Oncofusion

start / graph / localization-polycomb-bodies-kmt2a-r

Mechanismmoderate confidence 3 of 5: One strong study, or several smaller consistent ones.Q4b-localization

Polycomb bodies — indirect evidence via CBX8, undercut by the finding that CBX8-PRC1 condensates are dissolved rather than built by the rest of PRC1

also called Polycomb body · PRC1 condensate · CBX2 IDR · CBX8 chromodomain · PcG foci

Polycomb bodies are dense spots where the machinery that keeps developmental genes switched off collects. KMT2A fusions physically bind CBX8, a component of that machinery, which is the only reason to think the fusion goes near these spots — but the protein that actually builds the droplets is a different family member, and adding the rest of the complex makes CBX8's own droplets dissolve.

The same thing, for a scientist

Evidence grade for KMT2A: **indirect, via a partner, and weakened on inspection.** The partner link is solid. CBX8 interacts with MLL-AF9 and with TIP60, is required for MLL-AF9 transcriptional activation and leukemogenesis, and point mutations in MLL-AF9 that specifically disrupt the CBX8 interaction abrogate HOXA upregulation and abolish transformation — while Cbx8-null mice are viable with no apparent haematopoietic defect [E1]. A selective, cell-permeable CBX8 chromodomain inhibitor (SW2_110A, Kd 800 nM, ~5-fold paralogue selectivity) inhibits CBX8-chromatin association, blocks proliferation of MLL-AF9-driven THP1 cells and reduces HOXA9, establishing the chromodomain as necessary [E2]. For MLL-ENL the relationship is antagonistic rather than cooperative: ENL contacts both the elongation machinery and CBX8, the interactions are mutually exclusive in vitro, and MLL-ENL neutralises CBX8-mediated inhibition of elongation [E3]. The **compartment** link is where it falls apart. PRC1 phase separation is driven by CBX2, not CBX8: reconstituted PRC1 demixes at low concentration and physiological salt, the behaviour is CBX2-driven, and point mutations in the CBX2 internal domain that abolish demixing also abolish puncta in cells and were previously shown to abolish nucleosome compaction and cause axial patterning defects in mice [E4][E5]. A scaffold-client analysis then found that PRC1 composition **selectively promotes CBX4-PRC1 condensates while dissolving condensates of CBX6-, CBX7- and CBX8-PRC1** [E6]. So the one CBX paralogue the fusion engages is the one whose condensates the rest of the complex takes apart. The honest grade is: strong partner biochemistry, no compartment evidence, and a specific structural reason to doubt the compartment inference.

Go deeper
The partner link, which is genuinely strong

Tan et al. showed CBX8 interacts with MLL-AF9 and TIP60, is required for MLL-AF9-induced transcriptional activation and leukemogenesis, and — the discriminating experiment — that point mutations in MLL-AF9 that specifically disrupt the CBX8 interaction abrogate HOX upregulation and abolish transformation, while Cbx8-deficient mice are viable with no apparent haematopoietic defect 1. A viable knockout with an essential role in the leukemia is the therapeutic-window profile the field looks for, and Wang et al. converted it into a chemical probe: SW2_110A binds the CBX8 chromodomain at Kd 800 nM with about 5-fold selectivity over other CBX paralogues, inhibits CBX8-chromatin association in cells, inhibits THP1 (MLL-AF9) proliferation and reduces HOXA9 2.

Note the direction difference between fusions. Maethner et al. found ENL scaffolds both the elongation machinery and CBX8, that the two interactions are mutually exclusive in vitro, that CBX8 inhibits elongation in a reporter assay, and that this inhibition is neutralised by direct association with ENL — MLL-ENL therefore transforms partly by inhibiting Polycomb rather than by recruiting it 3. Whether KMT2A::AFF1 sits on the Tan side (recruit CBX8) or the Maethner side (neutralise CBX8) has never been determined, and the two make opposite predictions for whether a CBX8 inhibitor helps or harms in t(4;11).

Why the compartment inference does not follow1 deeper

'CBX8 is a PRC1 protein, PRC1 lives in Polycomb bodies, therefore the fusion visits Polycomb bodies' is the chain, and the middle link is paralogue-specific.

Plys et al. reconstituted PRC1 and found it phase-separates readily in vitro at low concentration and physiological salt, driven by the CBX2 subunit; point mutations in an internal CBX2 domain eliminate phase separation, eliminate puncta in cells, and are the same mutations previously shown to eliminate nucleosome compaction in vitro and to cause axial patterning defects in mice 4. Tatavosian et al. independently showed CBX2 condensates form by LLPS with liquid-like properties, concentrate DNA and nucleosomes, are driven by conserved IDR residues, are largely independent of H3K27me3, and that CBX2-PRC1 subunit condensation depends on CBX2 rather than the reverse 5.

