Non-REM and REM/Paradoxical sleep dynamics across phylogeny

Abstract

All animals carefully studied sleep, suggesting that sleep as a behavioral state exists in all animal life. Such evolutionary maintenance of an otherwise vulnerable period of environmental detachment suggests that sleep must be integral in fundamental biological needs. Despite over a century of research, the knowledge of what sleep does at tissue, cellular or molecular levels remain cursory. Currently, sleep is defined based on behavioral criteria and physiological measures, rather than at the cellular or molecular level. Physiologically, sleep has been described as two main states: Non-REM and REM/Paradoxical sleep, which are defined in the neocortex by synchronous oscillations and paradoxical wake-like activity, respectively. For decades, these two sleep states were believed to be defining characteristics of only mammalian and avian sleep. Recent work has revealed slow oscillation, silencing and paradoxical/REM like activities in reptiles, fish, flies, worms and cephalopods suggesting that these sleep dynamics and associated physiological states may have emerged early in animal evolution. Here, we discuss these recent developments supporting the conservation of neural dynamics (silencing, oscillation, paradoxical activity) of sleep states across phylogeny.

Introduction

Behavioral sleep is a fundamental state ubiquitously observed in animals ranging from humans to hydra^1–3 (Fig 1A). Recent advances have shown that many critical biological processes are influenced by, or dependent on sleep. These include synaptic restoration, waste clearance, DNA repair, immune function, metabolic regulation, endocytosis, development, growth, endocrine response and more^4–11. Despite these advances, the central role(s) of sleep remains ambiguous, and unlike many other biological processes, sleep does not have clearly defined cellular and molecular functions². But the evolutionary conservation of sleep supports the notion that sleep serves fundamental needs³. In order to properly determine the purposes of sleep, further studies are needed to characterize sleep at the cellular level. Behaviorally, sleep can be considered as a single state of quiescence. However, physiologically, sleep can be defined globally, but separated into at least two main states, Non-REM sleep and REM/Paradoxical sleep (see below and Fig 1B). While for over 50 years it was thought that only mammals and birds have these distinct neurological sleep stages, exciting new research has shown that reptiles, fish, drosophila, octopus, and other invertebrates also display sleep states with analogous features^{12,13,22–24,14–21}.

Figure 1.

What is sleep? A. Sleep is found in all animals carefully studied so far and shares common traits of quiescence, sensory detachment, metabolic reduction, and homeostasis/circadian regulation. B. Neural signatures of sleep have been found in animals ranging from c. elegans to humans. Three different sleep signatures are known: REM/PS is a state of sleep characterized by wake-like activity in the forebrain, accompanied by muscle twitches, eye movements, and cardiorespiratory fluctuations. Non-REM/SWS is defined by slow, high amplitude bursts in the forebrain, with muscle atonia, and reduced cardio-respiratory rate. Silencing is perhaps the core signature of neural sleep physiology, as it accompanies both REM/PS and Non-REM/SWS in large areas of the brain outside the cortex. C. Sleep physiology in tissues across the body. Sleep can be broken down into tissue/cellular signatures that associate with sleep, which have been found from neurons to muscle. Additionally, sleep can be considered for the genetic regulation that occurs in many tissue types across the body. Lastly, sleep can be examined by the tissue, cellular, and molecular processes that are dependent on proper sleep regulation, which have been found in all major tissue types. See ⁷⁸ for review of sleep genetics.

Behavioral sleep is universal

Sleep is an obvious behavioral state, with clearly defined hallmarks that are common among all animals (Fig 1A)^25,26. These defining aspects of sleep were first described over a century ago by Dr. Henri Pieron, and are among the first definitions of animal behavior in scientific literature²⁵. The formalized behaviors of sleep include specific postures (such as laying down), a place preference (such as a bed or den), reduction of locomotor activity (saving energy and metabolic reduction), and elevated sensory arousal threshold (to provide sensory dissociation from the environment). Dr. Irene Tobler, whose pioneering work demonstrated that even simple animals such as insects sleep, expanded these original definitions to include a fifth criteria, that animals will recover sleep after deprivation, suggesting that sleep is a homeostatic process^26,27. Therefore, behavioral sleep can be described as a largely homogenous state of sensory detachment and quiescence, interrupted sporadically with active episodes of muscle twitches and cardiorespiratory fluctuations.

