Oncofusion

start / graph / interface-akap8-dpy30

Interfacemoderate confidence 3 of 5: One strong study, or several smaller consistent ones.q166-the-akap95-dpy30-axis

AKAP95-DPY30: a direct contact whose binding determinant two groups map to OPPOSITE ends of AKAP95, and neither has been structurally resolved

also called AKAP8-DPY30 · AKAP95-DPY30 · O43823-Q9C005

AKAP95 grips DPY30, the small shared part of every machine that writes the activating H3K4 mark. Two labs have tried to find which piece of AKAP95 does the gripping, and they got answers at opposite ends of the protein - nobody has yet solved the structure that would settle it.

The same thing, for a scientist

The contact itself is well supported: AKAP95 is a major protein co-purifying with FLAG-HA-DPY30 from Flp-In-293 nuclear extract, endogenous DPY30 antibody co-IP from HeLa nuclear extract recovers it, and **bacterially purified GST-AKAP95 binds bacterially purified FLAG-HA-DPY30 on beads** - both sides recombinant, so the contact is direct [E1]. **What is not settled is where.** Roeder's group maps the determinant to the **N-terminus**: AKAP95(1-210) is sufficient for a Myc-DPY30 co-IP from cells, AKAP95(101-692) is not, and in the purified assay AKAP95(101-692) binds *much more weakly* than full length. UniProt records the feature as **Interaction with DPY30, 1-210** [E2]. **Poreba's group maps it to the PKA-binding (RII) domain at the far C-terminus, 572-589**, and shows on the DPY30 side that the C-terminal Dpy-30 motif is required and that the single substitution **DPY30 L69D** abolishes both dimerisation and AKAP8 binding [E3]. The 2013 paper explicitly says the two AKAP95 regions 'seem to be different'. **There is no structure of this interface and no Kd.**

Go deeper
What the direct-binding experiment actually was, because it is routinely over-stated

This project's previous entry on AKAP8, and the brief that dispatched this beat, both rendered the result as 'purified AKAP95 residues 1-100 bind purified DPY30 directly'. That is not what was done, and the difference matters.

The in vitro binding assay (Fig 5F, Methods) is:

  • Bait: FLAG-HA-DPY30 purified from bacteria, captured on M2 beads.
  • Prey: GST-AKAP95 full length or GST-AKAP95 (101-692), expressed in bacteria from pGEX and purified on glutathione-Sepharose; HeLa nuclear extract runs as a comparison lane.
  • Result: full-length binds; 101-692 binds 'much weaker', not zero.

No AKAP95(1-100) fragment was ever expressed, purified or tested for binding, in that paper or anywhere else this beat could find. The 1-210 sufficiency comes from a different experiment (Fig 5E), a Myc-DPY30 co-IP from transfected G-293T cells, where other proteins are present.

So the defensible statement is: the contact is direct, and removing AKAP95's first 100 residues substantially weakens it. The minimal sufficient fragment is 1-210, in cells. A claim of the form 'residues 1-100 bind DPY30' attributes a sufficiency to a region that was only ever shown to be necessary, and a purity to an experiment that was done in a lysate.

The disagreement, set out rather than resolved
Jiang 2013 (Roeder) [E1][E2]Bieluszewska 2018 (Poreba) [E3]
determinant on AKAP8N-terminus; 1-100 necessary, 1-210 sufficient in cellsPKA-binding (RII) domain, i.e. 572-589
determinant on DPY30not mappedC-terminal Dpy-30 motif (the dimerisation domain)
point mutantI582P (an RII-binding mutant) has little or marginal effect on AKAP95 coactivationDPY30 L69D abolishes dimerisation and AKAP8 binding outright
assaybacterial GST pull-down + cell co-IPnot inspected (full text not obtained)

These are not necessarily incompatible, and the most interesting reading is that they are not. A 2:1 ASH2L-DPY30 arrangement means DPY30 is a dimer in the complex, and a dimer has two faces; AKAP95 is 692 residues with a long IDR and could contact both. Jiang 2013 measured coactivation of a reporter when it called the RII region dispensable, not binding; the I582P result is a functional negative, not a binding negative. Bieluszewska measured binding.

