Oncofusion

start / graph / mechanism-calm-af10-focal-hoxa-h3k79-hypermethylation

Mechanismstrong confidence 4 of 5: Replicated in peer-reviewed work by more than one group.Q55-calm-af10

The focal half: H3K79 hypermethylation at the HOXA cluster and MEIS1 - but which cluster member differs between CALM-AF10 and MLL-AF10

also called CALM-AF10 HOXA activation

At a handful of growth-control genes, the CALM-AF10 fusion piles on the DOT1L chemical tag instead of removing it, and those genes switch on. Switching them on is what stops the blood cells maturing.

The same thing, for a scientist

Against the global depletion, CALM-AF10 produces **focal H3K79 hypermethylation at HOXA cluster genes and MEIS1**, and those genes are transcriptionally upregulated [E1][E2][E3]. Transgenic mice expressing CALM-AF10 overexpress Hoxa5/7/9/10 and develop acute leukaemia [E2]; patient T-ALL samples overexpress HOXA5/9/10, MEIS1 and BMI1 [E4]. Dot1l knockout in CALM-AF10-transformed murine bone marrow removes H3K79me2, restores differentiated colonies and reduces Hoxa and Meis1 expression [E3]. **The surprising control point is the CALM nuclear export signal**: mutating it abolishes both the HOXA overexpression and the focal hypermethylation, and fusing the bare NES to AF10 reconstitutes both at levels comparable to full-length CALM-AF10 [E5]. So a ten-residue export motif, not a chromatin-reading module, is the switch. CRM1 itself binds HOXA chromatin in a pattern resembling CALM-AF10's, and a **CRM1-AF10 chimera** - CRM1 in place of CALM entirely - causes leukaemia with HOXA activation [E6]. **One correction to the usual summary, added after reading Okada et al. in full text:** in that murine system **Hoxa9 was undetectable in CALM-AF10-transformed cells** while **Hoxa5** carried the phenotype, and **Hoxa5 knockout attenuated CALM-AF10 transformation while having no effect on MLL-AF10** [E8]. So the two AF10 fusions are dissociable inside the HOXA cluster, and 'CALM-AF10 drives the same HOXA9 programme' should not be written without that qualification.

Go deeper
The oddity worth sitting with

In a KMT2A fusion, the chromatin address is written into the protein: AT-hooks, CXXC, the menin-binding motifs. In CALM-AF10 the corresponding function is performed by an export receptor. Conway et al. propose that CRM1 is itself prebound at HOXA chromatin and that the fusion's NES is a docking signal rather than an export signal in this context 6, which is the same architecture reported for NUP98-HoxA9 and chromatin-prebound Crm1. If that is right, the CALM-AF10 'address' is a nucleoporin/exportin address, and the analogy to KMT2A fusions is output-level, not mechanism-level.

This also makes the disease's overexpression of XPO1 (the CRM1 gene) in PM-AML and PM-T-ALL interesting rather than incidental 7.

Which HOXA gene, and why it matters here

The literature summarises this fusion as HOXA-driven, and that is right at cluster level and loose at gene level.

  • Okada et al., murine bone marrow and U937: Hoxa5 tracks CALM-AF10, Hoxa5 knockout attenuates CALM-AF10 transformation and does nothing to MLL-AF10, and Hoxa9 was undetectable in the transformed cells 8.
  • Caudell et al., CALM-AF10 transgenic mice: Hoxa5, Hoxa7, Hoxa9 and Hoxa10 overexpressed 2.
  • Conway et al., murine PICALM::MLLT10 leukaemia: H3K79 hypermethylation at Hoxa7, Hoxa9, Hoxa10 and Hoxa11 promoters 5.
  • Patients (T-ALL): HOXA5, HOXA9, HOXA10, MEIS1 and BMI1 elevated 4.

