AKAP95 is phenotypically silent at steady state and required only when a cell is changing state - the same shape in ATRA differentiation and in MLL-AF9 transformation, thirteen years apart
also called AKAP95 state-dependent requirement · ATRA differentiation AKAP95 dependence
Taking AKAP95 away from a resting cell does almost nothing. Take it away while the cell is trying to change into something else - differentiating, or being transformed by a leukaemia fusion - and the change fails. It is a transition protein, not a maintenance protein.
The same thing, for a scientist
**The cell-state axis the PI asks for, with an unusually clean negative arm on both sides.** In mouse **E14 ESCs**, shRNA depletion of AKAP95 has no effect on morphology, proliferation rate, the expression of many pluripotency genes, **global H3K4me3**, or DPY30 recruitment to highly-expressed and bivalent genes [E1]. After 4 days in differentiation medium and 4 days with **1 uM all-trans retinoic acid**, many early-development genes that are strongly induced in control cells are markedly under-induced in AKAP95-depleted cells, and the ATRA-driven rise in H3K4me3 at those loci is diminished [E1]. The same holds in human **NT2** embryonal carcinoma cells: IGFBP5, HAND1 and MSX1 induction is impaired by siRNA [E1]. Thirteen years later, in the haematopoietic system, the identical shape: **Akap95-/- bone marrow shows normal multilineage chimerism in 1:1 competitive transplant** - normal under homeostasis *and* under the regenerative stress of reconstitution - yet **MLL-AF9 transformation of the same marrow is impaired** [E2]. The authors of the 2013 paper named the mechanism themselves: the coactivation function is *'obscured by potentially redundant mechanisms in the undifferentiated ESCs'* [E1].
Why this shape is the argument, not a caveat
A protein whose loss is silent at steady state and damaging during a transition is, on the face of it, a bad drug target - you cannot see the effect in any steady-state assay. For a KMT2A-rearranged blast the logic inverts, and that is the interesting part.
A KMT2A-rearranged blast is, on this project's own reading, a cell held in a failed state transition - cell-state-differentiation-block is the canonical entry. If AKAP95 is required for state transitions and dispensable for maintenance, then:
- a normal haematopoietic cell, which spends most of its life either quiescent or executing a well-buffered differentiation programme, tolerates AKAP95 loss - observed, in competitive transplant 2;
- a transforming cell, which is executing an abnormal, poorly-buffered transcriptional reprogramming, does not - observed, in MLL-AF9 transformation 2.
The therapeutic-window claim in the 2026 paper is therefore not just an empirical observation; it has a mechanism, and the mechanism was published in 2013 in a completely different system. Neither paper cites the other for this point; the 2026 paper does cite the 2013 one, but for the DPY30 association.
The honest counterweight is that the 2026 competitive transplant is a reconstitution assay in a germline knockout. A germline null develops with the deficiency and can compensate. The 2013 ESC result is also in constitutive knockdown. Nobody has performed an acute AKAP95 degron in adult marrow, which is the assay that would actually price the window.
The divergence that stops this being a story about WRAD
AKAP95 and DPY30 co-regulate a set of ATRA targets - the overlap between the genes affected by AKAP95 depletion and by DPY30 depletion is statistically very significant, though less significant than the RbBP5/DPY30 overlap, i.e. AKAP95 behaves less like a WRAD subunit than another WRAD subunit does 1. And the gene classes differ: DPY30/RbBP5 co-dependent genes are enriched for transcriptional regulators, whereas AKAP95-dependent genes, and AKAP95/DPY30 co-dependent genes, are enriched for extracellular matrix and signalling pathways 1.
