Adenosine is a metabolic intermediate of the energy-rich molecule adenosine-tri-phosphate (ATP) and a pleiotropic bioactive molecule with potent neuromodulatory properties because of its ability to cross the blood-brain barrier; it was proposed as a signaling molecule between the periphery and the brain (Chiu and Freund, 2014). Adenosine receptors are ubiquitously distributed throughout brain, heart, immune, and inflammatory cells and many other tissues. In the central nervous system (CNS), adenosine plays important functions such as modulation of neurotransmitter release, synaptic plasticity, and neuroprotection in ischemic, hypoxic, and oxidative stress events (Rodrigues et al., 2015). Numerous studies indicate that endogenous adenosine is a candidate for the sleep regulation and it induced sleep after prolonged wakefulness (Basheer et al., 2004). Moreover, adenosine interplays with several neurotransmitters that regulate sleep (Weber and Dan, 2016). Adenosine promotes sleep, at least in part, by inhibiting wakefulness-promoting neurons localized to the basal forebrain, lateral hypothalamus, and tuberomammillary nucleus of the posterior hypothalamus. There are two forms of sleep, non-rapid eye movement (NREM) sleep and rapid eye movement (REM) sleep. In humans, NREMS occupies about 80% and REMS about 20% of sleep time. These two stages alternate with each other with about a 90–100 min periodicity, with more NREM sleep occurring during the first half and more REM sleep occurring during the second half of the night (Krueger et al., 2008). The formation of adenosine in the brain changes in activity-dependent manner and is linked to the intracellular depletion of ATP (Krueger et al., 2013). The sleep inducing effects of extracellular adenosine are mediated through only two of the four adenosine receptor subtypes (A₁, A_2A, A_2B, and A₃): the inhibitory G protein-coupled adenosine A1 receptor (A₁R) and the excitatory G protein-coupled adenosine A2 receptor (A₂R) (Landolt et al., 2012). This author has suggested that both A₁and A_2A receptor subtypes probably mediate the effects of adenosine on vigilance and sleep, and these effects appear to be brain regions and receptor-dependent. Adenosine seems to promote sleep by simultaneously activating sleep-active neurons through A_2AR and suppressing wake-active neurons through A₁R (Watson et al., 2010). The A₁R activation in histaminergic neurons of the tuberomammillary nucleus increases NREM sleep (Oishi et al., 2008) while the activation of A_2AR in the pontine reticular formation increases acetylcholine release and NREM sleep (Coleman et al., 2006). Extracellular adenosine levels in basal forebrain and frontal cortex increased during 6 or 11 h of sleep deprivation but not in four other subcortical structures, and then decreased during recovery sleep (Porkka-Heiskanen et al., 2000; Kalinchuk et al., 2011). After 5 days of sleep restriction, A₁R mRNA levels and density were increased in the basal forebrain region and several brain regions respectively, whereas A_2AR mRNA levels and density were reduced in the frontal cortex and olfactory tubercle (Kim et al., 2012, 2015). Total sleep deprivation also upregulates A₁R mRNA levels in the rat basal forebrain and increased its density in cortical and subcortical brain regions (Basheer et al., 2007; Elmenhorst et al., 2009). In the human brain, a single night of total sleep deprivation leads to a significant increase of the A₁R binding (Elmenhorst et al., 2007). The same research group recently showed that variation of A₁R regional availability is associated with the A₁R and A_2AR gene polymorphisms in the human brain (Hohoff et al., 2014). In addition, a role for the A_2AR gene polymorphism has been particularly related to be an important determinant of psychomotor vigilance in rested and in acute and chronic sleep-deprived state (Bodenmann et al., 2012; Rupp et al., 2013). Indeed several studies have previously shown that sleep deprivation or restriction is associated with increased daytime sleepiness, decreased cognitive and physical performance, increased inflammatory markers (pro-inflammatory cytokines IL-1β and TNF-α) and hormonal/metabolic disturbances (Rupp et al., 2009; Chennaoui et al., 2011, 2015; Arnal et al., 2015, 2016; Irwin et al., 2015; Sauvet et al., 2015).

