The role of peripheral vs. central circadian rhythms and Clock in the maintenance of metabolic homeostasis and with aging was examined by using ClockΔ19+MEL mice. These have preserved suprachiasmatic nucleus and pineal gland rhythmicity but arrhythmic Clock gene expression in the liver and skeletal muscle. ClockΔ19+MEL mice showed fasting hypoglycemia in young-adult males, fasting hyperglycemia in older females, and substantially impaired glucose tolerance overall. ClockΔ19+MEL mice had substantially reduced plasma insulin and plasma insulin/glucose nocturnally in males and during a glucose tolerance test in females, suggesting impaired insulin secretion. ClockΔ19+MEL mice had reduced hepatic expression and loss of rhythmicity of gck, pfkfb3, and pepck mRNA, which is likely to impair glycolysis and gluconeogenesis. ClockΔ19+MEL mice also had reduced glut4 mRNA in skeletal muscle, and this may contribute to poor glucose tolerance. Whole body insulin tolerance was enhanced in ClockΔ19+MEL mice, however, suggesting enhanced insulin sensitivity. These responses occurred although the ClockΔ19 mutation did not cause obesity and reduced plasma free fatty acids while increasing plasma adiponectin. These studies on clock-gene disruption in peripheral tissues and metabolic homeostasis provide compelling evidence of a relationship between circadian rhythms and the glucose/insulin and adipoinsular axes. It is, however, premature to declare that clock-gene disruption causes the full metabolic syndrome.
there is increasing evidence that key physiological systems are strongly influenced by cellular rhythmicity, including metabolic homeostasis. The capacity to generate biological rhythmicity and to become entrained to the external environment is a property of most cells throughout the body and not solely that of the suprachiasmatic nucleus (SCN) (12). This new understanding of rhythmicity has important implications for the way that we view and address certain diseases, including diabetes. Cellular rhythmicity is generated by a highly conserved suite of transcription factors (27), with the role of the SCN being to entrain the rhythmic expression of transcription factors in peripheral tissues via neural and hormonal pathways (5). Recent microarray studies have shown that a large number of genes, including some that play a major role in the maintenance of metabolic homeostasis, exhibit high-amplitude rhythms in expression across the day in the liver and other tissues (2, 23). These observations may be physiologically relevant to the approximately twofold higher prevalence of diabetes in shift workers compared with day workers, which is exacerbated with increasing exposure to shift work (17). Shift workers also have a 40% excess risk of cardiovascular disease compared with day workers and twice the prevalence of gastrointestinal disease (17). Together, these observations raise the question of whether abnormal rhythmicity provoked by shift-work schedules may be a causal factor in the metabolic syndrome seen in these people. People engaged in shift work over the long term are forced to regularly shift their work/rest period by up to 12 h for prolonged periods, and this lifestyle disrupts sleep/wake rhythmicity, eating patterns, and light exposure (31). Direct evidence for the involvement of disrupted central and peripheral rhythmicity in impaired metabolic homeostasis in humans and the possible mechanisms responsible are currently lacking, however. Recent studies in nonhuman species have shown that persistent phase shifting of the light/dark cycle disrupts rhythmicity in peripheral tissue clock-gene expression (7), causes abnormal weight gain (33), and may cause earlier death (25), but there is little direct evidence that disrupted cellular rhythms impair metabolic homeostasis.
Animals with mutations in clock genes that disrupt cellular rhythmicity may provide a novel experimental model for investigating the impact of arrhythmicity on physiological processes. One such mouse, the ClockΔ19 (C57BL/6J) mutant (36), has a mutation in Clock that results in the truncation of transcription of exon 18 and deletion of exon 19. The resultant protein can bind to its partner BMAL1, but the heterodimer is incapable of initiating transcription through the CACGTG E-box (10) in the promoter of other clock genes (per and cry), clock-controlled transcription factors (e.g., dec1), and functional genes (e.g., vasopressin and plasminogen activator inhibitor 1). Recent evidence suggests that ClockΔ19 mutant mice have disrupted transcription of a wide range of genes throughout the periphery (23) and, when placed on continuous darkness, exhibit transient behavioral (wheel running) rhythmicity with a period of ∼27 h, then arrhythmicity after a few weeks. ClockΔ19 (C57BL/6J) mutant mice, like the majority of laboratory mice, are also melatonin deficient due to mutations in the genes for the melatonin-synthesizing enzymes, AA-NAT and HIOMT (6). ClockΔ19 (C57BL/6J) mutant mice are reported to express a metabolic syndrome (34). Recently, however, mutations in nicotinamide nucleotide transhydrogenase and markedly decreased glucokinase enzyme activity were reported in the parent C57BL/6J strain of the original ClockΔ19 mutant mice in association with impaired glucose tolerance (8, 9, 32). Whether the metabolic syndrome in ClockΔ19 (C57BL/6J) mutant mice is due to the absence of melatonin, which is involved in the transmission of central rhythmicity to the periphery and is also aberrant in this model, due to peripheral rhythm disruption, or due to the interactions of these defects with mutations that directly impair their glucose tolerance cannot be determined. Similarly, the extent to which loss of central and/or peripheral rhythms can impair glucose tolerance against a more resistant genetic background in contrast to the diabetes-prone C57BL/6J mouse is unclear.
