Negative Impacts of Disrupted Sleep During Mammalian Development

One of the strongest pieces of evidence to suggest a privileged role for sleep during development is the unique consequence of disturbed sleep within this period. Testing this idea has proved difficult in humans and other mammals. Most work in humans has taken advantage of a common and naturally occurring sleep disruptor, obstructive sleep apnea (OSA). OSA in adults and children alike results in numerous complications, including cognitive dysfunction (Findley et al. 1986; Gozal 1998; Beebe et al. 2003; Archbold et al. 2004). Indeed in children, sleep-disordered breathing has been associated with reduced cognitive function and general intelligence, as well as neuroimaging-based evidence for prefrontal cortical abnormalities (Schechter 2002; Gottlieb et al. 2004; O’Brien et al. 2004; Halbower et al. 2006); severity of apnea also correlates with worsened neuropsychological deficits (O’Brien et al. 2004). A caveat to utilizing OSA as a means to study the impact of sleep fragmentation is the difficulty in distinguishing the effect of sleep loss from that of poor oxygenation or even obesity, which could independently impair developmental processes. However, snoring with sleep fragmentation in the absence of abnormal oxygenation also negatively affects multiple cognitive measures (Gottlieb et al. 2004), lending credence to the hypothesis that sleep itself is of primary importance.

If sleep has a special role in brain development, disturbing sleep processes in children should have a more severe impact than in adults, although it is challenging to obtain direct evidence for this idea in humans. Suggestively, while attentional and cognitive measures are impaired in both, a correlation between OSA and general intelligence has been observed only in children (O’Brien et al. 2004; Alchanatis et al. 2005; Gozal 2008). Thus, sleep fragmentation during a heightened period of brain maturation might cause more profound and long-lasting impairments. However, most studies use premorbid IQ as a control between OSA and non-OSA populations and new onset of OSA in adults is difficult to predict and study.

Work in mammalian models has reached similar conclusions to those in humans, although also with constraints (Frank 2011). REM sleep deprivation in rodents with pharmacologic agents (such as tricyclic antidepressants, which inhibit serotonin–norepinephrine reuptake) during the perinatal period results in myriad adult deficits, such as increased anxiety and social behaviors, as well as abnormal sleep itself (Mirmiran et al. 1981, 1983a). Can REM sleep disturbances be dissociated from persistent alterations in monoaminergic signaling during a sensitive developmental period? REM sleep deprivation by distinct pharmacologic agents seems to cause inconsistent impairments (Frank and Heller 1997), and more specific REM inhibiting agents such as selective serotonin reuptake inhibitors do not result in adult sleep impairments (in contrast to tricyclic antidepressants) (Farabollini et al. 1988). Notably, gentle mechanical sleep deprivation likewise does not recapitulate tricyclic antidepressant exposure (Mirmiran et al. 1983b).

Other experiments have sought to disturb regions of the brain involved in pontine–geniculate–occipital wave generation, which is a core feature of REM sleep and thought to be critical for memory consolidation during sleep in adults. Lesions that eliminate pontine–geniculate–occipital waves in neonate kittens result in aberrant development of the lateral geniculate nucleus (Davenne and Adrien 1984). These results support the notion that REM sleep in early life could be crucial for normal brain patterning, although nonspecific or non-sleep-dependent effects of the lesion are difficult to rule out. In line with these findings, REM sleep deprivation in adolescent mice delays the critical period of visual cortical development, suggesting REM sleep disturbances somehow impinge upon brain maturation (Shaffery et al. 2002; Blumberg et al. 2013). Compelling evidence for an active role of sleep in brain sensorimotor development has also come from rodent pups. Like human infants, rat pups spend a large amount of time in REM/active sleep, characterized by muscle atonia with sudden bursts of muscle twitches and resultant limb jerks. Blumberg et al. (2013) have shown that this twitching is highly structured and that sleep-dependent twitches may provide a source of patterned neural activity with a role in sensory map development. Whether sleep disruptions during specific developmental times permanently disturb map establishment remains unknown.

A different approach seeks to exploit differences between wild-derived inbred strains to identify genetic factors that regulate sleep and wakefulness. For instance, inbred mouse strains are generally nocturnal but can exhibit phenotypic differences in total amounts of sleep, distributions of sleep throughout the day, and consolidation of sleep periods. These can be correlated with differences in whole brain monoamine contents, which are under elaborate genetic control (Hiyoshi et al. 2014). However, this approach has not been widely applied to ontogenetic changes in sleep. Moreover, some of the caveats in interpreting the results of pharmacological perturbations amid the complexity of mammalian brain development and sleep ontogeny may still apply to genetic studies.

