Oncofusion

start / graph / localization-dpy30-trans-golgi-network-pool

Mechanismmoderate confidence 3 of 5: One strong study, or several smaller consistent ones.Q57-localization2

DPY30 has a second job at the trans-Golgi network, and the authors who found it could not put ASH2L or RBBP5 there

also called mDpy-30 TGN · DPY30 Golgi · WRAD outside the nucleus · BIG1 DPY30 · CIMPR endosomal transport

DPY30 is known as a small piece of the machine that puts an activating mark on DNA packaging proteins in the nucleus. It also sits on a membrane compartment in the cytoplasm, where it helps ferry a receptor back from endosomes - and when it is removed, that traffic goes wrong. The same group looked for two of its nuclear partners there and could not find them.

The same thing, for a scientist

UniProt Q9C005 gives DPY30 two experimental locations from one source: `Nucleus` and `Golgi apparatus, trans-Golgi network`, both ECO:0000269|PubMed:19651892 [E1]. Xu et al. found by immunofluorescence in multiple cell types that mDpy-30 has a **dual nuclear and cytoplasmic distribution, the cytoplasmic part enriched at a perinuclear site**, reproduced by an HA-tagged construct; TGN targeting is mediated by **BIG1**, a TGN-localised ARF guanine-nucleotide exchange factor. Altering mDpy-30 levels changes the distribution of the **cation-independent mannose-6-phosphate receptor (CIMPR)** without affecting TGN46 or transferrin receptor, and knockdown enriches internalised CIMPR near cell protrusions - a defect in endosome-to-TGN transport [E2]. Knockdown of **ASH2L or RBBP5 produces a similar CIMPR redistribution**, yet the same authors state they **could not detect Golgi localisation of ASH2L or RBBP5 with multiple antibodies or GFP fusions** [E3]. The chromatin address annotated for DPY30 - H3K4me3-enriched promoters and 5'UTRs - is itself **by similarity** [E1].

Go deeper
A WRAD subunit on a membrane

The result is unusual enough to state carefully. In HeLa and CV-1 cells, endogenous mDpy-30 shows both nuclear and cytoplasmic staining, with the cytoplasmic pool concentrated perinuclearly; an HA-tagged construct reproduces it, which argues the dual distribution is intrinsic rather than an antibody artefact. The TGN pool is placed there by BIG1, and DPY30 loss produces a cargo-selective trafficking defect - CIMPR is redistributed while TGN46 and transferrin receptor are not 2.

Cargo selectivity is the part that makes this hard to dismiss. A generic secretory-pathway disruption moves everything; a defect confined to one receptor's endosome-to-TGN retrieval is a specific phenotype.

Rule 73: HeLa and CV-1 cells, a cell-biology system. Nothing haematopoietic and nothing leukaemic. And the mechanism is mDpy-30 - the mouse/mammalian protein - in a paper framed around trafficking, not chromatin.

The published negative that keeps this honest

The natural reading - the H3K4 methyltransferase module has a Golgi branch - is one the authors explicitly declined to make. In their discussion:

"we have not been able to detect the Golgi localization of either Ash2L or RbBP5 using multiple antibodies or their GFP fusions (unpublished data)"

So: ASH2L and RBBP5 knockdown reproduces the CIMPR phenotype, but neither protein could be put at the Golgi. The authors offer the alternative that the nuclear H3K4MT complex controls methylation of a trafficking regulator, or that the TGN pool of DPY30 coordinates with nuclear H3K4MT.

This is a bounded absence reported by the people best placed to find the positive, with the reagent classes named (multiple antibodies, GFP fusions). It is worth more than a search-based absence and it is recorded here as such. It also means the correct statement for ASH2L and RBBP5 is not 'also at the Golgi'.

ASH2L's nuclear annotation is a curator inference, not a measurement

Fetched 2026-09-20: UniProt Q9UBL3 (ASH2L, 628 aa) reads SUBCELLULAR LOCATION: Nucleus {ECO:0000305|PubMed:17355966}. ECO:0000305 is 'inferred by curator' - the curator's reading of a paper that did not directly assert the location - not ECO:0000269, 'experimental evidence used in manual assertion'. For a subunit of the complex this graph's entire WRAD position rests on, the nuclear annotation is a curator's inference plus the obvious biochemistry.

The same beat found DOT1L's cytoplasm transferred by similarity, ENL's nucleus uncited, and ASH2L's nucleus curator-inferred. Three of the core interactome's localisation annotations are not measurements, and only the evidence code distinguishes them from the ones that are.

Note also the standing residue-numbering trap for this protein, already in the brief: ASH2L papers number from the 534-residue isoform while Q9UBL3 is 628, a constant offset of 94. That is a different error class, but it is the same protein and worth carrying in one place.

