start / graph / hyp-the-kmt2a-breakpoint-cluster-is-an-rpa-loading-site-and-the-rpa-reserve-is-a-therapy-related-leukaemia-risk-modifier

HypothesisspeculativeQ192-the-rpa-heterotrimer

If RPA loading at the KMT2A breakpoint cluster region is the licensing step for the etoposide-induced translocation, then the nuclear RPA reserve is a modifier of therapy-related KMT2A-rearranged leukaemia risk - and an ATR inhibitor, which kills by exhausting RPA, would raise that risk rather than lower it

also called RPA reserve and t-AL risk · the KMT2A BCR as an RPA-loading site

After chemotherapy with etoposide, the DNA-coating protein RPA piles up on the exact stretch of the KMT2A gene where the cancer-causing break happens - and it piles up more, and for longer, when the damage-sensing kinase ATM is missing. If that piling-up is what allows the translocation to form, then how much RPA a cell has spare becomes a risk factor, and drugs that use up a cell's RPA could make therapy-caused leukaemia more likely, not less.

The same thing, for a scientist

**Statement.** RPA loading at the KMT2A 11q23 breakpoint cluster region is a required intermediate in the formation of etoposide-induced KMT2A translocations, and the *duration and spatial extent* of that loading - not merely its occurrence - determine whether the lesion is repaired faithfully or recombined illegitimately. If so, the **nuclear RPA reserve is a quantitative modifier of therapy-related leukaemia risk**, and any agent that depletes it during topoisomerase-II-poison exposure shifts the balance toward translocation. **The two published legs.** (1) **RPA, RAD51 and INO80 all enrich at the BCR after etoposide**; in ATM-proficient cells RPA enrichment is transient and back to baseline within 30 minutes, while in ATM-deficient cells it is **spatially extended and persistent** - and ATM deficiency independently increases the split-MLL FISH signal and the number of BCR-containing translocation junctions recoverable by LDI-PCR [E1]. (2) **The nuclear RPA pool is finite and exhaustible**; lowering RPA accelerates fork breakage and raising it delays replication catastrophe even without ATR activity [E2]. **Origin: data-pattern plus literature-gap.** The two papers have never been cited together, on this beat's check; PMID 24267891 was uncited anywhere in this corpus before cycle 160 and PMID 21048951 is cited once, in `gene-rad51`. **Prior probability: 0.15.** The occupancy result is real and the exhaustion result is real; the *join* - that reserve size modulates translocation outcome - is unsupported by any direct measurement, and the authors of the occupancy paper propose a different sufficient mechanism (excess RAD51 loading) that does not require the reserve to be limiting.

Go deeper

What is actually established, so the hypothesis is not mistaken for it

Established 1: etoposide induces RPA enrichment at the BCR; ATM deficiency makes that enrichment more extensive and more persistent; ATM deficiency increases 11q23 translocations; RAD51 over-expression also increases them; INO80 shows the same ATM-dependent pattern; bleomycin, which is not associated with 11q23 translocations, does not produce the effect.

Established 2: nuclear RPA is finite; its exhaustion converts stable stalled forks into nucleus-wide breakage; RPA levels set the threshold.

NOT established, and this is the hypothesis:

that RPA loading is required for the translocation rather than merely correlated with damage there;
that the duration of loading, rather than the amount of damage, is the variable;
that the nuclear reserve is ever limiting at a single locus (the exhaustion work is a genome-wide, high-ssDNA regime and a single 8.3 kb locus is a very different demand);
that any of this transfers to a haematopoietic progenitor, the cell in which therapy-related KMT2A-rearranged leukaemia actually arises.

And the authors' own alternative is sufficient without the reserve. Sun et al. propose that ATM deficiency permits excessive RAD51 loading, which makes the damaged chromatin hyper-recombinogenic. On that model RPA is an intermediate whose level does not need to be limiting at all. This hypothesis is a competitor to a published model, not an extension of it, and the discriminating experiment is in data.experiments.

