Oncofusion

start / graph / drug-silmitasertib

Drugmoderate confidence 3 of 5: One strong study, or several smaller consistent ones.Q-subtraction-kmt2a-counter

Silmitasertib (CX-4945) - a CKII inhibitor that works on KMT2A-rearranged leukemia by strengthening the allele the fusion does not have

also called CX-4945 · silmitasertib · CK2 inhibitor CX-4945

This drug blocks a kinase that has nothing to do with the leukemic fusion protein. It works - in mice - by making the healthy copy of KMT2A last longer, and the extra healthy protein crowds the leukemic one off the genes it was switching on.

The same thing, for a scientist

CX-4945 (silmitasertib) is an orally available ATP-competitive casein kinase II inhibitor that has been studied clinically in several cancers. Its relevance here is mechanistic and specific: CKII phosphorylates KMT2A at T2724/S2726, adjacent to the second taspase1 cleavage site, and that phosphorylation licenses cleavage; cleavage in turn primes MLL1 for proteasomal degradation [E1]. Inhibiting CKII therefore **stabilises full-length wild-type MLL1 and raises its chromatin occupancy**. In SEM and MV4-11 cells - both KMT2A::AFF1 - CX-4945 increased full-length MLL1 without changing MLL-AF4 protein levels, inhibited growth of SEM, MV4-11 and RS4;11, and **decreased genome-wide MLL-AFF1, AFF4 and ELL2 chromatin occupancy** with reduced target-gene mRNA. In an MLL-AF9 syngeneic mouse model it significantly delayed progression and extended survival by oral gavage (50 and 75 mg/kg twice daily) and by intraperitoneal injection [E1]. This is, as far as this graph can establish, the only published in-vivo demonstration of the allele-directed strategy that `hyp-fusion-is-exempt-from-restraint-not-free-of-it` and `subtraction-kmt2a-win-wrad-interface` both argue for on theoretical grounds. It is also a single paper, with a pleiotropic kinase inhibitor, and the attribution to MLL1 stabilisation is not formally proven.

Go deeper
Why this matters strategically

Every targeted agent in KMT2A-rearranged leukemia acts on the fusion or on machinery the fusion recruits: menin inhibitors on the fusion's own N-terminal interface, DOT1L inhibitors on the partner-recruited methyltransferase, BET and CDK9 inhibitors on the elongation apparatus. CX-4945 does none of that. It never touches the fusion. It raises the abundance of the wild-type KMT2A protein, and the extra wild-type protein competes the fusion off chromatin 1.

That makes it non-overlapping with every existing agent by construction, and it is the strongest existing argument that the intact allele is a drug target in this disease.