The sleeping brain: lessons from man to dragon

While behavioral sleep provides a simple framework to characterize the waking or sleeping state of an animal, it does not provide information about the internal physiology of that animal. Electrophysiological recordings of brain activity associated with behavior date back to the 19^th century, and the modern electroencephalogram (EEG) was founded in the early 20^th century^28–30. EEG records the electrical activity of large ensembles of neurons at the surface of the neocortex. By the 1950s, EEG was combined with electromyogram (EMG) to record muscle tone, electrocardiogram (ECG) for heart rate, and electrooculogram (EOG) for eye movement, which together make up polysomnography (PSG)³¹. Using such methods, two dominant stages of sleep have been described: slow wave sleep (SWS), the deepest stage of Non-REM sleep, which is defined by synchronized slow cortical activity conjoined with reduced muscle activity throughout the body; and rapid eye moment sleep (REM) also fittingly described as paradoxical sleep (PS), characterized by atonia, desynchronized EEG (reminiscent of waking activity), along with episodes of myoclonic tremors including rapid eye movements.

While polysomnography was first conducted in humans, other early sleep studies applied this technique to a range of mammalian and avian species such as cats, dogs, goats, mice, rats, chickens, pigeons and more^31–36(Fig 1B). These studies, without failure, found both Non-REM/SWS and REM/PS states during behavioral sleep that share many similarities to polysomnography studies performed in humans. Since then, it has been widely accepted that mammals and birds have similar sleep states^26,37. More recently, polysomnography has been applied to non-avian reptiles, including the bearded dragon, Argentine tegu, and crocodile^14,15,38. All of these reptiles show similar SWS activity to birds and mammals, termed high-voltage sharp waves (HShW). In the bearded dragon, HShW occur across the dorsal ventricular ridge of the pallium, a telencephalic structure homologous to the mammalianand avian layered pallium (neocortex), and arise out of the deeply conserved claustrum, providing some of the first organization for the neural organization of slow wave sleep³⁹.

REM/PS is an inherently perplexing state with no single label that does justice to describe its full context⁴⁰. Common descriptors such as desynchronized neuronal activity, myoclonic tremors (including rapid eye movement), rhombencephalic activation, all lend support to the paradoxical wake-like activity that underlies REM/PS. In addition to Non-REM/SWS state, both the bearded dragon and tegu display bouts of REM/PS activity, though with significant differences in distribution and timing^14,15. During REM/PS in the bearded dragon, a wide band of activity from 10–30 Hz can be found which mirrors waking activity, during which rapid movements of the eyes were also observed. However, in the Argentine tegu, a single band of 15 Hz was found during sleep that was absent in wake¹⁴. Still, eye, toe, and head tremors were also observed during this sleep state, suggesting its relation to REM/PS. Finding that paradoxical sleep activity is so significantly altered even in closely related species of lizards suggests that the timing and frequency of REM/PS telencephalic activity may matter less than the physiology that characterizes them or may be limited to currently unknown functions.

The overall architecture of physiological activity associated with sleep varies significantly across species. Humans experience bouts of Non-REM/SWS and REM/PS that can last 90–120 minutes, whereas mice sleep in bouts lasting as little as 5–10 seconds^41,42. Brain size and larger cortical structure found in humans have previously been thought to drive the long bouts of consolidated sleep experienced by humans, as even among primates we have the largest brain and cortex in relation to body size⁴³. However, sleep consolidation as a function of brain size or complexity does not appear to correlate throughout nature⁴¹. Several recent studies in bird species show that both ostriches and budgerigars spend more than 25% of sleeping time in REM/PS, similar to the amount of REM found in humans^44,45. Alternations between Non-REM/SWS and REM/PS in the bearded dragon occurs in rigidly consistent 80 second bouts, while in the tegu lizard REM/PS episodes last only 4 seconds, and their sleep cycle displays no periodicity^14,15.

Why the telencephalic pallium/neocortex undergoes different activities during sleep is not clearly understood, but it is thought that each of these states has unique functional relevance⁴⁶ (Fig 1B). One such contention is the Non-REM/SWS is associated with cellular processes, such as growth, stress regulation and repair, metabolic clearance, and synaptic homeostasis. While REM/PS activity is associated with circuit-level processing, such as neuronal plasticity, which is critical for learning and emotional processing^40,47. When considering animals that sleep in bouts lasting only seconds such as mice or tegu, one wonders what functional relevance these short periods could entail, since critical molecular and cellular process such as transcription, translation, and endocytosis needed for synaptic remodeling and potentiation occur over minutes rather than seconds in cells⁴⁸.

