microRNA genes silenced by CpG hypermethylation in t(4;11) infant ALL — the non-coding genome the fusion switches OFF
also called hypermethylated miRNAs MLL-rearranged infant ALL · miR-152 CpG island · zebularine-reactivated miRNAs
Most attention goes to the genes a fusion switches on, but in t(4;11) infant leukemia a large set of small regulatory RNAs is switched off instead, by chemical tags placed on their DNA. Drugs that strip those tags put the RNAs back in the dish, and how heavily one of them is tagged predicts how badly the child does - but when a trial actually stripped the tags in these infants alongside chemotherapy, it did not help them live longer.
The same thing, for a scientist
t(4;11) and t(11;19) infant ALL are **extensively CpG-hypermethylated**, while t(9;11) infant ALL and KMT2A-wild-type infant ALL epigenetically resemble normal bone marrow; within the hypermethylated group, the degree of promoter hypermethylation tracks relapse-free survival, and the demethylating agent zebularine reverses methylation and induces apoptosis [E1]. Extending this to the non-coding genome, 11 microRNA genes are downregulated in t(4;11) infant ALL as a consequence of CpG island hypermethylation, 7 of them re-activated by zebularine; five are associated with KMT2A or KMT2A fusions, and for miR-152 both KMT2A itself and DNMT1 were identified as candidate targets - creating a self-reinforcing methylation circuit. Heavy methylation of the miR-152 CpG island correlated strongly with poor clinical outcome [E2]. This is the mirror image of the HOXA/miR-196b story: the same fusion that drives a small set of non-coding RNAs up silences a larger set. **Two counters added cycle 35, and they bound the claim in different directions.** *Therapeutically*, the demethylation strategy this literature motivates has now been tested in the right patients and did not improve survival: COG AALL15P1 primed infants with newly diagnosed KMT2A-r ALL with azacitidine before chemotherapy, confirmed demethylation by WGBS in 87% of tested samples, and reported event-free survival similar to prior studies with no survival impact [E1 counter, PMID 38867582]. *Mechanistically*, CpG island hypermethylation in cancer preferentially strikes islands already carrying Polycomb H3K27me3 from the cell of origin, with the EZH2-containing complex recruiting the DNA methyltransferases [E2 counter, PMID 17200670] - so the methylation may be a consequence of the progenitor's pre-existing repressive state rather than something the fusion does, and the observation of 11 methylated miRNA promoters does not by itself implicate KMT2A::AFF1. What survives intact is the descriptive core: the partner-specific methylation phenotype, the reversibility of the silencing in vitro, and the single-cohort prognostic association.
The finding, and why it is underrated1 deeper
The KMT2A-r field is dominated by gain-of-function: the fusion recruits DOT1L and SEC and drives HOXA9/MEIS1. The methylation work says something orthogonal and, for therapy, possibly more useful: t(4;11) blasts also carry a large, reversible programme of non-coding silencing.
The numbers are specific. Eleven miRNAs downregulated through CpG island hypermethylation in t(4;11); seven reactivated by zebularine; five of the eleven linked to KMT2A or KMT2A fusions 2. And the methylation phenotype is partner-specific: t(4;11) and t(11;19) are extensively hypermethylated whereas t(9;11) looks like normal marrow 1. That is one of the cleanest examples in the whole graph of different KMT2A fusion partners producing genuinely different epigenomes, which matters directly for the assumption that a single menin inhibitor covers all KMT2A-r disease equally.
What cycle 35 removed from this. The therapeutic corollary - that an approved hypomethylating agent could cash this biology in - has been tested. COG AALL15P1 primed 31 evaluable infants with newly diagnosed KMT2A-r ALL with 5 days of azacitidine before each of four post-induction chemotherapy courses. It was safe, it demethylated (WGBS, 87% of tested samples), and event-free survival was no better than prior studies [E1 counter-evidence]. Nothing about that result is ambiguous, and it means this entry can no longer be read as pointing at an untried therapy. What it does not close: the trial did not select by methylation subgroup, did not measure miRNA re-expression in patients, and did not pair the drug with an immune effector. The reasonable position is that the unselected HMA-plus-chemotherapy strategy is refuted and a biomarker-selected or immunotherapy-paired version is untested.