We have, therefore, developed a modified ClockΔ19 mutant mouse that, unlike the original strain, retains central neuroendocrine rhythmicity (melatonin) (16) while showing evidence of liver and muscle arrhythmicity (14). In addition, the genetic background (CBA) of this modified mutant ensures that they are not prone to spontaneous insulin resistance or obesity independent of the ClockΔ19 mutation. As a consequence, this ClockΔ19+MEL mouse provides a model for investigating the specific impact of liver and muscle arrhythmicity in the presence of maintained central neuroendocrine rhythmicity on glucose homeostasis.
Adult male and female ClockΔ19+MEL mutants (16) and their wild-type control mice were obtained from our breeding colony and were maintained under a lighting schedule of 12 h of light and 12 h of darkness with lights off at 2000. Animals were fed normal rodent chow containing 22.7% protein, 7.0% fat, and 5.0% fiber. The ClockΔ19+MEL mutant strain was established by selective breeding of ClockΔ19 mutants (36) (kindly provided by Drs. J. Takahashi and M. Vitatema) with melatonin-proficient (11) CBA/6CaH mice. The resulting mutant mice have functional pineal AA-NAT and HIOMT enzymes and synthesize melatonin. The inheritance of the ClockΔ19 mutation was monitored in the breeding program by PCR of tail DNA (29). Briefly, mouse genomic DNA was extracted from tail biopsies and was subjected to PCR analysis by using primers reported previously (29). The products were digested with Hinc II and were electrophoresed on a 1.5% agarose gel. The mutant allele produced a 460-bp product, whereas the wild-type allele gave a 398-bp product.
All experiments were approved by the University of Adelaide Animal Ethics Committee and were conducted in compliance with the Australian code of practice for the care and use of animals for scientific purposes. ClockΔ19+MEL mutant mice and their wild-type controls (3–7 mice of each sex per time point) aged 2 mo were killed by decapitation every 4 h across 24 h at 0800, 1200, 1600, 2000, 2400, and 0400. An additional group of 6-mo-old ClockΔ19+MEL and wild-type mice (3–7 mice of each sex per time point) were killed at 0800, 1200, and 1600. Blood was collected into heparinized tubes, and plasma was harvested for metabolite and hormone assays. Liver and gastrocnemius muscle were rapidly dissected and were immediately placed in RNAlater (Ambion, Austin, TX) at −20°C until processing (15).
Hormone and metabolite assays.
Plasma glucose and free fatty acids were measured by colorimetric enzymatic analysis on a COBAS Mira automated centrifugal analyzer with the use of kits from Roche Diagnostics and Wako Pure Chemical Industries, respectively. Plasma insulin, adiponectin, and leptin were assayed by using RIA kits obtained from Linco Research (St. Charles, MO).
Intraperitoneal glucose tolerance test.
Wild-type and ClockΔ19+MEL mutant mice aged 2 mo (male and female; n = 6) were fasted overnight and were injected with glucose (1 mg/g body wt; Sigma, St. Louis, MO). Blood (5 μl) was obtained from the tail vein before and 30 and 60 min after glucose administration. Plasma glucose was analyzed by the dehydrogenase method (HemoCue, Angelholm, Sweden). The same animals were subjected to an intraperitoneal glucose tolerance test (IPGTT) 1 wk later but were killed at 30 min postinjection for plasma glucose and insulin determinations. Additional wild-type and ClockΔ19+MEL mutant mice aged 6 mo (n = 6–9 of each sex and each genotype) and 12 mo (females only; n = 6–7) were also subjected to the IPGTT. All IPGTT were conducted 2–3 h after lights were turned on.
Intraperitoneal insulin tolerance test.
After overnight access to food, wild-type (n = 14) and ClockΔ19+MEL (n = 16) male mice aged 6 mo were injected 2–3 h after lights on with insulin (0.75 IU/kg body wt; Actrapid). Blood was obtained as described for glucose determination before and at 30, 60, 90, and 120 min after insulin administration. Additional wild-type (n = 4) and ClockΔ19+MEL (n = 4) female mice aged 6 mo and wild-type (n = 5) and ClockΔ19+MEL(n = 5) female mice aged 24 mo were subjected to an intraperitoneal insulin tolerance test (IPITT).
