Experiment: bait the KMT2A breakpoint cluster region and sequence everything it joins to, in a fetal-like human progenitor
also called HTGTS with a KMT2A BCR bait · translocation capture from the MLL hotspot · unbiased partner map for KMT2A
Instead of asking which partners are found in patients, put a break in KMT2A on purpose and sequence every other place in the genome it sticks to. The clinical partner list then becomes a prediction you can check rather than the only thing you know.
The same thing, for a scientist
This graph's `mechanism-class-switch-recombination-programmed-breaks` names the experiment in its own open questions: translocation-capture sequencing has never been applied to the KMT2A breakpoint cluster region as a bait in a fetal-like haematopoietic progenitor, and doing so would give the first unbiased map of what KMT2A actually joins to, against the clinical partner distribution as a predicted outcome. The instrument exists and is mature: linear amplification-mediated HTGTS detects DSBs genome-wide by their translocation to a chosen bait DSB, extended the known off-target set of characterised nucleases more than tenfold, and detects translocations between targets on homologous chromosomes [E1]. The comparator is published and large: seven partners plus partial tandem duplication account for more than 90% of KMT2A recombinations across 3401 patients, with 37 recurrent partners and 63 seen once [E2]. The cells exist: human fetal liver and cord-blood CD34+ progenitors support KMT2A rearrangement assays [E3], and a donor-independent CRISPR protocol for cutting KMT2A in cord-blood HSPC is published [E4]. What has never been done is putting the three together. The experiment's single most important design decision is the bait chemistry, because a Cas9 bait makes a break that a TOP2 poison does not, and the graph already holds the evidence that these two break types are joined by pathways with opposite effects on translocation [E5].
Design1 deeper
Bait. A Cas9-induced DSB placed inside the KMT2A major breakpoint cluster region, at or adjacent to the intron 9 and intron 11 positions that dominate the clinical distribution. Run at least two bait positions, because a single bait confounds partner spectrum with bait position.
Cells. Human fetal liver CD34+ HSC and LMPP where obtainable; cord-blood CD34+ as the accessible substrate; an adult CD34+ arm and a non-haematopoietic line as the lineage controls. The lineage controls are not optional: if the partner spectrum is identical in a fibroblast, the spectrum is a property of genome geometry and not of the disease's cell of origin.
Arms. Untreated; etoposide; and a TOP2-poison-free genotoxic control. The etoposide arm is what connects the result to therapy-related disease; the untreated arm is the one relevant to the de novo infant case.
Readout. The full junction library: partner locus, junction position within the partner, and junction structure (microhomology class, insertion, deletion), which feeds directly into ana-junction-microhomology-signature-mining.
Primary comparison. Rank correlation between the HTGTS partner spectrum and the clinical partner ranking 2, run per cell type.
The interpretation that makes this worth doing
The clinical partner distribution is the product of three things: how often each partner breaks, how often it is near KMT2A when it does, and whether the resulting fusion makes a leukaemia that presents. HTGTS measures the first two together and is blind to the third by construction, because it reads junctions out of a population of cells regardless of whether they proliferate.
That blindness is the point. The difference between the HTGTS spectrum and the clinical spectrum is an estimate of selection. If the two rank alike, partner choice is set at the DNA level and selection adds little. If a partner is abundant in HTGTS and rare in patients, that partner makes a fusion that does not work - which is a functional statement about the fusion protein, obtained from a DNA experiment. No other design in this beat can separate those terms.
The three ways it goes wrong
-
Bait chemistry. A Cas9 break is blunt-or-small-overhang with no protein adduct; a TOP2 break carries a covalent adduct and a 4-nt 5'-cohesive overhang, and the end-joining route that handles it SUPPRESSES translocation rather than generating it 5. A Cas9-baited partner spectrum is therefore a map of what KMT2A is near and what else is broken, not a faithful model of a TOP2-made fusion. Mitigation: run an etoposide arm without a Cas9 bait using an endogenous-break HTGTS variant, and compare.
