PLoS Biology | www.plosbiology.org 0756
obvious selective advantages. Sleep is a time of reduced body
and brain metabolic rate [1,2], allowing energy conservation,
particularly if a warm place is available, as can be provided by
a compliant parent or sibling. The sleeping, quiescent infant
is also less likely to attract predators and is easier to transport.
At the earliest ages, infants who have not yet opened their
eyes and whose cortex is not yet developed have limited
learning opportunities from interactions with the outside
world: another reason for reduced waking.
But sleep comes in many forms. Evolutionary arguments
may make sense for slow-wave deep sleep patterns at birth
that are associated with a general shutdown of the brain,
but may not provide such an obvious explanation for the
relative predominance of rapid eye movement (REM) sleep.
For example, in human neonates REM sleep constitutes
approximately eight hours per day, or 50% of the total sleep
time, whereas human adults devote less than two hours per
day, or 20% of their seven to eight hours of sleep time, to
REM sleep . REM sleep is characterized by high brain
metabolic and neuronal activity rates , reduced muscle
tone, irregular and relatively automatic respiration uncoupled
from its usual regulatory mechanisms , and diminished
thermoregulation . These properties seem maladaptive,
which suggests that there must be some compensatory survival
benefi t for REM sleep to have persisted. Could REM sleep
play a particularly important role in development?
Interesting evidence for this hypothesis has come
from studying the effects of REM sleep deprivation on
the development of the visual system. It is known that
the occlusion of one eye during the maturation of visual
connections that occurs after birth causes the open eye
to acquire more central connections than the closed eye.
This disproportionate representation seems to result from
a difference of activity in the optic nerve between the open
eye and closed eye . Although early ideas that REM sleep
was necessary for brain plasticity might suggest that REM
sleep deprivation would prevent this reorganization, just
the reverse occurs. REM sleep deprivation accelerates the
shift of connections to favor the open eye [8,9]. Rather than
facilitating change , REM sleep may therefore be a source
of endogenous activity that tends to prevent altered sensory
stimulation from causing abnormal connections to form.
REM sleep may prevent the programmed cell death and the
pruning of connections that occurs when critical synapses are
Another possible role for neonatal REM sleep might
be in thermoregulation of the growing brain. It is known
lthough frazzled new parents may beg to differ, infants
do sleep more than adults. This sleep pattern is seen
in a wide variety of mammalian species, with some
that nonREM sleep tends to cool the brain, reducing its
thermoregulatory set point . In contrast, REM sleep
tends to heat certain brain regions . The nonREM–REM
alternation comprises a thermoregulatory oscillation.
May 2005 | Volume 3 | Issue 5 | e178
Functional Implications of Sleep
Jerome M. Siegel
Open access, freely available online
Primers provide a concise introduction into an important aspect of biology
highlighted by a current PLoS Biology research article.
Citation: Siegel JM (2005) Functional implications of sleep development. PLoS Biol
Copyright: © 2005 Jerome M. Siegel. This is an open-access article distributed
under the terms of the Creative Commons Attribution License, which permits
unrestricted use, distribution, and reproduction in any medium, provided the
original work is properly cited.
Abbreviation: REM, rapid eye movement
Jerome Siegel is at the Veterans Administration Greater Los Angeles Health Care
System in Sepulveda, California, and the Department of Psychiatry at the University
of California, Los Angeles, California, United States of America. E-mail: jsiegel@ucla.
Figure 1. Model of Some of the Major Systems Involved in Regulating
Muscle Activity in REM Sleep Drawn on a Sagittal Section of the
Cholinergic (ACh) neurons in the pons, which are under the
inhibitory control of noradrenergic (NE) and serotonergic
(5-HT) neurons, trigger REM sleep. They activate descending
glutamatergic neurons, which in turn activate glycinergic and
GABAergic neurons. Other glycinergic interneurons in the
spinal cord are also activated by unknown descending inputs.