During wakefulness, the increase of extracellular ATP in brain may either be hydrolyzed to adenosine or activate purine type 2 receptors such as P2X₇ receptors (P2RX₇) that affect sleep directly or indirectly through cytokines release or other sleep regulatory substances (Krueger, 2008; Krueger et al., 2010). The gene expression of P2RX₇ has been evidenced in immune cells (North, 2002) and one study has shown increased expression in peripheral monocytes after 24 h of sleep deprivation in healthy young volunteers (Backlund et al., 2012).

The A₁ and A_2A and P2X₇ receptors are expressed in human leukocytes and were reported to play a central role in mechanisms of inflammation associated to various neurological pathologies (Fortin et al., 2006; Varani et al., 2010; Chiu and Freund, 2014; Casati et al., 2016; Krügel, 2016). Moreover, leukocytes have been proposed as a useful peripheral model to study inflammatory processes in the CNS (Sullivan et al., 2006; Varani et al., 2010; Light et al., 2012).

In this study, we determined for the first time whether advancing bedtime by 1.5 h per night over six nights (i.e., going to bed at 21.00 vs. 22.30 and getting up at 07.00 in both conditions) influences A₁R, A_2AR, and P2RX₇ mRNA levels in leukocytes of healthy subjects during total sleep deprivation (TSD) followed by one night of sleep recovery. The gene expressions of the pro-inflammatory somnogenic cytokines IL-1β and TNF-α were also determined. We hypothesized that longer bedtime for sleep and/or sleep deprivation may altered peripheral purine type 1 and type 2 receptors.

Our present study shows that A_2AR mRNA levels in leukocytes of healthy subjects increase after 24 h-SD and recover basal levels after a night of sleep recovery, while A₁R has not changed, and this in two initial sleep conditions (21.00–07.00 and 22.30–07.00 bedtimes). The gene expression of P2RX₇ is increased after the night of sleep recovery compared to 24 h-SD in both sleep conditions, while IL-1β and TNF-α were not changed after 24 h-SD and sleep recovery.

In this study, we analyzed gene expression in leukocytes because it may be a useful surrogate for gene expression in the CNS, when relevant genes are expressed in blood and brain (Chennaoui et al., 2005; Sullivan et al., 2006). The leukocyte gene expression of A₁ and/or A_2A receptors, as that of P2X₇ receptor and the pro-inflammatory cytokines TNF-α and IL-1β was demonstrated altered in various neurological pathologies or in the stress conditions of total sleep deprivation and sleep restriction; (Varani et al., 2010; Backlund et al., 2012; Chennaoui et al., 2014; Sauvet et al., 2015; Casati et al., 2016). For Hurtado-Alvarado et al. (2016), sleep loss may induce blood-brain barrier disruption through a systemic low-grade inflammation characterized by the release of several molecules including cytokines, adenosine, and hormones.

