METTL3 turns out to be part of mitosis, which makes its inhibitors a scheduling problem

The uncomfortable arithmetic sits between two of the study's own results. Cut the CDK1-METTL3-7SK chain with CRISPR and cells start missegregating chromosomes, with lagging chromosomes and mutations as the readout ^[6]. Kabeche's own framing is that aneuploidy in cancer cells tracks with poor prognosis, more metastasis and more drug resistance ^[8]. A compound that blocks METTL3 is, at the level of this mechanism, doing something adjacent to what the genetic lesion does: cells killed by it count as efficacy, and cells that survive it carry away the karyotype the clinic likes least ^[16].

The account we have describes genetic perturbation, not compound dosing ^[12]. So that is a question the paper hands to translational teams, not one it settles.

The attribution problem is the sharper one. Work on METTL3 in cancer, neurodegeneration and stem cell biology has historically explained the enzyme's effects through post-transcriptional mechanisms, meaning changes to mRNA ^[11]. If a real part of its output is elongation control through the scaffolding RNA 7SK, which sequesters elongation factors until METTL3 tags it ^[3][4], then a pharmacodynamic panel built on mRNA fate can look perfectly informative while the consequential activity is happening on chromatin inside a window that lasts about an hour ^[9].

There is also a sign problem in the source itself. The paper's mitotic mechanism has CDK1 activating METTL3 at the moment a cell commits to division ^[13], while Kabeche describes CDK1 as a switch in which high enough activity turns METTL3 off, flipping back as CDK1 falls at the end of mitosis ^[14]. Both statements appear in the same account, and they cannot describe the same regulatory event with the same sign without detail the article does not supply ^[15]. Which one holds decides whether an inhibitor imitates the high-CDK1 state or blocks the pulse that clears RNA off the chromosomes before they condense ^[5].

The combination idea rests on a premise worth naming. Chromosomal instability is present in more than 90% of solid tumours ^[7], and the argument is that such cells sit closer to their tolerance limit for missegregation than normal tissue does, so pushing everything past that limit is selective ^[10]. That premise is the entire therapeutic index, and here it is carried by an analogy to desert heat rather than by a measurement. The number to ask the authors for is not the effect size in cancer lines. It is the missegregation rate in untransformed dividing cells at the same exposure.