The authors have declared that no competing interests exist.
Conceived and designed the experiments: GY TY. Performed the experiments: GY TA ZJ TY. Analyzed the data: GY TA ZJ TY. Contributed reagents/materials/analysis tools: DM RM. Wrote the paper: GY TY.
Compelling evidence from both human and animal studies suggests a physiological link between the circadian rhythm and metabolism but the underlying mechanism is still incompletely understood. We examined the role of PPARγ, a key regulator of energy metabolism, in the control of physiological and behavioral rhythms by analyzing two strains of whole-body PPARγ null mouse models. Systemic inactivation of PPARγ was generated constitutively by using Mox2-Cre mice (MoxCre/flox) or inducibly by using the tamoxifen system (EsrCre/flox/TM). Circadian variations in oxygen consumption, CO2 production, food and water intake, locomotor activity, and cardiovascular parameters were all remarkably suppressed in MoxCre/flox mice. A similar phenotype was observed in EsrCre/flox/TM mice, accompanied by impaired rhythmicity of the canonical clock genes in adipose tissues and liver but not skeletal muscles or the kidney. PPARγ inactivation in isolated preadipocytes following exposure to tamoxifen led to a similar blockade of the rhythmicity of the clock gene expression. Together, these results support an essential role of PPARγ in the coordinated control of circadian clocks and metabolic pathways.
Most living organisms display behavioral and physiological rhythms in response to the daily changes imposed by rotation of the earth. The rhythms are driven by internal molecular clocks and can be reset by environmental light-dark cycles. The core molecular clock is composed of transcriptional activators and repressors that are assembled into feedback loops
The master regulator of circadian rhythms resides in the suprachiasmatic nucleus (SCN) of the hypothalamus in mammals
Growing evidence has emerged to support a physiological link between the circadian rhythms and metabolism. Epidemiological studies showed that perturbations in circadian rhythms in humans involving a shift-working population of 27,485 people are associated with increased risk of obesity and hyperlipidemia
Peroxisome proliferator-activated receptor gamma (PPARγ) is a nuclear receptor that heterodimerizes with the retinoid X receptor (RXR) and binds to PPAR responsive elements in the regulatory region of target genes involved in various aspects of metabolism. PPARγ is most abundantly expressed in the adipose tissue where it plays a pivotal role in driving adipocyte differentiation and maintaining adipocyte specific functions, such as lipid storage in the white adipose tissue and energy dissipation in the brown adipose tissue
We generated MoxCre/flox mice by crossing floxed PPARγ mice with a transgenic line expressing Cre recombinase under the control of Mox-2 promoter as previously described
The canonical diurnal metabolic parameters including VO2 (A), VCO2 (B), heat production (C), food (D) and water (E) intake, and locomotor activity (F) were measured in MoxCre/flox mice. MAP (G&H), HR (I& J) and locomotor activity (K&L) were recorded using radiotelemetry. N = 5–6 in each group. Data are mean ± SE. *,
In light of the high lethality rate in MoxCre/flox mice, we generated an inducible whole-body PPARγ deletion by using the tamoxifen system. Mating of germ-line floxed PPARγ mice with tamoxifen-inducible Cre-expressing mice produced offspring with inducible homozygous EsrCre/flox mice, which had normal phenotype. To inactivate PPARγ gene, we treated adult EsrCre/flox mice with daily tamoxifen injections for 5 days. We performed PCR analysis of DNA recombination in various tissues from PPARγf/f and EsrCre/flox mice with or without tamoxifen treatment. We performed PCR analysis of DNA recombination in various tissues from these mice. The DNA recombination was reflected by the loss of the 2193-bp products derived from the floxed allele and appearance of the 260-bp products derived from the recombined allele. The untreated EsrCre/flox mice exhibited partial DNA recombination in most of the tissues possibly reflecting the endogenous steroid activity. After tamoxifen treatment, EsrCre/flox mice had nearly complete DNA recombination in all tissues examined (termed EsrCre/flox/TM) (
(A) Schematic illustration of primer design for detection of the foxed PPARγ allele (primers S1 and A1). (B) PCR analysis of the floxed PPARγ allele in various tissues of PPARγf/f/TM (left), EsrCre/flox (middle) and EsrCre/flox/TM (right) mice using primers S1 and A1.
