MLLT10 (AF10): the discarded PZP reader — the one partner where loss is proven to be the oncogenic event
also called AF10 PZP loss · gain by subtraction MLLT10 · PZP add-back CALM-AF10
Normal AF10 has a gripper that grabs the DOT1L enzyme and a separate sensor that decides where DOT1L is allowed to work. The leukemia fusion keeps the gripper and throws away the sensor, and when researchers put the sensor back, the cancer-causing activity disappears.
The same thing, for a scientist
KMT2A::MLLT10 and CALM::MLLT10 retain the AF10 octapeptide-motif/leucine-zipper (OM-LZ, aa 704-779) that binds the DOT1L coiled coil, and discard the N-terminal PZP module (aa 22-198) that reads unmodified H3K27 and licenses DOT1L-mediated H3K79 methylation. The discard is causal rather than incidental: Klein et al. (Nat Commun 2021) re-supplied a functional PZP in cis to CALM-AF10 and abolished transformation of HSPCs in vitro and in vivo, downregulated Hoxa genes and reversed established transformation. This is the only KMT2A-partner-family case in which the in-cis add-back experiment has actually been performed, and it was positive.
What PZP does in the intact protein
AF10 residues 22-198 fold as a single PHD-Zn-knuckle-PHD module (not three separate fingers) that binds histone H3 residues 22-27, an interaction abolished by any modification of H3K27. In cells, PZP recognition of H3 is required for H3K79 dimethylation, for expression of DOT1L target genes, and for proliferation of DOT1L-addicted leukemic cells 1. PZP also engages the nucleosome core particle through multivalent H3-tail and DNA contacts, colocalises with active methylation marks, and discriminates against H3K27me3 2.
Intact AF10 is therefore a two-part device: the C-terminal OM-LZ recruits DOT1L 3; the N-terminal PZP licenses where that recruitment is productive.
Is it genuinely discarded at the observed breakpoints?
Yes, and this is a sequence fact rather than an inference. This graph records the paediatric KMT2A::MLLT10 architecture (bp-kmt2a-intron-9-mllt10-omlz) as retaining OM, LZ and the AF10 C-terminal region and losing the PZP module, the NLS and the AT-hook region. The MLLT10 break falls in the roughly 500 residues of predicted disorder between PZP (ends 198) and OM-LZ (starts 704), so essentially every leukemogenic AF10 translocation lands N-terminal to OM-LZ and C-terminal to PZP. Klein et al. state directly that the PZP domain is consistently impaired or deleted in leukemogenic AF10 translocations 2.
Caveat worth keeping: MLLT10 rearrangements are structurally heterogeneous and frequently cryptic, so no single exon pair describes all cases. What is uniform is the module-level outcome, not the nucleotide.
Does the loss contribute to transformation? Yes — and the experiment has been done1 deeper
This is the distinction the whole subtraction beat turns on. Absence of a domain is cheap to establish; absence being oncogenic is not. For MLLT10 it has been established.
Klein et al. incorporated a functional AF10 PZP into the CALM-AF10 fusion in cis and found it prevents transforming activity in bone-marrow HSPCs in vitro and in vivo, abrogates CALM-AF10-mediated leukemogenesis in vivo, downregulates Hoxa genes and reverses cellular transformation. Their stated conclusion is that disruption of PZP function in the fusion directly leads to transformation 2.
That is precisely the in-cis re-supply design the King's framing audit proposed as the decisive test, and it was published in 2021.
A correction to the entry that commissioned this beat
mech-retained-half-framing-audit states, in its counter-evidence block and again in its open questions, that re-supplying PZP in cis 'as far as this graph knows has not been done'. It has been done — for CALM-AF10, in Klein et al. 2021 2, a paper the graph already cites as E6 of protein-mllt10-domains without connecting it to the framing argument. Two entries in this graph held the same citation and drew opposite conclusions about whether the experiment existed.
The residual gap is narrower and more specific than the King stated: the add-back has been done for CALM-AF10, not for KMT2A-AF10. The two fusions differ in their 5' partner and therefore in targeting and localisation, so the result does not transfer automatically. Closing that gap is what exp-pzp-cis-resupply-kmt2a-mllt10 is for.
What it means for drugging this partner
There is no retained neo-surface created by the deletion; you cannot inhibit an absence. Two consequences follow, and they point in different directions.
- Against retained-module scanning: the target-selection logic used everywhere else in this graph would return, for MLLT10, only the OM-LZ/DOT1L interface. That is a real target, structurally resolved (PDB 6CKN, 6CKO, 6JN2), and its disruption abrogates MLL-AF10 transformation 3. So retained-module scanning is not empty here; it is blind to half the mechanism.
