Oncofusion

start / graph / ana-kmt2a-lost-module-ledger

Analysisstrong confidence 4 of 5: Replicated in peer-reviewed work by more than one group.Q-subtraction-kmt2a-counter

Ledger: every module the KMT2A half loses, with coordinates, how each was confirmed, and whether its loss has been shown to matter

also called KMT2A subtraction ledger · lost module table · what the fusion loses from its own half

A single table of everything the broken half of KMT2A throws away, where each piece sits on the protein, and whether anyone has shown that losing it changes anything. Re-checked in September 2026: far more of these losses matter than the table first said, and most of what they do is stop the protein from being cleared away on schedule.

The same thing, for a scientist

Fifteen modules or sites C-terminal to the KMT2A breakpoint cluster region are absent from every der(11) KMT2A::partner fusion protein. This ledger records for each: residue coordinates, the source that establishes the coordinates, the binding partner or activity lost, and a loss class. **Corrected in cycle 31 by a counter-evidence pass.** The previous headline - that 9-10 of 15 lost modules had no demonstrated functional consequence - is **withdrawn**: it was wrong by roughly a factor of three, and wrong in the direction absence claims in this project usually fail. Five rows were overturned by literature that already existed when the ledger was written: the two taspase1 sites [E6], FYRN [E8], PHD1 and PHD4 [E8], FYRC [E9] and the MOF contact [E7]. The residual no-consequence bucket is **three rows** - the bromodomain, the pre-PHD C2HC finger and the HCF-binding motif - each searched for specifically this cycle with the searches recorded. The overturns also point one way, which is the more important result: **every lost module whose loss has a measured consequence makes the protein more abundant and more chromatin-bound, not more active.** The taspase1 sites, the FYRN degron and the PHD2 E3 ligase are three independent degradation determinants, all C-terminal to the break, all absent from the fusion [E6, E8, E12]. What the KMT2A half principally loses is control of its own abundance. The second axis - whether the lost function is supplied in trans by the intact allele - now rests on a genetic requirement rather than a recruitment model [E13].

Go deeper
CORRECTION (cycle 31): what was withdrawn from this ledger, and why3 deeper

This entry previously stated that nine to ten of fifteen lost modules had 'no demonstrated functional consequence whatsoever', and presented that as the beat's headline number. That claim is withdrawn. It did not survive a search aimed specifically at finding a phenotype for each row. Five rows changed class, on papers all published before this ledger was written:

RowWasNowSource
Taspase1 CS1 + CS2unmeasureddemonstrated, in trans - cleavage primes MLL1 for degradation; the fusion lacks the sites and is correspondingly stable and chromatin-heavy; stabilising the wild-type copy displaces MLL-AFF1/AFF4/ELL2 from chromatin in SEM cells and prolongs survival in MLL-AF9 mice6
FYRNunmeasureddemonstrated - FYRN is the target of the mechanism that destroys an unpartnered MLL N-terminal fragment, i.e. a degron8
PHD1'unknown - no verified independent function'verified function - mediates the MLL(N)-MLL(C) holocomplex; its natural deletion in some T-cell leukemia is a loss-of-function allele8
PHD4'unknown; routinely omitted'verified function - also mediates holocomplex formation8
FYRCunmeasuredretained by the reciprocal product and druggable there - FYRN/FYRC heterodimerisation activates AF4-MLL and protects it from SIAH1/2 ubiquitinylation; a FYRC decoy destroys it9
MOF contact'assessed only on the wild-type protein'assessed in a fusion-driven leukemia - but the result does not say what this ledger implied7

Two further corrections of scope rather than class:

  • 'Lost in every fusion' was scoped wrongly. Every row here is lost from der(11) KMT2A::partner. The reciprocal der(4) AFF1::KMT2A product carries FYRN, FYRC, the taspase1 sites and the SET domain, and in that protein the FYRN/FYRC clamp is the activating step and a proposed drug target 9. A ledger of 'what the fusion loses' that does not name the derivative is ambiguous at exactly the point where the answer flips.
  • 'Highest functional evidence' for the MOF contact is withdrawn. See below.

What the overturned rows have in common is the real finding of this pass.

The pattern the corrected rows make: the KMT2A half loses abundance control, not activity control

Three independent determinants of MLL1 turnover sit C-terminal to the breakpoint cluster region and are absent from every der(11) fusion:

  1. The PHD2 E3 ligase (1479-1533). Disabling it stabilises MLL1, increases its transactivation and increases its recruitment to target loci 12.
  2. The taspase1 cleavage sites (2666/2667, 2718/2719). Cleavage primes MLL1 for proteasomal degradation; blocking it stabilises MLL1 and raises its genome-wide chromatin occupancy 6.
  3. The FYRN degron (2018-2074). Free MLL(N) - an N-terminal fragment with no C-terminal partner, which is structurally what a fusion's KMT2A half is - is destroyed by a mechanism targeting FYRN 8.

All three perturbations point the same way: remove the module, get more protein, longer-lived, more chromatin-bound. That is a coherent story, and it is not the story this beat was framed around. The beat's organising metaphor has been restraint in the sense of brakes on activity - the PHD3/CyP33 repressor switch, the loss of a repression domain. The better-evidenced reading is that what the KMT2A half principally throws away is brakes on its own abundance, and that the fusion's defining biochemical property may be less that it does something new than that it does not go away.

This reframing is written up separately as hyp-kmt2a-half-loses-abundance-control-not-activity-control, and it is falsifiable in one experiment: measure fusion protein half-life against wild-type KMT2A half-life in the same cell. If the fusion is not the more stable species, the synthesis fails.

It also has a therapeutic shape that is already half-demonstrated. CX-4945 does not touch the fusion. It restores the wild-type allele's abundance, and the restored wild-type protein competes the fusion off chromatin 6. That is the same strategic move as MM-401 - act on the allele the fusion does not have - reached from the opposite direction.

The MOF row, corrected

The previous ledger ranked the MOF contact as the highest functional evidence of any KMT2A-side loss, on the strength of Mll1-deletion experiments in normal cells. That ranking is withdrawn, for two reasons drawn from a paper that tested MOF directly in a fusion-driven leukemia 7.

