Sleep to Survive Predators

Sleep is a reversible behavioral state characterized by a higher arousal threshold and transient disconnection from the external world. Typically, sleeping organisms show lowered behavioral responsiveness to sensory stimuli (such as sound, touch, and smell) in the experimental environment [1] that contrasts with heightened auditory-evoked neuronal responses recorded in the cortex in rodents and non-human primates [2, 3]. These reports support the hypothesis of sensory gating, as found during sleep spindles, that ultimately preserves sleep from external perturbations. However, this strongly contrasts to the high vulnerability to predation associated with sleep states in the natural environment.

In the wild, a consolidated sleep is essential for the survival of organisms and the adaptation of the species. The strong conservation of sleep across evolution argues for an important physiological function as revealed by its implication in brain development, brain metabolic clearance, synaptic plasticity, and the optimization of behaviors [4]. A common view is that sleep has evolved as an optimal adaptation to the needs of the species (e.g., energy resources and reproduction) and the environmental constraints (e.g., predation, climate, and resources). In fact, predation is an important determinant of sleep architecture as species living in high-risk environments spend less time asleep. In an original comparative study, Allison and Cicchetti [5] showed that species subjected to higher risks of predation, including mice and rabbits, display less non-rapid-eye movement (NREM) and rapid-eye-movement (REM) sleep than predator species such as carnivores. In agreement with these findings, species sleeping in open environments and herbivores exhibit less REM sleep than those sleeping in safe places or shelters [6–8]. As an evolutionary adaptation, some mammals and birds transiently shift to relatively ‘light’ or no sleep in response to increased predation risk or environmental constraints (e.g., migration). For instance, capuchin monkeys increase vigilance when behaving near the ground level where they are exposed to terrestrial predators, and sleep, or rest, in high forest strata. An interpretation of this adaptation to the temporal and spatial features of vulnerable states—e.g., a reduction of sleep during the foraging period—further reflects an evolutionary strategy to minimize sleep-related exposure to threats such as predation.

Snyder [9] proposed that “the REM state serves a "sentinel" function, bringing about brief but periodic awakenings to prepare the organism for immediate fight or flight”. It is postulated that “such physiological mechanism would provide maximal security from external danger compatible with minimal disturbance to the continuity of sleep”. This hypothesis is supported by the reports that animals, including humans, are more alert after awakening from REM than from NREM sleep, as revealed by a difference in sleep inertia [10]. Nevertheless, the neural mechanisms of the sentinel hypothesis are neither established nor experimentally investigated.

Recently, Tseng and collaborators provided experimental evidence in support of this hypothesis [11]. The authors reported that corticotropin-releasing hormone (CRH)-expressing neurons in the medial subthalamic nucleus (mSTN) are strongly activated during predation. These neurons promote rapid arousal in response to olfactory predatory cues during REM, but not NREM, sleep and successful defense against life-threatening events (Fig. 1A, B). In vivo chemogenetic and optogenetic inhibition of mSTN-CRH neurons prolongs the latency to awakening from REM sleep and increases the reaction time to predatory stimuli. On the other hand, optogenetic stimulation of mSTN-CRH activity during REM sleep promotes waking in response to olfactory predatory cues (Fig. 1B). Strikingly, the same manipulation stabilizes REM sleep in the absence of predatory odor (Fig. 1A). This suggests that putative inhibitory inputs to the downstream targets of the mSTN-CRH circuit maintain REM sleep, and presumably its functions, under safe conditions. Similar behavior occurs when mice are chronically exposed to threatening conditions. Interestingly, this sleep adaptation consists of enhanced total REM sleep time but shorter duration of REM sleep episodes and global sleep fragmentation (Fig. 1C). These findings are interesting because they reconcile the sentinel function hypothesis with previous reports of shortened sleep in species living in hostile environments. In this context, sleep fragmentation may represent an important evolutionary strategy to preserve the continuity of sleep and its functions, while increasing the probability of escape from predation in particular for gregarious species. Hence, sleep fragmentation, together with the increased duration and number of REM sleep episodes promotes higher arousability or a higher arousal level at awakening. This mechanism ultimately optimizes the behavioral reaction to threats. In the present study, the authors hypothesized that, when facing an inescapable but not imminent threatening condition (e.g., sustained rat exposure), animals tend to maintain sleep, possibly to satisfy physiological sleep needs. However, in contrast to NREM sleep, more time is allocated to REM sleep presumably due to its protective function against acute threatening cues (as discussed above) without awakening the animal and disturbing sleep [12]. This interpretation raises two important questions as to why this alerting mechanism is absent during NREM sleep and what the causes/consequences of increased REM sleep propensity are in regards to the other functions controlled by this state.