Finally, sleep plays an important role in synaptic plasticity (Abel et al. 2013), which has been most closely studied leveraging a form of visual–cortical plasticity that is highest in early development (Frank 2011). This body of work perhaps speaks more broadly to a role for sleep in consolidating changes in synaptic strength rather than a specific or limited function in development, per se (Frank 2015). In sum, experiments in humans and mammalian models highlight how a mechanistic link between ontogenetic sleep changes and brain development has remained tantalizing but elusive and present an opportunity to leverage powerful genetic models to facilitate progress.

Worm Sleep

The nematode Caenorhabditis elegans sleeps during lethargus, a 2- to 3-hr long developmental stage that precedes ecdysis at the termination of each larval stage (Cassada and Russell 1975; Raizen et al. 2008). Lethargus is notably characterized by behavioral quiescence—a cessation of feeding and episodic suspension of head and body motion (Cassada and Russell 1975; Singh and Sulston 1978; Raizen et al. 2008), with distinct posture (Figure 1) (Schwarz et al. 2012; Tramm et al. 2014). During lethargus, responses to external stimuli are reduced or delayed, and deprivation of quiescence by mechano- or optogenetic stimulation is followed by rebound sleep (Raizen et al. 2008; Cho and Sternberg 2014; Nagy et al. 2014). Weaker disruption to the architecture of quiescence also evokes homeostatic compensation, suggesting a deep complexity to worm sleep (Driver et al. 2013; Iwanir et al. 2013; Nagy et al. 2014).

• Download figure
• Open in new tab
• Download powerpoint

Figure 1

Postural/positional changes during sleep (right) and wakefulness (left) in model organisms. Worms assume a hockey-stick posture (Tramm et al. 2014). Flies have been observed to show a head “droop” and preference to sleep near food (Hendricks et al. 2000). Zebrafish exhibit two postures: floating with head droop or remaining horizontal near the bottom of the tank (Zhdanova et al. 2001).

The developmental nature of lethargus inherently emphasizes a conserved role for sleep in maturation, but its coupling to molting is a key difference from adult sleep (Singh and Sulston 1978). This naturally leads to the suggestion that mechanical restriction is the primary cause of lethargus quiescence; however, such an assertion is inconsistent with several lines of evidence, most notably that the onset of quiescence is abrupt, that quiescence is rapidly reversible, and that brief periods of normal locomotion are interspersed with quiescent bouts (Raizen et al. 2008; Nelson and Raizen 2013; Nagy et al. 2013; Tramm et al. 2014). Cessation of locomotion and feeding can also be induced in adult C. elegans by satiety, stress, or genetic manipulations. Examples of the latter include heat-shock-induced activation of epidermal growth factor signaling (Van Buskirk and Sternberg 2007), mutations affecting cGMP, cAMP, or TGF-β signaling (Raizen et al. 2008; You et al. 2008; Iwanir et al. 2013), specific rescue of a peptidergic or insulin/insulin-like signaling pathway (Driver et al. 2013; Hill et al. 2014; Nagy et al. 2014; Nelson et al. 2014), or optogenetic activation of an avoidance response-promoting neuron (Cho and Sternberg 2014). Such findings potentially decouple worm sleep from molting. Moreover, a recent study suggests that the regulation of stress-induced quiescence is distinct from that of developmentally timed sleep (Trojanowski et al. 2015).

What function might lethargus quiescence serve? The molt is a major biosynthetic event (Singh and Sulston 1978; Nelson and Raizen 2013), so the involvement of sleep in this process lends credence to a hypothesized role for sleep in macromolecular biosynthesis (Mackiewicz et al. 2007; Nelson and Raizen 2013). In addition, large-scale synaptic formation and rearrangement in C. elegans occur during development. Eighty neurons (mostly motoneurons) of the 302 in the adult hermaphrodite develop postembryonically; moreover, synaptogenesis maintains synaptic density through the fivefold increase in body length during larval development (White et al. 1986). Interestingly, many synaptic formation and rearrangement events are temporally correlated with periods of lethargus (White et al. 1986). For instance, the dorsal GABAergic motor neurons (DD) remodel their synaptic pattern (from ventral to dorsal) during late first larval stage and its corresponding lethargus period (White et al. 1978; Walthall et al. 1993; Hallam and Jin 1998); migration of axons and synaptic formation in the egg-laying circuit likewise occur during this period (Shen et al. 2004). Therefore, lethargus provides the opportunity to study the link between sleep and development in this simple organism.

