TIFAB — a 161-residue NF-kB adaptor-layer protein with TWO direct KMT2A-fusion papers, which is more disease-specific evidence than any receptor or TRAF in this beat has
also called TIFAB · TIFA-like protein · TRAF-interacting protein with FHA domain-containing protein B · Q6ZNK6
TIFAB is a small protein that dampens the NF-kB alarm signal, and in mouse MLL leukaemia it makes the disease worse rather than better. It is one of very few upstream signalling proteins with experiments done in this exact leukaemia, and it has the unusual distinction of being tested in two different KMT2A fusions.
The same thing, for a scientist
TIFAB (UniProt Q6ZNK6, 161 aa, primary gene name read back) is a TRAF-interacting FHA-domain protein described as an **inhibitor of NF-kB signalling**. It earns an entry over the receptors and adaptors this beat considered and rejected for one reason: **it has two papers in KMT2A-rearranged leukaemia specifically, using two different fusions.** In **MLL-AF9**: TIFAB is upregulated in human AML and correlates negatively with patient survival; forced TIFAB expression reverses NIK-induced impairment of AML by downregulating RelB and upregulating HOXA9; and TIFAB-driven transcription blocks myeloid development, raises an LSC signature and resembles HOXA9-MEIS1 and HOXA9-NUP98 patterns [E1]. In **KMT2A::MLLT3**: Tifab deletion impairs leukaemia stem/progenitor engraftment, glucose uptake and mitochondrial function, downregulating MYC, HOXA9/MEIS1, mTORC1 and glycolysis/OXPHOS genes, and acting through **HNF4A**, itself regulated by TIFAB's inhibition of RelB [E2]. **The unresolved discrepancy is worth naming:** the 2018 paper's RelB ChIP reports binding at Mef2c and Dnmt3a but **not** at the Tifab promoter, while the 2021 paper calls TIFAB a direct RelB target [E1][E3]. Both are from the same laboratory. This graph should not pick one.
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.
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.
Why this entry and not TRAF6, MYD88 or TNFRSF1A
The brief's instruction was to write upstream receptors and adaptors only where there is evidence in this disease, and to record the rest as bounded absences. Running that test:
| candidate | KMT2A/MLL title-abstract records | verdict |
|---|---|---|
| TIFAB | 3, two of them direct KMT2A-fusion functional studies | entry |
| TRAF6 | 3, none TRAF6-focused; reached only as part of a TLR-TRAF6-NF-kB axis in an IMPDH-inhibitor paper | bounded absence |
| MYD88 | 7, all CLL/lymphoma/mutation-panel | bounded absence |
| MAP3K7 (TAK1) | 3, none on topic | bounded absence |
| TRAF2 | 2, neither on topic; and its DepMap signal failed a permutation null at p=0.1618 | bounded absence |
| TNFRSF1A | 2, neither KMT2A-specific | bounded absence |
| IL1R1 | 25 raw, one on topic and it is AML-wide with one MLL-AF9 experiment | see mechanism-il1-irak-axis-in-leukaemia-stem-cells |
The full enumeration with queries is in ana-nfkb-upstream-enumeration-and-bounded-absences.
The direction, and what it implies for the NIK proposal
TIFAB is pro-leukaemic and it works by removing RelB, the protective subunit 1. That makes it a named, already-documented resistance route to any therapy that works by raising non-canonical output: a leukaemia that upregulates TIFAB escapes NIK stabilisation, and TIFAB is already high in human AML 1. HOXA9 does the same thing by the same route 1. So the NIK proposal has two published escape mechanisms before any drug exists — an unusually well-specified bite-back.
Localization: a three-instrument bounded absence, and the cell type that IS known (Q87, 2026-09-21)
Subcellular: UNKNOWN. UniProt Q6ZNK6 has no SUBCELLULAR LOCATION comment; HPA v25.1 has no IF data; the whole PubMed literature is 21 records and the 4 that co-mention a localisation term were retrieved and contain no localisation experiment. Absence class: never-measured.
Subnuclear: not answerable. The subcellular answer does not exist, so a subnuclear one would be a guess about a guess. A cytoplasmic prediction from TIFA/TRAF6 is available and is deliberately not recorded, because it is inference from a complex partner.
Cell type and state. Leukaemia stem/progenitor cells in KMT2A::MLLT3-induced AML (mouse), defined by engraftment plus glucose uptake and mitochondrial function, with MYC and HOXA9/MEIS1 programmes downstream.
