Humoral mechanisms of sleep–wake regulation: Historical review of prostaglandin D2 and related substances


  • Author's note: This review article is based upon the Honorary President Lecture, presented at the Hayaishi Memorial Symposium at the Joint Congress of the 6th Congress of Asian Sleep Research Society, the 34th Annual Meeting of Japanese Society of Sleep Research, and the 16th Annual Meeting of Japanese Society of Chronobiology, held on 26 Ootober 2009 at the Osaka International Convention Center, Osaka, Japan. I express my most sincere thanks to President Masako Okawa, President Yoshihiro Urade, and President Hitoshi Okamura for their kind invitation extended to me to present this lecture and to contribute this review article to Sleep and Biological Rhythms. The original title of the lecture was “Humoral mechanisms of sleep–wake regulation – commemorating the centennial anniversary of the discovery of endogenous sleep substances.”

Dr Osamu Hayaishi, Osaka Bioscience Institute, 6-2-4 Furuedai, Suita, Osaka 565-0874, Japan. Email:


Sleep is perhaps one of the most important and yet least understood of the physiological functions of the brain. Sleep is essential for life, but we still cannot answer even the simplest questions about sleep, such as “what is sleep?,”“why do we need to sleep?,” and most importantly, “where and how are sleep and arousal regulated?” In the mean time, the number of sleep-disorder patients has recently been increasing exponentially and now exceeds more than 25% of the total population in most countries. More than 107 different sleep disorders have now been described, but in most instances, their etiologies are not yet clearly understood, simply because basic sleep science research has really only just begun. In the early 1980s, my colleagues and I at Kyoto University serendipitously discovered that prostaglandin D2 induced physiological sleep in rats and monkeys, and subsequently we elucidated the molecular mechanisms underlying sleep–wake regulation by prostaglandins D2 and E2. In this review, I start with a brief historical account, follow it by our ongoing work on prostaglandins and sleep, and finally close with a few remarks on the future prospects of sleep science.


In Japan, the history of modern scientific research on sleep dates as far back as 100 years ago. The humoral theory of sleep regulation, the concept that sleep is regulated by a hormone-like chemical substance rather than by a neural network, was initially proposed by Kuniomi Ishimori, Professor of Physiology at Nagoya University, Japan,1 and independently and concurrently, by a French neuroscientist, Henri Piéron of Paris,2 around the first decade of the last century (Fig. 1). They took samples of homogenized and dialyzed brain extracts from sleep-deprived dogs and infused them into the brains of normal dogs. The recipient dogs soon fell asleep. In contrast, the same extracts prepared from normal dogs without sleep deprivation had no effect at all. Thus these researchers became the first to demonstrate the existence of so-called “endogenous sleep-promoting substances.” However, the chemical nature of their sleep substances was not identified. Subsequently, more than 30 endogenous sleep substances were reported from various research groups, but their physiological relevance remained uncertain in most instances. Ishimori's first paper on the subject was published in Japanese in the Tokyo Igakkai Zasshi in 1909 with the title of “True cause of sleep – a hypnogenic substance as evidenced in the brain of sleep-deprived animals.”1 Thus, the year 2009 marks the centennial anniversary of the discovery of the endogenous sleep substance by Ishimori. Ishimori went one step further and suggested that continued wakefulness may bring about the accumulation of such a factor in the brain, which is dissipated during sleep, a concept that is now being called the homeostatic regulation of sleep. I should also remind you that the existence of a blood-borne messenger was evidenced by the discovery of “secretin” by Bayliss and Starling in 1902,3 and that the term “hormone” and the new concept of “humoral control of body function” became popular and even quite fashionable in those days.

Figure 1.

Photographs of Ishimori and Piéron. Courtesy of Dr Tatsuro Ohta, Nagoya University, and Dr Kunio Kitahama, Université Claude Bernard Lyon, respectively.

About 10 years passed, and in the early 1920s, Hans Berger, a neurologist in Jena, Germany, invented the electroencephalogram (EEG) which is the recording of brain waves.4 This invention made it possible to determine sleep both quantitatively and qualitatively. Up to then, sleep had always been considered as an inaccessible phenomenon, simply because there was no way to quantify it scientifically.