To investigate the expression of clock and clock-controlled genes throughout a light/dark cycle, total RNA from ∼100 mg liver and muscle from a subset of wild-type and ClockΔ19+MEL mutant mice (2 mo old; n = 3 males and n = 3 females at each time point) was extracted by using 1 ml Tri Reagent (Sigma) according to the manufacturer's protocol. RNA (2 μg in 20 μl) was incubated with 0.4 μg of random hexamer primers (GeneWorks, Adelaide, SA, Australia) at 70°C for 10 min. The tubes were cooled on ice for 5 min, and 8 μl of 5× RT buffer (Invitrogen, Carlsbad, CA), 4 μl of 0.1 M DTT (Invitrogen), and 4 μl of 2′-deoxynucleoside 5′-triphosphates (10 mM; Amersham Pharmacia Biotech, Piscataway, NJ) were added. The tubes were incubated at 43°C for 2 min, 200 units of Superscript III RT (Invitrogen) were added, and tubes were further incubated for 90 min at 43°C and 5 min at 95°C. Ultrapure water (61 μl) was added to each sample to make a final volume of 100 μl cDNA, which was stored at −20°C until further use. Primers were designed with the ABI Prism Primer Express program (Applied Biosystems, Foster City, CA; see Table 1 for primers used). Amplification of cDNA was performed on a GeneAmp 7000 sequence-detection system (Applied Biosystems) by using 5 μl cDNA, 2 μl of 0.625 μM forward and reverse primers, 1 μl water, and 10 μl SYBR green (Applied Biosystems) per well. The samples were amplified in duplicate for 1 cycle of 50°C for 2 min and 95°C for 10 min, followed by 40 cycles of 95°C for 15 s and 60°C for 1 min. An arbitrary threshold of fluorescence was set within the exponential phase of the amplification, and the cycle at which the signal exceeded this threshold was designated as the cycle threshold (Ct).
The expression of each gene within each sample was normalized against β-actin and was expressed relative to a calibrator sample with the use of the formula 2−(ΔΔCt), as described by K. Livak (PE-ABI, Sequence Detector User Bulletin 2). The expression of β-actin did not vary significantly across the six time points studied (ANOVA; P > 0.05). The calibrator sample was designated as the most highly expressed time point for each gene of interest and therefore has a relative expression of 1. Expression data was compared with the wild-type calibrator sample set at 1 to determine differences in the time of peak amplitude and relative differences in total expression of the genes between genotypes.
The effects of IPGTT, IPITT, sex, genotype, time of day, and age and their interactions on plasma hormone, metabolite, and gene-expression data were analyzed by univariate ANOVA with SPSS for Windows (v. 13.0). Where appropriate, estimated marginal means are reported for the various comparisons.
Plasma glucose, free fatty acids, and hormones: 2 mo.
Plasma glucose was unaffected by genotype or time of day (P > 0.05) but varied with sex (P = 0.001; Table S11 and Fig. 1A). Plasma glucose was higher in males than females overall (10.4 ± 0.15 mM vs. 9.4 ± 0.2 mM; P = 0.001) and with wild-type mice (P = 0.013) and ClockΔ19+MEL mice (P = 0.003) separately. Furthermore, in the ClockΔ19+MEL males, plasma glucose was increased in the evening and reduced in the morning compared with wild type, whereas ClockΔ19+MEL females had reduced plasma glucose at midnight compared with wild type (sex × genotype × time of day; P = 0.037; Table S1 and Fig. 1A).
Plasma free fatty acids were reduced in ClockΔ19+MEL mice compared with wild type (1.16 ± 0.05 meq/l vs. 1.53 ± 0.05 meq/l; P = 0.001). Plasma free fatty acids also varied across 24 h (time of day, P = 0.001) and were highest in both wild-type and ClockΔ19+MEL mice 4 h before lights on and in both males and females (Table S1 and Fig. 1B). There were no interactions of sex with time of day or genotype (P > 0.05). Whereas overall females had higher plasma free fatty acids than males (1.4 ± 0.05 meq/l vs. 1.27 ± 0.04 meq/l; P = 0.043), both male and female wild-type mice had similar plasma free fatty acid levels, whereas female ClockΔ19+MEL mice had higher levels than mutant males (P = 0.006).
