AKAP95 reaches this disease by two separate routes - through WRAD/DPY30 and through the retained KMT2A fragment - and only the second one survives the fusion
also called AKAP95 two routes · AKAP95 WRAD versus condensate route
AKAP95 touches the KMT2A machinery in two different ways, using two different parts of itself. One way needs the enzyme that the leukaemia fusion has thrown away; the other grips the part of KMT2A the fusion keeps. Only the second one can matter in the disease.
The same thing, for a scientist
**Route A (2013, Roeder).** AKAP95's N-terminus binds **DPY30**; the AKAP95/DPY30 association enhances **H3K4 methylation by the MLL2/KMT2B complex** on recombinant chromatin. Enzymatic. Requires a SET domain. **Route B (2026, Yan).** AKAP95's **IDR1, residues 101-390**, co-condenses with the **MLL1 translocated fragment** - the N-terminal portion every KMT2A fusion retains - and AKAP95's zinc-finger and phase-separation mutants fail to rescue MLLr line growth. Non-enzymatic. Requires no catalysis at all. **KMT2A::AFF1 loses the SET domain, so Route A cannot operate on the fusion; it can only act on the retained wild-type KMT2A allele. Route B operates on the fusion by construction.** The two routes use regions of AKAP95 that overlap only in **101-210**, which makes them separable in principle and unseparated in practice: no published experiment tests a DPY30-binding mutant in a leukaemia cell, and no published experiment tests a condensation mutant for H3K4 methylation enhancement.
Side by side
| Route A - WRAD/DPY30 | Route B - retained KMT2A fragment | |
|---|---|---|
| AKAP95 region | 1-210 (or the RII domain 572-589; disputed) | IDR1, 101-390 |
| partner | DPY30 (Q9C005, 99 aa) | MLL1T, the retained N-terminal KMT2A fragment |
| nature of contact | direct protein-protein, bacterially purified both sides | co-condensation, charge-complementarity driven, imaging + co-IP |
| functional readout | H3K4 me1/me2/me3 on recombinant chromatin, KMT2B complex only | MLLr line growth, MLL-AF9 transformation, joint-target expression |
| needs the SET domain? | yes | no |
| survives the fusion? | no - KMT2A::AFF1 has no SET domain; only the wild-type allele is in scope | yes - MLL1T is the retained fragment |
| disease evidence | none | MLL-AF9 mouse, THP-1, MOLM-13, MV4-11 |
| source | PMID 23995757 | PMID 41501053 |
The graph held Route A for one cycle and drew the correct conclusion from it - interface-kmt2a-akap8-a-prey-only-grade-on-a-direct-mll-coactivator reasoned that 'AKAP95 acts on the enzyme, not on the fusion, which makes it a candidate modifier of the wild-type allele's output in a fusion-driven cell rather than a fusion partner'. That reasoning is still right about Route A. Route B is simply a second contact it had no way to know about.
The one place the two routes touch, and why it blocks the obvious experiment
The regions overlap in 101-210. Inside that overlap sit the six tyrosines whose mutation to serine (the PS-mutant) abolishes AKAP95-MLL1T co-condensation.
So the PS-mutant is not a clean separation-of-function reagent: it is inside the region Route A also needs, and its effect on DPY30 binding has never been measured.
The 2013 paper's other workhorse construct runs the opposite way, and this is a correction to an earlier draft of this entry, which stated it backwards. AKAP95(101-692) begins exactly where IDR1 begins, so it retains the whole of IDR1 (101-390) and deletes only 1-100 - the DPY30 determinant, which the 2026 fragment series independently shows does not co-localise with MLL1T. AKAP95(101-692) is therefore a clean Route-A-dead, Route-B-intact reagent, not a double mutant. That makes the 2013 data stronger than they were credited with being: if 101-692 condenses normally and still fails to enhance MLL2C methylation, the methylation enhancement is attributable to the DPY30 route and not to condensation. It has never been assayed for condensation, which is the one thing needed to close that argument.
Two constructs would separate them and neither has been made:
- AKAP95(1-100) - the Route A determinant without IDR1's bulk. Tests whether the N-terminal 100 residues are sufficient for DPY30 binding (currently only necessity is shown) and whether they are dispensable for MLL1T co-condensation (the 2026 fragment series says aa1-100 does NOT co-localise with MLL1T, which is half the answer already).
