Adenosine as a sleep factor


Tarja Porkka-Heiskanen, Institute of Biomedicine, Biomedicum Helsinki, University of Helsinki, FIN-00014, Helsinki, Finland. Email:


What puts us to sleep? This question has bothered the mankind for thousands of years, but we still have no definite answer. After abandoning philosophical and religious explanations, science has adopted this question and started to examine it with experimental methods. Two early pioneers in this field, Dr. Ishimori from Japan and Drs. Pieron and Legendre from France developed the concept of hypnotoxin – a factor that accumulates during waking and puts animals and humans to sleep. They were able to show that, indeed, during deprivation of sleep, something accumulates in body – something that can be removed and will induce sleep in another individual. Later research has identified many substances that affect sleep. One of them is adenosine, which fulfils the criteria of a physiologic sleep factor.

Dr Ishimori from Japan and Drs Pieron and Legendre from France were the first researchers to start experimental work on sleep factors.1,2 They were able to show that, during deprivation of sleep, some substance accumulates in the body – a substance that can be removed and will induce sleep in another individual. After the preliminary discovery, the idea of sleep factors was forgotten for years as sleep research was dominated by electrophysiological studies.


The pioneers defined the basic concept of a sleep factor: a molecule that accumulates in the brain during wakefulness and eventually forces the brain into the state of sleep. More recently, the definition has been refined to include more criteria:3 the concentration of the sleep factor must be higher during waking than during sleep, the concentration must decrease during sleep, and it must increase steadily during prolonged wakefulness. The substance must be able to inhibit neuronal activity: sleep is induced when activity in the waking-promoting neural networks decreases.

Sleep homeostasis means that a prolonged period of wakefulness is followed by a prolonged period of sleep, which is intensified by increased slow-wave activity/delta power (recovery sleep). The homeostatic process is mathematically formalized in the two-process model of sleep regulation as process S.4 The model is based on experimental EEG data collected from humans and several other species. Interestingly, the model introduces the concept of sleep propensity, which accumulates during wakefulness and finally forces the brain to sleep. However, the biological mechanisms that induce sleep propensity are unknown, leading to the suggestion that sleep factor(s) may represent the biological substrate for the component of sleep propensity that is defined as process S in the two-process model of sleep regulation.


The idea of sleep factors was revisited by some laboratories in the 1970s when Pappenheimer et al. started to extract sleep factors from goat cerebrospinal fluid (CSF).5 This group later found that the sleep-inducing factor was a muramyl peptide6– a finding that was the foundation of a completely new view in sleep research: the connection of the immune system and sleep regulation.7 A Japanse group worked on extracting sleep-inducing factors from the brainstems of sleep-deprived rats8 and a Mexican group experimented with protein changes in the CSF during SD.9,10


Adenosine is an ubiquitous nucleoside which forms the core of the basic energy molecule, adenosine triphosphate (ATP).11 In short, adenosine is the A of ATP, meaning a state where all energy-bearing phosphate groups are absent. Adenosine, and particularly an increasing adenosine concentration, thus often represents a state of increased energy consumption reflected as ATP hydrolysis.

Importantly, in the central nervous system (CNS), adenosine is generally an inhibitory neuromodulator, inhibiting neuronal activity of excitatory (e.g. cholinergic and glutamatergic) as well as inhibitory (e.g. GABAergic) neurons.12 The rate of adenosine formation is regulated by neuronal activity: the adenosine levels start to rise when the supply-to-demand ratio for oxygen decreases under circumstances like ischemia, hypoxia and hypoperfusion (for review, see Porkka-Heiskanen et al.).13 It has been suggested that this is a self-controlling neuroprotective mechanism to minimize cell damage: adenosine decreases cellular activity and thus decreases energy need.14 Adenosine can be directly released upon neuronal activation or co-released with several neurotransmitters (reviewed in Dunwiddie and Masino).15


In one laboratory, the effect of adenosine on vigilance states was examined using adenosine-receptor agonists and antagonists administered either systemically or intracerebroventricularly (ICV).16 These studies showed that ICV infusion of adenosine increased sleep and reduced wakefulness17 and that local injections of adenosine A1 receptor agonists in the preoptic area of the rat induced sleep, while an A2A agonist did not.18 Systemic, as well as ICV, administrations of A1 receptor agonists induced sleep.19,20 Furthermore, the well-known stimulants of wakefulness, caffeine and theophylline, are antagonists of adenosine receptors.21 The results of these experiments led the researchers to the conclusion that adenosine may be a physiological sleep-regulating factor.18

