Oncofusion

start / graph / fusion-pml-rara

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

PML::RARA, t(15;17)(q24;q21)

also called PML-RARA · PML-RARalpha · PML/RARA · MYL-RARA · t(15;17) APL · acute promyelocytic leukemia driver

A swap between chromosomes 15 and 17 joins a protein that builds little clumps inside the nucleus to the receptor for vitamin A, and the hybrid jams white-blood-cell development at one stage. Because the vitamin A pocket is still there and the clumping part can be poisoned by arsenic, two non-chemotherapy drugs destroy the hybrid protein outright - turning the most lethal leukaemia into the most curable one.

The same thing, for a scientist

t(15;17) fuses most of **PML** (P29590) — including its **RBCC/TRIM motif (RING, B-boxes, coiled coil)** and the cysteine-rich zinc-finger region — to a large part of **RARA** (P10276) **retaining both its DNA-binding and hormone(ligand)-binding domains**, producing a predicted 106 kDa chimera with altered transactivating properties relative to wild-type RARA, with identical fusion points recurring across patients [E1]. Mechanistically the fusion is a **ligand-dependent transcription factor corrupted into a constitutive repressor**: the retained RARA moiety binds retinoic-acid-response elements but, because PML's coiled coil forces it into homo-multimers, it recruits corepressor/HDAC complexes at physiological retinoid concentrations and blocks promyelocyte differentiation. The therapeutic breakthrough came from two independent, structurally distinct handles on the *same* protein. **ATRA** occupies the retained RARA **ligand-binding pocket** and both switches the fusion to an activator and triggers its catabolism [E2][E5]. **Arsenic trioxide binds directly to cysteine residues in the zinc fingers of PML's RBCC domain**, inducing PML oligomerisation, increased UBC9 recruitment, enhanced SUMOylation and degradation [E3]. Critically, **degradation rather than differentiation is what eradicates the leukaemia-initiating cell**: ATRA clears LICs in PML-RARA mouse APL, the two agents synergise for LIC clearance through cooperative PML-RARA degradation without enhancing differentiation, and PLZF-RARA APL differentiates fully under ATRA yet loses no LICs and does not remit — formally uncoupling differentiation from cure [E5]. Clinically, ATRA plus arsenic trioxide gives **100% complete remission and 97% 2-year event-free survival in low-to-intermediate-risk APL, with no chemotherapy at all** [E4].

What it looks like
loading 3KMR…
PDB 3KMR ↗
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

Kept from PML (5-prime partner): most of the PML sequence 1, comprising the RBCC / TRIM motif:

  • the RING finger (PDB 1BOR, the zinc-loaded PML RING) 6;
  • B-box 1 and B-box 2 (PDB 2MVW, the TRIM19/PML B-box-1 solution structure; PDB 6IMQ, PML B1-box multimers) 6;
  • the alpha-helical coiled-coil (CC) domain, which drives homodimerisation and higher-order multimerisation;
  • the cysteine-rich zinc-coordinating residues that turn out to be the arsenic binding site 3;
  • SUMO-interacting and SUMOylation sites (PDB 4WJN, SUMO1 with phosphorylated PML; 6UYO, acetylated SUMO1 with PML-SIM) 6.

Lost from PML: the C-terminal region including the nuclear-localisation and SUMO-interaction-motif-containing tail of some isoforms.

Kept from RARA (3-prime partner): a large part of RARA including BOTH its DNA-binding domain and its hormone-binding domain 1. This is the decisive retention. The DNA-binding domain is the C4 zinc-finger pair that reads retinoic-acid response elements as a heterodimer with RXR; the hormone-binding domain is the ligand-binding pocket that binds all-trans retinoic acid (structures 3KMR, RARA LBD with an agonist and a coactivator fragment; 1DKF, the RAR-RXR LBD heterodimer) 6.

Lost from RARA: the A domain / N-terminal AF-1 region.

Summary architecture: a self-multimerising, arsenic-sensitive, SUMOylation-prone scaffold fused to an intact nuclear-receptor DNA-binding plus ligand-binding module. Every element needed for both therapies is present in the fusion by construction.

WHY THIS ONE BECAME CURABLE (PI depth request)1 deeper

APL went from the most rapidly fatal leukaemia (death from haemorrhage within days) to the most curable acute leukaemia in adults, treatable without any chemotherapy 4. The reasons are structural and they are different from BCR::ABL1's.

1. The retained RARA ligand-binding pocket is a natural small-molecule site — the drug already existed. RARA is a nuclear receptor. Its whole biological purpose is to bind a small lipophilic ligand (all-trans retinoic acid) in a deep hydrophobic pocket and change conformation. The fusion keeps that pocket intact 1. So the 'drug discovery' step for ATRA was not discovery at all — the endogenous ligand of the retained domain is the drug, and pharmacological doses shift the fusion from corepressor-bound to coactivator-bound. Compare the rest of this beat: ETV6::RUNX1, TCF3::PBX1, TCF3::HLF, RUNX1::RUNX1T1, CBFB::MYH11, CBFA2T3::GLIS2, DUX4::IGH and both NUP98 fusions retain no ligand pocket of any kind. The only two curable fusions in the set are the two whose retained oncogenic module has a natural small-molecule binding site — an ATP pocket (BCR::ABL1) or a hormone pocket (PML::RARA). That is the central comparative finding of this beat.