Plasma insulin varied with time of day (P = 0.038) and sex (P = 0.001) but not genotype (P > 0.05; Table S1 and Fig. 1C). Plasma insulin, however, varied with genotype differently in males and females (sex × genotype; P = 0.009). In ClockΔ19+MEL males, plasma insulin was 0.83 ± 0.13 ng/ml compared with 1.43 ± 0.12 ng/ml in wild-type males, whereas plasma insulin was 0.54 ± 0.14 ng/ml in ClockΔ19+MEL females and 0.40 ± 0.16 ng/ml in wild-type females. Plasma insulin also varied differently between genotype with time of day (genotype × time; P = 0.037), such that it did not change across 24 h in ClockΔ19+MEL males and females and wild-type females (P > 0.05) but did in wild-type males. Thus plasma insulin in wild-type male mice varied across 24 h (P = 0.009), with a threefold increase within 4 h of lights off, whereas in ClockΔ19+MEL males it did not (P > 0.05), remaining constant at ∼0.7 ng/ml.
Plasma insulin-to-glucose ratios were increased in males compared with females (P < 0.0001) and were altered by genotype differently with sex (genotype × sex; P = 0.006), varied with time of day (P = 0.015) and differently with sex (sex × time; P = 0.02), and with genotype differently with sex (sex × genotype; P = 0.006). In males, plasma insulin/glucose varied with genotype (P = 0.006) and varied with time of day (P = 0.001) and differently with genotype (genotype × time; P = 0.06), being substantially reduced in ClockΔ19+MEL males nocturnally.
Plasma adiponectin was increased in ClockΔ19+MEL mice compared with wild-type mice overall (7.0 ± 0.3 μg/ml vs. 5.0 ± 0.3 μg/ml; P = 0.001) and for both sexes (P = 0.001) but did not vary with time of day (P > 0.05; Table S1 and Fig. 1D). Plasma adiponectin levels were higher in females than in males for both genotypes (7.8 ± 0.3 μg/ml vs. 4.2 ± 0.3 μg/ml; P = 0.001). Plasma leptin was not altered by genotype, time of day, or sex (P > 0.05; Table S1, Fig. 1E).
Plasma adiponectin and insulin were related in a curvilinear fashion in wild-type male mice, with higher plasma insulin levels associated with low plasma adiponectin (Fig. 2). This relationship was evident in part because of the high nocturnal insulin levels and the lack of rhythmicity in adiponectin in wild-type male mice. This relationship was not evident in wild-type females and ClockΔ19+MEL male and female mice (Fig. 2).
Plasma glucose, free fatty acids, and hormones: 6 mo.
Because plasma glucose and hormone levels did not change across the 24 h at 2 mo of age (except for the nocturnal insulin increase in wild-type males), only three time points during the light period (0800, 1200, and 1600) were studied in 6-mo-old animals. Plasma metabolite and hormone levels at these times of day in both 2- and 6-mo-old animals were analyzed by ANOVA with genotype, sex, time of day, and age as between-subject factors.
Plasma glucose varied with genotype (ClockΔ19+MEL, 9.5 ± 0.3 mM vs. wild type, 10.6 ± 0.3 mM; P = 0.004) and sex (male, 10.9 ± 0.3 mM vs. female, 9.2 ± 0.3 mM; P = 0.001) but not age (P > 0.05). Plasma glucose also varied with genotype differently in males and females (sex × genotype; P = 0.003), with lower plasma glucose in ClockΔ19+MEL males compared with wild type (9.7 ± 0.4 mM vs. 12.0 ± 0.4 mM), but there was no difference between genotypes in females (ClockΔ19+MEL, 9.3 ± 0.4 mM vs. wild type, 9.2 ± 0.4 mM).
Plasma free fatty acids varied by genotype (P = 0.001) and sex (P = 0.004) but not age (P > 0.05). Plasma free fatty acids also varied with genotype differently with age (genotype × age; P = 0.006). Plasma free fatty acids in ClockΔ19+MEL mice were 1.0 ± 0.1 meq/l and 1.0 ± 0.1 meq/l at 2 and 6 mo of age, respectively, whereas in wild-type mice, plasma free fatty acids decreased from 1.3 ± 0.1 meq/l to 1.0 ± 0.1 meq/l with increasing age.
Plasma insulin varied with genotype (ClockΔ19+MEL, 0.97 ± 0.16 ng/ml vs. wild type, 1.53 ± 0.16 ng/ml; P = 0.021), sex (male, 0.59 ± 0.17 ng/ml vs. female, 1.93 ± 0.15 ng/ml; P = 0.001), and age (2 mo, 0.68 ± 0.16 ng/ml vs. 6 mo, 1.83 ± 0.16 ng/ml; P = 0.001). Plasma insulin also varied differently with sex and age (P = 0.001), increasing from 0.48 ± 0.24 to 0.70 ± 0.24 ng/ml with age in females and to a much greater extent in males, from 0.88 ± 0.22 to 2.97 ± 0.21 ng/ml.