Tamoxifen-treated PPARγf/f mice (termed PPARγf/f/TM) served as controls. EsrCre/flox/TM mice had normal body weight and were grossly indistinguishable from the floxed controls. Under regular light/dark cycle, PPARγf/f/TM, EsrCre/flox, and EsrCre/flox/TM were placed in metabolic cages (Hatteras Instruments) for measurement of diurnal variations of food and water intake, and feces and urine production. Both PPARγf/f/TM and EsrCre/flox groups displayed obvious day-night variations in food intake and feces production. In contrast, EsrCre/flox/TM mice nearly lost the rhythms of these parameters (
Food (A), and feces (B) were measured during the light and dark phases. MAP (C&D), HR (E&F) and locomotor activity (G&H) were recorded using radiotelemetry for consecutive 2 days. N = 5–6 in each group. Data are mean ± SE. *,
MAP (A&B), HR (C&D) and locomotor activity (E&F) were recorded using radiotelemetry for consecutive 2 days under constant darkness. Black bars correspond to the period of darkness, and the gray bars indicate the period of subjective light under constant darkness. N = 5–6 in each group. Data are mean ± SE. *,
We performed qRT-PCR analyses of canonical clock genes in the fat, liver, hypothalamus and skeletal muscle of PPARγf/f/TM and EsrCre/flox/TM mice at various circadian time points under regular light/dark cycles. As expected, adipose expression of canonical clock genes in PPARγf/f/TM mice exhibited robust variations, with Bmal1 and MOP4 peaking at CT20, and Per1, Cry2, and Rev-erbα at CT8, and Per2 and Per3 at CT14 (
PPARγf/f/TM and EsrCre/flox/TM mice were sacrificed at 6-h intervals. The epididymal fat and liver were harvested for qRT-PCR analysis of canonical clock gene expression. For each gene, the lowest level of mRNA expression was set to 1. N = 6–8 per group. Data are mean ± SE. *
PPARγf/f/TM and EsrCre/flox/TM mice were sacrificed at 6-h intervals. The hypothalamus and skeletal muscle were harvested for qRT-PCR analysis of canonical clock gene expression. For each gene, the lowest level of mRNA expression was set to 1. N = 6–8 per group.
Fat | Liver | |||
PPARγf/f/TM | EsrCre/flox/TM | PPARγf/f/TM | EsrCre/flox/TM | |
Bmal1 | 8.59±1.22 | 2.74±0.35 |
27.23±4.33 | 7.89±1.77 |
CLOCK | 1.47±0.12 | 1.37±0.21ns | 2.17±0.36 | 2.30±0.22ns |
MOP4 | 5.671±1.453 | 3.191±0.5312ns | 48.75±3.96 | 26.95±4.49 |
Per1 | 1.66±0.14 | 1.16±0.28ns | 7.72±0.53 | 2.48±0.69 |
Per2 | 6.49±0.91 | 4.57±1.01ns | 14.03±1.69 | 2.76±0.40 |
Per3 | 2.84±0.66 | 2.00±0.54ns | 26.58±2.26 | 4.08±0.77 |
Cry1 | 3.36±0.71 | 3.96±0.58ns | 5.48±0.71 | 1.99±0.37 |
Cry2 | 2.32±0.22 | 1.37±0.27 |
2.58±0.44 | 2.12±0.18ns |
Rev-erbá | 19.89±2.80 | 6.58±1.11 |
39.88±5.99 | 49.59±9.82ns |
Shown are mean ± SE.
ns: no significant difference.
To investigate whether PPARγ directly regulated the clock system, we used the tamoxifen system to produce PPARγ deletion in primary preadipocytes and examined the consequence in expression of the clock genes. Exposure of EsrCre/flox preadipocytes to 4-hydroxytamoxifen (4-OHT) for 2 days resulted in 83% decrease of total PPARγ mRNA level (
(A) qRT-PCR analysis of total PPARγ expression in preadipocytes. (B) qRT-PCR analysis of PPARγ2 expression in preadipocytes.
4-OHT-treated primary preadipocytes from PPARγf/f and EsrCre/flox mice (termed PPARγf/f/OHT and EsrCre/flox/OHT, respectively) were stimulated with 50% horse serum (A) or 10 µM 15d-PGJ2 (B) for 2 h. Cells were harvested at 4-h intervals for 48 h and subjected to RNA extraction and qRT-PCR analysis of the canonical clock gene expressions. For each gene, the lowest level of mRNA expression was set to 1. N = 4 in each group/time point. Shown are mean ± SE. *
15-deoxy-Δ12,14-prostaglandin J2 (15d-PGJ2), a natural ligand of PPARγ, has been reported as an entrainment factor for the circadian clocks
We employed ELISA to determine urinary excretion of 15d-PGJ2. The specificity of 15d-PGJ2 ELISA has been validated by testing cross activities with other prostanoids; the values were: 0.11% for PGD2, 0.1% for delta12-PGJ2, 0.05% for PGA2, and 0% for TXA2, PGI2, and PGE2. Urinary 15d-PGJ2 excretion was found to exhibit robust diurnal variation that was effectively attenuated by indomethacin and the COX-2 inhibitor SC-560; the COX-2 inhibitor NS-398 reduced the baseline level of urinary 15d-PGJ2 but failed to influence the magnitude of the diurnal variation (
The compounds were administered from diet and dosing was based on estimated food intake. N = 4–7 in each group. Data are mean ± SE.
