Oncofusion

start / graph / fusion-ewsr1-fli1

Fusionestablished confidence 5 of 5: Multiple independent lines of evidence; treated as settled in the field.Q11-comparative

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

also called EWS-FLI1 · EWS::FLI1 · EWSR1-FLI1 · EWS-FLI-1 · t(11;22)(q24;q12) · type 1 / type 2 EWS-FLI1

In Ewing sarcoma, a bone and soft-tissue cancer of teenagers, half of a gene called EWSR1 - a floppy, sticky, low-complexity stretch of protein - is fused to the DNA-gripping end of FLI1. The sticky half recruits chromatin machinery to short repeated DNA sequences that are normally silent junk, switching on hundreds of genes that should never be on. There is no pocket anywhere on this protein, which is why after thirty years there is still no targeted drug.

The same thing, for a scientist

EWSR1::FLI1 results from t(11;22)(q24;q12) and was the founding example of a gene fusion joining a transactivating domain to an ETS DNA-binding domain [E1]. The chimera retains the **EWSR1 N-terminal low-complexity, SYGQ-rich, prion-like domain** and the **FLI1 C-terminal ETS DNA-binding domain**, losing EWSR1's RNA-recognition motif and zinc finger and FLI1's own N-terminal regions. Its defining mechanism is neomorphic DNA targeting: EWS/FLI binds **GGAA microsatellites**, which function as response elements only for the fusion and not for wild-type FLI1, with transactivation increasing as a function of GGAA repeat number [E2]. The low-complexity domain then does the work: it **retargets BAF (SWI/SNF) chromatin-remodelling complexes to those microsatellites, converting non-functional repeat DNA into de novo active enhancers** - a mechanism that depends on the prion-like character of the domain rather than on any folded surface [E3]. The same low-complexity interactions form transcriptional hubs whose valency can be tuned to modulate oncogenic transcription, and phase-separation behaviour of the fusion is itself modulated by the DNA-binding domain [E4]. Clinically, Ewing sarcoma is a disease of adolescents with 5-year EFS of 69% +/- 3% for non-metastatic disease on intensified chemotherapy and no benefit from that intensification in metastatic disease [E5]. Fusion architecture varies (type 1 n=296, type 2 n=133, non-type-1/2 n=91 of 565 informative Euro-E.W.I.N.G. 99 patients) but prospectively has **no prognostic value** [E6]. **Druggability: undrugged.** This is the clearest example in the graph of the same problem as KMT2A::AFF1 - no enzyme active site, no ligand-binding pocket - and, unlike KMT2A::AFF1, no menin equivalent either.

What it looks like
loading 9MWY…
PDB 9MWY ↗
shape

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.

Go deeper
What is kept and what is lost on each side1 deeper

Kept from EWSR1 (the 5' partner): the N-terminal transactivation domain, a low-complexity, SYGQ-rich, prion-like region composed largely of serine, tyrosine, glycine and glutamine repeats. It is intrinsically disordered - there is no fold, no cleft and no defined surface 3.

Lost from EWSR1: the RNA-recognition motif (RRM), the RGG boxes and the C-terminal zinc finger - i.e. all of EWSR1's RNA-binding chemistry. The PDB contains a solution structure of the EWS RRM (2CPE); it describes a module that is not in the oncoprotein, and citing it as a structure of EWS-FLI1 would be a mistake.

Kept from FLI1 (the 3' partner): the C-terminal ETS DNA-binding domain (a winged-helix-turn-helix fold), which is the only folded module in the entire fusion, plus the FLI1 C-terminal region.

Lost from FLI1: its N-terminal regions including its own transactivation domain and the PNT/SAM-like region.