Historically this was demonstrated before the mechanism was understood: 24 APL patients, including eight refractory to chemotherapy, all attained complete remission on all-trans retinoic acid without bone-marrow hypoplasia, with differentiation to functional granulocytes and — strikingly — no coagulation abnormalities during therapy 2. Side effects were dry lips and headache. Eight of 24 relapsed at 2-5 months, which is exactly why ATRA alone is not the answer 2.

2. The PML RBCC/coiled coil makes the fusion an obligate multimer, and multimers are degradable. PML's coiled coil forces the fusion into homo-multimers. That has two consequences. It is how the fusion represses (multivalent corepressor recruitment at RARE sites, at retinoid concentrations that would not repress through a normal RAR-RXR heterodimer). And it is why the fusion is degradable: arsenic binds directly to cysteine residues in the zinc fingers within the RBCC domain of PML-RARalpha and PML, and arsenic binding induces PML oligomerisation, which increases interaction with the SUMO-conjugating enzyme UBC9, giving enhanced SUMOylation and degradation 3. Arsenic is therefore not a generic cytotoxin here — it is a direct, covalent-style ligand of a specific retained structural module, and the authors note that identifying PML as the direct target explains the drug's specificity for APL 3.

This is the cleanest natural example in oncology of a degrader-sensitive fusion target: the fusion's own oligomerisation architecture is the vulnerability. Note the difference from NUP98 fusions, whose FG-repeat multivalency is interchangeable disorder with no binding site — PML's multimerisation runs through defined, zinc-coordinating, ligandable cysteines.

3. Degradation, not differentiation, is what cures — and this was proved by dissociation. This is the deepest point and it is often got wrong. The phenotypic hallmark of ATRA is dramatic differentiation of APL blasts. But 5 showed:

  • ATRA triggers growth arrest of leukaemia-initiating cells (LICs) ex vivo and clears them in PML-RARA mouse APL in vivo;
  • in PLZF-RARA mouse APL, ATRA triggers full differentiation but NO LIC loss and no disease remission — differentiation and LIC loss are therefore uncoupled;
  • ATRA and arsenic synergise to clear LICs through cooperative PML-RARA degradation, and this combination does NOT enhance differentiation;
  • a cAMP-dependent phosphorylation site in PML-RARA is required for ATRA-induced degradation and LIC clearance, and activating cAMP signalling enhances LIC loss.

The authors' conclusion is the generalisable statement: transcriptional activation of PML-RARA likely controls differentiation, but its catabolism triggers LIC eradication and long-term remission, and therapy-triggered degradation of oncoproteins could be a general strategy to eradicate cancer stem cells 5.

So PML::RARA is the founding proof-of-concept for targeted protein degradation in cancer, two decades before PROTACs and molecular glues became a field. For this beat that is the most important transferable idea: for the eight fusions with no pocket, degradation of the fusion may be a more realistic objective than inhibition of it, provided a handle on the fusion's scaffold can be found.

4. The clinical result, to be concrete. In a randomised phase 3 trial in low-to-intermediate-risk APL, ATRA plus arsenic trioxide gave complete remission in 77/77 evaluable patients (100%) versus 75/79 (95%) for ATRA plus chemotherapy, and 2-year event-free survival of 97% versus 86% (p<0.001 for non-inferiority, p=0.02 for superiority), with better overall survival, less haematologic toxicity and fewer infections but more hepatic toxicity 4. A chemotherapy-free cure for an acute leukaemia. Nothing else in this beat is close.

5. What did NOT make it curable. Not the PML RING finger, not the B-boxes, not the SUMO sites as drug targets in their own right — those are structural features that enable the arsenic mechanism rather than being targeted directly. And notably, the PML moiety's own tumour-suppressor and nuclear-body functions were not the therapeutic lever.

Side-by-side with BCR::ABL1: two curable fusions, two different reasons
BCR::ABL1PML::RARA
retained oncogenic moduleABL1 tyrosine kinaseRARA nuclear receptor
natural small-molecule siteATP pockethormone (retinoid) pocket
what the drug doesinhibits catalysisswitches the receptor AND triggers catabolism
selectivity comes froma distinctive inactive conformation (DFG-out)the pocket's natural ligand specificity, plus a second, orthogonal target (PML cysteines)
5-prime partner's rolecoiled coil forces oligomerisation → constitutive kinasecoiled coil forces multimerisation → repression and degradability
second drug axisallosteric myristate pocket (asciminib) - a pocket the fusion VACATEDarsenic on PML RBCC cysteines - a site the fusion RETAINED 3
cure or control?control: lifelong TKI, LSC persistscure: chemotherapy-free, 97% 2-y EFS 4
whykinase inhibition suppresses but does not destroy the driver or the stem celldegradation destroys the driver and eradicates the LIC 5

The last row is the important one. Both fusions had an ideal small-molecule site. Only PML::RARA is actually cured, and the difference is degradation versus inhibition. BCR::ABL1 patients take a pill indefinitely because imatinib occupies the kinase; APL patients finish treatment because ATRA plus arsenic destroys the protein and with it the leukaemia-initiating cell 5. For every undruggable fusion in this beat, that comparison argues for pursuing a degrader rather than an inhibitor even where a pocket does exist.

