Hypothalamic Orexin Prevents Hepatic Insulin Resistance via Daily Bidirectional Regulation of Autonomic Nervous System in Mice

Maintaining glucose and energy homeostasis throughout the day is essential for survival. Therefore, different metabolic functions are timely activated according to the circadian rhythm. Recently, it has also become clear that fine tuning of daily metabolic rhythms is crucial for preventing metabolic abnormalities. For instance, chronic disruption of circadian rhythm in humans, as seen in shift workers, increases the risk of obesity and type 2 diabetes (1,2), and circadian rhythms of glucose regulation are impaired in diabetic animals and patients with diabetes (3).

Circadian rhythms of peripheral tissues, including liver, are entrained to the central biological clock located in the suprachiasmatic nuclei (SCN) via autonomic and hormonal pathways (4). The liver plays a pivotal role in maintaining optimal glucose levels via hepatic glucose production (HGP) and glucose uptake. The daily rhythm of HGP is regulated by the autonomic nervous system under SCN control, leading to generation of a daily rhythm in basal blood glucose levels independently of the feeding rhythm (5). The central clock is also crucial for preventing metabolic abnormalities in the liver, since lesions of the SCN caused severe hepatic insulin resistance in mice (6); however, the underlying mechanisms remain uncertain. It is of note that nutrient excess causes endoplasmic reticulum (ER) stress in hepatocytes, a main cause of hepatic insulin resistance, and the cells need “resting time” to recover from the ER stress (7,8). These raise a possibility that daily changes in blood glucose levels are required for maintaining ER homeostasis and insulin sensitivity in the liver.

The perifornical hypothalamic area is one of the primary sites for mediating the functional interplay between the brain and liver (9). Orexin A and B, a pair of neuropeptides, are produced in neurons in the perifornical and lateral hypothalamus. Importantly, the SCN sends signals to orexin-producing neurons via the dorsomedial hypothalamus (10,11) and controls the orexin secretion (12). The central orexin A concentrations exhibit a daily rhythm with the highest levels during wakefulness (13,14), although the orexin expression was downregulated in diabetic db/db and ob/ob mice (15). Orexin regulates the sleep-wake rhythm, energy homeostasis, and autonomic nerve balance through orexin receptor-1 (OX1R) or -2 (OX2R) (16). Moreover, orexin appears to be involved in the maintenance of insulin sensitivity, because orexin-deficient male mice fed normal chow diet exhibited age-related development of systemic insulin resistance despite normal body weight (17,18), and because orexin-deficient narcoleptic patients with cataplexy showed BMI-independent metabolic alterations, including insulin resistance (19). However, the role of orexin in glucose homeostasis remains elusive, since orexin A is reported to cause the blood glucose–elevating and –lowering effects in rodents depending on the experimental conditions (20–22). We hypothesized that orexin has the ability to promote generation of the glucose rhythm via bidirectional regulation of HGP. Moreover, the orexin-controlled glucose rhythm, if any, would help to prevent hepatic ER stress and insulin resistance.

The aim of the current study was therefore to clarify, for the first time, whether hypothalamic orexin acts as a timekeeper in daily glucose regulation and whether it contributes to the maintenance of hepatic insulin sensitivity. To this end, we examined the influence of orexin deficiency and orexin A administration on daily changes in blood glucose levels and the regulation of hepatic gluconeogenesis in mice. Furthermore, we investigated whether orexin deficiency causes excessive ER stress in the liver, leading to hepatic insulin resistance during aging.

