The reciprocal fusion (AFF1::KMT2A and friends): a second oncoprotein that opens chromatin for the first
also called AF4-MLL · A4M · der(4) product · reciprocal MLL fusion · X-MLL fusion
A translocation makes two hybrid genes, not one: the famous KMT2A::AFF1 and its mirror image AFF1::KMT2A. The mirror-image protein is usually ignored, but there is evidence it pries chromatin open so the famous one can reach roughly ten times more target genes.
The same thing, for a scientist
A balanced t(4;11) produces both der(11) KMT2A::AFF1 (MA4) and der(4) AFF1::KMT2A (A4M). A4M retains the AFF1 N-terminal homology/ALF region fused to the KMT2A C-terminus (PHD fingers, bromodomain, FYRN/FYRC, TAD and the SET methyltransferase domain) [E1]. Marschalek's group showed that retrovirally expressed AF4-MLL alone was sufficient to cause pro-B ALL, biphenotypic B/T acute leukaemia or mixed-lineage leukaemia in mice transplanted with Lin-/Sca1+ progenitors, whereas MLL-AF4 alone produced no disease over 13 months [E2]. The current model from the same group is that the reciprocal protein acts as a chromatin-opening complex that expands the direct fusion's target-gene pool roughly tenfold, establishing the pre-leukaemic program, and may be dispensable once the oncogenic expression pattern is fixed [E1][E3]. Two clinical observations are consistent: t(4;11) pro-B ALL patients expressing both alleles reportedly do better than those expressing only the direct fusion [E1], and about 50% of KMT2A::MLLT1 cases (241/460) break upstream of MLLT1 exon 1, producing only a spliced direct fusion and no reciprocal product at all [E1]. Counter-evidence is substantial: AF4-MLL expressed in human cord-blood CD34+ HSPCs transiently boosted multilineage engraftment but did not initiate leukaemia within 36 weeks [E4]. Two harder bounds have since been added: in human CD34+ cells the DIRECT fusion alone, with no reciprocal present, produces a pro-B ALL that recapitulates the immunophenotypic and molecular features of t(4;11) [E3-ce], and in patients the reciprocal AF4-MLL transcript is expressed in only about half of t(4;11) cases [E5-ce].
Why a mirror-image protein could matter more than it looks2 deeper
Two things make A4M unusual among reciprocal fusion products.
It is not a nonsense fragment. It carries the KMT2A C-terminus: PHD1-3 (an H3K4me3 reader), the bromodomain, the FYRN/FYRC pair, the transactivation domain that binds CREBBP, and the SET H3K4 methyltransferase domain. Marschalek describes reciprocal X-MLL fusions as sequestering the histone acetyltransferases CREBBP and MOF and carrying a histone methyltransferase domain at their C-terminus 15. That is a chromatin-writing machine with no menin-binding module and no CXXC.
It is transcribed even when out of frame. KMT2A carries a gene-internal promoter upstream of exon 12, so most out-of-frame reciprocal alleles still express a 5'-truncated KMT2A protein ('KMT2A*') that lacks menin, LEDGF and MYB binding but keeps the enzymatic C-terminus and reportedly has oncogenic potential in a focus-formation assay 1. In the DCAL cohort, 426 patients had complex rearrangements, of which 40 reciprocal fusions were in-frame and 386 were out-of-frame or non-functional at the DNA level - yet most of those 386 can still express KMT2A* 1. So 'the reciprocal allele is out of frame' does not mean 'nothing is made'.
The tenfold target-gene expansion claim
Marschalek's group reports that co-expression of the reciprocal protein (AFF1::KMT2A or AFDN::KMT2A) allows the corresponding direct KMT2A fusion to activate roughly ten times more target genes, with the deregulated gene set evolving over time under selection toward the final oncogenic signature 1. A 2026 follow-up using engineered model systems reports the more nuanced finding that MA4 alone drives aberrantly high expression of only a few target genes and that A4M actually counteracts that excessive activation, with both reciprocal transcripts bound by IGF2BP3, which lowers MA4 protein abundance and raises A4M levels 3. The authors still conclude A4M is crucial through its chromatin-opening activity. The direction of the effect on individual genes is therefore not settled, but both papers agree the reciprocal protein is not a bystander.
Does loss of the reciprocal matter clinically?