However, to gain evidence for adenosine as a sleep factor, it was critically important to show that its concentration in the brain varied according to vigilance state and that during wakefulness it accumulated in the brain. This was accomplished in 1997 when, using in vivo microdialysis, we measured adenosine levels during spontaneous sleep–wake cycles and during prolonged wakefulness in cats.22,23


The measurements of extracellular adenosine concentrations during spontaneous sleep–wake cycles were made in several brain areas of cats.22,23 Basal levels of adenosine in the basal forebrain (BF) and thalamus (= 10) were 32.8 ± 3.0 nmol/L (mean ± SEM); hence, from these data we estimate that extracellular adenosine concentrations are in the range of 165 to 300 nM, based on an in vitro microdialysis probe recovery of 10–20%. Adenosine levels in samples collected during spontaneous waking were higher than those collected during periods of NREM sleep for all structures studied to date. The different brain regions had approximately the same reduced adenosine level in NREM sleep, which was approximately 75 to 80% of waking values. Another group has reported diurnal variation in adenosine levels, being approximately at the level of 60 nM during the lights-on period and about 70 nM during the lights-off period.24


In the first experiments in cats, adenosine concentrations increased in the BF steadily in the course of 6-h SD, reaching at 6 h twice the values that were measured in the beginning of the SD period.22 After the SD, during the recovery sleep period, adenosine levels gradually decreased during the next 2–3 h. This pattern was specific for the BF: the rise in adenosine levels in the cortex paralleled those in the BF during the first 5 h, although the rise was not as steep as in the BF and started to decline in the last hour of SD. In the thalamus, dorsal raphe nucleus, pedunculopontine tegmental nucleus, and preoptic area adenosine levels did not rise during 6-h SD, though many of these regions are also known to be active in sleep regulation. Several later studies from our group25–28 and from other groups24 have confirmed that adenosine increases in the BF in the course of SD as well as sleep fragmentation29 in young rats. This unique pattern of regionally specific rise in adenosine in response to SD supports the hypothesis that adenosine promotes sleep through inhibition of the wake-active cortically projecting neurons of the BF, and possibly also by inhibiting neurotransmitter release from the cortical terminals of these BF neurons. It is possible that adenosine in BF is particularly important in regulation of the homeostatic sleep responses, representing process S in the two-process model, rather than in regulation of spontaneous sleep–wake cycles.


It is our view that increased neuronal activity is the basic signal for adenosine increase, but as astrocytes provide neurons with energy during increased energy demand,30 adenosine can originate either from neurons themselves or from neighbouring astrocytes. It has recently been shown that sleep homeostasis is impaired in animals with genetically engineered deficiency in gliotransmission, resulting in decreased adenosine levels,31 implying that the increased adenosine levels during SD may originate from astrocytes. The neuronal activity that could increase extracellular adenosine levels could originate from different types of BF neurons, including the GABEergic and glutamatergic cortically projecting neurons. However, data from our laboratory and others indicate that intact BF cholinergic neurons are necessary for SD-induced increase in adenosine level.32,33 Interestingly, results from two laboratories show that the activity of the enzymes that metabolize adenosine or equilibrative adenosine transporters are not significantly modified by SD.34,35


The basal forebrain

Several lines of evidence have indicated that adenosine in the BF affects sleep through A1 receptors (for review, see Strecker et al.12 and Basheer et al.): 36 when the rise in adenosine levels in the BF during prolonged wakefulness was prevented by A1 receptor antagonist (but not A2A receptor antagonist)37 or by introducing an A1 receptor antisense,38 recovery sleep was attenuated. We have put forward the hypothesis that the effect of adenosine in the BF is based on inhibition (through A1 receptors) of the cortically projecting cholinergic cells.12,35,36 Direct evidence to support this view was provided by a series of experiments where we showed that specific destruction of the cholinergic cells in the BF not only prevented increase in extracellular adenosine during SD, but also abolished recovery sleep.32 Interestingly, when adenosine was infused into the BF of the rats with destroyed cholinergic neurons, it was not able to promote sleep. Another group has also shown that locally induced saporin lesion will decrease recovery sleep.39 Results of the study by a third group31 are difficult to interpret because of differences in methodological approach.

