User:Eugene M. Izhikevich/Proposed/Ontogeny of sleep

From Scholarpedia
Jump to: navigation, search

The states of arousal, wake (W), rapid eye movement (REM) sleep, and non-rapid eye movement (NREM) sleep, are physiologically and behaviorally well defined in adult mammals, and their temporal expression is clearly organized by the circadian system. The situation is not clear, however, in developing mammals. Comparing studies of the development of sleep/wake and circadian rhythms in different species of mammals is complicated because of the huge differences in maturity at the time of birth. But, understanding the development of these systems can contribute to understanding their basic biology and their possible contributions to certain postnatal pathologies such as sudden infant death syndrome (SIDS).


3 stages of sleep development

Altricial mammals are less fully developed at birth than are precocial mammals. Whereas precocial species have brain activity at birth that shows characteristic differences associated with adult arousal states, the electrical activity recorded from brains of altricial species is undifferentiated. This difference is not surprising given the fact that in altricial brains at the time of birth most axons have not yet reached their targets and much synaptogenesis is yet to come. However, in both altricial and precocial species the development of sleep/wake can be divided into three stages some of which take place in utero in the precocial species. These stages are:

  1. dissociation,
  2. concordance, and
  3. maturation.

Dissociation is the stage of development when the EEG can not be reliably used to determine states, but other correlates of sleep and wakefulness such as changes in heart and respiratory rate and motor activity are present. However, these correlates are not coordinated with each other. Hamburger referred to this developmental stage in human premature infants as presleep. A similar presleep stage has been observed in utero in precocial species such as lambs and baboons. In altricial species such as the kitten, the absence of sleep EEG patterns and dissociation of the other hallmarks of sleep states are seen postnatally.

Concordance refers to the stage of development when differentiated EEG patterns characteristic of adult arousal states emerge and begin to align with other correlates of wake and sleep states. In precocial species this process begins in late gestation and EEG definable arousal states are observable at parturition. Concordance in humans begins approximately between gestational weeks 28 and 32 and is completed sometime in the last month of fetal life in full term babies. Throughout the period of concordance in precocial species, there is no evidence of intermediate transitional sleep states that are homologous with adult sleep states. Rather, all features of sleep and wake emerge at about the same time.

Concordance in altricial species is not as clear and has generated some controversy. EEG patterns are undifferentiated in the newborn altricial animal, but motor patterns that appear similar to those associated with arousal states are evident. Specifically, these motor patterns are coordinated movements, jerky movements (twitches), or quiesence. Periods of tonic, coordinated motor activity are considered wake, and the rest of the time is considered to be sleep. Sleep is then divided into behavioral quiet sleep (BQS) characterized by quiescence or behavioral active sleep (BAS) characterized by phasic twitches. Because the undifferentiated EEG pattern is relatively fast and low amplitude, it resembles the activated EEG that in adults is associated with wake and REM sleep. This association of an apparent activated EEG with phasic muscle activity (BAS) has been assumed to be a precursor of mature REM sleep while quiescent periods (BQS) have been assumed to be the precursor of mature NREM sleep.

We will look more critically at the assumption that BAS is the precursor of REM sleep and that BQS is the precursor of NREM sleep later on in this article. We should note, however, that the terms active sleep and quiet sleep used to describe sleep states in premature and term human infants are referring to different suites of variables than are described by these same terms in altricial species. The difference is that the EEG patterns of active and quiet sleep are clearly differentiated in human preterm and full term infants. In contrast, BAS and BQS in altricial species are not distinguished by any differential EEG activity, but rather these states are distinguished primarily by the presence or absence of myoclonic twitches.

After the emergence of distinct sleep states, there are dramatic maturational changes in the amounts and timings of those states and in some of the EEG characteristics that define them. Early in life, the amount of REM sleep is much higher than adult levels and declines while the amounts of NREM sleep decrease to a lesser degree and wake increases. The characteristic brain activities that define the sleep states in the adult become increasingly well defined and in some cases altered. In altricial species for example, the activated EEG of REM sleep precedes the development of the slower, highly regular theta rhythm which has a frequency in the 4 to 5 Hz range early on and gradually increases to the characteristic adult frequency of 7 to 8 Hz. Pontine-geniculate-occipital (PGO) waves do not appear until a week or more after the emergence of REM sleep, and the characteristic muscle atonia of REM sleep is even slower to develop. The two major EEG correlates of NREM sleep, slow waves (delta band, 0.5 to 4.5 Hz) and sleep spindles (7 to 14 Hz) develop at different rates. First to appear is slow wave activity which in the rat steadily increase in amplitude from about postnatal age day 9 to day 21 reaching higher levels than are seen in the adult. Sleep spindles appear in the third postnatal week. Relatively late to mature are the regulatory mechanisms governing the expression of sleep – the ultradian, circadian, and homeostatic controls.

