SLEEP

REVIEW So, what is sleep for? A widespread, but in

our view mistaken, belief is that if a detriment
The inescapable drive to sleep: in behavior or well-being can be identified as a
consequence of sleep deprivation, then this
Overlapping mechanisms of sleep and sedation provides evidence for the reason that we sleep.
However, it seems more likely that sleep pro-
Nicholas P. Franks and William Wisden vides some fundamental “housekeeping” or
metabolic functions that are necessary for a
Common human experience is that a long period without sleep is unsustainable, and it is also detrimental to healthy brain and body, so that there will be
health and behavior. The powerful and primal urge to sleep after sleep deprivation is intense and seems multiple deficits of one sort or another that
inescapable. The longer we stay awake, the more we feel the need to sleep, and however much we resist, we will arise when sleep is denied (Fig. 1A). By “house-
inevitably succumb. Although it is obvious what benefits derive from other common and strong physiological keeping,” we mean basic structural or meta-
drives, such as hunger, sex, and thirst, it is less obvious what drives us to sleep and what benefits accrue. bolic processes that are essential for the normal
Understanding the biochemical or circuit basis for the sleep drive could enable the benefits of sleep to be functioning of the brain. The most important
artificially stimulated with a new generation of sedative drugs. constraint when considering what these func-
tions might be is that they cannot, apparently,

W
occur in the waking brain; restful waking does
hy do we sleep? The evolutionary ad- Because we all experience sleep, there is a not satisfy our need for sleep. Much like the
vantages of remaining permanently common understanding of what the sleeping cleaning teams who move into empty offices
awake and alert seem so obvious that state entails for us subjectively, which we are during the night and whose work would be al-
sleep must provide some essential need. comfortable extrapolating to other animals, at most impossible during the daytime bustle, some
Dolphins, fur seals, and whales alter- least to other mammals. Sleep can be mea- essential and restorative process is underway
nate their sleep in their cortical hemispheres sured using the external electrical signatures after we slip into sleep, when normal brain func-
so that they can keep swimming: One half of of brain and muscle (5) and can, in vertebrate tion is at least partly suspended. The need for un-
the brain “sleeps” while the other is awake (1). animals, be categorized as non–rapid eye move- consciousness is perhaps the key thing to explain.
This observation alone seems to highlight the ment (NREM) and REM sleep, or at least NREM- Mechanisms that stand out for being in-
importance of sleep, at least for mammals. De- like and REM-like. However, depending on compatible with consciousness would be those
spite intense research, why we sleep remains species, these sleep states differ in their phys- that involve structural changes in the brain.
one of the most baffling questions in neuro- iological properties and timing (5). The detailed For example, the glymphatic hypothesis, that
science. In this article, we discuss why we might properties of any one sleep state, electroenceph- NREM sleep allows the clearance of metabo-
sleep, what mechanisms force us to sleep when alographic (EEG) pattern, or calcium-oscillatory lites from the brain (11), posits that the end feet
we are sleep deprived, and whether certain an- activity are unlikely to be critical. Some other of astrocytes shrink to enhance glymphatic flow
esthetics and sedatives hijack this natural drive common, as yet unknown, property of the sleep (11). Similarly, any remodeling of synapses to
to exert their effects on human consciousness. state is probably the key ingredient for why rebalance network activity acquired during
sleep is essential. Much as birds and mammals wakefulness might preclude cognitive activity
