Oncofusion

start / graph / gene-nfkb2

Proteinmoderate confidence 3 of 5: One strong study, or several smaller consistent ones.Q36-menin-interactome

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

also called NFKB2 · p100 · p52 · LYT10 · NF-kappa-B2 · Q00653

NFKB2 belongs to the second, slower NF-kB circuit. Menin binds it, and in mouse MLL leukaemia switching that second circuit on actually shrinks the disease rather than feeding it. So the two NF-kB arms menin touches pull in opposite directions, and a drug that let go of both at once would be doing two contradictory things.

The same thing, for a scientist

NFKB2 (UniProt Q00653, 900 aa, primary gene name read back) is the p100 precursor processed to p52, the DNA-binding subunit of the non-canonical NF-kB pathway, activated through NIK and IKKa. Menin binds it: the 2001 paper recovered **p52 alongside p50 and p65** as specific, direct menin interactors, with the N-terminus of the NF-kB protein sufficient and menin 305-381 essential [E1]; IntAct curates the MEN1-NFKB2 pair from that paper as a **GST pull-down only**, the thinnest of the three [E2]. The reason NFKB2 gets its own entry rather than a line in a summary is a directional result. **Stabilising NIK activates non-canonical NF-kB, represses canonical NF-kB, and suppresses MLL-AF9 AML**, upregulating Dnmt3a and downregulating Mef2c; the authors propose developing NIK stabilisers as an AML therapy [E3]. That is the opposite sign to the canonical arm, where RELA occupancy sustains the MLL programme [E4]. **So 'menin restrains NF-kB' does not translate into a single prediction about menin inhibition.** Menin binds all three subunits; the canonical ones support the fusion programme and the non-canonical one opposes it; and no published work measures what a menin inhibitor does to either arm. **Counted absence:** `NFKB2`, `NF-kappaB2`, `p100` or `lyt10` against `KMT2A`, `MLL` or `"mixed-lineage leukemia"` returns **0 records** [E5].

Go deeper
Why the two arms matter separately
canonicalnon-canonical
subunitsRELA (p65) with NFKB1 (p50)RELB with NFKB2 (p52)
activationIKKb/IKKg, rapid, IkB degradationNIK and IKKa, slow, p100 processing
menin binds?yes, p65 and p50 1yes, p52 1
effect in MLL leukaemiasustains the LSC programme; required for fusion retention on chromatin 4suppresses MLL-AF9 AML when activated by NIK stabilisation 3

The NIK result also reports that activating the non-canonical arm represses the canonical one 3, so the two are coupled and an intervention at one is not silent at the other.

The practical consequence for this beat: any hypothesis of the form menin inhibition liberates NF-kB and that is a resistance mechanism has to say which NF-kB, and has to explain why the net of liberating three subunits with two opposing downstream consequences should favour the leukaemia. The liberation hypothesis (hyp-menin-inhibition-liberates-rela-and-reinforces-the-fusion-programme) is therefore written on RELA specifically, and this entry is the reason it is not written on NF-kB generically.

A counted absence with no inflation problem

Unusually for this project, the NFKB2 absence is clean. p100 and lyt10 are short aliases of the kind the brief warns inflates counts, and here the raw count is already zero, so there is nothing to title-check and nothing to over-count 5. NFKB2 has never been named in a KMT2A or MLL paper's title or abstract.

The limitation is the usual one and it is real: the NIK paper 3 is about the non-canonical arm in MLL-AF9 leukaemia and would not be returned by a NFKB2 symbol query, because it frames itself at the level of the pathway and the kinase. This is the same trap that cost this project a wrong absence claim for MED23. Recorded here so the zero is not misread as "nobody has studied the non-canonical arm in MLL leukaemia" — somebody has, and the paper is cited in this entry.

