The test bed nobody names: KMT2A rearrangement can be induced and counted in primary human CD34+ and fetal liver stem cells, and it happens without any drug at all
also called primary human CD34+ MLL rearrangement assay · inverse PCR MLL breakage readout · fetal liver HSC breakage model
You can take real human blood stem cells, treat them, and count how often the KMT2A gene breaks and rejoins wrongly. The most uncomfortable result from that assay is that the breaks also appear in untreated cells that are simply being told to divide.
The same thing, for a scientist
Every preventive hypothesis in this beat needs a system in which KMT2A rearrangement frequency can be measured in the right human cell. That system exists and has since 2005, and this graph had no entry naming it. Libura et al. established inverse PCR on primary human CD34+ cells and found etoposide-promoted rearrangements originating in the 8.3 kb hotspot at readily detectable frequency, with the stable post-expansion spectrum dominated by KMT2A tandem duplications plus translocations, deletions, inversions and insertions [E1]; they extended it to long-term repopulating HSC assayed by NOD/SCID engraftment [E2]. Moneypenny et al. induced KMT2A rearrangements in cultured HUMAN FETAL liver HSC at low nanomolar etoposide, with strand breaks at 140 nM and, importantly, no detectable in-vitro TOP2 inhibition below 25 microM - the damage appeared at doses where the enzyme assay was silent [E3]. Barjesteh van Waalwijk van Doorn-Khosrovani et al. showed quercetin, genistein and kaempferol induce dose-dependent DSBs and KMT2A translocations in primary human CD34+ cells, with junctions formed mostly by microhomology-mediated end joining and SINE/Alu or LINE/L1 elements at one side of all but one junction [E4]. Lu et al. did the same for chlorpyrifos in human fetal liver CD34+ HSC, implicating both TOP2 poisoning and caspase-3-dependent apoptotic signalling [E5]. And Harada et al. supply the control that reframes the whole field: KMT2A and RUNX1 rearrangement bands appeared in NON-etoposide-treated CD34+ cells under cytokine stimulation - MORE of them at day 7 than in etoposide-treated cells - and in cytokine-mobilised peripheral blood stem cells from donors [E6].
What the assay is1 deeper
Primary human CD34+ cells (cord blood, mobilised peripheral blood, or fetal liver), cultured with cytokines, exposed to a candidate agent, then interrogated for rearrangement originating in the KMT2A breakpoint cluster region. Three readouts have been used and they are not equivalent:
- Inverse PCR from a fixed primer in the BCR, which recovers the sequence of whatever the region joined to. This is the only readout that gives a junction and therefore the only one that can feed
ana-junction-microhomology-signature-mining1. - Break-apart FISH, which scores separation of probes flanking KMT2A. Cheap, blind to junction structure, and cannot distinguish a translocation from a large local rearrangement 3.
- Flow immunophenotyping of translocation-positive events, used as a screen and confirmed by FISH 3.
The engineered counterpart is now also published: a donor-independent CRISPR-Cas9 protocol reconstructs specific KMT2A translocations with different partner genes in K562 and in cord-blood HSPC 7. That is the positive control this assay never had - a way to make the exact rearrangement on purpose and ask what it takes to stop it.
Fetal liver, not cord blood, when the question is infant disease
Two of these studies used human fetal liver HSC rather than cord blood 35. That matters because this graph's cell-of-origin position places KMT2A::AFF1 initiation in a fetal liver HSC/LMPP compartment, and cord blood is a later, different cell. A protective effect demonstrated in cord blood - as the APE1 result in the replication-stress entry was - is a result in the wrong cell by one developmental step. Fetal liver tissue is harder to obtain and the studies using it are correspondingly older and smaller.
The result that should change how this beat is framed
Harada et al. set out to test whether a proliferation signal alone could do what a topoisomerase poison does, on the explicit reasoning that fetal exposure to topoisomerase inhibitors is rare. Using inverse PCR on cytokine-cultured CD34+ cells they found variable-sized rearrangement bands in RUNX1 and KMT2A at 3 hours in etoposide-treated cells which then decreased by day 7, while more rearrangement bands were present in the NON-treated cells at day 7; the same bands appeared in cytokine-mobilised peripheral blood stem cells collected for transplantation 6.
Two things follow. First, the background rate of KMT2A rearrangement in proliferating normal human progenitors is not zero, so any prevention strategy is trying to reduce a rate that already exists without an exogenous cause. Second, and this is the authors' own conclusion, none of the rearranged genes they recovered encoded a leukemogenic oncogene and the cells carrying them did not expand - so the limiting step is not the break. It is which break, in which cell, and whether that cell survives and proliferates. A prevention strategy aimed at the break is aimed at the abundant step, not the rare one.
