GATAD2B / p66-beta (Q8WXI9) - the NuRD scaffold with the broadest presence in the pan-MLL-fusion interactome, five of seven baits and still not the index fusion, and a haploinsufficiency syndrome that caps any therapeutic idea
also called GATAD2B · p66beta · p66-beta · hp66beta · P66B_HUMAN · KIAA1150 · GATA zinc finger domain-containing protein 2B
GATAD2B is the near-twin of GATAD2A and the NuRD scaffold protein that turns up most often in the published MLL-fusion pulldown - in five of seven experiments. It still does not turn up in the one that matters here, and losing one copy of the gene causes a recognisable intellectual-disability syndrome, which sets a hard ceiling on any idea of drugging it.
The same thing, for a scientist
GATAD2B (Q8WXI9, 593 aa, UniProt entry name P66B_HUMAN, synonym KIAA1150) is the p66-beta paralogue of GATAD2A. It binds MBD2 and MBD3 directly, is a potent transcriptional repressor, and acts in synergy with p66-alpha such that knocking down either impairs the other's repressive function - but it binds MBD2 with lower affinity than p66-alpha and through a single interaction domain rather than two [E1][E2]. In this project's re-scoring of the Skucha 2018 pan-MLL-fusion network it is the **most broadly recovered NuRD scaffold subunit: five of seven bait columns** (MLL-AF1p p = 0.0689, MLL-AF9 p = 0.0266, MLL-CBP p = 0.0217, MLL-EEN p = 0.042, MLL-ENL p = 0.00581), negative control 0|0 - and it is **absent from the MLL-AF4 (= KMT2A::AFF1) column**, as is its paralogue [E3]. So the two most broadly present NuRD scaffolds in the panel are both missing from the index fusion. In DepMap it ranks 2,141 of 17,728, behind GATAD2A's 866 [E4]. The literature ties it to no KMT2A fusion under either name [E5]. What it does have, and GATAD2A does not, is a human loss-of-function phenotype: heterozygous loss-of-function causes a recognisable intellectual-disability syndrome, and 1q21.3 microdeletion encompassing it is a recurrent microdeletion syndrome in which GATAD2B loss is the major contributor to the characteristic facies and ID [E6].
The paralogue asymmetry, stated as a question rather than a finding
| GATAD2A | GATAD2B | |
|---|---|---|
| Skucha bait columns | 1 (MLL-AF9) | 5 (all but MLL-AF4 and MLL-GAS7) |
| in the index fusion's column | no | no |
| negative control | 0|0 | 0|0 |
| DepMap KMT2A-selectivity rank | 866 | 2,141 |
| MBD2 binding | higher affinity, two interaction domains | lower affinity, one |
| human LoF phenotype | none reported | intellectual-disability syndrome |
The asymmetry runs in opposite directions on the two instruments: GATAD2B is recovered more often by AP-MS and GATAD2A ranks better in the dependency scan. Nothing distinguishes biology from detectability here. Protein abundance, peptide ionisation and antibody-free AP-MS recovery differ between paralogues for reasons unrelated to function, and a re-search of the public PXD009338 raw files comparing the two paralogues' spectral counts head to head would settle it. That is one of the named uses of ana-q193-planned-re-search-of-pxd009338-for-the-nurd-deacetylase-module.
What does not need settling: neither paralogue is in the KMT2A::AFF1 column, so for the index fusion the question is moot on present data.
The human loss-of-function phenotype, and why it belongs in this entry rather than a footnote
Heterozygous loss-of-function of GATAD2B causes a recognisable syndrome with intellectual disability, and an infant with a de novo 1.06 Mb 1q21.3 deletion encompassing GATAD2B, TPM3 and HAX1 presented with feeding difficulty, developmental delay, hypotonia and dysmorphic features; across eight patients with 1q21.3 microdeletion the authors conclude that loss of function of GATAD2B is the major contributor to the characteristic facies and intellectual disability 6.
This graph has a standing caution that a constraint statistic is not a safety signal - gnomAD pLI measures intolerance to lifelong heterozygous loss from conception, not the effect of partial reversible occupancy in an adult. This is not that. It is an observed human phenotype from losing one copy of this specific gene, in the developmental window, in exactly the patient population this project treats: infants. A NuRD-scaffold-directed agent given to an infant would be acting on a gene whose 50% dose reduction has a named neurodevelopmental syndrome attached.
