TCF3::HLF (E2A-HLF), t(17;19)(q22;p13)

This is real experimental data — atoms placed by crystallography or cryo-EM, deposited by the people who did the experiment.

• Venetoclax resistance in BCL2-dependent leukaemias is well documented and arises by BCL2 mutation (for instance G101V), by switching dependency to MCL1 or BCL-xL, and by MCL1 amplification. A subtype whose entire targeted rationale is a single BH3 node is maximally exposed to that switch, and combination with an MCL1 inhibitor brings cardiac toxicity.
• The anti-apoptotic mechanism cuts both ways for therapy. SLUG protects NORMAL haematopoietic progenitors from radiation-induced apoptosis in vivo [E3], so any attempt to inhibit SLUG rather than BCL2 would be expected to sensitise the patient's own marrow to the cytotoxic therapy given alongside it - an on-target, essential-function toxicity.
• Venetoclax plus intensive chemotherapy in children produces prolonged, profound myelosuppression and febrile neutropenia; in a subtype that also presents with coagulopathy and hypercalcaemia [E4], the supportive-care burden compounds.
• The evidence base is a handful of patients. Because the subtype is well under 1% of paediatric B-ALL [E5], no randomised trial will ever be powered in it, so the therapeutic standard will rest indefinitely on xenograft pharmacology plus case series - which is precisely the evidence regime in which apparent breakthroughs most often fail to replicate.
• Somatic mutation of the non-translocated TCF3 allele and PAX5 deletions are part of the lesion set [E5]. That means the leukaemia has already compromised the B-lineage transcription factor network, which is the same network that blinatumomab and CAR-T target through CD19; lineage switch and antigen loss are plausible escape routes, as they are in KMT2A-rearranged disease.
• TCF3::HLF has the structured dimeric interface this beat prizes, no structure of it, no molecule against it, direct mutagenesis showing that disabling that interface does not abolish the oncogenic output [E7], and a working drug lead aimed somewhere else entirely. That combination is the honest base rate for interface-first drug discovery in rare fusions.

1. Is the venetoclax sensitivity a consequence of the SLUG/anti-apoptotic axis, or an independent BCL2 dependency? If the fusion raises the apoptotic threshold via SLUG and BCL2 sits in the same bottleneck, then SLUG expression should predict venetoclax response across TCF3::HLF samples - directly testable and immediately clinically relevant.
2. TCF3::PBX1 and TCF3::HLF share an identical 5-prime moiety and differ only in the imported DNA-binding domain, yet differ in cell of origin, aggressiveness and drug sensitivity. What exactly does the bZIP take the fusion to that the homeodomain does not? Matched ChIP/CUT&RUN in isogenic human B progenitors expressing one or the other fusion would be one of the highest-yield experiments in this whole beat.
3. Why does importing a liver/kidney bZIP specificity into a pro-B cell produce a stem-cell/myeloid-skewed state [E5]? Is the fusion actively dedifferentiating the cell, or selecting a rare pre-existing hybrid progenitor?
4. No experimental structure of the HLF bZIP exists [E6]. Would a structure of the fusion's bZIP on its DNA site reveal a junction-proximal or dimer-interface surface not present in wild-type HLF, and is that surface accessible to a covalent or stapled-peptide approach?
5. Is the ANXA2 induction [E4] sufficient to explain the bleeding diathesis, and does it predict which patients will haemorrhage - i.e. is annexin II a usable biomarker for supportive-care intensification at presentation?

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

   t(17;19) produces a chimeric protein consisting of the amino-terminal transactivation domain of E2A linked to the carboxy-terminal basic region-leucine zipper of HLF, a PAR-family bZIP factor normally expressed in liver and kidney but not in lymphoid cells and related to DBP and TEF.

   Fusion of the leucine zipper gene HLF to the E2A gene in human acute B-lineage leukemia. · Inaba T, Roberts WM, Shapiro LH, Jolly KW, Raimondi SC, Smith SD, Look AT · Science 1992
   PMID 1386162DOI
   ▸What this evidence cannot show

   Cloned from cell lines and a small number of primary cases; HLF tissue distribution was assessed by Northern blot across a limited panel, so 'not expressed in lymphoid cells' means below detection in the tissues tested rather than proven absence.

