Export or docking? A locus-resolved H3K79me2 ChIP under an XPO1 inhibitor separates the two CALM-AF10 models, and it has never been run
also called H3K79me2 under selinexor · CALM-AF10 export-versus-docking discriminator
Two explanations for how CALM-AF10 causes leukaemia make opposite predictions about what happens to a specific chemical mark on DNA-packaging proteins when you block nuclear export. Measuring that mark gene by gene, with and without the drug, tells you which explanation is right.
The same thing, for a scientist
**The question this settles.** `mechanism-calm-af10-crm1-nuclear-export-requirement` records an unresolved disagreement with real therapeutic consequences: - **Export model.** The CALM NES exports the fusion; blocking CRM1 traps it in the nucleus and that is protective. Consistent with the reported leptomycin B sensitivity of CALM-AF10 cells. - **Docking model.** The NES docks the fusion onto chromatin-prebound CRM1; blocking CRM1 removes the chromatin anchor, which is also protective, but by a different route. Both predict that the cells die. **They predict opposite things about the mark.** CALM-AF10's signature is **global H3K79 hypomethylation with focal H3K79 HYPERmethylation at HOXA**. Under the docking model, removing the fusion from HOXA chromatin should **collapse the focal HOXA H3K79me2 peak** while leaving the global deficit unchanged or worsening it. Under the export model, a more nuclear fusion has no reason to lose its HOXA peak - the Okada reading would even predict it **increases**. **The absence is counted.** `(selinexor OR "XPO1 inhibitor" OR "XPO1 inhibition" OR eltanexor OR KPT-330) AND (H3K79 OR "H3-K79" OR "lysine 79")` returns **0** on PubMed, with operands of 829 and 654 and a positive control through the identical shape returning 3 [E1]. **Bounded by titles and abstracts; supporting figures not searched** - and the memo names two specific papers that could contain the panel. **The design.** CALM-AF10-positive cells (U937, which carries the fusion, plus a murine CALM-AF10 bone-marrow model for the in vivo arm) treated with selinexor or leptomycin B at a displacement-not-degradation exposure, then: - **H3K79me2 ChIP-seq or CUT&RUN**, spike-in normalised, reported **both** as the HOXA cluster profile and as a genome-wide global level - the two halves of the CALM-AF10 signature move independently and a single number hides that; - **CALM-AF10 occupancy** at the same loci in the same cells; - **DOT1L occupancy**, because the Okada model routes through DOT1L; - **HOXA5/HOXA9/HOXA10 expression**. **The control that makes it interpretable:** an NES-mutant CALM-AF10 in the same system. If the drug and the mutant produce the same H3K79me2 pattern, the drug is acting through the NES-CRM1 contact; if they differ, it is acting through something else, which for a common-essential target is entirely possible.
Why nobody has done it, and why that is a reason to
The two literatures are separate. The H3K79 measurements in CALM-AF10 come from the DOT1L field (Okada 2006, Conway 2013) and the CRM1 inhibitor experiments come from the transport field. Both use the same cells. The counted zero is not a hard problem going unsolved, it is two communities not running each other's assay, which is the cheapest kind of gap to close.
And the consequence is not academic. mechanism-calm-af10-crm1-nuclear-export-requirement carries the bite-back that 'if the docking model is right, an export inhibitor might stabilise the fusion on chromatin rather than displace it, and the intervention could point the wrong way.' PICALM::MLLT10 patients are excluded from menin and DOT1L trials (trial-menin-and-dot1l-eligibility-excludes-picalm-mllt10), so XPO1 inhibition is one of the few mechanistic options on the table for them, and the field does not know which direction it pushes.
