RAD21 - the cohesin subunit with the most break-prone sites in KMT2A-rearranged patients, and the degron experiment this graph keeps asking for

{
  "STAG2": "MDS-stage acquisition, secondary AML, adverse prognosis",
  "RAD21_SMC1A_SMC3": "de novo AML features, better prognosis",
  "pmid": "39033241",
  "title": "Subunit-specific analysis of cohesin-mutant myeloid malignancies reveals distinct ontogeny and outcomes.",
  "authors": "Jann JC, Hergott CB, Winkler M, et al.",
  "venue": "Leukemia",
  "year": 2024,
  "title_provenance": "Title/authors/venue backfilled from PubMed esummary on 2026-09-20 because this citation stored a bare PMID with no title, which made it unscoreable by verify-citations.py. BACKFILLING IS NOT VERIFICATION: it confirms the PMID resolves to a real paper, not that this is the right paper for the claim. That still requires reading."
}

{
  "chronos_mean_kmt2a_rearranged": -1.4553,
  "chronos_mean_other_leukaemia": -1.4672,
  "difference": 0.0119,
  "pan_essentiality_all_lines": -1.3556,
  "selectivity_rank": 10707,
  "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."
}

{
  "subcellular": "NUCLEAR AND CHROMOSOMAL, WITH A SPINDLE-POLE POOL, AND - THE PART THAT MATTERS FOR THIS GRAPH - A SEPARATE, EXPERIMENTALLY CODED BLOCK FOR A CLEAVAGE FRAGMENT. UniProt O60216 fetched 2026-09-21 carries TWO molecule-scoped comments. The full-length protein: Nucleus (ECO:0000269, three references), Nucleus MATRIX (two), Chromosome (two), Chromosome centromere (one), and Cytoplasm/cytoskeleton/spindle pole (two), with a fully coded note on the prophase pathway. The 64-kDa C-TERMINAL PRODUCT: Cytoplasm cytosol (ECO:0000269, two references) AND Nucleus (two). Every line in both blocks is experimentally coded - there is no ECO:0000250 or ECO:0000305 anywhere [E_q77_loc1].",
  "subnuclear": "THREE EXPERIMENTALLY CODED SUBNUCLEAR ADDRESSES, AND MY DISPATCH BRIEF'S PREMISE THAT SUBNUCLEAR LOCALISATION IS THE WRONG QUESTION FOR COHESIN IS WRONG FOR RAD21 - the literature answers it. (1) NUCLEAR MATRIX, and the evidence is the founding paper for one of RAD21's own UniProt synonyms: Sadano et al. 2000 characterised 'NXP-1' - nuclear matrix protein 1 - as a Rad21/Scc1/Mcd1-related component OF THE NUCLEAR MATRIX, and NXP1 is still listed as a RAD21 gene synonym today [E_q77_loc2]. (2) CHROMOSOME and specifically METAPHASE CENTROMERE, with hRad21 scattered along chromosome arms before prophase, dissociating from arms during prophase and prometaphase, and persisting at centromeres until anaphase [E_q77_loc3]. (3) SPINDLE POLE - a CYTOSKELETAL, non-nuclear address, from a study of cohesin in mitotic spindle aster assembly [E_q77_loc4]. The nuclear-matrix address is the one this graph should care about, because it holds localization-nuclear-matrix-mitotic-scaffold-kmt2a as an open KMT2A compartment and the two have never been put in the same sentence.",
  "how_shown": "nuclear-matrix fractionation and characterisation [E_q77_loc2]; immunofluorescence of metaphase chromosome spreads with M-phase phosphorylation [E_q77_loc3]; immunolocalisation to spindle poles with aster-assembly assays [E_q77_loc4]; caspase-cleavage mapping with fragment-specific localisation [E_q77_loc5]",
  "systems": "human cultured cells throughout. NO haematopoietic localisation system, no primary blast, and nothing KMT2A-rearranged - despite this entry resting on a patient break-mapping study in KMT2A-rearranged AML remission material.",
  "shown_for_fusion_localisation": "NO - nobody has imaged or fractionated RAD21 in any fusion system. What HAS been mapped in KMT2A-rearranged patients is RAD21's DNA BINDING SITES, not its protein compartment: 517 RAD21 sites were preferentially break-prone in normal cells from KMT2A-rearranged AML remission patients, the largest of four feature sets [E2]. Genomic occupancy and subnuclear compartment are different measurements and this entry keeps them apart.",
  "node": "localization-nuclear-matrix-mitotic-scaffold-kmt2a",
  "the_fragment_block_is_the_rule_91_positive_control": "Brief rule 91 documents that KMT2A's UniProt record carries three location comments - full-length plus 'MLL cleavage product N320' and 'MLL cleavage product C180' - and that BOTH FRAGMENT COMMENTS ARE UNCODED, so filtering on ECO:0000269 discards the taspase1 two-fragment reality the graph argues about constantly. RAD21 IS THE SAME STRUCTURE WITH THE OPPOSITE OUTCOME: its cleavage-product comment is fully coded on two references, so a filter keeps it. That makes RAD21 the positive control for the KMT2A complaint - the format supports coded fragment-scoped localisation, and KMT2A's record simply does not use it. The biology is also parallel: a site-specific protease cuts a chromatin protein in two and the C-terminal piece acquires a different distribution from the parent. For KMT2A that protease is taspase1 and the consequence is the fusion's architecture; for RAD21 it is caspase and the consequence is apoptosis.",
  "a_correction_to_my_own_dispatch": "The brief that dispatched this beat said of RAD21 and STAG2 that they are 'a chromatin-architecture story rather than a compartment one' and invited a bounded rejection if subnuclear localisation were the wrong question. FOR RAD21 IT IS NOT THE WRONG QUESTION AND THE REJECTION WOULD HAVE BEEN WRONG: three coded subnuclear addresses exist, one of them - the nuclear matrix - is the founding observation behind a current gene synonym, and a fragment-scoped coded block exists on top of that. The literature wins over the premise. See the STAG2 entry for the half of the premise that survives."
}

