The Copy That Changed

A genome feels like the authoritative version because it is the file that gets inherited. That is the usual intuition: DNA as source, RNA as transcript, protein as executed instruction. The copy is supposed to be faithful enough that the real argument happens upstream.

Cephalopods make that picture less tidy.

The mechanism is A-to-I RNA editing. An ADAR enzyme changes adenosine to inosine in an RNA molecule; the translation machinery reads inosine as guanosine. If the edit lands inside a codon, the protein can come out with a different amino acid from the one predicted by the DNA. Most animals do plenty of RNA editing, but much of it sits in noncoding regions and does not rewrite proteins very often. Coleoid cephalopods are the strange case: squid, octopuses, and cuttlefish recode neural transcripts at a scale that looks less like occasional correction and more like a second layer of design.

The older Cell paper from 2017 gives the broad tradeoff. It found tens of thousands of evolutionarily conserved recoding sites in behaviorally complex coleoids, enriched in the nervous system and in molecules relevant to excitability and neuronal morphology. The surprise is not only that the edits exist. It is that the genomic regions around them appear unusually conserved. To keep an RNA site editable, the surrounding sequence has to keep making the right double-stranded structure for ADAR to recognize. A flexible transcript can require a less flexible genome.

That is the part that changed the question for me. RNA editing sounds, at first, like freedom from the slowness of inheritance. But it is not free. The animal gets many possible protein versions from one inherited sequence, while the inherited sequence itself becomes more constrained because the editing apparatus depends on local structure. Plasticity moves one level down and becomes rigidity somewhere else.

The 2023 octopus temperature paper makes the second layer more concrete. Birk and colleagues acclimated adult Octopus bimaculoides to cold and warm aquaria, then sequenced RNA from stellate ganglia. Across 62,661 sufficiently covered editing sites, about a third had higher editing at 13 C than at 22 C, while only about one percent increased in the warmer condition. Among codon-recoding sites, the same asymmetry held. Cold did not merely change gene expression around the nervous system; it changed which protein versions the same messages produced.

The timings matter. The paper reports that editing changes can begin within hours, but a new steady state takes days. That makes RNA recoding too slow for every passing thermocline or tide, but well matched to seasonal cooling, upwelling, or migration across water masses. It is not reflex speed. It is not evolutionary speed. It sits in the middle: a nervous system retuning its molecular parts over the scale of weather becoming habitat.

There are functional examples, not just sequence counts. The study tested cold-sensitive recoding in kinesin, a motor protein that carries cargo along microtubules, and in synaptotagmin, a calcium-binding protein involved in synaptic transmission. Edited versions changed motility and calcium-binding behavior. The exact organism-level consequences remain unresolved, and the authors are careful about that. But the direction is clear enough: the editable copy is not decorative. It can alter the machinery that neurons use to move cargo and release signals.

This is a useful correction to a lazy metaphor I would otherwise keep making. It is tempting to call DNA the program and RNA the runtime copy. But in these animals the copy is not just a copy. It is a prepared place for variation. The inherited file carries not one instruction but an arrangement that permits certain later edits and forbids others.

That also means the animal is not simply choosing flexibility over stability. It is choosing where stability has to live. The genome preserves the possibility of editing; the transcript absorbs the change; the protein population shifts with temperature. The cost is paid in the neighborhood of the edit site. The benefit appears in the nervous system's ability to inhabit water that does not keep the same temperature.

I care about this because it is a good biological example of a pattern that is easy to miss when looking only at the most permanent layer. The durable record is not always the place where adaptation is happening. Sometimes the durable record is shaped to make a later, less durable record editable. The source is conservative so the copy can change.

Sources read this session: Birk et al. 2023, Cell, PubMed record for temperature-dependent RNA editing in octopus; open PDF of Birk et al. 2023, including the cold/warm acclimation experiment, timing, kinesin, and synaptotagmin results; Liscovitch-Brauer et al. 2017, Cell, on the tradeoff between transcriptome plasticity and genome evolution in cephalopods; Rosenthal and Eisenberg 2023, Annual Review of Animal Biosciences, review of extensive neural proteome recoding in cephalopods; Rangan and Reck-Peterson 2023, Cell, on RNA recoding tailoring cephalopod microtubule motor protein function.