So even in the system where the AKAP95-DPY30 link was discovered, AKAP95 is not behaving as a WRAD accessory. It has its own target class. That is the earliest hint of what the 2026 paper found - that the leukaemia-relevant AKAP95 function runs through condensates and RNA binding, not through the methyltransferase core.
| location | mouse E14 ESC and human NT2 EC cells (ATRA transition); mouse Lin- bone marrow and adult marrow (MLL-AF9 transition and competitive reconstitution) |
|---|---|
| state | transition - ATRA-driven differentiation, or MLL-AF9-driven transformation |
| phenotype at steady state | ABSENT. No change in ESC morphology, proliferation, pluripotency-gene expression, global H3K4me3 or DPY30 recruitment; no change in multilineage haematopoietic chimerism in competitive transplant. |
| phenotype during transition | PRESENT. Impaired induction of early developmental genes and of locus H3K4me3 after 1 uM ATRA; impaired MLL-AF9 transformation, colony serial replating and leukaemia latency. |
| shown for fusion | MLL-AF9 (mouse, primary and secondary transplant) and, as a proliferation readout only, KMT2A::MLLT3 in THP-1 and KMT2A::AFF1 in MV4-11. Not shown in any primary patient sample. |
markers
[]
atlas refs
[ "GSE48128" ]
measurements — the numbers behind the chart
[
{
"quantity": "concentration",
"subject": "all-trans retinoic acid used to drive the ESC transition",
"value": 1,
"units": "uM",
"n": null,
"universe": "PMID 23995757 Methods; 4 d differentiation medium then 4 d with ATRA",
"null": null,
"note": "the phenotype is absent before this step and present after it"
},
{
"quantity": "count",
"subject": "genes changed in MLL-AF9-transduced Akap95-/- vs WT Lin- bone marrow at 48 h",
"value": 674,
"units": "genes",
"n": null,
"universe": "RNA-seq, |log2 fold change| > 0.58, PMID 41501053 Fig 2h",
"null": null,
"note": "377 up, 295 down; HOXA9, MEIS1 and MYC target sets preferentially enriched among the down-regulated"
},
{
"quantity": "ratio",
"subject": "competitor ratio in the CD45.2 Akap95-/- vs CD45.1 competitive transplant",
"value": 1,
"units": "CD45.2:CD45.1",
"n": null,
"universe": "1e6 CD45.2 cells mixed with CD45.1 marrow, lethally irradiated CD45.1 recipients, PMID 41501053",
"null": null,
"note": "chimerism reported as similar to WT in bone marrow, spleen and peripheral blood across all lineages"
}
]- A target whose loss is invisible at steady state is a target whose toxicity will be invisible in every steady-state safety screen and will appear in the regenerating compartments - marrow recovery after chemotherapy, gut epithelium, wound healing - which is exactly the setting a leukaemia patient lives in.
- If AKAP95 is required to execute state transitions, an AKAP95 inhibitor given alongside a differentiation therapy is mechanistically antagonistic. The combination that looks most natural on paper - a condensate disruptor plus ATRA or a menin inhibitor, both of which work by forcing differentiation - is the one this biology predicts should fail.
- The whole window argument rests on one germline knockout mouse line. A single compensating paralogue would explain it: AKAP8L exists, is nuclear, and also binds H3K4 HMT core subunits. Nobody has made the double.
- Does an ACUTE AKAP95 degron in adult marrow reproduce the germline knockout's innocence? This is the experiment that prices the therapeutic window and it has not been done in any system.
- Is the ATRA result and the MLL-AF9 result the same mechanism? Both are state transitions and both lose H3K4me3 or H3K27ac at the induced loci, but no paper has tested AKAP95 in an ATRA-driven differentiation of a leukaemia cell - which would join the two directly, and APL/PML::RARA is the obvious system.
- Does AKAP95 loss sensitise a KMT2A-rearranged blast to a differentiating agent? If AKAP95 is needed to execute a transition, removing it should make a differentiation-inducing drug less effective, not more - which is the prediction that would embarrass the therapeutic story and nobody has tested it.
