Oncofusion

start / graph / risk-q187-yeats-inhibition-touches-an-imprinting-complex-in-an-infant

Riskspeculative confidence 1 of 5: Our own inference or analogy. Not shown to be true.Q187-the-sec-family-and-yeats-selectivity

The shared reader sits in a complex whose only well-defined normal job is parent-of-origin silencing — and the published safety package for YEATS inhibition is entirely haematopoietic

also called YEATS inhibition and imprinting · SEC-L3 imprinting liability

The protein a YEATS drug grabs also sits inside a machine whose normal job is keeping parent-of-origin gene silencing in place, and another one built on the gene that causes an inherited learning disability. The safety testing done so far only looked at blood stem cells, which cannot see either of those things, and the patients are babies.

The same thing, for a scientist

**Class:** essential-function toxicity in a non-haematopoietic tissue, currently unmeasured rather than observed. ENL/MLLT1 and AF9/MLLT3 are subunits of SEC-L2 and SEC-L3 as well as SEC [E1]. Of the three AFF-family scaffolds outside the fusion, **AFF3/SEC-L3 has the best-defined physiological function in the literature and it is allele-specific epigenetics**: AFF3 binds gametic differentially methylated regions and the Meg3 enhancer in the imprinted *Dlk1-Dio3* cluster allele-specifically and is required for the ~200 kb Meg3 polycistron [E2]; it binds the DMR downstream of the *XIST* promoter in a DNA-methylation-dependent manner, and AFF3 knockdown de-represses the **inactive** XIST allele in terminally differentiated cells [E3]. **AFF2 is FMR2**, silenced by CCG expansion in FRAXE intellectual disability [E4]. The published therapeutic-window measurement for a YEATS probe in this disease is an in vitro human HSPC colony assay plus a long-term mouse HSPC transplant arm [E5] — **neither of which can detect an allele-specific-expression change, an imprinting defect or a neurodevelopmental effect, by construction.** The target population is infants. This is a **liability with no measurement in either direction**, and its cheapest first test is a few hundred dollars of allele-specific RT-PCR on cells already treated in published experiments.

Go deeper
Why a clean haematopoietic safety result does not address this
endpoint reportedwhat it can detectwhat it cannot
human HSPC colony-forming potential in vitro 5acute progenitor cytotoxicity, differentiation blockanything allele-specific; anything outside blood; anything slower than a colony assay
long-term in vivo mouse HSPC function 5self-renewal loss, clonal exhaustionthe same three

The objection is not that the window measurement is bad. It is a good measurement of the liability the field expected — ENL is expressed in whole blood at about 40 TPM and MLLT3/AF9 governs human HSC self-renewal (target-phase2-enl-yeats-dossier), so haematopoietic toxicity was the right first worry and it has been addressed.

The composition result relocates the worry. A drug can spare every blood cell and still perturb a monoallelic expression state, and nothing in a colony count or a transplant readout is sensitive to that. ana-q187-the-scaffold-blindness-argument-does-not-survive-the-published-therapeutic-window argues that the fusion-selectivity version of this objection is refuted; this entry is what survives of it.

What is actually known, and the large gap in the middle

Known:

  • ENL and AF9 are in SEC-L3 1.
  • AFF3 is required for allele-specific expression at Dlk1-Dio3 and at XIST, and the authors of those papers describe AFF3 as the central factor of SEC-L3 23.
  • AFF2 is the FRAXE gene 4.

Not known, and this is the whole gap:

  • Nobody has ChIPped ENL or AF9 at a single imprinted control region, in any cell, ever. The imprinting papers blot AFF3 and its recruiters ZFP57 and ZFP281; they do not blot a single other SEC-L3 subunit.
  • Whether AFF3 does its imprinting job inside SEC-L3 or outside it. ENL and AF9 exist outside SEC when bound to DOT1L, so scaffold membership is not automatic at every locus.
  • Whether the YEATS contact is load-bearing at those loci even if the readers are present.

So the risk is conditional on a chain of three unmeasured links. Stating it as though it were established would be wrong; stating it as "not a concern because the HSPC data are clean" would be wrong in the other direction and is the failure mode this entry exists to prevent.