BAS and BQS may not be homologous with REM and NREM sleep

In altricial species such as the rat, the undifferentiated EEG is the same in the behaviorally defined states of BAS and BQS. Since the undifferentiated EEG lacks high amplitude wave patterns, it resembles somewhat the activated EEG of REM sleep. This resemblance, however, is superficial as might be expected by the fact that most axons traveling to the cortex have not yet reached their targets and most synaptogenesis has not yet occurred. Since REM sleep is a cholinergically mediated state, it is especially important to note that the cholinergic system is very immature at the time BAS is maximally expressed. There is an inverse relationship between BAS expression and (1) expression of muscarinic receptors and their signal transduction mechanisms, (2) the development of cholinergic nuclei and their efferents, and (3) neuronal responsiveness to cholinergic drugs. In fact, cholinergic blockade which eliminates adult REM sleep has no effect on BAS in the rat until late in the 2nd postnatal week after differentiated EEG patterns have developed. Lesions of brainstem structures essential for mature REM sleep have no effect on BAS in neonatal rats or kittens. Even transection of the neonatal rat’s spinal cord below the brainstem does not completely eliminate the phasic twitches that define BAS. In adult mammals, such transactions eliminate all episodes of myoclonia that normally accompany REM sleep. These results pose the obvious question of what is the nature of the myoclonia of BAS. It is likely that the twitches of BAS are non-specific fetal activity that is associated with the development of spinal reflex circuits prior to the imposition of descending inhibitory control. This is consistent with the fact that inhibitory mechanisms that normally inhibit motor activity during REM sleep are excitatory prior to the second postnatal week (in rodents). Although some have argued that BAS is in fact REM sleep without an EEG, it is difficult to reconcile that point of view with the above facts.

There is also no compelling evidence for assuming that BQS is a precursor state to NREM sleep. The comparison of different studies of BQS are complicated by the lack of a clear set of criteria that define BQS. If the definition of BQS is quiescent periods with regular respiration and no twitches, it amounts to only 2% of total recording time in one week old rats. However, in other studies in which small movements or startles are allowed in BQS, it can amount to over 30% of total recording time. Similar variances in the designation of BQS in kittens exist, all depending on the behavioral criteria used by the investigators.

The first EEG hallmark of sleep states to appear in altricial species is slow waves, and when they appear, they are not restricted to episodes of BQS. In fact, when slow waves first appear in the neonatal rat in the second postnatal week they occur equally in both BAS and BQS. By the third week of postnatal life, slow wave activity is mostly restricted to periods of motor quiescence and regular respiration, and the amplitude of the slow waves steadily grows until they reach maximum values around the beginning of the 4th week. EEG theta activity which marks REM sleep appears in the neonatal rat at about the same time as slow waves, but it matures to adult levels more slowly than do slow waves.

The view most consistent with existing data is that both REM and NREM sleep emerge from an undifferentiated brain state, and they do so at about the same time even though NREM sleep patterns seem to mature more rapidly than REM sleep patterns. Thus, in both precocial and altricial species of mammals, there are not precursor sleep states that precede EEG differentiation, rather the development of the EEG hallmarks of sleep reflect the development of the underlying neuronal mechanisms, and presumably therefore the yet-to-be-discovered functions of the sleep states.

Sleep regulation

There are three components of sleep regulation that develop after the emergence of EEG defined states: ultradian, circadian, and homeostatic. Ultradian refers to the cyclicity of sleep states, or the sleep cycle. In most mammals the distribution of total sleep time is about 80% NREM sleep and 20% REM sleep, but these sleep times are distributed into bouts that alternate with each other. The sleep cycle in the human adult is about 90 min and the sleep cycle in a rat is about 10 min. This cyclicity of sleep states is poorly developed at the time the states first emerge, and it develops gradually with the cycles becoming more regular and longer.

Circadian refers to the 24-hour rhythm of sleep and wake which is dependent on a biological clock that resides in the suprachiasmatic nucleus of the hypothalamus (SCN). Although the pacemaker is probably oscillating in the fetus and is probably entrained by the periodicity of the mother, its coupling to the systems controlling sleep states comes considerably after the emergence of sleep states. In rats the circadian organization of sleep begins to appear about postnatal day 15. In humans circadian organization of sleep emerges in the second to third month after birth, but takes longer to become entrained to the environmental cycle of light and dark.

