Oncofusion

start / graph / exp-replication-stress-readout-kmt2a-r

Experimentmoderate confidence 3 of 5: One strong study, or several smaller consistent ones.Q-fa-replication

Measure replication stress directly in KMT2A-rearranged cells instead of inferring it from a dependency ranking

also called DNA fibre assay KMT2A · decisive replication stress experiment

The whole Fanconi story rests on an argument that KMT2A leukaemia cells must be struggling to copy their DNA, inferred from which genes they need. Nobody has watched them copy their DNA. This is how you would: stretch out single DNA molecules from the cells, measure how fast and how evenly the copying machinery moves, and compare against leukaemia cells without the fusion.

The same thing, for a scientist

The transcription–replication-conflict hypothesis has never been tested in this disease. A PubMed search for MLL-AF4, MLL-AF9, MLL-rearranged or KMT2A-rearranged against DNA fibre, DNA combing, fork speed, fork progression or origin firing returns one record, which on title-checking is unrelated [E1]. The experiment is a single-molecule replication assay (DNA fibre / molecular combing, CldU–IdU dual pulse) on KMT2A-rearranged versus comparator leukaemia cells, with three readouts that fail independently — fork speed, fork asymmetry as a measure of stalling, and inter-origin distance as a measure of compensatory origin firing — plus S9.6 or RNase-H1-catalytically-dead-GFP imaging for R-loop burden and FANCD2 foci for pathway engagement. The comparator must be lineage-matched, because this project has shown the FA signal has both a B-lymphoid component and a KMT2A component [E2], and the isogenic arm — fusion knockdown in the same line — is what ties any phenotype to the fusion rather than to the line. The kill condition is explicit: **if KMT2A-rearranged cells show no fork-speed deficit, no excess asymmetry and no reduced inter-origin distance relative to lineage-matched comparators, the replication-stress mechanism is dead** and the FA dependency has to be explained some other way, most plausibly by the ALDH2/aldehyde route [E3].

Go deeper
Design

Cell material, in ascending order of what it costs and what it proves.

tiermaterialwhy
1SEM and RS4;11 (KMT2A::AFF1, B-ALL) and MV4-11 and MOLM-13 (KMT2A::MLLT3 / KMT2A-r, AML) against lineage-matched non-KMT2A lines: REH or NALM-6 for B-ALL, HL-60 or U937 for AMLthe same lines the DepMap signal was measured in, so a negative here kills the hypothesis in its own system
2isogenic: doxycycline-inducible shRNA or degron knockdown of the fusion in SEM, timepoints before proliferation collapsesties any phenotype to the fusion, not to the line — this is the arm that makes the experiment interpretable
3primary KMT2A-rearranged infant B-ALL blasts against primary non-KMT2A infant/childhood B-ALL blasts, short ex vivo culturethe only tier that speaks to risk-fa-signal-cell-line-artefact, and the only one with a real chance of contradicting tiers 1 and 2

Assay. Dual-pulse CldU (20 min) then IdU (20 min), spread by DNA fibre or molecular combing, ≥200 tracks per condition, three biological replicates, scored blind to genotype.

Primary readouts, which fail independently.

  • Fork speed (kb/min from IdU track length). Prediction under the hypothesis: slower in KMT2A-rearranged.
  • Fork asymmetry (ratio of the two arms of a bidirectional origin). Prediction: more asymmetric, indicating stalling.
  • Inter-origin distance. Prediction: shorter, indicating dormant-origin firing to compensate.

Secondary readouts.

  • R-loop burden by S9.6 immunofluorescence with an RNase-H1 digestion control, or by catalytically dead RNase-H1-GFP, which avoids the well-known S9.6 specificity problems.
  • FANCD2 foci and FANCD2-Ub western, as a direct measure of whether the FA pathway is engaged at baseline — a dependency with no pathway engagement would be a strong hint the requirement is non-canonical, which mech-fa-core-complex-heterogeneity-kmt2a already suspects.
  • 53BP1 nuclear bodies in G1, which report under-replicated DNA inherited through mitosis.