Mitigation, cheapest first
  1. Allele-specific expression readout in an already-treated cell line. Any female human line with a known XIST-inactive allele, treated with SGC-iMLLT, SR-0813 or TDI-11055, assayed by allele-specific RT-PCR at XIST and at MEG3. This is the experiment AFF3 knockdown was scored by 3 and it costs days. exp-q187-allele-specific-expression-under-a-yeats-antagonist states it.
  2. ENL/AF9 ChIP at the AFF3-bound control regions. The regions are named in 2 and 3; the antibodies exist and are in routine use. This decides whether the readers are there at all, and a negative closes the whole liability.
  3. Intersect the public YEATS-inhibitor RNA-seq with an imprinted-gene list. GSE185091 and GSE162592 are public (absence-q187-no-yeats-inhibitor-has-been-tested-against-sec-l2-or-sec-l3). This is free and can be done this cycle; it cannot give a clean negative, because the cells are AML lines, but a positive would be decisive.
  4. Only if 1-3 point somewhere: a developmental readout. Nothing short of that is warranted by the current evidence.
The counter-case, stated at full strength

Infant KMT2A-rearranged ALL has among the worst outcomes in paediatric oncology, which is the reason this project exists. A liability that rests on three unmeasured links should not be allowed to slow a class with a published in vivo window, and this graph has a documented tendency to let a well-written caveat do the work of a finding.

Three specific deflations:

  • The pool argument. AFF2 and AFF3 are about 10-fold lower expressed than AFF4 in the one background where it was measured 1. If most ENL and AF9 is in SEC, the sibling complexes absorb little occupancy.
  • Imprinting is maintained by DNA methylation, not by an elongation complex. AFF3 reads a methylated DMR; it does not write one. Transient pharmacological perturbation of a reader that responds to a methylation mark is not obviously the same as erasing the mark, and no evidence here says the effect would persist after washout.
  • FRAXE is a loss-of-expression syndrome from a CCG expansion silencing the whole gene, not from occupancy of a partner's reader pocket. Invoking it as a toxicity model for a reversible small molecule is an analogy several steps removed, and this entry does not claim otherwise.
Facts and figures
kindessential-function
severityunknown — not observed, not excluded. The mechanism requires three unmeasured links (readers present at imprinted control regions; YEATS contact load-bearing there; occupancy sufficient at achievable exposure) and none has been tested in any system.
mechanismENL/AF9 are shared subunits of SEC-L3, whose scaffold AFF3 is required for allele-specific expression at the Dlk1-Dio3 gametic DMR, the Meg3 enhancer and the XIST DMR; a YEATS-pocket ligand disables the reader's histone contact without regard to which scaffold it is attached to
population at riskinfants and young children treated for KMT2A-rearranged leukaemia — a developing population, which is the group in which a perturbation of imprinting or of an AFF2-dependent neurodevelopmental programme would matter most and be detected latest
detection trapEvery published safety endpoint for this class is haematopoietic (HSPC colony formation, long-term mouse HSPC transplant). Neither assay can report allele-specific expression, imprinting status or neurodevelopment, so a clean safety package is fully compatible with the liability being real. A negative result from an instrument that cannot see the effect is not evidence of absence.
mitigationAllele-specific RT-PCR at XIST and MEG3 in a YEATS-antagonist-treated human line (days); ENL/AF9 ChIP at the AFF3-bound gametic DMRs and the XIST DMR (the decisive test, and a negative closes the liability outright); intersection of the public GSE185091 / GSE162592 YEATS-inhibitor RNA-seq with an imprinted-gene list (free, positive-only)
escalation thresholdNothing beyond the three cheap tests is warranted on current evidence. Escalate to a developmental readout only if the ChIP places a YEATS reader at an imprinted control region OR the allele-specific assay moves.
epigenetic readoutthe hazard is itself epigenetic — DNA-methylation-dependent allele-specific binding at gametic DMRs and the XIST DMR, plus the YEATS acyl-lysine readout that would be pharmacologically blocked
shown for fusionNO — nobody has looked in any fusion, and nobody has looked in any system. The efficacy arm has been run in KMT2A::AFF1 (PMID 42242459); the imprinting arm has never been run anywhere.
absence classnever-measured
absence class basisNo deposit anywhere contains ENL or AF9 occupancy at an imprinted control region, and no YEATS-antagonist experiment has an allele-specific readout. Unlike the sibling absence entry, this one cannot be closed by reanalysis of public data — it needs a ChIP. The partial exception is item 3 of the mitigation list, which is free and can produce a positive but not a clean negative.
affects
[
  "drug-sgc-imllt",
  "target-phase2-enl-yeats-dossier"
]
marks named
[
  "DNA methylation (IG-DMR, Meg3 enhancer, XIST DMR)",
  "H3K27ac",
  "H3K9ac"
]
measurements
[
  {
    "quantity": "published_safety_endpoints_for_a_yeats_antagonist_that_could_detect_an_imprinting_effect",
    "subject": "PMID 42242459 abstract | human HSPC colony-forming assay and long-term mouse HSPC in vivo function",
    "value": 0,
    "units": "endpoints of 2",
    "n": 2,
    "universe": "the only published therapeutic-window measurement for a YEATS antagonist in KMT2A::AFF1; abstract read 2026-09-24, full text not in PMC and not read",
    "null": null,
    "note": "0 of 2. This is a statement about assay sensitivity, not about the compound. An assay that cannot report allele-specific expression returns the same clean result whether the effect exists or not."
  },
  {
    "quantity": "experiments_placing_a_yeats_reader_at_an_imprinted_control_region",
    "subject": "ENL or AF9 ChIP, CUT&RUN or equivalent at a gametic DMR, the Meg3 enhancer or the XIST DMR, any organism",
    "value": 0,
    "units": "experiments",
    "n": 3,
    "universe": "the three AFF3 imprinting papers (PMIDs 26728555, 28180295, 30535390), each read at abstract level; none blots or ChIPs any SEC-L3 subunit other than AFF3",
    "null": null,
    "note": "The single measurement that would confirm or close this entire risk, and it has never been made."
  }
]
How this could bite us back
  • This entry is itself a bite-back and therefore has to be held to the standard it is invoking. It rests on three unmeasured links, an analogy from a knockdown to a pocket ligand, and a disease (FRAXE) caused by a completely different lesion class. Quoting it as 'YEATS inhibitors risk causing an imprinting disorder in infants' would be a serious overstatement and the entry says so in its own severity field.
  • The cost of over-caution is real and asymmetric here. Infant KMT2A-rearranged leukaemia is among the worst-outcome paediatric cancers, the class has an in vivo window and an on-target resistance mutation confirming mechanism, and a well-written speculative liability can stall a programme more effectively than a measured one. The right output of this entry is three cheap experiments, not a recommendation.
  • If the three tests come back negative, this entry should be amended in place to say so rather than left standing as an open worry — a caveat that names a gap is a task, and an unresolved liability outlives the evidence that raised it.
Still open
  1. Are ENL and AF9 present at the AFF3-bound gametic DMRs, the Meg3 enhancer and the XIST DMR? One ChIP experiment with commercially available antibodies decides whether this risk exists at all, and it has never been done.
  2. Does AFF3 perform its imprinting function inside SEC-L3 or outside it? No imprinted-locus experiment has blotted any SEC-L3 subunit other than AFF3, so the complex's involvement is an inference from the scaffold's name.
  3. Does a YEATS antagonist change allele-specific expression at XIST or MEG3 in any treated human cell? This is a days-long assay that nobody has run, and a negative would close the liability far more cheaply than arguing about it.
  4. Is AFF3 expressed at all in the tissues an infant leukaemia therapy would reach at exposure? A GTEx lookup bounds the geography of the risk and has not been done in this beat.
Every source behind this entry (5)
  1. E1strong confidence 4 of 5: Replicated in peer-reviewed work by more than one group.paper

    ENL/MLLT1 and AF9/MLLT3 are subunits of SEC-L2 (AFF2) and SEC-L3 (AFF3) as well as of SEC, and AFF2 and AFF3 mRNA are about 10-fold lower than AFF4 in the HEK293T background in which the three complexes were compared.