| design | CALM-AF10-positive cells treated with selinexor or leptomycin B at a displacement-not-degradation exposure, with spike-in-normalised H3K79me2 ChIP-seq or CUT&RUN reported as BOTH the HOXA cluster profile and the genome-wide global level, alongside CALM-AF10 occupancy, DOT1L occupancy and HOXA5/9/10 expression |
|---|---|
| readout | focal HOXA H3K79me2 peak height versus global H3K79me2 level; fusion and DOT1L occupancy at the same loci |
| controls | an NES-mutant CALM-AF10 in the same system, which should phenocopy the drug if the drug acts through the NES-CRM1 contact; vehicle; a CALM-AF10-negative line; XPO1 western in the same lysates to exclude degradation |
| discriminates | docking model predicts collapse of the focal HOXA H3K79me2 peak with the global deficit unchanged; export model predicts the focal peak is preserved or increased |
| cost tier | med |
| timeline | 8-12 weeks |
| clinical stake | PICALM::MLLT10 patients are excluded from menin and DOT1L trials, so XPO1 inhibition is one of the few mechanistic options available and the field does not know which direction it pushes |
| check first | open PMID 23487024 and PMID 32733011 in full text before running anything - rule 80 says the counted zero does not bound supporting figures |
cell lines
[ "U937 (carries CALM-AF10)", "murine CALM-AF10 bone-marrow transplantation model for the in vivo arm" ]
- Whichever model wins, XPO1 remains common-essential, so the therapeutic conclusion is constrained by toxicity regardless of which mechanism is correct.
- If the drug and the NES mutant give different H3K79me2 patterns, the experiment has shown that the inhibitor does not act through the NES contact - which invalidates the pharmacological route to this biology without telling you what does work.
- A global H3K79me2 change under an XPO1 inhibitor could be a cell-cycle or differentiation consequence rather than a direct effect, and spike-in normalisation controls the technical half of that but not the biological half.
- Does the panel already exist in a supplementary figure? Two candidate papers are named and neither was opened in this beat.
- Is U937 an adequate model? It carries CALM-AF10 endogenously but is a decades-old monocytic line and the graph already records that the CALM-AF10 model space is thin.
- Should the experiment use the NES-mutant fusion as the primary comparator rather than the drug, given that a common-essential target's inhibitor has hundreds of off-pathway effects? That version is cleaner biology and answers a slightly different question.
- E1strong confidence 4 of 5: Replicated in peer-reviewed work by more than one group.database
A PubMed esearch on 2026-09-20 for (selinexor OR 'XPO1 inhibitor' OR 'XPO1 inhibition' OR eltanexor OR KPT-330) AND (H3K79 OR 'H3-K79' OR 'lysine 79') returned 0 records, with the SINE name list alone returning 829, the H3K79 name list alone returning 654, and the identical query shape substituting (H3K4me3 OR H3K27ac OR 'chromatin immunoprecipitation') returning 3.
PubMed E-utilities esearch, counted-absence query for XPO1 inhibition and H3K79 · Oncofusion Knowledge Graph project (Q61-export-axis) · NCBI E-utilities 2026What this evidence cannot show
Bounded by titles and abstracts; supporting figures not searched (rule 80). Two candidate papers that measure H3K79 AND use a CRM1 inhibitor - PMID 23487024 and PMID 32733011 - were not opened in this beat and could contain the panel.
Evidence that argues against this (1)
- The experiment assumes the XPO1 inhibitor acts on this biology through the NES-CRM1 contact. XPO1 inhibition also induces P2RY2-AKT signalling and hundreds of other cargo-level changes, any of which could move H3K79me2 indirectly through differentiation or cell-cycle change. That is why the NES-mutant comparator, not the drug, is the interpretable arm. Lin KH et al. 2022
- E2strong confidence 4 of 5: Replicated in peer-reviewed work by more than one group.paper
CALM residues 544-553 form a CRM1-recognised NES necessary for CALM-AF10 transformation; the same element controls HOXA overexpression and focal H3K79 hypermethylation; and CALM-AF10 leukaemia cells are selectively sensitive to leptomycin B.
A CALM-derived nuclear export signal is essential for CALM-AF10-mediated leukemogenesis. · Conway AE, et al. · Blood 2013What this evidence cannot show
Murine bone marrow plus HEK293 imaging; leptomycin B is a tool compound, not a clinical agent. Recorded in full in `mechanism-calm-af10-crm1-nuclear-export-requirement`. Cited here because it establishes that the NES controls the focal H3K79 signature - which is precisely what makes the mark the right readout for the discriminating experiment. This paper is also a candidate location for an unpublished-in-abstract H3K79 measurement under leptomycin B.
