Ten affinity measurements for this contact and not one uses a fusion protein, full-length KMT2A or a nucleosome
also called the construct absence · everything is an isolated domain on naked DNA
Every published number for how tightly the fusion grips DNA was measured on a small piece of the protein, alone, against a short piece of naked DNA. Nobody has ever measured the actual fusion protein binding actual chromatin.
The same thing, for a scientist
Across the five primary papers that report a dissociation constant for the KMT2A CpG-binding region - PMIDs 11842107, 16990798, 18483194, 20010842 and 23990460 - all nine measurements use an isolated 57- to 98-residue domain, five of them tagless and four GST-tagged, against synthetic 12- to 52-bp duplex DNA. None uses full-length KMT2A, either taspase1 fragment, any KMT2A fusion protein, nucleosomal DNA, a reconstituted chromatin template or a nuclear extract. This is not a bounded PubMed absence - abstract-level queries for these methods return zero even for the papers that contain them (rule 80, demonstrated below) - it is an exhaustive reading of the construct table of every paper that reports a number. The consequence is that the whole affinity argument in this project, in either direction, concerns a molecular species that does not exist in a leukaemia cell.
The construct table, exhaustively
| paper | protein species used | DNA species used | fusion? | chromatin? |
|---|---|---|---|---|
| Birke 2002 | GST-MLL 1147-1244 | 52-bp synthetic duplex | no | no |
| Allen 2006 | MLL 1146-1214 (tagless), 1136-1208 for EMSA | 12-, 16-, 20-bp synthetic duplexes | no | no |
| Erfurth 2008 | MLL 1147-1203 (tagless) | five 12- to 16-bp synthetic duplexes | no | no |
| Cierpicki 2010 | MLL 1147-1203 (tagless) | 12-bp synthetic palindrome | no | no |
| Risner 2013 | GST-MLL 1147-1203 and three paralogue domains | one 12-bp synthetic duplex | no | no |
| Liang 2021 | chemically synthesised CXXC domain, untagged | 12-bp synthetic duplex and a real 187-bp Hoxa9 CpG island on long DNA | no | no - naked DNA, no nucleosome |
AMENDED 2026-09-21 within the same cycle. A sixth paper was found after this entry was first written - Liang 2021, PMID 34280370 - and it partly weakens one half of the claim. It uses a genuine 187-bp CpG island from the Hoxa9 gene on long DNA rather than a designed short duplex, so 'all on short synthetic oligonucleotides' is no longer exactly true. The claim that matters is untouched: not one row has a fusion protein, a full-length protein, a taspase1 fragment, a nucleosome or a chromatin template. The largest KMT2A species anyone has titrated against DNA is 98 residues of a 3,969-residue protein - 2.5%.
Nor is there a measurement of the multivalent species the project argues about. The fusion presents, on one polypeptide: three AT-hooks, MBM1 and MBM2, the LEDGF-contacting region around 140-160, the CXXC/MT domain, and whatever the partner contributes. Every published number is of one of those in isolation. The avidity of the assembled N-terminus for a surface presenting menin, LEDGF and a CpG island simultaneously has never been measured in any form, and it is the only quantity from which the project's central claims actually follow.