The three later results are not in conflict with the first as measurements; they are in conflict with it as a requirement claim, and only Okada et al. tested requirement gene by gene. Nobody has repeated the Hoxa5-knockout experiment, and nobody has asked which cluster member is required in a human CALM-AF10 cell. For this beat the consequence is sharp: the brief's premise that CALM-AF10 drives 'the same HOXA/MEIS1 programme' is supported at cluster and output level and contradicted at the level of which gene is necessary, in the one experiment that asked.

Facts and figures
mechanism classepigenetic activation / transcriptional
markH3K79me2 at HOXA promoters (Hoxa7, Hoxa9, Hoxa10, Hoxa11 measured)
control elementCALM NES, aa 544-553 (Q13492-1 numbering) - necessary and sufficient
proposed addressing partnerCRM1/XPO1, reported prebound at HOXA chromatin
orthogonal supporta CRM1-AF10 chimera causes leukaemia with HOXA activation
menin involvementnone reported; the addressing element is an exportin, not menin
target genes
[
  "HOXA5",
  "HOXA7",
  "HOXA9",
  "HOXA10",
  "HOXA11",
  "MEIS1",
  "BMI1"
]
How this could bite us back
  • Targeting CRM1/XPO1 blocks the export of hundreds of cargoes; selinexor's toxicity profile is the cost of that breadth, and the CALM-AF10 rationale does not buy any selectivity.
  • If the NES is a docking signal rather than an export signal, an export inhibitor could stabilise the fusion on chromatin rather than displace it - the intervention could point the wrong way.
Still open
  1. Do CALM-AF10 and menin occupy the same HOXA elements? Nobody has run the co-occupancy experiment, and it is the one that decides whether the analogy to KMT2A fusions is mechanistic or merely output-level.
  2. Is the NES acting as an export signal or as a CRM1 docking signal on chromatin? Conway et al. favour docking; the two make opposite predictions about whether leptomycin B should help or harm.
  3. Does selinexor or another XPO1 inhibitor do anything in CALM-AF10 leukaemia? CALM-AF10 cells are reported selectively sensitive to leptomycin B, and XPO1 is highly expressed, but no clinical XPO1 inhibitor result in this fusion was found.
Every source behind this entry (8)
  1. E1strong confidence 4 of 5: Replicated in peer-reviewed work by more than one group.paper

    CALM-AF10 transformation involves upregulation of Hoxa5 through hDOT1L-mediated H3K79 methylation, and the fusion is both necessary and sufficient for leukaemic transformation in the system tested.

    PMID 16921363DOI
    What this evidence cannot show

    Murine bone-marrow transduction plus U2OS imaging. The paper's model - that hDOT1L acts by PREVENTING nuclear export of CALM-AF10 - is in direct tension with the later NES work [E5].

    Evidence that argues against this (1)
    • Okada et al. conclude hDOT1L contributes by preventing nuclear export of CALM-AF10; Conway et al. show that a nuclear-restricted CALM-AF10 fails to transform and that export is necessary and sufficient. The two readings of the same shuttling cannot both be straightforwardly correct. Conway AE et al. 2013
  2. E2strong confidence 4 of 5: Replicated in peer-reviewed work by more than one group.paper

    A CALM-AF10 transgene in mice drives Hoxa cluster overexpression (Hoxa5, Hoxa7, Hoxa9, Hoxa10) and acute leukaemia.

    PMID 17804713DOI
    What this evidence cannot show

    Transgenic mouse with long latency, implying required cooperating events.

    Evidence that argues against this (1)
    • PUBLISHED CONFLICTING RESULT on how much a transgene phenotype constrains. A second mouse model of the same fusion shows that which disease appears depends on which part of CALM the construct contains: excluding the C-terminal portion of CALM required for optimal clathrin binding gave a MYELOPROLIFERATIVE DISEASE, whereas including it gave ACUTE MYELOID LEUKAEMIA with changes in Pim1 and Crebbp. So 'a CALM-AF10 transgene drives Hoxa overexpression and acute leukaemia' is a statement about one construct in one promoter context, and the long latency this item's caveat notes is compatible with the transgene selecting cooperating events rather than the fusion instructing the Hoxa programme directly. Stoddart A et al. 2012
  3. E3strong confidence 4 of 5: Replicated in peer-reviewed work by more than one group.paper

    Dot1l knockout in CALM-AF10-transformed murine bone-marrow cells abolishes H3K79 dimethylation, promotes differentiated colonies, and reduces Hoxa5/7/9/10 and Meis1 expression; EPZ004777 reproduces the expression effect.