In MOF's favour: Mof is genuinely required. Conditional deletion in MLL-AF9-transformed cells abolishes colony formation, lowers tumour burden and extends median survival from 22-24 to 31 days, and an intact HAT domain is required to rescue. So the ledger's statement that the module had been 'assessed only on the wild-type protein' was simply false.

Against the ranking, decisively:

  • MOF loss does not downregulate the fusion's target programme. The authors state that MOF loss does not clearly downregulate MLL-AF9 target genes such as Meis1 and the Hoxa cluster. Whatever MOF does for this leukemia, it is not sustaining the oncogenic transcriptional output that a lost MOF contact would predict. This is consistent with what the graph already records elsewhere and is now sourced to a direct experiment.
  • It is not KMT2A-specific. Mof is equally required for NUP98-fusion-driven AML, and MG149 kills three non-KMT2A-rearranged leukemia lines. The mechanism the authors reach is global H4K16ac loss, DNA-damage-repair gene downregulation and gammaH2AX accumulation - a general leukemia dependency.

So the correct statement is: MOF is a dependency of the leukemic cell; the fusion's inability to contact MOF has still never been shown to matter. Those are different claims, and collapsing them is precisely the error this ledger exists to prevent. The row's class is now dependency-of-the-cell-not-of-the-fusion, the same class as the WIN/WRAD interface.

The three rows where the absence held, and what was searched

Confirming an absence is worth something only with the search attached, so:

  • Bromodomain (1635-1765). Searched PubMed for KMT2A/MLL bromodomain function, structure and phenotype, and for the PHD3-bromo cassette. Nothing beyond its role as the partner of PHD3, where the functional handle is CyP33-driven proline isomerisation of the linker (PMID 20541251, already in this graph). No phenotype attaches to losing the bromodomain itself. Absence holds.
  • pre-PHD C2HC zinc finger (1870-1910). Nothing found under any query. Absence holds, and this is the least-studied feature in the table - which is also the reason to distrust the absence most here, since the feature's name is unstable in the literature.
  • HCF-binding motif (1799-1802). Searched HCFC1 together with MLL and leukemia; the only MLL hits were the founding Yokoyama 2004 paper 2 and an unrelated Epstein-Barr virus study. Nobody has tested what losing the HBM costs. Absence holds, which is what makes exp-hbm-reinsertion-kmt2a-aff1 worth running.

Three rows out of fifteen, not nine or ten.

The ledger1 deeper
ModuleResidues (Q03164)Coordinates confirmed byWhat is lostLoss class
PHD11431-1482UniProtmediates the MLL(N)-MLL(C) holocomplex; naturally deleted in some T-ALL as a loss-of-function allele 8verified function; loss-in-fusion untested
PHD2 (E3 ligase)1479-1533UniProtCDC34-dependent self-limitation; disabling it stabilises MLL1 and raises its occupancy 12demonstrated on wild-type MLL1; untested in a fusion
PHD31566-1627UniProtCyP33/PPIE binding; H3K4me3 reading (region 1584-1600)demonstrated (add-back blocks replating; one study)
Bromodomain1635-1765UniProtclamp on PHD3; target of CyP33 linker isomerisationunmeasured - searched cycle 31, nothing found
pre-PHD C2HC1870-1910UniProtunknownunmeasured - searched cycle 31, nothing found
PHD41931-1978UniProtmediates the MLL(N)-MLL(C) holocomplex 8; routinely omitted from 'the three PHD fingers'verified function; loss-in-fusion untested
HCF-binding motif1799-1802 (DHNY)primary full text + unique sequence matchHCF-1 and HCF-2 recruitmentunmeasured - searched cycle 31, nothing found
FYRN2018-2074UniProthalf of the N320/C180 clamp, and the degron targeted by the mechanism that destroys free MLL(N) 8demonstrated (degradation determinant)
Taspase1 site 12666/2667UniProtproteolytic maturation, which primes MLL1 for degradation 6demonstrated in trans; reversal is therapeutic 6
Taspase1 site 22718/2719UniProtproteolytic maturation; CKII phosphorylates T2724/S2726 to license it 6demonstrated in trans; reversal is therapeutic 6
Transactivation domain2829-2883 (9aaTAD 2847-2855)primary abstract, verbatimCBP KIX bindingswapped - partner supplies elongation-based activation
FYRC3666-3747UniProtother half of the clamp; retained by the reciprocal AFF1::KMT2A product, where it is the activating heterodimerisation surface and a decoy target 9wild-type-and-reciprocal-only; druggable there
WIN motif3762-3767 (Arg3765)crystal structure + cellular co-IP + UniProtWDR5/RBBP5/ASH2L/DPY30 assemblywild-type-only - druggable (MM-401)
SET3829-3945UniProtH3K4 methyltransferase activitytolerated genetically; contradicted pharmacologically
MOF contactC-terminal, range unresolvedprimary abstract (no range given)KAT8/MOF recruitment, H4K16ac at targetsdependency of the cell, not of the fusion - MOF is required for MLL-AF9 leukemia but not via Hoxa/Meis1, and equally required in NUP98-fusion AML 7

Coordinates in bold differ from what the graph currently stores. The bromodomain is stored in gene-kmt2a as 1703-1809; UniProt gives 1635-1765.

Scope, added cycle 31. Every row in this table is lost from the der(11) KMT2A::partner product. The reciprocal der(4) AFF1::KMT2A product retains FYRN, FYRC, the taspase1 sites and SET 9.

