TRANSLATION OF BRAIN ACTIVITY INTO SLEEP

Abstract

Cytokines including tumor necrosis factor alpha (TNF) play a role in sleep regulation in health and disease. Hypothalamic and cerebral cortical levels of TNF mRNA or TNF protein have diurnal variations with higher levels associated with greater sleep propensity. Sleep loss is associated with enhanced brain TNF. Central or systemic TNF injections enhance sleep. Inhibition of TNF using the soluble TNF receptor, or anti-TNF antibodies, or a TNF siRNA reduces spontaneous sleep. Mice lacking the TNF 55 kD receptor have less spontaneous sleep. Injection of TNF into sleep regulatory circuits, e.g. the hypothalamus, promotes sleep. In normal humans, plasma levels of TNF co-vary with EEG slow wave activity (SWA) and in multiple disease states plasma TNF increases in parallel with sleep propensity. Downstream mechanisms of TNF-enhanced sleep include nitric oxide, adenosine, prostaglandins and activation of nuclear factor kappa B. Neuronal use induces cortical neurons to express TNF and if applied directly to cortical columns TNF induces a functional sleep-like state within the column. TNF mechanistically has several synaptic functions. TNF-sleep data led to the idea that sleep is a fundamental property of neuronal/glial networks such as cortical columns and is dependent upon past activity within such assemblies. This view of brain organization of sleep has profound implications for sleep function that are briefly reviewed herein.

INTRODUCTION

The accumulation of sleep regulatory substances (SRSs) in cerebrospinal fluid during prolonged waking was described a century ago by Ishimori¹⁾ and a few years later by Legendre and Pieron²⁾. Multiple modern studies have confirmed these pioneering investigations (reviewed³⁾). These studies provided strong evidence for the hypothesis that sleep is regulated, in part, by humoral agents^3–12). Many substances can affect sleep³⁾ and some SRSs meet all the criteria for SRSs (for criteria see^{10, 11, 13, 14)}; Table 1). The list includes tumor necrosis factor alpha (TNF), interleukin-1 beta (IL1), growth hormone releasing hormone (GHRH), prostaglandin D², and adenosine for non-rapid eye movement sleep (NREMS) and vasoactive intestinal polypeptide for rapid eye movement sleep (REMS) (reviewed³⁾). Additional putative SRSs include prolactin⁹⁾, hypocretin^{12, 15, 16)}, oleamide¹⁷⁾, nerve growth factor^18–20), brain derived neurotrophic factor (BDNF)^{21, 22)}, and NO^23–37). The various SRSs affect each other’s production and act in concert with each other to affect sleep^{3, 4, 5, 8)}. For instance, TNF induces IL1β, prostaglandins, NO, adenosine, and the gluR1 component of the glutamate AMPA receptor (reviewed⁴⁾). This review focuses on TNF because it is a fully characterized SRS and TNF sleep-related findings have led to a new view of how the brain is organized to produce sleep. Further, this work led to answers to the question; what exactly is it that sleeps? Our view of brain organization of sleep^{4, 38)} and the role that TNF plays in the translation of CNS cellular activity into sleep^{5, 39)} has led to revolutionary hypotheses about sleep regulation and brain organization of sleep. These hypotheses are presented herein. This essay ends with a brief summary of what the reviewed SRS literature implies for sleep function.

Table 1.

Criteria for Sleep Regulatory Substances

If injected sleep is enhanced If inhibited, spontaneous sleep is reduced Brain levels vary with sleep propensity Acts on known sleep regulatory circuits Levels vary with pathologies

TNF IS A SRS

TNF is well characterized for its role in several CNS functions including fever^{40, 41)}, food intake⁴²⁾, brain development ⁴³⁾, neurotransmission^44–47) and synaptic scaling^{48, 49)}. TNF increases during sleep deprivation (SD) ^{4, 5, 50)}. Sleep deprivation (SD) is associated with sleepiness, sleep rebound, cognitive dysfunction, and enhanced sensitivity to pain and kindling stimuli. All of these sleep loss symptoms can be induced in animals or humans by administration of exogenous TNF to normal subjects (reviewed⁵⁾); thus, TNF treatment mimics the increased TNF observed during SD. TNF is expressed by microglia, astrocytes, and neurons^{39, 51–54)}. It can mediate both brain damage and neuroprotection. Whether TNF is protective or damaging may depend upon the TNF receptor type present, TNF-55 kD receptor or the TNF 75 kD receptor^55–57), as well as the stimulus context and the presence or absence of substances that modify TNF activity^{58, 59)}.