Why this is not written as a bounded PubMed absence, and the rule-80 demonstration that came out of trying
The obvious form for this claim is a counted absence. It cannot be written that way, and the attempt produced a clean demonstration of brief rule 80. Measured 2026-09-21, one tag per term, querytranslation read back for each 2:
| query | translation | records | true? |
|---|---|---|---|
MLL[tiab] AND CXXC[tiab] AND "surface plasmon resonance"[tiab] | all three [Title/Abstract] | 0 | FALSE - Birke 2002's SPR is in its own abstract |
MLL[tiab] AND CXXC[tiab] AND "isothermal titration calorimetry"[tiab] | all three [Title/Abstract] | 0 | FALSE - Allen, Erfurth and Cierpicki all did ITC |
KMT2A[tiab] AND "isothermal titration calorimetry"[tiab] | both [Title/Abstract] | 0 | FALSE, same reason |
(MLL[tiab] OR KMT2A[tiab]) AND CXXC[tiab] AND avidity[tiab] | correct | 0 | uninformative |
CXXC[tiab] AND "ionic strength"[tiab] | correct | 0 | uninformative |
Operand controls through the identical shape - "surface plasmon resonance"[tiab] alone returns 29,126, "isothermal titration calorimetry"[tiab] alone 8,992, CXXC[tiab] AND "surface plasmon resonance"[tiab] returns 6, CXXC[tiab] AND "isothermal titration calorimetry"[tiab] returns 5. The instrument works. The zeros are false. A biophysical method appears in a paper's Methods, not its abstract, and one of these zeros is false against a paper whose abstract literally contains the phrase - because the same abstract never says CXXC (see concept-mt-domain-and-cxxc-domain-are-two-names-for-one-region-and-the-literatures-do-not-intersect).
So this absence is bounded by five papers read in full, not by a query, and it should be quoted that way. Adding a sixth paper would change it; a sixth exists and could not be read (PMID 29276034).
The general lesson, which is the reusable part: a counted absence cannot be written about an experimental method at all in this literature. Methods live in Methods sections. Any claim of the form 'nobody has measured X by technique Y' must be bounded by papers read, and must name them.
| claim type | absence bounded by papers read in full, NOT by a counted query |
|---|---|
| largest kmt2a species ever titrated AA | 98 |
| kmt2a length AA | 3969 |
| fraction of protein PCT | 2.5 |
| never measured in any form | the avidity of the assembled KMT2A fusion N-terminus - three AT-hooks, MBM1, MBM2, the LEDGF-contacting region, and the CXXC/MT domain on one polypeptide - for a surface presenting menin, LEDGF and a CpG island simultaneously. Every central claim in this project follows from that quantity and nobody has approached it. |
| shown for fusion | NO - nobody has looked in any fusion. |
| n measurements | 10 |
| amendment 2026 09 21 | One of the six papers - Liang 2021 - uses a genuine 187-bp genomic CpG island rather than a designed short duplex, so the DNA half of this absence is weaker than first written. The protein half (no fusion, no full-length, no taspase1 fragment) and the chromatin half (no nucleosome anywhere) are unchanged. This entry was amended within the cycle that created it. |
| absence class | never-measured |
| absence class basis | No fusion protein, full-length KMT2A, taspase1 fragment or nucleosome has ever been titrated; the largest species measured is 98 of 3,969 residues. |
| absence class assigned | 2026-09-21 (cycle 47, king) |
papers read in full
[ "11842107", "16990798", "18483194", "20010842", "23990460", "34280370" ]
papers on topic that could not be read
[ "29276034" ]
construct table — the numbers behind the chart
[
{
"pmid": "11842107",
"protein": "GST-MLL 1147-1244",
"largest_kmt2a_fraction_pct": 2.5,
"dna": "52-bp synthetic duplex, 6 CpG",
"fusion_protein": false,
"chromatin": false
},
{
"pmid": "16990798",
"protein": "MLL 1146-1214 tagless (1136-1208 for EMSA)",
"largest_kmt2a_fraction_pct": 1.7,
"dna": "12-, 16-, 20-bp synthetic duplexes",
"fusion_protein": false,
"chromatin": false
},
{
"pmid": "18483194",
"protein": "MLL 1147-1203 tagless",
"largest_kmt2a_fraction_pct": 1.4,