How similar is lethargus to sleep in mammals and other organisms? Genetically, deep cross-species conservation is found between lethargus, fruit fly sleep, and mammalian sleep (Singh et al. 2014). The C. elegans homolog of the circadian clock protein Period is required for clocking larval development and thus the timing of lethargus—its messenger RNA (mRNA) levels track the molting cycle (Jeon et al. 1999; Tennessen et al. 2006; Monsalve et al. 2011). Multiple additional conserved signaling pathways exhibit functional similarities in mammalian, insect, and nematode sleep, including epidermal growth factor signaling, protein kinase A (PKA) and G (PKG) activity, dopamine signaling, and pigment dispersing factor (PDF) signaling (Graves et al. 2003; Kramer et al. 2003; Snodgrass-Belt et al. 2005; Van Buskirk and Sternberg 2007; Foltenyi et al. 2007; Raizen et al. 2008; Langmesser et al. 2009; Choi et al. 2013, 2015; Iwanir et al. 2013; Singh et al. 2014).

Circuit mechanisms of C. elegans sleep and arousal likewise reveal parallels to those in flies, fish, and mammals. The nematode sleep circuitry includes multiple sensory circuits and classes of interneurons that display sleep-specific changes in activity and/or functional correlations (Schwarz et al. 2011; Choi et al. 2013, 2015; Iwanir et al. 2013; Cho and Sternberg 2014). Glutamate and PDF neuropeptides both mediate excitatory outputs from sensory neurons to promote arousal (Choi et al. 2013, 2015), consistent with pharmacological manipulations of glutamate in mammals (Juhász et al. 1990; Alam and Mallick 2008; Li et al. 2011) and the function of neuropeptides in vertebrates (orexin/hypocretin) and flies (PDF) (Sutcliffe and de Lecea 2002; Prober et al. 2006; Parisky et al. 2008; Choi et al. 2015). As both molecular and circuit mechanisms can be conserved across phyla, C. elegans lethargus can provide insight into core functions and regulation of sleep (Nelson and Raizen 2013; Singh et al. 2014).

Fly Sleep

Drosophila melanogaster, the fruit fly, was the first simple genetic system in which sleep was described (Hendricks et al. 2000; Shaw et al. 2000). Unbiased genetic screens have revealed numerous sleep-regulatory genes, pointing toward a deeper understanding of the genetics of sleep (Sehgal and Mignot 2011). Flies exhibit most behavioral features of sleep found in mammals, including a prolonged period of behavioral quiescence with increased arousal threshold, regulation by the circadian timing system, and homeostatic rebound in response to sleep loss (Hendricks et al. 2000; Shaw et al. 2000). In addition, sleep in flies as in mammals is suppressed with drugs that are wake promoting to humans, specifically caffeine and modafinil (Hendricks et al. 2000, 2001; Shaw et al. 2000). Remarkably, Drosophila recapitulate mammalian changes to sleep throughout their life span: young flies demonstrate increased sleep amount and depth, while old flies exhibit more fragmented sleep (Shaw et al. 2000; Koh et al. 2006; Seugnet et al. 2011a; Kayser et al. 2014). Drosophila life span is ∼60–80 days, and the features distinguishing juvenile sleep from that of the mature adult occur only during the first 3–5 days after eclosion (Seugnet et al. 2011a; Kayser et al. 2014). During this juvenile period, flies are far more somnolent than adult flies: they sleep more during the day, are more difficult to arouse from sleep at night, and are notably resistant to sleep deprivation (Kayser et al. 2014).