In 1953, rapid eye movement (REM) sleep in humans was discovered and characterized by Kleitman and coworkers in Chicago.5 Subsequently, Jouvet and colleagues in Lyon,6 using transection experiments, studied its mechanism in animals and proposed to term it paradoxical sleep because the whole body is in deep sleep but brain cells are awake. These authors clearly recognized that sleep is not a uniform phenomenon but consists of at least two fundamentally different stages. Wakefulness and sleep are very complex phenomena. Wakefulness and the two major types of sleep, namely, non-REM (NREM) sleep and REM sleep, can be assessed by examining the animal's behavior. However, a more precise qualitative and quantitative assessment of the sleep–wake pattern is made by use of the EEG, the electrooculogram (EOG), which is the recording of eye movement, and the electromyogram (EMG), being the recording of muscle tension.

For example, Figure 2 illustrates the sleep profile of a healthy adult during the entire night. We normally go to bed at about 11 o'clock, and then start the NREM and REM sleep cycle, which begins with NREM sleep that becomes progressively deeper. After four to five cycles, each lasting about 90 min, the final REM sleep is followed by arousal. This phenomenon has been known for years, yet its physiological meaning and regulatory mechanisms have remained a complete mystery.

Figure 2.

Sleep profile of a healthy adult during the entire night. The black area represents wakefulness; the shaded area, non-rapid eye movement (NREM); and the gray area, rapid eye movement (REM) sleep. Sleep begins with the so-called “sleep staircase.” Five complete NREM/REM sleep cycles are schematically presented.

In 1982, Alexander Borbély of the University of Zürich in Switzerland proposed his famous two-process model of sleep regulation.7 A simplified version of Borbély's two-process model of sleep regulation is shown on the left side of Figure 3, in which sleep propensity is plotted against the time of day, and the gray areas represent sleep. The “homeostatic” process is controlled by the sleep propensity or sleep pressure, which accumulates during the course of wakefulness (W) and dissipates during sleep (S). So this process may be called the Ishimori-Piéron type. In contrast, the “circadian” process, namely, the sleep–wake cycle during the day and night, is controlled by an internal pacemaker or a biological clock and is independent of prior sleep and waking. Subsequent studies revealed that an ultradian process generates the alternation of NREM and REM sleep as illustrated on the right side of Figure 3. However, the molecular mechanisms underlying the sleep–wake regulation in all these processes have thus far remained almost completely unknown.

Figure 3.

A two-process model of sleep regulation. Modified from AA Borbély. A two-process model of sleep regulation. Human Neurobiology (1982). NREM, non-rapid eye movement; REM, rapid eye movement.


Prostaglandins (PGs) are so-called lipid mediators. More than 30 different kinds of prostanoids are known. These compounds are widely distributed in virtually all mammalian tissues and organs, and have numerous and diverse biological effects on a wide variety of physiological and pathological activities and therefore are some times referred to as tissue or local hormones. However, relatively little had been known about PGs in the mammalian brain until about 40 years ago. In the early 1980s, we showed PGD2 to be the most abundant prostanoid in the brains of rats and other mammals including humans.8 Since PGD2 had long been considered as a minor and biologically inactive prostanoid, our findings suggested that PGD2 might be a unique constituent of the brain and might have some important and special function in this organ. We soon found out that PGD2 induced sleep when microinjected into the brains of rats.9 We were all delighted by this unexpected discovery and decided to extend our studies to the physiological significance of the somnogenic activity of PGD2, as well as to its molecular mechanisms.