Meyer et al. report that two publications found better outcomes for t(4;11)/pro-B ALL patients when both the direct and reciprocal fusion alleles were expressed, and offer the interpretation that a reciprocal-driven, broader transcriptome exposes more druggable targets and so responds better to chemotherapy, whereas a restricted transcriptome without the reciprocal may be more resistant 1. We were unable to independently verify the two primary clinical papers behind that statement within this cycle, so this is recorded at the confidence of the recombinome paper's own assertion and flagged as needing verification. The structurally solid version of the claim is the KMT2A::MLLT1 case: 241 of 460 KMT2A::MLLT1 patients break upstream of MLLT1 exon 1, generating a spliced fusion with no reciprocal transcript at all 1 - a natural experiment in reciprocal-free KMT2A-rearranged leukaemia that is fully leukaemogenic. (See the correction section below: this entry previously said MLLT10.)
A second, larger reciprocal-free population weakens the clinical claim further. In t(4;11) the direct MLL-AF4 transcript is expressed in every patient but the reciprocal AF4-MLL transcript in only about half, and those patients carry an intact der(4) rather than a spliced fusion 5. If expressing both alleles really predicted better chemotherapy response, that contrast should be visible in essentially every t(4;11) cohort ever assembled, and it has not been reported as such.
The honest counter-case
The strongest challenge to reciprocal-protein importance is species-specific. Prieto et al. expressed AF4-MLL in human cord-blood CD34+ HSPCs and transplanted immunodeficient mice: the cells gave enhanced long-term multilineage reconstitution in primary recipients but had no additional self-renewal on serial transplant, and no phenotypic or haematological sign of disease appeared within 36 weeks 4. In other words, the mouse-progenitor result 2 did not transfer to human cells. Combined with the fact that half of KMT2A::MLLT1 leukaemias have no reciprocal product at all 1, that about half of t(4;11) patients express no reciprocal transcript despite an intact der(4) 5, and that MLL fused to Af4 ALONE in human CD34+ cells reproduces t(4;11) pro-B ALL with its immunophenotypic and molecular features 3, the defensible position is: the reciprocal protein is a modifier of the direct fusion's reach, sufficient in some murine contexts, unnecessary in some human contexts, and not a universal requirement.
Correction, 2026-09-19: the reciprocal-free partner is MLLT1, not MLLT10
This entry previously asserted, in its summary, in evidence item E1 and in E1's own counter-evidence note, that '~50% of KMT2A::MLLT10 cases break upstream of MLLT10 exon 1' and that 'roughly half of all KMT2A::MLLT10 leukaemias express no reciprocal product'. The partner is MLLT1 (ENL). The primary text reads: 'Spliced fusions are rare events except for MLLT1 gene fusions (n = 460), where about 50% of the breakpoints (n = 241) localize upstream of MLLT1 exon 1', and 'In these cases, only KMT2A::MLLT1 fusion transcripts but no reciprocal one are expressed' 1. MLLT10 contributes 80 of 2,182 ALL cases in the same cohort, so a denominator of 460 cannot be MLLT10's. The sibling entry breakpoint-kmt2a-spliced-fusion was corrected in an earlier cycle and this one was not, so for a cycle the graph carried both readings.
The substance of the argument is unchanged - a large, constitutively reciprocal-free population of KMT2A leukaemias exists - but the partner identity matters, because MLLT1 is a YEATS-domain reader that engages the super elongation complex and DOT1L directly, and MLLT10 is an octapeptide/leucine-zipper DOT1L recruiter. Which one the reciprocal-free cases carry changes what the comparison controls for.
