EWSR1 — domain architecture and residue coordinates

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.

• The EAD is intrinsically disordered with no pocket, so structure-based drug design has nothing to bind. Approaches that target condensate behaviour risk hitting the many wild-type FET-family and low-complexity proteins that use the same physics.
• EWSR1 itself is broadly expressed and its loss impairs meiosis and B-cell development in mice; degrading EWSR1 to remove the fusion would remove a housekeeping RNA-binding protein.
• Because the transforming property is compositional rather than structural, a partial inhibitor may simply shift a phase boundary without abolishing the behaviour, giving apparent target engagement with no efficacy.

1. The type 1 junction leaves out EAD residues 265-285. Does that 21-residue difference explain any part of the prognostic difference between type 1 and longer fusion types, or are those two facts unrelated?
2. If short EWSR1 fragments suffice for BAF retargeting, what is the minimal transforming fragment in residue terms, and does it have a characteristic tyrosine spacing rather than a characteristic length?
3. The fusion has no EWSR1 NLS and relies on FLI1's nuclear targeting. Is nuclear import a rate-limiting and therefore targetable step for this oncoprotein, unlike for fusions that carry two NLSs?
4. Is there any experimental boundary for the EWS activation domain, or only conventions? No profile model covers residues 1-225 and no structure exists. Since the EAD is the oncogenic module of EWSR1::FLI1, the graph is drawing the most important bar in this node from a compositional judgement call.

1. E1established confidence 5 of 5: Multiple independent lines of evidence; treated as settled in the field.database

   EWSR1 is 656 residues with an EAD/Gln-Pro-Thr-rich region at 1-285 containing 31 approximate tandem repeats over 8-285, an IQ motif at 256-285, an RRM at 361-447, a RanBP2-type zinc finger at 518-549 and an NLS at 639-656; UniProt annotates a translocation site at residue 265 (EWSR1/ATF1) and another at 348/349 (EWSR1-FEV).

   UniProtKB entry Q01844 (EWS_HUMAN): reviewed sequence, length and positional feature annotations · UniProt Consortium · UniProtKB/Swiss-Prot 2026
   ▸What this evidence cannot show

   No EWSR1::FLI1 site is annotated; the residue-265 site is for a different fusion that happens to use the same exon boundary. UniProt also predicts disorder at 123-360, 448-525 and 547-656, i.e. most of the protein.

   ▸Evidence that argues against this (2)

   • UniProt's annotated translocation site in EWSR1 is at residue 265 and is labelled for the EWSR1/ATF1 fusion, not for EWSR1::FLI1; a second site at 348/349 is labelled for EWSR1-FEV. There is no UniProt-curated EWSR1::FLI1 site. This node's statement that the exon-arithmetic boundary at 264 is 'consistent with the UniProt translocation site at 265' is therefore weaker corroboration than it reads: the site corroborates the exon-7 boundary as a recurrent junction point across EWSR1 fusions, not the FLI1 fusion specifically. UniProt Consortium 2026
   • The RRM extends 32 residues further C-terminally under the RNA-binding-domain superfamily models (SUPERFAMILY SSF54928 and InterPro IPR035979 both 350-479) than under UniProt's 361-447, and the NMR construct 2CPE was built as 353-453. The stored bar is the narrow end of a 350-479 spread. InterPro Consortium; EMBL-EBI 2026
2. E2strong confidence 4 of 5: Replicated in peer-reviewed work by more than one group.database

   Domain databases reproduce the RRM at 361-447 (InterPro/PROSITE), 363-441 (Pfam PF00076), 362-443 (SMART) and 360-444 (CATH-Gene3D), and the RanBP2 zinc finger at 518-549 (InterPro/PROSITE) and 518-548 (Pfam PF00641).

   InterPro/Pfam/CATH-Gene3D/SUPERFAMILY domain matches for UniProt Q01844 · InterPro Consortium; EMBL-EBI · InterPro 2026
   ▸What this evidence cannot show

   Model-match extents. Notably no database annotates the EAD at all, because low-complexity regions are masked by profile methods, so the EAD boundary rests on UniProt curation alone.