What the brief asked me to check, and what I found

The dispatch listed five residue claims from the previous beat as unverified leads. Outcome:

  • HCF-1/HCF-2 motif ~1799-1802 - CONFIRMED, twice: stated in the Yokoyama 2004 full text with a Delta-1799/1803 mutant, and the only [DE]Hx Y match in the whole 3,969-residue sequence.
  • CyP33/PPIE at PHD3 ~1566-1627 - CONFIRMED as the PHD3 boundary (UniProt PHD-type 3, 1566-1627). The CyP33 contact surface within PHD3 is not resolved to residues by any source I read.
  • CBP at the TAD ~2829-2883 - CONFIRMED, stated verbatim in the Ernst 2001 abstract, with CBP KIX at 581-687.
  • MOF at the C-terminus - NOT CONFIRMED to a residue range. Dou 2005's abstract says 'MLL1 C-terminal' and gives no numbers. Recorded with no coordinates rather than a guess.
  • WRAD near the SET domain - CONFIRMED and sharpened: it is not 'near the SET domain', it is the WIN motif at 3762-3767 with Arg3765 in the WDR5 arginine pocket, ~65 residues N-terminal to the SET domain, and required for WRAD assembly in cells.

Additionally confirmed and not in the brief: taspase1 sites at 2666/2667 and 2718/2719; FYRN 2018-2074; FYRC 3666-3747; a fourth PHD finger at 1931-1978 and a pre-PHD C2HC finger at 1870-1910 that the graph's 'three PHD fingers' phrasing omits.

The count that matters (recounted, cycle 31)

Fifteen rows. Classified by what has actually been shown, after the correction pass:

  • Demonstrated consequence of loss, in cis (add-back into a fusion): 1 - PHD3, on a single unreplicated 2008 in-vitro study in one fusion (ana-phd3-addback-replication-audit).
  • Demonstrated consequence of loss, measured on wild-type KMT2A or in trans: 4 - the two taspase1 sites 6, FYRN 8 and PHD2 12. All four are degradation determinants and all four point the same way.
  • Verified molecular function, consequence of loss untested: 2 - PHD1 and PHD4, which mediate the MLL(N)-MLL(C) holocomplex 8.
  • Dependency of the leukemic cell through the allele the fusion does not have: 2 - the WIN/WRAD interface (MM-401) and the MOF contact 7.
  • Retained and druggable on the reciprocal product: 1 - FYRC 9.
  • Demonstrated tolerance of loss: 1 - the SET domain, genetically, though contested pharmacologically.
  • Swapped for a partner-supplied equivalent: 1 - the TAD.
  • No demonstrated functional consequence whatsoever: 3 - the bromodomain, the pre-PHD C2HC finger and the HCF-binding motif. Each was searched for specifically this cycle, and the searches are recorded above so a future reader can tell a real absence from an unasked question.

The previous headline of 9-10 untested rows was wrong by roughly a factor of three. The error was not in any row's sourcing. It was that the ledger asked 'has anyone added this back into a fusion?' and recorded 'no' as 'no demonstrated consequence'. Those are different questions, and this project has now made that mistake in both directions - once by over-claiming a consequence, and here by over-claiming an absence.

Second axis, added the same cycle: is the function supplied in trans?

The coordinator flagged mid-cycle that retained/discarded is not exhaustive, and the ledger was rebuilt with a second column in data.modules[].supplied_in_trans.

Only three module groups have primary-source grounds for 'supplied-in-trans':

  • PHD1-PHD3 - the PHD fingers sit inside an essential MLL1 recruitment domain, and MLL1 fusion proteins, lacking them, require prebinding of a wild-type MLL1 complex for stable HoxA9 association 5. This is the clearest case in the whole beat and it substantially changes how the PHD3 'lost brake' story should be read.
  • WIN/WRAD and SET - MM-401 blocks an interface the fusion does not possess and selectively kills MLL-rearranged cells, which only makes sense if the wild-type allele's assembled complex is doing work the leukemic cell needs.

For the other eleven rows the value is unknown: the wild-type allele certainly encodes the module, but nobody has shown it acting on the fusion's behalf. 'Encoded by the other allele' is not 'supplied in trans', and collapsing the two would repeat, one level up, exactly the retained/required confusion this beat was sent to fix.

The consequence for the ledger's headline count: of fifteen rows, one is demonstrated-and-outsourced (PHD3), one is tolerated, one is wild-type-only-but-druggable, one is swapped, and eleven remain untested on at least one of the two axes.

Cycle 31 update. The PHD-finger 'supplied-in-trans' value no longer rests only on a recruitment model. Thiel et al. showed that the wild-type Mll allele is genetically required for MLL-AF9 leukemogenesis and for maintenance of transformed cells, with menin recruiting both wild-type MLL and the fusion to HOX loci and wild-type MLL controlling a distinct component of the fusion-induced methylation 13. That is a functional requirement, not an occupancy observation, and it upgrades the whole second axis.

It also creates a tension the beat should hold rather than resolve. Thiel says the leukemia needs the wild-type allele's product 13. Zhao says that increasing the wild-type allele's product displaces the fusion from chromatin and is therapeutic 6. Both can be true if the relationship is dose-dependent and non-monotonic - some wild-type MLL1 is required for the fusion to load, too much competes it off - but nobody has measured that curve, and if it is real it is the most actionable thing in this ledger.

