How HOXA9 binds DNA: a hexapeptide into a PBX pocket, a 1.9 A structure that is mouse HoxA9 plus human Pbx1 homeodomains, and a second route that makes the motif redundant
also called HOX-PBX dimer · hexapeptide motif · HXXPWM · 1PUF · YPWM motif
HOXA9 on its own is a poor, unfussy DNA binder. It gets its grip by docking a short peptide loop into a pocket on a partner protein called PBX, and the pair then reads a longer stretch of DNA together. There is a crystal structure of that pair - but it is mouse HOXA9 with human PBX1, and only the small DNA-binding pieces of each, which matters for anyone designing against it.
The same thing, for a scientist
Cooperative HOX-PBC DNA binding requires a conserved motif N-terminal to the HOX homeodomain, the hexapeptide, which inserts into a pocket on the PBC-class TALE protein. The reference structure is PDB **1PUF**, 1.9 A, HoxA9 and Pbx1 homeodomains bound to a 20-bp duplex [E1]. It establishes that the **posterior** hexapeptide adopts an altered conformation compared with anterior HOX/PBC structures, and that additional non-specific contacts and altered DNA conformation account for posterior HOX proteins' stronger affinity and different specificity - and that N-terminal-arm residues are critical for sequence recognition **despite making no direct base contacts** [E1]. Two things qualify any use of this for design. **The deposit's HoxA9 chain is annotated to P09631, which is MOUSE Hoxa9, paired with human PBX1 (P40424); both chains are homeodomain-only fragments, 77 and 73 residues** [E2]. And the interaction is **redundant**: human HOXA9's TALE binding rests on the hexapeptide *and* two paralog-specific homeodomain residues, either of which suffices, while MEIS presence globally remodels the interaction and makes the hexapeptide dispensable for trimer formation in most HOX proteins [E3].
This is real experimental data — atoms placed by crystallography or cryo-EM, deposited by the people who did the experiment.
I have never read one of these pictures before
A protein is one long chain of amino acids that folds up. The ribbon traces that chain from one end to the other; it is not a wire, it is a simplification of where the backbone runs. Drag to rotate, scroll to zoom.
Domains are the stretches of chain that do a particular job, and they are what the coloured blocks above the picture mark out. Click one and the view flies to it. In this project the important question about any domain is usually whether the fusion still has it — the labels say lost where it does not.
Anything in yellow sticks is a bound small molecule, which is usually the drug.
What the structure shows, and what it is made of
PDB 1PUF, Crystal Structure of HoxA9 and Pbx1 homeodomains bound to DNA, 1.9 A, four polymer entities 2:
| entity | description | length | UniProt |
|---|---|---|---|
| 1 | DNA, ACTCTATGATTTACGACGCT | 20 nt | - |
| 2 | DNA, TAGCGTCGTAAATCATAGAG | 20 nt | - |
| 3 | Homeobox protein Hox-A9 | 77 aa | P09631 = Mus musculus Hoxa9 |
| 4 | Pre-B-cell leukemia transcription factor-1 | 73 aa | P40424 = Homo sapiens PBX1 |
Three consequences that a design programme has to carry:
- It is not human HOXA9. P09631 is mouse Hoxa9 (271 aa); human HOXA9 is P31269 (272 aa). Verified against UniProt on 2026-09-20. The graph's rule never trust recall for an identifier applies to species as much as to gene symbols, and a structure title reading HoxA9 gives no signal about which one.
- It is not PBX3. The PBC partner is PBX1 (430 aa), not PBX3 (434 aa) - four residues apart, accessions two digits apart, and the literature's entire point is that PBX3 rather than PBX1 is the leukaemia-relevant paralogue (
gene-pbx3). There is no HOXA9-PBX3 structure in anything examined this cycle. - Both protein chains are homeodomain fragments - 77 and 73 residues from proteins of 271 and 430. The structure models the DNA-binding module and not the proteins. It says nothing about the MEIS/PREP contacts, the PBC-A/PBC-B regions, or the nuclear-export machinery, all of which are outside the construct.
