MTA2 / MTA1L1 (O94776) - the NuRD scaffold that clamps HDAC1, recruited in T lymphocytes by BCL11B, and absent together with MTA1 and MTA3 from every column of the pan-MLL-fusion interactome

No three-dimensional model is available for this protein, so the map above is the whole picture. That is itself informative: AlphaFold does not model chains this long, and the parts that have been crystallised are held under the entries for the individual domains.

• Germline Mta2 loss in mouse causes T-cell HYPERproliferation with hyperinduction of IL-2, IL-4 and IFN-gamma, plus lupus-like autoimmunity. A drug removing MTA2 function in a child would be pushing lymphocytes in the direction of more division and autoimmunity, which is the opposite of the intended effect and a genuine toxicity signal rather than a constraint statistic.
• MTA2's function is to make HDAC1/2 processive on chromatin. Removing it does not silence a fusion programme; it disables a repressive machine, and de-repression in a leukaemia with a lineage-switch escape route documented in this exact fusion is the wrong direction.
• There is no MTA2 chemical matter and ZERO experimental structures. The ELM2-SANT clamp is known only from the paralogue, so even a structure-based campaign would start by solving the protein.
• The only reason MTA2 is in this graph at all is complex membership plus a bounded absence. If a future beat reads this entry as evidence that NuRD's deacetylase arm is at the index fusion, it will have committed exactly the inference this entry was written to prevent: complex membership does not transfer an interaction, and neither does paralogue homology.

1. Does MTA2 have zero spectra in the MLL-AF4 pulldown, or non-zero spectra that the top-300 filter discarded? PXD009338 is public and this is the single cheapest experiment in the beat. The answer decides whether the deacetylase module's absence is a fact about the biology or about the filter.
2. Is there a per-gene MTA2 score in the Schneider 2023 epigenome-wide CRISPR screen in SEM and ALL-PO? MBD3 was a hit from that library; MTA2 was almost certainly in it, and a supplementary table would say whether it scored.
3. Does BCL11B recruit MTA2-NuRD in a t(4;11) B-ALL cell? BCL11B is a T-lineage factor with a documented direct MTA2 contact, and KMT2A::AFF1 leukaemias include lineage-ambiguous and switching cases, so the recruitment mechanism may be present in exactly the cases that matter most.
4. Do MTA1, MTA2 and MTA3 differ in KMT2A-rearranged lines? Only MTA2's rank was extracted here. A paralogue split of the kind this graph has found for EP300/CREBBP and GATAD2A/GATAD2B would be a new lead and is a single lookup.
5. Is MTA2 anywhere in a leukaemic nucleus that matters? No subnuclear assignment exists in any cell type, so even the resting address of the protein that clamps HDAC1 onto nucleosomes is unmeasured.

1. E1established confidence 5 of 5: Multiple independent lines of evidence; treated as settled in the field.paper

   MTA2, a novel polypeptide highly related to metastasis-associated protein 1, is a subunit of the NuRD complex; it modulates the enzymatic activity of the HDAC1/HDAC2-RbAp48/RbAp46 histone deacetylase core complex, whose activity in isolation is severely compromised, and MBD3 mediates MTA2's association with that core.

   Analysis of the NuRD subunits reveals a histone deacetylase core complex and a connection with DNA methylation. · Zhang Y, Ng HH, Erdjument-Bromage H, Tempst P, Bird A, Reinberg D · Genes & Development 1999
   PMID 10444591DOI
   ▸What this evidence cannot show

   1999 biochemistry from HeLa nuclear extract, read from the abstract. 'Modulates the enzymatic activity' is stated without a direction or magnitude in the abstract and neither was verified here. Nothing in this paper concerns leukaemia, MLL or any fusion protein - the expected_absent_terms record that deliberately.