Sleep mechanisms are complex and multiple brain areas have been implicated in controlling the switch between wakefulness and the general state of sleep. Sleep-active neurons in the preoptic hypothalamus area inhibits wake-active neurons in the lateral hypothalamus (histamine neurons) and in the brainstem (serotonin and norepinephrine neurons). Sleep-active GABAergic neurons in the preoptic hypothalamus are inhibited by norepinephrine, acetylcholine and a subgroup by serotonin (Weber and Dan, 2016). In addition, adenosine, that accumulates during wakefulness under sleep pressure, has been extensively studied as a somnogen. Adenosine regulates also multiple physiological functions such as arousal, neuroprotection, learning and memory, cerebral blood circulation as well as pathological phenomena such as epilepsy. It is produced primarily from the metabolism of ATP and exerts pleiotropic functions throughout the body. Intra- and extra-cellular adenosine levels rise in response to physiological stimuli and with metabolic/energetic perturbations (such as prolonged wakefulness), inflammatory challenges, and tissue injury. In the CNS, several pieces of evidence support a role for the A₁R and A_2AR in mediating the sleep promoting actions of adenosine. In the human brain, a previous study evidenced that a single night of sleep deprivation leads to a significant increase of the A₁R availability in cortical and subcortical brain regions (Elmenhorst et al., 2007), and there is any data on A_2AR at the CNS or at the peripheral level during sleep deprivation. Our study is the first showing an increase in the gene expression of A_2AR in leukocytes of healthy subjects after 24 h of sleep deprivation and a return to basal level after a night of recovery sleep. No change was observed in the gene expression of A₁R. Another sleep regulatory signal is initiated by extracellular ATP changes that are detected by the purinergic P2X₇ receptors on glia, and causes in turn the release of the somnogenic TNF-α and IL-1β pro-inflammatory cytokines (Krueger et al., 2008, 2010, 2013). Indeed, the ATP acting at P2X₇ receptors was initially demonstrated to serve as an efficient secondary stimulus for the generation and release of IL-1β from proinflammatory cells (Ferrari et al., 1997). Krueger et al. demonstrated that mice lacking the P2X₇ receptor exhibit reduced duration of non-REM sleep and reduced EEG δ wave power during non-REM sleep compared to control mice after sleep deprivation (Krueger et al., 2010). There is only one study showing increase of P2RX₇ expression in monocytes of healthy subjects (eight participants, four males, and four females) after total sleep deprivation, and authors suggested that is a stress response to sleep deprivation (Backlund et al., 2012). In our study, we showed significant increase of P2RX₇ mRNA after the night of sleep recovery compared to 24 h-SD, and no change IL-1β, and TNF-α neither after 24 h-SD nor after the night of sleep recovery. This does not preclude a TNF-α response to TSD, as we previously revealed increased whole-blood gene expression of TNF-α after 25 h of sleep deprivation and increased circulating levels at 34 and 37 h of sleep deprivation (Chennaoui et al., 2011, 2014). Also Irwin et al. have evidenced an increase of the spontaneous TNF-α monocytic expression after a night of sleep loss, which persisted in monocytes even after a night of recovery sleep (Irwin et al., 2006, 2015).

In the periphery, significant levels of the A_2A or P2X₇ receptors expression have been evidenced in immune cells such as lymphocytes, monocytes, peripheral macrophages, neutrophils (North, 2002; Fortin et al., 2006; Fredholm et al., 2007), and it has become clear that they play critical role in the modulation of inflammatory reactions through complex mechanisms (Skaper et al., 2010; Sheth et al., 2014). Human neutrophils expressed significant amounts of A_2AR transcripts that are maximally up-regulated after 4 h stimulation with low TNF-α concentration and maintained up to 21 h (Fortin et al., 2006). In this study, TNF-α also increased A_2AR protein expression but the kinetic of induction is different from that of A_2AR mRNA as the amounts of A_2AR protein had returned to baseline after 12 h of TNF-α exposure. In line with most studies showing an increase in monocytic TNF-α expression during total sleep deprivation, increased expression of A_2AR in our study (additionally in positive correlation with TNF-α) could be the result of low-grade inflammation and related TNF-α generated by the total sleep deprivation.