Controls. Hydroxyurea or aphidicolin as a positive control for each readout. A menin inhibitor arm (revumenib) as a mechanistic probe: if replication stress is downstream of the fusion's transcriptional programme, shutting that programme down should relieve it, and if it does not, the stress is not transcription-coupled.

The second experiment, which is cheaper and should be done first

Before any fibre spreading, two things can be done at a desk.

A. Does the FA dependency track ALDH2 rather than KMT2A? ALDH2 expression is available for every DepMap line. Regress FA-gene Chronos on ALDH2 expression and on KMT2A status across the ~54-line leukaemia panel and see whether KMT2A status adds anything after conditioning. If it does not, mech-aldehyde-fa-dependency-rival replaces the replication story and this experiment should not be run 3.

B. Do the fusion's target loci overlap the loci where FANCD2 acts? Published MLL-AF4 ChIP-seq and published FANCD2 ChIP-seq both exist. FANCD2 accumulates at the central region of large fragile genes under replication stress 4. If the fusion's occupied domains do not overlap those genes, the spatial premise of the mechanism fails and no cell needs to be grown.

Both are days of analysis against months of bench work, and either could make the bench work unnecessary.

What each outcome would mean
resultreading
slower forks, more asymmetry, shorter inter-origin distance, all reversed by fusion knockdownthe mechanism is supported and the FA dependency has an explanation
replication phenotype present but not reversed by fusion knockdownthe stress is a property of the lines, not of the fusion — risk-fa-signal-cell-line-artefact wins
no replication phenotype, but FA dependency confirmed by knockdownthe hypothesis is dead and the aldehyde route or a non-canonical FANCL/UBE2T substrate is the explanation
replication phenotype present but no FANCD2 engagementthe FA genes are being used for something other than the canonical pathway, consistent with the core-complex spread
primary blasts negative where lines are positivethe whole finding is a culture artefact and the beat should be withdrawn

The fourth row is the one to plan for, because the core-complex spread already predicts it.

Facts and figures
designDual-pulse CldU/IdU single-molecule replication assay (DNA fibre or molecular combing) in KMT2A-rearranged versus lineage-matched non-rearranged leukaemia cells, with an isogenic fusion-knockdown arm and a primary-blast tier; secondary R-loop, FANCD2-foci and G1 53BP1-body readouts.
readoutfork speed (kb/min), fork asymmetry, inter-origin distance; S9.6 or dead-RNaseH1-GFP R-loop signal; FANCD2 foci and FANCD2-Ub; G1 53BP1 nuclear bodies
controlshydroxyurea and aphidicolin positive controls; lineage-matched non-KMT2A lines (REH/NALM-6, HL-60/U937); isogenic dox-inducible fusion knockdown; revumenib arm to test whether the phenotype is downstream of the fusion's transcriptional programme; blinded scoring, >=200 tracks per condition, 3 biological replicates
cost tiermed
timelinecheap desk analyses (ALDH2 conditioning; fusion-target vs large-fragile-gene overlap) in days; tiers 1-2 in 3-4 months; tier 3 dependent on primary sample access
kill conditionno fork-speed deficit, no excess fork asymmetry and no reduced inter-origin distance in KMT2A-rearranged cells against lineage-matched comparators kills the transcription-replication-conflict mechanism regardless of what the FA dependency does
do first
[
  "regress FA-gene Chronos on ALDH2 expression and KMT2A status across the DepMap leukaemia panel",
  "intersect published MLL-AF4 ChIP-seq domains with published large fragile genes / FANCD2 sites",
  "search published in vivo and PDX CRISPR screen supplements in KMT2A-rearranged leukaemia for FA genes"
]
How this could bite us back
  • If replication stress is confirmed and relieved by menin inhibition, then combining a menin inhibitor with an FA-pathway inhibitor would be antagonistic, and the two most promising ideas in this graph would be pulling against each other.
  • A positive fibre result would invite ATR-inhibitor combinations, and ATR inhibitors already carry marrow toxicity that adds to the FA pathway's own.
  • The experiment cannot distinguish replication stress caused by the fusion from replication stress caused by whatever else these lines carry unless the isogenic arm works, and inducible fusion knockdown collapses proliferation within days, which is the same axis the assay reads out.
Still open
  1. Is there an isogenic KMT2A-fusion-inducible system in a B-lymphoid background good enough for the tier-2 arm, or would it have to be built?
  2. Can enough primary infant KMT2A::AFF1 blasts be obtained for fibre analysis, which needs a lot of cells relative to most primary-sample assays?
  3. Should the revumenib arm be run first, given that a menin inhibitor relieving replication stress would be a striking and cheap result?
Every source behind this entry (5)
  1. E1strong confidence 4 of 5: Replicated in peer-reviewed work by more than one group.database