   ▸Evidence that argues against this (2)

   • TCF3::HLF is the beat's most informative negative control and the entry does not say so. The fusion retains a bZIP LEUCINE ZIPPER -- a compact, structured, obligate dimerisation interface of exactly the kind `hyp-obligate-partner-interface-generalises` predicts will be productive -- and in the 34 years since the fusion was cloned no molecule against that interface has been reported, while the disease remained uniformly fatal. Inaba T et al. 1992
   • The domain vocabulary this entry uses for the TCF3 side — AD2, 'N-terminal transactivation region', 'N-terminal two-thirds' — is literature language with no counterpart in UniProt P15923, which annotates neither AD1 nor AD2. Any statement here about which transactivation domains are retained rests on the primary literature's own boundaries, and those differ between papers. UniProt Consortium 2026
2. E2strong confidence 4 of 5: Replicated in peer-reviewed work by more than one group.paper

   E2A-HLF transactivates SLUG, a Snail-family zinc-finger repressor homologous to C. elegans CES-1 (which acts downstream of the bZIP factor CES-2 in a cell-death pathway), and SLUG is nearly as active as BCL2 or BCL-xL in promoting survival of cytokine-deprived pro-B cells.

   SLUG, a ces-1-related zinc finger transcription factor gene with antiapoptotic activity, is a downstream target of the E2A-HLF oncoprotein · Inukai T, Inoue A, Kurosawa H, Goi K, Shinjyo T, Ozawa K, Mao M, Inaba T, Look AT · Mol Cell 1999
   PMID 10518215DOI
   ▸What this evidence cannot show

   Murine IL-3-dependent pro-B cell lines with enforced expression; the CES-2/CES-1 analogy is an evolutionary-homology argument, and the survival comparison against BCL2/BCL-xL is in a cytokine-withdrawal assay that may not model chemotherapy-induced apoptosis in human blasts. SLUG has since been shown to have wide-ranging EMT and stemness roles, so its anti-apoptotic function may not be the only relevant output.

   ▸Evidence that argues against this (1)

   • A major E2A-HLF output is produced WITHOUT the fusion binding DNA, which undercuts the transactivation model the SLUG result instantiates and supplies a competing explanation for the subtype's chemoresistance. Baudis et al. used representational difference analysis to find ABCB1 (MDR1, P-glycoprotein) differentially expressed between t(17;19)-positive and -negative ALL lines, confirmed high ABCB1 protein and a drug-efflux-positive phenotype in the t(17;19) lines, and then showed that E2A-HLF did NOT activate reporter genes under ABCB1 promoter elements in transient transfection, yet induced ABCB1 in a DNA-binding-independent manner - as did E2A-PBX1 and truncated E2A polypeptides on their own. If truncated E2A alone reproduces the induction, the HLF-directed 'transactivates SLUG' chain is not the only, and may not be the dominant, route from this fusion to a drug-resistant phenotype. Baudis M et al. 2006
3. E3strong confidence 4 of 5: Replicated in peer-reviewed work by more than one group.paper

   SLUG is a genuine anti-apoptotic factor in normal haematopoietic progenitors, not merely an overexpression artefact: Slug-null mice are markedly radiosensitive with increased TUNEL-positive marrow progenitors, microhaemorrhages and disseminated microabscesses.

   Slug, a highly conserved zinc finger transcriptional repressor, protects hematopoietic progenitor cells from radiation-induced apoptosis in vivo · Inoue A, Seidel MG, Wu W, Kamizono S, Ferrando AA, Bronson RT, Iwasaki H, Akashi K, Morimoto A, Hitzler JK, Pestina TI, Jackson CW, Tanaka R, Chong MJ, McKinnon PJ, Inukai T, Grosveld GC, Look AT · Cancer Cell 2002
   PMID 12398892DOI
   ▸What this evidence cannot show

   A mouse germline knockout assessed after whole-body irradiation; it establishes that Slug is anti-apoptotic in normal progenitors but does not show that SLUG induction is necessary for TCF3::HLF leukaemogenesis in human cells. Also a direct bite-back signal: removing SLUG function would be expected to sensitise normal marrow progenitors too.