Clinical phenotype, age and prognosis

Disease: acute promyelocytic leukaemia (APL, FAB M3) — a differentiation arrest at the promyelocyte stage, giving hypergranular blasts with faggots of Auer rods (or the microgranular M3v variant).

The defining emergency is coagulopathy. APL presents with a life-threatening bleeding diathesis from combined disseminated intravascular coagulation and hyperfibrinolysis, with annexin II on the promyelocyte surface a proposed contributor. Historically this is what killed patients within days of diagnosis, before any therapy could work; it is still the main cause of early death. ATRA's original 24-patient series specifically noted no abnormalities in coagulation parameters during therapy, i.e. differentiation therapy also fixes the coagulopathy rather than worsening it as cytotoxic lysis does 2. ATRA is started on clinical suspicion, before genetic confirmation — one of very few places in oncology where that is correct practice.

Age: occurs across the age range. It is roughly 5-10% of paediatric AML, with the proportion rising through adolescence into young adulthood, and it is notably rare in infants — the opposite developmental distribution from CBFA2T3::GLIS2 or KMT2A-rearranged infant leukaemia. Paediatric APL is biologically the same disease as adult APL and is treated on the same ATRA-plus-arsenic principle, with the important caveat of higher pseudotumour cerebri risk with ATRA in children.

Cytogenetically visible — t(15;17)(q24;q21) is a balanced translocation detectable by karyotype, FISH or RT-PCR.

Prognosis: the best of any acute leukaemia. Complete remission 100% and 2-year event-free survival 97% with ATRA plus arsenic trioxide in low-to-intermediate-risk disease, versus 95% CR and 86% EFS with ATRA plus chemotherapy 4. ATRA plus chemotherapy already produced cure rates exceeding 80% before arsenic was added 4. Risk stratification is by presenting white-cell count (the trial's low-to-intermediate group was WBC <=10 x 10^9/L); high-count APL retains a higher early-death and relapse risk and generally still receives some cytotoxic therapy.

Residual failures are almost entirely early deaths from haemorrhage before or during the first days of ATRA, plus differentiation syndrome, rather than refractory disease.

Druggability: clinically validated, twice over, on two different modules

Status: clinically validated. This is the only fusion in this beat with two independently approved, mechanism-matched agents acting on two different retained structural modules of the same fusion protein.

ATRA — occupies the retained RARA ligand-binding pocket; converts the fusion from repressor to activator, drives differentiation, and triggers degradation via a cAMP-dependent phosphorylation site 25. Approved. ATRA alone gives universal remission but frequent early relapse 2, so it is never used alone.

Arsenic trioxide — binds cysteines in the PML RBCC zinc fingers, induces oligomerisation, UBC9 recruitment, SUMOylation and degradation 3. Approved. Synergises with ATRA specifically for LIC clearance via cooperative degradation 5.

cAMP agonism is a third, mechanism-derived and untested-in-humans idea from 5: the cAMP-dependent phosphorylation site is required for ATRA-induced degradation and LIC clearance, and activating cAMP signalling enhanced LIC loss in mouse APL. This has not become a clinical strategy and is a concrete, cheap, mechanistically justified thing to test.

Resistance is well characterised and instructive: point mutations in the RARA ligand-binding domain cause ATRA resistance, and mutations in the PML B2 domain (notably A216V) cause arsenic resistance. Each resistance mechanism maps to the module its drug binds — the same on-target resistance logic as BCR::ABL1 T315I, and further confirmation that both agents act where the mechanism says they do.

Variant RARA fusions matter clinically and mechanistically. PLZF(ZBTB16)::RARA is ATRA-unresponsive for cure despite differentiating 5, and STAT5B::RARA and NPM1::RARA have their own sensitivities. The graph treats these as a family: retaining the RARA ligand pocket is necessary but not sufficient — the 5-prime partner must also make the fusion degradable. PLZF-RARA is the negative control that proves it.

