Separating CK2's two routes into KMT2A leukaemia, which every published experiment has confounded
also called CX-4945 route separation · IBM versus taspase1 route
Blocking CK2 helps in this leukaemia for two possible reasons, and no experiment has ever told them apart. A short time-course with the right readouts would.
The same thing, for a scientist
CK2 inhibition in KMT2A-rearranged cells could act by (A) **loosening the fusion's own LEDGF grip** through S136/S142 on the retained fragment, or (B) **stabilising wild-type MLL1** by blocking taspase1 cleavage at T2724/S2726, which then competes the chimera off chromatin. Both predict reduced fusion chromatin occupancy and both are consistent with every published CX-4945 result in this disease. **They are distinguishable on three axes: time, wild-type allele, and readout.** Route A should be fast (a phospho-mark lifetime) and independent of the wild-type allele; route B is slow (a protein-accumulation timescale) and strictly requires it. The experiment is a CX-4945 time-course in a t(4;11) line and in an isogenic wild-type-KMT2A-null derivative, with fusion-LEDGF proximity ligation, full-length MLL1 western and fusion CUT&RUN at each time point.
Design
Cells: SEM or RS4;11 (KMT2A::AFF1), plus an isogenic derivative in which the wild-type KMT2A allele is disrupted. The isogenic pair is the whole experiment; without it the routes cannot be separated.
Treatment: CX-4945, and TTP22 as a structurally unrelated CK2 inhibitor, at two doses. Time points at 0.5, 1, 2, 4, 8, 24 and 30 h. Published work used 30 h, which is far past the point where route B dominates.
Readouts at every time point:
| readout | route A predicts | route B predicts |
|---|---|---|
| phospho-S136/S142 (phospho-specific antibody or targeted MS) | falls first, within hours | falls, but irrelevant to the phenotype |
| fusion-LEDGF proximity ligation signal | falls early | unchanged until MLL1 accumulates |
| full-length MLL1 western | unchanged early | rises, slowly |
| fusion CUT&RUN at HOXA9/MEIS1 | falls early | falls late, tracking MLL1 |
| the same in the wild-type-null derivative | all effects preserved | all effects abolished |
The wild-type-null arm is the decisive one and it is a single knockout line.
Why it is worth the effort
If route B is the whole effect, then CK2 inhibition in this disease is a wild-type-allele-dependent therapy, with a one-step resistance mechanism (lose the allele) and no activity in patients who have already lost it. That is a serious limitation and it should be known before a trial, not after.
If route A contributes, then there is a fusion-directed, menin-pocket-independent mechanism, the medicinal chemistry target becomes the IBM-IBD interface rather than CK2 itself, and the strategy survives MEN1 resistance mutations.
Every published CX-4945 experiment in KMT2A-rearranged leukaemia is compatible with both, so the literature cannot answer this and will not by accumulation.
| design | CX-4945 and TTP22 time-course (0.5-30 h, two doses) in SEM or RS4;11 and in an isogenic wild-type-KMT2A-disrupted derivative |
|---|---|
| readout | phospho-S136/S142 (phospho-specific antibody or targeted MS), fusion-LEDGF proximity ligation, full-length MLL1 western, fusion CUT&RUN at HOXA9/MEIS1, DEK phospho-S301/303/306/307 as a third-route control |
| controls | structurally unrelated second CK2 inhibitor; CSNK2A1/CSNK2A2 knockdown arm; isogenic wild-type-null pair; vehicle time-matched |
| cost tier | med |
| timeline | 4-6 months including isogenic line generation |
tests hypotheses
[ "hyp-ck2-is-the-second-handle-on-the-retained-fragment" ]
- If the answer is route B only, the experiment will have shown that a promising preclinical strategy depends on an allele the disease can lose - a negative that is more useful than the ambiguity, and less welcome.
- A pan-CK2 inhibitor time-course cannot exclude routes through substrates nobody has identified; the experiment separates the two named routes, not all routes.
- Generating a wild-type-KMT2A-null line changes the cell in ways beyond the experiment's variable, and a difference in the knockout arm may reflect adaptation rather than mechanism.
- Does a phospho-specific antibody to KMT2A S136/S142 exist? If not, targeted mass spectrometry on immunoprecipitated KMT2A is the fallback and adds cost.
- Is a wild-type-KMT2A-null t(4;11) line viable? KMT2A haploinsufficiency phenotypes suggest it may be, but complete loss in a leukaemic line is untested here.
- Should the design include a menin-inhibitor arm to test whether the two mechanisms are additive?
- E1strong confidence 4 of 5: Replicated in peer-reviewed work by more than one group.paper
CX-4945 treatment of SEM cells increased full-length MLL1 without altering MLL-AF4 protein levels, decreased genome-wide MLL-AFF1, AFF4 and ELL2 occupancy, and the authors attribute chimera displacement to competition from stabilised wild-type MLL1; CKII inhibitors CX-4945 and TTP-22 inhibited MLL wild-type peptide phosphorylation with IC50 values of 0.011 and 0.38 micromolar.
Regulation of MLL/COMPASS stability through its proteolytic cleavage by taspase1 as a possible approach for clinical therapy of leukemia. · Zhao Z, et al. · Genes Dev 2019What this evidence cannot show
Full text read via Europe PMC fullTextXML (PMC6317322) on 2026-09-19. Treatments were 30 h, a single late time point for this purpose, and no wild-type-KMT2A-null control was used, which is precisely why the routes remain confounded. The IC50 values are for peptide phosphorylation in vitro, not cellular target engagement.
Evidence that argues against this (1)
- The alternative route has its own functional evidence at the residue level: S136A/S142A substitution reduces MLL1-fusion transformation capacity and phosphomimetic substitution rescues it, which is a fusion-intrinsic phenotype that the wild-type-competition model does not explain. Sharma S et al. 2018
- E2moderate confidence 3 of 5: One strong study, or several smaller consistent ones.paper
A third CK2 route exists in the same disease - CK2 phosphorylation of DEK at Ser301/303/306/307 sustaining leukaemia stem cells in MLL-AF9 models, with CX-4945 selectively depleting LSCs - so a CX-4945 phenotype in KMT2A-rearranged cells has at least three candidate mediators, not two.
Phosphorylated DEK sustains leukemia stem cells by enabling PBX3-driven transcriptional reprogramming. · Lei Y, et al. · Blood 2026What this evidence cannot show
Abstract read 2026-09-19; 2026 publication, no independent replication. DEK is not a KMT2A contact, so this route would not be detected by any of the readouts in the design above and would appear as unexplained residual effect. The design should therefore include a DEK phospho-readout or accept that a third route is uncontrolled.
Evidence that argues against this (1)
- That a pleiotropic constitutive kinase has three published routes into one disease is itself an argument that no single-route attribution will succeed, and that the therapeutic effect may be the sum of many substrates rather than any one - in which case this experiment returns a partial answer at best. Zhao Z et al. 2019
isogenic wild-type-null arm separates the IBM route from the taspase1 route
- tests →CSNK2A1 (casein kinase 2 alpha) - the one kinase that turns up three separate times in KMT2A biology
CX-4945 and TTP22 time-course with a knockdown arm
route B of the separation
time-course and isogenic wild-type-null arm
DEK-4S is the uncontrolled third mediator of any CX-4945 phenotype in this disease