Oncofusion

start / graph / ana-gray-2023-cites-both-sides-in-one-sentence-and-never-cites-canela

Analysisstrong confidence 4 of 5: Replicated in peer-reviewed work by more than one group.Q105-ctcf-adjudication

What the CTCF-protects paper actually cites: it names Gothe and Atkin in ONE direction-free sentence and declares the direction unresolved, and the two papers that ran the reciprocal experiment are absent from all 81 of its references

also called Gray 2023 citation audit · the load-bearing premise, checked · the reciprocal experiment is uncited

This graph said the CTCF-protects paper cites the CTCF-breaks papers and ignores the disagreement. Read directly, the paper cites them in a single sentence that deliberately leaves the direction open and then says the direction has never been settled - and the two papers that did settle it, in the opposite direction, it never cites at all.

The same thing, for a scientist

The premise carried into this beat was that PMID 37788669 cites Gothe (PMID 31202576) and Atkin (PMID 34405474) as its references 57 and 56 and addresses neither conflict. **The reference numbers are right and the characterisation needs sharpening.** From the paper's own JATS reference list, read from PMC10591855: Atkin is **ref 56**, Gothe is **ref 57**, Lehman is **ref 62**, Broeker is **ref 50** [E1]. Both 56 and 57 are cited exactly once, together, in one sentence that opens the CTCF results section: *'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.'* [E1] **That is not an evasion, it is a direction-free absorption**: 'could impact' takes no sign, and the second clause is an explicit claim that the sign was open. Gray et al. never assert that CTCF occupancy is not a fragility marker; they assert that nobody had tested the direction. **What the paper does not do is much more specific than 'addresses neither conflict'**: the two papers whose experiments run the other way - Canela et al. 2017 (PMID 28735753, CTCF/RAD21 co-occupied sites 48-fold enriched for DSBs) and Canela et al. 2019 (PMID 31202577, acute SMC3 or RAD21 depletion causes a proportional genome-wide DROP in TOP2 DSB signal) - **appear nowhere in its 81 references** [E2]. And its body contains **no DSB measurement and no cohesin at all**: zero occurrences of 'DSB', 'double-strand break', 'H2AX', '53BP1', 'END-seq', 'comet', 'loop anchor', 'fragil-', 'cohesin', 'RAD21' or 'SMC3', against 49 'FISH' and 28 'break apart' [E3]. Lehman is cited once, only for protein loss, with the ChIP-seq result that genomic binding is largely unchanged not mentioned [E1][E4]. Twenty works cite Gray as of 2026-09-22 and none is a test of the direction [E5]; and a KMT2A/MLL-scoped CTCF query - the query anyone would run - **cannot see either Canela paper**, because neither abstract names MLL or KMT2A [E6]. A reader using that query sees a three-paper dispute. It is a five-paper dispute.

Go deeper
The one sentence, and why it changes the charge

Verbatim, from the opening of the results section titled 'Reduced CTCF occupancy promotes KMT2A copy gains and rearrangements' 1:

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. Upon evaluating multiple cell lines and tissues from ENCODE, we observed a highly conserved occupancy for CTCF at exon 11 within the BCR of KMT2A, directly overlapping with KDM3B binding.

Ref 56 is Atkin, ref 57 is Gothe 1.

Three things follow and they are not the same thing.

  1. Gray et al. do read Gothe and Atkin as support, not as opposition. They are cited for the proposition that CTCF binding matters at this locus - which both papers do establish. Nothing in the sentence is a misreading.
  2. The sign is deliberately unstated. 'Could impact genome integrity, rearrangement, or duplication' is compatible with protective and with fragile. The paper then says the direction 'has not been resolved', which is a claim to priority on direction, not a rebuttal.
  3. So the accusation of non-engagement has to be narrowed. Gray et al. did not fail to notice a contradiction they cited; they framed their own result as the first directional test. What they did not do is cite the work that had already run a directional test with the opposite outcome - and that work is Canela, not Gothe. Gothe's and Atkin's results are correlative maps of where breaks fall; Canela 2019's is an acute degron depletion 2.

The corpus statement that needs correcting is this graph's, not Gray's. mechanism-h3k9me1-2-ctcf-balance-gates-kmt2a-rearrangement says 'PMID 37788669 cites both Gothe and Atkin (its refs 57 and 56) and addresses neither conflict'. The numbers are right, the sentence is defensible as written, and the sharper and more useful version is: it cites both in one direction-free sentence, declares the direction unresolved, and does not cite either Canela paper.