- For restoration rather than inhibition: Klein et al. showed that restoring the lost constraint reverses transformation. That is a different therapeutic modality from an inhibitor — closer to forcing the retained OM-LZ back under H3K27-dependent control — and nothing in this project's target work has considered it.
| partner | gene-mllt10 |
|---|---|
| fusion | fusion-kmt2a-mllt10 |
| discarded region | PZP module (PHD1-Zn knuckle-PHD2) |
| retained oncogenic module | OM-LZ (octapeptide motif-leucine zipper), aa 704-779, binds the DOT1L coiled coil |
| intact function of discarded region | Single folded reader module binding unmodified H3K27 (H3 aa 22-27); licenses DOT1L-mediated H3K79 methylation; multivalent nucleosome engagement through H3 tail and DNA; discriminates against H3K27me3 |
| genuinely discarded at observed breakpoints | true |
| loss shown to contribute to transformation | true |
| verdict | subtraction |
| verdict confidence | strong for CALM-AF10, moderate for KMT2A::MLLT10 by extrapolation |
| experiment to settle | exp-pzp-cis-resupply-kmt2a-mllt10 |
| already done for | CALM-AF10 (Klein et al. 2021, PMID 34226546) |
| still open for | KMT2A::MLLT10 |
discarded residues
{
"start": 22,
"end": 198,
"uniprot": "P55197",
"numbering": "P55197-1 (1068 aa)"
}- The therapeutic implication of a proven subtraction mechanism is restoration, not inhibition, and restoring a deleted chromatin reader in cis is not a druggable modality with current technology. Recognising the mechanism does not hand you a drug.
- If PZP loss releases DOT1L from targeting control, DOT1L inhibition should be especially effective here — yet pinometostat failed clinically across KMT2A-r disease. Either the inference is wrong, or the drug failed for pharmacological reasons a mechanism-level argument cannot rescue.
- Restoring an H3K27-reading module would couple the fusion's activity to H3K27 methylation status, which is itself therapeutically manipulated. A restoration strategy could be undone by any co-treatment that raises H3K27me3.
- Does in-cis PZP re-supply suppress KMT2A::MLLT10 transformation as it does CALM-AF10, or does KMT2A's own targeting apparatus make the fusion insensitive to restored H3K27 licensing?
- Klein et al. found PZP also drives nuclear localisation of CALM-AF10. How much of the suppression is licensing and how much is mislocalisation? A PZP mutant that binds H3 but cannot drive localisation, or the reverse, would separate them.
- Wild-type AF10 with an intact PZP is still expressed from the untranslocated allele in these cells. Why does it not supply the missing constraint in trans? If the constraint is strictly cis-acting, that is itself a testable mechanistic claim.
- If PZP loss releases an H3K27-dependent brake, does H3K27me3 status modulate KMT2A::MLLT10 leukemia? EZH2 inhibition should then have a partner-specific effect that pooled trials would not detect.
- E1strong confidence 4 of 5: Replicated in peer-reviewed work by more than one group.paper
An AF10 PHD-Zn-knuckle-PHD (PZP) region folds into a single module that recognises H3 residues 22-27, the interaction is abrogated by H3K27 modification, and in cells PZP recognition of H3 is required for H3K79 dimethylation, DOT1L-target gene expression and proliferation of DOT1L-addicted leukemic cells.
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 2015What this evidence cannot show
Characterises PZP in intact AF10 and in DOT1L-addicted leukemia cells. It does not itself test the fusion context; that is E2. Verified against PubMed on 2026-09-18 — title, author list, journal, year and abstract all match the claim.
Evidence that argues against this (1)
- The module's parent protein is a POSITIVE requirement for leukaemogenesis in several fusions, which is awkward for reading PZP as a brake. Deshpande et al. showed that H3K79me2 is converted to me1 at HOX loci as haematopoietic cells mature, coinciding with falling HOX expression; that AF10 regulates both H3K79 methyltransferase activity and the me1-to-me2 conversion; and that AF10 inactivation reverses leukaemia-associated epigenetic profiles, precludes abnormal HOXA gene expression and impairs the transforming ability of MLL-AF9, MLL-AF6 and NUP98-NSD1. On that reading, PZP-containing AF10 supplies a licensing step that leukaemia needs, and what the fusion does by discarding PZP is relocate where that licensing happens rather than remove a constraint. The two readings are distinguishable - the licensing model predicts that endogenous AF10 remains required in KMT2A::MLLT10 cells - and nobody has run that test. Deshpande AJ et al. 2014
- E2strong confidence 4 of 5: Replicated in peer-reviewed work by more than one group.paper
The PZP domain of AF10 is consistently impaired or deleted in leukemogenic AF10 translocations and blocks malignant transformation: incorporating functional PZP into CALM-AF10 in cis prevents transformation of bone-marrow HSPCs in vitro and in vivo, abrogates leukemogenesis in vivo, downregulates Hoxa genes and reverses cellular transformation, leading the authors to conclude that disruption of PZP function in the fusion directly causes transformation.