Do we have to sleep and if so, why? use different circuitry and brain organizations or awareness and might only be possible “off-
Although it is often stated that “all animals to carry out cognitive tasks with high effective- line” (12). Another factor to consider is tem-
sleep,” it is not obvious, or proven, that all types ness, sleep is likely to be an essential emergent perature. During sustained NREM sleep, brain
of animals sleep or sleep in the same way or for property regardless of how the brain is or- neocortex temperature in mice drops by ~2°C
the same purpose. Nevertheless, it seems clear ganized (6). The requirement for sleep could (13, 14) (Fig. 1B). In fact, specific hypothalamic
that sleep is good for us humans and is in some originate at the cellular level, with neurons fa- circuitry exists to cool the brain and simulta-
way restorative. Moreover, we know from per- tiguing with time spent awake, analogous to neously induce NREM sleep (15). It has been
sonal experience that missing sleep is detrimen- the metabolic fatigue of skeletal muscle with proposed that a key function of sleep is to pre-
tal. In the short term, total sleep deprivation exercise. But it is not obvious whether all parts serve energy (16). For humans, however, this
decreases cognitive ability and mood (2). Twenty- of the brain need to sleep; circuits that drive temperature decrease with sleep would result
four hours of enforced wakefulness has a greater breathing, for example, do not seem to rest. in only a modest saving in metabolic energy,
detrimental effect on driving ability than con- Notwithstanding the many variations of equivalent to about two slices of bread (16). A
suming the legal limit for alcohol (3). There- sleeping, is there a core reason to sleep that is lower temperature with sleep could nonethe-
fore, something is clearly deteriorating in our so fundamental that sleep is obligatory? Al- less be required for an unknown restorative
brain function when sleep is missing. What though there are many activities that occur process. This was first suggested >30 years ago
about the long-term effects of sleep loss? A during sleep but occur less often during waking, (17) and might be linked to synaptic remodel-
recent study tracked nearly 8000 UK govern- few of these activities are likely to be the fun- ing. Cooling during NREM sleep induces the
ment employees for 25 years, from middle age damental and primal reason that we must sleep. expression of the cold-inducible RNA-binding
through to their early seventies. The results A telling fact is that of the hundreds of mouse protein and RNA-binding motif protein 3 genes
showed a higher dementia incidence in people mutants generated, no mutant mouse has yet (13), and the proteins encoded by these genes
between 50 and 60 years of age who had been been discovered that does not exhibit NREM could be required for structural remodeling
chronically short sleepers (6 hours a night or sleep (7). Many mutations affect the depth of (they are also induced in some hibernators).
less, self-reported) (4). Thus, in the long term, NREM sleep or its timing (e.g., circadian mu- Furthermore, cooling even by only a degree or
less sleep correlates with increased pathology. tants), but so far all have NREM sleep (7). The two would prolong inhibitory postsynaptic cur-
same picture emerges with circuit manipula- rents by about the same extent as do many
tions. Lesions of sleep-promoting centers can general anesthetics at sedative doses (18) (Fig.
Department of Life Sciences and UK Dementia Research permanently reduce the amount of sleep in ro- 1B); this would make normal cognition essen-
Institute, Imperial College London, London SW7 2AZ, UK.
Correspondence: [email protected] (N.P.F.); w.wisden@ dents, yet some sleep always remains (8–10). So tially impossible, so a sleeping state would be
imperial.ac.uk (W.W.) far, no cell or circuit lesion has removed all sleep. necessary for these processes to occur. Finally,