The clinical shape, which is what makes it worth testing

Therapy-related KMT2A-rearranged leukaemia after etoposide is a known, feared and mechanistically unexplained complication. This hypothesis makes it a quantitative risk with a measurable modifier, and it makes three predictions that cut against current therapeutic instinct:

An ATR inhibitor given with or near a topoisomerase-II poison should INCREASE therapy-related translocation frequency, because ATR inhibition kills by exhausting RPA 2 and the BCR needs RPA handled correctly 1. ATR/ATM inhibition already shows antitumour activity in MLL-rearranged AML mouse models 3, so the combination is not hypothetical.
ATM-inhibitor exposure should do the same, more directly, since ATM deficiency is the condition under which the persistent loading and the extra translocations were both measured 1.
Conversely, HR inhibition during etoposide might REDUCE therapy-related leukaemia - which is the direction Sun et al. themselves propose, and it is the opposite of the usual 'inhibit repair to sensitise the tumour' argument 1.

Prediction 3 is not this hypothesis's own prediction - it follows from the occupancy paper regardless of whether the reserve matters - and it is listed because it is the clinically loudest consequence of the underlying observation and a reader should not attribute it here.

Facts and figures

statement	RPA loading at the KMT2A 11q23 breakpoint cluster region is a required intermediate in etoposide-induced KMT2A translocation, and the duration and spatial extent of that loading - governed by ATM and by the available nuclear RPA reserve - determine whether the lesion is repaired faithfully or recombined illegitimately. Therefore the nuclear RPA reserve is a quantitative modifier of therapy-related KMT2A-rearranged leukaemia risk.
origin	data-pattern + literature-gap
prior probability	0.15
why that prior	Both legs are individually solid published results, which is why the prior is not lower. It is not higher because (a) the two papers have never been cited together and nobody has proposed the join, (b) the scale mismatch between genome-wide exhaustion and an 8.3 kb locus is severe and unaddressed, (c) the occupancy paper's own BrdU control removes the fork-stalling route that would make the locus an RPA sink, and (d) the authors offer a sufficient alternative mechanism - excessive RAD51 loading - in which RPA supply need not be limiting.
scale problem	THE WEAKEST JOINT, stated explicitly. The exhaustion result is a genome-wide regime with thousands of unscheduled origins; an 8.3 kb locus after a 10-minute pulse is a vastly smaller demand. The hypothesis needs LOCAL scarcity or a threshold effect that neither paper supports. If a reader discards this hypothesis on one ground, it should be this one.
what would make it much more interesting	If partial RPA reduction moved the junction rate, the hypothesis would predict that ANY replication-stress-inducing co-medication is a therapy-related-leukaemia risk factor, which is a testable population-level claim and would be checkable against existing t-AL cohort data without new wet work.
absence class	never-measured
absence class note	No RPA-dose experiment at any translocation breakpoint has ever been done, on this beat's reading. But the cost is moderate rather than years: the assay exists in full (PMID 21048951 supplies primer design, dose, timing and controls) and the perturbations are standard siRNA and over-expression.

predictions

[
  "In a KMT2A-rearranged or normal haematopoietic cell, etoposide induces RPA enrichment at the BCR relative to a control locus, with the same transient kinetics seen in ATM-proficient fibroblasts (the basic replication; if this fails the hypothesis is dead at step one).",
  "Partial RPA knockdown (a level that does not itself kill, as in the exhaustion work) during etoposide exposure INCREASES the frequency of BCR translocation junctions recoverable by LDI-PCR or of split-MLL FISH signals.",
  "RPA over-expression during etoposide exposure DECREASES that frequency.",
  "An ATR inhibitor co-administered with etoposide increases BCR translocation frequency, in the same direction as ATM loss.",
  "Cells with a larger baseline chromatin-bound RPA reserve, measured by single-cell imaging, show fewer etoposide-induced BCR junctions than matched cells with a smaller reserve."
]