| mechanism class | reciprocal fusion protein / chromatin opening / pre-leukaemic priming |
|---|---|
| direct product | KMT2A::AFF1 (MA4), der(11) |
| reciprocal product | AFF1::KMT2A (A4M), der(4) |
| kmt2a internal promoter | upstream of KMT2A exon 12; drives expression of 5'-truncated KMT2A* from out-of-frame reciprocal alleles |
| reciprocal absent natural experiment | KMT2A::MLLT1: 241 of 460 cases break upstream of MLLT1 exon 1, giving a spliced direct fusion and no reciprocal transcript. A second and larger reciprocal-free population exists independently of promoter capture: the reciprocal AF4-MLL transcript is expressed in only about half of t(4;11) patients (PMID 30679325). |
| spliced fusion total cases | 275 |
| clinical correlate | Better reported outcome in t(4;11)/pro-B ALL when both direct and reciprocal alleles are expressed (asserted in the recombinome Discussion; primary sources unverified in this cycle) |
reciprocal domain content
[ "AFF1 N-terminal homology/ALF region", "KMT2A PHD1-3", "KMT2A bromodomain", "KMT2A ePHD4", "KMT2A FYRN/FYRC", "KMT2A TAD (CREBBP-binding)", "KMT2A WIN motif", "KMT2A SET" ]
reciprocal lacks
[ "menin-binding motifs MBM1/MBM2", "CXXC", "AT-hooks", "LEDGF-binding region" ]
complex rearrangement counts — the numbers behind the chart
{
"total_complex": 426,
"reciprocal_in_frame": 40,
"reciprocal_out_of_frame_or_nonfunctional": 386,
"loci": 167,
"partner_genes": 219
}- **This is the clearest threat to the menin-interface hypothesis in this beat.** The reciprocal fusion protein has no menin-binding motif, no CXXC and no AT-hooks, yet carries the SET methyltransferase, the CREBBP/MOF-sequestering TAD and a PHD H3K4me3 reader. A menin inhibitor is structurally blind to it. If the reciprocal product contributes to chromatin state or to maintenance in any subset, menin inhibition leaves half the lesion untouched.
- The gene-internal KMT2A promoter upstream of exon 12 means a KMT2A* protein with full enzymatic capacity but no menin/LEDGF/MYB binding can be expressed from most out-of-frame reciprocal alleles - i.e. a menin-inhibitor-invisible KMT2A species is present in a large fraction of complex-rearrangement patients by default.
- If patients expressing both alleles do better on chemotherapy because a broader transcriptome exposes more druggable targets, then a therapy that narrows the transcriptome (such as effective menin inhibition) could paradoxically select for the restricted, reciprocal-negative, chemo-resistant state.
- Do KMT2A::MLLT1 spliced fusions (no reciprocal) differ in menin-inhibitor sensitivity, HOXA signature or outcome from reciprocal-expressing KMT2A::MLLT1? This is a ready-made, adequately powered natural experiment inside the existing DCAL cohort (about 241 versus 219 patients), and holding the partner constant is what makes it interpretable.
- Is the reciprocal protein required for maintenance or only for onset? If only for onset, it is a prevention/early-detection target, not a treatment target - a completely different translational path.
- The reciprocal protein carries the SET domain and the CREBBP-binding TAD but no menin-binding motif. If it is functionally important, is the leukaemia partly menin-independent by construction - and is that a resistance reservoir for menin inhibitors?
- Does IGF2BP3 inhibition shift the MA4:A4M protein ratio enough to change the phenotype, and is IGF2BP3 therefore an indirect handle on the fusion pair?
- What silences the reciprocal AF4-MLL allele in the roughly half of t(4;11) patients who carry an intact der(4) but express no reciprocal transcript? That population is larger and less selected than the spliced fusions and nobody has reported what distinguishes it.
- E1moderate confidence 3 of 5: One strong study, or several smaller consistent ones.paper
Reciprocal KMT2A fusion proteins act as chromatin-opening complexes that expand the direct fusion's target gene set roughly tenfold; KMT2A has a gene-internal promoter upstream of exon 12 so out-of-frame reciprocal alleles still express a 5'-truncated KMT2A* protein; 241/460 KMT2A::MLLT1 cases break upstream of MLLT1 exon 1 and express no reciprocal transcript.
The KMT2A recombinome of acute leukemias in 2023. · Meyer C, Larghero P, Almeida Lopes B, et al. · Leukemia 2023What this evidence cannot show
The tenfold-target-expansion and better-outcome-with-reciprocal statements appear in the Discussion as citations to the authors' own prior work; we could not resolve and verify those primary references in this cycle, so they are carried at review-statement confidence. The MLLT10 spliced-fusion counts are primary data from this cohort and are firmer. CORRECTION 2026-09-19: this claim and the counter-evidence note beneath it previously said MLLT10 and '241/460 KMT2A::MLLT10'. The primary text reads: 'Spliced fusions are rare events except for MLLT1 gene fusions (n = 460), where about 50% of the breakpoints (n = 241) localize upstream of MLLT1 exon 1.' MLLT10 in that cohort has 80 ALL cases, not 460. The partner is MLLT1 (ENL).