A large body of evidence from human and animal studies has demonstrated that the regulation of molecular clocks is linked to pathways of energy metabolism. A better understanding of the molecular basis of the relationship between the molecular clocks and metabolism may shed light on the etiologies as well as therapies of metabolic diseases. PPARγ is a key regulator of energy metabolism and is best known for serving as a therapeutic target for management of type 2 diabetes. Despite the intensive investigation, the mechanism of how PPARγ achieves an integrative control of energy metabolism is not fully understood. We hypothesize that PPARγ may function as an integrator of the molecular clocks and metabolism. Since this function may involve the multi-faceted interaction of PPARγ in multiple tissues, the use of generalized knockout models is necessary. The germline knockout of PPARγ produces the embryonic lethality due to abnormal placenta vascularization, hepatic dysfunction and multiple hemorrhages
The most novel finding of the present study was the robust alteration of circadian rhythms in a spectrum of physiological, metabolic and behavioral parameters of the two strains of systemic PPARγ null mice. Under regular light/dark cycles, MoxCre/flox mice displayed a nearly complete loss of circadian rhythms of food and water intake, metabolism (VO2, VCO2, and heat production), cardiovascular parameters (BP and HR) and locomotor activity. The variations of most of these parameters in EsrCre/flox/TM mice were blunted under both light/dark or constant darkness conditions with an exception for the locomotor activity. The rhythm of the locomotor activity in these mice remained intact under light/dark cycle but was diminished under constant darkness. The reason for the difference in the rhythm of the locomotor activity between the genotypes is unclear but one confounding factor may come from the high lethality rate in young MoxCre/flox mice. Despite this limitation, the circadian phenotypes of the two strains of PPARγ null models generated by different methods are largely consistent, establishing an essential role of PPARγ in the control of rhythmicity of behavior and physiology. Emerging evidence has demonstrated a physiological link between the circadian rhythms and metabolism
The robust circadian phenotype of the two strains of whole-body PPARγ null mice suggests a non-redundant role of this nuclear receptor in the circadian regulation. Indeed, emerging evidence supports a direct coupling of PPARγ with Bmal1. Our previous study demonstrates that PPARγ directly regulates Bmal1 transcription in the vascular cells, thereby regulating the cardiovascular rhythms
Our results also suggest that besides direct transcriptional regulation of Bmal1, PPARγ may determine the robustness of Bmal1 oscillation via Rev-erbα, a negative regulator of Bmal1
The study of Nakahata et al. employed an unbiased approach, namely the in vitro real-time oscillation monitoring system to identify unknown entrainment factors for clock genes in cultured 3T3 cells (Nakahata et al. 2006). Among 299 peptides and bioactive lipids tested in this study, 15d-PGJ2 was identified as a novel entrainment factor that produces the most robust effects on rhythmicity. In agreement with this observation, we found that a single treatment with 15d-PGJ2 produced robust rhythmicity. However, a difference between the two studies exists concerning the involvement of PPARγ. The present study demonstrated that tamoxifen-induced PPARγ deletion remarkably blunted the rhythmicity in preadipocytes exposed to 15d-PGJ2. This finding argues against the Nakahata's study reporting independence of the 15d-PGJ2 action from PPARγ based on the use of the PPARγ antagonist DW9662. Of note, the similar blockade of clock gene expression was observed in PPARγ-deficient preadipocytes exposed to 50% horse serum and 15d-PGJ2 with a few exceptions. For example, the blockade of Cry1 and MOP4 by PPARγ deletion was observed after serum shock but not after 15d-PGJ2. These results suggest a different mechanism responsible for regulation of Cry1 and MOP4 under the current experimental condition.
15d-PGJ2 was initially identified as an endogenous PPARγ ligand based on data from several
In summary, the two strains of whole-body PPARγ null mice consistently develop blunted physiological and behavioral rhythms. The impaired rhythmicity of the canonical clock genes in the null mice was found in adipose tissues and liver but not skeletal muscles or the kidney. PPARγ inactivation in isolated preadipocytes resulted in a similar blockade of the rhythmicity. Together, our studies have defined PPARγ as a key integrator of molecular clocks and metabolism.