Fig. 1.

A Optogenetic stimulation (10 Hz) of mSTN neurons does not induce wakefulness during REM sleep, but stabilizes it, in the absence of predator cues. B Stimulation of mSTN neurons induces rapid arousal from REM sleep, when a predator cue is presented conjointly. C Sustained predator exposure induces a significant increase in total REM sleep. The REM sleep adaptation consists of increased total REM sleep time but shorter durations of individual REM sleep episodes and sleep architecture fragmentation. The fragmentation results from an increase in transitions between wake, NREM, and REM sleep states, with increased episodes of each state.

In the present study, the authors showed that olfactory predatory cues induce awakening during REM but not during NREM sleep. This is a surprising result as previous studies evidenced either similar sensitivity of both states to auditory-evoked awakenings [2], or a lower response to olfactory [13] or tactile [14] stimuli during REM sleep than during NREM sleep.

A recent study on the neural mechanisms mediating reduced responsiveness to external sensory events showed that noradrenaline release from the locus coeruleus (LC) plays a key role in mediating the sensory-evoked arousal threshold during sleep [2]. Minimal optogenetic LC activation increases the probability of auditory-evoked awakening during both NREM and REM sleep, while it is decreased upon optogenetic silencing during NREM sleep. Due to the evolutionary importance of survival, awakening upon external threat or stimuli is likely to implicate other, possibly overlapping, mechanisms including systems producing arousal transmitters/modulators (e.g. histamine, acetylcholine, hypocretins/orexins, and dopamine). Interestingly, auditory-evoked responses in the primary auditory cortex are largely preserved, if not increased, across sleep states, in particular NREM sleep [3]. Although this and other studies only exposed animals to neutral, non-threatening cues, they suggested that alerting systems, similar to mSTN-CRH neurons during REM sleep, also contribute to successful defense against life-threatening events during NREM sleep. Furthermore, the olfactory cues used by Tseng et al.[11], and odor in general, are not relayed via the thalamus as are other somatosensory stimuli (i.e., touch, vision, and audition) suggesting an alternative neural integration during sleep. Interestingly, exposure to odorants during high-frequency EEG activity elicits robust spike responses in olfactory cortex neurons while weak responses occur during slow EEG activity, under anaesthetized conditions [15]. In this view, it is reasonable to hypothesize that the activity of olfactory circuits have a functional bias for specific brain states (e.g. REM sleep), although further work is needed to understand whether those alerting systems are modality-specific or not, as shown for mSTN-CRH neurons [11].

Last, whether the REM sleep gain is the consequence of an evolutionary strategy to improve reaction time during awakening upon chronic threat, or whether this is a pathologically-relevant adaptation remains unclear. A large body of evidence has shown that sleep, in particular REM sleep, plays a role in the long-term consolidation of recently-acquired memories. Accordingly, increased REM sleep duration may facilitate an adaptive dampening of emotional reactivity to stressors and may represent a common sleep adaptation in animals and human subjects with stress-related psychiatric disorders. Thus, understanding the neuronal underpinnings of the relationship between REM sleep and chronic stress is a major challenge in biomedical research.

Future investigations of the brain mechanisms underlying sleep-wake control in naturally relevant environments, in particular during danger, migration, or hibernation are crucial for the understanding of behavioral adaptations to environmental niches, and ultimately, natural selection in species. Translations of this knowledge to humans and society will undoubtedly improve the detection, prevention, and treatments of sleep disorders together with sleep perturbations associated with other brain disorders including dementia, schizophrenia, and depression.

Conflict of interests

No conflict to declare.