Conserved ontogenetic sleep changes between flies and mammals have presented an opportunity to leverage the powerful genetic toolkit of the fly toward understanding the mechanisms and purpose of juvenile sleep. Although developmental changes to sleep quality/quantity are conserved across multiple species, mechanisms controlling such changes remain underexplored. Recently, work in the fly has demonstrated that a small group of dopamine neurons undergo developmentally regulated changes in activity that likely control sleep ontogeny (Kayser et al. 2014). These few neurons project to and inhibit a sleep-promoting center in the fly brain, the dorsal fan-shaped body (Liu et al. 2012; Ueno et al. 2012). In young flies, the dopaminergic neurons are less active, allowing increased activity of the dorsal fan-shaped body (dFSB) and increased sleep; with maturity, dopaminergic activity increases, dFSB activity is reduced, and mature sleep quantity and quality are achieved (Kayser et al. 2014). Dopamine serves a wake-promoting function in mammals as well, so this basic developmental paradigm may represent a conserved logic for control of sleep ontogeny (Figure 2).

• Download figure
• Open in new tab
• Download powerpoint

Figure 2

Sleep ontogeny changes across phylogeny. Flies, fish, and worms all exhibit developmentally regulated changes to sleep amount/timing in a manner akin to that of humans. Fruit flies and zebrafish have increased sleep amount in early life compared to mature adults. Nematodes similarly exhibit developmental timing to sleep patterns.

Fish Sleep

Like roundworms and fruit flies, zebrafish (Danio rerio) also sleep. Fish show brief periods during which they stop swimming and are immobile, with increased arousal threshold; these behaviors occur nearly exclusively during the night (zebrafish are diurnal) (Zhdanova et al. 2001; Yokogawa et al. 2007; Appelbaum et al. 2009). Recent work leveraging the genetic tractability of this vertebrate model organism has uncovered a role for melatonin in conveying the circadian control of sleep timing (Zhdanova et al. 2001; Gandhi et al. 2015). Moreover, zebrafish utilize the hypocretin/orexin signaling system (Zhdanova 2006; Appelbaum et al. 2010; Panula et al. 2010; Yelin-Bekerman et al. 2015), abnormalities in which are known to cause the human sleep disorder narcolepsy. Similar to its mammalian analog, the zebrafish hypocretin system is located in the hypothalamus and regulates sleep–wake transitions (Prober et al. 2006; Yokogawa et al. 2007; Appelbaum et al. 2009; Elbaz et al. 2013). It comprises merely 20–60 hypocretin neurons, several orders of magnitude fewer than those found in mice (Kaslin et al. 2004). Zebrafish genetics were exploited to identify a minimal promoter region capable of mimicking the native expression pattern of fluorescent protein in hypocretin neurons, implicate an AMPA receptor clustering protein in regulation of structural synaptic plasticity, tag and sequence hypocretin neurons, and use a CRISPR-mediated gene inactivation system to implicate a voltage-gated potassium channel in regulating sleep (Faraco et al. 2006; Appelbaum et al. 2010; Yelin-Bekerman et al. 2015). Therefore, the zebrafish hypocretin system can serve as a simplified model, allowing sleep–wake regulation to be studied at nearly single-neuron resolution (Kaslin et al. 2004; Faraco et al. 2006).

Zebrafish share with C. elegans the optical advantage of transparency, allowing for the ability to monitor neuronal function/structure in real time in an organism that is anatomically similar to mammals. Capitalizing on this advantage, Appelbaum et al. (2009, 2010) showed that synaptic outputs of the hypocretin system are under circadian control, perhaps offering a mechanism linking the sleep and circadian systems. Although the neural circuits that regulate sleep in fish are not well understood at this time, the innervation pattern of the zebrafish hypocretin system resembles mammalian circuits for regulating sleep and wakefulness. For instance, serotonergic nuclei are densely innervated by hypocretin immunoreactive neurons (Kaslin et al. 2004; Panula et al. 2010). Detailed behavioral studies revealed conservation between fish and mammalian responses to pharmacological agents that promote or antagonize sleep (Rihel et al. 2010; Sigurgeirsson et al. 2011).

Do zebrafish exhibit ontogenetic sleep changes? Research in fish sleep has been undertaken both at the larval stage and in mature adults; however, until recently, a direct comparison of sleep features throughout development has not been addressed. By examining four different developmental time points within the same monitoring system, it is now clear that zebrafish undergo dramatic sleep changes throughout development, with more sleep early in life. The excess of early sleep derived primarily from an increase in the number of sleep bouts (defined as periods of immobility >6 sec in duration) (Sorribes 2013). Thus, in addition to conserved anatomical and molecular sleep substrates between zebrafish and humans, sleep ontogenetic patterns are also conserved. These similarities support the notion of conservation of neural mechanisms for regulating sleep and wakefulness.