And a mechanistic caution about causality. In cancers generally, the CpG islands that acquire de novo methylation are the ones already marked with Polycomb H3K27me3 in the precursor cell, with the EZH2 complex recruiting the DNA methyltransferases [E2 counter-evidence]. That is an alternative and sufficient explanation for a set of hypermethylated miRNA promoters in a fetal-origin leukaemia, and it does not require the fusion to have done anything. This entry's third open question - cell of origin versus fusion - is therefore not a housekeeping question but the one on which the whole causal story turns.
The miR-152 / DNMT1 / KMT2A circuit
miR-152 is reported to target both DNMT1 and KMT2A 2. If that holds, hypermethylating the miR-152 promoter removes a brake on DNMT1, which increases methylation, which keeps miR-152 off. A self-sustaining loop like this is attractive because it is exactly the kind of state a hypomethylating agent can break, and because it predicts that the response to a hypomethylating agent should be non-linear rather than dose-proportional.
The caution is that miR-152 target claims rest on reporter and prediction evidence in this paper, n is small, and the prognostic correlation is from a single cohort. Nobody has replicated the miR-152 outcome association in an independent Interfant cohort, which would be a straightforward reanalysis of existing data. Cycle 35 adds a further caution: if miR-152's promoter was Polycomb-marked in the fetal progenitor before any methylation arrived, the 'self-sustaining loop' may be the closing of a circuit that was already open rather than one the fusion created, and breaking it pharmacologically would not necessarily return the cell to a normal state.
Intervention assessment
Reactivating silenced miRNAs is a de-repression strategy, which is unusual in this disease and has an existing drug class: azacitidine and decitabine, both used in children and both already combined with chemotherapy in infant ALL trials. Zebularine itself was the tool compound and is not a clinical agent 12.
The attraction relative to the menin axis is that this is not a fusion-interface strategy, so it is not vulnerable to MEN1 interface mutations. The unattraction is that global demethylation is the least selective intervention imaginable: it reactivates whatever was silenced, including transposable elements (see hyp-transposable-element-reactivation-kmt2a-r), and in an infant the developmental consequences of genome-wide demethylation are the central unknown.
| class | microRNA (multiple loci) |
|---|---|
| direction in kmt2a r | down, via promoter CpG island hypermethylation |
| n mirnas downregulated by hypermethylation | 11 |
| n reactivated by zebularine | 7 |
| n linked to mll or mll fusions | 5 |
| intervention assessment | Tractable in principle with an approved drug class (azacitidine/decitabine); tool compound in the primary papers was zebularine. **Tested and negative as an unselected chemosensitiser**: COG AALL15P1 (NCT02828358) confirmed demethylation by WGBS in 87% of tested samples in infants with newly diagnosed KMT2A-r ALL and found no survival impact (PMID 38867582). Least selective option available: global demethylation reactivates everything silenced, including repeats. Not vulnerable to MEN1 interface resistance mutations, which was its main strategic attraction; that attraction is unchanged but the unselected version of the strategy has already failed. |
| genome build note | The source papers report array probe identities rather than coordinates; no hg38 intervals asserted. |
| causality caution | CpG islands that acquire de novo methylation in cancer are preferentially those already carrying Polycomb H3K27me3 in the cell of origin, with the EZH2 complex recruiting DNMTs (PMID 17200670). Hypermethylation of 11 miRNA promoters in t(4;11) is therefore compatible with a pre-existing fetal-progenitor repressive state and does not on its own implicate the fusion. |
exemplar
{
"mirna": "miR-152",
"candidate_targets": [
"MLL/KMT2A",
"DNMT1"
],
"clinical_correlate": "heavy CpG island methylation correlates with poor outcome"
}partner specificity
{
"hypermethylated": [
"t(4;11) KMT2A::AFF1",
"t(11;19) KMT2A::MLLT1"
],
"normal_marrow_like": [
"t(9;11) KMT2A::MLLT3",
"KMT2A wild-type infant ALL"
]
}clinical test
{
"trial": "COG AALL15P1",
"nct": "NCT02828358",
"agent": "azacitidine priming before chemotherapy",
"population": "infants with newly diagnosed KMT2A-rearranged ALL",
"n_evaluable": 31,
"dose_limiting_toxicity": "2/31 (6.5%)",
"pharmacodynamics": "WGBS showed decreased DNA methylation in 87% of samples tested",
"outcome": "event-free survival similar to prior studies; azacitidine did not impact survival",
"not_done": [
"methylation-subgroup patient selection",
"miRNA re-expression measurement in patients",
"pairing with an immune effector",
"TE/interferon read-out"
],
"pmid": "38867582",
"title": "Azacitidine as epigenetic priming for chemotherapy is safe and well-tolerated in infants with newly diagnosed KMT2A-rearranged acute lymphoblastic leukemia: Children's Oncology Group trial AALL15P1.",
"authors": "Guest EM, Kairalla JA, Devidas M, et al.",
"venue": "Haematologica",
"year": 2024,
"title_provenance": "Title/authors/venue backfilled from PubMed esummary on 2026-09-20 because this citation stored a bare PMID with no title, which made it unscoreable by verify-citations.py. BACKFILLING IS NOT VERIFICATION: it confirms the PMID resolves to a real paper, not that this is the right paper for the claim. That still requires reading."