{
  "evidence_class_key": "concept-expressed-required-maintains-three-evidence-classes",
  "identifier_note": "RAD21 is O60216, 631 aa, synonyms HR21, KIAA0078, NXP1 and SCC1. data/symbol-collisions.json was grepped for RAD21 and SCC1 before the first query and has NO row for either. THE ADJACENT ROW THAT DOES EXIST IS THE RELEVANT WARNING: the registry's STAG1 row records that gene_exact:STAG1 returns PMEPA1 first, and notes that SMC1A (1,233 aa) and SMC3 (1,217 aa) in this same complex are only 16 residues apart so length is not a general defence within cohesin. RAD21 at 631 aa is safely separated from all of them. One retrieval note worth carrying: the nuclear-matrix literature for this protein is published under NXP-1, and a RAD21-only query does not reach it.",
  "expressed_in": {
    "class": 1,
    "summary": "ubiquitous and DepMap near-pan-essential - pan_essentiality_all_lines -1.3556 in this entry's own scan block, with essentially no difference between KMT2A-rearranged and other leukaemia lines (+0.0119, rank 10,707 of 17,728). Per brief rule 84 that rank is reported as a rank and not as membership of any half."
  },
  "required_in": {
    "class": 2,
    "claim": "IN MICE, RAD21 HAPLOINSUFFICIENCY HAS A LINEAGE-SELECTIVE PHENOTYPE AND THE LINEAGE IT HITS IS THE LYMPHOID ONE. Conditional Rad21 haploinsufficient cells show decreased haematopoietic reconstitution in competitive bone-marrow transplantation, and the reduction in peripheral blood chimerism is SPECIFIC TO THE LYMPHOID COMPARTMENT while the myeloid compartment is unaffected; the HSPC and myeloid-progenitor compartments are altered, with significant accumulation of GRANULOCYTE-MACROPHAGE PROGENITORS in the bone marrow [E_q77_ct1]. Separately, inducible shRNA silencing of cohesin members in mice did NOT impair viability, increased progenitor replating capacity, altered stem-cell homeostasis and myelopoiesis with widespread chromatin-accessibility changes, and produced a myeloproliferative-neoplasm-like disease in aged animals [E5].",
    "pmids": [
      "37534584",
      "26438359"
    ],
    "system_named_in_the_claim": "murine conditional Rad21 knockout and murine inducible cohesin shRNA, both in normal haematopoiesis with no fusion present",
    "nodes": [
      "cell-type-hsc-bone-marrow",
      "cell-type-gmp"
    ],
    "why_the_lymphoid_specificity_matters_here": "This graph's disease is B-lineage and its cell-type evidence is predominantly myeloid - brief rule 73 measures 21% myeloid markers against 18% B-lineage in the load-bearing evidence. A cohesin dosage phenotype that is specific to the LYMPHOID compartment is therefore unusually relevant, and it sits beside the STAG2 result in the sibling entry, where Stag2 loss reduces HSPC commitment to the B-cell lineage by closing chromatin at Ebf1 and Pax5. Two cohesin subunits, two independent murine studies, both pointing at B-lymphoid commitment."
  },
  "the_feasibility_answer_for_the_degron": "This entry already argues that the RAD21 degron experiment is more feasible than the hypothesis assumed, because cohesin loss in mice gives expansion rather than death. The Rad21-specific conditional knockout sharpens that and complicates it: haploinsufficiency is tolerated but DOES reduce lymphoid reconstitution, so a degron readout in a B-lineage system has a real baseline effect to control for, which a myeloid readout would not.",
  "context": "murine normal haematopoiesis, two independent models, no fusion"
}

