Knock S395A into the MLLT3 half of a KMT2A::MLLT3 model and ask whether it becomes chemotherapy-resistant
also called S395A knock-in · AF9 phospho-dead fusion · MLL-AF9 S395A
Change one letter in the cancer fusion so the DNA-damage enzyme can no longer tag it, then give the mice chemotherapy. If the untagged version survives treatment better, then part of what chemotherapy does in this leukaemia is switch the fusion off, which nobody currently believes.
The same thing, for a scientist
The discriminating test for the chemotherapy-off-switch hypothesis. Introduce a serine-to-alanine substitution at MLLT3 S395 in a KMT2A::MLLT3 retroviral or knock-in model, confirm loss of damage-induced phospho-S/T-Q signal on the fusion, and compare genotoxic sensitivity to the wild-type fusion in vitro and in transplant. The phosphomimic S395D is the complementary arm. Because S395 is the only S/T-Q motif in AF9, the mutation is as clean a perturbation as this system allows: there is no second PIKK site to compensate.
Design and the reason the controls are the hard part
Constructs. KMT2A::MLLT3 wild-type, S395A, S395D. Retroviral transduction of murine HSPCs is the fast route; a knock-in at the endogenous locus in a human line by base editing is the rigorous one.
Confirm the perturbation first. Damage-induced phospho-S/T-Q signal on the immunoprecipitated fusion must be present in wild-type and absent in S395A. Without that, a downstream phenotype is uninterpretable. This step depends on exp-fusion-immunoprecipitation-phosphoproteomics having succeeded.
Primary readout. Dose-response to cytarabine, an anthracycline and ionising radiation; colony formation; and in vivo survival after transplant with and without a genotoxic regimen.
Secondary readout. Fusion-SEC co-immunoprecipitation (CDK9, AFF1, ELL) with and without damage, and HOXA9/MEIS1 expression at 1, 6 and 24 hours after treatment. The prediction is that wild-type fusion transiently loses SEC binding and target-gene output and S395A does not.
The control that makes or breaks it. S395A must not change untreated transformation, proliferation or target-gene expression. If it does, the mutation is affecting the fusion's baseline function and any difference under chemotherapy is confounded. An S395D arm showing the opposite direction is the internal consistency check.
Power. A survival difference of the size worth acting on is large; 8-10 mice per arm is the usual starting point for this design in this model, and a pilot in vitro dose-response should precede any transplant.
| design | KMT2A::MLLT3 wild-type, S395A and S395D in murine HSPC retroviral transformation and, if feasible, base-edited knock-in in a human KMT2A::MLLT3 line. Confirm loss of damage-induced phospho-S/T-Q on the immunoprecipitated fusion before phenotyping. |
|---|---|
| readout | Genotoxic dose-response (cytarabine, anthracycline, IR), colony formation, transplant survival with and without a genotoxic regimen; fusion-SEC co-IP and HOXA9/MEIS1 at 1, 6 and 24 h |
| controls | untreated transformation, proliferation and target-gene expression must be unchanged by S395A; S395D as the opposite-direction arm; empty vector; a KMT2A::MLLT3 line with an unrelated silent mutation as the editing control |
| cost tier | high |
| timeline | 9-12 months including the in vivo arm |
| depends on | exp-fusion-immunoprecipitation-phosphoproteomics |
| prediction | S395A fusion is less sensitive to genotoxic chemotherapy than wild-type fusion, with no baseline difference |
| falsifier | no difference in genotoxic sensitivity, or a baseline difference that confounds the comparison |
- If S395A makes the leukaemia chemo-resistant, the corollary is that anything reducing DNA-PK activity in a patient - including a DNA-PK inhibitor given as a sensitiser, and possibly some off-target kinase inhibition - does the same. That is the finding with clinical weight and it would come out of this experiment as a side effect of the main comparison.
- A murine transplant arm with a genotoxic regimen is a substantial animal-welfare cost for a hypothesis at 0.3 prior probability. The in vitro dose-response and the fusion-SEC co-IP should be required to clear a pre-specified bar before any mouse is used.
- Should the murine model use the human MLLT3 sequence, given that the experiment turns on a single human residue? Murine Mllt3 numbering must be checked before designing the edit.
- Does the KMT2A half's own PIKK phosphorylation swamp the partner's single site? A KMT2A-side S/T-Q mutant arm would control for this but multiplies the work.
- Is transient loss of SEC binding over hours even measurable as a survival difference over weeks?
- E1strong confidence 4 of 5: Replicated in peer-reviewed work by more than one group.paper
An AF9 S395A mutant is phospho-defective after irradiation and retains SEC interaction, and cells expressing it show enhanced global nascent transcription after irradiation relative to wild-type; the S395D phosphomimic reduces SEC and TFIID interaction.
Post-translational modification-dependent oligomerization switch in regulation of global transcription and DNA damage repair during genotoxic stress. · Talukdar P, et al. · Nat Commun 2024What this evidence cannot show
Established for wild-type AF9 in HEK293T, not for the fusion and not in a haematopoietic cell. Several panels are n=1 or n=2. The S395D arm also reduces TFIID binding, so the mimic is not a clean SEC-specific perturbation.
Evidence that argues against this (1)
- The fusion lacks the Poly-Ser domain and the YEATS domain that mediate AF9's TFIID interaction, so the TFIID half of the S395D phenotype cannot transfer to the fusion. That is an argument that the mutation would be CLEANER in the fusion than in wild-type AF9 - which is convenient, and should be treated with the suspicion that convenience deserves until measured. Kabra A et al. 2022
- E2strong confidence 4 of 5: Replicated in peer-reviewed work by more than one group.database
S395 is the only S/T-Q motif in AF9's 568 residues and lies C-terminal of residue 376, so it is present in the fusion protein in roughly 92% of KMT2A::MLLT3 patients and has no paralogous site within the protein that could compensate for its mutation.
UniProtKB P42568 canonical FASTA, [ST]Q motif scan 2026-09-19: one match at S395; breakpoint boundary from this graph's subtraction-mllt3-yeats entry · UniProt Consortium 2026What this evidence cannot show
Motif uniqueness constrains PIKK action within AF9 but does not exclude a non-canonical site, and the KMT2A half of the fusion carries its own S/T-Q motifs which this experiment does not control for. The breakpoint figure is a pooled review percentage.
Evidence that argues against this (1)
- The KMT2A half is large and carries its own PIKK sites, and this graph already records an ATR-dependent switch at KMT2A S516. A phospho-dead mutation at the partner's single site may be swamped by phosphorylation on the 5' half, which would produce a false negative. Casado P et al. 2023
discriminating in vivo test
phospho-dead and phosphomimic arms of S395
perturbs the retained partner residue in the fusion
- depends on →Immunoprecipitate the KMT2A fusion and map its phosphosites - the experiment this whole field has never done
needs the fusion phospho-blot to confirm the perturbation
- ← related toQ25 enumeration: what was searched, found, rejected and could not be read across the five 3' partners
the discriminating in vivo test