    PMID 23138183DOI
    What this evidence cannot show

    Murine. Tool compound rather than clinical DOT1L inhibitor.

    Evidence that argues against this (1)
    • PUBLISHED CONFLICTING RESULT about which downstream pathway carries the DOT1L dependence. A genetic Dot1l knockout removes H3K79me2 everywhere and collapses HOXA, MEIS1 and everything else at once, so it cannot tell you which arm matters. At sub-saturating exposures it is not the HOXA arm: in MLL-rearranged leukaemia, low-dose pinometostat is cytotoxic far below the concentrations that downregulate HOXA9 and MEIS1, the major pathway is transcriptional downregulation of FLT3, and overexpression of constitutively active STAT5A largely rescues the cytotoxicity and the expression changes. If the same is true here, a DOT1L inhibitor in CALM-AF10 would act through FLT3/STAT5 at achievable exposures and the focal HOXA hypermethylation would be the wrong biomarker. Richter WF et al. 2021
  4. E4strong confidence 4 of 5: Replicated in peer-reviewed work by more than one group.paper

    CALM-AF10-positive patient T-ALL overexpresses HOXA genes and BMI1.

    PMID 16107895DOI
    What this evidence cannot show

    Patient expression profiling; associative. Small numbers of CALM-AF10 cases.

    Evidence that argues against this (1)
    • PUBLISHED CONFLICTING RESULT that bounds how far BMI1 elevation discriminates. This study's contrast is CALM-AF10-positive T-ALL against other T-ALL and against MLL-translocated acute leukaemia, and it is real. But in a separate unselected paediatric T-ALL series of 43 patients, upregulated BMI1 was observed by quantitative real-time PCR in ALL PHENOTYPIC SUBGROUPS - and none of those patients carried the CALM-AF10 fusion transcript. So high BMI1 is not rare in fusion-negative T-ALL, and how much of the CALM-AF10 contrast is fusion-driven rather than cohort- or platform-driven has not been settled. The comparison that would settle it - BMI1 level in CALM-AF10-positive versus CALM-AF10-negative T-ALL, matched for maturation stage and measured on one platform - is in neither paper. Sayitoğlu M et al. 2012
  5. E5strong confidence 4 of 5: Replicated in peer-reviewed work by more than one group.paper

    The CALM NES is necessary and sufficient for HOXA overexpression and for focal H3K79 hypermethylation: a mutated NES abolishes both, and fusing only the NES to AF10 reconstitutes both at levels comparable to full-length CALM-AF10; H3K79 hypermethylation is detectable at the Hoxa7, Hoxa9, Hoxa10 and Hoxa11 promoters in a murine PICALM::MLLT10 leukaemia model.

    PMID 23487024DOI
    What this evidence cannot show

    Murine. 'Sufficient' here means sufficient in combination with AF10, not sufficient alone.

    Evidence that argues against this (1)
    • PUBLISHED CONFLICTING RESULT that breaks the causal chain between export and the mark. The chain here is NES to export to focal hypermethylation. In leukaemia cells the fusion is NUCLEAR, and it suppressed H3K79 methylation regardless of the presence of clathrin - so the localisation state the NES is supposed to establish is not what is observed in a haematopoietic cell, and the H3K79 effect does not require the clathrin arm of the same region. An NES that is necessary for transformation in murine bone marrow can be necessary for something other than steady-state export, and 'necessary and sufficient for focal H3K79 hypermethylation' has been measured only in fibroblasts and a murine leukaemia model, never in a cell where the fusion's localisation was independently confirmed. Stoddart A et al. 2012
  6. E6strong confidence 4 of 5: Replicated in peer-reviewed work by more than one group.paper

    CRM1 binds HOXA chromatin in a pattern resembling CALM-AF10 binding in MEFs and human leukaemia cells, including in the absence of CALM-AF10, and a CRM1-AF10 chimeric protein causes leukaemia with transcriptional activation of HOXA genes.