The honest counterweight is that a drug is not a germline deletion and the comparison is still indirect. But for a target in infant leukaemia the argument runs closer than it usually does, and it belongs in the entry rather than in a footnote.
| uniprot | Q8WXI9 |
|---|---|
| length AA | 593 |
| uniprot entry name | P66B_HUMAN |
| uniprot recommended name | Transcriptional repressor p66-beta |
| accession resolution | Resolved live 2026-09-24: gene_exact:GATAD2B AND organism_id:9606 AND reviewed:true returns exactly one reviewed row, gene_primary GATAD2B, Q8WXI9, P66B_HUMAN, 593 aa, synonym KIAA1150. Matches the accession in Skucha Supplementary Table 5. |
| coordinate provenance | NO RESIDUE NUMBER IS QUOTED IN THIS ENTRY. The conserved-region coordinates available to this beat (CR1 144-178, CR2 340-480) are GATAD2A's, from human UniProt Q86YP4, and are deliberately NOT transferred to the paralogue - a 633-residue and a 593-residue protein do not share a numbering. If a future beat needs GATAD2B's region boundaries, they must be fetched from Q8WXI9 and string-matched against its own sequence. |
| localization | Nucleus |
| localization evidence | UniProt Q8WXI9 was resolved for accession, length and synonyms on 2026-09-24; its subcellular-location comments and evidence codes were NOT fetched in this cycle, and this entry therefore makes no ECO-coded localisation claim. Recorded as unfetched rather than assumed from the paralogue. |
| subnuclear localization | NO ASSIGNMENT RECORDED HERE. Note the paralogue asymmetry in the database: GATAD2A carries a 'Nucleus speckle' annotation (which gene-gatad2a shows overstates its two cited papers) and no equivalent was found for GATAD2B in this beat's fetches. |
| subnuclear absence class | never-measured |
| localization shown for fusion | NO - nobody has looked in any KMT2A fusion. |
| shown for fusion | MLL-AF1p, MLL-AF9, MLL-CBP, MLL-EEN and MLL-ENL by AP-MS co-purification in Jurkat (five of seven baits, one call above p = 0.05, negative control 0|0); ABSENT from MLL-AF4 (= KMT2A::AFF1) and MLL-GAS7. Nothing functional in any KMT2A fusion. |
| epigenetic readout | ACETYLATION-SENSITIVE HISTONE-TAIL BINDING, shared with the paralogue: p66-beta binds the tails of all four octamer histones in vitro and acetylation of those tails interferes with binding [E2]. In vitro, non-haematopoietic, no fusion. No mark has been measured after GATAD2B perturbation in any KMT2A-rearranged cell. |
| epigenetic readout absence class | never-measured |
| absence class | measured-not-reported |
| absence class note | For GATAD2B's spectral count in the MLL-AF4 column: measured-not-reported, upgradeable to measured-not-analysed against PXD009338 - and this is the most informative single case for that re-search, because GATAD2B scored in five of the other six columns, so a genuine zero in MLL-AF4 would be striking rather than merely uninformative. For any functional GATAD2B experiment in a KMT2A-rearranged cell: never-measured. |
| relationship to existing entries | NEW. Written as a separate entry from gene-gatad2a rather than folded into it, per the one-entry-per-thing rule and because the two paralogues differ on every instrument in this beat and in opposite directions. gene-gatad2a carries the CR1/CR2 architecture, the ETO2/CBFA2T3 structural link and the speckle-annotation correction; this entry carries the five-of-seven interactome breadth and the human loss-of-function syndrome, neither of which applies to the alpha paralogue. protein-nurd-complex states the paralogue choice once. |
cell types and states
[ "Jurkat (human T-ALL) - AP-MS host, GATAD2B recovered in five of seven MLL-fusion bait columns, not MLL-AF4 [E3]", "unnamed human and mouse lines - p66 synergy, MBD2 dependence, in vitro histone-tail binding [E1][E2]", "human patients with 1q21.3 microdeletion or GATAD2B loss-of-function - developmental delay, hypotonia, intellectual disability, characteristic facies [E6]", "NO KMT2A-rearranged cell and no haematopoietic cell of any kind" ]
depmap
{
"kmt2a_selectivity_rank": 2141,
"of_genes": 17728,
"percentile": 12.077,
"difference": -0.0715,
"mean_kmt2a_rearranged": -0.1828,
"mean_other_leukemia": -0.1113,
"pan_essentiality_all_lines": -0.1452,
"release": "NOT RECORDED BY THE SCAN - generation date 2026-09-19 only (rule 83).",
"paralogue_note": "GATAD2A ranks 866 and GATAD2B 2,141, on the instrument where GATAD2B is the WEAKER of the pair - the opposite ordering from the AP-MS, where GATAD2B appears in five bait columns and GATAD2A in one. A single-gene knockout screen is the worst instrument for a synergistic paralogue pair, and no double-knockout arm exists."