PPARγf/f mice contain two loxP sites inserted into intron 1 and 2 of the PPARγ gene flanking the critical exon 2 (Akiyama et al. 2002). The floxed mice were crossed with MoxCre mice
Tamoxifen stock solution was prepared as previously described
DNA recombination of the PPARγ gene was evaluated in the brain, heart, lung, liver, pancreas, stomach, intestine, spleen, kidney, muscle and fat from PPARγf/f/TM, EsrCre/flox and EsrCre/flox/TM mice. Primers flanking the 2 loxP sites and exon 2 were used to amplify a product of 2193 bp from the floxed allele and 260 bp from the recombined allele.
Regular metabolic cages (Hatteras Instruments, Cary, NC) were used for urine and feces collections and also for measurement of food and water intake during the light and dark phases. Indirect calorimetry was performed with a four-chamber Oxymax system (Columbus Instruments, Columbus, OH). Animals were allowed to adapt to the metabolic chamber for 4 h and then food and water intake, movement, oxygen consumption (VO2), carbon dioxide output (VCO2) and heat production were measured every 15 min for 3 days from individually housed mice.
Under general anesthesia, the radiotelemetric device (model No. TA11PA-C20, DSI, MN) was implanted through catheterization of the carotid artery as previously described
PPARγf/f/TM and EsrCre/flox/TM mice were killed at 6 hr intervals of 24 hr. The fat, liver, skeletal muscle and kidney were harvested for qRT-PCR analysis of canonical clock genes including Bmal1, CLOCK, MOP4, Cry1–2, Per1–3 and Rev-erbα. The primer sequences are listed in supplemental
Gene name | Sense Primer (5′-3′) | Antisense Primer (5′-3′) | Accession No. |
Bmal-1 |
|
|
NM_007489 |
CLOCK |
|
|
NM_007715 |
Cry1 |
|
|
NM_007771 |
Cry2 |
|
|
NM_009963 |
Per1 |
|
|
NM_011065 |
Per2 |
|
|
NM_011066 |
Per3 |
|
|
NM_011067 |
MOP4 |
|
|
NM_008719 |
Rev-erbα |
|
|
NM_145434 |
PPARγ |
|
|
NM_001127330 |
PPARγ2 |
|
|
NM_011146 |
GAPDH |
|
|
M32599 |
White adipose tissues from 3-week old PPARγf/f and EsrCre/flox mice were used for preadipocyte culture. The epididymal and inguinal fat depots were dissected, minced, and transferred to a Krebs-Ringer buffer (Sigma, K4002) containing 15 mM sodium bicarbonate, 10 mM HEPES, 2 mM sodium pyruvate and 1% BSA (pH 7.4). Collagenase type I (Worthington Biochemical) was added at 2 mg/ml. The tissues were shaken at 100 rpm, 37°C for 60 min. The digest was filtered through a 70-µm nylon filter (BD Falcon). The flow-through was centrifuged (100× g, 5 min) and the cell pellet was suspended in the DMEM. The cells were recentrifuged, resuspended in regular medium, 5% fetal bovine serum (FBS)-DMEM and cultured in flasks. Culture medium was exchanged after 24 hours and every 2 days thereafter. Before experiments, cells were seeded into 24-well plates at 1×105/well. After confluence, cells were treated with 100 nM 4-hydroxytamoxifen (4-OHT, Sigma) for 2 days followed by medium exchange with fresh regular medium. Then the cells were challenged with serum shock or 15d-PGJ2. Briefly, at time = 0, the medium was exchanged with DMEM supplemented with 50% horse serum or 10 µM 15d-PGJ2, and after 2 hr, this medium was replaced with regular medium. At the indicated times, the cells were harvested in TRI Reagent (Applied Biosystems) and applied for RNA extraction. These RNA samples were used for qRT-PCR analysis of circadian genes.
PPARγf/f mice treated with DMSO (vehicle), indomethacin (Indo) (5 mg/kg/d), SC-560 (30 mg/kg/d), or NS-398 (5 mg/kg/d). The compounds were administered from diet and dosing was based on estimated food intake. After treatment for 3 days, urine was collected during the light phase (ZT0–12) and dark phase (ZT12–24) and was stored at −80°C before the assays. Urinary 15d-PGJ2 were measured by using a commercial EIA kit (Assay Designs, Ann Arbor, MI).
All values are presented as mean ± SE. ANOVA and Bonferroni post-tests were used for comparisons among multiple groups and the unpaired Student's t test for comparisons between two groups. Differences were considered to be significant when the