Break formation, break persistence and break fate are three quantities, and every assay in this dispute reports the middle one - which is why the triad's third premise is ambiguous rather than weak
also called formation is not persistence · what END-seq actually counts · the triad's hidden fourth premise · (iii-a) versus (iii-b)
There are three different things you can count: how often DNA snaps, how many snapped ends are still open when you look, and how many of those turn into a rearrangement. Every method used in this argument counts the middle one, and the argument has been treating it as the first.
The same thing, for a scientist
The Q105 triad is (i) occupancy up implies DSBs up, (ii) occupancy up implies rearrangement down, (iii) DSBs up implies rearrangement up - and it argues that all three cannot hold. **The word 'DSBs' in (i) and (iii) does not denote the same quantity as any assay in the dispute measures.** END-seq, sBLISS and C-Fusion 3D all report the number of breaks **present at the moment of fixation**, which is formation rate multiplied by mean lifetime. So (iii) splits: **(iii-a)** *formation* up implies rearrangement up, and **(iii-b)** *persistence* up implies rearrangement up. **Only (iii-b) has evidence, and it is strong** - TDP2 and MRE11-nuclease loss, which act on rejoining and adduct removal rather than on cleavage, raise *MLL* breakage and *MLL-AF4* fusions together in isogenic human TK6 cells [E1]. **(iii-a) has no direct evidence at any locus in this dispute**, because no perturbation in any of the three studies changes cleavage rate while holding repair constant. The one assay contrast that could isolate formation - Canela's **ExoT-only versus ExoVII+ExoT** END-seq, which separates trapped TOP2 cleavage complexes from irreversible DSBs - was run for **every mouse condition and no human condition** [E2]. **Consequence for the adjudication: (iii) is not the triad's weak link. It is its ambiguous link, and disambiguating it is cheaper than testing (i).**
The three quantities, and which instrument sees which
| quantity | what it is | which assay reports it |
|---|---|---|
| formation | cleavage events per cell per unit time | nothing here directly. The nearest proxy is trapped-TOP2cc occupancy, read as the ExoVII+ExoT minus ExoT-only difference in END-seq |
| persistence | breaks open at fixation = formation x mean lifetime | END-seq, sBLISS, gammaH2AX, C-Fusion 3D probe separation - i.e. every break readout in the dispute |
| fate | fraction of breaks resolved as an interchromosomal junction | C-Fusion 3D fusion scoring; HTGTS; patient junction sequencing |
A perturbation that lengthens lifetime raises persistence and fate together without touching formation. That is what TDP2 loss does, by the mechanism its own paper names - "TDP2 rejoins DSBs". It is also what a proteasome inhibitor does (Canela 2019's epoxomicin arm) and what an NHEJ mutant does. A perturbation that raises formation - more TOP2 loading, more elongating polymerase - raises persistence too, so the two are only separable by assay design, never by perturbation direction.
Why this matters for the CTCF question specifically
If cohesin's role at the element is to hold DNA in a configuration that TOP2 cleaves, removing cohesin lowers formation. If its role is to hold the two ends together so they religate locally, removing cohesin raises lifetime and raises the chance the ends find a partner on another chromosome. These predict opposite signs for the same intervention, and they are indistinguishable by a break map alone. A cohesin degron that lowered END-seq signal would be read as 'fewer breaks form' when it could equally be 'breaks form and are rejoined faster because the anchor is gone' - which is physically odd but not excluded - or, far more plausibly in the other direction, a degron that raised the fusion rate while lowering the break map would look like a paradox and would simply be formation down, lifetime up.
This is exactly the shape of the Gray-versus-fragility dispute, and it is why hyp-ctcf-occupancy-splits-break-formation-from-rearrangement-fate is the right hypothesis and its name is the right name. What this entry adds is that the split it proposes has a measurement requirement, not only a design requirement: the experiment must include a time course and a repair-kinetics arm, or it will report a persistence change and be quoted as a formation change.
