RARA — domain architecture and residue coordinates
also called RARA · RAR-alpha · retinoic acid receptor alpha · NR1B1 · P10276
RARA is a receptor that reads vitamin-A signals and, when it does, tells blood cells to finish maturing. In acute promyelocytic leukaemia its front end is chopped off and replaced by PML, which jams the receptor in the off position until a high dose of vitamin-A acid forces it back on.
The same thing, for a scientist
RARA (UniProt P10276, 462 aa) has the canonical nuclear-receptor layout: an A/B modulating region carrying AF-1 (1-87), a C4-type zinc-finger DNA-binding domain (88-153, two zinc fingers at 88-108 and 124-148), a hinge D region (154-182), a ligand-binding domain E (183-417) containing the AF-2 9aaTAD at 408-416 and an NCOR1-binding region at 404-419, and a disordered F region (419-462). UniProt annotates the PML::RARA (and PLZF-RARA) translocation site between residues **60 and 61**, so the fusion retains RARA 61-462: the DNA-binding domain, hinge, ligand-binding domain and AF-2 are all intact, and what is destroyed is AF-1. RARA is a drug target in the strict sense — ATRA binds this exact ligand-binding domain — which is why this node carries a substantial `bite_back`.
This is a prediction, not a measurement. No one has seen this protein in this shape. The colours say how much the software trusts itself, and that is worth reading before anything else — the orange stretches are places the model is effectively saying this part has no fixed shape.
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.
The break removes AF-1 and nothing else structured1 deeper
UniProt annotates the RARA breakpoint at 60/61 for PML-RARA, PLZF-RARA and the reciprocal RARA1-PLZF product 1. That number is doing a lot of work, so note what it implies: of the A/B modulating region at 1-87, residues 1-60 are lost and 61-87 survive. A partial A/B region is not a functional AF-1, so the array marks the A/B entry lost and explains the partial survival in its note rather than pretending the domain is cleanly excised.
Everything else is retained: both zinc fingers, the full DBD, the hinge, the whole ligand-binding domain and the AF-2 activation helix. PML::RARA is a complete, ligand-competent retinoic acid receptor with a tetramerising scaffold bolted on in place of AF-1. That is why it is a dominant negative at physiological retinoid concentrations and an agonist-responsive activator at pharmacological ones.
The co-repressor site sits inside the retained region
UniProt annotates residues 404-419 as required for binding the co-repressor NCOR1, immediately overlapping the AF-2 9aaTAD at 408-416 1. Both are retained. The competition between NCOR1/SMRT and coactivators for this same short stretch of the LBD surface is the molecular seesaw that ATRA tips, and the fact that the fusion keeps both halves of the seesaw is the structural reason differentiation therapy works at all.
Two structures, two boundary opinions for the LBD
UniProt gives the ligand-binding domain as 183-417. The crystal structures disagree slightly and among themselves:
| Structure | RARA chain extent |
|---|---|
| 1DKF (RAR/RXR LBD heterodimer) 2 | 182-416 |
| 3KMR, 7AOS, 7WQQ | 176-421 |
| 5K13 | 181-426 |
| 1DSZ (RXR/RAR DBD on a DR1 element) | 82-167 |
So the LBD is good to about +/-8 residues and the DBD construct in 1DSZ (82-167) is wider at both ends than UniProt's 88-153. Neither disagreement is large, but both are recorded rather than smoothed away.
Bourguet and colleagues solved 1DKF with RARalpha bound to a selective antagonist and showed that helix H12 adopts an antagonist position when pushed by a bulky ligand extension 2 — the structural basis of how the same pocket can be an on-switch or an off-switch depending on ligand shape.
