The founding WRAD negative was never run on a fusion protein
also called Yokoyama 2004 WRAD negative · MLL delta-3820 · the 2004 negative
Everyone in this field cites a 2004 paper for the fact that the leukemia fusion protein keeps menin and loses the rest of the normal partners. Read the actual experiments and the fusion proteins were only ever tested against menin and one other protein - the WRAD proteins were tested against a shortened normal KMT2A, never against a fusion.
The same thing, for a scientist
Yokoyama et al. 2004 (Mol Cell Biol 24:5639) is the source of the statement that oncogenic MLL retains menin 'but not other identified complex components'. The paper contains two distinct experiments and this graph has been reading them as one. Figure 4 maps ASH2L, WDR5 and RBBP5 onto the SET domain using a C-terminal DELETION MUTANT of otherwise wild-type MLL (construct 34/3820, 'MLL delta-3820') transfected into 293 cells. Figure 6 tests FUSION proteins - FLAG-tagged MLL-p300 in transfected 293 cells, and endogenous MLL-ENL in the HB leukaemia line - and blots those fusion immunoprecipitates for HIS, MLL-C, HCF-1 and menin only. No WRAD subunit is blotted in any fusion experiment in the paper, and DPY30 does not appear in the paper at all. The field's canonical WRAD-versus-fusion negative is therefore an INFERENCE from a deletion mutant, not a measured co-immunoprecipitation negative on a fusion protein. That is the distinction between 'tested and negative' and 'not tested', and this project has a standing rule that the two need different entries.
What the paper actually did, experiment by experiment
Figure 4 (domain mapping, wild-type backbone). 293 cells were transiently transfected with HIS/FLAG-tagged MLL deletion mutants and anti-FLAG immunoprecipitates were blotted for HCF-1, HCF-2, ASH2L, WDR5 and RBBP5. Endogenous ASH2L, WDR5 and RBBP5 coprecipitated with full-length MLL. The mutant 'lacking C-terminal sequences spanning the SET domain (delta-3820)' lost ASH2L, WDR5 and RBBP5 while retaining HCF. The authors conclude ASH2L, WDR5 and RBBP5 'interact with the p180 MLL(C) subunit through its SET domain' 1.
Figure 5 (HCF mapping). Deleting MLL aa 1799-1802 abolished HCF-1/HCF-2 binding and did NOT affect ASH2L or WDR5 1.
Figure 6 (the only fusion experiments). (a) FLAG-tagged MLL-p300 in transfected 293 cells; the immunoblot panel is anti-His, anti-MLL-C, anti-HCF-1C and anti-menin. (b) Endogenous MLL-ENL in HB cells, menin immunoprecipitation, blotted anti-MLL-N, anti-MLL-C and anti-ENL, with REH as the non-translocated control. Neither panel contains a WRAD blot 1.
Figure 7 (function). siRNA in HeLa cells: >80% knockdown of ASH2L, WDR5, HCF-1 and HCF-2 had no measurable effect on HoxA9 expression; menin knockdown phenocopied MLL knockdown 1. HeLa is a cervical carcinoma line with no KMT2A rearrangement, so this is not a statement about leukaemia either.
What the paper's own sentence claims, and what it can carry
The abstract says: 'Oncogenic mutant forms of MLL retain an ability to interact with menin but not other identified complex components.' The Results say menin 'is the only identified component of the MLL/HCF complex that associates with mutant MLL proteins'. Both are true as written and both are supported for HCF-1/HCF-2, which were blotted on a fusion and map to aa 1799-1802, past the breakpoint cluster.
For WRAD the support is different in kind. The only WRAD measurement is on delta-3820, an otherwise intact MLL truncated after residue 3820. That construct retains the WIN motif at 3762-3773, which is the surface WDR5 is now known to bind 2 - yet WDR5 coprecipitation was lost. Two readings follow, and 2004 could not distinguish them: either WRAD's stable association requires the assembled SET-domain module rather than the WIN motif alone, or the truncation destabilised the fold. Either way, delta-3820 is not a KMT2A fusion protein: it has no partner moiety, retains the PHD fingers, the bromodomain and FYRN/FYRC, and lacks only the last 149 residues. A real fusion discards roughly 2,500 residues and adds a partner.
DPY30 is outside the claim entirely
The string 'DPY' appears zero times in the full text of Yokoyama 2004 (checked against the PMC record, 68,159 characters of flattened text). DPY30 was not described as a shared subunit of the human SET1/MLL complexes until 2007 3 and the acronym WRAD was not coined until 2012 4. A 2004 negative cannot bound a subunit that had not yet been placed in the complex. Any entry that cites Yokoyama 2004 against DPY30 is citing it for something it does not say.
