In humans, sleep spindles are generated by local thalamic pacemakers

In the 1930s, Alfred Loomis performed electroencephalographical (EEG) recordings on his guests during a nap or early night sleep in his private lab at Tuxedo Park and stumbled upon a peculiarity during sleep: bouts of oscillations at 10–15 cycles per second, lasting for up to 3 seconds, which he called 'spindles'. We now know that spindles occur specifically during non-rapid eye movement (NREM) sleep, which constitutes the majority of the first sleep cycles. In EEG recordings, spindle oscillations seem to co-occur in synchrony over large portions of the scalp. A new report by Bastuji and collaborators in this issue (Bastuji et al. 2020) presents new evidence that spindles are sometimes localized within restricted brain networks and are much less widespread than previously believed. The cellular and network properties leading to the emergence of spindles during NREM sleep are now well established (see Fernandez & Lüthi, 2020 for review). The pacemaker of this oscillation is the thalamus, a subcortical structure that is crucial for routing sensory information to the neocortex and for coordination of activity across different cortical areas. Thalamic neurons rhythmically modulate their neocortical targets. In turn, the neocortex sends divergent feedback to the thalamus and these reverberating loops have long been assumed to underlie the synchronization of spindles over a large portion of the neocortex. One caveat of this model is that most evidence so far has been collected from anaesthetized animals, in which spindles may be artificially over-synchronized. Several observations of humans had previously suggested that spindles might instead be a local phenomenon. Specifically, recordings using magnetoencephalography, which is believed to capture the activity of more localized electrical sources than EEG, revealed that spindles were desynchronized in the neocortex (Dehghani et al. 2010). In fact, EEG may reflect a summation of electrical waves over large areas of the neocortex, obscuring the fine-grained spatial distribution of spindles. Direct evidence then came from intracerebral recordings of humans. In some cases, patients suffering from drug-resistant forms of epilepsy are implanted with depth electrodes to localize the epileptogenic zone. For some patients, seizures are rare and their brain signals between seizures, especially far from the epileptogenic zone, are assumed to be broadly similar to the activity in healthy subjects. By leveraging this unique opportunity to get access to the inner dynamics of the brain, it was shown that spindles (as well as slow waves) in the neocortex were often observed only on nearby electrodes, indicating that at least some spindles may be largely local phenomena (Nir et al .2011). Bastuji and collaborators went one step further (Bastuji et al. 2020). Using depth recordings directly from the thalamus in epileptic patients, they report what had long been suspected: that spindles are also sometimes local in the thalamus. This was the case for about half of the spindles detected, whereas other spindles occurred in concert across multiple thalamic locations. Interestingly, local thalamic spindles were more often than not associated with local spindles in the neocortex. These observations provide one of the most compelling pieces of evidence so far that during sleep, different modalities in the brain, associated with different thalamocortical loops, can act independently of each other. The sleeping brain is thus not so much of an orchestra playing the harmonious symphony that one can listen to with scalp EEG but, instead, thalamocortical loops are soloists playing their own scores. Of course, these networks are not totally independent of each other, as reflected by the occasional synchronous non-local spindles, and this coupling may be what enables the binding of multiple information streams during wakefulness, giving rise to our conscious life. Conversely, the relative independence of these networks gives rise to the flexibility and versatility of the complex cognitive functions supported by the thalamus and the neocortex. Bastuji and collaborators have analysed recordings from the posterior thalamus, which is mostly associated with sensory networks. It remains to be answered whether the proportion of local versus non-local spindles is the same in other parts of the thalamus; for example, in its anterior pole (associated with limbic functions). Furthermore, how often spindles co-occur between distal parts of the thalamus is also unknown. Further studies, both in animal models and human epileptic patients, will be necessary to provide a full picture of the phenomenon. Although they have now been observed for almost 90 years, the function of spindles remains as much of a mystery as, some would argue, most oscillations in the brain. Yet an emerging view, supported by ample correlative, interventional and computational studies, is that they are crucial for plasticity processes both during development and learning at adult age (Peyrache & Seibt, 2020). For example, learning a new motor skill leads to an increase in slow waves and spindles during subsequent NREM sleep specifically over the motor cortex and not elsewhere, and this is possible only if neuronal activity in different brain networks is relatively independent, as suggested by the observation of Bastuji et al. One promising line of research is that the spatiotemporal distribution of spindles during NREM sleep is a potential biomarker of various neuronal disorders (Fernandez & Lüthi, 2020), reflecting a disorganization in the underlying brain networks at the neuronal and circuit levels. It is possible that overnight sleep EEG will become a standard exam in yearly check-ups, especially since cheap and high-quality EEG recording systems are becoming available. Still, much work is needed to bridge the gap between these various levels of observation. The study of intracerebral signals in human epileptic patients is a big step forward. The author has no competing interest. This work was supported by a Canadian Research Chair in Systems Neuroscience (245716), a CIHR Project Grant (155957), a NSERC Discovery Grant (RGPIN-2018-04600), an IRDC grant (108877-001) and an Azrieli Foundation-Brain Canada Early-Career Capacity Building Grant (80070).

Récupéré en direct depuis OpenAlex et désinversé. Les résumés ne sont pas conservés dans cette base de données : les index inversés représentent 8,6 Go des 9,3 Go de texte de la base, et le serveur dispose de 13 Go libres.