   ▸Evidence that argues against this (1)

   • The human loss-of-function phenotype is not haematopoietic. Sanchez-Martin et al. identified two unrelated patients with Waardenburg syndrome type 2 carrying HOMOZYGOUS SLUG (SNAI2) deletions resulting in complete absence of the SLUG product; the phenotype reported is auditory-pigmentary - deafness with pigmentary abnormalities, a neural-crest defect - with no marrow failure, cytopenia or radiosensitivity described, and the paper frames SLUG's essential role as being in neural-crest-derived lineages. Mouse Slug-null radiosensitivity after whole-body irradiation therefore does not establish that human haematopoietic progenitors require SLUG to survive. That cuts both ways: it weakens the claim that SLUG is a validated anti-apoptotic factor in normal human progenitors, and it weakens the on-target-toxicity argument the entry uses against ever targeting SLUG. Sanchez-Martin M et al. 2002
4. E4moderate confidence 3 of 5: One strong study, or several smaller consistent ones.paper

   E2A-HLF expression is associated with hypercalcaemia and haemorrhagic complications, and the fusion induces annexin II (ANXA2) in cytokine-independent fashion, apparently substituting for RAS/PI3K signalling downstream of IL-3; annexin II is also a proposed cause of the haemorrhagic complications of acute promyelocytic leukaemia.

   Regulation of annexin II by cytokine-initiated signaling pathways and E2A-HLF oncoprotein · Matsunaga T, Inaba T, Matsui H, Okuya M, Miyajima A, Inukai T, Funabiki T, Endo M, Look AT, Kurosawa H · Blood 2003
   PMID 15070701DOI
   ▸What this evidence cannot show

   Four t(17;19)-positive cell lines plus enforced-expression experiments; annexin II induction is correlative with respect to the clinical bleeding, and the mechanistic link from ANXA2 to coagulopathy is imported from the APL literature rather than demonstrated in t(17;19) patients. The hypercalcaemia association is stated as background in the introduction and is not measured in this study.

   ▸Evidence that argues against this (1)

   • The hypercalcaemia has a measured mechanism and it is not annexin II. Inukai et al. retrospectively analysed 22 children with ALL and hypercalcaemia above 12 mg/dl reported in Japan between 1990 and 2005 and confirmed PTHrP (parathyroid hormone-related peptide)-mediated hypercalcaemia in 11 of the 16 patients in whom serum PTHrP or PTHrP immunohistochemistry was available; t(17;19) or add(19)(p13) was present in 5 of the 17 patients with karyotype data and E2A-HLF was confirmed in all five, every one of whom relapsed very early, while hypercalcaemia itself carried no adverse prognosis once t(17;19) cases were excluded. Placing hypercalcaemia and ANXA2 induction in one sentence implies a shared mechanism that the clinical series does not support - the calcium phenotype runs through PTHrP. Inukai T et al. 2007
5. E5strong confidence 4 of 5: Replicated in peer-reviewed work by more than one group.paper

   TCF3-HLF-positive ALL is currently incurable and genomically distinct from treatment-responsive TCF3-PBX1 ALL, carrying recurrent intragenic PAX5 or VPREB1 deletions, somatic mutation of the non-translocated TCF3 allele, reduced PAX5 dosage and a stem-cell/myeloid-skewed signature; matched patient-derived xenografts are resistant to conventional chemotherapeutics but show striking on-target sensitivity to the BCL2 inhibitor venetoclax.