   ▸Evidence that argues against this (1)

   • The zinc finger's spread is nearly twofold, not the one-residue difference quoted. The fetched InterPro record for Q01844 puts the RanBP2-type zinc finger at 518-549 (IPR001876 / PROSITE PS50199), 518-548 (Pfam PF00641), 520-546 (SMART SM00547), 522-543 (PROSITE signature PS01358), 512-551 (CATH-Gene3D G3DSA:4.10.1060.10) and 511-550 (SUPERFAMILY SSF90209 / IPR036443) - domain lengths from 22 to 41 residues. The RRM has the same problem at the C-terminus: SUPERFAMILY SSF54928 and IPR035979 give 350-479, a 130-residue window against UniProt's 87-residue 361-447. Citing the two narrowest, closest calls as reproduction selects the agreeing pair out of six. InterPro Consortium; EMBL-EBI 2026
3. E3strong confidence 4 of 5: Replicated in peer-reviewed work by more than one group.paper

   EWSR1 interacts with the BAF complex; EWS-FLI1 recruits BAF to tumour-specific enhancers as a neomorphic property not shared with wild-type FLI1; this depends on tyrosines required for phase transitions of the EWSR1 prion-like domain; and short EWSR1 fragments fused to FLI1 suffice to recapitulate BAF retargeting and EWS-FLI1 activity.

   Cancer-Specific Retargeting of BAF Complexes by a Prion-like Domain · Boulay G, Sandoval GJ, Riggi N, et al. · Cell 2017
   PMID 28844694DOI
   ▸What this evidence cannot show

   Cell-line and mesenchymal-stem-cell models; phase-transition behaviour was assayed in vitro and inferred in cells. The 'short fragments suffice' result implies the EAD has no sharp functional boundary, which limits how much a single coordinate pair can mean.

   ▸Evidence that argues against this (1)

   • The tyrosine requirement is real, but phase transition is the wrong physical description of what endogenous EWS::FLI1 does in a Ewing sarcoma cell. Knocking a HaloTag into the endogenous EWS::FLI1 locus in A673 cells and imaging by PALM shows it forms sub-diffraction hubs (cluster correlation length 29.87 plus/minus 1.89 nm; mean hub diameter 94.17 plus/minus 1.65 nm) that are explicitly supersaturated clusters rather than bona fide liquid-liquid phase-separated droplets, with a nucleation free-energy barrier preventing macroscopic phase separation. Hub formation is neomorphic - EWSR1 and FLI1 puncta were significantly smaller. Independent work reinforces this by showing that pushing low-complexity-domain interactions further, toward putative phase separation, actually REPRESSES EWS::FLI1 GGAA-microsatellite target genes. Yoshida S et al. 2026
4. E4established confidence 5 of 5: Multiple independent lines of evidence; treated as settled in the field.paper

   In the commonest (type 1) EWS-FLI1 fusion, EWS exon 7 is linked in frame with exon 6 of FLI1; in a series of 99 evaluable Ewing sarcoma patients 64 carried the type 1 fusion, which was an independent favourable prognostic factor.

   EWS-FLI1 fusion transcript structure is an independent determinant of prognosis in Ewing's sarcoma · de Alava E, Kawai A, Healey JH, et al. · Journal of Clinical Oncology 1998
   PMID 9552022DOI
   ▸What this evidence cannot show

   A 1998 retrospective series with 26-month median follow-up; the prognostic advantage of type 1 has not consistently replicated in later trial-era cohorts treated with modern regimens. The exon assignment, which is what this node uses, is not in dispute.