Facts and figures
statusrun
result summaryCORRECTED cycle 31. Of 15 lost modules: 1 has an in-cis add-back phenotype (PHD3, unreplicated); 4 have a demonstrated consequence measured on wild-type KMT2A or in trans (taspase1 CS1 and CS2, FYRN, PHD2), all of them degradation determinants pointing the same way - loss makes the protein more abundant and more chromatin-bound; 2 have a verified molecular function with the consequence of loss untested (PHD1, PHD4); 2 are dependencies of the leukemic cell through the intact allele (WIN/WRAD, MOF); 1 is retained and druggable on the reciprocal AFF1::KMT2A product (FYRC); 1 is tolerated (SET); 1 is swapped (TAD); and 3 have no demonstrated consequence after a specific search (bromodomain, pre-PHD C2HC, HBM). The previous headline of 9-10 untested rows is withdrawn - it was wrong by about a factor of three.
second axis noteAdded cycle 29 after the coordinator flagged that retained/discarded is not exhaustive. 'supplied-in-trans' is asserted only where a primary source shows the wild-type allele performing the function in the leukemic context: the PHD fingers via PMID 20541448 (fusion requires prebinding of a wild-type MLL1 complex), and the WIN/WRAD and SET modules via PMID 24389101 (pharmacological block of a wild-type-only interface kills MLL-r cells). Everything else is 'unknown' - the wild-type allele encodes the module, but nobody has shown it acts for the fusion.
scope noteEvery row is lost from der(11) KMT2A::partner. The reciprocal der(4) AFF1::KMT2A product carries FYRN, FYRC, the taspase1 sites and SET, and there the FYRN/FYRC clamp is an activating, druggable surface (PMID 21233834). Statements of the form 'the fusion loses X' in this graph should name the derivative.
modules
PHD11,431PHD1: 1431PHD21,479PHD2: 1479PHD31,566PHD3: 1566Bromodomain1,635Bromodomain: 1635HCF-binding motif (HBM)1,799HCF-binding motif (HBM): 1799pre-PHD C2HC1,870pre-PHD C2HC: 1870PHD41,931PHD4: 1931FYRN2,018FYRN: 2018Taspase1 cleavage site 12,666Taspase1 cleavage site 1: 2666Taspase1 cleavage site 22,718Taspase1 cleavage site 2: 2718TAD (transactivation doma…2,829TAD (transactivation domain): 2829FYRC3,666FYRC: 3666WIN (WDR5 interaction mot…3,762WIN (WDR5 interaction motif): 3762SET3,829SET: 3829
inputs
[
  "UniProt Q03164 REST record",
  "PMID 15199122 full text (PMC480881)",
  "PMID 11259575",
  "PMID 15960975",
  "PMID 18829459",
  "PMID 22665483",
  "PMID 12482972",
  "PMID 14636557",
  "PMID 23129768",
  "PMID 20541251",
  "PMID 24813891",
  "PMID 30573454 full text (PMC6317322)",
  "PMID 28202522 full text (PMC5501293)",
  "PMID 21670200",
  "PMID 21233834",
  "PMID 20506279",
  "PMID 16951254",
  "PMID 20159607"
]
modules — the numbers behind the chart
[
  {
    "name": "PHD1",
    "start": 1431,
    "end": 1482,
    "loss_class": "verified-function-loss-untested",
    "supplied_in_trans": "supplied-in-trans"
  },
  {
    "name": "PHD2",
    "start": 1479,
    "end": 1533,
    "loss_class": "demonstrated-on-wild-type",
    "supplied_in_trans": "supplied-in-trans"
  },
  {
    "name": "PHD3",
    "start": 1566,
    "end": 1627,
    "loss_class": "demonstrated",
    "supplied_in_trans": "supplied-in-trans"
  },
  {
    "name": "Bromodomain",
    "start": 1635,
    "end": 1765,
    "loss_class": "unmeasured-searched-nothing-found",
    "supplied_in_trans": "unknown"
  },
  {
    "name": "HCF-binding motif (HBM)",
    "start": 1799,
    "end": 1802,
    "loss_class": "unmeasured-searched-nothing-found",
    "supplied_in_trans": "unknown"
  },
  {
    "name": "pre-PHD C2HC",
    "start": 1870,
    "end": 1910,
    "loss_class": "unmeasured-searched-nothing-found",
    "supplied_in_trans": "unknown"
  },
  {
    "name": "PHD4",
    "start": 1931,
    "end": 1978,
    "loss_class": "verified-function-loss-untested",
    "supplied_in_trans": "unknown"
  },
  {
    "name": "FYRN",
    "start": 2018,
    "end": 2074,
    "loss_class": "demonstrated-degron",
    "supplied_in_trans": "unknown"
  },
  {
    "name": "Taspase1 cleavage site 1",
    "start": 2666,
    "end": 2667,
    "loss_class": "demonstrated-in-trans",
    "supplied_in_trans": "supplied-in-trans"
  },
  {
    "name": "Taspase1 cleavage site 2",
    "start": 2718,
    "end": 2719,
    "loss_class": "demonstrated-in-trans",
    "supplied_in_trans": "supplied-in-trans"
  },
  {
    "name": "TAD (transactivation domain)",
    "start": 2829,
    "end": 2883,
    "loss_class": "swapped",
    "supplied_in_trans": "unknown"
  },
  {
    "name": "FYRC",
    "start": 3666,
    "end": 3747,
    "loss_class": "retained-by-reciprocal-product",
    "supplied_in_trans": "unknown"
  },
  {
    "name": "WIN (WDR5 interaction motif)",
    "start": 3762,
    "end": 3767,
    "loss_class": "wild-type-only",
    "supplied_in_trans": "supplied-in-trans"
  },
  {
    "name": "SET",
    "start": 3829,
    "end": 3945,
    "loss_class": "tolerated",
    "supplied_in_trans": "supplied-in-trans"
  }
]
module without coordinates
{
  "name": "MOF/KAT8 recruitment contact",
  "note": "primary source gives 'MLL1 C-terminal' with no residue range; deliberately not assigned; and MOF is required for MLL-AF9 leukemia maintenance but not via Hoxa/Meis1, and is equally required in NUP98-fusion AML (PMID 28202522, full text), so this row is a dependency of the cell, not a demonstrated cost of the fusion's missing contact",
  "loss_class": "dependency-of-the-cell-not-of-the-fusion",
  "supplied_in_trans": "supplied-in-trans"
}
corrections owed
[
  {
    "entry": "gene-kmt2a",
    "field": "data.domains[] 'Bromodomain'",
    "stored": "1703-1809",
    "should_be": "1635-1765",
    "source": "UniProt Q03164 Domain: Bromo"
  },
  {
    "entry": "gene-kmt2a",
    "field": "data.domains[]",
    "issue": "omits PHD4 (1931-1978) and the pre-PHD C2HC finger (1870-1910); the graph's recurring phrase 'the three PHD fingers' undercounts",
    "source": "UniProt Q03164 zinc-finger features"
  },
  {
    "entry": "dispatch brief premise (not a graph file)",
    "field": "the premise that the fusion retains FYRN",
    "issue": "the cycle-31 dispatch stated that FYRN (2018-2074) and FYRC (3666-3747) are split by the breakpoint, so the fusion keeps FYRN and loses FYRC",
    "should_be": "FYRN and FYRC are split by the TASPASE1 CLEAVAGE at 2666/2718, not by the translocation breakpoint. The breakpoint cluster region ends around residue 1400-1445, so der(11) fusions lose BOTH halves of the clamp. There is no orphaned FYRN in the fusion; the orphan is the wild-type MLL(N) fragment, and its fate is degradation through a mechanism that targets FYRN.",
    "source": "UniProt Q03164 feature table (translocation-breakpoint site 1444/1445; FYRN 2018-2074; cleavage sites 2666/2667 and 2718/2719), with PMID 18235233 stating that 11q23 translocations fuse the common MLL N-terminal ~1400 aa in frame, and PMID 21670200 for the FYRN-targeted degradation of free MLL(N)"
  }
]
How this could bite us back
  • A ledger invites the reader to treat every row as equivalent. They are not: the evidence behind the rows spans from a 1.7 A crystal structure to an unreplicated 2008 replating assay to nothing at all, and the loss_class column exists to stop that flattening.
  • Confirming a coordinate is not confirming a consequence. Every coordinate in this table is now well sourced, which could easily be mistaken for the beat having established that these losses matter. It has established the opposite for most of them.
  • The correction pass moved five rows, and the pattern it found - that the losses are degradation determinants - is itself a story that could be over-told. Three determinants pointing the same way is a pattern, not a proof, and the crucial measurement (fusion half-life versus wild-type half-life in one cell) has not been made.
  • 'Searched and found nothing' is only as good as the searches, which were PubMed queries in a single session. Two of the three surviving absences concern features whose names are unstable in the literature - the pre-PHD C2HC finger and the bromodomain - which is exactly how a real result hides from a query.
Still open
  1. Is the fusion actually longer-lived than wild-type KMT2A in the same cell? Three independent degradation determinants are missing from the fusion [E6, E8, E12] and the conclusion is asserted in review and discussion rather than measured head-to-head. A cycloheximide chase or degron washout comparing the two species in one KMT2A::AFF1 line would settle the beat's new organising claim in about a week.
  2. Is the wild-type MLL1 dose-response on fusion chromatin occupancy non-monotonic? Thiel says the leukemia requires the wild-type allele [E13]; Zhao says stabilising it displaces the fusion and prolongs survival [E6]. Titrating wild-type MLL1 abundance against fusion occupancy in one line would test whether there is a therapeutic window on the allele the fusion does not have.
  3. Does taspase1 cleavage activate MLL1 [E11] or destabilise it [E6]? Two careful studies reach opposite conclusions on the same perturbation, and every 'maturation' statement downstream in this beat depends on which is right.
  4. Where does MOF bind MLL1 in residues? Still unresolved - but the question now matters less, because MOF turns out to be a dependency of the cell rather than of the contact [E7].
  5. Three rows survive as genuine absences - bromodomain, pre-PHD C2HC, HBM. A single in-cis add-back panel in one KMT2A::AFF1 line would cover all three, and nobody has built it.
Every source behind this entry (13)
  1. E1established confidence 5 of 5: Multiple independent lines of evidence; treated as settled in the field.database