falsifiers

[
  "SCOREABLE FORM: In a human CD34+ haematopoietic progenitor or a KMT2A-germline B-lineage line, partial RPA1 or RPA2 knockdown to 40-60% of control protein level (western, quantified) during a 100 uM / 10 min etoposide pulse produces NO INCREASE in BCR translocation junction frequency - defined as a fold-change below 1.5x versus scrambled control, measured by LDI-PCR junction counts over at least 3 independent experiments, with viability matched to within 10%. If knockdown does not move the junction rate, the reserve is not a modifier and the hypothesis is refuted.",
  "SCOREABLE FORM: RPA enrichment at the BCR after etoposide, measured by ChIP-qPCR against a beta-globin control in an ATM-proficient haematopoietic cell, fails to exceed 1.5x the control locus at the t56/bt56 primer sets at the 0-minute timepoint. That would mean the founding observation does not transfer out of fibroblasts and the whole hypothesis has no locus.",
  "SCOREABLE FORM: An ATR inhibitor at a concentration that measurably depletes chromatin-bound RPA (single-cell imaging, median intensity reduced by at least 30%) co-administered with etoposide produces no increase, or a decrease, in BCR junction frequency (fold-change below 1.5x). This is the clinically decisive falsifier and it also discriminates against the RAD51-excess model, which predicts no RPA-dose dependence.",
  "SCOREABLE FORM: If RPA over-expression and RPA knockdown move BCR junction frequency in the SAME direction, the effect is not a reserve effect and is more likely a general toxicity artefact."
]

experiments

[
  "exp-q192-rpa-chip-at-the-kmt2a-breakpoint-cluster-in-a-kmt2a-rearranged-blast"
]

analyses

[
  "ana-q192-the-complex-form-query-the-rpa2-decline-left-unrun-finds-rpa-at-the-kmt2a-breakpoint-hotspot"
]

competing model

{
  "name": "excessive RAD51 loading (the authors' own model)",
  "statement": "ATM deficiency permits excessive RAD51 loading at the BCR, making the damaged chromatin hyper-recombinogenic; RPA is an intermediate whose abundance need not be limiting.",
  "how_to_discriminate": "The RPA-dose experiments. The RAD51-excess model predicts NO dependence of junction frequency on RPA level, because RPA is not rate-limiting in it; this hypothesis predicts a monotone dependence in both directions. A single knockdown/over-expression pair discriminates them.",
  "why_it_is_favoured_a_priori": "It is the published model, it is supported by a direct RAD51 over-expression experiment in the same paper, and it does not require the scale argument this hypothesis needs."
}

measurements

[
  {
    "quantity": "count",
    "subject": "entries in this corpus citing PMID 24267891 before cycle 160",
    "value": 0,
    "units": "entries",
    "n": 3086,
    "universe": "scripts/check-citation-reuse.py 24267891 over all node files, 2026-09-24",
    "null": null,
    "note": "the RPA-exhaustion paper was uncited here; PMID 21048951 was cited once (gene-rad51). The two have never appeared in the same entry, which is why this join had not been proposed."
  }
]

How this could bite us back

THE CENTRAL BITE-BACK, and it runs against current therapeutic instinct: if this hypothesis holds, an ATR inhibitor - a drug class already being taken into MLL-rearranged AML - would RAISE the risk of therapy-related KMT2A-rearranged leukaemia when given near a topoisomerase-II poison, because it kills by exhausting the very reserve the breakpoint needs handled correctly. That risk would fall on surviving normal progenitors, not on the treated blasts, and would appear years later.
The same argument applies to ATM inhibitors more directly, since ATM loss is the exact condition under which the persistent BCR loading and the extra translocations were both measured.
And the mirror-image bite-back on the opposite strategy: Sun et al.'s own proposal is that inhibiting HR during etoposide might REDUCE secondary leukaemia. If that is right, an HR-restoring or HR-enhancing intervention would do the reverse. Any agent touching this axis has a late, population-level toxicity that no short-term efficacy endpoint can see.
Because the predicted harm is a second cancer appearing years after treatment, no conventional trial endpoint would detect it, and the hypothesis is therefore one whose refutation is much cheaper than its confirmation. That asymmetry is an argument for running the falsifiers early rather than for treating the concern as speculative and deferring it.

