Originally, I planned to share these thoughts under the heading ‘Brainless animals still sleep’, but the insights are deeper than that. Still, a recent study showing that cnidarians, ‘lowly’ animals with neural networks but without brains, such as jellyfish and sea anemones, exhibit sleep, is a good place to start 1. Thus, sleep actually evolved before a centralized nervous system, which had already become clear with the discovery of sleep-like states in jellyfish almost 10 years ago 2. The latest, more specific, findings bolster a theory that sleep evolved, at least in part, to protect the DNA in individual nerve cells, helping to repair damage that builds up while animals are awake. So, what are these specific findings? It has been known for over a decade that sleep is critical for maintaining neuronal genomic integrity, as it reduces DNA damage, which normally accumulates in nerve cells of (in)vertebrates (e.g., flies and mice) during wakefulness. This stems from, among other things, higher metabolic activity and associated reactive oxygen species (ROS). Therefore, one of the benefits of sleep, and probably its oldest, is protecting neurons from accumulating DNA damage. The jellyfish model, Cassiopea andromeda, sleeps at night, while the sea anemone model, Nematostella vectensis, sleeps at dawn. But indeed, in both cases, sleep deprivation and mutagens increased neuronal DNA damage and sleep pressure, while sleep, both spontaneous and induced, improved genome stability. As an author of 1, Lior Appelbaum, states: “Neurons are very precious”, “they don't divide, so you need to keep them intact” 3. Which brings us to the deeper question: why don't neurons (normally) divide anymore? At the dawn of metazoan evolution, multicellularity and differentiation towards nerve cells allowed further encoding of the state of the world, including its history. This introduced concepts such as critical periods during embryogenesis, which allowed neural networks to start monitoring different parts of the body. Such mirroring evolved into a full representation of the body in the central nervous systems of ever more complex animals. Alongside this, the development of highly dynamic connections with varying strengths between ever-increasing numbers of neurons enabled the storage of exponentially growing amounts of information about the environment these bodies inhabit. In vertebrates, a single neuron can have between 103 and 104 synaptic connections. Thus, while we can easily replace dedicated cells such as muscle, skin, liver, or blood cells, neurons are exceptional because the information they help to encode is eroded upon their loss. As such, they will become ‘irreplaceable’. The new results with cnidarians demonstrate that such a threshold is reached surprisingly quickly. This all fits nicely with an informative recent review about “signaling roles for astrocytic lipid metabolism in brain function” 4. Here, the authors show how the well-established metabolic model for brain energy homeostasis, the astrocyte-neuron lactate shuttle hypothesis, is being extended with signaling roles for lactate and fatty acid oxidation (FAO) derived ROS in proper brain functioning and maintenance. In passing, they also implicitly demolish two long-standing popular ‘explanations’ for the relative absence of mitochondrial (mt) FAO in neurons: (i) FAs cannot pass the blood brain barrier (but astrocytes do abundantly oxidize them); (ii) starved neurons might ‘eat’ their own protective myelin sheaths if they used mt FAO (but the myelin is on the outside covering their axons and made by oligodendrocytes which themselves use this pathway to generate abundant ATP for myelin production). Because glial cells, such as oligodendrocytes and astrocytes (themselves also heterogeneous), can be replenished, while the death of neurons leads to irreplaceable loss, an association of mt FAO with higher rates of internal ROS formation and cell death seems a much more likely explanation 5. Indeed, the review mentions the controlled generation of ROS by mt complex I for signaling purposes by astrocytes using FAO, though it stresses the role of weaker complex I—III interactions, instead of reverse electron transport from higher levels of reduced ubiquinone. Both probably play a role; see also 6, 7. Nature hardly ever deals in absolutes, and neurons seem to display occasional mt FAO under special conditions (e.g., when confronted with ‘peak demand’ or in their nerve terminals), but “…astrocytes are overall much more competent in FAO than neurons” 4. The review sketches an almost poetic image of oligodendrocytes and astrocytes protectively ‘coddling’ neurons, in a highly complex, finely tuned, balancing act of mt FAO and associated ROS generation. Vertebrate brains have come a long way indeed, but some of the crucial selection forces can already be discerned in their humble beginnings.
Dave Speijer (Thu,) studied this question.