   ▸Evidence that argues against this (1)

   • The same purification lineage later showed that NuRD assembles from PREFORMED, INDEPENDENTLY ACTIVE sub-modules, with the deacetylase module functional on its own and CHD4 associating transiently. On that picture MTA2's role is to build a competent deacetylase module rather than to complete a single holo-complex, and an experiment that finds the remodelling arm without the MTA arm - which is what the MLL-fusion interactome does - is finding a normal cellular species rather than a truncated complex. Zhang W et al. 2016
2. E2established confidence 5 of 5: Multiple independent lines of evidence; treated as settled in the field.paper

   In the crystal structure of HDAC1 in complex with MTA1 from the NuRD complex, the ELM2-SANT domains of MTA1 wrap completely around HDAC1 and occupy both sides of the active site, with the adjacent BAH domain ideally positioned to recruit nucleosomes to the enzyme's active site, and inositol-tetraphosphate is a bona fide conserved regulator of class I HDAC complexes rather than a passive structural cofactor.

   Class I HDACs share a common mechanism of regulation by inositol phosphates. · Millard CJ, Watson PJ, Celardo I, Gordiyenko Y, Cowley SM, Robinson CV, Fairall L, Schwabe JWR · Molecular Cell 2013
   PMID 23791785DOI
   ▸What this evidence cannot show

   THIS STRUCTURE IS MTA1, NOT MTA2 - the expected_absent_terms say so explicitly. MTA2 has ZERO PDB entries (UniProt O94776 cross-references, fetched 2026-09-24), so every architectural statement about MTA2 in this entry is a PARALOGUE-HOMOLOGY INFERENCE from an MTA1 structure plus the fact that MTA2 carries BAH, ELM2 and SANT domains at comparable positions in its own UniProt feature table. The paralogues diverge in their C-terminal halves and in their transcription-factor partners, which is where such an inference would fail. Read from the abstract; the construct boundaries and resolution were not verified.

   ▸Evidence that argues against this (1)

   • A structure existing is not a structure suitable for design, and this one is a corepressor-fragment complex: the ELM2-SANT construct is a domain pair from a 715-residue protein, and nothing here says what a full MTA scaffold does on a nucleosome. More to the point for this entry, MTA1 and MTA2 are the paralogue pair, and this graph's own experience with paralogues - EP300 at rank 108 against CREBBP at 16,950, GATAD2A at 866 against GATAD2B at 2,141 - is that paralogues routinely behave differently despite identical domain architecture. Millard CJ et al. 2013
3. E3strong confidence 4 of 5: Replicated in peer-reviewed work by more than one group.paper

   Endogenous BCL11B associates with the NuRD complex in T lymphocytes; the BCL11B complexes possess trichostatin-A-sensitive histone deacetylase activity; BCL11B interacts directly with the metastasis-associated proteins MTA1 and MTA2 through its amino-terminal region; and MTA1 is functionally required for BCL11B-mediated repression - overexpression enhances it, siRNA knockdown inhibits it, and MTA1 is specifically recruited to a BCL11B target promoter.

   BCL11B functionally associates with the NuRD complex in T lymphocytes to repress targeted promoter. · Cismasiu VB, Adamo K, Gecewicz J, Duque J, Lin Q, Avram D · Oncogene 2005
   PMID 16091750DOI
   ▸What this evidence cannot show

   THE FUNCTIONAL ARM IS MTA1, NOT MTA2. MTA2's role here is a direct binding result (amino-terminal region of BCL11B); the overexpression, siRNA and promoter-recruitment experiments are all MTA1. So this establishes that MTA2 is a NuRD subunit a lymphoid transcription factor binds, and does NOT establish that MTA2 is required for the repression. T-LINEAGE, and BCL11B disruption is linked to T-ALL, not to B-lineage KMT2A::AFF1 disease. Read from the abstract. Nothing here involves MLL or any fusion.