Mice were housed at 23–25°C with free access to normal chow diet and water under a 12 h–12 h light-dark cycle (lights on from 0700 to 1900 h). Zeitgeber time (ZT) 0 and 12 are defined as lights-on and -off times, respectively. The animal experiments were performed between 1000 and 1600 h, unless otherwise indicated. All experimental procedures used in this study were approved by the Committee of Animal Experiments at the University of Toyama or Kanazawa University. Male C57BL/6J mice, male C57BLKS/J Iar-m+/m+ mice (m+/m+), and male C57BLKS/J Iar-+Lepr^db/+Lepr^db mice (db/db) were purchased from Japan SLC (Shizuoka, Japan). Wild-type (WT) mice, preproorexin-deficient (Orexin^−/−) mice, Ox1r-deficient (Ox1r^−/−) mice, and Ox2r^−/− mice (N5-N6 backcross to C57BL/6J) were prepared as described previously (17,23). Continuous measurements of the amounts of food intake, energy consumption, and locomotor activity were performed using metabolic cages (MK-5000RQ; Muromachi Kikai, Tokyo, Japan).

Orexin A was purchased from the Peptide Institute (Osaka, Japan). Human regular insulin Novolin R was provided by Novo Nordisk (Copenhagen, Denmark). Anti–insulin receptor substrate 1 (IRS1) antibody and anti-IRS2 antibody were purchased from Millipore (Temecula, CA), and an anti–phospho-inositol-requiring enzyme 1α (IRE1α) (Ser⁷²⁴) antibody was from Abcam (Cambridge, MA). Anti–insulin receptor β antibody and anti-Akt1 antibody were purchased from Santa Cruz Biotechnology (Santa Cruz, CA). Anti-IRE1α antibody, anti–phospho-SAPK/Jun N-terminal kinase (JNK) (Thr¹⁸³/Tyr¹⁸⁵) antibody, anti–phospho-Akt (Ser⁴⁷³) antibody, anti–β-actin antibody, anti–cAMP response element binding protein (CREB) antibody, anti–phospho-CREB (Ser¹³³) antibody, anti–mammalian target of rapamycin (mTOR) antibody, anti–phospho-mTOR (Ser²⁴⁴⁸) antibody, anti–signal transducer and activator of transcription 3 (STAT3) antibody, and anti–phospho-Stat3 (Tyr⁷⁰⁵) antibody were from Cell Signaling Technology (Beverly, MA).

Blood samples were collected from the tail vein or orbital sinus of mice under anesthesia. Serum levels of insulin (Takara Bio, Shiga, Japan), glucagon (Wako Pure Chemical, Osaka, Japan), and corticosterone (Cayman Chemical Company, Ann Arbor, MI) were measured with specific ELISA. Fasting serum levels of triglyceride, nonesterified fatty acids, and cholesterol were determined using colorimetric kits (Wako). Blood glucose levels were measured using a glucose meter (FreeStyle Freedom; Nipro, Osaka, Japan). Tissue contents of glycogen were measured using a glucose assay kit (Sigma-Aldrich).

Intracerebroventricular (ICV) administration of orexin A (ICV–orexin A) was performed as described previously (17). In brief, a guide cannula was implanted into the lateral ventricle of mice. After a 7-day recovery period, the mice were ICV injected with 3 μL saline or orexin A using a Hamilton syringe. For the glucose clamp study, as described below, an ICV injection cannula with extension tubing (Plastics One, Roanoke, VA) was used for the drug injection. Hepatic parasympathectomy of C57BL/6J mice at 7 weeks of age was performed immediately before the ICV cannulation using standard protocols as described previously (24). Furthermore, orexin A was administered continuously according to a previously described method (25). In brief, an osmotic pump (Alzet 1007D; Durect, Cupertino, CA) containing saline was subcutaneously placed on the back, and an ICV cannula connected to the pump (Alzet Brain Infusion Kit 3; Durect) was implanted into the lateral ventricle in 7-week-old db/db mice under anesthesia. After a 7-day recovery period, the mice were administered either orexin A (0.3 nmol/day) or saline by replacing the pump containing each solution.

