot and *suppresses* and *reduces* TOP2-induced translocations. It makes no claim that breakage is unchanged, and failure to rejoin leaves breaks persisting - so the result is CONSISTENT with the third premise in its persistence reading. See E_k58_tdp2_correction.",
  "same_element": true,
  "same_readout": false,
  "formally_contradictory": false,
  "contradictory_given_the_standard_auxiliary_premise": true
}
falsifiable triad
[
  {
    "label": "i",
    "statement": "At GRCh38 chr11:118,488,159-118,489,206, TOP2-mediated DSB rate increases with CTCF/cohesin occupancy.",
    "best_support": "PMID 31202577 acute SMC3/RAD21 depletion drops END-seq proportionally; PMID 28735753 CTCF+RAD21 sites 48-fold DSB-enriched",
    "best_objection": "the acute arm is cohesin not CTCF, and nothing was measured at this human element"
  },
  {
    "label": "ii",
    "statement": "At the same element, inherited focal amplification and rearrangement rate decreases with CTCF occupancy.",
    "best_support": "PMID 37788669 siCTCF sufficient, CTCF overexpression rescues Dox-induced gain",
    "best_objection": "one siRNA arm in one epithelial line against a near-pan-essential protein; no DSB measured"
  },
  {
    "label": "iii",
    "statement": "At the same element, rearrangement rate increases with DSB rate.",
    "best_support": "the field's default premise, and the basis of using break-proneness as a susceptibility biomarker (PMID 34405474)",
    "best_objection": "never tested at this element. **CORRECTED CYCLE 58.** This field previously read 'falsified at this locus for TDP2 (PMID 28794467), where translocations rise because rejoining changes, not because breakage does'. That was wrong: the paper reports that TDP2 REJOINS TOP2-induced DSBs at the MLL hotspot and SUPPRESSES translocations, and says nothing about breakage being unchanged - unrejoined breaks persist, which raises break burden. The real objection is that this premise CONFLATES TWO QUANTITIES: break FORMATION and break PERSISTENCE. Only the persistence reading has evidence, and in isogenic TK6 (Gothe, bioRxiv 10.1101/485763) MLL breakage and MLL-AF4 translocations rise TOGETHER in TDP2-deficient cells. So (iii) is the AMBIGUOUS link, not the weak one, and the cheapest attack on the triad is (i).",
    "status_cycle_58": "ambiguous between formation and persistence; only persistence is evidenced. Not falsified."
  }
]
hypotheses
[
  "hyp-ctcf-occupancy-splits-break-formation-from-rearrangement-fate"
]
experiments
[
  "exp-ctcf-degron-breaks-and-rearrangements-same-cells-kmt2a"
]
analyses
[
  "ana-kmt2a-exon-11-ctcf-anchor-is-one-cohesin-bound-element",
  "ana-gray-2023-cites-both-sides-in-one-sentence-and-never-cites-canela"
]
who measured what
[
  {
    "paper": "PMID 28735753",
    "system": "activated mouse B and T cells, human lines, M. musculus vs M. spretus",
    "perturbation": "etoposide; natural CTCF-motif variation",
    "readout": "END-seq DSB maps, Hi-C, CTCF/RAD21/TOP2B ChIP",
    "question": "DSB formation",
    "measures_rearrangement": false
  },
  {
    "paper": "PMID 31202577",
    "system": "mouse embryonic fibroblasts, human HCT-116",
    "perturbation": "etoposide; acute SMC3 CRE and RAD21 auxin degron",
    "readout": "END-seq; TOP2cc trapping vs conversion",
    "question": "DSB formation and conversion",
    "measures_rearrangement": false
  },
  {
    "paper": "PMID 31202576",
    "system": "human leukaemia lines",
    "perturbation": "etoposide; TOP2 isoform loss; elongation inhibition",
    "readout": "TOP2 DSB maps; MLL fusion formation",
    "question": "DSB formation and fusion formation",
    "measures_rearrangement": "partly, not at this element"
  },
  {
    "paper": "PMID 34405474",
    "system": "human AML patients' normal remission cells, n = 3 vs 3, adult myeloid",
    "perturbation": "low-dose etoposide ex vivo; none on CTCF",
    "readout": "low-input DSBCapture DSB maps",
    "question": "constitutional break-proneness",
    "measures_rearrangement": false
  },
  {
    "paper": "PMID 37788669",
    "system": "human RPE epithelium, KG1a, primary CD34+ HSPCs, mice",
    "perturbation": "siKDM3B, siCTCF, siG9a, JDI-12, UNC0642, doxorubicin, etoposide, overexpression rescues",
    "readout": "DNA FISH copy gain and break-apart; locus qPCR; CTCF and H3K9me1/2 ChIP",
    "question": "whether the locus amplifies and rearranges",
    "measures_rearrangement": true,
    "measures_dsb": false
  }
]
compatibility routes tested
{
  "different_places": "DEAD - 0 bp apart in coordinates",
  "site_versus_occupied_site": "WEAKENED - ENCODE clusters the element as CTCF+RAD21+SMC3 occupied at maximum score in two assemblies; what survives is that occupancy in a haematopoietic cell is unmeasured and that a stress-labile CTCF pool can be lost without global ChIP change",
  "different_quantities": "THIS IS THE ANSWER - formation versus fate, with a published precedent at this locus (TDP2)",
  "one_of_them_is_measuring_cohesin": "NOT PREVIOUSLY LISTED AND LIVE - the only acute directional experiment depletes cohesin, not CTCF, and no head-to-head exists"
}
corrections owed to entries this beat does not own
[
