Oncofusion

start / graph / ana-canela-2019-never-names-mll-and-its-hct116-arm-is-lovastatin-arrested-in-g1

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

Canela 2019's full text never names MLL or KMT2A, and its HCT116 cohesin arm is lovastatin-arrested in G1 - so the only directional cohesin experiment was run in a cell where transcription-coupled fragility should be lowest

also called Canela 2019 full text · PMC8170508 · lovastatin G1 arrest · no MLL panel · premise (i) cell-cycle mismatch

The one experiment that removes the cohesin ring and then counts DNA breaks was done in cells deliberately frozen before they divide, using a drug that stops them growing. That is the state in which this stretch of DNA is least likely to break in the first place.

The same thing, for a scientist

Q107 audited this paper's deposit and did not read its text, and flagged that as the discrepancy to chase: if PMC8170508 contains a cohesin-dependent HCT116 *MLL* panel that Q107's percentile readout does not reproduce, that contradiction would be the finding. **It does not, and the reason is stronger than 'no panel found'.** **The full text never mentions the gene.** Retrieved by `efetch db=pmc` (63,630 characters of `<body>` after tag-stripping and entity-unescaping), the text returns **zero** hits for MLL, KMT2A, Kmt2a, 'mixed lineage', 'mixed-lineage', 11q23, HRX, ALL-1, AF4, AFF1, AF9, MLLT3, t(4;11) and t(9;11). The single apparent `ENL` hit is the substring inside *enlarged*. Positive controls prove the document was loaded: TOP2 **216**, END-seq **46**, cohesin **37**, RAD21 **22**, etoposide **18**. The paper's cohesin claim is explicitly **genome-wide and in supplementary figures**: in MEFs the END-seq drop was *"proportional to the decrease in cohesin levels (27% and 26% reduction in RAD21 and END-seq levels)"*, and *"RAD21 depletion resulted in a precipitous drop in END-seq signal in HCT-116 genome-wide (Figures S1B-S1D)"*. So there is no locus-specific KMT2A claim to contradict, and **Q107's premise-(i) conclusion is not in tension with this paper's text.** **THE FINDING IS IN THE METHODS, AND IT ANSWERS A QUESTION I HAD QUEUED FOR NEXT CYCLE.** Verbatim: *"HCT-116-CMV-OsTir1 and HCT-116-RAD21-mAID-Clover cells were obtained from (Natsume et al., 2016), **arrested in G1 by the treatment with 20 μM of lovastatin for 24 h** and treated with auxin for 4 h before 50 μM etoposide treatment for 30 min."* Q107 proposed an ENCODE lookup to ask whether *KMT2A* is transcribed in lovastatin-arrested G1 HCT116. The G1 arrest itself is confirmed from the paper, so the premise of that question is real - and it **upgrades the problem with premise (i)**. Q107 showed the element is a top-0.05%-of-chr11 etoposide DSB site in every blood-lineage cell measured and only the 94.004th percentile in HCT116, and read that as a cell-type mismatch. It is a cell-type **and cell-cycle and transcriptional-state** mismatch: this graph's own measurement of Gothe's DRB arm puts the element at the 98.0th percentile untreated and the **71.0th under transcriptional-elongation blockade**, so fragility here is transcription-coupled - and a lovastatin-arrested G1 cell is close to the condition in which it should be lowest. **It engages the fusion-hotspot literature without naming the gene**, which is exactly the retrieval failure this project has a registry row for: *"TOP2 colocalized with CTCF/RAD21 binding at recurrent translocation hotspots found in ETO therapy-related acute myeloid leukemia (t-AML)"*, citing Ashour 2015, Cowell 2012 and Gómez-Herreros 2017. That row was previously verified at abstract level; it is now verified at **full text**.

