GATAD2A / p66-alpha (Q86YP4) - the subunit that bridges NuRD's two modules, the best-ranked NuRD gene in this project's KMT2A-selectivity scan, and present in the pan-fusion interactome under MLL-AF9 but NOT the index fusion

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

• The one available chemical tool against this subunit - the p66alpha-MBD2 coiled-coil disrupting peptide - was designed to DE-REPRESS a silenced locus and it works: it elevates gamma-globin. A de-repressing agent in a leukaemia whose problem is an aberrantly ACTIVE transcriptional programme is pointing the wrong way, and the same peptide could plausibly open loci the fusion wants open.
• GATAD2A and GATAD2B are ubiquitously coexpressed and function in synergy, with knockdown of one affecting the other's repressive function. A GATAD2A-selective agent would therefore have an unpredictable net effect, and a pan-p66 agent would remove a protein present in every cell.
• GATAD2A's chromatin binding is blocked by histone-tail acetylation. Anything that raises global acetylation - including an HDAC inhibitor given for a NuRD hypothesis - would EVICT this subunit from chromatin, so two interventions aimed at the same complex would oppose each other.
• The evidence placing GATAD2A near a KMT2A fusion is a single AP-MS column at p = 0.036 under a fusion this project does not study, on a table whose strongest complex signal is a known contaminant. A therapeutic hypothesis built on that row is built on one marginal call.

1. Are GATAD2A's nuclear puncta actually nuclear speckles? A two-colour image against SRSF2/SC35 in any cell would settle a database annotation that two papers do not support, and if they are speckles it puts GATAD2A at the same address as AFF1 and AFF4 - the index fusion's own C-terminal partner and its paralogue.
2. Does the ETO2 NHR4-GATAD2A disrupting peptide do anything in a KMT2A-rearranged cell? It was built for erythroid differentiation, it is already a molecule, and it targets the only NuRD scaffold contact in this beat with a structure - but nobody has put it in a t(4;11) line.
3. Why GATAD2B in five bait columns and GATAD2A in one? A spectral-count comparison in the public PXD009338 raw files would say whether the paralogue asymmetry is biology or detectability, and the answer changes which paralogue a future experiment should target.
4. Is GATAD2A's MLL-AF9-only appearance a lineage effect? MLL-AF9 is the myeloid-associated fusion in the panel and GATAD2A's one documented functional axis is erythroid; whether a myeloid KMT2A fusion recruits the NuRD bridge and a B-lineage one does not is testable by co-IP in matched lines.
5. Does GATAD2A occupy KMT2A::AFF1 target genes? No occupancy experiment on GATAD2A exists in any leukaemia, so the most basic question - is it even at the loci - is unanswered.

1. E1strong confidence 4 of 5: Replicated in peer-reviewed work by more than one group.paper

   hp66-alpha (GATAD2A) and hp66-beta (GATAD2B) are two highly related 66-kDa proteins identified in a yeast two-hybrid screen with MBD2b, components of the MeCP1/Mi-2-NuRD complex; hp66-alpha is the human orthologue of Xenopus p66, is a potent transcriptional repressor reducing gene activity about 100-fold, binds directly to both MBD2 and MBD3, binds with higher affinity than hp66-beta using two interaction domains rather than one, co-precipitates with MBD2 in vivo, and colocalises with hp66-beta and MBD2 in a speckled nuclear pattern by confocal microscopy.

   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 2002
   PMID 12183469DOI
   ▸What this evidence cannot show

   Read from the abstract. The cell line behind the confocal images is not named in the abstract and was not verified; nothing here is haematopoietic and nothing involves a fusion protein. THE 'SPECKLED PATTERN' PHRASE IS A MORPHOLOGICAL DESCRIPTION - the expected_absent_terms above are deliberate: the paper does NOT report a speckle-marker colocalisation, and UniProt's 'Nucleus speckle' annotation for Q86YP4 rests on this paper and PMID 16415179. The 100-fold repression figure is a transfected reporter, not an endogenous gene.