The reference list, checked as a set

Gray et al.'s bibliography, read from the JATS <ref id=...> blocks of PMC10591855, contains 81 reference entries 2. Mapping the ones that matter:

paperPMIDGray's ref number
Broeker et al. 1996, 11q23 breakpoint distribution / SARs / topo II sites863443950
Atkin et al. 2021, DSB mapping in patients at TOP2 and CTCF/cohesin sites3440547456
Gothe et al. 2019, spatial folding and transcription drive MLL translocations3120257657
Lehman et al. 2021, CTCF stability and sub-nuclear localisation under stress3341170462
Canela et al. 2017, Genome Organization Drives Chromosome Fragility28735753ABSENT
Canela et al. 2019, TOP2 breakage determined by chromosome architecture31202577ABSENT

Both Canela PMIDs were searched as substrings across the whole reference block and return zero hits 2.

Why those two absences are the load-bearing ones. Gothe and Atkin establish an association - breaks are enriched at CTCF/cohesin-bound anchors, and CTCF sites are more break-prone in people who developed a KMT2A fusion. An association is compatible with CTCF being protective at a site that is intrinsically exposed. Canela 2019 is not an association. It removes cohesin acutely - SMC3 by CRE in mouse embryonic fibroblasts, RAD21 by auxin-inducible degron in human HCT-116 - and reports that END-seq signal 'dropped precipitously genome-wide', proportional to the loss ('27% and 26% reduction in RAD21 and END-seq levels, respectively') 2. Canela 2017 adds the CTCF-side version genetically: comparing Mus musculus and Mus spretus, which differ at ~5,000 CTCF motif variants, 'sites bound by CTCF/RAD21 in C57BL/6 but not in Spretus were more likely to harbor DSBs in C57BL/6 compared to Spretus (and visa-versa)' 7. Those are loss-of-occupancy experiments with the opposite outcome to siCTCF in Gray, and neither is cited.

What Gray measured, counted in its own words

Word counts over the full body text of PMC10591855 3:

termoccurrences
FISH49
break apart28
Dox / doxorubicin83 / 21
RPE48
KG1a12
HSPC9
CD343
etoposide3
topoisomerase2
DSB / 'double-strand break' / H2AX / 53BP1 / END-seq / BLISS / comet0 each
cohesin / RAD21 / SMC30 each
'loop anchor' / fragil-0 each
Gothe / Atkin / Canela0 each

Both uses of 'topoisomerase' are clinical framing, not mechanism: 'Therapy-related acute myeloid leukemia (t-AML) is a clinical syndrome occurring long after chemotherapy treatment with agents such topoisomerase II (topo II) inhibitors' and 'amplified and rearranged MDS and AML are generated after topoisomerase II (topo II) inhibitor treatment' 3.

This is the substantive finding of the audit. Gray et al. never measure a DNA break. Their readouts are DNA FISH copy gain and probe break-apart, plus locus qPCR, in RPE cells with extension to KG1a and primary HSPCs. Gothe, Atkin and both Canela papers measure DNA breaks and never measure an inherited copy gain. The two sides do not share a single readout, which is the technical form of the compatibility argument and is why the dispute has never been decided by anybody's data.

Lehman, cited for half of itself

Verbatim 1:

Our observations are consistent with a prior report noting a loss of CTCF protein in Dox-treated patient-derived mammary epithelial cells.62

That is the only mention. Lehman et al.'s abstract states that in human mammary epithelial cells a subset of CTCF forms complexes localising to SC-35 nuclear speckles, that stress rapidly downregulates this species through protein stability, and that 'Our ChIP-seq analysis indicated that CTCF binding to genomic DNA is largely unchanged', with the pool and its speckle localisation restored by proteasome inhibition 4.

The lead carried into this beat was that this undercuts Gray. It does not, and the reason matters. 'Dox lowers CTCF protein' and 'Dox lowers CTCF occupancy where it matters' are indeed different claims - but Gray et al. do not rely on the inference. They measure CTCF ChIP at the KMT2A site under Dox directly and report that occupancy falls there 8. So Lehman is a reason the western blot cannot carry the argument, and a reason to expect the effect to be site-selective rather than global, which is what a locus-specific mechanism needs anyway. It is a sharpening, not a refutation. What it does undercut is any reading in which Dox produces a genome-wide CTCF loss.

The real gap Lehman exposes is in this graph, not in Gray. gene-ctcf's subnuclear_localization field records exactly two measured compartments - nucleolar tethering via nucleophosmin (PMID 14759373) and mitotic pericentromere via CENP-E (PMID 26321640) - and does not record the SC-35 nuclear-speckle pool, although the same entry's own bounded-absence block counts CTCF[tiab] AND "nuclear speckles"[tiab] at 7 records 49. A third measured, condition-scoped subnuclear compartment for CTCF exists, is cited elsewhere in this corpus, and is missing from the field built to hold it.

