The PBX3-MEIS1 dimer, not HOXA9, is the unit that reproduces the MLL-fusion transcriptome - and it turns the HOXA cluster on from underneath
also called PBX3/MEIS1 cooperation · TALE-TALE dimer · MEIS1-PBX3 interface
Everyone writes that KMT2A-fusion leukemia runs on HOXA9 plus its helper MEIS1. When a group actually tested the combinations, the pair that caused leukemia fastest and looked most like the real disease was MEIS1 plus PBX3 - with no HOXA9 added at all. The helpers turned the HOXA genes on by themselves.
The same thing, for a scientist
The field's shorthand, HOXA9/MEIS1, encodes an assumption about which node is upstream. Two independent murine studies argue it is wrong in a specific and testable way. (1) Co-expressing PBX3 and MEIS1 in murine HSPCs, with no ectopic HOX gene, transforms cells in vitro and causes AML in vivo with a latency similar to MLL-AF9, and - uniquely among the pairs tested - recapitulates the MLL-fusion core transcriptome including upregulation of the endogenous *Hoxa* genes. HOXA9/MEIS1, HOXA9/PBX3 and HOXA9 alone do not [E1]. (2) Pbx3 and Meis1 must dimerise: without Pbx3, Meis1 protein is degraded by ubiquitination through a motif coinciding with the Pbx-binding domain; Pbx3 also induces endogenous *Meis1* transcription; and Pbx3/Meis1 dimers are rate-limiting for Hoxa9-induced leukaemia [E2]. Together these put the two TALE proteins upstream of the HOX protein rather than beside it. **Both are murine and myeloid, neither is replicated in a B-lineage or human system, and the human data point the other way** [E3].
The combinatorial table, which is the whole argument
From the one experiment that varied the combination systematically in murine haematopoietic stem/progenitor cells 1:
| construct | transforms HSPC in vitro | AML in vivo | reproduces the MLL-fusion core transcriptome |
|---|---|---|---|
| HOXA9 alone | no | no | no |
| HOXA9 + MEIS1 | yes | yes, rapid | no |
| HOXA9 + PBX3 | - | - | no |
| PBX3 + MEIS1 (no ectopic HOX) | yes | yes, latency similar to MLL-AF9 | yes, incl. endogenous Hoxa upregulation |
The transcriptome column is the one that matters. HOXA9/MEIS1 transforms and causes AML - nobody disputes that - but it does not produce the gene-expression programme of the disease. PBX3/MEIS1 does, and it does so by switching the endogenous Hoxa cluster on.
Disrupting the MEIS1-PBX3 binding diminished both the transformation and the HOX upregulation, which ties the phenotype to the interface rather than to either protein's abundance.
The wiring that makes it more than a curiosity
An independent laboratory, working on murine Hoxa9-transformed cells, found the biochemistry that would make PBX3 upstream of MEIS1 rather than parallel to it 2:
- Protein stability. In the absence of Pbx3, Meis1 is highly unstable. Degradation depends on a motif coinciding with the Pbx-binding domain. Deleting that motif, or binding Pbx3, prolongs Meis1's half-life by preventing ubiquitination; proteasome inhibition blocks the breakdown independently.
- Transcription. Overexpressing Pbx3 induces endogenous Meis1 transcription.
- DNA binding. Meis1 must bind Pbx3 to form high-affinity DNA/Hoxa9/Meis1/Pbx3 complexes in vitro.
- Rate limitation. Kinetics and severity of disease in transplantation indicate that Pbx3/Meis1 dimers are rate-limiting for Hoxa9-induced leukaemia, and blocking dimerisation abrogated proliferation and colony formation in Hoxa9-transformed primary cells, with the Meis1 targets Flt3 and Trib2 dependent on the dimer.
So one interface does three jobs: it holds MEIS1 up against the proteasome, it raises MEIS1 transcript, and it is required for high-affinity ternary DNA binding. A fourth job is predicted but not measured here - family-level work shows MEINOX-class binding is what retains PBC-class proteins in the nucleus by blocking export (see gene-pbx3, section on cell biology).
An interface with four separable functions is a much better target than a transcription factor, and it is a different interface from the one the only existing chemical matter attacks.
What this does NOT establish, stated plainly
Four limits, and the third is the serious one.
