Three measurements the export axis needs and nobody has published: XPO1 occupancy in KMT2A-r, H3K79 under an XPO1 inhibitor, and CRM1 occupancy under a menin inhibitor
also called export axis bounded absences · missing XPO1 chromatin measurements
Three specific experiments would settle most of the open arguments about XPO1 in leukaemia. We searched for each one properly, checked that the search could find things when they exist, and found none of them published.
The same thing, for a scientist
Three counted absences, each with the exact query, the date, the tool, the name list and an operand control run through the identical query shape (rule 67). All run 2026-09-20 on PubMed E-utilities esearch, with `querytranslation` read back. **Absence 1 - no H3K79 measurement under any XPO1 inhibitor, anywhere.** | query | records | |---|---:| | `(selinexor OR "XPO1 inhibitor" OR "XPO1 inhibition" OR eltanexor OR KPT-330) AND (H3K79 OR "H3-K79" OR "lysine 79")` | **0** | | operand: the SINE name list alone | **829** | | operand: `(H3K79 OR "H3-K79" OR "lysine 79")` alone | **654** | | **control through the identical shape:** the SINE list `AND (H3K4me3 OR H3K27ac OR "chromatin immunoprecipitation")` | **3** | The control returns ones, so the shape composes and the zero is a measurement. **This is the experiment `mechanism-calm-af10-crm1-nuclear-export-requirement` names as the one that would separate the export model from the docking model, and it has never been done in any system.** **Absence 2 - no XPO1 or CRM1 occupancy measurement in a KMT2A-rearranged cell.** | query | records | |---|---:| | `(selinexor OR eltanexor OR "XPO1 inhibitor") AND (KMT2A OR MLL) AND (ChIP OR occupancy OR CUT&RUN)` | **0** | | operand: `(selinexor OR eltanexor OR "XPO1 inhibitor") AND (KMT2A OR MLL)` | **4** | | operand: `(selinexor OR eltanexor OR "XPO1 inhibitor") AND (ChIP OR occupancy)` | **6** | | `(CRM1[tiab] OR XPO1[tiab]) AND (ChIP[tiab] OR "CUT&RUN"[tiab] OR occupancy[tiab] OR "CUT and RUN")` | **16**, none KMT2A-rearranged on title check | Both operands are small but non-zero, and the 16-record XPO1-occupancy literature exists and was title-checked: it is nucleoporin fusions, NPM1c, UBTF-TD and unrelated biology. **No XPO1 ChIP in a KMT2A-rearranged cell was found.** **Absence 3 - no CRM1 occupancy measurement under a menin inhibitor.** | query | records | |---|---:| | `(CRM1 OR XPO1) AND (menin inhibitor OR menin inhibitors OR menin inhibition OR revumenib OR ziftomenib OR VTP-50469 OR MI-503 OR MI-2-2)` | **15** | | `menin[tiab] AND XPO1[tiab]` (the bare-operand check rule 72 demands) | **16** | Both counts are healthy, so this is **not** a rule-72 name-list failure. **Every one of the 15 was title-checked**: nine are NPM1-mutated-AML reviews, one is a venetoclax-resistance review, one is a p27Kip1 review, one is the C-body commentary, one is the HOXA9-regulators review, one is the UBTF-TD paper and one is the ziftomenib-plus-selinexor preprint. **None reports CRM1 or XPO1 chromatin occupancy measured under a menin inhibitor.** The preprint (PMID 41959134) reports the converse - menin's chromatin binding measured under an XPO1 inhibitor - which is the mirror experiment and not this one.
Rule 80: what these zeros are bounded by
All three are bounded by titles and abstracts, not by papers. Supporting figures were not searched, and in this field the measurements that matter are almost always supporting panels - this graph's own rule 80 was written after an H3K79me2 ChIP was found in Figure 6A of a paper whose abstract never names the mark.