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 2021What this evidence cannot show
The add-back was done on CALM-AF10, not KMT2A::MLLT10. Mouse bone-marrow HSPC and transplant models. PZP also promotes nuclear localisation of CALM-AF10, so part of the suppression may be mislocalisation rather than restored H3K27-dependent licensing; the authors report both effects and do not fully separate them. Verified against PubMed on 2026-09-18.
Evidence that argues against this (1)
- Because PZP re-addition also changes the subcellular localisation of CALM-AF10, an alternative reading is that the add-back works by mislocalising the fusion rather than by restoring a chromatin-licensing constraint. For KMT2A::MLLT10 the 5' partner already supplies nuclear localisation and chromatin targeting, so that escape route would not apply — which is exactly why the KMT2A version of the experiment is informative and not redundant. Klein BJ et al. 2021
- E3strong confidence 4 of 5: Replicated in peer-reviewed work by more than one group.paper
Crystal structures of apo AF10 OM-LZ and of its complex with the DOT1L coiled-coil domain (PDB 6CKN, 6CKO; AF10 chains 704-779) define the recognition interface, and disrupting that interface abrogates MLL-AF10-associated leukemic transformation.
Structural and functional analysis of the DOT1L-AF10 complex reveals mechanistic insights into MLL-AF10-associated leukemogenesis · Zhang H, Zhou B, Qin S, Xu J, Harding R, Tempel W, Nayak V, Li Y, Loppnau P, Dou Y, Min J · Genes & Development 2018What this evidence cannot show
Structure plus interface mutagenesis in transformation assays; not a small-molecule inhibitor result. Verified against PubMed on 2026-09-18.
Evidence that argues against this (1)
- Disrupting the OM-LZ/DOT1L interface and inhibiting DOT1L are the same axis, so the structure supports an addition mechanism at least as strongly as a subtraction one. Chen et al. showed by conditional Dot1l knockout that MLL-AF10 and CALM-AF10 transformation is Dot1l-dependent - abolishing in vitro transformation of murine bone-marrow cells and both initiation and maintenance in vivo - and that the selective inhibitor EPZ004777 suppressed Hoxa cluster genes and Meis1, selectively impaired proliferation of AF10-fusion-transformed cells, and reduced their spleen-colony-forming ability after pretreatment. An interface whose mutation abrogates transformation because it stops H3K79 methylation is the fusion's gain of function; the structure work says nothing about what the discard cost. Chen L et al. 2013
Counter-evidence for the PZP subtraction: H3K79 hypermethylation and Hoxa9 upregulation occur with no AF10 sequence at all - fusing hDOT1L directly to MLL transforms in a methyltransferase-dependent manner (Okada 2005, PMID 15851025). And a competing account puts the decisive element on the CALM side: NES motifs from four unrelated proteins fused to AF10 are SUFFICIENT to immortalise progenitors (Conway 2013, PMID 23487024).
the same subtraction, tested by cis restoration in CALM-AF10
Counter-evidence (Chen 2013, PMID 23138183): MLL-AF10 and CALM-AF10 transformation is abolished by conditional Dot1l knockout and suppressed by EPZ004777, so the retained OM-LZ -> DOT1L addition mechanism is sufficient to explain initiation and maintenance without invoking PZP loss
- ← depends onSubtraction is a partner class, not a curiosity — and the class is defined by compartment, not by sequence
Counter-evidence (Chen 2013, PMID 23138183): MLL-AF10 and CALM-AF10 transformation is Dot1l-dependent for initiation and maintenance, so a sufficient addition mechanism already explains the founding member of the discarded-module sub-type
the PZP module lost in MLL-AF10 mediates AF10 oligomerisation, a possible nuclear-retention mechanism nobody has tested
the OM-LZ/DOT1L interface argument is an H3K79 argument: conditional Dot1l knockout and EPZ004777 abolish MLL-AF10 and CALM-AF10 transformation (Chen 2013)
discarded PZP module of the partner protein; both entries now bounded by AF10's positive requirement in other fusions (Deshpande 2014)
supplies the causal evidence the audit said was missing, and corrects its claim that the in-cis add-back had never been done
mechanism class of this fusion: gain by subtraction
PZP module (aa 22-198) discarded from the fusion
domain coordinates and fate assignments for AF10
discarded PZP normally licenses where the retained OM-LZ takes DOT1L
PZP loss releases H3K79 methylation from H3K27-dependent targeting
same architecture — retain the recruiter, discard the targeting reader — but the causal test is missing here
- ← testsRe-supply the AF10 PZP module in cis to KMT2A::MLLT10 and ask whether transformation is suppressed
transfers the CALM-AF10 add-back design to the KMT2A fusion, with a binding-dead PZP control
per-partner entry feeding the synthesis