   ▸Evidence that argues against this (1)

   • The paralogue finding cuts against treating GATAD2A as the NuRD p66. The two proteins are 'ubiquitously coexpressed' with 'overlapping functions', and the follow-up work from the same group shows they function in SYNERGY - knocking down p66-alpha affects p66-beta's repressive function and vice versa. So a GATAD2A-specific claim in a cell that also expresses GATAD2B is underdetermined, and in the Skucha network it is GATAD2B, not GATAD2A, that appears in five of seven baits. Brackertz M et al. 2006
2. E2strong confidence 4 of 5: Replicated in peer-reviewed work by more than one group.paper

   p66-alpha and p66-beta function in synergy and both interact with MBD2 functionally and biochemically; a single-amino-acid mutation in p66-alpha abolishes in vivo MBD2 binding, interferes with MBD2-mediated repression and converts p66-alpha's speckled nuclear distribution to a diffuse one, and wild-type subnuclear distribution of both paralogues depends on the presence of MBD2; both proteins bind the tails of all octamer histones in vitro, ACETYLATION of the histone tails interferes with p66 binding, and conserved region 2 of p66-alpha is required both for histone-tail interaction and for wild-type subnuclear distribution - supporting a two-interaction forward-feedback model in which stable chromatin association occurs only after histone deacetylation.

   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 2006
   PMID 16415179DOI
   ▸What this evidence cannot show

   Read from the abstract; cell lines are human and mouse but not named there, and none is haematopoietic. The histone-tail binding and acetylation-sensitivity are IN VITRO with recombinant tails, so 'stable chromatin association follows deacetylation' is a model the authors propose, not a measurement in chromatin. The residue identity of the single-amino-acid mutant is not in the abstract and is NOT quoted here - this entry deliberately states no mutant residue number, because it could not be confirmed against the sequence. The subnuclear claim is punctate-versus-diffuse morphology, not a named compartment.

   ▸Evidence that argues against this (1)

   • An acetylation-sensitive histone-tail interaction is exactly what a bulk chromatin binding assay would report for any basic, disordered, abundant nuclear protein, and the paper's own model requires deacetylation to precede stable binding - which makes GATAD2A's chromatin association DOWNSTREAM of the deacetylase module that this project's interactome analysis finds entirely absent from the MLL-fusion network. On that reading, a fusion-associated GATAD2A with no HDAC1/2 nearby would not be stably chromatin-bound at all, and the interactome signal would be a soluble-pool interaction. This repository 2026
3. E3strong confidence 4 of 5: Replicated in peer-reviewed work by more than one group.paper

   ETO2 (CBFA2T3) helps recruit the NuRD complex to the globin locus through its nervy homology region 4 (NHR4) domain binding multiple polyproline-leucine motifs within GATAD2A; oligomerisation of ETO2's NHR3 enhances affinity for peptides containing at least two such motifs, replacing the native GATAD2A motifs with a higher-affinity sequence from NCoR yields a peptide that disrupts the interaction between ETO2 and its target proteins, and enforced expression of that peptide elevates gamma-globin expression and induces differentiation of HUDEP-2 and K562 cells.

   The role of multivalency in the association of the eight twenty-one protein 2 (ETO2) with the nucleosome remodeling and deacetylase (NuRD) complex. · Dan-Dukor G, Shang S, Leighton GO, Travis CR, Schwochert TD, Agrawal P, Ajasa O, Li T, Waters ML, Ginder GD, Williams DC · Nucleic Acids Research 2025
   PMID 40421803DOI
   ▸What this evidence cannot show

   ERYTHROID, not leukaemic: the functional readouts are gamma-globin induction and differentiation in HUDEP-2 and K562. The paper's own framing that ETO2 'is a potential therapeutic target for acute leukemia' is a motivation sentence, not a result in this paper. It is ETO2, the CBFA2T3 gene product - the 5' partner of CBFA2T3::GLIS2 - and NOT the fusion protein, so nothing here shows a fusion binding GATAD2A. The expected_absent_terms are deliberate: neither KMT2A nor MLL appears, and this item must never be cited for anything about the index fusion.

   ▸Evidence that argues against this (1)

   • Whether this GATAD2A contact survives in CBFA2T3::GLIS2 is unresolved in this graph's own entry for that fusion, which notes that retention of NHR4/MYND DEPENDS ON THE EXACT JUNCTION - so the NuRD-recruiting motif-binding domain may or may not be present in the oncoprotein. The same entry also cautions that PDB 9DE2 is a single covalently tethered chain, not two molecules in complex, which limits what the structure establishes about the affinity of a genuine bimolecular interaction. RCSB PDB 2024
4. E4moderate confidence 3 of 5: One strong study, or several smaller consistent ones.database

   GATAD2A is present in the Skucha 2018 960-protein pan-MLL-fusion AP-MS network (UniProt Q86YP4, negative control 0|0) in exactly one of seven bait columns - MLL-AF9, p = 0.0361 with 10|12 spectral counts - and is absent from the MLL-AF4 (= KMT2A::AFF1) column; its paralogue GATAD2B is present in five bait columns and also absent from MLL-AF4.