The specific risk for each:
- Absence 1 is the most likely of the three to be wrong this way. An H3K79me2 blot under leptomycin B could easily sit in a supplementary panel of the CALM-AF10 NES paper (PMID 23487024) or the CRM1-AF10 chimera paper (PMID 32733011), both of which measure H3K79 and both of which use CRM1 inhibitors. Neither was opened in this beat. That is the single highest-value full-text check left undone here, and it is named in the memo.
- Absence 2 is less likely to be hiding, because an XPO1 ChIP in a KMT2A-rearranged line would be a headline result, not a control panel.
- Absence 3 is a measurement nobody has a reason to make unless they already believe XPO1 and menin are on one assembly, which is a 2026 idea.
Stated as the rule requires: zero in titles and abstracts; supporting figures not searched.
Why absence 3 is now the most interesting of the three
Until 2026 there was no reason to measure CRM1 occupancy under a menin inhibitor - the two systems were thought to be independent addressing mechanisms. Two 2026 sources changed that:
- a preprint reporting XPO1 stabilises menin's binding to chromatin and its interactions with KMT2A and KMT2A::MLLT3 (
mechanism-xpo1-stabilises-menin-chromatin-binding); - a commentary reporting menin and XPO1 co-partitioning in the same NPM1c condensate (
mechanism-nes-crm1-chromatin-addressingE5).
If either is right, the reciprocal experiment - does menin inhibition change chromatin-bound XPO1? - becomes a direct test of whether the two are one assembly or two. A symmetric result (each inhibitor displaces the other's protein) argues one assembly; an asymmetric result (XPO1 inhibition moves menin, menin inhibition does not move XPO1) argues that XPO1 is upstream, which is what the preprint implies.
This graph already holds a menin-inhibitor ChIP infrastructure in its LEDGF and menin layers, so the experiment is cheap where those datasets exist.
| script | manual; PubMed E-utilities esearch via scratchpad helper q61_es.py, 2026-09-20 |
|---|---|
| status | run |
| date | 2026-09-20 |
| tool | NCBI E-utilities esearch (db=pubmed), querytranslation read back on every query |
| result summary | Three named measurements are absent from the titles and abstracts of the published literature as of 2026-09-20, each with operand controls and, for the first, a positive control through the identical query shape. |
absences — the numbers behind the chart
[
{
"measurement": "H3K79 methylation under any XPO1 inhibitor, any system",
"query": "(selinexor OR \"XPO1 inhibitor\" OR \"XPO1 inhibition\" OR eltanexor OR KPT-330) AND (H3K79 OR \"H3-K79\" OR \"lysine 79\")",
"records": 0,
"operand_a": 829,
"operand_b": 654,
"positive_control_same_shape": "SINE list AND (H3K4me3 OR H3K27ac OR \"chromatin immunoprecipitation\") = 3",
"bound": "titles and abstracts; supporting figures not searched",
"unchecked_candidates": [
"PMID 23487024 (Conway, CALM NES, measures H3K79 and uses leptomycin B)",
"PMID 32733011 (Aumann, CRM1-AF10 chimera, measures HOXA)"
]
},
{
"measurement": "XPO1 or CRM1 chromatin occupancy in any KMT2A-rearranged cell",
"query": "(selinexor OR eltanexor OR \"XPO1 inhibitor\") AND (KMT2A OR MLL) AND (ChIP OR occupancy OR CUT&RUN)",
"records": 0,
"operand_a": 4,
"operand_b": 6,
"broader_occupancy_literature": 16,
"bound": "titles and abstracts; 16 occupancy records title-checked, none KMT2A-rearranged"
},
{
"measurement": "CRM1/XPO1 chromatin occupancy under a menin inhibitor",
"query": "(CRM1 OR XPO1) AND (menin inhibitor OR menin inhibitors OR menin inhibition OR revumenib OR ziftomenib OR VTP-50469 OR MI-503 OR MI-2-2)",
"records": 15,
"on_topic_after_title_check": 0,
"bare_operand_check": "menin[tiab] AND XPO1[tiab] = 16",
"bound": "titles and abstracts; all 15 title-checked"
}
]- A counted zero on a measurement is the weakest kind of absence this graph records, because the measurement can exist in a figure the index cannot see. All three are labelled accordingly and none should be quoted as 'nobody has looked'.