Evidence that argues against this (1)
- Roughly half of all KMT2A::MLLT1 leukaemias express no reciprocal product whatsoever yet are fully leukaemogenic, which means the reciprocal protein cannot be generally required. Meyer C et al. 2023
- E2moderate confidence 3 of 5: One strong study, or several smaller consistent ones.paper
AF4.MLL (AFF1::KMT2A) expressed in murine Lin-/Sca1+ progenitors induced pro-B ALL, B/T biphenotypic acute leukaemia or mixed-lineage leukaemia, whereas MLL.AF4 alone caused no disease over 13 months.
The AF4.MLL fusion protein is capable of inducing ALL in mice without requirement of MLL.AF4 · Bursen A, Schwabe K, Ruster B, Henschler R, Ruthardt M, Dingermann T, Marschalek R · Blood 2010What this evidence cannot show
Retroviral overexpression at non-physiological levels in murine progenitors; the negative MLL-AF4 arm is a well-known artefact of retroviral MLL-AF4 models generally, so 'A4M works and MA4 does not' partly reflects the difficulty of modelling MLL-AF4 rather than a true ranking. Single laboratory.
Evidence that argues against this (1)
- The same fusion expressed in human cord-blood CD34+ HSPCs failed to initiate leukaemia within 36 weeks, indicating the murine result is species-dependent. Prieto C et al. 2017
- E3weak confidence 2 of 5: Preliminary, small, indirect, or not yet peer reviewed.paper
In engineered model systems, KMT2A::AFF1 alone drives aberrantly high expression of only a few target genes while AFF1::KMT2A counteracts that effect; both reciprocal transcripts are bound by IGF2BP3, which reduces KMT2A::AFF1 protein abundance and increases AFF1::KMT2A levels.
Influence of IGF2BP3 on t(4;11) fusion mRNA and protein abundance · Hanewald T, Benz T, Marschalek R · Leukemia Research 2026What this evidence cannot show
Engineered cell-line model systems, single laboratory, very recent and not independently replicated. Note the finding that A4M counteracts rather than amplifies MA4 target-gene activation partially contradicts the simpler 'tenfold expansion' narrative from the same group.
Evidence that argues against this (1)
- The direct fusion needs no reciprocal partner to build the full disease. MLL fused to Af4 and delivered alone into human CD34+ cells - no reciprocal AF4-MLL anywhere in the system - generates a pro-B acute lymphoblastic leukaemia that recapitulates the immunophenotypic and molecular features of t(4;11), with fusion-specific genomic binding and gene expression distinct from MLL-AF9. A model in which the reciprocal is what expands the target-gene pool tenfold has to explain why its absence costs nothing measurable here. Lin S et al. 2016
- E4moderate confidence 3 of 5: One strong study, or several smaller consistent ones.paper
AF4-MLL expressed in human cord-blood CD34+ HSPCs enhanced long-term multilineage engraftment in primary immunodeficient recipients but conferred no serial self-renewal and initiated no leukaemia within 36 weeks.
The AF4-MLL fusion transiently augments multilineage hematopoietic engraftment but is not sufficient to initiate leukemia in cord blood CD34+ cells · Prieto C, Marschalek R, Kuhn A, Bursen A, Bueno C, Menendez P · Oncotarget 2017What this evidence cannot show
A negative result over a 36-week window in a xenograft; infant t(4;11) may need a fetal rather than neonatal cell of origin, and cord-blood CD34+ cells may be the wrong target cell. Absence of leukaemia is not proof the protein is inert.
Evidence that argues against this (1)
- A negative retroviral result in human cells is confounded by a known technical problem with human AF4 constructs. The group that built the working human model reported that they had to use MURINE Af4, 'highly conserved with human AF4', because that is what 'produces high-titer retrovirus permitting efficient transduction of human CD34 cells'. An AF4-MLL construct carrying the human AF4 N-terminus is subject to the same constraint, so failure to initiate leukaemia in cord-blood CD34+ cells over 36 weeks may report construct expression rather than protein function. We flag this as an inference: the cord-blood paper's own titres were not compared with the murine-Af4 ones. Lin S et al. 2016
- E5moderate confidence 3 of 5: One strong study, or several smaller consistent ones.review
Reciprocal X-MLL fusion proteins exhibit a PHD H3K4me3-reader domain, sequester the histone acetyltransferases CREBBP and MOF, and carry the SET histone methyltransferase domain at their C-terminus.