   Genomics and drug profiling of fatal TCF3-HLF-positive acute lymphoblastic leukemia identifies recurrent mutation patterns and therapeutic options. · Fischer U, Forster M, Rinaldi A, Risch T, Sungalee S, Warnatz HJ, Bornhauser B, Gombert M, Kratsch C, Stutz AM, Sultan M, Tchinda J, Worth CL, Amstislavskiy V, et al., Borkhardt A, Bourquin JP, Franke A, Korbel JO, Stanulla M, Yaspo ML · Nat Genet 2015
   PMID 26214592DOI
   ▸What this evidence cannot show

   The cohort is necessarily tiny because the subtype is so rare, so the 'recurrent' cooperating lesions rest on a handful of cases and the mutation-pattern comparison against TCF3-PBX1 is underpowered for anything but the strongest differences. The venetoclax result is an ex vivo and xenograft drug-response finding, not a clinical outcome; no randomised or even single-arm clinical demonstration of venetoclax benefit in TCF3::HLF patients is cited here.

   ▸Evidence that argues against this (1)

   • The one therapeutic lead that did emerge is not an interface drug. Venetoclax targets BCL2, a downstream effector, and the result is ex vivo and patient-derived-xenograft drug profiling rather than a clinical trial. The lesson TCF3::HLF teaches this beat is that when interface-first fails, effector-first can still work -- which argues for hedging the menin-interface strategy with downstream-effector strategies rather than deepening it. Fischer U et al. 2015
6. E6strong confidence 4 of 5: Replicated in peer-reviewed work by more than one group.database

   No experimental structure of the HLF bZIP domain retained in the fusion exists; the only structures relevant to this fusion cover the TCF3 modules, including the p300 TAZ2 complex with the retained E2A activation domain and the bHLH domain the fusion deletes.

   RCSB PDB search by UniProt accession: Q16534 (HLF) returns zero entries; P15923 (TCF3) returns 2MH0, 2YPA, 2YPB, 3U5V, 6MGN · RCSB Protein Data Bank · RCSB PDB 2026
   ▸What this evidence cannot show

   Verified by querying the RCSB search API on the reference-sequence accession field. Absence of a PDB entry is an absence of a deposited experimental structure only; bZIP domains are structurally well understood by homology and AlphaFold models exist, so 'no structure' means no experimental coordinates for this specific protein, not structural ignorance.

   ▸Evidence that argues against this (1)

   • The coordinates that are missing are for the module that was tested and found dispensable; the modules the functional data implicate are the ones already solved. Inukai et al. expressed a panel of E2A-HLF mutants from a zinc-inducible vector in IL-3-dependent murine pro-B cells and found that different combinations of disabling mutations within the HLF bZIP domain had little effect on the chimera's ability to prolong survival after IL-3 withdrawal, provided the amino-terminal portion of E2A was intact; in a bZIP background defective in DNA binding, mutants retaining EITHER AD1 or AD2 still extended survival, while E12, E47 and wild-type HLF each protected not at all. 'No experimental structure of the HLF bZIP exists' is therefore a real gap but probably not the limiting one - the surface with an experimental structure (E2A AD1 bound to the p300 TAZ2 domain, 2MH0) is the surface the mutagenesis implicates. Inukai T et al. 1998
7. E7strong confidence 4 of 5: Replicated in peer-reviewed work by more than one group.paper

   Drug-response profiling of matched patient-derived xenografts showed TCF3-HLF ALL is resistant to conventional chemotherapeutics but sensitive to glucocorticoids, anthracyclines and agents in clinical development, with striking on-target sensitivity to the BCL2-specific inhibitor venetoclax (ABT-199). The same study found recurrent intragenic deletions of PAX5 or VPREB1 in constellation with TCF3-HLF, somatic mutations in the non-translocated TCF3 allele, and a stem-cell and myeloid-feature enrichment in the TCF3-HLF signature.

   Genomics and drug profiling of fatal TCF3-HLF-positive acute lymphoblastic leukemia identifies recurrent mutation patterns and therapeutic options. · Fischer U, Forster M, Rinaldi A, et al. · Nature Genetics 2015
   PMID 26214592DOI
   ▸What this evidence cannot show

   Read in full from PMC4603357 on 2026-09-19. Previously carried at moderate confidence from the abstract, bundled with an undruggability claim this paper does not make — see E10, where that half now lives. All drug data are ex vivo on xenograft-derived cells and in PDX models; there is no randomised clinical evidence in TCF3-HLF, and the cohort is small because the disease is rare. Surfaced by `scripts/reading-queue.py`.