  "mechanism-h3k9me1-2-ctcf-balance-gates-kmt2a-rearrangement: 'THE CTCF SITE IS AT THE FUNCTIONAL CUT POINT OF THE BCR' is false (3,752 bp away); the 'addresses neither conflict' characterisation of PMID 37788669 should be replaced with the verbatim sentence and the Canela omission; the contradiction section should be rewritten around the triad.",
  "genome-kmt2a-exon-11-ctcf-site: four of seven unknowns in data.what_is_NOT_known_about_this_element are now answered; the functional-cut-point convergence claim must be withdrawn.",
  "ana-kmt2a-bcr-ctcf-site-coordinate-overlap: status planned -> run; pre-registered most-likely outcome did not occur; secondary test answered.",
  "gene-ctcf: data.subnuclear_localization should add the SC-35 nuclear-speckle pool (PMID 33411704), a third measured condition-scoped compartment currently absent from a field that already counts seven speckle records in its own bounded absence.",
  "breakpoint-kmt2a-major-bcr: data.exon_convention is null and should record DCAL = Ensembl canonical + 1 with the three base-pair identities.",
  "mechanism-loop-anchor-dsb-hotspots: should record that its model has a directional rival at one named element, and that Canela 2017's mouse Kmt2a exon labels are a fourth numbering frame."
]
How this could bite us back
  • DOXORUBICIN. This adjudication touches an anthracycline that sits in this graph's infant-ALL backbone, and it must be read narrowly. The evidence on the table is: locus copy gain and probe break-apart in immortalised human epithelial cells, extension to a myeloid line and to primary CD34+ cells, and a mouse arm with a cytogenetic endpoint at n = 3-4 per group and no leukaemia-incidence readout. That is cell and animal data about an 8 kb interval. It is NOT a clinical recommendation about anthracycline use, dose or substitution in any patient, and nothing in this entry supports withholding a curative backbone drug. Therapy-related AML after topoisomerase II poisons is a real and long-recognised risk with a latency of up to fifteen years, and it is already weighed in protocol design; this work proposes a mechanism for part of it, not a change in practice.
  • G9a INHIBITORS. UNC0642 is an in vivo chemical probe, not a clinical agent, and the protective result rests on a small cytogenetic mouse experiment. Worse for the idea as an intervention: G9a/EHMT2 sets H3K9me1/2 genome-wide, and a drug given alongside chemotherapy to protect one locus would be altering a global silencing mark in a child during cytotoxic treatment. The compensation risk is obvious - GLP/EHMT1 is the paralogue, dual G9a/GLP inhibition is what the probe actually does, and the study's own result is that EHMT1 depletion does not phenocopy. Any prophylaxis hypothesis here needs a leukaemia-incidence endpoint before it is spoken about as a clinical idea.
  • THE PROTECTIVE DIRECTION IS THE AWKWARD ONE. If the reconciliation is right, CTCF/cohesin occupancy causes the break as well as guarding its repair, so 'increase occupancy to protect the locus' may buy fewer rearrangements at the cost of more breaks - and the anchor is built from proteins this project's own DepMap scan puts near pan-essentiality (CTCF -1.827 across all lines). The tractable target in this mechanism is the methyltransferase, not the anchor, and even that is a genome-wide enzyme.
  • ESSENTIAL-FUNCTION TOXICITY OF THE EXPERIMENT ITSELF. A CTCF degron collapses TAD structure across the genome within hours. Any rearrangement or breakage measured after it must be controlled against the possibility that the cell is dying or mis-segregating, which is exactly the objection to the existing siCTCF arm. The settling experiment has to report viability, cell cycle and a control locus or its answer will be uninterpretable in the same way.
Still open
  1. Did Gothe et al. perturb CTCF or cohesin at the MLL locus? That paper has no reachable PMC deposit and its full text was not read here; if it contains such an arm the triad should be rebuilt around it, and this is the single largest gap in the adjudication.
  2. What is the rearrangement-per-break ratio at this element under any perturbation? Nobody has reported it, in any system, for any factor, and it is the quantity the whole dispute turns on.
  3. Is the anchor occupied by CTCF and cohesin in a human fetal liver or CD34+ progenitor - the cell of origin for the index disease? Every occupancy statement here is from ENCODE cross-cell-type clusters or from RPE epithelium.
  4. Does cohesin removal change rearrangement outcome at this element in the same direction as CTCF removal, or the opposite? This is the arm that distinguishes 'CTCF is a repair chaperone' from 'the whole anchor is the repair chaperone', and it is a one-arm addition to the settling experiment.
  5. Would the reconciliation predict that a cohesin-mutant leukaemia - STAG2 or RAD21 mutant AML, which this graph already holds entries for - shows a distinctive KMT2A rearrangement rate or junction structure? That is a cohort question answerable without a new experiment.
Every source behind this entry (9)
  1. E1strong confidence 4 of 5: Replicated in peer-reviewed work by more than one group.dataset