Facts and figures
paperCanela et al. 2019, PMID 31202577, open access at PMC8170508
retrieval routeefetch db=pmc id=PMC8170508 retmode=xml; <body> present; tags stripped then HTML entities unescaped; 63,630 characters analysed
why this was runQ107 audited this paper's GEO deposit and explicitly did not read its text, naming that as the highest-value unread item and as the one place its premise-(i) conclusion could still be wrong.
absence classmeasured-not-analysed
mll kmt2a token census
{
  "MLL": 0,
  "KMT2A": 0,
  "Kmt2a": 0,
  "mixed lineage": 0,
  "mixed-lineage": 0,
  "11q23": 0,
  "HRX": 0,
  "ALL-1": 0,
  "AF4": 0,
  "AFF1": 0,
  "AF9": 0,
  "MLLT3": 0,
  "t(4;11)": 0,
  "t(9;11)": 0,
  "ENL": "1, and it is the substring inside 'enlarged' - checked in context, not counted as a gene mention",
  "positive_controls_proving_the_document_loaded": {
    "TOP2": 216,
    "END-seq": 46,
    "cohesin": 37,
    "RAD21": 22,
    "etoposide": 18,
    "G1": 6,
    "lovastatin": 1
  }
}
the cohesin claim is genome wide
{
  "MEFs": "the END-seq drop was proportional to the cohesin drop - 27% RAD21 and 26% END-seq",
  "human": "'RAD21 depletion resulted in a precipitous drop in END-seq signal in HCT-116 genome-wide (Figures S1B-S1D)'",
  "consequence": "there is no locus-specific KMT2A panel in this paper, so nothing here contradicts Q107's percentile readout at the element"
}
the hct116 arm is g1 arrested
{
  "verbatim_methods": "HCT-116-CMV-OsTir1 and HCT-116-RAD21-mAID-Clover cells were obtained from (Natsume et al., 2016), arrested in G1 by the treatment with 20 microM of lovastatin for 24 h and treated with auxin for 4 h before 50 microM etoposide treatment for 30 min.",
  "parental_lines": "Natsume et al. 2016",
  "auxin_exposure": "4 h before etoposide",
  "etoposide": "50 microM for 30 min",
  "how_often_stated": "once, in a methods sentence - 'lovastatin' occurs exactly once in the whole body"
}
what this does to premise i
{
  "premise": "(i) at this anchor, TOP2-mediated DSB rate is an increasing function of CTCF/cohesin occupancy; remove occupancy and breaks fall",
  "Q107_status": "strong correlational leg, zero directional evidence - the element is not a hotspot in HCT116 (94.004th chr11 percentile) so the depletion cannot remove a phenotype that is not there",
  "upgraded_here": "the mismatch is not only cell TYPE. The HCT116 arm is lovastatin-arrested in G1, and this graph's own measurement of Gothe's DRB arm puts the element at the 98.0th chr11 percentile untreated and the 71.0th under elongation blockade - so fragility at this element is transcription-coupled, and the one directional cohesin experiment was performed in a growth-arrested cell.",
  "what_would_settle_it": "a cohesin degron with a DSB readout in a cycling human B-cell precursor - which is exp-cohesin-degron-break-map-in-a-human-b-precursor-at-kmt2a"
}
it cites the fusion hotspot literature without naming the gene
{
  "verbatim": "TOP2 colocalized with CTCF/RAD21 binding at recurrent translocation hotspots found in ETO therapy-related acute myeloid leukemia (t-AML)",
  "cited": [
    "Ashour 2015",
    "Cowell 2012",
    "Gomez-Herreros 2017"
  ],
  "and": "'Translocated genes found in t-AML are frequently transcribed within the same transcription factory'",
  "registry_significance": "the row 'a gene-scoped query cannot reach the experiment that would settle a gene-scoped dispute' was verified at ABSTRACT level. It is now verified at FULL TEXT: 63,630 characters engaging the t-AML fusion-hotspot literature with zero mentions of the gene."
}
How this could bite us back
  • Nothing here nominates a target or supports an intervention. It is a reading of one paper's text and methods, and its only clinical bearing is negative: it removes a reason to believe the cohesin-occupancy arm of this graph's break model has been tested in a relevant cell.
  • The G1-arrest observation must not be used to dismiss Canela 2019. The arrest is a defensible design for isolating transcription-and-architecture-driven breakage from replication-associated breakage, and the paper's genome-wide claim is unaffected by it.
Still open
  1. Is KMT2A transcribed in lovastatin-arrested G1 HCT116? The arrest is now confirmed from the paper, so the question Q107 proposed is well-posed - and an ENCODE or GTEx lookup cannot answer it, because the condition is drug-arrested HCT116 and not a standard cell state. RNA-seq in that exact condition is the answer and may not exist.
  2. Does the element behave like the genome average under cohesin depletion? The paper's claim is genome-wide and proportional; Q107's percentile data say the element is not a hotspot in HCT116 at all. Those are compatible and jointly uninformative about the element.
  3. Which side of the transcription-dependence contradiction is right at this locus? The G1 explanation for the HCT116 non-hotspot requires Gothe's transcription-dependent model; Canela 2017's transcription-independent model predicts fragility should persist in G1.
  4. Do the deposited HCT116 END-seq tracks show the element at all in the untreated arm? Q107 measured 92.893rd percentile untreated and 94.004th with etoposide - a 1.18-fold ratio against the paper's own >10-fold threshold - which is consistent with the locus being silent in this condition rather than with the assay failing.
Every source behind this entry (2)
  1. E1strong confidence 4 of 5: Replicated in peer-reviewed work by more than one group.paper