   ▸Evidence that argues against this (1)

   • The type 1 survival advantage did not replicate in the prospective trials built to test it. Euro-E.W.I.N.G. 99 registered 1,957 patients before 1 January 2007, obtained 703 primary tumours and analysed 565 informative patients - 296 type 1, 133 type 2, 91 other EWS-FLI1 and 45 EWS-ERG - with median follow-up 4.5 years, and found no significant prognostic value of fusion type for progression or relapse; the only signal was a non-significant excess for non-type-1/non-type-2 fusions (relative risk 1.38, 95% CI 0.96-2.0, P = 0.10). That is 5.7 times the 99 evaluable patients of the founding series, where type 1 gave a relative risk of 0.37 (P = 0.014). The Children's Oncology Group replicated the null independently in 119 fusion-positive tumours diagnosed 2001-2005. Le Deley MC et al. 2010
5. E5strong confidence 4 of 5: Replicated in peer-reviewed work by more than one group.database

   Recomputing the Ensembl EWSR1 canonical coding exon structure (ENST00000397938, 656 aa) places the end of coding exon 7 at codon 264, consistent with the UniProt translocation site at residue 265.

   Ensembl transcript ENST00000397938 (EWSR1): exon structure and coding-sequence coordinates, used to convert exon breakpoints into residue numbers · Ensembl; EMBL-EBI · Ensembl REST API 2026
   ▸What this evidence cannot show

   Computed by us; exon 7 ends mid-codon (+1 nt), so the junction codon is chimeric and '1-264' versus '1-265' is a labelling convention rather than a factual difference.

   ▸Evidence that argues against this (1)

   • That arithmetic is one of ninety-nine possible answers for this gene. The fetched Ensembl record for ENSG00000182944 lists 99 protein-coding EWSR1 transcripts with translation lengths spanning at least 584-684 aa, including ENST00001117210 at 684 and ENST00001117209 at 683 - 28 residues longer than the canonical - and a second transcript, ENST00001132407, that also encodes exactly 656 aa but is not the canonical one. With that many alternative structures and a 100-residue spread in translation length, an exon-to-codon conversion needs its transcript stated as part of the result; and the consistency with UniProt's 265 is not an independent check, because UniProt's site is curated from the same EWS-FLI1 breakpoint literature the conversion is meant to interpret. Ensembl / EMBL-EBI 2026
6. E6established confidence 5 of 5: Multiple independent lines of evidence; treated as settled in the field.database

   An AlphaFold model AF-Q01844-F1 covers all 656 residues; the only experimental structure is the RRM (PDB 2CPE, residues 353-453).

   AlphaFold Protein Structure Database entry for Q01844 (model AF-Q01844-F1) · Jumper J, Varadi M, et al.; EMBL-EBI/DeepMind · AlphaFold DB 2026
   ▸What this evidence cannot show

   The EAD, which is the entire transforming element, is intrinsically disordered and has no meaningful predicted structure. Any three-dimensional depiction of the EWSR1 half of the fusion is decorative.

   ▸Evidence that argues against this (1)

   • The AlphaFold model is the worst of the protein entries in this beat and the single experimental structure has never been published. AF-Q01844-F1 (v6) has a global pLDDT of 50.41 with 70.9% of residues at pLDDT below 50 and only 2.4% at 90 or above - 465 of 656 residues in the band AlphaFold advises against interpreting, and fewer than 16 residues modelled with high confidence. 2CPE is a 20-model NMR ensemble whose RCSB primary citation is recorded as 'To be Published'; it deposits 113 polymer monomers of which SIFTS attributes only 101 (PDB residue numbers 8-108) to Q01844 353-453, so 12 residues are non-native. Neither source is strong enough to anchor a domain coordinate. PDBe / SIFTS et al. 2026
7. E7strong confidence 4 of 5: Replicated in peer-reviewed work by more than one group.database

   InterPro confirms that the EWSR1 activation domain has no profile-database support at all - the earliest family match (InterPro IPR034870, PANTHER PTHR23238) starts at residue 226, and no model covers 1-225 - while the RRM's C-terminal boundary is disputed by 32 residues: SUPERFAMILY SSF54928 and InterPro IPR035979 give 350-479 against UniProt's 361-447, with Pfam PF00076 at 363-441, SMART at 362-443, CDD cd12533 at 361-444 and the NMR construct 2CPE at 353-453.