Facts and figures
gene5PML
gene3RARA
cytobandt(15;17)(q24;q21)
cytoband noteOlder literature writes this as t(15;17)(q22;q21); the q24 assignment for PML reflects current genome builds. Both refer to the same translocation.
frequency notePML::RARA is present in essentially all true APL (recorded as 98 to leave room for the variant RARA fusions - PLZF/ZBTB16::RARA, STAT5B::RARA, NPM1::RARA - which together account for a small percentage of RARA-rearranged APL). The paediatric and adult AML shares of ~8% and ~10% are standard literature ranges we did NOT verify in a single cited cohort and should be treated as approximate.
prognosisThe best of any acute leukaemia and the only chemotherapy-free cure: complete remission 100% (77/77) and 2-year event-free survival 97% with ATRA plus arsenic trioxide in low-to-intermediate-risk APL (WBC <=10 x 10^9/L), versus 95% CR and 86% EFS with ATRA plus chemotherapy, with better overall survival and less haematologic toxicity [E4]. ATRA plus chemotherapy already exceeded 80% cure before arsenic [E4]. Residual mortality is dominated by EARLY HAEMORRHAGIC DEATH before or during the first days of therapy, and by differentiation syndrome, rather than by refractory disease; high presenting white count remains the adverse risk factor.
uniprot5P29590
uniprot3P10276
PDB noteThere is NO structure of the PML::RARA fusion, none of ATRA bound to the fusion's ligand-binding domain, and - notably - none of arsenic bound to PML: the arsenic site on the RBCC cysteines is defined biochemically, not crystallographically [E3][E6]. 3KMR is the wild-type RARA ligand-binding domain with an agonist (Am580, not ATRA) and a coactivator fragment - the retained pocket that makes ATRA work. 1DKF is the RAR-RXR LBD heterodimer. 1BOR is the zinc-loaded PML RING, 2MVW the PML B-box-1, 6IMQ PML B1-box MULTIMERS (the closest structural evidence for the multimerisation that underlies both repression and degradability), and 4WJN SUMO1 bound to phosphorylated PML (the modification arsenic enhances). Every entry is an isolated wild-type domain.
mechanism class noteA LIGAND-DEPENDENT transcription factor converted into a constitutive repressor: the retained RARA moiety binds retinoic-acid response elements but PML-driven multimerisation makes it recruit corepressor/HDAC complexes at physiological retinoid levels. Distinctive within the class because the corruption is REVERSIBLE by ligand - the only fusion in this beat where a small molecule can flip the transcriptional sign rather than merely block the protein. Also the founding case of therapy-triggered oncoprotein DEGRADATION [E5], for which the vocabulary has no term.
druggability classclinically-validated
druggability noteThe only fusion in this beat with TWO approved, mechanism-matched agents acting on TWO different retained modules of the same protein: ATRA on the retained RARA ligand-binding pocket [E2][E5], and arsenic trioxide on the retained PML RBCC cysteines [E3]. Together they give a chemotherapy-free cure [E4]. Resistance maps module-wise: RARA LBD mutations cause ATRA resistance, PML B2 mutations (A216V) cause arsenic resistance. cAMP agonism is a third mechanism-derived, clinically untested idea [E5].
stage of hematopoiesisCommitted myeloid precursor arrested at the PROMYELOCYTE stage - late in granulocytic differentiation, which is why the blasts are hypergranular and why differentiation therapy can push them all the way to functional granulocytes [E2]. The leukaemia-initiating cell is a distinct, less differentiated compartment that responds to fusion DEGRADATION rather than to differentiation [E5]. One of the latest cells of origin in this beat; notably rare in infants, the opposite of CBFA2T3::GLIS2 and KMT2A-rearranged infant leukaemia
age peakoccurs across the age range with a young-adult to middle-age peak; roughly 5-10% of pediatric AML with the proportion rising through adolescence, and rare in infants
developmental order rank8
frequency
pediatric_aml_pct8.000pediatric_aml_pct: 8adult_aml_pct10.0adult_aml_pct: 10apl_pct98.0apl_pct: 98
diseases
[
  "acute promyelocytic leukemia (APL, FAB M3 and M3v microgranular variant)",
  "pediatric APL",
  "adult APL"
]
frequency — the numbers behind the chart
{
  "apl_pct": 98,
  "pediatric_aml_pct": 8,
  "adult_aml_pct": 10
}
breakpoints
[
  {
    "id": "bp-pml-intron6-rara-intron2-bcr1",
    "gene5_exon_last": 6,
    "gene3_exon_first": 3,
    "frequency_pct": 55,
    "population": "APL (bcr1 / long-form isoform)",
    "domains_retained_5": [
      "RING finger",
      "B-box 1",
      "B-box 2",
      "alpha-helical coiled-coil (homo-multimerisation)",
      "cysteine-rich zinc-coordinating residues that constitute the arsenic binding site",
      "SUMOylation and SUMO-interaction sites",
      "nuclear-body targeting"
    ],
    "domains_lost_5": [
      "PML C-terminal region including isoform-specific tails"
    ],
    "domains_retained_3": [
      "RARA C4 zinc-finger DNA-binding domain (retinoic-acid response elements, RXR heterodimerisation)",
      "RARA hormone/ligand-binding domain (the ATRA pocket)",
      "AF-2 coactivator-binding surface",
      "corepressor-binding surface"
    ],
    "domains_lost_3": [
      "RARA A domain / N-terminal AF-1 region"
    ],
    "phenotype_notes": "The long (bcr1) isoform, breaking in PML intron 6. Retains BOTH the RARA DNA-binding and hormone-binding domains [E1] and the full PML RBCC, i.e. every module required for both ATRA and arsenic to work. Frequency 55% is an approximation of the commonly reported bcr1 predominance and was not verified in a cited cohort."
  },
  {
    "id": "bp-pml-intron3-rara-intron2-bcr3",
    "gene5_exon_last": 3,
    "gene3_exon_first": 3,
    "frequency_pct": 40,
    "population": "APL (bcr3 / short-form isoform)",
    "domains_retained_5": [
      "RING finger",