The dose problem in the flavonoid and etoposide work
Moneypenny et al. report the detail most often dropped from citations of this literature: DNA strand breaks and KMT2A rearrangement appeared at low nanomolar etoposide, while measurable in-vitro TOP2 inhibition required 25 microM or more 3. A four-order-of-magnitude gap between the concentration that damages the genome and the concentration at which the proposed mechanism is detectable is not a small caveat; either the enzyme assay is far less sensitive than the cellular readout, or something other than bulk TOP2 inhibition is producing the breaks at low dose.
The flavonoid work has the mirror-image problem in the other direction: the authors argue their concentrations are biologically relevant and that fetal levels may exceed maternal ones, and they close by calling for public guidelines on flavonoid supplements 4. This graph's target-prevention-exposure-avoidance-infant-leukemia records that the epidemiology does not support that call - the largest case-control study found little evidence of association and the most consistent finding was a protective association with maternal fruit and vegetable intake. A cell-culture demonstration of a mechanism is not an exposure risk estimate, and here the two point in opposite directions.
| design | primary human CD34+ or fetal liver HSC, cytokine culture, agent exposure, rearrangement recovery from the KMT2A breakpoint cluster region |
|---|---|
| controls | untreated cytokine-only arm at MATCHED timepoints is mandatory, because the untreated day-7 background exceeds the etoposide-treated day-7 signal in at least one study (E6) |
| dose discordance note | DNA damage and KMT2A rearrangement at low nanomolar etoposide versus measurable in-vitro TOP2 inhibition only at >=25 microM (E3) |
| cost tier | med |
| timeline | weeks per arm; fetal liver tissue access is the rate-limiting step |
readouts
[ "inverse PCR from a fixed BCR primer - the only readout that returns a junction sequence", "break-apart FISH across KMT2A", "flow immunophenotyping of translocation-positive events", "neutral comet assay for DSB induction", "NOD/SCID engraftment and long-term culture-initiating assay for stemness of the rearranged clone" ]
cell sources used
[
{
"source": "human fetal liver CD34+ HSC",
"refs": [
"E3",
"E5"
]
},
{
"source": "cord blood / adult mobilised CD34+",
"refs": [
"E1",
"E4",
"E6"
]
},
{
"source": "long-term repopulating HSC via xenograft",
"refs": [
"E2"
]
},
{
"source": "cord blood HSPC, CRISPR-engineered",
"refs": [
"E7"
]
}
]agents shown to induce kmt2a rearrangement in this system
[ "etoposide (low nanomolar in fetal liver HSC)", "quercetin, genistein, kaempferol", "chlorpyrifos", "cytokine stimulation alone, with no genotoxin" ]
- The assay counts breaks, and the Harada result says breaks are the abundant step while leukaemogenic breaks in surviving cells are the rare one. An intervention that halves the break rate may not halve anything that matters.
- Fetal liver tissue is the correct substrate and is ethically and practically constrained, so the cheapest version of every experiment here is run in the wrong cell and will be cited as if it were not.
- Demonstrating that a common dietary compound induces KMT2A translocations in culture creates public-health pressure that the epidemiology does not support, and the 2007 flavonoid paper closes by calling for exactly that. A graph entry that repeats the mechanism without the null epidemiology does harm.
- What is the spontaneous KMT2A rearrangement rate per cell division in cytokine-stimulated human FETAL LIVER HSC? Harada measured a non-zero background in adult CD34+ cells; nobody has measured it in the compartment that matters, and it is the denominator every prevention claim needs.
- Does APE1 inhibition, the one intervention shown to protect cord-blood HSPC from MLLbcr cleavage, reduce actual rearrangement frequency by inverse PCR in this assay rather than cleavage signal? That is the experiment the replication-stress entry asks for and this is the system to run it in.
- Do the junctions recovered by inverse PCR from etoposide-treated CD34+ cells contain the 4 bp perfect-homology class that marks a TOP2/TDP2 origin? The sequences exist in published supplementary material and have never been classified that way.
- Why do DNA damage and KMT2A rearrangement appear four orders of magnitude below the etoposide concentration at which TOP2 inhibition is measurable? Either the enzyme assay is the wrong instrument or the low-dose mechanism is not bulk TOP2 poisoning.
- E1strong confidence 4 of 5: Replicated in peer-reviewed work by more than one group.paper
In primary human CD34+ cells, etoposide promotes stable rearrangements originating within the KMT2A translocation hotspot at readily detectable frequency; after 10-14 days of proliferation the majority of illegitimate events are KMT2A tandem duplications, with translocations, deletions, inversions and insertions also recovered, and only a small subset of translocations occurs in cells that proliferate significantly.
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 is sensitive to template abundance and to primer placement, so the recovered spectrum is conditioned on the assay design. Adult mobilised or cord CD34+ cells, not fetal liver. Etoposide doses are pharmacological. The dominance of tandem duplications in the surviving population is a survival-filtered spectrum, not the spectrum of initiating events.
