ETV6 — domain architecture and residue coordinates
also called ETV6 · TEL · TEL1 · ETS translocation variant 6 · P41212
ETV6 is a switch-off protein that normally sticks to DNA and quietens genes. Its front end is a self-stacking module that links copies of itself into a long chain, and in leukaemia that chain-making module is welded onto other proteins, which is what makes the fusions dangerous.
The same thing, for a scientist
ETV6 (UniProt P41212, 452 aa) is an ETS-family transcriptional repressor with two structured modules: an N-terminal PNT/SAM domain (UniProt 40-124) that self-associates head-to-tail into an open-ended helical polymer, and a C-terminal ETS DNA-binding domain (339-420). UniProt annotates four distinct translocation breakpoint sites in ETV6, including one between residues 336 and 337 for the ETV6::RUNX1 fusion of childhood B-ALL. That break sits immediately upstream of the ETS domain, so ETV6::RUNX1 retains the polymerising PNT domain and the central repression region but discards ETV6's own DNA-binding domain entirely: the chimera is targeted by RUNX1's Runt domain and repressed by ETV6's polymer.
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.
A polymer, not a dimer1 deeper
The PNT (pointed/SAM) domain of ETV6 was the first SAM domain shown to polymerise. Kim and colleagues solved a head-to-tail helical polymer held together by strong intermolecular contacts and proposed that this is how repression complexes spread along a chromosome 2. Crystallisation required a V80E mutation to break the polymer into a tractable species; the native interface was later shown to be essentially the same, and notably to be flat, with no obvious small-molecule pocket 3.
For fusion biology the consequence is stark. A dimerisation module doubles a partner's local concentration; a polymerisation module can multimerise it without limit. Every ETV6 fusion — to kinases such as PDGFRB, ABL1, JAK2, NTRK3, and to transcription factors such as RUNX1 — inherits that property 2.
Where the PNT domain starts and stops
UniProt annotates 40-124. Every one of roughly forty PDB entries mapped to this region covers approximately 47-123 (a few start at 36 or 43). The crystallographers' construct boundaries are not the same claim as a domain boundary — a construct is trimmed for behaviour, not for biology — but the consistency of 47-123 across independent groups suggests the folded core is a little narrower than UniProt's 40-124. Both numbers are recorded; the array uses 40-124 and the note carries the structural range.
What ETV6::RUNX1 keeps
UniProt annotates the ETV6::RUNX1 (TEL-AML1) breakpoint between residues 336 and 337 1. Golub and colleagues described the fusion as joining the helix-loop-helix (now called PNT/SAM) domain of TEL to the DNA-binding and transactivation domains of AML1 4.
So on the ETV6 side: PNT retained, the central region (which includes a poorly delimited internal repression element) retained, and the ETS domain at 339-420 lost by three residues. That is a striking piece of positioning — the break falls in the two-residue gap between the end of the disordered linker and the start of the DNA-binding domain. It is the clearest example in this comparator set of a breakpoint that is selected precisely because it separates a protein-protein interaction module from its own DNA-targeting module.
Other breakpoints in the same protein
UniProt records three further ETV6 translocation sites: 11/12 (CHIC2-ETV6), 54/55 (ETV6-MDS2) and 55/56 (PAX5-ETV6) 1. The 54/55 and 55/56 breaks cut into the PNT domain rather than downstream of it, which means those fusions cannot use the polymer mechanism and must work some other way. That contrast is exactly the kind of natural experiment that tests whether polymerisation is necessary or merely sufficient.