The biology the structure explains
The HOX/HOM superfamily binds DNA through a homeodomain that, alone, recognises a short and highly degenerate site. Specificity comes from the partner. Most HOX proteins bind DNA cooperatively with PBC-family TALE proteins, and that cooperation requires the hexapeptide, which differs significantly in sequence between anterior- and posterior-regulating HOX proteins 1.
1PUF is the posterior case. Its findings 1:
- The posterior hexapeptide adopts an altered conformation relative to previously solved anterior HOX/PBC structures.
- Additional non-specific interactions and an altered DNA conformation account for the stronger DNA-binding affinity and different specificity of posterior HOX proteins.
- DNA-binding studies of wild-type and mutant HoxA9 and HoxB1 show that residues in the N-terminal arm of the homeodomains are critical for proper sequence recognition despite not contacting the bases directly - specificity by indirect readout.
That last point is the one with a practical edge: sequence specificity here is partly a property of DNA shape rather than of protein-base contacts, so a design effort that models only the protein-protein interface is modelling half the system.
The redundancy that undercuts single-interface competition
This is the finding that bears directly on whether a hexapeptide-mimetic peptide can work.
Dard et al. dissected human HOXA9's interaction with PBX1 and MEIS1 across different DNA-binding sites in vitro and by bimolecular fluorescence complementation in different cell lines 3. Two results:
- The HOXA9-TALE interaction relies on the redundant activity of the hexapeptide motif AND two paralog-specific residues of the HOXA9 homeodomain. Either route supports the interaction.
- More generally, the additional presence of MEIS remodels the interaction, so the hexapeptide is globally dispensable for trimeric complex formation in the large majority of HOX proteins.
If the hexapeptide is one of two sufficient routes, then competing for the PBX hexapeptide pocket does not necessarily dissociate the complex - and in a cell where MEIS1 is abundant, which is precisely the KMT2A-rearranged cell, the motif is the more dispensable of the two.
This is a mechanistic argument against the design of HXR9/HTL001 (target-hox-pbx-interface-peptide-hxr9) and it is not decisive: those peptides demonstrably kill cells, and the redundancy result is from binding assays and BiFC rather than from leukaemia cells. Both are recorded; neither is dismissed. What the redundancy does establish is that the peptides' cellular activity should not be assumed to be on-mechanism.
| mechanism class | cooperative protein-DNA recognition; protein-protein interface |
|---|---|
| no structure exists for | HOXA9 with PBX3, in any species; and HOXA9/MEIS1/PBX3 as a ternary complex on DNA. Neither was found in the RCSB records examined this cycle. This is a bounded structural absence limited to what a single entry-level query and its polymer entities showed, NOT a systematic RCSB search - a full search was not run, so it is recorded as unchecked rather than as zero. |
| domains note | NO residue numbers are asserted. Per the brief's PDB rule, a span must be derived by aligning the fetched entity sequence to a UniProt sequence fetched the same day, and that alignment was not performed. The hexapeptide and homeodomain boundaries are therefore deliberately absent rather than estimated. |
| specificity source | partly indirect readout - N-terminal-arm homeodomain residues are critical for sequence recognition without contacting bases, so DNA shape carries part of the specificity |
| redundancy of the interface | hexapeptide motif OR two paralog-specific HOXA9 homeodomain residues; MEIS presence makes the hexapeptide globally dispensable for trimer formation in most HOX proteins |
PDB IDS
[ "1PUF" ]
structure composition
{
"pdb": "1PUF",
"title_as_deposited": "Crystal Structure of HoxA9 and Pbx1 homeodomains bound to DNA",
"resolution_A": 1.9,
"protein_entities": 2,
"nucleic_acid_entities": 2,
"hox_chain": {
"description": "Homeobox protein Hox-A9",
"length_aa": 77,
"uniprot": "P09631",
"species": "Mus musculus",
"note": "NOT human HOXA9 (P31269)"
},
"pbc_chain": {
"description": "Pre-B-cell leukemia transcription factor-1",
"length_aa": 73,
"uniprot": "P40424",
"species": "Homo sapiens",
"note": "PBX1, NOT the leukaemia-relevant paralogue PBX3 (P40426)"
},
"dna": [
"ACTCTATGATTTACGACGCT",
"TAGCGTCGTAAATCATAGAG"
],
"construct_warning": "both protein chains are homeodomain-only fragments; the structure models the DNA-binding module, not either protein"
}domains
[]
- Designing against 1PUF means designing against a mouse protein paired with the wrong human paralogue, using only the homeodomain fragments of both. Every one of those three mismatches is the kind the brief's PDB rule exists to catch.