   ▸Evidence that argues against this (1)

   • THIS PAPER'S SYSTEM IS AN UNPLANNED CONTROL THAT CUTS AGAINST THE INTERACTOME DATA, and it is the reason this item is in the entry. The Skucha AP-MS was run in JURKAT, a T-ALL line - exactly the cell type in which MTA2-containing NuRD is documented here to be recruited by a lymphoid transcription factor - and MTA2 is absent from all seven bait columns. Either MTA2-NuRD exists in that nucleus and is not near any MLL fusion, or the bait-specificity filter removed an abundant, promiscuously recruited corepressor scaffold. The data do not choose. This repository 2026
4. E4strong confidence 4 of 5: Replicated in peer-reviewed work by more than one group.paper

   Mta2-null mice show partial embryonic lethality, and survivors develop lupus-like autoimmunity with skin lesions, bodyweight loss, glomerulonephritis, liver inflammation and autoantibodies; transplantation of Mta2-null bone marrow recapitulates skin lesions and bodyweight loss in recipients; Mta2-null T lymphocytes develop normally but hyperproliferate on stimulation with hyperinduction of IL-2, IL-4 and IFN-gamma, T-cell-specific knockout reproduces the hyperproliferation but not the other autoimmune features, and IL-4 is a direct target gene of Mta2.

   Inactivation of NuRD component Mta2 causes abnormal T cell activation and lupus-like autoimmune disease in mice. · Lu X, Kovalev GI, Chang H, Kallin E, Knudsen G, Xia L, Mishra N, Ruiz P, Li E, Su L, Zhang Y · Journal of Biological Chemistry 2008
   PMID 18353770DOI
   ▸What this evidence cannot show

   MOUSE, GERMLINE, and constitutive - the developmental phenotype of a complete knockout from conception is not a model of pharmacological inhibition in an adult or an infant, and this graph's standing rule on constraint measures applies: it says the gene matters in development, not that partial reversible inhibition would be unsafe. No leukaemia phenotype is reported and none was looked for. The mouse gene symbol is Mta2; no residue numbers are taken from this paper, so no murine-versus-human coordinate enters this entry. Read from the abstract.

   ▸Evidence that argues against this (1)

   • The T-cell-specific knockout reproduces ONLY the hyperproliferation and none of the autoimmune disease, and the paper's own conclusion is that gene expression in NON-T cells drives the autoimmunity. So the tissue in which Mta2 loss matters most is not the lymphoid compartment, which weakens any argument from this mouse to a lymphoid leukaemia - and the direction of the T-cell phenotype is hyperproliferation, i.e. loss of Mta2 makes lymphocytes divide MORE, which is the opposite of what an anti-leukaemic target should do. Lu X et al. 2008
5. 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 for MTA2 with KMT2A and ZERO for MTA2 with MLL in title or abstract, against a positive control of 55 for MEN1 with MLL; MTA2 with leukemia returns 2, both rejected on title check; and the retired symbol MTA1L1 returns 4 records, none leukaemic.

   NCBI esearch via scripts/pubmed_safe.py on 2026-09-24. MTA2[tiab] AND KMT2A[tiab] -> 0 (translation "MTA2"[Title/Abstract] AND "KMT2A"[Title/Abstract]). MTA2[tiab] AND MLL[tiab] -> 0. MTA2[tiab] AND leukemia[tiab] -> 2 (PMIDs 23466416, 16091750). MTA1L1[tiab] -> 4 (24055374, 16309230, 12684393, 9929979). MTA2[tiab] AND NuRD[tiab] -> 54. Positive control MEN1[tiab] AND MLL[tiab] -> 55. · NCBI PubMed E-utilities 2026
   ▸What this evidence cannot show

   NAMES TRIED, which a reader needs in order to judge the absence: MTA2, MTA1L1. NOT tried as bare tokens: 'PID' (a three-letter symbol that matches pelvic inflammatory disease and hundreds of unrelated uses - excluded on precision grounds, per this corpus's own alias-precision rule) and 'MTA1-L1' (hyphenation variant; PubMed strips punctuation into phrases so it is covered by MTA1L1). One tag per term throughout (rule 94) with the querytranslation recorded, run through pubmed_safe so an empty translation raises rather than returning a fake zero - the two zeros here are real zeros. The 54 records for MTA2 with NuRD show the instrument reaches this protein perfectly well, so the zeros are about the disease link and not about the query.