Has anyone else adjudicated this? No, and the obvious query cannot see half the dispute

Who cites Gray. Europe PMC returns 20 citing works as of 2026-09-22, all screened by title 5. They are reviews of histone demethylases, of extrachromosomal and inherited amplification, of epigenetic modulators and of fusion transcripts; a KMT2A case report; chromatin-domain-detection methodology; and unrelated primary papers in head and neck cancer, ovarian cancer, AML aging and NF-kB amplification. None is a directional test of CTCF's effect at the KMT2A locus, and none cites Gray alongside Canela to adjudicate.

The literature intersection, bounded. Running the two gene names separately as rule 94 requires 6:

queryquerytranslationcountwarninglist
CTCF[tiab] AND KMT2A[tiab]"CTCF"[Title/Abstract] AND "KMT2A"[Title/Abstract]7none
CTCF[tiab] AND MLL[tiab]"CTCF"[Title/Abstract] AND "MLL"[Title/Abstract]9none
union of the two ID SETS-14 (overlap 2)-
CTCF[tiab] AND KMT2A[tiab] AND "double-strand break"[tiab]full translation, non-empty1none
CTCF[tiab] AND MLL[tiab] AND "double-strand break"[tiab]full translation, non-empty1none
KDM3B[tiab] AND CTCF[tiab]"KDM3B"[Title/Abstract] AND "CTCF"[Title/Abstract]1none
CTCF[tiab] AND "loop anchor"[tiab]full translation20none
CTCF[tiab] AND "loop anchors"[tiab]full translation57none

All 14 records of the union were read by title, not counted 6. Three are on topic: 37788669 (Gray), 34405474 (Atkin), 31202576 (Gothe). The other eleven are HOXA CTCF-boundary work (33057942, 30985763, 35180428), the m6A/TAD paper (41430607), a RET enhancer paper (38226414), and six unrelated records - prostate methylation, short stature, NPM/SET mutations, nasopharyngeal epigenome, IDH/KDM review, CFTR topology.

And here is the trap. Neither Canela 2017 nor Canela 2019 is in that union, because neither abstract contains 'MLL' or 'KMT2A' - Canela 2017's Kmt2a analysis is Figure 1A and its abstract says only 'breakpoint clusters commonly translocated in cancer'; Canela 2019 contains no MLL analysis at all. This is rule 80 in an adjudication: a disease- or gene-scoped query bounds abstracts, and the reciprocal experiment lives in a supporting panel of a paper about something broader. Any beat that scoped to KMT2A would have concluded, correctly by its own instrument and wrongly in fact, that this is a dispute between three papers.

Also recorded: "loop anchor" returns 20 and "loop anchors" returns 57 - the same singular/plural split rule 72 documents for drug names, on the central term of this model's vocabulary.