The release of glycine and GABA inhibits motoneurons. The
descending glutamatergic pathway also activates GABAergic
interneurons, which inhibit noradrenergic and serotonergic
neurons. The reduction in norepinephrine and serotonin release
during REM sleep disfacilitates motoneurons. Descending
glutamatergic neurons that connect directly to motoneurons
produce phasic excitation during REM sleep. The net result of
the action of this network is an absence of muscle tone in the
“antigravity” muscles in REM sleep, interrupted by twitches (see
text for references).
PLoS Biology | www.plosbiology.org 0757
It is often assumed that the amount of time spent in
different sleep states is determined by processes controlled
by the cerebral cortex. The emphasis on the cortical role in
sleep may result more from the technical ease of recording
electroencephalograms from the cortex than from persuasive
functional evidence. At birth, cortical metabolism and
neuronal fi ring are minimal , yet this is the time of
greatest sleep. In adults, damage to the cortex produces little
or no change in sleep, indicating that the signal for sleep
does not originate in or at least does not require the cortex
. Animals with proportionally larger cortices do not have
more REM or nonREM sleep time than animals with relatively
little cortex . The effects of long-term sleep deprivation
have been shown to be largely autonomic in nature, including
elevated body temperature, skin lesions, and increased food
intake . Such effects cannot be duplicated by any cortical
lesions. However, many of these symptoms appear to be
consistent with hypothalamic dysfunction [17,18].
Evolutionary evidence also suggests that the cortex may be
a relatively recent participant in REM sleep. Plesiomorphic
(primitive) mammals such as the egg-laying echidna and
platypus have large amounts of REM-sleep-like activity in
brainstem structures at birth [19,20]. The brainstem is the
key region for REM sleep generation, being both necessary
and suffi cient for its occurrence . However, the cortex of
these animals scarcely changes activity during these states,
showing slow-wave patterns during the REM sleep state. In
this respect the sleep of placental mammals may represent
ontogeny recapitulating phylogeny, since a reduction in
electroencephalogram power is a late-developing component
of REM sleep.
A prominent feature of REM sleep is the rapid eye
movements and associated twitches that defi ne the state.
These are particularly marked and vigorous in neonates.
It has been shown that twitches with some resemblance to
REM sleep activity are present in the isolated spinal cord
of neonates and diminish in the transected cord of older
animals . This has suggested to some that a primal phasic
activity of the central nervous system transforms postnatally
over an extended time period into the very different
brainstem-generated pattern seen in adults. But in this issue
of PLoS Biology, Karlsson et al.  show that this is not the
case. In a set of technically demanding experiments, they
demonstrate a remarkable similarity between sleep control
mechanisms in the one-week-old rat and those in the adult
cat, and by implication throughout the mammalian line.
By severing the connections to and from the forebrain
(cerebral cortex and associated structures), Karlsson et al.
were able to study sleep-related activity in the midbrain and
brainstem. They described the rat homologs of the medullary
neurons that induce the atonia seen in sleeping adult cats
and narcoleptic dogs [22,23,24] (Figure 1). More rostrally,
they identifi ed neural activity in the region of the locus
coeruleus that facilitates movement and report contrasting
inhibitory activity in the adjacent subcoeruleus region, again
paralleling studies in the cat [25,26,27]. They also found cells
that appear to generate or at least contribute to the twitches
of REM sleep.
The similarities to the adult cat’s REM sleep control
mechanisms are so striking that what becomes interesting are
the small differences that are reported. The locus coeruleus
REM “sleep-off cells,” which are active in waking, reduce
activity in nonREM sleep, and cease activity in REM sleep,
appear to not have long-duration waveforms in the neonatal
animals examined by Karlsson et al., unlike the case of the
adult rat and cat [28,29]. Another difference is the apparent
absence of the cessation of dorsal raphe (serotonin) unit
discharge in REM sleep. Although the authors speculate that
this is due to the absence of forebrain connections in their
experimental preparation, it has been shown that forebrain
mechanisms are not necessary for this cessation of raphe
activity in adult cats . However, identifi cation of the
narrow dorsal raphe nucleus is diffi cult even in adult cats, and
it is certainly possible that these neurons were overlooked in
the neonatal rat.