Shown for fusion. KMT2A::MLLT3 and MLL-AF9 for function, and the two must not be merged; nobody for localisation. The single fusion-scoped localisation hit is the KMT2A::MLLT3 metabolic paper, where the word 'nuclear' belongs to hepatocyte NUCLEAR factor 4 alpha rather than to a compartment.
| uniprot | Q6ZNK6 |
|---|---|
| length AA | 161 |
| alphafold ID | AF-Q6ZNK6-F1 |
| resolution note | gene_exact:TIFAB AND organism_id:9606 AND reviewed:true returned 1 reviewed row; gene_primary read back as TIFAB; length 161 aa asserted because the alias 'TIFA-like protein' invites a paralogue swap. |
| pathway arm | adaptor layer; described as an NF-kB inhibitor, acts here by suppressing RelB |
| direction in disease | pro-leukaemic in MLL-AF9 and KMT2A::MLLT3 mouse models |
| subnuclear localization | BOUNDED ABSENCE (never-measured). TIFAB has no subcellular localisation record at all: UniProt Q6ZNK6 carries no SUBCELLULAR LOCATION comment, HPA v25.1 has no immunofluorescence data for it, and of the 21 PubMed records for TIFAB[tiab] the 4 that co-mention a localisation term were each retrieved and none reports the protein's compartment. A subnuclear answer cannot be given because the subcellular answer has never been measured, and inferring one from TIFA, TRAF6 or USP15 would be inference from a complex partner. |
| cell state | LEUKAEMIA STEM/PROGENITOR STATE with a metabolic signature - high glucose uptake, intact mitochondrial function, MYC and HOXA9/MEIS1 programmes on. Tifab loss collapses that state in KMT2A::MLLT3-driven AML. Recorded as a state because it is defined by engraftment capacity plus measured metabolic readouts rather than by a marker label alone. |
| shown for fusion | KMT2A::MLLT3 and MLL-AF9 for FUNCTION; NOBODY for LOCALISATION. This entry already records two direct KMT2A-fusion papers using two different fusions, and the naming matters: the metabolic/HNF4A work is in KMT2A::MLLT3 and the NIK/RelB/HOXA9 work is in MLL-AF9. Neither measures TIFAB's compartment. The fusion-scoped localisation query returns exactly 1 record, PMID 39626355, which is the KMT2A::MLLT3 metabolic paper and contains no localisation experiment - the word 'nuclear' in it belongs to hepatocyte NUCLEAR factor 4 alpha, the transcription factor, and not to a compartment. Screened count on topic: 0. |
fusions tested
[ "MLL-AF9", "KMT2A::MLLT3" ]
downstream routes claimed
[ "HOXA9", "USP15/p53", "HNF4A" ]
depmap kmt2a selectivity
{
"chronos_mean_kmt2a_rearranged": -0.1135,
"chronos_mean_other_leukaemia": -0.0126,
"difference": -0.1009,
"pan_essentiality_all_lines": -0.0213,
"selectivity_rank": 1125,
"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": "UNKNOWN - and this is the cleanest bounded absence in the tranche, because THREE independent instruments return nothing. (1) UniProt Q6ZNK6 carries NO SUBCELLULAR LOCATION comment at all - not an uncoded one, not an ECO:0000305 curator inference, none; fetched 2026-09-21. (2) HPA v25.1 has no immunofluorescence data for TIFAB: 'Reliability (IF)' is null and all three subcellular-location fields are null. (3) The whole PubMed literature for the symbol is 21 records (TIFAB[tiab], 2026-09-21); of those, 4 co-mention any localisation term, and all four were retrieved and read - a KMT2A::MLLT3 metabolic study, a TIFA-TIFAB heterodimer structure paper, an unrelated miR-626 oral-squamous-carcinoma paper, and a NEUROG1 microdeletion case report in which TIFAB is merely one of three genes inside the deleted interval. None reports where TIFAB is in a cell.",
"subnuclear": "BOUNDED ABSENCE, and not even a compartment-level answer exists to build on. Because the subcellular answer is itself unknown, a subnuclear answer would be a guess about a guess. TIFAB's known partners - TIFA, TRAF6, USP15 - are signalling proteins whose own biology is cytoplasmic, and the FHA domain is a phosphothreonine-reading module common to both compartments, so a cytoplasmic prediction is available and is deliberately NOT recorded here as a finding: it would be inference from a complex partner, which this tranche was instructed not to do.",
