Anti-AFF1 antibodies: attributable on a gel, not attributable in a microscope, and the classic localisation result shows exactly why
also called anti-AF4 antibody · AF4(C) ChIP · AF4 antibody specificity in t(4;11)
Antibodies against AF4, the KMT2A partner in t(4;11), see the normal AF4 protein and the fusion at the same time. On a gel that is fine because the two run at different sizes; under a microscope it is not, and the 1998 paper that established where the fusion sits in the nucleus shows the same speckled pattern in cells that have no fusion at all.
The same thing, for a scientist
AFF1 is expressed ubiquitously, so in a t(4;11) cell an anti-AFF1 antibody raised against retained (C-terminal) sequence binds three proteins: wild-type AFF1 (reported as 125 and 145 kDa species), the direct KMT2A::AFF1 fusion (~240 kDa), and - where an in-frame reciprocal allele exists - nothing, because the reciprocal product carries the AFF1 **N**-terminus, not the C-terminus. The classic characterisation is explicit about both halves of this [E1]: the antibodies detect the anticipated 240 kDa MLL-AF4 protein in all t(4;11) lines and the fusion is present in **lesser quantity than wild-type AF4** - an attributable, quantitative western result - and in the same paper the immunofluorescence shows *"a similar pattern of punctate nuclear staining in all cell lines tested"* in **both t(4;11) and non-t(4;11) leukaemic cells**. The microscopy therefore reports the distribution of the antibody's targets collectively, and the widely cited statement that the KMT2A::AFF1 fusion localises to subnuclear speckles is, on the evidence in that paper, a statement about wild-type AF4 that the fusion does not visibly perturb. **The size dimension is the whole of the attribution**, and it is present in a blot and absent in microscopy, immunoprecipitation, ChIP and flow cytometry.
The flagship occupancy dataset and what rescues it
The canonical KMT2A::AFF1 chromatin map - 4,427 binding sites in SEM cells, the basis of the spreading model and of the downstream gene-regulatory network - was generated with MLL(N) and AF4(C) ChIP-seq 2. Both antibodies are shared-epitope reagents, so on reagent grounds alone the peaks could be wild-type KMT2A, wild-type AFF1, or the fusion.
What rescues it is not the antibody but a loss-of-function control: the authors performed MLL-AF4 siRNA knockdown coupled to ChIP-seq and report loss of signal at 85% of target gene promoters. If the siRNA is fusion-specific, that is a genuine attribution - the signal that disappears when the fusion transcript is removed was fusion-derived - and it is one of the few genome-wide attributions in this literature. Two caveats keep it from being airtight. First, this beat could not verify from the retrievable text that the siRNA spans the junction rather than targeting shared KMT2A or AFF1 sequence; the sequences sit in supplemental procedures. An siRNA against shared sequence would deplete the wild-type parent as well and the control would be uninformative. Second, the fusion-specific siRNA in general use in this field was introduced in a paper whose full text is not open 3.
The practical instruction is narrow and useful: when reading any 'MLL-AF4 binding' claim, look for the knockdown-coupled ChIP, and then look at what the knockdown reagent targets.
What an anti-AFF1 antibody cannot do, stated positively
- It cannot assign a subnuclear localisation to the fusion, because the wild-type protein is in the same compartment and, by the only published test, in the same pattern in cells without a fusion.
- It cannot assign a complex composition to the fusion, because an AF4-C immunoprecipitate contains wild-type AFF1 in excess of the fusion, and wild-type AFF1 is itself a bona fide SEC component with its own interactome.
- It cannot assign chromatin occupancy without a fusion-specific loss-of-function control, for the same reason.
- It can measure fusion protein abundance relative to wild-type on a blot, and it can serve as the detection arm of a sandwich assay whose capture arm is an anti-KMT2A-N antibody, which is what makes the immunobead assay fusion-specific.
The last point is the constructive one: two shared-epitope antibodies used in series are fusion-specific even though neither is fusion-specific alone.
| reagent class | antibodies against AFF1/AF4 C-terminal (retained) sequence |
|---|---|
| fusion less abundant than wild type partner | true |
| canonical dataset | 4,427 KMT2A::AFF1 binding sites in SEM (PMID 28076791) |
species recognised
[ "wild-type AFF1 (125 and 145 kDa reported species)", "KMT2A::AFF1 direct fusion (~240 kDa)" ]
species not recognised
[ "AFF1::KMT2A reciprocal fusion (retains the AFF1 N-terminus, not the C-terminus)" ]
attributable in
[ "western blot (size-resolved)", "sandwich immunoassay with an anti-KMT2A-N capture arm", "ChIP when paired with a fusion-specific loss-of-function control" ]
not attributable in
[ "immunofluorescence / subnuclear localisation", "immunoprecipitation for complex composition", "ChIP without a fusion-specific knockdown" ]
- If the punctate nuclear pattern attributed to the fusion is wild-type AFF1's pattern, then condensate and phase-separation arguments built on fusion localisation in t(4;11) cells are built on a wild-type observation - and this graph already records that hexanediol and FRAP cannot distinguish phase separation from clustered binding in any case.