Facts and figures
scriptnone - reference-list and full-text extraction from PMC JATS, plus Europe PMC and NCBI E-utilities calls, all recorded inline
statusrun
run date2026-09-22
result summaryThe premise is confirmed in its reference numbers and needs sharpening in its charge. Gray et al. cite Atkin as ref 56 and Gothe as ref 57, once, together, in a single direction-free sentence that then declares the direction of CTCF's effect unresolved - so they absorb both as precedent rather than contradicting them. The unengaged work is Canela 2017 and Canela 2019, which are absent from all 81 of Gray's references and which carry the reciprocal loss-of-occupancy results. Gray's body contains no DSB measurement and no mention of cohesin, RAD21 or SMC3; the other side measures DSBs and never measures an inherited copy gain, so the two sides share no readout. Lehman is cited only for CTCF protein loss and not for its finding that genomic binding is largely unchanged - which sharpens rather than refutes Gray, because Gray measures occupancy at the locus directly, but does expose that gene-ctcf's subnuclear_localization field omits the SC-35 speckle pool. Twenty citing works screened; none adjudicates. And a KMT2A/MLL-scoped CTCF query cannot see either Canela paper, so the dispute looks like three papers and is five.
name list noteCTCF[tiab] AND "loop anchor"[tiab] = 20; CTCF[tiab] AND "loop anchors"[tiab] = 57. Singular and plural are different strings on the central term of this model. Any bounded absence about loop anchors in this corpus that ran only the singular form is a floor.
inputs
[
  "PMC10591855 JATS XML via NCBI efetch db=pmc (Gray et al. 2023)",
  "Europe PMC references endpoint for MED/37788669",
  "Europe PMC citations endpoint for MED/37788669",
  "PMC6133249 JATS XML (Canela et al. 2017)",
  "PMC8170508 JATS XML (Canela et al. 2019)",
  "PMID 33411704 abstract (Lehman et al. 2021)",
  "NCBI esearch, 15 bounded queries with querytranslation and warninglist read back"
]
premise audit
{
  "brief_premise": "PMID 37788669 cites both Gothe and Atkin (its references 57 and 56) and addresses neither.",
  "verdict": "PARTLY CONFIRMED, SHARPENED",
  "what_is_right": "Both are cited. The reference numbers are exactly right: Atkin is 56 and Gothe is 57 in the paper's own numbering.",
  "what_is_wrong_or_imprecise": "'Addresses neither' overstates the omission in one direction and understates it in another. Gray et al. cite both in one sentence, as support for the premise that CTCF binding matters at this locus, with the sign left open by the word 'impact', and immediately state that 'the direct role of CTCF in controlling amplification and rearrangement has not been resolved' - a claim to priority on direction, not a failure to notice a contradiction. The papers they do not cite at all are Canela 2017 and Canela 2019, and those are the ones that ran the reciprocal experiment.",
  "identifier_trap_found_while_checking": "Europe PMC's citedOrder field is NOT the reference number printed in the paper. For this article citedOrder gives Gothe as 2 and Atkin as 57; the article's own numbered list, read from the JATS <ref id> blocks, gives Atkin as 56 and Gothe as 57. An agent quoting citedOrder would have reported the brief's premise as wrong when it was right."
}
gray reference map
[
  {
    "pmid": "8634439",
    "gray_ref": 50,
    "cited_for": "the 8.3 kb BCR spanning exons 8-14, enriched for KMT2A rearrangements"
  },
  {
    "pmid": "34405474",
    "gray_ref": 56,
    "cited_for": "one direction-free sentence, jointly with 57, on CTCF binding impacting genome integrity at the KMT2A locus"
  },
  {
    "pmid": "31202576",
    "gray_ref": 57,
    "cited_for": "the same single sentence"
  },
  {
    "pmid": "33411704",
    "gray_ref": 62,
    "cited_for": "loss of CTCF protein in Dox-treated patient-derived mammary epithelial cells - protein level only"
  },
  {
    "pmid": "28735753",
    "gray_ref": null,
    "cited_for": "NOT CITED - absent from all 81 references"
  },
  {
    "pmid": "31202577",
    "gray_ref": null,
    "cited_for": "NOT CITED - absent from all 81 references"
  }
]
readout disjunction
{
  "gray_2023": "DNA FISH copy gain and two-colour probe break-apart, locus qPCR; RPE cells, extended to KG1a, primary HSPCs and mice. No DSB map, no gammaH2AX, no 53BP1, no END-seq.",
  "gothe_2019": "TOP2 DSB mapping plus MLL translocation formation; transcription and elongation dependence. No inherited copy-gain readout.",
  "atkin_2021": "low-input DSBCapture DSB mapping in patients' normal remission cells. No perturbation of CTCF, no rearrangement readout.",
  "canela_2017": "END-seq DSB mapping with Hi-C loop anchors and CTCF/RAD21/TOP2B ChIP; natural CTCF-motif variants between two mouse species. No rearrangement readout at KMT2A.",
  "canela_2019": "END-seq with acute SMC3 (CRE) and RAD21 (auxin degron) depletion. No MLL/Kmt2a analysis at all.",
  "consequence": "No paper on either side measures BOTH a DSB at this element and a rearrangement outcome at this element in the same cells. That is the technical reason the dispute is undecided, and it is what the settling experiment has to fix."
}
bounded absences
[
  {
    "claim": "No published work adjudicates the direction of CTCF's effect on KMT2A breakage or rearrangement.",
    "instrument": "Europe PMC citations of MED/37788669 (20 records, all screened by title) plus the union of CTCF[tiab] AND KMT2A[tiab] and CTCF[tiab] AND MLL[tiab] (14 records, all screened by title)",
    "bound": "titles and abstracts only; supporting figures not searched",
    "absence_class": "never-measured",
    "known_blind_spot": "the query cannot reach Canela 2017 or 2019, whose abstracts name neither gene - so this absence was established only because both papers were reached from Gray's citation neighbourhood and from this graph's own mechanism-loop-anchor-dsb-hotspots entry, not from the scoped query"
  },
  {
    "claim": "gene-ctcf's subnuclear_localization does not record the SC-35 nuclear-speckle pool.",
    "instrument": "direct read of data.subnuclear_localization in data/graph/nodes/break-repair-machinery/gene-ctcf.json on 2026-09-22",
    "bound": "exact - the field names two compartments, nucleolar tethering and mitotic pericentromere, plus two uncoded hints, and no speckle",
    "absence_class": "measured-not-recorded"
  }
]
How this could bite us back
  • Nothing in this entry is a clinical claim about doxorubicin or about G9a inhibitors. It is an audit of who cited whom. The only intervention-adjacent content is that Gray et al.'s doxorubicin arm rests on a direct locus ChIP rather than on a western blot, which makes their mechanism narrower and better supported, not more actionable - a locus-level event in RPE cells and in mice is not evidence about anthracycline dosing in a child.
  • The absence of a G9a-inhibitor adjudication in the literature must not be read as a green light. UNC0642 is an in vivo chemical probe, not a clinical agent, and the mouse arm supporting it is a cytogenetic endpoint at n = 3-4 per group with no leukaemia-incidence readout. An entry establishing that nobody has contradicted a result is not an entry establishing that the result is ready to act on.
  • A citation audit is the weakest kind of evidence about who is right. Establishing that Gray did not cite Canela says nothing about whether Gray's result is correct, and using this entry to discount Gray would be a fallacy. Its use is to show that the two literatures have never met, which is a reason to run the experiment, not a reason to pick a side.
Still open
  1. Does Nat Struct Mol Biol 2026 PMID 41131337, primary work on epigenetic priming and oncogene amplification that cites Gray, contain a CTCF occupancy or breakage measurement at an amplified locus? It is the closest candidate for an existing directional test and this beat screened it by title only.
  2. Did Gothe et al. perturb CTCF or cohesin at all, or is their MLL result purely a break-map plus a translocation assay? That paper has no PMC deposit, so its full text was not read here, and if it contains a CTCF-depletion arm at the MLL locus the falsifiable pair in this beat would need rewriting around it.
  3. Where do the four mouse Kmt2a exon labels in Canela 2017 ('near exon 10', 'at exon 12') map in human canonical coordinates? Until that is done there is a fourth exon-numbering frame in this dispute that nobody has reconciled.
  4. Canela 2017 attributes to Cowell et al. 2012 the prior observation that 'the translocation breakpoint in exon 12 of human MLL co-localized with DNase I hypersensitive sites associated with CTCF occupancy'. If that is right, a CTCF site at the human MLL breakpoint was published in 2012 and this graph does not hold it - which exon 12, in whose numbering, and is it the same element?
Every source behind this entry (9)
  1. E1established confidence 5 of 5: Multiple independent lines of evidence; treated as settled in the field.paper