Minimum design to disambiguate, all available off the shelf: both END-seq variants (ExoT-only and ExoVII+ExoT) at three or more timepoints including a wash-out, in the same cells, alongside the fusion readout. Canela 2019 ran precisely this design - time course plus reversibility plus wash-out plus epoxomicin plus a TDP2 arm - in mouse pre-B cells, 18 paired libraries, and reduced the human arm to three unpaired ones 2.
The vocabulary this graph should use from here
Anywhere the graph writes 'breaks' in a claim about this locus, it should write one of:
- 'trapped TOP2 cleavage complexes' - what ExoVII+ExoT END-seq minus ExoT END-seq reports;
- 'unrepaired DSBs' or 'DSB burden' - what END-seq, sBLISS and gammaH2AX report;
- 'break formation rate' - a quantity no entry in this graph currently cites a measurement of at KMT2A.
And a claim of the form 'X increases breakage at KMT2A' should carry the timepoint, because at 30 min and at two days after release those are different sentences about different quantities. The deposited protocols in this dispute span 30 minutes (HCT116 END-seq), 4 hours (Gothe sBLISS) and two days after release (C-Fusion 3D fusion scoring) - a 96-fold range that the graph has been treating as one readout.
three quantities
[ "formation - cleavage events per cell per unit time; no direct measurement exists at KMT2A in any study in this dispute", "persistence - breaks open at fixation, equal to formation times mean lifetime; what END-seq, sBLISS, gammaH2AX and C-Fusion 3D probe separation all report", "fate - fraction of breaks resolved as an interchromosomal junction; C-Fusion 3D fusion scoring, HTGTS, patient junction sequencing" ]
triad premise iii split
{
"iii_a": "break FORMATION up implies rearrangement up - no direct evidence at any locus in this dispute",
"iii_b": "break PERSISTENCE up implies rearrangement up - supported at MLL in isogenic human TK6 by TDP2 and MRE11-nuclease loss"
}assay timepoints in the dispute
{
"hct116_end_seq": "30 min etoposide, G1-arrested",
"gothe_sbliss": "4 h etoposide",
"c_fusion_3d_fusion_scoring": "two days after release from etoposide",
"range": "96-fold, treated by the graph as one readout"
}- If the graph adopts this vocabulary, several existing claims become weaker as stated and will need rewording rather than deleting - which is a cost, and the reason to pay it is that a therapeutic argument built on 'fewer breaks' when the measurement was 'fewer open ends' would predict the wrong sign for any intervention that acts on repair.
- The distinction also cuts against this beat's own headline: the element being a top-0.05% persistence site in CD34+ cells does not establish that it is cleaved more often there than elsewhere - it may simply be repaired more slowly, which would be a different and arguably more actionable mechanism.
- Is break formation rate measurable at a single element with any current method? If not, premise (iii-a) is permanently untestable as stated and the triad should be rewritten around persistence, which would change which of the three premises is the target.
- Would an ExoT-only END-seq library in a human haematopoietic cell, paired with the ExoVII+ExoT variant, separate trapped complexes from breaks at the KMT2A element? Canela ran that pairing 18 times in mouse and never once in human.
- How many claims in this graph of the form 'X increases breakage at KMT2A' carry a timepoint? A corpus scan would say how far the conflation has spread, and this beat did not run one.
- E1strong confidence 4 of 5: Replicated in peer-reviewed work by more than one group.paper
Perturbations acting on rejoining rather than on cleavage raise the measured break burden and the rearrangement rate together at MLL in the same human cells: PMID 28794467 states its mechanism as 'TDP2 rejoins DSBs induced during transcription-dependent TOP2 activity in breast cancer cells and at the translocation hotspot, MLL', and PMID 31202576 reports that in isogenic TK6 'a similar increase in the percentage of cells with MLL breakage and MLL-AF4 translocations was observed upon ETO-treatment in cells deficient for TDP2 or MRE11 nuclease activity', with 'TDP2 prevented the accumulation of ETO-induced DSBs within the MLL locus and the formation of MLL fusions'.