Why this matters more than a pedantic reading
This graph's cycle-28 argument - that menin is very nearly the only handle the fusion retains - rests in part on the 2004 negative being a measured negative across the whole complex. It is a measured negative for HCF, an inference for WDR5/RBBP5/ASH2L, and silent on DPY30. The inference happens to be well supported by twenty years of later structural work on where WRAD binds, which is why the conclusion survives for the der(11) direct fusion. But the evidence class is wrong in several entries, and the difference shows up the moment you ask about a fusion protein that keeps the SET domain - which two classes of KMT2A rearrangement do.
| founding paper pmid | 15199122 |
|---|---|
| founding paper pmcid | PMC480881 |
| wrad mapping construct | MLL 34/3820 ('delta-3820'), C-terminal truncation of wild-type MLL lacking the SET domain, transfected 293 cells |
| win motif within delta3820 | true |
| dpy30 mentions in full text | 0 |
| full text route | PMC HTML (fullTextXML returns HTTP 500 and efetch db=pmc returns a body-less record for this publisher) |
fusion constructs tested
[ "MLL-p300 (HIS/FLAG-tagged, transient, 293 cells)", "endogenous MLL-ENL (HB cell line)" ]
proteins blotted in fusion experiments
[ "HIS tag", "MLL-C (mmC2.1)", "HCF-1C (H12)", "menin (C19)", "MLL-N (mmN4.4)", "ENL" ]
wrad subunits blotted in fusion experiments
[]
- If the WRAD-versus-fusion negative is an inference rather than a measurement, then striking WRAD subunits from this project's target list on the strength of it is a strike resting on an untested premise, and a wrong strike is silent and self-perpetuating.
- Conversely, treating the 2004 statement as merely unproven could licence re-promoting WDR5 as a fusion target, which the structural and pharmacological literature says is the wrong reading: WIN-site inhibitors act through ribosome protein gene suppression and p53, not through a KMT2A-fusion programme.
- Any WRAD-directed intervention hits all six SET1/MLL complexes and the NSL complex at once; the module is shared, so target engagement and target selectivity are different questions here.
- Did the 2004 authors run a WRAD blot on MLL-p300 or MLL-ENL and not show it? The corresponding author could still be asked, and the answer changes whether this is an untested question or an unpublished negative.
- Would delta-3820 coprecipitate WDR5 if the truncation were placed after the SET domain instead of before it? That single construct separates 'the WIN motif is insufficient' from 'the truncation misfolded', and it has never been reported.
- How many other complex-membership negatives in this field are deletion-mutant inferences read as fusion co-IP negatives? The HCF-1 case in this same paper is a measured negative; the WRAD case is not, and nothing in the citation distinguishes them.
- E1established confidence 5 of 5: Multiple independent lines of evidence; treated as settled in the field.paper
In Yokoyama 2004, ASH2L, WDR5 and RBBP5 were tested for coprecipitation only against MLL deletion mutants in transfected 293 cells (the SET-domain-deleted construct delta-3820 lost them), while the two fusion-protein experiments - FLAG-MLL-p300 in 293 cells and endogenous MLL-ENL in HB cells - were blotted for HIS, MLL-C, HCF-1 and menin only, with no WRAD subunit assayed on any fusion.
Leukemia proto-oncoprotein MLL forms a SET1-like histone methyltransferase complex with menin to regulate Hox gene expression. · Yokoyama A, Wang Z, Wysocka J, et al. · Mol Cell Biol 2004What this evidence cannot show
This is a reading of the published figures and figure legends of an open-access record (PMC480881); the underlying gels cannot be reinspected. It is possible that unshown data exist. The paper's own summary sentence is broader than its figures, and the authors may have had grounds not presented. Absence of a blot is not evidence that WRAD does associate with a fusion - it is evidence that this paper did not test it.
Evidence that argues against this (2)
- The inference the 2004 paper drew has been repeatedly confirmed structurally: the WDR5 contact is the WIN motif at approximately 3762-3773 with Arg3765 in the WDR5 pocket, roughly 2,400 residues past the breakpoint cluster region. So the conclusion is very likely right for the direct fusion even though the cited experiment does not establish it. Patel A et al. 2008
- The functional half of the 2004 claim is directly contradicted for WDR5 by later work: genetic deletion of Wdr5 impairs growth, colony formation and in vivo leukemogenesis of MLL-AF9 murine leukaemia cells. Whatever the physical association, WDR5 is not functionally irrelevant to a KMT2A-fusion-driven leukaemia. Liu L et al. 2023
- E2established confidence 5 of 5: Multiple independent lines of evidence; treated as settled in the field.paper
WDR5 binds the MLL1/KMT2A WDR5-interaction (WIN) motif, with Arg3765 of MLL1 occupying the WDR5 arginine pocket - a surface that lies inside the delta-3820 construct that nevertheless lost WDR5 binding in 2004.