Facts and figures
uniprotQ00653
length AA900
uniprot resolutiongene_exact:NFKB2 AND organism_id:9606 AND reviewed:true; exactly one reviewed record; primary gene name read back as NFKB2; length 900 aa checked against the entry on 2026-09-20
nfkb armnon-canonical; p100 precursor processed to p52, downstream of NIK and IKKa
subnuclear localizationUNPUBLISHED for both p100 and p52 in any cell type. The informative localisation fact for this gene is the cytoplasm-to-nucleus step gated by p100-to-p52 processing, not a subnuclear compartment.
shown for fusionNO for localisation. Nobody has measured p100 processing or p52 nuclear translocation in any KMT2A-rearranged cell; the bounded query returned 0. The functional result this entry holds - that activating the non-canonical arm suppresses MLL-AF9 leukaemia - is murine MLL-AF9 and myeloid, and it is a phenotype, not a localisation measurement. NOTE FOR THE NEXT READER: brief rule 67, the rule about never trusting a zero from an uncontrolled query, was discovered on a counted absence about THIS GENE. The absence recorded here is controlled - the fusion clause alone returns 8,743 and the gene clause alone 5,191 - and the query form is one tag per term.
menin contact
{
  "methods": [
    "GST pull-down"
  ],
  "source_pmid": "11526476",
  "n_source_publications_in_intact": 1,
  "coip_exists": false,
  "binding_region_mapped": false
}
arm sign in mll leukaemia
{
  "canonical": "supports - RELA occupancy sustains the LSC programme and IKK/NF-kB is required for fusion retention on chromatin (PMID 24054986)",
  "non_canonical": "opposes - NIK stabilisation suppresses MLL-AF9 AML (PMID 29320732)",
  "coupling": "NIK stabilisation also represses canonical signalling, so the two arms are not independent"
}
literature
{
  "kmt2a_records_raw": 0,
  "kmt2a_records_title_checked": 0,
  "aliases_searched": [
    "NFKB2",
    "NF-kappaB2",
    "p100",
    "lyt10"
  ],
  "kmt2a_query": "(NFKB2[tiab] OR \"NF-kappaB2\"[tiab] OR p100[tiab] OR lyt10[tiab]) AND (KMT2A[tiab] OR MLL[tiab] OR \"mixed-lineage leukemia\"[tiab])",
  "pathway_level_paper_missed_by_that_query": "29320732"
}
depmap kmt2a selectivity
{
  "chronos_mean_kmt2a_rearranged": -0.2096,
  "chronos_mean_other_leukaemia": -0.1398,
  "difference": -0.0699,
  "pan_essentiality_all_lines": -0.1622,
  "selectivity_rank": 2215,
  "rank_denominator": 17728,
  "provenance": "Joined 2026-09-21 from analyses/depmap-kmt2a-selective-scan-full (generated 2026-09-19T10:12:06Z), the project's own genome-wide scan over 11 KMT2A-rearranged and 43 other leukaemia DepMap lines. NOT re-fetched and NOT independent of that scan: every caveat it carries applies here, including that the KMT2A-rearranged arm is a small number of lines drawn from nine individuals, that the rank denominator is 17,728 genes, and that a difference is not a dependency. Added because only 26 of 308 gene entries carried this number, which made any KMT2A-selectivity question answerable for 8% of the corpus.",
  "how_to_read": "A negative difference means MORE required in KMT2A-rearranged lines than in other leukaemia lines. A rank near 17,728 is ANTI-selective. Neither is evidence of a fusion-driven dependency on its own: check pan_essentiality_all_lines before calling anything selective, and note that a paralogue-redundant gene can score near zero for reasons that have nothing to do with this disease - see ana-the-paralogue-redundancy-excuse-is-not-tested-and-my-proxy-decided-the-answer."
}
localization
{
  "subcellular": "CYTOPLASM AS p100, NUCLEUS AS p52 - the localisation IS the activation step. UniProt Q00653 lists Nucleus and Cytoplasm with NO EVIDENCE CODES AT ALL (both lines uncoded), noting only that the protein is nuclear but also found in the cytoplasm in an inactive form complexed to an inhibitor. HPA immunofluorescence gives main locations Nucleoplasm AND Cytosol at reliability ENHANCED - HPA's highest tier, and the strongest subcellular call in this tranche.",
  "the_mechanism_is_the_localisation": "The p100 precursor holds its own ankyrin-repeat inhibitory domain in cis and is retained in the cytoplasm; NIK/IKKa-driven partial proteasomal processing removes that domain to give p52, which translocates to the nucleus with RelB. So for NFKB2 there is no such thing as a resting address independent of pathway state, and 'nucleus and cytoplasm' without the processing condition conveys nothing.",
  "subnuclear": "UNPUBLISHED. No subnuclear compartment has been reported for p100 or p52 in any cell type; the bounded subnuclear query returned 2 records, neither an NFKB2 subnuclear study.",