}bounded literature
{
"queries_and_translations": [
"GATAD2B[tiab] AND MLL[tiab] -> 0 | \"GATAD2B\"[Title/Abstract] AND \"MLL\"[Title/Abstract]",
"GATAD2B[tiab] AND KMT2A[tiab] -> 1 | PMID 38114583 - REJECTED on title check, a 46-gene de novo variant list",
"GATAD2B[tiab] AND leukemia[tiab] -> 2 | PMIDs 39696784, 35685361 - counted, NOT read",
"p66beta[tiab] AND NuRD[tiab] -> 6"
],
"positive_control": "MEN1[tiab] AND MLL[tiab] -> 55",
"date": "2026-09-24",
"tool": "scripts/pubmed_safe.py",
"names_tried": [
"GATAD2B",
"p66beta"
],
"repeat_trap_note": "PMID 38114583 is the SECOND gene-list co-occurrence in this beat, after PMID 29158550 put SIN3A and KMT2A in one Exome Pool-Seq variant list. At counts of one or two records, a neurodevelopmental gene panel listing KMT2A alongside anything is the single commonest false positive, and every such hit must be title-checked."
}measurements
[
{
"quantity": "bait_columns_containing_the_protein",
"subject": "GATAD2B in the Skucha 2018 pan-MLL-fusion network",
"value": 5,
"units": "bait columns",
"n": 7,
"universe": "seven MLL-fusion baits in Jurkat, top ~291-293 preys retained per bait",
"null": null,
"note": "MLL-AF1p 0.0689 (3|2) - ABOVE 0.05, MLL-AF9 0.0266 (5|6), MLL-CBP 0.0217 (2|8), MLL-EEN 0.042 (6|7), MLL-ENL 0.00581 (9|8). Neg ctrl 0|0. Absent from MLL-AF4 and MLL-GAS7. Paralogue GATAD2A is in 1 column."
},
{
"quantity": "kmt2a_selectivity_rank",
"subject": "GATAD2B in the DepMap KMT2A-rearranged versus other-leukaemia scan",
"value": 2141,
"units": "rank",
"n": 17728,
"universe": "17,728 genes; 11 KMT2A-rearranged versus 43 other leukaemia lines, arms fixed from the first gene with data, no DepMap release recorded, no double-knockout arm for the paralogue pair",
"null": 8864.5,
"note": "percentile 12.1 against a 50th-percentile null; difference -0.0715 on arm means of -0.183 and -0.111. Paralogue GATAD2A ranks 866 - the reverse of the AP-MS ordering."
}
]- Heterozygous loss of GATAD2B causes a recognisable intellectual-disability syndrome in humans. This project treats INFANTS, in the developmental window, so a NuRD-scaffold-directed agent here is not covered by the usual 'constraint is not a safety signal' argument - the phenotype is observed, human, and in the right age group.
- GATAD2A and GATAD2B are synergistic and mutually dependent for repression, so a paralogue-selective drug would have an unpredictable net effect and a pan-p66 drug would remove a protein whose loss has a named syndrome. Neither option is clean.