   ▸Evidence that argues against this (2)

   • Direct mutagenesis says the retained zipper is not a functional requirement of this fusion, which is the thing a druggable obligate interface has to be. Inukai et al. report that 'different combinations of disabling mutations within the HLF bZIP domain had little effect on the antiapoptotic property of the chimeric protein, so long as the amino-terminal portion of E2A remained intact', and conclude that 'neither DNA binding nor protein dimerization through the bZIP domain of HLF is required for this effect', proposing instead that protein-protein interactions with the E2A amino-terminus let the chimera act as a transcriptional cofactor. A zipper whose dimerisation can be disabled without losing the oncogenic output is an obligate dimer in the biophysical sense and NOT an obligate partner interface in the therapeutic sense. This entry's claim asserted the latter and has been corrected (see the 2026-09-19 correction section). This is the clearest cross-family test of the lab's obligate-partner-interface hypothesis available in this beat, and it goes against it. Inukai T et al. 1998
   • bZIP interfaces as a class have produced chemical matter, so the absence here is specific to HLF rather than intrinsic to the fold. Aikawa et al. designed de novo a small-molecule inhibitor of c-Fos/activator protein-1 using three-dimensional pharmacophore modelling built on the three-dimensional structure of the basic region-leucine zipper domain of the AP-1-DNA complex; administered orally it prevented collagen-induced arthritis when started at day 21 before onset and RESOLVED established arthritis when started at day 27, reducing inflammatory cytokines and matrix metalloproteinases in serum, joints, synovial cells and chondrocytes and synergising with anti-TNF-alpha. The same compound was later shown to block TGF-beta-driven fibroblast activation and to prevent cartilage destruction in vivo. 'No chemical matter in three decades' therefore records the absence of a solved HLF structure and of any programme aimed at a sub-1% paediatric leukaemia, not a demonstration that bZIP interfaces are undruggable. Aikawa Y et al. 2008
8. E10weak confidence 2 of 5: Preliminary, small, indirect, or not yet peer reviewed.paper

   No chemical matter has been reported against the retained HLF bZIP leucine zipper. It is a structured, constitutively dimeric protein-protein interface of the class that has repeatedly resisted small-molecule inhibition, and the actionable lead in this disease targets a downstream effector (BCL2) instead.

   Genomics and drug profiling of fatal TCF3-HLF-positive acute lymphoblastic leukemia identifies recurrent mutation patterns and therapeutic options · Nature Genetics 2015
   PMID 26214592DOI
   ▸What this evidence cannot show

   **This is an argument from silence and is now recorded as its own item because the citation attached to it does not support it.** PMC4603357 was read in full on 2026-09-19 and does not discuss the bZIP or the leucine zipper anywhere: the string 'bZIP' occurs once in the paper, inside a reference title, and 'leucine zipper' not at all. The paper is a genomics and drug-profiling study and is properly cited for the venetoclax result in E7. The absence of anti-bZIP chemistry rests on a literature search, not a systematic screen, and this entry's E6 records that no experimental structure of the HLF bZIP exists, which is part of why no chemistry was attempted. Found by `scripts/reading-queue.py`, which flagged the combined item for low claim-to-abstract overlap — low because half the claim was not in the paper at all.