Evidence that argues against this (1)
- PUBLISHED CONFLICTING RESULT about what leptomycin B sensitivity can mean. Leptomycin B is elactocin, and it has been given to humans: in a phase 1 trial across five schedules - 1 h infusion 3-weekly, 24 h infusion 3-weekly, 1 h daily x 5, 1 h weekly and continuous 5-day infusion - the dose-limiting toxicity was profound anorexia and malaise on EVERY schedule, no partial or complete responses were seen, and the authors recommended that no further trials be performed. Global XPO1 blockade is therefore known to be broadly and non-selectively damaging in vivo. 'CALM-AF10 leukaemia cells are selectively sensitive to leptomycin B' is a relative-potency statement between cultured lines under an agent whose measured in-human effect is generalised toxicity, and it cannot separate loss of a docking interaction from collapse of the whole nucleocytoplasmic economy. Newlands ES et al. 1996
- E3strong confidence 4 of 5: Replicated in peer-reviewed work by more than one group.paper
Fusing the CRM1 nuclear export receptor to AF10 in place of CALM causes leukaemia with transcriptional activation of HOXA genes, and mutating the CRM1 motifs that contact NUP214 abolishes self-renewal and prevents leukaemia.
Fusion of the CRM1 nuclear export receptor to AF10 causes leukemia and transcriptional activation of HOXA genes. · Aumann WK, et al. · Leukemia 2021What this evidence cannot show
Murine fetal-liver haematopoietic cells and transplantation; a chimera argues about sufficiency, not about what the natural fusion does. Recorded in full in `mechanism-calm-af10-crm1-nuclear-export-requirement`. Cited here as the second candidate location for an unchecked H3K79 panel and as the strongest existing argument for the docking model.
Evidence that argues against this (1)
- PUBLISHED CONFLICTING RESULT about what the chimera argues. Conway et al. report that CRM1 itself localises to HOXA loci and recruits CALM-AF10 there, and that genetic or pharmacological disruption of the CALM-CRM1 interaction prevents CALM-AF10 enrichment at HOXA chromatin with immediate loss of transcription - they describe CRM1's chromatin-binding ability as a NOVEL function of the receptor. If CRM1 is itself a chromatin-binding factor at HOXA, then fusing CRM1 to AF10 supplies a chromatin-binding moiety directly, and the chimera's leukaemogenicity is equally well explained by CRM1 acting as the chromatin anchor IN the fusion as by CRM1 being a pre-existing address the fusion docks onto. The chimera cannot discriminate the two, which is the same distinction this experiment is designed to make. Conway AE et al. 2015
separates the export model from the docking model
the focal HOXA peak is the discriminating readout
- related to →PICALM::MLLT10 (CALM-AF10), t(10;11)(p13;q14) - the DOT1L-dependent HOXA fusion that contains no KMT2A
the genotype with the clinical stake, excluded from menin and DOT1L trials
DOT1L occupancy is a parallel readout because the dissenting model routes through it
THE MARK THIS EXPERIMENT READS OUT, and the entry to consult for what a di-methyl measurement can and cannot show: H3K79me2 is the state the leukaemia runs on, and the one menin governs only partially.
THE SECOND, AND THE REASON THE READOUT MUST BE LOCUS-RESOLVED: the same fusion STRIPS H3K79 methylation genome-wide while adding it at HOXA, so a bulk measurement averages two opposite effects and can move either way for the wrong reason.
if CRM1 is itself a chromatin-binding factor at HOXA (PMID 25027513), a CRM1-AF10 chimera supplies the chromatin anchor directly and cannot discriminate 'CRM1 as pre-existing address' from 'CRM1 as the chromatin-binding moiety inside the fusion' - the same distinction this experiment exists to make
THE DESIGNED-AND-UNRUN EXPERIMENT THAT WOULD SUPPLY THIS ENTRY'S EPIGENETIC CONTENT: spike-in-normalised locus-resolved H3K79me2 under selinexor, which in CALM-AF10 discriminates the export from the docking model (opposite predictions for the focal HOXA peak) and in a KMT2A-rearranged line would answer whether blocking export moves a mark at all.