| uniprot | P41212 |
|---|---|
| length AA | 452 |
| gene symbol | ETV6 |
| isoform note | UniProt lists no alternative products for P41212, so ETV6 numbering is unambiguous: 452 aa, one canonical sequence. This is the one protein in this comparator set with no isoform trap. |
| assumed fusion | ETV6::RUNX1 (TEL-AML1), t(12;21). ETV6 contributes residues 1-336 per the UniProt-annotated breakpoint site at 336/337. |
| alphafold ID | AF-P41212-F1 |
| domain coordinate caveat | CORRECTED 2026-09-19: neither domain is good to +/-8. The PNT domain spreads about 32 residues at the N-terminus across sources (SUPERFAMILY 24-122 to CDD 56-123) and the ETS domain spreads from 315 to 442 (SUPERFAMILY 315-439, CATH 322-439, UniProt/PROSITE 339-420, NMR construct 2DAO 338-442). The ETS spread matters for this node specifically, because the superfamily boundaries start upstream of the ETV6::RUNX1 breakpoint at 336/337. The 'central repression region' 125-336 is NOT a verified domain: it is the interval between the two folds, included so that the viewer draws the retained portion of the protein to scale, and its boundaries carry no structural claim. The breakpoint at 336/337 comes from UniProt's curated cDNA-junction annotation and was not independently recomputed from exon coordinates in this pass. |
| fusion boundary residue | 336 |
PDB IDS
[ "1JI7", "1LKY", "2DAO", "5L0P" ]
domains — the numbers behind the chart
[
{
"name": "PNT (SAM polymerisation domain)",
"start": 40,
"end": 124,
"note": "head-to-tail helical polymer, not a dimer. Structural constructs consistently span ~47-123 across ~40 PDB entries, so the folded core may be slightly narrower than the UniProt 40-124 annotation. N-terminal spread ~32 residues: SUPERFAMILY 24-122, InterPro IPR013761 24-127, SMART 38-124, UniProt/PROSITE 40-124, Pfam 40-123, CDD 56-123; crystal constructs 1JI7/1LKY 47-123 and 5L0P 47-124 [E6][E7].",
"fate": "retained"
},
{
"name": "CRD (central repression region)",
"start": 125,
"end": 336,
"note": "SOFT BOUNDARIES: this is the interval between the two structured domains and is not itself a UniProt-annotated domain. UniProt predicts disorder at 158-262. Drawn as the retained linker, not as a fold. Do not quote these numbers as a domain. Confirmed to have zero database support: InterPro returns no match of any kind between residues 127 and 315 in P41212 [E6]. The bar exists only so the viewer draws the retained portion to scale.",
"fate": "retained"
},
{
"name": "ETS (DNA-binding domain)",
"start": 339,
"end": 420,
"note": "lost by three residues: the ETV6::RUNX1 break is annotated at 336/337. PDB 2DAO covers 338-442. BOUNDARY DISPUTE that touches the fate call: UniProt and PROSITE give 339-420, Pfam 341-420, SMART/InterPro 338-424, the NMR construct 2DAO 338-442, and the winged-helix superfamily models 315-439 (SUPERFAMILY) and 322-439 (CATH). The last two start BEFORE the ETV6::RUNX1 break at 336/337, so 'the fusion loses the whole ETS domain' holds for the DNA-binding domain as UniProt draws it but not for the fold as the structural classifications draw it [E6][E7].",
"fate": "lost"
}
]other annotated breakpoints
[
{
"site": "11/12",
"fusion": "CHIC2-ETV6"
},
{
"site": "54/55",
"fusion": "ETV6-MDS2"
},
{
"site": "55/56",
"fusion": "PAX5-ETV6"
},
{
"site": "336/337",
"fusion": "ETV6-RUNX1"
}
]counter evidence pass
{
"queen": "Q2b-atlas2",
"cycle": 31,
"date": "2026-09-19",
"checked": [
"UniProt P41212 live record (primary gene name ETV6 confirmed, 452 aa; PNT 40-124 and ETS DNA-binding 339-420 match, and all four annotated translocation sites including ETV6-AML1 at 336/337 confirmed)",
"InterPro entry/all for P41212",
"SIFTS mappings for 1JI7, 1LKY, 2DAO, 5L0P"
]
}- The PNT polymer interface is identical in wild-type ETV6, which is a haematopoietic tumour suppressor; anything that disrupts the polymer disrupts the tumour suppressor too, and ETV6 germline loss causes thrombocytopenia with leukaemia predisposition.
- Because polymerisation is open-ended rather than a fixed stoichiometry, partial inhibition may not be partially effective: shortening a polymer from 20 subunits to 5 could leave the transforming activity intact.
- Two of the four annotated ETV6 breakpoints (54/55 and 55/56) cut through the PNT domain rather than downstream of it. If polymerisation is the essential mechanism, how do those fusions transform at all? A side-by-side transforming assay would be a clean test.
- The central 125-336 region is retained in ETV6::RUNX1 and has no structure at all. Does it contribute repression activity independent of the PNT polymer, and where exactly does that activity map?
- Is the flat, pocketless TEL-SAM interface reported in 2002 still undruggable given covalent and molecular-glue chemistry? Nobody appears to have re-evaluated it with modern methods.
- Does the ETV6 moiety retained by ETV6::RUNX1 end inside the winged-helix fold or cleanly before it? The breakpoint at 336/337 sits between UniProt's ETS start (339) and the superfamily fold start (315-322). If 15-22 residues of the fold's first element are retained, they are retained as an exposed fragment on the fusion, and nobody appears to have asked whether that fragment does anything.