TNF is synthesized as a 26 kD membrane-associated protein⁶⁰⁾. Soluble TNF, a 17 kD protein, is cleaved from the 26 kD membrane associated protein by TNF converting enzyme. TNF production is tightly regulated in a tissue-specific manner⁶¹⁾. The 5′ flanking region of the TNF gene contains several regulatory sites. TNF induces its own expression, in part, via a nuclear factor kappa B (NFκB) regulatory site.

There are two TNF cell-surface receptors, the 55 kD TNF receptor and the 75 kD TNF receptor. The intracellular domains of these receptors lack intrinsic enzymatic activity. Rather, both receptors signal by recruitment of cytosolic proteins via protein-protein interaction domains. The diversity of these adaptor proteins and their ability to interact with other members of the TNF receptor superfamily helps explain the pleiotropic actions of TNF. The TNF receptor family is divided into 2 large groups based on the adaptor proteins recruited in response to ligand binding^62–66). One major TNF activated signaling pathway prevents cell death, while another leads to cell death⁶²⁾. The ectodomains of both receptors are shed to form soluble receptors. The TNF transmembrane protein has biological activity and it can signal intracellularly and thus acts as a receptor for the soluble TNF receptors⁶⁷⁾.

TNF and its receptors have a role in sleep and sleep disorders. Plasma levels of TNF correlate with EEG δ wave power, an index of sleep intensity, in normal individuals^{68, 69)}. Further, changes in plasma TNF or TNF receptors are linked to sleep disturbances in many diseases including; sleep apnea^70–75), insomnia⁷⁶⁾, Parkinson’s disease^{77, 78)}, myocardial infarction ⁷⁹⁾, preeclampsia ⁸⁰⁾, AIDS⁶⁸⁾, alcoholism⁸¹⁾, infectious diseases⁸²⁾ and many more (reviewed⁵⁾). Clinically approved inhibitors of TNF, e.g. etanercept, reverse the sleepiness and fatigue associated with sleep apnea⁸³⁾ rheumatoid arthritis⁸⁴⁾, ankylosing spondylitis⁸⁵⁾ and alcoholism⁸⁶⁾. Surgical treatment for ^87–89), or CPAP treatment of^{90, 91)} obstructive sleep apnea reduces TNF/TNF receptor plasma levels. Polymorphic variants of TNF are linked to narcolepsy^92–94), obesity⁹⁵⁾, sleep apnea⁹⁶⁾ and poor performance associated with sleep loss⁹⁷⁾. Given these clinical and experimental foundations showing the involvement of TNF in sleep in health and disease it is likely that the importance of TNF to the sleep clinic will grow.

TNF enhances duration of NREMS in all species thus far tested: rabbits⁹⁸⁾, mice ⁹⁹⁾, rats¹⁰⁰⁾, and sheep¹⁰¹⁾. The effects are big, e.g. mice receiving 3 μg TNF ip spend about 90 minutes extra in NREMS during the first 9 hours post-injection⁹⁹⁾. TNF also enhances EEG δ power during NREMS⁹⁹⁾. TNF affects REMS; with low NREMS-promoting TNF doses REMS is not affected, however, higher doses inhibit REMS. Sleep following TNF treatment appears normal (reviewed^{3, 5)}).

Inhibition of TNF inhibits spontaneous NREMS whether anti-TNF antibodies¹⁰²⁾, the full-length soluble TNF receptor¹⁰³⁾, or TNF soluble receptor fragments containing the TNF recognition site¹⁰⁴⁾ are used. Pretreatment of animals with TNF inhibitors prior to SD reduces the expected sleep rebound¹⁰⁴⁾. Substances that inhibit TNF action or production, directly or indirectly, also inhibit spontaneous sleep, e.g., IL4, IL10, TGF and IL13. Furthermore, inhibition of TNF also blocks the increases in NREMS observed in response to an acute mild increase in ambient temperature¹⁰⁵⁾.