    In genomic coordinates, Gray et al.'s conserved exon-11 CTCF site and Atkin et al.'s hg38 chr11:118,488,500-118,488,800 break-prone window are the same element: the window lies wholly inside an ENCODE clustered CTCF site at GRCh38 chr11:118,488,159-118,489,206 (score 1000) that is co-occupied by RAD21 (118,488,428-118,488,933, score 1000) and SMC3 (118,488,454-118,488,896, score 1000) and contains all 147 bp of Ensembl canonical exon 11 (ENST00000534358, 118,488,614-118,488,760), reproduced independently in GRCh37; the DCAL/recombinome exon convention is Ensembl canonical plus one, proved by three exact base-pair identities (exon9+intron9 = 1,761; exon10+intron10 = 679; exon11-to-exon13 = 5,026 bp); the DCAL 'functional cut point' at the intron-10/exon-11 border is therefore canonical intron-9/exon-10 at chr11:118,484,861|118,484,862, which is 3,752 bp centromeric of canonical exon 11; and the anchor sits at 70.4% along the Broeker breakpoint cluster region, in the telomeric half that carries 75% of therapy-related and 26% of de novo breakpoints.

    What this evidence cannot show

    The one inferential link is that Gray et al.'s 'exon 11' is canonical exon 11 rather than DCAL exon 11 (= canonical exon 10); Gray publishes no coordinate and the assignment rests on ENCODE occupancy strength, since the only maximum-score CTCF cluster in the BCR spans canonical exon 11 while a DCAL reading lands near a much weaker score-514 cluster. It is settleable from GSE210480's deposited CTCF coverage, which nobody has read. ENCODE clusters are cross-cell-type summaries, not occupancy in a haematopoietic cell. The centromeric/telomeric half assignment applies Broeker's 4.57/3.87 kb split proportionally to an exon span because the BamHI fragment's coordinates are not published.

