TRAF2 - a cytoplasmic NF-kB brake reported as an AFF4 two-hybrid hit, 97 partners in the screen that found it, and the live instance of the brief's own accession trap

This is a prediction, not a measurement. No one has seen this protein in this shape. The colours say how much the software trusts itself, and that is worth reading before anything else — the orange stretches are places the model is effectively saying this part has no fixed shape.

• If the Xiu result transfers, non-canonical NF-kB activation is protective here - so any agent that inhibits NIK or stabilises the TRAF2/TRAF3/cIAP brake would push in the wrong direction, and IAP antagonists (which do the opposite) would be the rational class. The sign has never been established in B-lineage disease and getting it backwards is the concrete hazard on this entry.
• TRAF2 is essential for TNF-induced survival signalling in normal cells; anything that degrades it sensitises normal tissue to TNF-driven apoptosis. That is an essential-function toxicity, not a resistance mechanism, and it is why TRAF2 is a poor target even if the biology pointed the right way.

1. Is there a nuclear pool of TRAF2 in any haematopoietic cell? The whole compartment argument against this interaction rests on an annotation's silence, and one fractionation blot in a KMT2A-rearranged line would either close the question or make the AFF4 row worth re-examining.
2. Does TRAF2 loss protect against KMT2A::AFF1 B-lineage leukaemia the way NIK stabilisation protects against murine MLL-AF9? The prediction has a sign, it is cheap to test in an existing B-lineage model, and nothing in the literature has attempted it.
3. Is the non-canonical NF-kB brake intact in infant KMT2A::AFF1 blasts? TRAF2, TRAF3, BIRC2 and BIRC3 lesions are recurrent in lymphoid malignancy and this graph's cooperating-lesion enumeration does not cover them.

1. E1weak confidence 2 of 5: Preliminary, small, indirect, or not yet peer reviewed.database

   IntAct records AFF4 and TRAF2 as interacting in three rows, all from PMID 25416956 / IMEx IM-23318, in host Saccharomyces cerevisiae, annotated two hybrid prey pooling approach, two hybrid array and validated two hybrid, MI-score 0.56, with no other source publication.

   IntAct molecular interaction database, PSICQUIC MITAB 2.7, query id:Q12933 AND id:Q9UHB7, 3 rows retrieved 2026-09-21 · IntAct (EMBL-EBI) 2026
   ▸What this evidence cannot show

   Three rows, one publication, one screen, one assay class. Not recovered in HuRI (Luck 2020), although IntAct cannot distinguish 'tested and failed' from 'not in the search space'. No region mapped, no affinity, no structure, no mammalian-cell evidence, and the bait is wild-type AFF4 rather than AFF1 or any KMT2A fusion.

   ▸Evidence that argues against this (1)

   • Rolland et al. present their map as at least as accurate as the literature-curated interactome, so a blanket discount of the assay class is not what the source supports. The discount applied here is for promiscuity and compartment, which are properties of this pair rather than of the method. Rolland T et al. 2014
2. E2strong confidence 4 of 5: Replicated in peer-reviewed work by more than one group.analysis

   TRAF2 has 97 distinct partners within PMID 25416956 and 204 within HuRI (PMID 32296183), against AFF4's 5 in the former and 0 rows in the latter.

   The four AFF4 two-hybrid partners are three rows of one screen, not three experiments · Oncofusion knowledge graph, ana-the-aff4-two-hybrid-quartet-is-one-screen-and-three-of-the-four-preys-are-hubs 2026
   ▸What this evidence cannot show

   Degrees computed from IntAct's import of each publication on 2026-09-21, deduplicated to distinct accessions, not from the papers' supplementary tables. No base rate was computed for the screen's degree distribution, so '97 partners' is compared against AFF4's own 5 in the same assay rather than tested against the screen. A high degree flags a pair as uninformative about the bait; it does not prove the pair false.

   ▸Evidence that argues against this (1)

   • TRAF2 is a genuine signalling hub with a large true interactome across TNFR-family receptors, so part of its degree is biology rather than artefact, and a promiscuity argument penalises real hubs exactly as hard as sticky ones. Huttlin EL et al. 2021
3. E3established confidence 5 of 5: Multiple independent lines of evidence; treated as settled in the field.database

   UniProt annotates TRAF2 with exactly one subcellular location, Cytoplasm (ECO:0000269), and describes it as an E3 ubiquitin-protein ligase regulating NF-kB and JNK activation that catalyses Lys-63-linked ubiquitination of targets including BIRC3, IKBKE, MLST8, RIPK1 and TICAM1; AFF4 is annotated Nucleus and Chromosome, both ECO:0000269.

   UniProtKB entries Q12933 (TRAF2_HUMAN) and Q9UHB7 (AFF4_HUMAN), subcellular location and function comments with evidence codes · UniProtKB/Swiss-Prot 2026
   ▸What this evidence cannot show

   Retrieved 2026-09-21 with the evidence codes read, per brief rule 91. ECO:0000269 on 'Cytoplasm' means a curator read an experiment showing cytoplasmic localisation; it does not mean anyone looked for and excluded a nuclear pool, and the cited experiments were not opened. The absence of a nuclear annotation is therefore weaker than a measured absence of nuclear TRAF2, and this entry claims only the weaker statement.

   ▸Evidence that argues against this (1)

   • Signalling adaptors with no curated nuclear annotation do sometimes have functional nuclear pools that the annotation lags; the same criticism this graph makes of UniProt localisation elsewhere - that a quarter of its entries rest on uncoded annotations - cuts against leaning too hard on an annotation's silence. Lin WW et al. 2013
4. E4strong confidence 4 of 5: Replicated in peer-reviewed work by more than one group.database

   No literature connects TRAF2 to KMT2A or MLL: TRAF2[tiab] AND (MLL[tiab] OR KMT2A[tiab]) returns exactly 1 record, PMID 32598477, a systematic review and meta-analysis of genetic mutations in mantle cell lymphoma, which is off-beat.