    In Gray et al. 2023 the results section 'Reduced CTCF occupancy promotes KMT2A copy gains and rearrangements' opens: '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. Upon evaluating multiple cell lines and tissues from ENCODE, we observed a highly conserved occupancy for CTCF at exon 11 within the BCR of KMT2A, directly overlapping with KDM3B binding.' - where refs 56 and 57 are Atkin et al. (PMID 34405474) and Gothe et al. (PMID 31202576); refs 56 and 57 are cited nowhere else in the paper; ref 62 is Lehman et al. (PMID 33411704), cited once, in 'Our observations are consistent with a prior report noting a loss of CTCF protein in Dox-treated patient-derived mammary epithelial cells.62'; and ref 50 is Broeker et al. (PMID 8634439), cited for 'the 8.3kb breakpoint cluster region (BCR) spanning exons 8-14, which is enriched for KMT2A rearrangements'.

    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

    Read from the PMC JATS XML of PMC10591855 on 2026-09-22, resolving reference ids to PMIDs from the <ref> blocks and locating every in-text <xref rid> for R56, R57, R62 and R50. The reference-number-to-PMID mapping is the article's own numbering, not Europe PMC's citedOrder, which differs. Only PMC's deposited version was read; if the published article differs in reference numbering this mapping would differ with it. 'Cited nowhere else' is bounded by the xref markup in that deposit - an uncited discussion of the same work would not be found this way.

    Evidence that argues against this (1)
    • Reading a paper's engagement from its citation markup can understate it: the same corpus has recorded that a curator's evidence code tells you an experiment was seen somewhere in a paper, not that the paper is about the fact, and the inverse failure is equally available - a paper can engage a rival model in prose without an xref, and figure panels are not marked up at all. Gray ZH et al. 2023
  2. E2established confidence 5 of 5: Multiple independent lines of evidence; treated as settled in the field.paper

    Gray et al.'s bibliography as deposited in PMC10591855 contains 81 <ref> entries; Atkin et al. (PMID 34405474) is ref 56, Gothe et al. (PMID 31202576) is ref 57, Broeker et al. (PMID 8634439) is ref 50 and Lehman et al. (PMID 33411704) is ref 62; and the PMIDs 28735753 (Canela et al. 2017, Genome Organization Drives Chromosome Fragility, Cell) and 31202577 (Canela et al. 2019, Topoisomerase II-Induced Chromosome Breakage and Translocation Is Determined by Chromosome Architecture and Transcriptional Activity, Mol Cell) do not occur anywhere in that reference block. Canela et al. 2019 reports that in mouse embryonic fibroblasts with CRE-mediated SMC3 depletion the etoposide END-seq signal 'dropped precipitously genome-wide', 'proportional to the decrease in cohesin levels (27% and 26% reduction in RAD21 and END-seq levels, respectively)', and that the result was confirmed in human HCT-116 cells where RAD21 was depleted with an auxin-inducible degron.