But note which way the ambiguity cuts for this project. The 2013 paper's own headline analogy - the DPY30 dimerisation domain is strikingly homologous to the PKA RII dimerisation domain (their Fig 7G) - predicts Bieluszewska's result, not its own. If a nuclear AKAP binds DPY30 because DPY30's dimerisation domain looks like RII's, then it should bind through the same AKAP surface that binds RII. The 2013 paper states the homology and then maps the interaction somewhere else, and does not remark on the tension.

Is this contact in a region the fusion keeps? The question does not apply here, and that is the point

This project asks of anything structural: is the contact in a region KMT2A fusions retain? (data/symbol-collisions.json records that all eleven KMT2A-H3C1 structural records map to the PHD3-bromodomain cassette every fusion deletes.)

For AKAP95-DPY30 the question is malformed, because neither partner is KMT2A. DPY30 makes no direct contact with KMT2A at all - it docks ASH2L, at 2:1 stoichiometry, and concept-dpy30-what-is-established-for-wild-type-kmt2a-and-what-for-the-fusion already records that the retained/lost framing does not reach it. AKAP95 reaches a KMT2A fusion, if it does, through a different surface: IDR1 (101-390) co-condensing with the retained MLL1 N-terminal fragment 4. See interface-akap8-kmt2a-n-terminal-fragment.

So this graph now holds two AKAP95 routes into the same axis, using non-overlapping regions of AKAP95 (1-210 vs 101-390 - they overlap only in 101-210), and they are separable by mutant. That is the most useful thing in this entry.

Facts and figures
partner agene-akap8
partner bgene-dpy30
partner a uniprotO43823
partner a length AA692
partner b uniprotQ9C005
partner b length AA99
motif bDPY30 C-terminal Dpy-30 dimerisation motif (PF05186); L69D abolishes both dimerisation and AKAP8 binding (PMID 29288530)
druggabilityNot currently druggable and not currently defined. There is no structure, no Kd and no agreed determinant. The tractable object in the neighbourhood is the DPY30 dimerisation surface itself, which is already the target of ASH2L-derived peptides reported to suppress blood-cancer cell growth (PMID 31251903, not read in this beat) - and DPY30 L69D shows that a single substitution on that surface removes AKAP8 as well as BIG1, so any DPY30 dimerisation ligand would be expected to strip AKAP95 too, not selectively.
shown for fusionNO for this specific contact - nobody has tested whether the AKAP95-DPY30 contact is required in any KMT2A-rearranged cell. The AKAP95 requirement in MLLr leukaemia (PMID 41501053) was mapped to the IDR and the zinc fingers, and that paper does not test a DPY30-binding mutant at all.
evidence statemeasured-and-contradicted
evidence state noteThis is not an unasked question. Two groups asked it and got different answers; the gap is a structure, not an experiment.
evidence state why not absence classThis entry is not an absence. It reached for absence_class to record an evidence state that vocabulary cannot express - 'measured-and-contradicted' - so the value moved to evidence_state rather than being folded into an absence token that would misdescribe it. See q190: only 69 of the 529 entries carrying absence_class were absence entries.
interface residues a
[]
structures
[]
ligands
[]
disputed mapping
{
  "roeder_2013": {
    "region_on_akap8": "1-210 (1-100 necessary)",
    "pmid": "23995757"
  },
  "poreba_2018": {
    "region_on_akap8": "572-589, the PKA RII-binding domain",
    "pmid": "29288530"
  },
  "status": "UNRESOLVED. No structure exists. The two are not strictly incompatible - DPY30 is a dimer with two faces and one paper measured coactivation while the other measured binding - but the corpus must not quote either as settled."
}
measurements
[
  {
    "quantity": "length",
    "subject": "DPY30 Q9C005",
    "value": 99,
    "units": "aa",
    "n": 1,
    "universe": "UniProtKB reviewed human entry, retrieved live 2026-09-24",
    "null": null,
    "note": "L69 sits in the C-terminal Dpy-30 motif of a 99-residue protein"
  },
  {
    "quantity": "count",
    "subject": "published structures of the AKAP8-DPY30 interface",
    "value": 0,
    "units": "structures",
    "n": null,
    "universe": "no PDB entry is cited by either primary paper and none was found; a bounded absence, not an exhaustive PDB query",
    "null": null,
    "note": "the interface is defined entirely by deletion and by one point mutant"
  }
]
How this could bite us back
  • The only druggable object here is the DPY30 dimerisation surface, and DPY30 L69D removes AKAP8 and BIG1 together. A ligand for that surface is a general WRAD-assembly disruptor, not an AKAP95 tool - it would hit all six SET1/MLL complexes and the Golgi BIG1 pool at once.
  • If the Poreba mapping is right, the AKAP95 determinant for DPY30 is the same eighteen residues that bind PKA RII. Any peptide designed to block one blocks the other, and PKA anchoring is a housekeeping function in every tissue.
  • A disputed interface invites a separation-of-function mutant that turns out to separate nothing. Before building AKAP95 point mutants, the two published mappings need to be run head to head in one hand, or the mutant will be interpreted against whichever paper the reader happens to believe.
Still open
  1. Which AKAP95 surface binds DPY30? A single co-IP panel with AKAP95 dRII (1-571 + 590-692) against Myc-DPY30 - the exact construct Jiang 2013 already built for a different purpose - would settle the two published mappings against each other in one lane.
  2. Does DPY30 L69D, which abolishes AKAP8 binding, phenocopy AKAP95 loss in an MLLr line? If it does not, the DPY30 route is not the leukaemia-relevant one and this entry drops out of the disease argument entirely.
  3. Is the AKAP95 binding site on DPY30 the same as the ASH2L SDI site? If it is, AKAP95 and ASH2L compete and AKAP95 cannot be in an assembled WRAD module - which would contradict the gel-filtration co-elution. Nobody has asked.
Every source behind this entry (4)
  1. E1established confidence 5 of 5: Multiple independent lines of evidence; treated as settled in the field.paper