Brown et al. then applied a scaffold-client framework: CBX2 is the scaffold, other PRC1 proteins are clients, and composition selectively promotes CBX4-PRC1 condensates while dissolving condensates of CBX6-, CBX7- and CBX8-PRC1; CBX8-PRC1 is enriched into CBX2-PRC1 condensates rather than forming its own 6.

The consequence for this beat is concrete. If the fusion's Polycomb contact is CBX8, then in a cell where CBX2-PRC1 is the condensate scaffold, CBX8 is a client that partitions in — so the fusion could be dragged into a CBX2-scaffolded body without any condensate property of its own, or equally could hold CBX8 out of it. Both are testable and neither has been tested. What cannot be claimed is that CBX8 binding places the fusion in a Polycomb body.

One awkward fact about CBX2 in this disease

CBX2 is the scaffold and the fusion engages CBX8. If Polycomb bodies mattered to KMT2A-fusion biology through condensation, CBX2 rather than CBX8 would be the expected genetic dependency — and it is CBX8 that came out of the screens and CBX8 for which a probe was built. That mismatch is itself weak evidence that the relevant CBX8 function here is chromodomain-mediated chromatin reading and protein contact, not compartment membership.

Facts and figures
compartmentPolycomb body / PRC1 condensate
evidence class for kmt2aindirect via partner CBX8; no compartment-level evidence and a paralogue-specific reason to doubt it
direct evidencefalse
indirect evidence via partnertrue
condensate scaffold paralogueCBX2
paralogue engaged by mll fusionCBX8
cbx8 condensates dissolved by prc1 compositiontrue
chemical probeSW2_110A (CBX8 chromodomain, Kd 800 nM)
kd NM800
marker proteins
[
  "CBX2",
  "CBX8",
  "RING1B (RNF2)",
  "BMI1",
  "H3K27me3"
]
How this could bite us back
  • A CBX8 chromodomain inhibitor at 5-fold paralogue selectivity will engage CBX7 and CBX4 at achievable exposures; CBX7 has tumour-suppressor roles in some lineages, so paralogue spillover is a plausible route to accelerating rather than suppressing disease.
  • If MLL-ENL's relationship to CBX8 is inhibitory, then in fusions with that behaviour a CBX8 inhibitor is mechanistically redundant with the oncoprotein — it would do what the fusion already does, and could not help.
  • Cbx8-null mice being viable is reassuring for a therapeutic window in mice; the human paralogue redundancy has not been tested, and paediatric patients need decades of safety, not weeks.
Still open
  1. Does KMT2A::AFF1 bind CBX8 at all? The interaction is established for MLL-AF9 via the AF9 AHD; AFF1 has no AHD, so the t(4;11) fusion may have no Polycomb contact and the whole indirect argument may be specific to the YEATS-partner fusions.
  2. Does the fusion behave like MLL-AF9 (requires CBX8) or like MLL-ENL (neutralises CBX8)? These predict opposite clinical effects for a CBX8 inhibitor in t(4;11), and the graph currently carries the question unresolved under target-phase2-cbx8-dossier.
  3. In a leukemia cell, is CBX8 inside CBX2-scaffolded condensates or excluded from them, and does fusion binding change that partition? A three-colour live-imaging experiment settles whether the fusion has any compartment relationship to Polycomb at all.
  4. Is CBX2 a dependency in KMT2A-rearranged lines? If the condensate scaffold is not required while the client is, condensation is not the mechanism and CBX8 is acting as a reader and adaptor.
Every source behind this entry (6)
  1. E1strong confidence 4 of 5: Replicated in peer-reviewed work by more than one group.paper

    CBX8 interacts with MLL-AF9 and TIP60 and is required for MLL-AF9-induced transcriptional activation and leukemogenesis; CBX8 ablation and point mutations in MLL-AF9 that specifically disrupt the CBX8 interaction both abrogate HOX gene upregulation and abolish MLL-AF9 leukemic transformation, while Cbx8-deficient mice are viable and display no apparent haematopoietic defects.

    CBX8, a polycomb group protein, is essential for MLL-AF9-induced leukemogenesis · Tan J, Jones M, Koseki H, Nakayama M, Muntean AG, Maillard I, Hess JL · Cancer Cell 2011
    PMID 22094252DOI
    What this evidence cannot show

    MLL-AF9 in mouse models; KMT2A::AFF1 was not tested and AFF1 is not an AHD-containing YEATS partner, so the CBX8 contact surface may differ entirely. No subnuclear localisation was assessed in this paper.