  "how_shown": "the UniProt lines are uncoded; HPA antibody immunofluorescence at Enhanced reliability is the best-supported measurement available",
  "uncoded_annotation_note": "Brief rule 91 item 2 in practice: this is a location comment with no ECO code whatsoever. A pipeline filtering UniProt on ECO:0000269 would record NFKB2 as having NO localisation. The uncoded text is nevertheless the more informative statement, because it carries the inhibitor-complex condition.",
  "tissue_of_evidence": "HPA's non-haematopoietic immunofluorescence panel; the processing biology is from lymphoid and transfected systems",
  "node": "localization-chromatin-bound-versus-soluble-fraction-kmt2a"
}
cell type
{
  "haematopoietic_answer": "B CELLS, by human genetics. Germline NFKB2 mutations cause common variable immunodeficiency through the non-canonical NF-kB pathway (PMID 24140114), and patients with NFKB2 mutations have INTRINSIC functional defects in their B cells (PMID 39405181). HPA Blood Atlas classes NFKB2 as Immune cell enhanced, detected in some, with neutrophil listed at 2.9 nTPM.",
  "why_this_matters_here": "The index disease is B-lineage. NFKB2's human loss-of-function phenotype is a B-cell phenotype, which is a stronger and more lineage-specific cell-type statement than anything the interaction literature provides for this protein.",
  "grade": "human germline genetics (class 2) plus an expression atlas (class 1). No NFKB2 localisation experiment in any leukaemic cell."
}
cell state
{
  "state": "NIK/IKKa-ACTIVATED (non-canonical arm on) versus resting. This is a signalling state and it is exactly what decides whether NFKB2 is nuclear.",
  "sign_in_this_disease": "This entry already records that ACTIVATING the non-canonical arm SUPPRESSES MLL-AF9 leukaemia. The localisation reading of that is unusual and worth stating: the therapeutic direction here is to drive NFKB2 INTO the nucleus, not to keep it out - the opposite of the canonical NF-kB logic in most cancers."
}
localization bounded absence
{
  "date": "2026-09-21",
  "tool": "NCBI E-utilities esearch, db=pubmed",
  "shape": "gene clause AND localisation clause AND KMT2A/MLL-fusion clause, one [tiab] tag per term (brief rule 94), every cytogenetic term quoted (brief rule 95)",
  "querytranslation_fusion_clause": "\"KMT2A\"[Title/Abstract] OR \"MLL\"[Title/Abstract] OR \"MLL1\"[Title/Abstract] OR \"MLL-AF4\"[Title/Abstract] OR \"MLL-AF9\"[Title/Abstract] OR \"MLL-ENL\"[Title/Abstract] OR \"KMT2A-rearranged\"[Title/Abstract] OR \"MLL-rearranged\"[Title/Abstract] OR \"t 4 11\"[Title/Abstract] OR \"t 9 11\"[Title/Abstract] OR \"MLL fusion\"[Title/Abstract]",
  "count": 0,
  "controls": {
    "fusion_clause_alone": 8743,
    "gene_clause_alone": 5191
  },
  "reading": "Counted in TITLES AND ABSTRACTS only; supporting figures were not searched (brief rule 80)."
}
How this could bite us back
  • NFKB2 loss-of-function in humans causes common variable immunodeficiency with adrenal insufficiency; perturbing this arm in children on immunosuppressive therapy is not free.
  • Because the two NF-kB arms have opposite signs in MLL leukaemia and are reciprocally coupled, a drug that releases all three menin-bound subunits at once could produce a net effect in either direction, which makes the outcome unpredictable rather than merely unknown.
Still open
  1. Does a menin inhibitor change p100 processing to p52 in a KMT2A-rearranged cell, and in which direction?
  2. If NIK stabilisation is anti-leukaemic partly by repressing the canonical arm, does it combine with or duplicate a menin inhibitor?
  3. Is the p52 contact with menin real in a cell, or an artefact of recombinant GST pull-down? A single endogenous co-IP would settle twenty-five years of it being carried forward on one in vitro method.
  4. This entry's own finding is that ACTIVATING the non-canonical arm suppresses MLL-AF9 leukaemia - i.e. the therapeutic direction is to drive p52 INTO the nucleus. Nobody has measured p100-to-p52 processing or p52 nuclear content in a KMT2A-rearranged blast, so the graph does not know whether the arm is already off in this disease or merely unexploited. A p52 nuclear/cytoplasmic western in SEM, RS4;11 and a KMT2A-r AML line would settle it.
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 NF-kB proteins p50, p52 and p65 interact specifically and directly with menin in vitro and in vivo; the N-terminus of the NF-kB proteins is sufficient for binding menin and amino acids 305-381 of menin are essential.