"dna": "five 12- to 16-bp synthetic duplexes",
"fusion_protein": false,
"chromatin": false
},
{
"pmid": "20010842",
"protein": "MLL 1147-1203 tagless",
"largest_kmt2a_fraction_pct": 1.4,
"dna": "12-bp synthetic palindrome",
"fusion_protein": false,
"chromatin": false
},
{
"pmid": "23990460",
"protein": "GST-MLL 1147-1203 plus three paralogue CXXC domains and the MBD1 MBD",
"largest_kmt2a_fraction_pct": 1.4,
"dna": "one 12-bp synthetic duplex",
"fusion_protein": false,
"chromatin": false
},
{
"pmid": "34280370",
"protein": "chemically synthesised CXXC domain, untagged; span in an unretrieved supplementary table",
"largest_kmt2a_fraction_pct": 1.4,
"dna": "12-bp synthetic palindrome for EMSA; a real 187-bp Hoxa9 CpG island in a 236-bp hairpin for magnetic tweezers",
"fusion_protein": false,
"chromatin": false
}
]false zero queries — the numbers behind the chart
[
{
"query": "MLL[tiab] AND CXXC[tiab] AND \"surface plasmon resonance\"[tiab]",
"querytranslation": "\"MLL\"[Title/Abstract] AND \"CXXC\"[Title/Abstract] AND \"surface plasmon resonance\"[Title/Abstract]",
"records": 0,
"false_because": "Birke 2002's abstract contains 'surface plasmon resonance' but never contains 'CXXC'",
"date": "2026-09-21"
},
{
"query": "MLL[tiab] AND CXXC[tiab] AND \"isothermal titration calorimetry\"[tiab]",
"querytranslation": "\"MLL\"[Title/Abstract] AND \"CXXC\"[Title/Abstract] AND \"isothermal titration calorimetry\"[Title/Abstract]",
"records": 0,
"false_because": "three papers in the set report ITC and none names the method in its abstract (rule 80)",
"date": "2026-09-21"
},
{
"query": "KMT2A[tiab] AND \"isothermal titration calorimetry\"[tiab]",
"querytranslation": "\"KMT2A\"[Title/Abstract] AND \"isothermal titration calorimetry\"[Title/Abstract]",
"records": 0,
"false_because": "same",
"date": "2026-09-21"
}
]operand controls — the numbers behind the chart
[
{
"query": "\"surface plasmon resonance\"[tiab]",
"records": 29126
},
{
"query": "\"isothermal titration calorimetry\"[tiab]",
"records": 8992
},
{
"query": "CXXC[tiab] AND \"surface plasmon resonance\"[tiab]",
"records": 6
},
{
"query": "CXXC[tiab] AND \"isothermal titration calorimetry\"[tiab]",
"records": 5
},
{
"note": "the instrument can return a one through the identical query shape, so the zeros above are false negatives and not real absences"
}
]- If every affinity in the literature is for 2.5% of the protein, then a drug-discovery assay built on that construct is measuring engagement with a fragment, and the correlation between biochemical potency and cellular displacement would be unknown at the start of any campaign.
- An absence of this shape is also an argument for humility about the correction this beat is making: the ladder resolves a discrepancy between two measurements of a species that does not exist, and a future measurement on the real species could sit outside the whole ladder in either direction.
- Could the avidity question be approached at all? A reconstituted CpG-island nucleosome array, immobilised menin and LEDGF, and a purified fusion N-terminal fragment is a hard but not impossible single-molecule or BLI experiment, and nobody has proposed it in print.
- What is the DNA affinity of the taspase1 N320 fragment, the species that actually carries the CXXC domain in the cell? UniProt models N320 and C180 as separate entities and no affinity exists for either.
- Does the AT-hook contribution to DNA binding have a published dissociation constant, and if so under what conditions? If not, the same absence applies to the other half of the DNA-binding argument.
- E1strong confidence 4 of 5: Replicated in peer-reviewed work by more than one group.paper
All ten published dissociation constants for the KMT2A CpG-binding region use an isolated 57- to 98-residue domain against naked DNA - nine of them synthetic 12- to 52-bp duplexes and one a 187-bp genomic CpG island on long DNA; none uses full-length KMT2A, a taspase1 cleavage fragment, any KMT2A fusion protein, a nucleosome, a reconstituted chromatin template or a nuclear extract.