   ▸Evidence that argues against this (2)

   • An absence of testing is not a negative result, and this project has a standing rule saying so. 'No chemical matter reported' is compatible with 'nobody has tried', which is the likelier explanation for a fusion this rare: TCF3-HLF occurs in well under 1% of paediatric ALL, so the absence may measure market size rather than tractability. Nature Genetics 2015
   • Constitutively dimeric coiled-coil interfaces are not uniformly undruggable, and the generalisation is doing the work here rather than any TCF3-HLF-specific evidence. Stapled peptides and covalent approaches have produced tool compounds against bZIP and coiled-coil targets in other systems, so 'this class resists small molecules' is a prior, not a finding, and should not be cited as a reason to stop looking. Oncofusion knowledge graph (internal tooling) 2026

• ← related toTCF3 (E2A) — domain architecture and residue coordinates

  Same isoform and construct-numbering traps apply to the second TCF3 fusion: both fusions take TCF3 1-483, which lies upstream of the E12/E47 divergence at residue 529, so the TCF3 moiety is isoform-independent -- but any bHLH coordinate quoted for wild-type TCF3 is not.

• ← part ofTCF3 (E2A) — domain architecture and residue coordinates

  same TCF3 5-prime contribution in TCF3::HLF, t(17;19)

• related to →Hypothesis — the menin lesson generalises: every non-enzymatic fusion has an obligate partner interface that is more druggable than the fusion

  The beat's cleanest negative control: TCF3::HLF retains a bZIP leucine-zipper obligate dimerisation interface, has no experimental structure of it and no molecule against it 34 years after cloning, while the one working lead (venetoclax) targets a downstream effector (Fischer 2015, PMID 26214592). Interface-first failed; effector-first worked.

• related to →TCF3 (E2A) — domain architecture and residue coordinates

  Counter-evidence: Baudis 2006 (PMID 16206189) shows E2A-HLF induces ABCB1 DNA-binding-independently, as do E2A-PBX1 and truncated E2A alone, undercutting the bZIP-driven transactivation model

• related to →Of the menin co-dependencies, which are actually safe to hit? (run)

  Counter-evidence: Sanchez-Martin 2002 (PMID 12444107) shows humans with homozygous SNAI2 deletion have an auditory-pigmentary phenotype, not marrow failure, bounding the mouse radiosensitivity result

• related to →The parathyroid does not need the rest of the body: deleting menin only there causes neoplasia and hypercalcaemia

  Counter-evidence: Inukai 2007 (PMID 17183364) attributes hypercalcaemia in t(17;19) ALL to PTHrP, confirmed in 11 of 16 assessed patients, not to annexin II

• related to →HLF (hepatic leukemia factor) domain architecture

  Counter-evidence: Inukai 1998 (PMID 9742120) shows the unsolved HLF bZIP is dispensable for the fusion's anti-apoptotic output, so the missing structure is not the limiting gap

• related to →Hypothesis — the menin lesson generalises: every non-enzymatic fusion has an obligate partner interface that is more druggable than the fusion

  FALSIFYING: Inukai 1998 (PMID 9742120) found that disabling DNA binding and dimerisation in the HLF bZIP did not abolish E2A-HLF's anti-apoptotic activity, so a compact structured partner interface is not automatically an obligate one

• related to →Analysis (planned) — comparative druggability scan of every obligate fusion-partner interface in the atlas

  Counter-evidence: Aikawa 2008 (PMID 18587386) designed an orally active c-Fos/AP-1 inhibitor from the bZIP-DNA structure, so bZIP interfaces are not intrinsically undruggable

• ← analogous toTCF3::PBX1 (E2A-PBX1), t(1;19)(q23;p13.3)

  SAME 5' PARTNER, DIFFERENT 3' HALF, and radically different outcome - t(1;19) is curable where t(17;19) is among the worst-prognosis B-ALL lesions. The pair isolates the 3' contribution while holding the TCF3 activation domains constant.

• depends on →Mechanism class 2 — transcription-factor corruption (RUNX1::RUNX1T1, PML::RARA, ETV6::RUNX1, CBFB::MYH11) — a balance, not a pure dominant negative
• ← part ofHLF (hepatic leukemia factor) domain architecture

  reference domain coordinates for the HLF side of this fusion

• ← related toThe lymphoid-restricted DAXX dependency survives three nulls and two comparator restrictions - and ATRX, MEN1 and DOT1L do not survive the same test in the same lines

  YCUB-2 (TCF3::HLF) is the second non-KMT2A line in the deep DAXX group, at -0.792