The Dolphin Dilemma

Increased sleep in early life is a widespread phenomenon, but is it universal, and, if not, how does that inform a hypothesized essential role for sleep during development? While controversial, some data suggest cetaceans (bottlenose dolphins and killer whales) actually show less typical sleep (rest on the surface) during the first month following birth; mothers likewise suppress sleep during this time, perhaps to remain vigilant in the postpartum period (Lyamin et al. 2005). The observed activity was not accompanied by stress as assayed by levels of the hormone cortisol. The authors hypothesize that constant activity could be advantageous to newborns both to avoid predation and to maintain body temperature while development ensues (Lyamin et al. 2005). This research was soon followed by two additional studies reaching the opposite conclusion (Gnone et al. 2006; Sekiguchi et al. 2006). Dolphins exhibit unihemispheric sleep, where slow-wave electroencephalogram (EEG) patterns restricted to a single hemisphere are linked to closure of the contralateral eye (Mukhametov 1987; Lyamin et al. 2002, 2004). Swimming and unihemispheric sleep are compatible: the underwater sleeping behavior termed “swim rest” is associated with continuous activity. The initial work may have underestimated swim rest (and perhaps alternative sleep strategies), as both follow-up studies concluded that bottlenose dolphin mothers and calves experience significant cumulative periods of swim rest during the postpartum period.

What are the potential implications of these findings for theories of the role of sleep in development? In utero sleep is prominently exhibited by all mammalian prenatals that have been carefully observed (Ruckebusch 1972; Drucker-Colín et al. 1979; Awoust and Levi 1984; Szeto and Hinman 1985). Human fetal sleep begins early in the third trimester of pregnancy. Synchronous waves of neuronal activity occur during REM sleep and are widely thought to be essential for synaptic formation and thus for the proper development of neurosensory systems, motor systems, and memory circuits (Marks et al. 1995; Mirmiran 1995; Segawa 1999). In contrast, there have been no reports of REM sleep in cetaceans. Disambiguating sleep states based on cetacean behavior is challenging, but even twitches (observed in adults) that may correspond to REM sleep are extremely sporadic (Siegel 2005). Thus, the minimal amounts or absence of REM sleep may reflect peculiarities of the cetacean brain (Siegel 2005; Lyamin et al. 2008; Dell et al. 2015; Patzke et al. 2015). A detailed understanding of prenatal cetacean sleep patterns would provide context to the observations of postpartum behavior. A better mechanistic understanding of the roles of sleep in developing fish, flies, and nematodes would likewise inform our perspective.

Conservation of Sleep-Related Functions with “Developmental” Genes

Using unbiased and candidate-based genetic screens in model organisms, a cluster of genes with essential roles in basic developmental processes have been found to influence sleep in adulthood. Although a dual function in these processes is not unassailable evidence for a core role of sleep in development, it does suggest the two may be intimately linked at a genetic level. For example, several lines of evidence implicate Notch signaling in regulating nematode and fly sleep. In worms, perturbing the functions of either Notch receptors (LIN-12 or GLP-1) or ligands (OSM-7 and OSM-11) altered the total quiescence observed during L4 lethargus, albeit in complex fashion (Singh et al. 2011). Moreover, decreasing Notch signaling impaired sensory gating, i.e., reduced the arousal threshold of quiescent L4 lethargus larvae. Overexpression of the Notch ligand OSM-11 induced anachronistic sleep-like quiescence in adult nematodes. Interestingly, Notch signaling regulates quiescence through pathways that were previously implicated in regulating sleep in invertebrates and vertebrates alike, such as epidermal growth factor (EGF) (see below) and cGMP-dependent kinase (PKG) signaling (Singh et al. 2011). In Drosophila, a specific role for Notch in sleep rebound, i.e., in compensating for lost sleep following a delay in sleep onset or a disruption to sleep quality, has been proposed (Seugnet et al. 2011b). Following sleep deprivation, a negative regulator of Notch (bunched) is upregulated, while overexpression of Notch or its ligand Delta reduces sleep rebound. This works suggest that sleep deprivation normally suppresses Notch via upregulation of bunch, which permits the homeostatic sleep response.