Why a leukaemia graph should care

WDR5, ASH2L, RBBP5 and DPY30 are the WRAD module, and this graph holds a full dossier on whether WRAD reaches KMT2A fusions at all (hyp-wrad-reaches-kmt2a-fusions-only-as-a-second-shell-passenger, absence-no-proximity-degron-or-crispr-study-of-wrad-in-kmt2a-fusion-leukaemia). A DPY30 dependency in a leukaemia screen is routinely read as a COMPASS dependency. On this evidence it could in part be a trafficking dependency:

  • CIMPR retrieval failure affects lysosomal enzyme delivery;
  • DPY30 is small (99 aa), abundant, and shared between complexes;
  • nothing in a knockdown or CRISPR screen distinguishes the two pools.

The discriminating experiment is a separation-of-function allele - a DPY30 mutant that cannot bind BIG1 but still assembles into COMPASS, or the converse - and none exists. Until it does, 'DPY30 is required in KMT2A-rearranged leukaemia' is a statement about the protein, not about the complex.

Facts and figures
compartment subcellularDPY30: nucleus AND Golgi apparatus / trans-Golgi network, both experimentally coded from one source. ASH2L: nucleus, by curator inference (ECO:0000305). RBBP5: no Golgi localisation detected.
compartment subnuclearDPY30 is annotated as associated with H3K4me3-enriched chromatin, highly enriched at promoters and 5'UTRs and not at downstream regions or 3'UTRs - but that note carries ECO:0000250, BY SIMILARITY, not an experimental code. No nuclear-body or compartment-level observation exists for any WRAD subunit.
targeting factorBIG1, a TGN-localised ARF guanine-nucleotide exchange factor, mediates DPY30 TGN targeting
cargo specificityCIMPR (cation-independent mannose-6-phosphate receptor) redistributed; TGN46 and transferrin receptor unaffected
published negativeThe same authors could not detect Golgi localisation of ASH2L or RBBP5 with multiple antibodies or GFP fusions (stated as unpublished data in the discussion)
shown for fusionNO - no WRAD subunit has been localised in a KMT2A-rearranged cell in any source read for this beat
patient material datafalse
residue numbering trapASH2L papers commonly number from the 534-residue isoform while UniProt Q9UBL3 is 628 aa, a constant offset of 94; the DPY30-contact residue published as L513 is Q9UBL3 L607
genes
[
  "gene-dpy30",
  "gene-ash2l",
  "gene-rbbp5"
]
uniprot evidence codes
{
  "DPY30 Nucleus": "ECO:0000269|PubMed:19651892",
  "DPY30 Golgi/TGN": "ECO:0000269|PubMed:19651892",
  "DPY30 chromatin distribution note": "ECO:0000250 - BY SIMILARITY",
  "ASH2L Nucleus": "ECO:0000305|PubMed:17355966 - INFERRED BY CURATOR"
}
methods
[
  "immunofluorescence of endogenous protein in multiple cell types with antibody characterisation",
  "HA- and GFP-tagged constructs",
  "siRNA knockdown with two independent siRNAs",
  "antibody-uptake assays for endogenous and CD8-chimeric CIMPR",
  "GFP-trap immunoprecipitation with tandem mass spectrometry"
]
species and systems
[
  "human HeLa",
  "CV-1 (African green monkey kidney)"
]
subunits
{
  "DPY30": {
    "accession": "Q9C005",
    "length_aa": 99
  },
  "ASH2L": {
    "accession": "Q9UBL3",
    "length_aa": 628
  }
}
How this could bite us back
  • DPY30 is a 99-residue shared subunit with at least two distinct cellular jobs. Any degrader or inhibitor aimed at its COMPASS function necessarily removes the trafficking function too, and the toxicity of the second is unmeasured.
  • A CRISPR or shRNA DPY30 dependency cannot be attributed to chromatin, so a therapeutic rationale built on 'DPY30 is required for KMT2A-fusion H3K4me3' is currently under-determined by its own genetic evidence.
  • CIMPR retrieval failure impairs delivery of lysosomal enzymes - a mechanism with obvious consequences in normal tissue and none of them leukaemia-selective.
  • If ASH2L and RBBP5 produce the same trafficking phenotype without being at the Golgi, then the WRAD module's non-chromatin effects may run through gene expression, which means a chromatin-targeted intervention does not avoid them.
Still open
  1. How much of a DPY30 dependency in KMT2A-rearranged leukaemia is COMPASS and how much is trafficking? No separation-of-function allele exists - a BIG1-binding-dead DPY30 that still assembles into COMPASS would answer it.
  2. Does the DPY30 TGN pool exist in a haematopoietic cell at all? Every observation is in HeLa or CV-1, and a 99-residue protein's minority pool could be cell-type specific.
  3. Is the ASH2L/RBBP5 CIMPR phenotype nuclear in origin - i.e. does the H3K4 methyltransferase control the expression or methylation of a trafficking regulator? The authors raise it and nobody has tested it.
  4. Does CIMPR retrieval failure matter in leukaemia? CIMPR delivers lysosomal hydrolases, and lysosomal function is an active area in leukaemia stem-cell biology, but the two literatures have never met.
Every source behind this entry (3)
  1. E1established confidence 5 of 5: Multiple independent lines of evidence; treated as settled in the field.database

    UniProt Q9C005 (human DPY30, 99 aa) annotates Nucleus and Golgi apparatus/trans-Golgi network, both with ECO:0000269|PubMed:19651892, while the chromatin-distribution note - association with H3K4me3-enriched regions, highly enriched in promoters and 5'UTRs but not in downstream regions or 3'UTRs - carries ECO:0000250, i.e. by similarity; and UniProt Q9UBL3 (human ASH2L, 628 aa) annotates Nucleus with ECO:0000305|PubMed:17355966, i.e. inferred by curator rather than experimentally asserted.