Fusion architecture. Breakpoints vary: type 1 (EWSR1 exon 7 to FLI1 exon 6) and type 2 (EWSR1 exon 7 to FLI1 exon 5) are the commonest, with a substantial non-type-1/non-type-2 group. Distribution in the prospective Euro-E.W.I.N.G. 99 molecular cohort was type 1 n=296, type 2 n=133, other EWS-FLI1 n=91 and EWS-ERG n=45 of 565 informative patients 6. In roughly 85-90% of Ewing sarcomas the ETS partner is FLI1; ERG accounts for most of the remainder 6.

The honest structural situation

Of the two halves, one is a disordered polymer of repeats and the other is a 90-residue DNA-binding fold. The FLI1 ETS domain has been crystallised extensively, including - directly relevant here - bound to DNA containing three and four contiguous GGAA sites (PDB 9MX8, 9MWY), which is the structural counterpart of the microsatellite mechanism, and on a canonical single GGAA site (5JVT). Nothing exists, and arguably nothing can exist, for the EWSR1 low-complexity half in its functional state, because its function is its disorder. Any picture of the whole fusion is a schematic.

That asymmetry is the druggability problem in one sentence: the only folded part of the oncoprotein is a DNA-binding domain, and DNA-binding domains are notoriously poor small-molecule targets.

Mechanism: enhancer creation at junk DNA, driven by a prion-like domain

Class: tf-repression-corruption plus condensate-rewiring plus enhancer-hijack. All three labels apply, and the combination is what makes this fusion distinctive.

1. Neomorphic DNA targeting. EWS/FLI binds GGAA microsatellites - tandem repeats of the core ETS motif that are dispersed through the genome and are, in normal cells, transcriptionally inert. Gangwal and colleagues showed that these microsatellites act as EWS/FLI response elements, that transactivation scales with repeat number, and that they account for the activation of key EWS/FLI target genes 2. Wild-type FLI1 does not do this. The specificity is emergent from the fusion, which is why this is a neomorph and not a hyperactive transcription factor.

2. Enhancer creation. Binding alone would achieve little. Boulay and colleagues showed that the EWSR1 prion-like domain recruits the BAF (SWI/SNF) nucleosome-remodelling complex to those GGAA microsatellites, converting them into de novo active enhancers with the full complement of enhancer marks. The retargeting depends on the prion-like, low-complexity character of the domain - a mutant lacking it fails to recruit BAF - and it is cancer-specific, because normal cells have no protein that does this 3.

3. Condensate biology. The low-complexity domain drives multivalent, weak, transient interactions that assemble transcriptional hubs. Chong and colleagues showed that the level of low-complexity-domain interaction can be tuned - up or down - to modulate endogenous oncogenic transcription, implying the hub has an optimum rather than a simple on/off state 4. More recent biophysics shows that the fusion's phase-separation behaviour is modulated by its DNA-binding domain, so the condensate and the DNA-binding functions are coupled rather than independent [E4 counter-evidence].

Why this is the sharpest statement of the lab's problem. KMT2A::AFF1 has no pocket but does have an obligate, specific, folded partner - menin - whose peptide-binding cleft is drugged. EWSR1::FLI1 has no pocket and no obligate folded partner. Its 'partner' is a chromatin-remodelling complex recruited through a disordered surface with no defined binding site, and the relevant interaction is a distributed, multivalent, low-affinity network rather than a 1:1 interface. There is nothing shaped like a menin here. If an obligate-partner-interface strategy is to generalise, this is the fusion it has to explain, and at present it cannot.

Clinical phenotype, cell of origin and age

Disease. Ewing sarcoma: a small round blue cell tumour of bone (commonly femur, pelvis, chest wall) and soft tissue, with pain and a palpable mass, frequently metastatic to lung, bone and marrow at diagnosis.

Age. Adolescence, with a peak in the second decade; the Grier trial enrolled patients aged 30 years or younger 5. Rare under 5 and over 30.

Cell of origin: genuinely unresolved. Candidates are a mesenchymal stem/stromal cell and a neural-crest-derived progenitor; the tumour expresses neuroectodermal markers (CD99, NKX2-2) but arises in bone and soft tissue, and EWS-FLI1 is poorly tolerated by most primary human cells, which is precisely why identifying a permissive cell has been so hard. This uncertainty is itself informative: like the KMT2A fusions, this oncoprotein appears to need a narrow permissive developmental window.