- AKAP95(Δ1-100) i.e. 101-692 assayed for co-condensation rather than for DPY30 binding. The 2026 paper's fragment series implies it should still condense with MLL1T, since IDR1 is 101-390 - which would mean the 2013 paper's negative-control construct is condensation-competent, and every 2013 result read as 'the DPY30 route is required' is also consistent with 'the DPY30 route is required and the condensation route is intact'.
That last point is the sharpest thing in this entry. It is a re-reading of a thirteen-year-old control in the light of a 2026 mapping, it costs nothing to check, and it would change how the 2013 coactivation data should be interpreted.
routes
[
{
"route": "A",
"akap8_region": "1-210",
"partner": "DPY30",
"mechanism": "direct binding, enhances KMT2B H3K4 methylation",
"requires_set_domain": true,
"survives_kmt2a_fusion": false,
"disease_evidence": "none",
"pmid": "23995757"
},
{
"route": "B",
"akap8_region": "101-390 (IDR1)",
"partner": "MLL1 translocated fragment",
"mechanism": "co-condensation, charge complementarity",
"requires_set_domain": false,
"survives_kmt2a_fusion": true,
"disease_evidence": "MLL-AF9 mouse, THP-1, MOLM-13, MV4-11",
"pmid": "41501053"
}
]overlap region
{
"start": 101,
"end": 210,
"note": "contains the six tyrosines of the PS-mutant; this is why no published construct separates the two routes"
}constructs that would separate them
[ "AKAP95(1-100) - never made; tests Route A sufficiency and Route B dispensability", "AKAP95(101-692) assayed for MLL1T co-condensation - the 2013 negative control, predicted by the 2026 mapping to be condensation-COMPETENT, never tested" ]
measurements — the numbers behind the chart
[
{
"quantity": "length",
"subject": "overlap between the Route A region (1-210) and the Route B region (101-390) on AKAP8",
"value": 110,
"units": "aa",
"n": null,
"universe": "residues 101-210 of O43823 (692 aa)",
"null": null,
"note": "15.9% of the protein; the six PS-mutant tyrosines lie inside it"
},
{
"quantity": "count",
"subject": "published experiments testing a DPY30-binding-deficient AKAP95 mutant in a leukaemia cell",
"value": 0,
"units": "experiments",
"n": null,
"universe": "the complete 92-record AKAP8/AKAP95 title-abstract literature plus the full texts of PMID 23995757 and 41501053",
"null": null,
"note": "the experiment that would tell you whether Route A contributes to the disease phenotype"
},
{
"quantity": "count",
"subject": "published experiments testing an AKAP95 condensation mutant for H3K4 methylation enhancement",
"value": 0,
"units": "experiments",
"n": null,
"universe": "same",
"null": null,
"note": "the mirror-image gap"
}
]- The enzymatic route is the memorable one and the one that cannot work on a SET-less fusion. If a downstream reader takes 'AKAP95 enhances MLL methyltransferase activity' out of a summary and designs a KMT2A::AFF1 experiment around H3K4 methylation, the experiment has no mechanistic basis and a null result will be mis-attributed to AKAP95 rather than to the design.
- Both routes converge on wild-type KMT2A as well as on the fusion, because MLL1T is a fragment of the wild-type protein. Any AKAP95-directed agent hits both alleles, and the retained wild-type allele is a live variable in this graph's own model of the disease.
- Is AKAP95(101-692) - the 2013 paper's negative control - condensation-competent? The 2026 fragment mapping says IDR1 is 101-390, so it should be. If it is, every 2013 result interpreted as 'the DPY30 route is required for coactivation' was obtained against a construct that retained the condensation route, and the interpretation needs revisiting.
- Does the DPY30 route contribute anything to MLLr leukaemia? A DPY30-binding-deficient, condensation-competent AKAP95 that fully rescues an AKAP95 knockdown in THP-1 would retire Route A from the disease argument in one experiment.
- Do the two routes compete? DPY30 lives in assembled WRAD modules; MLL1T lives in condensates. If AKAP95 cannot do both at once, the ratio of the two pools is a regulated variable and nothing is known about what sets it.