Evidence that argues against this (1)
- The same assay run without etoposide produces MORE rearrangement bands at day 7 than with it, under cytokine stimulation alone. An etoposide-versus-untreated comparison at a single early timepoint therefore measures the drug's effect on a background that is itself substantial and time-dependent, and a study that did not include a day-7 untreated arm would attribute that background to the drug. Harada Y et al. 2020
- E2moderate confidence 3 of 5: One strong study, or several smaller consistent ones.paper
Etoposide-initiated KMT2A rearrangements including duplications, inversions and translocations are recovered at significant frequency from clones derived from primary human long-term repopulating haematopoietic stem cells, whose proliferative capacity is increased rather than decreased by etoposide exposure as judged by NOD/SCID engraftment and long-term culture-initiating assays.
Etoposide-initiated MLL rearrangements detected at high frequency in human primitive hematopoietic stem cells with in vitro and in vivo long-term repopulating potential · Libura J, Ward M, Solecka J, Richardson C · European journal of haematology 2008What this evidence cannot show
Same laboratory as E1. 'Proliferative capacity dramatically increased following etoposide' is a striking claim that would benefit from independent replication; a selection effect in which only resistant clones engraft is an alternative reading the paper does not exclude. Xenograft engraftment is a surrogate for stemness, not a measure of leukaemogenesis.
Evidence that argues against this (1)
- Carrying a KMT2A rearrangement and being able to repopulate are not sufficient for leukaemia: in normal CD34+ cells the authors of the cytokine-stimulation study recovered rearranged KMT2A and RUNX1 alleles but none encoded a leukaemogenic oncogene and the cells carrying them did not expand. Recovering a rearrangement from a repopulating clone therefore does not establish that a pre-leukaemic clone has been made. Harada Y et al. 2020
- E3moderate confidence 3 of 5: One strong study, or several smaller consistent ones.paper
In cultured human FETAL LIVER haematopoietic stem cells, etoposide caused dose-dependent viability loss at low nanomolar concentrations, DNA strand breaks at 140 nM, and a significant increase in KMT2A translocation-positive events confirmed by FISH, while in-vitro topoisomerase II inhibition was only observed at 25 microM or above.
MLL rearrangements are induced by low doses of etoposide in human fetal hematopoietic stem cells · Moneypenny CG, Shao J, Song Y, Gallagher EP · Carcinogenesis 2006What this evidence cannot show
Immunophenotyping of translocations by flow is an unusual readout and is confirmed here only by FISH, not by sequence. No junctions were recovered, so nothing from this study can enter a junction-signature analysis. The dose discordance between DNA damage and measurable TOP2 inhibition is stated by the authors and is not explained.
Evidence that argues against this (1)
- The mechanism attributed to low-dose etoposide here is not supported by the paper's own enzyme data, and an alternative account exists in which the KMT2A break is TOP2-independent: apoptosis-induced cleavage of the KMT2A BCR is unaffected by the catalytic TOP2 inhibitor merbarone and a 367 bp episomal fragment of the BCR is sufficient for site-specific cleavage. At nanomolar etoposide an apoptotic-nuclease route is at least as plausible as a topoisomerase route. Betti CJ et al. 2005
- E4moderate confidence 3 of 5: One strong study, or several smaller consistent ones.paper
Quercetin, genistein and kaempferol induce dose-dependent double-strand breaks and KMT2A translocations in primary human CD34+ cells; most of the resulting translocations were formed by microhomology-mediated end joining, and in all but one translocation a SINE/Alu or LINE/L1 repetitive element was present on at least one side of the breakpoint junction.
Dietary flavonoids induce MLL translocations in primary human CD34+ cells · Barjesteh van Waalwijk van Doorn-Khosrovani S, Janssen J, Maas LM, et al. · Carcinogenesis 2007What this evidence cannot show
A small junction set from one exposure class in one cell type; the repeat-element observation is an annotation of recovered junctions and not an enrichment test against a matched background. 'Biologically relevant concentrations' is the authors' judgement. The paper's closing call for public guidelines on flavonoid supplements goes well beyond what a CD34+ culture can support.
Evidence that argues against this (1)
- The epidemiology points the other way. The largest and best-designed case-control study of maternal diet and infant leukaemia, 240 incident cases and 255 controls, reported little evidence of association between the dietary topoisomerase II inhibitor index and leukaemia overall or by subtype, with the most consistent finding a PROTECTIVE association with maternal fruit and vegetable intake. A mechanism demonstrated in culture is not an exposure risk, and this graph's prevention entry records the verdict as not feasible. Spector LG et al. 2005
- E5weak confidence 2 of 5: Preliminary, small, indirect, or not yet peer reviewed.paper
In human fetal liver-derived CD34+ haematopoietic stem cells, 10 microM chlorpyrifos reduced viability, inhibited proliferation, induced DNA double-strand breaks and KMT2A rearrangements, caused G0/G1 arrest in contrast to etoposide's G2/M arrest, acted as a topoisomerase II poison, and induced early apoptosis with active caspase-3.