Plasma adiponectin varied with genotype (ClockΔ19+MEL, 9.1 ± 0.5 μg/ml vs. wild type, 6.7 ± 0.5 μg/ml; P = 0.001), sex (male, 4.5 ± 0.5 μg/ml vs. female, 11.4 ± 0.5 μg/ml; P = 0.001), and age (2 mo, 6.4 ± 0.5 μg/ml vs. 6 mo, 9.4 ± 0.5 μg/ml; P = 0.001). Plasma adiponectin was altered differently by genotype in males and females (sex × genotype; P = 0.036). Plasma adiponectin in females was 13.3 ± 0.7 μg/ml in ClockΔ19+MEL mice and 9.4 ± 0.8 μg/ml in wild type, whereas plasma adiponectin in male ClockΔ19+MEL mice was 4.9 ± 0.7 and 4.0 ± 0.6 μg/ml in wild type. Plasma adiponectin varied differently with age in males and females (sex × age; P = 0.001), increasing from 8.4 ± 0.8 to 14.3 ± 0.8 μg/ml between 2 and 6 mo of age in females but not changing in males (4.4 ± 0.6 vs. 4.5 ± 0.6 μg/ml).
Plasma leptin varied with age (0.8 ± 0.6 ng/ml at 2 mo vs. 9.0 ± 0.6 ng/ml at 6 mo; P = 0.001) but not with genotype (P > 0.05) or sex (P > 0.05).
At 2 mo of age, ClockΔ19+MEL males had fasting hypoglycemia compared with wild-type mice (Fig. 3). At 6 mo of age, fasting plasma glucose levels in ClockΔ19+MEL and wild-type males were similar (Fig. 3).
Fasting plasma glucose was similar in 2-mo-old ClockΔ19+MEL and wild-type female mice (P > 0.05; Fig. 3b). At 6 mo of age female ClockΔ19+MEL mice had significantly higher fasting plasma glucose than wild type mice (P < 0.05), whereas at 12 mo of age fasting plasma glucose levels were similar in both genotypes (Fig. 3).
Analysis of males at 2 and 6 mo of age and females at 2, 6, and 12 mo of age revealed a greater plasma glucose response (area under the curve from 0 to 60 min) to a glucose challenge in ClockΔ19+MEL mice compared with wild-type mice (ClockΔ19+MEL, 347 ± 21 units vs. wild type, 252 ± 22 units; P = 0.001) and sex (P = 0.047). Glucose tolerance, as indicated by the area under the glucose curve following a glucose challenge, varied with age (P = 0.003) and varied differently with age depending on sex (females, 151 ± 34, 360 ± 34, and 337 ± 36 units at 2, 6, and 12 mo, respectively, and males, 313 ± 34 and 336 ± 31 units at 2 and 6 mo, respectively). The plasma insulin/glucose ratio 30 min after glucose administration in 2-mo-old female ClockΔ19+MEL mice was higher (P < 0.05) than in wild-type females but was similar in ClockΔ19+MEL and wild-type male mice (data not shown). Table 2 shows the body weights of the mice that underwent glucose tolerance tests, which did not vary with genotype or age for males and females.
At 6 mo of age, administration of insulin reduced plasma glucose in mice by 30 min after injection (Fig. 4). In wild-type male mice, plasma glucose reached a nadir of 15% of the baseline after 30 min, returned to preinjection levels by 60 min, and increased above these over the next 60 min (Fig. 4A). In contrast, in ClockΔ19+MEL male mice, plasma glucose reached a nadir of 42% of preinjection levels 60 min after injection and remained low for another 60 min (Fig. 4A). In contrast, at 6 mo of age, insulin tolerance was similar in wild-type and ClockΔ19+MEL female mice (Fig. 4B). By 24 mo of age, however, insulin tolerance was increased in ClockΔ19+MEL female mice compared with wild-type (Fig. 4C).
Hepatic gene expression.
Expression of liver gck mRNA varied with genotype (0.35 ± 0.06 units in ClockΔ19+MEL mutants vs. 0.67 ± 0.05 units in wild-type mice; P = 0.001) and time of day (P = 0.029; Table S2 and Fig. 5). Peak expression of gck mRNA in female wild-type mice occurred between 2400 and 0400, whereas in male wild-type mice the peak occurred earlier (between 2000 and 2400). Expression of gck in ClockΔ19+MEL mutants changed across the day in females (P = 0.002), with elevated levels between 2000 and 0400, but did not change in males (P > 0.05). The maximum level of expression for male and female ClockΔ19+MEL mice was similar to that at the nadir of expression in the wild-type mice.
Expression of pfkfb3 mRNA varied with genotype (0.12 ± 0.03 units in ClockΔ19+MEL mice vs. 0.38 ± 0.03 units in wild-type mice; P = 0.001) and time of day (P = 0.001) but not sex (P > 0.05). Expression of pfkfb3 across time varied with genotype (genotype × time; P = 0.001; Table S2 and Fig. 5). Peak expression of pfkfb3 occurred in the late light period (2000) in both male and female wild-type mice. Expression of pfkfb3 in the ClockΔ19+MEL mutants did not change across the day and was similar to the nadir of expression of the wild-type mice.