Notably, spatial restriction of enhanced Notch signaling or overexpression of its ligand in a known sleep-promoting region (the mushroom body) is sufficient to reduce rebound sleep, arguing for anatomical specificity in Notch’s role. Despite apparent opposing effects on sleep for Notch in flies and worms, these works emphasize how a canonical developmental gene can adopt a critical role for sleep behavior (Wu and Raizen 2011).

In addition to Notch signaling, EGF (Foltenyi et al. 2007), Cyclin A (Rogulja and Young 2012), and ecdysone (Ishimoto and Kitamoto 2010) all appear to regulate sleep as well as function in their more conventional developmental roles. Overexpression of EGFR ligands in worms and flies induces sleep-like behaviors (Foltenyi et al. 2007; Van Buskirk and Sternberg 2007; Raizen et al. 2008), and this behavioral state change acts through EGFR in both animals. The initial description of nematode sleep onset as molt specific was influential in the examination of a role for ecdysone—the major Drosophila steroid hormone—in fly sleep (Ishimoto and Kitamoto 2010). Ecdysone has a critical role in developmental molting transitions, including emergence of the adult fly during eclosion, after which sleep amounts/depth are highest. In mature adults, exogenous administration of ecdysone’s active metabolite 20-hydroxyecdysone results in a dose-dependent increase in sleep amount, while sleep is reduced in ecdysone steroid or ecdysone receptor mutants (Ishimoto and Kitamoto 2010). Ecdysone thus represents an attractive candidate in Drosophila for coupling developmentally regulated processes to sleep need.

Ramification of Sleep Loss During Development

Ontogenetic sleep changes are conserved across phylogeny, and even in humans it is clear that sleep disturbances during critical developmental windows can have outsized neurobehavioral effects later in life. So, what happens to structural maturation of the brain when sleep in early life is perturbed, and are there global rules governing why certain brain regions might be affected more than others? In adolescent mouse development, most studies have focused on the role for sleep in structural plasticity after the phase of robust synaptogenesis (which occurs in first ∼3 postnatal weeks), during which time synaptic pruning is, on balance, predominant. In vivo studies suggest that changes to dendritic spine number (and likely synapses) depend on sleep–wake state (Maret et al. 2011; Yang and Gan 2012), whereas in adults, sleep influences synaptic strength without changing the number of synaptic contacts (Maret et al. 2011). More recent work has closely examined the idea that sleep ontogeny in mice, which like in humans is characterized by a notable decline in slow wave activity beginning in adolescence, reflects pruning of synaptic contacts. Interestingly, the decrease in slow wave activity is uncoupled from any measureable changes to synapse number or size, unlike in more mature stages in which slow wave activity is hypothesized to be a marker for and perhaps driver of synaptic homeostasis (de Vivo et al. 2014). The data overall support that idea that during highly plastic developmental periods, sleep may sculpt synapses in a manner distinct from adulthood and thereby play a prominent role patterning neural circuits in early life. Indeed, during developmentally timed, robust stages of synaptogenesis, sleep may function specifically to drive addition of new synapses (Kayser et al. 2014). Studying sleep longitudinally during this time frame in mice is technically challenging, but work in fruit flies has provided novel insights.

Behavioral work in Drosophila has established that, as in mammals, sleep deprivation during sensitive developmental windows has long-lasting ramifications compared to sleep loss in adulthood. Mechanical sleep deprivation in juvenile flies leads to deficits in both courtship behaviors and performance in learning/memory assays as mature adults (Seugnet et al. 2011a; Kayser et al. 2014). Might these changes simply result from heightened stress susceptibility during development? Leveraging the mechanistic understanding of circuits controlling sleep ontogeny in the fly, a specific wake-promoting circuit that is normally quiescent in young flies can be experimentally hyperactivated at inappropriate developmental times to more specifically probe the question. With this approach, sleep deprivation of juvenile flies has been associated with abnormal social behaviors in adulthood, without similar defects when sleep loss occurs in maturity (Kayser et al. 2014). Importantly, this effect does not appear to stem from a general “monoaminergic flood” during development, as restricted manipulation of sleep–wake circuits impairs behaviors arising through unrelated brain regions (Kayser et al. 2014). This finding, however, resurrects an older question: How are sleep disturbances in early life linked to later behavioral abnormalities? In the fly, courtship deficits in adult flies that were sleep deprived when young are accompanied by abnormal development of a specific olfactory glomerulus, known as VA1v, that plays a pheromone-related role in social behaviors (Dweck et al. 2015; Lone et al. 2015). Surprisingly, VA1v grows more rapidly during early posteclosion development compared to its neighboring glomeruli (Kayser et al. 2014), likely reflecting a higher rate of synapse addition. Together these findings suggest that sleep in early life is particularly crucial for areas undergoing high levels of synaptogenesis. This hypothesis can be tested across species and developmental time windows and could provide a global rule for why sleep amount/depth is highest (and impact of sleep disruptions most severe) during periods of explosive brain growth.