Franks et al., Science 374, 556–559 (2021) 29 October 2021 1 of 4

the delta and theta electrical oscillations that REM sleep always follows NREM sleep, their particularly the higher end of this band of fre-
occur during NREM and REM sleep could be functions are linked. For example, REM could quencies (14), markedly increases after sleep
important for rebalancing the network during be an algorithm that tests whether the restor- deprivation but rapidly rebounds. However,
sleep (19), and such oscillations may also be ative function of NREM has been completed. many hours more are needed for the actual
incompatible with normal cognitive function. If it has, we wake up. An important question sleep loss to be restored, implying that addi-
A good example of how oscillations can block to solve is, what determines the timing and tional, potentially restorative processes that are
cognition is provided by the anesthetic keta- sequence of the NREM and REM cycles? not reflected in EEG delta power per se are un-
mine. This compound elicits a highly selective, derway (Fig. 2). The high delta power immedi-
delta-like rhythm (1 to 3 Hz) in layer 5 pyram- The homeostatic drive to sleep ately after sleep deprivation might nonetheless
idal neurons of the retrosplenial cortex, thus In many animal species, sleep deprivation in- be permissive for the restorative benefits that
inducing a dissociation of stimulus detection creases the drive for both NREM and REM follow, perhaps through brain cooling.
and response (20), which is obviously inconsis- sleep (2). This homeostatic sleep drive strength- We can consciously fight sleepiness to a cer-
tent with a normally functioning waking state. ens the idea that some important restorative tain extent, a top-down control from the neocor-
In contrast to NREM sleep, REM sleep may process is underway and is the reason that we tex imposing its will on the “sleep centers,” and
not be essential for life. It is possible to genet- must “catch up” on lost sleep. This catching up, we might be determined to stay awake, but even-
ically ablate all REM sleep in laboratory mice however, is not just about quantity but also tually we give way to sleep. After a certain period
with no apparent ill effects (21). On the other quality, at least for NREM sleep. If we miss one of intense cortical activity, local delta oscillations
hand, and perhaps unexpectedly, people who night’s sleep of 7 hours, say, our next night of “break out” in areas of the neocortex even if ani-
report good and refreshing sleep have good sleep is likely to be subjectively “deeper,” as well mals are behaving as if they are awake (23). One
REM sleep continuity, whereas consolidated as potentially longer. Although it remains mys- might then ask, if this restricted “sleep” is pro-
NREM appears to be less important for the per- terious what this deep sleep is providing phys- viding restorative benefits locally, then why does
ception of a good night's sleep (22). Theories for iologically, a pervasive concept is that a reliable the whole animal need to shut down and enter
the function of REM sleep are not constrained marker of NREM sleep depth in both humans global sleep? Why could the brain not be contin-
by the need for the brain to be offline because and other terrestrial vertebrates is the power ually having smaller domains entering “sleep” so
the brain is active during REM (5) and brain exhibited in the EEG in the delta range of fre- that normal function is still possible at the whole-
temperature rises (13). It could be that because quencies (~0.5 to 4 Hz). The EEG delta power, animal level? This could be because it is necessary
to cool the brain for the restoration to occur, and
this cannot be achieved locally, only centrally. In
A our view, the term “sleep” is best restricted to a
Sleep onset after
SD
brain function

whole-animal behavior, and local short periods

Deteriorating

sleep deprivation (SD)

of increases in delta power are clearly insuffi-
cient for whatever purpose global sleep serves.
37˚C 35.5˚C The sleep drive can also be assessed inde-
pendently of EEG delta power by changes in
behavior. When we are strongly sleep deprived,
we become highly motivated to find a way to
sleep, in the same way that strong thirst and
Normal
sleep onset hunger motivate us to drink and eat. If sleep
WAKE NREM deprived enough, we will do almost anything to
200 200 sleep. Afternoon naps provide additional evi-
EEG Power

EEG Power

dence for an accumulating sleep drive during

100 100 the time spent awake and show that this drive
can be partially dissipated by short periods of
0 0
0 10 20 0 10 20 acute sleep. An afternoon nap delays the onset
Frequency (Hz) Frequency (Hz) of later NREM sleep, which starts off with a
lower EEG delta power than it would have
done had no nap been taken (2). If naps are
Waking Sleeping too long, it can be difficult to get to sleep at all
Time (hours)
in the evening. This implies that there must be
some feedback between the process(es) that
B
Brain temperature

38 Waking
GABAergic IPSC

provides restorative benefits and the mech-

Waking NREM sleep
37 anism that tracks the time awake.
NREM sleep Classical views of how the hunger drive is or-
36 and sedation ganized implicated hunger centers in the hy-
35 pothalamus; for the sleep drive, the preoptic
Time (minutes) Time (ms) hypothalamus was shown to have a particularly
Fig. 1. The most probable reason for sleep is to allow fundamental housekeeping processes to occur. strong sleep-promoting role (24). However, it is
(A) During waking, there is a progressive deterioration of brain function until a point is reached when sleep is now clear that motivational drives for food and
triggered, consciousness is lost, and repair is initiated. This change in vigilance state is reflected in the EEG power water emerge in a distributed way throughout
spectrum, which moves from a low-power broad spectrum to a high-power spectrum that peaks at delta the brain (25). Similarly, the basis for the sleep
frequencies (0.5 to 4 Hz) and is characteristic of NREM sleep. With prolonged NREM sleep, the brain cools. If sleep drive is likely to be more distributed than orig-
is delayed, brain deterioration continues until a point is reached when sleep onset is inevitable. (B) During inally anticipated (Fig. 3A). Indeed, in mice, the
sustained NREM sleep, the brain cools by ~1.5°C, which would prolong inhibitory postsynaptic currents by about circuitry that induces both NREM and REM
the same extent as sedative doses of anesthetics. sleep is itself widely distributed (26). It could