- The interface is redundant [E3]. A competitor at the hexapeptide pocket may leave a complex that forms through the homeodomain route instead, and MEIS1 - abundant in exactly this disease - is what makes the hexapeptide the more dispensable route.
- Specificity is partly in the DNA shape, not in protein-base contacts [E1]. A protein-protein-only design model cannot predict which sites a disrupted complex stops binding, which makes on-target pharmacodynamic readouts hard to design.
- The hexapeptide is conserved across HOX paralogues, so any agent built on it is paralogue-blind by construction - it cannot distinguish HOXA9 from HOXA10, or leukaemic HOX from the HOX proteins running normal marrow, gut and limb.
- Which two paralog-specific homeodomain residues carry the alternative TALE route in human HOXA9? The abstract does not name them and the full text was not retrieved. They are the residues a structure-based programme would have to engage alongside the hexapeptide pocket, and they are a one-paper lookup away.
- Is there any structure of a PBC-class protein with PBX3 rather than PBX1, and any ternary HOX/MEIS/PBX structure on DNA? A systematic RCSB search was not run this cycle and should be - the entry records the gap as unchecked, not as zero.
- Does the posterior hexapeptide conformation seen in 1PUF hold when MEIS1 is present? E3 says MEIS remodels the interaction, and no structure of the remodelled state was located.
- If specificity is partly DNA shape, does the KMT2A fusion's own chromatin context change which HOX-PBX sites are bound? Nothing read this cycle connects the indirect-readout finding to fusion-occupied loci.
- E1established confidence 5 of 5: Multiple independent lines of evidence; treated as settled in the field.paper
The 1.9 A structure of the posterior HOX protein HoxA9 complexed with Pbx1 and DNA shows that the posterior Hox hexapeptide adopts an altered conformation compared with previously determined anterior HOX/PBC structures; the additional non-specific interactions and altered DNA conformation account for the stronger DNA-binding affinity and altered specificity of posterior HOX proteins; and DNA-binding studies of wild-type and mutant HoxA9 and HoxB1 show that residues in the N-terminal arm of the homeodomains are critical for proper DNA sequence recognition despite those residues making no direct contact with the DNA bases.
Structure of HoxA9 and Pbx1 bound to DNA: Hox hexapeptide and DNA recognition anterior to posterior · LaRonde-LeBlanc NA, Wolberger C · Genes & Development 17:2060-2072 2003What this evidence cannot show
Abstract read this cycle. A crystal structure of isolated homeodomain fragments on a synthetic duplex; it is a model of the DNA-binding module, not of either full-length protein, and it contains no MEIS/PREP component. The structural-database rule applies: a PDB entry existing is not a structure suitable for design - ask what the density covers, at what resolution, in what state, and whether the construct is the molecule you care about. Here the answer to the last question is no on two counts, species and paralogue (see E2).