Why this one is curable and the others are not
APL is the standing counterexample to 'fusion oncoproteins are undruggable' 3. The reason is visible in the domain map: the fusion retains an intact, evolved, high-affinity small-molecule binding pocket. Nothing about being a fusion made the pocket druggable; the pocket was already there because RARA is a ligand-activated receptor. None of BCR::ABL1's curability logic, and certainly none of KMT2A::AFF1's, transfers from this. The honest lesson from PML::RARA is narrow: if a fusion happens to retain a nuclear-receptor ligand pocket, differentiation therapy is on the table.
| uniprot | P10276 |
|---|---|
| length AA | 462 |
| gene symbol | RARA |
| isoform note | Coordinates are UniProt canonical P10276, 462 aa. RARA has alpha-1 and alpha-2 promoter-driven forms that differ only in the extreme N-terminal A region; because the PML::RARA break at residue 60/61 removes that region anyway, the isoform question is moot for the fusion. Numbering here is the standard one used in the nuclear-receptor and APL literature. |
| assumed fusion | PML::RARA bcr1. The RARA contribution is residues 61-462 per the UniProt-annotated translocation site at 60/61; the RARA portion is the same for bcr1, bcr2 and bcr3 and for the PLZF-RARA variant. |
| alphafold ID | AF-P10276-F1 |
| domain coordinate caveat | The DBD and LBD are among the best-determined coordinates in this whole comparator set, but the crystal constructs are systematically a few residues wider than the UniProt annotations (DBD 82-167 vs 88-153; LBD 176-426 vs 183-417), so read both as +/-8. The A/B and F regions have no experimental coverage at all and their boundaries are definitional. The 9aaTAD and NCOR1 regions are short functional motifs mapped by mutagenesis, not by structure. |
| fusion boundary residue | 61 |
| fusion boundary note | RARA contributes residues 61-462. The same boundary applies to PLZF-RARA and the other RARA fusions with UniProt-annotated sites. |
PDB IDS
[ "1DKF", "1DSZ", "3KMR", "5K13", "7AOS" ]
domains — the numbers behind the chart
[
{
"name": "A/B (AF-1 modulating region)",
"start": 1,
"end": 87,
"note": "TRUNCATED: the break at 60/61 removes residues 1-60 and leaves 61-87. A partial A/B region is not a functional AF-1, so marked lost. No structure covers this region in any case.",
"fate": "lost"
},
{
"name": "DBD (C, nuclear receptor DNA-binding domain)",
"start": 88,
"end": 153,
"note": "the DBD construct in PDB 1DSZ spans 82-167, wider than the UniProt annotation at both ends.",
"fate": "retained"
},
{
"name": "ZnF1 (C4-type zinc finger)",
"start": 88,
"end": 108,
"fate": "retained"
},
{
"name": "ZnF2 (C4-type zinc finger)",
"start": 124,
"end": 148,
"fate": "retained"
},
{
"name": "D (hinge)",
"start": 154,
"end": 182,
"fate": "retained"
},
{
"name": "LBD (E, ligand-binding domain)",
"start": 183,
"end": 417,
"note": "the ATRA pocket and the RXR heterodimer interface. Structural chain extents range 176-426 across entries, so treat as +/-8.",
"fate": "retained"
},
{
"name": "NCOR1-binding region",
"start": 404,
"end": 419,
"note": "overlaps AF-2; the co-repressor/coactivator seesaw that ATRA tips.",
"fate": "retained"
},
{
"name": "AF-2 (9aaTAD activation helix)",
"start": 408,
"end": 416,
"fate": "retained"
},
{
"name": "F (C-terminal disordered region)",
"start": 419,
"end": 462,
"fate": "retained"
}
]- ATRA acts on wild-type RARA in every tissue that expresses it, which is why differentiation syndrome, pseudotumour cerebri and severe teratogenicity are the dose-limiting realities of this therapy. The target's druggability and its toxicity come from the same physiology.
- Because the fusion retains a fully functional ligand-binding domain, resistance can arrive as point mutations in that pocket that preserve the fusion's oncogenic repression while abolishing drug binding — the pocket is a liability as well as an opportunity.