    The UniProt Q03164 feature table (primary gene name KMT2A, 3,969 aa) places: PHD1 1431-1482, PHD2 1479-1533, PHD3 1566-1627, H3K4me3-interaction region 1584-1600, Bromo 1635-1765, pre-PHD C2HC 1870-1910, PHD4 1931-1978, FYRN 2018-2074, taspase1 cleavage sites at 2666/2667 and 2718/2719, 9aaTAD 2847-2855, FYRC 3666-3747, WIN motif 3762-3767 with residue 3765 important for WDR5 binding, SET 3829-3945 and post-SET 3953-3969.

    What this evidence cannot show

    Accession resolved by fetching Q03164 directly and reading back the PRIMARY gene name and sequence length, not by a gene_exact query. UniProt feature boundaries are curatorial calls from sequence profiles and structural literature; they are the best available consistent ruler but are not experimental measurements, and constructs in individual papers can differ by several residues. UniProt annotates no HCF-binding motif and no transactivation domain spanning 2829-2883, so those two rows come from primary literature, not from this record.

    Evidence that argues against this (1)
    • A UniProt feature table is a curated model, not a measurement, and this ledger treats its boundaries as if they were the domains. UniProt's PHD and bromo calls for KMT2A are largely propagated from sequence profiles; Pfam, SMART, InterPro and CATH give different edges for the same regions, and the sibling atlas beat corrected several entries last cycle for quoting one source's range as though it were observed. A lost-module ledger built on one annotation set inherits that annotation set's choices about where a module stops. Protein Science 2010
  2. E2established confidence 5 of 5: Multiple independent lines of evidence; treated as settled in the field.paper

    HCF-1 and HCF-2 bind a conserved (D/E)HXY motif at MLL amino acids 1799-1802, a Delta-1799/1803 mutant loses HCF binding while retaining ASH2L and WDR5, menin interaction maps within the first 1,406 amino acids of MLL-N, and menin is the only identified component of the MLL/HCF complex that associates with mutant MLL fusion proteins.

    Leukemia proto-oncoprotein MLL forms a SET1-like histone methyltransferase complex with menin to regulate Hox gene expression · Yokoyama A, Wang Z, Wysocka J, Sanyal M, Aufiero DJ, Kitabayashi I, Herr W, Cleary ML · Molecular and Cellular Biology 24(13):5639-5649 2004
    PMID 15199122DOI
    What this evidence cannot show

    FULL TEXT read via PMC480881 for the residue numbers. This is the source for the two rows UniProt does not cover and for the retained/lost boundary itself.

    Evidence that argues against this (1)
    • A deletion mutant losing one interaction while retaining others is good evidence that the motif is necessary for HCF binding, and weak evidence that HCF binding is what the motif does in a cell. Delta-1799/1803 removes a region in a large scaffold, and retention of ASH2L and WDR5 shows the complex has not collapsed globally - it does not show that nothing else at that surface changed. The entry treats a binding result as a module assignment. Molecular and Cellular Biology 2004
  3. E3established confidence 5 of 5: Multiple independent lines of evidence; treated as settled in the field.paper

    The minimal MLL region binding the CBP KIX domain is amino acids 2829-2883, a potent transcriptional activation domain, with CBP residues 581-687.