Hyperinsulinemic-euglycemic clamp studies were performed as described previously with minor modification (26). After 7 days of recovery from ICV cannulation surgery, 8-week-old C57BL/6J mice were fasted for 16 h and injected with orexin A or saline through the ICV cannula. [3-³H]glucose infusion at 0.1 μCi/min was started 60 min after the ICV injection. For measurement of the basal HGP rate, blood samples were collected 120 min after the ICV injection. Subsequently, insulin was infused at 1.25 mU/kg/min with variable amounts of 40% glucose solution to maintain a blood glucose level of 90–120 mg/dL. The glucose infusion rate was measured between 90 and 120 min after the initiation of insulin infusion to calculate the HGP rate. For the pyruvate tolerance test to evaluate the activity of HGP, fasted mice were intraperitoneally injected with pyruvate (2 g/kg), and blood samples were collected from the tail vein.

The tissue samples dissected from mice were homogenized, lysed, and subjected to immunoblotting, as previously described (17). For the analysis of insulin signaling, mice fasted overnight were injected with insulin or PBS. Fifteen minutes after injection, tissues were rapidly isolated.

RNA extraction, reverse transcription, and quantitative real-time PCR were performed using a Mx3000p real-time PCR system (Stratagene, La Jolla, CA) with SYBR Premix Ex Taq II (Takara Bio), as described previously (25). Expression level of the gene transcript of interest was calculated as a ratio to that of S18 ribosomal protein in each sample. Primer pairs are listed in Supplementary Table 1.

The liver tissues were isolated at ZT 8 from male WT and Orexin^−/− mice at 6 months of age under fed conditions. Total RNA pooled from two mice was used to synthesize double-strand cDNA for each experimental group. The gene transcript profiles were analyzed using a GeneChip system with Mouse Genome 430 2.0 Array (Affymetrix, Santa Clara, CA), GeneSpring software (Agilent Technologies, Santa Clara, CA), and Ingenuity Pathway Analysis software (Ingenuity Systems, Mountain View, CA), as described previously (27).

The liver tissues isolated from mice were immediately fixed with 2% glutaraldehyde in 0.1 mol/L PBS (pH 7.4) for 90 min at 4°C and postfixed with 1% osmium tetroxide for 90 min at 4°C. The specimen was dehydrated, embedded in Epoxy resin, and sectioned using standard protocols. Ultrathin sections (70-nm thickness) were stained with uranyl acetate and lead citrate. Electron micrographs were taken with a transmission electron microscope (JEOL JEM-1400TC, Tokyo, Japan).

Data are expressed as the means ± SEM. The significance of differences between two groups was assessed by Student t test, and differences between multiple groups were assessed by one-way ANOVA followed by the Bonferroni test. Daily rhythm in blood glucose levels was determined using Time Series Analysis Serial Cosinor 6.3 software (Expert Soft Technologie, Esvres, France). P < 0.05 was considered significant.

Blood glucose levels exhibited daily changes that peaked at ZT 14 in the dark phase in 3- and 6-month-old WT mice under ad libitum–fed conditions; however, Orexin^−/− mice at 3 and 6 months of age did not show the increase in blood glucose at ZT 14, when compared with ZT 5 (Fig. 1A and B). Cosinor analysis showed the presence of a significant daily rhythm in 3- and 6-month-old WT mice and 3-month-old Orexin^−/− mice (P < 0.01) but not in 6-month-old Orexin^−/− mice (P = 0.725). In addition, the glucose levels from ZT 5 to 11 and ZT 20 to 2 were significantly higher in 6-month-old Orexin^−/− mice than those in WT mice (Fig. 1B). The timing of feeding and locomotion in Orexin^−/− mice was identical to that of WT mice at both ages, although their absolute activities in the dark phase were lower in Orexin^−/− mice (Fig. 1C–F). Body weights and 24-h energy expenditure were not different between the two groups (data not shown). Serum levels of insulin and glucagon in Orexin^−/− mice under fed condition were almost comparable with those of WT mice; the detected changes in the hormone levels were not reflected in the glucose changes (Fig. 1G–J). Serum levels of corticosterone, triglyceride, nonesterified fatty acids, and total cholesterol were not different between WT and Orexin^−/− mice (data not shown). Thus, orexin deficiency preferentially affected daily changes in glucose levels.