Homeostatic regulation refers to the relationship between sleep and the duration of prior wake. In adult mammals prolonged wakefulness produces compensatory increases in sleep time, but most importantly, it produces an increase in NREM sleep EEG slow wave activity. A similar intensity dimension in REM sleep is not well-established; instead, REM sleep amounts rebound following prolonged total sleep deprivation or selective REM sleep deprivation. One striking developmental change is an increase in the tolerance of sleep pressure. The amount of wake is very low in neonatal humans and altricial species, and they are unable to sustain long episodes of wake. A second interesting developmental event is that the way that infants respond to sleep deprivation changes abruptly (in the rat) between the 3rd and 4th postnatal weeks. Short periods of sleep deprivation that would be insignificant in the adult result in clear compensatory changes in infant sleep. But, the nature of this compensatory response changes with age. Early on when there are robust rebounds in NREM sleep time, there are none in REM sleep. In the rat, responses in REM sleep time to prior sleep deprivation first appear toward the end of the 3rd postnatal week. Increases in NREM slow wave activity also appear around this time. We do not know the significance of this sudden shift, but it comes after the time that normal delta power levels have reached their maximum and are declining and at the time that the ability of the animal to sustain longer episodes of wake are rapidly increasing. Nor is it known if similar developmental events occur in humans, although there is some indication that similar events occur in other animals. These observations indicate that either the cost of wakefulness is different or the ability to pay sleep debt is different in infants and adults.

The development of the circadian system

The circadian system of mammals is practically synonymous with the suprachiasmatic nucleus of the hypothalamus, but it also involves the pineal gland which produces the hormone melatonin. In contrast to the complexity of the neural structures involved in sleep, the SCN and the pineal are discrete structures that can be studied anatomically, neurochemically, and functionally. Therefore, we can ask questions about the development of the circadian system at these several levels.

The SCN is a compact but heterogenous cluster of cells with a definite structure, yet the relationship between the organization of the SCN and its function as the circadian pacemaker is not understood. SCN cells arise from the germinal epithelium of the anterior, ventral diencephalon. In the rat this begins about embryonic day 14 (E14) and is complete by several days before birth. The fate determination of putative SCN cells is quite robust in that deafferentation of the area does not alter their differentiation and even if they are excised and transplanted into the chamber of the eye, they still develop into SCN cells. And when SCN tissue is extracted in early postnatal life and placed in culture, the cells continue to express the properties of SCN cells including circadian rhythmicity.

There are three major afferent pathways to the SCN, and their development has been described in the rodent. Information from photoreceptors reaches the SCN by a direct and an indirect route. The direct route is via the retinohypothalamic tract in which the neurotransmitters are glutamate and pituitary adenylate cylcase-activating polypeptide (PCAP), and these connections develop between P1 and P10. The indirect pathway for photic information through the intergeniculate leaflet (IGL) uses the neurotransmitter neuropeptide Y and develops between P4 and P11. The third set of projections are serotonergic from the midbrain raphe nuclei, and they develop between the day of birth and P10. These pathways are the major mechanisms of entrainment in the adult mammal, but as we will see, entrainment of the fetal SCN occurs considerably before birth. Thus additional mechanisms must be responsible for the initial entrainment of the fetal SCN. Strong evidence shows that melatonin and dopamine may be two such factors that mediate the maternal entrainment of the circadian systems of her pups, and fetal SCN cells do express the receptors for these two neurochemicals.

Functional development of the circadian system involves: 1. pacemaker function, or the generation of rhythms, 2. entrainment of rhythms, and 3. the coupling of SCN rhythmicity to drive other variables such as body temperature and sleep/wake. The first of these properties to develop is, of course, pacemaker function. Circadian cycles are observable in the SCN within days after the cells become postmitotic. The first evidence of rhythmicity in the fetal SCN was documeted in rats by the differences in uptake of 2 deoxyglucose by SCN cells when the 2DG was injected at two times of day. The earliest these oscillations were observed in rats was about 3 days before birth. Of course these changes in fetal SCN metabolism could have been solely imposed by a maternal influence. But subsequent studies revealed daily changes in 2DG uptake and in electrical activity of the SCN in brain slices taken from fetuses. Such rhythms have also been observed in fetuses or in brain slices from fetuses after the dam received SCN lesions early in gestation and was herself arrhythmic. The evidence supports the view that pacemaking function develops in the SCN prior to birth.