- E1established confidence 5 of 5: Multiple independent lines of evidence; treated as settled in the field.paper
Route A: AKAP95 binds DPY30 directly (bacterially purified on both sides), the N-terminal region 1-100 is required and 1-210 sufficient in cells, and recombinant AKAP95 but not AKAP95(101-692) enhances MLL2/KMT2B-complex H3K4 mono-, di- and tri-methylation on recombinant chromatin; AKAP95 has no methyltransferase activity of its own and does not bind free H3. The paper's Discussion states that whether AKAP95 enhances H3K4 methylation by other MLL complexes remains to be examined.
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 expected_absent_terms are deliberate: the 2013 paper contains no condensate or phase-separation concept at all, which is why its AKAP95(101-692) control was never considered as a possibly condensation-competent construct. Supplementary figures were unavailable.
Evidence that argues against this (1)
- The inference drawn from this item in the entry's table - that Route A is out of scope for the disease - is stronger than the item supports. What follows from the SET-domain loss is that Route A cannot act on the FUSION PROTEIN. A KMT2A-rearranged blast still carries a wild-type KMT2A allele and a full complement of KMT2B, and both are legitimate Route A substrates. So Route A could contribute to the leukaemic phenotype without ever touching the fusion, and every sentence in this evidence item would remain true. The table's 'survives the fusion: no' row is about the protein, not about the cell. Jiang H et al. 2013
- E2moderate confidence 3 of 5: One strong study, or several smaller consistent ones.paper
Route B: a GFP-fragment series (aa1-100, 101-390, 391-550, 551-692) co-transfected with mCherry-MLL1T in 293T cells identifies aa101-390 (IDR1) as the only fragment that co-localises with MLL1T and forms large nuclear aggregates; aa1-100 does not. Mutating six tyrosines within aa101-210 to serine abolishes the co-condensation. Endogenous AKAP95 co-IPs with Flag-HA-MLL1T; in rescued THP-1 cells only AKAP95 WT, not the ZF- or PS-mutants, co-localises with endogenous MLL1 N-terminal fragments and rescues the growth defect.
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; supplementary unavailable. The fragment mapping is imaging co-localisation, not binding. The expected_absent_terms are the finding that motivates this entry: the 2026 paper tests no DPY30-binding mutant and performs no methyltransferase assay, so it never asks whether Route A contributes to the leukaemia phenotype it measures.
Evidence that argues against this (1)
- The PS-mutant and ZF-mutant fail to rescue, but both also fail to condense at all, so neither separates 'AKAP95 must condense' from 'AKAP95 must condense WITH MLL1T'. And neither has been tested for DPY30 binding, so a residual Route A contribution to the rescue failure cannot be excluded. Yan X et al. 2026
- E3established confidence 5 of 5: Multiple independent lines of evidence; treated as settled in the field.dataset
KMT2A::AFF1 retains the N-terminal portion of KMT2A including the AT-hooks, CXXC and menin-binding region, and loses the PHD fingers, bromodomain, FYRN, transactivation domain and the SET domain, so the fusion protein has no H3K4 methyltransferase activity of its own.
fusion-kmt2a-aff1, data.breakpoints[].domains_retained_5 and domains_lost_5, this project's fusion atlas · this project 2026What this evidence cannot show
An internal record, cited as the graph's own standing position rather than as a primary source, which is why support_location is 'this-project'. The SET-domain loss in KMT2A fusions is not in dispute anywhere in the field; this citation is a pointer to where the graph states it, not an argument for it.
Evidence that argues against this (1)
- This graph's domain-loss record is written per breakpoint for reciprocal KMT2A translocations, and not every KMT2A rearrangement is one. KMT2A partial tandem duplication retains the SET domain, and a KMT2A-PTD cell would therefore be a Route-A-competent counterexample to the whole framing of this entry. The claim as written should be read as 'KMT2A fusions of the translocation class indexed here', which is the population the project cares about but is not the same set as 'KMT2A-rearranged'. Jiang H et al. 2013
Route A, the enzymatic route, which cannot operate on a SET-less fusion
Route B, the co-condensation route, which survives the fusion by construction
- related to →KMT2A::AFF1 (MLL-AF4), t(4;11)(q21;q23)
the SET-domain loss is what decides which route can reach the index fusion
the same discipline applied to a different WRAD-adjacent protein: which allele, which complex, which region
the hypothesis is the claim that Route B carries everything