The bioassay system for sleep analysis, originally designed by Honda and Inoué10 of Tokyo Medical and Dental University, was modified and used in my laboratory. This bioassay system can be described as follows: by means of a microinjection pump, the desired substance, such as PGD2, is infused continuously and slowly through a cannula chronically implanted into the third ventricle of a freely moving rat. The sleep stages are then determined on the basis of polygraphic recordings of EEG and EMG. Brain temperature as well as food and water intake is also monitored, and the behavior of the rat is recorded by use of a video-recording system under infra-red light. The very first experimental result, which was carried out in collaboration with Inoué's group is shown in Figure 4, in which the amounts of NREM and REM sleep are plotted against clock time. Unlike typical human beings, rats are nocturnal creatures and sleep most of the time during the day but are awake and active during the night. When PGD2 (0.6 pmol/min) was infused continuously for 10 h into the third ventricle of a freely moving rat during the night, both NREM and REM sleep increased significantly during the time of infusion.11 This effect was specific to PGD2, as other PGs were much less effective or totally inactive. Furthermore, it was dose dependent, and as little as a picomolar amount of PGD2 delivered per minute was effective in inducing excess sleep. The amount of PGD2 needed to induce sleep corresponded quite closely to the natural concentration in the brain. This finding meant that “pharmacological” high doses are not necessary, suggesting that the variations in the concentrations of PGD2 that occur naturally in the brain cells may play a role in the regulation of sleep under physiological conditions. Most importantly, however, the sleep induced by PGD2 was indistinguishable from physiological sleep as judged by several behavioral and electrophysiological criteria, including power spectral data. In contrast to PGE2, PGD2 is not pyrogenic but actually caused small amounts of decrease in body temperature, as is observed to occur during physiological sleep.

Figure 4.

I.c.v. infusion of prostaglandin (PG) D2 induces sleep in rats. Open and closed circles indicate the control and experimental values (n= 4–10). NREMS, non-rapid eye movement sleep; REMS, rapid eye movement sleep.

These and other lines of experimental evidence carried out with Japanese monkeys (Macaca mulatta)12 clearly showed that PGD2 induces physiological or natural sleep. In contrast, under the same experimental conditions, sleep induced by sleeping-pill drugs was clearly different from natural or physiological sleep, indicating that PGD2 is indeed a sleep hormone.


In our early experiments mentioned above, we infused chemically synthesized PGD2. Our subsequent studies revealed that in the brain of mammals, PGD2 is produced from the substrate PGH2 by the action of PGD synthase (PGDS). Two types of PGDS have been purified from the brain, their genes cloned, their proteins crystallized, and their detailed structures were delineated in my own and other laboratories. One of them, the lipocalin type, or L-PGDS, is involved in the sleep induction. Surprisingly, as shown by the results of in situ hybridization for the detection of L-PGDS mRNA, L-PGDS was mainly, if not exclusively, detected in the leptomeninges, which is the membrane system surrounding the brain and spinal cord, as well as in the choroid plexus rather than in the brain parenchyma.13 These somewhat unexpected results were further confirmed by immunostaining and direct enzyme activity determination, and essentially the same results were obtained except that a small amount of L-PGDS activity was found to be present in oligodendrocytes. These results clearly showed that PGD2 is an endogenously produced sleep-inducing compound and that it is biosynthesized in non-neural cells. In that sense, PGs are not typical neurotransmitters but should be regarded as “local hormones.”


Having been encouraged by these somewhat unexpected discoveries, we then proceeded to investigate the molecular mechanisms underlying the sleep–wake regulation by PGD2 and E2 in collaboration with a number of investigators in my own and other laboratories in Japan, the USA, Europe, and elsewhere. Our tentative conclusion, initially published in the Philos. Trans. R. Soc. Lond. B. in the year 2000, is briefly summarized in Figure 5.14 L-PGDS, the key enzyme in sleep induction, is present mainly in the arachnoid membrane, and also in the choroid plexus. Once produced, PGD2 is secreted into the cerebrospinal fluid (CSF), and circulates throughout the ventricular and subarachnoidal spaces. In contrast, PGD receptors, abbreviated as DPR in Figure 5, are exclusively localized in a small area on the ventro-rostral surface of the basal forebrain. PGD2 circulating in the CSF then binds to this receptor, where a signal for sleep is generated.

Figure 5.

Molecular mechanisms of sleep regulation by prostaglandin (PG) D2. From reference 14 with permission. DPR, prostaglandin D receptor; L-PGDS, lipocalin-type prostaglandin D synthase; TMN, tuberomammillary nucleus; VLPO, ventrolateral preoptic area.