      "B-box 1",
      "B-box 2",
      "coiled-coil (homo-multimerisation)",
      "arsenic-binding cysteines within the RBCC"
    ],
    "domains_lost_5": [
      "PML exons 4-6 encoded region",
      "PML C-terminus"
    ],
    "domains_retained_3": [
      "RARA DNA-binding domain",
      "RARA hormone/ligand-binding domain",
      "AF-2"
    ],
    "domains_lost_3": [
      "RARA A domain / AF-1"
    ],
    "phenotype_notes": "The short (bcr3) isoform, breaking in PML intron 3, associated with higher presenting white count, the microgranular M3v morphology and FLT3-ITD. Critically, the RBCC and the RARA ligand pocket are retained in bcr3 as well, which is why both drugs work across isoforms - a useful contrast with KMT2A, where breakpoint position is argued to change biology. Frequency 40% is an approximation and was not verified in a cited cohort.",
    "frequency_pct_note": "bcr1 ~55%, bcr3 ~40%, bcr2 (variable intra-exon-6 breaks) ~5% are the commonly cited proportions; we did not verify them against a specific cohort and flag all three as approximate."
  }
]
PDB IDS
[
  "3KMR",
  "1DKF",
  "1BOR",
  "2MVW",
  "6IMQ",
  "4WJN"
]
mechanism class
[
  "tf-repression-corruption"
]
How this could bite us back
  • Differentiation syndrome (retinoic acid syndrome) is the direct consequence of the therapy working: differentiating promyelocytes cause fever, hypoxia, pulmonary infiltrates, effusions and renal failure, and it is potentially fatal. The better the differentiation, the greater the risk - an unavoidable coupling of efficacy and toxicity.
  • The coagulopathy is worst at presentation and can be aggravated by cytotoxic lysis. ATRA's original series specifically noted no coagulation abnormalities during therapy [E2], which is a therapeutic advantage, but the window before ATRA takes effect remains the highest-mortality period in the whole disease and no targeted agent shortens it.
  • Arsenic is arsenic. It causes QT prolongation with torsades risk, hepatotoxicity (the pivotal trial found MORE hepatic toxicity in the ATRA-arsenic arm [E4]), peripheral neuropathy, and carries long-term second-malignancy concerns that matter especially in a cured paediatric population with decades of life ahead.
  • On-target resistance is module-specific and predictable: RARA ligand-binding-domain mutations abolish ATRA response, and PML B2 mutations (A216V) abolish arsenic-induced degradation. Dual therapy mitigates this, which is precisely why neither agent should be used alone [E2].
  • In children, ATRA carries a distinct pseudotumour cerebri risk with no adult analogue, and ATRA is teratogenic - a serious consideration in adolescent and young-adult patients of childbearing potential.
  • The generalisation risk for this whole beat: PML::RARA's curability came from an unusual coincidence - a natural hormone pocket retained AND a degradable oligomerisation scaffold retained AND an existing approved ligand for each. Treating it as a template for every fusion is the error this node is written to prevent. The transferable lesson is the strategy (degrade, do not inhibit; target the LIC, not the bulk blast), not the expectation of a similarly lucky structure.
  • Both arms of the APL cure have a published escape route: RARA ligand-binding resistance to ATRA, and PML B2-domain mutation against arsenic. The fusion that the field treats as solved is not resistance-proof; it is resistance-outrun by a two-drug combination.
Still open
  1. The cAMP-dependent phosphorylation site required for ATRA-induced PML-RARA degradation and LIC clearance, and the finding that cAMP agonism ENHANCES LIC loss in mouse APL [E5], have not been translated clinically in eighteen years. Why not, and would a cAMP agonist add anything on top of a regimen already at 97% EFS? It might matter most in high-count APL, which the pivotal trial excluded [E4].
  2. PLZF::RARA differentiates fully under ATRA but neither loses LICs nor remits [E5], while PML::RARA does both. The difference must lie in the 5-prime partner's degradability. What exactly does PML's RBCC provide that PLZF's BTB-POZ does not - and could a degrader be engineered for PLZF::RARA to make the variant APLs curable too? This is the most directly actionable question in the node.
  3. There is no structure of arsenic bound to PML [E6]; the site is defined biochemically as cysteines in the RBCC zinc fingers [E3]. A structure would enable rational design of a non-arsenical PML-directed degrader - removing arsenic's toxicity while keeping its mechanism, and providing a template for degrading other coiled-coil-containing fusions.
  4. PML::RARA is the founding proof that therapy-triggered oncoprotein degradation eradicates leukaemia-initiating cells [E5]. Can that principle be transferred to any of the eight pocket-less fusions in this beat - for instance by finding a ligandable cysteine or a defined oligomerisation interface on ETV6's SAM polymer, ETO's NHR2 tetramer or MYH11's coiled coil - and engineering a degrader rather than an inhibitor?
  5. Residual APL mortality is dominated by early haemorrhagic death before therapy takes effect, which no amount of target biology addresses. What actually reduces it - faster diagnosis, pre-emptive ATRA on suspicion, better transfusion algorithms - and is the annexin-II mechanism shared with TCF3::HLF (see fusion-tcf3-hlf) a usable biomarker for who will bleed?
  6. How often do PML B2-domain mutations arise under front-line ATRA plus arsenic, as opposed to in the relapsed/refractory setting where they were discovered? If they are rare front line, the two-drug combination is doing the work, which is itself the transferable lesson.
Every source behind this entry (8)
  1. E1established confidence 5 of 5: Multiple independent lines of evidence; treated as settled in the field.paper