Outcome. In 518 patients aged 30 or under with Ewing sarcoma or PNET of bone, adding ifosfamide and etoposide to vincristine-doxorubicin-cyclophosphamide-dactinomycin improved 5-year EFS from 54% +/- 4% to 69% +/- 3% and OS from 61% +/- 3.6% to 72% +/- 3.4% in the 398 patients with non-metastatic disease, but produced no significant improvement whatsoever in the 120 patients with metastatic disease 5. That dichotomy has barely moved in two decades and defines the unmet need.

Fusion type and prognosis: a corrected story. Retrospective studies suggested type 1 EWS-FLI1 carried a better prognosis. The prospective Euro-E.W.I.N.G. 99 analysis of 565 informative patients found no significant prognostic value of fusion type; the only signal, non-type-1/non-type-2 fusions at HR 1.38 (95% CI 0.96-2.0), was not significant (p=0.10) 6. Worth recording alongside the parallel correction in KMT2A-r AML, where t(9;11)'s supposed favourable prognosis also failed prospective confirmation - two instances of the same methodological trap.

Druggability: undrugged, and the record of what has been tried

There is no approved targeted therapy for EWSR1::FLI1. Standard care remains multiagent cytotoxic chemotherapy plus local control.

What has been tried, and what happened:

  • Direct interaction inhibition. YK-4-279 and its clinical analogue TK216 were developed to block the EWS-FLI1 interaction with RNA helicase A. Three registered trials exist: NCT02657005 (phase 1/2 in relapsed/refractory Ewing sarcoma, TERMINATED), NCT05046314 (phase 2 in relapsed/refractory Ewing sarcoma, COMPLETED without an approval following), and NCT03752138 (phase 1 with decitabine in AML, WITHDRAWN). This is the single most direct test of 'drug the disordered fusion's protein interaction' in oncology, and it has not produced a medicine.
  • Transcription-directed cytotoxics. Trabectedin and lurbinectedin interfere with the EWS-FLI1 transcriptional program; activity is modest and non-specific.
  • Synthetic lethality. Ewing lines are strikingly sensitive to PARP inhibitors in vitro; single-agent clinical trials were disappointing.
  • Epigenetic co-factors. LSD1, BET and BAF-directed agents follow from the enhancer-creation mechanism and remain preclinical or early-phase.
  • CD99 (MIC2) is a near-universal surface marker in Ewing sarcoma and an immunotherapy target under investigation - the same strategic move as targeting CD19 in KMT2A::AFF1 ALL: give up on the oncoprotein, hit the surface phenotype.

Status: undrugged. The comparative reading: EWSR1::FLI1 shows what KMT2A::AFF1 would look like without menin. Two fusions, both pocketless transcriptional neomorphs, both differentiation-distorting, both driven by low-complexity partner sequence - and the one with an obligate folded adaptor has a drug while the one without has nothing. That single contrast is the strongest argument in this graph for the obligate-partner-interface hypothesis being worth pursuing, and it simultaneously shows its limit: it only works where such an adaptor exists.