-
HTGTS measures joined events. It convolves break rate with joining propensity 1, so a partner that breaks often and rejoins locally is under-counted. This cannot be fully removed; it can be bounded by comparing against a break map that does not require joining, which is why
ana-dsb-mapping-assays-at-the-kmt2a-locusbelongs beside this experiment. -
Cell number. Translocation is rare, so the library depth needed to see the 63-single-occurrence tail of the clinical distribution is large, and fetal liver material is limited. Power for the head of the distribution (the seven commonest partners) is achievable; power for the tail probably is not, and the analysis should pre-specify which it is testing.
| design | LAM-HTGTS with a Cas9 bait DSB in the KMT2A major breakpoint cluster region, at two or more bait positions, in human fetal liver and cord-blood CD34+ progenitors |
|---|---|
| readout | genome-wide partner spectrum plus junction structure per event |
| primary comparison | rank correlation between HTGTS partner spectrum and the 3401-patient clinical partner ranking, per cell type |
| what the difference estimates | the gap between the HTGTS spectrum and the clinical spectrum is an estimate of SELECTION on the fusion product, because HTGTS is blind to whether the cell proliferates |
| cost tier | high |
| timeline | 12-24 months including fetal tissue access |
arms
[ "untreated", "etoposide", "non-TOP2 genotoxic control" ]
controls
[ "two or more bait positions, to separate partner spectrum from bait position", "adult CD34+ and a non-haematopoietic line, as lineage controls", "untreated arm, because untreated progenitors already generate KMT2A rearrangements" ]
design risks
[ "Cas9 bait chemistry differs from TOP2 break chemistry and the two are joined by pathways with opposite effects on translocation", "HTGTS convolves break rate with joining propensity", "power for the 63-single-occurrence tail of the clinical distribution is probably unachievable at fetal-tissue cell numbers" ]
- Engineering a double-strand break in KMT2A in primary human haematopoietic stem cells creates the founding lesion of an aggressive infant leukaemia in a dish, and the resulting cells would be capable of engrafting an immunodeficient mouse. The biosafety and ethical review is not a formality.
- The result is descriptive: a partner map changes no treatment and, if it matches the clinical distribution, simply confirms what patient data already show.
- If the experiment shows partner choice is set at the DNA level, it strengthens the case that nothing downstream can be intervened on to change which fusion forms - a negative for the whole prevention framing rather than a lead.
- Does the unbiased partner spectrum from a KMT2A bait resemble the clinical partner distribution? If it does not, the clinical distribution is mostly selection and the field's mechanistic focus on breakage geometry is misplaced.
- Is the partner spectrum different in a fetal progenitor than in an adult one, or than in a fibroblast? A cell-type-invariant spectrum would place all of the disease's lineage specificity downstream of the break.
- Can an endogenous-break HTGTS variant, with no engineered bait, be run at the KMT2A BCR after etoposide? That would remove the bait-chemistry objection entirely and is the version worth attempting first.
- Do the junctions recovered carry the 4 bp perfect-homology class that marks a TOP2/TDP2 origin, and does that fraction differ between the etoposide and untreated arms? This experiment would generate the first purpose-built dataset for `ana-junction-microhomology-signature-mining`.
- E1established confidence 5 of 5: Multiple independent lines of evidence; treated as settled in the field.paper
Linear amplification-mediated HTGTS robustly detects DNA double-strand breaks across the human genome based on their translocation to other endogenous or ectopic DSBs, extended the known off-target set for characterised nucleases more than tenfold, and detected translocations between bona fide nuclease targets on homologous chromosomes.
Genome-wide detection of DNA double-stranded breaks induced by engineered nucleases · Frock RL, Hu J, Meyers RM, Ho YJ, Kii E, Alt FW · Nature biotechnology 2015What this evidence cannot show
The method needs a bait break and reports partners only insofar as they join to it, so it measures break frequency times joining propensity rather than break frequency alone. Developed and validated for nuclease off-target detection, not for modelling endogenous oncogenic translocation.
Evidence that argues against this (1)
- The break chemistry the method relies on is the wrong one for this locus's dominant proposed mechanism. Translocations induced by recombinant nucleases in human cells implicate classical NHEJ in FORMING translocations, whereas at TOP2-induced breaks the TDP2-dependent end-joining route SUPPRESSES them - the authors of the TDP2 study state the contrast explicitly. A Cas9-baited partner spectrum is therefore a model of a different lesion. Gomez-Herreros F et al. 2017
- E2established confidence 5 of 5: Multiple independent lines of evidence; treated as settled in the field.paper
Across 3401 acute leukaemia patients, the seven most frequent KMT2A translocation partner genes plus partial tandem duplications account for more than 90% of all recombinations, with 37 partners recurring and 63 identified only once.
The KMT2A recombinome of acute leukemias in 2023. · Meyer C, Larghero P, Almeida Lopes B, et al. · Leukemia 2023What this evidence cannot show
A single referral-based diagnostic centre; partner frequencies reflect its referral pattern and its assay's partner coverage. The comparison this experiment makes is against an ascertained distribution, not a population one.