    What this evidence cannot show

    Database records. Both accessions were resolved by gene_exact with organism_id:9606 and reviewed:true and matched on gene_primary; DPY30 length 99, ASH2L length 628 (note the standing 94-residue offset trap: ASH2L papers commonly number from the 534-residue isoform). The point of this item is the evidence-code asymmetry, which is invisible in the flat text rendering that a bulk import reads.

    Evidence that argues against this (1)
    • ECO:0000305 does not mean 'doubtful'. ASH2L is recovered in nuclear extracts and in chromatin-bound COMPASS complexes in essentially every biochemical study of it, so the nuclear assignment is very likely right. The finding recorded here is that the graph should cite that biochemistry rather than the database line, because the database line is a curator's inference.
  2. E2strong confidence 4 of 5: Replicated in peer-reviewed work by more than one group.paper

    In HeLa and CV-1 cells, mDpy-30 - a subunit of several H3K4 methyltransferase complexes - shows a dual nuclear and cytoplasmic distribution with the cytoplasmic pool enriched perinuclearly and localised at the trans-Golgi network, reproduced by an HA-tagged construct; TGN targeting is mediated by BIG1, a TGN-localised ARF guanine-nucleotide exchange factor; altering mDpy-30 levels changes the distribution of the cation-independent mannose-6-phosphate receptor without affecting TGN46 or transferrin receptor, and knockdown enriches internalised CIMPR and recycling endosomes near cell protrusions, indicating a function in endosome-to-TGN transport of CIMPR.

    A role of histone H3 lysine 4 methyltransferase components in endosomal trafficking. · Xu Z, Gong Q, Xia B, Groves B, Zimmermann M, Mugler C, Mu D, Matsumoto B, Seaman M, Ma D · J Cell Biol 2009
    PMID 19651892DOI
    What this evidence cannot show

    RULE 73: HeLa and CV-1 cell-biology systems; no haematopoietic and no leukaemic cell, and no in-vivo work. The TGN pool is a minority of total DPY30 and the paper does not quantify the split. Knockdown phenotypes on membrane traffic can be indirect through the protein's nuclear function - which the authors themselves raise as an alternative. Cargo selectivity (CIMPR affected, TGN46 and TfnR not) is the strongest argument against a generic secretory disruption. Read: full text via Europe PMC PMC2728403, 2026-09-20.

    Evidence that argues against this (1)
    • The authors offer two alternative explanations in their own discussion: that the nuclear H3K4 methyltransferase controls the methylation of a trafficking regulator, and that the TGN pool coordinates with the nuclear complex. Neither is excluded, so 'DPY30 works at the TGN' and 'DPY30 works in the nucleus on a trafficking gene' both fit the data, and no separation-of-function allele exists to distinguish them.
  3. E3moderate confidence 3 of 5: One strong study, or several smaller consistent ones.paper

    Knockdown of ASH2L or RBBP5, two other H3K4 methyltransferase subunits, leads to a CIMPR redistribution similar to that caused by mDpy-30 depletion; the same authors state that they were unable to detect Golgi localisation of either ASH2L or RBBP5 using multiple antibodies or their GFP fusions.

    A role of histone H3 lysine 4 methyltransferase components in endosomal trafficking. · Xu Z, Gong Q, Xia B, Groves B, Zimmermann M, Mugler C, Mu D, Matsumoto B, Seaman M, Ma D · J Cell Biol 2009
    PMID 19651892DOI
    What this evidence cannot show

    The negative is stated in the discussion as UNPUBLISHED DATA with no figure, so its stringency cannot be assessed - 'multiple antibodies or their GFP fusions' names the reagent classes but not the reagents. It is nonetheless a negative reported by the group with the strongest incentive to find a positive, and it is recorded as a bounded absence on that basis. RULE 73: HeLa and CV-1 again. Read: full text via Europe PMC PMC2728403, 2026-09-20.

    Evidence that argues against this (1)
    • A failure to detect a protein at a compartment by antibody or GFP fusion is a weak negative in general: a small minority pool below detection is exactly what DPY30's own TGN fraction looks like, and a GFP fusion can be excluded from a compartment its untagged counterpart reaches. The honest reading is that ASH2L and RBBP5 have a CIMPR phenotype with no demonstrated Golgi localisation, not that they are absent from the Golgi.
Connected to