The caveats, which are substantial
  • One paper. No independent replication of the leukemia result is known to this graph.
  • CKII is pleiotropic. It has hundreds of substrates, and CX-4945 has off-target activity beyond CKII. The survival benefit in MLL-AF9 mice is consistent with MLL1 stabilisation but is not formally attributed to it; the discriminating experiment - CX-4945 in a non-cleavable or wild-type-allele-null background - has not been run.
  • The mechanism has a floor. The wild-type Mll allele is genetically required for MLL-AF9 leukemogenesis and maintenance (PMID 20159607, in ana-kmt2a-lost-module-ledger E13), so the relationship between wild-type KMT2A abundance and fusion occupancy may be non-monotonic. exp-wildtype-allele-dosage-vs-fusion-occupancy is the experiment that would map it.
  • An obvious resistance route. Losing or silencing the wild-type allele defeats the mechanism entirely, and this graph already records wild-type KMT2A silencing in a related context.
  • No paediatric leukemia trial. The clinical development of silmitasertib has been in other indications; this graph makes no claim about registered studies in KMT2A-rearranged disease and has not checked the registry for this entry.
Facts and figures
targetCSNK2A1 / CSNK2A2 (casein kinase II catalytic subunits) - acting indirectly on KMT2A by blocking the T2724/S2726 phosphorylation that licenses taspase1 cleavage
modalitysmall molecule, ATP-competitive, orally available
statusclinical-stage in other indications; NOT in clinical development for KMT2A-rearranged leukemia as far as this graph has established, and the registry was not queried for this entry
mechanism in kmt2a rstabilises full-length wild-type KMT2A, raising its chromatin occupancy, which displaces the fusion and SEC components (AFF4, ELL2) from fusion target loci - an allele-directed rather than fusion-directed mechanism
preclinical evidenceSEM, MV4-11, RS4;11 growth inhibition; genome-wide MLL-AFF1/AFF4/ELL2 displacement in SEM; MLL-AF9 syngeneic mouse survival benefit by two routes of administration (PMID 30573454)
key aenot assessed in this entry; CKII is pleiotropic and the safety profile should be taken from the drug's own clinical literature, which this graph has not reviewed
resistance
[
  "loss or silencing of the wild-type KMT2A allele defeats the mechanism entirely",
  "the mechanism is dose-ceilinged by how much wild-type protein can be stabilised",
  "if the wild-type-abundance/fusion-occupancy relationship is non-monotonic, under-dosing could be neutral or harmful"
]
trials
[]
How this could bite us back
  • This is a single paper with a pleiotropic kinase inhibitor, and the graph should not present it as an established allele-directed therapy. It is the best existing proof of concept for a strategy, which is a much weaker statement than it can be made to sound.
  • Any agent whose mechanism runs through the wild-type allele is defeated by losing that allele - and unlike resistance to a fusion-directed drug, that escape requires no new mutation in a drug-binding site, only loss of a copy.
  • If the wild-type-abundance curve is non-monotonic, a trial dosing below the crossover could accelerate disease. That risk is specific to this mechanism and has no analogue in the menin-inhibitor programme.
Still open
  1. Is the MLL-AF9 survival benefit actually mediated by wild-type MLL1 stabilisation? CX-4945 in a Kmt2a wild-type-allele-null or non-cleavable background would attribute it, and that experiment has not been run.
  2. Has anyone replicated the leukemia result? One paper, one group, and the mechanism is strategically important enough that replication should precede any translational enthusiasm.
  3. Does CX-4945 combine additively or synergistically with a menin inhibitor? The two act on different alleles of the same gene - one evicting the fusion, the other reinforcing the wild-type protein that would occupy the vacated site - which is a combination logic this graph has not previously considered.
  4. Is CX-4945 sensitivity across KMT2A-rearranged lines proportional to wild-type KMT2A dosage? If the mechanism is competition it must be, and the prediction is testable across the handful of available lines today.
Every source behind this entry (2)
  1. E1strong confidence 4 of 5: Replicated in peer-reviewed work by more than one group.paper

    CKII phosphorylates MLL1 at T2724/S2726 proximal to the second taspase1 cleavage site and thereby facilitates cleavage; CX-4945 inhibits that phosphorylation in vitro with an IC50 of 0.011 microM, blocks taspase1-dependent MLL1 processing, and selectively increases full-length MLL1 protein without affecting MLL2 or TFIIA cleavage; in SEM and MV4-11 cells CX-4945 raised full-length MLL1 without altering MLL-AF4 levels, caused dose-dependent growth inhibition of SEM, MV4-11 and RS4;11, and substantially decreased genome-wide MLL-AFF1, AFF4 and ELL2 occupancy with decreased target-gene mRNA; and in an MLL-AF9 syngeneic mouse model CX-4945 significantly delayed leukemic progression and improved survival by oral gavage (50 or 75 mg/kg twice daily, 20 doses) and by intraperitoneal injection, with reduced tumour burden by IVIS.