    The PML-RARA mRNA generated by t(15;17) encodes a predicted 106 kDa chimeric protein containing most of the PML sequences, including a novel zinc finger motif, fused to a large part of RARA INCLUDING BOTH its DNA-binding and hormone-binding domains; the hybrid shows altered transactivating properties versus wild-type RARA, and identical fusion points recur across patients.

    PMID 1652369DOI
    What this evidence cannot show

    1991 cloning study with transient-expression reporter assays; 'altered transactivating properties' was as far as the functional characterisation went, and the constitutive-corepressor mechanism was established later. The 106 kDa size is a prediction from sequence. Breakpoint heterogeneity (bcr1/bcr2/bcr3 isoforms) was recognised subsequently, so 'identical fusion points' holds within the cases examined rather than universally.

    Evidence that argues against this (1)
    • t(15;17) makes two fusion genes, and the reciprocal one is not inert. In transgenic mice, a bcr-3 isoform of RARalpha-PML expressed alone in early myeloid cells altered nothing and caused no leukaemia, but co-expressed with a bcr-1 PML-RARalpha it raised APL penetrance from 15% to 57% (p<0.001) and produced less mature leukaemias - without shortening latency, so it is not behaving as a classical second hit either. Describing the lesion as 'the PML-RARA mRNA' therefore omits a co-expressed product, present in a large fraction of patients, that measurably changes disease frequency and phenotype. The same transgenic work also shows why 'identical fusion points recur across patients' is a within-class statement: bcr-1 and bcr-3 are different fusion points giving different-sized proteins with different biology. Pollock JL et al. 1999
  2. E2established confidence 5 of 5: Multiple independent lines of evidence; treated as settled in the field.paper

    All-trans retinoic acid induced complete remission in all 24 APL patients treated, including eight previously nonresponsive or chemotherapy-resistant, without bone-marrow hypoplasia, with morphological and functional differentiation to granulocytes, no coagulation abnormalities during therapy and only mild side effects; eight of 24 relapsed after 2-5 months of remission.

    Use of all-trans retinoic acid in the treatment of acute promyelocytic leukemia · Huang ME, Ye YC, Chen SR, Chai JR, Lu JX, Zhoa L, Gu LJ, Wang ZY · Blood 1988
    PMID 3165295
    What this evidence cannot show

    A 24-patient single-institution uncontrolled series from 1988, before the molecular lesion was known; no randomisation, short follow-up, and one-third relapsed within 5 months, which is the key limitation - ATRA monotherapy induces remission but does not cure. No DOI is registered for this 1988 Blood article, so the PubMed record is the citable anchor. Also note that differentiation syndrome, later recognised as a major ATRA toxicity, was not reported here.

    Evidence that argues against this (1)
    • 'Only mild side effects' did not survive a larger consecutive series. Among 35 patients with APL given all-trans retinoic acid at 45 mg/m2/day at Memorial Sloan-Kettering, 9 (26%, 95% CI 9-52%) developed the retinoic acid syndrome - fever and respiratory distress with weight gain, lower-extremity oedema, pleural or pericardial effusions and episodic hypotension, starting 2 to 21 days after treatment began. Three died; post-mortem examination in two showed pulmonary interstitial infiltration by maturing myeloid cells. In three of the nine the syndrome was not preceded by leukocytosis, so a white-count trigger does not protect. Leukapheresis, stopping ATRA and moderate-dose chemotherapy were all useless once respiratory distress was established. The 24-patient Shanghai experience predates recognition of the syndrome that ATRA's own mechanism causes. Frankel SR et al. 1992
  3. E3established confidence 5 of 5: Multiple independent lines of evidence; treated as settled in the field.paper

    Arsenic binds directly to cysteine residues in zinc fingers located within the RBCC domain of PML-RARalpha and PML; arsenic binding induces PML oligomerisation, which increases its interaction with the SUMO-conjugating enzyme UBC9, resulting in enhanced SUMOylation and degradation - identifying PML as the direct molecular target of arsenic trioxide and explaining the drug's specificity for APL.