    PMID 32733011DOI
    What this evidence cannot show

    Murine leukaemia induction plus chromatin occupancy. A chimera that substitutes an export receptor for the 5' partner is a strong argument that the 5' partner's job is CRM1 engagement; it does not prove that is ALL the 5' partner does.

    Evidence that argues against this (1)
    • PUBLISHED ALTERNATIVE for the same substitution. A CRM1-AF10 chimera shows that a CRM1 moiety can replace CALM, and that is consistent with at least two accounts: CRM1 supplies the chromatin address, or CRM1 supplies an oligomerising, chromatin-adjacent bulk that any such moiety would supply. The address question has since been answered differently and directly on this fusion: the MOZ/ENL complex recruits the DOT1L/AF10 fusion complex and maintains chromatin retention through MOZ/MORF acetyltransferase activity, with ENL's YEATS-MOZ interaction critical for CALM-AF10 transformation. A chimera cannot distinguish the two accounts; a requirement experiment on the natural fusion can, and it names an acetylation reader. Komata Y et al. 2023
  7. E7moderate confidence 3 of 5: One strong study, or several smaller consistent ones.paper

    Expression profiling of 20 PICALM::MLLT10 cases confirms HOXA activation and shows relatively high XPO1 expression in both PM-AML and PM-T-ALL.

    PMID 38429501DOI
    What this evidence cannot show

    n=20 and the comparator is haematopoietic stem cells. High XPO1 in leukaemia is common and is not by itself evidence that the CRM1 mechanism is operating.

    Evidence that argues against this (1)
    • METHODOLOGICAL DEFEATER with a published comparator for what the missing evidence looks like. Expression level is not dependency: where an XPO1 dependency in a leukaemia with a nucleoporin partner has actually been established, it took CRISPR DELETION of XPO1 in the fusion-positive line, pharmacological inhibition with a SINE, direct measurement of XPO1-fusion co-localisation at chromatin, downregulation of fusion target genes, and a patient-derived xenograft. Relatively high XPO1 in 20 profiled cases against a haematopoietic stem cell comparator is none of those things, and XPO1 is a common-essential gene whose transcript is high in many proliferating cells. This item cannot support the CRM1 mechanism being operative in patients. Charles Cano F et al. 2025
  8. E8strong confidence 4 of 5: Replicated in peer-reviewed work by more than one group.paper

    In murine bone marrow and in U937, Hoxa5 expression tracked CALM-AF10 and was upregulated relative to an OM-LZ deletion mutant, Hoxa9 expression was undetectable in CALM-AF10-transformed cells (not a primer artefact, since the same primers detected Hoxa9 in MLL-ENL-transduced cells), and Hoxa5 knockout attenuated the transformation capability of CALM-AF10 while having no effect on that of MLL-AF10.

    PMID 16921363DOI
    What this evidence cannot show

    Murine bone-marrow colony-replating plus U937. Read in the paper's full text on 2026-09-20 via the PMC HTML route (PMC4425349); none of this is in the abstract. Because Hoxa5 knockout is embryonic-lethal in C57BL/6, the knockout and wild-type littermates used were 129Sv/Ev-MF1, so the CALM-AF10 and MLL-AF10 arms of that specific comparison sit on a different genetic background from the rest of the paper. Colony numbers did not increase between the second and third plating rounds even for wild-type CALM-AF10, so the assay's dynamic range is narrow.

    Evidence that argues against this (1)
    • Three other systems report Hoxa9 elevation or Hoxa9-promoter H3K79 hypermethylation in CALM-AF10 - transgenic mice, a murine PICALM::MLLT10 leukaemia model, and patient T-ALL - so the undetectable-Hoxa9 result is specific to Okada et al.'s retrovirally transduced bone-marrow system and should not be generalised to the disease. Caudell D et al. 2007
Connected to