- GATAD2B is the broadest NuRD signal in the pan-fusion interactome and is absent from the one fusion this project studies. If a future entry cites 'GATAD2B is a pan-fusion NuRD partner' as support for a KMT2A::AFF1 mechanism, it will have used the one bait column where the protein does not appear.
- Is the five-versus-one paralogue split across bait columns biology or detectability? A head-to-head spectral-count comparison of GATAD2A and GATAD2B peptides in the public PXD009338 raw files would settle it, and GATAD2B's absence from MLL-AF4 is the most informative single cell in that table - a genuine zero there, in a protein that scored in five of the other six, would be a real asymmetry.
- Does a GATAD2A/GATAD2B double knockout do anything in a KMT2A-rearranged line? Neither single-gene DepMap rank can mean much for a synergistic paralogue pair, and no double-knockout arm exists anywhere in this project's data.
- Does GATAD2B have a subnuclear address? Its paralogue carries a speckle annotation this beat found overstated, and nothing equivalent was found for GATAD2B - so the asymmetry may be in the literature rather than in the cell, and one two-colour image would say.
- What is HAX1 doing in the 1q21.3 microdeletion region? It is a severe congenital neutropenia gene sitting beside GATAD2B, which is an odd neighbourhood for a haematology project to leave unexamined, and no entry in this graph mentions it.
- E1strong confidence 4 of 5: Replicated in peer-reviewed work by more than one group.paper
hp66-alpha and hp66-beta are two highly related 66-kDa proteins identified in a yeast two-hybrid screen with MBD2b and are components of the MeCP1/Mi-2-NuRD complex; both bind directly to MBD2 and MBD3, both are potent transcriptional repressors, and hp66-alpha binds with higher affinity than hp66-beta and via two interaction domains where hp66-beta has one; both are ubiquitously coexpressed in cell lines and in fetal and adult tissues.
Two highly related p66 proteins comprise a new family of potent transcriptional repressors interacting with MBD2 and MBD3. · Brackertz M, Boeke J, Zhang R, Renkawitz R · Journal of Biological Chemistry 2002What this evidence cannot show
Read from the abstract; the cell lines behind the co-precipitation and confocal work are not named there. Note that hp66-beta was ALREADY KNOWN as a component of the human MeCP1 complex before this paper and is the better-established of the pair historically, which is the opposite of the impression the alpha-first naming gives. Nothing haematopoietic, nothing with a fusion protein.
Evidence that argues against this (1)
- 'Overlapping functions' and ubiquitous coexpression cut against any GATAD2B-specific claim in a cell that also expresses GATAD2A, and the companion paper shows the two act in SYNERGY - knocking down one impairs the other's repressive function. So neither paralogue can be perturbed cleanly, and a GATAD2B result in a cell expressing both is underdetermined. Brackertz M et al. 2006
- E2strong confidence 4 of 5: Replicated in peer-reviewed work by more than one group.paper
p66-alpha and p66-beta function in synergy - knocking down p66-alpha affects the repressive function of p66-beta and vice versa - both interact with MBD2 functionally and biochemically, both bind the tails of all octamer histones in vitro with acetylation of the tails interfering with binding, and wild-type subnuclear distribution of both depends on the presence of MBD2.
p66alpha and p66beta of the Mi-2/NuRD complex mediate MBD2 and histone interaction. · Brackertz M, Gong Z, Leers J, Renkawitz R · Nucleic Acids Research 2006What this evidence cannot show
Read from the abstract; cell lines human and mouse, not named, none haematopoietic. THE MUTANT AND CONSERVED-REGION EXPERIMENTS IN THIS PAPER ARE ON p66-ALPHA, not p66-beta - the single-amino-acid MBD2-binding mutant and the CR2 requirement are alpha results, and this entry does not transfer them to beta. UniProt records no subnuclear body for Q8WXI9, unlike Q86YP4, so even the punctate-distribution result is paralogue-asymmetric in the annotation.
