Oncofusion

start / graph / exp-pzp-cis-resupply-kmt2a-mllt10

Experimentmoderate confidence 3 of 5: One strong study, or several smaller consistent ones.Q11b-subtraction

Re-supply the AF10 PZP module in cis to KMT2A::MLLT10 and ask whether transformation is suppressed

also called PZP add-back MLL-AF10 · in-cis restoration experiment

Put the missing sensor back into the fusion protein and see whether the leukemia stops. This has already worked for a related fusion; the point is to test it on the one that matters here.

The same thing, for a scientist

Direct transfer of the Klein et al. CALM-AF10 design to KMT2A::MLLT10. Construct MLL-AF10 variants that restore the AF10 PZP module (residues 22-198) in cis, in four arms: patient-like (no PZP), wild-type PZP, H3-binding-dead PZP point mutant, and PZP with the interstitial region restored as a length control. Score myeloid progenitor immortalisation, Hoxa9/Meis1 output, H3K79me2 distribution by CUT&RUN, subcellular localisation, and leukemia latency on transplant.

Go deeper
Why this experiment and not another

The subtraction claim for MLLT10 currently rests on a positive result in CALM-AF10 [see subtraction-mllt10-pzp]. CALM and KMT2A are very different 5' partners: CALM is a cytoplasmic endocytic protein whose fusion depends on nuclear export signal biology, while KMT2A supplies AT-hooks, CXXC, menin and LEDGF tethering. Klein et al. explicitly found that PZP promotes nuclear localisation of CALM-AF10, so part of their suppression could be localisation rather than chromatin licensing — a confound that KMT2A's own targeting apparatus would remove.

Running the experiment in the KMT2A context therefore does two things at once: it tests whether the mechanism generalises to the fusion this project cares about, and it subtracts the localisation confound from the original result.

Design

Arms (retroviral constructs, murine myeloid progenitors, matched expression verified by immunoblot):

  1. MLL-AF10 patient-like — KMT2A exon 9 junction to AF10 retaining OM-LZ, no PZP. Positive control for transformation.
  2. MLL-PZP-AF10 — PZP (aa 22-198) restored in cis, N-terminal to the AF10 segment.
  3. MLL-PZP(mut)-AF10 — the same, with PZP point mutations that abolish H3(22-27) binding while preserving the fold. This is the arm that makes the result interpretable: it separates 'restoring a folded domain that sterically interferes' from 'restoring a chromatin-reading constraint'.
  4. MLL-linker-AF10 — an inert spacer of equivalent length. Length control for arm 2.

Readouts: serial replating and colony morphology; Hoxa9 and Meis1 transcript levels; H3K79me2 CUT&RUN genome-wide, scored specifically for redistribution towards versus away from H3K27-unmodified regions; fusion protein localisation by immunofluorescence; and survival on transplant into sublethally irradiated recipients.

The decisive comparison is arm 2 versus arm 3. If wild-type PZP suppresses and the binding-dead mutant does not, the constraint is chromatin reading and MLLT10 is confirmed as a subtraction partner in the KMT2A context. If both suppress equally, the effect is steric or structural and the subtraction interpretation fails.

Facts and figures
designFour-arm retroviral structure-function study in murine myeloid progenitors with transplant follow-on: MLL-AF10 (patient-like, no PZP), MLL-PZP-AF10 (wild-type PZP restored in cis), MLL-PZP(H3-binding-dead)-AF10, and MLL-linker-AF10 length control. Expression matched by immunoblot across arms.
readoutSerial replating and colony morphology; Hoxa9/Meis1 transcript levels; genome-wide H3K79me2 by CUT&RUN scored for redistribution relative to H3K27 modification state; fusion protein subcellular localisation; leukemia-free survival on transplant.
controlsBinding-dead PZP point mutant (separates chromatin reading from steric/structural effects); inert linker of matched length; empty vector; matched expression levels verified by immunoblot. NOT controlled for: change in fusion self-association/oligomerisation state caused by inserting a folded module (So 2003, PMID 12957285, showed synthetic dimerisation modules alone convert MLL into an oncoprotein), and vector-capacity effects of adding ~530 bp of coding sequence to an already large cassette (Wachter 2014, PMID 25016062).
decisive comparisonarm 2 (wild-type PZP) versus arm 3 (binding-dead PZP). Suppression in arm 2 only confirms a chromatin-licensing constraint; equal suppression in both refutes it.
cost tiermed
timeline9-15 months including transplant follow-up
falsifieshyp-subtraction-is-a-partner-class
prior result to beatKlein et al. 2021, positive in CALM-AF10 (PMID 34226546)
added arms recommended
[
  "Direct measurement of fusion self-association across all four arms (size-exclusion chromatography, BiFC or in-cell crosslinking), so that an equal suppression in arms 2 and 3 can be attributed to oligomerisation rather than recorded as uninformative 'steric' effect.",
  "Sleeping Beauty or other size-unconstrained delivery in parallel with the retroviral arm, so that a null in arm 2 cannot be a packaging artefact.",
  "An endogenous-Mllt10-deleted background, to reconcile a positive result with the requirement for intact AF10 in other fusions (Deshpande 2014)."
]
How this could bite us back
  • A positive result would identify a mechanism with no inhibitor-shaped therapeutic handle, which is a scientifically valuable and clinically frustrating combination. The programme should decide in advance what it would do with a positive answer.
  • A negative result does not restore the retained-half framing; it would only narrow the subtraction class back to CALM-AF10 while leaving the AFDN sequestration and KMT2A-PTD allelic mechanisms untouched.
  • The decisive arm-2-versus-arm-3 comparison controls for histone binding but not for self-association, and MLL chimeras are known to be activated by dimerisation alone (So 2003). Equal suppression in both arms would be scored 'steric' and the cycle would end with no answer, having spent 9-15 months.
Still open
  1. Would a KMT2A::MLLT10 patient-derived xenograft be a better test bed than murine progenitors, given that MLLT10 rearrangements are structurally heterogeneous and the retroviral construct necessarily picks one architecture?
  2. Should a fifth arm supply PZP in trans (co-expressed, not fused) to test whether the constraint is genuinely cis-acting? Wild-type AF10 is already present in these cells and evidently does not suppress, which is itself unexplained.
  3. How will a null result in arm 2 be distinguished from a delivery failure? MLL fusion cassettes already strain retroviral packaging and only 1 of 4 direct MLL fusions was oncogenic in a size-unconstrained comparison (Wachter 2014). A pre-registered positive control for cassette size - the patient-like construct padded with the same number of inert codons - would settle it and is cheap.
Every source behind this entry (2)
  1. E1strong confidence 4 of 5: Replicated in peer-reviewed work by more than one group.paper