Franks et al., Science 374, 556–559 (2021) 29 October 2021 2 of 4

 SLEEP

well be that all parts of the sleep circuitry dis- reveal the functions of sleep. One way that tumor necrosis factor–a, the levels of which in-
tributed throughout the brain can implement sleep drive might track wakefulness has been crease in the neocortex during sleep deprivation
homeostasis. Hypothalamic preoptic galanin discovered in fruit flies and involves a changing (6). The brain is not the whole story. Certainly,
neurons (27, 28), basal forebrain cholinergic redox state with time spent awake and a change peripheral tissue can signal sleep; unidentified
neurons (29), midbrain ventral tegmental area in the activity of potassium channels, resulting systemic factors from skeletal muscle contribute
(VTA) GABA neurons (30), raphe serotonin neu- in an increased firing rate of sleep-promoting to the sleep drive (39), perhaps explaining why
rons (31), neocortical layer 5 pyramidal neurons neurons (35). In principle, such a mechanism sleepiness can often be induced by exercise (2).
(10), and astrocytes (32) have so far been pro- could work in mammals too. In vertebrate spe- Phosphorylation tracks sleep need. In mice, a
posed as mediators of sleep homeostasis. It cies, however, the actual biochemical processes gain-of-function mutation in the salt-inducible
is unclear what “tiredness signals” are being that track the time spent awake and lead to kinase 3 (Sik3) gene reduces NREM sleep ho-
sensed by this distributed circuitry (see below). increased NREM sleep pressure remain unclear. meostasis as defined by delta power and in-
Because NREM sleep can be dispensed with The simple and compelling idea that there is a creases the amount of sleep (7). This particular
temporarily, for example, if we stay up all night, substance that accumulates during sleep, tracks kinase is widely expressed, including in pe-
this implies that any restorative function of time awake, provides feedback to induce sleep, ripheral organs, and could also act throughout
NREM sleep need not be immediate (e.g., mice and is then degraded during sleep has been pro- the brain, in keeping with the distributed sleep
and humans catch up on lost sleep over many posed many times, starting with Rosenbaum in homeostasis–promoting circuitry. It is not known
hours, if not days). Similarly, frigate birds typ- 1892 (36). Adenosine is the best-known candi- how wakefulness activates Sik3, but the targets
ically sleep very little (0.7 hours/day) during date molecule: It accumulates with time spent of this kinase are numerous. In mice, as wakeful-
10-day periods on the wing yet sleep nearly awake in a few select brain areas, such as the ness progresses, there are steady increases in
13 hours per day when on land (33). Thus, basal forebrain (37), where it (or adenosine tris- phosphorylation (and dephosphorylation) of
whatever is being tracked during sleep depri- phosphate) could be secreted from cholinergic many proteins in the forebrain, including ion
vation and leading to the drive to sleep can be neurons in the basal forebrain and astrocytes channels (40, 41), which might increase or de-
partially overridden (34); therefore, this is not (29, 38). In addition to its many metabolic roles, crease the firing rate of sleep-promoting neu-
the same as faster breathing and heart rate adenosine can act through metabotropic recep- rons or inhibit wake-promoting neurons.
after exertion to immediately resupply oxygen. tors to induce NREM sleep (36), but appealing One generally ignored question is, what de-
Hunger and thirst are also drives that can be though this idea is, definitive proof that aden- termines when we wake up? Although this is
similarly put on hold by the executive control osine is the primary cause of the sleep drive is partly a circadian influence, waking from a re-
centers in the brain, at least for a while. A grad- lacking, especially because the adenosine in- freshing night’s sleep suggests that the ho-
ual process of recovering lost sleep over many creases that correlate with prolonged waking meostatic process of sleep has been completed.
hours suggests anabolic or catabolic housekeep- seem to happen in only a few brain regions, yet Therefore, presumably, the restorative process of
ing processes, i.e., synthesis or degradation. the sleep homeostasis circuitry is widely distrib- sleep is continuously monitored to determine
A better understanding of the sleep drive uted. Many other possible sleep-inducing factors that the brain is now ready to wake. This could
and how the need for sleep is “clocked” might have been identified, including interleukin 1b and be related to the NREM-REM cycle discussed
above because we tend to wake from REM sleep.
Sleep
Sleep deprivation Catching up on lost sleep
Can drugs provide the restorative
deprivation
benefits of natural sleep?
0
Cumulative sleep loss (h)