Evidence that argues against this (1)
- Human HOXA9's TALE interaction is redundant between the hexapeptide and two paralog-specific homeodomain residues, and MEIS presence remodels the interaction so the hexapeptide becomes globally dispensable for trimer formation - so a structure of the hexapeptide contact describes one of two sufficient routes, not the route. Dard A et al. 2019
- E2established confidence 5 of 5: Multiple independent lines of evidence; treated as settled in the field.database
PDB entry 1PUF, 'Crystal Structure of HoxA9 and Pbx1 homeodomains bound to DNA', resolution 1.9 A, contains four polymer entities: two 20-nucleotide DNA strands, a 77-residue Homeobox protein Hox-A9 chain annotated to UniProt P09631, and a 73-residue Pre-B-cell leukemia transcription factor-1 chain annotated to UniProt P40424. P09631 is MOUSE Hoxa9 (Hoxa9, Mus musculus, 271 aa) and P40424 is HUMAN PBX1 (430 aa); human HOXA9 is P31269 (272 aa) and human PBX3 is P40426 (434 aa).
RCSB PDB entry 1PUF, polymer entity records 1-4; and UniProt REST records P09631, P40424, P31269, P40426 · RCSB Protein Data Bank; UniProt Consortium; queried by this project 2026-09-20 · Structural and sequence databases 2026What this evidence cannot show
Fetched from data.rcsb.org polymer_entity endpoints and rest.uniprot.org on 2026-09-20, following the brief's instruction to fetch polymer entities rather than reading the entry title. The RCSB primary-citation PubMed field returned null for this entry, so the paper (E1) was located separately and its title taken from PubMed, not from RCSB. RESIDUE NUMBERS ARE NOT ASSERTED anywhere in this entry: the brief's rule is that indices read straight off a deposit are often wrong and that a span must be derived by aligning the fetched entity sequence to a same-day UniProt sequence. That alignment was not performed this cycle, so no hexapeptide or homeodomain coordinates are recorded.
Evidence that argues against this (1)
- A species and paralogue mismatch does not make the structure wrong or useless - homeodomains are extremely conserved and mouse and human HOXA9 differ by one residue in overall length - but it does mean the deposit cannot be cited as 'the human HOXA9-PBX3 structure', which is how a structure of this kind tends to get quoted. UniProt Consortium 2026
- E3strong confidence 4 of 5: Replicated in peer-reviewed work by more than one group.paper
The human HOXA9-TALE interaction relies consistently on the redundant activity of the hexapeptide 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; and the additional presence of MEIS induces a remodelling of the HOX-PBX interaction leading to a global dispensability of the hexapeptide motif 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; full text not retrieved, so the identity of the two paralog-specific residues is NOT asserted here. Human proteins, but in heterologous cell lines and in vitro DNA-binding assays - not leukaemia cells, not a KMT2A-rearranged context, and not a functional leukaemia readout. BiFC reports proximity and complementation, not affinity. The claim is about complex FORMATION; it does not establish that a complex formed by the alternative route is transcriptionally equivalent.
Evidence that argues against this (1)
- Hexapeptide-based peptides do kill leukaemic cells overexpressing HOXA and PBX3 genes, so whatever the binding-assay redundancy shows, competing at this motif produces a cellular effect - though that effect may not be on-mechanism. Li Z et al. 2013
- related to →HOXA9 — required for survival of established MLL-rearranged leukemia, and dispensable for starting it
how HOXA9 achieves DNA-binding specificity: hexapeptide into a PBC pocket, plus indirect readout
the structure is PBX1, not PBX3; no HOXA9-PBX3 structure was located
PDB 1PUF chain 4 is human PBX1 (P40424), the protein that entry covers, as a 73-residue homeodomain fragment
the hexapeptide is paralogue-shared, which is why an interface agent cannot distinguish HOXA9 from HOXA10
the peptide's design premise is the hexapeptide contact this mechanism entry describes
PBX3 is the PBC partner in the dimer this mechanism describes, though the solved structure uses PBX1
third PBC-class member of the HOX-PBX dimer, previously unwritten in this graph