- E1established confidence 5 of 5: Multiple independent lines of evidence; treated as settled in the field.database
ETV6 is 452 residues with a PNT/SAM domain at 40-124 and an ETS DNA-binding domain at 339-420, and UniProt annotates the ETV6-AML1 (ETV6::RUNX1) translocation breakpoint between residues 336 and 337.
UniProtKB entry P41212 (ETV6_HUMAN): reviewed sequence, length and positional feature annotations · UniProt Consortium · UniProtKB/Swiss-Prot 2026What this evidence cannot show
The breakpoint site annotation is a curated summary of cDNA junctions, not a single measured protein; alternative ETV6::RUNX1 junctions exist at low frequency.
Evidence that argues against this (2)
- The clean statement that ETV6::RUNX1 'loses the ETS domain' is true of the DNA-binding domain as UniProt draws it (339-420) but is boundary-dependent. Under SUPERFAMILY (315-439) and CATH-Gene3D (322-439) the winged-helix fold starts 14-21 residues BEFORE the breakpoint at 336/337, so the fusion retains the first turn or two of it. That does not make the fusion DNA-binding - the recognition helix and the wing are gone either way - but a figure that shows the break falling cleanly between two domains is showing UniProt's boundary, not a consensus one. InterPro Consortium; EMBL-EBI 2026
- The PNT domain's N-terminal boundary is also softer than the node states: SUPERFAMILY SSF47769 and InterPro IPR013761 give 24-122 and 24-127 respectively, sixteen residues N-terminal of UniProt's 40, while CDD cd08535 gives 56-123, sixteen residues inside it. The spread across sources is about 32 residues at the N-terminus, not the +/-8 the node claims. InterPro Consortium; EMBL-EBI 2026
- E2established confidence 5 of 5: Multiple independent lines of evidence; treated as settled in the field.paper
The TEL/ETV6 SAM (PNT) domain forms a head-to-tail helical polymer rather than a dimer, and this self-association is essential for transformation by TEL fusion proteins.
Polymerization of the SAM domain of TEL in leukemogenesis and transcriptional repression · Kim CA, Phillips ML, Kim W, et al. · The EMBO Journal 2001What this evidence cannot show
The crystallised species carried a V80E substitution introduced specifically to prevent runaway polymerisation, so the deposited structure is of a mutant interface. PDB 1JI7 ('Crystal Structure of TEL SAM Polymer', title verified) covers residues 47-123.
Evidence that argues against this (1)
- The crystal-structure paper asserts transformation but never tests it, and when the commonest TEL fusion was put into mice, polymerisation-competent TEL/AML1 did not transform anything. Expressing TEL/AML1 in a bone-marrow transplantation model gave accumulation of an Sca1+/Kit-high/CD11b+ early-progenitor compartment with increased self-renewal and a pro-B differentiation block, but no leukaemia induction over an observation period of more than a year - a preleukaemic state, not transformation. The same study undercuts the polymerisation interpretation specifically: the pointed domain was essential for the phenotype but could NOT be replaced with another oligomerisation domain, so its role is partner-specific corepressor recruitment rather than generic multimerisation, which the crystal structure cannot exclude. Fischer M et al. 2005
- E3strong confidence 4 of 5: Replicated in peer-reviewed work by more than one group.paper
The native (non-mutant) TEL-SAM polymer interface closely matches the mutant structure, and the interacting surfaces are relatively flat with no obvious pocket for small-molecule inhibitors.
Native interface of the SAM domain polymer of TEL · Tran HH, Kim CA, Faham S, Siddall MC, Bowie JU · BMC Structural Biology 2002What this evidence cannot show
Druggability was assessed by visual and geometric inspection in 2002, long before covalent, molecular-glue and degrader approaches; 'no pocket' is not the same conclusion today as it was then.
Evidence that argues against this (1)
- A flat SAM-SAM interface is not the same as an undruggable one - designed peptides bind exactly this kind of surface with measurable affinity and cellular effect. Peptides built from the EphA2-Sam interface yielded a positively charged penta-amino-acid motif that, repeated three times in tandem, binds the mid-loop interface of Ship2-Sam with a dissociation constant of 100 micromolar by surface plasmon resonance kinetics and 83 plus/minus 8 micromolar by binding isotherm, confirmed by NMR chemical-shift perturbation and microscale thermophoresis, and at 50 micromolar killed PC-3 cells with less effect on normal dermal fibroblasts. That bounds the flat-surface argument to SMALL MOLECULES only. Honest scope caveat, which cuts the other way: for the ETV6 PNT domain itself the negative result still stands - more than 75 screening hits were tested by NMR and none bound the purified domain - although alanine scanning shows the interface is not energetically featureless, with hot spots above 6 kJ/mol including L96. Mercurio FA et al. 2017
- E4established confidence 5 of 5: Multiple independent lines of evidence; treated as settled in the field.paper
The t(12;21) of childhood pre-B ALL fuses the self-associating N-terminal domain of TEL/ETV6 to the DNA-binding and transactivation domains of AML1/RUNX1.