    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

    The reference-list membership test is a substring search for each PMID across the JATS reference block of the PMC deposit. A reference carrying no PMID in the deposit would be missed by that test; the Canela papers are both PMID-carrying and heavily indexed, so a missing-identifier explanation is unlikely but not excluded, and a title-level search for 'Canela' also returned nothing. Canela 2019's own depletion experiments are in MEFs and HCT-116 colorectal cells and contain no MLL or Kmt2a analysis at all (zero occurrences of either string in its full text), so its relevance to this locus is by class membership, not by measurement here.

    Evidence that argues against this (1)
    • Cohesin loss is not CTCF loss, and the escape hatch is real: CTCF depletion does not remove cohesin from chromatin but redistributes it, so an acute cohesin degron reducing TOP2 breakage does not by itself predict what siCTCF does at one site. The same paper's framing - that cohesin binding and transcriptional activity alone predict TOP2 DSB kinetics - makes cohesin, not CTCF, the quantity its experiment tested. Canela A et al. 2019
  3. E3established confidence 5 of 5: Multiple independent lines of evidence; treated as settled in the field.paper

    Case-insensitive counts over the full body text of PMC10591855 (76,926 characters as extracted) give: FISH 49, 'break apart' 28, Dox 83, doxorubicin 21, RPE 48, KG1a 12, HSPC 9, CD34 3, etoposide 3, topoisomerase 2, ecDNA 1, translocation 1; and zero occurrences of each of 'DSB', 'double-strand break', 'double strand break', 'H2AX', '53BP1', 'END-seq', 'BLISS', 'comet', 'loop anchor', 'fragil', 'cohesin', 'RAD21', 'SMC3', 'Gothe', 'Atkin', 'Canela', 'Nussenzweig', 'Roukos' and 'speckle'. Both 'topoisomerase' occurrences are clinical framing of therapy-related AML as following treatment with topoisomerase II inhibitors, not mechanism.

    PMID 37788669DOI
    What this evidence cannot show

    The extracted body includes figure legends and the STAR Methods but NOT the reference list, so author-name counts of zero are counts over prose and cannot be read as absence from the bibliography - that test is [E2]. Supplementary figures, tables and their legends beyond what PMC deposits inline were not searched, so a DSB assay reported only in a supplement would be invisible; this is the rule-80 bound applied to a single paper rather than to an index. The gene symbol 'TOP2A'/'TOP2B' was searched as 'TOP2' and returned zero, but a paper writing only 'topo II' would be caught by the topoisomerase count, which it was.

    Evidence that argues against this (1)
    • A word count is not a readout inventory: FISH break-apart of a two-colour probe set IS a physical detection of a chromosomal break, so 'zero DSB measurements' must be read as 'no assay that maps or quantifies double-strand breaks at nucleotide or kilobase resolution', not as 'no evidence of breakage'. Gray et al.'s break-apart counts are evidence of breakage without being a DSB map. Gray ZH et al. 2023
  4. E4strong 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 on stress this species is rapidly downregulated by changes in protein stability with loss from the speckles and altered CTCF-RNA interactions; that 'Our ChIP-seq analysis indicated that 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 stress 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 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 read in full on 2026-09-22; the body and the ChIP-seq figures were not read, so 'largely unchanged' is the authors' own summary characterisation and its quantitative extent, the stressors used, and whether any individual site changed are not established here. The system is mammary epithelium and neuronal commitment, not haematopoiesis, and the stressors are not stated in the abstract as being doxorubicin specifically - Gray et al. cite the paper for Dox-treated HMECs, which this beat could not confirm from the abstract alone.

    Evidence that argues against this (1)
    • Gray et al. do not depend on inferring occupancy from protein level: they report a direct CTCF ChIP measurement at the KMT2A site showing occupancy falls under Dox and under KDM3B depletion without any change in total CTCF protein, so a globally unchanged ChIP-seq profile is compatible with a site-selective occupancy loss and does not refute the mechanism. Gray ZH et al. 2023
  5. E5moderate confidence 3 of 5: One strong study, or several smaller consistent ones.database