In our study, we determined if basal sleep duration influences A₁R, A_2AR, and P2RX₇ mRNA levels and IL-1β and TNF-α during total sleep deprivation and sleep recovery, because down-regulation of A₁R was suggested to mediate beneficial effect of sleep extension on cognitive performance during total sleep deprivation and recovery (Arnal et al., 2015). We observed higher total sleep time and higher duration time of N1, N2, and REM sleep stages in the extended basal sleep condition, but no effect on the expression of A₁R which partially responds to our hypothesis (Arnal et al., 2015), no effect on A_2AR, and P2RX₇, nor on IL-1β and TNF-α. However, we did not measure their circulating protein level or function nor adenosine levels which limits the responses to our hypothesis (Arnal et al., 2015) and that of Rupp et al. who suggested that changes of extracellular adenosine/adenosine receptor ratio may explain behavioral consequences of sleep restriction, recovery, and sleep extension (Rupp et al., 2009). We confirmed that expression of A₁R gene is low compared A_2AR as previously highlighted in neutrophils (Fortin et al., 2006), and we can suggest thus that A_2AR expression in leukocytes could be a reliable peripheral marker of sleep deprivation in human. Our results also show high and opposite correlations between A₁R and A_2AR expressions at baseline before sleep deprivation and latency of stage 3 sleep during the previous night, which give meaning to the suggestion of Sheth et al. (2014) concerning the involvement of adenosine receptors in the regulation of sleep. From their studies on mice with a global deletion in the p50 subunit of NF-κB, Sheth et al. (2014) speculated that the increase in cortical A_2AR could explain the increases in slow wave and REM sleep, in addition to their rapid recovery following sleep deprivation. In our study, TNF-α expression was also negatively correlated with the latency of stage 3 sleep in the EXT sleep condition, which may emphasize Krueger et al. who suggested that TNF-α plays a role in sleep regulation and in particular enhances non-REM sleep (Krueger et al., 2008). Our results showed that high levels of stage 3 sleep latency coincided with high A₁R (r = 0.703, p < 0.05) and low A_2AR and TNF-α mRNA levels (r = −0.650, and r = −0.720, p < 0.05) in the EXT and HAB sleep conditions, respectively. In addition, these correlations were not significant when examined after the night of sleep recovery, which may be related to a buffering effect of the stage 3 rebound during the recovery night after TSD (Arnal et al., 2015). The correlations we observe could add information on the relationship between A₁R and A_2AR and TNF-α gene expression and stage 3 sleep which is the sleep stage that mediates at least some of the beneficial functions of sleep on the brain, such as memory consolidation (Tononi and Cirelli, 2014). We also consider that it would be interesting to study the effects of sleep deprivation and/or sleep extension on other neurotransmitters that play wake-promoting action such as norepinephrine and serotonin in the leukocytes from healthy subjects. Indeed, monoamine's receptors are expressed by leukocytes (Chennaoui et al., 2005), and sleep deprivation has been evidenced to increase serotonergic turnover in the basal forebrain (Zant et al., 2011) and to induce alterations in the norepinephrine system in the basal forebrain and cingulate cortex (Kim et al., 2013).

Some limitations of this study need to be noted. We focused on determination of peripheral gene expression of five biomarkers involved in the sleep regulation, but we did not include determination of their protein levels. Future studies are needed to determine whether sleep deprivation-induced changes in leukocyte gene expression are paralleled by altered protein levels. In addition, we conducted a highly controlled laboratory study with a small sample size of healthy young subjects, so future studies should be made on large healthy population or clinical cases with sleep disorders.

The presented results showed for the first time that 24 h of sleep deprivation increased gene expression of the A_2AR in leukocytes of healthy subjects and did not change A₁R, and this even if bedtime was initially increased by 1.5 h per night for six nights. A delayed increase of the P2RX₇ expression after the night of sleep recovery is observed, and TNF-α and IL-1β expressions were not changed. Interestingly the latency of stage 3 sleep was positively correlated with A₁R and negatively with A_2AR and TNF-α expression. The correlation analysis supported the hypothesis that adenosine receptors in the periphery may reflect sleep pattern in healthy subjects. Further, studies are warranted to determine if A_2A receptor changed at the protein level and functionality.

This work was performed in the French armed forces biomedical research institute (IRBA), Brétigny sur Orge, France.

Authors have made substantial contributions to the following: Conception and design of the study: MC, PA, FS, and DG. Acquisition and analysis of data: MC, PA, CD, and FS. Interpretation of data and final approval of the version to be submitted: MC, PA, CD, DL, FS, and DG. Writing the manuscript: MC, PA, DL, FS, and DG. Revisiting the manuscript: MC, PA, DL, FS, and DG.

At the IRBA (French Armed Forces Biomedical Research Institute, ministry of defense) studies are funded in operating cost by the DGA (General Directorate for Armament, ministry of defense), contracts 14Ca703 and PDH-1-SMO-2-508.