TDP2 suppresses chromosomal translocations induced by DNA topoisomerase II during gene transcription (abstract) read alongside Spatial chromosome folding and active transcription drive DNA fragility and formation of oncogenic MLL translocations (full text, Results and Figure 7) · Gomez-Herreros F, Zagnoli-Vieira G, Ntai I, Martinez-Macias MI, Anderson RM, Herrero-Ruiz A, Caldecott KW; and Gothe HJ, Bouwman BAM, Gusmao EG, et al., Roukos V · Nature Communications 8:233; and bioRxiv/Molecular Cell 75:267-283.e12 2017What this evidence cannot show
The Gomez-Herreros claim is from its abstract; its full text was not read. The Gothe claim is from the December 2018 preprint, and its C-Fusion 3D readouts are imaging measurements at whole-gene probe resolution taken two days after etoposide release, so 'breakage' there is specifically persistent probe separation. Neither paper attempts to measure a cleavage rate, so the formation/persistence distinction drawn here is an inference from the stated mechanisms and the assay timings, not a measurement either paper reports. The systems are breast cancer cells and human lymphoblastoid TK6 - neither is B-precursor, infant, or KMT2A-rearranged.
Evidence that argues against this (1)
- A determined defence of the original reading is available: MRE11 nuclease activity removes TOP2-DNA adducts, which is a step BEFORE a free DSB exists, so MRE11-nuclease loss arguably reduces DSB formation while raising adduct burden - in which case 'MLL breakage' rising in MRE11-/H129N cells is the assay scoring adducts as breaks, and the co-movement is an assay artefact rather than a biological co-movement. That reading strengthens this entry's central point (the assay does not report what its name says) while weakening its use of MRE11 as an example. Gothe HJ et al. 2018
- E2established confidence 5 of 5: Multiple independent lines of evidence; treated as settled in the field.dataset
The one assay contrast in this literature that separates trapped TOP2 cleavage complexes from irreversible double-strand breaks - Exonuclease VII plus Exonuclease T blunting 'to detect all TOP2 lesions' versus Exonuclease T alone 'to detect only DSBs, ExoT END-seq' - was run as matched pairs for every mouse condition in GSE129524, including an 0.5-to-12.5-hour time course (GSM3714871-GSM3714882), a wash-out reversibility series, a TDP2 arm and an epoxomicin proteasome-inhibition arm, and was not run for any human sample: the three HCT116 libraries GSM3714903, GSM3714904 and GSM3714905 exist only in the ExoVII+ExoT variant with no ExoT-only counterpart.
NCBI GEO GSE129524 full sample listing (48 !Sample_title lines) and the GSM3714904 extract protocol quoting both END-seq exonuclease variants, retrieved 2026-09-22 · NCBI Gene Expression Omnibus; submitter Yaakov Maman · NCBI GEO 2019What this evidence cannot show
Bounds the deposit, not the experiments performed: an ExoT-only HCT116 library could have been generated and not deposited. The mouse time-course and reversibility arms are in LIG4-deficient pre-B cells, which cannot rejoin by NHEJ - so those paired libraries measure persistence in a background where persistence is artificially maximal, which is excellent for detecting breaks and poor for inferring normal lifetimes. The ExoVII/ExoT contrast is the submitters' own stated interpretation of their two blunting protocols and has not been independently validated here.
Evidence that argues against this (1)
- Treating the ExoT-versus-ExoVII contrast as a formation readout overstates it. Trapped TOP2cc abundance is also a product of trapping rate and reversal rate, so it is itself a steady-state quantity, not a rate. A genuine formation-rate measurement would need a pulse-chase or a rapid-induction system, which no study in this dispute has. So the honest statement is that the contrast gives an EARLIER steady state, not a rate - and that (iii-a) may not be measurable with current tools at all. NCBI Gene Expression Omnibus 2019
the split this hypothesis proposes needs a measurement design - both END-seq variants plus a time course - or it will report persistence and be quoted as formation
the TDP2 co-movement is the worked example that splits premise (iii)
- related to →Chromatin loop anchors are DNA-break hotspots - and the two papers that established it disagree about why
assay timings in this dispute span 30 min to two days after release, a 96-fold range treated as one readout