Mice lacking the TNF 55 kD receptor fail to exhibit NREMS responses if given TNF, thereby implicating this receptor in TNF-enhanced sleep⁹⁹⁾. These mice have less NREMS and REMS than corresponding control strains. Mice lacking both TNF receptors also have less spontaneous sleep¹⁰⁶⁾. One report¹⁰⁷⁾ showed the changes in REMS we described in TNF receptor-deficient mice but failed to show changes in NREMS. However, in that study inappropriate controls were used and there was no demonstration that the mice were deficient in the TNF receptor.

Hypothalamic levels of TNF^{5, 108)} and the TNF mRNA^109–111) vary diurnally. The highest levels in rats occur at daybreak. The amplitude of the day-night changes in TNF protein is about 10-fold and TNF mRNA about 2-fold. This reflects the predominate post-transcriptional regulation of TNF. After SD, hypothalamic TNF mRNA also increases^{110, 111)}. SD also increases brain expression of the 55 kD TNF receptor mRNA¹¹¹⁾. TNF serum levels increase in mice after SD, but not after stress⁵⁰⁾. In normal humans, blood levels of TNF correlate with EEG δ wave power⁶⁸⁾. After SD, circulating levels of TNF¹¹²⁾ and the 55 kD soluble TNF receptor, but not the 75 kD TNF soluble receptor, increase^{69, 113)}. The 55 kD soluble receptor is a component of normal cerebrospinal fluid¹¹⁴⁾.

Systemic TNF, like IL1, likely signals the brain via multiple mechanisms; one involves vagal afferents since vagotomy attenuates ip-TNF-induced NREMS responses¹¹⁵⁾. The effects of systemic bacterial products such as endotoxin may also involve TNF¹¹⁶⁾. For instance, in humans, endotoxin doses that induce transient increases in sleep induce concomitant increases in circulating TNF¹¹⁷⁾.

The sites of action of TNF-induced NREMS include the hypothalamic preoptic area and the locus coeruleus. Microinjection of TNF into these areas promotes sleep^118–119). In contrast, injection of a soluble TNF receptor fragment into the hypothalamus inhibits spontaneous NREMS¹¹⁸⁾. Microinfusion of TNF into the subarachnoid space beneath the basal forebrain enhances NREMS and reduces REMS¹²⁰⁾. Unilateral application of TNF onto the surface of the somatosensory cortex induces unilateral state-dependent increases in EEG δ wave power¹²¹⁾. Conversely, unilateral cortical application of the soluble TNF receptor reduces EEG δ power during the NREMS occurring after SD¹²¹⁾. Associated with the changes in the TNF-altered EEG δ power are enhancements of Fos immuno-reactivity (IR) and IL1-IR in the somatosensory cortex and reticular thalamus¹²²⁾.

Several other cytokines also promote sleep; the list includes BDNF, IL1, IL2, IL6, IL8, IL15, IL18, epidermal growth factor, acidic fibroblast growth factor, colony stimulating factors, and interferons (reviewed⁵⁾). In contrast, other cytokines inhibit NREMS including the IL1 receptor antagonist (IL1RA), IL4, IL10, IL13, insulin-like growth factor, transforming growth factor beta and the soluble IL1 and TNF receptors (reviewed^{3, 5, 11)}). Collectively these data strongly suggest that the brain cytokine network participates in sleep regulation (reviewed^{3, 5, 11)}). Some cytokine-associated substances, such as the IL1RA and the TNF and IL1 soluble receptors act as endogenous antagonists and inhibit spontaneous sleep^{123, 124)}. Anti-somnogenic cytokines act, in part, by inhibiting production of pro-somnogenic cytokines (reviewed⁵⁾). These include IL4 and IL13.

IL1 and TNF may also provide a bridge between the circadian rhythm and the sleep homeostat. There are daily rhythms in brain cytokines including IL1 and TNF (reviewed^{3, 11)}). Removal of the type I IL1 receptor results in sleep deficits that are limited to the night hours¹²⁵⁾. Finally, IL1 and TNF inhibit expression of some clock genes via interfering with CLOCK-BMAL1-induced activation of E-box regulatory elements¹²⁶⁾.