The MT domain of the proto-oncoprotein MLL binds to CpG-containing DNA and discriminates against methylation. · Birke M, Schreiner S, García-Cuéllar MP, et al. · Nucleic Acids Res 2002What this evidence cannot show
Established by reading the Methods of the six papers that report a number (PMIDs 11842107, 16990798, 18483194, 20010842, 23990460, 34280370), all six read in full this cycle via the PMC HTML route. This is an absence bounded by six papers, not by a query - abstract-level queries for these methods return false zeros, demonstrated in the second section. A seventh relevant paper, PMID 29276034, could not be retrieved by any of the five documented routes and might contain a construct or ligand not in this table. The sixth paper was itself missed by this beat's queries and found only by reading an existing entry's counter-evidence, so the bound should be read as six papers FOUND rather than six papers that exist.
Evidence that argues against this (2)
- Gel-shift and ChIP experiments on larger KMT2A species do exist - Ayton 2004 uses larger MLL fragments in EMSA and cells - so it is not true that nobody has studied DNA binding by anything bigger than a domain. What is absent is a quantitative dissociation constant, which is a narrower and more defensible claim than 'nobody has looked at the fusion'. Ayton PM et al. 2004
- Liang 2021 uses a real 187-bp Hoxa9 CpG island on long DNA and profiles binding along it at 1-bp resolution, which is a substantially more physiological ligand than the rest of the table and weakens the 'all short synthetic duplexes' half of this claim. The protein side is still an isolated synthesised domain and the DNA is still naked, so the claim about the protein species and about chromatin stands. Liang L et al. 2021
- E2established confidence 5 of 5: Multiple independent lines of evidence; treated as settled in the field.database
Abstract-level PubMed queries for the biophysical methods used in this literature return well-formed zeros that are false: MLL[tiab] AND CXXC[tiab] AND "surface plasmon resonance"[tiab] returns 0 although Birke 2002's abstract contains the phrase, and MLL[tiab] AND CXXC[tiab] AND "isothermal titration calorimetry"[tiab] returns 0 although three papers in the set report ITC, while the operand controls CXXC[tiab] AND "surface plasmon resonance"[tiab] and CXXC[tiab] AND "isothermal titration calorimetry"[tiab] return 6 and 5 respectively.
PubMed E-utilities esearch, live queries with querytranslation recorded and operand controls run through the identical query shape · NCBI Entrez 2026What this evidence cannot show
Counts from one index on 2026-09-21 and they drift. Every query written one tag per term per rule 94 with the querytranslation string read back; none shows an [All Fields] fallback. The two false zeros have two distinct causes - the SPR one is a naming problem (Birke's abstract never says CXXC) and the ITC one is a rule-80 problem (the method is in Methods) - and they should not be conflated.
Evidence that argues against this (1)
- Europe PMC full-text search would reach these papers where esearch cannot, so the correct conclusion is not that method-level absences are unwriteable but that they are unwriteable with an abstract index. A beat with a full-text budget could bound this claim properly. Erfurth FE et al. 2008
one of only two entries in 39 bounded by reading the primary source in full - the only class safe by construction
Every rung of the ladder is an isolated domain on naked DNA; none uses a fusion protein, full-length KMT2A, chromatin or a nucleosome.
The same retrieval failure is why a method-level absence cannot be written as a counted query in this literature.
The free nuclear concentration term is unmeasured for the same reason the affinity is measured only on fragments.
- related to →KMT2A / MLL1 — the gene that breaks
The largest KMT2A species ever titrated against DNA is 98 of 3,969 residues.
- related to →KMT2A::AFF1 (MLL-AF4), t(4;11)(q21;q23)
No affinity measurement uses a fusion protein, so the whole affinity argument concerns a species that does not exist in a leukaemia cell.
The matched titration is a first step out of the absence, though it too uses isolated domains.