    Arsenic trioxide controls the fate of the PML-RARalpha oncoprotein by directly binding PML · Zhang XW, Yan XJ, Zhou ZR, Yang FF, Wu ZY, Sun HB, Liang WX, Song AX, Lallemand-Breitenbach V, Jeanne M, Zhang QY, Yang HY, Huang QH, Zhou GB, Tong JH, Zhang Y, Wu JH, Hu HY, de The H, Chen SJ, Chen Z · Science 2010
    PMID 20378816DOI
    What this evidence cannot show

    Binding was demonstrated largely on recombinant PML fragments using mass spectrometry and biophysical methods plus cell-line experiments; no crystal structure of arsenic bound to PML is reported here. Arsenic is a promiscuous thiol-reactive agent with many cellular targets, so 'direct target' establishes a specific high-affinity interaction rather than exclusivity, and the clinical specificity for APL is an inference consistent with - but not proven by - the binding data.

    Evidence that argues against this (1)
    • Arsenic's action at the PML RBCC is escapable in patients. Two of fifteen relapsed/refractory APL patients treated with arsenic were clinically resistant and both carried PML B2-domain missense mutations; the mutant PML-RARA no longer SUMOylated, no longer oligomerised on arsenic and no longer partitioned into the insoluble fraction, while remaining an intact oncoprotein. The drug's target site is a mutable drug-binding surface like any other. Goto E et al. 2011
  4. E4established confidence 5 of 5: Multiple independent lines of evidence; treated as settled in the field.paper

    In a randomised phase 3 trial in low-to-intermediate-risk APL (WBC <=10 x 10^9/L), ATRA plus arsenic trioxide achieved complete remission in 77/77 evaluable patients (100%) versus 75/79 (95%) with ATRA plus chemotherapy, with 2-year event-free survival 97% versus 86% (p<0.001 non-inferiority, p=0.02 superiority), better overall survival, less haematologic toxicity and fewer infections but more hepatic toxicity - a chemotherapy-free cure.

    Retinoic acid and arsenic trioxide for acute promyelocytic leukemia · Lo-Coco F, Avvisati G, Vignetti M, Thiede C, Orlando SM, Iacobelli S, Ferrara F, Fazi P, Cicconi L, Di Bona E, Specchia G, Sica S, Divona M, Levis A, Fiedler W, Cerqui E, Breccia M, Fioritoni G, Salih HR, Cazzola M, Melillo L, Carella AM, Brandts CH, Morra E, von Lilienfeld-Toal M, Hertenstein B, Wattad M, Lubbert M, Hanel M, Schmitz N, Link H, Kropp MG, Rambaldi A, La Nasa G, Luppi M, Ciceri F, Finizio O, Venditti A, Fabbiano F, Dohner K, Sauer M, Ganser A, Amadori S, Mandelli F, Dohner H, Ehninger G, Schlenk RF, Platzbecker U · N Engl J Med 2013
    PMID 23841729DOINCT00482833
    What this evidence cannot show

    Restricted to LOW-TO-INTERMEDIATE-RISK APL (presenting white count <=10 x 10^9/L); high-count APL was excluded and does not have the same evidence base. Median follow-up 34.4 months, so 2-year EFS is the mature endpoint reported. Designed as a non-inferiority trial, with superiority as a secondary finding. Adults; paediatric practice extrapolates from this. More hepatic toxicity with arsenic, and the trial does not capture early haemorrhagic deaths occurring before enrolment - the residual mortality in APL is largely pre-treatment.

    Evidence that argues against this (1)
    • The APL0406 cure rate is a statement about who reaches randomisation. In the unselected Swedish Adult Acute Leukemia Registry, 30 of 105 APL patients (29%) died within 30 days of diagnosis at a median of 4 days, 41% of those deaths were haemorrhagic, and 35% of early-death patients never received ATRA at all. Lehmann S et al. 2011
  5. E5established confidence 5 of 5: Multiple independent lines of evidence; treated as settled in the field.paper

    Retinoic acid triggers growth arrest and in vivo clearance of leukaemia-initiating cells in PML-RARA mouse APL, whereas in PLZF-RARA APL it triggers full differentiation without LIC loss or remission - uncoupling differentiation from cure; retinoic acid and arsenic synergise to clear LICs through cooperative PML-RARA degradation WITHOUT enhancing differentiation; a cAMP-dependent phosphorylation site in PML-RARA is required for retinoic-acid-induced degradation and LIC clearance; therapy-triggered oncoprotein degradation is proposed as a general strategy to eradicate cancer stem cells.

    Eradication of acute promyelocytic leukemia-initiating cells through PML-RARA degradation · Nasr R, Guillemin MC, Ferhi O, Soilihi H, Peres L, Berthier C, Rousselot P, Robledo-Sarmiento M, Lallemand-Breitenbach V, Gourmel B, Vitoux D, Pandolfi PP, Rochette-Egly C, Zhu J, de The H · Nat Med 2008
    PMID 19029980DOI
    What this evidence cannot show

    Transgenic mouse APL models and ex vivo LIC assays, not human patients; LIC quantification depends on serial transplantation readouts with their own limitations. The PLZF-RARA comparison is a different fusion in a different mouse model, so the differentiation/LIC dissociation, while compelling, rests on cross-model comparison. The cAMP finding has not been translated clinically in the 18 years since.