Evidence that argues against this (1)
- If the two paralogues are synergistic and mutually dependent for repression, then their DIFFERENT behaviour across the seven Skucha bait columns - five for beta, one for alpha - is hard to read as a functional difference at all, and is more consistent with a detectability difference. That weakens the only observation that makes GATAD2B distinctive in this beat. This repository 2026
- E3moderate confidence 3 of 5: One strong study, or several smaller consistent ones.database
GATAD2B is present in the Skucha 2018 960-protein pan-MLL-fusion AP-MS network (UniProt Q8WXI9, negative control 0|0) in five of seven bait columns - MLL-AF1p p = 0.0689 (3|2), MLL-AF9 p = 0.0266 (5|6), MLL-CBP p = 0.0217 (2|8), MLL-EEN p = 0.042 (6|7), MLL-ENL p = 0.00581 (9|8) - making it the most broadly recovered NuRD scaffold subunit in the panel, and it is absent from the MLL-AF4 (= KMT2A::AFF1) and MLL-GAS7 columns.
NuRD sub-module split and Sin3/HDAC scoring over Skucha 2018 Supplementary Table 5 (PMID 29777171), run 2026-09-24: analyses/q193-nurd-module-split/run.py, per-member accessions, bait columns, p-values, spectral counts and negative-control counts persisted in results.json. · This repository (script and results.json committed) 2026What this evidence cannot show
THE COUNTS ARE SMALL AND ONE CALL IS ABOVE 0.05: MLL-AF1p at p = 0.0689 with 3|2 spectra is retained only because the paper kept the top ~300 preys per bait rather than applying a threshold, so 'five of seven' is four significant calls plus one. The accession Q8WXI9 matches UniProt's live reviewed row for gene_exact:GATAD2B AND organism_id:9606 (2026-09-24). Absence from MLL-AF4 is a rank truncation, not a zero - measured-not-reported, upgradeable to measured-not-analysed against PXD009338. Jurkat T-ALL host, doxycycline-induced Strep-HA bait, nuclear lysate, so this is co-purification and not a direct contact.
Evidence that argues against this (1)
- Five of seven bait columns at 2-9 spectra per replicate is broad but shallow, and on the same table the CCT/TRiC contaminant control reaches 8 of 9 subunits at p = 2.0e-10 including 8 of 8 in the MLL-AF4 column. Breadth across baits on this dataset is characteristic of non-specific nuclear background as much as of a shared partner - which is precisely what a PAN-fusion signal cannot distinguish. Skucha A et al. 2018
- E4weak confidence 2 of 5: Preliminary, small, indirect, or not yet peer reviewed.database
In this project's genome-wide DepMap KMT2A-selectivity scan GATAD2B ranks 2,141 of 17,728 (percentile 12.1) with a difference of -0.0715, mean CHRONOS -0.1828 in KMT2A-rearranged lines against -0.1113 in other leukaemia lines and a pan-essentiality of -0.1452 - behind its paralogue GATAD2A at rank 866, and with arm means close enough to zero that the effect is small in absolute terms.
DepMap KMT2A-selective dependency scan over 17,728 genes: analyses/depmap-kmt2a-selective-scan-full/results.json, generated 2026-09-19. GATAD2B rank 2,141, difference -0.0715, mean_kmt2a_r -0.1828, mean_other -0.1113, pan_essentiality -0.1452. · DepMap via this repository 2026What this evidence cannot show
Percentile 12.1 is quoted against an explicit 50th-percentile null. BOTH ARM MEANS ARE SMALL (-0.183 and -0.111), so this is a modest rank on a weak absolute effect and should not be read as a dependency. The standing scan caveats apply in full: both arms are fixed from the first gene with data, dropping 5 of 16 eligible KMT2A-rearranged lines from every comparison; a uniform n_all_lines of 1,208 is reported for every gene; and no DepMap release is recorded, only a generation date. Re-read from results.json on 2026-09-24. A paralogue pair is also the worst case for a single-gene knockout screen, because the surviving paralogue buffers the loss.