This signal is then transmitted across the pia membrane and through the brain parenchyma to the ventrolateral preoptic area (VLPO), a sleep center, where sleep neurons are activated, as evidenced by the result of cFos experiments.15 This transduction is mediated via adenosine by way of the A2A adenosine receptor. The VLPO projects to the tuberomammillary nucleus (TMN), a wake center located in the posterior hypothalamus, through GABAergic and galaninergic neurons that send inhibitory signals to down-regulate the wake neurons. Interestingly, Oishi and coworkers recently found that adenosine produced in the TMN induced NREM sleep by inhibiting the histaminergic system via A1 receptors.16

Thus PGD2 induces sleep by up-regulating the sleep neurons via adenosine in the VLPO and at the same time down-regulating the wake neurons in the TMN through GABA and galanin by a flip-flop mechanism. Likewise, wakefulness is promoted by wake substances, such as PGE2 or orexin, via the histamine system. These schemes are still our working hypothesis; and much more effort is needed to clarify the details, to validate these mechanisms in vivo, i.e. in whole live animals and also to investigate the exact role of PGD2 in the circadian and homeostatic control processes of sleep.


In an attempt to find out if PGD2 is indeed involved in the circadian processes in vivo under physiological conditions, we used an enzyme inhibitor and/or a receptor antagonist. After a long and extensive search for an inhibitor of the PGDS, Islam and coworkers found inorganic tetravalent (4+) selenium compounds to be potent, specific, and reversible inhibitors of the brain PGDS.17 These compounds seem to interact with the free sulfhydryl group in the active site of the enzyme, because the inhibition can be reversed by the addition of excess amounts of SH compounds such as glutathione or dithiothreitol. When we administered selenium chloride to a sleeping mouse, it also inhibited the enzyme in vivo, resulting in the cessation of the circadian sleep (Fig. 6).18 When SeCl4 (5 mg/kg) was administered to these mice by an intraperitoneal bolus injection at 11 o'clock in the morning, both NREM and REM sleep were inhibited promptly and effectively; and almost complete inhibition was observed in about 1 h as shown by the closed circles in Figure 6. This inhibitory effect lasted about 5 to 6 h under these conditions, after which a sleep rebound was observed during the night. This inhibition of sleep by SeCl4 could be reversed by sulfhydryl reagents just as in the case of the in vitro enzyme activity. These results clearly showed that the enzyme PGDS plays a crucial role in the maintenance of the sleep state under physiological conditions.

Figure 6.

SeCl4 inhibits sleep in wild-type (WT) mice. Values are means ± SEM (n= 6). *P < 0.05, **P < 0.01 by the paired t-test. NREM, non-rapid eye movement; REM, rapid eye movement.

Likewise, when a PGD receptor (DPR) antagonist, ONO-4127, was infused into the subarachnoid space in the rostral basal forebrain of a sleeping rat, where the receptors are densely localized, both NREM and REM sleep were promptly inhibited dose- and time-dependently and reversibly during the period of infusion. (Fig. 7).18

Figure 7.

Prostaglandin D receptor (DPR)-antagonist inhibits sleep in rats. ONO-4127 was given between 09.00 and 15.00 hours for 6 h, as indicated by the horizontal bars. Values are means ± SEM (n= 6). *P < 0.05, **P < 0.01 by the paired t-test. NREM, non-rapid eye movement; REM, rapid eye movement.

All of these results taken together clearly showed that the PGD2 system, namely, the enzyme PGDS, its product PGD2, and DPR, the receptor of this product, play an essential role in the circadian regulation of sleep in rodents under physiological conditions.


Finally, in order to find out if PGD2 is also involved in the homeostatic regulation of sleep, we then subjected mice to sleep deprivation (SD) experiments, an up-to-date version of the Ishimori-Piéron type of experiment. Since most procedures used by previous investigators including Ishimori and Piéron were usually stressful, in our case, we gently touched the back of mice with a soft tissue paper ball while monitoring sleep by EEG and behavior. The sleep–wake pattern was determined under normal conditions on the first day; and on day 2, the mice were subjected to sleep deprivation during the last 6 h of the light phase (Fig. 8).

