There are several neurochemical systems with diffuse projections that promote and maintain wakefulness. These include the noradrenergic cells in the locus coeruleus (LC); cholinergic cells in the pedunculopontine tegmental nucleus (PPTN), lateral dorsal tegmental nucleus (LDTN), and basal forebrain; histaminergic cells in the tuberomamillary nucleus (TMN) of the posterior hypothalamus; and glutamatergic neurons in various structures in the central nervous system (3). These cells fire at a higher rate during waking than during non–rapid eye movement (NREM) sleep. Dopamine-containing neurons in the substantia nigra and ventral tegmental area (VTA) also modulate arousal. VTA cells fire tonically during quiet wakefulness and NREM sleep, and in bursts (inducing massive release of dopamine) during appetitive waking behavior and REM sleep (4). Psychostimulants, such as amphetamines and cocaine, that block the reuptake of monoamines including norepinephrine, dopamine, and serotonin promote prolonged wakefulness and increase both cortical activation and behavioral arousal (5). The most recent wakefulness-promoting system to be discovered is the hypocretin system. The peptide hypocretin (also known as orexin) is produced by cells in the posterior hypothalamus that provide excitatory input to all remaining wakefulness-promoting areas, including the LC, PPTN, LDTN, VTA, basal forebrain, and TMN (6) (Figure 1). Hypocretin neurons are most active during waking, especially in relation to motor activity and exploratory behavior, and almost stop firing during both NREM and REM sleep.

Figure 1 The hypocretin system in the posterior hypothalamus (dots) and its projections (arrows) to most brain regions.

Wakefulness-promoting systems appear to be redundant to some extent, which may explain why specific lesions that affect one or a few of these brain areas do not result in a complete and permanent suppression of cortical activation and wakefulness (7). A common mechanism through which these systems produce cortical activation is by closing potassium channels on the cell membrane of cortical and thalamic neurons, thus keeping cells depolarized and ready to fire (8). At sleep onset, wakefulness-promoting neuronal groups are actively inhibited by antagonistic neuronal populations located in the hypothalamus (9, 10) and the basal forebrain (11, 12), most of which are GABAergic (transmit or secrete γ-aminobutyric acid). Exactly how the switch from wakefulness to sleep occurs, however, remains unclear.

What is clear is that sleep is regulated through both circadian and homeostatic mechanisms (13). The circadian regulation is responsible for the change in sleep propensity depending on the time of day, with obvious adaptive advantages. Humans, for instance, are a diurnal species (active during the daytime and sleeping mainly at night), while mice and rats are nocturnal and sleep mainly during the day. The homeostatic regulation of sleep, on the other hand, is responsible for the fact that the longer one stays awake, the stronger the pressure to go to sleep becomes (sleepiness increases). Also, the subsequent sleep is longer and more consolidated (with fewer brief awakenings). Influential models of sleep regulation (13) have suggested that the homeostatic component predicts the existence in the brain of a distinct physiological/biochemical/molecular process that builds up during wakefulness and declines during sleep, but the nature of such a process is still unclear.

In this issue of the JCI, Rao et al. (14) have made an important contribution to our understanding of how the activity of the hypocretin wakefulness-promoting system changes during prolonged wakefulness. The authors postponed, for 2–4 hours, the normal sleep period in mice either by injecting the animals with the wake-promoting drug modafinil or by sleep depriving them using gentle handling. They then sacrificed the mice to measure miniature excitatory postsynaptic currents (mEPSCs) in brain slices and found that this increase in waking duration led to long-term potentiation (LTP; a long-lasting improvement in the communication between 2 neurons, via their synapses, that usually results from simultaneous neuronal stimulation) of glutamatergic synapses on identified hypocretin neurons. Both the frequency and the amplitude of mEPSCs were increased, suggesting changes in both the pre- and postsynaptic components of synaptic transmission. Pretreatment with a selective D1 antagonist, SCH23390, blocked the modafinil-induced increase in mEPSC frequency, but not in amplitude, indicating that dopaminergic transmission is involved in the potentiation of the presynaptic component, while other neuromodulators, most likely norepinephrine and histamine, may play a role in the potentiation of the postsynaptic component. By measuring evoked EPSCs in the same hypocretin neurons, the authors also found that the contribution to EPSCs of glutamatergic α-amino-3-hydroxy-5-methyl-4-isoxazolepropionic acid (AMPA) receptors was increased relative to that of N-methyl-d-aspartic acid receptors. Moreover, in mice chronically treated with modafinil (once daily for 1 week) the authors found not only an increase in frequency and amplitude of mEPSCs, but also an increase in the number of asymmetric synapses on hypocretin cells. Finally, Rao et al. show that the LTP-like changes observed 2 hours after acute modafinil treatment partially occlude the ability of forskolin, a drug that increases the intracellular levels of cyclic AMP and normally induces synaptic potentiation (the long-lasting strengthening of the response of a neuron after it has received a message from another neuron across the synapse), to further potentiate synapses. Thus, this elegant and comprehensive study shows that even a few hours of wakefulness trigger synaptic potentiation in one of the wakefulness-promoting systems, suggesting that hypocretin cells can somehow sense how long the brain has been awake. The authors speculate that the increase in synaptic strength in these neurons may be at least one of the mechanisms by which prolonged wakefulness is maintained despite an increase in sleep pressure.

In future studies, other wakefulness-promoting cells — for instance, those present in the LC, PPTN, or TMN — should be examined. It could be that wakefulness-induced synaptic potentiation may take place in other wakefulness-promoting systems. If so, synaptic changes in the input to neuromodulatory systems could represent one of several possible mechanisms underlying the homeostatic regulation of sleep. Obviously, future causal experiments are needed to address these questions. Another important result shown by Rao et al. (14) is the partial occlusion of LTP observed after acute modafinil treatment. This clearly suggests that the ability of glutamatergic synapses on hypocretin cells to undergo synaptic potentiation in vivo can become saturated. How quickly and easily can this happen? Intriguingly, in vitro studies in the hippocampus have shown that prolonged waking impairs the induction of LTP but favors the induction of long-term depression (e.g., refs. 15–17). Moreover, in both cerebral cortex and hippocampus, learning during wakefulness can strengthen synapses to near the maximum of their modification range, impairing the further induction of LTP (18, 19). These experiments suggest that wakefulness-induced synaptic potentiation may be a more widespread phenomenon that could extend beyond neuromodulatory systems to some of their targets in the brain (20). Finally, the results of Rao et al. raise an intriguing question: if wakefulness is associated with the progressive potentiation of synapses on hypocretin neurons, there must be a mechanism by which a synaptic balance is maintained — obviously, synapses cannot simply keep strengthening indefinitely. The most straightforward possibility, of course, would be that such balance is achieved in subsequent sleep periods. Thanks to this provocative study, we will probably know the answer soon.

This work was partially supported by an NIH Director’s Pioneer Award to Giulio Tononi.