Facts and figures
gene5EWSR1
gene3FLI1
cytobandt(11;22)(q24;q12)
prognosis5-year EFS 69% +/- 3% and OS 72% +/- 3.4% for non-metastatic disease with ifosfamide/etoposide-intensified chemotherapy (versus 54% +/- 4% EFS without), and no significant benefit from intensification in metastatic disease (n=120). Fusion type (type 1 versus type 2 versus other) has no significant prognostic value prospectively.
uniprot5Q01844
uniprot3Q01543
uniprot5 verifiedQ01844 = EWS_HUMAN, gene EWSR1, RNA-binding protein EWS, 656 aa, Homo sapiens (UniProt REST, checked 2026-09-18)
uniprot3 verifiedQ01543 = FLI1_HUMAN, gene FLI1, Friend leukemia integration 1 transcription factor, 452 aa, Homo sapiens (UniProt REST, checked 2026-09-18)
exon numbering noteType 1 and type 2 designations and their exon junctions follow the standard Ewing-sarcoma diagnostic convention used in the cited trials; EWSR1 exon numbering differs between annotation sets, so treat the numbers as the conventional labels rather than as coordinates.
PDB noteNo structure of the intact EWSR1::FLI1 fusion protein exists and, for the EWSR1 half, essentially none is possible: its function is its disorder. Every entry here covers the FLI1 ETS DNA-binding domain, the only folded module in the oncoprotein. 9MWY and 9MX8 = FLI1 DBD bound to DNA containing four and three contiguous GGAA sites respectively - the structural counterpart of the GGAA-microsatellite mechanism; 5JVT = FLI1 DBD on a single canonical GGAA-containing 11-mer; 1FLI = the apo FLI1 DNA-binding domain. Deliberately EXCLUDED: 2CPE, the solution structure of the EWS RNA-recognition motif - it resolves and is genuine EWSR1, but the RRM is lost from the fusion, so citing it as a fusion structure would mislead. All entries verified against the RCSB entry API on 2026-09-18.
mechanism class noteAll three apply and the combination is the mechanism: a corrupted ETS transcription factor (tf-repression-corruption) whose low-complexity domain forms multivalent transcriptional hubs (condensate-rewiring) and converts inert GGAA microsatellites into de novo enhancers by recruiting BAF (enhancer-hijack). The enhancer-hijack here is neomorphic enhancer CREATION rather than the capture of a pre-existing enhancer.
druggability classundrugged
druggability noteNo approved targeted therapy after more than thirty years. No enzyme active site, no ligand-binding pocket, and - unlike KMT2A::AFF1 - no obligate folded partner adaptor equivalent to menin. The only folded module is a DNA-binding domain. The one clinical programme aimed at an EWS-FLI1 protein interaction (TK216) has three registered trials, one terminated, one withdrawn and one completed without an approval.
stage of hematopoiesisNot haematopoietic. Tissue cell of origin is genuinely unresolved: the leading candidates are a mesenchymal stem/stromal cell of bone or soft tissue and a neural-crest-derived progenitor. The tumour expresses neuroectodermal markers (CD99, NKX2-2) yet arises in bone and soft tissue, and EWSR1::FLI1 is poorly tolerated by most primary human cells, which is itself evidence for a narrow permissive developmental window - the same constraint the KMT2A fusions show.
age peaksecond decade (adolescence); rare under 5 and over 30
developmental order noteNull because this is a non-haematopoietic comparator with no position on the haematopoietic developmental axis, and because the cell of origin is itself unresolved. If ranked on a mesenchymal axis it would sit at the progenitor end, but that would be an assertion the literature does not support.
drugs noteNo approved targeted agent exists. TK216 (and its parent compound YK-4-279) is the only clinical-stage molecule aimed at the fusion's protein interactions; trabectedin/lurbinectedin, PARP inhibitors, LSD1 and BET inhibitors and anti-CD99 immunotherapy are the other explored routes. None has a node in this graph.
trials noteTK216 trials, included specifically to document the negative therapeutic record: NCT02657005 terminated, NCT05046314 completed without approval, NCT03752138 withdrawn. Status verified against the ClinicalTrials.gov API on 2026-09-18.
diseases
[
  "Ewing sarcoma of bone",
  "extraskeletal Ewing sarcoma",
  "peripheral primitive neuroectodermal tumour"
]
frequency
{
  "ewing_sarcoma_ews_fli1_pct": 85,
  "ewing_sarcoma_ews_fli1_pct_note": "Conventionally quoted as 85-90% of Ewing sarcomas, with EWS-ERG accounting for most of the remainder; in the Euro-E.W.I.N.G. 99 molecular cohort 520/565 informative cases were EWS-FLI1 and 45/565 (8.0%) EWS-ERG, i.e. 92% FLI1 among ETS-fusion-positive cases typed.",
  "type1_n_of_565": 296,
  "type2_n_of_565": 133,
  "non_type1_non_type2_ews_fli1_n_of_565": 91,
  "ews_erg_n_of_565": 45
}
breakpoints
[
  {
    "id": "bp-ewsr1-e7-fli1-e6-type1",
    "gene5_exon_last": 7,
    "gene3_exon_first": 6,
    "frequency_pct": 52.4,
    "population": "type 1 EWS-FLI1; 296 of 565 informative patients in the prospective Euro-E.W.I.N.G. 99 molecular cohort",
    "domains_retained_5": [
      "N-terminal transactivation domain (SYGQ-rich low-complexity, prion-like)"
    ],
    "domains_lost_5": [
      "RGG1 box",
      "RNA-recognition motif (RRM)",
      "RGG2/RGG3 boxes",
      "C-terminal zinc finger",
      "nuclear localisation signal region"
    ],
    "domains_retained_3": [
      "ETS DNA-binding domain (winged helix-turn-helix)",