Chlorpyrifos Induces MLL Translocations Through Caspase 3-Dependent Genomic Instability and Topoisomerase II Inhibition in Human Fetal Liver Hematopoietic Stem Cells · Lu C, Liu X, Liu C, et al. · Toxicological sciences : an official journal of the Society of Toxicology 2015What this evidence cannot show
A single 10 microM concentration in the headline result, with no junction sequencing - the authors state explicitly that future work will elucidate the type of CPF-induced rearrangement. Two mechanisms (TOP2 poisoning and caspase-3 apoptosis) are asserted in the same system without a dissociation experiment separating them. Confidence is set to weak accordingly.
Evidence that argues against this (1)
- The same field's exposure epidemiology does not support acting on this: the dietary and environmental topoisomerase-inhibitor hypothesis has not been confirmed in the largest case-control test, and this graph's prevention entry records the verdict as not feasible with the original signal resting on 84 case-control pairs and a 95% confidence interval spanning nearly two orders of magnitude. Ross JA et al. 1996
- E6moderate confidence 3 of 5: One strong study, or several smaller consistent ones.paper
Inverse PCR detected KMT2A and RUNX1 rearrangement bands in cytokine-cultured CD34+ cells without any etoposide, with MORE bands in non-treated cells at day 7 than in etoposide-treated cells, and the same rearrangements in cytokine-mobilised peripheral blood stem cells; none of the rearranged genes encoded a leukaemogenic oncogene and the cells carrying them did not expand.
Gene rearrangements of MLL and RUNX1 sporadically occur in normal CD34+ cells under cytokine stimulation · Harada Y, Shingai N, Ding Y, et al. · Cancer science 2020What this evidence cannot show
Inverse PCR bands are not, on their own, verified rearrangements at single-molecule resolution, and the comparison between treated and untreated arms is across timepoints. Adult CD34+ cells and mobilised donor products, not fetal liver. 'None encoded a leukaemogenic oncogene' is a statement about what was recovered, bounded by the sensitivity of the assay.
Evidence that argues against this (1)
- A background of non-oncogenic rearrangement does not make exogenous exposure irrelevant. In the same cell type, etoposide promotes rearrangement of the KMT2A hotspot at readily detectable frequency with a spectrum matching the full range of oncogenic events seen in patients, and those rearrangements persist through 10-14 days of clonal expansion. The two results constrain each other rather than one refuting the other. Libura J et al. 2005
- E7established confidence 5 of 5: Multiple independent lines of evidence; treated as settled in the field.paper
A donor-independent CRISPR-Cas9 protocol exists for reconstructing KMT2A chromosomal translocations with different partner genes in vitro, covering patient-specific sgRNA design, sgRNA in vitro transcription, purification of haematopoietic stem and progenitor cells from umbilical cord blood, and editing of both K562 and cord-blood HSPC.
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 paper, so it documents feasibility rather than a biological result. A nuclease-induced break is chemically unlike a TOP2-induced one - blunt or small-overhang versus 4-nt 5'-cohesive with a protein adduct - so an engineered translocation models the product, not the process, and its junctions should not be used to infer how patient junctions were made.
Evidence that argues against this (1)
- The chemistry difference is load-bearing and documented: designer-nuclease breaks in human cells implicate classical NHEJ in MAKING translocations, whereas for TOP2-induced breaks the equivalent end-joining route is protective and its loss increases translocation. A CRISPR-reconstructed KMT2A translocation therefore cannot serve as a positive control for a mechanism question about TOP2-made fusions, only for a downstream question about the fusion product. Gomez-Herreros F et al. 2017
KMT2A rearrangement induced and counted in cultured human fetal liver HSC
the human primary-cell system in which TOP2-poison-induced KMT2A rearrangement is actually measured
supplies the mechanism half whose epidemiology that entry judges not feasible; the two point opposite ways
- supports →Apoptotic nucleases cut KMT2A, and cells that survive the attempt can carry a translocation
chlorpyrifos-induced KMT2A rearrangement in fetal liver HSC runs partly through caspase-3 activation
SINE/Alu or LINE/L1 present on at least one side of all but one flavonoid-induced KMT2A junction
inverse PCR is the only readout in this system that returns junction sequence, and flavonoid junctions were mostly MMEJ
the system in which the rearrangement-frequency version of the APE1 result could be run
the existing assay measures rearrangement; these three measure the break itself and have never been applied here
the same cell system, with the enzyme identified rather than assumed
the marker that defines that experiment's starting material, and the caveat that it is a mixture in this disease