    The tumor suppressor protein menin interacts with NF-kappaB proteins and inhibits NF-kappaB-mediated transactivation. · Heppner C, Bilimoria KY, Agarwal SK, Kester M, Whitty LJ, Guru SC, Chandrasekharappa SC, Collins FS, Spiegel AM, Marx SJ, Burns AL · Oncogene 2001
    PMID 11526476DOI
    What this evidence cannot show

    `established` for the reported system only; no haematopoietic cell appears, and the functional repression the paper demonstrates concerns p65, not p52. Full text unreadable through all five routes: no PMCID, and the DOI resolves to a 303-character stub.

    Evidence that argues against this (1)
    • p52 evidence in that paper is the thinnest of the three subunits - a GST pull-down with no reciprocal co-immunoprecipitation - so the claim that menin binds the non-canonical subunit rests on one in vitro method in one paper. IntAct (EMBL-EBI) 2026
  2. E2established confidence 5 of 5: Multiple independent lines of evidence; treated as settled in the field.database

    IntAct curates the MEN1-NFKB2 pair from PMID 11526476 as a single GST pull-down read by western blot, with no co-immunoprecipitation and no mapped binding region on either partner; it holds no other source publication for the pair.

    What this evidence cannot show

    A structured transcription of one paper, not independent evidence. A GST pull-down with recombinant protein establishes that two purified proteins can associate; it does not establish that they do so in a cell at physiological stoichiometry.

    Evidence that argues against this (1)
    • A curated interaction inherits every limitation of the single experiment behind it. Anti-bait co-immunoprecipitation and affinity-capture mass spectrometry cannot distinguish a direct contact from a shared-complex association, tagged-bait overexpression can create contacts that do not occur at endogenous stoichiometry, and a partner curated from one publication has by definition never been independently replicated. The enumeration is accurate about the database; the database is not a statement about the blast. Heppner C et al. 2001
  3. E3established confidence 5 of 5: Multiple independent lines of evidence; treated as settled in the field.paper

    Stabilisation of NF-kB-inducing kinase (NIK) suppresses AML: it activates NF-kB non-canonical signalling and represses NF-kB canonical signalling, upregulates Dnmt3a and downregulates Mef2c, and suppresses MLL-AF9-induced AML development. Canonical NF-kB signalling is constitutively activated in AML stem cells and required for their self-renewal, while the non-canonical arm had been largely overlooked.

    Stabilization of NF-κB-Inducing Kinase Suppresses MLL-AF9-Induced Acute Myeloid Leukemia. · Xiu Y, Dong Q, Li Q, Li F, Borcherding N, Zhang W, Boyce B, Xue HH, Zhao C · Cell Rep 2018
    PMID 29320732DOI
    What this evidence cannot show

    Mouse MLL-AF9, a myeloid model; nothing lymphoid and nothing human primary. NIK stabilisation is a genetic manipulation and its effect is attributed to the non-canonical arm by inference from the pathway it activates - the paper does not knock out NFKB2 to prove the effect runs through p52. The connectivity-MAP-derived verteporfin repurposing claim is a separate and much weaker part of the same paper.

    Evidence that argues against this (1)
    • Because NIK stabilisation also REPRESSES canonical signalling, the anti-leukaemic effect could be entirely a canonical-suppression effect with the non-canonical activation as a bystander. The paper's own data cannot separate the two, which matters because only the canonical arm has a direct MLL-chromatin mechanism attached to it. Kuo HP et al. 2013
  4. E4established confidence 5 of 5: Multiple independent lines of evidence; treated as settled in the field.paper

    RELA occupies promoter regions of crucial MLL target genes and sustains the MLL-dependent leukaemia stem cell programme; IKK/NF-kB signalling is required for retention of wild-type and fusion MLL protein on chromatin, and suppressing NF-kB phenocopies loss of the functional MLL oncoprotein.

    Epigenetic roles of MLL oncoproteins are dependent on NF-κB. · Kuo HP, Wang Z, Lee DF, Iwasaki M, Duque-Afonso J, Wong SH, Lin CH, Figueroa ME, Su J, Lemischka IR, Cleary ML · Cancer Cell 2013
    PMID 24054986DOI
    What this evidence cannot show

    Mouse MLL-AF1p, MLL-AF10 and MLL-AF9 plus human AML cohorts; myeloid, not infant B-ALL. Cited in this entry only to establish the sign of the canonical arm, for contrast with E3.

