Clock Genes and Sleep Homeostasis: A Fundamental Link Within the Two-Process Model?

The two-process model of sleep regulation posits that separate circadian (C) and sleep homeostatic (S) processes combine to produce our subjective and objective levels of sleepiness,¹ and presumably, are also tied to our physiological need for sleep, whatever that may be. The sleep homeostat appears to respond to prior sleep-wake history by increasing sleepiness and sleep propensity with time spent awake and decreasing these variables with time spent asleep, although it does not always accurately predict our neurobehavioral responses to extended wakefulness.² The underlying physiological mechanisms are not well understood; however, molecules such as adenosine are believed to play a role. It is even less clear how the circadian system influences our subjective sleepiness, for example, promoting high alertness in the evening in the face of an escalating homeostatic drive. Further, misalignment of these processes can have negative effects on health and cognitive performance, as seen in shift workers.³ Thus, a better understanding of these C and S processes may allow us to address the deleterious consequences of poor sleep architecture or insufficient sleep at an adverse circadian phase.

The extent to which the two processes are linked is still unclear. Studies in humans using forced desynchrony, such as a 28-h day⁴ or SCN lesions in experimental organisms,⁵ support the idea that the two processes run independently, and then, through largely undefined output pathways, combine to influence sleep propensity. Examples of this independence include certain hormone rhythms that follow a circadian pattern with or without sleep⁶ and EEG variables such as delta power that appear to be driven almost exclusively by prior wake duration.⁴ However, a growing body of evidence over the past ten years suggests that the circadian and sleep homeostatic processes are not completely independent even in their core components. In this issue of SLEEP, Curie and colleagues⁷ provide new insights into this issue by examining the time-of-day effects of sleep deprivation on the expression of a core clock gene (Per2), a clock-controlled gene (Dbp), a wake-induced transcript thought to index homeostatic sleep need (homer1a), and the dominant marker of sleep homeostasis, EEG delta power. Franken and colleagues previously established that Per2 cannot be considered a pure marker, or state-variable, of the circadian clock, at least in brain regions outside the SCN (see review in Franken⁸). Indeed, they have shown that Per2 mRNA generally increases with wake and decreases with sleep in most brain regions, while Dbp does the opposite. This evidence combined with alterations in sleep homeostasis found in many “clock” gene mutants^9,10 supports the idea that in brain areas outside the SCN, the expression of these circadian genes is influenced more by sleep-wake patterns than circadian timekeeping. In the present study, Curie et al.⁷ also find that the expression of Per2 and Dbp consistently increases and decreases, respectively, following sleep deprivation, but that the magnitude of the change depends on the time of day that the sleep deprivation is performed, suggesting that Per2 in particular may provide a critical link between sleep-wake dependent and circadian processes.

Most studies of nocturnal rodents (our primary model organisms) perform sleep deprivations at or near ZT0 (light onset), so the influence of time-of-day on the effects of sleep deprivation cannot be thoroughly assessed. Curie and colleagues performed sleep deprivation at four times of day on three strains of mouse, and measured the expression of several core clock genes, as well as Dbp and Homer1a. Per2, Dbp, and Homer1a showed signifi-cant time-of-day dependent changes in gene expression following sleep deprivation. Interestingly, while Per2 and Dbp expression increased or decreased, respectively, relative to the baseline expression of these genes at the end of sleep deprivation, at certain times of day their expression changed in the opposite direction during the course of the sleep deprivation. Further analysis revealed that the baseline slope, or change, in gene expression at a particular time of day for Per2 and Dbp, and the baseline expression level at the beginning of the sleep deprivation for Homer1a, could explain most of the variance in these results.

Curie and colleagues simulated gene expression as a function of sleep-wake distribution and found that their model accurately predicts observed results for Dbp and Homer1a. However, the model of Per2 was phase delayed by 6 h relative to observations. Since corticosterone is known to influence Per2 expression, they used plasma corticosterone levels to set the phase of Per2. While this model fits the data, the mechanism by which Per2 influences sleep is not known. Mice can become sleepy after one or two hours of sleep deprivation during the day, which is probably too fast to be influenced by the molecular network of Per2, which requires multiple rounds of transcription and translation before altering physiological variables, such as resting membrane potential. A better understanding of the temporal dynamics of these processes, especially at the protein level, would be beneficial.

The authors also analyzed the effects of sleep deprivation on the sleep homeostatic variable, NREM EEG delta power. Under baseline conditions, EEG delta power peaks around ZT18, at the end of the largest wake period in mice, and decreases gradually until the last few hours of the next light period. All sleep deprivations increased EEG delta power, but the largest increase was observed after the ZT12-18 SD and the smallest after the ZT0-6 SD. These results make sense when we consider prior sleep-wake history and latency to sleep in these periods. Mice took an average of 2.5 h longer to fall asleep after the ZT12-18 SD, allowing sleep pressure to accumulate. Simulations of delta power as a function of sleep-wake distribution accurately predicted these results. However, a surprising feature of the findings of Curie et al. was the great difficulty in performing sleep deprivation in the latter half of the day (ZT6-12), despite considerable sleep time in these mice from ZT0-6.

Overall, this study adds to the case that clock genes not only underlie circadian and homeostatic aspects of sleep regulation, but that there is a fundamental link between the two processes. Further evidence to support the interaction between these processes might come from continuous monitoring of these genes and proteins during both sleep deprivation and recovery sleep periods, and in the following 24 hours under constant conditions (such as DD). However, it is already clear from this work that the interaction of the S and C processes is discoverable at a molecular and physiological level. Perhaps of even greater interest, the links between redox state, energy metabolism, and the clock-gene network,^8,11 suggest the possibility that Per2 not only integrates the two processes of sleep regulation, but that it may further integrate functional aspects of energy balance and sleep.

CITATION

Striz M; O'Hara BF. Clock genes and sleep homeostasis: a fundamental link within the two-process model? SLEEP 2013;36(3):301-302.

DISCLOSURE STATEMENT

Dr. O'Hara has a 45% ownership in Signal Solutions, LLC. In addition to ownership, he consults and does research for the company. Mr. Striz consults for Signal Solutions, LLC.