JD-PI95 - a genetically encoded bridging peptide that drags AKAP95 to HSP70 and dissolves its condensates, with in vivo activity in THP-1
also called JD-PI95 peptide · PI95-J domain fusion
Rather than blocking AKAP95 with a drug, the authors built a short protein that grabs AKAP95 with one end and the cell's own protein-disassembly machine with the other. Forced together, the machine takes AKAP95's droplets apart, and the leukaemia cells grow more slowly.
The same thing, for a scientist
**A modality, not a compound.** JD-PI95 fuses **PI95**, an AKAP95-binding peptide, to the conserved **J domain (aa 1-76) of HSP40/DNAJB1**, which loads substrates onto HSP70 and stimulates its ATPase [E1]. Two matched controls are the strength of the design: **H32Q-PI95**, a J-domain point mutant that cannot engage HSP70, and **JD-PI95scr**, a scrambled PI95 that cannot bind AKAP95. In cells, JD-PI95 - and neither control - hooks HSP70 onto AKAP95 by co-IP, accelerates AKAP95 release from chromatin under MNase, shrinks AKAP95-LacI puncta at a LacO array in U2OS, suppresses AKAP95 coactivation in a G-293T dual-luciferase assay (P = 0.0002, 0.0002, 0.006 across conditions), reduces THP-1 proliferation in liquid culture (P = 0.006, 0.001, 0.001), and reduces tumour burden after tail-vein injection of transduced THP-1 into NCG mice [E1]. **Endogenous AKAP95 protein levels are unchanged** - the mechanism is condensate dissolution, not degradation. Estimated intranuclear peptide concentration in THP-1 is **19-38 nM** [E1].
What this is and is not, for anyone tempted to call it a drug
It is not a drug. It is a lentivirally expressed protein, delivered by transducing cells before they are injected into mice. There is no systemic administration, no pharmacokinetics, no formulation, and no experiment in which an animal with established leukaemia receives anything. The in vivo result is that THP-1 cells transduced with JD-PI95 before injection form less tumour burden than THP-1 cells transduced with the controls - which is an on-target genetic result presented in an animal, not a therapeutic experiment.
What it is, and why it belongs in this graph: a worked demonstration that a condensate can be selectively dissolved by forced proximity to a disaggregase, with the two orthogonal negative controls that make the mechanism interpretable. The 'bridge to a quality-control machine' modality is the condensate analogue of a PROTAC, and this is one of the first instances applied to a fusion-leukaemia dependency.
The controls deserve their own line, because they are unusually good. H32Q-PI95 binds AKAP95 but not HSP70; JD-PI95scr engages HSP70 but not AKAP95. Both are inactive on every readout. A phenotype that requires both halves of a bridge is hard to explain by off-target binding of either half.
What was not tested
- No KMT2A::AFF1 cell. Every JD-PI95 experiment is in THP-1 (KMT2A::MLLT3), G-293T, HeLa or U2OS. The index fusion is absent from the entire peptide section.
- No normal haematopoietic cell. The selectivity argument for JD-PI95 is imported entirely from the Akap95-/- mouse, a different reagent with a different mechanism (absence versus mislocalisation). A peptide that drags AKAP95 to HSP70 also perturbs whatever else AKAP95 is holding - the paper concedes 'we cannot rule out the possibility that it may affect the activities of its interacting proteins'.
- No patient sample. The authors state this explicitly as a limitation.
- No resistance experiment. Nothing tests whether AKAP95 point mutants that escape PI95 binding, or HSP70 pathway changes, restore growth.