    Evidence that argues against this (1)
    • RELA occupancy at promoters is close to ubiquitous in an activated haematopoietic cell, so finding it at MLL target promoters is weak evidence that it is doing anything specific there. Occupancy is not requirement, and the leukaemia-stem-cell claim rests on a separate experiment with its own assumptions. Xiu Y et al. 2018
  5. E5strong confidence 4 of 5: Replicated in peer-reviewed work by more than one group.database

    A PubMed title/abstract search over NFKB2, NF-kappaB2, p100 and lyt10 against KMT2A, MLL and 'mixed-lineage leukemia' returns 0 records. NFKB2 has never been named in the title or abstract of a KMT2A or MLL paper.

    What this evidence cannot show

    A zero at the symbol level is not a zero at the pathway level, and in this case demonstrably is not: the NIK paper studies the non-canonical arm in MLL-AF9 leukaemia and this query does not return it. This is the MED23 trap and it is flagged rather than hidden.

    Evidence that argues against this (1)
    • The paper this query cannot see is the most relevant one that exists, which is the clearest possible demonstration that a subunit-keyed counted absence is not an absence of study. Xiu Y et al. 2018
  6. E6moderate confidence 3 of 5: One strong study, or several smaller consistent ones.database

    UniProt Q00653 records NFKB2 as nuclear and also cytoplasmic in an inactive form complexed to an inhibitor, and BOTH location lines carry no evidence code at all, so a pipeline filtering on ECO:0000269 would record this protein as unlocalised.

    What this evidence cannot show

    An uncoded comment is a curator's summary of the field rather than a pointer to an experiment, so it cannot be traced to a source. It is nevertheless the more useful of the two available statements because it carries the inhibitor-complex condition, which the compartment names alone do not.

    Evidence that argues against this (1)
    • The uncoded record's implicit model - cytoplasmic p100 equals inactive - is wrong, and repeating it would propagate the error. p100 has signalling roles as a precursor: a TNF-p100 pathway subverts noncanonical NF-kB signalling in inflamed secondary lymphoid organs, so accumulated cytoplasmic p100 is an active participant rather than a stored reservoir awaiting processing. 'Inactive form complexed to an inhibitor' is UniProt's phrasing and it understates the precursor's role. EMBO J 2017
  7. E7strong confidence 4 of 5: Replicated in peer-reviewed work by more than one group.database

    Human Protein Atlas v25.1 assigns NFKB2 main subcellular locations Nucleoplasm AND Cytosol at reliability ENHANCED, HPA's highest tier, and classes it Immune cell enhanced in the Blood Atlas.

    What this evidence cannot show

    Even at Enhanced reliability this is antibody immunofluorescence in non-haematopoietic lines under basal conditions, and a dual nucleoplasm/cytosol call cannot distinguish p100 from p52 unless the antibody epitope is C-terminal to the processing site - which HPA does not state. Blood Atlas values are RNA.

    Evidence that argues against this (1)
    • Even at HPA's highest reliability tier this call cannot do the work the entry wants from it, because an antibody cannot separate p100 from p52 unless its epitope lies C-terminal to the processing site - and HPA does not state the epitope. A dual Nucleoplasm/Cytosol pattern is therefore compatible with an entirely unprocessed p100 pool, an entirely processed p52 pool, or any mixture. For a gene whose localisation IS its activation state, an antibody that cannot see the processing state reports nothing about pathway activity. UniProt Consortium 2026
  8. E8strong confidence 4 of 5: Replicated in peer-reviewed work by more than one group.paper

    Germline NFKB2 mutations cause common variable immunodeficiency through the non-canonical NF-kB pathway, and patients carrying them have intrinsic functional defects in their B cells - making B lineage the human loss-of-function phenotype of this gene.

    PMID 24140114
    What this evidence cannot show

    Human germline disease, not leukaemia and not a KMT2A fusion. Title-level reading of the founding paper; the B-cell-intrinsic point comes from a separate 2025 study (PMID 39405181), read at title level. An immunodeficiency phenotype identifies the lineage that needs the gene, which is not the same as identifying where the protein sits in a blast.

    Evidence that argues against this (1)
    • The germline phenotype may not report the normal function of the gene. Pathogenic NFKB2 alleles are frequently dominant-negative PROCESSING mutants rather than nulls, so the B-cell defect can reflect the behaviour of an un-processable p100 poisoning the pathway rather than the absence of nuclear p52 - two different molecular lesions with the same gene name. The phenotype is also immunodeficiency, not leukaemia, so it identifies a lineage that needs the pathway intact and says nothing about where the protein sits in a blast. Clin Exp Immunol 2025
Connected to