Since daily glucose rhythm largely depends on HGP, profiles of hepatic gluconeogenic gene expression in Orexin^−/− mice were investigated. At 3 months of age, Orexin^−/− mice exhibited a decrease in the mRNA levels of peroxisome proliferator–activated receptor γ coactivator 1α (Pgc1a) at ZT 14 (Fig. 1K). The expression of glucose-6-phosphatase (G6Pase) was also deranged in the light phase (Fig. 1L), although the Pepck expression was unaltered (data not shown). At 6 months of age, Orexin^−/− mice exhibited a significant increase in the expression of Pepck at ZT 8 and a trend of increase in the expression of G6Pase (Fig. 1M and N) and Pgc1a (data not shown) at ZT 8. Thus, orexin deficiency perturbed the daily rhythms of hepatic gluconeogenic gene expression.

To characterize the effect of orexin A on blood glucose levels, orexin A at 3 nmol was ICV injected into C57BL/6J mice. After the injection, blood glucose increased and peaked at 60 min (Fig. 2A). This effect was inhibited by combined pretreatment of α- and β-adrenergic antagonists (phenoxybenzamine and propranolol) (Fig. 2B). ICV–orexin A at 3 nmol caused an increase in the basal rate of HGP assessed by a tritiated glucose infusion method in C57BL/6J mice, although orexin A had no effect on HGP during hyperinsulinemic-euglycemic clamp (Fig. 2C).

Diabetic db/db mice exhibited hyperglycemia throughout the day, although there existed a significant daily rhythm in blood glucose (P < 0.01, determined by Cosinor analysis) that peaked at ZT 17 (Supplementary Fig. 1A). In addition, preproorexin mRNA levels in the hypothalamus in db/db mice were lower than those in C57BL/6J mice (Supplementary Fig. 1B). We therefore examined the influence of nighttime orexin A supplementation on blood glucose. In db/db mice fasted for 6 h, ICV–orexin A at 0.3 nmol did not significantly affect the glucose levels (Fig. 2D), whereas ICV–orexin A at 3 nmol acutely increased blood glucose (Fig. 2E). In nondiabetic m+/m+ mice, both doses of orexin A significantly elevated the glucose levels (Fig. 2D and E). To further reveal the impact of orexin A on daily glucose rhythm, orexin A was administered daily to db/db mice at ZT 14 under ad libitum–fed condition, since cerebrospinal fluid levels of orexin A increase in the dark phase in rodents (14). Intriguingly, the first nighttime administration of orexin A at 0.3 nmol had no acute effect on blood glucose but gradually lowered it in the subsequent light phase (Fig. 2F). Moreover, the second and third administrations caused the rapid glucose elevation at ZT 15 to 16, followed by the lowering of glucose at ZT 2, thereby generating an apparent glucose oscillation. Similarly, in 20-week-old db/db mice, ICV–orexin A at 0.3 nmol reduced fed glucose levels at ZT 14 (Fig. 2G) without affecting body weight (Fig. 2H). It also reduced fasting glucose and hepatic Pgc1α, Pepck, and G6Pase without affecting serum insulin at ZT 8 (Fig. 2I–K). Under these conditions, increased phosphorylations of CREB, JNK, and mTOR in the liver of db/db mice, relevant to hyperglycemia, were reduced to nondiabetic control levels (Supplementary Fig. 2A–C). ICV–orexin A at ZT 14 for 3 days had no effect on insulin-induced Akt phosphorylation in the skeletal muscle of db/db mice dissected at ZT 6 (Supplementary Fig. 3A). Daytime administration (at ZT 2) of orexin A at 0.3 nmol was obviously less effective at lowering glucose in db/db mice (Fig. 2L) than its nighttime administration (at ZT 14) as shown above (Fig. 2F). In addition, continuous ICV infusion of orexin A (0.3 nmol/day) using an osmotic pump did not affect blood glucose levels and body weight in db/db mice (Fig. 2M), whereas hepatic G6Pase was increased by the orexin infusion (Fig. 2N). These results indicate that daily increase in central orexin levels in the dark phase contributes to the onset of glucose oscillation via adrenergic pathway.