    MLL and CREB bind cooperatively to the nuclear coactivator CREB-binding protein · Ernst P, Wang J, Huang M, Goodman RH, Korsmeyer SJ · Molecular and Cellular Biology 21(7):2249-2258 2001
    PMID 11259575DOI
    What this evidence cannot show

    ABSTRACT ONLY; the residue range is stated verbatim in the abstract so the coordinate is confirmed, but the underlying mapping data were not inspected.

    Evidence that argues against this (1)
    • A minimal interacting region mapped in vitro is not evidence that the contact operates at that boundary in a cell. Residues 2829-2883 is the smallest fragment that bound CBP KIX in the assay used, which sets an upper bound on the required region and says nothing about affinity, stoichiometry or competition in a nucleus where CBP KIX is bound by many partners including CREB - the same paper's other subject. Molecular and Cellular Biology 2001
  4. E4established confidence 5 of 5: Multiple independent lines of evidence; treated as settled in the field.paper

    The 2023 KMT2A recombinome analysed 3,401 acute leukaemia patients diagnosed between 2003 and 2022 and, with the published literature, identifies 107 in-frame KMT2A fusion partners, plus 16 out-of-frame rearrangements, 18 patients with no partner fused to 5'-KMT2A and 2 with a 5'-KMT2A deletion. Breakpoints concentrate overwhelmingly in the major cluster region BCR1, between KMT2A intron 7 and exon 13: 3,336 patients (98%) against 47 (1.4%) in the minor BCR2 between intron 20 and exon 24. The seven most frequent partners plus partial tandem duplications account for more than 90% of all KMT2A recombinations. The composition of the fusion protein's KMT2A half is therefore nearly constant regardless of partner.

    The KMT2A recombinome of acute leukemias in 2023. · Meyer C, Larghero P, Almeida Lopes B, et al. · Leukemia 37(5):988-1005 2023
    PMID 37019990DOI
    What this evidence cannot show

    Read in full from PMC10169636 on 2026-09-19; the figures above are from the abstract and the breakpoint-distribution section, not recalled. This item previously said only 'more than 100 documented fusion partners across thousands of patients' and was flagged by `scripts/reading-queue.py` for low claim-to-abstract overlap. The series is a referral cohort from a diagnostic reference laboratory, so partner frequencies reflect what is sent for characterisation and the 98% BCR1 figure is conditioned on a KMT2A rearrangement having been identified by that laboratory's assays.

    Evidence that argues against this (2)
    • The partner catalogue is a referral series and its composition tracks assay mix, not biology: spliced fusions are recoverable only from RNA, cryptic partners need molecular work, and visible reciprocal translocations are over-represented. More importantly for a LOST-MODULE ledger, partner count says nothing about what the KMT2A side loses, which is set by breakpoint position - and that varies across introns 9 to 11 with age. Molecular Cell 2010
    • 'Nearly constant' is doing real work and the same paper undercuts it. Within BCR1 the breakpoint position is not neutral: the authors note that outcome has been linked to breakpoint distribution, with patients breaking in KMT2A intron 11 faring worse than those with upstream breakpoints, the cut point falling between intron 10 and exon 11. A KMT2A half that is constant in domain composition is not thereby constant in consequence. Leukemia 2023
  5. E5strong confidence 4 of 5: Replicated in peer-reviewed work by more than one group.paper

    An essential MLL1 recruitment domain includes the CXXC domain and the PHD fingers; MLL1 fusion proteins lack the PHD fingers and require prebinding of a wild-type MLL1 complex, together with CXXC-domain recognition of DNA, for stable HoxA9 association.

    Multiple interactions recruit MLL1 and MLL1 fusion proteins to the HOXA9 locus in leukemogenesis · Milne TA, Kim J, Wang GG, Stadler SC, Basrur V, Whitcomb SJ, Wang Z, Ruthenburg AJ, Elenitoba-Johnson KSJ, Roeder RG, Allis CD · Molecular Cell 38(6):853-863 2010
    PMID 20541448DOI
    What this evidence cannot show

    ABSTRACT ONLY. Recruitment and stable-association assays at a single locus (HoxA9); it establishes a trans dependency for binding, not for transformation. Used here solely to justify the 'supplied-in-trans' value on the PHD rows of the ledger.

    Evidence that argues against this (1)
    • If the fusion REQUIRES prebound wild-type MLL1 for stable HoxA9 association, then the modules listed as lost are not simply absent from the system - they are supplied in trans by the wild-type allele. A ledger of what the fusion lacks therefore overstates what the leukaemic cell lacks, and the two are different questions. Molecular Cell 2010
  6. E6strong confidence 4 of 5: Replicated in peer-reviewed work by more than one group.paper

    Taspase1-dependent cleavage of MLL1 destabilises it: loss of taspase1 leaves MLL1 uncleaved, more resistant to cycloheximide-chase degradation and markedly increased in genome-wide chromatin occupancy; because KMT2A breakpoints lie N-terminal to the cleavage sites the chimeras are resistant to ubiquitin-proteasome degradation and are more stable than the wild-type MLL1 copy in leukemic cells; and stabilising uncleaved wild-type MLL1 - by taspase1 knockout or by CKII inhibition with CX-4945 - displaces MLL-AFF1, AFF4 and ELL2 from chromatin in SEM cells, lowers target-gene mRNA, and delays leukemic progression and prolongs survival in an MLL-AF9 mouse model.