- The APL success story is repeatedly used to argue that fusion oncoproteins are tractable. The domain map shows why that generalisation is weak: the druggable pocket was pre-existing nuclear-receptor biology, not something the fusion created. A fusion that retains no evolved ligand pocket inherits none of this.
- Residues 61-87 of the A/B region survive the break. Do those 27 orphaned residues do anything in the fusion, or are they inert? Nobody appears to have tested a 61-87 deletion in a PML::RARA background.
- The NCOR1-binding region (404-419) and AF-2 (408-416) overlap almost completely. Is there any ligand shape that displaces co-repressor without recruiting coactivator, i.e. a pure de-repressor rather than an agonist?
- How much of APL's curability is attributable to the retained ligand pocket versus to arsenic's independent action on the PML moiety? The two are always given together, so the counterfactual is poorly constrained.
- E1established confidence 5 of 5: Multiple independent lines of evidence; treated as settled in the field.database
RARA is 462 residues with A/B modulating region 1-87, DNA-binding domain 88-153 (zinc fingers 88-108 and 124-148), hinge 154-182, ligand-binding domain 183-417, NCOR1-binding region 404-419, AF-2 9aaTAD 408-416 and disordered F region 419-462; the PML-RARA translocation site is annotated between residues 60 and 61.
UniProtKB entry P10276 (RARA_HUMAN): reviewed sequence, length and positional feature annotations · UniProt Consortium · UniProtKB/Swiss-Prot 2026What this evidence cannot show
The same 60/61 site is annotated for PLZF-RARA and for the reciprocal product, so it represents a curated consensus over several rearrangements rather than a measurement specific to PML::RARA.
Evidence that argues against this (1)
- Both boxes move substantially depending on who draws them. For the DNA-binding domain the live InterPro call returns Pfam PF00105 87-155, SMART SM00399 85-156, PROSITE PS51030 and InterPro IPR001628 85-160, SUPERFAMILY SSF57716 85-168, CDD cd06964 82-166 and CATH-Gene3D G3DSA:3.30.50.10 83-179 - a C-terminal spread of 26 residues against UniProt's 153. For the ligand-binding domain the disagreement is larger at the N-terminus: Pfam PF00104 gives 227-399 and SMART SM00430 gives 230-388, i.e. 44-47 residues shorter at the front and 18-29 shorter at the back than UniProt's 183-417, while CDD cd06937 gives 186-416 and SUPERFAMILY SSF48508 gives 137-415. The stored 183-417 is the PROSITE PS51843 answer, one of six. InterPro Consortium; EMBL-EBI 2026
- E2established confidence 5 of 5: Multiple independent lines of evidence; treated as settled in the field.paper
The crystal structure of the RARalpha ligand-binding domain heterodimerised with RXRalpha shows helix H12 driven into an antagonist position by a bulky ligand extension, and defines the RAR-RXR heterodimer interface.
Crystal structure of a heterodimeric complex of RAR and RXR ligand-binding domains · Bourguet W, Vivat V, Wurtz JM, Chambon P, Gronemeyer H, Moras D · Molecular Cell 2000What this evidence cannot show
The RXR partner carried a constitutively active F318A mutation and its pocket contained an adventitious fatty acid; the RARA chain in PDB 1DKF is modelled over residues 182-416, slightly narrower than the UniProt LBD annotation of 183-417.
Evidence that argues against this (1)
- Other RARA LBD structures (3KMR, 7AOS, 7WQQ) model the chain from 176 to 421 and 5K13 from 181 to 426, so the LBD boundary is uncertain by roughly +/-8 residues. UniProt Consortium 2026
- E3established confidence 5 of 5: Multiple independent lines of evidence; treated as settled in the field.review
ATRA-based regimens raised APL complete remission rates to 90-95% and arsenic trioxide improved outcomes further, converting the most rapidly fatal acute leukaemia into the most curable.
Acute promyelocytic leukemia: from highly fatal to highly curable · Wang ZY, Chen Z · Blood 2008What this evidence cannot show
A historical review aggregating heterogeneous cohorts; does not itself report the structural basis of the response.