    Canela et al. 2019's full text contains no mention of MLL or KMT2A under any spelling. Retrieved via efetch db=pmc for PMC8170508 and analysed as 63,630 characters of body text, the token counts are zero for MLL, KMT2A, Kmt2a, 'mixed lineage', 'mixed-lineage', 11q23, HRX, ALL-1, AF4, AFF1, AF9, MLLT3, t(4;11) and t(9;11); the single ENL match is the substring inside 'enlarged'. Positive controls in the same document return TOP2 216, END-seq 46, cohesin 37, RAD21 22 and etoposide 18, establishing that the document was loaded and searched. The paper's cohesin-depletion claim is genome-wide and carried in supplementary figures: in MEFs the END-seq reduction was proportional to the cohesin reduction at 27% and 26%, and RAD21 depletion produced a precipitous genome-wide END-seq drop in HCT-116 in Figures S1B-S1D.

    PMID 31202577
    What this evidence cannot show

    This is a token census over the retrievable PMC body, which is an author manuscript. A result present only inside a figure image, or only in a supplementary data file not carried in PMC, would not be seen - so this bounds what the paper SAYS, not what its data contain. The data are genome-wide, so the KMT2A locus IS within the assay's scope; the negative is about the paper's claims. Counts were taken case-insensitively and the only non-zero match was read in context before being discarded.

    Evidence that argues against this (2)
    • A genome-wide claim is arguably STRONGER support for premise (i) than a locus-specific panel would be, which cuts against using this absence as reassurance. If cohesin depletion drops END-seq signal genome-wide and proportionally, the general mechanism is supported and the question at this element becomes whether the element behaves like the genome average - which Q107's percentile data says it does not in HCT116 and does in blood-lineage cells. The absence of a KMT2A panel removes a contradiction; it does not add support to the element-specific reading.
    • Reading a zero token count as 'this paper makes no claim about our locus' assumes the authors would have named the gene had they analysed it. They engage the t-AML translocation-hotspot literature explicitly while naming no hotspot gene at all, so their convention is to discuss the class rather than the locus - which means a KMT2A-relevant analysis could exist in their figures under a generic label.
  2. E2established confidence 5 of 5: Multiple independent lines of evidence; treated as settled in the field.paper

    The HCT116 RAD21-degron arm - the only directional cohesin-depletion experiment with a DSB readout in a human cell in this dispute - was performed in G1-arrested cells. The methods state that HCT-116-CMV-OsTir1 and HCT-116-RAD21-mAID-Clover cells, from Natsume et al. 2016, were arrested in G1 by treatment with 20 microM lovastatin for 24 hours, then treated with auxin for 4 hours before 50 microM etoposide for 30 minutes. The word lovastatin appears exactly once in the body, in that sentence.

    PMID 31202577
    What this evidence cannot show

    A single methods sentence, quoted as written; no figure or supplementary methods file was consulted, so whether the G1 arrest was verified by flow cytometry in this study, and how complete it was, is not established here. Nothing in the sentence states transcriptional output, so the step from 'G1-arrested' to 'transcription-coupled fragility should be lowest' is an inference built on this graph's own DRB measurement and not on anything the paper asserts.

    Evidence that argues against this (3)
    • This cuts against itself under the authors' own model. Canela 2017 - the companion paper - argues that loop-anchor breakage is largely transcription-INDEPENDENT, and under that model a G1 arrest would not be expected to abolish fragility at a loop anchor, so the arrest would not explain the HCT116 non-hotspot. The G1 explanation only works if Gothe's transcription-dependent model is right, and the two models are the unresolved contradiction already recorded in mechanism-loop-anchor-dsb-hotspots. So this finding does not settle premise (i); it makes the answer depend on which side of that contradiction is correct.
    • Lovastatin is an HMG-CoA reductase inhibitor with pleiotropic effects well beyond cell-cycle arrest - depleting isoprenoids and blocking protein prenylation among them - so attributing the HCT116 result specifically to G1 or to transcriptional state is underdetermined by this observation. A G1 arrest achieved by a different method, or a cycling-cell degron arm, is what would separate them.
    • The arrest may be the correct experimental choice rather than a confound. Etoposide-induced END-seq signal in cycling cells is contaminated by replication-associated breakage, and arresting in G1 isolates transcription-and-architecture-driven breakage - which is the question the paper is asking. Reading the arrest as a limitation assumes our question, not theirs.
Connected to