   PubMed counted-absence search for TRAF2 against MLL and KMT2A, run 2026-09-21 · PubMed (NCBI) 2026
   ▸What this evidence cannot show

   Operand controls: TRAF2[tiab] AND (leukemia[tiab] OR leukaemia[tiab]) returns 37 and TRAF2[tiab] AND NIK[tiab] AND (degradation OR ubiquitin*) returns 53, so the instrument returns ones. Querytranslation expanded as written. Aliases TRAP3 and the complex-level phrase 'TRAF complex' were not searched. Brief rule 64 is a live gap here: TRAF2 belongs to a family routinely framed as 'TRAF' or 'TNFR signalling' in abstracts, and a subunit query cannot reach that framing. Zero in titles and abstracts; supporting figures not searched.

   ▸Evidence that argues against this (1)

   • The NF-kB literature in this disease is repeatedly framed at the pathway level rather than by adaptor symbol - this project has already recorded that a paper knocking down Ikka, Ikkb, Ikkg and IRAK3 is unreachable from any of those four symbols because its abstract says 'IKK/NF-kB'. A TRAF2 symbol query is exposed to exactly that failure. Xiu Y et al. 2018
5. E5moderate confidence 3 of 5: One strong study, or several smaller consistent ones.paper

   In murine MLL-AF9 acute myeloid leukaemia, stabilisation of NF-kB-inducing kinase (NIK) suppresses the disease, activating non-canonical and repressing canonical NF-kB signalling, upregulating Dnmt3a and downregulating Mef2c.

   Stabilization of NF-kappaB-Inducing Kinase Suppresses MLL-AF9-Induced Acute Myeloid Leukemia. · Xiu Y, Dong Q, Li Q, et al., Zhao C · Cell Rep 2018
   PMID 29320732DOI
   ▸What this evidence cannot show

   Murine MLL-AF9 myeloid leukaemia, not human B-lineage KMT2A::AFF1 - the exact boundary this project's commonest documented synthesis failure crosses. Read as abstract, not full text. The paper manipulates NIK, not TRAF2; TRAF2's connection to it is the upstream degradation complex and is inference from pathway architecture, not an experiment in this paper. No TRAF2 genotype is reported in the abstract.

   ▸Evidence that argues against this (1)

   • If non-canonical NF-kB activation suppresses this leukaemia, then TRAF2 loss - which stabilises NIK - would be protective, which is the opposite of the direction a reader primed by TRAF2's tumour-suppressor role in lymphoid malignancy would assume. This graph's own entry mechanism-nfkb-lineage-split-myeloid-story-lymphoid-silence records that the NF-kB evidence here is myeloid and the lymphoid side is unmeasured, so the sign of this effect in B-lineage KMT2A::AFF1 is genuinely unknown. Lin WW et al. 2013
6. E6established confidence 5 of 5: Multiple independent lines of evidence; treated as settled in the field.database

   gene_exact:TRAF2 AND organism_id:9606 AND reviewed:true returns two reviewed human records with the wrong one first - Q92844 (TANK_HUMAN, primary gene TANK, synonyms ITRAF and TRAF2, 425 aa) ahead of Q12933 (TRAF2_HUMAN, primary gene TRAF2, synonym TRAP3, 501 aa).

   UniProt REST search, gene_exact:TRAF2 AND organism_id:9606 AND reviewed:true, fields accession,id,gene_primary,gene_synonym,protein_name,length, run 2026-09-21 · UniProtKB/Swiss-Prot 2026
   ▸What this evidence cannot show

   This reproduces the standing brief's rule 62 exactly, two years after it was written, so it is a rediscovery rather than a new trap - and the row is absent from data/symbol-collisions.json, which is the part worth reporting. The 76-residue gap between the two records means verify-accessions.py would catch a swap on length, so the practical exposure here is lower than for the PBX or RAS families.

   ▸Evidence that argues against this (1)

   • Asserting an accession and a length defends against a swap and nothing more. It does not establish that TRAF2 protein is expressed in KMT2A-rearranged blasts or that it does anything in this disease, and the entry's confidence does not rest on it. Rolland T et al. 2014

• binds →AFF4 — the textbook SEC scaffold that KMT2A::AFF1 cells need LESS than other leukemias

  yeast two-hybrid hit in Rolland 2014; prey degree 97 in the same screen, cytoplasmic, not usable

• supports →The four AFF4 two-hybrid partners are three rows of one screen, not three experiments - and in that same screen GOLGA2 has 249 partners while AFF4 has five

  third hub prey of the four

• regulates →RELB — the NF-kB subunit that SUPPRESSES MLL-AF9 leukaemia, with a direct ChIP target list and a genetic rescue, which makes it the best-evidenced anti-leukaemic node in this beat

  TRAF2 with TRAF3 and the cIAPs degrades NIK, restraining the NFKB2/RELB non-canonical arm; TRAF2 loss releases it

• regulates →NFKB2 (p100/p52) — menin binds the non-canonical subunit too, and activating that arm SUPPRESSES MLL-AF9 leukaemia

  upstream brake on non-canonical NF-kB processing of NFKB2/p100 to p52

• related to →Canonical versus non-canonical NF-kB in KMT2A-rearranged leukaemia — the two arms have opposite signs, and the two papers usually read as contradicting each other actually agree

  the upstream node of the non-canonical arm; NIK stabilisation suppresses murine MLL-AF9 AML, so TRAF2 loss is predicted protective there