BRAIN ORGANIZATION OF SLEEP

Our theory^{4, 38, 127)} and others^128–131), posits that sleep is a local network process dependent upon past neuronal use. There are several well-characterized neuronal circuits involved in sleep regulation (reviewed e.g.^132–136)). An inherent fundamental premise of the sleep regulatory circuitry literature is that such circuits impose sleep on the brain; hence sleep is viewed as being initiated by these circuits[¹]. The sleep regulatory circuit paradigm lacks explanation for, or is silent on, the mechanisms underlying many well-known sleep phenomena, e.g. performance decrements associated with prolonged wakefulness, sleep inertia, sleep homeostasis, many sleep parasomnias such as sleep-walking, and the reoccurrence and reorganization of sleep after lesions. In contrast, the local use-dependent sleep theory addresses these issues to the extent that it is easy to envision how these deficits could arise if local areas of the brain were asleep while other areas were awake^{4, 129)}.

IL1¹³⁷⁾, TNF¹²¹⁾, GHRH¹³⁸⁾, and BDNF¹³⁹⁾ have the capacity to act locally within the cortex to alter a sleep phenotype. Unilateral application of IL1, TNF or GHRH to the surface of the cortex enhances EEG δ power during NREMS, but not during REMS or waking, on the ipsilateral side but not on the contralateral side suggesting that sleep is more intense on the side receiving these SRSs¹⁴⁰⁾. These unilateral changes in EEG δ power are associated with changes in cFos and IL1 immuno-reactivity ( IR) in the corresponding cortical areas and reticular thalamus¹⁴⁰⁾, suggesting the involvement of the biochemical sleep regulatory cascade and known corticothalamic sleep regulatory circuitry¹³²⁾. These data coupled with what is known about use-dependent production of IL1, TNF, and BDNF strongly support the idea that sleep is targeted to active circuits and is initiated at a local network level. Further, local application of TNF to cortical columns is associated with cortical column state changes³⁹⁾. Finally, dispersed co-cultures of neurons and glia form networks that exhibit spontaneous activity and oscillate between states¹⁴¹⁾. If such cultures are stimulated, cytokine expression is enhanced¹⁴²⁾. These data suggest that state oscillations result from activity-induced cytokine expression and that sleep, manifests in the whole animal, as an emergent property of multiple local networks.

If a subject survives a brain lesion, whether experimental or pathological, for a few days or more, it sleeps. There are no reported cases of subjects with complete lack of sleep, including those with fatal familial insomnia¹⁴³⁾. This is an important meta-finding for sleep research because it indicates that sleep is a property of any surviving group of neurons. From comparative studies, it appears that many species of birds and marine mammals exhibit unihemispheric sleep^144–146). A defining characteristic of NREMS, EEG δ waves, has in part a local cortical origin¹³²⁾. Further, isolated cortical islands that retain blood flow, wax and wane through periods of high amplitude δ waves¹⁴⁷⁾. Clinical evidence also indicates that the brain can be awake and asleep simultaneously, e.g. parasomnias such as sleep walking¹⁴⁸⁾.

The idea that sleep is a local process is directly supported by the finding that cortical columns oscillate between sleep-like and wake-like states¹⁴⁹⁾. Further sleep intensity, a sleep phenotype determined from EEG δ power, is dependent upon prior use and is targeted and localized to areas disproportionately used during prior wakefulness. EEG δ power is enhanced in the left somatosensory cortex compared to the right during NREMS after prolonged right hand stimulation prior to sleep onset¹⁵⁰⁾. Other evidence is consistent with the idea that sleep is a regional property of the brain that is dependent upon prior activity. In mice, rats, chickens, pigeons, humans and cats, if a localized area is disproportionately stimulated during waking, EEG δ power in that area is enhanced during subsequent NREMS^151–158). There are also several findings showing that cerebral blood flow during sleep is enhanced in those areas disproportionately stimulated during prior waking^159–160). Finally, the developmental plasticity literature^{131, 161–163)} and the learning literature demonstrating replay of neuronal electrical patterns associated with waking learning tasks¹⁶⁴⁾, indicate that changes in the EEG during sleep are targeted to areas activated during prior waking.