   InterPro member-database domain matches for UniProt Q01844 (Pfam, PROSITE, SMART, CDD, CATH-Gene3D, SUPERFAMILY) · InterPro Consortium; EMBL-EBI · InterPro 2026
   ▸What this evidence cannot show

   Absence of a profile match over 1-225 is exactly what one expects for a low-complexity prion-like region and is not evidence that the EAD is not functional; it is evidence that its boundaries are a compositional convention with no independent check, which is what this node already says and this item now confirms rather than assumes. Retrieved 2026-09-19 (database record; no full text read).

   ▸Evidence that argues against this (1)

   • UniProt does annotate a named domain inside the activation domain, and no InterPro member database reproduces it - so the region is not annotation-free, it is annotated by exactly one source the others contradict by silence. The fetched Q01844 record carries Domain 'IQ' at 256-285, a calmodulin-binding helical motif, sitting in the last 30 residues of the EAD and spanning the type-1 EWS-FLI1 breakpoint at 265; none of the 20 InterPro matches for Q01844 includes an IQ entry (no PROSITE PS50096, no Pfam PF00612, no SMART SM00015). The absence of profile coverage over 1-225 is therefore not the only thing the databases say about the EAD, and the one thing they do say at 256-285 is a disagreement rather than a gap. UniProt Consortium 2026

• refutes →EWSR1::FLI1 (EWS-FLI1), t(11;22)(q24;q12)

  Counter-evidence added cycle 41 on two fronts. Prognosis: the type 1 EWS-FLI1 advantage did not replicate - Euro-E.W.I.N.G. 99 analysed 565 informative patients and found no significant prognostic value of fusion type (Le Deley 2010, PMID 20308673). Physics: endogenous HaloTag-EWS::FLI1 in A673 cells forms supersaturated sub-diffraction hubs explicitly NOT liquid-liquid phase-separated droplets (Yoshida 2026, PMID 42417395).

• related to →EWSR1::FLI1 (EWS-FLI1), t(11;22)(q24;q12)

  Counter-evidence (InterPro Q01844): the EWS activation domain, which is the oncogenic module of this fusion, has no profile-database support of any kind -- no model covers residues 1-225. Its bar is a compositional convention. Separately, UniProt's translocation site at 265 is annotated for EWSR1/ATF1, not for EWSR1::FLI1, so it corroborates exon 7 as a recurrent junction rather than this fusion specifically.

• part of →EWSR1::FLI1 (EWS-FLI1), t(11;22)(q24;q12)

  5-prime partner domain architecture; the EAD low-complexity activation domain is retained, the RRM and Zn finger lost

• fuses with →FLI1 — domain architecture and residue coordinates

  EWSR1::FLI1, t(11;22)

• supports →Mechanism class 5 — enhancer hijacking (inv(3) GATA2/EVI1 — and DUX4-r, which only half belongs here)

  the retained EAD is the element implicated in building de novo enhancers at GGAA microsatellites

• analogous to →NUP98 — domain architecture and residue coordinates (with the 1817/920 numbering trap resolved)

  both contribute an intrinsically disordered, phase-separating low-complexity region as the transforming module

• ← fuses withFLI1 — domain architecture and residue coordinates

  t(11;22)(q24;q12), type 1

• ← related toProximity-labelling map of KMT2A::AFF1 — the experiment nobody has run in twenty years of interactome work

  the only published proximity-labelling maps of a chromatin-bound fusion oncoprotein are on EWS-FLI1/EWS-ETS; the top validated hit in one was a Golgi-endosome cargo receptor

• ← part ofEWSR1::FLI1 — the comparator that shows condensate formation has an optimum, and that forcing a fusion into the nucleolus switches it off

  the tyrosine-rich prion-like domain is the phase-competent module

• ← analogous toTAF15 (RBP56, TAF2N) - the FET-family, prion-like partner that donates a low-complexity N-terminus to ZNF384

  FET-family paralogue: both donate an N-terminal prion-like low-complexity region to a DNA-binding partner

• ← part ofEWSR1::ZNF384 - the same EWSR1 exon 7 fragment that drives Ewing sarcoma, fused instead to a zinc-finger protein, in B-ALL

  EWSR1 exon 7, the same junction as type 1 EWSR1::FLI1 - on this graph's own arithmetic, residues 1-264