"absence_class": "never-measured",
"how_bounded": "UniProt REST Q6ZNK6.json (no SUBCELLULAR LOCATION comment); HPA v25.1 search_download API (all subcellular fields null, IF reliability null); PubMed esearch TIFAB[tiab] = 21, and TIFAB[tiab] AND (localization[tiab] OR localisation[tiab] OR cytoplasm[tiab] OR cytoplasmic[tiab] OR nuclear[tiab]) = 4, all four retrieved by efetch and read",
"what_would_close_it": "one immunofluorescence or fractionation panel on endogenous TIFAB in a myeloid line - an afternoon's work, and nobody has published it in the twelve years since the microdeletion report first named the gene"
}cell type
{
"haematopoietic_answer": "LEUKAEMIA STEM/PROGENITOR CELLS in KMT2A::MLLT3-induced AML, in mouse. Tifab deletion in KMT2A::MLLT3-induced AML impairs LSPC engraftment, glucose uptake and mitochondrial function, and downregulates MYC, HOXA9/MEIS1, mTORC1 signalling and glycolysis/OXPHOS gene sets; HNF4A is identified as the key TIFAB target. This is the most cell-type-specific statement available for TIFAB and it is a FUNCTIONAL one, not a localisation one.",
"expression_atlas": "HPA v25.1: RNA tissue specificity 'Tissue enhanced'; no protein IF data."
}- TIFAB is already upregulated in human AML, and its documented action is to remove RelB. Any therapy that works by raising non-canonical output has a pre-existing, expressed escape mechanism in the patients it would be given to.
- HOXA9 does the same thing by the same route, and HOXA9 is the defining target of this disease. A KMT2A-rearranged leukaemia is, by construction, a leukaemia with the NIK-escape mechanism switched on.
- A 161-residue adaptor with no enzymatic activity and no pocket is not a tractable target; the therapeutic value here is as a biomarker or as an argument against a strategy, not as a target.
- Is TIFAB a direct RelB ChIP target? Two papers from the same group say opposite things and neither cites the other on this point.
- Is TIFAB expressed or functional in KMT2A-rearranged B-ALL? Both models are myeloid and TIFAB's known biology is myelodysplasia-centred.
- Which of the three proposed downstream routes carries the phenotype - HOXA9, USP15/p53, or HNF4A? Three mechanisms for one gene across three papers is a warning, not a synthesis.
- Would TIFAB expression predict failure of a non-canonical-activating therapy? It is the cleanest pre-specified biomarker anywhere in this beat and nobody has proposed it as one.
- E1established confidence 5 of 5: Multiple independent lines of evidence; treated as settled in the field.paper
In MLL-AF9 AML, RelB suppresses its direct target TIFAB, which is upregulated in human AML and correlates negatively with patient survival; forced TIFAB expression reverses NIK-induced impaired AML development through downregulation of RelB and upregulation of HOXA9, blocks myeloid cell development, upregulates a leukaemia stem cell signature, induces gene expression patterns similar to HOXA9-MEIS1 and HOXA9-NUP98, and upregulates oxidative phosphorylation. Forced HOXA9 expression also largely releases NIK's inhibitory impact via downregulation of RelB and upregulation of RelA.
TIFAB accelerates MLL-AF9-Induced acute myeloid leukemia through upregulation of HOXA9. · Zhao J, Xiu Y, ... Zhao C · iScience 2021What this evidence cannot show
Abstract read; full text not retrieved. Mouse MLL-AF9, overexpression-based, from the same laboratory as the NIK paper it extends. The human survival correlation is a database association and its cohort, method and multiple-testing correction were not checked here. The claim that TIFAB is a direct RelB target conflicts with the earlier paper's ChIP (see E3).