The variation in the organization of Non-REM/SWS and REM/PS bouts found across species suggests that the architecture of these states may be less important than the dynamics that distinguish them. For example, the rigid 80 second period of Non-REM/SWS to REM/PS found in the bearded dragon differs significantly from the lack of periodicity and shorter sleep bouts in the tegu lizard. Additionally, two distinct sleep states analogous to Non-REM/SWS and REM/PS that have no apparent periodicity have recently been characterized in zebrafish¹⁶ (see below). Intriguingly, these studies show that among vertebrates, and even in two closely related species of lizards, evolution drives significant changes in sleep organization, while leaving the dynamics largely unaltered. This suggests that ultradian cycling and periodicity may matter less compared to the cellular dynamics themselves.

Neural and muscular signatures of sleep in fish

With over 33,000 species, fishes are by far the most numerous group of extant vertebrates, and are present in vastly different ecological niches ranging from small desert springs to the deep ocean. While behavioral sleep has been well-described in numerous fish species for many decades, the presence of cellular/brain sleep activity has, until recently, remained unknown^49–53. Because of mammalian, avian and non-avian body opacity, polysomnography in these species only allows the superficial recording of brain and body activities by placing electrodes at the surface of the cortex/pallium and skin or by invasive surgical placement to access deeper neural structures. In contrast, several species of fish are transparent allowing total access to every cell in the body during sleep in a vertebrate⁵⁴. Leveraging the advantage of whole body transparency, the recently developed fluorescence-based polysomnography (fPSG) in zebrafish uses brain wide and voluntary muscle Ca²⁺ imaging combined with heartbeat and eye movement recording to allow for constant monitoring of muscle tone, heart rate, and eye movement in conjunction with whole brain imaging¹⁶. fPSG of sleeping zebrafish revealed two main types of sleep dynamics termed slow bursting sleep (SBS) and propagating wave sleep (PWS) sharing commonalities with Non-REM/SWS and REM/PS respectively. Like mammalian slow wave sleep, slow bursting sleep in fish is characterized by slow synchronous oscillation of neuronal firing in the dorsal pallium of the telencephalon, combined with low muscle tone and slow regular heartbeat. Similar to mammalian SWS, SBS activity is proportional to sleep deprivation and can be induced by H1R antagonists¹⁶. In contrast to SBS, propagating wave sleep is distinguished primarily by muscle atonia and twitches, brainstem activation and a posterior-anterior wave of neural activity propagating from the pons. Once this wave has dissipated, PWS is characterized by wake-like dysynchronous low amplitude neural dynamics within the pallium, muscle atonia and irregular heartbeat. These features are reminiscent of the ponto-geniculo-occipital waves observed at the onset of REM/PS episodes as well as the loss of voluntary muscle tone and cardiac arrhythmia observed during REM/PS. Further, like REM/PS, fish PWS is modulated by the melanin concentrating hormone signaling pathway and can be induced or blocked by cholinergic agonists and antagonists respectively, suggesting conservation of neurotransmitter control of this state.

The strong parallels observed between SWS vs. SBS and REM/PS vs. PWS suggest a broader conservation of the sleep states across vertebrates. Further studies in other fish species will shed more light on the potential generalization of these first observations. For instance, Astyanax mexicanus is a growing model system to explore behavioral and morphological evolution⁵⁵. These fish exist in drastically divergent populations that include both long sleeping surface fish and short sleeping cavefish. Cavefish are naturally short sleeping, likely due to enhanced hypocretin (HCRT) signaling and HCRT receptor mutations^56,57. Tol2 transgenesis has recently been applied to these fish, and pan neuronal Ca²⁺ indicators now exist, making it possible to apply fPSG to this species and test if cavefish have similar sleep signatures^58,59. This species could facilitate cellular studies of sleep biology, evolution and functions such as resiliency mechanisms for sleep loss and sleep efficiency.