- The PI95 peptide sequence is not given in the main text, so it cannot be checked against O43823 or re-synthesised from this record.
| target | gene-akap8 |
|---|---|
| modality | genetically encoded bridging peptide (lentivirally expressed); J domain aa 1-76 of HSP40/DNAJB1 fused to an AKAP95-binding peptide, recruiting HSP70 |
| mechanism | forced proximity to a disaggregase - dissolves AKAP95 condensates without changing AKAP95 protein level |
| status | preclinical, research reagent only. Not a compound: no administered agent, no PK, no formulation, no dosing study, no patient sample. |
| shown for fusion | KMT2A::MLLT3 only (THP-1, in vitro and in NCG mice). NOT tested in KMT2A::AFF1, in any primary sample, or in any normal haematopoietic cell. |
trials
[]
key ae
[]
resistance
[]
controls
[ "H32Q-PI95 (J-domain mutant, cannot engage HSP70)", "JD-PI95scr (scrambled PI95, cannot bind AKAP95)" ]
measurements — the numbers behind the chart
[
{
"quantity": "concentration",
"subject": "estimated intranuclear JD-PI95 peptide concentration in THP-1",
"value": 28.5,
"units": "nM",
"n": null,
"universe": "immunoblot of lysates from a known cell number against purified quantified Flag-JD-PI95, assuming a 220 fL HeLa nuclear volume; PMID 41501053 Methods",
"null": null,
"note": "reported as a 19-38 nM range; the midpoint is recorded here and the range is the real result. The nuclear volume used is HeLa's, applied to THP-1."
},
{
"quantity": "p_value",
"subject": "AKAP95 coactivation in the G-293T dual-luciferase assay, JD-PI95 vs control",
"value": 0.0002,
"units": null,
"n": null,
"universe": "two-tailed Student t test, PMID 41501053 Fig 6f; three comparisons reported as 0.0002, 0.0002, 0.006",
"null": null,
"note": "the smallest of the three reported values"
},
{
"quantity": "count",
"subject": "transcripts differing between JD-PI95-treated and untreated THP-1 by PRO-seq",
"value": 916,
"units": "transcripts",
"n": null,
"universe": "PRO-seq, PMID 41501053 Supplementary Fig",
"null": null,
"note": "the metagene comparison against the SCRAMBLED control was not significant; the 916 is against untreated"
},
{
"quantity": "count",
"subject": "animals per arm in the THP-1 xenograft",
"value": 4,
"units": "mice",
"n": 15,
"universe": "NCG tail-vein xenograft, PMID 41501053 Fig 6h: 4 WT, 4 JD-PI95, 3 H32Q-PI95, 4 JD-PI95scr",
"null": null,
"note": "cells were transduced before injection; this is a genetic experiment in an animal, not a treatment experiment"
}
]- Delivery is unsolved and is the whole problem. This is a nuclear-acting peptide delivered by lentiviral transduction ex vivo. Nothing in the paper addresses how an equivalent activity would reach a marrow blast in a patient, and 'design a small molecule that does this' is not a step, it is a programme.
- Forced proximity to HSP70 is not selective for one client. AKAP95 arrives at HSP70 carrying whatever it is bound to - hnRNPs, Pol II, DPY30, MLL1 fragments - and the authors explicitly decline to rule out effects on its interacting proteins. The resulting toxicity would not look like AKAP95 loss.
- Resistance has an obvious route and was not tested: any AKAP95 variant that loses the PI95 epitope while keeping IDR1 escapes the bridge entirely, and unlike an enzyme inhibitor there is no catalytic requirement constraining which residues can change.
- The selectivity data and the efficacy data come from different reagents. If the germline-knockout window does not transfer to acute condensate dissolution, the therapeutic index is unmeasured rather than favourable, and nothing in this paper would have detected that.
- Does JD-PI95 work in a KMT2A::AFF1 line? Every peptide experiment is in THP-1. SEM, RS4;11 and MV4-11 are standard and MV4-11 already has an AKAP95 knockdown growth curve in the same paper, so the reagent gap is one transduction.
- What is the PI95 sequence, and does it map to the region of AKAP95 that the leukaemia phenotype needs? If PI95 binds IDR1 (101-390) the bridge and the MLL1T contact compete directly, which would be a second, simpler mechanism for the phenotype than condensate dissolution.
- Does JD-PI95 spare normal CD34+ progenitors? The selectivity claim is borrowed from a germline knockout mouse; a peptide that mislocalises AKAP95 into an HSP70 pathway is a different perturbation and has never been given to a normal haematopoietic cell.