   ▸Evidence that argues against this (1)

   • A ZERO IN PUBMED IS A BOUND ON REPORTING AND IS DEMONSTRABLY NOT A BOUND ON MEASUREMENT HERE. MTA2 has been screened genome-wide in DepMap's panel with a rank of 1,037, and it was in the lysate of an AP-MS run against seven MLL fusions including the index one. Neither measurement appears in any MTA2-plus-MLL literature record. A reader who reads these zeros as 'never tested' would be wrong twice. DepMap via this repository 2026
6. E6moderate confidence 3 of 5: One strong study, or several smaller consistent ones.database

   MTA1, MTA2 and MTA3 are all absent from the Skucha 2018 960-protein pan-MLL-fusion AP-MS network in all seven bait columns, as are HDAC1, HDAC2, RBBP4 and RBBP7 - the entire seven-member NuRD deacetylase module scores 0 of 7, while the remodelling module scores 5 of 7 (Fisher two-sided p = 0.0210).

   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, with load-time assertions on sheet layout and a 960-row count, three backgrounds, and a background-free Fisher test on the module 2x2. · This repository (script and results.json committed) 2026
   ▸What this evidence cannot show

   ABSENCE HERE IS NOT A ZERO SPECTRAL COUNT. The paper retains the ~300 most significant preys per bait (291 for MLL-AF4), so MTA2's per-bait counts were generated and never tabulated - measured-not-reported, upgradeable to measured-not-analysed against the public PXD009338 raw files. The subunit lists are my curation, written out in the script. The Jurkat host and the doxycycline-induced Strep-HA bait apply as they do to every claim from this dataset.

   ▸Evidence that argues against this (1)

   • THE ABSENT ARM IS THE ABUNDANT, PROMISCUOUS ARM, and that is a competing explanation I cannot exclude. HDAC1/2 sit in at least five corepressor complexes and RBBP4/7 in four, so a bait-specificity model applied to a top-300 cut removes them first whether or not the biology is there. MTA2 inherits that: an abundant scaffold recruited by many transcription factors behaves like background in exactly this kind of filter. Bantscheff M et al. 2011
7. E7weak confidence 2 of 5: Preliminary, small, indirect, or not yet peer reviewed.database

   In this project's genome-wide DepMap KMT2A-selectivity scan MTA2 ranks 1,037 of 17,728 (percentile 5.85) with a difference of -0.1047, mean CHRONOS -0.3976 in KMT2A-rearranged lines against -0.2929 in other leukaemia lines and a pan-essentiality of -0.2103; this corpus's earlier interactome enumeration rejected MTA2 on exactly this number, calling it 'zero literature and rank 1,037 inside noise'.

   DepMap KMT2A-selective dependency scan over 17,728 genes: analyses/depmap-kmt2a-selective-scan-full/results.json, generated 2026-09-19. MTA2 rank 1,037, difference -0.1047, mean_kmt2a_r -0.3976, mean_other -0.2929, pan_essentiality -0.2103. · DepMap via this repository 2026
   ▸What this evidence cannot show

   Re-read from results.json on 2026-09-24 rather than quoted from ana-interactome4-enumeration, which carries the same figure. Percentile 5.85 is stated against an explicit 50th-percentile null. THREE STANDING CAVEATS APPLY IN FULL: the scan fixes both comparison arms from the first gene with data and therefore drops 5 of 16 eligible KMT2A-rearranged lines from every comparison; it reports a uniform n_all_lines of 1,208 for every gene when DepMap screens each gene in a gene-specific set; and it records no DepMap RELEASE, only a generation date, so per rule 83 drift cannot be distinguished. Roughly half the effect (-0.210 of the -0.398 KMT2A-r mean) is lineage-independent fitness.