Evolutionary conservation of cellular sleep dynamics

In addition to the recent reptile and fish reports, there is now growing evidence that physiological sleep states also exist in invertebrates, and that they share many similarities with mammalian/vertebrate sleep^{12,13,17–20,23,24,60,61}. During behavioral sleep, the crustacean arthropod, crayfish, switch from an active posture to laying on their sides²⁰. During these behavioral epochs, they display Non-REM/SWS-like activity of slower and high-amplitude electrical activity compared to waking state. Similarly, Non-REM/SWS and REM/PS-like activity observations have also been made during sleep in terrestrial arthropods, Drosophila. Several independent studies have demonstrated, through both behavior and physiology, that Drosophila have two stages of sleep that converge with mammalian sleep definitions. During a deep behavioral phase of sleep reminiscent of Non-REM/SWS, Drosophila are able to increase efficiency of waste clearance similar to mammals^8,62. Further, Drosophila have been shown to have brain-wide down state during sleep that is accompanied by slow oscillations of neuronal activity within the ellipsoid body, a brain area likely analogous to the vertebrate pallium^61,63. These oscillations occur in the 7–10Hz, a range shared with mammalian Non-REM. Very recently, using calcium imaging of thousands of brain neurons, both slow-wave delta band activity and paradoxical sleep have been reported in Drosophila during bouts of behavioral quiescence^18,64. During spontaneous sleep in Drosophila, brain activity transitions from wake-like (paradoxical, active sleep) to silence (quiet sleep), and neural connectivity is reduced¹⁸. Together, these studies uncovering neural oscillation, silencing and paradoxical activity reveal that different species of arthropods have unique stages of sleep within defined sets of neural circuits, suggesting that these cellular features of Non-REM/quiet and REM/active sleep are likely conserved in invertebrate species.

In contrast to the fruit fly and crayfish, C. elegans is not an arthropod but an evolutionarily ancient nematode with a simple nervous system made up of 302 neurons. A developmental sleep state, termed lethargus, occurs at the conclusion of every molt stage which fulfills the behavioral definition of sleep⁶⁵. During lethargus, there is a global reduction in the activity of the nervous system, with 75% reduction in active cells²³. Importantly, cells that remained active were largely sleep promoting, such as GABAergic neurons, and the sleep-promoting interneuron (RIS), indicating a positive sleep valence for neuronal silencing at the network level. Sleep can also be found in adult stages of C elegans, and recently platforms have been developed to simultaneously record behavior and neural activity, providing the opportunity to track single neurons during behavior in fully developed adult worms, with similar findings of network-wide silencing⁶⁶. The global down state of C. elegans sleep could be a circuit-wide version of down state of Non-REM/SWS found in vertebrates and suggests that this system-wide neural silencing “down” state may be an ancient precursor to the “up-down” state present in vertebrate pallial Non-REM/SWS.

Neural silencing could be the unifying feature of sleep from C elegans and Drosophila to zebrafish and mammals^16,18,23,67. In mice undergoing SWS, there is large overall downscaling of neural activity, yet with cell-type specific increases in activity⁶⁷. However, during REM/PS, the same neural circuits are even less active overall, suggesting that even within cortical regions, a core function of sleep is silencing, despite the long held supposition of elevated activity in REM/PS^14,68–70. It is likely that widespread silencing allows for metabolic recovery, waste clearance, DNA repair, and synaptic pruning^{7,8,62,71–73}. It is provocative to consider that neuronal silencing could be the salient feature of sleep function, creating a coordinated “quiet” cellular environment throughout the central nervous system allowing for coordinated cellular repair, homeostasis, and modification.

In addition to well established models such as Drosophila and C. elegans, octopus and cuttlefish have also shone a new light on the possible conservation of sleep states in invertebrates. While PSG-like sleep measurements have not yet been reported in cephalopod studies, behavioral sleep methods have shown the striking commonalities of sleep physiology, suggesting the broad conservation of distinct sleep states in all animals^19,24. In cuttlefish and octopus, sleep can be divided into two broad categories of “quiet sleep” and “active sleep” that are reminiscent of Non-REM/SWS and REM/PS^12,13,19. Cephalopods are capable of camouflaging themselves to match their environment, this phenomenon is under neuromuscular control, raising the possibility that chromophore activity can serve as a proxy to underlying neural activity⁷⁴. Cuttlefish display two phases of sleep: quiet sleep with little to no chromophore activity, and active sleep which is characterized by body surges, eye movements, and intense non-camouflaging chromophore patterns¹⁹. Similar states have recently been described in octopus^12,13,24. Video recordings showed that octopuses spend over 50% of the day in quiet sleep with little to no movement, and no chromophore activity, which was sometimes followed, in an ultradian manner, by bouts of active sleep which made up 0.5% of the recording period¹³. The functional relevance of active sleep in the octopus is puzzling as it makes up less than 8 minutes of the entire day and occurs in bouts lasting as little as 30 seconds. This order of magnitude difference between quiet and active sleep further supports the biological importance of conserved quiet sleep compared to active sleep.