- E1moderate confidence 3 of 5: One strong study, or several smaller consistent ones.paper
JD-PI95 is a fusion of an AKAP95-binding peptide (PI95) to the conserved J domain (aa 1-76) of HSP40, designed to bridge AKAP95 to HSP70; controls are H32Q-PI95 (J-domain mutant that fails to interact with HSP70) and JD-PI95scr (scrambled PI95 that fails to bind AKAP95). Co-IP shows HSP70 hooked to AKAP95 with JD-PI95 but not H32Q-PI95. Endogenous AKAP95 protein levels are unaffected by any of the three. AKAP95 dissociates from chromatin faster after MNase with JD-PI95 only; AKAP95-LacI puncta at a LacO array in U2OS are significantly smaller with JD-PI95 only (n = 9 control, 11 JD-PI95, 9 H32Q-PI95, 11 JD-PI95scr); dual-luciferase AKAP95 coactivation in G-293T is significantly reduced (P = 0.0002, 0.0002, 0.006); THP-1 proliferation in liquid culture is reduced (P = 0.006, 0.001, 0.001); and tail-vein-injected transduced THP-1 give significantly less tumour burden in NCG mice (n = 4 WT, 4 JD-PI95, 3 H32Q-PI95, 4 JD-PI95scr). qPCR shows reduced static mRNA of several MLL-AF9 targets; PRO-seq shows 916 transcripts differing, with RUNX2, SMC4 and NEK7 lowest in ongoing transcription although the metagene difference versus the scrambled control was not significant. Estimated intranuclear peptide concentration in THP-1 is 19-38 nM.
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 figures and Source Data not available, and several of the peptide panels (the HSP70 co-IP, the AKAP95 protein-level control, the PRO-seq metagene) are supplementary. The animal cohorts are n = 3-4 per arm. The peptide is lentivirally expressed in cells BEFORE injection - there is no administered agent, no dosing and no PK. The PI95 sequence is not disclosed in the main text. The expected_absent_terms record three things the paper does not contain and that a reader would otherwise assume.
Evidence that argues against this (1)
- The paper's own limitations paragraph concedes the effect 'still needs to be tested on clinical patient samples' and that although AKAP95 protein levels are unchanged, effects on its interacting proteins cannot be excluded. Separately, the PRO-seq metagene comparison against the scrambled control was NOT significant, so the on-going-transcription mechanism is supported by three named genes rather than by a global shift. Yan X et al. 2026
- E2strong confidence 4 of 5: Replicated in peer-reviewed work by more than one group.paper
The AKAP95 dependency this peptide exploits is leukaemia-restricted in genetic experiments: Akap95-/- bone marrow reconstitutes normally in 1:1 competitive transplant while MLL-AF9 transformation of the same marrow is impaired.
AKAP95 condensates regulate transcription and can be targeted in MLL-fusion driven oncogenesis. · Yan X, Luan Y, Zhu Y, Cai Y, et al. · Nat Commun 2026What this evidence cannot show
This is the selectivity argument for the peptide and it is IMPORTED from a different reagent. A germline null and a peptide that mislocalises the protein are not the same perturbation, and no normal haematopoietic cell was exposed to JD-PI95 in this study.
Evidence that argues against this (1)
- The selectivity import is not even conservative, because the direction of its error is unknown. A bridging peptide could be SAFER than the knockout - it is partial, tunable and leaves AKAP95 protein levels intact - or MORE toxic, because it drags whatever AKAP95 is bound to into an HSP70 disaggregase, an effect a null allele cannot produce and which the authors explicitly decline to exclude. Nothing in the germline-knockout data constrains which. The window is unmeasured for this agent, not favourable. Yan X et al. 2026
bridges AKAP95 to HSP70 and dissolves its condensates without changing its protein level
the condensate route is what the peptide dissolves; the DPY30 route is never assayed
- treated by →KMT2A::MLLT3 (MLL-AF9), t(9;11)(p21.3;q23.3)
reduces THP-1 proliferation in liquid culture and tumour burden in NCG mice; transduced ex vivo, not administered
its entire selectivity argument is imported from the germline Akap95-/- mouse, a different reagent
the mechanism the peptide is designed to break