    In PMID 23995757, bacterially expressed and purified GST-AKAP95 binds bacterially purified FLAG-HA-DPY30 captured on M2 beads, while GST-AKAP95(101-692) binds much more weakly; separately, AKAP95(1-210) but not AKAP95(101-692) co-immunoprecipitates Myc-DPY30 from transfected G-293T cells. No AKAP95(1-100) fragment was expressed or tested in either assay. AKAP95 is also a major protein co-purifying with FLAG-HA-DPY30 by MALDI-MS from Flp-In-293 nuclear extract, is recovered by an endogenous DPY30 antibody IP from HeLa nuclear extract, and is co-precipitated by WDR5 and ASH2L IPs reciprocally.

    Regulation of transcription by the MLL2 complex and MLL complex-associated AKAP95. · Jiang H, Lu X, Shimada M, Dou Y, Tang Z, Roeder RG · Nat Struct Mol Biol 2013
    PMID 23995757DOI
    What this evidence cannot show

    Full text and Methods read via PMC3813012; the figure panels themselves were not inspected, so 'much weaker' is the authors' word and no densitometry is available. The expected_absent_terms are load-bearing and were checked: the paper contains no Kd, no crystal or NMR structure, and no construct beginning and ending inside 1-100.

    Evidence that argues against this (1)
    • The authors' own framing is that AKAP95 is 'not an integral core subunit of MLL complexes like DPY30 or RbBP5', and that purified DPY30 binds AKAP95 less well than endogenous DPY30 in HeLa nuclear extract does - i.e. the direct contact is weaker than the cellular association, so something else in the extract contributes. Jiang H et al. 2013
  2. E2established confidence 5 of 5: Multiple independent lines of evidence; treated as settled in the field.database

    UniProtKB O43823 records a feature 'Region 1-210, Interaction with DPY30' with evidence ECO:0000269 sourced to PubMed 23995757, and separately a feature 'Region 572-589, RII-binding'. Its SUBUNIT comment reads: 'Interacts with DPY30; mediating AKAP8 association with at least the MLL4/WBP7 HMT complex (PubMed:23995757)'.