Evidence that argues against this (1)
- TIFAB is not only an NF-kB-axis protein in haematopoiesis: it also regulates USP15-mediated p53 signalling during stressed and malignant haematopoiesis, so a TIFAB phenotype cannot be attributed to NF-kB without excluding the p53 route. The MLL-AF9 paper's own companion work invokes that route. Wang Y et al. 2025
- E2established confidence 5 of 5: Multiple independent lines of evidence; treated as settled in the field.paper
TIFAB is described as an inhibitor of NF-kB signalling with roles in haematopoiesis, myelodysplastic neoplasms and leukaemia. Tifab deletion in KMT2A::MLLT3-induced AML impairs leukaemia stem/progenitor cell engraftment, glucose uptake and mitochondrial function, and downregulates MYC, HOXA9/MEIS1, mTORC1 signalling, glycolysis and oxidative phosphorylation genes. HNF4A was identified as a key TIFAB target regulated through inhibition of the NF-kB component RelB, which suppresses Hnf4a in leukaemia cells; HNF4A rescues the metabolic defects of Tifab deletion and Hnf4a knockdown attenuates TIFAB-mediated enhancement of LSPC function. Tifab had previously been shown to enhance KMT2A::MLLT3-driven AML by upregulating Hoxa9 or by USP15-mediated downregulation of p53 signalling.
TIFAB modulates metabolic pathways in KMT2A::MLLT3-induced AML through HNF4A. · Wang Y, Xiu Y, ... Zhao C · Blood Advances 2025What this evidence cannot show
Abstract read; full text not retrieved. Mouse KMT2A::MLLT3 model, same laboratory as E1, so the two KMT2A fusions are two models in one group rather than independent replication. The TIFAB phenotype is now attributed to three different downstream routes across three papers - HOXA9, USP15/p53, and HNF4A - which is a sign the mechanism is not settled.
Evidence that argues against this (1)
- Impairing leukaemia stem/progenitor enrichment on deletion shows requirement in that model, not that the metabolic route through HNF4A is how it acts - the two are separate claims and the second needs the epistasis. The system is also KMT2A::MLLT3, a myeloid fusion, and this graph's index disease is infant B-lineage KMT2A::AFF1. Xiu Y et al. 2018
- E3established confidence 5 of 5: Multiple independent lines of evidence; treated as settled in the field.paper
In the earlier MLL-AF9 study, chromatin immunoprecipitation found that RelB binds directly to the Mef2c promoter and to a Dnmt3a promoter upstream of exon 7, but NOT to the Tifab promoter.
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 Reports 2018What this evidence cannot show
Read in the full text this session (PMC5810947); the Tifab negative is stated in the results and no primer coordinates are given in the flattened text, so the discrepancy with E1's 'direct target' language cannot be adjudicated from what is available here. Recorded as an unresolved discrepancy rather than as a correction to either paper.
Evidence that argues against this (1)
- A negative ChIP result is weak on its own - the antibody, the crosslinking and the primer placement all decide it - but here it is load-bearing, because it contradicts the claim elsewhere in this beat that TIFAB is RelB's direct target. The contradiction is recorded rather than resolved, and resolving it needs the two datasets side by side. Zhao J et al. 2021
- E4established confidence 5 of 5: Multiple independent lines of evidence; treated as settled in the field.database
UniProt reviewed human record Q6ZNK6 (TIFAB_HUMAN) has primary gene name TIFAB and length 161 aa; gene_exact:TIFAB AND organism_id:9606 AND reviewed:true returned exactly one row and gene_primary read back as TIFAB.
UniProtKB entry Q6ZNK6 (TIFAB_HUMAN), TRAF-interacting protein with FHA domain-containing protein B · UniProt Consortium · UniProt 2026What this evidence cannot show
Resolved 2026-09-20 with gene_primary requested and all rows read. The paralogue TIFA was not resolved; TIFAB's alias 'TIFA-like protein' is exactly the shape Rule 63 warns about, so the accession is asserted here for verify-accessions.py to check the 161-aa length.
Evidence that argues against this (1)
- A reviewed database record fixes the canonical sequence and primary symbol on the date it was read, and nothing more. It does not establish that the protein is expressed in KMT2A-rearranged blasts, which isoform is present, or that the subunit acts in this disease. Accession and length are asserted so that a swap fails verify-accessions.py - a defence this graph needs because IKK-family and NF-kB subunit names collide heavily in the older literature - not because the record carries biological weight. Xiu Y et al. 2018
- E5strong confidence 4 of 5: Replicated in peer-reviewed work by more than one group.database
TIFAB has no published subcellular localisation. UniProt Q6ZNK6 carries no SUBCELLULAR LOCATION comment of any kind (fetched 2026-09-21), and HPA v25.1 has no immunofluorescence data for the protein - IF reliability and all three subcellular-location fields are null.