   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) 2026
   ▸What this evidence cannot show

   The accession in the table, Q86YP4, matches the accession returned live by UniProt for gene_exact:GATAD2A AND organism_id:9606 AND reviewed:true on 2026-09-24, so this is the right protein. ABSENCE FROM THE MLL-AF4 COLUMN IS NOT A ZERO: the table keeps the ~291 most significant preys for that bait, so GATAD2A's MLL-AF4 spectral count was generated and never tabulated (measured-not-reported; raw files public at PXD009338). Host is JURKAT T-ALL with a doxycycline-induced Strep-HA bait, not a B-lineage or infant proteome.

   ▸Evidence that argues against this (1)

   • A single bait column at p = 0.0361 is inside the weak tail of this table - the paper retained the top 300 interactions per bait rather than applying a p threshold, and 26.6% of the tabulated per-bait p-values exceed 0.05 - so a one-of-seven call at p = 0.036 is a marginal entry on a rank-truncated list, not a demonstrated interaction. The corpus's own distinctiveness analysis also found the MLL-AF4 column reaches deepest into the weak tail, which cuts the other way and means an MLL-AF4 absence is a stronger absence than another bait's would be. Skucha A et al. 2018
5. E5weak confidence 2 of 5: Preliminary, small, indirect, or not yet peer reviewed.database

   In this project's genome-wide DepMap KMT2A-selectivity scan GATAD2A ranks 866 of 17,728 (percentile 4.9) with a difference of -0.1144, mean CHRONOS of -0.4062 in KMT2A-rearranged lines against -0.2918 in other leukaemia lines and a pan-essentiality of -0.3811 - the best rank of any NuRD subunit in the scan, ahead of MTA2 at 1,037 and GATAD2B at 2,141, and far ahead of CHD4 at 13,245.

   DepMap KMT2A-selective dependency scan over 17,728 genes: analyses/depmap-kmt2a-selective-scan-full/results.json, generated 2026-09-19. GATAD2A rank 866, difference -0.1144, mean_kmt2a_r -0.4062, mean_other -0.2918, pan_essentiality -0.3811. · DepMap via this repository 2026
   ▸What this evidence cannot show

   A RANK IS NOT A DEPENDENCY, and three things cap this one. (1) Most of the signal is general fitness: pan-essentiality -0.381 against a KMT2A-r-versus-other difference of only -0.114, so roughly three quarters of the effect is present in every lineage. (2) The scan FIXES BOTH ARMS from the first gene with data and therefore drops 5 of 16 eligible KMT2A-rearranged lines from every comparison. (3) It reports a uniform n_all_lines of 1,208 for every gene and records NO DepMap RELEASE, only a generation date, so per rule 83 drift cannot be told from gene-specific screen coverage. Figures re-read from results.json on 2026-09-24, not quoted from an earlier entry. Percentile 4.9 is quoted against an explicit 50th-percentile null.

   ▸Evidence that argues against this (1)

   • The corpus's own interactome enumeration REJECTED the neighbouring NuRD subunits on precisely this kind of number - MTA2 at rank 1,037 was called 'inside noise', and HDAC1, HDAC2, SIN3A, CHD4 and SMARCA4 were all rejected as anti-dependencies. GATAD2A was never scored in that round at all. If rank 1,037 is noise then rank 866 is very nearly noise too, and the honest reading is that no NuRD subunit has a dependency shape in DepMap that would justify a therapeutic hypothesis. This repository 2026
6. E6moderate confidence 3 of 5: One strong study, or several smaller consistent ones.database

   A bounded literature search on 2026-09-24 finds no record at all linking GATAD2A to a KMT2A fusion: GATAD2A with MLL returns 0, GATAD2A with KMT2A returns 0, and GATAD2A with leukemia returns 2 (the ETO2-NuRD paper and an AML prognostic RT-qPCR study); the retired name returns 0 with leukemia and 13 with NuRD, against a positive control of 55 for MEN1 with MLL.