The upshot of these fi ndings is a picture of a largely mature
REM sleep generator mechanism at birth. The developmental
progression of REM sleep signs, particularly the reduction
in sleep duration and the development of the characteristic
reduction in electroencephalogram voltage to a waking-like
pattern in REM sleep, may result from the maturation of
the targets of these brainstem systems, the modulation of
these generator mechanisms by developing systems, or a
relatively subtle maturing of connections within the REM
sleep generator systems. This work pushes the probable
organization of the REM sleep generator system in rats back
to before one week of age, possibly to an in utero stage.
What does all this say about the function of REM sleep?
Although we are left with the same initial speculations,
the neonatal model provides a different perspective for
approaching these functions. It is particularly useful to
know that key elements of the REM sleep system are present
in neonatal rats, since these animals are ideal subjects for
in vitro studies of tissue slices [31,32]. It is not practical
to perform in vitro experiments on the adult brainstem.
However, there has always been some question as to whether
studies of neonatal brainstems would be applicable to
the question of adult REM sleep mechanisms. One can
now imagine examining the metabolism and membrane
characteristics of these critical cell groups as a means of
gaining better insight into REM sleep function. However, as
Karlsson et al.’s work demonstrates , most of the neurons
of interest are not homogenously concentrated in any easily
targeted region. Identifying the individual neurons of interest
in vitro remains a challenge. This challenge will have to be
surmounted in order to identify the control mechanism and
better understand the function of REM sleep. ?
Research was supported by the Medical Research Service of the
Department of Veterans Affairs and National Institutes of Health
grants NS14610, HL41370, MH64109, and HL060296.
1. Nofzinger EA, Buysse DJ, Germain A, Price JC, Miewald JM, et al. (2004)
Functional neuroimaging evidence for hyperarousal in insomnia. Am J
Psychiatry 161: 2126–2128.
2. Maquet P (1995) Sleep function(s) and cerebral metabolism. Behav Brain
Res 69: 75–83.
3. Roffwarg HP, Muzio JN, Dement WC (1966) Ontogenetic development of
the human sleep-dream cycle. Science 152: 604–619.
4. Siegel JM (2000) Brainstem mechanisms generating REM sleep. In:
Kryger MH, Roth T, Dement WC, editors. Principles and practice of sleep
medicine. Philadelphia: W. B. Sanders. pp. 112–133.
5. Baker T, McGinty D (1977) Reversal of cardiopulmonary failure during
active sleep in hypoxic kittens: Implications for sudden infant death.
Science 199: 419–421.
6. Parmeggiani PL, Azzaroni A, Calasso M (2002) Systemic hemodynamic
changes raising brain temperature in REM sleep. Brain Res 940: 55–60.
May 2005 | Volume 3 | Issue 5 | e178
PLoS Biology | www.plosbiology.org 0758
7. Wiesel TN (1999) Early explorations of the development and plasticity of
the visual cortex: A personal view. J Neurobiol 41: 7–9.
8. Shaffery JP, Roffwarg HP, Speciale SG, Marks GA (1999) Ponto-geniculo-
occipital-wave suppression amplifi es lateral geniculate nucleus cell-size
changes in monocularly deprived kittens. Brain Res Dev Brain Res 114:
9. Shaffery JP, Oksenberg A, Marks GA, Speciale SG, Mihailoff G, et al. (1998)
REM sleep deprivation in monocularly occluded kittens reduces the size of
cells in LGN monocular segment. Sleep 21: 837–845.
10. Siegel JM (2001) The REM sleep-memory consolidation hypothesis. Science
11. McGinty D, Szymusiak R (1990) Keeping cool: A hypothesis about the
mechanisms and functions of slow-wave sleep. Trends Neurosci 13: 480–487.
12. Wehr TA (1992) A brain-warming function for REM sleep. Neurosci
Biobehav Rev 16: 379–397.
13. Corner MA (1985) Ontogeny of brain sleep mechanisms. In: McGinty DJ,
Drucker-Colin R, Morrison A, Parmeggiani PL, editors. Brain mechanisms
of sleep. New York: Raven Press. pp. 175–198.
14. Villablanca J, Marcus R (1972) Sleep-wakefulness, EEG and behavioral
studies of chronic cats without neocortex and striatum: The ‘diencephalic’
cat. Arch Ital Biol 110: 348–382.