    No published study measures replication dynamics in KMT2A-rearranged cells: a title/abstract search for MLL-AF4, MLL-AF9, MLL-rearranged or KMT2A-rearranged against DNA fibre, DNA combing, fork speed, fork progression or origin firing returns one record, which on title-checking is an AML lncRNA study not about KMT2A rearrangement.

    What this evidence cannot show

    Fibre assays are frequently a single panel inside a paper about something else and would not reach the title or abstract, so this is an absence of a headline result rather than proof the measurement has never been made.

    Evidence that argues against this (1)
    • ATR inhibition does kill MLL-rearranged AML in vivo, which is indirect functional evidence that a replication-stress liability exists even though nobody has measured forks. Morgado-Palacin I et al. 2016
  2. E2moderate confidence 3 of 5: One strong study, or several smaller consistent ones.dataset

    The comparator must be lineage-matched, because the FA dependency has both a B-lymphoid component (non-KMT2A B-ALL headline-eight mean d -0.196) and a KMT2A component that survives holding lineage constant (-0.334 in AML, -0.308 in B-ALL).

    What this evidence cannot show

    Four KMT2A-rearranged B-ALL lines against nine comparators. A lineage-matched design inherits the same small numbers if it reuses the same lines.

    Evidence that argues against this (1)
    • Matching lineage does not match culture history, proliferation rate or TP53 status, and REH and NALM-6 differ from SEM and RS4;11 in all three. Only the isogenic fusion-knockdown arm removes that class of objection. Ben-David U et al. 2018
  3. E3strong confidence 4 of 5: Replicated in peer-reviewed work by more than one group.paper

    The rival hypothesis to exclude first is aldehyde genotoxicity: UBE2T and FANCL are AML dependencies because ALDH2 is silenced by hypermethylation, which predicts the dependency tracks ALDH2 status rather than KMT2A rearrangement and is testable from existing DepMap data without any experiment.

    PMID 33893150DOI
    What this evidence cannot show

    Its comparator is other cancer lineages rather than other leukaemias, so it does not directly contradict the KMT2A finding; the two have not been tested against each other.

    Evidence that argues against this (1)
  4. E4strong confidence 4 of 5: Replicated in peer-reviewed work by more than one group.paper

    FANCD2 accumulates at the central region of large fragile genes under replication stress and facilitates replication through common fragile sites, which is what makes the overlap between fusion-occupied domains and those genes a computable prediction from published ChIP-seq.

    PMID 29394375DOI
    What this evidence cannot show

    Aphidicolin-induced stress in non-leukaemia human cells, and the FANCD2 localisation is for the substrate, not for the core-complex subunits that carry this project's signal.

    Evidence that argues against this (1)
    • FANCD2 is one of the weakest members of the set in DepMap (d -0.150, line-permutation p 0.071), so designing the spatial test around FANCD2 localisation may be aiming at the wrong protein. Madireddy A et al. 2016
  5. E5strong confidence 4 of 5: Replicated in peer-reviewed work by more than one group.paper

    In vivo and patient-derived CRISPR screening in KMT2A-rearranged leukaemia is established, so tier 3 of this design is routine rather than exploratory.

    PMID 37036970DOI
    What this evidence cannot show

    That screen concerns bromodomain inhibitor resistance, and primary infant B-ALL blasts are scarce and hard to culture even briefly, which is the real constraint on tier 3.

    Evidence that argues against this (1)
    • Short ex vivo culture of primary blasts itself imposes stress and selects; a primary-sample negative could be an artefact of the handling rather than a refutation. Vujovic A et al. 2023
Connected to