}- Global demethylation in an infant reactivates far more than the intended miRNAs, including transposable elements and imprinted loci, with unknown developmental consequences during a period of active haematopoietic and somatic maturation.
- Reactivating a miRNA that targets KMT2A would also reduce wild-type KMT2A, which is required for normal haematopoiesis; the therapeutic window for this axis is undefined.
- If the methylation phenotype is a property of the cell of origin rather than the fusion, a hypomethylating strategy would be expected to fail in exactly the KMT2A-r subtypes that are not hypermethylated, and there is currently no assay in use to select patients.
- The prognostic association is single-cohort; building a risk-stratification claim on it before replication would be the kind of error this project exists to avoid.
- The unselected version of this strategy has already been taken into a COG trial and returned no survival benefit despite confirmed target engagement, so a further hypomethylating-agent proposal in infant KMT2A-r ALL inherits a failed precedent and needs a selection biomarker or a different combination partner to be worth running.
- Does the miR-152 methylation-outcome association replicate in an independent Interfant or COG infant ALL cohort? The samples and the methylation arrays largely exist already.
- Which of the eleven silenced miRNAs is functionally required, as opposed to merely silenced? A miRNA-mimic add-back screen in SEM/RS4;11 would separate driver from passenger.
- Why are t(4;11) and t(11;19) hypermethylated while t(9;11) is not? Cycle 35 sharpens this: if the hypermethylated islands are the ones already Polycomb-marked in the cell of origin (PMID 17200670), the answer is cell of origin, not partner, and the discriminating experiment is H3K27me3 ChIP in fetal-liver versus postnatal pro-B progenitors checked against the 11 miRNA promoters - existing data may already answer it.
- Does menin inhibition change DNA methylation, or only chromatin and transcription? If the hypermethylated state persists through a menin inhibitor, adding a hypomethylating agent is a rational and untested combination.
- AALL15P1 demethylated these patients and banked material. Did the 11 hypermethylated miRNAs actually re-express in vivo? If they did and outcome was unchanged, the miRNA silencing is a passenger; if they did not, the in-vitro zebularine result does not transfer to clinical azacitidine dosing. Either answer settles this entry.
- E1strong confidence 4 of 5: Replicated in peer-reviewed work by more than one group.paper
Distinct promoter CpG island methylation patterns separate genetic subtypes of MLL-rearranged infant ALL: t(4;11) and t(11;19) are extensively hypermethylated whereas t(9;11) and KMT2A-wild-type infant ALL resemble normal bone marrow; among t(4;11)/t(11;19) cases the degree of hypermethylation influences relapse-free survival, and zebularine reverses aberrant methylation and induces apoptosis in MLL-rearranged ALL cells.
Specific promoter methylation identifies different subgroups of MLL-rearranged infant acute lymphoblastic leukemia, influences clinical outcome, and provides therapeutic options · Stumpel DJPM, Schneider P, van Roon EHJ, Boer JM, de Lorenzo P, Valsecchi MG, de Menezes RX, Pieters R, Stam RW · Blood 2009What this evidence cannot show
Differential methylation hybridisation, a pre-bisulfite-sequencing platform with limited CpG coverage; subtype group sizes in infant ALL are inherently small, so the survival association needs independent replication. Zebularine is a tool compound, not a clinical agent.