Mechanistically our hypothesis is summarized as follows⁴⁾: 1) Neuronal activity is associated with glio- and neurotransmission co-release of ATP; 2) the consequent increase in extracellular ATP thus provides an index of prior local neuronal activity; 3) the ATP is detected by nearby purine type 2 receptors causing the release of sleep regulatory cytokines such as TNF, IL1 and BDNF and this provides for the translation of prior neuronal activity into local levels of SRSs; 4a) these substances in turn, by a slow process (gene transcription/translation), alter electrical properties of nearby neurons by altering their own production and that of receptor populations, such as glutamate and adenosine receptors; 4b) the SRSs also, by a fast process (diffusion for short distances), directly interact with their receptors on neurons and alter electrical properties; 4c) further, ATP itself breaks down, releasing extracellular adenosine that in turn acts on adenosine receptors - again altering electrical potentials on the nearby neurons. These events are happening locally and the collective electrical changes result in a shift in input-output relationships within the local neuronal assemblies that originally exhibited the increase in activity, i.e., a state shift. In a mathematical model, the local state of neuronal assemblies rapidly synchronize, or phase lock, with each other because they are loosely connected to each other via neurons and humoral substances¹⁶⁵⁾. Well characterized sleep regulatory circuits and associated activation networks play a critical role in both sleep and waking by ensuring the synchronization of neuronal assembly state for niche-adaptation purposes.

Consistent with this model, ATP agonists promote NREMS while ATP antagonists inhibit sleep¹⁶⁶⁾. Further, after sleep deprivation, mice lacking the P2X7 receptor have attenuated duration of NREMS and EEG δ wave power during NREMS compared to control mice¹⁶⁶⁾.

EXTRACELLULAR ATP PROVIDES AN INDEX OF PRIOR CELL ACTIVITY

ATP is present in neuronal synaptic vesicles. The concentration of ATP in the vesicles is 10–50 times higher than in the cytosol. In the brain, ATP is co-released in GABAergic, cholinergic, noradrenergic and glutamatergic synapses (reviewed^167–171)). ATP is also considered a gliotransmitter. Once released, some extracellular ATP is converted to adenosine. Adenosine, in turn, binds to the other major purine receptor type, P1 receptors. The action of adenosine is fast, occurring within milliseconds to seconds and it results in increased K⁺ permeability and hyperpolarization. Some of the released extracellular ATP acts on glial P2Rs and causes the release of IL1, TNF and BDNF as well as additional ATP. IL1 precursor is processed by caspase-1 and this is triggered via ATP activation of P2X7 receptors^{4, 172, 173)}. TNF and IL1 released from glia act more slowly (minutes to hours) on adjacent neurons leading to the activation of NFkB. NFkB promotes transcription of receptor mRNAs such as the adenosine A1R and the glutamate AMPA receptor-gluR1 mRNAs. Translation of those mRNAs into their respective proteins and their subsequent expression on the cell membrane would change sensitivity of the postsynaptic neuron over longer periods. This is a prototypical scaling effect since the expression of postsynaptic receptors is modulated by the activity of the presynaptic neuron (i.e., the amount of released ATP) or, in other words, the sensitivity of the postsynaptic neuron is scaled to the prior use of the synapse. Indeed, both TNF and BDNF are the two molecules most firmly linked to synaptic scaling (reviewed¹⁷⁴⁾).

The breakdown of extracellular ATP to adenosine is regulated enzymatically in brain¹⁷⁵⁾. ATP and ADP are converted to 5′ AMP by the enzyme ectonucleoside triphosphate diphosphohydrolase1 (also known as CD39) ¹⁷⁶⁾. Ecto-5′-nucleotidase (also known as CD73) is a 70-kDa glycosylphophatidylinositol anchored ectoenzyme catalyzes 5′ adenosine monophosphate to adenosine¹⁷⁵⁾, although it has secondary functions and alters inflammation and immune responses in vitro and in vitro^177–181). However, CD73 knockout mice have intact immune systems¹⁸¹⁾. CD73 is found on many CNS cells, including astrocytes, oligodendrocytes, and microglia^{177, 178, 182, 183)}. Mice lacking CD73 have more spontaneous NREMS yet fail to exhibit sleep responses after sleep deprivation¹⁸⁴⁾. These data point to a role for CD73, and thus the extracellular ATP – adenosine catabolic pathway, in sleep regulation.

SLEEP FUNCTION

Sleep probably has many functions analogous the multiple functions of the lungs, i.e. gas exchange, acid-base balance and speech. However, when sleep first evolved it probably did so for a crucial primordial function. At the behavioral level, some functions of sleep are clear: calories are saved and peak performance is restored. These observations led to the universal recognition that sleep is restorative. However, at cellular and molecular levels no one has determined whether any restoring takes place. This deficit of evidence has led many theorists to posit that optimizing neuronal efficacy/connectivity is the primordial function of sleep^{4, 38, 128, 129, 161)}.