- Sufficiency is not necessity. Showing PBX3/MEIS1 transforms without ectopic HOXA9 does not show HOXA9 is dispensable. The construct upregulates the endogenous Hoxa cluster, so the transformed cells have endogenous Hoxa9. A cleaner test is PBX3/MEIS1 transformation in Hoxa9-null cells, and it was not found this cycle.
- Species and lineage. Both studies are murine, and the readouts are myeloid. The index disease is infant B-lineage KMT2A::AFF1. Rule 73 applies.
- The human data point the other way. In human MLL-rearranged leukaemia, HOXA9 suppression kills MLL-rearranged and not MLL-germline cells, and co-down-regulates PBX3 among other genes 3. That places PBX3 downstream of HOXA9. This graph holds both directions and the likeliest resolution is a feed-forward loop - HOXA9 sustains PBX3 transcript, PBX3 sustains MEIS1 protein, the trimer sustains the HOXA cluster - in which asking which is upstream is the wrong question and asking which node breaks the loop is the right one.
- Neither study is independent of its own laboratory's prior. 1 comes from the group that proposed PBX3 as the critical cofactor in the first place. 2 is independent of that group, which is what makes the convergence worth anything at all.
Why this changes what to aim at
If the load-bearing unit is the MEIS-PBX interface rather than the HOX-PBX interface, then the only peptide that exists is aimed at the wrong contact.
HXR9 and its successor HTL001 are built on the conserved HOX hexapeptide and antagonise HOX/PBX binding (target-hox-pbx-interface-peptide-hxr9). That is the contact 1 shows is not sufficient to reproduce the disease transcriptome, and the contact 4 shows is redundant with homeodomain residues on the HOXA9 side.
The MEIS-PBX contact, by contrast, is the one 1 disrupted to abolish the phenotype and the one 2 shows is rate-limiting. No chemical matter against it was found this cycle - a bounded absence recorded in hyp-meis1-pbx3-interface-is-the-unattempted-target.
| mechanism class | transcription-factor cofactor assembly; protein stability control; feed-forward transcriptional loop |
|---|---|
| the unit | PBX3-MEIS1 heterodimer (PBC-class TALE plus MEINOX-class TALE) |
| what is NOT the unit | the HOX-PBX hexapeptide contact, which is the only contact with chemical matter against it and which HOXA9 can bypass using paralog-specific homeodomain residues |
| index disease coverage | NONE of the four is infant B-lineage KMT2A::AFF1 in human cells or patients. The closest is E3's SEMK2 xenograft arm, and that arm is about HOXA9 rather than about the dimer. |
| unresolved direction | E1/E2 put PBX3-MEIS1 upstream of the HOXA cluster; E3 puts PBX3 downstream of HOXA9. Both are recorded. A feed-forward loop reconciles them and has not been tested. |
functions of the one interface
[ "protects MEIS1 protein from ubiquitin-mediated degradation (motif coincides with the Pbx-binding domain)", "required for high-affinity ternary DNA/HOXA9/MEIS1/PBX3 complex formation in vitro", "PBX3 induces endogenous MEIS1 transcription, so the interface is reinforced transcriptionally as well", "PREDICTED, NOT MEASURED HERE: MEINOX-class binding retains PBC-class protein in the nucleus by blocking export (family-level, PBX1/PREP1)" ]
systems
{
"E1": "murine HSPC, retroviral overexpression, myeloid AML readout",
"E2": "murine Hoxa9-transformed primary cells and transplantation, myeloid",
"E3": "human AML/ALL lines plus t(4;11) SEMK2 xenograft, B-lineage for the xenograft arm",
"E4": "human proteins in heterologous cell lines and in vitro DNA binding"
}- This mechanism argues for targeting a transcription-factor-transcription-factor interface with no enzyme and no published chemical matter. The honest status is that it identifies a better target than the one being pursued and supplies nothing to pursue it with.
- If PBX3 works by stabilising MEIS1, an interface inhibitor acts at the rate of MEIS1 protein turnover. Pharmacodynamic lag is a real failure mode in a disease that relapses in weeks.
- MEIS1 and PBX3 are both required for normal haematopoiesis and for normal development. An interface inhibitor does not distinguish the leukaemic dimer from the physiological one, and the patients are infants.
- The whole argument rests on murine myeloid overexpression experiments. If the direction is reversed in B-lineage - as the one human dataset in this entry suggests - then the interface is downstream of HOXA9 and a menin inhibitor already reaches it more cheaply.