    Evidence that argues against this (1)
    • If the Gray site is DCAL exon 11 - canonical exon 10 - then it is the weak score-514 CTCF cluster at chr11:118,485,122-118,486,000 and not the break-prone window at all, findings 1 and 2 invert, and the two sides really are describing two elements 3.5 kb apart. The deposited GSE210480 CTCF tracks would show this in one query and are coverage-only, with no peak calls in the series supplement. NCBI Gene Expression Omnibus 2024
  2. E2moderate confidence 3 of 5: One strong study, or several smaller consistent ones.paper

    Atkin et al. mapped DNA breaks with a low-input adaptation of DSBCapture in normal, non-blast cells from AML patients in remission and found breakage at the KMT2A intron 10 / exon 11 region (hg38 chr11:118,488,500-800) 2- to 10-fold higher in each of three KMT2A-rearranged patients than in each of three non-fusion patients, a significant 3.2-fold group average, with 304 TOP2-sensitive, 404 CTCF, 517 RAD21 and 256 SMC3 sites preferentially break-prone in the fusion-driven patients, 59 regions shared across all four datasets and 34 with a break ratio at least as large as the KMT2A region; low-dose etoposide ex vivo raised breakage at these regions in the KMT2A-rearranged patients only, while genome-wide median breakage did not differ between any samples; and they propose that mapping DSBs in patients enables monitoring of individuals susceptible to chromosomal abnormalities and to therapy-related malignancy after TOP2 poisons.

    PMID 34405474DOI
    What this evidence cannot show

    n = 3 per arm, adult, myeloid; the design is cross-sectional and cannot distinguish a constitutional set point from a consequence of having had leukaemia and been treated. It performs no CTCF perturbation and measures no rearrangement, so it constrains proposition (i) only through the occupancy correlation and supplies proposition (iii) as an assumption rather than a measurement. Read from abstract and metadata on 2026-09-22; the hg38 window is as recorded in this graph's gene-ctcf from the paper.

    Evidence that argues against this (1)
    • The same paper's partner-side result argues against the two-sided loop-anchor reading this graph also holds: none of the three patients' own partner genes appeared among the 34 regions carrying all four break-prone features, and only the 3' end of AFDN - not its breakpoint region - carried a break-prone CTCF site. Atkin ND et al. 2021
  3. E3strong confidence 4 of 5: Replicated in peer-reviewed work by more than one group.paper

    Canela et al. 2017 report that CTCF/cohesin-bound loop anchors are vulnerable to TOP2B-mediated DSBs, that sites co-occupied by CTCF and RAD21 are 48-fold enriched in DSBs over matched random regions and 62-fold when TOP2B is also present, that RAD21 occupancy is the single most predictive parameter for breakage (rho = 0.7) and its levels alone suffice to predict DSB intensity, that sites bound by CTCF/RAD21 in C57BL/6 but not in Mus spretus were more likely to harbour DSBs in C57BL/6 and vice versa, and that loop-anchor DSBs are largely transcription-, replication- and cell-type-independent; and Canela et al. 2019 report that acute cohesin loss - SMC3 by CRE in mouse embryonic fibroblasts and RAD21 by auxin-inducible degron in human HCT-116 - causes etoposide-induced END-seq signal to drop 'precipitously genome-wide' in proportion to the cohesin loss (27% RAD21 reduction, 26% END-seq reduction), while separating TOP2 cleavage-complex trapping, which is transcription-independent and requires pre-existing cohesin, from its conversion into an irreversible DSB, which is about two-fold faster at transcribed loci and depends on proteasomal degradation and TDP2 phosphodiesterase activity.