Expression of pepck mRNA varied with genotype (0.18 ± 0.03 units in ClockΔ19+MEL mutants vs. 0.40 ± 0.03 units in wild-type mice; P = 0.001), sex (0.18 ± 0.03 units in males vs. 0.40 ± 0.03 units in females; P = 0.001), and time of day (P = 0.001; Table S2 and Fig. 5). There were also significant genotype × time (P = 0.001), genotype × sex (P = 0.001), or sex × time interactions (P = 0.001). Expression of pepck mRNA varied with time only in wild-type mice (P = 0.001), with peak expression during the late light period (2000) in both males and females. Female wild-type mice had higher levels of pepck expression than males (0.60 ± 0.04 units vs. 0.20 ± 0.04 units; P = 0.001), but there was no effect of sex in ClockΔ19+MEL mutants (0.19 ± 0.04 units vs. 0.16 ± 0.05 units; P > 0.05). Expression of pepck in the ClockΔ19+MEL mutants did not change across the day and was similar to the nadir expression of the wild-type mice.
Skeletal muscle gene expression.
The was no effect of sex on the expression of glut4, pparα, or cpt1 mRNA in the gastrocnemius muscle (P > 0.05; Table S3). Expression of glut4 mRNA in gastrocnemius muscle varied with genotype (0.61 ± 0.04 units in ClockΔ19+MEL mice vs. 0.82 ± 0.04 units in wild-type mice; P = 0.002) and time (P = 0.015; Table S3 and Fig. 6). Expression of glut4 mRNA showed a similar pattern of expression across 24 h of the day in both genotypes, but the change was significant only in ClockΔ19+MEL mice (P = 0.015). Expression of pparα mRNA in muscle did not vary with genotype or time of day (P > 0.05; Fig. 6).
Expression of cpt1 mRNA did not vary with genotype (P > 0.05) but did vary with time in both genotypes (P = 0.001) and was biphasic in both, with peaks at 2000 and 0800. At 2000, the increase in cpt1 mRNA expression was less prominent in ClockΔ19+MEL mice (Fig. 6).
Here we show directly for the first time that impaired peripheral cellular rhythmicity is characterized by impaired glucose tolerance and altered gene expression of molecular determinants of metabolic homeostasis in liver and skeletal muscle. The ClockΔ19+MEL mutant mouse strain utilized in this study is unique in being a melatonin-producing line with a defective Clock gene (16). We have previously shown that the ClockΔ19+MEL mutant mouse produces melatonin rhythmically at night, with peak production occurring just prior to lights on (16). In a recent study (14), we showed that ClockΔ19+MEL mutant mice sustain melatonin rhythmicity in constant darkness for at least 14 days and that the melatonin rhythm can be entrained to a skeleton photoperiod. Furthermore, we reported significant rhythmicity of per2 and PK2 mRNA expression in the SCN of ClockΔ19+MEL mutant mice (14). The detection of significant levels of npas2 mRNA expression in the SCN of both wild-type and ClockΔ19+MEL mutant mice suggested that neuronal PAS domain protein 2 (NPAS2) may be responsible for this rhythmicity, because it has been shown to substitute for Clock in the forebrain (26). Despite the retention of central neuroendocrine rhythmicity in the ClockΔ19+MEL mutant mice, the liver and muscle Bmal1, per1, and per2 gene expression was not rescued by NPAS2 and was arrhythmic (14). Thus the ClockΔ19+MEL mutant provides a model for investigating the effects of retained central neuroendocrine rhythmicity but disrupted peripheral rhythmicity and here is linked to substantially impaired glucose tolerance.