    Evidence that argues against this (1)
    • The generalisation in the last clause is the weak step, and the founding fusion of the targeted-therapy era is the counter-example. In newly diagnosed CML, imatinib inhibited BCR-ABL activity to the same degree in CD34+CD38- and CD133+ stem cells as in CD34+CD38+ progenitors, and in quiescent as in cycling cells - yet with cytokine support those primitive cells grew and survived in the absence of BCR-ABL activity as well as their normal counterparts did. Human CML stem cells are not oncogene-addicted. Removing an oncoprotein's function, and by extension removing the oncoprotein, therefore does not by itself clear the leukaemia-initiating compartment. What the APL work establishes is that degradation clears LICs in APL; 'therapy-triggered oncoprotein degradation as a general strategy to eradicate cancer stem cells' is a cross-family extrapolation that the CML data already contradict. Corbin AS et al. 2011
  6. E6established confidence 5 of 5: Multiple independent lines of evidence; treated as settled in the field.database

    Both drug-relevant retained modules are structurally characterised: the RARA ligand-binding domain with an agonist and coactivator fragment (PDB 3KMR) and in the RAR-RXR LBD heterodimer (1DKF), and the PML RBCC elements including the zinc-loaded RING (1BOR), the B-box-1 (2MVW), PML B1-box MULTIMERS (6IMQ) and SUMO1 bound to phosphorylated PML (4WJN) - but no structure of the PML::RARA fusion itself exists.

    What this evidence cannot show

    Verified by fetching each RCSB entry record and by searching RCSB on UniProt P29590 (20 PML entries) and P10276 (RARA). Every entry is an isolated domain of the WILD-TYPE protein; there is no structure of the fusion, no structure of ATRA bound to the fusion's LBD, and - importantly - no structure of arsenic bound to PML among these entries, so the arsenic binding site is defined biochemically [E3] rather than crystallographically. 3KMR contains the synthetic agonist Am580 rather than ATRA itself.

    Evidence that argues against this (1)
    • The structural inventory here should also record that the PML oligomerisation module is larger and more distributed than the deposited fragments suggest: cryo-EM and crosslinking of the PML RBCC dimer show a coiled-coil-mediated assembly in which each CC helix of one dimer engages a CC helix of a neighbouring dimer, so the self-association surface is not a single compact site. Tan Y et al. 2024
  7. E7strong confidence 4 of 5: Replicated in peer-reviewed work by more than one group.paper

    PML B2-domain missense mutations in PML-RARA confer clinical arsenic resistance, abolishing arsenic-induced SUMOylation, oligomerisation and insolubility while the oncoprotein persists.

    Missense mutations in PML-RARA are critical for the lack of responsiveness to arsenic trioxide treatment. · Goto E, Tomita A, Hayakawa F, Atsumi A, Kiyoi H, Naoe T · Blood 2011
    PMID 21613260DOI
    What this evidence cannot show

    Abstract only. 2 of 15 relapsed/refractory APL patients, mechanism reconstructed in HeLa overexpression. Small n; PML A216V is now trackable by ddPCR before overt resistance (PMID 30651632).

    Evidence that argues against this (1)
    • Arsenic resistance is not confined to the fusion allele, so sequencing the fusion transcript alone can return a falsely reassuring wild-type result. Yu et al. describe a relapsed APL patient resistant to both arsenic and chemotherapy who carried an A216V point mutation in the UNREARRANGED PML allele rather than in PML::RARA. The mutant wild-type PML impeded arsenic-induced destabilisation and degradation of an otherwise wild-type PML::RARA protein, and deleting the coiled-coil domain of the PML mutant abolished the PML/PML-RARA interaction and completely reversed the resistance. The mechanism is therefore trans-acting and runs through oligomerisation between the fusion and the normal allele - which also means the resistance is a property of the oligomer, not of the B2 domain of the fusion as such. Yu PH et al. 2025
  8. E8established confidence 5 of 5: Multiple independent lines of evidence; treated as settled in the field.paper

    In an unselected population-based cohort, 29% of APL patients died within 30 days of diagnosis, 41% of those deaths haemorrhagic, and a third of early-death patients never received ATRA.

    PMID 21502956DOI
    What this evidence cannot show

    Abstract only. Swedish adults 1997-2006, n=105; predates ATRA-on-suspicion policy and front-line ATO, so modern rates are lower but not near zero.

    Evidence that argues against this (1)
    • Two unselected populations give substantially different numbers, and the size of the gap is itself the finding. In SEER, 1400 patients diagnosed with APL between 1992 and 2007 had an overall early-death rate of 17.3% - not 29% - rising to 24.2% at age 55 or older, with only a modest change across the 16-year period even as 3-year survival rose from 54.6% to 70.1%. Registry early-death estimates depend heavily on ascertainment: deaths occurring before a diagnosis is confirmed are captured differently in a national haematology registry than in a cancer registry keyed to a coded diagnosis. 17% and 29% therefore bound the same quantity rather than refute each other, and the durable claim is the trial-versus-population gap, not either percentage. Park JH et al. 2011
Connected to