Fusion of the TEL gene on 12p13 to the AML1 gene on 21q22 in acute lymphoblastic leukemia · Golub TR, Barker GF, Bohlander SK, et al. · PNAS 1995What this evidence cannot show
Two index cases in the original report; the domain assignment is from cDNA sequence, and the paper uses the then-current 'helix-loop-helix' name for what is now the PNT/SAM domain.
Evidence that argues against this (1)
- Calling the retained RUNX1 region a transactivation domain describes the parent protein, not the fusion - in the fusion that region does not transactivate. The 93-kDa TEL/AML-1B protein binds the enhancer core motif TGTGGT and core-binding factor beta and retains the C-terminal AML-1B domain required for transactivation of the T-cell receptor beta enhancer, yet fails to activate transcription and instead INHIBITS the basal activity of that enhancer, dominantly interfering with AML-1B-dependent transactivation in a way co-expressed wild-type TEL does not reverse; the N-terminal TEL pointed domain was essential for the repression. Follow-up work mapped the mechanism: the TEL pointed domain is a portable repression motif and TEL/AML-1B binds the mSin3A corepressor far more stably than either wild-type protein. Hiebert SW et al. 1996
- E5established confidence 5 of 5: Multiple independent lines of evidence; treated as settled in the field.database
An AlphaFold model AF-P41212-F1 covers all 452 residues, and experimental structures exist for both the PNT domain (e.g. 1JI7, 47-123) and the ETS domain (2DAO, 338-442).
AlphaFold Protein Structure Database entry for P41212 (model AF-P41212-F1) · Jumper J, Varadi M, et al.; EMBL-EBI/DeepMind · AlphaFold DB 2026What this evidence cannot show
No structure covers the central 125-338 region, which UniProt predicts to be largely disordered; the AlphaFold model there is low confidence and the relative orientation of the two domains in the model is not evidence.
Evidence that argues against this (1)
- The two halves are not remotely comparably supported, and the model's own confidence is poor. UniProt P41212 cross-references 43 PDB entries: 42 cover the PNT domain, mostly 47-123/47-124 and at resolutions down to 1.19 A, and exactly one - 2DAO - covers the ETS domain, an NMR ensemble whose RCSB primary citation is recorded as 'To be Published', i.e. it has no peer-reviewed source. The AlphaFold record for AF-P41212-F1 (v6) gives a global pLDDT of 61.81 with 47.8% of residues in the very-low band and only 19.7% very high, so 216 of 452 residues sit where AlphaFold advises against interpreting. The caveat that the central 125-338 region is largely disordered per UniProt also overstates it: UniProt annotates disorder only at 158-262, 105 of those 214 residues. PDBe / SIFTS et al. 2026
- E6strong confidence 4 of 5: Replicated in peer-reviewed work by more than one group.database
The winged-helix fold that contains the ETV6 ETS domain begins UPSTREAM of the ETV6::RUNX1 breakpoint under the structural classifications: SUPERFAMILY SSF46785 and InterPro IPR036390 place it at 315-439 and CATH-Gene3D G3DSA:1.10.10.10 with InterPro IPR036388 at 322-439, against the breakpoint at 336/337 and against the stored ETS bar at 339-420 (UniProt and PROSITE PS50061); Pfam PF00178 gives 341-420 and SMART SM00413 with InterPro IPR000418 give 338-424.
InterPro member-database domain matches for UniProt P41212 (Pfam, PROSITE, SMART, CDD, CATH-Gene3D, SUPERFAMILY) · InterPro Consortium; EMBL-EBI · InterPro 2026What this evidence cannot show
Profile-model extents, not measurements, and the superfamily models deliberately capture the whole winged-helix architecture including flanking basic and N-terminal elements that are not part of the ETS domain proper. Retrieved 2026-09-19 (database record; no full text). The claim here is about where the fold as modelled starts, not that ETV6::RUNX1 retains a functional DNA-binding module - it plainly does not.