The drive to sleep after a certain amount of sleep

-1 deprivation is so intense that it resembles drug-
induced sedation. “I'm just going to give you
LPO-Gal something to make you sleep” is an almost uni-
-2
versally used and comforting metaphor when
-3 we undergo an investigative procedure or are
LPO-ΔGal being prepared for surgery. This is probably
Delta power more than a metaphor. Although deep anesthe-
-4
rebound
2.5 sia suitable for surgery might be best described
as a reversible coma, lower concentrations of an-
Normalized delta power

SD Delta rebound 2 esthetics produce a sedative state. The overall

pattern of brain activity in humans, as measured
1.5
using functional magnetic resonance imaging, is
1
markedly similar during NREM sleep and after
Control
EEG
sedation with dexmedetomidine or propofol
0.5 Control delta (Fig. 3B) (42). Humans given the a2 adrenergic
power agonist dexmedetomidine enter a state resem-
0 bling stage 3 NREM sleep, the deepest type of
0 6 12 18 24 sleep (43). We now know that this is because
Time (h)
sedatives and anesthetic agents can act in the
Fig. 2. After sleep deprivation with novel objects, delta oscillations in mice rebound rapidly, but sleep-wake circuitry (18, 44–46), especially in the
recovering lost sleep takes much longer. After 6 hours of sleep deprivation using novel objects, lost sleep midbrain circuitry comprising the lateral haben-
is recovered over the following 24 hours, but the rebound in delta oscillations comes back to baseline after ula, VTA, and hypothalamus, to mimic NREM
only about 6 hours. Selective genetic ablation of certain neuronal populations can greatly decrease sleep sleep (although there are no drugs known that
homeostatic responses. For example, if galanin neurons are selectively ablated in the lateral preoptic mimic REM sleep). For example, the preoptic hy-
hypothalamus, both measures of sleep homeostasis are significantly blunted. Graphs are replotted from data pothalamic galanin neurons that contribute to
in (27) with permission under the Creative Commons CC-BY license. NREM sleep homeostasis (27, 28) partly mediate