The primary aim of the current study was to use the ClockΔ19+MEL mutant strain to address the role of peripheral circadian rhythmicity and Clock in the maintenance of metabolic homeostasis. The outcomes show that Clock has an important role in metabolic homeostasis, because ClockΔ19+MEL mutant mice had impaired glucose tolerance, and although younger males were hypoglycemic, older females developed hyperglycemia. This poor glucose control may be due in part to impaired insulin secretion, because ClockΔ19+MEL mice had substantially reduced plasma insulin, with reduced plasma insulin/glucose nocturnally in males and during glucose tolerance testing in females. There was additional evidence of altered insulin secretion in ClockΔ19+MEL mice. In wild-type males in the current study, there was a nocturnal surge of insulin that was not reflected by any significant change in plasma glucose, suggesting decreased nocturnal insulin sensitivity. In male NMIR and CBA/T6 mice, a similar sudden two- to threefold elevation in insulin occurred in the late light-early dark period that was sustained through the dark period until around lights on (1, 4). By contrast, in females there was a more modest increase in insulin during the night (1). No circadian change was observed in plasma glucose in males (1, 4) or females (1). Decreased glucose tolerance and lowered insulin sensitivity have been shown to occur during the early rest period in rats (18) and humans (35). The nocturnal insulin surge does not occur in fasted male mice (1), which is an indication that the increase may not be exclusively clock driven but a reflection of nocturnal feeding. Although we did not study feeding rhythmicity in the ClockΔ19+MEL mice, the activity patterns previously reported (16) suggest that feeding would have occurred predominantly at night, yet there was no nocturnal increase in insulin in male ClockΔ19+MEL mice. In ClockΔ19 (C57BL/6J) mutants, there was a significant change in feeding patterns, with a greater proportion of feeding occurring during the light period (34). We hypothesize that the apparent lack of feeding-induced insulin secretion in the ClockΔ19+MEL male mice may be due to an insulin-secretory defect and/or loss of pancreas rhythmicity (19). Consistent with this, the male mutant mice did not mount a hyperinsulinemic response to glucose, in contrast to females.
The ClockΔ19 mutation also had an impact on the rhythmic expression of clock-sensitive genes in both liver and skeletal muscle. In the liver, the expressions of gck, pfkfb3, and pepck mRNA were arrhythmic in ClockΔ19+MEL mutant male and female mice. Furthermore, the overall levels of expression of these key metabolic enzyme genes were reduced, implying that both glycolysis and gluconeogenesis would be impaired in the mutants, and the latter may partly explain the hypoglycemia observed in male mutant mice at least. In skeletal muscle, the expression of glut4 mRNA was reduced, whereas the patterns of pparα and cpt1 mRNA expression were unaltered in ClockΔ19+MEL mutant mice. This may limit insulin responsiveness of glucose uptake by skeletal muscle and may contribute to the impaired glucose tolerance. One contribution to reduced glut4 mRNA in skeletal muscle of ClockΔ19+MEL mice could be impaired insulin signaling; however, whole body insulin tolerance was enhanced in males at 6 mo of age and in females at 24 mo of age. This suggests that if skeletal muscle is insulin resistant in ClockΔ19+MEL mutant mice as indicated by reduced GLUT4 expression, then insulin sensitivity in terms of glucose metabolism may be enhanced elsewhere, presumably in liver. If so, the latter could help explain the reduced pepck expression and hypoglycemia observed at least in male ClockΔ19+MEL mutant mice. Direct measurement of the impact of the ClockΔ19+MEL mutation on tissue-specific insulin sensitivity is needed to resolve these possibilities. Importantly, these adverse changes occur in the absence of obesity and with other metabolic and hormonal changes that normally enhance insulin sensitivity and secretion, that is, reduced plasma free fatty acids and increased plasma adiponectin.
The results of the current study have some similarities to, as well as some important differences from, previous reports on homeostasis in ClockΔ19 mutant mice (20, 28, 34). Rudic et al. (28) first reported that ClockΔ19 mutant mice had impaired conversion of pyruvate to glucose, together with alterations in liver PEPCK enzyme activity, which is indicative of altered gluconeogenesis. Insulin administration caused significantly greater hypoglycemia in their ClockΔ19 mutant mice than the wild type. The mice used in these studies were male ClockΔ19 (C57BL/6J) mutant mice (D. Rudic, personal communication). As shown in the current study, 6-mo-old male ClockΔ19+MEL mutant mice and 24-mo-old female ClockΔ19+MEL mutant mice were also extremely sensitive to exogenous insulin and had a poor counterregulatory response compared with the wild-type mice. Turek et al. (34) reported that male ClockΔ19 (C57BL/6J) mutant mice were hyperphagic and obese and developed a metabolic syndrome of hyperlipidemia, hepatic steatosis, hyperglycemia, hypoinsulinemia, and mild hyperleptinemia. Recently, Oishi et al. (20) reported that ClockΔ19 (Jcl:ICR) mice had low plasma free fatty acid levels and normal plasma glucose, insulin, and leptin levels. Interestingly, these ClockΔ19 (Jcl:ICR) mutants appeared to be protected from high-fat-diet-induced obesity through disrupted fat absorption. It was of interest that a rhythm in plasma free fatty acid levels persisted in both the wild-type and ClockΔ19 mutant mice of both sexes in our study. A similar nocturnal increase in plasma free fatty acids and triacylglycerol has been reported previously in 129/SV mice (24), and an sp-nocturnal increase in triacylglycerol in CBA/T6 mice (4), but not Jcl:ICR mice (20), has also been reported. These differences may again be a reflection of strain differences in feeding time. In the current study, unlike ClockΔ19 (C57BL/6J) mutant mice (34), neither male nor female ClockΔ19+MEL mutant mice were obese, and they mostly had low or normal fasting plasma glucose rather than hyperglycemia, low plasma free fatty acids rather than hyperlipidemia, and normal plasma leptin rather than hyperleptinemia. Nevertheless, the ClockΔ19+MEL mutant mice did exhibit signs of metabolic dysfunction, particularly with respect to their impaired glucose tolerance. Preliminary experiments with male ClockΔ19+MEL mutant mice maintained on a high-fat diet (21% fat by weight) from weaning to 2 mo of age have shown that the mutants have significantly increased epigonadal fat pad weight (D. J. Kennaway, unpublished results). The relative weight of epigonadal fat compared with body weight was, however, not significantly different between male wild-type and ClockΔ19+MEL mutant mice. Thus it is unlikely that the metabolic phenotype we report is due to altered fat adsorption as occurred in the ClockΔ19 (Jcl:ICR) mutants.