Structure of WDR5 bound to mixed lineage leukemia protein-1 peptide. · Patel A, Dharmarajan V, Cosgrove MS · J Biol Chem 2008What this evidence cannot show
A peptide co-crystal structure at 1.7 A; it establishes the contact but says nothing about the stoichiometry or stability of the interaction in cells, and nothing about whether a WIN motif in an otherwise truncated protein would suffice for co-immunoprecipitation.
Evidence that argues against this (1)
- WDR5's WIN site is engaged by several partners besides KMT2A, including SETD1A and KMT2B, so occupancy of that pocket is not specific to the KMT2A axis and a WIN-directed reagent does not report on KMT2A alone. Ernst P et al. 2012
- E3established confidence 5 of 5: Multiple independent lines of evidence; treated as settled in the field.paper
DPY30 (hDPY-30) was identified as a component of the human Set1-like H3K4 methyltransferase complexes alongside ASH2L, RBBP5 and WDR5 in 2007, three years after the founding paper.
PTIP associates with MLL3- and MLL4-containing histone H3 lysine 4 methyltransferase complex. · Cho YW, Hong T, Hong S, et al. · J Biol Chem 2007What this evidence cannot show
This paper's primary subject is the MLL3/MLL4 complex and PTIP; it names hDPY-30 as a shared Set1-like complex component rather than characterising it. An earlier incidental report cannot be excluded, but it would still postdate 2004.
Evidence that argues against this (1)
- Naming a protein in a purified complex is not the same as establishing its stoichiometric membership; a systematic biophysical reconstitution of the MLL1 core complex shows the holocomplex assembles hierarchically and that subunit occupancy depends on temperature and component concentration, so complex membership is conditional rather than fixed. Namitz KEW et al. 2023
- E4established confidence 5 of 5: Multiple independent lines of evidence; treated as settled in the field.review
The acronym WRAD for the WDR5-RBBP5-ASH2L-DPY30 module was introduced in 2012, and the module is described there as the activating unit without which SET1-family enzymes are not fully active.
WRAD: enabler of the SET1-family of H3K4 methyltransferases. · Ernst P, Vakoc CR · Brief Funct Genomics 2012What this evidence cannot show
A review, used here only to date the terminology and to state the module's accepted composition, not as evidence for any experimental claim.
Evidence that argues against this (1)
- The module's activating role is genetically dispensable in the disease this graph is about: MLL1 methyltransferase activity is not required for haematopoiesis or for MLL-AF9-driven leukemogenesis in mouse. Mishra BP et al. 2014
- refutes →First systematic enumeration of the KMT2A fusion interactome, and the diff against this graph
Corrects the enumeration's characterisation: Yokoyama 2004 reports directly only for HCF-1; for ASH2L, WDR5 and RBBP5 it infers from the SET-domain-deleted construct delta-3820, and no WRAD subunit was blotted on either fusion protein.
- related to →WDR5 — the partner that is not a partner: it binds the half of KMT2A the fusion throws away
Evidence-class correction to this entry's E_yokoyama2004 block: the WRAD measurement in that paper is on a deletion mutant, not on a fusion.
Same evidence-class correction to E_yokoyama2004.
- related to →ASH2L — the WRAD subunit that has DPY30's essentiality profile and none of its co-dependency signal
Same evidence-class correction to E_yokoyama2004.
- refutes →DPY30 — the smallest COMPASS subunit, and the only one that behaves like a selective dependency
Scope error: Yokoyama 2004 does not mention DPY30 anywhere (zero occurrences of 'DPY' in the full text) and DPY30 was not placed in the human SET1-like complexes until 2007, so E_yokoyama2004 cannot carry a negative about DPY30.
- supports →Lost from the KMT2A half: the WIN motif and the whole WRAD interface (aa 3762-3767, Arg3765)
The WIN/WRAD interface is indeed absent from the direct fusion; this entry supplies the correct evidence class for that claim, which is structural rather than a measured fusion co-IP negative.
The verdict rests on the full-text reading of the founding paper recorded there.