    Evidence that argues against this (1)
    • If the CBX8 requirement is real, the Polycomb BODY is probably not the unit through which it acts, which is what this localisation entry needs it for. Eeftens et al. showed that PRC1 forms multicomponent condensates through hetero-oligomerisation, that these preferentially seed at H3K27me3 and subsequently write H2AK119Ub, that inducing Polycomb phase separation can cause chromatin compaction - and that Polycomb condensates are DISPENSABLE for maintenance of the compacted state, with the repressive output recorded in histone marks rather than sustained by the condensate. A CBX8 interaction that MLL-AF9 needs is therefore evidence about a protein-protein contact and a mark, not evidence that the fusion visits or requires a Polycomb body; no subnuclear localisation was assessed in the source paper, and this entry's compartment claim remains unsupported at the compartment level. Eeftens JM et al. 2021
  2. E2strong confidence 4 of 5: Replicated in peer-reviewed work by more than one group.paper

    SW2_110A, a selective cell-permeable CBX8 chromodomain inhibitor with a dissociation constant of 800 nM and minimally 5-fold selectivity over other CBX paralogues, specifically inhibits CBX8 association with chromatin in cells, inhibits proliferation of MLL-AF9-driven THP1 leukemia cells and significantly decreases expression of MLL-AF9 target genes including HOXA9.

    Optimization of Ligands Using Focused DNA-Encoded Libraries To Develop a Selective, Cell-Permeable CBX8 Chromodomain Inhibitor · Wang S, Denton KE, Hobbs KF, Weaver T, McFarlane JMB, Connelly KE, Gignac MC, Milosevich N, Hof F, Paci I, Musselman CA, Dykhuizen EC, Krusemark CJ · ACS Chemical Biology 2019
    PMID 31755685DOI
    What this evidence cannot show

    Five-fold paralogue selectivity is modest and the 800 nM affinity is a probe-grade, not drug-grade, number; a single MLL-AF9 line (THP1) was used for the cellular phenotype, with no t(4;11) line tested.

    Evidence that argues against this (1)
    • A single-line cellular phenotype for a probe-grade compound is the weakest link in this chain, and cell lines are less stable instruments than the design assumes. Ben-David et al. genomically characterised 106 lines grown in two laboratories and 27 strains of one commonly used line, found extensive clonal diversity and rapid diversification driven by culture-condition-sensitive positive selection with differential activation of expression programmes, and - testing the 27 strains against 321 anti-cancer compounds - found at least 75% of compounds that strongly inhibited some strains completely inactive in others. SW2_110A's antiproliferative effect and HOXA9 reduction were measured in THP1 alone, at 800 nM affinity with about 5-fold paralogue selectivity. Neither the potency nor the single-line readout is sufficient to call the chromodomain necessary in this disease, and no t(4;11) line was tested at all. Ben-David U et al. 2018
  3. E3strong confidence 4 of 5: Replicated in peer-reviewed work by more than one group.paper

    ENL scaffolds both the elongation machinery and the PRC1 component CBX8, these interactions are mutually exclusive in vitro, CBX8 inhibits elongation in a reporter assay and this inhibition is neutralised by direct association with ENL, so MLL-ENL transforms in part by inhibiting Polycomb-mediated silencing.

    MLL-ENL inhibits polycomb repressive complex 1 to achieve efficient transformation of hematopoietic cells · Maethner E, Garcia-Cuellar MP, Breitinger C, Takacova S, Divoky V, Hess JL, Slany RK · Cell Reports 2013
    PMID 23623499DOI
    What this evidence cannot show

    MLL-ENL, biochemistry in HEK293 plus haematopoietic transformation assays; the mutual exclusivity is an in-vitro binding result. This paper's conclusion (fusion inhibits PRC1) is in tension with Tan et al.'s (fusion requires CBX8), and the tension has not been resolved.

    Evidence that argues against this (1)
    • The opposite conclusion for a different fusion: CBX8 is required for MLL-AF9 transcriptional activation and leukemogenesis. Tan J et al. 2011
  4. E4strong confidence 4 of 5: Replicated in peer-reviewed work by more than one group.paper

    Reconstituted PRC1 phase-separates into droplets in vitro at low concentration and physiological salt, the behaviour is driven by the CBX2 subunit, and point mutations in an internal CBX2 domain eliminate phase separation, eliminate puncta in cells, and are the same mutations previously shown to eliminate nucleosome compaction in vitro and to generate axial patterning defects in mice.

    Phase separation of Polycomb-repressive complex 1 is governed by a charged disordered region of CBX2 · Plys AJ, Davis CP, Kim J, Rizki G, Keenen MM, Marr SK, Kingston RE · Genes & Development 2019
    PMID 31171700DOI
    What this evidence cannot show

    In-vitro reconstitution at chosen concentrations plus puncta in NIH3T3 cells; the mouse patterning phenotype long predates the phase-separation interpretation and is equally explained by the compaction defect, which the authors acknowledge. CBX2, not CBX8.