Evidence that argues against this (1)
- Comparing an unbiased HTGTS spectrum against an assay-limited clinical spectrum risks attributing to biology what is an artefact of detection: a targeted panel returns negative for every partner it was not designed around. Partners abundant in HTGTS and absent clinically are therefore ambiguous between 'makes a non-viable fusion' and 'was never looked for', and the analysis must state which partners the clinical assay could have detected. Meyer C et al. 2024
- E3strong confidence 4 of 5: Replicated in peer-reviewed work by more than one group.paper
Primary human CD34+ haematopoietic cells support recovery of stable rearrangements originating within the KMT2A translocation hotspot at readily detectable frequency after etoposide, with translocations, tandem duplications, deletions, inversions and insertions among the recovered events.
Therapy-related acute myeloid leukemia-like MLL rearrangements are induced by etoposide in primary human CD34+ cells and remain stable after clonal expansion · Libura J, Slater DJ, Felix CA, Richardson C · Blood 2005What this evidence cannot show
Inverse PCR rather than HTGTS, so the partner spectrum recovered is conditioned on primer design and is not genome-wide. Adult or cord CD34+, not fetal liver. Establishes feasibility of the cell system, not of the assay in that system.
Evidence that argues against this (1)
- The same cell system produces KMT2A rearrangement bands with no genotoxin at all under cytokine stimulation, more at day 7 than in etoposide-treated cells, and none of the recovered rearrangements encoded a leukaemogenic oncogene. An HTGTS library from such cells will therefore contain a substantial background of junctions that have nothing to do with disease, and the untreated arm is a necessary control rather than an optional one. Harada Y et al. 2020
- E4established confidence 5 of 5: Multiple independent lines of evidence; treated as settled in the field.paper
A donor-independent CRISPR-Cas9 protocol is published for inducing KMT2A rearrangements with different partner genes in K562 and in umbilical cord blood haematopoietic stem and progenitor cells, including patient-specific sgRNA design and HSPC purification.
Protocol for CRISPR-Cas9-mediated induction of KMT2A rearrangements in cell line and umbilical cord blood hematopoietic stem and progenitor cells · Benz T, Larghero P, Meyer C, et al. · STAR protocols 2025What this evidence cannot show
A protocol for making a SPECIFIED translocation with paired guides, not for placing a single bait break and reading out unbiased partners. The sgRNA design and HSPC handling transfer; the experimental logic does not.
Evidence that argues against this (1)
- A protocol demonstrating that a chosen KMT2A translocation can be engineered says nothing about whether an unbiased partner spectrum can be recovered at achievable cell numbers. Translocation is rare, and the clinical distribution's informative tail is 63 partners each seen once across 3401 patients, so library depth rather than protocol availability is the binding constraint. Meyer C et al. 2023
- E5strong confidence 4 of 5: Replicated in peer-reviewed work by more than one group.paper
TDP2 protects the KMT2A locus from TOP2-induced breakage and suppresses etoposide-induced translocations, and the authors contrast this explicitly with recombinant-nuclease-induced breaks, where end joining instead promotes translocation - a difference they attribute to TOP2 breaks having 4-nt cohesive overhangs that religate without further processing.
TDP2 suppresses chromosomal translocations induced by DNA topoisomerase II during gene transcription · Gomez-Herreros F, Zagnoli-Vieira G, Ntai I, et al. · Nature communications 2017What this evidence cannot show
Full text read via Europe PMC fullTextXML (PMC5550487). hTERT-RPE-1 and MCF7 cells; the KMT2A breakage readout is FISH probe separation rather than sequence. Cited here as a design constraint on bait chemistry.
Evidence that argues against this (1)
- The same paper reports that TDP2 status did NOT significantly change the distribution of KMT2A translocation partners by Circos analysis, which they read as TDP2 not influencing where TOP2 cuts. If partner distribution is robust to a large change in the joining pathway, then a bait-chemistry difference may perturb junction structure without perturbing the partner spectrum - which would make the Cas9-bait objection to this experiment weaker than it first appears. Gomez-Herreros F et al. 2017
names HTGTS as the only one of the three assays that returns the partner
the translocation-capture experiment that entry names in its own open questions
supplies an unbiased partner spectrum to place beside contact frequency and clinical rank
would generate the first purpose-built junction library for the signature classification
- tests →Each partner prefers a different KMT2A exon: partner-specific break position in 3,401 patients
an unbiased partner spectrum tests whether observed partner and exon preference is breakage or selection