GSK3 is required by KMT2A-rearranged leukaemia and nobody has shown it touches KMT2A
also called GSK3 MLL leukaemia · glycogen synthase kinase 3 dependency
An enzyme called GSK3, normally thought of as a brake on cancer, turns out to be needed by KMT2A-rearranged leukaemia specifically. But the reason runs through a cell-cycle brake protein, not through KMT2A itself.
The same thing, for a scientist
Wang et al. reported an **oncogenic requirement for GSK3 in the maintenance of MLL-rearranged leukaemia**, genetically defined by MLL mutation, with the mechanism running through **destabilisation of p27Kip1** and with efficacy in a preclinical murine model. It is the clearest kinase dependency specific to this genotype and it is in this beat because of what it is **not**: there is no reported GSK3 phosphosite on KMT2A, on a fusion, on AFF1, AFF4 or MEN1. The dependency is **genotype-specific and substrate-indirect**, which is a distinct category from every other entry here. It is recorded as such so that a future enumeration does not count it as a phospho-gated contact on the fusion, and does not miss it either.
Why it is in the beat and what it is not
In: a kinase whose requirement is specific to the KMT2A-mutant genotype, with a preclinical efficacy result, in a field where genotype-specific kinase dependencies are rare.
Not: a phosphorylation-dependent contact on KMT2A or any fusion component. The paper's mechanism is p27Kip1 destabilisation; no KMT2A residue is named, and no phosphosite on KMT2A is attributed to GSK3 in UniProt or in the phosphoproteomics resource this beat queried.
The distinction matters because a reader scanning for 'kinases documented in KMT2A leukaemia' will find this paper first - it is in Nature and it is about MLL leukaemia by name - and could reasonably but wrongly record it as a kinase acting on the fusion.
A later, independent phosphoproteomic observation partly re-raises the question from a different direction: releasing PP2A with FTY720 in KMT2A-rearranged cells reduced GSK3-beta activity among four kinases that moved together. That places GSK3-beta inside the phospho-network of this disease without placing it on KMT2A 2.
What would move it from indirect to direct
GSK3 is a processive, priming-dependent kinase: it phosphorylates S/T four residues N-terminal to an already-phosphorylated S/T, giving the consensus S/T-x-x-x-pS/pT. The 56 retained-fragment phosphosites catalogued in ana-kmt2a-phosphosite-atlas-retained-vs-lost are exactly the priming sites a GSK3 search would need.
A consensus scan of the retained fragment against the observed phosphosite list would produce a candidate list in an afternoon, and it has not been done. If a primed GSK3 site sat in the retained fragment, this entry would move from 'genotype-specific and indirect' to a direct phospho-gated contact - and GSK3 inhibitors are abundant, well-characterised and already tested in this disease.
| kinase | GSK3A/GSK3B |
|---|---|
| mechanism | indirect - destabilisation of p27Kip1; genotype-specific dependency in MLL-rearranged leukaemia |
| phospho contact on fusion | false |
| category | genotype-specific kinase dependency, substrate-indirect |
| untested lead | GSK3 consensus is S/T-x-x-x-pS/pT; the 56 observed retained-fragment phosphosites are the priming sites a candidate scan would use, and no such scan has been run |
- GSK3 inhibition stabilises beta-catenin, and Wnt/beta-catenin self-renewal signalling supports KMT2A-rearranged leukaemia stem cells. The same drug may hit the dependency and feed the stem-cell compartment.
- GSK3 has hundreds of substrates across metabolism and neuronal function; systemic inhibition in children is a substantial ask for a dependency whose mechanism is one cell-cycle inhibitor.
- Recording an indirect dependency in a beat about phospho-gated contacts risks it being read as a contact. It is not one, and this entry exists partly to prevent that misreading.
- Is there a primed GSK3 consensus site in the retained fragment? The priming phosphosites are now catalogued and the scan has not been run.
- Why has an eighteen-year-old Nature genotype-specific kinase dependency produced no clinical programme in this disease? The answer is likely in the Wnt counter-mechanism and is not written down anywhere in this graph.
- Does GSK3 inhibition combine with or antagonise menin inhibition in KMT2A-rearranged cells?
- E1strong confidence 4 of 5: Replicated in peer-reviewed work by more than one group.paper
Pharmacological, physiological and genetic experiments demonstrate an oncogenic requirement for GSK3 in the maintenance of MLL-rearranged leukaemia, a mechanism that ultimately involves destabilisation of the cyclin-dependent kinase inhibitor p27Kip1, with GSK3 inhibition showing efficacy in a preclinical murine model of MLL leukaemia.
Glycogen synthase kinase 3 in MLL leukaemia maintenance and targeted therapy. · Wang Z, et al. · Nature 2008What this evidence cannot show
Abstract read 2026-09-19; full text not retrieved. A 2008 result; GSK3 inhibitors have not since become a therapy in this disease, and the abstract's own framing is that GSK3's role here is paradoxical relative to its established tumour-suppressive functions. The mechanism is indirect - no KMT2A substrate is identified - and 'MLL-rearranged' in 2008 covers a set defined less precisely than current KMT2A-rearranged criteria.
Evidence that argues against this (1)
- GSK3 is tumour-suppressive in most contexts and its inhibition promotes Wnt/beta-catenin signalling, which is itself implicated in the self-renewal of KMT2A-rearranged leukaemia stem cells. A GSK3 inhibitor could therefore act against the dependency this paper reports and in favour of the leukaemia through a second pathway in the same cell, and eighteen years without a clinical programme in this indication is consistent with that difficulty. Di Mambro A et al. 2023
- E2moderate confidence 3 of 5: One strong study, or several smaller consistent ones.paper
Phosphoproteomic profiling after PP2A release by FTY720 in KMT2A-rearranged leukaemic models showed reduced activity of PP2A-regulated kinases including GSK3-beta, alongside ERK1, Aurora B and PLK1, placing GSK3-beta within the phosphorylation network of this disease without placing it on KMT2A.
SET-PP2A complex as a new therapeutic target in KMT2A (MLL) rearranged AML. · Di Mambro A, et al. · Oncogene 2023What this evidence cannot show
Abstract-level. Kinase activities are inferred from substrate phosphorylation signatures rather than measured, and FTY720 has S1P-receptor pharmacology independent of SET, so the attribution of any single kinase change to the PP2A axis is weak. Four kinases moving together is a systems response.
Evidence that argues against this (1)
- Four kinases moving in the same direction after one drug is the signature of a broad phosphatase release rather than of a specific regulatory relation, and the same paper's proposal is a feedback loop among PP2A, Aurora B, PLK1, MYC and SET in which GSK3-beta is a follower rather than a driver. Di Mambro A et al. 2023
genotype-specific kinase dependency; mechanism indirect via p27Kip1
GSK3-beta is one of four kinases moved by PP2A release in this disease
recorded as indirect so it is not miscounted as a contact