The hypothalamus

The hypothalamus is an important sleep-regulating area where the actions of adenosine have been described in both wake-active and sleep-active nuclei. Generally, adenosine, through A1 receptors,40 inhibits the activity of the wake-promoting cells.41 Adenosine promotes NREM sleep by inhibiting the wake-active histaminergic neurons in the tuberomamillar nucleus.42 Interestingly, in addition to A1 receptor–mediated modulation, the same group has reported that adenosine A2A receptor agonist can also increase sleep by inhibiting histamine release in the tuberomamillar nucleus, an effect which appeared to be mediated through induction of GABA release in the tuberomamillar nucleus.43 In the lateral hypothalamus, another area where wake-promoting cells are located, adenosine A1 receptor agonist and antagonist promote and suppress NREM/REM sleep, respectively.44

Sleep-active cells have been described in the preoptic area (POA) of the hypothalamus.45 Decreasing the activity of these cells by infusion of A1 receptor agonist, or increasing extracellular adenosine concentration by nitrobenzylthioinosine (NBTI) in the lateral preoptic area, induced waking, while activation of these cells with A2A receptor stimulation induced sleep.44 Similarly, A2A receptor agonist increased sleep when infused into the ventrolateral preoptic nucleus.46

Infusion of adenosine agonists/antagonists into the subarachnoidal space that lies immediately under the BF and hypothalamus modulates sleep through A2A receptors.47

Other brain areas

Administration of adenosine to the prefrontal cortex also modifies the vigilance state. Consistent with results from the BF, A1 receptor agonists reduced signs of wakefulness (including a decrease in acetylcholine release).48 However, in this area A2A agonists increased wakefulness and acetylcholine release, pointing out the importance of the anatomical site, the distribution of different types of neurons and the distribution of different adenosine receptors when evaluating the effects of adenosine on behavioural state. In addition, a specific A1 antagonist, or caffeine, increased acetylcholine release and waking while A2A antagonist was ineffective,48 suggesting that caffeine at the cortical level increases arousal through A1 rather than A2A receptors. Interestingly, administration of adenosine agonists and antagonists in the prefronatal cortex modulated acetylcholine release in the pontine reticular formation, suggesting that adenosine, via A1 receptors, is able to modulate behavioural arousal via descending input to the pontine brainstem.48

Also the pons region, which is the central regulator of REM sleep, is affected by adenosine: adenosinergic agonist, when microinjected into the region of the caudal oral pontine reticular formation, resulted in significant, long-lasting elevation in REM sleep.49


Adenosine has biochemical properties that make it a good candidate for a physiological sleep factor: it is an ubiquitous and small molecule located both intra- and extracellularly; the inhibitory A1 receptor is also ubiquitous, as are the enzymes needed to produce and metabolize adenosine. The close connection to energy metabolism on one hand and the action as a neuromodulator on the other hand provide adenosine with both a cause and means to promote sleep.

It is clear that adenosine has many routes to modulate vigilance states, and that it works through both A1 and A2A receptors. The key areas are the BF- hypothalamus region and pons, where also the classical centers for vigilance regulation are located.50

Our hypothesis is that during prolonged wakefulness, adenosine concentration, triggered by increased neuronal activation, elevates in the extracellular space of the BF and increases sleep propensity. Adenosine promotes sleepiness via inhibition of wake-active cells in the BF, including cholinergic neurons, through A1 receptors. When the activity of the waking-active cells decreases sufficiently, sleep is initiated. The long-term effects of prolonged wakefulness are, at least partially, mediated by high extracellular adenosine concentrations that initiate intracellular signal-transduction cascades resulting in gene transcription.51

Two brain structures appear to be central in accumulation of sleep propensity: the cortex and the BF. Numerous experiments from other laboratories indicate that in the cortex there is a local, use-dependent accumulation of sleep propensity.52,53 The role of adenosine in this process has so far remained under-explored.

To conclude, it must be emphasized that our model concentrates on the global effects of prolonged wakefulness, and on mechanisms that regulate this aspect of sleep regulation. We have identified a relatively specific area, the BF, which appears to be central in the regulation/execution of recovery sleep: increase in adenosine in this area induces sleep, while prevention of this increase abolishes recovery sleep. So far we have found no condition where the connection between BF adenosine concentration and induction of (recovery) sleep would have been disrupted.


This work was supported by a grant from Academy of Finland.


The authors indicated no potential conflict of interests.