KAT2B / PCAF - named in an existing entry's own id and never written, and the only acetyltransferase in this graph whose substrate is a fusion partner protein

{
  "depmap_common_essential": null,
  "note": "Not queried here. The preprint reports SAGA as a selective chromatin dependency in AML from genome-scale data; this graph's paralogue analysis found acetyltransferase-module selectivity splits by paralogue with rank spreads in the thousands, so a module-level claim must be tested per gene before it is believed."
}

{
  "chronos_mean_kmt2a_rearranged": 0.0162,
  "chronos_mean_other_leukaemia": -0.0365,
  "difference": 0.0526,
  "pan_essentiality_all_lines": 0.02,
  "selectivity_rank": 14947,
  "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": "MAINLY NUCLEAR, WITH A DIFFERENTIATION-STATE-DEPENDENT CYTOPLASMIC POOL AND A CENTROSOMAL ONE. UniProt Q92831 fetched 2026-09-21 annotates 'Nucleus' at ECO:0000269 on three references, 'Cytoplasm' on one and 'Cytoplasm, cytoskeleton, microtubule organizing center, centrosome' on one, with a fully coded note: 'Mainly localizes to the nucleus. Also localizes to centrosomes in late G1 and around the G1/S transition, coinciding with the onset of centriole formation. SUBCELLULAR LOCATION MAY VARY DEPENDING UPON CELL DIFFERENTIATION STATE. Cytoplasmic at the very stages of keratinocyte differentiation, becomes nuclear at later differentiation stages. Cytoplasmic in BASAL epithelial cells (undifferentiated) and nuclear in PARABASAL cells (differentiated). Localizes to sites of DNA damage.' [E_q77_loc1]",
  "subnuclear": "SITES OF DNA DAMAGE, AND NOTHING ELSE - no nuclear body, no matrix, no speckle. UniProt Q92831 records that KAT2B 'localizes to sites of DNA damage' at ECO:0000269, and the cited source is a bromodomain-protein recruitment SCREEN rather than a PCAF study [E_q77_loc3]. THAT SCREEN IS THE SAME PAPER BRIEF RULE 91 NAMES AS ONE OF UNIPROT'S MIS-POINTED CITATIONS FOR KMT2A: PMID 25593309, the ZMYND8/NuRD bromodomain DNA-damage screen, supplies the damage-site annotation for KMT2A and for KAT2B in this graph, and presumably for every other bromodomain protein it scored. One screen is therefore generating 'localizes to sites of DNA damage' across the corpus, and it is a recruitment assay at induced breaks, not a description of where a protein lives. Outside that, KAT2B's genuinely interesting spatial fact is not subnuclear at all - it is that the NUCLEUS-VERSUS-CYTOPLASM partition tracks DIFFERENTIATION STATE [E_q77_loc2].",
  "how_shown": "keratinocyte differentiation time courses with Rb acetylation-site mutants [E_q77_loc2]; a bromodomain-protein recruitment screen at induced DNA double-strand breaks [E_q77_loc3]; TBX5 acetylation with cell-cycle-resolved centrosomal imaging in cardiac and limb development models [E_q77_loc4]",
  "systems": "human keratinocytes, human U2OS-type DNA-damage reporter lines, and murine and chick cardiac and limb development. NO haematopoietic system, no primary blast, nothing KMT2A-rearranged - which is stark for the WRITER in this graph's only complete writer-reader-eraser cycle.",
  "shown_for_fusion_localisation": "NO for KAT2B's own compartment in any fusion system. The fusion-adjacent spatial content here is entirely about the SUBSTRATE and the ERASER rather than the writer: PCAF acetylates AF9 at K339, and HDAC5's CaMKII-driven NUCLEAR EXPORT is what allows that mark to accumulate after ionising radiation. Nobody has asked whether PCAF itself moves. And K339 is absent from the fusion protein in about 92% of KMT2A::MLLT3 patients, so the cycle operates on the wild-type allele.",
  "node": "localization-af9-cytoplasmic-pool-and-hsp90-dependent-nuclear-residence",
  "the_shared_screen_is_a_finding": "PMID 25593309 is UniProt's ECO:0000269 source for 'localizes to sites of DNA damage' on KAT2B, and brief rule 91 records it as one of the three experimental citations on KMT2A's own localisation - where it is a ZMYND8/NuRD bromodomain DNA-damage screen rather than a KMT2A study. The same is true here. A single recruitment screen is therefore the experimental backing for a damage-site annotation on at least two of this graph's proteins, and the annotation it generates is a statement about RECRUITMENT AFTER AN INDUCED BREAK, not about resting distribution. Anyone comparing KMT2A's and KAT2B's damage-site behaviour from UniProt would be comparing two rows of one screen and would not know it."
}