Neurons spontaneously form connections in brain and culture. Gene expression affects this process and alters functional properties of the network. Expression of certain genes, including many involved in connectivity, e.g. TNF, and synaptic efficacy e.g. glutamate receptor populations, is activity-dependent. The activity-dependency of network efficacy/connectivity-genes provides a bottom-up means to target changes in the network to those synapses that are disproportionately used. The use-dependent gene expression efficacy/connectivity mechanism is an established epigenetic brain plasticity mechanism. It is also a positive feedback loop because enhanced efficacy and connectivity lead to re-use. Such reinforced repetitive use and re-use eventually would lead to a network locked in one firing mode and thus be a rigid non-plastic network. This argument is fully developed by Kavanau¹²⁹⁾ and acknowledged by us³⁸⁾. As a consequence, negative feedback-damping mechanisms are needed to stabilize the existing networks that prima fascia are adaptive because the animal is alive. These negative feedback mechanisms lead to state oscillations and altered local network input-output relationships. Further, the linking of positive and negative feedback loops leads to electrical and gene expression oscillations (Fig 1). At different phases of the oscillations, a given input leads to a different network output; the network can thus be considered in different states depending upon the oscillation phase. Synchronization of multiple network (e.g. cortical columns) state phase transitions emerges as sleep-wake cycling in whole animals.

Figure 1.

Cell activity drives oscillations between wake and sleep. Environmental and internal inputs induce cell activity; during waking this induces output 1 that is adaptive to the inputs. The waking-induced cell activity also induces release of extracellular ATP. The extracellular ATP provides a way for the brain to track prior activity; it induces via purine type 2 receptors (P2) release of cytokines, growth factors and neurotrophins that in turn activate nuclear factor kappa B to induce expression of glutamate and adenosine receptors. The change in receptor populations changes sensitivities of the neurons within the diffusible range of the extracellular ATP and this will result in output 2. Output 2, is delayed in time due to transcription and translation steps and thus is divorced, in time, from environmental inputs. Since that could cause non-adaptive behavior it requires a state (sleep) within which quiescence is imposed. Output 1, waking, has a positive feedback, whereas Output 2 has a negative feedback, onto cell input (e.g. reduced sensory input). The whole brain contains thousands of such neuronal/glial networks, e.g. cortical columns are a good example of a highly interconnected network. The synchrony of state between networks is a consequence of there being neuronal and chemical connections to other such networks (165) and from the influence of the activation systems projecting throughout the cortex; whole organism sleep is thus emerges from the activity-dependent local events.

Embedded within this logic is the idea that sleep is a consequence of epigenetic use-dependent changes in synaptic efficacy/connectivity and functions to serve it. Epigenetic plasticity has great evolutionary value. The functional importance of sleep is illustrated by the fact that during sleep one gives up opportunities to reproduce, eat, drink or socialize and is subject to predation. Sleep could only have evolved despite these high evolutionary costs if it serves a crucial, primordial function. Maintaining adaptable flexible neural connectivity may be sufficiently important for the brain to allow the persistence of such a periodic, disadvantaged state^{4, 38)}.

If we assume that afferent input, both internal and sensory, during waking induces an environmentally relevant local brain network output (i.e., network electrical and chemical patterns signaling to other brain areas contribute to whole-animal mental and motor responses), then after prolonged activation of glia/neurons within a neuronal assembly the resulting SRS released would induce changes in the network’s outputs and responsiveness to inputs (Fig 1). These changes would likely summate and result in a state shift in the local network. Thus sleep can be viewed as being initiated within local neuronal assemblies as a function of prior use. This altered state (different output in response to the same input), because it is qualitatively and quantitatively different from the original output, is likely not relevant to the environmentally driven input. Thus, if the initial environmental inputs induce adaptive outputs ( waking), then a shift in input–output relationships would result in outputs that lack the necessary network-activity patterns to induce environmentally relevant cognitive or motor outputs (i.e. sleep) (Fig. 1). This creates an adaptive need to prevent the animal from behaving. A potential role for the traditional sleep–wake regulatory circuits is to ensure the absence of behavior at such times. Thus, not only are local sleep mechanisms inseparable from the synaptic efficacy/connectivity functions of sleep, they also are causal of the network outputs that require the altered consciousness that typically pervades sleep. Previously, it was posited that the environmentally disconnected outputs occurring during sleep serve to stabilize the networks and thereby preserve them^{38, 129)}.