      "FLI1 C-terminal region"
    ],
    "domains_lost_3": [
      "FLI1 N-terminal region",
      "FLI1 PNT/SAM-like region",
      "FLI1 own transactivation domain"
    ],
    "phenotype_notes": "The commonest architecture. Retrospectively associated with better prognosis; prospectively NOT prognostic. The retained EWSR1 half contains the prion-like domain that recruits BAF complexes, and the retained FLI1 half contains the only folded module in the entire oncoprotein."
  },
  {
    "id": "bp-ewsr1-e7-fli1-e5-type2",
    "gene5_exon_last": 7,
    "gene3_exon_first": 5,
    "frequency_pct": 23.5,
    "population": "type 2 EWS-FLI1; 133 of 565 informative patients in the Euro-E.W.I.N.G. 99 molecular cohort",
    "domains_retained_5": [
      "N-terminal transactivation domain (SYGQ-rich low-complexity, prion-like)"
    ],
    "domains_lost_5": [
      "RGG1 box",
      "RNA-recognition motif (RRM)",
      "RGG2/RGG3 boxes",
      "C-terminal zinc finger"
    ],
    "domains_retained_3": [
      "ETS DNA-binding domain",
      "FLI1 C-terminal region",
      "an additional FLI1 exon-5-encoded segment relative to type 1"
    ],
    "domains_lost_3": [
      "FLI1 N-terminal region",
      "FLI1 PNT/SAM-like region"
    ],
    "phenotype_notes": "Type 2 includes one extra FLI1 exon relative to type 1 and shows differential transactivation strength in reporter assays, but no clinical prognostic difference in prospective analysis. Non-type-1/non-type-2 architectures accounted for a further 91 of 565 patients and carried a non-significant HR of 1.38."
  }
]
PDB IDS
[
  "9MWY",
  "9MX8",
  "5JVT",
  "1FLI"
]
mechanism class
[
  "tf-repression-corruption",
  "condensate-rewiring",
  "enhancer-hijack"
]
drugs
[]
key interactors
[
  "BAF / SWI/SNF complex (recruited by the prion-like domain)",
  "RNA helicase A / DHX9",
  "GGAA microsatellite DNA",
  "RNA polymerase II",
  "LSD1/NuRD"
]
trials
[
  "NCT02657005",
  "NCT05046314",
  "NCT03752138"
]
How this could bite us back
  • Delivery / tractability: the mechanistically essential half of the oncoprotein is intrinsically disordered, with no pocket, no defined interface and no stable conformation to design against. This is not a hard target, it is a differently shaped problem, and thirty years of failure should be taken as evidence rather than as insufficient effort.
  • Essential-function toxicity: every candidate route hits machinery the body needs. BAF/SWI-SNF is required in essentially every cell; LSD1 and BET proteins are general transcriptional regulators; DNA-binding-domain inhibition would also hit wild-type FLI1, which is required for haematopoiesis and vascular development.
  • Compensation: the fusion activates hundreds of targets through created enhancers, so partial inhibition is likely to shift rather than shut down the program. Chong and colleagues' finding that the hub has an interaction optimum raises the specific hazard that a partial inhibitor could tune transcription in an unhelpful direction.
  • Resistance: EWS-ERG and other EWS-ETS variants provide a ready-made escape from any FLI1-specific agent, and about 8% of cases already use ERG.
  • Modelling bite-back: GGAA microsatellites are not conserved in mouse, so preclinical models may systematically mis-predict both efficacy and toxicity - a structural flaw in the evidence base rather than a property of any one drug.
  • Clinical bite-back: with non-metastatic disease already ~70% curable and metastatic disease stubbornly ~20-25%, a new agent must be tested in the metastatic setting where the biology may be different and the patients most fragile.
Still open
  1. EWSR1::FLI1 is the same problem as KMT2A::AFF1 - a pocketless transcriptional neomorph - but with no menin-equivalent adaptor. Is there any obligate, folded, non-redundant partner of this fusion that has simply not been found, or is the absence real? A proteome-scale search for a single required folded interactor, benchmarked against what menin looks like for KMT2A, is the decisive experiment and would either extend the lab's hypothesis to solid tumours or falsify its generality.
  2. BAF recruitment is mediated by a disordered domain with no defined interface. Can a distributed, multivalent, low-affinity recruitment event be drugged at all - by a molecular glue, a condensate-partitioning agent, or a valency disruptor - or is the interface concept simply inapplicable here?
  3. GGAA microsatellite length is human-polymorphic and not conserved in mouse, so the disease's core response elements cannot be modelled in standard mice. How much of the field's failure to find a drug is attributable to this modelling gap rather than to the target's intrinsic properties?
  4. The cell of origin is still unknown after three decades, and EWSR1::FLI1 is toxic to most primary human cells. Is identifying the permissive cell a prerequisite for a drug, as the KMT2A::MLLT1 age paradox suggests it may be for leukemia?
  5. Metastatic Ewing sarcoma outcomes have not improved since 2003 despite intensification. Is the correct near-term target the oncoprotein at all, or the surface phenotype (CD99) - the same strategic choice that blinatumomab represents in KMT2A::AFF1 ALL?
Every source behind this entry (6)
  1. E1established confidence 5 of 5: Multiple independent lines of evidence; treated as settled in the field.paper