    Topoisomerase II-Induced Chromosome Breakage and Translocation Is Determined by Chromosome Architecture and Transcriptional Activity. · Canela A, Maman Y, Huang SN, Wutz G, Tang W, Zagnoli-Vieira G, Callen E, Wong N, Day A, Peters JM, Caldecott KW, Pommier Y, Nussenzweig A · Molecular Cell 2019
    PMID 31202577DOI
    What this evidence cannot show

    Both papers read in full text from PMC6133249 and PMC8170508 on 2026-09-22. The 2017 CTCF-side evidence is a natural-variant comparison between two mouse species, not a depletion - no siRNA, shRNA, auxin or degron experiment appears in that paper. The 2019 acute perturbation is of COHESIN, not CTCF, and contains no MLL or Kmt2a analysis whatever (zero occurrences of either string). The 48-fold and 62-fold figures are genome-wide class statements, not measurements at the human KMT2A element, and ENCODE records no TOP2B at that element, so only the 48-fold class applies. Systems are mouse B cells, MEFs and a human colorectal line - none haematopoietic human, none B-lineage human.

    Evidence that argues against this (1)
    • The transcription-independence half of the 2017 model is contradicted by the other founding paper of the same model, which reports TOP2 DSBs enriched at loop anchors with HIGH transcriptional activity, break frequency correlating with transcriptional output and directionality, and MLL fusion formation dependent on transcriptional elongation as well as on both TOP2 isoforms. Gothe HJ et al. 2019
  4. E4strong confidence 4 of 5: Replicated in peer-reviewed work by more than one group.paper

    Gray et al. report that H3K9me1/2 balance at the MLL/KMT2A locus, set by crosstalk between KDM3B and G9a/EHMT2, regulates amplification and rearrangement of that locus; that KDM3B depletion increases H3K9me1/2 and reduces CTCF occupancy at the locus, promoting amplification and rearrangements; that depleting CTCF alone is sufficient to generate these focal alterations; that doxorubicin suppresses KDM3B and CTCF protein levels; that KDM3B and CTCF overexpression rescues Dox-induced MLL/KMT2A alterations; and that G9a inhibition in human cells or mice also suppresses MLL/KMT2A events accompanying Dox treatment - with readouts that are DNA FISH copy gain and probe break-apart, locus qPCR and ChIP, in immortalised near-diploid human RPE cells extended to KG1a, primary HSPCs and mice, and with no double-strand-break assay and no mention of cohesin, RAD21 or SMC3 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

    Full text read from PMC10591855 on 2026-09-22; the zero counts for DSB, cohesin, RAD21 and SMC3 are over the deposited body and figure legends, not the supplement. 'Depleting CTCF is also sufficient' is a single siRNA arm against a protein whose pan-essentiality in this project's own DepMap scan is -1.827 across all lines, so the perturbation is not clean. Every CTCF occupancy and chromatin measurement is in RPE epithelium; the haematopoietic and murine extensions are cytogenetic phenotype only. A FISH break-apart is a physical break detection but not a DSB map, so 'no DSB measurement' means no assay that quantifies breaks at kilobase resolution, not no evidence of breakage.

    Evidence that argues against this (1)
    • The direction of this result is opposed at the level of the same class of element by acute cohesin depletion, which reduces TOP2-induced DSB signal genome-wide in proportion to the occupancy lost - so if breaks fall when the anchor is disassembled, a rise in rearrangement after CTCF loss requires the conversion of break to rearrangement to change, which Gray et al. do not measure. Canela A et al. 2019
  5. E5strong confidence 4 of 5: Replicated in peer-reviewed work by more than one group.paper

    Gomez-Herreros et al. show that TDP2 rejoins DSBs induced during transcription-dependent TOP2 activity in breast cancer cells and at the translocation hotspot MLL; that TDP2 suppresses chromosome rearrangements induced by TOP2 and reduces TOP2-induced chromosome translocations arising during gene transcription; and that TDP2-dependent non-homologous end-joining is nonetheless implicated in the formation of a rare subclass of translocations previously associated with therapy-related leukaemia and characterised by junction sequences with 4 bp of perfect homology - establishing at this locus that whether a TOP2-induced break becomes a rearrangement is governed by the rejoining step and can move independently of, and even opposite to, the number of breaks formed.