Evidence that argues against this (1)
- The clinical test of the 'provides therapeutic options' half of E1's title has been run in this exact population and was negative on survival. Children's Oncology Group AALL15P1 gave 5 days of azacitidine priming before chemotherapy in four post-induction courses to infants with newly diagnosed KMT2A-r ALL: 2 of 31 evaluable patients had dose-limiting toxicity, whole-genome bisulfite sequencing showed decreased DNA methylation in 87% of samples tested, event-free survival was similar to prior studies of newly diagnosed infant ALL, and the authors state that incorporating azacitidine to enhance cytotoxicity did not impact survival. The pharmacodynamic premise of this entry held; the therapeutic inference from it did not. Note what the trial did not do: it did not select patients by methylation subgroup, did not measure miRNA re-expression, and did not pair the HMA with an immune effector - so it refutes the unselected chemosensitisation strategy rather than the underlying biology. Guest EM et al. 2024
- E2moderate confidence 3 of 5: One strong study, or several smaller consistent ones.paper
Eleven microRNA genes are downregulated in t(4;11) infant ALL as a consequence of CpG island hypermethylation, seven of which are re-activated by zebularine; five are associated with MLL or MLL fusions, miR-152 has both MLL and DNMT1 as candidate targets, and heavy methylation of the miR-152 CpG island correlates strongly with poor clinical outcome.
Hypermethylation of specific microRNA genes in MLL-rearranged infant acute lymphoblastic leukemia: major matters at a micro scale · Stumpel DJPM, Schotte D, Lange-Turenhout EAM, Schneider P, Seslija L, de Menezes RX, Marquez VE, Pieters R, den Boer ML, Stam RW · Leukemia 2011What this evidence cannot show
Small single-centre cohort from the Erasmus MC/Interfant material, same group as E1 so not independent. miR-152 target assignment is candidate-level rather than fully validated in this disease. The paper itself calls for in vivo testing, which as far as we can find was never reported for the miRNA arm.
Evidence that argues against this (1)
- CpG island hypermethylation in cancer is often a passenger rather than a cause of silencing. Schlesinger et al. showed that genes methylated in cancer cells are specifically packaged with H3K27me3-containing nucleosomes, that this Polycomb mark is established on the unmethylated CpG islands early in development and maintained by an EZH2-containing complex in differentiated cells, and that in cancer cells the presence of that complex recruits DNA methyltransferases and drives de novo methylation. On this model, a CpG island is methylated in a tumour *because it was already Polycomb-silenced in the cell of origin*. Two consequences for this entry. (i) Finding 11 hypermethylated miRNA promoters in t(4;11) does not establish that methylation caused their downregulation, nor that the fusion caused the methylation; a pre-existing fetal-progenitor Polycomb state predicts the same observation. (ii) Re-expression of 7 of them by zebularine shows the methylation is now load-bearing for expression in the leukaemic cell, which is a real result, but says nothing about how it got there. The entry's own open question about cell of origin versus fusion is the right question, and this is the mechanism that makes it pressing. Schlesinger Y et al. 2007
Counter-evidence (Guest 2024, COG AALL15P1, PMID 38867582): azacitidine priming in infants with newly diagnosed KMT2A-r ALL achieved demethylation (WGBS, 87% of tested samples) with 2/31 dose-limiting toxicities but produced event-free survival similar to prior studies and no survival impact - the demethylation therapy this methylation literature motivates has been tested unselected and failed
Causality counter (Schlesinger 2007, PMID 17200670): CpG islands that acquire de novo methylation in cancer are preferentially those already carrying Polycomb H3K27me3 from the cell of origin, with the EZH2 complex recruiting DNMTs - so 11 hypermethylated miRNA promoters in t(4;11) are compatible with a pre-existing fetal-progenitor repressive state and do not on their own implicate the fusion
Shared clinical counter (PMID 38867582): both entries rest on the idea that reversing hypermethylation in t(4;11) infant ALL is therapeutically useful; COG AALL15P1 demethylated these patients successfully and changed nothing about survival, so both must now argue from a biomarker-selected or immunotherapy-paired design rather than from demethylation alone
Counter-evidence: the demethylation therapy this miRNA-silencing literature motivates was tested unselected in infant KMT2A-r ALL, reached its pharmacodynamic target, and did not change event-free survival - so a further hypomethylating-agent proposal needs a selection biomarker or a different combination partner
miR-9 is among the microRNAs reported silenced by CpG hypermethylation in t(4;11)
a silencing class described specifically in t(4;11) infant disease
both rest on the same premise, that t(4;11) is a globally hypermethylated genome