    What this evidence cannot show

    In UniProt's nomenclature MLL4/WBP7 is KMT2B (Q9UMN6) - the same gene the source paper calls MLL2. `data/symbol-collisions.json` records the MLL2/MLL4 token as an alias shared by two genes whose protein names carry each other's number. So UniProt's SUBUNIT line, read carefully, scopes the AKAP8-DPY30-mediated complex association to KMT2B and NOT to KMT2A - which is the same scope limit the source paper's own discussion states. The expected_absent_terms entry is the assertion that UniProt does not claim an MLL1/KMT2A complex here.

    Evidence that argues against this (1)
    • UniProt is not independent evidence here - the region feature is ECO:0000269 sourced to the very paper it is being used to adjudicate, and the curator chose the cell co-IP fragment (1-210) over the direct-binding result. A different curator could defensibly have written 1-100. And this entry reads 'at least the MLL4/WBP7 HMT complex' as a scope LIMIT when 'at least' is a floor: the curator may have intended to leave other complexes open rather than to exclude them. UniProt Consortium 2026
  3. E3moderate confidence 3 of 5: One strong study, or several smaller consistent ones.paper

    Bieluszewska et al. report that the PKA-binding domain of AKAP8 and the C-terminal Dpy-30 motif of DPY30 are crucial for the AKAP8-DPY30 interaction, that DPY30 L69D disrupts dimerisation and completely abolishes binding to both AKAP8 and BIG1, that AKAP8 interacts with DPY30 and with RII-alpha in both interphase and mitosis, and that AKAP8L also interacts with H3K4 HMT core subunits. They frame their result as reinforcing the previously postulated analogy between the AKAP8-RIIalpha and AKAP8-DPY30 interactions.

    PKA-binding domain of AKAP8 is essential for direct interaction with DPY30 protein. · Bieluszewska A, Weglewska M, Bieluszewski T, Lesniewicz K, Poreba E · FEBS J 2018
    PMID 29288530DOI
    What this evidence cannot show

    Abstract only; the paper is not in PMC and the full text was not obtained in this beat, so the constructs, the assay and the controls behind the claim were not inspected. Confidence is therefore capped at moderate and this must not be reported as having overturned the 2013 mapping - only as conflicting with it.

    Evidence that argues against this (1)
    • Jiang 2013 found that deleting AKAP95 572-589 does not impair coactivation and that the I582P RII-binding point mutant is near-fully active. If the RII domain were essential for DPY30 binding and DPY30 binding were essential for coactivation, I582P should have been inactive. Either the RII domain is not the DPY30 determinant, or coactivation does not require the DPY30 contact. Jiang H et al. 2013
  4. E4moderate confidence 3 of 5: One strong study, or several smaller consistent ones.paper

    A separate surface of AKAP95 - IDR1, residues 101-390 - is the one that co-condenses with the MLL1 translocated fragment, and the DPY30-binding N-terminal fragment aa1-100 does not co-localise with MLL1T in that assay.

    AKAP95 condensates regulate transcription and can be targeted in MLL-fusion driven oncogenesis. · Yan X, Luan Y, Zhu Y, Cai Y, Ye F, Yang L, Zhu K, Hu J, Luo S, Lin J, Wang Y, Huang Y, Sun P, Yang Z, Shi B, Jiang H, et al. · Nat Commun 2026
    PMID 41501053DOI
    What this evidence cannot show

    The fragment series (aa1-100, 101-390, 391-550, 551-692) was assayed as GFP fusions co-transfected with mcherry-MLL1T and scored by co-localisation and aggregate formation in 293T cells - an imaging readout, not a binding measurement. A fragment can fail to co-localise for expression or folding reasons. No co-IP of individual fragments with MLL1T is reported in the main text.

    Evidence that argues against this (1)
    • A negative co-localisation result for a 100-residue GFP fusion is weak evidence that the region does not contact MLL1T. A short fragment can be freely diffusible, mislocalised, or expressed at a level at which enrichment into a condensate is undetectable by imaging, and no expression control for the individual fragments is reported in the main text. The aa1-100 negative is used in this entry to argue the two routes are separable, and it is the least robust observation in that argument. Yan X et al. 2026
Connected to