Still open

Is the nuclear RPA reserve ever limiting at a single 8.3 kb locus? This is the scale question the hypothesis stands or falls on, and neither cited paper addresses it. A biophysical estimate - RPA molecules per nucleus versus ssDNA generated by a 10-minute etoposide pulse - could be made from published numbers and would sharpen or kill the prior before any wet work.
Does the Sun 2010 result transfer out of SV40-transformed fibroblasts into a haematopoietic progenitor? Every arm of the founding experiment is a fibroblast pair or a CML line, and therapy-related KMT2A-rearranged leukaemia arises in a progenitor.
Are there existing therapy-related leukaemia cohorts with germline variation in RPA1, RPA2 or RPA3, or with measurable RPA expression, that could be interrogated without new experiments? Not searched by this beat.
Does the KMT2A BCR load RPA because of topoisomerase-II cleavage specifically, or would any DSB there do it? Bleomycin did NOT produce the effect (PMID 21048951), which is a striking agent specificity and suggests the lesion type matters - and that is unexplained.

Every source behind this entry (3)

E1strong confidence 4 of 5: Replicated in peer-reviewed work by more than one group.paper
ATM deficiency increases etoposide-induced 11q23 translocations and simultaneously changes RPA behaviour at the KMT2A breakpoint cluster region: ChIP with anti-RPA antibodies showed etoposide-induced RPA enrichment in the BCR of ATM-proficient 11-4 cells that returned to normal within 30 minutes of recovery, while in ATM-deficient AT5BIVA cells the enrichment extended into regions adjacent to bt56/t56 and remained high at 30 minutes; RAD51 and INO80 showed the same ATM-dependent over-accumulation; RPA, RAD51 and INO80 expression levels were unchanged by etoposide or ATM status; bleomycin, which is not associated with 11q23 translocations, did not produce a significant increase; ATM-deficient cells showed more split-MLL FISH signals and more BCR translocation junctions by long-distance inverse PCR; and RAD51 over-expression independently increased split signals after etoposide.
ATM modulates the loading of recombination proteins onto a chromosomal translocation breakpoint hotspot. · Sun J, Oma Y, Harata M, Kono K, Shima H, Kinomura A, et al. · PLoS One 5(10):e13554 2010
PMID 21048951 DOI
What this evidence cannot show
Full text read from PMC 2026-09-24; figures not viewed, so all quantitative statements are the authors' prose and no effect size is recorded. NO FUSION IN ANY ARM and no haematopoietic progenitor: AT5BIVA is an SV40-transformed ATM-deficient fibroblast line, 11-4 its chromosome-11-complemented derivative, BV173 a CML-derived B-cell line. Etoposide 100 uM for 10 min is pharmacological. Candidate-locus ChIP-qPCR with a single control region (beta-globin, 11p15) - it cannot say the BCR is unusual genome-wide. Three independent experiments for the RPA panels. Single group, little follow-up in sixteen years. The antibody is named anti-RPA34 in Methods and anti-RPA32 in figure legends, with no clone (rule 74).
Evidence that argues against this (1)
- The paper's own BrdU control removes the mechanism this hypothesis would most naturally invoke: BrdU incorporation after etoposide did NOT differ significantly between the BCR and the beta-globin control locus, so replication-fork stalling is not elevated at the BCR and 'the BCR is a fragile site that consumes RPA' is not supported. The authors also state that the nature of the etoposide-induced damage in the BCR is still unclear, and they offer excessive RAD51 loading - not RPA supply - as the sufficient mechanism. Occupancy is not translocation and the causal link is their proposal. Sun J et al. 2010
E2established confidence 5 of 5: Multiple independent lines of evidence; treated as settled in the field.paper
The nuclear RPA pool is finite and exhaustible: stalled forks in ATR-deficient cells are initially stable but break nucleus-wide once unscheduled origin firing generates enough ssDNA to exhaust nuclear RPA; partial reduction of RPA accelerated fork breakage; forced elevation of RPA delayed replication catastrophe even without ATR activity; and unscheduled origin firing broke stalled forks even in cells with active ATR.
ATR prohibits replication catastrophe by preventing global exhaustion of RPA. · Toledo LI, Altmeyer M, Rask MB, Lukas C, Larsen DH, Povlsen LK, Bekker-Jensen S, Mailand N, Bartek J, Lukas J · Cell 155(5):1088-103 2013
PMID 24267891 DOI
What this evidence cannot show
Abstract level; full text not opened. Human tumour cell lines, single-cell imaging of chromatin-bound RPA. NO leukaemia, NO fusion, NO topoisomerase-II poison, and crucially NO LOCUS-LEVEL component - this is a genome-wide, massive-ssDNA regime. The expected_absent_terms field records that KMT2A and MLL do NOT appear in this paper: the join to the KMT2A locus is this hypothesis's own, not the paper's. This paper was uncited anywhere in this corpus before cycle 160.
Evidence that argues against this (1)
- The scale mismatch is the strongest argument against this hypothesis and it should be read first: exhaustion in this paper requires unscheduled firing of thousands of origins generating genome-scale ssDNA. An 8.3 kb breakpoint cluster region after a 10-minute etoposide pulse is a demand many orders of magnitude smaller, and there is no reason from this paper to think the nuclear reserve is limiting at a single locus. The hypothesis therefore requires a LOCAL rather than global scarcity argument, which neither paper supports, and that is its weakest joint. Toledo LI et al. 2013
E3moderate confidence 3 of 5: One strong study, or several smaller consistent ones.paper
Targeting the kinase activities of ATR and ATM exhibits antitumoral activity in mouse models of MLL-rearranged AML, so an ATR or ATM inhibitor is a realistic co-medication in this disease rather than a hypothetical one.
Targeting the kinase activities of ATR and ATM exhibits antitumoral activity in mouse models of MLL-rearranged AML. · Morgado-Palacin I, Day A, Murga M, Lafarga V, et al. · Science Signaling 9(445):ra91 2016
PMID 27625305 DOI
What this evidence cannot show
Title-level claim; already cited four times in this corpus and those entries hold its content. Mouse models of MLL-rearranged AML, not the index B-lineage disease, and not a therapy-related-leukaemia model. It is cited here ONLY to establish that the drug class is being taken into this disease - it measures nothing about RPA and nothing about translocation formation.
Evidence that argues against this (1)
- This citation could easily be over-read into support for the hypothesis and it is not: an antitumour effect of ATR inhibition says nothing about translocation risk, and the two effects would occur in different cells (leukaemic blasts versus surviving normal progenitors). A drug that kills fusion-positive blasts efficiently could still raise the translocation rate in normal haematopoiesis, and no experiment cited anywhere in this entry measures that. This repository 2026