    Europe PMC's citations endpoint for PMID 37788669 returns a hitCount of 20 as of 2026-09-22, comprising two reviews of histone demethylases (40185081 Cell 2025, 39938867 Mol Cells 2025), a review of epigenetic regulation of extrachromosomal and inherited DNA amplification (40154613 J Biol Chem 2025), reviews of epigenetic modulators (38914477 Genes Dev 2024), JMJD proteins (38926050 Trends Biochem Sci 2024), non-canonical chromatin species (38253662 Nat Struct Mol Biol 2024), fusion transcripts (42671651 Clin Exp Med 2026), pancreatic cancer immune microenvironment (39303859) and reproductive-health epigenetics (39345884); primary work on tyrosine-kinase-inhibitor resistance and oncogene amplification (41131337 Nat Struct Mol Biol 2026 and its preprint PPR970821), head and neck chromatin factors (40974867), a chromatin-associated circRNA (40908373), NF-kB transient amplifications (PPR989879), del(5q) AML epigenetics (PPR940374), AML aging profiling (38422020), ovarian cancer KMT2A (39589456), a chromatin-domain-detection framework (39617879), a doxorubicin nanoparticle delivery paper (39549369) and a KMT2A AML case report (41897553) - and none of the twenty is a directional test of CTCF's effect on breakage or rearrangement at the KMT2A locus.

    What this evidence cannot show

    Screened by TITLE only; none of the twenty was opened. A directional CTCF experiment reported as a supporting panel inside any of them - most plausibly inside 41131337, which is primary work on amplification from an overlapping author group - would not have been seen. Europe PMC citation counts lag and are incomplete for very recent and for non-indexed work, and two of the twenty are preprints (PPR ids) whose published versions may be double-counted.

    Evidence that argues against this (1)
    • One of the twenty is primary work on oncogene amplification from a Nature Structural and Molecular Biology 2026 paper on epigenetic priming, which this beat did not open; if it contains a CTCF occupancy or breakage measurement at an amplified locus it would be the closest existing test and this bounded absence would need narrowing. Europe PMC Consortium 2026
  6. E6established confidence 5 of 5: Multiple independent lines of evidence; treated as settled in the field.database

    NCBI esearch over PubMed on 2026-09-22, with querytranslation and warninglist read back from every response, returns: CTCF[tiab] AND KMT2A[tiab] = 7 (translation '"CTCF"[Title/Abstract] AND "KMT2A"[Title/Abstract]', no warninglist); CTCF[tiab] AND MLL[tiab] = 9; the union of the two ID SETS = 14 records with 2 in both; CTCF[tiab] AND KMT2A[tiab] AND "double-strand break"[tiab] = 1; CTCF[tiab] AND MLL[tiab] AND "double-strand break"[tiab] = 1; CTCF[tiab] AND KMT2A[tiab] AND rearrangement[tiab] = 2; CTCF[tiab] AND MLL[tiab] AND rearrangement[tiab] = 3; CTCF[tiab] AND KMT2A[tiab] AND fragility[tiab] = 1; CTCF[tiab] AND MLL[tiab] AND fragility[tiab] = 2; KDM3B[tiab] AND CTCF[tiab] = 1; cohesin[tiab] AND KMT2A[tiab] AND "double-strand break"[tiab] = 1; CTCF[tiab] AND doxorubicin[tiab] AND occupancy[tiab] = 2; CTCF[tiab] AND "nuclear speckles"[tiab] = 7; CTCF[tiab] AND SC-35[tiab] = 1; CTCF[tiab] AND "loop anchor"[tiab] = 20 while CTCF[tiab] AND "loop anchors"[tiab] = 57. All 14 records of the KMT2A/MLL union were read by title: only 37788669, 34405474 and 31202576 concern CTCF at the KMT2A locus in relation to breakage or rearrangement, and neither 28735753 nor 31202577 is in the union because neither abstract contains the string MLL or KMT2A.

    What this evidence cannot show

    Every count here is bounded by titles and abstracts; supporting figures were not searched, which is precisely the bound that hides Canela 2017's Kmt2a panel. Two calls in the first pass returned count=null with querytranslation=null - an API rate-limit error object, not a zero - and were re-run with pacing; a null translation is an instrument failure and no count was recorded from those responses. Single-digit counts are screened, not statistical: the low numbers reflect how few abstracts name both a chromatin protein and this gene, not how much relevant work exists. One term per tag was used throughout; no parenthesised grouped-tag form appears in any of these queries.