- Neither PBX3 nor MEIS1 is a dependency in any human leukaemia line in DepMap (PBX3 +0.0418, MEIS1 -0.1791 across 57 leukaemia lines), so there is no in vitro potency assay to optimise an interface inhibitor against.
- Does PBX3/MEIS1 transform Hoxa9-null cells? This is the one experiment that would turn a sufficiency claim into a necessity claim, and it has apparently never been done.
- Does any of this hold in B-lineage? Every positive result in this entry is myeloid. A PBX3/MEIS1 co-transduction of human cord-blood CD34+ cells under B-lineage conditions, or in a KMT2A::AFF1 background, is the obvious first test.
- Is MEIS1 protein stabilised by PBX3 in human KMT2A-rearranged cells? A cycloheximide chase with and without PBX3 knockdown in SEM and RS4;11 costs a week and would establish whether the murine stability mechanism transfers.
- What was the reagent that blocked Meis1/Pbx3 dimerisation in E2, and can it be used as a tool compound? The abstract does not say, and if it is a peptide or a point mutant it is the starting point for the interface programme this beat argues for.
- E1strong confidence 4 of 5: Replicated in peer-reviewed work by more than one group.paper
In murine haematopoietic stem/progenitor cells, co-expression of PBX3 and MEIS1 without ectopic expression of a HOX gene is sufficient for transformation in vitro and causes AML in vivo with a leukaemic latency similar to MLL-AF9; gene-expression profiling shows PBX3/MEIS1 overexpression - but not HOXA9/MEIS1, HOXA9/PBX3 or HOXA9 alone - recapitulates the MLL-fusion-mediated core transcriptome, particularly upregulation of the endogenous Hoxa genes; and disrupting the binding between MEIS1 and PBX3 diminished PBX3/MEIS1-mediated cell transformation and HOX gene upregulation.
PBX3 and MEIS1 Cooperate in Hematopoietic Cells to Drive Acute Myeloid Leukemias Characterized by a Core Transcriptome of the MLL-Rearranged Disease · Li Z, Chen P, Su R, Hu C, Li Y, Elkahloun AG, Zuo Z, Gurbuxani S, Arnovitz S, Weng H, Wang Y, Li S, Huang H, Neilly MB, Wang GG, Jiang X, Liu PP, Jin J, Chen J · Cancer Research 76:619-629 2016What this evidence cannot show
Abstract read this cycle; full text not retrieved. MURINE HSPCs, MYELOID readout, retroviral overexpression. Sufficiency without an ECTOPIC HOX gene is not dispensability of HOX: the construct upregulates the endogenous Hoxa cluster, so endogenous Hoxa9 is present in the transformed cells and the experiment that would separate them - PBX3/MEIS1 in Hoxa9-null cells - was not found this cycle. Same senior group as the 2013 PBX3 cofactor paper, so this is not independent confirmation of the PBX3 position.
Evidence that argues against this (1)
- In human MLL-rearranged leukaemia cells, HOXA9 suppression induces apoptosis specifically in MLL-rearranged cells and co-down-regulates PBX3, placing PBX3 downstream of HOXA9 - the opposite direction to this claim. Faber J et al. 2009
- E2strong confidence 4 of 5: Replicated in peer-reviewed work by more than one group.paper
In murine Hoxa9-transformed cells, Pbx3 and Meis1 must dimerise to support Hox-induced leukaemia: without Pbx3, Meis1 protein is highly unstable, with degradation contingent on a motif coinciding with the Pbx-binding domain, and either deleting that sequence or binding Pbx3 prolongs Meis1's half-life by preventing ubiquitination; proteasome inhibition also blocks the breakdown; Pbx3 overexpression induces endogenous Meis1 transcription; blocking Meis1/Pbx3 dimerisation abrogates enhanced proliferation and colony-forming cell numbers in Hoxa9-transformed primary cells and the expression of the Meis1 targets Flt3 and Trib2; Meis1 must bind Pbx3 to form high-affinity DNA/Hoxa9/Meis1/Pbx3 complexes in vitro; and Pbx3/Meis1 dimers are rate-limiting for Hoxa9-induced leukaemia in transplantation assays.