To investigate whether adrenaline substitutes for orexin to induce daily glucose oscillation, adrenaline was injected intraperitoneally into db/db mice at ZT 14 under fed conditions. Adrenaline acutely caused short-term elevation of blood glucose; however, it failed to affect glucose levels in the light phase (Fig. 3A). Thus, the glucose elevation by sympathetic signals was not enough for subsequent lowering of blood glucose. After ICV administration of orexin A at 3 nmol, hepatic STAT3 phosphorylation, indicative of parasympathetic input to the liver (28), was increased at 120 min in db/db mice compared with respective saline controls (Supplementary Fig. 2D). We therefore conducted hepatic parasympathectomy experiments to investigate the involvement of parasympathetic pathway in the glucose-lowering effect of orexin. In sham-operated C57BL/6J mice, ICV–orexin A (3 nmol) at ZT 14 enhanced the daily changes in fed glucose levels (Fig. 3B); the rapid glucose elevations in the dark phase by the second and third administrations were followed by lowering of glucose in the subsequent light phase, when compared with the saline control. No change was observed in body weight and food intake by daily orexin A administration (data not shown). After the orexin treatment, blood glucose levels at 15 min in pyruvate tolerance test were slightly decreased compared with the saline control (Fig. 3C). In contrast, hepatic parasympathectomy abolished the glucose-lowering effect of orexin A in the light phase (Fig. 3D), whereas acute increases in glucose levels by orexin A in the dark phase were not statistically different between sham-operated and parasympathectomized mice (Fig. 3B and D). Moreover, in the subsequent pyruvate tolerance test, the glucose elevation at 15 min was greater in the orexin-treated group than the saline control group (Fig. 3E). Therefore, hepatic parasympathetic activity appears to be required for the glucose-lowering effect of orexin A. Neither ICV–orexin A at ZT 14 for 3 days nor hepatic parasympathectomy affected insulin-induced Akt phosphorylation in the skeletal muscle at ZT 6 (Supplementary Fig. 3B).

To further reveal the mode of action of orexin A on hepatic glucose regulation, the dose-response relationship was examined. Under fasting condition, blood glucose levels were elevated by ICV–orexin A at 3 nmol but lowered at 3 fmol (Fig. 4A). Serum levels of insulin were not altered by these treatments (Fig. 4B). Consistently, the expression levels of hepatic gluconeogenic genes, Pepck and G6Pase, were increased by ICV–orexin A at 3 nmol but decreased at 3 fmol (Fig. 4C and D). In the liver of Ox1r^−/− mice, the decreasing effect of ICV–orexin A at 3 fmol on the Pepck mRNA levels disappeared, whereas ICV–orexin A at 3 nmol normally increased the Pepck levels (Fig. 4E). In contrast, ICV–orexin A at both 3 fmol and 3 nmol decreased the hepatic Pepck levels in Ox2r^−/− mice (Fig. 4F). The orexin A (3 nmol)–induced increase in Pepck expression was inhibited by pretreatment with an adrenergic α-blocker phenoxybenzamine (Fig. 4G) but not a muscarinic antagonist atropine (data not shown), whereas the orexin A (3 fmol)–induced decrease in Pepck expression was inhibited by pretreatment with atropine (Fig. 4H) but not phenoxybenzamine (data not shown). Therefore, it is likely that the high dose of orexin A enhances the Pepck levels via the OX2R and sympathetic nerve pathways, whereas the low dose of orexin A suppresses them via the OX1R and parasympathetic pathways.