    Evidence that argues against this (1)
    • The 'loop anchor' 20 versus 'loop anchors' 57 split shows that these counts are governed by the name list rather than by the literature: a singular-only query on the central term of the model under dispute misses about two thirds of it, and the same failure mode applies to every term here that was run in only one form. NCBI 2026
  7. E7strong confidence 4 of 5: Replicated in peer-reviewed work by more than one group.paper

    Canela et al. 2017 report, in the section 'Polymorphisms in CTCF/Cohesin binding rewire DSBs to novel sites': 'To directly determine the effects of CTCF/Cohesin binding on TOP2B activity, we examined the repertoire of DNA breaks in activated B cells from Mus musculus (C57BL/6) and Mus spretus (Spretus), which exhibit approximately 5000 single nucleotide variants within CTCF DNA binding motifs... Analysis of ETO induced damage revealed that DSBs followed CTCF/RAD21 binding; that is, sites bound by CTCF/RAD21 in C57BL/6 but not in Spretus were more likely to harbor DSBs in C57BL/6 compared to Spretus (and visa-versa)... Among the CTCF/RAD21 binding sites that were conserved, DSBs were generally present in both species'; and that 'the most predictive parameter for DNA breakage was RAD21 occupancy (rho=0.7)', with RAD21 levels alone sufficient to predict DSB intensity.

    Genome Organization Drives Chromosome Fragility. · Canela A, Maman Y, Jung S, Wong N, Callen E, Day A, Kieffer-Kwon KR, Pekowska A, Zhang H, Rao SSP, Huang SC, Mckinnon PJ, Aplan PD, Pommier Y, Aiden EL, Casellas R, Nussenzweig A · Cell 2017
    PMID 28735753DOI
    What this evidence cannot show

    Read in full text from PMC6133249 on 2026-09-22. This is a natural-variant genetic COMPARISON between two mouse species, not a depletion: no siRNA, shRNA, auxin or degron experiment appears anywhere in the paper (zero occurrences of each string). It is therefore a correlative loss-of-occupancy argument, stronger than co-localisation and weaker than an acute perturbation, and the acute version is in Canela 2019 and is of cohesin rather than CTCF. Two mouse species differing at 5,000 CTCF motifs differ at very much else besides.

    Evidence that argues against this (1)
    • The paper's own framing makes RAD21 rather than CTCF the predictive variable - RAD21 occupancy is the most predictive parameter at rho=0.7 and its levels alone suffice to predict DSB intensity - so the species comparison supports a cohesin-dependence claim more strongly than a CTCF-dependence claim, and CTCF and cohesin occupancy are confounded in it. Canela A et al. 2017
  8. E8strong confidence 4 of 5: Replicated in peer-reviewed work by more than one group.paper

    Gray et al.'s abstract states that 'KDM3B depletion increases H3K9me1/2 levels and reduces CTCF occupancy at the MLL/KMT2A locus, in turn promoting amplification and rearrangements', that 'Depleting CTCF is also 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' - so the occupancy claim at the locus is made as a direct measurement and not as an inference from total protein.

    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 occupancy measurement is in immortalised near-diploid human RPE epithelial cells, with the cytogenetic phenotype extended to KG1a and primary HSPCs and to mice, but no CTCF occupancy measurement in any haematopoietic cell. 'Depleting CTCF is also sufficient' is a single siRNA arm against a protein this project's own DepMap scan scores at -1.827 pan-essentiality, so a CTCF knockdown perturbs a great deal besides this locus.

    Evidence that argues against this (1)
    • The direction of this whole result is opposed by the model in which CTCF/cohesin occupancy is what makes a loop anchor breakable: in that model removing occupancy should reduce TOP2-mediated breakage at the anchor, and an acute cohesin degron does exactly that, genome-wide and proportionally. Canela A et al. 2019
  9. E9established confidence 5 of 5: Multiple independent lines of evidence; treated as settled in the field.database

    UniProt P49711 (CTCF_HUMAN, 727 aa) records subcellular locations 'Nucleus, nucleoplasm' (ECO:0000269 PubMed:28319062), 'Chromosome' and 'Chromosome, centromere' (both ECO:0000269 PubMed:26321640), with an uncoded note that CTCF 'May translocate to the nucleolus upon cell differentiation', associates with centromeres and chromosomal arms during metaphase, associates with the H19 ICR in mitotic chromosomes and 'May be preferentially excluded from heterochromatin during interphase' - and records no nuclear-speckle or SC-35-associated species.

    What this evidence cannot show

    Quoted from this graph's own gene-ctcf entry, which records the field as fetched on 2026-09-21, rather than re-fetched by this beat - so this item establishes what the graph holds and what UniProt held a day earlier, not what UniProt holds today. The absence of a speckle annotation is a statement about UniProt's curation, not about the biology, and rule 91 applies: the nucleoplasm citation is an endometrial carcinoma genetics paper whose abstract describes no localisation experiment.

    Evidence that argues against this (1)
    • A missing UniProt annotation is a weak absence for exactly the reason this corpus already documented: the coded locations frequently point at the wrong paper and the informative statements are often in the uncoded notes, so the speckle pool's absence from the feature table says little beyond that no curator entered it. Lehman BJ et al. 2021
Connected to