Sleep in Neurodevelopmental Mutants

Further highlighting the important link between sleep and development, sleep disturbances are pervasive across neurodevelopmental disorders in humans (Angriman et al. 2015). While difficult in young children to disentangle whether sleep abnormalities precede other neurocognitive symptoms, it is clear that sleep disturbances exacerbate an already challenging constellation of symptoms. Can model systems elaborate our understanding of sleep in neurodevelopmental disorders? Work in Drosophila indicates abnormalities in sleep or circadian rhythms are highly conserved across species. Fly models of fragile X syndrome (FXS) display aberrant circadian output (Dockendorff et al. 2002) and sleep. In particular, flies with a loss-of-function mutation in dFmr1 display high sleep need (Bushey et al. 2009), perhaps with a specific abnormality of sleep ontogenetic change. Recent work indicates abnormal sleep in a fly model of neurofibromatosis 1 (NF1) (Bai and Sehgal 2015), and NF1 has previously been implicated in circadian output circuitry (Williams et al. 2001). Likewise, rest–activity rhythms are weakened in a Drosophila Angelman syndrome model (Wu et al. 2008) although sleep itself has yet to be examined. These fly models of neurodevelopmental disorders provide an opportunity to dissect sleep by development interactions with the goal of a deeper genetic/molecular understanding of sleep in neurodevelopmental disorders. Of note, this line of work will not be limited to the fly going forward: a zebrafish model of Allan–Herndon–Dudley syndrome (ADHS) exhibits both abnormal (high) sleep levels and responses to light–dark transitions (Zada et al. 2014), and it is likely that additional models will continue to emerge. There are multiple questions uniquely suited to examination in simple genetic systems: Are specific sleep/circadian neural loci perturbed in neurodevelopmental disorders? Are common sleep-relevant genetic targets affected across these models? And can restoration of normal sleep improve performance in other affected behaviors?

Complex Questions, Simple Models

Simple models were foundational to the development of molecular biology (see, e.g., Cairns et al. 2007), and newly available molecular approaches enabled the launch of neurogenetics, using invertebrate models to investigate the role of genetic pathways in regulating behavior and neurophysiology (Brenner 1974; Harris 2008). A prominent example invoked fly genetics to resolve the question of whether circadian rest–activity rhythms are endogenous or exogenously controlled. The molecular mechanisms underlying circadian clocks, first identified in a D. melanogaster screen (Konopka and Benzer 1971), were later found to be largely conserved in mammals (Takahashi et al. 2008); moreover, a human circadian rhythm disorder can be caused by mutations in the homolog of per, the first identified clock gene (Toh et al. 2001).

Perhaps the study of sleep—in particular during development—can similarly benefit from tractable models (Table 1). For example, beyond questions of development alone, an integrated understanding of the neural circuitry regulating sleep and arousal remains unsolved in any organism. The nervous systems of D. melanogaster and C. elegans are amenable to systematic analyses toward providing a basic neural logic of sleep–wake balance (Choi et al. 2015; Sitaraman et al. 2015). Another key set of questions concerns the role played by sleep in reorganization of neural circuits during development. In contrast to numerous studies in vertebrates, data on lasting deficits incurred by invertebrate sleep loss during development are sporadic. Sleep loss in young adult flies leads to lasting deficits in social courtship behaviors traced to impaired development of a specific brain region (Kayser et al. 2014). Similarly, behavioral defects can be observed after partial sleep deprivation of C. elegans: nonlethal yet severe sleep deprivation during the fourth larval sleep period results in deficits in feeding behavior that persist in the adult animals (D. Biron, unpublished results). The degree of sleep-dependent reorganization in invertebrate neural circuits, activity in these circuits during sleep, and how such activity might promote appropriate neurodevelopment invite careful examination, which we suggest is best accomplished initially in species amenable to sophisticated genetics.