While still incomplete, our current definitions of sleep have allowed for sleep to be captured among vertebrates and invertebrates alike. Basic analogues of quiet and active sleep, as well as Non-REM/SWS and REM/PS have now been found from humans to fish to Drosophila to octopus, which suggests that a universal definition of sleep maybe established based on conserved sleep specific cellular dynamics rather than physiological measures and behavioral criteria. If true, one would expect to find such sleep-specific cellular landmarks even in animal species devoid of brains.

Broader aspects: does sleep require a brain?

Cnidarians are simple diploblastic animals that do not possess a central brain, whose nervous system is instead made up of a diffuse net⁷⁵. If sleep is present within members of this phylum, it is reasonable to conclude that sleep predates the evolution of a centralized brain and raises questions such as: what is the simplest form of nervous system required for sleep? Is a brain required for sleep at all? And what is the role of sleep is outside of the nervous system? Several recent studies have found a sleep state in two different species of cnidaria^21,22. Cassiopea, the upside-down jellyfish, and Hydra, the simplest animals within the cnidarian phylum both show all behavioral definitions of sleep. Further, pharmacological experiments in both species have shown that several molecular pathways are conserved in the regulation of their sleep. Histaminergic antagonism and melatonin signaling both increase sleep in Cassiopea, while melatonin and GABAergic pathways both promote sleep in Hydra, in a similar fashion to mammalian phenotypes. The logical next question is whether distributed/non-cephalized nervous systems share similar sleep-related global neural and muscular dynamics as animals with centralized brains.

The diffuse nerve nets that make up the nervous system of cnidaria do not form a central ganglion, but neural principles such as action potentials, peptidergic, and transmitter signaling are already in place^76,77. This suggests that the need for sleep does not rely upon a centralized brain. Rather, the need for sleep is clearly encoded within cells, and may be driven by nodal arrays, or even by single cells across many different tissue types that require homeostatic refreshment following the energy expenditure and stress accrued during wake (Fig 1C). Simple organisms such as hydra or jellyfish may be essential to dissect what sleep is at a cellular level by determining if they share any core neural features such as silencing, oscillation, or paradoxical activity.

Conclusion

Behavioral sleep has been observed in all animals examined so far, even in species that do not possess a brain. For decades, distinct physiological sleep states were believed to be limited to mammals and birds. Over the past five years, Non-REM and REM/PS-like dynamics have been more broadly reported across vertebrates and invertebrates alike, challenging our current conception and definition of sleep states. Since even the physiological definition of REM sleep is debated⁴⁰, using cellular dynamics as a comparison across evolutionary distant species could provide a more precise understanding of sleep states. Common features found across the animal kingdom, such as silencing, slow synchronous oscillation, wake-like/paradoxical neuronal and muscular activity allow the field to move from behavioral/physiological toward a cellular definition of sleep. Importantly, while the telencephalon is dispensable for sleep, most current physiological sleep definitions are biased towards a neorcortex, /pallium expression of sleep where oscillation and paradoxical activity are measured in vertebrates. Comprehension of sleep cellular dynamics across the entire vertebrate and invertebrate brain/nervous system and body will reveal whether sleep is only a two-state period or a more complex phenomenon that can be compartmentalized by brain region, neuronal network, and distinct cellular activity. Finally, it remains to be seen if sleep is present in animals that possess no nervous system at all, such as Profera sponges or placozoans which are known to alter their behavior in a circadian-dependent manner. A comprehensive view of sleep as an event relevant to the entire body, and not just the brain/nervous system, will be integral for the future of understanding what sleep actually is.

Highlights.

• Quiescence and sensory gating epitomize behavioral sleep in all animals.

• Common neural patterns of sleep states exist from mammals to invertebrates. These include silencing, oscillations, and paradoxical wake-like activity.

• Animals lacking a centralized nervous system such as hydra and jellyfish have been shown to sleep, indicating that the need for sleep predates a centralized brain and opens the question of whether similar neural dynamics exist in nerve nets.