    The super elongation complex family of RNA polymerase II elongation factors: gene target specificity and transcriptional output. · Luo Z, Lin C, Guest E, Garrett AS, Mohaghegh N, Swanson S, Marshall S, Florens L, Washburn MP, Shilatifard A · Molecular and Cellular Biology 2012
    PMID 22547686DOI
    What this evidence cannot show

    HEK293/293T, over-expressed tagged baits, one laboratory. The purification establishes co-recovery, not stoichiometry, and says nothing about whether the readers accompany AFF3 to an imprinted control region in a cell that is doing imprinting. No brain, no germline, no developing tissue.

    Evidence that argues against this (1)
    • The 10-fold expression gap is the single strongest argument against this whole risk entry, and it is in the same paper as the composition result the entry rests on. A complex present at a tenth the abundance of its paralogue absorbs correspondingly little of any pocket-directed drug, so the liability could be real in kind and negligible in size. Nothing anywhere measures the reader-pool partition, so neither reading can be excluded. Luo Z et al. 2012
  2. E2strong confidence 4 of 5: Replicated in peer-reviewed work by more than one group.paper

    In mouse embryonic stem cells AFF3 — described by the authors as the central factor of SEC-L3 — binds gametic differentially methylated regions and enhancers within imprinted loci allele-specifically, and AFF3 at the Meg3 enhancer is required for expression of the approximately 200 kb Meg3 polycistron within the imprinted Dlk1-Dio3 locus.

    Regulation of the imprinted Dlk1-Dio3 locus by allele-specific enhancer activity. · Luo Z, Lin C, Woodfin AR, Bartom ET, Gao X, Smith ER, Shilatifard A · Genes & Development 2016
    PMID 26728555DOI
    What this evidence cannot show

    Mouse embryonic stem cells, not human and not an infant tissue. `expected_absent_terms` carries the load-bearing negative: the paper does not measure ENL or AF9 at any of these regions, so the link from 'AFF3 does imprinting' to 'a YEATS reader is involved in imprinting' is unmade. It is also the same laboratory that defined SEC-L3, so the scaffold's identification with the complex comes from one group throughout.

    Evidence that argues against this (1)
    • AFF3 could be doing this job outside SEC-L3 entirely. ENL and AF9 are known to exist outside SEC when bound to DOT1L, and no imprinted-locus experiment in any paper has blotted a single non-AFF3 subunit of SEC-L3. If AFF3 acts as a monomer or in some other assembly at gametic DMRs, this entire risk entry has no mechanism, and the experiment that decides it is the one in the mitigation list. Wang Y et al. 2017
  3. E3strong confidence 4 of 5: Replicated in peer-reviewed work by more than one group.paper

    AFF3, previously identified as the central factor of SEC-L3, binds the differentially methylated region downstream of the XIST promoter in a DNA-methylation-dependent way, and knockdown of AFF3 de-represses the INACTIVE allele of XIST in terminally differentiated cells — so loss of AFF3 function measurably breaks a monoallelic expression state in human cells.

    AFF3-DNA methylation interplay in maintaining the mono-allelic expression pattern of XIST in terminally differentiated cells. · Zhang Y, Wang C, Liu X, Yang Q, Ji H, Yang M, Xu M, Zhou Y, Xie W, Luo Z, Lin C · Journal of Molecular Cell Biology 2019
    PMID 30535390DOI
    What this evidence cannot show

    Knockdown, not pharmacology, and AFF3 loss is not the same perturbation as occupying a partner subunit's reader pocket. Terminally differentiated cells, one laboratory, and again no ENL or AF9 measurement anywhere in it. This is the clearest available demonstration that an AFF3 perturbation breaks monoallelic expression, and it is three steps removed from what a YEATS antagonist does.