   ▸Evidence that argues against this (1)

   • THE CORPUS'S OWN JUDGEMENT ON THIS EXACT NUMBER IS THAT IT IS NOISE, and I have not found a reason to overturn it. Note also the internal inconsistency it creates: GATAD2A ranks 866 - BETTER than MTA2 - and was never scored in that round at all, while CHD4, HDAC1, HDAC2, SIN3A, SMARCA4 and ARID1A were rejected as anti-dependencies. If 1,037 is inside noise then so is 866, and the honest position is that no NuRD subunit has a DepMap dependency shape that would carry a therapeutic hypothesis. This repository 2026
8. E8moderate confidence 3 of 5: One strong study, or several smaller consistent ones.paper

   MTA2 depletion by shRNA reduces proliferation, migration and invasion and attenuates xenograft growth and lung metastasis in human breast cancer (MDA-MB-231) and gastric cancer (SGC-7901, AGS) models, with MTA2 expressed in 55.9% (71 of 127) of gastric cancer samples and its promoter bound and activated by Sp1 between -1043 and -843 bp.

   MTA2 promotes gastric cancer cells invasion and is transcriptionally regulated by Sp1. · Zhou C, Ji J, Cai Q, Shi M, Chen X, Yu Y, Liu B, Zhu Z, Zhang J · Molecular Cancer 2013
   PMID 24010737DOI
   ▸What this evidence cannot show

   SOLID EPITHELIAL TUMOURS, and cited here ONLY to establish that MTA2 is a perturbable, phenotype-producing gene in cancer cells - nothing about leukaemia, haematopoiesis or any fusion. The promoter coordinates -1043 to -843 bp are relative to the human MTA2 transcription start site as defined by that paper and are NOT genome coordinates in any assembly; they are recorded here as the paper states them and are not used for any calculation. Read from the abstract. A metastasis-and-invasion phenotype is of limited relevance to a leukaemia.

   ▸Evidence that argues against this (1)

   • The metastasis literature on MTA2 runs the opposite way from the Mta2 knockout mouse, in which loss of the gene causes T-cell HYPERproliferation - so 'MTA2 loss reduces proliferation' is a property of particular epithelial lines and not of the gene. Since the target tissue here is lymphoid, the mouse result is the more relevant of the two and it points the wrong way for a therapeutic hypothesis. Lu X et al. 2008

• 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

  MTA2 modulates the HDAC1/2 core's activity; the paralogue MTA1's ELM2-SANT wraps HDAC1 on both sides of the active site (homology inference for MTA2)

• 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

  MBD3 mediates the association of MTA2 with the HDAC1/2-RBBP4/7 core

• part of →RBBP4 - the shared histone-binding subunit that every leukaemia line needs, and whose identical-length paralogue cannot replace it

  MTA2 is the NuRD-specific scaffold on the shared HDAC1/2-RBBP4/RBBP7 core

• ← 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

  MTA1, MTA2 and MTA3 all absent from all seven bait columns; the deacetylase module is 0 of 7

• related to →Interactome round 4: what was counted, what was written, what was rejected, and the two alias traps this beat fell into and caught

  this entry confirms rather than overturns that round's rejection of MTA2 (0 records, rank 1,037 inside noise)

• 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

  NuRD-specific scaffold of the deacetylase module

• ← 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 binding experiment; absent from all seven bait columns

• ← testsPROPOSED: a side-by-side CHD4 co-IP in a t(4;11) B-ALL line and a KMT2A-wildtype B-ALL line, probing one subunit from each NuRD module - the first experiment that would ask whether a KMT2A fusion sees the whole complex

  MTA2 is the deacetylase-module probe