- Two of the three absences would be closed by one experiment each, which means they are cheap for a competitor to close too.
- Does an H3K79me2 measurement under leptomycin B sit in a supplementary figure of PMID 23487024 or PMID 32733011? Rule 80 says the abstract-level zero cannot answer it and both papers are open candidates.
- Is the menin-XPO1 relationship symmetric? Only one direction has been measured, and it is in a preprint.
- Does any deposited ChIP dataset contain an XPO1 antibody in a KMT2A-rearranged line? A GEO elink sweep from the four KMT2A-plus-XPO1 papers would answer it and was not run in this beat.
- 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, while the SINE name list alone returned 829, the H3K79 name list alone returned 654, and the identical query shape with (H3K4me3 OR H3K27ac OR 'chromatin immunoprecipitation') in place of the H3K79 clause returned 3.
PubMed E-utilities esearch, counted-absence query set 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 only; supporting figures not searched (rule 80). The name list is stated in the claim and is the instrument here (rules 70-72). The positive control ran through the identical query shape and returned 3, so the shape composes. The zero says no paper ANNOUNCES an H3K79 measurement under an XPO1 inhibitor; it does not say no such blot exists in a supplementary figure, and two specific candidate papers are named in the memo as unchecked.
Evidence that argues against this (1)
- A counted zero on a MEASUREMENT is the weakest class of absence this graph records. PMID 23487024 measures focal HOXA H3K79 hypermethylation AND uses leptomycin B in the same paper; if an H3K79 panel under the inhibitor exists there it would be invisible to every query run here. The zero is bounded by the index, not by the papers, and two named candidates remain unopened. The founding platform paper cited here is the second such candidate, since it too combines a CRM1 inhibitor with chromatin readouts. Oka M et al. 2016
- E2strong 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 eltanexor OR 'XPO1 inhibitor') AND (KMT2A OR MLL) AND (ChIP OR occupancy OR CUT&RUN) returned 0, with the two operand pairs returning 4 and 6 respectively; and a broader query for (CRM1[tiab] OR XPO1[tiab]) AND (ChIP[tiab] OR 'CUT&RUN'[tiab] OR occupancy[tiab] OR 'CUT and RUN') returned 16 records, none of which is in a KMT2A-rearranged system on title check.
PubMed E-utilities esearch, counted-absence query set for XPO1 chromatin occupancy in KMT2A-rearranged leukaemia · Oncofusion Knowledge Graph project (Q61-export-axis) · NCBI E-utilities 2026What this evidence cannot show
Titles and abstracts only. The 16 records were title-checked but not read; a KMT2A-rearranged comparator arm inside one of them would be invisible to a title check. The ampersand in 'CUT&RUN' is a query-parser hazard and the broader query was run with both the ampersand form and the spelled-out form to cover it.
Evidence that argues against this (1)
- PUBLISHED CONFLICTING RESULT about what the absence covers. A genome-wide XPO1 chromatin occupancy measurement DOES exist and post-dates the framing this entry inherits: Chen et al. show that XPO1 and p300 co-occupy hundreds of chromatin loci in human cells, and that degrading XPO1 with C646 or S109 diminishes the chromatin occupancy of both. The absence recorded here is therefore not 'nobody has mapped chromatin-bound XPO1' - it is the much narrower 'nobody has done it in a KMT2A-rearranged cell'. A chemical-biology paper framed on p300 and molecular glues is unreachable from any query built from KMT2A, MLL or a leukaemia clause (brief rule 77), so this bounded zero is accurate and materially understates what has been measured. Chen YF et al. 2024
- E3strong confidence 4 of 5: Replicated in peer-reviewed work by more than one group.database
A PubMed esearch on 2026-09-20 for (CRM1 OR XPO1) AND (menin inhibitor OR menin inhibitors OR menin inhibition OR revumenib OR ziftomenib OR VTP-50469 OR MI-503 OR MI-2-2) returned 15 records and the bare-operand check menin[tiab] AND XPO1[tiab] returned 16; all 15 were title-checked and none reports CRM1 or XPO1 chromatin occupancy measured under a menin inhibitor.