Evidence that argues against this (1)
- Population data put a floor under that framing that the trial numbers do not show. In an unselected Swedish registry cohort of 105 APL patients, 29% died within 30 days and 26% within 14 days (median 4 days), with 35% of the early deaths in patients who never received ATRA at all; complete remission was 97% only among those who survived the early window. In California, 30-day mortality in 772 patients aged under 40 fell from 26% to 14% across the ATRA era but 7-day mortality did not improve at all (P = 0.229), and lack of health insurance (early-death OR 2.67, P = 0.031) and Hispanic ethnicity (OR 2.13, P = 0.014) predicted early death. The most-curable framing describes patients who reach treatment, not patients who are diagnosed. Abrahao R et al. 2015
- E4established confidence 5 of 5: Multiple independent lines of evidence; treated as settled in the field.database
An AlphaFold model AF-P10276-F1 covers all 462 residues; experimental structures cover the DBD on DNA (1DSZ, 82-167) and the LBD in several entries that each span only part of 176-426 (1DKF 182-416, 3KMR/3KMZ 176-421, 5K13 181-426).
AlphaFold Protein Structure Database entry for P10276 (model AF-P10276-F1) · Jumper J, Varadi M, et al.; EMBL-EBI/DeepMind · AlphaFold DB 2026What this evidence cannot show
No structure covers residues 1-81 (the A/B region including AF-1) or 427-462 (the F region); both are predicted-disordered and drawn schematically. The 1DSZ and 1DKF entry titles were retrieved from RCSB and confirmed. CORRECTION 2026-09-19: this item previously gave the LBD structural coverage as '1DKF, 3KMR and others, ~176-426', which no single entry spans. 1DKF maps to 182-416 with UniProt 211-213 unobserved; the widest single entry is 5K13 at 181-426.
Evidence that argues against this (1)
- The stored ligand-binding-domain range is wrong at both ends, and the correction is fetchable. PDBe SIFTS maps 1DKF chain B to P10276 182-416, not 176-426, and the RCSB instance record shows entity residues 30-32 unobserved, i.e. UniProt 211-213 has no coordinates; 3KMR and 3KMZ map to 176-421 and the widest LBD entry in the whole P10276 set is 5K13 at 181-426. No single structure spans 176-426. (1DSZ 82-167 checks out exactly, with no unobserved residues in chain A.) On the AlphaFold side, per-residue pLDDT from the v6 model gives 112 of 462 residues (24%) below 50 with a mean of 78.1, the worst-modelled residues being precisely the A/B and F regions. PDBe / SIFTS et al. 2026
- related to →PML::RARA, t(15;17)(q24;q21)
Counter-evidence added cycle 41: the most-curable framing describes patients who reach treatment. In 772 Californians under 40, 30-day mortality fell from 26% to 14% across the ATRA era but 7-day mortality did not improve at all (P = 0.229), with no health insurance (OR 2.67) and Hispanic ethnicity (OR 2.13) predicting early death (Abrahao 2015, PMID 26264598). Structural correction: no single RARA structure spans 176-426.
Dispute recorded rather than resolved: the PML RING tetramer model (Wang 2018) and the RBCC dimer + coiled-coil-pairing model (Tan 2024) disagree about what the basic PML unit is. Anything downstream that assumes a fixed stoichiometry for PML::RARA should treat the oligomeric state as open.
PML::RARA, t(15;17)
- part of →PML::RARA, t(15;17)(q24;q21)
3-prime partner domain architecture; DBD and ligand-binding domain retained, A/B region lost
residue coordinates for the ATRA-binding pocket inside the retained ligand-binding domain
the retained corepressor-binding surface is what converts the receptor into a constitutive repressor
both fusions retain an intact, pre-existing small-molecule binding pocket; neither pocket was created by the fusion
t(15;17)(q24;q21)
the fusion retains an intact, ligand-competent retinoic acid binding pocket (LBD 183-417)