Next we investigated the impact of long-term deficiency of the daily orexin action on hepatic glucose metabolism. First, the expression profiles of core clock components in the liver were compared between WT and Orexin^−/− mice, since the biological clock system serves to control metabolic efficacy (2). The mRNA levels of Period 2 (Per2) at ZT 8 were significantly, and those of Per1 were slightly, increased in 6-month-old Orexin^−/− mice, whereas no changes were detected at 3 months of age (Supplementary Fig. 4). Moreover, since the glucose levels and hepatic Pepck levels were deranged mainly during daytime in Orexin^−/− mice along with aging (Fig. 1), we explored the underlying mechanism in the light phase. In the pyruvate tolerance test, the glucose response was greater in both male and female Orexin^−/− mice at 6 months of age (Fig. 5A and B) but not 3 months of age (data not shown), when compared with age-matched WT controls. Glycogen contents in the liver and skeletal muscle were not different between WT and Orexin^−/− mice at 6 months of age (data not shown). Concerning the insulin signaling, phosphorylation of Akt at Ser⁴⁷³ induced by insulin was significantly reduced in the liver but not skeletal muscle of Orexin^−/− mice at 6 months of age compared with WT controls (Fig. 5C). The mRNA levels of Irs2 and the protein levels of IRS1 and IRS2 (Fig. 5D and E), but not insulin receptor (not shown), were significantly decreased in the liver of 6-month-old Orexin^−/− mice under fed conditions. The GeneChip and pathway analyses indicated that the gene expression profile in the liver of 6-month-old Orexin^−/− mice at ZT 8 under fed conditions was highly related to diabetes and glucose metabolism disorder (Supplementary Table 2). On the other hand, the mRNA levels of tumor necrosis factor α (Tnfα) and interleukin-1β (Il1β) in epididymal fat tissues were not different between WT and Orexin^−/− mice (data not shown). Thus, chronic deficiency of daily orexin action perturbs the clock system in the liver and impairs the regulation of HGP in the light phase.

ER is an organelle functionally linked to the circadian clock and insulin sensitivity (7,29). Under 5-h fasting conditions, the mRNA levels of C/EBP-homologous protein (Chop), an ER stress marker, were increased in the liver of Orexin^−/− mice at 6 months of age (Fig. 6A) but not at 3 months of age (data not shown). When ER stress was induced by 1-h refeeding after the 24-h fasting period, according to a previously reported ER stress paradigm (30), abnormal increases in the phosphorylation of IRE1α (Fig. 6B) and JNK (Fig. 6C) were observed in the liver of Orexin^−/− mice at 6 months of age compared with those of WT controls. Similarly, excessive phosphorylation of JNK was induced by refeeding after fasting for 24 h in the liver of Ox1r^−/− and Ox2r^−/− mice at 4 months of age compared with those of WT mice (Fig. 6D). Electron microscope analysis demonstrated that Orexin^−/− mice at 6 months of age displayed typical features of ER stress, i.e., the expansion and dilation of ER in the liver (Fig. 6E). Thus, lack of daily orexin action caused age-related development of ER stress in the liver, relevant to hepatic insulin resistance.

The SCN-controlled biological clock rhythm is crucial for maintaining hepatic insulin sensitivity (6). In the current study, we investigated the underlying mechanisms and found that 1) hypothalamic orexin bidirectionally regulates hepatic gluconeogenesis via control of autonomic balance, leading to generation of the daily blood glucose oscillation. We further found that 2) the orexin-induced periodical activation and inactivation of hepatic gluconeogenesis reduces ER stress in the liver, thereby preventing hepatic insulin resistance. These results provide the first evidence that orexin plays a key role in the maintenance of hepatic insulin sensitivity by circadian system.