    The t(11;22) of Ewing sarcoma creates a gene fusion joining EWSR1 to an ETS DNA-binding domain gene, FLI1.

    PMID 1522903DOI
    What this evidence cannot show

    A cloning and characterisation paper. It establishes the fusion architecture but not the GGAA-microsatellite targeting, the BAF-retargeting mechanism, the cell of origin or any therapeutic implication - all of which took a further 15-25 years.

    Evidence that argues against this (1)
    • FLI1 is one of at least two interchangeable ETS partners, established two years after the cloning paper and visible in this entry's own trial cohort. Sorensen et al. identified a second Ewing's sarcoma translocation, t(21;22)(q22;q12), fusing the IDENTICAL EWS nucleotide sequences to ERG on 21q22 and producing a hybrid EWS/ERG protein, and noted that t(11;22) accounts for 85% of Ewing's sarcoma and related tumours, leaving a substantial remainder. Their conclusion is that fusion of EWS to different ETS family members produces similar disease phenotypes. The comparative lesson is the opposite of partner specificity - which is also why the Euro-E.W.I.N.G. 99 series in [E6] contains 45 EWS-ERG patients alongside 520 EWS-FLI1 patients, with no prognostic difference between them. Sorensen PH et al. 1994
  2. E2strong confidence 4 of 5: Replicated in peer-reviewed work by more than one group.paper

    GGAA microsatellites function as EWS/FLI response elements in Ewing sarcoma: EWS/FLI binds them, transactivation increases with the number of GGAA repeats, and they account for activation of key EWS/FLI target genes - an activity not shared by wild-type FLI1.