PubMed E-utilities esearch, counted-absence query set for CRM1 occupancy under a menin inhibitor · Oncofusion Knowledge Graph project (Q61-export-axis) · NCBI E-utilities 2026What this evidence cannot show
Rule 72 satisfied: the bare-operand check (menin AND XPO1 = 16) returns a healthy number, so the drug-name list is not the failure point. Titles and abstracts only; a ChIP panel inside one of the fifteen reviews is impossible by construction (reviews do not contain primary ChIP), but a panel inside the UBTF-TD paper or the ziftomenib preprint is possible and neither full text was opened.
Evidence that argues against this (1)
- METHODOLOGICAL DEFEATER on what a published CRM1 occupancy measurement would have to survive before it could be reported at all, which changes how this absence should be read. Jain et al. found that roughly 3,000 chromosomal loci, mostly ACTIVE PROMOTERS, are prone to non-specific enrichment in ChIP and appear as 'Phantom Peaks', and that these were obtained with several polyclonal AND monoclonal antibodies yet were unchanged in mutant embryos not expressing the target. Their prescription is that such profiles must be validated with chromatin from cells in which the protein has been depleted. An XPO1/CRM1 ChIP under a menin inhibitor would land on the HOXA/HOXB clusters - the most highly transcribed loci in the cell being studied - so a group obtaining such a signal without a depletion control has a strong reason not to publish it. Part of this zero may be a publication filter on an artefact-prone assay rather than an absence of attempts. Jain D et al. 2015
- E4weak confidence 2 of 5: Preliminary, small, indirect, or not yet peer reviewed.preprint
A 2026 preprint reports the MIRROR of absence 3 - menin's chromatin binding measured under XPO1 inhibition - stating that XPO1 stabilises menin's binding to chromatin and its interactions with KMT2A and KMT2A/MLLT3, disrupted by XPO1 inhibitor treatment.
Combined Menin and XPO1 inhibition drive synergistic antileukemic activity in KMT2Ar and NPM1-m AML. · Uddin MH, Dhiman S, Han Y, et al. · bioRxiv 2026What this evidence cannot show
PREPRINT, abstract-level. Recorded here to make the asymmetry explicit: one direction of the experiment now has a (weak, unreviewed) result and the other direction has none. Full entry at `mechanism-xpo1-stabilises-menin-chromatin-binding`.
Evidence that argues against this (1)
- PUBLISHED CONFLICTING RESULT offering a non-specific explanation for the preprint's mirror result. If XPO1 stabilises menin on chromatin, the finding is specific to the menin-KMT2A axis. But Chen et al. show that degrading XPO1 reduces the chromatin occupancy of p300 across hundreds of co-occupied loci - i.e. loss of XPO1 detaches a general chromatin-associated cofactor on a wide scale. Menin coming off chromatin under an XPO1 inhibitor is equally consistent with that general effect, and the preprint's abstract does not report a non-menin chromatin factor as a specificity control. The asymmetry this evidence item exists to record is real, but the direction that has a result is the one with the weaker specificity claim. Chen YF et al. 2024
three counted absences the hypothesis depends on
no H3K79 measurement under any XPO1 inhibitor exists in titles or abstracts
the counted absence is not 'nobody has mapped chromatin-bound XPO1' but 'nobody has done it in a KMT2A-rearranged cell': a genome-wide XPO1 occupancy map exists in a chemical-biology paper framed on p300 and is unreachable from any KMT2A-scoped query (brief rule 77)
no XPO1 localisation measurement of any kind exists in a KMT2A-rearranged cell