15. Zepelin H, Siegel JM, Tobler I (2005) Mammalian sleep. In: Kryger MH,
Roth T, Dement WC, editors. Principles and practice of sleep medicine.
Philadelphia: Elsevier Saunders. pp. 91–100.
16. Rechtschaffen A, Bergmann BM (2002) Sleep deprivation in the rat: An
update of the 1989 paper. Sleep 25:18–24.
17. Everson CA, Crowley WR (2004) Reductions in circulating anabolic
hormones induced by sustained sleep deprivation in rats. Am J Physiol
Endocrinol Metab 286: E1060–E1070.
18. Everson CA, Nowak TSJ (2002) Hypothalamic thyrotropin-releasing
hormone mRNA responses to hypothyroxinemia induced by sleep
deprivation. Am J Physiol Endocrinol Metab 283: E85–E93.
19. Siegel JM, Manger P, Nienhuis R, Fahringer HM, Pettigrew J (1996) The
echidna Tachyglossus aculeatus combines REM and nonREM aspects in a
single sleep state: Implications for the evolution of sleep. J Neurosci 16:
20. Siegel JM, Manger PR, Nienhuis R, Fahringer HM, Shalita T, et al. (1999)
Sleep in the platypus. Neuroscience 91: 391–400.
21. Karlsson KÆ, Gall AJ, Mohns EJ, Seelke AMH, Blumberg MS (2005) The
neural substrates of infant sleep in rats. PLoS Biol 3: e143. DOI: 10.1371/
22. Sakai K (1985) Anatomical and physiological basis of paradoxical sleep. In:
McGinty DJ, Drucker-Colin R, Morrison AR, Parmeggiani PL, editors. Brain
mechanisms of sleep. New York: Raven Press. pp. 111–138.
23. Siegel JM, Wheeler RL, McGinty DJ (1979) Activity of medullary reticular
formation neurons in the unrestrained cat during waking and sleep. Brain
Res 179: 49–60.
24. Siegel JM, Nienhuis R, Fahringer H, Paul R, Shiromani P, et al. (1991)
Neuronal activity in narcolepsy: Identifi cation of cataplexy related cells in
the medial medulla. Science 262: 1315–1318.
25. Kodama T, Lai YY, Siegel JM (2003) Changes in inhibitory amino acid
release linked to pontine-induced atonia: An in vivo microdialysis study. J
Neurosci 23: 1548–1554.
26. Lai YY, Kodama T, Siegel JM (2001) Changes in monoamine release in the
ventral horn and hypoglossal nucleus linked to pontine inhibition of muscle
tone: An in vivo microdialysis study. J Neurosci 21: 7384–7391.
27. Mileykovskiy BY, Kiyashchenko LI, Kodama T, Lai YY, Siegel JM
(2000) Activation of pontine and medullary motor inhibitory regions
reduces discharge in neurons located in the locus coeruleus and the
anatomical equivalent of the midbrain locomotor region. J Neurosci
28. Aston-Jones G, Bloom FE (1981) Activity of norepinephrine-containing
locus coeruleus neurons in behaving rats anticipates fl uctuations in the
sleep-waking cycle. J Neurosci 1: 876–886.
29. Hobson JA, McCarley RW, Wyzinski PW (1975) Sleep cycle oscillation:
Reciprocal discharge by two brainstem neuronal groups. Science 189:
30. Hoshino K, Pompeiano O (1976) Selective discharge of pontine neurons
during the postural atonia produced by an anticholinesterase in the
decerebrate cat. Arch Ital Biol 144: 244–277.
31. van den Pol AN, Ghosh PK, Liu RJ, Li Y, Aghajanian GK, et al. (2002)
Hypocretin (orexin) enhances neuron activity and cell synchrony in
developing mouse GFP-expressing locus coeruleus. J Physiol 541: 169–185.
32. Greene RW, Gerber U, McCarley RW (1989) Cholinergic activation of
medial pontine reticular formation neurons in vitro. Brain Res 476: 154–
May 2005 | Volume 3 | Issue 5 | e178