    Evidence that argues against this (1)
    • The in-vitro droplets and cellular puncta in this item were diagnosed with assays that were later shown not to discriminate mechanism. Muzzopappa et al. benchmarked LLPS diagnostics against controlled model systems in vitro and in cells and found 1,6-hexanediol treatment and classical FRAP both unable to separate phase separation from molecules binding to spatially clustered binding sites, while preferential internal mixing in half-bleach experiments does separate them; their MOCHA-FRAP measurements show interfacial barrier strength varying substantially between heterochromatin foci, nucleoli, stress granules and nuage granules. Reconstituted PRC1 droplets at chosen concentrations plus puncta in NIH3T3 cells are therefore compatible with CBX2 driving avid multivalent nucleosome binding, and the authors' own acknowledgement that the mouse patterning phenotype is equally explained by the compaction defect makes the same point genetically. Muzzopappa F et al. 2022
  5. E5strong confidence 4 of 5: Replicated in peer-reviewed work by more than one group.paper

    CBX2 undergoes phase separation to form condensates with liquid-like properties in live cells, driven by conserved residues within its intrinsically disordered region; the condensates concentrate DNA and nucleosomes, H3K27me3 has minimal effect on their formation, and condensation of other CBX2-PRC1 subunits depends on CBX2 rather than the reverse.

    Nuclear condensates of the Polycomb protein chromobox 2 (CBX2) assemble through phase separation · Tatavosian R, Kent S, Brown K, Yao T, Duc HN, Huynh TN, Zhen CY, Ma B, Wang H, Ren X · Journal of Biological Chemistry 2018
    PMID 30514760DOI
    What this evidence cannot show

    Fluorescence live-cell imaging of tagged CBX2 in mouse embryonic stem cells, an overexpression-sensitive design of exactly the class flagged in the condensate-evidence-standards entry.

    Evidence that argues against this (1)
    • This is the evidence class the field's own methodological review identifies as insufficient. McSwiggen, Mir, Darzacq and Tjian analysed the in vivo evidence for phase separation across cellular processes and concluded that it is often phenomenological and inadequate to discriminate phase separation from other mechanisms, that causal relationships and functional consequences in vivo are more elusive still, and that what is needed is quantitative measurement of proteins in their ENDOGENOUS state at physiological abundance. Fluorescence live-cell imaging of tagged CBX2 in mouse embryonic stem cells, with condensates concentrating DNA and nucleosomes, meets none of those conditions; that H3K27me3 has minimal effect on formation is itself a hint that what is being imaged is driven by the tagged protein's concentration rather than by the chromatin state it is supposed to read. McSwiggen DT et al. 2019
  6. E6strong confidence 4 of 5: Replicated in peer-reviewed work by more than one group.paper

    PRC1 condensates follow a scaffold-client model in which CBX2 is the scaffold and the other PRC1 proteins are clients; composition selectively promotes formation of CBX4-PRC1 condensates while dissolving condensates of CBX6-, CBX7- and CBX8-PRC1, and controls enrichment of CBX4-, CBX7- and CBX8-PRC1 within CBX2-PRC1 condensates.

    Principles of assembly and regulation of condensates of Polycomb repressive complex 1 through phase separation · Brown K, Chew PY, Ingersoll S, Espinosa JR, Aguirre A, Espinoza A, Wen J, Astatike K, Kutateladze TG, Collepardo-Guevara R, Ren X · Cell Reports 2023
    PMID 37756159DOI
    What this evidence cannot show

    Combined in-vitro, cellular and coarse-grained simulation work; the dissolution of CBX8-PRC1 condensates is a composition-dependent effect in a reconstituted or tagged system, not a measurement in a leukemia cell. No MLL fusion present.

    Evidence that argues against this (1)
    • The scaffold-client architecture is compatible with the condensates being downstream of the repression rather than upstream of it, which is what the direct test found. Eeftens et al. showed PRC1 forms multicomponent condensates by hetero-oligomerisation that preferentially SEED at pre-existing H3K27me3 and then write H2AK119Ub, that inducing Polycomb phase separation can cause compaction, and that the condensates are not required to maintain the compacted state - proposing that the time integral of phase separation is recorded in histone marks which then drive compaction. In that model, composition-dependent dissolution of CBX8-PRC1 condensates is a statement about which complexes co-assemble where marks already are, and cannot be read as a mechanism by which a KMT2A fusion would be excluded from, or captured by, a Polycomb body. Eeftens JM et al. 2021
Connected to