Evidence that argues against this (1)
- No structure has ever observed residues 315-337 of ETV6, so the part of the claim that does the work is pure profile extrapolation. The only ETS-domain structure in the whole P41212 cross-reference set is 2DAO, whose SIFTS construct starts at UniProt 338 - one residue past the breakpoint - and the other 42 entries all stop at or before residue 132. The interval 133-337 therefore has no experimental coverage of any kind, and the assertion that the fold begins upstream of the breakpoint rests entirely on SUPERFAMILY and CATH extending a hidden Markov model into sequence nobody has ever seen folded. The competing profiles in the same fetch, Pfam PF00178 341-420 and PROSITE PS50061 339-420, start AFTER the breakpoint. PDBe / SIFTS et al. 2026
- E7established confidence 5 of 5: Multiple independent lines of evidence; treated as settled in the field.database
Deposited construct boundaries are wider than the stored bars at both ends of ETV6: the ETS-domain NMR structure 2DAO maps to P41212 338-442, twenty-two residues past the stored ETS end of 420, while the PNT-domain crystals 1JI7 and 1LKY map to 47-123 and 5L0P to 47-124, seven residues short of the stored PNT start of 40.
SIFTS UniProt residue-level mappings for PDB entries 2DAO, 1JI7, 1LKY, 5L0P (construct ranges read back from the PDBe mappings API) · PDBe / SIFTS; EMBL-EBI · PDBe SIFTS 2026What this evidence cannot show
SIFTS reports the deposited sequence range, not the observed model, so these are construct choices. Read back 2026-09-19 from the PDBe mappings API (database record; no full text). 5L0P additionally maps a second ETV6 segment at 10-39, which the stored array does not represent at all.
Evidence that argues against this (1)
- Thirteen of 2DAO's residues are not ETV6 at all, and 5L0P is a designed chimera with a bacterial protein in the same chain. RCSB reports 2DAO with 118 deposited polymer monomers while SIFTS attributes only PDB residue numbers 8-112 (105 residues) to P41212 338-442, so the deposit carries 13 non-native residues - and because an NMR entry deposits coordinates for every construct residue regardless of order, 'extends 22 residues past 420' measures the expression tag, not the fold. For 5L0P, SIFTS maps P41212 10-39 and 47-124 to chain A together with P45599 (FUR_KLEPN) residues 3-83 in the same chain, i.e. ETV6 segments fused to a Klebsiella ferric-uptake regulator as a crystallisation scaffold, and PDBe coverage shows UniProt residues 47 and 124 themselves are not observed. PDBe / SIFTS et al. 2026
Counter-evidence added cycle 41: polymerisation-competent TEL/AML1 in a bone-marrow transplantation model produced a preleukaemic Sca1+/Kit-hi compartment and a pro-B block but NO leukaemia over more than a year, and the pointed domain could not be replaced with another oligomerisation domain - so its role is partner-specific corepressor recruitment, not generic multimerisation (Fischer 2005, PMID 16044150). The retained RUNX1 'transactivation domain' does not transactivate in the fusion; it represses (Hiebert 1996, PMID 8657108).
- related to →ETV6::RUNX1 (TEL-AML1), t(12;21)(p13;q22)
Boundary dispute that touches a fate call: SUPERFAMILY (315-439) and CATH-Gene3D (322-439) start the ETV6 winged-helix fold 14-21 residues BEFORE the breakpoint at 336/337, so 'the fusion loses the whole ETS domain' is true of UniProt's 339-420 bar but not of the fold as the structural classifications draw it.
5-prime partner domain architecture; the PNT/SAM polymerisation domain is retained
ETV6::RUNX1, t(12;21)
the retained ETV6 PNT domain converts RUNX1 from activator to repressor
- ← analogous toFLI1 — domain architecture and residue coordinates
two ETS-family proteins entering fusions from opposite ends: FLI1 contributes its ETS domain, ETV6 its PNT domain
the native TEL-SAM interface was reported flat and pocketless, arguing against small-molecule tractability
- analogous to →FLI1 — domain architecture and residue coordinates
same ETS-family two-domain layout, opposite halves retained by their fusions
t(12;21)(p13;q22)
- ← related toETV6::RUNX1 (TEL-AML1), t(12;21)(p13;q22)
Counter-evidence: Uchida 1999 (PMID 10023677) shows repression by TEL-AML1 needs three domains, two of them from RUNX1, so the SAM polymer is necessary but not the whole mechanism
- ← related toETV6::NTRK3 (ETV6-TRKC), t(12;15)(p13;q25)
Counter-evidence: Wai 2000 (PMID 10702799) tested dimerisation, not polymerisation, in ETV6-NTRK3 and found heterodimers with wild-type ETV6, which the open-ended homopolymer model does not predict