Figure 8.

Effect of SD on non-rapid eye movement (NREM) sleep in wild-type (WT) and lipocalin-type prostaglandin D synthase (L-PGDS)-KO mice. Open circles indicate the control and closed circles experimental values. *P < 0.05 **P < 0.01; n= 6. Compared with the baseline day by the paired t-test.

We then compared the effect of sleep deprivation in wild-type mice and knockout (KO) mice for PGDS. These L-PGDS KO mice, generated by Eguchi and coworkers in my laboratory, were supposedly unable to produce PGD2.19 Since sleep is essential for life, the loss of PGDS would be a lethal mutation. However, these mutant mice appeared to be quite healthy and to sleep normally. As shown in Figure 8, the circadian profiles of the sleep–wake patterns of the wild-type and the mutant mice, indicated by open circles, were essentially identical under macroscopic examination. These results suggest that the sleep-regulatory system is composed of a complicated network with built-in redundancies; and thus the deficiency of one system may be effectively compensated or bypassed by other systems during early ontogenic development.

When the WT mice were subjected to sleep deprivation for 6 h immediately before the onset of their wake period, a pronounced rebound was observed in NREM sleep. The total amount of NREM sleep rebound during 12 h after sleep deprivation exceeded more than 60 min. In contrast, little, if any, rebound occurred in NREM sleep in the L-PGDS KO mice under the same experimental conditions. These results clearly show that the PGD2 system plays a crucial role in the homeostatic regulation of NREM sleep, too.


Thus far I have mentioned experimental results obtained with rodents and some times with monkeys. However, in 1985 Roberts and coworkers at the Vanderbilt University in the USA reported that the endogenous production of PGD2 increased up to 150-fold in systemic mastocytosis patients during deep sleep episodes.20 Then, in 1990, Pentreath and coworkers at the University of Salford in the UK reported that the PGD2 concentration was elevated progressively and selectively up to 1000-fold in the CSF of patients with African sleeping sickness caused by Trypanosoma.21 These clinical observations are consistent with the notion that excessive endogenous production of PGD2 induces sleep in man under certain pathological conditions.


  • 1PGD2 is essential for the initiation and maintenance of the sleep state under physiological conditions.
  • 2The adenosine and A2A receptor system is a link between the humoral and neural mechanisms of sleep–wake regulation in the PGD2 system.
  • 3The PGD2 system, namely, the enzyme PGDS, its product PGD2, and DPR, the receptor of this product, plays a crucial role in the homeostatic as well as circadian regulation of NREM sleep.

It is possible that PGD2 may be most likely the endogenous sleep substance that was described by Ishimori and Piéron almost 100 years ago, although it is no longer possible to prove it.


I would like to express my most sincere thanks to all my collaborators, past and present, both in and out of my laboratory, whose names are too numerous to mention here. My special thanks are expressed to Dr Yoshihiro Urade, Dr Naomi Eguchi, and Dr Zhi-Li Huang in my laboratories in Kyoto and Osaka as well as Dr Clifford B. Saper and Dr Thomas E. Scammell of the Harvard Medical School and Dr Michel Jouvet and Dr Pierre-Hervé Luppi of the Université Claude Bernard for their collaboration in some of the crucial experiments. This is the end of my swan song; but studies are still in progress in my own and other laboratories to attain our final goal of understanding in biochemical terms the entire mechanism of sleep–wake regulation by PGs. We realize that we have a long way to go, but hopefully, we are on the right track. The works presented in this paper were supported by Grants-in-Aid from the Ministry of Health, Labour, and Welfare of Japan; the Ministry of Education, Culture, Sports, Science, and Technology of Japan; the Program for Promotion of Basic and Applied Researches for Innovations in Bio-oriented Industry; Ono Pharmaceutical Co., Ltd.; Takeda Pharmaceutical Co., Ltd.; and the Osaka Bioscience Institute. Finally, thanks are due to Ms Naoko Ueda for her secretarial assistance.


The author indicated no potential conflict of interests.