Connected to

depends on →The replication protein A heterotrimer (RPA1/RPA2/RPA3) - the ssDNA clamp that every resection and replication-stress claim in this corpus silently depends on, and which had no entry until cycle 160
rests on the finite-pool property and on the locus-level RPA loading measurement
related to →KMT2A / MLL1 — the gene that breaks
the locus whose breakpoint cluster region loads RPA after etoposide
related to →Etoposide (VP-16) - the drug that makes this disease and cures several others, and the only entry in this graph whose main finding is that you must not stop using it
the agent that induces RPA loading at the hotspot, and the agent associated with therapy-related KMT2A-rearranged leukaemia
related to →RAD51 - the one repair protein measured AT the KMT2A translocation hotspot, and ATM controls how much of it loads there
the competing published model - excessive RAD51 loading - in which RPA supply need not be limiting
related to →ATR - the replication-stress kinase with the best preclinical case in this disease and the best published reason not to inhibit it
an ATR inhibitor kills by exhausting RPA, so this hypothesis predicts it RAISES therapy-related translocation risk
← testsRepeat the one published RPA-at-the-KMT2A-breakpoint experiment in a haematopoietic cell, with an RPA-dose arm - the protocol is fully specified in a 2010 paper whose every arm used fibroblasts and a CML line, and whose primer sets, dose and controls are all published
repeats a fully published protocol in the cell type it has never been used in, and adds the RPA-dose arm that discriminates the two models