    TDP2 suppresses chromosomal translocations induced by DNA topoisomerase II during gene transcription. · Gomez-Herreros F, Zagnoli-Vieira G, Ntai I, Martinez-Macias MI, Anderson RM, Herrero-Ruiz A, Caldecott KW · Nature Communications 2017
    PMID 28794467DOI
    What this evidence cannot show

    Abstract read in full on 2026-09-22; the body and figures were not read, so the magnitude of the translocation increase on TDP2 loss, whether breakage was quantified in the same cells, and which MLL breakpoint region was assayed are not established here. The system is breast cancer cells, not haematopoietic. This paper concerns TDP2, not CTCF, so it establishes that proposition (iii) is not a law at this locus - it does not establish that CTCF is the factor that breaks it. Notably this paper IS in Gray et al.'s reference list (their ref 53) and is not used there to separate formation from fate.

    Evidence that argues against this (1)
    • The same paper cuts both ways and says so: TDP2-dependent end-joining is implicated in CREATING a rare therapy-related translocation subclass with 4 bp of perfect homology at the junction, so a repair factor can be protective in aggregate and causal for a specific rearrangement class - which means 'occupancy guards repair' could be true on average and false for the junction type that actually produces KMT2A fusions. Gomez-Herreros F et al. 2017
  6. E6strong confidence 4 of 5: Replicated in peer-reviewed work by more than one group.paper

    Gothe et al. report that TOP2-mediated DSBs are enriched at chromatin loop anchors with high transcriptional activity, that recurrent DSBs occur at CTCF- and cohesin-bound sites at the bases of chromatin loops with frequency positively correlating with transcriptional output and directionality, that genes recurrently translocating to drive leukaemias are highly transcribed and enriched at loop anchors, and that these genes accumulate DSBs at recurrent hotspots giving rise to chromosomal fusions dependent on both TOP2 isoforms and on transcriptional elongation - concluding that transcription and 3D chromosome folding jointly threaten genome stability.

    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, Nielsen CF, Wagner EM, Takeda S, Sasanuma H, Hudson DF, Kindler T, Baranello L, Papantonis A, Crosetto N, Roukos V · Molecular Cell 2019
    PMID 31202576DOI
    What this evidence cannot show

    ABSTRACT AND METADATA ONLY. This paper has no PubMed Central deposit reachable by this beat, so its full text was not read and it is the one paper on either side whose methods this adjudication has not inspected. In particular, whether Gothe et al. performed any CTCF or cohesin depletion at the MLL locus is NOT established here - if they did, the falsifiable triad in this entry would need rewriting around their arm rather than around Canela's. That gap is recorded as an open question and should be closed before this adjudication is quoted as complete.

    Evidence that argues against this (1)
    • The transcription-dependence emphasis is contradicted by the other founding paper of the same model, which reports loop-anchor DSB formation as largely transcription-, replication- and cell-type-independent and driven by cohesin occupancy, with transcription accelerating only the later conversion step. Canela A et al. 2017
  7. E7strong confidence 4 of 5: Replicated in peer-reviewed work by more than one group.paper

    Lehman et al. report that CTCF is exquisitely sensitive to diverse forms of stress in normal patient-derived human mammary epithelial cells; that a subset of CTCF protein forms complexes localising to SC-35-containing nuclear speckles; that stress rapidly downregulates this species through changes in protein stability with loss from the speckles and altered CTCF-RNA interactions; that their ChIP-seq analysis indicated CTCF binding to genomic DNA is largely unchanged; that the stress-sensitive pool and its speckle localisation are restored by inhibiting proteasome-mediated degradation; and that the same response occurs at a specific stage of neuronal commitment from human pluripotent stem cells but not in differentiated neurons, while being uncoupled in persistently stressed 'variant' HMECs and in certain cancer cell lines.

    Dynamic regulation of CTCF stability and sub-nuclear localization in response to stress. · Lehman BJ, Lopez-Diaz FJ, Santisakultarm TP, Fang L, Shokhirev MN, Diffenderfer KE, Manor U, Emerson BM · PLoS Genetics 2021
    PMID 33411704DOI
    What this evidence cannot show

    Abstract only; the ChIP-seq figures were not read, so 'largely unchanged' is the authors' characterisation and its quantitative extent and per-site exceptions are unknown. System is mammary epithelium and neuronal commitment, not haematopoiesis. The abstract does not name doxorubicin as one of the stressors, although Gray et al. cite the paper for Dox-treated HMECs, and this beat could not confirm that from the abstract.