Franks et al., Science 374, 556–559 (2021) 29 October 2021 3 of 4

 out the brain, and this circuitry eventually trig-
A Time
Local sensing gers global sleep. In principle, drugs could be
of sleep need
awake developed that more selectively target the re-
Global induction storative properties of natural sleep.
CTX of sleep
REFERENCES AND NOTES
Time
1. O. I. Lyamin, L. M. Mukhametov, J. M. Siegel, Curr. Opin. Neurobiol.
awake
44, 144–151 (2017).
Mouse brain VTA 2. D. J. Dijk, H. P. Landolt, Handb. Exp. Pharmacol. 253, 441–481
Time (2019).
awake Raphe 3. J. Lowrie, H. Brownlow, BMC Public Health 20, 980 (2020).
4. S. Sabia et al., Nat. Commun. 12, 2289 (2021).
PO/BF 5. M. S. Blumberg, J. A. Lesku, P.-A. Libourel, M. H. Schmidt,
N. C. Rattenborg, Curr. Biol. 30, R38–R49 (2020).
6. J. M. Krueger, J. T. Nguyen, C. J. Dykstra-Aiello, P. Taishi, Sleep
Med. Rev. 43, 14–21 (2019).
B 7. H. Funato et al., Nature 539, 378–383 (2016).
8. J. Lu, M. A. Greco, P. Shiromani, C. B. Saper, J. Neurosci. 20,
3830–3842 (2000).
9. X. Yu et al., Nat. Neurosci. 22, 106–119 (2019).
10. L. B. Krone et al., Nat. Neurosci. 24, 1210–1215 (2021).
11. M. Nedergaard, S. A. Goldman, Science 370, 50–56 (2020).
12. C. Cirelli, G. Tononi, Semin. Cell Dev. Biol. 10.1016/
Propofol Dexmedetomidine Deep NREM sleep j.semcdb.2021.02.007 (2021).
13. M. M. Hoekstra, Y. Emmenegger, J. Hubbard, P. Franken, eLife
8, e43400 (2019).
Fig. 3. Distributed circuitry senses local sleep need, which triggers global sleep, and sedatives hijack
14. J. Hubbard et al., Nat. Commun. 11, 3130 (2020).
some of these pathways. (A) Current data support a model in which there is distributed circuitry that 15. E. C. Harding et al., Curr. Biol. 28, 2263–2273.e4 (2018).
locally senses the need to sleep and then sleep is initiated globally, perhaps by the lateral hypothalamus 16. C. M. Jung et al., J. Physiol. 589, 235–244 (2011).
17. D. McGinty, R. Szymusiak, Trends Neurosci. 13, 480–487
(red sphere). In addition to this circuitry that registers “tiredness,” signals from glia and muscle also (1990).
contribute to the sleep drive. (B) Overall brain activity in humans during deep NREM sleep is very similar to 18. N. P. Franks, Nat. Rev. Neurosci. 9, 370–386 (2008).
that caused by general anesthetics and sedatives. Images show changes in regional cerebral blood flow 19. A. R. Adamantidis, C. Gutierrez Herrera, T. C. Gent, Nat. Rev.
Neurosci. 20, 746–762 (2019).
measured with positron emission tomography. Figure is adapted from (42) with permission under the 20. S. Vesuna et al., Nature 586, 87–94 (2020).
Creative Commons CC-BY license. BF, basal forebrain; CTX, neocortex; PO, preoptic hypothalamic area. 21. Y. Niwa et al., Cell Rep. 24, 2231–2247.e7 (2018).
22. C. della Monica, S. Johnsen, G. Atzori, J. A. Groeger, D.-J. Dijk,
Front. Psychiatry 9, 255 (2018).
the arousable NREM-like sleep induced by example, the anesthetics propofol and isoflurane 23. V. V. Vyazovskiy et al., Nature 472, 443–447 (2011).
24. C. B. Saper, T. E. Scammell, J. Lu, Nature 437, 1257–1263 (2005).
dexmedetomidine (27). Propofol and isoflurane potentiate the actions of GABA at GABAA re- 25. J. E. Allen, T. E. Sutherland, Science 364, 738–739 (2019).
both induce a sedative state (47, 48), in part by ceptors (18), and this will reduce EEG frequency 26. D. Liu, Y. Dan, Annu. Rev. Neurosci. 42, 27–46 (2019).
activating the lateral habenula, a midbrain nu- independently of sleep. Therefore, a direct com- 27. Y. Ma et al., Curr. Biol. 29, 3315–3322.e3 (2019).
28. S. Reichert, O. Pavón Arocas, J. Rihel, Neuron 104, 370–384.e5
cleus important for ensuring NREM sleep stabil- parison between the EEG during sleep and (2019).
ity (47). Selectively activating GABA neurons by during sedation is not straightforward with 29. A. V. Kalinchuk, T. Porkka-Heiskanen, R. W. McCarley,
chemogenetics in the VTA also produces a par- these drugs. A better approach is to determine R. Basheer, Eur. J. Neurosci. 41, 182–195 (2015).
30. X. Yu et al., Mol. Psychiatry 10.1038/s41380-020-0810-9 (2020).
ticularly deep, sedative-like NREM state for 4 to whether sedative drugs can “repay” sleep debt 31. G. Oikonomou et al., Neuron 103, 686–701.e8 (2019).