The SCN controls daily glucose rhythm independently of feeding rhythm. The daily glucose oscillation peaks at the time of awakening mainly through increase in HGP (9,31). The balance between sympathetic and parasympathetic output to the liver is crucial for maintenance of the glucose rhythm (32), although the integrating mechanism has long been uncertain. We found that central action of orexin A at the awake period increased the amplitude of daily glucose oscillations via HGP in mice. Moreover, deficiency of endogenous orexin disrupted the daily glucose rhythm. These suggest that orexin mediates the regulation of HGP rhythm by circadian system.

Hypothalamic orexin neurons project to both sympathetic and parasympathetic preganglionic neurons in the central nervous system (33,34), and orexin activates both types of neurons (35,36). In fact, renal sympathetic nerve activity and blood pressure were increased by ICV–orexin A at high doses (3 pmol to 3 nmol) but decreased at low dose (3 fmol) in rats (37). Consistently, we found that ICV–orexin A at 3 nmol enhanced hepatic gluconeogenic gene expression via the α-adrenergic pathway, whereas ICV–orexin A at 3 fmol suppressed it via cholinergic pathway in fasted mice. These raise a possibility that when local concentrations of orexin A increase and then decrease after ICV administration, the mode of action may be converted from a sympathetic to a parasympathetic nerve–mediated one. Indeed, high dose (3 nmol) of orexin A at the awake period caused sympathetic pathway–dependent blood glucose elevation that was accompanied by parasympathetic pathway–dependent lowering of glucose at resting period. Therefore, daily changes in the orexinergic tones under control of circadian system appear to serve as a master “on-off” signal for HGP.

In diabetic mice such as db/db and ob/ob mice, orexin expression is suppressed due to hyperglycemia (15,38). We found that ICV–orexin A more profoundly reduced blood glucose levels in db/db than C57BL/6J mice. These suggest that physiological orexin A action plays a fundamental role in the maintenance of glucose homeostasis, and supplementation of orexin A to ameliorate its suppressed level in the diabetic state can improve hyperglycemia. In addition, unlike normal daily glucose oscillation that peaked at ZT 14, the highest glucose levels were observed at ZT 17 in db/db and 6-month-old Orexin^−/− mice, implying that the peak-time change in db/db mice may be related, at least in part, to orexin insufficiency. Importantly, the present chronopharmacological approach demonstrated that the timing of orexin A administration is crucial for glucose regulation; its administration at the sympathetic-dominant nighttime awake period, which mimics the physiological orexin secretion pattern, was more effective to lower blood glucose than parasympathetic-dominant daytime resting period. The diabetic db/db mice have been reported to exhibit the increased sympathetic and decreased parasympathetic tones in autonomic cardiovascular regulation (39). Increased sympathetic tone enhances parasympathetic control, leading to an accentuated antagonism in heart (40). By similar mechanism, the parasympathetic control of HGP by orexin may be enhanced, when the sympathetic tone is increased in the diabetic state and/or during the active phase.

Interestingly, ICV–orexin A at both high and low doses exclusively suppressed hepatic Pepck expression in Ox2r^−/− mice, whereas they only increased it in Ox1r^−/− mice. Therefore, hepatic gluconeogenesis may be enhanced by OX2R/sympathetic signals and suppressed by OX1R/parasympathetic signals (Fig. 6F). However, since OX1R and OX2R are widely and differentially distributed throughout the brain (41), more precise study is required to clarify the contribution of each receptor subtype in various brain regions to hepatic glucose regulation.

In type 2 diabetic subjects, fasting plasma glucose levels were elevated due to the increase in basal HGP, which is entirely explained by an increase in hepatic gluconeogenesis (42). In our study, Orexin^−/− mice at 6 months of age exhibited hyperglycemia associated with increased activity of hepatic gluconeogenesis at daytime resting period, despite normal feeding behavior. These abnormalities appear to be due to hepatic insulin resistance, since 6-month-old Orexin^−/− mice exhibited impairment of insulin signaling in the liver but not skeletal muscle. Systemic insulin resistance, glucose intolerance, and impaired insulin signaling in the skeletal muscle become apparent in Orexin^−/− mice older than 9 months of age, as reported previously (17). Thus, orexin deficiency appears to cause insulin resistance preferentially in the liver during the aging process.