- A drug screened on 'fusion relocalisation' as a readout would be scored on the behaviour of wild-type AFF1, which is present in every cell in the body.
- Is the MLL-AF4 siRNA used as the attribution control in the canonical ChIP-seq study junction-spanning? The whole attributive force of the field's flagship occupancy dataset turns on this, and the answer sits in a supplemental file.
- Does the KMT2A::AFF1 fusion occupy a different subnuclear compartment from wild-type AFF1? The only published microscopy cannot answer it, and the experiment that could - endogenous tagging of the fusion allele - has never been done in any KMT2A-rearranged cell.
- E1strong confidence 4 of 5: Replicated in peer-reviewed work by more than one group.paper
Polyclonal antibodies against AF4 detect wild-type AF4 at 125 and 145 kDa and the MLL-AF4 fusion at 240 kDa in all t(4;11) lines tested, with the fusion present in lesser quantity than wild-type AF4; by confocal immunofluorescence both t(4;11) and non-t(4;11) leukaemic cells show a similar punctate nuclear pattern.
AF4 encodes a ubiquitous protein that in both native and MLL-AF4 fusion types localizes to subnuclear compartments. · Li Q, Frestedt JL, Kersey JH · Blood 1998What this evidence cannot show
Abstract read verbatim 2026-09-20; full text not retrieved. Rabbit polyclonals raised against GST fusion immunogens, 1998 reagents, semi-quantitative densitometry. The paper's own framing is that the fusion 'had a similar subcellular distribution as AF4' - the attribution point made here is a reading of its control, not a claim the authors dispute.
Evidence that argues against this (1)
- The paper's western blot does distinguish the species by size and does detect the fusion, so 'anti-AFF1 antibodies cannot see the fusion' would be wrong. The limitation is assay-specific, not reagent-specific, and the same antibody is attributable in one experiment in the paper and not in another. Li Q et al. 1998
- E2strong confidence 4 of 5: Replicated in peer-reviewed work by more than one group.paper
The canonical KMT2A::AFF1 binding-site set in SEM cells was defined by MLL(N) and AF4(C) ChIP-seq, with attribution supplied by an MLL-AF4 siRNA knockdown coupled to ChIP-seq that removed signal at 85% of target gene promoters.
MLL-AF4 Spreading Identifies Binding Sites that Are Distinct from Super-Enhancers and that Govern Sensitivity to DOT1L Inhibition in Leukemia. · Kerry J, Godfrey L, Repapi E, et al. · Cell Rep 2017What this evidence cannot show
Read in full via PMC5263239. The siRNA sequences are in Supplemental Experimental Procedures, which this beat did not retrieve, so whether the knockdown reagent is junction-spanning was not verified. If it targets shared sequence the control loses its attributive force, because wild-type KMT2A or AFF1 would be depleted alongside the fusion.
Evidence that argues against this (1)
- Knocking down the fusion could reduce ChIP signal indirectly - by closing chromatin, by removing a cooperating factor, or by killing the cells over the knockdown window - so signal loss on fusion depletion is consistent with, but does not prove, direct occupancy by the fusion protein. Thomas M et al. 2005
- E3strong confidence 4 of 5: Replicated in peer-reviewed work by more than one group.paper
siRNA depletion of MLL-AF4 in SEM and RS4;11 impairs proliferation and clonogenicity, induces apoptosis, and reduces HOXA7, HOXA9 and MEIS1 expression, establishing the reagent used as the loss-of-function control in later ChIP studies.
Targeting MLL-AF4 with short interfering RNAs inhibits clonogenicity and engraftment of t(4;11)-positive human leukemic cells. · Thomas M, Gessner A, Vornlocher HP, et al. · Blood 2005What this evidence cannot show
Abstract only. Europe PMC returns no PMCID for this record (isOpenAccess: N), so routes 2-4 of the unreadable test are unavailable and the siRNA target sequence could not be read. The abstract does not state whether the siRNAs span the fusion junction. Anyone relying on the attribution in E2 should obtain this paper.
Evidence that argues against this (1)
- Because the fusion is strongly required for survival in these lines, any knockdown-coupled assay is measuring a dying cell at later timepoints; the specificity of the reagent and the specificity of the phenotype are separable questions and only the second is established by this paper. Kumar AR et al. 2011
the wild-type protein whose signal is pooled with the fusion's by every C-terminal AFF1 reagent
- related to →KMT2A::AFF1 (MLL-AF4), t(4;11)(q21;q23)
detected at 240 kDa on a blot, unattributable in microscopy and in IP
the two shared-epitope reagents used in series are fusion-specific even though neither is alone
the reagent property that drives the verdict is established there
the anti-AF4 reagent class is what makes the fractionation lane self-controlled