    Evidence that argues against this (1)
    • This result does not undercut the protective model as the dispatch lead suggested, because Gray et al. measure CTCF occupancy at the KMT2A element directly under doxorubicin and under KDM3B depletion rather than inferring it from total protein - a globally stable ChIP-seq profile with a site-selective loss is exactly what a locus-specific mechanism requires. Gray ZH et al. 2023
  8. E8established confidence 5 of 5: Multiple independent lines of evidence; treated as settled in the field.paper

    Gray et al. cite Atkin as reference 56 and Gothe as reference 57 exactly once, together, in a single sentence - 'Prior studies suggest that CTCF binding could impact genome integrity, rearrangement, or duplication, especially at the KMT2A locus;56,57 however, the direct role of CTCF in controlling amplification and rearrangement has not been resolved' - and the two papers carrying the reciprocal loss-of-occupancy results, Canela et al. 2017 (PMID 28735753) and Canela et al. 2019 (PMID 31202577), are absent from all 81 references in the PMC deposit.

    PMID 37788669DOI
    What this evidence cannot show

    Reference numbers are the article's own, read from the JATS <ref id> blocks; Europe PMC's citedOrder field gives different numbers (Gothe 2, Atkin 57) and must not be used for this purpose. The reference-membership test for the two Canela papers is a PMID substring search over the reference block plus a title-level search for 'Canela'; a reference deposited without a PMID would be missed. 'Cited once' is bounded by xref markup and would miss an unlinked prose discussion.

    Evidence that argues against this (1)
    • Absence from a reference list is weak evidence of non-engagement and no evidence at all about correctness: a citation audit cannot show who is right, and reading it as a discount against Gray et al. would be a fallacy. Its only use is to show the two literatures have never met. Europe PMC Consortium 2026
  9. E_k58_tdp2_correctionstrong confidence 4 of 5: Replicated in peer-reviewed work by more than one group.paper

    The third premise of this entry's triad was never falsified, and two fields of this entry said it was. PMID 28794467 reports that TOP2-induced double-strand breaks are rejoined in part by TDP2-dependent non-homologous end-joining, that TDP2 rejoins DSBs induced during transcription-dependent TOP2 activity in breast cancer cells and at the translocation hotspot MLL, that TDP2 suppresses chromosome rearrangements induced by TOP2, and that it reduces TOP2-induced translocations arising during gene transcription. It makes NO claim that breakage is unchanged; loss of rejoining leaves breaks persisting, which raises break burden rather than holding it constant. The paper additionally implicates TDP2-dependent NHEJ in the FORMATION of a rare subclass of translocations characterised by junctions with 4 bp of perfect homology and associated previously with therapy-related leukaemia - so TDP2 suppresses most rearrangements and promotes a rare class.

    PMID 28794467
    What this evidence cannot show

    Abstract-level: the full text and figures were not read, so whether the paper measured breakage separately from translocation frequency - and if so what it found - is not established here. What IS established is that the abstract does not support the claim this entry made from it. The corroborating co-movement of MLL breakage and MLL-AF4 translocation in isogenic TDP2-deficient TK6 comes from Q107's read of the bioRxiv preprint (10.1101/485763) and is relayed rather than independently verified by the king, because bioRxiv blocks programmatic fetching from this machine.

    Evidence that argues against this (2)
    • The 4-bp-microhomology subclass cuts against the corrected reading as much as against the original one. If TDP2-dependent NHEJ FORMS a class of translocations, then TDP2 is not a pure suppressor and the direction of its effect depends on which junction class is counted - so neither 'TDP2 loss raises rearrangements' nor 'breaks and rearrangements co-move' is safe as a general statement about this locus.
    • This correction does not rescue the triad; it moves where the triad is weak. The entry's argument was that all three premises cannot hold and that (iii) was the one already refuted. With (iii) merely ambiguous, the entry no longer has a published refutation of ANY premise, which means its central claim - that the dispute is resolved by rejecting (iii) - is weaker than when it was written.
Connected to