MLL leukemia and future treatment strategies · Marschalek R · Archiv der Pharmazie 2015What this evidence cannot show
Review by the group that generated most of the reciprocal-fusion literature; the review itself states that except for AF4-MLL the functional consequences of reciprocal fusion proteins are not well understood.
Evidence that argues against this (1)
- Half of the patients this domain architecture is supposed to matter in do not express the protein at all. The direct MLL-AF4 fusion is expressed in every t(4;11) B-ALL patient, but the reciprocal AF4-MLL is expressed in only about half of them - and these are patients with a structurally intact der(4), not spliced fusions. A reciprocal protein whose PHD reader, CREBBP/MOF sequestration and SET domain are decisive for leukaemogenesis should not be dispensable in half of the disease it is invoked to explain. Bueno C et al. 2019
- E6moderate confidence 3 of 5: One strong study, or several smaller consistent ones.paper
A CRISPR/Cas9-engineered t(4;11) in primary human haematopoietic stem/precursor cells, in which both reciprocal fusion proteins are concomitantly present, shows plasticity to differentiate into either the myeloid or lymphoid lineage.
CRISPR/Cas9-mediated t(4;11) translocation in human hematopoietic stem/precursor cells demonstrates plasticity to differentiate into either the myeloid or lymphoid lineage · Benz T, Larghero P, Meyer C, Hanewald T, Bruggmann D, Hentrich AE, Louwen F, Marschalek R · Leukemia 2025What this evidence cannot show
In vitro expansion over 5-6 months with IL-7 supplementation; no leukaemia readout in vivo, so this establishes lineage plasticity of the translocated cell, not transformation. Single laboratory, very recent.
Evidence that argues against this (1)
- Lineage plasticity of a translocated cell does not predict what leukaemia, if any, follows. A germline Mll-AF4 knock-in mouse showed exactly this mixed lymphoid/myeloid deregulation early, with bone-marrow cells forming unique mixed pro-B/myeloid colonies in IL-7 and Flt3L - and then developed mostly B-cell LYMPHOMAS after prolonged latency, leading the authors to state that Mll-AF4-induced lymphoid/myeloid deregulation alone is insufficient to produce malignancy. An in vitro plasticity readout at 5-6 months is therefore compatible with no leukaemia and with the wrong disease. Chen W et al. 2006
Counter-evidence to the reciprocal-required model: MLL fused to Af4 ALONE in human CD34+ cells generates a pro-B ALL recapitulating the immunophenotypic and molecular features of t(4;11), with no reciprocal present (Lin 2016, PMID 27846391); and in patients the reciprocal AF4-MLL transcript is expressed in only about half of t(4;11) cases (Bueno 2019, PMID 30679325).
Two hard bounds added this cycle: the direct fusion alone suffices in human cells (PMID 27846391), and half of t(4;11) patients express no reciprocal transcript despite an intact der(4) (PMID 30679325). The defensible position is that the reciprocal is a modifier, not a requirement.
Both entries carried the same MLLT1/MLLT10 spliced-fusion error; mech-breakpoint-position-consequences was corrected earlier and this one was not. Corrected here 2026-09-19 in summary, E1, E1's counter-note, the data payload and the open questions.
- ← related toSpliced KMT2A fusions join to partner exon 2 — with the per-partner counts, and a correction
the 650-fold MLLT1-vs-AFF1 spliced-fusion rate difference bears directly on whether the reciprocal product is partner-specifically required
loss of the reciprocal is a structural second event invisible to mutation counts
- related to →KMT2A::AFF1 (MLL-AF4), t(4;11)(q21;q23)
the second oncoprotein made by the same translocation
- depends on →Spliced KMT2A fusions: breaks upstream of the partner's exon 1, and the reciprocal-free natural experiment
only exists when the break leaves a reciprocal allele
Supplies the temporal half of the chromatin-opening account: 48 hours of expression, 28 days of effect.
Supplies the arm counts behind that entry's proposed natural experiment.
Adds the WRAD consequence the existing entry did not draw: because the reciprocal product carries the WIN motif and SET domain, it is a KMT2A fusion protein with an intact WRAD interface.