    The equivalent experiment in CALM-AF10 gave a positive result: incorporating functional PZP into the fusion prevented transformation of bone-marrow HSPCs in vitro and in vivo, abrogated leukemogenesis, downregulated Hoxa genes and reversed cellular transformation, and PZP also promoted nuclear localisation of CALM-AF10.

    The role of the PZP domain of AF10 in acute leukemia driven by AF10 translocations · Klein BJ, Deshpande A, Cox KL, Xuan F, Zandian M, Barbosa K, Khanal S, Tong Q, Zhang Y, Zhang P, Sinha A, Bohlander SK, Shi X, Wen H, Poirier MG, Deshpande AJ, Kutateladze TG · Nature Communications 2021
    PMID 34226546DOI
    What this evidence cannot show

    Establishes feasibility and expected effect size for the CALM context only. Verified against PubMed on 2026-09-18.

    Evidence that argues against this (2)
    • The design's discriminating comparison is not as clean as it looks, because MLL fusions can be switched on and off by OLIGOMERISATION state rather than by anything the inserted module reads. So et al. showed that the coiled-coil oligomerisation domains of AF1p/Eps15 and GAS7 are necessary AND SUFFICIENT for leukemogenic transformation by those MLL fusions, and that homodimerising MLL with a synthetic dimerisation module - a domain with no chromatin function at all - reproduces bona fide MLL fusion behaviour, activating Hox genes and enhancing progenitor self-renewal. Inserting a folded 177-residue PZP module between MLL and AF10 changes the spacing, valency and self-association geometry of the chimera. An H3-binding-dead point mutant controls for histone reading but NOT for that: both arm 2 and arm 3 carry the same fold in the same place, so if the insertion suppresses transformation by disrupting an oligomerisation-dependent architecture, both arms suppress equally and the design reads that as 'steric' - correctly - but the experiment then returns no information rather than a negative. The arm that is missing is a direct measurement of fusion self-association (size-exclusion, BiFC or crosslinking) across all four arms. So CW et al. 2003
    • MLL-fusion transformation assays fail for construct and vector reasons often enough that a null arm is weak evidence. Wachter et al. built a Sleeping Beauty system specifically because MLL fusion cDNA cassettes are large and difficult to package into retro- or lentiviral particles, which imposes size limits and integration-site preferences on exactly the kind of study proposed here. Testing eight MLL alleles in that size-unconstrained system, only 1 of 4 direct MLL fusions showed oncogenic function, while 3 of 4 reciprocal fusions did. Adding a 177-residue PZP module to an already large MLL-AF10 cassette pushes it further against the packaging limit of a retroviral vector, and a failure to transform in arm 2 is therefore consistent with reduced titre, truncated product or silenced integration as well as with the hypothesis. Immunoblot expression matching across arms is necessary but does not control for integration-site or product-integrity effects. Wachter K et al. 2014
  2. E2strong confidence 4 of 5: Replicated in peer-reviewed work by more than one group.paper

    PZP recognition of H3 residues 22-27 is abrogated by H3K27 modification and is required in cells for H3K79 dimethylation, DOT1L-target gene expression and proliferation of DOT1L-addicted leukemic cells, which supplies the binding-dead point mutations needed for the discriminating arm.

    The PZP Domain of AF10 Senses Unmodified H3K27 to Regulate DOT1L-Mediated Methylation of H3K79 · Chen S, Yang Z, Wilkinson AW, Deshpande AJ, Sidoli S, Krajewski K, Strahl BD, Garcia BA, Armstrong SA, Patel DJ, Gozani O · Molecular Cell 2015
    PMID 26439302DOI
    What this evidence cannot show

    The specific cage residues to mutate must be taken from the structures (PDB 5DAG, 5DAH, 7MJU) rather than from the abstract. Verified against PubMed on 2026-09-18.

    Evidence that argues against this (1)
    • The experiment's premise - that PZP-containing AF10 is a constraint the fusion escapes - is in tension with published loss-of-function data on endogenous AF10. Deshpande et al. found that AF10 inactivation impairs the transforming ability of MLL-AF9, MLL-AF6 and NUP98-NSD1 and precludes abnormal HOXA expression, so intact AF10 is required for leukaemogenesis in three other settings. If a positive result is obtained in arm 2, the interpretation 'PZP constrains the fusion' will have to be reconciled with 'AF10 is needed by other fusions', and the cleanest way to do that is to include an arm in which endogenous Mllt10 is deleted, which the current design does not. Deshpande AJ et al. 2014
Connected to