    Regulation of MLL/COMPASS stability through its proteolytic cleavage by taspase1 as a possible approach for clinical therapy of leukemia · Zhao Z, Wang L, Volk AG, Birch NW, Stoltz KL, Bartom ET, Marshall SA, Rendleman EJ, Nestler CM, Shilati J, Schiltz GE, Crispino JD, Shilatifard A · Genes & Development 33(1-2):61-74 2019
    PMID 30573454DOI
    What this evidence cannot show

    FULL TEXT read via PMC6317322. The degradation and occupancy measurements are on wild-type MLL1 in 293T, MCF7 and HCT116; the leukemia arms are the SEM/MV4-11/RS4;11 CX-4945 experiments and the MLL-AF9 syngeneic transplant. The statement that the chimeras are more stable than wild-type MLL1 is asserted in the discussion with citations rather than measured here, and CX-4945 is a multi-target CKII inhibitor whose survival benefit is not formally proven to run through MLL1 stabilisation. The direction of this result contradicts Takeda et al. 2006 [E11]; the conflict is recorded, not adjudicated.

    Evidence that argues against this (1)
    • Takeda et al. reached the opposite directional conclusion on the same perturbation: uncleaved precursor MLL has REDUCED H3K4 methyltransferase activity in vitro, Taspase1-null cells show decreased H3K4me3 at Cyclin E1 and E2, and MLL(nc/nc;2nc/nc) MEFs are impaired in proliferation - i.e. cleavage activates rather than destabilises. Zhao et al. additionally note a third study in which non-cleavable knock-in mice had no apparent defect. The field is genuinely unresolved. Takeda S et al. 2006
  7. E7strong confidence 4 of 5: Replicated in peer-reviewed work by more than one group.paper

    Conditional Mof (Kat8) deletion in MLL-AF9-transformed cells abolishes colony formation and, in vivo, lowers tumour burden and extends median survival from 22-24 to 31 days; rescue requires an intact MOF HAT domain (a G327E point mutant and a CoA-site deletion both fail); MOF loss depletes GLOBAL H4K16 acetylation and downregulates DNA-damage-repair genes with increased gammaH2AX foci; and, explicitly, MOF loss does not lead to clear downregulation of MLL-AF9 target genes such as Meis1 and the Hoxa cluster. Mof is equally required for NUP98-fusion-driven AML, and the MYST HAT inhibitor MG149 kills three non-KMT2A-rearranged leukemia lines as well as MLL-rearranged ones.

    Histone Acetyltransferase Activity of MOF Is Required for MLL-AF9 Leukemogenesis · Valerio DG, Xu H, Chen CW, Hoshii T, Eisold ME, Delaney C, Cusan M, Deshpande AJ, Huang CH, Lujambio A, Zheng YG, Zuber J, Pandita TK, Lowe SW, Armstrong SA · Cancer Research 77(7):1753-1762 2017
    PMID 28202522DOI
    What this evidence cannot show

    FULL TEXT read via PMC5501293. This is the paper that makes 'the MOF contact has been assessed only on the wild-type protein' false. But it measures MOF the enzyme in the leukemic cell, not the fusion's inability to contact MOF: no arm restores a MOF contact to a fusion. The authors' own preferred reading is that MLL1 is a scaffold recruiting MOF to the targets of the oncogenic fusion - i.e. the wild-type allele acting in trans. MG149 is a weak, non-selective first-generation tool (IC50 19.6-50.9 microM).

    Evidence that argues against this (1)
    • The in vitro and in vivo magnitudes disagree and only one is quoted here. Colony formation is abolished, but median survival moves from 22-24 days to 31 - a real extension and not an abolition. Reading MOF as a required module from the colony result overstates what the animal shows. Cancer Research 2017
  8. E8strong confidence 4 of 5: Replicated in peer-reviewed work by more than one group.paper

    The stability of MLL(N) depends on an intramolecular interaction with MLL(C) mediated through PHD1, PHD4 and the FYRN domain of MLL(N); free MLL(N) is destroyed by a mechanism that targets the FYRN domain, while free MLL(C) is exported to the cytoplasm and degraded by the proteasome; and PHD1, encoded by an alternatively spliced exon occasionally deleted in T-cell leukemia, is required for holocomplex formation, so its absence is a loss-of-function allele.

    Proteolytically cleaved MLL subunits are susceptible to distinct degradation pathways · Yokoyama A, Ficara F, Murphy MJ, Meisel C, Naresh A, Kitabayashi I, Cleary ML · Journal of Cell Science 124(Pt 13):2208-2219 2011
    PMID 21670200DOI
    What this evidence cannot show

    ABSTRACT ONLY. A genetically engineered mouse line lacking MLL(C) plus construct biochemistry; the degradation routes are inferred from fragment stability and the specific ubiquitin machinery acting on FYRN is not identified. Cited here for three ledger corrections: PHD1 and PHD4 have verified functions, and FYRN behaves as a degron on an unpartnered N-terminal fragment.

    Evidence that argues against this (1)
    • This is a statement about WILD-TYPE MLL architecture, and the fusion has no MLL(C) to interact with, so the intramolecular stabilisation described cannot operate in it. That makes the finding evidence about what the wild-type allele does, not a module the fusion has lost - and the ledger's framing conflates the two. Journal of Cell Science 2011
  9. E9moderate confidence 3 of 5: One strong study, or several smaller consistent ones.paper

    In t(4;11) the RECIPROCAL AFF1::KMT2A (AF4-MLL) product does contain FYRN and FYRC: it is activated by taspase1-mediated hydrolysis into AF4-MLL-N and MLL-C, which heterodimerise through the FYRN and FYRC interaction domains, and that heterodimer induces a high-molecular-weight complex protected against SIAH1/2-mediated polyubiquitinylation; expressing FYRC, or smaller portions of it, blocks heterodimer formation and destroys the AF4-MLL oncoprotein.

    The heterodimerization domains of MLL-FYRN and FYRC--are potential target structures in t(4;11) leukemia · Pless B, Oehm C, Knauer S, Stauber RH, Dingermann T, Marschalek R · Leukemia 25(4):663-670 2011
    PMID 21233834DOI
    What this evidence cannot show

    ABSTRACT ONLY. Bacterial two-hybrid interaction mapping plus a mammalian biosensor assay in HeLa cells - construct-level destruction, not a leukemia model, and the premise that AF4-MLL is itself oncogenic rests on the same group's transplantation work. Cited to correct a scope error in this ledger: its rows are lost from der(11) KMT2A::partner, not from every product of the translocation.