Pbx3 and Meis1 cooperate through multiple mechanisms to support Hox-induced murine leukemia · Garcia-Cuellar MP, Steger J, Fuller E, Hetzner K, Slany RK · Haematologica 100:905-913 2015What this evidence cannot show
Abstract read this cycle; full text not retrieved. MURINE Hoxa9-driven leukaemia, which is the fusion's output rather than the fusion, and the readout is myeloid. The surname Fuller is written without its umlaut; PubMed renders it Fuller E. The dimerisation-blocking reagent is not identified in the abstract, so what 'blocking Meis1/Pbx3 dimerisation' means experimentally is not known from what was read.
Evidence that argues against this (1)
- If Pbx3's main job is stabilising Meis1, then MEIS1 is the effector and PBX3 is a regulator of its abundance - which would make a PBX3-directed agent an indirect and slow way of lowering MEIS1 rather than a target in its own right. Li Z et al. 2016
- E3strong confidence 4 of 5: Replicated in peer-reviewed work by more than one group.paper
In human MLL-rearranged acute leukaemia, RNA-interference suppression of HOXA9 co-down-regulates a programme including HOXA10, MEIS1, PBX3 and MEF2C, and induces proliferation arrest and apoptosis specifically in the 7 MLL-rearranged of 17 human AML/ALL lines tested (P=0.007), with reduced leukaemia burden in mice transplanted with HOXA9-depleted t(4;11) SEMK2 cells.
HOXA9 is required for survival in human MLL-rearranged acute leukemias · Faber J, Krivtsov AV, Stubbs MC, Wright R, Davis TN, van den Heuvel-Eibrink M, Zwaan CM, Kung AL, Armstrong SA · Blood 113:2375-2385 2009What this evidence cannot show
Abstract read this cycle. HUMAN, and the in vivo arm is a t(4;11) KMT2A::AFF1 B-lineage line, which makes it the most index-relevant item in this entry - and it is a KNOCKDOWN, so it reports a requirement for HOXA9 transcript, not the topology of the network. Co-down-regulation of PBX3 could be direct HOXA9 regulation or a consequence of the cells dying.
Evidence that argues against this (1)
- Live DepMap gives HOXA9 knockout a mean gene effect of only -0.223 across 57 leukaemia lines, so a CRISPR knockout does not reproduce the survival requirement this knockdown reports - a discrepancy that is unresolved and that bears on how much weight the downstream direction can carry. Broad Institute DepMap; Open Targets Platform 2026
- E4strong confidence 4 of 5: Replicated in peer-reviewed work by more than one group.paper
The HOXA9-TALE interaction relies consistently on the redundant activity of the hexapeptide (HX) motif AND two paralog-specific residues of the HOXA9 homeodomain, measured on different DNA-binding sites in vitro and by bimolecular fluorescence complementation in different cell lines; more generally, the presence of MEIS induces a remodelling of the HOX-PBX interaction that makes the HX motif globally dispensable for trimeric complex formation in the large majority of HOX proteins.
The human HOXA9 protein uses paralog-specific residues of the homeodomain to interact with TALE-class cofactors · Dard A, Jia Y, Reboulet J, Bleicher F, Lavau C, Merabet S · Scientific Reports 9:5664 2019What this evidence cannot show
Abstract read this cycle. HUMAN proteins but in heterologous cell lines and in vitro DNA-binding assays, not in leukaemia cells and not in a KMT2A-rearranged context. 'Redundant activity' is the authors' word and means either route suffices - which is precisely what makes single-interface competition an incomplete strategy, but it is an in vitro and BiFC result, not a demonstration that a peptide fails in cells.
Evidence that argues against this (1)
- The group that developed the hexapeptide-based peptides reports selective killing of leukaemic cells overexpressing HOXA/PBX3 genes with HXR9, so the HX-motif route is evidently sufficient to produce a cellular effect whatever the redundancy in binding assays says. Li Z et al. 2013
the mechanism entry for the dimer this gene belongs to
PBX3/MEIS1 without an ectopic HOX gene reproduces the MLL-fusion core transcriptome
partially opposes the HOXA9/MEIS1 framing: HOXA9/MEIS1 transforms but does NOT reproduce the fusion transcriptome, while PBX3/MEIS1 does
- related to →Step 5 — The output program: HOXA9/HOXA10/MEIS1/PBX3 plus MEF2C, MYB, BCL2, FLT3 and PROM1/CD133
the combinatorial test of the output-programme claim; the pair that reproduces the programme contains no HOX gene
the contact the peptide attacks is not the contact that reproduces the fusion transcriptome
rests on the combinatorial and dimer-stability results
would add a fourth separable function to the MEIS1-PBX3 interface if it holds