    Microsatellites as EWS/FLI response elements in Ewing's sarcoma · Gangwal K, Sankar S, Hollenhorst PC, et al. · Proceedings of the National Academy of Sciences USA 2008
    PMID 18626011DOI
    What this evidence cannot show

    Established in Ewing sarcoma cell lines with reporter assays and ChIP; microsatellite length is polymorphic between individuals and between species, which limits mouse modelling and means the human-specific repeat landscape is part of the mechanism. The paper does not explain how binding is converted into enhancer activity.

    Evidence that argues against this (1)
    • Activation at GGAA repeats is half of what the fusion does, and the other half runs through exactly the canonical ETS sites the microsatellite model sets aside. Riggi et al. showed that EWS-FLI1 uses DIVERGENT chromatin-remodelling mechanisms: at GGAA repeats that lack evolutionary conservation and regulatory potential in other cell types, EWS-FLI1 multimers induce chromatin opening and create de novo enhancers that physically contact target promoters; but at CONSERVED enhancers carrying canonical ETS motifs it INACTIVATES them by displacing wild-type ETS transcription factors, repressing tumour suppressors and mesenchymal lineage regulators. An account in which GGAA microsatellites 'account for activation of key EWS/FLI target genes' is accurate about the activated half and silent about a repressive arm at ordinary ETS sites - the same sites wild-type FLI1 uses, which also qualifies 'an activity not shared by wild-type FLI1': the fusion competes with wild-type ETS factors on their own ground as well as inventing new ground. Riggi N et al. 2014
  3. E3strong confidence 4 of 5: Replicated in peer-reviewed work by more than one group.paper

    The EWSR1 prion-like low-complexity domain retargets BAF (SWI/SNF) chromatin-remodelling complexes to GGAA microsatellites, converting inert repeat DNA into de novo active enhancers, and this retargeting depends on the prion-like character of the domain rather than on a folded binding surface.

    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 xenograft work with degron/knockdown and domain-swap constructs; the interaction between a disordered domain and a multi-subunit remodeller has no defined stoichiometry or interface, so 'retargeting' is a functional rather than a structural description. Whether BAF recruitment is necessary in a primary human tumour, as opposed to established lines, is not directly shown.

    Evidence that argues against this (1)
    • Graded against this graph's own condensate evidence standards (`localization-condensate-evidence-standards`), 'depends on the prion-like character of the domain' is not what a domain swap establishes. Chong et al. imaged transcription-factor low-complexity domains, including the EWS LCD, by live-cell single-molecule microscopy and found they form local high-concentration interaction HUBS at synthetic and endogenous genomic loci which stabilise DNA binding, recruit RNA Pol II and activate transcription; the LCD-LCD interactions are highly dynamic, selective for binding partner, DIFFERENTIALLY sensitive to hexanediols between LCDs, and under physiological conditions occur WITHOUT DETECTABLE PHASE SEPARATION. Recovering BAF retargeting after swapping in another prion-like domain therefore shows that selective multivalent LCD interaction is sufficient - it does not show that prion-like or phase-separating character is the operative property, and hexanediol-based evidence for that property is not interpretable across different LCDs. Chong S et al. 2018
  4. E4moderate confidence 3 of 5: One strong study, or several smaller consistent ones.paper

    Low-complexity-domain interactions of EWS::FLI1 form transcriptional hubs whose interaction level can be tuned up or down to modulate endogenous oncogenic transcription.