    Evidence that argues against this (1)
    • The reciprocal AFF1::KMT2A product exists in only a subset of t(4;11) patients - the reciprocal transcript is expressed in roughly half of those with an intact der(4) - so a module 'restored' by the reciprocal is restored in some patients and not others. A ledger entry that notes FYRN/FYRC return via the reciprocal should say in how many. Leukemia 2011
  10. E10strong confidence 4 of 5: Replicated in peer-reviewed work by more than one group.paper

    FYRN and FYRC are not two domains but two halves of a single folded module: the crystal structure of the TBRG1 FYR region shows FYRN and FYRC each forming part of one novel alpha+beta fold, and in proteins such as trithorax and MLL, where the two motifs are separated by hundreds of amino acids, the FYR domain is expected to carry a large insertion between two strands of the same beta-sheet.

    The structure of the FYR domain of transforming growth factor beta regulator 1 · Garcia-Alai MM, Allen MD, Joerger AC, Bycroft M · Protein Science 19(7):1432-1438 2010
    PMID 20506279DOI
    What this evidence cannot show

    ABSTRACT ONLY, and the structure solved is TBRG1, not MLL - the MLL statement is the authors' explicit extrapolation from sequence homology, not a measurement on MLL. Cited for one structural point: an unpartnered FYRN is half a beta-sheet rather than a domain, which is why treating an orphaned half-module as inert was never a safe default.

    Evidence that argues against this (1)
    • The crystal structure is of TBRG1, not MLL. Concluding that KMT2A's FYRN and FYRC are two halves of one fold is an inference across proteins, and the same paper notes the MLL arrangement carries a large insertion between two strands of one sheet - which is precisely the kind of difference that can make a shared fold behave differently. No KMT2A FYR structure exists. Protein Science 2010
  11. E11strong confidence 4 of 5: Replicated in peer-reviewed work by more than one group.paper

    Taspase1-null mice show noncleavage of MLL and MLL2 with homeotic transformations and impaired MEF proliferation; the uncleaved precursor MLL displays REDUCED histone H3 methyltransferase activity in vitro, with markedly decreased H3K4 trimethylation at the Cyclin E1 and E2 genes; MLL(nc/nc;2nc/nc) MEFs are also impaired in proliferation; and Taspase1-null cells are resistant to oncogenic transformation.

    Proteolysis of MLL family proteins is essential for taspase1-orchestrated cell cycle progression · Takeda S, Chen DY, Westergard TD, Fisher JK, Rubens JA, Sasagawa S, Kan JT, Korsmeyer SJ, Cheng EH, Hsieh JJ · Genes & Development 20(17):2397-2409 2006
    PMID 16951254DOI
    What this evidence cannot show

    ABSTRACT ONLY. Mouse genetics and MEFs; no leukemia model and no fusion protein. Listed as the counterweight to E6 - the same perturbation read in the opposite direction. The ledger records the conflict rather than picking a winner, because every 'maturation' statement downstream in this beat depends on which is right.

    Evidence that argues against this (1)
    • Zhao et al. 2019 find the opposite: cleavage primes MLL1 for degradation, and loss of taspase1 increases MLL1 stability and chromatin occupancy without changing bulk H3K4me1/2/3 or MLL/COMPASS assembly. Zhao Z et al. 2019
  12. E12strong confidence 4 of 5: Replicated in peer-reviewed work by more than one group.paper

    The second PHD finger (PHD2) of MLL1 is an E3 ubiquitin ligase in the presence of the E2-conjugating enzyme CDC34, and mutation of PHD2 leads to MLL1 stabilisation, increased transactivation ability and increased MLL1 recruitment to target gene loci, so PHD2 negatively regulates MLL1 abundance and activity.

    A subset of mixed lineage leukemia proteins has plant homeodomain (PHD)-mediated E3 ligase activity · Wang J, Muntean AG, Wu L, Hess JL · Journal of Biological Chemistry 287(52):43410-43416 2012
    PMID 23129768DOI
    What this evidence cannot show

    ABSTRACT ONLY. HEK293 biochemistry with mutant constructs; no haematopoietic model and no fusion protein. Cited for the DIRECTION of the consequence, which is the point: it is the same direction as loss of the taspase1 sites [E6] and of FYRN [E8].

    Evidence that argues against this (1)
    • PHD2's E3 activity was shown with CDC34 in a reconstituted system, and stabilisation on mutation is consistent with it but does not establish that CDC34 is the physiological E2 or that this is the dominant route to MLL1 turnover in a leukaemic cell. This ledger already lists two other independent degradation determinants, which means no one of them is demonstrated to be rate-setting. Journal of Biological Chemistry 2012
  13. E13strong confidence 4 of 5: Replicated in peer-reviewed work by more than one group.paper

    The wild-type Mll allele is required for MLL-AF9-induced leukemogenesis and for maintenance of MLL-AF9-transformed cells; menin recruits both wild-type MLL and the MLL-AF9 fusion protein to HOX loci to activate their transcription, and wild-type MLL controls a distinct component of the MLL-AF9-induced histone methylation.

    MLL-AF9-induced leukemogenesis requires coexpression of the wild-type Mll allele · Thiel AT, Blessington P, Zou T, Feather D, Wu X, Yan J, Zhang H, Liu Z, Ernst P, Koretzky GA, Hua X · Cancer Cell 17(2):148-159 2010
    PMID 20159607DOI
    What this evidence cannot show

    ABSTRACT ONLY. Mouse genetics in a single fusion; 'required' is a genetic requirement in a transplantation model, not a statement about human patients, and no comparable human allele-status dataset exists. Cited as the second, functional pillar under the 'supplied-in-trans' column, which previously rested on PMID 20541448's recruitment data alone.

    Evidence that argues against this (1)
    • Five separate items in this ledger rest on this one mouse MLL-AF9 result, which makes them a single point of failure rather than five supports. It is also a myeloid mouse model standing in for infant KMT2A::AFF1 B-lymphoid disease, and the requirement was demonstrated for leukemogenesis and maintenance rather than for any specific module's function. Cancer Cell 2010
Connected to