Transposons — mobile genetic elements that copy or cut themselves from one location in the genome and paste into another — make up roughly half the human genome. When a transposon inserts into a gene, it should disrupt the coding sequence. If the disruption survives into the mRNA, the protein is garbled. Many transposon insertions are tolerated, and the standard explanation was redundancy: backup copies of the gene compensate, or the insertion falls in a non-critical region, or the damaged mRNA is degraded before it causes harm.
Zhao, Nardone, Chang, Paulo, Elledge, and Kennedy (Nature, January 2026) discovered that the tolerance isn't passive. Cells have a dedicated system — which the authors call SOS splicing — that detects transposon insertions in mRNA and cuts them out. The excision operates independently of the spliceosome, the known RNA splicing machinery. It recognizes the structural signature of the transposon: inverted terminal repeats, which are short sequences at each end of the transposon that fold into a recognizable shape in the RNA. When the system detects this shape, it excises the transposon and ligates the flanking mRNA back together.
Three proteins are required. AKAP17A binds the transposon-containing mRNA. CAAP1 bridges the binding protein to RTCB, an RNA ligase. RTCB joins the mRNA fragments after excision. Deleting the internal transposon sequences — everything between the inverted terminal repeats — doesn't affect the system's ability to detect and excise. The trigger is purely structural: the shape formed by the repeats, not the content between them.
The system was identified through a genetic screen in C. elegans, where it excises Tc1 transposons from host mRNAs. It is conserved in human cells. The key proteins — AKAP17A, CAAP1, RTCB — have human orthologs that perform the same function. The defense has existed across hundreds of millions of years of evolutionary distance.
The structural insight: what looked like tolerance was active repair. Transposon insertions that don't kill the gene aren't surviving because the organism can absorb the damage. They're surviving because the organism removes the damage at the RNA level, after transcription but before translation. The mRNA is corrupted and then fixed, silently, by a system nobody knew existed. The resilience isn't in the genome's redundancy. It's in an RNA-level quality control mechanism that erases the corruption before it reaches the ribosome.
Note (2026-11-16; checked against the abstract of the Nature paper, 'An RNA splicing system that excises DNA transposons from animal mRNAs', via Europe PMC, not the full text): the abstract confirms SOS splicing, a pattern-recognition system triggered by base-pairing of inverted terminal repeats, the three factors AKAP17A (binds the transposon-containing mRNAs), RTCB (the ligase) and CAAP1 (bridges the two), and that they are required in both C. elegans and human cells. It words several points more carefully than the essay: SOS splicing 'seems to operate independently of the spliceosome', the authors 'propose' it is a new conserved, RNA-structure-directed mode of splicing, and 'an identified function' (the preprint says 'one function') is to genetically buffer animals from DNA-transposon-mediated gene perturbation. It is also specific to DNA transposons, which are a small part of the genome, where the essay opens with transposons making up about half of the human genome (mostly other classes) and then speaks of the system detecting transposon insertions in general. The essay's 'what looked like tolerance was active repair' and 'the resilience isn't in the genome's redundancy' set that one function against redundancy as alternatives, which the abstract does not do, and the 'hundreds of millions of years' of conservation is the essay's inference; I did not check the Tc1 detail or the deletion experiment.