    Tuning levels of low-complexity domain interactions to modulate endogenous oncogenic transcription · Chong S, Graham TGW, Dugast-Darzacq C, Dailey GM, Darzacq X, Tjian R · Molecular Cell 2022
    PMID 35483357DOI
    What this evidence cannot show

    Single-molecule imaging and engineered interaction-tuning in cell lines; 'tunable' is demonstrated with genetic and optogenetic-style perturbations, not with drug-like molecules, so tractability by a small molecule remains unestablished. The relationship between in-cell hubs and classical in vitro phase separation is contested.

    Evidence that argues against this (2)
    • The one clinical programme that attempted to drug an EWS-FLI1 protein interaction (TK216, targeting the EWS-FLI1/RNA helicase A interaction) has produced no approved medicine: NCT02657005 was terminated, NCT05046314 completed without approval, and NCT03752138 was withdrawn. Tunability in cells has not translated into druggability in patients. ClinicalTrials.gov 2026
    • The fusion's phase-separation behaviour is modulated by its DNA-binding domain, so the condensate and DNA-binding functions are coupled; a model treating the low-complexity domain as an independent, tunable module is an oversimplification. Selig EE et al. 2025
  5. E5established confidence 5 of 5: Multiple independent lines of evidence; treated as settled in the field.paper

    Adding ifosfamide and etoposide to standard chemotherapy improved 5-year event-free survival from 54% +/- 4% to 69% +/- 3% and overall survival from 61% +/- 3.6% to 72% +/- 3.4% in 398 patients with non-metastatic Ewing sarcoma or PNET of bone, but produced no significant improvement in the 120 patients with metastatic disease.

    PMID 12594313DOI
    What this evidence cannot show

    A trial of the 1990s reporting 2003; patients aged 30 or younger with bone primaries, so it does not describe extraskeletal Ewing sarcoma or adults over 30. Fusion status was not an eligibility criterion, so a small number of enrolled 'PNET of bone' cases may not have carried EWSR1::FLI1. Absolute numbers are historical; the metastatic-disease result, however, has not been meaningfully improved since.

    Evidence that argues against this (1)
    • The randomised European comparison does not reproduce ifosfamide-plus-etoposide as the operative variable, and its absolute numbers are lower. EURO EWING 2012 randomised 640 patients aged 2-49 with newly diagnosed Ewing sarcoma 1:1 between the European VIDE-based regimen and the US VDC/IE-based regimen - BOTH of which contain ifosfamide and etoposide - and found 3-year event-free survival of 61% in the European arm versus 67% in the US arm (adjusted HR 0.71, 95% credible interval 0.55-0.92 favouring the US regimen), with grade 3-5 febrile neutropenia in 74% versus 58%. Benefit therefore tracks the whole regimen, its drug pairing and its scheduling rather than the addition of two agents, and a 3-year EFS of 61-67% across all stages including metastatic disease is the number to read against the 69% five-year figure quoted here for non-metastatic bone primaries in patients aged 30 or under. Brennan B et al. 2022
  6. E6strong confidence 4 of 5: Replicated in peer-reviewed work by more than one group.paper

    In 565 informative patients from the prospective Euro-E.W.I.N.G. 99 trial the fusion-type distribution was type 1 (n=296), type 2 (n=133), non-type-1/non-type-2 EWS-FLI1 (n=91) and EWS-ERG (n=45), and fusion type had no significant prognostic value for progression or relapse, contradicting earlier retrospective reports.

    PMID 20308673DOI
    What this evidence cannot show

    Only 703 of 1,957 registered tumours were accessible for molecular study and 565 were informative for the prognostic analysis, so selection by tissue availability is possible; median follow-up was 4.5 years, and a modest effect on late events could be missed. The non-type-1/2 group's HR of 1.38 (95% CI 0.96-2.0, p=0.10) is a non-significant signal that larger cohorts could resolve either way.

    Evidence that argues against this (1)
    • Earlier retrospective work reported that EWS-FLI1 fusion transcript structure was an independent determinant of prognosis, with type 1 faring better. The prospective analysis did not confirm it - the same pattern as t(9;11) in KMT2A-rearranged AML. de Alava E et al. 1998
Connected to