5 hours (9). The activation of these VTA GABAergic after sleep deprivation. Some anesthetics can 32. A. M. Ingiosi et al., Curr. Biol. 30, 4373–4383.e7 (2020).
neurons enhances isoflurane and sevoflurane repay NREM debt but few repay REM debt (51). 33. N. C. Rattenborg, Interface Focus 7, 20160082 (2017).
anesthesia (49, 50), and their inhibition pro- Even if current sedatives act on natural sleep 34. A. Eban-Rothschild, W. J. Giardino, L. de Lecea, Curr. Opin.
Neurobiol. 44, 132–138 (2017).
motes arousal from anesthesia (49). These VTA pathways, they certainly have other targets. The 35. A. Kempf, S. M. Song, C. B. Talbot, G. Miesenböck, Nature 568,
GABA neurons are also needed for sleep ho- receptors are not restricted to sleep-inducing 230–234 (2019).
meostasis (30). Therefore, these particular cells, pathways. Moreover, the natural pattern of 36. M. Lazarus, J.-F. Chen, Z.-L. Huang, Y. Urade, B. B. Fredholm,
Handb. Exp. Pharmacol. 253, 359–381 (2019).
or a subset of them, could be a potential target for NREM-REM cycles does not occur during se- 37. T. Porkka-Heiskanen et al., Science 276, 1265–1268 (1997).
new sedatives that provide restorative benefits. dation, so current drugs clearly do not exactly 38. M. G. Frank, Handb. Exp. Pharmacol. 253, 83–96 (2019).
What is the evidence that current sedatives mimic natural sleep. Therefore, until a drug is 39. J. C. Ehlen et al., eLife 6, e26557 (2017).
40. F. Brüning et al., Science 366, eaav3617 (2019).
provide any restorative benefits? As many of developed that selectively acts on sleep path- 41. Z. Wang et al., Nature 558, 435–439 (2018).
us have experienced, recovery from surgery or ways, we cannot be sure that the restorative ben- 42. A. Scheinin et al., J. Neurosci. 41, 1769–1778 (2021).
43. O. Akeju et al., Clin. Neurophysiol. 129, 69–78 (2018).
a same-day procedure can be exhausting and efits of sleep can be provided artificially by drugs. 44. H. C. Hemmings Jr et al., Trends Pharmacol. Sci. 40, 464–481
feels the opposite of restorative. However, the (2019).
trauma experienced during these operations Conclusions 45. L. F. Jiang-Xie et al., Neuron 102, 1053–1065.e4 (2019).
46. M. B. Kelz, G. A. Mashour, Curr. Biol. 29, R1199–R1210 (2019).
could easily overwhelm any beneficial restor- The overwhelming urge to sleep shapes the 47. C. Gelegen et al., Curr. Biol. 28, 580–587.e5 (2018).
ative benefits that the sedatives themselves lives of all humans. The most likely reason we 48. C. Liu et al., Front. Mol. Neurosci. 14, 628996 (2021).
might provide. More importantly, because we sleep is to allow fundamental housekeeping 49. L. Yin et al., Front. Neural Circuits 13, 73 (2019).
50. K. Vlasov et al., Anesth. Analg. 132, e50–e55 (2021).
do not know what these restorative benefits of processes to occur, and these processes require 51. D. Pal, W. J. Lipinski, A. J. Walker, A. M. Turner, G. A. Mashour,
sleep are, we have no way of directly measur- an unconscious brain. The lower brain temper- Anesthesiology 114, 302–310 (2011).
ing whether they accrue during sedation. In- atures during sustained NREM sleep would
AC KNOWLED GME NTS
stead, we have to rely on other measures such slow inhibitory postsynaptic currents by about
We thank D.-J. Dijk (University of Surrey, UK) for helpful
as the EEG delta power; however, as mentioned the same extent as sedatives and thus be incon- discussion. Funding: Work in the Franks-Wisden laboratory is
above, this is more likely to correlate with sleep sistent with consciousness. Multiple deficits in funded by the Wellcome Trust (grants 107839/Z/15/Z and
depth rather than any restorative processes. An brain function result if sleep is denied; the time 107841/Z/15/Z) and the UK Dementia Research Institute (grant
UK DRI-5004). Competing interests: The authors declare no
additional confound is that many drugs will course of catching up on lost sleep implies an-
competing financial interests.
affect the shape and timing of oscillations in the abolic or catabolic processes. Sleep need is
EEG independently of what drives them. For sensed locally by circuitry distributed through- 10.1126/science.abi8372

Franks et al., Science 374, 556–559 (2021) 29 October 2021 4 of 4