{
  "evidence_class_key": "concept-expressed-required-maintains-three-evidence-classes",
  "identifier_note": "KAT2B is Q92831, 832 aa; its paralogue KAT2A is Q92830, 837 aa - FIVE residues apart, so verify-accessions.py's length check gives essentially no protection on this pair, which the entry already records. data/symbol-collisions.json was grepped for KAT2B and PCAF before the first query and has NO row for either. A row for the KAT2A/KAT2B pair IS PROPOSED in this beat's memo on the rule-68 pattern - near-identical lengths, adjacent accessions differing in one digit, shared complexes, and a literature that writes 'GCN5/PCAF' as a unit - and the registry was NOT edited, per the dispatch.",
  "expressed_in": {
    "class": 1,
    "summary": "broadly expressed. The notable expression-level fact is spatial rather than tissue-level: PCAF's own subcellular distribution tracks differentiation state, being cytoplasmic in undifferentiated basal epithelial cells and nuclear in differentiated parabasal cells [E_q77_loc2]."
  },
  "required_in": {
    "class": 2,
    "claim": "NO NORMAL HAEMATOPOIETIC CELL TYPE HAS BEEN SHOWN TO REQUIRE KAT2B, and the requirement that IS established is in a non-haematopoietic differentiation programme: PCAF acetyltransferase activity is required for normal human keratinocyte differentiation, acting through acetylation of Rb at sites lying WITHIN Rb's nuclear localisation sequence [E_q77_loc2]. In the haematopoietic setting what exists is a LINEAGE result inside a fusion: swapping the PCAF HAT domain into MLL-CBP conferred enhanced proliferative capacity but lost the myeloid surface markers of MLL-CBP, and reconstituted mice developed different disease including LYMPHOID disease [E3].",
    "pmids": [
      "20940255",
      "17047066"
    ],
    "system_named_in_the_claim": "human keratinocyte differentiation for the requirement; murine bone-marrow reconstitution with domain-swapped MLL-CBP for the lineage result",
    "nodes": []
  },
  "bounded_absence_on_cell_type": {
    "date": "2026-09-21",
    "tool": "NCBI esearch db=pubmed",
    "query": "(PCAF[tiab] OR KAT2B[tiab]) AND (hematopoietic[tiab] OR erythroid[tiab] OR myeloid[tiab] OR leukemia[tiab])",
    "count": 61,
    "on_topic_after_title_check": 0,
    "what_they_are": "the 61 are dominated by inhibitor chemistry, vascular ageing, renal fibrosis, osteoclastogenesis and p53-acetylation cell biology; none reports a normal haematopoietic cell type in which KAT2B is required. The nearest on-beat items are the 2006 MLL-CBP domain swap this entry already holds and a 2026 preprint on KAT2A/B degradation in AML, also already held.",
    "bound": "titles and abstracts only, brief rule 80 - and this is precisely the case where that bound bites, because a PCAF requirement in a progenitor would most likely appear as one panel of a paper about something else. Aliases searched: KAT2B, PCAF."
  },
  "context": "human keratinocytes; murine MLL-CBP reconstitution; no normal haematopoietic requirement published"
}

{
  "evidence_class_key": "concept-expressed-required-maintains-three-evidence-classes",
  "maintains_state": {
    "class": 3,
    "states": [
      "cell-state-differentiation-block",
      "cell-state-cell-cycle-g1-cdk6"
    ],
    "claim": "KAT2B'S LOCATION IS ITSELF A DIFFERENTIATION READOUT, WHICH IS AN UNUSUAL AND USEFUL SHAPE. In human keratinocytes PCAF is CYTOPLASMIC in undifferentiated basal cells and NUCLEAR in differentiated parabasal cells, and its acetyltransferase activity is required for differentiation to proceed - through Rb acetylation at sites lying inside Rb's own nuclear localisation sequence, with Rb depletion causing failure of differentiation, delayed cell-cycle exit and permitted cell-cycle re-entry [E_q77_loc2]. The cell-cycle arm is separately anchored at the CENTROSOME: PCAF localises there in late G1 and around the G1/S transition, coinciding with the onset of centriole formation [E_q77_loc4].",
    "pmids": [
      "20940255",
      "29174768"
    ],
    "system_named_in_the_claim": "human keratinocyte differentiation and cardiac/limb development models - no haematopoietic cell and no fusion"
  },
  "why_this_is_worth_carrying_into_a_leukaemia_graph": "This graph's central phenotype is a differentiation block, and here is an acetyltransferase whose own compartment moves WITH differentiation and whose catalytic activity is required FOR differentiation, acting on a substrate whose acetylation sites sit in its nuclear localisation signal. Every element of that is measurable in a KMT2A-rearranged blast under a menin inhibitor - where differentiation is exactly what the drug is trying to induce - and none of it has been measured. Two antibodies and a fractionation would say whether PCAF relocates as blasts differentiate.",
  "confidence": "moderate",
  "caveats": "Everything in this block is epithelial or developmental. Keratinocyte differentiation is not haematopoietic differentiation, and the Rb-acetylation mechanism is specific to a substrate whose role differs between the two. The centrosomal localisation comes from a cardiac and limb development study. No fusion, no blast, no leukaemia. The one fusion-relevant lineage result [E3] is a 2006 domain-swap in MLL-CBP, which is not a fusion this graph otherwise holds.",
  "counter_evidence": [
    {
      "note": "The paralogue makes every KAT2B-specific claim provisional, and the tools cannot separate them. KAT2A is 837 aa against KAT2B's 832, they share the SAGA and ATAC complexes, the preprint this entry cites at [E4] treats KAT2A/KAT2B as one degradation target, and the 2006 domain swap found that BOTH the P/CAF and the GCN5 HAT domains changed the disease when substituted into MLL-CBP. So the lineage result this entry treats as evidence about PCAF is equally evidence about GCN5, and the differentiation-state localisation has not been checked for KAT2A at all.",
      "citation": {
        "title": "Bromodomain and histone acetyltransferase domain specificities control mixed lineage leukemia phenotype.",
        "authors": "Santillan DA, Theisler CM, Ryan AS, Popovic R, Stuart T, Zhou MM, Alkan S, Zeleznik-Le NJ",
        "year": 2006,
        "venue": "Cancer Res",
        "pmid": "17047066",
        "doi": "10.1158/0008-5472.CAN-06-2597",
        "type": "paper"
      }
    }
  ]
}