The ER stress response is a short-term adaptive system to cope with abnormal increase in unfolded proteins in the ER lumen; in the liver, protein synthesis for metabolic reprogramming is promoted during postprandial state or by refeeding after fasting, resulting in the development of ER stress (43). However, prolonged ER stress is detrimental and leads to the development of insulin resistance and type 2 diabetes (44,45). In this deleterious process, the IRE1α/JNK signaling pathway plays a major role. In our study, abnormal increase in hepatic IRE1α/JNK signaling was induced by refeeding after 24-h fasting in the liver of Orexin^−/− mice at 6 months but not 3 months of age. Since the glucose rhythm was already disrupted in 3-month-old Orexin^−/− mice, nonrhythmic HGP throughout the day may gradually impair the machinery for protecting ER stress in the liver. Ox1r^−/− and Ox2r^−/− mice exhibited similar abnormalities, suggesting that the absence of daily action of orexin, rather than the complete lack of orexin, is the main cause of the increased ER stress. Periodical action of endogenous orexin under the SCN control may provide the time for recovery from ER stress in the liver by suppressing HGP at resting period and prevent hepatic insulin resistance (Fig. 6F).

Although glucose variability is supposed to be a causal factor in the development of vascular complications or mortality, a larger study did not find any relation between them; therefore, so far, there is little consensus on the importance of glucose variability (46). Rather, it has been reported that nighttime-restricted feeding accompanied by nocturnal increase in carbohydrate utilization improved metabolic state in mice fed a high-fat diet (47). Moreover, ER stress can be mitigated by low-level preconditioning stimuli (48), and circadian clock–dependent rhythmic activation of the IRE1α signaling in the ER of liver is required for normal metabolic functions in mice (29). Therefore, daily oscillation of basal glucose levels under orexin control is considered to be beneficial for preventing metabolic disorders.

In conclusion, the current study indicated that orexin functions as a timekeeper to regulate the daily on-off rhythm in HGP along the sleep-wake cycle. Importantly, this mechanism was required for not only supplying energy at the time of awakening but also preventing the development of hepatic insulin resistance (Fig. 6F). Therefore, the present results may provide a basis for the development of novel chronotherapy against metabolic diseases associated with circadian rhythm disturbance, including type 2 diabetes. Pharmacological or nonpharmacological intervention to augment the amplitude of daily change in orexinergic tone might be therapeutically valuable.

Acknowledgments. The authors thank K. Yamaguchi, C. Sugawara, T. Nagata, K. Nishio, and Dr. H. To (University of Toyama, Toyama, Japan) for technical assistance.

Funding. This work was supported by the Japan Society for the Promotion of Science Grants-in-Aid for Scientific Research (KAKENHI) grants 24591317 and 24390056 and grants from the Takeda Science Foundation (H.T.), the Smoking Research Foundation (H.T.), the Research Foundation for Pharmaceutical Sciences (H.T.), the Japan Diabetes Foundation (T.S.), the Foundation for Growth Science (T.S.), and the Tamura Science and Technology Foundation (T.S.).

Duality of Interest. No potential conflicts of interest relevant to this article were reported.

Author Contributions. H.T. researched the data and wrote the manuscript. E.T., Y.N., K.T., M.F., T.A., K.K., Y.A., I.T., K.K., and H.I. researched the data. T.W., M.Y., T.Sak., and T.Sas. contributed to discussion and wrote the manuscript. H.T. and T.Sas. are the guarantors of this work and, as such, had full access to all the data in the study and take responsibility for the integrity of the data and the accuracy of the data analysis.

Prior Presentation. Part of this study was presented at the 74th Scientific Sessions of the American Diabetes Association, San Francisco, CA, 13–17 June 2014.