   NCBI esearch via scripts/pubmed_safe.py on 2026-09-24. GATAD2A[tiab] AND MLL[tiab] -> 0 (translation "GATAD2A"[Title/Abstract] AND "MLL"[Title/Abstract]). GATAD2A[tiab] AND leukemia[tiab] -> 2 (PMIDs 40421803, 35681732). p66alpha[tiab] AND leukemia[tiab] -> 0. p66alpha[tiab] AND NuRD[tiab] -> 13. Positive control MEN1[tiab] AND MLL[tiab] -> 55. · NCBI PubMed E-utilities 2026
   ▸What this evidence cannot show

   One tag per term throughout (rule 94), with the querytranslation recorded beside every count, and run through pubmed_safe so an empty translation raises rather than returning a false zero. BOTH THE CURRENT AND THE RETIRED SYMBOL WERE RUN, because at the complex level in this beat the two gave disjoint sets: NuRD[tiab] AND KMT2A[tiab] returns 0 while NuRD[tiab] AND MLL[tiab] returns 6. The two leukaemia records were title-checked: PMID 40421803 is the ETO2-NuRD paper in E3, PMID 35681732 is an AML refractory-versus-responsive RT-qPCR marker panel whose abstract, through the MCP PubMed tool, has its italicised gene symbols deleted (the raw PubMed record keeps them: GATAD2A, HDAC2, TET1 and others sit inside <i>, efetch XML 2026-10-03), and which I did not read past the abstract.

   ▸Evidence that argues against this (1)

   • This bound is on REPORTING and is silent on measurement, and here the gap between the two is large and demonstrable: GATAD2A HAS been assayed against seven MLL fusions by AP-MS and scored under one of them, and it is screened in DepMap's whole panel - neither measurement appears in any GATAD2A-plus-MLL literature record. A reader who takes the 0 as 'never tested against a KMT2A fusion' would be wrong. This repository 2026

• binds →CHD4 (Q14839) - the NuRD ATPase, a pan-MLL-fusion co-purifier, and the gene whose disruption drives lineage switching in KMT2A::AFF1 relapse

  CR2 (340-480, human Q86YP4) interacts with CHD4; GATAD2A is the bridge between NuRD's two modules

• binds →MBD3 (O95983) - the methyl-CpG-binding subunit that does NOT bind methylated DNA, and a validated KMT2A::AFF1-selective CRISPR vulnerability that is absent from every column of the pan-fusion interactome

  p66-alpha binds directly to both MBD2 and MBD3 (yeast two-hybrid and co-precipitation)

• analogous to →CBFA2T3::GLIS2, inv(16)(p13.3q24.3)

  ETO2/CBFA2T3 NHR4 binds polyproline-leucine motifs in GATAD2A (PDB 9DE2); the NuRD contact in a different fusion family

• binds →KMT2A::MLLT3 (MLL-AF9), t(9;11)(p21.3;q23.3)

  present in the MLL-AF9 column of the Skucha AP-MS network, p = 0.0361, 10|12 spectra, neg ctrl 0|0

• binds →HDAC1 - a class I eraser that has no dependency claim of its own in this disease, because removing it alone does nothing until HDAC2 goes too

  CR1 (144-178, human Q86YP4) interacts with HDAC1, HDAC2, MBD2 and MTA2

• ← supportsThe MLL-fusion interactome recovers corepressor SCAFFOLDS and not their HDAC1/2 catalytic core: NuRD splits 5/7 remodelling arm versus 0/7 deacetylase arm (Fisher p = 0.021), and HDAC1, HDAC2, RBBP4 and RBBP7 are absent from all seven baits

  GATAD2A is the one NuRD scaffold subunit recovered, and only under MLL-AF9

• part of →The NuRD complex - two separable modules, three paralogue choices that change what it does, and a graph that carried its ATPase and its shared core but never the scaffolds that make it NuRD

  the bridge between NuRD's two modules

• ← supportsNo experiment has ever ASKED whether MTA2, MBD3, GATAD2A or SIN3A binds a KMT2A fusion - four of them were nonetheless in the lysate of a seven-fusion pulldown, and the class of that absence is measured-not-reported, not never-measured

  no targeted experiment under either name; scored under MLL-AF9 only

• ← co-occurs withGATAD2B / 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

  synergistic paralogues: knockdown of either impairs the other's repressive function; opposite orderings on AP-MS and DepMap

• ← related toNo histone-mark readout has ever been measured after perturbing ANY NuRD or Sin3 subunit in a KMT2A-rearranged cell - the two complexes whose entire job is acetylation have zero acetylation data in this disease

  the nearest mark result - acetylation blocks p66-alpha histone-tail binding - is in vitro and non-leukaemic