Lasting deficits from developmentally specific sleep loss naturally lead to asking how such ill effects might be mitigated; i.e., what mechanisms facilitate compensation for disruptions or aid recovery on longer timescales. Stress response pathways, e.g., insulin/insulin-like (IIS) or heat-shock factor (HSF)-dependent signaling, are activated by disrupting sleep in flies and worms (Shaw et al. 2000, 2002; Cirelli et al. 2006; Naidoo 2009; Brown and Naidoo 2010; Varshavsky 2012; Driver et al. 2013; Nagy et al. 2014). These findings indicate that, like mammalian sleep, invertebrate sleep can be a period of heightened vulnerability when otherwise benign stimuli act as stressors (Shaw et al. 2002; Driver et al. 2013; Nagy et al. 2014). In addition to the mechanistic links between sleep loss and other stressors, broader parallels can be drawn between research performed in the two fields. Typically, protective responses, e.g., to heat shock, are complex and employ both cell-autonomous and cell-nonautonomous signaling pathways (Carvalhal Marques et al. 2015; Mardones et al. 2015). Likewise, pathways that protect sleep and aid recovery after deprivation can span scales from the cellular to the organismal, involve intertissue communication, and employ complex cellular signaling. Notably, tractable models prominently contribute to understanding classical protective responses (Carvalhal Marques et al. 2015; Mardones et al. 2015). Precisely because susceptibility to sleep loss is elevated in early life, considering the protective responses evoked by sleep disruptions during development is a natural approach for shedding light on generalizable functions of sleep. A deeper understanding of how developing organisms cope with sleep loss at a cellular level would likewise inform why sleep is so tightly protected in immature brains and delineate cellular signatures of sleep states throughout life.

A Link Between Development and the Ubiquity of Sleep?

Sleep remains a mysterious behavior, from its core functions at the cellular and systems level, to how sleep itself is adaptive (or maladaptive), to mechanisms compensating for sleep loss, to even whether sleep is universally essential (Campbell and Tobler 1984; Cirelli and Tononi 2008; Siegel 2009; McNamara et al. 2010). There are many arguments in support of the universality of sleep, but conservation of ontogenetic sleep changes with an active role in the developing nervous system is emerging as one of the strongest (Roffwarg et al. 1966; Jouvet-Mounier et al. 1970; McGinty et al. 1977; Shaw et al. 2000; Kirov and Moyanova 2002; Raizen et al. 2008; Hasan et al. 2012; Todd et al. 2012; Sorribes 2013; Kayser et al. 2014). The observation that sleep amount is highest during developmental periods across species has led to extensive and ongoing explorations of the ontogenetic hypothesis, suggesting that sleep promotes normal brain development by providing necessary endogenous activity (Roffwarg et al. 1966; Jouvet-Mounier et al. 1970; Oksenberg et al. 1996; Shaffery et al. 1999; Frank 2011; Blumberg et al. 2013; Kayser et al. 2014; Tononi and Cirelli 2014). The self-evident fact that all nervous systems originate from a single cell can be viewed as a truly universal constraint imposed by the very nature of development. Within a restricted period, which cannot be trivially prolonged, all nervous systems must undergo structural plasticity on a scale that is macroscopic with respect to their own size (Figure 3). Thus, the feat facing developing nervous systems (and perhaps other tissues) across the animal kingdom shares a common barrier, for which sleep may provide a vital solution. Genetically tractable organisms are particularly useful for understanding such ancient functions of sleep during development and the conserved (or convergent) mechanisms that regulate them.

• Download figure
• Open in new tab
• Download powerpoint

Figure 3

The omnipresence of sleep may be a signature of its contribution to satisfying a constraint inherent to development. Should the time pressure of development be thought of as a universal constraint that can be mitigated by a sleep–wake cycle? All nervous systems originate from a single cell. In a restricted period of time, they must undergo structural plasticity (e.g., synaptogenesis) on a scale that is comparable to their full size. If sleep provides vital aspects of the solution to this challenge, its ubiquity in the animal kingdom may be related to the common feat of development (or neurodevelopment in particular). Much like the contributions of simple organisms to understanding developmental processes, facile genetics and tractable anatomy are key to understanding ontogenetic changes, core functions, and evolutionary origins of sleep.