UniProtKB Q6ZNK6 (TIFAB) SUBCELLULAR LOCATION comment, fetched 2026-09-21 · UniProt Consortium 2026What this evidence cannot show
This is an absence in two curated databases, which is weaker than a literature sweep: UniProt records a compartment only when a curator has read an experiment reporting one, and HPA images only the proteins for which it has a validated antibody, so 'no row' can mean 'no usable antibody' rather than 'nobody looked'. The literature bound is supplied separately, and it is small enough to have been read in full: 21 records total for TIFAB[tiab].
Evidence that argues against this (1)
- A cytoplasmic answer is available by inference and is deliberately refused here, but the inference is not frivolous and a reader should know how strong it is before treating the absence as complete. The one structural study of TIFAB shows it forms a complex with MONOMERIC TIFA rather than with the TIFA dimer, acting as a negative regulator of the ALPK1-TIFA-TRAF6 route to NF-kB - machinery that operates in the cytosol. So the prediction is specific and well-motivated, which makes recording it as a finding more tempting and not more correct: a heterodimer partner's compartment is not a measurement of this protein's compartment, and TIFAB's other documented activity, USP15-mediated downregulation of p53 signalling, points the other way. Nakamura T et al. 2024
- E6strong confidence 4 of 5: Replicated in peer-reviewed work by more than one group.database
The entire PubMed literature for TIFAB is 21 records (TIFAB[tiab], 2026-09-21). Of these, 4 co-mention any localisation term: TIFAB[tiab] AND (localization[tiab] OR localisation[tiab] OR cytoplasm[tiab] OR cytoplasmic[tiab] OR nuclear[tiab]) returns PMIDs 39626355, 38442163, 35549438 and 23419067. All four were retrieved by efetch and read: none reports TIFAB's subcellular compartment. Two are off topic entirely - an miR-626 radiosensitivity study in oral squamous cell carcinoma, and a case report of a boy with a homozygous 5q31.1 microdeletion in which TIFAB is simply one of three genes inside the deleted interval. The absence class is never-measured.
What this evidence cannot show
Bounded by titles and abstracts, so a fractionation blot buried in a supplementary figure would be invisible (rule 80); with a 21-record literature, however, the number of places such a panel could hide is very small. TIFAB is not an alias-rich symbol - it has no retired HGNC symbol in play and the only variant spelling in use is the expansion 'TRAF-interacting protein with FHA domain-containing protein B' - so the rule-71 name-list failure mode is a small risk here compared with the septin case.
Evidence that argues against this (1)
- The absence is being recorded as never-measured, and the strongest argument against that class is that TIFAB has been handled repeatedly as a tagged, overexpressed construct in leukaemia work, which means somebody has almost certainly SEEN where it goes down a microscope without writing it down - which would make the class measured-not-reported rather than never-measured, and would change the cost of closing it from wet work to opening a supplementary figure. The KMT2A::MLLT3 study is the specific place to look: it works with TIFAB-overexpressing leukaemia stem/progenitor cells throughout and compares them against Tifab-deleted cells, so a construct and its controls exist. That argues the gap is cheaper to close than 'never-measured' implies. Wang Y et al. 2025
TIFAB inhibits RelB, which suppresses Hnf4a; HNF4A rescues the metabolic defects of Tifab deletion
RelB suppresses TIFAB (but the 2018 ChIP found no RelB binding at the Tifab promoter)
forced TIFAB downregulates RelB and reverses NIK's anti-leukaemic effect
TIFAB accelerates MLL-AF9 AML through upregulation of HOXA9
Tifab deletion in KMT2A::MLLT3 AML impairs LSPC engraftment and metabolism
TIFAB has NO subcellular localisation record in UniProt, HPA or the 21-record literature; a cytoplasmic answer inferred from TIFA/TRAF6 was refused (Q87 2026-09-21)
- expressed in →The stem-like self-renewal state — a limited stem-cell module grafted onto a committed cell
Tifab loss in KMT2A::MLLT3-induced AML collapses leukaemia stem/progenitor engraftment and the MYC/HOXA9-MEIS1 metabolic programme (PMID 39626355, mouse)
ITS TWO EVIDENCE ITEMS DISAGREE: E2 carries RelB suppressing TIFAB as a direct target while E3 records the companion study's ChIP as negative at the Tifab promoter.