The ability of the SCN oscillations to be entrained also develops before birth in rodents, but the entraining stimuli come from the mother. Under normal circumstances, the new born pups are in synchrony with each other and with the maternal circadian rhythm. If the dam has received an SCN lesion early in gestation and is therefore arrhythmic, the newborn pups express circadian rhythms, but they are not in synchrony with each other. Most telling are cross fostering experiments in which the newborn pups are presented to mothers that are 180 degrees out of phase from their natural mothers. In these litters, when rhythmicity is expressed in the pups, it is in phase with the rhythm or their birth mothers and not their foster mothers. The fetal SCN seems to be highly sensitive to entraining signals including melatonin, dopamine, and probably other compounds such as the cholinergic drug nicotine. Whether or not disruption of fetal rhythms by exposure to such compounds in utero have effects on postnatal health is not known. However, when pregnant ewes were kept in constant light, their circadian rhythms were disrupted as evidenced by vasopressin levels in the CSF and their fetuses did not survive. When kept on a light/dark cycle, their circadian rhythms were normal and their fetuses survived. In the postnatal period, the sensitivity of the SCN to the entraining influence of light develops as the afferent pathways bringing photic information to the SCN develop.

The coupling of SCN oscillations to various physiological and behavioral properties of the newborn emerge at different postnatal ages. In the rat, a rhythm of pineal N-acetyltranferase (the rate limiting enzyme for melatonin synthesis) activity develops in the first post-natal week, a rhythm of body temperature develops in the second postnatal week, and a rhythm of sleep/wake appears in the third postnatal week. In the human newborn, a clear day/night organization of sleep/wake develops – to the relief of parents – by the third month of life.

Sleep and circadian development continues through life

In humans we know that many characteristics of sleep and circadian rhythmicity change from birth, through childhood, adolescence, adult, middle, and old age. Total sleep time (TST) is maximal in the neonate and may reach 17 to 18 hrs a day with almost half of that being REM sleep. Circadian organization of sleep is absent in the neonate, but, gradually develops between the 2nd and 3rd months of life. When circadian organization appears, it is in phase with the environment, so entrainment of the pacemaker occurs before the coupling of the circadian system to sleep control. By age 5, TST declines to about 10 to 12 hrs per day with only about 20% of that being REM sleep. Thus, the major change in sleep amounts over early childhood is the decrease in REM sleep.

Adolescence is a time of significant changes in characteristics of sleep and circadian rhythms. The spectral properties of the NREM sleep EEG show considerable change at this time. The power density in the slow wave band (delta power) is high during childhood and at a maximum around ages 9 to 10 years. Delta power then declines significantly (about 25%) between ages 12 and 14. After that, delta power continues to fall slowly and steadily with age. Since delta power is the definitive characteristic of stages 3 and 4 NREM sleep, those stages decrease linearly with age and are practically non-existent in the elderly. REM sleep amounts, in contrast, show no major change with adolescence and just follow a modest, linear decline from an average of about 22% in childhood to about 18% in old age.

The major circadian changes that occur after infancy are a phase advance of the timing of sleep onset at adolescence, and a gradual dampening of the amplitude of circadian rhythms with advanced age. To a certain extent, this leveling of circadian oganization of sleep/wake may be due to declining levels of physical activity during the day. Sleep during the day leads to a lessening of sleep pressure at night, and hence lessening quality of nocturnal sleep.


  • Blumberg, Mark S., Karlsson, Karl A., Seelke, Adele M. H., and Mohns, Ethan J. (2005) The ontogeny of mammalian sleep: a response to Frank and Heller (2003). J. Sleep Research. 14:91-98.
  • Carskadon, Mary A. and Dement, William C. (2005) Normal Human Sleep: an Overview. In. Principles and Practice of Sleep Medicine. Kryger, Meir H., Roth, Thomas, and Dement, William C. eds. Elsevier Saunders. New York, NY. Pp 13-23.
  • Davis, Fred C., Frank, Marcos G., Heller, H. Craig (1999). Ontogeny of Sleep and Circadian Rhythms. In. Regulation of Sleep and Circadian Rhythms. Turek, Fred. W. and Zee, Phyllis C. eds. Marcel Dekker, Inc. New York, NY. Pp 19-79.
  • Frank, Marcos G. and Heller, H. Craig (2003) The ontogeny of mammalian sleep: a reappraisal of alternative hypotheses. J. Sleep Res. 12:25-34.
  • Frank, Marcos G. and Heller, H. Craig (2005) Unresolved issues in sleep ontogeny: a response to Blumberg et al. J. Sleep Research. 14:98-101.

Recommended reading

External links

See also

Dreaming, Molecular biology of sleep, Narcolepsy, Phylogeny of sleep, REM (paradoxical) sleep, Sleep, Sleep and learning, Sleep deprivation, Sleep homeostasis, Sleep-walking, Slow wave sleep

Personal tools

Focal areas