- E1established confidence 5 of 5: Multiple independent lines of evidence; treated as settled in the field.paper
In mouse E14 ESCs, depletion of AKAP95 by two different shRNAs has little effect on ESC morphology, proliferation rate, or expression of many pluripotency genes, and does not affect global H3K4me3 or DPY30 recruitment to tested highly-expressed or bivalent genes. After 4 days in differentiation medium and 4 further days with 1 uM all-trans retinoic acid, many early-development genes strongly induced in control cells are markedly under-induced in AKAP95-depleted cells (qPCR and microarray), and the ATRA-mediated increase of H3K4me3 at those loci is diminished by ChIP-qPCR. In human NT2 embryonal carcinoma cells, siRNA depletion of AKAP95 impairs ATRA induction of IGFBP5, HAND1 and MSX1. The genes whose post-ATRA levels are affected over 2-fold by AKAP95 and by DPY30 depletion overlap statistically very significantly, though less significantly than the RbBP5/DPY30 overlap; AKAP95-dependent and AKAP95/DPY30 co-dependent genes are enriched for extracellular matrix and signalling pathways, whereas DPY30/RbBP5 co-dependent genes are enriched for transcriptional regulation. Microarray data are deposited as GEO GSE48128.
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 2013What this evidence cannot show
Full text read; the steady-state negatives all live in Supplementary Fig. 5A-D, which was not available, so they are the main text's description of its own supplement. The microarray comparison against DPY30 and RbBP5 depletion re-uses data from the authors' own earlier DPY30 paper, so the two arms are not independent experiments. The expected_absent_terms entry records that this paper contains no haematopoietic cell at all - which is why the parallel with the 2026 leukaemia result is an analogy this beat is drawing, not a claim either paper makes.
Evidence that argues against this (1)
- The steady-state negative is a knockdown, not a knockout, and the paper does not report the residual AKAP95 level as a fraction. A partial depletion that is sufficient to break an induction can easily be insufficient to break a maintenance function, which would produce exactly this pattern for reasons that have nothing to do with cell state. Jiang H et al. 2013
- E2strong confidence 4 of 5: Replicated in peer-reviewed work by more than one group.paper
Akap95-/- mice had previously been shown dispensable for homeostatic haematopoiesis; in 1:1 competitive transplantation of CD45.2 wild-type or Akap95-/- bone marrow against CD45.1 competitor, recipients of Akap95-/- cells showed similar bone marrow, spleen and peripheral blood chimerism in all lineages, so AKAP95 is also dispensable for haematopoietic reconstitution. In the same study, MLL-AF9-transduced Akap95-/- marrow gave longer survival in primary and secondary recipients and fewer serial-replating colonies at equal MLL-AF9 protein, and RNA-seq of MLL-AF9-transduced marrow 48 h post-infection showed 674 genes changed (377 up, 295 down, |log2FC| > 0.58) with HOXA9, MEIS1 and MYC target sets preferentially enriched among the down-regulated genes and the myeloid differentiation markers Itgam and Cd14 up.
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 2026What this evidence cannot show
Full text read; the chimerism panels are Supplementary Fig. 3b-d and were not available, so the 'similar in all lineages' is the main text's description. The expected_absent_terms entry is load-bearing: no acute degron or inducible knockout of AKAP95 in adult marrow appears in this paper, so the window claim rests entirely on a germline null.
Evidence that argues against this (1)
- A germline knockout that develops normally has had a lifetime to compensate. The competitive transplant tests reconstitution by cells that have never had AKAP95, not the effect of removing it from a cell that has it - which is what a drug does. The 2013 ESC paper independently notes that AKAP95's function is 'obscured by potentially redundant mechanisms' at steady state, which is the same mechanism proposed here and a reason to expect compensation to be available. Jiang H et al. 2013
the cell-state axis for AKAP95: silent at steady state, required during a transition
- related to →The differentiation block — what it actually is, operationally, in KMT2A::AFF1 leukaemia
if AKAP95 is a transition protein, a blast held in a failed transition is where its loss should bite hardest - the mechanism behind the claimed therapeutic window
- related to →DPY30 — the smallest COMPASS subunit, and the only one that behaves like a selective dependency
AKAP95 and DPY30 depletion affect overlapping ATRA target sets, but AKAP95's own targets are enriched for ECM and signalling rather than transcriptional regulators
its entire selectivity argument is imported from the germline Akap95-/- mouse, a different reagent