These discrepancies highlight the impact of the genetic background on metabolic function in laboratory mouse strains. C57BL/6J mice are widely used in metabolism studies because they are obesity prone (3) and are more glucose intolerant than other strains (13), which is exacerbated by a high-fat diet. Recently, Toye et al. (32) showed that the glucose and insulin response to glucose in C57Bl/6J mice was significantly poorer than in C3H/HeH mice. Furthermore, insulin secretion in response to glucose was impaired in isolated islets from C57Bl/6J mice compared with C3H/HeH mice (32). Toye et al. (32) suggested that the strain differences were due in part to altered function of at least three genes compared with mice that are considered normal (i.e., C3H/HeH), including nicotinamide nucleotide transhydrogenase (nnt) and glucokinase (gck). In the case of gck, C57BL/6J mice had significantly lower glucokinase enzyme activity than the C3H/HeJ mice despite the failure to identify any functional sequence changes that might account for it. This rate-limiting enzyme is expressed rhythmically in liver (e.g., this study) and is modulated by Clock and hepatocyte nuclear factor-3 (23). The defect in gck in C57BL/6J mice (32) therefore further complicates the phenotype of the ClockΔ19 (C57BL/6J) mutant mice (28, 34). Not only is there likely to be loss of rhythmicity of gck mRNA expression and enzyme activity, but lower-than-normal enzyme activity is also likely. The ClockΔ19+MEL mice used in the current study were derived from a ClockΔ19 (BALB/c) line that was selectively crossed with a CBA strain, which has a similar origin to C3H mice and is likely to have comparably robust metabolic homeostasis that was nevertheless impaired by clock disruption.
The limited impact on fed and fasting plasma glucose levels and the enhanced whole body insulin tolerance of clock disruption observed here may be a consequence in part of the elevated adiponectin secretion detected in both male and female ClockΔ19+MEL mutants. Adiponectin is an insulin sensitizer, and therefore in an environment of impaired insulin secretion the hormone may compensate for any insulin-secretory defect. Male and female ClockΔ19+MEL mutants have higher adiponectin levels than wild-type mice (30 and 60%, respectively), whereas leptin is unaltered. These circulating adipocytokine patterns suggest the presence of predominantly small immature adipocytes in the mice lacking a functional Clock gene. Together with the recent outcomes of in vitro studies (30), this suggests that the ClockΔ19 mutation impairs adipocyte differentiation and maturation.
The studies on clock-gene disruption and metabolic homeostasis provide compelling evidence of a relationship between circadian rhythms and the glucose/insulin and adipoinsular axes. Here we show directly for the first time that disruption of peripheral rhythms impairs glucose tolerance. People are increasingly being exposed to employment conditions that may disrupt circadian rhythms in hormone secretion, sleep, and organ function. There is emerging evidence that these shift workers are at higher risk of obesity, cardiovascular disease, and metabolic syndrome. Not every shift worker develops health complications as a result of the disruptive lifestyle, and it may be that rhythm disruption, together with polymorphisms in enzymes, hormones, receptors, or intracellular signaling molecules, increases the risk. Together with previous findings in other murine models, there are indications that the interactions between genetic background and circadian rhythm disruption are clearly important in determining the severity of the metabolic impairment.
These studies were supported in part by a Strategic Research Development grant to D. J. Kennaway from the Faculty of Health Sciences, University of Adelaide.
We thank Dr. Miles DeBlasio for performing the plasma glucose and free fatty acid analyses.
↵1 Supplemental tables are available online at the American Journal of Physiology-Regulatory, Integrative, and Comparative Physiology website.
The costs of publication of this article were defrayed in part by the payment of page charges. The article must therefore be hereby marked “advertisement” in accordance with 18 U.S.C. Section 1734 solely to indicate this fact.
- Copyright © 2007 the American Physiological Society