Evidence that argues against this (1)
- PARALOGUE BUFFERING MAKES BOTH GATAD2 RANKS UNINTERPRETABLE AS THEY STAND. The two proteins are ubiquitously coexpressed and functionally synergistic, so a single-gene CRISPR knockout of either is partially compensated by the other, and the scan contains no double-knockout arm. This graph has already seen paralogue pairs behave in ways a single-gene scan cannot explain - EP300 at rank 108 against CREBBP at 16,950 - and the honest reading here is that neither GATAD2 rank measures the module's importance. Brackertz M et al. 2002
- E5moderate confidence 3 of 5: One strong study, or several smaller consistent ones.database
A bounded literature search on 2026-09-24 returns zero records tying GATAD2B to MLL and one to KMT2A, which fails a title check - it is a neurodevelopmental cohort study in which GATAD2B and KMT2A are two of 46 genes in a de novo variant list with no relationship to each other - against a positive control of 55 for MEN1 with MLL; the retired name p66beta returns 6 with NuRD and the gene returns 2 with leukemia.
What this evidence cannot show
NAMES TRIED: GATAD2B, p66beta. NOT tried: KIAA1150 (a clone designation, low yield). One tag per term, querytranslation recorded, run through pubmed_safe so the zero is a real zero. THE SINGLE KMT2A HIT WAS TITLE-CHECKED AND REJECTED and is worth naming because it is the SECOND instance of the same trap in this beat: PMID 38114583 lists 46 genes with de novo variants in an Indian neurodevelopmental cohort, GATAD2B and KMT2A among them, exactly as PMID 29158550 put SIN3A and KMT2A in one Exome Pool-Seq gene list. A gene-list co-occurrence is not a relationship, and at these counts it is the majority of the hits. The two leukemia records were NOT read.
Evidence that argues against this (1)
- The bound is on reporting and understates measurement: GATAD2B was in the lysate of an AP-MS against seven MLL fusions and scored in five of them, and it is screened genome-wide in DepMap - neither measurement appears in any GATAD2B-plus-MLL literature record. A reader taking the zero as 'never tested against a KMT2A fusion' would be wrong. This repository 2026
- E6strong confidence 4 of 5: Replicated in peer-reviewed work by more than one group.paper
Loss-of-function mutations of GATAD2B cause a recognisable syndrome with intellectual disability; an infant with a de novo 1.06 Mb 1q21.3 deletion encompassing GATAD2B, TPM3 and HAX1 presented with feeding difficulty, developmental delay, hypotonia and dysmorphic features, and across eight patients with 1q21.3 microdeletion the authors conclude that loss of function of GATAD2B is the major contributor to the characteristic facies and intellectual disability.
1q21.3 deletion involving GATAD2B: An emerging recurrent microdeletion syndrome. · Tim-Aroon T, Jinawath N, Thammachote W, Sinpitak P, Limrungsikul A, Khongkhatithum C, Wattanasirichaigoon D · American Journal of Medical Genetics Part A 2017What this evidence cannot show
ONE INDEX PATIENT plus seven from literature review and clinical databases (DECIPHER, ISCA/ClinGen) - a case report with a review, not a cohort study. The deleted region in the index case encompasses 35 genes including TPM3 and HAX1, so attributing the facies and ID to GATAD2B is the authors' inference across the eight patients and not a measurement in this one. The statement that GATAD2B loss-of-function point mutations cause a recognisable ID syndrome is background cited by this paper, not established in it. Read from the abstract. HAX1 in the deleted region is itself a severe congenital neutropenia gene, which is relevant context for a haematology project and is not pursued here.
Evidence that argues against this (1)
- A germline heterozygous deletion present from conception is not a model of partial, reversible pharmacological inhibition in an adult or even an infant - the same objection this graph applies to gnomAD constraint scores. The developmental phenotype says GATAD2B matters for brain development; it does not by itself establish that a drug occupying the protein for weeks would be neurotoxic, and treating it as a hard safety verdict would be the constraint-statistic error in a different costume. Pande S et al. 2023
the p66-beta bridge paralogue; broadest NuRD scaffold in the pan-fusion interactome at five of seven baits
synergistic paralogues: knockdown of either impairs the other's repressive function; opposite orderings on AP-MS and DepMap
p66-beta binds directly to both MBD2 and MBD3, at lower affinity and through one interaction domain
MLL-AF9 column, p = 0.0266, 5|6 spectra, neg ctrl 0|0
MLL-ENL column, p = 0.00581, 9|8 spectra - its strongest bait
the most broadly recovered NuRD scaffold subunit, and still absent from the index fusion's column