    Regulation of MLL/COMPASS stability through its proteolytic cleavage by taspase1 as a possible approach for clinical therapy of leukemia · Zhao Z, Wang L, Volk AG, Birch NW, Stoltz KL, Bartom ET, Marshall SA, Rendleman EJ, Nestler CM, Shilati J, Schiltz GE, Crispino JD, Shilatifard A · Genes & Development 33(1-2):61-74 2019
    PMID 30573454DOI
    What this evidence cannot show

    FULL TEXT read via PMC6317322. A single study from one group. CKII is a highly pleiotropic kinase and CX-4945 has activity beyond it, so the attribution of the in-vivo survival benefit to MLL1 stabilisation is a mechanistic inference supported by the ChIP-seq displacement data rather than a formal genetic attribution. No independent replication of the leukemia result is known to this graph, and no enantiomer-style inactive-analogue control was used for the in-vivo arms.

    Evidence that argues against this (1)
    • The wild-type Mll allele is genetically REQUIRED for MLL-AF9 leukemogenesis and maintenance, so a mechanism that works by increasing wild-type MLL1 cannot be linear and could in principle help the leukemia at some doses. The dose-response has not been measured. Thiel AT et al. 2010
  2. E2moderate confidence 3 of 5: One strong study, or several smaller consistent ones.paper

    A dedicated virtual screen of 30 million compounds against the taspase1 crystal structure yielded 26 candidates, none of which blocked taspase1 activity in vitro or in cells, and a 70% knockdown of taspase1 was insufficient to affect MLL1 or MLL2 cleavage - which is why CKII, acting on the substrate rather than the protease, is the tractable node in this pathway.

    Regulation of MLL/COMPASS stability through its proteolytic cleavage by taspase1 as a possible approach for clinical therapy of leukemia · Zhao Z, Wang L, Volk AG, Birch NW, Stoltz KL, Bartom ET, Marshall SA, Rendleman EJ, Nestler CM, Shilati J, Schiltz GE, Crispino JD, Shilatifard A · Genes & Development 33(1-2):61-74 2019
    PMID 30573454DOI
    What this evidence cannot show

    FULL TEXT read via PMC6317322; the screen result is reported within that paper as 'data not shown', so it is an author statement rather than published data. CORRECTED 2026-09-20 by the Q40 counter-evidence quality sweep: this caveat previously said the screen was 'the only substantive public information on taspase1's druggability'. That is false. A pharmacologic taspase1 inhibitor was published in 2012 (NSC48300, PMID 22166309) and nanomolar supramolecular inhibitors in 2024 (PMID 38958349); neither was cited anywhere in this graph before this sweep. See the two counter-evidence notes below.

    Evidence that argues against this (3)
    • The same paper, read against itself. A failed virtual screen is weak evidence that taspase1 is undruggable, and this entry uses it to motivate an indirect route. Twenty-six candidates from a single docking campaign against one crystal form is a small, method-bound sample, and the authors report it as 'data not shown'; it says the approach tried did not work, not that the site is intractable. The argument for silmitasertib should not rest on it. Independent literature on this exact point exists and is given in the two notes that follow, which this sweep added on 2026-09-20 after fetching both records from PubMed. Genes & Development 2019
    • INDEPENDENT SOURCE, added 2026-09-20 by the Q40 counter-evidence quality sweep. Taspase1 has an published pharmacologic inhibitor. Chen et al. screened the NCI diversity library in a cell-based dual-fluorescent proteolytic assay and identified NSC48300, a reversible NON-COMPETITIVE inhibitor of taspase1 with Ki = 4.22 uM, which inhibited tumour growth in the MMTV-neu breast model and the U251 brain xenograft; they also report that genetic deletion of Taspase1 in mouse produces no overt deficit, which is a therapeutic-index argument the silmitasertib entry does not make. This directly contradicts the premise that the only information available is one failed docking campaign, and it is the reason the E2 caveat was corrected. Chen DY et al. 2012
    • INDEPENDENT SOURCE, added 2026-09-20 by the Q40 counter-evidence quality sweep, and it is twelve years newer than the docking campaign this entry relies on. Multivalent molecular tweezers directed at taspase1's flexible surface loop reach ligand binding affinities in the NANOMOLAR range with efficient proteolytic inhibition, and a calix[4]arene scaffold improves on that further. Taspase1's undruggability is therefore a claim about small molecules docked into one crystal form, not about the protein. Hommel K et al. 2024
Connected to