{
  "evidence_class_key": "concept-expressed-required-maintains-three-evidence-classes",
  "maintains_state": {
    "class": 3,
    "states": [
      "cell-state-stem-like-self-renewal",
      "cell-state-unsynchronisable-cell-cycle"
    ],
    "claim": "COHESIN DOSE SETS SELF-RENEWAL, AND REDUCING IT INCREASES SELF-RENEWAL RATHER THAN KILLING THE CELL - which is why the therapeutic reading of this entry's break-mapping result has to be handled carefully. In mice, inducible cohesin silencing increases the replating capacity of haematopoietic progenitors, alters stem-cell homeostasis and myelopoiesis with widespread chromatin-accessibility changes, and gives a myeloproliferative-neoplasm-like disease in aged animals [E5]; Rad21 haploinsufficiency accumulates granulocyte-macrophage progenitors [E_q77_ct1]. The cell-cycle arm is structural rather than transcriptional: RAD21 is phosphorylated in M phase, leaves chromosome arms at prophase and prometaphase, persists at centromeres until separase cleaves it at anaphase [E_q77_loc3], and a SEPARATE caspase cleavage generates a 64-kDa C-terminal fragment that redistributes and promotes apoptosis [E_q77_loc5].",
    "pmids": [
      "26438359",
      "37534584",
      "11073952"
    ],
    "system_named_in_the_claim": "murine normal haematopoiesis for the self-renewal state; human cultured cells for the mitotic and apoptotic cleavage mechanisms - no fusion in any of them"
  },
  "the_two_proteases_are_the_state_switch": "RAD21 is cut by SEPARASE at anaphase, releasing chromosomes, and by CASPASE during apoptosis, releasing a 64-kDa C-terminal fragment that is found in BOTH cytosol and nucleus and that promotes apoptosis. So the same protein's proteolysis reports two different cell fates and the products have different distributions. Any experiment that depletes RAD21 - including the degron this graph keeps asking for - is entering a system where a cleavage FRAGMENT has its own activity and its own compartment, and a degron that removes the whole protein and a caspase that generates the fragment are not equivalent perturbations.",
  "confidence": "moderate",
  "caveats": "Both cell-state results are murine and in normal haematopoiesis with no fusion. The cohesin-silencing study perturbs several subunits, so the self-renewal phenotype is a complex-level statement rather than a RAD21 one - brief rule 64 in the other direction, where the complex result cannot be attributed to the subunit. The mitotic and apoptotic cleavage mechanisms are from human cultured cells in the early 2000s and have not been revisited in a leukaemic context. And this entry's own DepMap arm shows essentially no KMT2A selectivity, so none of this is a selectivity argument.",
  "counter_evidence": [
    {
      "note": "The most important published negative for this entry is already at [E4] and it constrains every state claim here: complete loss of Stag2 or haploinsufficiency of Smc3 were WELL TOLERATED in MLL-AF9-expressing murine HSPCs with minimal effect on in vitro self-renewal, so cohesin loss and an MLL fusion are NOT synthetic lethal, and the authors propose the two act through a COMMON HOXA-upregulation mechanism. If cohesin loss and the fusion install the same self-renewal state by different routes, then a cohesin-directed intervention in KMT2A-rearranged disease is pushing on a lever the fusion is already holding down - which predicts exactly the flat DepMap profile this entry records.",
      "citation": {
        "title": "Cohesin loss alters adult hematopoietic stem cell homeostasis, leading to myeloproliferative neoplasms.",
        "authors": "Mullenders J, Aranda-Orgilles B, Lhoumaud P, Keller M, Pae J, Wang K, Kayembe C, Rocha PP, Raviram R, Gong Y, Premsrirut PK, Tsirigos A, Bonneau R, Skok JA, Cimmino L, Hoehn D, Aifantis I",
        "year": 2015,
        "venue": "J Exp Med",
        "pmid": "26438359",
        "doi": "10.1084/jem.20151323",
        "type": "paper"
      }
    }
  ]
}