    Evidence that argues against this (1)
    • The perturbation gap cuts hard here. Removing AFF3 protein removes the scaffold, the P-TEFb it delivers and any reader it carries; a YEATS-pocket ligand removes only the histone contact of a shared subunit and leaves the scaffold bound. The AML mutagenesis work shows those are different perturbations even for ENL itself — a YEATS-dead ENL is still in its complex. So a strong AFF3-knockdown phenotype is a weak predictor of a YEATS-ligand phenotype, and treating the two as interchangeable would overstate this risk substantially. Wan L et al. 2017
  4. E4established confidence 5 of 5: Multiple independent lines of evidence; treated as settled in the field.paper

    AFF2/FMR2 is silenced in FRAXE intellectual disability, all four AFF family members localise to nuclear speckles and bind G-quadruplex RNA, and the authors propose that redundancy among AFF family members may compensate for loss of AFF2 activity and explain the relatively mild FRAXE phenotype.

    Functional characterization of the AFF (AF4/FMR2) family of RNA-binding proteins: insights into the molecular pathology of FRAXE intellectual disability. · Melko M, Douguet D, Bensaid M, Zongaro S, Verheggen C, Gecz J, Bardoni B · Human Molecular Genetics 2011
    PMID 21330300DOI
    What this evidence cannot show

    FRAXE arises from CCG-repeat expansion silencing the AFF2 gene — complete loss of expression from development onward — which is a different perturbation class from reversible pharmacological occupancy of a partner subunit's reader pocket in an infant. This item establishes that AFF2 loss has a human neurodevelopmental consequence; it does not establish that a YEATS antagonist produces any part of it. The independent instrument agreeing with the disease assignment is the MedGen gene-phenotype link for AFF2 (`ana-q187-aff1-is-the-only-aff-scaffold-with-no-named-mendelian-syndrome`).

    Evidence that argues against this (1)
    • This paper argues the OPPOSITE of the non-redundancy framing the rest of this beat relies on: it reports that AFF2, AFF3 and AFF4 all localise to speckles, all bind G-quadruplex RNA and all modulate the same mini-gene's splicing, and concludes there is 'functional redundancy among the AFF family members', offering that as the explanation for FRAXE's mild phenotype. The KINSSHIP paper reaches the opposite conclusion from transcriptome data, that the ALF factors are not redundant. The field disagrees with itself on AFF-family redundancy, and both this risk entry and the target-set analysis pick one side. Voisin N et al. 2021
  5. E5moderate confidence 3 of 5: One strong study, or several smaller consistent ones.paper

    The only published therapeutic-window measurement for a YEATS antagonist in KMT2A::AFF1 reports efficacy in vitro and in vivo alongside two haematopoietic safety endpoints — unaltered in vitro colony-forming potential of human HSPCs and unaffected long-term in vivo function of mouse HSPCs — and no non-haematopoietic endpoint of any kind.

    Inhibition of MLLT1 Limits Growth of KMT2A::AFF1 Leukemias Without Killing Healthy Hematopoietic Stem Cells. · Rajhansa S, Crump NT, Khoo HM, Dopico-Fernandez A, Bozhilov Y, Brennan PE, Fedorov O, Adams C, Farnie G, Milne TA, Wilkinson AC · Experimental Hematology 2026
    PMID 42242459DOI
    What this evidence cannot show

    ABSTRACT ONLY — the full text is not in PMC and this beat did not read it. The `expected_absent_terms` above assert the load-bearing negative for THE ABSTRACT; the full text may contain endpoints the abstract does not mention, and this entry's central claim would weaken if it did. Anyone acting on this should obtain the paper. One study, one group overlapping the probe's originating consortium.

    Evidence that argues against this (1)
    • An absence of non-haematopoietic endpoints in an Experimental Hematology paper is close to what publication convention guarantees, and this entry should not treat it as a finding about the compound. The informative version of this claim would be a bound over a regulatory toxicology package or a broader profiling dataset, neither of which exists for a chemical probe. What the item genuinely establishes is narrow: the widely quoted 'therapeutic window' for this class rests on two blood endpoints. Moustakim M et al. 2018
Connected to